AD VANCES IN
AGRONOMY VOLUME 28
CONTRIBUTORS TO THIS VOLUME
RODNEY J. ARKLEY A.
v. BARKER
JOHN
E. BEGG
J. M. BR...
31 downloads
1726 Views
21MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
AD VANCES IN
AGRONOMY VOLUME 28
CONTRIBUTORS TO THIS VOLUME
RODNEY J. ARKLEY A.
v. BARKER
JOHN
E. BEGG
J. M. BREMNER
C.M. DONALD L. T. EVANS E. A. N. GREENWOOD J. HAMBLIN
R. D . HAUCK D . N . MAYNARD
P. L. MINOTTI N.
H. PECK
G . C.M.SAGE NEILC. TURNER INDRA
K.VASIL
I. F. WARDLAW
ADVANCES IN
AGRONOMY Prepared under the Auspices of the
AMERICANSOCIETY OF AGRONOMY
VOLUME 28 Edited by N. C . BRADY International Rice Research Institute Manila, Philippines
ADVISORY BOARD
w. L COLVILLE, CHAIRMAN G. W. KUNZE D. G. BAKER D. E WEIBEL G. R DUTT H J. GORZ M, STELLY, EX OFFICIO, ASA Headquarters 1976
ACADEMIC PRESS 0 New York
San Francisco
London
A Subsidiary of Harcourt Brace Jovanovich, Publishers
COPYRIGHT 0 I 976,w ACADEMIC PRESS, INC. ALL RIGHTS RESERVED. NO PART O F THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN A N Y FORM OR BY ANY MEANS, ELECTRONIC O R MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR A N Y INFORMATION STORAGE AND RETRIEVAL SYSTEM, WlTHOUT PERMISSION IN WRITING FROM THE PUBLISHER.
ACADEMIC PRESS, INC.
111 Piftb Avenue, New York, New York 10003
United Kingdom Edition published by ACADEMIC PRESS, INC. (LONDON) LTD. 24/28 Oval Road,
London N W l
LIBRARY OF CONGRESS CATALOG CARD
NUMBER:50-5598
ISBN 0-12-000728-2 PRINTED IN THE UNITED STATES OF AMERICA
CONTENTS
............................ PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CONTRIBUTORS TO VOLUME 28
ix xi
NITROGEN STRESS IN PLANTS
E . A. N . Greenwood
I . Introduction .......................................... 1 Quantitative Concepts of Nutrient Deficiency . . . . . . . . . . . . . . . . . 2 Nitrogen Stress ........................................ 7 Factors Affecting Nitrogen Stress .......................... 14 Alternative Evaluators of Nitrogen Stress .................... 18 Applications .......................................... 28 Conclusions and Aspirations .............................. 34 References ............................................ 34
I1. 111. IV . V. VI . VII .
STATISTICAL METHODS IN SOIL CLASSIFICATION RESEARCH
Rodney J . Arkley
I . Introduction: Objectives and Problems of Soil Classification
.....
I1. Numerical Taxonomy or Cluster Analysis of Soils .............. I11. Ordination of Soils .....................................
IV. Soil as an Anisotropic Entity .............................. V . Statistical Methods for Comparing Classification ............... VI . Conclusions and Evaluation ............................... References ............................................
37 39 54 59 63 64 68
NITRATE ACCUMULATION IN VEGETABLES
D. N . Maynard. A. V. Barker. P . L.Minotti. and N . H .Peck I . Introduction .......................................... 71 I1. Hazards of Nitrate and Nitrite to Human Health ............... 72 I11. Factors Affecting Nitrate Accumulation ..................... 77
IV. Nitrate Concentrations in Vegetables ....................... V . Conclusions ........................................... References ............................................ V
99 113 114
vi
CONTENTS
THE PROGRESS. PROBLEMS. AND PROSPECTS O F PLANT PROTOPLAST RESEARCH
Indra K . Vasil I. I1. 111. IV. V.
Introduction .......................................... Isolation of Protoplasts .................................. Culture of Protoplasts ................................... Protoplasts and the Genetic Modification of Plants ............. TheFuture ........................................... References ............................................
119 120 126 135 152 153
CROP WATER DEFICITS
John E . Begg and Neil C.Turner
.......................................... ..................................... Measurement of Crop Water Status ......................... Effects of Water Deficits on Crop Growth and Development ..... Adaptation to Water Deficits .............................. Effects of Water Deficits on Yield .......................... Water Use Efficiency ....................................
I . Introduction
I1. Evapotranspiration
.
111
161 163 167 170 182 188 195
IV. V. VI . VII . VIII. Difference in Response of Plants Grown Under Controlled Conditions and in the Field ............................... 202 205 IX. Summary and Conclusions ............................... References ............................................ 207
USE OF TRACERS FOR SO1 L AND FERTILIZER NITROGEN RESEARCH
R . D. Hauck and J .M . Bremner I . Introduction .......................................... I1. Assumptions .......................................... 111. Advantages and Disadvantages of Nitrogen Tracer Techniques .... IV . Determination of Nitrogen Isotopes ........................ V. Sources and Cost of Nitrogen Tracer Materials ................ VI . Use of Nitrogen Tracer Materials ........................... VII . Perspective ........................................... References ............................................
219 223 225 226 239 242 260 261
CONTENTS
vii
NUCLEO-CYTOPLASM IC RELATIONSHIPS IN WHEAT
. .
G. C M Sage 1. Introduction
.......................................... I1. Cytoplasmic Variation in Wheat ........................... 111. The Genetics of Fertility Restoration ....................... IV . Cytoplasmic Effects Other than Male Sterility ................ V . The Biological Basis of Nucleo-cytoplasmic Interactions ......... VI . Cytoplasmic Variation in the Absence of Male Sterility ......... VII . Conclusion ........................................... References ............................................
267 268 274 286 290 295 296 297
ASPECTS OF THE COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
. .
L T Evans and I . F.Wardlaw
I . Introduction .......................................... I1. Origins and Adaptation .................................. 111. Reproductive Development ............................... IV. Root Growth and Nutrient Use ............................ V . Canopy Growth ........................................ VI . Leaf Photosynthesis .................................... VII . Canopy Photosynthesis .................................. VIII . Translocation ......................................... IX . GrainGrowth ......................................... X . Limiting Stages in the Life Cycle ........................... XI . Conclusion ........................................... References ............................................
301 303 305 310 315 317 323 329 335 340 349 350
THE BIOLOGICAL YIELD AND HARVEST INDEX OF CEREALS AS AGRONOMIC AND PLANT BREEDING CRITERIA
C. M . Donald and J . Hamblin
I . Introduction .......................................... 361 I1. The Relationship of Biological Yield. Grain Yield. and Harvest Index to Each Other and to Other Plant Characteristics ......... 364
viii
CONTENTS
111. The Influence of Environmental Factors ..................... IV. Biological Yield and Harvest Index as Criteria in Cereal Breeding . . V . Concluding Comments .................................. References ............................................ SUBJECTINDEX
.......................................
375 390 402 404 407
CONTRIBUTORS Numbers in parentheses indicate the pages on which the authors’ contributionsbegin.
RODNEY J. ARKLEY (37), Department of Soils and Plant Nutrition, College of Natural Resources, University of California,Berkeley, California A. V . BARKER (71), Department of Plant and Soil Sciences, University of Massachusetts, Amherst, Massachusetts JOHN E. BEGG (161), CSIRO Division of Plant Industry, anbema, A.C.T., Australia J . M. BREMNER (219), Department of Agronomy, Iowa State University, Ames, Iowa C. M. DONALD (361), Waite Agricultural Research Institute, University of Adelaide, South Australia, and Department of Applied Biology. University of Cambridge, England L. T. EVANS (301), Division of Plant Industry, CSIRO, Canbema, A.C. T.,Australia E. A. N . GREENWOOD (l), Division of Land Resources Management, CSIRO, Floreat Park, Western Australia J. HAMBLIN* (361), Waite Agricultural Research Institute, University of Adelaide, South Australia, and Department of Applied Biology, University of Cambridge, England R. D. HAUCK (219), Division of Agricultural Development, Tennessee Valley Authority, Muscle Shoals, Alabama D. N. MAYNARD (71), Department of Plant and Soil Sciences, University of Massachusetts, Amherst, Massachusetts P. L. MINOTTI (7 l), Department of Vegetable Crops, Cornell University,Zthaca, New York N. H. PECK (71), Department of Seed and Vegetable Sciences, New York State Agricultural Experiment Station, Geneva, New York G. C. M. SAGE (2671, Plant Breeding Institute, Cambridge, England NEIL C. TURNER (161), CSIRO Division of Plant Industry, Canbema, A.C.T., Australia INDRA K. VASIL (119), Department of Botany, University of Florida, Gainesville, Florida I. F . WARDLAW (301), Division of Plant Industry, CSIRO, Canberra, A.C.T.. Australia
*Present address: Department of Agriculture, Perth, Western Australia.
ix
This Page Intentionally Left Blank
PREFACE During 1976, an event occurred which is of great significance to all mankind. At some location, probably in one of the developing countries where population growth is high, a child was born, the world’s four billionth human being. T h ~ s event, which was easily predictable considering national birth and death rates, calls our attention to the ever-present race between population and food supplies. It gives reason for the increased focus of the world’s scientific community on food production and it justifies the attention that has been given by crop and soil scientists to this important human problem. Rising costs of energy are a second area of worldwide human concern, an area that has marked effects on man’s ability to feed himself. Modern agricultural technology, developed over a long period of relatively low energy costs, is generally high in its energy requirements. Sudden increases in energy costs have caused a reexamination of the high energy demanding concepts of modern agriculture and have accelerated attempts to obtain high food production rates using practices with modest energy requirements. This volume contains reviews of the research contributions of crop and soil scientists to the solution of each of these worldwide problems. Attention is given to nitrogen, an element essential for crop production and an element whose supply is markedly affected by energy costs. Nitrogen is supplied either through biological fixation in the field or by commercial fertilizers. In either case, there is need for research findings on this element, its efficiency of utilization and its influence on crop production and crop physiological processes. Three review articles on nitrogen are found in this volume. One deals with nitrogen stress in plants, one with the use of tracers in nitrogen research, and a third with nitrate accumulation in plants. Cereal crops are carrying and will likely continue to carry a major share of the burden of food supply, especially in the developing countries. Three articles in this volume focus on the cereals. Emphasis is given to contributions relating t o their genetic improvement and to their physiology. A review of statistical techniques used in soil classification is very timely and should be useful in studies to integrate soil and crop performance information. Likewise, the article on crop-water relations summarizes work on moisture stress in plants and provides background information for draught-alleviating practices. Lastly, the review article on protoplasts calls attention to an exciting new area of research which will likely be pursued as we attempt to increase the yield potential of crop plants.
xi
xii
PREFACE
Volume 28 continues the tradition of the past in that it calls upon scientists from different national and disciplinary backgrounds and it covers a variety of topics of interest to crop and soil scientists.
NITROGEN STRESS IN PLANTS E.A.N. Greenwood Division of Land Resources Management. CSIRO. Floreat Park. Western Australia
..................................................
I . Introduction I1. Quantitative Concepts of Nutrient Deficiency ......................... A . Classic Approach to Nutrient Response ........................... B . Nutrientstress .............................................. 111. Nitrogen Stress ................................................ A . Definition .................................................. B . Measurement ............................................... IV. Factors Affecting Nitrogen Stress .................................. A Supply .................................................... B . Demand ................................................... V . Alternative Evaluators of Nitrogen Stress ............................. A . LeafNitrogenFractions ....................................... B . LeafElongation ............................................. C. LeafArea .................................................. D. Carbon Dioxide Exchange Rate ................................. E . Conclusions ................................................ VI . Applications ................................................... A . AgronomyandEcology ....................................... B . Stress Physiology ............................................ C. Modeling ................................................... VII . Conclusions and Aspirations ...................................... References ....................................................
.
1 2 3 6 7 7
a
14 14 16 18 18 21 25 27 28 28 29 32 33 34 34
I . Introduction
This article is about the quantitative measurement of nitrogen deficiency in plants . It is not a general review of the extensive field of nitrogen nutrition . It must occur to many agronomists to ask why water stress can appear to be so precisely measurable. whereas for nutrient stress we are limited to such broad terms as clinical. subclinical. severe. and mild Our dependence on these qualitative terms reflects an extraordinary weakness within the discipline of plant nutrition in quantifying nutrient deficit You will say at once that there are valid reasons why water deficit must be easier than nutrient deficit to determine And I will agree My criticism is not directed to the greater challenge. but to the meager evolution within the discipline to deal with it . To some extent the cause may be historical This
.
.
.
.
.
1
2
E. A. N. GREENWOOD
century has seen classic plant nutrition engrossed in the search for essential elements, the physiological role, availability in the soil, the clinical symptoms of deficiency and toxicity, and the response of plants to the quantity and chemical composition of nutrient supply. Plant water research was, of course, not concerned with essentiality and diagnostic critera. Whatever the reason, water deficit is commonly defined in terms of the physical stress of the water, whereas nutrient deficiency has come to be defined in terms of the plant’s response to the added nutrient. Consider the succinct definition proposed by Goodall and Gregory (1947): “a plant is deficient in a certain element if supplying that element to the plant in a suitable form causes an increase in yieM [my italics] , this effect being specific to the element in question.” It follows from this definition that in order to quantify the degree of deficiency one can measure the magnitude of the response of the plant. Hence the universal methods by which agronomists seek an estimate of nutrient deficit are based on plant response. If the increase in yield itself after application of fertilizer is not measured, then the soil or plant tissue might be analyzed in order to predict the response, the predictor having been previously calibrated against a response curve. Although the term nutrient stress has been in limited use for a number of years, I am not aware of any exposition of definitions and philosophy. Thus, an attempt is made in Section I1 to state in agronomic terms some basic considerations that agronomists may well accept as axiomatic in the field of water stress, but which have been ignored in the field of nutrient stress. They are my current, personal viewpoints and all of them are arguable (by author as well as reader). Regardless of whether you agree, your constructive participation will certainly assist the advancement of this aspect of agronomy.
II. Quantitative Concepts of Nutrient Deficiency
The term “stress” as used in plant science needs some clarification. It is a broad term that can be defined from more than one point of view; that is, as the status of the stress factor or as the effect of that status on the growth of the plant. For example, Taylor (1968) considered water stress to be the physiological condition of water in the plant, although a plant is often said to be under water stress whenever the conditions of water are unfavorable to plant growth. Because of this ambivalence of meaning, Taylor considered water stress too broad a term to quantify numerically. Hsiao (1973) confines the meaning of water stress to deficit, which is clearly separate from the plant responses which it induces. Central to the issue of evaluating nutrient stress, then, is to decide whether it is to be defined as deficit or as induced response. My current attitude (I have not always held this view) is that nutrient stress should remain a quantitative concept of nutrient deficiency-like growth, it is not measurable. Our efforts, as with water stress, should be directed toward
NITROGEN STRESS IN PLANTS
3
developing measurable parameters of stress which can be used for its evaluation. It would be no departure from convention to call such evaluations “nutrient stress,” for we do not hesitate to invest increments in biomass with the name of “growth rate.” A vital rule must be to indicate which evaluator of nutrient stress is being used. If we allow these first steps to be taken, some difficult problems in estimating nutrient stress can be resolved. But, as these steps are only preliminary, they can be readily retraced should the ground rules be found unacceptable. If the evaluation of nutrient stress is to be analogous to water stress, then we would have to measure such parameters as relative nutrient content and the chemical potential of the nutrient in the plant in some way. Assuming, for the argument, that it is feasible to do this, it would seem that the situation is rather different to what it is for water stress. We take for granted that the upper limit to the water content of a plant cell (full turgor) closely coincides with the water contents at which zero water potential and optimum plant growth occur. But, as is evident from the ability of plants to take upluxury and toxic quantities of nutrients, the upper limit to the nutrient content of a plant cell is well above that required for optimum plant growth. While the lack of a reference point is a major constraint to the concept of relative nutrient content, it is not an objection to the use of chemical potential. It simply implies that the relationship between plant performance and chemical potential would not be so simple, or perhaps as general, as it is in the special case for water stress. The main barrier to the use of chemical potential for estimation of nutrient stress seems to be that it is not yet a feasible proposition to obtain a meaningful measurement of activity of specific ions in plant tissue. And so, for the present, it becomes necessary to fall back to suitable parameters of current plant response. How appropriate to this specific purpose are the classic approaches to the evaluation of nutrient response, viz., nutrient response curves and critical concentrations?
A. CLASSIC APPROACH TO NUTRIENT RESPONSE
This section starts with response curves because the concept dominates the thinking of agronomists about the severity of nutrient deficiency. The only critical criterion of the severity of a deficiency which can be derived from a response curve is the magnitude of the reduction in yield below the maximum, This implies that, for the evaluation of a deficiency, it is unnecessary to know the form of the response curve. The practical significance of this point will be taken up again later in this section. With respect to our present purpose, the main weakness of the nutrient response curve is that it fails to deal with changes in response which inevitably occur in the field with time. The curve is the resultant of all the curves that might have been constructed between sowing and harvesting. Maybe an agron-
4
E. A. N.GREENWOOD
omist does not always want to know more than the final outcome. But sometimes he does need to know what has happened in order to understand how the final response developed-particularly if the results were unexpected. A moment’s consideration shows that responses do vary with time, even to the point of changing sign. These variations in response arise from changes in the supply of the nutrient and the demand for it by the plant. The nutrient supply to the plant may change with time because of uptake by the plant, or as a result of gaseous or leaching losses, chemical reactions, or microbial activity. When fluctuations in nutrient supply are large, the original levels of fertilizer applied inaccurately represent nutrient supply. Further, efficient uptake of fertilizer by the plant implies that the supply of applied fertilizer must approach zero with time, and so response might be expected to decline with this depreciation in treatment strength. The demand by the plant for a given nutrient also changes with time since it is influenced by the changes in all other environmental factors that control plant growth-other nutrients, water, radiation, etc. For example, the amount of nitrogen required by a plant now may be quite large if the environment is favorable for growth, whereas it may be small if, say, the temperature or the supply of available phosphorus is low. In addition to these influences on demand there is the compensating effect of plant size. Plants that response positively to the higher initial levels of the nutrient will, because they are bigger sinks, make larger demands on other nutrients or water. These may in turn become limiting, thereby curtailing the initial response. More certainly, the large plants will intercept so much light that the lower leaves will receive suboptimal illumination. Another important influence on the response of deficient plants is age. The responsivity of plants to a deficient nutrient declines with ontogenetic drift, particuarly in annuals where response approaches zero with maturity. Two conclusions follow from these remarks: nutrient response curves give very little information about the current intensity of deficiency, and in some cases they may also be quite misleading. From the foregoing discussion it is clear there will have to be some integration of supply and demand. A familiar approach has been to determine the concentration of the nutrient in the plant, arguing that the plant itself is the actual integrator of supply and demand. The graphic expression of this approach is to plot yield against nutrient concentration in the tissue (Fig. 1). Refinements are made by selecting the yield of a specific product and the chemical determination may be of some sensitive compound in sensitive tissue. In Fig. la the yield is beet weight, the compound determined is nitrate nitrogen, and the tissue is petiole. A frequent, but not universal, characteristic of the relationship between yield and chemical composition is that, for severe deficiency, yield increases linearly with increase in concentration with a well-defined steep slope, whereas at sufficiency, the curve flattens and becomes poorly defined (Fig. 1). If the change
"Mature" Petioles
A A
A
0
i 20
-
N0.25
I
I
0
I
1
1
16,000
8.000
I
1
1
24,000
Nitrate Nitrogen (ppm) b
100
'
0
128
0
8
8 256
A 80
A 96
0
A
B
0512 88
A
8
0
0
~4 L ~ l l l ~ ~ 1
0
0
2,500
1 .m
5,000
7,500
1
1
1
1
1 0 , ~12,500 15,000 17,500 20,OOO
Nitrate-N In Blade 1 (ppm)
FIG. 1. Relation of yield and nitrate nitrogen in plant tissue. (a) Weight of sugar beet and nitrate in petioles (after Ulrich, 1950). (b) Dry weight of Italian ryegrass and nitrate in leaf blade 1. dy/dx = 1 at approximately 1000 ppm nitrate-N. After Hylton etal. (1964).
6
NITROGEN STRESS IN PLANTS
in slope is abrupt, a “critical concentration” is discernible at the discontinuity, above which there is little increase in yield and below which yield is greatly reduced. It is not always easy to locate the critical point because of the variability of the data or because the curve may be broad and continuous. In the latter case, a numerical definition of the critical concentration can be based on the slope of the curve. For example in Fig. l b a slope of dy/dx = 1 has been adopted by Hylton et al. (1964) as critical. It is a step toward numerically quantifying the degree of deficiency, for a continuous change in slope can provide a continuous numerical scale. There are two serious weaknesses in the use of the slope of the yield concentration curve as a continuous indicator of the intensity of deficiency, as distinct from a single critical point. First, as seen in Fig. 1, the deficient arm of the curve shows little change in slope. Second, it is well known that the critical concentration declines with age and that in annuals this decline is precipitous after flowering. As for the response curve, a whole suite of curves like Fig. l a can be obtained for successive ages. Nevertheless it can be a most useful technique for estimating deficiencies in highly standardized crops (e.g., Ulrich, 1950). The critical concentrations of nutrients in plants is the main form of reference to nutrient status to be found in the literature. Extensive use is made of the reference tables compiled by Chapman (1966). For more specific information, agronomists turn to their colleagues in plant nutrition for interpretation of “spot” chemical analyses. The persistent attraction of such information lies in the certainty that below an accepted concentration in the plant a nutrient will be severely limiting, and that usually there is no better information available. It seems paradoxical to make such a claim in face of the vast literature on the use of plant analyses as a tool for assessing the nutritional status of plants. The problem is that although the concepts seem simple, the results must be qualified, as with response curves, by strong interactions with time. This aspect is discussed by Smith (1962) in his comprehensive and lucid review of tissue analysis.
B. NUTRIENT STRESS
It is evident that response curves and tissue analyses will not provide a satisfactory basis for using plant response as an indicator of nutrient status. We ought now to return to the definition of deficiency proposed by Goodall and Gregory (see Section I), and develop a proposition for nutrient stress. For it seems reasonable to argue that if it is acceptable to base a definition of nutrient deficiency on the yield response to a dose of the nutrient, then a quantification of this response should provide an acceptable basis for nutrient stress. Goodall and Gregory used yield (presumably of dry matter) as the criterion of response. But other criteria can be envisaged such as size (e.g., height, leaf area), or a
E. A. N. GREENWOOD
7
process (e.g., carbon dioxide exchange rate), or any other partial expression of plant growth which can be measured readily. The problem with this choice is that each parameter of growth has its own functional relationship with deficiency which would in turn give different estimates of nutrient stress. The use of the full expression of growth itself for these purposes is not possible since growth is not’ measurable, as pointed out by Arnott et al. (1974) in their introduction on the measurements of “growth.” Setting aside the choice of growth parameter until Section V, let me show what can be done with dry weight. The first task will have to be the development of a reference point to represent zero stress. It is implicit in the nutrient response curve and in the tissue concentration curve that the highest yield obtained under the circumstances represents the complete absence of the deficiency. This circumstantial maximum is the standard against which the degree of deficiency at lesser yields is judged. It was pointed out in Section 11, A that one does not need to know the form of the response curve in order to evalute a deficiency. What is required is the relationship of the yield of the deficient plant to the yield of the plant at the circumstantial maximum. It is necessary also to establish that maximum yield has been obtained. In other words, a numerical value for deficiency can be derived from the shortfall in yield relative to the circumstantial maximum yield. This is the crucial point on which the whole of the remainder of this article is based. There is a further point to recall from the criticism of yield response curves in Section 11, A-an inability to indicate current response. This shortcoming can be easily dealt with by measuring current growth rate instead of the accumulated yield of dry matter. If all these considerations are adopted, the current intensity of deficiency of a nutrient-nutrient stress-can be evaluated as the proportion by which the growth rate of the plant or crop is limited by that nutrient under the prevailing conditions. This definition needs to be qualified by the parameter of growth rate used-in this case, dry weight. The transformation of these ideas into a workable technique will be the objective of Section I11 which will be confined to the specific case of nitrogen stress. At this point also there will be a change in emphasis from a critique to a review of the subject. I l l . Nitrogen Stress
A. DEFINITION
Nitrogen stress is a quantitative estimate of the intensity of current nitrogen deficiency in a plant or crop. It can be evaluated as the proportion by which the growth rate of the plant falls short of maximum growth rate attained with a
8
E. A. N.GREENWOOD
nonlimiting supply of nitrogen over the period when stress is being measured. For this representation of nitrogen stress, Greenwood, Goodall, and Titrnanis (1965) used the symbol SN when it was made on a biomass basis. The relative shortfall can be expressed as a percentage, i.e., (prowth rate at maximum N response)-(growth rate at deficiency) SN = 100
growth rate at maximum N response
(1)
For example, if the growth rate of the deficient crop is 7 g/m2/day and 10 g/m2/day when it is given a nonlimiting dose of nitrogen, then nitrogen stress would be 100[(10-7)/10] = 30%. SN has some broad similarities with relative water content, for which the weight of a leaf deficient in water is compared with its weight after it has been brought to a standard, nonlimiting water content. In order to put this expression into practice, it is necessary to decide precisely what is meant by growth rate and how to find the current limitation of it by nitrogen.
B. MEASUREMENT The split-plot technique is basic to all methods so far devised for estimating
SN. At the time SN is to be measured, one subplot is left untreated and the other is given sufficient nitrogen to make nitrogen nonlimiting. Current growth rate is then measured in the untreated subplot, and the maximum growth rate that can be attained by adding nitrogen is measured in the other subplot. 1. Growth Response There are two conventional ways of expressing growth rate on a dry weight basis. Symbol Crop growth rate
C
1 dW Relative growth rate - Wdt
R
Working formula
In W2-ln W 1 t2 - t 1
Units
g/g/day
where W1 and W 2 are the dry weights at time tl and t 2 , respectively. The numerical value of SN obtained may be influenced by which expression is used. Broadly speaking, C is appropriate for crops or swards with a closed canopy and R is appropriate for spaced plants. In order to distinguish which form of growth rate has been used for estimating SN,the appropriate symbol can be used as a
NITROGEN STRESS IN PLANTS
9
subscript, i.e., SNC or SNR. The expressions for SN then become: SNC =
100 [(cM-C)/cMl
SNR = 100 [ ( R M - R ) / ~ V I
(2) (3)
where C or R is the growth rate in the control subplot and CM or RM is the growth rate in the subplot to which nitrogen has been added. Equation (3), the one originally proposed by Greenwood et LIZ. (1969, has been used consistently in publications on nitrogen stress. Nevertheless, as indicated above, it is not necessarily the most appropriate equation. Relative growth rates were originally adopted as a compromise in the interests of general application. The problem, briefly, is the exponential character of plant growth in unclosed canopies. Under steady conditions, the daily increment in weight increases with plant size, with the implication that the magnitude of the response to nitrogen is confounded with the weight of the plant. More specifically, when one wishes to compare the responsivity of two plants of different size due to different treatments or age, a bias is introduced. The bigger plant will have the bigger potential response in absolute terms. If the exponent of growth were constant then the bias could be exactly overcome by using relative growth rate. Although R does change less than C over the life of a crop by an order of magnitude, its change is appreciable. Operationally there is no difference between using SNC and SNR,for they both require the same primary data, W 1and W2.The real dilemma is that both C and R are nonideal over the whole life of the crop. And there seems no way of deciding which of them is to be preferred, for young crops behave as a community of spaced plants until the canopy is closed. Even if a working rule is adopted such that R is appropriate for crops with LA1 < 1 and C for crops with LA1 > 1 (LA1 = leaf area index), we would be well aware of a very broad transition zone in which neither R nor C would be fully appropriate. We must conclude that the approach to plant stress through dry weight response cannot be taken without some bias. The importance of the bias can be gauged by comparing SNC and SNR derived from the same dry weight data. This can be seen in Fig. 2c for a situation supposedly favoring SNR. The difference between the two curves is small at the extremes but large in the middle portion. In the case of wheat, the real situation does not seem to be as bad as it sounds from these considerations. In Fig. 3, SNR is plotted against SNC from two sets of data. One is the pot data used for Fig. 2c. The other is a reworking of the field data of Halse etal. (1969) in which SN was estimated at four stages from 4 to 16 weeks (ear emergence) with LAI ranging from about 0.1 to 2.5. The one curve fits both sets of data (R2 = 0.995). Three important points follow: age or LA1 has not affected the relationship between SNR and SNC;SNR is neither more nor less appropriate than SNC ;and it is a simple matter to convert SNRto SNC or vice versa. It can be seen from Fig. 3 that because the intercept is close
u
0
2
4
6
N Supply (mM1
8
1
0
.02 0
2
4
6
N Supply (mM)
8
10
0
2
4
6 N Supply (mM1
8
10
FIG. 2. Four expressions of response by wheat to nitrogen supply between the third and fiith week after emergence. (a) Yield of dry matter; (b) relative growth rate ( R ) and crop growth rate (0;(c) nitrogen stress based on R and on C. Data are from Greenwood (1966).
NITROGEN STRESS IN PLANTS
11
80
40
20
0 0
20
60
40
80
100
sNC(%)
FIG. 3. Relationship of nitrogen stress in wheat based on relative growth rate, Sm, to nitrogen stress based on crop growth rate, SNC. Points denoted as are from Fig. 2(c). Other points are derived from Halse et 01. (1969)from a field crop sampled between 4 and 7 weeks (A), 6 and 9 weeks (A), 10 and 13 weeks (+) and 13 and 16 weeks (x) after sowing. ~ derived from all points and accounts for The curve y = 3.55 + 0.25% + 0 . 0 0 6 0 6 ~was 99.5% of the variance.
to zero, and, for subclinical levels of nitrogen deficiency only, the quadratic term can be ignored, the relationship of SNR to SNCcan be simplified to SNR = 0 . 7 5 ~ These ~ . remarks apply to wheat; they may not hold for other species, particularly dicotyledons. A useful reference point is the value of SN at which nitrogen deficiency symptoms begin to develop. In all published work with grasses and cereals this Point Occurs at SNR = 40% (SNC = 60%). This section has dealt with growth response only in terms of dry weight. Other parameters of growth rate are considered in Section V.
2. Response Interval Ideally, growth rates on the control and the plus-nitrogen subplots should be measured as quickly as possible after the plants have responded to the addition of nitrogen. This is because current response is required and SN may be changing rapidly. In practice there is a lag period between application of the nonlimiting dose of nitrogen and its entry into the roots and leaves even in solution culture
12
E. A. N. GREENWOOD
where immediate uptake of nitrogen is ensured. Bouma (1970a) measuring both leaf area and carbon dioxide exchange rate (CER) on subterranean clover at 2-day intervals could clearly detect a nitrogen response after 2 days. Wolf and Greenwood (unpublished data) used a much shorter time interval and measured the leaf elongation rate (L), and CER of expanding and mature leaves of wheat seedlings in light and in the dark. They found that dark respiration responded to added nitrogen in 2-7 hours but the response was small and insensitive. The CER of mature and expanding leaves responded after 22 hours, and a similar time lag was taken by L for elongating leaves. Full response by AL occurred after 48 hours, whereas CER required more than 72 hours. There was no evidence of any temporary toxicity when the nitrogen supply (as ammonium nitrate) was raised from 1 'to 20 mEq N/liter. An example of one of the several response runs is given in Fig. 4. From the foregoing evidence, it seems that at least 2-3 days should elapse between applying nitrogen and commencing to measure the response. In practice, as agronomists will appreciate, it is difficult to get accurate estimates of dry weight increments in less than a week. This is because field sampling is imprecise and because the variance of the increment W2-W, is about twice the variance of a single dry weight measurement. In addition to this we are measuring differences between increments on one subplot and increments on another. Thus the variance of SN is high. 80
0 0
-8
I
60
f
G e
40
8
2 20
t
i Dark
0
20
Dark I
I
0
40
1
60
1
80
Hours
FIG. 4. Apparent change in nitrogen stress in deficient wheat seedlings immediately after the nonlimiting dose of nitrogen was given to alternate matched plants. Successive values of rates of elongation (L)and carbon dioxide exchange rate Q of the emerging second leaf were substituted into Eqs. (4) and (6) in the text. Developing response is indicateli by rising slopes and the attainment of full response by flat slopes. Replication was X 4 . From unpublished data by Wolf and Greenwood.
NITROGEN STRESS IN PLANTS
13
My experience is that a growth interval of at least 10-14 days is required to get meaningful and reliable differences between growth rates on the subplots when these are expressed as dry weights. This is far from being instantaneous. If SN is changing rapidly, only an average value over the time interval can be used. Rapid changes in SN do occur naturally. Power (1971) reported an increase in SNR from 4 to 73% in bromegrass within 3 weeks. On the other hand, there is an advantage in the field in having a fairly long time interval over which SN is being estimated, for it helps to integrate the day-to-day variations in weather which may be of little interest in themselves. Of course, where one wishes to establish relationships between stress and the environment a short-term response would be welcomed. Alternative, short-term, nondestructive methods for estimating stress are described and evaluated in Section V.
3. The Nonlimiting Supply of Nitrogen
SN can be estimated without invoking any assumptions as to the form of the response surface. It requires only the values for the actual growth rate and the growth rate with nitrogen nonlimiting. Since the response curve (i.e., the section of the response surface at prevailing levels of factors other than nitrogen supply) generally has an extensive plateau around the optimum, high precision in the choice of nonlimiting nitrogen levels is not necessary. In cases where there is adequate information available (Power, 1971), one estimate of a nonlimiting nitrogen supply level is sufficient. Otherwise, at least two different levels should be used in order to judge whether the deficiency has been completely removed and whether toxicity has been avoided. The following working rules are useful: when the two “nonlimiting” levels of nitrogen do not produce growth increments that are significantly different from each other, then the increments are averaged to give a best estimate of growth rate with nitrogen nonlimiting; when the difference between them is significant but small, the larger value is taken for nitrogen nonlimiting; when the difference between them is large, then the data are abandoned. Significant growth differences for the two levels of nitrogen selected can be easily avoided provided that some precautions are taken. These precautions can be generalized: very young seedlings and heavily defoliated plants require much lower levels of nitrogen for maximum response than do older and intact plants; for the former, temporary toxicity may occur, particularly if nitrogen is supplied in the ammonium form. 4. Operational Procedures For the estimate of SN by destructive dry weight harvest, a multi-split-plot technique is required. A quartet of matching subplots or quadrat areas is selected. Each of the four subplots is assigned at random to one of the following
14
E. A. N. GREENWOOD
procedures. Dry weight, W1,
is determined at the beginning of the test interval tl and, similarly, W z is obtained at the end of the interval tz . The actual growth increment of the crop is computed from these two values. Meanwhile, at r ,each of the remaining two subplots is given a different “nonlimiting” application of nitrogen and is harvested at tz . From the two dry weights for these subplots the value for W M is obtained. The growth increment of the crop with nitrogen nonlimiting is obtained from W land W M . In the field, if the soil is wet throughout the root zone, the nonlimiting dose of nitrogen can be applied in solution using the equivalent of say only 2 mm of artifical rain in order to avoid disturbing the water regime. A similar amount of nitrogen-free water must also be added to the W z subplot. Where the surface of the soil is dry due to a short and perhaps unimportant absence of rain, then a decision must be made either to wait for rain or to add sufficient water with the nitrogen to simulate rain which could be expected to fall. Again, if water is added it should include Wz.In either case the value for SN would apply to situations where water is not an important limiting factor. In cases where it is realized that water is an important limiting factor, albeit unevaluated, but where rain falls during the growing season, another technique is available. Since it is unlikely that in these circumstances an argonomist would want to know the importance of nitrogen as a limiting factor without similar information for water (though few have sought it), the technique used by Power (1971) can be most effectively employed. This involves the use of supplementary water as well as nitrogen, and it will be discussed in Section VI. IV. Factors Affecting Nitrogen Stress
Nitrogen stress can be considered as a concept that integrates the rate of nitrogen supply with all the other factors essential to growth: genetic, ontogenetic, nutritional, environmental, symbiotic, and other factors. It follows that a change in nitrogen stress may be brought about through variation in either the supply of nitrogen or in any of these other determinants of growth. The following is a review of the limited experimental evidence on factors affecting nitrogen stress. A. SUPPLY
Curves relating SNR and SNC to nitrogen supply for wheat between weeks 3 and 5 are given in Fig. 2c. In this example, both curves lead to a credible
NITROGEN STRESS IN PLANTS
15
extrapolation to 100%stress (no net increment in dry weight) at zero introgen supply, and, of course, they both reach zero stress (no response to nitrogen) as the nitrogen supply approaches the nonlimiting level. In nutrient culture work, it is usual to make some attempt to keep the concentration of nitrogen in the solution fairly steady by periodic or by continuous replacement. But where only a single application is given, say at the beginning of a pot experiment, the progressive depletion of the applied nitrogen must result in an increase in SN ,other factors being held unchanged. In the field, marked fluctuations in nitrogen supply in both directions may occur. On the one hand, these may be leaching and gaseous losses and, on the other hand, all those factors such as temperature and wetting and drying which may control the rate of microbial production of available nitrogen in the soil. Some practical perspective is given by the work of Halse et ul. (1969) who grew a wheat crop in a nitrogen-deficient sandy soil (average annual rainfall, 390 mm) and applied nitrogen at three rates at sowing: nil, 56 kg/ha, and 112 kg/ha, the last also receiving two further applications of 112 kg/ha. These treatments gave grain yields of 900, 1800, and 3000 kg/ha, respectively. SNR was determined on the nil and 56 kg/ha treatments. On the nil treatment SNR remained at 48% for several weeks and then fell to 14% at the late boot stage. The application of 56 kg (kglha) almost eliminated stress during the first few weeks, but by floral initiation it had reached 23%before falling to 5% (Table I). These results were obtained in the Mediterranean climatic zone of southern Australia during the growing seasons of autumn, winter, and spring.
TABLE I Effect of Nitrogen Fertilizer and Plant Age on Nitrogen Stress (SNR) in a Wlieat Crop' Nitrogen applied Weeks after sowing
Nil
56 kg/ha %i
4-7 6-9 10-13 13-16
48i1 47il 26i5 14i7
SE
6i2 23i3 11i4 5i8
'Adapted from Hake e l al. (1969).
16
E. A. N. GREENWOOD
B. DEMAND The age and ontogeny of a plant can greatly influence the magnitude of SN. Greenwood and Titmanis (1968) found that for annual ryegrass grown on a constant nitrogen concentration of 1 mM, S N increased ~ from 10% at 2 weeks after emergence to 11% at week 3, to 17% at week 4, and to 32% at week 5. The explanation could be that, as the daily increment in dry weight increases with age, the demand for nitrogen must increase. If the supply is fixed in concentration and inadequate, the shortfall in supply, whence stress, must also increase. At a much later stage, individual axes of the plant produce flowers and become less capable of response to nitrogen in the sense of net increment of dry weight-at least in more determinate species. This implies, particularly for annuals, that SN must approach zero as the plant approaches maturity. These trends are evident in Tables I and 111. Light will increase SN provided that light intensities are below the optimum for plant growth. Table I1 records some unpublished results for wheat seedlings. When the potential growth rate of a plant is limited by other nutrients as well as nitrogen, then an increase in the supply of those nutrients ought to increase SN. A very clear demonstration of this point with respect to sulfur and phosphorus can be inferred from the data of Bouma and Dowling (1967). They showed increasing responses (whence stress) of leaf area to nitrogen in TrifoZium subferraneum. For example, on a very low supply of inorganic nitrogen, values of stress are 25%, SO%, and 70% for sulfur supplies of 0.125,1.0, and 8.0 ppm. In computing these values, I have taken the highest supply of nitrogen as being nonlimiting, which is good enough for this exercise.
TABLE I1 Effect of Shading on Nitrogen Stress (SNC) at lS"/lO"C in Wheat
N fllpply (mM)
Unshaded
Shaded
2 4 6
42 17
31 11 4
13
NOTE: Plants were grown in a naturally lit glasshouse with noon light intensities of 45,000 (unshaded) and 18,000 lx (shaded). Replication was X 7.
17
NITROGEN STRESS IN PLANTS
Power (1971) studied the interaction between wafer sfress and nitrogen stress on bromegrass in the northern Great Plains of the United States. He found, over a range of conditions, that the plant stress caused by each of these two limiting factors was roughly additive. To give one instance, the value of nitrogen stress (SNR)as obtained by making nitrogen nonlimiting was 24%.The value of water stress (SWR) as obtained by making water nonlimiting was 32%. But when nitrogen and water were both made nonlimiting, the value of plant stress obtained was 54%. This implies that as water becomes more limiting the value of SN declines. Further results are displayed in Table VI. Defoliation, whether by cutting or by grazing, might be expected to reduce SN at low values of LAI when a reduction in photosynthetic area might be expected to limit potential growth during the recovery period. Greenwood and Titmanis (1968) established this point experimentally with young swards of annual ryegrass in pots with clipping, and in the field with sheep. The particular defoliation regimes used reduced SNR from 32 to 19% in the pot experiment and from 11 to 3%in the grazing experiment. These examples probably underestimate the immediate effect of defoliation on SN since the latter was, of necessity, estimated during the ungrazed recovery period. One of the general, and important, effects of cultivation is to increase the supply of available nitrogen to a crop. This should result in a reduction of stress provided that there is a deficiency of nitrogen and that cultivation does not affect other limiting factors. The situation is likely to vary with soil type and other circumstances. The integration of all these factors by the crop can be expressed, with reference to nitrogen, as SN. Table I11 shows that the effect of cultivating sandy loam was to reduce the subsequent values of SNR in a wheat crop. This will be discussed further in Section VI,A.
TABLE 111 Effect of Age and Cultivation on Nitrogen Stress (SNR) in Wheata Nitrogen stress (%) Cultivation treatment
Wk after emergence: 3-6 13
Nil Conventional
6-9 26,
9-12 13
*
***
ns.
5
17
10
Adapted from Greenwood e?al. (1970).
18
E. A. N. GREENWOOD V. Alternative Evaluators of Nitrogen Stress
The evaluation of nitrogen stress by dry weight increment (SN)as described in Section IV is a simple procedure, but it has three important disadvantages in that it requires destructive sampling, a high replication for precision, and a lengthy period for the response to manifest. Obviously it would be a great advantage if nondestructive or more rapid methods could be found. Alternatives are available and they fall into two classes: parameters of plant nitrogen status which are used as indices of SN, and parameters of plant growth which are used as direct alternatives to dry weight.
A. LEAF NITROGEN FRACTIONS
When attempting to relate leaf nitrogen to nitrogen stress, similar specifications to those which usually apply to the establishment of “critical levels” in tissue analysis must be considered. Decisions must be made as to the choice of nitrogen fraction to determine, organ or tissue to sample, and the time of sampling. Insight into these aspects can be gained first from reference tables such as in Goodall and Gregory (1947) and Chapman (1966) as well as individual research papers. For example, Ulrich (1950) gives critical values of nitrate nitrogen for leaves and for petioles at three physiological ages and at different dates of sampling of Beta vulguns; Hylton et al. (1964) give the value of nitrate nitrogen in several plant parts at which dy/& = 1 for Lolium multijlorum; and Rauschkolb el al. (1974a, b) evaluate the nitrogen status of maize and sorghum in terms of the total nitrogen and nitrate-nitrogen concentrations in the whole leaf, midrib, and basal section of the stem. We have examined the relationship of SN to the concentration of total nitrogen and of ninhydrin (mainly a-amino) nitrogen in the youngest fully expanded leaf of the tillers of annual ryegrass (Lolium rigidum) and in several genotypes of wheat. We used the youngest fully expanded leaf on the tiller at the time of sampling because its physiological age is constant and because it is easy to identify and sample. Total nitrogen was chosen because of its common use. Nitrate nitrogen was rejected, mainly because it does not accumulate in measurable quantities over the whole range of deficiency (e.g., Hylton et al., 1964), but also because it is highly variable in concentration, and it is sensitive to the form of nitrogen supplied and to time of day (Allen et al., 1961). Ninhydrin nitrogen was chosen as being intermediate between nitrate (unmetabolized) nitrogen, and total (mainly protein and therefore historic) nitrogen. Typical curves for total nitrogen and ninhydrin nitrogen in the youngest fully expanded leaves of wheat tillers are given in Fig. 5 .
19
NITROGEN STRESS IN PLANTS
K
z
*.
a
b
lJY
i .-
80
:\ 0 60 - \o 40
1
60
Total Nitrogen
\o
40
\
0
b
20
20
0 2
1
I
I
3
4
5
0 6
7
0.3
0.2
0.1
Leaf Nitrogen (%I FIG. 5 . The relation between nitrogen stress (SNR) and the concentration of nitrogen in the youngest expanded leaf of wheat between the third and fgth week after emergence. After Greenwood (1966).
To test the general applicability of leaf nitrogen fractions as estimators of S N , the constancy of the relationship must be investigated under a range of conditions. The results of some investigations are reviewed below. A direct comparison of the relationship of SN to leaf nitrogen was made between four commercial genotypes of wheat between the third and fifth week after emergence. Whereas some genotypes had similar calibration curves others differed (Titmanis and Greenwood, 1969), which leads to the conclusion that, in practice, it would be necessary to calibrate each genotype separately regardless of whether total nitrogen or ninhydrin nitrogen was to be used. Table IV gives the predicted value of SNR for a given value of the estimator in each of the four TABLE IV Comparison of Estimates of Nitrogen Stress (SNR) Given by Set Values of Leaf Nitrogen Fractions in Several Genotypes of Wheat' Nitrogen stress (%) Value of estimator Total N 4% 5% Ninhydrin N 1200 ppm 1500 ppm
Mendos
Gamenya
Emblem
Olympic
Gab0
25 8
25 9
30 -
45 17
21 9
23 10
23 10
20 -
35 11
33 20
'Adapted from Titmanis and Greenwood (1969) and Greenwood (1966).
0.4
20
E. A. N.GREENWOOD
genotypes, and also for Gabo at the same age but from another experiment (Greenwood, 1966). Since the relationship between SNR and leaf nitrogen vanes between close genotypes of the one species, it can also be expected to vary between closely related species. This certainly holds for the two species that have been investigated-annual ryegrass and wheat. At 28 days after seedling emergence the predicted value of SNR at 4% N content is 46% for ryegrass (cf. Fig. 6), which is a much higher value than four out of the five wheat genotypes at the same age just mentioned (Table IV). Work on the effect of age was conducted in two separate experiments by Greenwood and Titmanis (1966); the results have been brought together and reexpressed in Fig. 6. Over the first 5 weeks after emergence, a given concentration of total nitrogen in the youngest fully expanded leaf of the tiller gives a fairly constant estimate of SNR. Thereafter the predicted value of stress falls. For ninhydrin nitrogen, the two experiments gave inconclusive results.
60
0-0-0
50
-
40
-
-ap
3.5%N
-0
Total Nitrogen
\
-
-
6.MN
0-0-0-
10
DO
1
Spring Experiment
0 ,
I
1
0
I
Winter Experiment I
1
I
1
NITROGEN STRESS IN PLANTS
21
Insofar as one can generalize from two species, the period over which the concentration of total nitrogen in the youngest fully expanded leaf of the tiller can be used as a stable estimator of SN is short, i.e., 5-6 weeks at the most. No precise work has been recorded on the effects of either the composition of the nitrogen source or the relative supply of other nutrients on the regression of SN on leaf nitrogen. In experiments with young annual ryegrass, we examined the effects of defoliation on the relation of SN to leaf nitrogen (Greenwood and Titmanis, 1968). The various defoliation treatments used had very little effect on the relationship except at high concentrations of plant nitrogen (cf. Fig. 6). Total nitrogen and ninhydrin nitrogen performed simiiarly as estimators of stress. Light exerted a marked effect on the relationship between SNC and the total nitrogen concentration in the youngest fully expanded leaf of wheat tillers in an unpublished experiment by Greenwood. The treatments were full daylight and shading. Average noon light intensities for the unshaded and shaded treatments were 45,000 and 18,000 lx, respectively. Figure 7 and the accompanying statistics show that the estimation of SNC by total leaf nitrogen was good (R2> 0.98) and that the effect of shading was to produce greater slopes and intercepts than the unshaded treatments. If leaf nitrogen is to be used, then which fraction of nitrogen is to be preferred? There is usually little to choose between total nitrogen and ninhydrin nitrogen. The former is easier to deal with in the field and this may be the deciding factor. Nitrate nitrogen is unsatisfactory because it is not present in detectable amounts in severely deficient plants and because it shows diurnal and other variations. My own conclusion now is that any fraction of leaf nitrogen is rather unsatisfactory for general use as a quantitative estimator of SN.
B. LEAF ELONGATION The use of the rate of leaf or tiller elongation as an estimator of current growth rate in Gramineae has been briefly reviewed by Williams and Biddiscombe (1965). This reference also includes an excellent photograph of both continuous-recording and multi-point auxanometers, which are instruments used for the automatic measurement of tiller elongation rate. Leaf elongation has turned out to be the most useful of the estimators of SN so far evaluated for Gramineae. Leaf elongation rate is sensitive to nitrogen stress and, because of the morphology and pattern of leaf development in Gramineae it is simple and cheap to measure. Under steady conditions the daily rate of elongation of a given leaf of a Gramineae is constant from its first appearance until just before its full expansion. Since the latter coincides with the emergence of the next leaf this period approximates the leaf appearnce interval. Leaf elongation is obtained as
22
E. A. N. GREENWOOD 70
X
60
50
--8
40
0
z
v)
R2=0.995)
30
20
Unshaded R2=O!
10
\X I
0 2
I
1
3
I
4
1
1
5
Leaf Nitrogen(%)
FIG. 7. Effect of shading on the estimation of nitrogen stress (SNC) by total nitrogen in the leaf.
the difference between successive measurements of leaf length (in Gramineae this can be from the base of the ligule of the older leaf to the tip of the emerging leaf). Tiller elongation can be measured from a bench mark on the ground to the tip with a rule, or by an awanometer attached to the emerging leaf tip. More elaborate devices may be used for measuring elongation, such as the one used by Hsiao el al. (1970) for maize. The use of leaf elongation rate as an estimator of SN has not been evaluated in dicotyledons. It may well be suitable, for Wadleigh and Gauch (1948), for example, found leaf elongation to be a sensitive estimator of water stress in cotton plants. Leaf elongation rate, L (cm/day), can be used to estimate SN by substituting L for R or Cin the nitrogen stress equation SNL = 100 [(LM-L)/LM]
(4)
The ability of SNL to estimate SN in Lolium is seen in Fig. 8. For subclinical
NITROGEN STRESS IN PLANTS
23
100 28-42 Days 0 44-58Days
SNR
A
A
sNC
80
28-42 Days ~-58Dayr
60
-s?z v)
40
20
0 0
20
40
60
80
100
SNL(%)
FIG. 8. Linear regressionsof SNR on SNL and of SNC on SNL in Lolium rigidum. Points derived from Greenwood and Titmanis (1966) for plants between 2 8 4 2 days and 44-56 days after emergence. Values from plants with symptoms are not included in the regression.
levels of nitrogen deficiency, over 97%of variation in either SNR or SNC can be accounted for by SNL.Where deficiency symptoms are present there is a marked increase in slope. This seems to be caused by a rapid recycling of nitrogen from the older leaves to the emerging leaf. Consequently, even when SN approaches 100%(zero growth on a dry weight basis), leaf elongation continues. Unfortunately, annual ryegrass is the only species to have been studied with respect to leaf elongation and nitrogen stress. Some unpublished data of Power show that leaf elongation is sensitive to the effect of nitrogen supply on certain growth rate in a wide range of grasses from May to July in North Dakota (Table V) and therefore might also be useful as an estimator of stress. Power compared leaf elongation rate in swards that received either no nitrogen fertilizer or 220 kg Nlha in early spring. It is tempting to calculate SNL values from this array but to do so would be far from rigorous. This is because at each time of comparison,
24
E. A. N. GREENWOOD TABLE V Rate of Leaf Extension of Various Grass Species Prior to Inflorescence as Affected by Nitrogen Fertilization (Values Are Means of 6 Tiller@ Date
Species
N rate (kdha)
5/27
6/2
6/9
6/17
Average SE (%of mean)
17.8 22.0 7.2 11.3 10.5 25.6 11.3 32.3 6.0 2.7 6.7 12.5 9.0 32.2 9.7 9.5
10 12 17 24 18 19 15 20 26 27 17 18 14 17 14 19
mm/day Reed canary grass Smooth brome Western wheatgrass Russian wild
we Crested wheatgrass Green needle grass Garrison creeping foxtail Intermediate wheatgrass
0 180 0 180 0 180 0 180 0 180 0 180 0 180 0 180
8.7 14.0 10.1 8.2 7.2 8.1 4.8 8.5 7.0 3.8 5.2 4.4 5.5 2.5 10.6 8.5
13.5 18.2 7.8 14.5 9.7 12.5 7.3 18.2 11.0 20.0 9.0 10.0 6.8 21.5 12.3 19.2
14.0 25.3 7.5 18.2 9.3 21.2 10.2 25.0 7.6 10.7 8.4 16.0 6.3 36.2 11.3 18.8
‘Unpublished data of J. F. Power.
the No and Nzzo plots were not at all comparable, the latter having responded extensively to nitrogen applied in early April; consequently a split-plot design that is required for SN did not apply. The data have been presented here because of the rare insight they give into the sensitivity of leaf elongation under a variety of seasonal conditions. For example, soil water measurements showed that the fertilized plots contained 2-8 cm less water in the root zone in late May than did the check plots. This could account for the negative response to nitrogen obtained at that period. The dry matter responses obtained (to be published by Power elsewhere) reflected the variation in leaf elongation rates. The day-to-day change in temperature also produced a noticeable effect on elongation (Power, personal communication), an observation that supports the elegant results with Phalaris tuberosa L., P. arundinacea L., and Festuca arundinacea Schreb. obtained by Williams and Biddiscombe (1965). The standard errors in Table V give some idea of the adequacy of the fourfold replication used by Power. More comprehensive information on the number of
NITROGEN STRESS IN PLANTS
25
replicates required for a given level of precision in L is provided by Scott (1961) for tussock grasses. At the present stage of evaluation it appears that leaf elongation is a simply measured and promising estimator of SN, and its use should be encouraged. Accordingly, it would be profitable to investigate its performance early in any program where it might be applicable. Even in the case of annual ryegrass it has been shown that the relation of SNL and SNC does not hold indefinitely. From Fig. 8 it can be seen that a steady relationship holds at least for the first 8 weeks. But Greenwood et al. (1965) have shown that values of SNR begin to rise in relation to SNL shortly after 8 weeks. The application of leaf elongation is taken up again in Section VI, A.
C . LEAFAREA
Since leaf elongation rate is sensitive to nitrogen status, it follows that the rate of increase in leaf area should also reflect accurately the influence of deficiency on growth. Furthermore, lead area is both an expression of size and a partial expression of photosynthetic potential. In this context leaf area is taken to be total leaf area present either per plant or per unit ground area (LAI) as distinct from that of a selected expanding leaf as is used for leaf length. Nutrition can influence photosynthesis (whence growth) through affecting leaf area itself or through changes in photosynthetic rate per unit leaf area (net assimilation rate). These aspects of crop nutrition have been reviewed by Watson (1963) who concluded that, for nitrogen, the main effect on growth is through leaf area rather than through net assimilation rate. More specific evidence of the close relationship between leaf area and the influence of nitrogen on growth can be derived from the data of Bouma and Dowling (1966) and of Halse et al. (1969). Bouma and Dowling measured the dry weight and leaf area response of subterranean clover to nitrogen supply in water culture. I have computed the linear regression of the dry weight data on the leaf area data for each nitrogen level. It shows that 98% of the variation in dry weight is accounted for by the regression on leaf area. Halse measured photosynthetic area (green leaf plus stem area) and SNR in a wheat crop grown at three levels of nitrogen fertilizer: nil, 56 kg N/ha at sowing, and 336 kg N/ha split over three applications. Figure 9 shows my plotting of the change in leaf area against SNR for each period when stress was measured. The exercise shows a close relation between LA1 and nitrogen stress and a strong interaction with time. What is required for the estimation of stress is a set of leaf area data (derived from appropriate plots split for nitrogen) from a range of nitrogen levels and
26
E. A. N. GREENWOOD 2.5
-
2.0
4-7Weeks
-
1.5
0 6-9Weeks
-
A 10-13Weeks
4 C
X
i
-ic?l
1.0
-
0.5
.
0 -
1
1
L
0
10
20
13-16W~ks
I
I
30
40
00,
50
SN R (%I
FIG. 9. Relationship between increase in leaf area index (LAI) and nitrogen stress (Sm) in a crop of wheat over successive 3-weekly intervals. Nitrogen fertilizer treatments were nil, 56 kg N/ha at sowing, and 112 kg N/ha applied at sowing at 5 and at 10 weeks (taken here as producing zero stress). Data are derived from Halse et al. (1969).
which can be substituted into the expression SNA = 100 [(AM-A)/AMI
(5)
where A is the change in leaf area of the deficient plant over a given time interval and A M is the corresponding change in leaf area for a plant given a nonlimiting dose of nitrogen at the beginning of the interval. The adequacy of leaf area as an estimator of SN could then be tested by plotting SNA against SNC which would be derived from the corresponding dry weight values. The author is not aware of any set of data that completely fulfill these requirements other than a fragment from Bouma (1970a). In this instance, subterranean clover was grown in sterile culture solutions containing 4, 16, or 64 ppm N for 27 days. Among other treatments the 16 ppm plants were split into two groups: (1) the nitrogen level remained at 16 ppm; (2) the nitrogen level was raised to 64 ppm (assumed by me to be nonlimiting). Leaf area was estimated frequently. From Bouma’s Fig. 1 , at day 36, leaf area was 34.5 cm for 16 ppm
NITROGEN STRESS IN PLANTS
27
and 55.9 cm for 66 ppm. Substituting into Eq. (9,SNA = 38%. This value is very close to that for wheat at the same age and nitrogen supply (Greenwood, 1966) as also is the value of 59% derived by my extensive extrapolation from Bouma’s Fig. 1 for plants on 4 ppm N raised to 64 ppm. The technique for assessing the nutrient status of plants used extensively by Bouma in Canberra over the last decade has all the essentials of the split-plot approach for estimating nutrient stress. It is unfortunate for this review that most of his research has been concerned with nutrients other than nitrogen, for it has led to a more specific understanding of the physiology of nutrient response. Because of the importance of the work to the field of nutrient stress, the essence of the technique and results will be given. Bouma’s procedure is to grow subterranean clover seedlings in a nutrient solution containing all essential elements except the nutrient to be studied. The latter is provided as a pretreatment over a range of deficient levels. At a certain time some of the plants on each pretreatment are transferred to complete solutions, this time containing the previously deficient nutrient at a nonlimiting level. The treatment for the remainder is unchanged. On several occasions the leaf area of each plant is estimated using the photographic standards of Williams et al. (1964), and the carbon dioxide exchange rate (CER) is measured under standard conditions with an infrared gas analyzer. The results demonstrate the speed of response of leaf area and CER, and the relative importance of the contribution of leaf area and net assimilation rate. With these and other measurements, Bouma has studied what might be called the physiology of recovery from nutrient deficiencies such as nitrogen (Bouma, 1970a, b), phosphorus (Bouma, 1967a, b, 1969a, 1971, 1975; Bouma and Dowling, 1969a, b), sulfur (Bouma, 1967a, c, 1970c, 1971; Bouma et al., 1972), potassium and magnesium (Bouma 1970c), and boron (Bouma, 1969b). For most of these references it would be possible to compute nutrient stress in terms of leaf area, CER, and dry weight. With the advent of bench and portable photometric devices the measurement of leaf area has become a rapid operation. If the area of intact leaves is measured, by either photographic standards or, better, electronic scanner, then it is feasible to consider using the expansion rate of selected individual leaves. This would be much quicker than measuring all leaves. In conclusion, leaf area increment must rank as one of the simplest and most meaningful parameters of growth for estimating nitrogen stress of plants.
D. CARBON DIOXIDE EXCHANGE RATE (CER)
Photosynthesis being the major process for accumulation of dry matter in the plant, its rate is likely to be closely correlated with growth rate as controlled by nutritional deficiencies. The rate of CER may be considered as the most
28
E. A. N. GREENWOOD
comprehensive single measurement that can be taken to indicate instantaneous growth rate of the plant. So it would seem to be an ideal way of obtaining a rapid assessment of nitrogen stress, or any other nutritional stress. In practice it has some limitations. First, the initial cost of the apparatus for determining carbon dioxide concentration (infrared gas analyzer) is great, its use in the field is a little cumbersome, and it has a high maintenance requirement. Second, the rate of CER varies diurnally and between days and seasons. Hence, a near instantaneous determination of CER is not integrated over these variations and will give a biased result. Of course, a more representative result can be obtained by taking several CER readings but this is tedious. If the value of CER of the deficient plant is E , and the corresponding value for a plant given a nonlimiting dose of nitrogen is E M ,then the expression for nitrogen stress becomes
SNE = 100 [(EM-~IEMI
(6) It seems to me that the use of CER in studying nutritional stress has its greatest application in laboratory or controlled environment studies when CER for a large number of treatments can be compared under a standard environment. Bouma (Section V, C) has made great use of this approach. Wolf and Greenwood (unpublished) found that CER measurements in the laboratory gave very satisfactory results in the estimation of nitrogen stress in wheat seedlings. The application of these techniques to physiological studies in the laboratory is taken further in Section VI. E. CONCLUSIONS
Five plant parameters have been proposed for estimating nitrogen stress-leaf nitrogen, dry weight, leaf elongation, leaf area, and CER. What are the criteria for choosing which one of these parameters (or any other which might be proposed) should be used? It has been argued that it is not feasible to measure nitrogen stress in a way similar to water stress, e.g., chemical potential, and that there is no single measurement that can be made which can be called “growth.” Therefore, for the present, practical considerations should be given high priority. The following criteria are helpful: (1) the parameter that has the greatest meaning for the objectives of the project (what measurements are being made irrespective of the measurement of stress); (2) the equipment and expertise that are available; and (3) the measurement that uses the least resources. VI. Applications
It might be said of agronomists in general that they tend to apply treatments, obtain an end result, and then speculate as to how the measured outcome
NITROGEN STRESS IN PLANTS
29
occurred. Those who rigorously examine processes and their interactions come to be called crop physiologists! The foregoing remarks are indeed an oversimplification and are not intended as a criticism insofar as there are obvious and compelling reasons why agricultural research does concentrate on yield of products. But it must also be said that there is an aversion to studying the processes underlying yield, which is largely due to the daunting complexities and the lack of practical techniques to resolve them. The previous sections of this article have been directed to developing concepts and techniques that promise some easing of these constraints with respect to nutritional aspects of agronomy. This section deals with their practical application with the aim of suggesting the easiest way of doing the job in a variety of circumstances.
A. AGRONOMY AND ECOLOGY
Consider the basic operation of cultivation. Let us say that in certain situations it has been found to improve crop yield. And let us suppose that appropriate research showed that cultivation controlled weeds, thereby reducing competition for light, water, nitrogen, and other nutrients. On the other hand, the supply of available nitrogen and other nutrients was increased through microbial activity and through greater root exploration, . . ,and so on. In the event, how much of the response to cultivation can be ascribed to the nitrogen status of the plant during successive stages of growth? And from this, what deductions can be made about the mechanism and the timing of the nitrogen effect on final yield? Could we avoid cultivation by substituting an appropriately timed application of nitrogen and weedicide, and, if so, which would be the better solution? A comprehensive way of obtaining answers to such questions would be to have a large number of plots allocated to cultivation treatments and rates and times of application of nitrogen fertilizer. A shortcut to answering some of the questions posed, though certainly not all of them, would be to restrict the experiment to the two cultivation treatments upon which SN would be determined over successive periods. If, in addition to nitrogen stress, one were to determine other stresses such as water and phosphorus, then the agronomist would be in a strong position not only to answer some of the questions about nitrogen but also to comment on the importance of nitrogen vis-A-vis phosphorus and water stress. Cultivation has been chosen as an example because data already presented (Table 111) as a partial illustration can be used. A knowledge of nitrogen stress of a crop at a particular time does not provide a direct and accurate answer to the practical question: How much fertilizer is required to reduce stress to an acceptable level? This is the price to be paid for avoiding the response curve. Naturally it provides a partial answer; that is, whether the amount of fertilizer required is zero, a little, or a lot. In most cases
.
30
E. A. N.GREENWOOD
this may be good enough unless the agronomist has available an integrating model that qualifies the current biological requirement by economic and marketing factors and also accounts for change in requirement with time. A decade ago most agronomists worked exclusively in agricultural ecosystems. They could rely on a strong background of research literature and unpublished knowledge on which to base their future research policies. In the seventies, with increasing emphasis on nonagricultural ecology many agronomists have been assigned to work on ecosystems with which they are unfamiliar and for which little “hard” data exist. When one is confronted with a new ecosystem to be studied, the first questions to be asked are, What are the important limiting factors? How do they interact? The answers provide such penetrating insight into an ecosystem that they are almost a prerequisite for rationalizing research priorities. For example, there seems little point in embarking early, if at all, on a research program of plant nutrition or nutrient cycling in an ecosystem if it can be shown that nutrients are not important limiting factors. Similarly, if nutrition is limiting plant growth then which element is the most limiting? The technique of evaluating nutrient stress is an efficient way of rating nutrients as limiting factors. The techniques described here for estimating nitrogen stress can also be adapted for other nutrients (e.g., Bouma er al., 1969), for temperature (Greenwood er aL, 1976), and for water (Power, 1971). The most elegant example of this approach is the study by Power (1971) on the northern Great Plains of the United States. In these grassland ecosystems both nitrogen and water were known to be limiting plant growth. Since rainfall declined to the west it was considered that water stress would increase in that direction and that nitrogen stress would increase to the east. Further, as soil water declined during the growing season, water stress was expected to increase. At any one location and at any one time, what were the respective limitations to plant growth by the two factors? Power estimated SNR by using nitrogen fertilizer. Concurrently, he estimated water stress, SWR (Sw in his article), by a comparable technique in which water stress was removed by irrigation. An extract of the results is given in Table VI. There was a strong interaction with time, but the two stresses were, in general, additive. It became clear that, for the particular situation studied, nitrogen was more limiting than water. The dry weight approach to stress used by Power is obviously appropriate for the arable grasslands of North Dakota. It would not be satisfactory for arid rangelands where the spatial distribution of plants is exceedingly variable and sparse. Here it would seem better to use a nondestructive method such as SNL , which has the real advantage for remote areas that there is no equipment to break down. M. A. Ross and M. Friedel (personal communication) are currently evaluating leaf elongation rate of grasses in arid rangelands of Central Australia with a view to estimating SN and S, as did Power. Ross is studying the selection of
31
NITROGEN STRESS IN PLANTS
TABLE VI Percent Stress on Top Growth of Bromegrass Due to Nitrogen and Water Deficiencies' Sampling interval Stress due to
4/28 to 5/22
5/22 to 6/10
6/10 to 6/30
6/30 to 7/22
56 stress A. Low basal N (0-N) N Water N + Water
73
4 1
57 83
8
61 22 73
104 61 103
B. Medium b a d N (9@N) N Water N + Water
-1 2 4
3 32 49
24 32 54
35 10 58
C. High basal N (270-N)
N
0
Water N + Water LSD (0.5)
4 5
-7 31 23
ns.
18
7 24 33 17
13 32 59 25
'Reproduced from Power (1971).
suitable tillers relating leaf elongation rate to dry weight increments, and developing meaningful ways of applying nutrients and water in remote and a n d locations. He has observed an initial surge in elongation rate for a few days following a subsurface application of water to plants under long-term water stress. A similar study of leaf area increment is envisaged for dicotyledons. If these techniques are successful they will be used to estimate the changes in SN and S, which occur after effective rainfall. A technique for estimating phosphorus status using the dry weight response of detached leaves when placed in solutions with and without the element, has been developed by Bouma and Dowling (1976). It should be adaptable to evaluating nitrogen stress. The approach seems appropriate to those situations where water is nonlimiting and where the standard conditions of response imposed match the cultural conditions of the plant, as in the laboratory. In the field the technique would be most useful where the nutrient in question is the major limiting factor, otherwise some bias in the magnitude of response may develop. Further work on the development of this technique is proceeding (Bouma, personal communication).
32
E. A. N.GREENWOOD B. STRESS PHYSIOLOGY
If functional relationships between nutrient deficiency and specific plant processes are to be established, then two conditions must be met. First, the deficiency must be expressed in terms that arise from the plant itself as distinct from some external supply term. Second, the stress must be expressed numerically. The general failure of nutritional physiologists to meet these two conditions has led to a crippling weakness in research in, and also an avoidance of, this field. It is indicative that the abstracting journal Current Advances in Plant Science cites very few papers dealing with nutrient stress in Section 21 on Stress physiology. It must be admitted that some of the techniques of estimating nitrogen stress, as have been reviewed here, are barely adequate to meet the precision and speed often required by physiologists, particularly in the laboratory. Three suggestions follow. The most suitable technique for physiologists so far discussed is that by Bouma for the determination of CER using the infrared gas analyzer. Either the whole plant or a selected leaf can be used. CER determinations are made in the deficient plant and on a comparable plant for which nitrogen stress has been removed by an adequate addition to the nitrogen supply about 2 days prior to the measurements. The major requirements are a controlled environment chamber, an infrared gas analyzer, and an air-sealed photosynthesis chamber of appropriate geometry (Wolf et aL, 1969). Wolf and Greenwood (unpublished) have designed an extremely simple and sensitive arrangement for determining SNE in grasses and cereals on a relatively large scale. In this system the whole controlled environment chamber becomes the equivalent of the mixing chamber of the above-mentioned air-sealed device in which pots can be placed. The air in the large chamber is homogenized with one or more fans. The leaf is inserted in a glass tube through which the homogenized air is continuously drawn. Reference air is similarly sampled and both streams are led to the external analyzer which determines CER by the difference method. A large battery of plants can be harnessed prior to a run and the tubes may be left on the leaves indefinitely, provided that the leaf does not expand beyond the dimensions of the tube. Speed and precision can be obtained with this arrangement. A manifold of 2-way taps is installed outside the growth chamber so that when the air from a particular leaf is not being analyzed it will still be drawn over the leaf by another pump at the same rate. This reduces equilibration time within the tubing almost to zero and allows successive determinations of CER t o be made at about 2-minute intervals. Some simple precautions must be taken to isolate the air inside the growth chamber from the massive fluctuations in carbon dioxide which may occur outside. The resolution
33
NITROGEN STRESS IN PLANTS
and precision of this technique is demonstrated indirectly in Fig. 4,and, of course, it is nondestructive. Wolf and Greenwood used the technique only for one type of situation, but it seems capable of wide adaptation within a controlled environment. A further nondestructive laboratory technique is available for obtaining shortterm weight changes such as would be required by whole-plant physiologists. This is the weighmg technique of Amott et al. (1974), which permits separate live weighings of tops and roots of plants to be made at frequent intervals and with changes in nitrogen supply. Very good estimates of SN on a fresh weight basis can be obtained with this device while the plants themselves are available for other physiological measurements. Two units would be required for each estimate of stress.
C . MODELING
The concepts of nutrient stress can be of direct use to modelers of plant growth in at least two ways. First, a prior knowledge of the stress values for several nutrients allows the modeler to rank the elements in order of importance. He can then filter out the unimportant elements. Second, if the model first generates a potential growth rate which in turn is successively reduced according to the constraints from each important element, then the stress values themselves will be appropriate terms to govern the magnitude of those reductions without necessarily calling on a nutrient supply subroutine. Where the supply of the nutrient, say nitrogen, is also being generated then stress can be estimated through the following steps.
F;-iG J , ,T , , , -( _ty/ {-iziq = supply
controller
increment
CO, ,etc.
I Actual increment
Nitrogen demand is estimated by computing the product of the potential dry matter increment and the nitrogen concentration in the plant. The concentration is derived from a curve of the time course of nitrogen for a plant grown on a nonlimiting supply of nitrogen. This information requires minimal experimentation. The supply of nitrogen is generated by the supply subroutine. The final step is to reduce the potential dry matter increment by a proportion given by the ratio N supply/N demand, which is, of course, a term similar to nitrogen stress.
34
E. A. N. GREENWOOD VII. Conclusions and Aspirations
In this article the opinion has been expressed that the discipline of plant nutrition has badly neglected the quantification of deficiency. It has been suggested that the conventional approaches to deficiency through response curves and tissue analyses are inadequate bases on which to quantify deficiency rigorously except in special circumstances. By analogy with water stress, an attempt has been made to establish the basic requirements for a concept of nutrient stress and a proposal has been offered on how they might be put into practice. But the latter has proved too difficult to accomplish without some compromise and there has been insufficient experimentation to evaluate fully the several techniques available. It is hoped that this article will stimulate the evolution of plant nutrition whether it be through the development of the ideas presented or through more fruitful ideas arising from them.
ACKNOWLEDGMENTS The author is grateful to Dr.J. F. Power, Dr. M. A. Ross, and Dr. D. D. Wolf for supplying unpublished data and to Dr. N. J. Barrow, Dr. D. Bouma, and Mr.G. B. Taylor for their help with the manuscript. The use of leaf elongation was suggested by Dr. R. F. Williams. The work with Dr. Wolf was done while the author was a guest of Virginia Polytechnic Institute. Special thanks go to Dr.R. C. Rossiter for the many enjoyable arguments over the concepts of nutrient stress.
REFERENCES Allen, R. S., Worthington, R. E., Could, N. R., Jacobson, N. L., and Freeman, A. E. 1961. J. Agr. Food G e m . 9,406-408. Amott, R. A., Brockington, N. R., and Spedding, C. R. W. 1974. J. Exp. Bot. 25, 1124-1 136. Bouma, D. 1967a. Aust. J. Biol. Sci. 20, 51-66. Bouma, D. 1967b.Aust. J. Biol. Sci. 20,601-612. Bouma, D. 1967c. Aust. J. Biol. Sci. 20,613-621. Bouma, D. 1969a. Aust. J. Agric. Res. 20,435-445. Bouma, D. 1969b. Aust. J. Biol. Sci. 22,523-533. Bouma, D. 1970a. Ann. Bot. 34,1131-1142. Bouma, D. 1970b. Ann. Bot. 34,1143-1153. Bouma, D. 1970c. Proc. Int. Grassl. Congr., 11th. 1970 pp. 347-350. Bouma, D. 1971. Aust. J. Agric. Res. 22,723-730. Bourn, D. 1975. J. EXP.Bot. 26,42-59. Bouma, D., and Dowling. E J. 1966. Aust. J. Agric. Res. 17,647-655. Bouma, D., and Dowling, E. J. 1967. Aust. J. Agric. Res. 18,223-233. Bouma, D., and Dowling, E. J. 1969a. Aust. J. Biol. Sci. 22,505-514. Bouma, D., and Dowling, E. J. 1969b. Aust. J. Biol. Sci. 22,515-521.
NITROGEN STRESS IN PLANTS
35
Bouma, D., and Dowling, E. J. 1976.Aust. J. Agric. Res. 27,5342. Bouma, D., Spencer, K., and Dowling, E. J. 1969.Aust. J. Exp. Agric. Anirn. Husb. 9,
329-340. Bouma, D., Greenwood, E. A. N., and Dowling, E. J. 1972. Aust. J. Biol. Sci. 25,
1147-1 156. Chapman, H . D., ed. 1966. “Diagnostic Criteria for Plants and Soils.” Div. of Agric. Sci., Univ. of California. Goodall, D. W., and Gregory, F. G. 1947. Cornrn. Bur. Hort. Plant. Crop (G. B.) Tech. Cornmun. 17. Greenwood, E. A. N. 1966.Plant Soil 24,279-288. Greenwood, E. A. N., and Titmanis, Z. V. 1966.Plant Soil 24,379-389. Greenwood, E.A. N., and Titmanis, Z. V. 1968.Aust. J. Agric. Res. 19,9-14. Greenwood, E. A. N., Goodall, D. W.,and Titmanis, Z.V. 1965.Plant Soil 23,97-116. Greenwood, E. A. N., Boyd, W. J. R., Wiitchead, J. A., and Titmanis, Z. V. 1970.Ausf. J. Exp. Agric. Anirn. Husb. 10,763-767. Greenwood, E. A. N., Carbon, B. A., Rossiter, R. C., and Beresford, J. D. 1976.Aust. J. A g r k Res. (in press). Hake, N. J., Greenwood, E. A. N., Lapins, P., and Boundy, C. A. P. 1969. Aust. J. Agric. Res. 20, 987-998. Hsiao, T. C. 1973.Annu. Rev. Plant Physiol. 24,519-570. Hsiao, T. C., Acevedo, E.,and Henderson, D. W. 1970.Science 168,590-591. Hylton, L. O., Jr., Williams, D. E., Ulrich, A., and Cornelius, D. R. 1964. Crop Sci. 4,
16-19. Power, J. F. 1971.Crop Sci. 63,726-728. Rauschkolb, R. S., Brown, A. L., Quick, J., Prato, J. D., Pelton, R. E., and Kegel, F. R. 1974a. Calif: Agric. 28, 10-12. Rauschkolb, R. S., Brown, A. L., Sailsbury, R. L., Quick, J., Prato, J. D., and Pelton, R. E. 1974b.Calif:Agrk. 28, 12-13. Scott, D. 1961.N.Z.J. Agric. Res. 4,282-285. Smith, P. F. 1962.Ann. Rev. Plant Physiol. 13,81-108. Taylor 1968. In “Water Deficits and Plant Growth” (T. T. Kozlowski, ed.), Vol. 1, pp. 49-72. Academic Press, New York. Titmanis, Z. V., and Greenwood, E. A. N. 1969.Field Stn. Rec. 8,9-16. Ulrich, A. 1950.Soil Sci. 69,291-309. Wadleigh, C. H.,and Gauch, H.G. 1948.Plant Pfiysiol. 23,485-495. Watson, D. J. 1963.Proc. Easter Sch. Agric Sci., 10th pp. 233-247. Williams, C. N., and Biddiscombe, E. F. 1965.Aust. J. Agric. Res. 16,14-22. Williams, R. F.,Evans, L.T., and Ludwig, J. 1964.Aust. J. Agric. Res. 15,231-233. Wolf, D. D., Peace, R. B., Carbon, G.E., and Lee, D. R. 1969.Crop Sci. 9,24-27.
This Page Intentionally Left Blank
STATISTICAL METHODS IN SOIL CLASSIFICATION RESEARCH
.
Rodney J Arkley Department of Soils and Plant Nutrition. College of Natural Resources. University of California. Berkeley. California
............. ..................... ..............................................
I . Introduction: Objectives and Problems of Soil Classification
11. Numerical Taxonomy or Cluster Analysis of Soils
A. GeneralTheory B. Dataselection .............................................. C. Weighting and Standardization of Variables ........................ D. Measures of Similarity or Difference .............................. E Sorting Strategies ............................................ F . Presentation of Results of Sorting Procedures ...................... 111. Ordination of Soils .............................................. A . Q-TypeOrdination ........................................... B. R-Typeordination ........................................... C. Presentation of Results of Ordination ............................. IV . Soil as an Anisotropic Entity ...................................... A . Soil Profile as an Array of Soil Properties .......................... B Soil Data by Layers or Horizons ................................. C. Soil Profile as an Array of Depth Functions ........................ V . Statistical Methods for Comparing Classifications ...................... A . Cophenetic Correlation ........................................ B Coefficient of Association ..................................... C. Wilk's Criterion .............................................. VI . Conclusions and Evaluation ....................................... A The Choice of Methods ........................................ B. A Suggested Procedure for General Soil Classification ................ References ....................................................
.
.
.
.
37 39 39 40 41 45 47 52 54 55 56 58 59 60 61 63 63 64 64 64 64 65 67 68
I . Introduction: Objectives and Problems of Soil Classification
The purpose of classification is to organize the members of a large population of objects into groups or classes of objects so that the nature of. and relationships between. the objects can be more easily understood . This purpose may be limited to understanding related to a specific purpose such as the irrigability of soils and based upon only those attributes relevant to the purpose; or the purpose may be to develop a more general classification based upon as many 37
38
RODNEY J. ARKLEY
attributes as possible and useful for a wide range of purposes. Most of the research reported herein is of the latter kind with a few exceptions. Prior to 1955 when Hughes and Lindley first used a statistical procedure to reclassify members of six soil series, soil classification was based primarily on subjective judgment. That is not to say that the judgment applied has not been good; some of the best minds in soil science have been focused on soil classification. Nevertheless, both the selection of criterion variables for classification, their effective weighting by application at different categorical levels in a hierarchical classification, and of boundary values for separations have all been made primarily on the basis of fallible human judgment. Although some statistical analysis was carried out on soil data prior to 1955, the amount was limited primarily by the tedium of statistical analysis and the hand sorting of data. However, with the development of the electronic computer this tedium has been removed, and the application of numerical and statistical methods to soil classification has developed rapidly as indicated by the number of references cited in this paper, most of which deal directly with soil classification. To be both comprehensible and most effective, the differentiating characteristics or criterion variables used to form classes should contain the maximum possible information. That is, they should be those which have the most predictive value for the nature and behavior of the soil when subject to external influences. These criterion variables then should be those which are covariant with other properties (accessory characteristics) not used as criterion variables. The number of soil characteristics that might be used for soil classification is very large, and the selection of criterion variables from this list is very likely to be suboptimal by subjective methods. However, the covariance among all variables can readily be examined with the use of the computer by calculating the product-moment correlation coefficient between all pairs of variables and with direct examination or analysis of the resulting correlation matrix. On this basis an optimal set of differentiating characteristics can be chosen. Gibbons (1968) effectively argued the importance of covariance to the usefulness of soil classification. The extent to which soil properties are covariant will be discussed later in this paper. Soil classification in the past has been primarily hierarchical in nature, with one or more criteria used at each categorical level to divide soils into mutually exclusive classes. Such a classification is helpful in the understanding of relationships among soils, but some relationships may be seriously distorted. This occurs when a group of soils which is relatively similar in all other respects, is subdivided into two groups at all lower categorical levels by small differences in a particular differentiating characteristic. This is a general problem of hierarchical classification, in that it is a divisive procedure and the divisions are dichotomous; that is, they either have or have not a certain property or the value of a certain measured property is above or below a specified level. Avery (1968)
STATISTICAL METHODS IN SOIL CLASSIFICATION RESEARCH
39
points out that a successful hierarchical classification can be made only if the differentiating criteria can be ordered in accordance with the number of other attributes (accessory characteristics) associated with them. He points out further that soil variation is not of this character, presumably because soil characteristics result from the interaction of several factors. This is the crux of a major problem in soil classification. Are soils to be considered as made up of discrete natural individuals, or do the individual soil profiles represent points in a multivariate continuum, which considered as a whole contains no distinct boundaries for purposes of classification? Experience has shown that soils may occur as discrete, relatively homogeneous bodies when considered only within a local area. But as investigations extend to broader areas more and more soils of intermediate character are encountered which bridge the gaps between the original discrete soil bodies. This leads to problems in soil correlation. The recognition and identification of soil classes, such as soil series in the field, in the face of this kind of soil variation is the bane of the soil surveyor’s existnece. His decisions must be made primarily on the basis of field observations of a number of soil characteristics, and the soils classified on the basis of variations in one or many of these simultaneously. Avery (1968) argues forcefully that a coordinate system of classification is more appropriate for soils than a hierarchical system. The advantage to a coordinate system being that each differentiating criterion is given equal weight, at least a priori. It should be pointed out also that a coordinate system can be used to construct a number of hierarchical systems, by arranging the differentiating criteria in different categorical orders. In the following discussion of the various statistical methods applied to soil classification, it will be observed that much of the work has assumed that soils (mainly soil profiles) fall into natural clusters or groups which can then be ordered into a classification. For the limited sets of soils used, this appears to be a correct assumption. Although little work has been done toward the development of a coordinate classification scheme, it is made clear that some of the statistical methods described can be used effectively for this purpose assuming that soil characteristics vary in such a way as to form a continuum through the whole population of soils. A method is suggested by which soils can be classified using a set of well-separated centroids or conceptual modal soils and classes formed on the basis of the general affinity of real soils to the centroids. II. Numerical Taxonomy or Cluster Analysis of Soils
A. GENERAL THEORY
Numerical taxonomy is defined by Sneath and Sokal (1973) as the grouping by numerical methods of taxonomic units into taxa on the basis of their
40
RODNEY J. ARKLEY
character states. Groupings are formed using the following general procedures: First, data for a number of units, such as soil profiles, are assembled including a sizable number of selected variable properties for each unit. Because this discussion is confined to the classification of soil units, they will be simply referred to as soils. The data are commonly arranged into a matrix consisting of soils by columns and soil properties by rows. Next an over-all estimate of resemblance is obtained between pairs of soils by some mathematical function of all differences between the values for each property of the two soils. After numerical values for the estimate of resemblance (either estimates of similarity or of difference can be used) between all pairs of soils included in the study are obtained, the matrix of n(n-1)/2 values is subjected to a sorting strategy which forms groups of similar soils. The nature of the groups formed and their relationships or taxonomic structure can be presented in various ways; these may include dendrograms, reordered matrices, ordination, or simply tables of coordinates.
B. DATA SELECTION
The choice of soils to be included in the data should be such that the number of soils is large and the general kinds of soils included are well represented. For example if the soils included are mainly well drained and without evidence of wetness, then the inclusion of a very few poorly drained soils may interfere with the analysis because those. soil properties associated with wetness may not be representative of the range of variation in those properties. Tllis is particularly important in cluster analysis as the order of cluster formation is affected by the number of soils in the clusters or groups formed in some clustering techniques. The selection of soil properties is even more important to a successful analysis than the selection of soils. Although all kinds of both field and laboratory data can be used, there are certain kinds that should be excluded or that need special treatment. Some soil properties, such as the field moisture content, are generally irrelevant to soil classification and should be excluded. Logically correlated properties, such as dry and moist colors, are generally so highly covariant that one or the other should be included. Particle size distribution values for sand, silt, and clay always add up to 100% and so one of the three should be eliminated from the data. The inclusion of large numbers of logically related properties should be avoided, as they tend to create an inadvertent extra weight to such a group of properties in the classification. For example, in the initial list of properties used by Sarker et ul. (1966) were 6 particle size ratios, 5 of which were intercorrelated above the 0.90 level. This kind of redundancy among properties should be avoided or dealt with by analysis of variables as discussed later in this paper. Soil data obtained from laboratory analysis are almost always continuous
STATISTICAL METHODS IN SOIL CLASSIFICATION RESEARCH
41
variables as are a number of field measurements such as thickness and depths of recognizable characters. Hue, value, and chroma (soil color) are continuous variables even though the color chart commonly used is made up with discrete steps. However, field observations such as soil texture, structure, and consistency are usually discrete or multistate variables and need to be treated with special care. Data of this kind should be coded in such a way as to reflect their proper rank order reflecting their importance to soil behavior or development. For example, soil texture classes might be coded in order of their relative clay content or water retention characteristics. Structure is a particularly difficult soil property to code as a ranked multistate variable. For surface soils an appropriate order of structure type might be single grained, massive, platy, crumb, granular, subangular blocky and angular blocky; for subsoil layers the order might be single grained, massive, platy, granular, subangular blocky , angular blocky, prismatic, and columnar. Structure grade or distinctness such as weak, moderate, and strong coded 1, 2, 3 might well be multiplied by the code for structure type coded 0 to n and a value for size of peds added to give a single value for type-grade-size of structure. The system suggested is only one of many possible ways that soil structure might be treated. Arkley (1971) and Cipra et al. (1970) have used type and grade omitting structure size with some success. Barkham and Norris (1970) treated soil structure type, grade, and size as separate characters. Soil color mottling is troublesome to code. Cipra et al. (1970) used a combination of abundance, size, and contrast of mottles scaled from 0 to 8. Cuanalo and Webster (1970) used abundance percent and position on peds separately. Rayner (1966) used abundance, size, and contrast as separate variables. Dichotomies such as the presence or absence of earthworms, concretions, carbonates, iron pans, manganese stains are sometimes used (Rayner, 1969; Muir et al., 1970). Dichotomies require special treatment which will be discussed in relation to the standardization of variables. The same is true for unranked multistate variables.
C. WEIGHTING AND STANDARDIZATION OF VARIABLES
I . The Problem of Weighting of Variables Sneath and Sokal (1973) present cogent arguments in favor of weighting all variables equally, especially where a classification is intended t o be a “natural” or basic classification for general use rather than one for a specific objective. These arguments against “a priori” weighting appear to be on sound rational grounds. This is in direct opposition to the methods of orthodox or conventional
42
RODNEY J. ARKLEY
hierarchical classification wherein the differentiating characteristics used at higher categorical levels take precedence over those at lower levels, and therefore have greater effective “weight.” For example, in the Soil Classification of the Soil Survey Staff, U.S. Department of Agriculture (1960), certain diagnostic horizons are considered more important than others; a case in point is the use of the mollic epipedon at the “Order” level to separate the Mollisol soil order from other orders, irrespective of the nature of the subsoil horizons to a large degree, whereas most of the other orders are separated on the basis of the nature of subsoil horizons. Decision such as the one cited are based on intuition or human judgment, both of which are fallible. Sneath and Sokal (1973) also argue in favor of the use of a large number of variables (i.e., soil properties) in numerical taxonomy, on the grounds that the use of variables greatly evens out the effective weight which each one contributes. This argument presupposes that all pertinent groups of covariant properties are about equally represented in the data. In the data used for numerical classification research on soils, this is clearly not true in many cases. Some kinds of measurement on soil properties are more easily obtained than others, or have been of more interest to the investigator, and so are overrepresented and thus unduly weighted. Also the use of a large number of variables involves a great deal of time and expense, especially if the classification is intended to encompass a large number of individuals. However, in the first stages of analysis, the use of a large number of variables standardized so as to give equal weight to each is certainly a sound approach. For the final classification it may be possible to reduce the number of variables to a manageable but still effective size by analysis of the covariance among them.
2. Covariant Soil Variables In the past, covariance among soil variables was rarely analyzed, but with the advent of electronic computers the tedium of the calculation of correlation coefficients has been removed. In several papers involving numerical taxonomy of soils, correlation matrices have been published, revealing how much covariance exist among soil variables. Moore and Russell (1967) analyzed 10 trace elements in 28 soil profiles and found that the correlation matrix of 45 r-values contained 27 which were significant (P < 0.01) and ranging from 0.49 to 0.90; Moore et al. (1972) show 50 of 91 r-values significant (P < 0.01) ranging from 0.18 to 0.80 for 14 variables in 4 layers of 40 soils. Sacker et al. (1966) found that 39 of 61 soil properties were correlated with at least one other at the level of r > 0.50; Russell and Moore (1967) show 57 of 136 r-values significant (P>0.01) ranging from 0.40 to 0.80 for 17 variables and 43 soils. Reexamination of my own data sets used for analysis of variables also revealed similar levels of communality among variables as shown in Table I (Arkley, 1971).
STATISTICAL METHODS IN SOIL CLASSIFICATION RESEARCH
43
TABLE I Number of Variables SigniFcantly Correlated with other Variables in Analyses Reported by Arkley (1971) Variables with rvalues significant at P < 0.01
Number of
Soils
Variables
59
621 220 81 87
1+
>2
21 23 34
4 1 7 0 23
8 21
0
34
44
0
53
2
44 51
30 42 46
0
>4 1
20 26 40 38
The extensive covariance among soil variables in the widely differing data sets described in Table I is strong evidence that a long list of soil variables is not necessary to classify soils effectively by either conventional or numerical methods. Also, it is evident that analysis of variables should be among the first steps in the development of a classification system. There should be no objection to weighting variables according to their predictive values as revealed by the analysis of variables as this would not be considered a priori weighting.
3. Standardization of Variables Most procedures for obtaining an estimate of resemblance require that the variables be standardized to a common range of values. It is clearly inappropriate to compare differences in a variable with a range of 0.0 to 1.0 with those in a variable with a range of 100 to 1000. For continuous variables standardization may be by range, i.e.,
x' = (x- Xndn) I x,,
-~ m i 1 n
or by variance, i.e .,
X'=(X-X)/SDx The former gives each variable a range of 0.0 to 1.O, the latter a mean of 0.0 and a standard deviation of k1.0. These methods can also be applied to ranked multistate variables, but with more risk of injecting spurious information. For data containing both continuous and discrete variables, either dichotomous or multistate, Crigal and Arneman (1969) applied a method proposed by Talkington (1967): For a variable that can assume a number of discrete and mutually exclusive states (i.e., soil structure types) which is coded as 1.O for no
44
RODNEY J. ARKLEY
more than one of these states and 0.0 for all others, the maximum possible contribution to the differences or the sum of the square of the differences between two individuals is 2.0 (Table 11). Continuous variables are therefore standardized so that the maximum contribution to the squared difference is also 2.0; such variables are standardized by range and then multiplied by 21'2 or 1.414.
Another method for equalizing the contribution of discrete and continuous variables based upon information theory has been developed by Burr (1 968) but so far has not been applied to soils. Continuous variables are standardized to a mean of 0 and a standard deviation of +21n by the formula
x'=(X-X)/(1.414X SDx) And dichotomous and multistate variables by the formula
M 2= M(t-1) /
[2tp, (s, - I)]
where M' is the standardized variate, M is an unstandardized variate (as coded in Table 11), t is the total number of individuals (soils) with nonmissing data,p, = the proportion of t in state s(sn/t), sn is the number of individuals in state s, and ,S is the number of possible states of variate M. Burr proposes to call this procedure standardizing by reciprocal proportions since the weight of each state is weighted inversely to its frequency of occurrence @,). The formula given above weights a multistate variate M equally with a continuous variable. If one considers that each state s should be weighted equally with a continuous variable, then the parameter S (, - 1) can be omitted from the formula. The problem of highly skewed data should be considered in the standardization of variables. In some cases it would be appropriate to use a logarithmic or square root transform for known skewed distributions as was done by Moore and Russell (1967). Talkington (1967) advocates a slight truncation of the range for extreme values which occur very rarely as was done by Grigal and Arneman TABLE I1 Individuals Variable states (s) $1
Sl s3
84
Difference A
B
(4
0 1 0 0
0 0 0 1
0 1 0 1 2
d' 0 1 0 1
2
STATISTICAL METHODS IN SOIL CLASSIFICATION RESEARCH
45
(1969). In this connection it should be pointed out that the use of ratios as data variables may well lead to highly skewed distributions because of the hyperbolic nature of ratios; in general they should be avoided. In any case a careful examination of the data for errors or aberrant data should precede the analysis, simply as a good analytical practice. Where highly skewed variables are suspected, they should be examined in a frequency distribution and perhaps plotted against related variables in a scatter diagram before a decision is made as to their treatment. D. MEASURES OF SIMILARITY OR DIFFERENCE
Various procedures are available for calculating an over-all estimate of resemblance, which can be based upon measures of either similarities or differences. Sneath and Sokal (1973) use the term “similarity coefficient” t o cover coefficients both of similarity and of dissimilarity; the one being the complement of the other. They describe four somewhat fuzzy classes of similarity coefficients as (1) distance Coefficients, (2) association coefficients, (3) correlation coefficients, and (4) probabilistic coefficients, of which the first and third have been used in soil studies most commonly. 1. Distance Coefficients
The simplest practical form of “distance” measure called mean character difference (MCD) is: where Xi . . . n are standardized variates and j and k are two individuals such as soil profiles. However, this coefficient is rarely used and suffers from the fact that a large difference in a single variable is inadequately represented in the coefficient. MCD has been applied by Moore and Russell (1967) and Webster and Burrough (1972). A much more commonly used distance coefficient is the familiar Euclidean distance (d) in the form: d , k = [ a i (X,j-X,)’ i= 1
1
lR
The expression l / n is introduced into the equation in order to equalize differences introduced by missing data, or differing numbers of variates used. The average Euclidean distance coefficient has the advantage of being more readily visualized and can be plotted in two or three dimensiocc although not in n-space where n is greater than 3. It also has other statistical properties which are advantageous. This coefficient has been used in soil studies by Cipra et al. (1970), Crichton (1975), Grigal and Arneman (1969), Moore et al. (1972),
46
RODNEY J. ARKLEY
Lamp (19721, and Webster and Burrough (1972). Cuanalo and Webster (1970) used d2 rather than d. Another distance that has been used several times is referred to as the Canberra metric (dc). It was developed by Lance and Williams (1967b) and has the advantage of needing no prior standardization of variables. It is in the form of n
dc
=z
K
r - 8
lO0OC
I - 6
'001
- 4
0
- 2
L 150
0
200
250
300
350
J -50 '2
D A Y OF YEAR
FIG. 1. Changes in the dry weight of roots, shoots, and grain, in LA1 and in leaf water potential during growth of a wheat crop. (Adapted from Connor, 1975.)
1974), but the fall in LAI is often slower and more delayed (cf. Fig. 4) in both maize (Allison, 1964) and sorghum (Fischer and Wilson, 1975c; Soza, 1973).
VI. Leaf Photosynthesis
Photosynthesis provides most of the increase in crop dry weight as well as the metabolic energy required for crop development. The course of crop photosynthesis must therefore be a major determinant of crop yield. Throughout the early life of cereal crops the leaf blades are the main photosynthetic organs and crop growth rate depends on both the rate of expansion of leaf area and the rate of photosynthesis per unit leaf area. Once the leaf canopy has closed, leaf photosynthetic rate becomes the more important determinant, depending not only on weather conditions but also on the geometry of the canopy. Toward the end of the life cycle photosynthesis by the stems, leaf sheaths, and inflorescences tends to becomes increasingly important as the leaves senesce, especially in temperate small grain crops subject to drought stress.
A. DIFFERENCES BETWEEN CJ AND Cq CEREALS
The action spectrum for photosynthesis by leaves of all the major cereal groups was found by McCree (1972) to be closely similar in wheat, oats, barley, triticale, rice, maize, and sorghum. Nevertheless, there exists within the cereals a major difference in the carbon pathway in photosynthesis, which has profound
318
L. T. EVANS AND I. F. WARDLAW
implications for both their productivity and their range of adaptation. Rice and the temperate small grain cereals depend entirely on the Calvin (C,) cycle, whereas in maize, sorghum, Italian millet (Setaria itulicu), panic millet (Panicum miliaceurn), finger millet (Eleusine coracaw), and probably pearl millet (Pennisetum typhoides) the Calvin cycle is preceded by C 0 2 furation in the C4-dicarboxylic acids (Downton, 1971). Several characteristics comprising the “C4 syndrome” have been used to establish that the tropical cereals other than rice operate by the C4 pathway, such as the identification of the earliest products of photosynthesis (Hatch et al., 1967), a low C02 compensation point (e.g., Downton and Tregunna, 1968), or the lack of enhancement of the photosynthetic rate at low oxygen concentrations (Downes and Hesketh, 1968). Also associated with the C4 pathway in cereals are other characteristics such as the Kranz anatomy of the leaves and closer spacing of the veins which may influence translocation as well as photosynthesis. It now seems that the Michaelis-Menten constant for the carboxylating enzyme in C3 plants (RuDP-carboxylase) is not much greater than that of PEP carboxylase which mediates the primary carboxylation in C4 plants. More importantly, refixation of C 0 2 by RuDP carboxylase in the bundle sheath cells of C4 cereals, after transfer and decarboxylation of malate or aspartate from the mesophyll, probably takes place at a much higher C02 concentration, and is therefore less susceptible to photorespiratory decarboxylation. Consequently, the minimum values for mesophyll (or residual) resistance to C02 uptake by leaves appear to be substantially lower in the C4 cereals than in the Calvin cycle species. For example, r,,, values of 0.7-0.9 s cm-’ (Gifford and Musgrave, 1973) and 1.0 s cm-’ (El Sharkawy and Hesketh, 1965) for maize are to be compared with minimum values of 4.1 s cm-’ for oats (El Sharkawy and Hesketh, 1965) and 2.7-3.1 s cm-’ for a range of wheat species (Dunstone et al., 1973). The smaller r,,, of the leaves of C4 compared with C3 cereals tends to be associated with a greater stornatal resistance, r,, hence the greater efficiency of the Cq cereals in dry matter production per unit of water transpired. Stomatal opening in the C4 cereals increases up to very high flux densities of light, as in maize (Gifford, 1971). Consequently, whereas photosynthesis by single leaves of the C3 cereals tends to reach light saturation at - 4 full sunlight, that of the C4 cereals increases with increasing intensity up to full sunlight (Hesketh and Musgrave, 1962; Hesketh, 1963). Even so, the minimum recorded gas phase resistances to C02 uptake tend to be rather lower in the C3 cereals, e.g., 0.7-0.8 s cm-’ in wheat (Dunstone et ul.. 1973) compared with 1.5 s cm-’ in maize (El Sharkawy and Hesketh, 1965). The maximum photosynthetic rates achieved by the C4 cereals are distinctly greater than those of the C3 cereals. Rates up to 240-280 ng C 0 2 cm-2 s? have been recorded in maize (Heichel and Musgrave, 1969), sorghum (Downes, 1971), and pearl millet (McPherson and Slatyer, 1973), compared with up to 120 ng
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
319
cm-2 C1 in wheat cultivars at atmospheric C02 levels. However, rates up to 200 ng cm-2 S’ have been measured in the wild diploid wheats (Evans and Dunstone, 1970). The greater rates of photosynthesis in C4 plants, associated with their reduced photorespiratory losses and other characteristics listed above, have led to them being referred to as “efficient” plants, and to a search for nonphotorespiring forms among the temperate cereals such as oats and wheat (Moss and Musgrave, 197l), without success. Other characteristics of the C4 pathway in cereals should be considered, however. Although their greater photosynthetic rate may be of advantage at high light intensities, especially in view of their more efficient use of water, at low light intensities-in overcast weather or for leaves deep in the canopy-their photosynthetic rate may be less than that of C3 plants. This is particularly true when cooler temperatures are associated with low light, the conditions under which the temperate cereals usually make their early growth. Murata and Iyama (1963; Murata et ul., 196Sb) first drew attention to the poor photosynthetic performance of the tropical (C4) grasses at cool temperatures (-10°C) relative to temperate (C,) grasses and cereals such as barley and wheat. At high temperatures, on the other hand, photosynthesis by the C3 cereals falls off rapidly at temperatures above 30”C, as in wheat (Stoy, 1965; Friend, 1966; Sawada, 1970), whereas photosynthesis by the C4 cereals may reach its peak at temperatures of 30”-40”C, as in sorghum (El Sharkawy and Hesketh, 1964), maize (Hofstra and Hesketh, 1969), and pearl millet (McPherson and Slatyer, 1973). Although rice is a crop of tropical origin, its photosynthetic response to temperature resembles that of the other C3 cereals in having a broad optimum (Murata, 1961) with a rapid fall in rate at high temperatures. Cool temperatures, however, have a more adverse effect on photosynthesis in rice than in the temperate cereals, particularly on the indicu varieties (Kishitani and Tsunoda, 1974).
B. DIFFERENCES BETWEEN CULTIVARS
Between the C3 and C4 cereals, and to a smaller extent between indicu and juponicu rices, there are substantial differences in the rate of photosynthesis and in its environmental adaptation. To what extent are there comparable differences among the cultivars of each major cereal? What role have they played in crop evolution and selection for yield? In trying to answer these questions we are confronted with major problems of how to make valid comparisons of leaf photosynthesis rates, since these are so dependent not only on the conditions under which the rates are measured, but also on the rank, age, and previous treatment of the leaf, as well as on the demand within the plant for its photosynthetic products.
320
L. T. EVANS AND I. F. WARDLAW
1. Factors Complicating Comparisons
The ontogenetic rank of cereal leaves influences their peak photosynthetic rate under defined conditions of temperature, light intensity, and C02 concentration to a substantial degree, as shown for wheat by Sawada (1970) and Dunstone et al., (1973). Leaf age also has a pronounced effect and its time course seems to vary a great deal depending on conditions. In wheat, for example, the rate of flag leaf photosynthesis may fall progressively with time from anthesis, the fall being particularly rapid in the more primitive wheats (Evans and Dunstone, 1970). In more productive modern cultivars, on the other hand, the rate may fall after anthesis-when the demands of grain growth are low and stem reserves are being mobilized-only to rise again until grain growth slows (e.g., Evans and Rawson, 1970; Rawson and Evans, 1971). This pattern of change was also found by Rawson et al. (1976) for plants in which tillers were continuously defoliated, whereas flag leaves on intact plants of the same cultivar displayed a progressive fall in photosynthetic rate. Thus, the time course of photosynthetic rate in flag leaves of wheat appears to vary in response to the demand for assimilates from it. The nutritional condition of the crop also influences the time course of leaf photosynthesis, particularly the availability of N, but also the K content (e.g., Murata, 1961). A high correlation between % N in the leaf tissue and photosynthetic rate has been established in several cereals, e.g. wheat (Khan and Tsunoda, 1970b), barley (Natr, 1973), maize (Ryle and Hesketh, 1969), and rice (Takano and Tsunoda, 1971). Light and temperature conditions during leaf development may profoundly influence the subsequent photosynthetic capacity of leaves. In maize, growth under high temperature or light intensity increases the subsequent photosynthetic rate (Hesketh, 1968). Although photosynthesis in modern wheat cultivars was not greatly influenced by earlier light conditions, that in the primitive wheats was greatly affected (Dunstone et al., 1973).
2. Assimilate Demand and Photosynthetic Rate The important question to ask in this context is whether control of photosynthetic rate by the demand for assimilates can occur in cereals. The extent to which it does so will presumably depend on experimental and crop conditions. Thorne (1973) accepts that the phenomenon occurs in other crops, but not in cereals. Kiesselbach (1948) found that removal of maize ears at silking reduced final crop dry weight by 27%, which he considered was due to a feedback inhibition of photosynthesis in the absence of the major sink for assimilates. Moss (1962) subsequently demonstrated an initial decline in photosynthetic rate when fertilization of the ear was prevented in maize. With wheat plants, Birecka and
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
32 1
Dakik-Wlodkowska (1 963) recorded a substantial reduction in total photosynthesis following ear removal. King er al. (1967) found the rate of flag leaf photosynthesis in wheat to fall by about 40% within one day of ear removal, and to be restored to a high level when the flag leaf was made to support the whole plant because all other leaves were removed or shaded. Similarly, the low rates of flag leaf photosynthesis in plants held in continuous light were increased when ear photosynthesis was inhibited with DCMU. These experiments thus showed that leaf photosynthesis in wheat can be reversibly lowered and raised as the internal demand for assimilates is lowered and raised. The changing pattern of demand for assimilates was demonstrated by concurrent analysis of l4 C distribution within the plants. Recent experiments by Rawson er al. (1976) also found flag leaf photosynthesis to be decreased by partial grain removal and increased by inhibition of ear photosynthesis, but not to as great an extent as in the experiments of King et al. (1967). These effects appeared to be due mainly to changes in stomata1 resistance of the abaxial leaf surface. Puckridge (1969) also presents evidence of reversible control of photosynthetic rate in wheat according to demand, his measurements and treatments being on a crop canopy in the field. Romer (cited by Michael et al., 1973) found that shading of barley ears led to a doubling of the rate of flag leaf photosynthesis and an increase in the proportion of its assimilates exported to the ear from 51 to 86%. Flag leaf removal led to a rise in the rate of photosynthesis by the penultimate leaf in his experiments, and to a rise in ear photosynthesis in those of Johnson er al. (1975). With oats, also, removal of the lower leaves increased the photosynthetic rate of the flag leaf (Criswell and Shibles, 1972), although removal of spikelets did not reduce it. Thus, there is clear evidence, with many of the cereals, that the photosynthetic rate of leaves can respond to internal changes in the demand for assimilates. Photosynthetic rate not only decreases when demand is reduced and alternative sinks for assimilate are limited, but can also increase, indicating the existence of spare photosynthetic capacity within the plant. However, failure to find photosynthetic rate responsive to demand has also been reported for wheat (Lupton, 1968; Dantuma, 1973; Austin and Edrich, 1975) and barley (Apel and Peisker, 1973; Natr, 1967; Apel et al., 1973). Low photosynthetic rates were recorded in many of these experiments, and Austin and Edrich (1975) estimated that only low rates were required in their experiments. However, Rawson er al. (1976) found little evidence of response in photosynthetic rate to demand even when grain growth and photosynthesis were rapid, as long as tillers were allowed to grow, but not when they were continually defoliated. Clearly, whether or not such a response is observed depends on both experimental and crop growth conditions, but there should be little doubt that it can occur in cereals.
322
L. T. EVANS AND I. F. WARDLAW
3. Cultivar Differences Given the complications discussed above, it is not easy to make satisfactory comparisons between cultivars in their photosynthetic rate, and even harder to assess their relation to yield differences. From a survey of 22 races of maize ranging from ancient varieties to modern hybrids, Duncan and Hesketh (1968) found no evidence that improvement of maize over the centuries has been associated with an increase in photosynthetic rate. There were, however, differences between races in the way their photosynthesis responded to temperature, which appeared to be adaptive. High altitude races, for example, had relatively lower rates at high temperatures. Differences between maize cultivars in their photosynthetic rate have been reported, but they appear to be very dependent on environmental conditions. Those found by Heichel and Musgrave (1969) in the Philippines, for example, were not always apparent at Comell (Gifford, 1970). Similarly, although heterosis in photosynthetic rate has been reported in some instances (e.g., Fousova and Avratovschukova, 1967; Heichel and Musgrave, 1969; Derieux et al., 1973), there are others where it was not found (e.g., Duncan and Hesketh, 1968). Heichel (1971) associated the lower photosynthetic rate of one cultivar with greater respiratory loss. Substantial differences in photosynthetic rate per unit leaf area have been found in wheat. The highest rates have been found in the primitive and wild diploid wheats (Evans and Dunstone, 1970; Khan and Tsunoda, 1970a; Dunstone et al., 1973), with a progressive fall in the course of evolution. Both between species and among modern cultivars the relation between maximum photosynthetic rate and leaf area is negative (Evans and Dunstone, 1970; Gale et al., 1974). Photorespiration was proportional to photosynthetic rate across many lines (Dunstone et al., 1973), and the latter was negatively correlated with mesophyll cell size (Dunstone and Evans, 1974), but only inconsistently with specific leaf weight (Khan and Tsunoda, 1970b; Dunstone et al., 1973). Dantuma (1973) has recorded limited differences between cultivars of wheat and barley in their photosynthetic rate, as has Murata (1964) for rice. Murata considered there was some correlation between photosynthetic rate and yield but his claim is not convincing in view of the low and narrow range of rates presented. Correlations between photosynthetic rate and crop dry weight or relative growth rate among 12 rice varieties were highly inconsistent (Osada and Murata, 1965). Taken overall, there is no clear evidence among the cereals of any increase in leaf photosynthetic rate in the course of either their earlier domestication or their recent improvement in yield ability. This is rather surprising in view of the central role of photosynthesis in relation to crop growth and yield, and of its rate per unit leaf area as a determinant of the closed canopy rate. Possibly some
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
323
counterproductive feature may be associated with further increase in photosynthetic rate. There is, for example, the negative relation between cell and leaf size and photosynthetic rate in wheat, and the positive relation between leaf and grain size. Consequently, insofar as selection for increased yield involved increased grain size and therefore leaf size, this would have operated against any increase in photosynthetic rate. However, these negative correlations could probably be broken by selection as Wilson and Cooper (1970) have shown with ryegrass. Another possibly counterproductive association may be the inverse relation between the maximum rate and the duration of photosynthetic activity evident among the wheats (Evans and Dunstone, 1970). At present no clear answer can be given to this important question, which merits much more investigation among the cereals. VII. Canopy Photosynthesis
In the early stages of crop growth, the main determinant of its photosynthesis is the extent of leaf area development. As the leaf area index (LAI) increases, so does the extent of light interception, which exceeds 95% for most cereal crops with an LAI of about 4. Once the canopy is closed in this way, further increase in LAI has little effect on crop photosynthesis, which is then most influenced by the incident radiation and the structure of the canopy. Light interception by cereal crops is low during early establishment but then increases rapidly as the larger upper leaves expand and tillers develop. Earlier closure of the canopy can be effected by increase in sowing density, as Williams et al. (1968) have shown for maize. However, although this increases the early crop growth rate, in proportion to the increase in light interception (Williamser al., 1965), it may have little effect on subsequent crop growth rate and an adverse effect on grain yield. Canopy photosynthesis appears to increase asymptotically with increase in LAI, reaching a broad plateau at LA1 values above 4, as illustrated in Fig. 2 for wheat. This shows excellent agreement between field and phytotron-grown crops, and Puckridge and Ratkowsky (1971) found that, at a given level of incident radiation, the relation between net photosynthesis and LA1 was unaffected by either cultivar or sowing density. It is probably similar in the other cereals. The relation between canopy photosynthesis and incident radiation depends on LAI, species, and canopy structure. Both aerodynamic and enclosed canopy measurements suggest that the maximum rates of crop photosynthesis at high levels of incident radiation can be almost twice as high in maize as in wheat (cf. Baker and Musgrave, 1964; Puckridge and Ratkowsky, 1971; Gifford, 1974a). On the other hand, C3 crops may have the advantage at low radiation levels,
324
L. T. EVANS AND I. F. WARDLAW
-
2
4
I
I
6
8
I I 1 0 1 2
FIG.2. The relation between photosynthesis, respiration, and LAI in wheat crops (from Evans et al., 1975). The line is for a field crop, the circles for a crop in an artificially lit cabinet.
possibly because of greater stomatal closure in C4 cereals under weak light. Photosynthesis X radiation curves for crops of barley (Biscoe et al., 1975a), rice (Tanaka, 1972), and sorghum (Allen et al., 1974) were obtained in such different conditions as to preclude comparisons. The maximum crop growth rates (CGR) recorded for the various cereals over periods of a week or more may be more validly compared. There are many records of very high CGR for the C4 cereals. Up to 78 g m-' 6' has been recorded for maize by Haggar and Couper (1972), and rates of more than 50 g mm2 d-' by Kowal and Kassam (1973), Takeda and Akiyama (1973) and Williams el al. (1965). With sorghum, CGR of up to 46-51 g m-' d-' are recorded by Fischer and Wilson (1975~)and Williams et al. (1965). Pearl millet has also been found to sustain a high CGR, 54 g m-' d-' (Begg, 1965), and in a comparative study in the Nigerian savannah, Kassam and Kowal (1973) measured rates of 4 6 4 8 g m-' d-' in millet, 30-50 g m-' d-' in maize, and 3 0 4 0 g m-' d-' in sorghum. Crop growth rates for C3 plants are usually, but not always, substantially lower (Gifford,. 1974a; Evans, 1975). Among the C3 cereals, only in rice has a CGR of more than 50 g m-' d-' been recorded (Tanaka er al., 1966). In the very extensive Japanese IBP experiments, however, the highest CGR found for rice was 36 g m-' d-' whereas 52 g m-' d-' was found for maize, and the average values for the latter crop were also higher (Murata and Togari, 1975). Crop growth rates recorded for the temperate cereals are usually quite low, mostly
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
325
less than 20 g m-2 d-' , although a CGR of 30 g m-2 d-' has been measured in wheat (King and Evans, 1967). Thus, in terms of the maximum CGR recorded, the C4 cereals appear to have a major advantage over the temperate C3 cereals. In fact, this advantage is almost too great to be accounted for solely in terms of greater leaf photosynthetic capacity. Associated with the C4 syndrome are many other characteristics, such as better adaptation to high temperature conditions, faster and more complete export of leaf assimilates, and higher maximum relative growth rates, which may also contribute to their higher CGR. So too may differences in canopy architecture.
A. CANOPY ARCHITECTURE
Individual plants of many C4 cereals are much larger than those of the C3 cereals. The size difference is reflected in the optimum planting densities, about 4-8 m-2 for maize and 10-60 m-2 for sorghum, compared with 250-300 m-2 for wheat and rice crops under favorable conditions. Height, leaf size and the vertical separation of leaves is greater in the C4 cereals, features that affect not only light penetration into the crop, but also its ventilation with C 0 2 . With the more distantly spaced plants of the C4 crops, the advantage in having more horizontally inclined leaves in the early stages of crop growth is greater than in the denser stands of the small grains. At later stages, when the LAI exceeds 4, more vertically inclined leaves-with their reduced canopy extinction coefficient-may be advantageous in improving light penetration into the canopy while at the same time reducing the extent of light saturation of leaf photosynthesis when the sun is at high elevation. This advantage should be of greater importance to the C3 cereals, because of their readier light saturation, than to maize, sorghum, and millet. Also, canopy structure in maize can be controlled by the density of planting far more than in the C3 cereals, in which greater tillering compensates for sparser planting. Consequently, increased fertilizer inputs, especially of N, can result in extremely dense communities of the small grain cereals, in which more vertical leaves are likely to be of greater importance than in the more controllable canopies of maize. Similarly, we should expect cultivars bred for lower input systems and sparser communities, in drier areas for example, to be less subject to selection for vertically inclined leaves. With barley, Angus er al. (1972) found crop photosynthesis to be greater with erect leaves; grain yield was greater with erect leaves at high sowing density, but with lax leaves at low density. Tanner et al. (1966) found a strong association between high yield and erect leaves among barley, wheat, and oat cultivars, as did Tanaka et al. (1966) and Hayashi (1966) among
326
L. T. EVANS AND I. F. WARDLAW
rice cultivars. More vertically inclined leaves were found by Tanaka (1972) to increase canopy photosynthesis of rice by more than 50% under high radiation levels, and to be associated with higher yields (Tanaka et ul., 1969). For the C3 cereals, therefore, more upright leaves may be advantageous to both photosynthesis and yield in high density crops. Although maize is the cereal in which the effects of leaf inclination have been most fully explored, ironically they remain less clear. Simulation models of maize crop photosynthesis by Duncan er ul. (1967) and Duncan (1971) suggest that although more horizontal leaves may have an advantage at LA1 values less than 3, more erect leaves could result in up to a 2-fold increase in photosynthesis at high LAI. However, Sinclair (quoted by Wallace et ul., 1972) did not find greater photosynthesis in more erect leaved stands. With regard to grain yield, Russell (1972) and Ariyanayagam et ul. (1974) found that yield tended to be higher with less erect than with more erect leaves at several plant spacings. Whigham and Woolley (1974) found greater light interception with less erect leaves but little effect on yield, while Winter and Ohlrogge (1973) found a small yield advantage with more upright leaves at high LAI. Overall, these results suggest that leaf inclination has a greater influence on yield in the C3 cereals than it has in maize and other C4 cereals (cf. Trenbath and Angus, 1975). More vertical orientation of leaves among the C3 cereals could account for Yoshida’s (1972) observation that the LAI at which maximum crop growth rate is reached is rather higher in them (6-8.8 in wheat, 4-7 in rice) than in maize (3-5).
B. PHOTOSYNTHESIS BY STEMS AND EARS An important characteristic of some cereals is the substantial photosynthetic contribution made by both stems and inflorescences, particularly in the later stages of grain growth. The overlapping leaf sheaths that wrap and support the lower parts of the stem can be as photosynthetically active as the leaf blades, e.g., in barley (Thorne, 1959 ). Consequently, they can act as photosynthetic traps for C02 respired by stems during the day. In the same way, the bracts and glumes of the inflorescences may serve as efficient photosynthetic traps for the substantial amounts of C02 respired by the grains during their growth (cf. Abdul-Baki and Baker, 1970). Moreover, the stems and ears of the temperate cereals tend to remain green after many of the leaves have dried off. Although most noticeable in water-stressed crops, the phenomenon is not confined to such conditions, as in the wheat crops examined by Spiertz et ul. (1971). Toward the end of grain filling, therefore, stem and ear photosynthesis can become the major source of current assimilate. But even in the earlier stages of grain growth, stem photosynthesis can be a substantial component (Evans and Rawson, 1970). Puckridge (1969) found stems and leaf sheaths to account for 39-44% of
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
327
canopy photosynthesis in wheat crops. Net photosynthesis by the awned ears was rather less, but increased in relative importance as the elevation of the sun declined. In rice and the C4 cereals leaf senescence is less rapid, and stem photosynthesis is relatively less important. For example, Allison (1964) found leaf blades to comprise about 80% of the total green surface area in maize crops at both anthesis and maturity, whereas in wheat they comprised 50% at anthesis and only 10% at maturity. Nevertheless, leaf sheaths may contribute substantially to grain growth in sorghum under some conditions (Stickler and Pauli, 1961). The importance of photosynthesis by the inflorescence varies to a considerable extent among the cereals, depending on its position and structure. In those having a terminal inflorescence above the crop, with awned glumes, such as barley, rye, and many wheat cultivars, photosynthesis by the ears can be a major source of carbohydrates for grain growth, as Archbold (1945) first emphasized for barley. The ears are exposed to full light and high COz concentrations and their assimilates, being so close to the grains, are in the most favorable position to be used in grain growth. In barley, ear photosynthesis may comprise one quarter to one half of total crop photosynthesis, and contributes 24-84% of grain growth (Birecka et al., 1964, 1967; Biscoe et aL, 1973; Thorne, 1965). The large barley awns are responsible for three quarters or more of the photosynthesis and transpiration by ears (Biscoe et al., 1973; Johnson et al., 1975; Kjack and Witters, 1974). The awns of wheat, when present, are not so large (except in T. dumrn), and ear photosynthesis is not quite so important as in barley. Its contribution to grain growth vanes from 10 to 60% (Saghir et al., 1968; Lupton, 1969; Evans and Rawson, 1970). Cereal awns are not only active in photosynthesis, but may also increase the efficiency of water use by the crop (Johnson et al., 1975), modify the energy balance and turbulence above the crop (Benci et al., 1973; Ferguson et al., 1973), and may increase the movement of cytokinins to the grains (Michael and Seiler-Kelbitsch, 1972; Michael et al., 1973). They may increase the danger of lodging under humid conditions (Pinthus, 1973). The panicles of rice are not nearly so active in photosynthesis and do not even refix all the C 0 2 from grain respiration (Yoshida, 1972). For that reason they could, with advantage, be lowered into the canopy. The large, terminal inflorescences of sorghum can intercept 2 5 4 0 % of incoming radiation (Eastin, 1968). Their net photosynthesis is rather low, especially in the compact types, and confined to the early stages of grain development (Eastin and Sullivan, 1969). Thus, they might also be better recessed within the canopy, like those of rice. However, Fischer and Wilson (1971b) found inflorescence photosynthesis in sorghum to account for 6-18% of grain yield. In maize, the male tassels intercept 4-20% of the incoming radiation (Duncan et al., 19671, but the female inflorescence is positioned well down in the canopy. Although its husks are photosynthetic, their contribution to grain growth is probably slight.
328
L. T. EVANS AND I. F. WARDLAW
C . RESPIRATORY LOSSES
We consider here only that respiration, often called dark respiration, in which there is an efflux of COz as the metabolic cost of the cellular work done for growth or maintenance. Differences among the cereals in their “photorespiration,” as a result of the oxygenase activity of carboxydismutase, were discussed earlier in connection with photosynthesis. Respiratory losses represent a substantial fraction of the COz fixed in crop photosynthesis. For example, losses of 14C recorded over a period of weeks after its initial fixation include 20% for maize (Palmer er al., 1973), 25% for wheat (Rawson and Hofstra, 1969; Rawson and Evans, 1971), 33% for barley (Birecka er al., 1967), and 3 4 3 7 % in rice (Lian and Tanaka, 1967). These figures should not be taken as indicating differences among the cereals, because of differences in the stage and conditions in which the plants were exposed to l4COZ. But given the magnitude of the losses, we should ask whether there are differences among crops and cultivars that could be selected for. In early models of crop growth, respiration losses were assumed to be proportional to accumulated dry weight or leaf area of the crop. A consequence of this assumption was the prediction that there should be a clearly defined optimum LAI for crop photosynthesis and growth. Although sometimes recorded (cf. Yoshida, 1972), the usual situation is for net crop photosynthesis to approach an asymptote as LA1 increases. The explanation of this response is that crop respiration also approaches an asymptote, as may be seen from Fig. 2, rather than increasing in proportion to the LAI as earlier assumed. In fact, the respiration rate per unit dry weight of wheat crops falls asymptotically with time (Puckridge and Ratkowsky, 1971) to a minimum value of about 0.016 g g’ d-‘. As is evident from Fig. 2, crop respiration is approximately proportional to photosynthesis. However, it is more satisfactorily estimated on the basis proposed by McCree (1970) as the sum of two terms, one for maintenance and the other for growth. The distinction between these is operational rather than biochemical. Growth respiration represents the metabolic cost of converting the translocated products of photosynthesis to structural, cytoplasmic, or storage compounds. The theoretical efficiencies of the various conversions have been calculated by Penning de Vries (1972, 1975), who has shown them to be unaffected by temperature. Although there are differences between species and between organs, these probably reflect differences in their composition rather than differences in synthetic efficiency. There may, however, be substantial differences among crops and cultivars in the other term, for maintenance respiration. Whereas growth respiration represents the cost of producing new tissues, maintenance respiration is for the renewal or replacement of old structures. McCree (1970) related it to accumulated dry weight, and on that basis it becomes the dominant term in the
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
329
respiration of crops toward the end of their life cycle. The rate of maintenance respiration depends on temperature and on other conditions. There are still too few estimates of it for different crops to be compared. For maize they range from 0.008-0.022 g S' d-' (Penning de Vries, 1972, 1975), lower than those for sunflower. Similarly, McCree (1974) found those for sorghum, 0.0068 g g-' d-' at 20°C and 0.01 1 g g' d-' at 30°C, to be only about half as great as those for white clover, even when expressed on the basis of protein rather than dry weight. A rate of 0.12 g g-' d-' fits the crop respiration measurements of Biscoe et al. (1975a) for barley, similar to those for C4 cereals and to the minimum rate for wheat crops mentioned above. Thus, the cereals all appear to have rather low rates of maintenance respiration compared with other crops like sunflowers, white clover, or cotton (Hesketh et al., 1971). Differences that may exist among the cereals are unlikely to be revealed by such comparisons, however. They require more direct analysis. Among 6 cultivars of wheat of varying height, Rawson and Evans (1971) found a 4-fold variation in the respiration rate of stems, the rate increasing as stem length was reduced. Heichel (1971) found the faster growing of two maize cultivars to have lower respiration rates in its leaves and roots, which suggests there could be very different rates of maintenance respiration in the two cultivars. How could these differences arise? To a large extent maintenance respiration probably derives from the turnover of enzymes and membrane proteins (cf. Penning de Vries, 1975). It is likely to be tightly coupled in the biochemical sense, but not necessarily in the physiological sense in that the rate of turnover may be dependent on substrate levels. Indeed, Penning de Vries (1975) offers evidence of such dependence in maize. This could account for the observations of Birecka and Dakit-Wlodkowska (1963) and Birecka et al. (1969) that the removal of wheat ears led to a 2-fold increase in the rate of stem respiration. On this basis, Heichel's (1971) slower growing maize would have a higher respiration rate because it grew slowly, not the other way round. But we need far more varietal comparisons before any conclusions about the effect of differences in respiration rate on yield can be drawn. VIII.
Translocation
Limitations to yield in cereals have often been considered in terms of the relative importance of the source (photosynthesis) and sink (grain growth), with little or no account being taken of the system transporting assimilates from source to sink. The capacity of this system could limit grain growth however, and changes in the transport and partitioning of assimilates are likely to have been a major component in the evolution of the cereals.
330
L. T. EVANS AND I. F. WARDLAW
A. LOADING AND EXPORT
Differences among the cereals, especially between those with the C3 and C4 photosynthetic pathways, are apparent at the very first step in the translocation process. The distance between the parallel veins in leaves of C4 gramineae is only half or less that in the C3 species (Lush and Evans, 1974) and the distance traversed from the site of fixation to the phloem is correspondingly smaller (Wardlaw, 1976). However, there is an approximately 2-fold range in these distances within the two groups and even between cultivars of a species (Khan and Tsunoda, 1971; Hanson and Rasmussen, 1975). Presumably related to this is the much faster export of assimilates from the C4 cereals. Hofstra and Nelson (1969) first showed this in comparisons between maize, sorghum, and millet with various dicotyledonous crops, and Moss and Rasmussen (1969) in comparisons between maize and sugar beet. Subsequent experiments (Lush and Evans, 1974; Gallaher et al., 1975; Wardlaw, 1976) have extended these comparisons to several C3 and C4 gramineae. Not only is initial export more rapid in the C4 cereals, it is also more complete; what is not exported immediately is stored as starch in the bundle sheath chloroplasts by day, and almost entirely mobilized and exported during the following night (Rhoades and Carvalho, 1944; Lush and Evans, 1974). In the C3 grasses and cereals, on the other hand, a greater proportion of assimilates is accumulated in the leaves, and these are exported more gradually. C3 and C4 plants may also differ in the loading of their assimilates into the phloem. This appears to be an active process in C3 plants, requiring metabolic energy and accompanied by the hydrolysis and resynthesis of sucrose, usually associated with movement against a concentration gradient. Such hydrolysis does not occur in sugar cane (Hatch and Glasziou, 1964) and loading into the phloem of C4 gramineae from the bundle sheath cells may take place down a concentration gradient (Lush and Evans, 1974). Possibly for this reason, initial export of assimilate is more adversely affected by low light intensity in the C4 grasses. The high sugar content of the bundle sheath cells may be associated with their greater resistance to injury by water stress, as in maize (Giles et al., 1974).
B. PHLOEM CAPACITY AND SPEED OF MOVEMENT
A manyfold increase in the need by ears to import assimilates through the peduncle has occurred in the course of evolution in cereals. In wheat, this increased need has been matched by a comparable increase in the area of phloem in the peduncle, with the result that the rate of transfer of assimilates per unit cross-sectional area of phloem has not changed (Evans er d.,1970). In fact, the rate in all genotypes is close to the average rate of 3.6 g cm-' phloem h-' which
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
331
Canny (1960) found in a number of plant systems. This could be taken as indicating a limitation by phloem on the rate of translocation. However, the leaves of C4 gramineae maintain specific mass transfer rates 3 4 times higher than those in C3 grass leaves (Lush and Evans, 1974), and Passioura and Ashford (1974) have found specific mass transfer rates of up to 180 g cm-' sieve tube h-' in the upper root of wheat seedlings. Along with this greater rate of assimilate translocation in C4 gramineae, its speed may also be faster, although there are too few measurements to be sure. Speeds of up to 100 cm h-' have been measured in the peduncle of wheat (Wardlaw, 1965), up to 180 cm h-' in the leaves of rice (Troughton er al., 1974a), and up to 210 cm h-' in the leaves of maize (Troughton et al., 1974b; Moorby er al., 1974). Wardlaw and Marshall (1976) found translocation in Sorghum sudanense to be 2-3 times faster than in a C3 grass, u p to 200 cm h-'. However, the speed of movement appears to be very responsive to increased push from the sources, increased pull from the sinks, or restrictions in the vascular pathway. In both rice and maize the speed increases with increase in light intensity and leaf photosynthesis. In wheat culms it increases as the need for assimilates by the ear increases, whether due to more grains or the inhibition of ear photosynthesis (Wardlaw, 1965; Wardlaw and Moncur, 1976). Similarly, severing half of the vascular tissue in the stem soon leads to a compensating increase in transport through the remainder to the ear, both in sorghum (Fischer and Wilson, 1975a) and in wheat (Wardlaw and Moncur, 1976). These results suggest, therefore, that the capacity of the vascular system is unlikely to be a major limitation to yield in cereals. Diffusion of assimilates to the phloem in the leaves, and from the inflorescence phloem across the grains may, however, be a more significant limitation.
C. UNLOADING OF ASSIMILATES
In the earlier stages of crop growth much assimilate is unloaded from differentiating termini of the vascular system into meristematic tissue. In the case of the intercalary meristems of cereal leaves, they are clearly in a favored position to receive assimilates once their leaf blade begins photosynthesis, but the more peripheral root and shoot meristems are less favorably placed. The rate of diffusion of assimilates to them from the phloem may well limit their development, as Kirby and Rymer (1974) believe happens in the young barley inflorescence. At later stages in its development, the cereal inflorescence must compete with the stem and leaves for assimilates, and its size may be limited by the supply of assimilates it receives. During grain development, quite marked profiles in the supply of assimilates from leaves to the various parts of the inflorescence may be found. In wheat, for example, the more distal spikelets receive progres-
332
L. T. EVANS AND I. F. WARDLAW
sively less photosynthate from the leaves, which may reflect an inadequate supply to the upper parts of the ear. However, more labeled assimilates move there when there are more grains to attract them (Rawson and Evans, 1970). The pattern of demand is therefore a predominant influence on the distribution of assimilates, though modified to some extent by vascular patterns (e.g., Inosaka, 1957). Within the spikelet of wheat the continuity of the tracheary elements from the rachilla to the pericarp of the grain is interrupted by xylem discontinuities (Zee and O'Brien, 1970), and xylem and phloem transfer cells occur in the nodal plexus where the glumes and caryopsis are attached (Zee and O'Brien, 1971a). There is then no anatomical discontinuity in phloem cells from the rachilla to the vascular bundle up the crease of the grain and labeled assimilates from the leaves rapidly accumulate in the crease, according to Sakri and Shannon (1975). These workers believe sucrose hydrolysis and resynthesis is a prerequisite to the transfer of assimilates to the endosperm, which could imply loading against a concentration gradient, but Jenner (1973, 1974b) found no evidence of inversion. In fact, Jenner (1974a) has measured sucrose concentrations at various steps from the vascular bundle to the endosperm and shown them to fall progressively, so that movement into and across the grain could be by diffusion. The situation in the grains of the C4 cereals appears to be rather different. Phloem transfer cells are absent from the vascular traces leading to the caryopsis, at least in Japanese millet (Zee and O'Brien, 1971b), and the phloem terminates before the caryopsis. Assimilates presumably diffuse through the pigment strand cells in the gap of the seed coat to the nucellar projection, and from there to the nearby aleurone layer cells which are modified as transfer cells in the Seturia millets and also in maize (Kiesselbach and Walker, 1952). Inside these aleurone transfer cells in maize is a region of elongated, highly protoplasmic conducting cells with spirally thickened walls. According to Shannon (1968,1972), sucrose is hydrolyzed on leaving the conducting tissue, and diffuses through the endosperm as glucose and fructose at about 2.7 mm h-' . Since the rate of growth per kernel can be up to 5 times higher in maize than in the temperate cereals, the arrangements for unloading and transfer of assimilates within the caryopses of C4 cereals may be much more efficient than those in the C3 cereals.
D. DISTRIBUTION OF ASSIMILATES
The principles that govern the partitioning of assimilates among the organs of a plant are not fully understood. The pattern of distribution depends to a considerable extent on environmental conditions. Drought stress and low temperatures, for example, tend to favor root growth relative to shoot growth. Root growth is also relatively more important in the early stages of crop development.
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
333
Rawson and Hofstra (1969) have described the changes in the pattern of 14C distribution from the various leaves of wheat as plants age. The second youngest fully emerged leaf is the major exporter of assimilates prior to anthesis, as also in rice (Shen, 1960), and the major supplier to the young leaves and shoot apex. Older leaves export predominantly downward, to roots and tillers. Even after anthesis, the lower leaves of the main stem may continue to export much of their assimilate to the tillers. When the grains begin to grow, however, they rapidly become the dominant sink for assimilates from the upper leaves. The awned ears of the primitive wheat progenitors are largely self-sufficient for assimilates, but selection for greater yield has resulted in the flag leaf and the leaves below it playing a greater role in the supply of assimilates for grain growth (Evans and Dunstone, 1970). Distribution patterns for 14C in wheat and barley, combined with the results of many defoliation and shading experiments, indicate that the flag leaf, stem and ear-the organs closest to the grains-are the main source of carbohydrates for grain growth. Perhaps because ear photosynthesis is less important in rice, leaves below the top two also supply the developing grain (e.g., Tanaka, 1958; Lizandr and Brovtsyna, 1964). The role of the lower leaves is even more important in sorghum. Of the assimilate exported by the five uppermost leaves, approximately the same proportion went to the panicle (Eastin, 1972a). With spaced plants, Fischer and Wilson (197 1 b) estimated that 93% of grain yield was due to photosynthesis by the head and upper 4 leaves, each contributing 17-21%, although the flag and penultimate leaves were much smaller than the leaves below them. Even more leaves are involved in the supply of assimilates to the centrally placed maize cob. A substantial fraction of the ''C-labeled assimilates from the leaves above the cob and down to 5 leaves below it ends up in the grains (Palmer, 1969; Eastin, 1969; Tripathy et al., 1972; Palmer et aL, 1973). Movement from the upper leaves is more rapid, and because of the higher light intensity at the top of the canopy, they are undoubtedly the major suppliers of the substrates for grain growth. However, defoliation experiments clearly indicate that leaves in the middle of the canopy, and even the lower leaves, are also important in supplying the grain and maintaining high yields (Allison and Watson, 1966; Pendleton and Hammond, 1969). In this respect, maize differs considerably from the temperate cereals. However, there is great flexibility in the pattern of assimilate distribution. In wheat, for example, most flag leaf assimilates go to the grain after anthesis, while lower leaves support the roots and tillers. If the upper leaves are removed or shaded, the lower leaves assume the role of supporting grain growth. If the ear and lower leaves are removed, on the other hand, the flag leaf then supports the roots and tillers (King et al., 1967; Marshall and Wardlaw, 1973). Thus, a large potential sink for assimilates, whatever its position, dominates the pattern of their distribution. In this respect, neither the central position of the maize cob
334
L. T. EVANS AND I. F. WARDLAW
nor the terminal position of other cereal inflorescences offers an advantage in terms of assimilate distribution.
E. THE ROLE OF RESERVES
Although carbohydrate reserves accumulated by cereals before anthesis were previously assumed to play a major role in grain growth, Archbold’s (1945) experiments with barley led to the realization that most grain growth is based on concurrent photosynthesis. Nevertheless, reserves do contribute to grain yield, to an extent that varies greatly both between the cereals and depending on environmental conditions. In the C4 cereals there is a rapid initial export from leaves of more than half the assimilated I4C, and nearly all the remainder is exported during the following night. In the C3 cereals, assimilate export from the leaf blades is more gradual. Starch does not usually accumulate in the leaves of wheat, oats, barley, and rye (Gates and Simpson, 1968), but other polysaccharides are present (Wardlaw and Porter, 1967). These can accumulate to quite high levels in the leaf blades-up to 45% of dry weight in wheat (King et aL, 1967)-and also in leaf sheaths and stems. Assimilate from the lower leaves and earlier photosynthesis tends to be stored in the lower internodes of wheat, whereas assimilate from the flag leaves after anthesis tends to be stored in the upper internodes and peduncle (Rawson and Hofstra, 1969; Wardlaw and Porter, 1967). As a result of this storage, stem weight may continue to increase after its elongation has ceased, and then fall during grain growth. The extent of the fall varies a great deal between crops, cultivars, and conditions. Not ail the mobilized material is stored in the grains. Some is lost by respiration, the proportion varying from 14 to 49% among 5 wheat cultivars (Rawson and Evans, 1971). In rice, respiratory loss accounted for 24% of the reserves (Cock and Yoshida, 1972). Thus, half to three-quarters of the mobilized reserves may end up in the grain. Direct evidence of such movement has been provided in rice (Murayama et al., 1961; Oshima, 1966; Cock and Yoshida, 1972) and wheat (Stoy, 1963; Wardlaw and Porter, 1967; Rawson and Hofstra, 1969). The contribution of these pre-anthesis reserves to grain yield has been estimated by a variety of techniques, and reviewed by Yoshida (1972). They appear to be most important in rice, contributing up to 40% of final grain weight (cf. Cock and Yoshida, 1972). In barley they contribute 20% (Archbold and Mukherjee, 1942) to 30% (Biscoe el aZ., 1975b), but only 5 1 0 % in wheat plants not under stress (Asana and Saini, 1958; Wardlaw and Porter, 1967; Rawson and Evans, 1971). With sorghum Goldsworthy (1970) estimated that most of the grain assimilates were produced after anthesis. According to Fischer and Wilson
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
335
(1971a), pre-anthesis assimilates accounted for 12% of final grain weight in sorghum. However, the stem and leaf sheath reserves of carbohydrate in cereal crops are often not fully mobilized. Maize inbreds maintain a higher proportion of soluble carbohydrates in their stems than do more productive hybrids (Johnson and Tanner, 1972a), and up to 40% of stem weight may be sucrose not called on for storage in the grain (Campbell and Hume, 1970). Moreover, high stem sugar contents may be important in conferring greater resistance to stalk rot in maize and to frost injury in sorghum. In plants stressed by shading or defoliation, the reserves are mobilized to a greater extent in order to maintain grain growth, as in rice (Murayama et al., 1955), maize (Duncan et al., 1965) and wheat (Rawson and Evans, 1971). Under these circumstances, the reserves may contribute a much greater proportion of grain weight. On the other hand, greater applications of N fertilizer tend to increase growth at the expense of reserves, and may reduce the need to call on reserves (Yoshida, 1972). Thus, carbohydrate reserves in cereals can make a major contribution to grain yield in most crops under stress conditions, but where nutrition and water supply are favorable the reserves are often drawn on to only a limited extent, except in rice and perhaps hybrid maize. Although of value in adaptation and as an insurance against late stresses, they nevertheless represent unused yield potential. I X . Grain Growth
Under reasonably constant environmental conditions, there is a linear increase in the dry weight of kernels throughout most of the period of their growth, as may be seen from Fig. 3 for wheat. The period of linear increase is preceded by an initial lag after anthesis and may be terminated quite suddenly. Cessation of grain growth in maize is indicated by the formation of a black closing layer in the placental region (Kiesselbach and Walker, 1952), and Daynard and Duncan (1969) proposed the use of this criterion to determine physiological maturity and the duration of grain growth. In sorghum, also, the formation of the black layer coincides with cessation of the import of I4C-labeled assimilates by the kernels (Eastin, 1972a). Because of the linearity of increase throughout most of the period of kernel growth, not only the rate but also the effective duration of grain filling can be determined, in the latter case by extrapolation to the initial (at anthesis) and final kernel weights. We can therefore compare crops and cultivars in terms of the growth rate per kernel, the effective duration of growth, and the length of the lag period before linear growth begins.
336
L. T. EVANS AND I. F. WARDLAW TRIPLE DIRK
60 r-
18,
DAYS FROM ANTHESIS
FIG. 3. Increase in dry weight (left) and N content (right) in grains of Triple Dirk wheat with time from anthesis in three temperature regimes (Sofield et al., 1974).
A. THE LAG PERIOD
The initial lag period has been little studied, but may be important in relation to the number of kernels set in each inflorescence. Among wheat cultivars, the longer the initial lag period the more grains there were in each ear (Rawson and Evans, 1971; Sofield er al., 1974), especially at low temperatures. The length of the lag period in wheat is only a few days however, whereas in sorghum it appears to be about 14 days (Eastin, 1972b), and in maize 15-18 days, 47-85% as long as the effective duration of grain filing (Johnson and Tanner, 1972a,b). These much longer lag periods are associated with the presence of 10-100 times as many grains per inflorescence in maize and sorghum compared with wheat, and may be necessary to allow the later florets in the inflorescence to set grains. Assimilate requirements by the inflorescences during the lag period are relatively modest, and stem and leaf reserves tend to accumulate at that time. Because of the long lag period in maize and sorghum, these “post-anthesis reserves” are likely to represent a major component of grain yield, and deserve further study.
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
337
B. DURATION OF GRAIN GROWTH
Because of the relative ease with which cessation of grain growth can be determined in maize and sorghum by observation of the black layer, more is known of the relations between yield and duration of grain growth in those crops than in the temperate cereals. Daynard et al. (1971) and Funnah (1971) found yield among maize hybrids to be highly correlated with the effective duration of filling, more so than with its rate. In a comparison of inbreds and hybrids Johnson and Tanner (1972a,b) found the effective duration of grain filling to be much longer in the hybrids, but their rates of grain growth per unit ground area were also higher. Data presented by Eastin (1972a) shows that the period from anthesis to maturity, which varies among lines from 31 to 56 days, is consistently longer in the higher yielding hybrids, and that there is a positive correlation between the length of this period and grain yield. However, no data are presented on how this period was partitioned between the initial lag and the linear grain growth phases. Pronounced differences among wheat cultivars have not yet been found in the period from anthesis to maturity (Marcellos and Single, 1972) or in the effective grain-filling period (Sofield e t al., 1974), but the experiments of Asana and Joseph (1964) suggest that they may occur, as they do in rice ( S . Yoshida, unpublished data). Comparisons are made difficult by the very great effect of temperature on the duration of grain filling in wheat, evident in Fig. 3. Field and phytotron experiments agree in showing a major increase in the duration of grain filling as temperature falls, whereas light intensity during grain filling has no effect on its duration. High temperatures, by day or night, also shorten the duration of grain filling in sorghum (J. D. Eastin, unpublished data) and in japonica rice to such a degree that grain yields are reduced and more carbohydrate reserves are left in the plants (Nagato and Ebata, 1959;Nagato et al., 1961; Sato and Takahashi, 1971). However, in tropical rice cultivars such as IR 20 the increased rate of grain growth may compensate for the reduced duration at high temperatures ( S . Yoshida, unpublished data). Tsunoda (1965) argues that more prolonged grain development is needed in rice cultivars grown under heavy N fertilization. The data presented by Shaw and Thom (1951) and Peaslee e l QZ. (1971) suggest that temperature has little effect on grain-filling duration in maize. An important question that cannot be answered with confidence for any cereal is “What causes the grains suddenly to stop growing?” Many lines of evidence from wheat experiments suggest that it is not from lack of assimilates. The leaves and stems may still be green and there may be ample reserves in the stems, indeed stem weights may be increasing. Jenner and Rathjen (1975) have shown that the sucrose concentration in both the free space and endosperm cells is
338
L. T. EVANS AND 1. F. WARDLAW
higher in grains approaching full weight than at earlier stages, so the decline in starch accumulation cannot be due to lack of assimilate.
C. THE RATE OF GRAIN GROWTH
The rate of growth per kernel increases with temperature in rice (e.g., Nagato et al,, 1961; Sat0 and Takahashi, 1971), wheat (Thorne, 1970; Sofield ef al., 1974; Spiertz, 1974), and maize (Peaslee er al., 1971). With wheat andjaponica rice the increase in rate is not sufficient to compensate for the reduction in duration of filing, with the result that grain yield tends to be reduced at higher temperatures. The effect of light intensity on the rate of grain growth is more complex. Only at very low light intensities, combined with high temperatures, did an increase in intensity result in faster grain growth in some wheat cultivars (Sofield e t al., 1974). C02 enrichment had little effect on the rate of growth per kernel in 4 barley cultivars (Natr and Apel, 1974). It seems, therefore, that the rate of kernel growth in the temperate cereals is not markedly limited by the supply of current assimilates except at very low light intensities. In the experiments of Natr and Apel, the barley grains continued to grow rapidly even in the absence of C02 for 5 days. Similarly in maize, Duncan et al. (1965) found the rate of growth per kernel to be reduced only from 9.8 to 7.4 mg kernel-' d-' by complete defoliation of the plants and wrapping of the husks. Stem reserves were mobilized and used with high efficiency to sustain kernel growth in the defoliated plants. Growth rates per kernel in maize range from 6-10 mg d-' among cultivars, and are correlated to some degree with final kernel weight (Carter and Poneleit, 1973). Such rates are far higher than those of the temperate cereals, commonly less than 2 mg kernel-' d-'. Growth rates per kernel in sorghum are comparable to those of the temperate cereals, but there are of course far more kernels per inflorescence. Data to compare rates of grain growth on a per unit ground area basis among the various cereals are insufficient. Goldsworthy et al. (1974) record grain growth rates of up to 35 g m-* d-' in lowland tropical maize, but only up to 21 g m-2 d-' in highland crops (Goldsworthy and Colegrove, 1974), although the maximum crop growth rates were 35 g m-2 d-' at both stations. The data of Benoit ef al. (1965) indicate a grain growth rate of 30 g m-2 d-', but most of those recorded for maize crops are nearer 20 g m-2 d-' (e.g., Daynard et al., 1971; Johnson and Tanner, 197213). Eastin's (1972a) data for sorghum also yield rates of 15-19.1 g m-2 d-', which are probably comparable to those for many temperate cereal crops.
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
339
D. PROTEIN STORAGE Only a few aspects of this large and important subject are considered here. A negative relation between % protein in cereal grains and their yield has often been found, but is by no means universal even with a single cultivar (e.g., Fernandez and Laird, 1959). Major yield advance in some cereals has been associated with a progressive fall in % protein as in Triticale (Zillinsky, 1974) and sorghum (Collins and Pickett, 1972). Selection for high % protein may reduce yield, as in maize (Woodworth et aL, 1952; Dudley, 1973). Low light intensity 'during grain development may affect storage of starch more than that of protein, leading to reduced grain yield but a higher percentage protein, as in rice (Cruz et al., 1970; Sato, 1971). Similarly, protein content is enhanced relative to starch content at high temperatures in rice (Sato and Takahashi, 1971) and wheat (Campbell and Read, 1968). Drought stress also enhances the relative protein content (e.g., Petinov and Pavlov, 1955). Under such conditions, therefore, there can be a negative relation between yield and percentage protein. This is not an obligatory relation, however, and it is much influenced by the adequacy of nitrogen nutrition during grain development. Under conditions where N uptake continues throughout grain filling, both protein and starch content of the grains may increase linearly until near maturity, as in wheat (Jennings and Morton, 1963; Bremner, 1972; and Fig. 3) and rice (Paul er al., 1971). Under these conditions the % protein in the grain need not fall, and may even rise, as storage proceeds (Johnson et al., 1967). Moreover, fertilizer N may be used with very high efficiency, as in the experiments of Hucklesby et al. (197 1) with wheat, and much of the grain protein may derive from N taken up during grain fdling (Pavlov, 1969). Where soil N is severely depleted by heading time, on the other hand, concurrent uptake may be insufficient to sustain the rate of protein storage in the grain, and remobilization of proteins in the plants may occur, often to a greater extent than that of carbohydrate reserves (e.g., Peterson er al., 1975). Cultivars appear to differ considerably in this respect (e.g., Cataldo er al., 1975). In those where remobilization is extensive, there may be rapid leaf senescence and starch storage may be more adversely affected than protein storage, resulting in low grain yields of high protein content (Termanetal., 1969; McNeal et al., 1972). There may be little relation between the protein percentage in the vegetative parts and that in the grain during the later stages of development in such plants, as in wheat (Pavlov and Kolesnik, 1974) and oats (Peterson er al., 1975). These major differences between cultivars in the extent of protein remobilization to the grain make comparisons between the cereal crops difficult. The considerable range in their protein content may, to a large degree, reflect the
340
L. T. EVANS AND I. F. WARDLAW
traditional uses and selection pressures under which the various cereals have developed. Cereals like maize and sorghum which were commonly supplemented by legumes may have been under less pressure to maintain a high protein content than those like wheat. X. Limiting Stages in the Life Cycle
An important characteristic of the cereals is the relatively clear separation between the phases of their life cycle in the organs undergoing growth. Root and leaf growth dominate the vegetative stage together with tillering, the extent of which depends on assimilate supply within the plant. In the second (reproductive) stage, ushered in by inflorescence initiation, the rapidly elongating stems and the differentiating inflorescence are the main competitors for assimilates, although root growth continues and expansion of the upper leaves is completed. Stem growth ceases soon after anthesis, and the developing grains then become the dominant sink for assimilates in the final stage of the life cycle, although root growth and tillering may be renewed if conditions are favourable. Discussions of whether source or sink limits yield refer mostly to this final stage of the life cycle, since most grain growth is supported by concurrent photosynthesis rather than by stored reserves of carbohydrate. However, the sink or storage capacity for assimilates at this late stage is to a large degree determined by the extent of photosynthesis earlier in the life cycle, especially during the reproductive stage of the crop. Thus, whether source or sink limits grain development depends on the balance between the various stages of the life cycle, as determined by cultivar and the sequence of growing conditions. In productive cultivars growing in the conditions and seasonal sequence to which they are adapted, source and sink should be in balance. In other conditions one or other of them may be the more limiting. Quite often both may seem to be limiting (Gifford et al., 1973; Gifford, 1974b). Surplus storage capacity is suggested by the formation of far more tillers than eventually bear ears, e.g., in wheat (Bingham, 1969), more spikelets than bear grains in maize and barley, and more florets than filled grains in wheat and rice. Moreover, grain size in most cereals other than rice can usually be increased when grain number is reduced (e.g., Bingham, 1967; Rawson and Evans, 1970), which suggests the existence of spare storage capacity even at a late stage in most cereal crops. At the same time, however, many cereal crops also have substantial unmobilized reserves of carbohydrate even at the end of grain filling (e.g., Johnson and Tanner, 1972a; Rawson and Evans, 1971), such as the high stem sugar contents of maize and sorghum noted earlier. Similarly, leaf area reduction or shading treatments late in the life cycle may reduce crop gruwth without reducing grain yield. The existence of spare photosynthetic capacity is also
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
34 1
suggested by the finding that the low leaf rates late in grain filling may be raised when demand is raised (e.g., King et ul., 1967). How can both source and sink appear to have spare capacity? Gifford (1974b) presents a model that suggests that it is impossible to operate at both full source and full sink capacity simultaneously because of feedback inhibition of photosynthesis at sucrose levels required to maximize the rate of storage. Compartmentation may overcome this, however. Another possibility is that other processes besides the supply and storage of assimilates may be limiting, such as the capacity of the translocation processes, although evidence reviewed above indicates that this is unlikely. Rawson and Bremner (1976) have suggested that sink limitation may occur early in grain filling and source limitation toward the end, or sink limitation by day and source limitation by night, but this too is unlikely in many cases. For example, the presence soon after anthesis of additional grains, which can develop in wheat if growth correlations permit (Rawson and Evans, 1970; Evans et ul., 1972), suggests that storage capacity per se is not limiting at that time, just as the continuing presence of stem reserves late in grain filling suggests that source is not limiting at that stage. An evolutionary approach may be more helpful. Survival of plants in the wild, or in primitive agriculture, is likely to be aided by the accumulation of substantial reserves of both assimilates and potential seed sites. Recovery from adverse conditions, whether loss of photosynthetic capacity by drought or defoliation, or loss of seed sites by frost or insect attack, is then likely. The essence of cereal domestication and improvement has been to provide the conditions-through cultivation, fertilizer application, irrigation, herbicides, pesticides, etc.-which permit a reduction in the accumulation of reserves and a progressive increase in the proportion of assimilates invested in the grains (Evans, 1976a). Nevertheless, variation from year to year and from season to season in most agricultural areas is still sufficient for some selective advantage to accrue from the maintenance of surplus source and sink capacity. As control of the agricultural environment improves, so can the mobilization of these reserve capacities into additional grain production be increased, but the extent of seasonal variation still precludes their full exploitation, as may be seen in numerous analyses of crop yield in successive years (e.g., Campbell et ul., 1969; Murata and Matsushima, 1975; Fischer and Aguilar, 1976). A. DURATION OF LIFE CYCLE STAGES
The relative duration of the various stages, and their optimum balance, obviously depend to a very great degree on the seasonal conditions under which the crop is grown. Rice in the tropics, under intensive multiple cropping, may require only 25 days in the field for each of its three stages, although the first stage may be preceded by 20 days in the seed bed (Yoshida and Parao, 1976).
342
L. T. EVANS AND I. F. WARDLAW
Crops in cooler conditions may require longer at each stage, particularly for the vegetative stage in autumn-sown temperate cereals. Among 12 cultivars of sorghum in which days from planting to maturity varied only between 105 and 114 days, the percentage of the life cycle for each stage varied from 26-36%, 30-34%, and 34-41% for the vegetative, reproductive, and grain-filling stages, respectively (Eastin, 1972a). Within this group of cultivars, grown in Nebraska, higher yields were associated with shorter times in the vegetative stage and longer times in the grain-filling stage. Eastin emphasizes how important a few extra days of grain filling are to yield of sorghum in Nebraska, where grain growth is cut short by frost. In Texas, on the other hand, sorghum yields tend to increase as time to anthesis increases (Dalton, 1967). Clearly, the environment determines which stage of the life cycle it is more important to extend. In spring-sown wheat in Minnesota, with a total life cycle of 93 days, the grain-filling stage was the shortest, possibly reflecting the progressive rise in temperature (Krenzer and Moss, 1975). When the cultivars were grown under controlled temperature, the length of the vegetative stage was greatly reduced relative to the later stages, while the length of the grain-filling stage increased. Substantial year-to-year variation occurs in the relative length of the stages for crops even at one site (cf. Fischer and Aguilar, 1976). An adequately long vegetative stage is required to establish the root system and leaf canopy as a basis for later crop growth. It is also needed to establish the tillers and potential spikelet sites which contribute to eventual grain storage capacity, particularly in the temperate cereals in which tillering is more important and primordia accumulate at the shoot apex before inflorescence initiation occurs. Greater fertilizer use, early irrigation and denser planting all permit a shortening of the vegetative phase. Too little is known of the relations between yield and the duration of the reproductive stage in cereal crops for useful comparisons to be made, most attention having been given to the grain-filling stage. The integral of LAI with respect to time from anthesis to maturity, the leaf area duration (D), has been extensively analyzed in relation to yield in the temperate cereals. Earlier experiments indicated a reasonably close relation between D and grain yield in some temperate cereals (e.g., Watson et a l , 1963; Welbank et al., 1968). The aggregated data for wheat (Evans er al., 1975) indicate that the relationship tends to be closer at relatively low values of D and yield. At higher values, increases in D from additional N fertilizer are not associated with proportional increases in yield (Thorne and Blacklock, 1971). Indeed, Thorne (1973) has shown that there is little advantage to yield in raising the LAI at anthesis in wheat and barley crops above 6. The concept of leaf area duration involves several problems when comparisons are made between cereal crops. Because the LA1 falls, often sharply, with time from anthesis, D values tend to be dominated by the high LA1 values early in the
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
343
grain-filling stage. However, LA1 values above 4 have little effect on crop photosynthesis (cf. Fig. 2) in spite of their large effect on estimates of D. On the other hand, small differences in LA1 late in the grain-filling stage may have a substantial effect on yield (e.g., Watson et al., 1963). Yield in cereals often bears a close relation to LA1 at anthesis, as in wheat (Simpson, 1968), maize (Eik and Hanway, 1966), and rice (Yoshida et aL, 1972). In these conditions, yield is largely determined by the yield capacity which is in turn determined by the extent of canopy growth or nitrogen uptake, often related to the maximum LAI attained. In these cases, the element of duration is of less importance. Under other conditions, the time element of D may be predominant in yield determination. In a comparison of wheat and maize, for example, Allison (1964) found that although D for both crops was similar (-260 days), the much greater yield of maize was associated with more prolonged grain growth and a correspondingly longer retention of green leaves (Fig. 4). It is an open question, of course, whether the maize leaves stayed green longer because of the demands from more prolonged grain growth, or vice versa. Prolongation of grain growth, at lower temperatures for example, can increase yields substantially in spite of reduced growth rates, as shown in Fig. 3. The increased duration at low temperatures reduces the photosynthetic limitation on
1
0
4
8
12
WEEKS FROM EAR EMERGENCE OR HALF SILK
FIG. 4. Relation between LA1 and grain growth with time from ear emergence in wheat or half-silk in maize (0).Solid lines represent LAI, broken lines grain growth. (Adapted from Allison, 1964.)
(A)
344
L. T. EVANS AND I. F. WARDLAW
yield development, as evident in the results of Krenzer and Moss (1975) in which enrichment with C02 during either the reproductive or the grain-filling stage had no effect on yield for wheat crops grown at 13"/17"C although it enhanced yield substantially at 25"/20"C or in the field. Since both the duration of the grain-filling stage and D vary over a wide range for each crop, it is difficult to make useful comparisons between the major cereals. Very low D values have been reported for sorghum crops in the tropics (Krishnamurthy et al., 1973), but values for wheat and barley crops can be as low under hot dry conditions. In view of the comments made above, comparisons of D values for the various cereals and further analysis of the concept of D is unlikely to be rewarding.
B. PHOTOSYNTHETIC LIMITATIONS TO YIELD AT THE VARIOUS STAGES
The relative limitation to yield by photosynthesis at various stages of the life cycle, even in locally adapted cultivars, depends a great deal on the sequence of conditions for the particular crop examined. It varies not only from site to site, but also from year to year. With Yecora wheat grown under irrigation at Obregon, for example, shading reduced yield by up to 60% in 1973 when applied during the middle of the reproductive stage, whereas in 1974 the greatest yield reduction by the same degree of shading was only 16% and this occurred during the grain-filling stage (Fischer, 1975). Comparable results for the two preceding years are presented in Fig. 5 along with those of Yoshida and Parao (1976) for rice in the Philippines. These agree in showing that shading and reduced photosynthesis has least effect on yield when given during the vegetative stage, and most effect during the reproductive stage, with the grain-filling stage also being sensitive. For both crops early shading greatly reduced growth, and especially tillering, but not the number of tillers that produced ears. In Fischer's (1975) experiments, shading always reduced crop growth, in direct proportion to the reduction in radiation, even when it had little effect on grain yield. With spring wheat in a very different environment, Willey and Holliday (1971b) found results similar to those of Fischer (1975), yield being most affected by shading during the reproductive stage and least during the vegetative stage. With winter wheat, on the other hand, yield was equally sensitive to shading before and after heading (Pendleton and Weibel, 1965). Maize seems to be most sensitive to shading at the silking and grain set stages (Moss and Stinson, 1961). Shading treatments on crops of two barley cultivars had a pronounced effect on yield when applied before anthesis, but no effect after anthesis (Willey and Holliday, 1971a), which suggests that storage capacity was more limiting to yield
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
345
9 w
> 4
RICE
3
n
w>
za
a
u
u1
E
4 w n
I
I
100
200
’I
I 300
I
I
I
400
500
600
cal cm-*d-’
70
60
/,
0
20
WHEAT I
I
I
I
40
60
80
100
% SUNLIGHT
FIG. 5. Effect on grain yield of shading at various stages in the life cycle of crops of rice in the Philippines (Yoshida and Parao, 1976) and of wheat in Mexico (Fischer, 1975).
than was the supply of assimilates during grain filling. With a barley crop grown late in the season, however, Gifford et al. (1973) found the adverse effect of shading on grain yield to be about equal when given before or after anthesis. However, C02 enrichment increased grain yield much more when given before rather than after anthesis, because the effects of post-anthesis enrichment on grain number per ear and kernel weight were much smaller than those of pre-anthesis enrichment on the number of ears. Fischer and Aguilar (1976) enriched Yecora wheat crops, grown under irrigation and high fertilizer levels, with COz applied during each of four 1-month periods, over 3 years. Although COz enrichment increased crop growth, it increased grain yield consistently only when applied during the reproductive stage. Applied during the grain-filling stage, it increased grain yield significantly only in one year. Krenzer and Moss (1975) applied COzenrichment to spring wheat crops in Minnesota, and found it to increase grain yield about equally in the pre- and post-anthesis periods, through increase in the number of ears and grains per ear in the earlier treatments, and in kernel weight from enrichment after anthesis.
346
L. T. EVANS AND I. F. WARDLAW
Kernel weight in rice is not so readily increased by favorable conditions after anthesis as it is in wheat. Consequently, the main effect of C02 enrichment after anthesis for rice crops is an increase in the percentage of filled grains, with consequent increase in yield. C02 enrichment before anthesis causes a greater increase in yield of rice under tropical conditions, through effects on both grain number and kernel weight (Yoshida et al., 1972; Cock and Yoshida, 1973; Yoshida, 197313). Shading and COz-enrichment treatments can be applied for defined periods, and for that reason are more satisfactory than crop thinning treatments, which resemble C 0 2 enrichment in enhancing the photosynthetic rate of the plants, but right through to maturity rather than for a limited period. Early thinning of wheat crops led to an increase in ear number per plant; later pre-anthesis thinning increased the number of grains per spikelet, while post-anthesis thinning increased kernel weight (Fischer and Laing, 1976). The increase in kernel weight from crop thinning at anthesis ranged from 6% to 41%, being greater at higher temperatures. In sorghum, likewise, early thinning increased seed number while late thinning increased kernel size (Fischer and Wilson, 1975b). The shading, C02-enrichment and crop thinning treatments have indicated the very considerable variability between crops, cultivars, environments, and years in the relative limitation to yield by the pre- and post-anthesis stages. This is one reason why attempts to correlate yield and yield components with natural radiation levels at various stages of crop life cycles are unlikely to result in clear relations. There are, of course, many other problems in such work, one being correlations between pre- and post-anthesis radiation levels, and between radiation and temperature. The higher temperatures often associated with higher radiation levels may so shorten the duration of the reproductive or grain-filing stages as to mask any yield advantage from higher radiation and additional photosynthesis. Consequently, although grain yield may increase with increase in receipt of radiation during grain filling up to a certain level for temperate cereals, still higher radiation levels may not be associated with greater yields (e.g., Welbank et ul., 1968; Evans, 1973). When temperature is controlled, however, grain yield is higher the greater is the radiation receipt during grain filling (Evans, 1976b). In rice crops at higher latitudes, grain yield may be positively related to both temperature and radiation levels during grain filling, but negatively related to them at lower latitudes (e.g., Lee, 1972; Murata, 1975). Yoshida and Parao (1976) found spikelet number in rice crops to be related positively to solar radiation but negatively to temperature during the reproductive stage. Grain yield of maize in the Japanese IBP experiments was positively related to both radiation and temperature (Kudo, 1975). For each of the cereals, grain yield may be positively associated with both temperature and incident radiation in the lower part of their range, but nega-
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
347
tively associated with temperature when it is high. In the higher part of the temperature range, therefore, the effects of increasing radiation and temperature may cancel out, or the adverse effects of high temperature may become predominant. In this respect, the C4 cereals are likely to be more tolerant of high temperatures, and therefore more able to take advantage of very high radiation levels during grain filling.
C. YIELD COMPONENTS
One reason for the success of cereals as crops is their capacity for yield component compensation, i.e., for the later-determined components of grain yield to compensate for earlier losses or restriction of development or to take advantage of favorable conditions late in the crop life cycle. The major cereals differ, of course, in the extent to which such yield component compensation can occur in the later stages of the life cycle. For example, kernel size is more restricted by glume size in rice than in other cereals, with the result that kernel weight in rice is far less variable and unable to accommodate additional carbohydrate when conditions during grain filling favor more rapid or prolonged grain growth (Matsushima, 1970). In wheat and barley, on the other hand, kernel weight displays a substantial range. If grain number per ear is reduced, the remaining grains may grow to a greater size in wheat (e.g., Bingham, 1967; Rawson and Evans, 1970). This did not occur in barley (Buttrose and May, 1959) or maize (Duncan and Hatfield, 1964), which suggests that assimilate supply was not limiting grain growth in intact ears. Although variation in kernel weight makes a degree of yield compensation possible late in the life cycle, except in rice, the scope for compensation is much greater earlier in the life cycle. Grain number per unit ground area, the major determinant of yield, can be influenced by the number of inflorescences, the number of spikelets per inflorescnece, the number of florets per spikelet, and the proportion of florets actually setting grains. These yield components are determined in succession, as Matsushima (1970) has shown for rice. Limitations by adverse conditions on the earlier-determined yield components can be compensated for in the later ones. Poor or variable establishment, or a low density of sowing, can be compensated for in many cereals by abundant tillering and the development of more ears per plant. For example, whereas tillers may account for only 30%of the total grain yield in a dense stand of wheat (300 plants m-’), about half of the plants producing only the main stem ear, at half that plant density 5040% of the yield may come from tillers and a third of the plants may have as many as four ears (Bremner, 1969). Compensation for the halved plant density was such that grain yield was reduced by only 9%.The temperate cereals produce tillers to
348
L. T. EVANS AND I. F. WARDLAW
an extent determined by incident radiation prior to inflorescence initiation (e.g., Evans et al., 1975), often to an extent far in excess of the number eventually bearing ears. Consequently, up to two thirds of the tillers may be “wasted,” but this wastage ensures considerable scope for compensation early in the life cycle. There is less scope in sorghum, in this respect, and still less in modem sparsely tillering forms of maize. In the latter, however, the main stem may bear additional ears, particularly in cultivars that tend to have fewer kernels per ear (Duncan, 1975). Increases in grain number per ear and in kernel weight may also help compensate for low stand densities. A striking example is provided by the work of Kirby (1969) with barley crops in which final grain yield was virtually unaffected by density over the range from 50 to 800 plants m-’. The 16-fold increase in planting density resulted in only a 90% increase in ear number m-’ , combined with a 40% reduction in grain number per ear and an 18% reduction in kernel weight. As a result of this remarkable compensatory power, grain yield in cereals is relatively insensitive to planting density, and displays a wide variety of negative correlations among the yield components (e.g., Leng, 1963). For the same reason there is considerable variation from site to site, cultivar to cultivar, and year to year in the component most closely related to grain yield. In rice, for example, grain yield may be closely related to the number of spikelets m-’ and bear little relation to the percentage of filled grains, or vice versa, depending on conditions (Murata and Matsushima, 1975). Soil fertility and water supply, and the usual seasonal sequence of conditions, may favor a particular balance among the yield components, as Grafius and Okoli (1974) argue for barley. But there are considerable differences among plant breeders in the emphasis they give to the various components, and many paths to success as a crop. Given the magnitude of year-to-year variations in weather, too precise a specification of yield components could be harmful, and selection for yield by emphasizing particular componenets is not always effective (e.g., Rasmussen and Cannell, 1970). Perhaps the most important feature to preserve is some capacity to “overproduce” the yield components determined at successive stages in their life cycle, even into the grain-filling stage. This must of course be wasteful of resources, and there may be an optimal degree of overproduction at each stage which will depend on the climatic variability at each site and stage. In spite of the great extent of compensatory variation within each of the major cereals, they nevertheless differ substantially one from the other in their yield component balance. To some extent this is determined by individual plant size, as in the range of optimum planting densities m-’ from 4-8 for maize and 30-60 for sorghum up to about 300 for wheat and rice under favorable conditions. They also reflect kernel size, which displays more than a 10-fold. range from
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
349
20-25 mg for rice, 25-35 mg for sorghum, rather more for the temperate cereals, and up to 250-350 mg for maize. Grain number m-’ tends to vary inversely with kernel size from up to 5000 m-’ in maize to up to 25,000 m-’ in wheat and 45,000 m-’ in rice. Grain number per inflorescence shows the greatest range of all, from 15-50 in the temperate cereals, rather more in rice, 500-1 500 in maize, and from 1500 to more than 12,000 per panicle in sorghum (Pepper and Prine, 1972). Clearly, the balance among the yield components varies greatly not only between the major cereals, but also between cultivars of each, and for each cultivar according to the environment. Their potential overproduction at each stage in the life cycle is important in conferring adaptability to seasonal fluctuations, and tends to result in negative correlations among some of the yield components. Such negative correlations may be interpreted as implying that the supply of assimilate limits grain yield, and it may often do so. Paradoxically, evidence can also be adduced that spare assimilatory capacity and unmobilized reserves may be present even when negative correlations among yield components are found. The more the agricultural environment can be specified and controlled, the more these reserves of potential yield can be mobilized into actual grain yield. On this argument, it is the correlative and feedback mechanisms operating within the cereals that require modification to permit advance in yield quite as much as do the photosynthetic and storage systems.
XI. Conclusion
We have tried in this review to emphasize the comparative aspects of yield development in the major cereals. However, relatively few studies actually make physiological comparisons among them. One reason for this lack of comparative data on the major cereals may be that they have evolved to take advantage of such different environments that valid comparisons are difficult to make. Indeed, it is their capacity to exploit a wide range of complementary environments so effectively that makes the cereals so important and fascinating as crop plants. Our purpose in emphasizing their comparative behavior has been to obtain a better understanding of how yield develops and is limited in each of them. Many aspects have been left aside, but even among those we have examined, a quite remarkable range of behavior is apparent. This physiological diversity is to be cherished, as the source of adaptation to a wide range of environments. It may be confusing, and may seem unnecessary in the face of the progressive control and homogenization of our agricultural environments, But it should be preserved and explored in the search for increased level, stability and quality of yield from mankind’s most important source of food, the cereals.
3 50
L. T. EVANS AND I. F. WARDLAW ACKNOWLEDGMENTS
We are most grateful to Drs P. M. Bremner, R. M. Gifford, H. M. Rawson and S. Yoshida for their helpful comments on this review.
REFERENCES
Abdul-Baki, A., and Baker, I. E. 1970. Plant Physiol. 45,698-702. Adontsev, A. I., Bondarenko, V. I., Grinchenko, A. L., and Samoshkin, A. A. 1970. Fiziol. Rust. 17,107-111. Allen, L. H.,Hanks, R. J., Aase,J. K., and Gardner, H. R. 1974. Agron. J. 66, 35-41. Allison, I. C. S. 1964. J. Agrk ScL 63, 1-4. Allison, J. C. S., and Watson, D. J. 1966. Ann. Bot. (London) [N. S.] 30,366-381. Angus, J. F., Jones, R., and Wilson, J. H. 1972. Aust. J. Agric. Res. 23,945-957. Apel, P., and Peisker, M. 1973. In “Breeding and Productivity of Barley,” pp. 4 3 3 4 4 4 . Inst. Cereal Crops, Kromeriz. Apel, P., Tschape, M., Schalldach, I., and Aurich, 0.1973. Photosynrhetica 7, 132-139. Archbold, H. K. 1945. Nature (London) 156,70-73. Archbold, H. K., and Mukhejee, B. N. 1942.Ann. Bot. (London) [N. S.] 6,1-41. Ariyanayagam, R. P., Moore, C. L., and Carangal, V. R. 1974. Crop Sci. 14,551-556. Asana, R . D. 1961. Arid Zone Res. 16,183-190. Asana, R. D., and Joseph, C. M. 1964. Indian J. Plant Physiol. 7,86-101. Asana, R. D., and Saini, A. D. 1958. Physiol. Plant. 11,666474. Asana, R. D., and Singh, D. N. 1967. Indian J. Plant Physiol. 10,154-169. Aspinall, D., Nicholls, P. B., and May, L. H. 1964. Aust. J. Agric. Res. 15,729-745. Austin, R. B., and Edrich, J. 1975. Ann. Bot. (London) [N.S . ] 39,141-152. Baker, D. N., and Musgrave, R. B. 1964. Crop Sci. 4,127-131. Baldy, C. 1973. Ann. Agron. 24,241-276. Batch, J. J., and Morgan, D. G. 1974. Nature (London) 250,165-167. Beg, J. E. 1965. Nature (London) 205,1025-1026. Benci, J. F., Aase, J. K., and Ferguson, A. H. 1973. Agron. J. 65, 373-377. Bennett, M. D., Finch, R. A., Smith, J. B., and Rao, M. K. 1973. Proc. R. Soc. London, Ser. B 138,301-319. Benoit, C . R., Hatfield, A. L., and Ragland, J. L. 1965. Agron. J. 57, 223-226. Bhan, S., Singh, H. G., and Singh, A. 1973. Indian J. Agric. Sci. 43,828-830. Bingham, J. 1966. Ann. Appl. Biol. 47,365-377. Bingham, J. 1967. J. Agric. Sci. 68,411-422. Bingham, J. 1969.Agric. b o g . 4 4 , 3 0 4 2 . Birecka, H., and DakiC-Wlodkowska, L. 1963. Acta SOC.Bot. Pol. 32,631-650. Birecka, H., Skupinska, J., and Bernstein, I. 1964. Acta SOC.Bot. Pol. 33,601-618. Birecka, H., Skupinska, J., and Bernstein, I. 1967. Acta Soc. Bof. Pol. 36,387-409. Birecka, H., Skiba, T., and Kozlowska, Z. 1969. Bull. Acad. Pol. Sci., CI. 5 17,121-127. Biscoe, P. V., Littleton, E. J., and Scott, R. K. 1973. Ann. Appl. Biol. 75, 385-297. Biscoe, P. V., Scott, R. K., and Monteith, J. L. 1975a. J. Appl. Ecol. 12,269-291. Biscoe, P. V., Gallagher, J. M., Littleton, E. J., Monteith, J. L., and Scott, R. K. 1975b. J. Appl. E d . 12,295-318. Bonnett, 0. T. 1967. Ill., Agric. Exp. Stn., Bull. 721, 1-105. Bremner, P. M. 1969. J. Agric. Sci. 72, 273-280.
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
35 1
Bremner, P. M. 1972. Aust. J. Biol. Sci. 25,657-668. Brouwer, R. 1966. In “The Growth of Cereals and Grasses” (F.L. Milthorpe and J. D. lvins, eds.), pp. 153-166. Butterworth, London. Bruinsma, J., and Schuurman, J. J. 1966. Plant Soil 24,309-316. Burton, G . W., and Powell, J. B. 1968. Adv. Agron. 2 0 , 5 0 4 9 , Buttrose, M. S., and May, L. H. 1959. Aust. J. Biol. Sci. 12,40-52. Campbell, C. A., and Read, D. W. L. 1968. Can. J. Plant Sci. 48,299-311. Campbell, C. A., Pelton, W. L., and Neilson, K. F. 1969. a n . J. Plant Sci. 49,685-699. Campbell, D. K., and Hume, D. J. 1970. Crop Sci. 10,625-626. Canny, M. J. 1960. Biol. Rev. Cambridge Philos. SOC.35,507-532. Carpenter, R. W., Haas, H. J., and Miles, E. F. 1952. Agron. J. 44,420-423. Carter, M. W., and Poneleit, C. G . 1973. Crop Sci. 13,436439. Cataldo, D. A., Schrader, L. E., Peterson, D. M., and Smith, D. 1975. Crop Sci. 15, 19-23. Chandraratna, M. F. 1954. New Phytol. 53, 397-405. Chang, T. T.,and Oka, H. I. 1976. In “Climate and Rice,’’ lnt. Rice Res. Inst., Los Ban& (in press). Clarkson, D. T., Sanderson, J., and Russell, R. S. 1968. Nature (London) 220,805-806. Cock, J. H., and Yoshida, S. 1972. Proc. Crop Sci. SOC.Jpn. 41,226-234. Cock, J. H., and Yoshida, S. 1973. SoilSci. Plant Nutr. (Tokyo) 19,229-234, Collins, F. C., and Pickett, R. C. 1972. Crop Sci. 12,423425. Connor, D. J. 1975. Aust. J. Plant Physiol. 2,353-366. Cooper, J. P. 1956. J. Agric. Sci. 47, 262-279. Criswell, J. G., and Shibles, R. M. 1972. Iowa StateJ. Sci. 46,405-415. Cruz, L. J., Cagampang, G. B., and Juliano, B. 0. 1970. Plant Physiol. 46,743-747. Curtis, D. L. 1968. J. Appl. E d . 5, 215-226. Dalton, L. G. 1967. Crop Sci. 7,271. Dantuma, G . 1973. Neth. J. Agric. Res. 21, 188-197. Daynard, T. B., and Duncan, W. G. 1969. Crop Sci. 9,473476. Daynard, T. B., Tanner, J. W., and Duncan, W. G. 1971. Crop Sci. 1 1 , 4 5 4 8 . Denmead, 0. T., and Shaw, R. H. 1960. Agron. J. 52,272-274. Derieux, M., Kerrest, R., and Montalant, Y. 1973. Ann. Amelior. Plant. 23.95-107. de Wet, J. M. J., and Harlan, J. R. 1971. Econ. Bot. 25, 128-135. Doggett, H. 1970. “Sorghum.” Longmans, Green, New York. Dore, J. 1959. Nature (London) 183,413414. Downes, R. W. 1971. Aust. J. Biol. Sci. 24,843-852. Downes, R. W., and Hesketh, J. D. 1968. Planta 78, 79-84. Downton, W. J. S. 1971. In “Photosynthesis and Photorespiration” (M. D. Hatch, C. B. Osmond, and R. 0. Slatyer, eds.), pp. 554-558. Wiley, New York. Downton, W. J. S., and Tregunna, E. G. 1968. Can. J. Bot. 46,207-215. Dudley, J. W. 1973. Proc. Annu. Corn Sorghum Res. Con$ 28,123-136. Duncan, W. G. 1971. CropSci. 11,482485. Duncan, W. G. 1975. In “Crop Physiology” (L. T. Evans, ed.), pp. 23-50. Cambridge Univ. Press, London and New York. Duncan, W. G., and Hatfield, A. L. 1964. Crop Sci. 4,550-551. Duncan, W. G., and Hesketh, J. D. 1968. Crop Sci. 8,670-674. Duncan, W. G., Hatfield, A. L., and Ragland, J. L. 1965. Agron. J. 57, 221-223. Duncan, W. G., Loomis, R. S.,Williams,W.A., and Hanau, R. 1967. Hilgardiu 38,181-205. Dunstone, R. L., and Evans, L. T. 1974. Aust. J. Plant Physiol. 1, 157-165. Dunstone, R. L., Gifford, R. M., and Evans, L. T. 1973. Ausf. J. Biol. Sci 26, 295-307. Eastin, J. A. 1969. Proc. Annu. Corn Sorghum Res. Con5 24,81-89.
352
L. T. EVANS AND I. F. WARDLAW
Eastin, J. D. 1968. Proc. Annu. Corn Sorghum Res. Conf: 23, 129-136. Eastin, J. D. 1972a. In “Sorghum in the Seventies” (N. G . P. Rao and L. R. House, eds.), pp. 214-246. IBH & Oxford, New Delhi. Fastin, J. D. 1972b.Proc. Annu. Corn Sorghum Res. Conf 27,7-17. Eastin, J. D., and Sullivan, C. Y. 1969. Crop Sci. 9,165-166. Eberhart, S . A., and Sprague, G. F. 1973. Agron. J. 65,365-373. Eik, K., and Hanway, J. J. 1966. Agron. J. 58.16-18. El Sharkawy, M. A., and Hesketh, J. D. 1964. Oop Sci. 4,514-518. El Sharkawy, M. A., and Hesketh, J. D. 1965. Crop Sci. 5,517-521. Epstein, E., and Jefferies, R. L 1964. Annu. Rev. Plunt Physiol. 15, 169-184. Evans, L. T. 1972. In “Rice Breeding,” pp. 499-511. Int. Rice Res. Inst., Los Ban6s. Evans, L. T. 1973. In “Plant Response to Climatic Factors” (R. 0. Slatyer, ed.), pp. 21-35. UNESCO, Paris. Evans, L. T. 1975. In “Crop Physiology” (L. T. Evans, ed.). pp. 327-355. Cambridge Univ. Press, London and New York. Evans, L. T. 1976a. Proc. R. Soc. London (in press). Evans, L. T. 1976b. In “Climate and Rice.” Int. Rice Res. Inst., Los Banes (in press). Evans, L. T., and Dunstone, R. L 1970. Aust J. Biol. Sci. 23,725-741. Evans, L. T.,and Rawson, H. M. 1970. Aust. J. Biol. Sci. 23,245-254. Evans, L. T., Dunstone, R. L., Rawson, H. M.,and Williams, R. F. 1970. Aust. J. Biol. Sci. 23,743-752. Evans, L. T., Bingham, J., and Roskams, M. A. 1972. Aust. J. Biol. Sci. 25,l-8. Evans, L. T., Wardlaw, I. F., and Fischer, R. A. 1975. In “Crop Physiology” (L. T. Evans, ed.), pp. 101-149. Cambridge Univ. Press, London and New York. Ferguson, H., Eslick, R. F.,and Aase, J. K. 1973.Agron. J. 65,425-428. Ferguson, I. B., and Clarkson, D. T. 1975. New Phytol. 75,69-79. Fernandez, R.,and Laud, R. T. 1959. Agron. J. 41, 33-36. Fischer, K. S., and Wilson, G. L. 1971a. Aust. J. Agric. Res. 22,33-37. Fischer, K. S., and Wilson, G. L. 1971b. Aust. J. Agric. Res. 22,39-47. Fischer, K. S., and Wilson, G. L. 1975a. Aust. J. Agric. Res. 26, 11-23. Fischer, K. S., and Wilson, G. L. 1975b. Aust. J. Agric. Res. 26,25-30. Fischer, K. S., and Wilson, G. L. 1975c. Aust. J. Agric. Res. 26, 31-41. Fischer, R. A. 1973. In “Plant Response to Climatic Factors” (R. 0. Slatyer, ed.), pp. 233-241. UNESCO, Paris. Fischer, R. A. 1975. CropSci. 15,607-613. Fischer, R. A., and Aguilar, M. I. 1976. Agron. J. (in press). Fischer, R. A., and Kohn, G. D. 1966. Aust. J. Agric. Res. 17, 281-295. Fischer, R. A., and Laing,D. R. 1976. J. Agric. Sci. (in press). Fisher, J. E. 1972. Bot. Gaz. (Chicugo) 133,78-85. Fousova, S., and Avratovschukova, N. 1967. Photosyntheticu 1,3-12. Francis, C. A. 1972. Proc. Annu. Corn Sorghum Res. Con8 27,119-131. Friend, D. J. C. 1965. CanJ. Bot. 43,345-363. Friend, D. J. C. 1966. In “The Growth of Cereals and Grasses” (F. L. Milthorpe and J. D. Ivins, eds.), pp. 181-199. Butterworth, London. Funnah, S. M. 1971. M.Sc. Thesis, University of Florida, Gainesville. Gale, M. D., Edrich, J., and Lupton, F. G. H. 1974. J. Agric. Sci. 8 3 , 4 3 4 6 . Galinat, W. C., and Naylor, A. W. 1951. A m J. Bot. 38, 38-47. Gallaher, R. N., Ashley, D. A., and Brown, R. H. 1975. O o p Sci. 15,55-59. Gates, J. W., and Simpson, G. M. 1968. Clm. J. Bot. 46,1459-1462. Gifford, R. M. 1970. Ph.D.Thesis, Cornell University, Ithaca, New York.
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
353
Gifford, R. M. 1971. In “Photosynthesis and Photorespiration” (M. D. Hatch, C. B. Osmond, and R. 0. Slatyer, eds.), pp. 51-56. Wiley, New York. Gifford, R. M. 1974a.Aust. J. Plant Physiol. 1, 107-117. Gifford, R. M. 1974b.R. Soc. N Z., Bull. 12,887-893. Gifford, R. M., and Musgrave, R. B. 1973.Aust. J. Biol. Sci. 26,3544. Gifford, R. M.,Bremner, P. M., and Jones, D. B. 1973.Aust. J. Agric. Res. 24, 297-307. Giles, K. L., Beardsell, M. F., and Cohen, D. 1974.Plant Physiol. 54,208-21 2. Gingrich, J. R., and Russell, M. B. 1957.Soil Sci. 84,185-194. Goldsworthy, P. R. 1970.J. Agric. Sci. 75,523-531. Goldsworthy, P . R. 1974.J.Agric. Sci. 83,213-221. Goldsworthy, P. R., and Colegrove, M. 1974.J. Agric. Sci. 83,213-221. Goldsworthy, P. R., Palmer, A. F. E., and Sperling, D. W. 1974.J. Agric. Sci. 83, 223-230. Gott, M. B. 1961.Aust. J. Agric. Res. 12, 547-565. Grafius, J. E.,and Okoli, L. B. 1974.Crop Sci. 14, 353-355. Greenway, H. 1962.Aust. J. Biol. Sci. 15,16-38. Greenway, H., and Gunn, A. 1966.Planta 71,43-61. Greenway, H., and Pitman, M. G . 1965.Aust. J. Biol. Sci. 18,235-247. Hackett, C. 1969.New Phytol. 68,1023-1030. Hackett, C. 1971.Aust. J. Biol. Sci. 24,1057-1064. Hackett, C. 1973.Aust. J. Biol. Sci. 26,1211-1214. Haggar, R. J., and Couper, D. C. 1972.Exp. Agric. 8,251-263. Halse, N. J., and Weir, R. N. 1974.Aust. J. Agric. Res. 25, 687-695. Hake, N. J., Greenwood, E. A. N., Lapins, P., and Boundy, C. A. P. 1969.Aust. J. Agric. Res. 20,987-998. Hanson, J. C., and Rasmussen, D. C. 1975.Crop Sci. 15, 248-251. Hanway, J. J. 1962.Agron. J. 54,217-224. Harlan, J. R. 1971.Science 174,468-474. Harlan, J. R.,and Zohary, D. 1966.Science 153,1074-1080. Hasegawa, G . 1962.Mem. Hyogo. Univ. Agric. 11, Agron. Ser. 5, 1-154. Hatch, M. D., and Glasziou, K. T. 1964.Plant Physiol. 39,180-184. Hatch, M. D., Slack, C. R., and Johnson, H. S . 1967.Biochem. J. 102,417-422. Hayashi, K. 1966.Proc. Crop Sci. SOC.Jpn. 35,205-211. Heichel, G. H. 1971.Photosynthetica 5,93-98. Heichel, G . H., and Musgrave, R. B. 1969. O o p Sci. 9,483-486. Hesketh, J. D. 1963.Crop Sci. 3,493496. Hesketh, J. D. 1968.Aust. J. Biol. Sci. 21,235-241. Hesketh, J. D.,and Musgrave, R. B. 1962.Crop Sci. 2, 311-315. Hesketh, J. D., Baker, D. N., and Duncan, W. G. 1971. Crop Sci. 11,394-398. Hofstra, G., and Hesketh, J. D. 1969.PIanta 85, 228-237. Hofstra, G., and Nelson, C. D. 1969.PIanta 88, 103-112. Holmes, D.P. 1973.Can. J. Bot. 51,941-956. Hoshikawa, K. 1959.B o c . Crop Sci. SOC.Jpn. 28,291-295. Hsiao, T. C., and Acevedo, E. 1914.Agric. Meteorol. 14,59-84. Hucklesby, D. P., Brown, C. M., Howell, S. E., and Hageman, R. H. 1971.Agron. J. 63,
274-216. Humphries, E. C. 1968.Field Crop Abstr. 21,91-99. Hunter, R. B., Hunt, L. A,, and Kannenberg, L. W. 1974. Can. J. Plant Sci. 54,71-78. Hurd, E. A. 1968.Agron. J. 60,201-205. Hurd, E. A. 1974.Agric. Meteorol. 14, 39-55. Inosaka, M. 1957.Proc. CropSci. SOC.Jpn. 26,197-198.
354
L. T. EVANS AND I. F. WARDLAW
Jackson, W. A., Johnson, R. E., and Volk. R. J. 1974. Physiol. Plant. 32,3742. Jenner, C. F. 1973. J. Exp. Bot. 24, 295-306. Jenner, C. F. 1974a. R. SOC.N. Z.. Bull. 12,901-908. Jenner, C . F. 1974b. Aust. J. Plant Physiol. 1,319-329. Jenner, C . F., and Rathjen, A. J. 1975. Aust. J. Plant Physiol. 2,311-322. JeMhgS, A. C., and Morton, R. K. 1963. Aust. J. Biol. Sci. 16, 318-331. Johnson, D. R., and Tanner, J. W. 197%. C h p Sci. 12,482485. Johnson, D. R., and Tanner, J. W. 1972b. C h p Sci. 12,485486. Johnson, R. R., Willmer, C. M., and Moss, D. N. 1975. Crop Sci. 15,217-221. Johnson, V. A., Mattern, P. J., and Schmidt, J. W. 1967. Crop Sci. 7,664-667. Kassam, A. H., and Kowal, J. M. 1973. Savanna 2 , 3 9 4 9 . Katayama, T. C. 1964a. Jpn. J. Bot. 18,309-348. Katayama, T. C. 1964b. Jpn. J. Bot. 18,349-383. Katayama, T. C. 1970. Mem Fac. Agric., Kagoshima Univ. 7,219-241. Katayama, T. C. 1974. Proc. Crop Sci. SOC.Jpn. 43,224-236. Kellogg, C. E., and Orvedal, A. C. 1969. Adv. &on. 21,109-170. Khan, M. A., and Tsunoda, S. 1970a. Jpn J. Breed. 20,133-140. Khan, M. A., and Tsunoda, S. 1970b. Jpn. J. Breed. 20,305-314. Khan, M. A., and Tsunoda, S. 1970c. Tohoku J. Agric. Res. 21,47-59. Khan, M. A., and Tsunoda, S. 1971.Jpn. J. Breed 21,143-150. Kiesselbach, T. A. 1948. J. Am. Soc. Agron. 40, 216-236. Kiesselbach, T. A., and Walker, E. R. 1952. A m J. Bot. 39,561-569. King, R. W., and Evans, L. T. 1967. Aust. J. Biol. Sci. 20,623-635. King, R. W., Wardlaw, I. F., and Evans, L. T. 1967. Planta 77,261-276. Kinoch, H. G., Ramig, R. E., Fox, R. L., and Koehler, J. E. 1957. &on. J. 49,20-25. Kirby, E. J. M. 1969. Ann. Appl. Biol. 63,513-521. Kirby, E. J. M. 1974. J. Agric. Sci. 82,437447. Kirby, E. J. M.,and Rymer, J. L. 1974. Ann. Bot. (London) (N.S.] 38,565-573. Kishitani, S., and Tsunoda, S. 1974. Photosyntheticu 8,161-167. Kjack, J. L., and Witters, R. E. 1974. Crop Sci. 14,243-248. Kornilov, A. A. 1960. DokL Akad. Nauk SSSR 130,25-27. Kowal, J. M., and Kassam, A. H. 1973. Agric. Meteorol. 12,391406. Krenzer, E. G., and Moss, D. N. 1975. Crop Sci. 15,71-74. Krishnamurthy, K., Rajashekara, B. G., Jagannath, M. K., Bombe Gowda, A., Raghunatha, G., and Venugopal, N. 1973. Agron. J. 65,858-860. Kudo, K. 1975. Jpn. IBPSynth. 11,199-220. Langer, R. H. M. 1966. In “The Growth of Cereals and Grasses’’ (F.L. Milthorpe and J. D. lvins, eds.), pp. 21 3-226. Butterworth, London. Lebsock, K. L, Joppa, L. R., and Walsh, D. E. 1973. C h p Sci. 13,670-674. Lee, J. H. 1972. R o c . Crop Sci. Soc. Jpn. 41, 1-14. Lee, J. H., and Ota, Y. 1970. Roc. b p Sci. SOC.Jpn. 39,505-510. Leng, E. R. 1963. CropSci. 3,187-190. Levy, J., and Peterson, M. L. 1972. Crop Sci 12,487490. Lm,S., and Tanaka, A. 1967. PIant Soil 26,333-347. Lizandr, A. A., and Brovtsyna, V. L 1964. Sov. Plant Physiol. (Engl. Transl.) 11,333-338. Lupton, F. G. H. 1968. Ann. Appl. Biol. 61,109-119. Lupton, F. G. H. 1969. Ann. Appl. Biol. 64,363-374. Lupton, F. C. H., Oliver, R. H., Ellis, F. B., Barnes, B. T., Howse, K. R., Welbank, P. J., and Taylor, P. J. 1974. Ann. Appl. Biol. 77,129-144. Lush, W. M., and Evans, L. T. 1974. Aust. J. Plant Physiol. 1,417431. McCIure, J. W., and Harvey, C. 1962.Agron. J. 54,457459.
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
355
McCree, K. J. 1970. In “Prediction and Measurement of Photosynthetic Productivity,” pp. 221-229. Pudoc, Wageningen. McCree, K. J. 1972. Agric. Meteorol. 9, 191-216. McCree, K. J. 1974. CropSci. 14,509-514. McNeal, F. H., Berg, M. A., McCuire, C: F., Stewart, V. R., and Boldridge, D. E. 1972. Crop Sci. 12,599-602. McPherson, H. G., and Slatyer, R. 0. 1973. Aust. J. Bid. Sci. 26, 324-339. Malm, N. R., and Rachie, K. 0. 1971. “The Setaria Millets,” SB 513, pp. 1-133. Univ. of Nebraska Press, Lincoln. Mann, H. S. 1957. Indian J. Agron. 2, 13-26. Marcellos, H., and Single, W. V. 1972. Aust. J. Agric. Res. 23, 533-540. Marshall, C., and Wardlaw, I. F. 1973. Aust. J. Bid. Sci. 26, 1-1 3. Matsushima, S. 1962. Proc. Crop Sci. SOC.Jpn. 31,115-121. Matsushima, S. 1970. “Crop Science in Rice.” Fuji, Tokyo. May, L. H., Chapman, F. H., and Aspinall, D. 1965. Aust. J. Biol. Sci. 18, 25-35. May, L. H., Randles, F. H., Aspinall, D., and Paleg, L. G. 1967. Ausr. J. Biol. Sci. 20, 27 3-28 3. Mehrotra, 0. N., Sinha, N. S., and Srivastava, R. D. L. 1968. Phnt Soil 28,422-430. Mengel, D. B., and Barber, S. A. 1974. Agron. J. 66, 399-402. Meyer, R. E., and Gingrich, J. R. 1966. Agron. J. 58, 377-381. Michael, G., and Seiler-Kelbitsch, H. 1972. Crop Sci. 12, 162-165. Michael, G., Seiler-Kelbitsch, H., and Wagner, H. 1973. In “Breeding and Productivity of Barley,” pp. 397-41 2. Inst. Cereal Crops, Kromeriz. Miflin, B. J. 1967. Nuture (London) 214,1133-1 134. Minotti, P. L., and Jackson, W. A. 1970. Plnntu 95, 36-44. Misra, G., and Khan, P. A. 1973. Bor. Gar. (Chicago) 134, 93-99. Moorby, J., Troughton, J. H., and Currie, B. G. 1974, J. Exp. Bot. 25,937-944. Moss, D. N. 1962. Crop Sci. 2,366-367. Moss, D. H., and Musgrave, R. B. 1971. Adv. Agron. 23,317-336. Moss, D. N., and Rasmussen, H. P. 1969. Plant Physiol. 44,1063-1068. Moss, D. N.. and Stinson, H. T. 1961. Crop Sci. 1,416-418. Murata, Y. 1961. Bull. Natl. Inst. Agric. Sci., Ser. D 9, 1-169. Murata, Y. 1964. In “The Mineral Nutrition of the Rice Plant,’’ pp. 385400. Johns Hopkins Press, Baltimore, Maryland. Murata, Y. 1975.Agric. Meteorol. 15,117-131. Murata, Y., and Iyama, J. 1963. Roc. G o p Sci. Soc. Jpn. 31, 315-321. Murata, Y., and Matsushima, S. 1975. In “Crop Physiology” (L. T. Evans, ed.), pp. 73-99. Cambridge UNv. Press, London and New York. Murata, Y., and Togari, Y. 1975. Jpn. IBP Synrh. 11,9-19. Murata, Y., Iyama, J., and Honma, T. 1965a. Proc. Crop Sci SOC.Jpn. 34,148-153. Murata, Y., Iyama, J., and Honma, T. 1965b. Roc. Crop Sci. SOC.Jpn. 34,154-158. Mwayama, N., Yoshino, M., Oshima, M., Tsukahara, S., and Kawarazaki, K. 1955. BUN. Natl. Inst. Agric. Sci. Jpn., Ser. B 4, 123-166. Murayama, N., Oshima, M., and Tsukahara, S. 1961. J. Sci. Soil & Munure, Jpn. 32, 26 1-265. Nagai, T., Yoshida, S., and Tatsumichi, Y. 1962. Proc. Crop Sci. Soc. Jpn. 30,137-142. Nagato, K., and Ebata, M. 1959. Proc. Czop Sci. SOC.Jpn 28,275-278. Nagato, K., Ebata, M., and Kono, Y. 1961. Roc. Crop Sci. Soc. Jpn. 29,337-340. Nanda, K. K., and Chinoy, J. J. 1958. Curr. Sci. 27,141-143. Nanda, K. K., Grover, R., and Chinoy, J. J. 1957. Qyton 9,15-24. Natr, L. 1967. Photosynthetica 1,29-36.
356
L. T. EVANS AND I. F. WARDLAW
Natr, L. 1973.In “Breeding and Productivity of Barley,” pp. 417-432. Inst. Cereal Crops, Kromeriz. Natr, L., and Apel, P. 1974.Photosynthetica 8,53-56. Noggle, J. C., and Fried, M. 1960.Soil Sci. Soc. Am., Proc. 24, 33-35. Oka, H. I. 1958. Qyton 11,153-160. Onderdonk, T . J., and Ketcheson, J. W. 1973.Plant Soil 39,177-186. Osada, A., and Murata, Y. 1965.Proc. Ctop Sci. SOC.Jpn. 33,454459. Oshima, M. 1966.J. Sci. Soil & Manure, Jpn. 37,589-593. Owen, P. C. 1969.Exp. Agric. 5, 85-90. Palmer, A. F. E. 1969.Ph.D. Thesis, Comell University, Ithaca, New York. Palmer, A. F. E., Heichel, G. H.,and Musgrave, R. B. 1973. Crop Sci. 13, 371-376. Passioura, J. B. 1972.Aust. J. Agric. Res. 23, 745-152. Passioura, J. B., and Ashford, A. E. 1974.Aust. J. Plant Physwl. 1,521-527. Paul, A. K.,Mukheji, S., and Sircar, S. M. 1971.Physiol. Plant. 24,342-346. Pavlov, A. N. 1969. Biologh 4,230-235. Pavlov, A. N., and Kolesnik, T. I. 1974. Sov. Plant Physiol. (Engl. nand.) 21, 267-272. Pavlychenko, T. K. 1937.Ecology 18,62-79. Peaslee, D. E., Ragland, J.L., and Duncan, W. G. 1971.Agron. J. 63,561-563. Pendleton, J. W.,and Hammond, J. J. 1969.Agron. J. 61,911-913. Pendleton, J. W., and Weibel, R. 0. 1965.Agron. J. 57, 292-293. Penning de Vries, F. W. T. 1972.In “Crop Processes in Controlled Environments” (A. R. Rees et al., eds.), pp. 327-346. Academic Press, New York. Penning de Vries, F. W. T. 1975.In “Photosynthesis and Productivity in Different Environments” ( J. P. Cooper, ed.), pp. 459480. Cambridge Univ. Press, London and New York. Pepper, G. E., and Rine, G. M. 1972. O o p Sci. 12,590-593. Peterson, D. M., Schrader, L. E., Cataldo, D. A., Youngs, V. L., and Smith, D. 1975. Can. J. Plant Sci. 55,19-28. Petinov, N. S., and Pavlov, A. N. 1955.Fiziol. Rast. 2, 113-122. Pinthus, M.J. 1973.Adv. Agron. 25,209-263. Pinthus, M.J., and Eshel, Y. 1962.Isr. J. Agric. Res. 12, 13-20. Pitman, M. G. 1965.Aust. J. Biol. Sci. 18,541-546. Prine, G . M. 1971.Ctop Sci. 11,782-786. Puckridge, D. W. 1969.Aust. J. Agric. Res. 20,623-634. Puckridge, D. W. 1971.Aust. J. Agric. Res. 22, 1-9. Puckridge, D. W., and Ratkowsky, D. A. 1971.Aust. J. Agric. Res. 22,ll-20. Rasmussen, D. C.,and Cannell, R. Q. 1970. Crop Sci. 10,51-54. Rawson, H. M. 1970.Aust. J. Biol. Sci. 23,l-15. Rawson, H. M., and Bremner, P. M. 1976.In “Crop Physiology” (U. S. Gupta, ed.). IBH & Oxford, New Delhi (in press). Rawson, H. M., and Evans, L. T. 1970.Aust. J. Biol. Sci. 23,753-164. Rawson, H. M., and Evans, L. T. 1971.Aust. J. Agric. Res. 22,851-863. Rawson, H. M.,and Hofstra, G. 1969.Aust. J. Biol. Sci. 22, 321-331. Rawson, H.M., Gifford, R. M., and Bremner, P. M. 1976.Planta (in press). Razumov, V. I. 1955.Fiziol. Rast. 2,247-252. Rhoades, M. M., and Carvalho, A. 1944.Bull. Torrey Bot. Club 71, 335-346. Robards, A. W.,Jackson, S. M., Clarkson, D. T., and Sanderson, J. 1973.Protoplasma 77,
291-311. Ross, W. M.,and Eastin, J. D. 1972.Field Crop Abstr. 25, 169-174. Rovira, A. D.,and Bowen,G. D. 1973. Planta 114,101-107.
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
357
Roy, R. N., and Wright, B. C. 1974. Agron. J. 66,s-10. Russell, W. A. 1972. Crop Sci. 12,90-92. Ryle, G. J. A., and Hesketh, J. D. 1969. Crop Sci. 9,451-454. Saghir, A. R., Khan, A. R., and Worzella, W. 1968. Agron. J. 60,95-97. Saki, F. A. K., and Shannon, J. C. 1975. Plant Physiol. 55, 881-889. Salim, M. H., Todd, G. W., and Schlehuber, A. M. 1965. Agron. J. 57,603607. Samygin, G. A. 1946. K. A. Timiriazeva Tr. Akad. Nauk SSSR 3,129-262. Satake, T., and Hayase, H. 1970. Proc. Crop Sci. SOC.Jpn. 45,468-473. Sato, K. 1971. Tohoku J. Agric. Res. 22,69-79. Sato, K., and Takahashi, M. 1971. TohokuJ. Agric. Res. 22,5748. Sawada, S. 1970. J. Fac. Sci., Univ. Tokyo, Sect. 3 10, 233-263. Sayre, J. D. 1948. Plant Physiol. 23,267-281. Scott Russell, R. 1970. Endeavour 29,60-66. Shannon, J. C. 1968. Plant Physiol. 43,1215-1 220. Shannon, J. C. 1972. Plant Physiol. 49,198-202. Shaw, R. H., and Thom, H. C. S. 1951. Agron. J. 43,541-546. Shen, Gong-mou. 1960. Acta Agric. Sin. 11, 30-40. Simpson, G. M. 1968. Can J. Plant Sci. 48, 253-260. Sims, J. L., and Place, G. A. 1968. Agron. J. 60,692-696. Skazkin, F. D., and Lukomskaya, K. A. 1962. Sov. Plant Physiol. (Engl. Transl.) 9, 5 6 1-56 3. Slootmaker, L. A. J. 1974. Euphytica 23,505-513. Sofield, I., Evans, L. T., and Wardlaw, I. F. 1974. R. SOC.N Z., Bull. 12,909-915. Soza, R. F. 1973. Ph.D. Thesis, University of Nebraska, Lincoln. Spiertz, J. H. J. 1974. Neth. J. Agric. Sci. 22, 207-220. Spiertz, J. H. J., ten Hag, B. A., and Kupers, L. J. P. 1971. Nerh. J. Agric. Sci. 19, 211-222. Stevenson, J. C., and Goodman, M. M. 1972. Crop Sci. 12, 864-868. Stickler, F. C., and Pauli, A. W. 1961.Agron. J. 53,352-353. Stoy, V . 1955. Physiol. Plant. 8,963-986. Stoy, V . 1963. Physiol. Plant. 16,851-866. Stoy, V . 1965. Physiol. Plant., Suppl. 4,l-125. Sugimoto, K. 1965. Proc. Crop Sci. SOC.Jpn. 34,6-13. Sullivan, C. Y.,and Blum, A. 1970. Proc. Annu. Corn Sorghum Res. Conf: 25,55-66. Takano, Y., and Tsunoda, S. 1971. Jpn. J. Breed. 21,69-76. Takeda, T., and Akiyama, T. 1973. Proc. Crop Sci. SOC.Jpn. 42,302-306. Tanaka, A. 1958. J. Sci. Soil & Manure, Jpn. 29,327-333. Tanaka, A., Kawano, K., and Yamaguchi, J. 1966. Int. Rice Res. Inst., Tech. Bull. 7, 1-46. Tanaka, T. 1972. Bull. Natl. Inst. Agric. Sci., Ser. A 19, 1-100. Tanaka, T., Matsushima, S., Kojo, S., and Nitta, H. 1969. Proc. Crop Sci. SOC.Jpn. 38, 287-293. Tanner, J. W.,Gardener, C. J., Stoskopf, N. C., and Reinbergo, E. 1966. Can. J. Plant Sci. 46,690. Terman, G. L., Ramig, R. E., Dreier, A. F., and Olson, R. A. 1969.Agron. J. 61, 755-759. Thorne, G. N. 1959. Ann. Bot. (London) [N. S.] 23,365-370. Thorne, G. N. 1963. Ann. Bot. (London) [N. S.] 27,155-174. Thorne, G. N. 1965.Ann. Bot. (London) [N. S.] 29,317-329. Thorne, G. N. 1970. In “Prediction and Measurement of Photosynthetic Productivity,” pp. 399-404. Pudoc, Wageningen. Thorne, G. N. 1973. Rep. Rofhamsted Exp. Sfn., Harpenden, Engl. Part 2, pp. 5-25.
358
L. T. EVANS AND I. F. WARDLAW
Thorne, G. N.,and Blacklock, J. C. 1971. Ann. Appl. Biol. 68,93-111. Toda, M. 1962. Proc. Crop Sci. SOC.Jpn. 30,241-249. Trenbath, B. R., and Angus, J. F. 1975. Field Crop Absfr. 28,231-244. Tripathy, P. C., Eastin, J. A., and Schrader, L. E. 1972. Crop Sci. 12,495497. Troughton, A. 1962. Commonw. Bur. Pastures Field Oops, Mimeo. Publ. 2,1-91. Troughton, A., and Whittington, W. J. 1969. In “Root Growth” (W. J. Whittington, ed.), pp. 296-314. Butterworth, London. Troughton, J. H., Chang, F. H.. and Currie, B. G. 1974a. Pkrnf Sci. Left. 3,49-54. Troughton, J. H., Moorby, J., and Currie, B. G.1974b. J. Exp. Bof. 25,684-694. Tsunoda, S. 1965. In “The Mineral Nutrition of the Rice Plant,” pp. 401-418. Int. Rice Res. Inst., Johns Hopkins. Van Dobben, W. H. 1962. Nefh J. Agric. Sci. 10,377-389. Velasco, J. R., and de la Fuente, R. K. 1958. Philipp. Agric. 42,12-17. Vergara, B. S., and Lilis, R. 1967. Nature (London) 216,168. Vergara, B. S., Chang, T. T., and Lilis, R. 1969. Inf. RiceRes. Insf., Tech. Bull. 8,1-31. Wall, P. C., and Cartwright, P. M. 1974. Ann. Appl. Biol. 76,299-309. Wallace, D. H., Ozbun, J. L., and Munger, H. M. 1972. Adv. Agron. 24,97-146. Wang, T.-D., and Yan, R. H. 1965. Acra Phytophysiol. Sin. 1,9-13. Wardlaw, I. F. 1965. Ausf. J. Biol. Sci. 18,269-281. Wardlaw, I. F. 1968. Bof. Rev. 34,79-105. Wardlaw, I. F. 1970. Ausf. J. Biol Sci. 23,765-774. Wardlaw, I. F. 1976. Aust. J. Plant Phys&l. 3,377-387. Wardlaw, I. F., and Marshall, C. 1976. Ausf. J. Plant Physiol. 3,389-400. Wardlaw, I. F., and Moncur, L. 1976. Planfa 128,93-100. Wardlaw, I. F., and Porter, H. K. 1967. Ausf. J. Biol. Sci. 20, 309-318. Watson, D. J. 1952. Adv. Agron. 4,101-145. Watson, D. J., Thorne, G. N.,and French, S. A. W. 1963. Ann. Bof. (London) [N. S . ] 27, 1-22. Welbank, P. J., and Williams, E. D. 1968. J. Appl. Ecol. 5 , 4 7 7 4 8 1 . Welbank, P. J., Witts, K. J., and Thorne, G. N. 1968. Ann. Bof. (London) [N. S . ] 32, 79-95. Welbank, P. J., Gibb, M. J., Taylor, P. J., and Williams, E. D. 1973. Rep., Rothamsfed Exp. Sm., Harpenden, Engl. Part 2, pp. 26-66. Whigham, D. K., and Woolley, D. G. 1974. &on. J. 6 6 , 4 8 2 4 8 6 . Willey, R. W., and Holliday, R. 1971a. J. Agric. Sci. 7 7 , 4 4 5 4 5 2 . Willey, R. W., and Holliday, R. 1971b. J. Agric. Sci. 7 7 , 4 5 3 4 6 1 . Williams, R. F. 1955. Annu. Rev. Plant PhysioL 6 , 2 5 4 2 . Williams, W. A., Loomis, R. S., and Lepley, C. R. 1965. Crop Sci. 5,211-215. Williams, W. A., Loomis, R. S., Dundan, W. G., Dovrat, A., and F. Nunez, A. 1968. O o p Sci. 8, 303-308. Wilson, D., and Cooper, J. P. 1970. New Phytol. 69,233-245. Winter, S . R., and Ohlrogge, A. 3.1973. &on. J. 65,395-397. Woodworth,C. N., Leng, E. R., and Jugenheimer, R. W. 1952. Agron. J. 4 4 , 6 0 4 5 . Wu, C.-J., and Yao, Y.-T. 1974. Taiwania 19,1-6. Yoshida, S . 1972. Annu. Rev. Plant Physwl. 2 3 , 4 3 7 4 6 4 . Yoshida, S . 1973a. SoilSci. Plant Nun. (Tokyo) 19,299-310. Yoshida, S. 1973b. Soil Sci. Plant Nufr. (Tokyo) 19, 311-316. Yoshida, S., and Parao, F. T. 1976. In “Climate and Rice.” Int. Rice Res. Inst., Los BaAos (in press).
COMPARATIVE PHYSIOLOGY OF GRAIN YIELD IN CEREALS
3 59
Yoshida, S., Cock, J. H.,and Parao, F. T. 1972. In “Rice Breeding,” pp. 455-468. Int. Rice Res. Inst., Los Banes. Youngs, V. L., and Shands, H. L. 1974. O o p Sci. 14,578-580. Zee, S . Y., and O’Brien, T. P. 1970. Ausf. J. Bwl. Sci. 23, 783-791. Zee, S. Y., and O’Brien, T. P. 1971a. Ausf. J. Biol. Sci. 24, 35-49. Zee, S. Y., and O’Brien, T. P. 1971b. Ausf. J. Biol. Sci. 24, 391-395. Zillinsky, F. J. 1974. Adv. Agron. 26,315-348. Zohary, D., Harlan, J. R., and Vardi, A. 1969. Euphyficu 1 8 , 5 8 4 5 .
This Page Intentionally Left Blank
THE BIOLOGICAL YIELD AND HARVEST INDEX OF CEREALS AS AGRONOMIC AND PLANT BREEDING CRITERIA C. M. Donald and J. Hamblin’ Waite Agricultural Research Institute, University of Adelaide, South Australia, and Department of Applied Biology, University of Cambridge, England
I. Introduction .................................................. 11. The Relationship of Biological Yield, Grain Yield, and Harvest Index to Each Other and to Other Plant Characteristics ............................. 111. The Influence of Environmental Factors ............................. A. PopulationDensity ........................................... B. Water Availability ............................................ C. Nitrogen Nutrition ........................................... D. The Interaction of Environmental Factors, Yield, and Harvest Index ..... IV. Biological Yield and Harvest Index as Criteria in Cereal Breeding .......... V. ConcludingComments ........................................... References ....................................................
361 364 375 375 318 382 390 390 402
404
I. Introduction
The biological yield of a cereal crop is the total yield of plant material, and the harvest index is the ratio of the yield of grain to the biological yield. Expression of the “efficiency” of grain production through an index was proposed 60 years ago by E. S. Beaven, a barley breeder at Warminster, Wiltshire, England. He defined the “migration coefficient” of cereals as “the proportion of dry matter of the entire ripe plant, excluding the root, which is accumulated in the grain” (Beaven, 1914, 1920). The term “migration coefficient” was chosen because, lacking our present understanding of postflowering photosynthesis, Beaven believed “that the yield of dry grain, as distinct from the total weight of the entire plant, depends on the effective transmission of the material accumulated in the stems of the plant .” Beaven noted that the migration coefficient is much more constant for the individual plants of a variety than is the number of tillers or the size of heads and that in different seasons the separate varieties tend to maintain much the same ranking for the ratio, although the mean of all varieties will vary from one season to the next. He further wrote that “we shall do better in the long run if we attach special importance to this character of good migration for the purpose in the first place, of selecting Presenr address: Department of Agriculture, Perth, Western Australia. 361
362
C. M. DONALD AND J. HAMBLIN
individual plants and in the second place for the purpose of assessirig the comparative values of the small aggregates of each race (genotype) obtained within the first few generations.” He did not, however, present full data in support of these views. In the following years a few agronomists emphasized the usefulness of the migration coefficient; Engledow and Wadham (1923, 1924) wrote that “it seems possible that the coefficient might . . . prove to be of service as an index of the yielding power of single plants.” But it was not widely adopted. At the Rothamsted Experimental Station, the published results of many agronomic experiments included “the ratio of grain to total produce,” but apparently independently of Beaven’s writings (e.g., Russell and Watson, 1940). Among cereal breeders, Beaven’s advocacy of the migration coefficient was quite disregarded. The measurement of grain yield alone was simple and rapid and was believed to provide the significant information needed in breeding cereals for yield. This position continues substantially unchanged. After an interval of several decades, interest in the ratio among crop ecologists was stimulated when Niciporovic (1956), in a notable lecture to the Timijazev Institute of Plant Physiology, emphasized that successful crop production depends on the effective exploitation of photosynthesis to achieve maximum biological yield. But in most crops the economic product is not the whole crop but some particular part of it such as the grain or tubers (the “economic yield”). If economic yield is to be maximal, continued Niciporovic, there must be a correct distribution at the right time of the products of photosynthesis. Black and Watson (1960) comment that Niciporovic’s “discussion seems to assume that the final distribution of dry matter is the result of a phase of accumulation followed ‘at the right time’ by a phase of partition, but this is evidently not always true .” Niciporovic’s thoughts regarding economic yield were expressed in the following equation: Coefficient of effectiveness Biological yield X of formation of the economic = Economic yield part of the total yield ybiol
x Kecon = Yecon
It will be seen that Niciporovic’s “coefficient of effectiveness” is identical with Beaven’s “migration coefficient” except that he emphasized that it could be applied to any crop, whether the economic yield were grain, tubers, fiber, oil, or any other product. Unlike Beaven, Niciporovic placed equal emphasis on the attainment of high biological yield. In a discussion of these relationships as pertaining to plant breeding, Donald (1962) questioned whether plant breeders had a sufficiently positive approach toward increased grain production. Though they bred for the realization of yield
363
BIOLOGICAL YIELD AND HARVEST INDEX
potential through such attributes as disease resistance or drought escape through earliness, they did not specifically seek to breed plants capable of greater photosynthesis or able to render a greater part of their biological yield as grain. The only advance in basic yield potential was by the selection of high yielding segregates without prior nomination of their characters. Measurement of biological yield and its relation to grain yield in breeding programmes was strongly advocated, and the term “harvest index” was proposed for the ratio of grain yield to biological yield, a term identical in meaning with the earlier coefficients, but without physiological or teleological overtones. Thus for each component of the equation for cereals there are a number of terms, including: Biological yield or Total yield or Yield of dry matter or Yield of grain + straw or Yield of total produce
Migration coefficient or Ratio of grain to total produce or X Coefficient of effectiveness or Harvest index
=
Grain yield
The terms adopted in this article are “biological yield,’’ “harvest index,” and “grain yield.” True biological yield includes the weight of roots, but since these are normally nonrecoverable, the term is usually applied, like its counterparts listed above, to the total weight of the tops. Harvest index, by definition, is a factor less than unity, say 0 to 0.55, but some workers prefer to use “harvest index (%),” in which they express the index as 0 to 55%. A term closely related to harvest index, and used by many cereal agronomists, is the grainlstraw ratio, or, for maize, the grainlstover ratio. Assuming that grain t straw equals biological yield, the grainlstraw ratio can easily be converted to harvest index or vice versa. We suggest, despite this easy conversion, that there is clear advantage in the use of the ratio:harvest index. First and foremost, harvest index links grain yield to the valuable measurement, biological yield; the determination of biological yield is a necessary step in the derivation of the harvest index. A second advantage is that its values tend to be linear rather than exponential in their relationship to grain yield. Thus if biological yield were constant, a situation approached in varying degree in many comparisons of genotypes, harvest indices of 0.2, 0.4, and 0.6 would indicate grain yields in those proportions, but the corresponding grainlstraw ratios would be 0.25,0.67, and 1.50. A further disability of ratios between grain and straw is, that though they are usually calculated as “grain-to-straw,” they are not infrequently expressed as “straw-to-grain,” a confusing inversion. The determination of harvest index, the term adopted in this paper, is usually based on a harvest at maturity. Mature plants are cut at ground level, weighed to give total yield, and threshed to give the yield of grain. An assessment based on air-dry material is usually adequate but if the harvested material is unusually wet or variable in water content, oven drying may be necessary.
364
C. M. DONALD AND J. HAMBLIN
Some imprecision may also result from the determination of harvest index at maturity. Because of leaf loss and respiration, the apparent biological yield of the crop may be less at maturity than at a somewhat earlier stage. For example, in a study of wheat at widely different densities, Puckridge and Donald (1967) recorded the following values: Density @lants/m’ at 26 wk): Biol. yield @/ma)at 20 wk (early dough stage): Biol. yield @ma) at 26 wk (maturity):
1.4 105 126
7 466 483
35 891 812
154 932 891
447 852 738
At the density close to normal crop density (154/m2), the decrease in crop weight was 4.6%. There is not much information on this aspect in relation to different genotypes at a common density, but if they lose different amounts of leaf before maturity, their biological yields will be understated to an unequal degree. In the following pages we discuss the meaning and value of the biological yield and harvest index of cereals in agronomic studies and in cereal breeding. It is our view that each of these measures of performance by the crop or the single plant can contribute importantly to the advancement of cereal productivity. The lack of interest by plant breeders and many agronomists in biological yields-in the total production by their plants and crops-has seriously limited the understanding of cereal performance and biotype behavior. Though biological yield and the linking ratio to grain yield, harvest index, are extremely simplified statements of multiple and complex growth processes, they nevertheless permit a far more analytical interpretation of environmental and genotypic influences than is possible from grain yields alone. It. The Relationship of Biological Yield, Grain Yield, and Harvest Index to Each Other and to Other Plant Characteristics
Various models and actual relationships between biological yields and grain yields within a series of genotypes or agronomic treatments are shown in Fig. 1. Figure 1 [Graph l(a)] depicts the situation in which a number of varieties all have precisely the same biological yield but different grain yields. In Fig. 1 [Graph l(b)] these genotypes are ranked in order of increasing grain yield. In these two figures Grain yield is proportional to harvest index and their correlation is 1.OO.On the other hand, biological yield and harvest index are unrelated. This situation may be closely approached in varietal trials. In a series of trials in New South Wales, a comparison was made of six Australian wheats and the
365
BIOLOGICAL YIELD AND HARVEST INDEX
German variety, Opal, all of normal height and two Mexican semi-dwarf wheats (Syme, 1970). The biological yield of the nine varieties showed the small range of 11,040 to 12,540 kg/ha, while the grain yields were from 2920 to 4890 kg/ha. The harvest indices ranged from 0.243 to 0.390; the correlation between grain yield and harvest index was 0.96, while that between biological yield and harvest index was 0.08. In most varietal comparisons the correlations between grain yield and harvest index is a good deal below 1.00 because biological yield is considerably more variable than in Syme’s study, but as the following values indicate, correlations of 0.6 or higher are commonly reported or can be calculated from published yield data.
con. YgJH.1. Finlay (private communication)
Thorne er u1. (1969) Hamblin (1971)
Singh and Stoskopf (1971) Nass (1973) Bhatia (1975)
Barley, 1962 Barley, 1963 Barley, 1964 Spring wheat Barley, at low N Barley, at high N Winter wheat Spring wheat Oats Spring wheat, 1971 Spring wheat, 1972 Wheat
0.73
Oh8 0.57 0.65 0.36 0.89 0.62 0.66 0.50 0.62 0.75 0.71
The contrary situation is that in which grain yields show dependence on biological yield. Figure 1 (Graph 2) shows this relationship in its full expression, whereby grain yield is strictly proportional to the biological yield. In this instance:
H.I.= K
and
Ygr= K. Ybiol
(2)
Thus grain yield has a correlation of 1.00 with biological yield. However, this positive correlation is not unique to the situation in Fig. 1 (Graph 2), but also applies in other situations in which grain yield rises with biological yield (Fig. 1, Graphs 3 and 4).
366
C. M. DONALD AND J. HAMBLIN
YIELD AND H.I.
Varieties ranked
Varieties unranked YbiolL K Ygr a H.I.
A principal component of most comparisons of cultivars
H.I. = K ygr a ybiol As Ybiol increases,
YF increases proportionally. Tendency of genotypes in mixtures
Correlation ygr/ybiol Ygr/H.I. YbiodH.1.
0 1 0
1 0 0
FIG. 1. Model of relationships between biological yield, grain yield, and harvest index. (Ybiol is shown and discussed as constant or as increasing from left to right; however, the graphs can also be considered in mirror image.)
We believe that a strong positive relationship of biological yield and grain yield may be characteristic of genotypes competing in mixtures, as in a segregating population. In a barley F3 at crop density, Hamblin (1971) recorded correlations of Yv/Ybbl of 0.98 at low N and 0.95 at high N. The correlations of Y,/H.I. were 0.09 and 0.33 and of Ybbl/H.I. -0.10 and 0.03. Taller, leafy plants were successful in this competitive situation. In all comparisons of cdtivars, grain yields must by definition (Ybiol X H.I. = Y,) depend wholly on differences in their biological yields and differences in their harvest indices. The correlations of grain yield and harvest index in the several studies listed on p. 365 conform to the view (without proof of it, as we emphasize later) that the dominant relationship in field plot comparisons of genotypes in pure strands (that is to say, in crop situations) is a constant
367
BIOLOGICAL YIELD AND HARVEST INDEX
YIELD AND H.I.
AS Ybiol increases, Y p increases more than proportionally
AS Ybiol increases, Ygr increases less than proportionally
AS Ybiol increases Y, decreases
Typical of responses to water
Typical of responses to N.
Typical of responses to N when water is deficient
Correlation ygr/ ybiol YgJH.1. YbiodH.1.
1 1 1
1 -1 -1
-1 1 -1
FIG. 1 (cont'd)
biological yield with a variable harvest index [Fig. 1, Graph l(b)] ,as contrasted from the relationship depicted in Fig. 1 (Graph 2). Furthermore, as discussed in Section IV,most of the progress in breeding high-yielding cereal cultivars seems to be related to higher harvest indices with little change in biological yield. The foregoing discussion perhaps serves to illustrate that the calculation of correlations between harvest index and its components may be an inadequate means of examining relationships among cultivars. We illustrate this further by returning to the situation in Fig. 1 (Graph l(a) and (b)] in which biological yield is constant. It is evident from inspection of these graphs, or by consideration of Eq. (l), that random numbers bsed as simulated grain yields would, like any set of real yields, give a correlation H.I./Y, of 1.00. But this does not invalidate the implications in plant breeding of a constant biological yield as an
368
C. M. DONALD AND J. HAMBLIN
important relationship among a group of genotypes. It is the constancy of biological yield which is the significant feature of this model. We can further examine the structural relationships between biological yield, harvest index, and grain yield by the use of two symbols, namely; a = yield of grain; b = yield of vegetative parts (all nongrain organs). Then, (a t b) = biological yield, and a/(a t b) =harvest index. The equation relating the terms is
Each of the terms may be derived from the other two. Usually harvest index is derived from measurements of biological yield and grain yield, but sometimes the biological yield may usefully be derived from published values of harvest index and grain yield. Table I shows correlation data for 60 cultivars of barley grown at Adelaide in 1962, 1963, and 1964 (data from K. W. Finlay). A model was then constructed in respect of each season, comprising 100 populations, each of 60 entries; each population had random a values (simulated grain yields) and random b values (simulated vegetative yields), restricted to give means, relative values of means, and variances equal to those of Finlay’s barleys for that season. The values of a and b were chosen independently and were ips0 fact0 uncorrelated. The simulated grain yields, a, biological yields, (a t b), and harvest indices, a/(a t b), were correlated within each population. The mean correlations for the 100 populations for each year are shown in Table I; it will be seen that the mean correlations derived from the random values are strongly related to those from the field study. Although the random values were subject to several significant restraints, including, for example, the degree of constancy of the biological yield, it is evident that, within any varietal comparison and even where grain yields and vegetative yields are uncorrelated, biological yield, grain yield, and harvest index will always show partially predictable mutual correlations because of the common terms in their formulas. Turner (1959) has examined the correlations between a ratio (W/B, kg wool/kg body weight of sheep) and its numerator (W, wool weight), and between the ratio and its denominator (B, body weight), in terms of the variances and covariances of its components. She showed that with phenotypic correlations between W and B of -0.4 to 0.4 and genetic correlations between W and B of -0.2 to 0.4, as in the wool and body weights of these Merino sheep, the genetic correlation between W and W/B was positive and usually over 0.5 in value, while the genetic correlation between B and W/B was negative and frequently greater than 0.5 in magnitude. Similar relationships emerge in Table 1, in which Y , and Y,, (i.e., a and b, equivalent to W and B in Turner’s study) show phenotypic correlations from -0.46 to 0.35. Harvest index, a/(a t b), is similar to Turner’s W/B,where W a .
369
BIOLOGICAL YIELD AND HARVEST INDEX TABLE I Correlations between the Components of Harvest Index for 60 Barley Varieties Grown at Adelaide in Three Successive Years, Together with the Mean Correlations in Each Season for 100 Model Populations Each of 60 Random Value@ Ygr
Correlation:
Ybiol
Ygr H.I.
Ybiol H.I.
ygr
Yveg
~~~
1962 Actual Randoma
-0.25 0.24
0.73*** 0.35
-0.81*** -0.46*** -0.77 0.02
0.68***
-0.36** -0.37
0.02 0.02
-0.33** -0.47
0.35** 0.02
1963 Actual Random
0.36** 0.35
0.55
1964 Actual Random
0.56*** 0.57*** 0.30 0.66
“Based on random values of Ygr and Yveg restricted to give means, relative values of means, and variances equal to those of the actual material in each season (see text). Note: Throughout this article * = P