AOBPreview originally published online on August 3, 2004
Annals of Botany 2004 94(3):385-392; doi:10.1093/aob/mch154
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Annals of Botany 94/3, © Annals of Botany Company 2004; all rights reserved
The Use of a Principal Axis Model to Examine Individual Plant Harvest Index in Four Grain Legumes
1 Plant Sciences Group, Agriculture and Life Sciences Division, PO Box 84, Lincoln University, Canterbury, New Zealand and 2 DNRE, Victorian Institute for Dryland Agriculture, PMB 260, Horsham, Victoria 3401, Australia
* For correspondence. E-mail moot{at}lincoln.ac.nz
Received: 20 August 2003 Returned for revision: 20 February 2004 Accepted: 14 May 2004 Published electronically: 3 August 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
• Background and Aims A principal axis model (PAM) has been proposed to enable the selection of crop ideotypes. The PAM enables plant-to-plant variability within crops to be quantified and compared. The aim of this paper is to validate the PAM for four grain legumes.
• Methods Four grain legumes (Cicer arietinum, Lens culinaris, Lupinus angustifolius, Pisum sativum) were used to quantify the influence of plant-to-plant variability on crop yields. To create variability, populations of 10, 100 and 400 plants m–2 were established ‘on-the-square’ with sowing depths of 2, 5 and 10 cm. Further, a central plant was treated with nitrogen and the impact of this on its four neighbouring plants was examined. Seeds were sown and plants harvested individually by hand.
• Key Results Mean individual plant seed weight (SWT) and plant weight (PWT) decreased as plant population increased but there was a consistent and strong (R2 > 0·90) linear relationship between SWT and PWT, with a negative SWT-axis intercept in all species. These components form the basis of the principal axis model (PAM). The PAM was used to summarize the performance of individual plants within a crop and quantify the variability caused by N treatment and the lowest and highest yielding individual plants. A negative SWT-axis intercept indicated that a minimum plant weight (MPW) was required for seed production and therefore the relationship between plant harvest index (PHI) and PWT was asymptotic. The heaviest MPW was calculated for plants grown at the lowest plant population and it was species-dependent, being higher in the larger seeded species.
• Conclusions Agronomic or physiological characteristics that lead to variability in PWT within a population will decrease PHI, and crop yield. The PAM may be useful in breeding programmes to identify plant phenotypes that minimize this plant-to-plant variability.
Key words: Chickpea (Cicer arietinum), field pea (Pisum sativum), individual plants, lentils (Lens culinaris), narrow-leafed lupin (Lupinus angustifolius), plant harvest index, plant weight, principal axis model (PAM)
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Under favourable conditions grain legumes can produce high seed yields (Heath and Hebblethwaite, 1985
; Loss and Siddique, 1997). However, average seed yields are generally low due to poor yield stability within and between sites and seasons (Heath and Hebblethwaite, 1985). For field peas, seed yield and crop harvest index (CHI) differences in favourable agronomic conditions have been associated with variability in plant harvest index (PHI) of individual plants within a crop (Ambrose and Hedley, 1984). This is in contrast to the situation reported for soybeans when Spaeth et al. (1984) concluded this variability had little effect on crop harvest index. To quantify plant-to-plant variability, Moot (1997) developed a principal axis model (PAM) that described the relationship between individual total above-ground plant (PWT) and seed (SWT) weights of all plants within a crop. This allowed a detailed examination of the total biological yield (TBY) and total seed yield (TSY) of the crop. The PAM consists of a principal axis effectively determined by a least squares linear regression relationship (eqn 1) and an ellipse to quantify the deviation from the central axis (Moot, 1997).
 |
When the PAM has a negative y-axis intercept (a < 0) then a ‘minimum plant weight’ (MPW) is calculated to quantify the minimum biological yield an individual plant requires to produce seed yield (Moot et al., 1997).
 |
), although Prihar and Stewart (1991) considered the MPW was purely a consequence of stressed plants. Confirmation of an MPW for legumes is important in understanding the consequences of inter-plant competition on crop yields. The importance of an MPW is highlighted by the subsequent relationship between PHI and PWT, which is then asymptotic. The small plants in any population have the lowest PHI. These plants therefore utilize resources but produce a proportionately low SWT. Moot (1993) hypothesized that as environmental or plant population constraints reduce the available resources, the proportion of small plants in a crop will increase, CHI and seed yield will decrease and PHI will become more variable. An increase in inter-plant competition can be quantified by an increase in the skewness (g1) of the individual plant frequency distribution and a move away from the normal distribution. The aim of the current research was to test whether the PAM could be extended from field peas to describe plant populations of other grain legumes with different growth habits, namely chickpea, lentil and narrow-leafed lupin. This included an experiment to create inter-plant competition within these crops, using different plant populations, sowing depths, and N application to examine whether the effect on plant-to-plant variability in SWT, PWT and PHI could be quantified using the PAM. |
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
The experiment was set up at Lincoln University, Canterbury, New Zealand (43°38'S) in 1999/2000 as a split–split–split plot design. The four grain legumes (chickpea, lentil, narrow-leafed lupin and field pea) were the main plots, with three plant populations (10, 100 and 400 plants m–2) as sub-plots and three sowing depths (2, 5 and 10 cm) as sub-sub-plots. There were three replicates. Plots were hand sown, ‘on-the-square’ at equidistant spacings of 31·5 x 31·5 cm, 10 x 10 cm and 5 x 5 cm and in sub-sub-plots of 5·67 x 3·15 m, 1·5 x 2·0 m and 0·8 x 0·2 m for the low, medium and high populations, respectively. To achieve this each replicate was sown over two days between 19 and 24 Oct. 1999. Holes were made with a 10-mm-diameter dibber and seed was sown before each hole was back-filled with soil and the surface firmed. Additional details of crop management can be found in Ayaz (2001)
. Thirty-five days after sowing (DAS) a further split in the design was established to enhance inter-plant competition. Nitrogen (N, as urea, 46 % N) was added to five tagged plants (Nx) at a rate equivalent to 100 kg N ha–1 in each sub-sub-plot. For the applications, the urea (32·6 g) was dissolved in 3000 ml of water. A 30-ml syringe was used to apply N around the plants.
The immediate neighbour to the north (
), south (
), east (Ex) and west (Wx) of each plant was tagged to investigate the influence of the central N-treated plant on the SWT, PWT and PHI of its nearest neighbours. This gave a total of 2700 individual plants for assessment. A further 140 plants, free of any N effect, were used for comparison as control plants. To further examine the influence of plant position on plant growth, the ratio of the final dry weight of the neighbours to the treated plant was calculated for all variables. On this basis values <1·0 represent a proportional decrease for the untreated neighbouring plants or controls.
Measurements
Plants were individually harvested at crop maturity, defined as when >90 % leaves had senesced in any species. In most cases all leaves were still attached to the stem. Where leaves had abscised from the lowest two or three nodes of plants (
5 %) they were gathered from the ground, weighed and assigned as a mean weight to the harvested area to enable the estimation of PHI. Twenty-five individual plants per sub-sub-plot were hand harvested by cutting to ground level and then oven dried to constant weight. Plants from sub-sub-plots were harvested on the same day. Any physically damaged or diseased plants (<1 %) were excluded from the data analysis. For each plant the individual PWT and SWT were measured and PHI calculated before values were summed to give the total biological yield (TBY), total seed yield (TSY) and CHI for each crop.
Statistical analysis
The four-way interaction term was added to the random effects error term of the full split–split–split plot analysis of variance (ANOVA). For all measured variables (SWT, PWT, PHI) the analysis showed that sowing depth, as a main effect, and in the two- and three-way interactions was not significant. Results were therefore omitted from interaction tables. However, several two-way interactions, which involved legume species, were significant. Results from these interactions are reported but they were secondary to the main aim of using plant population and position to generate variability for testing the PAM. Further, the validity of concentrating on species interactions is questionable because of their morphological differences and an a priori knowledge that optimum plant populations differed among species. The PAM involved fitting least squares regression to the SWT against PWT data, and calculating the MPW from the resulting coefficients (eqn 2).
In addition, the normality and skewness (g1) of frequency distributions of populations were examined for each crop. All variates were analysed using the SYSTAT (SYSTAT 9 for windows) or the Genstat statistical packages.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Yield and crop harvest index
The interactions between the four legume species and the three plant populations for TBY, TSY and CHI are summarized in Table 1. As expected, the highest TBY occurred at the highest population and this also gave the highest TSY for all species. These trends were also reflected in the CHI (Table 1) and as plant population increased, CHI increased more (P < 0·05) for chickpea than narrow-leafed lupin.
|
Mean plant weight, seed weight and plant harvest index
A summary of the main effects of factors on PWT, SWT and PHI is presented in Table 2. The range in mean PHI was 0·408 to 0·463 for species but it was most affected by plant position. The N-treated plants were heaviest and had the highest SWT and PHI. Plants to the north of this central plant were least affected by the treated plant but those to the south always had the lowest PWT, SWT and a mean PHI approx. 30 % lower than the central plant.
|
The N-treated plants for all species and populations were also heavier (P < 0·05) (Table 3) than the control plants but this did not necessarily translate into a higher PHI (Table 4). For the southern plants the decrease in growth ratio for PWT and PHI was accentuated by the increase in plant population from 10 to 100 plants m–2 (Tables 3 and 4).
|
|
PAM analyses
The regression between SWT and PWT was positive, linear and had an R2
0·92 for all species by population combinations (Figs 1–4). All 12 models produced a negative SWT-axis intercept and as a consequence an MPW was calculated for each model (Table 5). The MPW changed with population, being lowest at the highest population for each species and ranged from 0·40 g plant–1 for lentil at 400 plants m–2 to 16·38 g plant–1 for lupin at 10 plants m–2.
|
|
|
|
|
The relationship between PHI and PWT was asymptotic with the lowest PHI values from the smallest plants in each population. This asymptotic trend was strongest at 400 plants m–2 for each species and there was a clear indication that these smallest, low PHI plants were disproportionately located to the south of the treated plant in each crop. Conversely, the larger plants with a high PHI were generally N-treated (Figs 1
–4).
Frequency distributions
In all species PWT and SWT were normally distributed at 10 plants m–2 but significantly positively skewed with a consequent increase in the CV at 400 plants m–2 (Table 6). In contrast, the PHI distribution was increasingly negatively skewed as population increased. This increased variability across populations was shown by an increase in the coefficient of variation (CV) of at least 10 % from 10 to 400 plants m–2 (Table 6). This means there were a few plants with a low PHI including some barren plants (PHI = 0) at higher populations.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Yield, mean PWT and SWT
From an agronomic perspective, TBY, TSY and CHI were all increased with increased plant population. This was explained by earlier canopy closure giving a higher green area index (GAI) and greater radiation absorption at high populations (Ayaz, 2001
). A consequence of the increased plant population was the decreased mean PWT (Table 2) that was consistent with previous results for field peas (Moot, 1997). This is consistent with an increase in the amount of random leaf overlapping and penumbral shading among neighbouring plants that occurs with increased GAI at higher plant populations. Plant-to-plant variability in DM accumulation per plant increased with leaf expansion, as shown by the increased coefficient of variation for PWT as plant population increased for each species (Table 6).
Classically, seed yield per unit area usually also increases with increased plant population to a plateau and may eventually decline (Holliday 1960a, b), while yield per plant decreases (Table 2). In this study, lower seed yield was associated with fewer pods per plant, and therefore fewer seeds per plant (Ayaz et al., 2001). The challenge for crop management, particularly in grain legumes, is to maximize DM accumulation without compromising crop seed yield through excessive plant-to-plant competition producing low yielding or barren plants (Hedley and Ambrose, 1985; Moot et al., 1997). Effectively, the increasing portion at the beginning of each asymptotic relationship in Figs 1–4 represents these plants with a low PHI. The difference between their individual PHI and the asymptote represents the potential yield lost through inter-plant competition.
In this study this was accentuated deliberately by the central N-treated plant that was heavier than its surrounding neighbours (Table 2). This was designed to represent an early emerging or dominant plant that had an advantage over its neighbours in resource capture. Results indicate that this treated plant probably suppressed the southern plant by disproportionate resource capture (e.g. light, space) and the severity of competition, which increased with plant population. This produced the approx. 30 % reduction in PWT in the southerly plants (Table 3). This reduced the SWT more than in the other neighbouring plants by reducing the number of branches (Ayaz, 2001) on which seeds were borne. These results support the hypothesis of Weiner et al. (1990) that competition among plants is asymmetric where larger plants are able to capture resources and suppress the growth of smaller plants. It also confirms the assumption of Moot (1993) that many small plants in a crop can compromise potential crop seed yield.
Plant harvest index
The effects of inter-plant competition were seen as variation in PHI among the different plant positions that increased with population (Table 4). The PHI is reported to be sensitive to management and environmental factors, and also differs considerably among species and cultivars (Snyder and Carlson, 1984). Variable partitioning of DM to seed, as quantified by PHI, has been identified as an important contributor to yield variability in grain legumes (Ambrose and Hedley, 1984; Moot and McNeil, 1995).
The PAM analyses: relationship between SWT and PWT
Having established that variation in PWT, SWT and PHI was induced by treatments, the PAM can be used to quantify differences among treatments. The PAM analyses indicated that, for each species and at each population, there was a strong linear relationship between SWT and PWT (Figs 1–4). The stability of this relationship, despite the 40-fold change in plant population and the contrasting growth habits of these four species, supports its use as the basis for the PAM. The implication of this relationship for describing PHI distributions and therefore crop seed yields depends on the position of the SWT-axis intercept (Moot, 1997).
The negative SWT-axis intercept for all treatment combinations enables an MPW to be estimated from the PAM (Table 5). The highest calculated MPW for each species was for plants sown at 10 plants m–2. Moot et al. (1997) reported a similar result and showed this resulted from an increased number of barren branches and partitioning to structural dry matter at low populations, which emphasized the occurrence of intra-plant competition. Further, species that had larger seeds also had higher MPWs.
The position of the SWT-axis intercept and therefore existence of an MPW is controversial (Moot, 1993). Gardner and Gardner (1983) and Moot (1997) considered the intercept was species dependent, but was generally negative, and was influenced by environmental factors. Prihar and Stewart (1991) proposed that negative intercepts were an artefact of the inclusion of stressed plants in the analysis of the relationship between SWT and PWT. They also concluded that an MPW was not species dependent.
The widely spaced plants (10 plants m–2) in this experiment were sown to produce plants in an environment which was free of stress from agronomic factors and inter-plant competition. The conclusions of Prihar and Stewart (1991) would lead to the expectation that the SWT-axis intercepts from this treatment should be zero or positive. However, the opposite result was observed. The largest negative intercepts and MPW-values were calculated from this treatment (Table 5).
The fact that all treatments gave negative SWT-axis intercepts also supports an MPW for these four grain legumes and that MPW-values are higher in large seeded species (Moot, 1997). It seems likely that the species dependent MPW was accentuated by stress, as shown by the disproportionate number of shaded southerly located plants near the PWT-axis.
By definition a negative SWT-axis intercept, and thus MPW, produces an asymptotic relationship between PHI and PWT (Moot, 1997; Moot et al., 1997). The location and dispersion of the frequency distribution for PWT were then dominant factors, which determine the effect of the MPW on the dispersion of PHI and consequently the implications for seed yield of a crop (Moot, 1997).
Plant harvest index and PWT
The indication from plants at 10 plants m–2 was that the PHI was close to the asymptote. This can be interpreted as the genetic maxima, as shown for pea genotypes by Moot et al. (1997). In contrast, there was increased plant-to-plant variability of PHI at the two highest plant populations as indicated by the higher CVs (Table 6), and a more defined asymptotic relationship (Figs 1–4). At 400 plants m–2, >95 % of plants had a high PHI but the smallest plants had the lowest PHIs. This trend has previously been reported in field peas (Hedley and Ambrose, 1985; Moot, 1993), and is consistent with analyses that include an MPW (Moot, 1997; Moot et al., 1997).
The frequency distribution results for PHI indicated that variability increased with plant population and as mean PWT decreased. The PHI distributions were all negatively skewed but as plant population increased the CVs increased (Table 6). The results indicate that at populations of 10 plants m–2 the range of PWT values was beyond the initial linear phase of the asymptotic relationship and as a consequence variability in PHI was low. At high populations, and with the competition intensified due to the presence of the central-N treated plant, the asymptotic relationship was apparent (Figs 1–4).
If the lowest mean PHI of 0·345 for a southerly plant was increased to the average PHI of 0·464 (Table 2), the overall CHI (=
PHI) would increase and the TSY would be increased by approx. 5 % for the same TBY (=PWT). The implication from these results is that any factors that induce plant-to-plant variability are also likely to decrease potential crop yield. To achieve higher yields, crop ideotypes have been proposed as a basic approach to plant breeding. Donald (1962) suggested that plant breeders should consider HI and morphological characters, which influence the photosynthetic ability of a plant and the effects of competition in the evaluation of early generation materials. Ambrose and Hedley (1984) believed that vigorous and highly competitive plants induced variation in PHI when grown in crop communities with a few dominant plants contributing the majority of the seed yield (e.g. as with N-treated plants in this work). Less competitive plants (e.g. southerly plants) with low PHI values would contribute the least, and consequently the CHI would be low (Figs 1–4). This contradicts the conclusion of Spaeth et al. (1984) who indicated these small plants only contributed a small amount to the yield potential of the crop. However, the existence of a predisposition for inter-plant competition may not be expressed fully until resources become limiting. In these circumstances the yields of grain legumes have been shown to collapse (Heath and Hebblethwaite, 1985).
Thus, to produce stable, high yields, most individual plants in a community should be ‘weak competitors’ as proposed by Donald and Hamblin (1976); i.e. weakly competitive with neighbours but maximizing utilization of resources within their immediate space. For field peas, Hedley and Ambrose (1985) defined such an ideotype but offered no quantitative method for selection in a breeding programme. Similarly, Moot et al. (1997) showed that conventionally leafed vigorous pea seedlings produced greater variability than semi-leafless low-vigour genotypes. They proposed a system of selection, based on the PAM, to identify appropriate genotypes in early generations of plant breeding programmes that accounted for the MPW. Results from the present study suggest a similar ideotype approach should be adopted for chickpea, lentil and narrow-leafed lupin programmes and the PAM may assist selection.
Ultimately, crop management or the selection of morphological traits that minimize variability and the number of small plants in a crop, should lead to an increase in crop yield and stability. Ideally this hypothesis should be tested in a grain legume-breeding programme with isogenic lines that differ only in morphological traits that affect their competitiveness as seedlings.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
The New Zealand Vice-Chancellors' Committee provided financial support for Shaukat Ayaz.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSLITERATURE CITED |
|---|
-
Ambrose MJ, Hedley CL. 1984. A population study to aid the selection of improved dried pea (Pisum sativum) crop plants. Annals of Botany 53: 655–662.
Ayaz S. 2001. Variability of harvest index in four grain legume species. PhD Thesis, Lincoln University, Canterbury, New Zealand.
Ayaz S, McNeil DL, McKenzie BA, Hill GD. 2001. Population and sowing depth effects on yield components of grain legumes. Proceedings of the Australian Agronomy Conference 10: 342–346.
Donald CM. 1962. In search of yield. Journal of the Australian Institute of Agricultural Science 28: 171–178.
Donald CM, Hamblin J. 1976. The biological yield and harvest index of cereals as agronomic and plant breeding criteria. Advances in Agronomy 28: 367–405.
Gardner WR, Gardner HR. 1983. Principals of water management under drought conditions. Agricultural Water Management 7: 143–155.
Heath MC, Hebblethwaite PD. 1985. Solar radiation interception by leafless, semi-leafless and leafed peas (Pisum sativum) under contrasting field conditions. Journal of Applied Biology 107: 309–318.
Hedley CL, Ambrose MJ. 1985. The application of plant physiology to the development of dried pea crop plants. In: Hebblethwaite PD, Heath MC, Dawkins TCK, eds. The pea crop—a basis for improvement. London: Butterworths, 95–104.
Holliday R. 1960a. Plant population and crop yield. Part I. Field Crop Abstracts 13: 159–167.
Holliday R. 1960b. Plant population and crop yield. Part II. Yield and plant population in British crops. Field Crop Abstracts 13: 247–254.
Loss SP, Siddique KHM. 1997. Adaptation of faba bean (Vicia faba L.) to dryland Mediterranean-type environments. I. Seed yield and yield components. Field Crops Research 52: 17–28.
Moot DJ. 1993. Harvest index variability within and between field pea (Pisum sativum L.) crops. PhD Thesis, Lincoln University, Canterbury, New Zealand.
Moot DJ. 1997. Theoretical analysis of yield of field pea (Pisum sativum L.) crop using frequency distributions for individual plant performance. Annals of Botany 79: 429–437.
Moot DJ, McNeil DL. 1995. Yield components, harvest index and plant type in relation to yield differences in field pea genotypes. Euphytica 86: 31–40.[CrossRef]
Moot DJ, Wilson DR, McNeil DL. 1997. Validation of the principal axis model (PAM) and its application to genotype selection in field pea (Pisum sativum) crops. Annals of Botany 79: 651–656.
Prihar SS, Stewart BA. 1991. Sorghum harvest index in relation to plant size, environment and cultivar. Agronomy Journal 83: 603–608.
Snyder FW, Carlson CM. 1984. Selecting for partitioning of photosynthetic products in crops. Advances in Agronomy 37: 47–72.
Spaeth SC, Randall HC, Sinclair TR, Vendeland JS. 1984. Stability of soybean harvest index. Agronomy Journal 76: 482–486.
Weiner J, Mallory EB, Kennedy C. 1990. Growth and variability in crowded and uncrowded populations of dwarf marigolds (Tagetes patula). Annals of Botany 65: 513–524.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||