- Email: [email protected]
Reproductive Allometry in Soybean, Maize and Sunflower
Annals of Botany 85: 461±468, 2000 doi:10.1006/anbo.1999.1084, available online at http://www.idealibrary.com on
Reproductive Allometry in Soybean, Maize and Sun¯ower C . R . C . V E G A * , V. O . S A D R A S , F. H . A N D R A D E and S . A . U H A R T INTA BalcarceÐUniversidad de Mar del Plata, CC 276, 7620 Balcarce, Argentina Received: 5 July 1999 Returned for revision: 5 October 1999 Accepted: 7 December 1999 We compared the relationship between grain yield per plant (YP) and shoot biomass per plant (SP) in three annual crops with contrasting reproductive strategies: sun¯ower, a determinate species with a single in¯orescence; maize, a determinate species with a limited capacity to adjust the number of ears in response to resource availability; and indeterminate soybean, a species with a large capacity to adjust the number of in¯orescences. Our working hypotheses were: H1 Ðthe relationship between YP and SP is linear; H2 Ðthe intercept of the model is zero, i.e. there is not a threshold plant mass for reproduction. A wide range of YP and SP was generated by manipulation of plant density; SP varied between 0.3 and 196 g per plant in soybean, between 6 and 873 g per plant in sun¯ower and between 23 and 697 g per plant in maize. Within these broad ranges of plant size, both hypotheses were rejected in ®ve out of six experiments, i.e. the relationship between YP and SP departed from linearity and there was a threshold for SP below which no grain set occurred. The SP threshold for grain set varied widely among species; it was close to 2 g per plant for soybean, 27 g per plant for sun¯ower and 43±71 g per plant for maize. Because of this size threshold and non-linearity, harvest index (HI YPSP ÿ1) was stable for mid-size plants, diminished slightly for large plants, and diminished sharply for smaller plants in all three crops. Harvest index stability was highest in soybean, intermediate in sun¯ower and lowest in maize. Dierential stability of reproductive partitioning partially derived from contrasting # 2000 Annals of Botany Company patterns of meristem allocation. Key words: Helianthus annuus L., Zea mays L., Glycine max (L.) Merrill, grain yield, harvest index, plant density, reproductive allocation, meristem allocation, plasticity.
I N T RO D U C T I O N Biomass accumulation and partitioning to reproductive structures are critical aspects of plant ®tness (Solbrig and Solbrig, 1985; Reekie and Bazzaz, 1987; Weiner, 1988; Hartnett, 1990). They are also key determinants of crop yield (Giord et al., 1984; Andrade et al., 1999). In many plant species, dry matter partitioning to reproductive organs is a stable and highly heritable trait (Spaeth et al., 1984; Johnson et al., 1985; Hay, 1995; Sadras, Bange and Milroy, 1997). The mechanisms underlying reproductive partitioning are, however, poorly understood (Evans, 1994). Linear relationships between reproductive and shoot biomass are often used to describe reproductive partitioning and grain yield (Gardner and Gardner, 1983; Samson and Werk, 1986; Sinclair, Bennett and Muchow, 1990; Prihar and Stewart, 1991; Moot, Wilson and McNeil, 1997). There is no agreement, however, about the x-intercept of this linear model, i.e. the minimum shoot biomass required for grain set. Gardner and Gardner (1983), Samson and Werk (1986) and Weiner (1988) argued that plants have a threshold size to produce ¯ower and fruit. Empirical evidence to support the existence of such a threshold has been found in a number of studies (Gardner and Gardner, 1983; Weiner, 1988; Hartnett, 1990; Thompson, Weiner and Warwick, 1991; Moot et al., 1997). In others, it has been suggested that a threshold plant mass is not a condition for reproduction (Rees and Crawley, 1989; Prihar and Stewart, 1991). This apparent disagreement * For correspondence. E-mail [email protected]
0305-7364/00/040461+08 $35.00/00
could be caused by a number of factors, including range of plant and grain biomass, the source of variation of plant size (e.g. water stress, nutrient de®cit), growth habit (determinate vs. indeterminate), life history (e.g. perennial vs. annual) and agronomic selection (wild vs. cultivated). We analysed the relationship between grain yield and shoot biomass and the eect of variation in plant biomass on the stability of dry matter partitioning to grain. Our null hypotheses were: H1 Ðthe relationship between grain yield and shoot biomass is linear; and H2 Ðthe intercept of the model is zero, i.e. a threshold plant mass is not a condition for reproduction. Our approach had three main components. Firstly, we compared three cultivated, annual species with contrasting reproductive strategies: sun¯ower (Helianthus annuus L.), a determinate species with a single in¯orescence; maize (Zea mays L.), a determinate species with a limited capacity to adjust its number of ears in response to availability of resources; and indeterminate soybean (Glycine max L. Merrill), a species with a large capacity to adjust the number of in¯orescences. Secondly, we generated a wide range of availability of resources by manipulation of plant population density. Thirdly, to account for plant-to-plant variation, we measured grain yield and shoot biomass of individual plants. M AT E R I A L S A N D M E T H O D S Crops and treatments Crops were grown on deep (51.5 m) Typical Argiudolls at Balcarce (378450 S, 588180 W; altitude 130 m), Argentina, # 2000 Annals of Botany Company
462
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower T A B L E 1. Summary of treatments in ®eld experiments at Balcarce Plant population ( plants m ÿ2)
Number of plants sampled
Species
Cultivar
Season
Date of emergence
Maize
DK 636 DK 639
1994±95 1997±98
17 Oct. 1994 2 Nov. 1997
2.2, 8.5 and 16.9 2.0, 3.6, 8.4, 13.0, 15.9 and 20.4
298 287
Sun¯ower
DK G-100 DK G-100
1994±95 1995±96
27 Oct. 1994 31 Oct. 1995
1.4, 5.8 and 10.3 1.4, 1.8, 2.2, 2.9, 4.1, 5.8 and 11.6
258 60
Soybean
Asgrow 3127 Asgrow 3205
1994±95 1997±98
13 Nov. 1994 14 Nov. 1997
7.9, 29.8 and 56.5 3.6, 7.1, 26.5 and 51.4
322 77
between 1994 and 1998 (Table 1). All three crops were fertilized with 35 kg P ha ÿ1 before sowing, and with 150 kg N ha ÿ1 at the V6 stage in maize (Ritchie and Hanway, 1982) and V4 in sun¯ower (Schneiter and Miller, 1981). Soybean seed was inoculated with Bradyrhizobium japonicum. Crops were irrigated to keep water content above 50% of maximum soil available water. Weeds and insects were adequately controlled. Each species was grown in a dierent section of the same ®eld. Plant density treatments were laid out within each crop species in a block design with three (1994±95 and 1995±96) or four replications (1997±98). Distance between rows was 0.7 m in all three crops and plant densities ranged from 3.6 to 56.5 plants m ÿ2 in soybean, 1.4 to 11.6 plants m ÿ2 in sun¯ower and 2.0 to 20.4 plants m ÿ2 in maize (Table 1). Maize and sun¯ower were sown by hand at the target density (Table 1) placing three seeds per hill and thinning to one plant per hill at stage V2 (Ritchie and Hanway, 1982) and V4±5 (Schneiter and Miller, 1981), respectively. Heavily sown soybean was thinned to achieve the appropriate density at stage V2 (Fehr and Cavinness, 1977). At physiological maturity, individual plants were harvested to determine shoot dry matter (SP) and grain yield (YP). In soybean, no attempt was made to recover fallen leaves. Plants were sampled randomly within each treatment and no plants were excluded except for some soybean individuals with heavy, broken branches at the lowest density. Sampling strategies varied between seasons. In the ®rst season, three plant population densities were established and large plant samples were taken to exploit variation within treatments as well as variation among treatments. In the second, we increased the number of treatments to four±seven and took fewer plants per treatment (Table 1). Analytical approach Despite some drawbacks (Samson and Werk, 1986; Klinkhamer et al., 1992) the relationship between SP and YP is an appropriate framework to investigate the pattern of reproductive allocation (Prihar and Stewart, 1991; Sadras et al., 1997). Furthermore, these two variables bring forth the variable harvest index, i.e. HI YPSP ÿ1, a concept widely used in agronomic and plant breeding research (Donald and Hamblin, 1976; Hay, 1995; Sinclair, 1998). Thus, our study is largely based on the relationships
between YP and SP , and between HI and SP . Additionally, we investigated the relationship between YP and vegetative shoot dry matter. A number of methods have been proposed for the analysis of size-dependent reproductive eort in plants (Klinkhamer et al., 1992; Schmid et al., 1994). In the nonlinear model of Klinkhamer et al. (1992) data points below the estimated threshold must be omitted from the analysis (Schmid et al., 1994). To deal with this problem and to properly analyse discontinuous relationships, Schmid et al. (1994) proposed a censored allometric model. Our approach in this study is based on: (1) no exclusion of data points with zero yield; (2) simple, biologically meaningful models; and (3) sequential analysis of YP vs. SP and HI vs. SP relationships. To test our hypotheses, we ®tted linear and non-linear models to the data, i.e. YP b1 SP
Model 1
YP a2 b2 SP
Model 2
YP
a3 SP 1 b3 SP
Model 3
YP
a4A SP ÿ ST 1 b4A SP ÿ ST
Model 4A
YP a4B
b4B SP
Model 4B
The implications of these models are given in Table 2. The linear models (1 and 2) are the simplest and more often used models to describe the relationship between YP and SP . Departure from linearity can be tested through regression of log YP on log SP (Klinkhamer, De Jong and Meelis, 1990; Thompson et al., 1991). However, as this test can produce misleading results when the y-intercept diers from zero, polynomial regressions were preferred, as in Thompson et al. (1991). Hypothesis 1 was rejected when second or third-order terms in polynomials were signi®cant (P 5 0.05). Models 3 and 4 are hyperbolic functions without and with x-intercepts, respectively; hyperbolic models have been widely used in studies of reproductive growth in maize (Edmeades and Daynard, 1979; Tollenaar, McCullough and Dwyer, 1994). In Model 4A, the parameter ST estimates the SP threshold for grain yield. Signi®cance of ST was used to test hypothesis 2. For
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower T A B L E 2. Generic models to describe the relationship between grain yield (YP) and shoot dry matter (SP) and consequences derived from each model for the relationship between harvest index (HI) and SP Model type
Slope
Intercept
Consequences for HI
1 2
Constant Constant
Zero Negative
3 4
Variable Variable
Zero Negative
HI constant If SP 4 ST then HI 0 If SP 4 ST then HI f(SP) HI f(SPÿ1) If SP 4 ST then HI 0 If ST 5 SP 5 SM then HI f(SP) If SP 4 SM then HI f(SP ÿ1)
ST is the SP threshold for grain yield, SM is the SP corresponding to maximum HI.
comparisons among species, ST and its standard error were expressed as a percentage of: (1) average SP at commercial density, and (2) maximum SP observed in the experiments. Model 4B was used to describe the relationship between YP and SP in proli®c maize plants. Each of the four models described has important implications for the relationship between HI and SP . Depending on whether the slope is constant or variable, and whether the intercept is zero or negative, the expected relationship between HI and SP is outlined in Table 2. As a further test of the modelsÐand therefore of our working hypothesesÐ we compared the actual relationship between HI and SP with the predictions resulting from each model.
463
Contrasting availability of resources per plant together with the use of individuals, rather than averages, contributed to the wide range of both YP and SP (Fig. 1). Shoot dry matter varied between 0.3 and 196 g per plant in soybean, between 6 and 873 g per plant in sun¯ower and between 23 and 698 g per plant in maize. Statistically, model 2 described the data very well (Table 3). Signi®cance of the second-order term in polynomial regressions, however, revealed departures from linearity in ®ve out of six cases (P 5 0.0001). Soybean in 1997±98 was the only exception. Several non-linear functions adequately ®tted the data for the three crops. Of them, we chose hyperbolic functions (Model 4) because they are both statistically sound and biologically meaningful as they include parameters that re¯ect SP thresholds for grain yield ( parameter ST , Table 4). The threshold of shoot dry matter required for grain set and yield was close to 2 g per plant in soybean (P 0.06), 27 g per plant in sun¯ower (P 5 0.0001) and 43±71 g per plant in maize (P 5 0.0001). To account for dierences in plant size among species (Fig. 1) thresholds were also calculated as a percentage of: (1) average shoot dry matter at commercial plant population density; and (2) maximum SP (Table 4). Normalised thresholds ranked soybean 5 sun¯ower 5 maize (Table 4). The large variation in thresholds in maize can be attributed, at least in part, to the use of dierent hybrids in both experiments (Gardner and Gardner, 1983). It is worthwhile noting that the parameter b4a was not dierent from zero for soybean in 1997±98, indicating that the relationship was adequately described by a linear model (Table 4). Some maize individuals were proli®c; i.e. set grain in a second ear. Hence, a second non-linear relationship was ®t to the data for proli®c plants (Fig. 1; Table 5).
R E S U LT S Relationship between YP and SP
Relationship between HI and SP
Figure 1 shows the relationship between YP and SP in individuals of soybean, sun¯ower and maize at physiological maturity. Yield and shoot biomass are not strictly independent variables (Charles-Edwards, 1982). The relationships between YP and vegetative shoot biomass per plant for all three crops (not shown), however, were similar to those found for YP and SP (Fig. 1).
In all three species, HI was stable for mid-size plants, diminished slightly for large plants, and diminished sharply for smaller plants (Fig. 2). The only exception to this pattern was soybean in 1997±98 in which HI did not decline with the largest SP (Fig. 2). The analysis of individuals allowed for the detection of a continuous variation in HI from maximum values to zero.
T A B L E 3. Parameters (+s.e.) of the linear regression (model 2) between grain yield (YP) and shoot dry matter (SP) for individuals of maize, sun¯ower and soybean Species
Season
a2
b2
R2{
N{
Maize
1994±95 1997±98
ÿ15.04 + 2.784 ÿ13.87 + 1.448
0.52 + 0.008 0.57 + 0.006
0.94 0.97
298 287
Sun¯ower
1994±95 1995±96
ÿ1.87 + 1.017 ÿ0.52 + 2.954
0.36 + 0.003 0.43 + 0.010
0.98 0.97
258 60
Soybean
1994±95 1997±98
0.78 + 0.360 ÿ1.00 + 0.362
0.49 + 0.006 0.60 + 0.009
0.95 0.98
322 77
Parameters are: intercept (a2), and slope (b2). All regressions were signi®cant at P 5 0.0001. { Coecient of determination. { Number of individuals.
464
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower 400 Maize 199798
Maize 199495 300 200 100 0
0
Grain yield per plant (g)
400
200
400
600
800
0
200
Sunflower 199495
400
600
800
Sunflower 199596
300 200 100 0
0
100
250
500
750
1000
0
250
500
750
1000
150
200
Soybean 199798
Soybean 199495
75 50 25 0 0
50
100
150
200
0
50
100
Shoot biomass per plant (g) FIG. 1. Relationship between grain yield per plant (YP) and shoot biomass per plant (SP) at physiological maturity in maize, sun¯ower and soybean grown under a wide range of plant population densities. In 1994±95, plant population densities were: low (j), intermediate (s) and high (m) as described in Table 1. For maize, (h) indicates proli®c plants.
Maximum HI was close to 0.7 in soybean, 0.43±0.47 in sun¯ower and 0.55±0.63 in maize. At high values of SP , soybean had the smallest reduction in HI and maize the greatest. In maize, only proli®c plants had HI close to the maximum (Fig. 2). The analysis in Fig. 2 (cf. SSi) further supports the nonlinearity of the relationship between YP and SP , as well as the existence of a non-zero threshold of shoot biomass, i.e. model 4 predicted HI better than models 1, 2 and 3. DISCUSSION We used the relationship between YP and SP and between HI and SP to investigate dry matter partitioning to grain in species with contrasting reproductive strategies (Figs 1 and
2). Linear models between YP and SP were statistically adequate to describe reproductive partitioning (Table 3). The relationship between these variables, however, departed from linearity signi®cantly at both ends of resource availability and plant size (Figs 1 and 2). Plant population density was the main source of variation in plant size; limiting individual plant biomass by other means, e.g. by manipulation of water supply, could give dierent relations between YP , SP and HI (Prihar and Stewart, 1991; Sadras and Connor, 1991). Weiner and Thomas (1992) pointed out the disagreement between static, inter-individual allometry at one point in time, and dynamic patterns of allometric growth for competing plants. Accordingly, the static patterns of reproductive allocation in our study should not be taken
Normalized ST Species
Season
a4A
ST (g per plant)
% of average SP at commercial plant population density}
% of maximum SP
b4A
R2 {
N{
Maize*
1994±95 1997±98
0.9 + 0.03 0.7 + 0.03
70.8 + 2.62 42.7 + 2.94
23.58 + 0.009 15.55 + 0.011
10.15 + 0.004 6.92 + 0.005
0.0019 + 0.0002 0.0009 + 0.0001
0.95 0.97
255 273
Sun¯ower
1994±95 1995±96
0.5 + 0.01 0.6 + 0.04
25.4 + 3.02 28.9 + 7.80
14.80 + 0.018 17.87 + 0.048
2.90 + 0.003 4.81 + 0.013
0.0004 + 0.0001 0.0005 + 0.0002
0.98 0.98
258 60
Soybean
1994±95 1997±98
0.6 + 0.02 0.6 + 0.03
1.5 + 0.79 1.6 + 0.83
5.73 + 0.030 6.54 + 0.034
0.77 + 0.004 1.35 + 0.007
0.0014 + 0.0003 0.0001 + 0.0001
0.95 0.99
322 77
Parameters are: a4A (initial slope), ST (SP threshold for grain yield) and b4A (slope at high values of SP). Normalized thresholds are also shown for comparison among species. All regressions were signi®cant at P 5 0.0001. { Coecient of determination. { Number of individuals. } Maize: 8.4±8.5 plants m ÿ2, sun¯ower: 5.8 plants m ÿ2, soybean: 26.5±29.8 plants m ÿ2. * Model ®tted to plants with one ear.
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower
T A B L E 4. Parameters (+s.e.) of model 4A ®tted to grain yield (YP) and shoot dry matter (SP) for individuals of maize, sun¯ower and soybean
465
466
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower 0.8
Maize 199495
Maize 199798
0.6 0.4 0.2 0
SS1 = 1.9 SS2 = 0.8 SS3 = 1.6 SS4 = 0.6
SS1 = 8.9 SS2 = 4.6 SS3 = 8.3 SS4 = 2.1
0
200
400
600
800
0
Harvest index per plant
0.6
200
400
600
800
Sunflower 199596
Sunflower 199495 0.4
0.2
0.0
SS1 = 0.09 SS2 = 0.09 SS3 = 0.11 SS4 = 0.06
SS1 = 2.7 SS2 = 1.5 SS3 = 3.0 SS4 = 0.4
0
250
0.8
500
750
1000
0
200
400
600
800
1000
Soybean 199798
Soybean 199495
0.6 0.4 SS1 = 4.0 SS2 = 45.5 SS3 = 7.0 SS4 = 2.6
0.2 0
0
50
100
150
200
SS1 = 1.2 SS2 = 2.6 SS3 = 1.0 SS4 = 2.5
0
50
100
150
200
Shoot biomass per plant (g) FIG. 2. Relationship between harvest index (HI) and shoot biomass (SP) at physiological maturity in individuals of maize, sun¯ower and soybean grown under a wide range of plant population densities. Curves are the expected relationship based on models 4A and 4B using the parameters in Tables 4 and 5. In 1994±95, plant population densities were: low (j), intermediate (s) and high (m) as described in Table 1. For maize, (h) indicates proli®c plants. SSi are sums of square errors between actual and predicted HI values for models 1 to 4 (Table 2).
as indication of dynamic trajectories of individuals. It is worth noting, however, that the patterns of reproductive allocation and size-dependence of HI in the ®rst seasonÐ when a substantial proportion of the variation in shoot mass and grain yield derived from hierarchies established within treatmentsÐwere very similar to those in the second year, when plant density treatments accounted for a larger proportion of the variation in SP and YP . Threshold for reproduction Our study showed that there is: (1) a threshold plant size for grain set; and (2) a substantial variation in the threshold
among species. The ranking of thresholds, i.e. soybean 5 sun¯ower 5 maize, re¯ects contrasting patterns of reproductive partitioning under poor growing conditions (Table 4). In correspondence with the ranking of SP thresholds, the ranking of the percentage of sterile plants in high-density stands in 1994±95 was 3% for soybean, 7% for sun¯ower and 26% for maize. The ¯oral biology of maize, a monoecious species with marked apical dominance (Paterniani, 1981; Fischer and Palmer, 1984; Doebley, Stec and Hubbard, 1997), underlies the high susceptibility of the plant to low availability of resources that causes sterility at SP values well above those observed for soybean and sun¯ower.
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower T A B L E 5. Parameters (+s.e.) of model 4B ®tted to grain yield (YP) and shoot dry matter (SP) of proli®c maize individuals Season
a4B
b4B
R2 {
N{
1994±95 1997±98
602.0 + 5.31 782.3 + 60.60
ÿ180000 + 30200 ÿ250000 + 28500
0.46 0.82
43 14
{ Coecient of determination. { Number of individuals.
Previous research also shows positive SP thresholds for reproduction (Gardner and Gardner, 1983; Weiner, 1988; Hartnett, 1990; Thompson et al., 1991; Moot et al., 1997). Others, however, showed zero or negative x-intercepts (Hartnett, 1990; Sinclair et al., 1990; Prihar and Stewart, 1991; Thompson et al., 1991). This apparent discrepancy may have derived from a number of reasons, including a narrow range of SP (see Introduction). In particular, lack or shortage of data at the lowest end of SP implies considerable extrapolation to estimate thresholds. Forcing linear relationships even in cases of slight curvilinearity may also generate unrealistic thresholds. For instance, the linear model for soybean in 1994±95 had a negative x-intercept which unrealistically implies that a plant with no mass could produce 0.78 g seed (Table 3); the poor ability of model 2 to predict HI in this case is also evident (see SS in Fig. 2). Stability of HI In all three species, curvilinear relationships between YP and SP showed positive ST indicating that YP depended not only on dry matter accumulation but also on dry matter partitioning to grain (Fig. 2; Table 4). Furthermore, Fig. 2 clearly demonstrates the non-linear association between HI and SP . In all three species, HI was stable for intermediate plant size. The stability of HI often described in the literature derives from studies involving predominantly mid-size plants grown under situations with low to moderate stress (Spaeth et al., 1984; Sinclair et al., 1990; Cox, 1996; Sadras et al., 1997). The apparent contradiction between the widespread view of `stable' HI in the agronomic literature, and the lower stability demonstrated in our study derives, therefore, from our consideration of plant-to-plant variation in reproductive partitioning within the population. Therefore, HI stability at crop level will clearly depend on the frequency distribution of SP and YP within the population. Our study also demonstrated that HI stability strongly varied among species; it was greatest in soybean, intermediate in sun¯ower and lowest in maize. The reasons for this ranking have to be found at both ends of the SP range, where the HI vs. SP curve bends down. Dierences among species at the lower end of plant size have been discussed in the previous section. The bending of the curve at the high end of SP was much less pronounced in soybean than in sun¯ower and non-proli®c maize. This is a re¯ection of the greater
467
reproductive plasticity of indeterminate soybean in comparison with determinate plants with no (sun¯ower) or limited (maize) capacity to adjust in¯orescence number in response to availability of resources (Loomis and Connor, 1996). However, it is valid to point out that growing soybean plants heavier than 200 g dry matter in the ®eld was very dicult because heavy branches broke easily. Comparison of soybean responses in both experimental seasons highlights how important it is to consider the range of SP in drawing conclusions about HI stability: the drop in HI with increasing SP was evident in 1995±96, when maximum shoot dry matter was close to 200 g per plant, but not in 1997±98 when it was only 120 g per plant (Fig. 2). The range of SP for soybean in this study, nonetheless, exceeds the range of sizes we can ®nd in crops at normal sowing densities. The morphological approach to resource allocation outlined by Marshall and Watson (1992), later formalized in the model of Bonser and Aarssen (1996), provides a framework to make explicit the contrasting strategies of the species in this study. First, Bonser and Aarssen (1996) considered that all initially dormant axillary meristems have one of three principal fates: growth (G) meristems produce a new shoot or branch; reproductive (R) meristems produce a ¯ower or in¯orescence; and inactive (I) meristems remain dormant. Then, they proposed to de®ne plant strategies according to the measure of reproductive eort, RE R/ (G I), branching intensity, BI G/(I R), and apical dominance, AD I/(G R). Within this frame, Sadras (1998) pointed out that the stability of reproductive allocation in soybean results from a strategy based on weak AD, large RE, and large BI. In contrast, the more restricted plasticity of maize and sun¯ower is the result of an extreme strategy with very strong AD and very reduced or nil BI. Both maize and sun¯ower can be regarded as species with low RE notwithstanding the substantial proportion of axillary buds that became reproductive, but remain inactive in mature maize plants (Kiesselbach, 1949). Comparison of maize plants with one and two ears in Fig. 2 shows how proli®cacy contributes to the stability of reproductive allocation in this species. Proli®cacy in maize, however, is largely in¯uenced by environmental and genetic factors (Prior and Russell, 1975; Motto and Moll, 1983; Otegui, 1995). The varying limits to reproductive plasticity among the species demonstrated in our study broaden Weiner's (1988) theory of linear, size-dependent reproductive output. CO N C L U S I O N Manipulation of plant density allowed us to generate a broad range of resource availability for individual plants, and therefore of plant size. Proportion of dry matter partitioned to reproduction was stable for mid size plants and decreased for small or extremely large plants. Harvest index stability was highest in soybean, intermediate in sun¯ower and lowest in maize. Dierences in reproductive strategies between soybean, maize and sun¯ower were re¯ected in dierences in both the threshold of plant dry matter for reproduction and the drop
468
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower
in reproductive partitioning with high availability of resources. Dierential stability of reproductive partitioning partially derives from contrasting patterns of meristem allocation. AC K N OW L E D G E M E N T S This work is part of a thesis submitted by CRC Vega in partial ful®llment for the requirements for the degree of MSc, Universidad de Mar del Plata. We thank Professor Weiner, Dr LT Evans and Dr P Laterra for helpful comments, and the ®nancial support from INTA, UMP, FundacioÂn Antorchas, Dekalb S.A, and CONICET, the Research Council of Argentina. CRC Vega held a scholarship from CIC (ComisioÂn de Investigaciones CientõÂ ®cas de la Provincia de Buenos Aires). VO Sadras, FH Andrade, and SA Uhart are members of CONICET. L I T E R AT U R E C I T E D Andrade FH, Vega CRC, Uhart SA, Cirilo AG, Cantarero M, Valentinuz O. 1999. Kernel number determination in maize. Crop Science 39: 453±459. Bonser SP, Aarssen LW. 1996. Meristem allocation: a new classi®cation theory for adaptive strategies in herbaceous plants. Oikos 77: 347±352. Charles-Edwards DA. 1982. Physiological determinants of crop growth. North Ryde: Academic Press. Cox WJ. 1996. Whole-plant physiological and yield responses of maize to plant density. Agronomy Journal 88: 489±496. Doebley J, Stec A, Hubbard L. 1997. The evolution of apical dominance in maize. Nature 386: 485±488. Donald CM, Hamblin J. 1976. The biological yield and harvest index of cereals as agronomic and plant breeding criteria. Advances in Agronomy 28: 361±405. Edmeades GO, Daynard TB. 1979. The relationship between ®nal yield and photosynthesis at ¯owering in individual maize plants. Canadian Journal of Plant Science 89: 585±601. Evans LT. 1994. Crop physiology: prospects for the retrospective science. In: Boote KJ, Bennett JM, Sinclair TR, Paulsen GM, eds. Physiology and determination of crop yield. Madison: ASA, CSSA, SSSA, 19±35. Fehr WR, Caviness CE. 1977. Stages of soybean development. Special Report 80. Ames, Iowa: Cooperative Extension Service, Agriculture and Home Economics Exp. Stn Iowa State University 11: 929±931. Fischer KS, Palmer AFE. 1984. Tropical maize. In: Goldsworthy PR, Fisher NM, eds. The physiology of tropical ®eld crops. Bath, Avon: J. Wiley & Sons Ltd., 213±248. Gardner WR, Gardner HR. 1983. Principles of water management under drought conditions. Agricultural Water Management 7: 143±155. Giord RM, Thorne JH, Hitz WD, Giaquinta RT. 1984. Crop productivity and photoassimilate partitioning. Science 225: 801±808. Hartnett DC. 1990. Size dependent allocation to sexual and vegetative reproduction in four clonal composites. Oecologia 84: 254±259. Hay RKM. 1995. Harvest index: a review of its use in plant breeding and crop physiology. Annals of Applied Biology 126: 197±216. Johnson EC, Fischer KS, Edmeades GO, Palmer AFE. 1985. Recurrent selection for reduced plant height in lowland tropical Maize. Crop Science 26: 253±260. Kiesselbach TA. 1949. The structure and reproduction of corn. Nebraska Agricultural Research Bulletin No 161. Lincoln: University of Nebraska. Klinkhamer PGL, De Jong TJ, Meelis E. 1990. How to test for proportionality in the reproductive eort of plants. American Naturalist 135: 291±300.
Klinkhamer PGL, Meelis E, de Jong TJ, Weiner J. 1992. On the analysis of size-dependent reproductive output in plants. Functional Ecology 6: 308±316. Loomis RS, Connor DJ. 1996. Crop ecology. Productivity and management in agricultural systems. Cambridge: Cambridge University Press. Marshall C, Watson MA. 1992. Ecological and physiological aspects of reproductive allocation. In: Marshall C, Grace J, eds. Fruit and seed production. Aspects of development, environmental physiology and ecology. Cambridge: Cambridge University Press, 173±202. Moot DJ, Wilson DR, McNeil DL. 1997. Validation of the principal axis model (PAM) and its application to genotype selection in ®eld pea (Pisum sativum L.) crops. Annals of Botany 79: 651±656. Motto M, Moll RH. 1983. Proli®cacy in maize: a review. Maydica 28: 53±76. Otegui ME. 1995. Proli®cacy and grain yield components in modern Argentinian maize hybrids. Maydica 40: 371±376. Paterniani E. 1981. In¯uence of tassel size on ear placement. Maydica 26: 85±91. Prihar SS, Stewart BA. 1991. Sorghum harvest index in relation to plant size, environment, and cultivar. Agronomy Journal 83: 603±608. Prior CL, Russell WA. 1975. Yield performance in non proli®c and proli®c maize hybrids at six plant densities. Crop Science 15: 482±486. Reekie EG, Bazzaz FA. 1987. Reproductive eort in plants. 1. Carbon allocation to reproduction. American Naturalist 129: 876±896. Rees M, Crawley MJ. 1989. Growth, reproduction and population dynamics. Functional Ecology 3: 645±653. Ritchie SW, Hanway JJ. 1982. How a corn plant develops. Special Report 48. Ames, Iowa: Cooperative Extension Service, Iowa State University of Science and Technology. Sadras VO. 1998. Variation in apical dominance and its implications for herbivory resistance, competitive ability and biomass partitioning. In: Otegui ME, Slafer GA, eds. Proceedings of the international workshop on physiological bases for maize improvement. Buenos Aires, 71±81. Sadras VO, Connor DJ. 1991. Physiological basis of the response of harvest index to the fraction of water transpired after anthesis. A simple model to estimate harvest index for determinate species. Field Crops Research 26: 227±239. Sadras VO, Bange MP, Milroy SP. 1997. Reproductive allocation of cotton in response to plant and environmental factors. Annals of Botany 80: 75±81. Samson DA, Werk KS. 1986. Size-dependent eects in the analysis of reproductive eort in plants. American Naturalist 127: 667±680. Schmid B, Polasek W, Weiner J, Krause A, Stoll P. 1994. Modeling of discontinuous relationships in biology with censored regression. American Naturalist 143: 494±507. Schneiter AA, Miller JF. 1981. Description of sun¯ower growth stages. Crop Science 21: 901±903. Sinclair TR. 1998. Historical changes in harvest index and crop nitrogen accumulation. Crop Science 38: 638±643. Sinclair TR, Bennett JM, Muchow RC. 1990. Relative sensitivity of grain yield and biomass accumulation to drought in ®eld-grown maize. Crop Science 30: 690±693. Solbrig O, Solbrig DJ. 1985. Size inequalities and ®tness in plant populations. Topics in Evolutionary Biology 1: 141±159. Spaeth SC, Randall HC, Sinclair TR, Vendeland JS. 1984. Stability of soybean harvest index. Agronomy Journal 76: 482±486. Thompson BK, Weiner J, Warwick SI. 1991. Size-dependent reproductive output in agricultural weeds. Canadian Journal of Botany 69: 442±446. Tollenaar M, McCullough E, Dwyer LM. 1994. Physiological basis of the genetic improvement of corn. In: Slafer GA, ed. Genetic improvement of ®eld crops. New York: Marcel Dekker, 183±236. Weiner J. 1988. The in¯uence of competition on plant reproduction. In: Lovett Doust J, Lovett Doust L, eds. Plant reproductive ecology: patterns and strategies. New York: Oxford University Press, 228±245. Weiner J, Thomas SC. 1992. Competition and allometry in three species of annual plants. Ecology 73: 648±656.
Reproductive Allometry in Soybean, Maize and Sun¯ower C . R . C . V E G A * , V. O . S A D R A S , F. H . A N D R A D E and S . A . U H A R T INTA BalcarceÐUniversidad de Mar del Plata, CC 276, 7620 Balcarce, Argentina Received: 5 July 1999 Returned for revision: 5 October 1999 Accepted: 7 December 1999 We compared the relationship between grain yield per plant (YP) and shoot biomass per plant (SP) in three annual crops with contrasting reproductive strategies: sun¯ower, a determinate species with a single in¯orescence; maize, a determinate species with a limited capacity to adjust the number of ears in response to resource availability; and indeterminate soybean, a species with a large capacity to adjust the number of in¯orescences. Our working hypotheses were: H1 Ðthe relationship between YP and SP is linear; H2 Ðthe intercept of the model is zero, i.e. there is not a threshold plant mass for reproduction. A wide range of YP and SP was generated by manipulation of plant density; SP varied between 0.3 and 196 g per plant in soybean, between 6 and 873 g per plant in sun¯ower and between 23 and 697 g per plant in maize. Within these broad ranges of plant size, both hypotheses were rejected in ®ve out of six experiments, i.e. the relationship between YP and SP departed from linearity and there was a threshold for SP below which no grain set occurred. The SP threshold for grain set varied widely among species; it was close to 2 g per plant for soybean, 27 g per plant for sun¯ower and 43±71 g per plant for maize. Because of this size threshold and non-linearity, harvest index (HI YPSP ÿ1) was stable for mid-size plants, diminished slightly for large plants, and diminished sharply for smaller plants in all three crops. Harvest index stability was highest in soybean, intermediate in sun¯ower and lowest in maize. Dierential stability of reproductive partitioning partially derived from contrasting # 2000 Annals of Botany Company patterns of meristem allocation. Key words: Helianthus annuus L., Zea mays L., Glycine max (L.) Merrill, grain yield, harvest index, plant density, reproductive allocation, meristem allocation, plasticity.
I N T RO D U C T I O N Biomass accumulation and partitioning to reproductive structures are critical aspects of plant ®tness (Solbrig and Solbrig, 1985; Reekie and Bazzaz, 1987; Weiner, 1988; Hartnett, 1990). They are also key determinants of crop yield (Giord et al., 1984; Andrade et al., 1999). In many plant species, dry matter partitioning to reproductive organs is a stable and highly heritable trait (Spaeth et al., 1984; Johnson et al., 1985; Hay, 1995; Sadras, Bange and Milroy, 1997). The mechanisms underlying reproductive partitioning are, however, poorly understood (Evans, 1994). Linear relationships between reproductive and shoot biomass are often used to describe reproductive partitioning and grain yield (Gardner and Gardner, 1983; Samson and Werk, 1986; Sinclair, Bennett and Muchow, 1990; Prihar and Stewart, 1991; Moot, Wilson and McNeil, 1997). There is no agreement, however, about the x-intercept of this linear model, i.e. the minimum shoot biomass required for grain set. Gardner and Gardner (1983), Samson and Werk (1986) and Weiner (1988) argued that plants have a threshold size to produce ¯ower and fruit. Empirical evidence to support the existence of such a threshold has been found in a number of studies (Gardner and Gardner, 1983; Weiner, 1988; Hartnett, 1990; Thompson, Weiner and Warwick, 1991; Moot et al., 1997). In others, it has been suggested that a threshold plant mass is not a condition for reproduction (Rees and Crawley, 1989; Prihar and Stewart, 1991). This apparent disagreement * For correspondence. E-mail [email protected]
0305-7364/00/040461+08 $35.00/00
could be caused by a number of factors, including range of plant and grain biomass, the source of variation of plant size (e.g. water stress, nutrient de®cit), growth habit (determinate vs. indeterminate), life history (e.g. perennial vs. annual) and agronomic selection (wild vs. cultivated). We analysed the relationship between grain yield and shoot biomass and the eect of variation in plant biomass on the stability of dry matter partitioning to grain. Our null hypotheses were: H1 Ðthe relationship between grain yield and shoot biomass is linear; and H2 Ðthe intercept of the model is zero, i.e. a threshold plant mass is not a condition for reproduction. Our approach had three main components. Firstly, we compared three cultivated, annual species with contrasting reproductive strategies: sun¯ower (Helianthus annuus L.), a determinate species with a single in¯orescence; maize (Zea mays L.), a determinate species with a limited capacity to adjust its number of ears in response to availability of resources; and indeterminate soybean (Glycine max L. Merrill), a species with a large capacity to adjust the number of in¯orescences. Secondly, we generated a wide range of availability of resources by manipulation of plant population density. Thirdly, to account for plant-to-plant variation, we measured grain yield and shoot biomass of individual plants. M AT E R I A L S A N D M E T H O D S Crops and treatments Crops were grown on deep (51.5 m) Typical Argiudolls at Balcarce (378450 S, 588180 W; altitude 130 m), Argentina, # 2000 Annals of Botany Company
462
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower T A B L E 1. Summary of treatments in ®eld experiments at Balcarce Plant population ( plants m ÿ2)
Number of plants sampled
Species
Cultivar
Season
Date of emergence
Maize
DK 636 DK 639
1994±95 1997±98
17 Oct. 1994 2 Nov. 1997
2.2, 8.5 and 16.9 2.0, 3.6, 8.4, 13.0, 15.9 and 20.4
298 287
Sun¯ower
DK G-100 DK G-100
1994±95 1995±96
27 Oct. 1994 31 Oct. 1995
1.4, 5.8 and 10.3 1.4, 1.8, 2.2, 2.9, 4.1, 5.8 and 11.6
258 60
Soybean
Asgrow 3127 Asgrow 3205
1994±95 1997±98
13 Nov. 1994 14 Nov. 1997
7.9, 29.8 and 56.5 3.6, 7.1, 26.5 and 51.4
322 77
between 1994 and 1998 (Table 1). All three crops were fertilized with 35 kg P ha ÿ1 before sowing, and with 150 kg N ha ÿ1 at the V6 stage in maize (Ritchie and Hanway, 1982) and V4 in sun¯ower (Schneiter and Miller, 1981). Soybean seed was inoculated with Bradyrhizobium japonicum. Crops were irrigated to keep water content above 50% of maximum soil available water. Weeds and insects were adequately controlled. Each species was grown in a dierent section of the same ®eld. Plant density treatments were laid out within each crop species in a block design with three (1994±95 and 1995±96) or four replications (1997±98). Distance between rows was 0.7 m in all three crops and plant densities ranged from 3.6 to 56.5 plants m ÿ2 in soybean, 1.4 to 11.6 plants m ÿ2 in sun¯ower and 2.0 to 20.4 plants m ÿ2 in maize (Table 1). Maize and sun¯ower were sown by hand at the target density (Table 1) placing three seeds per hill and thinning to one plant per hill at stage V2 (Ritchie and Hanway, 1982) and V4±5 (Schneiter and Miller, 1981), respectively. Heavily sown soybean was thinned to achieve the appropriate density at stage V2 (Fehr and Cavinness, 1977). At physiological maturity, individual plants were harvested to determine shoot dry matter (SP) and grain yield (YP). In soybean, no attempt was made to recover fallen leaves. Plants were sampled randomly within each treatment and no plants were excluded except for some soybean individuals with heavy, broken branches at the lowest density. Sampling strategies varied between seasons. In the ®rst season, three plant population densities were established and large plant samples were taken to exploit variation within treatments as well as variation among treatments. In the second, we increased the number of treatments to four±seven and took fewer plants per treatment (Table 1). Analytical approach Despite some drawbacks (Samson and Werk, 1986; Klinkhamer et al., 1992) the relationship between SP and YP is an appropriate framework to investigate the pattern of reproductive allocation (Prihar and Stewart, 1991; Sadras et al., 1997). Furthermore, these two variables bring forth the variable harvest index, i.e. HI YPSP ÿ1, a concept widely used in agronomic and plant breeding research (Donald and Hamblin, 1976; Hay, 1995; Sinclair, 1998). Thus, our study is largely based on the relationships
between YP and SP , and between HI and SP . Additionally, we investigated the relationship between YP and vegetative shoot dry matter. A number of methods have been proposed for the analysis of size-dependent reproductive eort in plants (Klinkhamer et al., 1992; Schmid et al., 1994). In the nonlinear model of Klinkhamer et al. (1992) data points below the estimated threshold must be omitted from the analysis (Schmid et al., 1994). To deal with this problem and to properly analyse discontinuous relationships, Schmid et al. (1994) proposed a censored allometric model. Our approach in this study is based on: (1) no exclusion of data points with zero yield; (2) simple, biologically meaningful models; and (3) sequential analysis of YP vs. SP and HI vs. SP relationships. To test our hypotheses, we ®tted linear and non-linear models to the data, i.e. YP b1 SP
Model 1
YP a2 b2 SP
Model 2
YP
a3 SP 1 b3 SP
Model 3
YP
a4A SP ÿ ST 1 b4A SP ÿ ST
Model 4A
YP a4B
b4B SP
Model 4B
The implications of these models are given in Table 2. The linear models (1 and 2) are the simplest and more often used models to describe the relationship between YP and SP . Departure from linearity can be tested through regression of log YP on log SP (Klinkhamer, De Jong and Meelis, 1990; Thompson et al., 1991). However, as this test can produce misleading results when the y-intercept diers from zero, polynomial regressions were preferred, as in Thompson et al. (1991). Hypothesis 1 was rejected when second or third-order terms in polynomials were signi®cant (P 5 0.05). Models 3 and 4 are hyperbolic functions without and with x-intercepts, respectively; hyperbolic models have been widely used in studies of reproductive growth in maize (Edmeades and Daynard, 1979; Tollenaar, McCullough and Dwyer, 1994). In Model 4A, the parameter ST estimates the SP threshold for grain yield. Signi®cance of ST was used to test hypothesis 2. For
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower T A B L E 2. Generic models to describe the relationship between grain yield (YP) and shoot dry matter (SP) and consequences derived from each model for the relationship between harvest index (HI) and SP Model type
Slope
Intercept
Consequences for HI
1 2
Constant Constant
Zero Negative
3 4
Variable Variable
Zero Negative
HI constant If SP 4 ST then HI 0 If SP 4 ST then HI f(SP) HI f(SPÿ1) If SP 4 ST then HI 0 If ST 5 SP 5 SM then HI f(SP) If SP 4 SM then HI f(SP ÿ1)
ST is the SP threshold for grain yield, SM is the SP corresponding to maximum HI.
comparisons among species, ST and its standard error were expressed as a percentage of: (1) average SP at commercial density, and (2) maximum SP observed in the experiments. Model 4B was used to describe the relationship between YP and SP in proli®c maize plants. Each of the four models described has important implications for the relationship between HI and SP . Depending on whether the slope is constant or variable, and whether the intercept is zero or negative, the expected relationship between HI and SP is outlined in Table 2. As a further test of the modelsÐand therefore of our working hypothesesÐ we compared the actual relationship between HI and SP with the predictions resulting from each model.
463
Contrasting availability of resources per plant together with the use of individuals, rather than averages, contributed to the wide range of both YP and SP (Fig. 1). Shoot dry matter varied between 0.3 and 196 g per plant in soybean, between 6 and 873 g per plant in sun¯ower and between 23 and 698 g per plant in maize. Statistically, model 2 described the data very well (Table 3). Signi®cance of the second-order term in polynomial regressions, however, revealed departures from linearity in ®ve out of six cases (P 5 0.0001). Soybean in 1997±98 was the only exception. Several non-linear functions adequately ®tted the data for the three crops. Of them, we chose hyperbolic functions (Model 4) because they are both statistically sound and biologically meaningful as they include parameters that re¯ect SP thresholds for grain yield ( parameter ST , Table 4). The threshold of shoot dry matter required for grain set and yield was close to 2 g per plant in soybean (P 0.06), 27 g per plant in sun¯ower (P 5 0.0001) and 43±71 g per plant in maize (P 5 0.0001). To account for dierences in plant size among species (Fig. 1) thresholds were also calculated as a percentage of: (1) average shoot dry matter at commercial plant population density; and (2) maximum SP (Table 4). Normalised thresholds ranked soybean 5 sun¯ower 5 maize (Table 4). The large variation in thresholds in maize can be attributed, at least in part, to the use of dierent hybrids in both experiments (Gardner and Gardner, 1983). It is worthwhile noting that the parameter b4a was not dierent from zero for soybean in 1997±98, indicating that the relationship was adequately described by a linear model (Table 4). Some maize individuals were proli®c; i.e. set grain in a second ear. Hence, a second non-linear relationship was ®t to the data for proli®c plants (Fig. 1; Table 5).
R E S U LT S Relationship between YP and SP
Relationship between HI and SP
Figure 1 shows the relationship between YP and SP in individuals of soybean, sun¯ower and maize at physiological maturity. Yield and shoot biomass are not strictly independent variables (Charles-Edwards, 1982). The relationships between YP and vegetative shoot biomass per plant for all three crops (not shown), however, were similar to those found for YP and SP (Fig. 1).
In all three species, HI was stable for mid-size plants, diminished slightly for large plants, and diminished sharply for smaller plants (Fig. 2). The only exception to this pattern was soybean in 1997±98 in which HI did not decline with the largest SP (Fig. 2). The analysis of individuals allowed for the detection of a continuous variation in HI from maximum values to zero.
T A B L E 3. Parameters (+s.e.) of the linear regression (model 2) between grain yield (YP) and shoot dry matter (SP) for individuals of maize, sun¯ower and soybean Species
Season
a2
b2
R2{
N{
Maize
1994±95 1997±98
ÿ15.04 + 2.784 ÿ13.87 + 1.448
0.52 + 0.008 0.57 + 0.006
0.94 0.97
298 287
Sun¯ower
1994±95 1995±96
ÿ1.87 + 1.017 ÿ0.52 + 2.954
0.36 + 0.003 0.43 + 0.010
0.98 0.97
258 60
Soybean
1994±95 1997±98
0.78 + 0.360 ÿ1.00 + 0.362
0.49 + 0.006 0.60 + 0.009
0.95 0.98
322 77
Parameters are: intercept (a2), and slope (b2). All regressions were signi®cant at P 5 0.0001. { Coecient of determination. { Number of individuals.
464
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower 400 Maize 199798
Maize 199495 300 200 100 0
0
Grain yield per plant (g)
400
200
400
600
800
0
200
Sunflower 199495
400
600
800
Sunflower 199596
300 200 100 0
0
100
250
500
750
1000
0
250
500
750
1000
150
200
Soybean 199798
Soybean 199495
75 50 25 0 0
50
100
150
200
0
50
100
Shoot biomass per plant (g) FIG. 1. Relationship between grain yield per plant (YP) and shoot biomass per plant (SP) at physiological maturity in maize, sun¯ower and soybean grown under a wide range of plant population densities. In 1994±95, plant population densities were: low (j), intermediate (s) and high (m) as described in Table 1. For maize, (h) indicates proli®c plants.
Maximum HI was close to 0.7 in soybean, 0.43±0.47 in sun¯ower and 0.55±0.63 in maize. At high values of SP , soybean had the smallest reduction in HI and maize the greatest. In maize, only proli®c plants had HI close to the maximum (Fig. 2). The analysis in Fig. 2 (cf. SSi) further supports the nonlinearity of the relationship between YP and SP , as well as the existence of a non-zero threshold of shoot biomass, i.e. model 4 predicted HI better than models 1, 2 and 3. DISCUSSION We used the relationship between YP and SP and between HI and SP to investigate dry matter partitioning to grain in species with contrasting reproductive strategies (Figs 1 and
2). Linear models between YP and SP were statistically adequate to describe reproductive partitioning (Table 3). The relationship between these variables, however, departed from linearity signi®cantly at both ends of resource availability and plant size (Figs 1 and 2). Plant population density was the main source of variation in plant size; limiting individual plant biomass by other means, e.g. by manipulation of water supply, could give dierent relations between YP , SP and HI (Prihar and Stewart, 1991; Sadras and Connor, 1991). Weiner and Thomas (1992) pointed out the disagreement between static, inter-individual allometry at one point in time, and dynamic patterns of allometric growth for competing plants. Accordingly, the static patterns of reproductive allocation in our study should not be taken
Normalized ST Species
Season
a4A
ST (g per plant)
% of average SP at commercial plant population density}
% of maximum SP
b4A
R2 {
N{
Maize*
1994±95 1997±98
0.9 + 0.03 0.7 + 0.03
70.8 + 2.62 42.7 + 2.94
23.58 + 0.009 15.55 + 0.011
10.15 + 0.004 6.92 + 0.005
0.0019 + 0.0002 0.0009 + 0.0001
0.95 0.97
255 273
Sun¯ower
1994±95 1995±96
0.5 + 0.01 0.6 + 0.04
25.4 + 3.02 28.9 + 7.80
14.80 + 0.018 17.87 + 0.048
2.90 + 0.003 4.81 + 0.013
0.0004 + 0.0001 0.0005 + 0.0002
0.98 0.98
258 60
Soybean
1994±95 1997±98
0.6 + 0.02 0.6 + 0.03
1.5 + 0.79 1.6 + 0.83
5.73 + 0.030 6.54 + 0.034
0.77 + 0.004 1.35 + 0.007
0.0014 + 0.0003 0.0001 + 0.0001
0.95 0.99
322 77
Parameters are: a4A (initial slope), ST (SP threshold for grain yield) and b4A (slope at high values of SP). Normalized thresholds are also shown for comparison among species. All regressions were signi®cant at P 5 0.0001. { Coecient of determination. { Number of individuals. } Maize: 8.4±8.5 plants m ÿ2, sun¯ower: 5.8 plants m ÿ2, soybean: 26.5±29.8 plants m ÿ2. * Model ®tted to plants with one ear.
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower
T A B L E 4. Parameters (+s.e.) of model 4A ®tted to grain yield (YP) and shoot dry matter (SP) for individuals of maize, sun¯ower and soybean
465
466
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower 0.8
Maize 199495
Maize 199798
0.6 0.4 0.2 0
SS1 = 1.9 SS2 = 0.8 SS3 = 1.6 SS4 = 0.6
SS1 = 8.9 SS2 = 4.6 SS3 = 8.3 SS4 = 2.1
0
200
400
600
800
0
Harvest index per plant
0.6
200
400
600
800
Sunflower 199596
Sunflower 199495 0.4
0.2
0.0
SS1 = 0.09 SS2 = 0.09 SS3 = 0.11 SS4 = 0.06
SS1 = 2.7 SS2 = 1.5 SS3 = 3.0 SS4 = 0.4
0
250
0.8
500
750
1000
0
200
400
600
800
1000
Soybean 199798
Soybean 199495
0.6 0.4 SS1 = 4.0 SS2 = 45.5 SS3 = 7.0 SS4 = 2.6
0.2 0
0
50
100
150
200
SS1 = 1.2 SS2 = 2.6 SS3 = 1.0 SS4 = 2.5
0
50
100
150
200
Shoot biomass per plant (g) FIG. 2. Relationship between harvest index (HI) and shoot biomass (SP) at physiological maturity in individuals of maize, sun¯ower and soybean grown under a wide range of plant population densities. Curves are the expected relationship based on models 4A and 4B using the parameters in Tables 4 and 5. In 1994±95, plant population densities were: low (j), intermediate (s) and high (m) as described in Table 1. For maize, (h) indicates proli®c plants. SSi are sums of square errors between actual and predicted HI values for models 1 to 4 (Table 2).
as indication of dynamic trajectories of individuals. It is worth noting, however, that the patterns of reproductive allocation and size-dependence of HI in the ®rst seasonÐ when a substantial proportion of the variation in shoot mass and grain yield derived from hierarchies established within treatmentsÐwere very similar to those in the second year, when plant density treatments accounted for a larger proportion of the variation in SP and YP . Threshold for reproduction Our study showed that there is: (1) a threshold plant size for grain set; and (2) a substantial variation in the threshold
among species. The ranking of thresholds, i.e. soybean 5 sun¯ower 5 maize, re¯ects contrasting patterns of reproductive partitioning under poor growing conditions (Table 4). In correspondence with the ranking of SP thresholds, the ranking of the percentage of sterile plants in high-density stands in 1994±95 was 3% for soybean, 7% for sun¯ower and 26% for maize. The ¯oral biology of maize, a monoecious species with marked apical dominance (Paterniani, 1981; Fischer and Palmer, 1984; Doebley, Stec and Hubbard, 1997), underlies the high susceptibility of the plant to low availability of resources that causes sterility at SP values well above those observed for soybean and sun¯ower.
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower T A B L E 5. Parameters (+s.e.) of model 4B ®tted to grain yield (YP) and shoot dry matter (SP) of proli®c maize individuals Season
a4B
b4B
R2 {
N{
1994±95 1997±98
602.0 + 5.31 782.3 + 60.60
ÿ180000 + 30200 ÿ250000 + 28500
0.46 0.82
43 14
{ Coecient of determination. { Number of individuals.
Previous research also shows positive SP thresholds for reproduction (Gardner and Gardner, 1983; Weiner, 1988; Hartnett, 1990; Thompson et al., 1991; Moot et al., 1997). Others, however, showed zero or negative x-intercepts (Hartnett, 1990; Sinclair et al., 1990; Prihar and Stewart, 1991; Thompson et al., 1991). This apparent discrepancy may have derived from a number of reasons, including a narrow range of SP (see Introduction). In particular, lack or shortage of data at the lowest end of SP implies considerable extrapolation to estimate thresholds. Forcing linear relationships even in cases of slight curvilinearity may also generate unrealistic thresholds. For instance, the linear model for soybean in 1994±95 had a negative x-intercept which unrealistically implies that a plant with no mass could produce 0.78 g seed (Table 3); the poor ability of model 2 to predict HI in this case is also evident (see SS in Fig. 2). Stability of HI In all three species, curvilinear relationships between YP and SP showed positive ST indicating that YP depended not only on dry matter accumulation but also on dry matter partitioning to grain (Fig. 2; Table 4). Furthermore, Fig. 2 clearly demonstrates the non-linear association between HI and SP . In all three species, HI was stable for intermediate plant size. The stability of HI often described in the literature derives from studies involving predominantly mid-size plants grown under situations with low to moderate stress (Spaeth et al., 1984; Sinclair et al., 1990; Cox, 1996; Sadras et al., 1997). The apparent contradiction between the widespread view of `stable' HI in the agronomic literature, and the lower stability demonstrated in our study derives, therefore, from our consideration of plant-to-plant variation in reproductive partitioning within the population. Therefore, HI stability at crop level will clearly depend on the frequency distribution of SP and YP within the population. Our study also demonstrated that HI stability strongly varied among species; it was greatest in soybean, intermediate in sun¯ower and lowest in maize. The reasons for this ranking have to be found at both ends of the SP range, where the HI vs. SP curve bends down. Dierences among species at the lower end of plant size have been discussed in the previous section. The bending of the curve at the high end of SP was much less pronounced in soybean than in sun¯ower and non-proli®c maize. This is a re¯ection of the greater
467
reproductive plasticity of indeterminate soybean in comparison with determinate plants with no (sun¯ower) or limited (maize) capacity to adjust in¯orescence number in response to availability of resources (Loomis and Connor, 1996). However, it is valid to point out that growing soybean plants heavier than 200 g dry matter in the ®eld was very dicult because heavy branches broke easily. Comparison of soybean responses in both experimental seasons highlights how important it is to consider the range of SP in drawing conclusions about HI stability: the drop in HI with increasing SP was evident in 1995±96, when maximum shoot dry matter was close to 200 g per plant, but not in 1997±98 when it was only 120 g per plant (Fig. 2). The range of SP for soybean in this study, nonetheless, exceeds the range of sizes we can ®nd in crops at normal sowing densities. The morphological approach to resource allocation outlined by Marshall and Watson (1992), later formalized in the model of Bonser and Aarssen (1996), provides a framework to make explicit the contrasting strategies of the species in this study. First, Bonser and Aarssen (1996) considered that all initially dormant axillary meristems have one of three principal fates: growth (G) meristems produce a new shoot or branch; reproductive (R) meristems produce a ¯ower or in¯orescence; and inactive (I) meristems remain dormant. Then, they proposed to de®ne plant strategies according to the measure of reproductive eort, RE R/ (G I), branching intensity, BI G/(I R), and apical dominance, AD I/(G R). Within this frame, Sadras (1998) pointed out that the stability of reproductive allocation in soybean results from a strategy based on weak AD, large RE, and large BI. In contrast, the more restricted plasticity of maize and sun¯ower is the result of an extreme strategy with very strong AD and very reduced or nil BI. Both maize and sun¯ower can be regarded as species with low RE notwithstanding the substantial proportion of axillary buds that became reproductive, but remain inactive in mature maize plants (Kiesselbach, 1949). Comparison of maize plants with one and two ears in Fig. 2 shows how proli®cacy contributes to the stability of reproductive allocation in this species. Proli®cacy in maize, however, is largely in¯uenced by environmental and genetic factors (Prior and Russell, 1975; Motto and Moll, 1983; Otegui, 1995). The varying limits to reproductive plasticity among the species demonstrated in our study broaden Weiner's (1988) theory of linear, size-dependent reproductive output. CO N C L U S I O N Manipulation of plant density allowed us to generate a broad range of resource availability for individual plants, and therefore of plant size. Proportion of dry matter partitioned to reproduction was stable for mid size plants and decreased for small or extremely large plants. Harvest index stability was highest in soybean, intermediate in sun¯ower and lowest in maize. Dierences in reproductive strategies between soybean, maize and sun¯ower were re¯ected in dierences in both the threshold of plant dry matter for reproduction and the drop
468
Vega et al.ÐReproductive Allometry in Soybean, Maize and Sun¯ower
in reproductive partitioning with high availability of resources. Dierential stability of reproductive partitioning partially derives from contrasting patterns of meristem allocation. AC K N OW L E D G E M E N T S This work is part of a thesis submitted by CRC Vega in partial ful®llment for the requirements for the degree of MSc, Universidad de Mar del Plata. We thank Professor Weiner, Dr LT Evans and Dr P Laterra for helpful comments, and the ®nancial support from INTA, UMP, FundacioÂn Antorchas, Dekalb S.A, and CONICET, the Research Council of Argentina. CRC Vega held a scholarship from CIC (ComisioÂn de Investigaciones CientõÂ ®cas de la Provincia de Buenos Aires). VO Sadras, FH Andrade, and SA Uhart are members of CONICET. L I T E R AT U R E C I T E D Andrade FH, Vega CRC, Uhart SA, Cirilo AG, Cantarero M, Valentinuz O. 1999. Kernel number determination in maize. Crop Science 39: 453±459. Bonser SP, Aarssen LW. 1996. Meristem allocation: a new classi®cation theory for adaptive strategies in herbaceous plants. Oikos 77: 347±352. Charles-Edwards DA. 1982. Physiological determinants of crop growth. North Ryde: Academic Press. Cox WJ. 1996. Whole-plant physiological and yield responses of maize to plant density. Agronomy Journal 88: 489±496. Doebley J, Stec A, Hubbard L. 1997. The evolution of apical dominance in maize. Nature 386: 485±488. Donald CM, Hamblin J. 1976. The biological yield and harvest index of cereals as agronomic and plant breeding criteria. Advances in Agronomy 28: 361±405. Edmeades GO, Daynard TB. 1979. The relationship between ®nal yield and photosynthesis at ¯owering in individual maize plants. Canadian Journal of Plant Science 89: 585±601. Evans LT. 1994. Crop physiology: prospects for the retrospective science. In: Boote KJ, Bennett JM, Sinclair TR, Paulsen GM, eds. Physiology and determination of crop yield. Madison: ASA, CSSA, SSSA, 19±35. Fehr WR, Caviness CE. 1977. Stages of soybean development. Special Report 80. Ames, Iowa: Cooperative Extension Service, Agriculture and Home Economics Exp. Stn Iowa State University 11: 929±931. Fischer KS, Palmer AFE. 1984. Tropical maize. In: Goldsworthy PR, Fisher NM, eds. The physiology of tropical ®eld crops. Bath, Avon: J. Wiley & Sons Ltd., 213±248. Gardner WR, Gardner HR. 1983. Principles of water management under drought conditions. Agricultural Water Management 7: 143±155. Giord RM, Thorne JH, Hitz WD, Giaquinta RT. 1984. Crop productivity and photoassimilate partitioning. Science 225: 801±808. Hartnett DC. 1990. Size dependent allocation to sexual and vegetative reproduction in four clonal composites. Oecologia 84: 254±259. Hay RKM. 1995. Harvest index: a review of its use in plant breeding and crop physiology. Annals of Applied Biology 126: 197±216. Johnson EC, Fischer KS, Edmeades GO, Palmer AFE. 1985. Recurrent selection for reduced plant height in lowland tropical Maize. Crop Science 26: 253±260. Kiesselbach TA. 1949. The structure and reproduction of corn. Nebraska Agricultural Research Bulletin No 161. Lincoln: University of Nebraska. Klinkhamer PGL, De Jong TJ, Meelis E. 1990. How to test for proportionality in the reproductive eort of plants. American Naturalist 135: 291±300.
Klinkhamer PGL, Meelis E, de Jong TJ, Weiner J. 1992. On the analysis of size-dependent reproductive output in plants. Functional Ecology 6: 308±316. Loomis RS, Connor DJ. 1996. Crop ecology. Productivity and management in agricultural systems. Cambridge: Cambridge University Press. Marshall C, Watson MA. 1992. Ecological and physiological aspects of reproductive allocation. In: Marshall C, Grace J, eds. Fruit and seed production. Aspects of development, environmental physiology and ecology. Cambridge: Cambridge University Press, 173±202. Moot DJ, Wilson DR, McNeil DL. 1997. Validation of the principal axis model (PAM) and its application to genotype selection in ®eld pea (Pisum sativum L.) crops. Annals of Botany 79: 651±656. Motto M, Moll RH. 1983. Proli®cacy in maize: a review. Maydica 28: 53±76. Otegui ME. 1995. Proli®cacy and grain yield components in modern Argentinian maize hybrids. Maydica 40: 371±376. Paterniani E. 1981. In¯uence of tassel size on ear placement. Maydica 26: 85±91. Prihar SS, Stewart BA. 1991. Sorghum harvest index in relation to plant size, environment, and cultivar. Agronomy Journal 83: 603±608. Prior CL, Russell WA. 1975. Yield performance in non proli®c and proli®c maize hybrids at six plant densities. Crop Science 15: 482±486. Reekie EG, Bazzaz FA. 1987. Reproductive eort in plants. 1. Carbon allocation to reproduction. American Naturalist 129: 876±896. Rees M, Crawley MJ. 1989. Growth, reproduction and population dynamics. Functional Ecology 3: 645±653. Ritchie SW, Hanway JJ. 1982. How a corn plant develops. Special Report 48. Ames, Iowa: Cooperative Extension Service, Iowa State University of Science and Technology. Sadras VO. 1998. Variation in apical dominance and its implications for herbivory resistance, competitive ability and biomass partitioning. In: Otegui ME, Slafer GA, eds. Proceedings of the international workshop on physiological bases for maize improvement. Buenos Aires, 71±81. Sadras VO, Connor DJ. 1991. Physiological basis of the response of harvest index to the fraction of water transpired after anthesis. A simple model to estimate harvest index for determinate species. Field Crops Research 26: 227±239. Sadras VO, Bange MP, Milroy SP. 1997. Reproductive allocation of cotton in response to plant and environmental factors. Annals of Botany 80: 75±81. Samson DA, Werk KS. 1986. Size-dependent eects in the analysis of reproductive eort in plants. American Naturalist 127: 667±680. Schmid B, Polasek W, Weiner J, Krause A, Stoll P. 1994. Modeling of discontinuous relationships in biology with censored regression. American Naturalist 143: 494±507. Schneiter AA, Miller JF. 1981. Description of sun¯ower growth stages. Crop Science 21: 901±903. Sinclair TR. 1998. Historical changes in harvest index and crop nitrogen accumulation. Crop Science 38: 638±643. Sinclair TR, Bennett JM, Muchow RC. 1990. Relative sensitivity of grain yield and biomass accumulation to drought in ®eld-grown maize. Crop Science 30: 690±693. Solbrig O, Solbrig DJ. 1985. Size inequalities and ®tness in plant populations. Topics in Evolutionary Biology 1: 141±159. Spaeth SC, Randall HC, Sinclair TR, Vendeland JS. 1984. Stability of soybean harvest index. Agronomy Journal 76: 482±486. Thompson BK, Weiner J, Warwick SI. 1991. Size-dependent reproductive output in agricultural weeds. Canadian Journal of Botany 69: 442±446. Tollenaar M, McCullough E, Dwyer LM. 1994. Physiological basis of the genetic improvement of corn. In: Slafer GA, ed. Genetic improvement of ®eld crops. New York: Marcel Dekker, 183±236. Weiner J. 1988. The in¯uence of competition on plant reproduction. In: Lovett Doust J, Lovett Doust L, eds. Plant reproductive ecology: patterns and strategies. New York: Oxford University Press, 228±245. Weiner J, Thomas SC. 1992. Competition and allometry in three species of annual plants. Ecology 73: 648±656.