| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Northwestern Agricultural Research Center, 4570 MT 35, Kalispell, MT 59901
b Dep. of Statistics, Univ. of Nebraska, Lincoln, NE 68583
* Corresponding author (rns{at}montana.edu)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: CFI, Bentler's comparative fit index • NFI, Bentler's Normed index • NNFI, Bentler & Bonnet's Non-Normed index • SR, Standardized residual
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Crop–weed competition is a complex, dynamic process largely influenced by the response of the competing species to external factors such as soil moisture and nutrients and their interactions with the phenologic and physiologic events occurring within them (Callaway, 1992; Deen and Swanton, 2001). In view of this, the use of a mechanistic approach to assess crop–weed competition should render a comprehensive description of the phenomena.
Grain yield in cereals is often defined in terms of inflorescence per area, seeds per inflorescence, and seed weight, which are collectively referred to as yield components (Evans et al., 1975). These components develop sequentially (i.e., in an ontogenic manner) with late-developing components controlled by early-developing ones (Adams, 1967; Garcia del Moral et al., 1991; Dofing and Knight, 1992; Garcia del Moral et al., 2003). The reproductive ability of wild oat can also be defined on the basis of the yield component concept since the developmental rates between wheat and this weed have been found to be similar (Cudney et al., 1989).
The degree to which these individual yield components contribute to final yield varies and is governed by intraspecific competition for available resources. Factors governing the source–sink relationship in the plant and the resource-driven interrelations among them exert compensatory effects among the yield components and alter their relative contribution to grain yield (Adams, 1967; Bingham, 1969). Similar processes occur when weed competition is considered, but inter-plus intraspecific competitive effects alter the compensatory dynamics.
Linear structural modeling (i.e., path analysis) is a statistical tool for testing hypotheses of frameworks of cause–effect relationships. When using yield component data, path analysis allows measurement of the direct influence of one yield component on yield from the indirect influences caused by the mutual relationships among yield components (Dewey and Lu, 1959). Since the method is based on standardized coefficients, they are independent of original units and so causal relationships may be directly compared.
The objective of this study was to determine the causal mechanisms responsible for the differential competitive ability previously observed between wheat seed size classes and wild oat using a mechanistic approach based on yield and yield component characteristics from an ontogenic perspective. Such an approach should provide clues as to which developmental stages are most susceptible to competition, and by extension, the resources that are most limiting, and the plant characteristics involved in the phenomena. The results ultimately may provide insights as to how to manipulate the agroecosystem to reduce wild oat interference and fecundity as well as rendering practical information in defining realistic goals in breeding for improving crop competitiveness against weeds.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Treatments were established as in Stougaard and Xue (2004); however, the present study considered a reduced treatment set consisting of three wild oat densities (85, 170, and 340 plants m–2), two spring wheat seed size classes (large and small), and two spring wheat seeding rates (175 and 280 plants m–2) arranged as a complete factorial. Seed size classes were obtained by passing preconditioned certified grade seed over 2.7-, 2.3-, and 1.9-mm sieves. Large seed was that retained on the 2.7-mm sieve and the small seed was that which passed through the 2.3-mm sieve but was retained on the 1.9-mm sieve. Thousand-kernel weights for the large and small seed size classes were 43 and 23 g in 1999, and 45 and 27 g in 2000, and 44 and 24 in 2001. Spring wheat seeding rates were adjusted for differences in percentage germination among size classes. Wild oat seed was obtained from a local elevator, with the same seed lot used during each year of the study. Since the seed had been obtained via mechanical harvest, the awns, lemma and, palea were absent. Wild oat populations were determined on a pure live seed basis by measuring replicated 1000-kernel weights and adjusting for germination percentage.
The experimental design each year was a split-plot with wild oat densities representing the whole plot factor and spring wheat seed size and seeding rate combinations representing the subplot effect. Whole plot dimensions were 18 x 5 m and subplot treatments measured 3 x 5 m. Treatments were replicated four times. The study areas were fertilized and the wild oat density treatments broadcast over the respective areas before establishing the spring wheat treatments. The wild oat seed was then distributed throughout the soil profile by incorporating with a field cultivator to a depth of 7.5 cm in an effort to better represent the variable depths from which wild oat seed typically germinates. The sites were cultipacked to firm the seedbed and the spring wheat treatments seeded. Both species were seeded the same day in each year of the study.
Two permanent 0.14-m2 quadrats were placed in separate areas within each plot after seeding, assuring that equal numbers of crop rows were located within each quadrat. Spring wheat and wild oat plants were hand-harvested from each quadrat and combined for each plot before wild oat seed shatter near the wheat soft dough stage. Spike and panicle numbers were counted. The panicles were put in separate bags and placed in a forced air drier for 3 d at 38°C, after which seed was threshed from the panicles, weighed, and counted. After maturity, plots were combine harvested and the grain cleaned to remove excess chaff and straw. Wild oat seeds were removed from a 200-g grain subsample to determine clean spring wheat grain yield and 1000-kernel weight, after adjusting to 130 g kg–1 moisture. The variables measured in this study were spikes m–2 (1), kernels spike–1 (3), kernel weight (5), and grain yield (g m–2) (7) in spring wheat, whereas panicles m–2 (2), seeds panicle–1 (4), seed weight (6), and seed yield (g m–2) (8) were measured for wild oat.
Statistical Analyses
The data was initially tested for normality by the Shapiro-Wilk test (PROC UNIVARIATE, SAS Inst., 1999). The test failed to reject the hypothesis of nonnormality for some of the response variables, but a square-root transformation normalized the data in most instances. Hence, further analyses were performed on square-root transformed data. Analyses of variance within years gave similar residual mean squares for each of the response variables; thus it was assumed that the residual variability across years was homogeneous.
Analysis of variance combined across all factors was performed followed by an individual analysis of each seed size class. In both instances, a linear mixed model was used where all the effects in the model, except replications, were considered fixed. In each seed size class, means for the response variables were used to calculate simple Pearson correlation coefficients among them. These analyses were performed by PROC MIXED and PROC CORR from the SAS software (SAS Inst., 1999).
Path models were developed for each seed size class from crop and weed data on the basis of a variance–covariance matrix among the response variables. Path analysis has been used in crop–weed interaction studies (Pantone et al., 1992; Ogg and Seefeldt, 1999; Ni et al., 2000); however, the validity of the cause-effect models used in these previous instances has generally received little attention. In these models, it was assumed that the yield components develop in an ontogenic manner following the sequence inflorescences m–2, seed inflorescence–1, seed weight, and that the relationships among them are linear (Garcia del Moral et al., 1991; Dofing and Knight, 1992). Since spring wheat spikes m–2 and wild oat panicles m–2 are the first yield components to develop, they were considered exogenous variables, i.e., the effect between them could be reciprocal. The correlations between the two variables were –0.65 and –0.69 (P < 0.01) in large and small seed classes, respectively.
These analyses were performed by the CALIS procedure from the SAS software (Hatcher, 1994). In this statistical algorithm, estimation of the parameters is made by the maximum likelihood method. A Chi-square goodness-of-fit test, Bentler's Comparative Fit Index (CFI), Bentler & Bonnett's Non-normed Index (NNFI), Bentler & Bonnett's Normed Index (NFI), first-moment analysis of asymptotically standardized residuals (SR), and amount of variance explained by each endogenous variable were used as criteria for the identification of models with acceptable fit to the data. Differences in path coefficient values between seed size classes were compared by a t test statistic in which it was assumed unequal variances existed between seed size classes.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
In the analyses of variance within each seed size class, interactions between year, seeding rate, and wild oat density effects were generally nonsignificant (Tables 1 and 2). In addition, with few exceptions the F values of these interactions were substantially lower compared to those of the main effects, indicating that interactions accounted for only a small fraction of the total variance in each analysis (Tables 1 and 2). On the basis of these initial results, path analysis in each seed size class was conducted using a cross-classification data set containing 18 observations (year x seeding rate x wild oat density).
|
|
|
|
|
|
Although each yield component had a direct positive effect on wheat yield, several indirect effects regulated the overall response. For example, while an increase in spikes m–2 contributed positively to grain yield, there was a corresponding reduction in kernels spike–1, and to a lesser extent, kernel weight, which tended to offset the positive contribution of spikes m–2 to grain yield (Table 4). As before, these intraspecific compensatory mechanisms tended to decrease in an ontogenic manner and were of greater magnitude for plants developed from large seed than for small.
Intraspecific Compensatory Processes in Wild Oat
As was the case with spring wheat, inflorescences per unit area (panicles m–2) had the greatest impact on wild oat yield (Table 5). However, the relative contributions of seeds panicle–1 and seed weight to wild oat yield were more similar in magnitude as compared with spring wheat. This demonstrates that the two later formed wild oat yield components contributed in a similar manner to yield and is indicative of greater overall compensatory plasticity relative to that observed with wheat.
|
Several negative indirect effects regulated the response of the individual yield components. The positive direct effect of panicles m–2 on wild oat yield was counterbalanced by a reduction in seeds panicle–1. And while seeds panicle–1 also had a positive direct effect on wild oat yield, there was a corresponding reduction in wild oat seed weight. In all instances, these values were of greater magnitude when wild oat was grown in the presence of wheat derived from small seed, further demonstrating that intraspecific compensation mechanism were stronger when wild oat was grown in competition with wheat derived from small seed.
The Effect of Spring Wheat Competition on Wild Oat
The mechanisms by which spring wheat affected the reproductive ability of wild oat differed between wheat established from large and small seed (Fig. 3 and 4). Wheat derived from small seed negatively affected wild oat yield during the early stages of development, whereas plants established from large seed reduced wild oat yield by affecting early as well as later developing yield components. Nonetheless, spikes m–2 was principally responsible for the observed response with both seed size classes (Table 6).
|
These compensatory mechanisms in response to reductions in panicles m–2 were more clearly evident with the direct effects of wheat spikes on seeds panicle–1 and seed weight. In particular, the reduction in wild oat seed weight figured prominently in the overall model, and was the only direct effect in the model to be significant for each seed size class. The impact of spikes m–2 on wild oat seed weight reduction was almost two-fold greater for wild oat grown in the presence of wheat established from large seed as compared to small. This suggests that the associated improvement in wheat competitive ability attributed to large seed size negatively affected wild oat throughout the entire life history of the weed (Table 6).
The negative direct effect of wheat spikes m–2 on wild oat seed weight was accentuated by the negative indirect effect spikes m–2 exerted on wild oat seed weight via seeds panicle–1 and kernels spike–1. In addition, there was a significant positive direct effect of kernels spike–1 on wild oat seed weight for wheat derived from small seed. The reduction in spikes m–2 because of severe wild oat competition, and the corresponding increase in kernels spike–1, resulted in a positive direct association between kernels spike–1 and seed weight when wild oat was grown in the presence of wheat derived from small seed.
Although spikes m–2 had substantial effects on wild oat seed weight, minor effects also were observed for seeds per panicle. As noted above, there was a positive indirect effect of spikes m–2 on wild oat seed yield via seeds panicle–1 (Table 6). As spikes per square meter increased, panicles m–2 decreased, and seeds panicle–1 increased in response. This same mechanism also was reflected in the positive direct effect of wheat spikes m–2 on wild oat seeds panicle–1. In both instances, coefficient values were greater for wild oat plants grown in the presence of wheat derived from small seed.
While spikes m–2 were largely responsible for reductions in wild oat seed yield, later-formed wheat yield components also contributed to the response. The effects were more apparent for plants derived from large seed and were reflected in the paths associated with wheat kernels spike–1 and kernel weight on wild oat seed yield. The direct effects were nonsignificant, but the indirect effects that contributed most to the total correlation were associated with wild oat seed weight, providing additional evidence as to the importance of seed weight as a determinant of wild oat yield.
Overall, wheat established from large seed had the greatest negative impact on wild oat seed yield and more completely suppressed wild oat compensatory mechanisms. Although the indirect effects of wheat spikes on wild oat yield were substantial, the direct effect of spikes on wild oat seed weight was of greater importance, demonstrating that wheat established from large seed negatively impacted wild oat at later stages of development.
The Effect of Wild Oat Competition on Spring Wheat
Wild oat negatively affected wheat derived from small seed by affecting early as well as later developing yield components. In contrast, plants established from large seed were generally affected only during early stages of development, and demonstrated a greater capacity to counter wild oat interference via compensatory mechanisms. As was the case for crop interference effects, the first-formed wild oat yield component, panicles m–2, was largely responsible for the reduction in wheat grain yield with both seed size classes (Table 7).
|
While plants established from large seed were able to compensate for the negative indirect effect of panicles on spike production, such was not the case for plants derived from small seed. Moreover, the indirect effects involving the three wheat yield components on grain yield were negatively affected by panicles m–2, suggesting that wheat plants derived from small seed were vulnerable to wild oat competition throughout their entire life history.
Panicles m–2 had a substantial negative direct effect on kernels spike–1. This path was the only direct effect in the model to be significant for both seed size classes and was of greater magnitude for plants established from small seed compared with large seed. The susceptibility of this yield component in the small seed size class also was substantiated by the negative direct effect of seeds panicle–1 on kernels spike–1. In total, the negative direct effects of wild oat panicles m–2 and seeds panicle–1 on wheat kernels spike–1 were numerically the largest values in the model for plants established from small seed.
Additional negative indirect effects associated with kernels spike–1 further contributed to this response within the small seed size class. These consisted of the negative indirect effects of panicles m–2 on wheat grain yield, panicles m–2 on kernel weight, and seeds panicle–1 on kernel weight. Although these indirect effects were relatively minor, the results provide additional evidence that wheat kernels spike–1 were highly sensitive to wild oat competition when established from small seed.
Kernels spike–1 were also affected in plants derived from large seed, but to a lesser extent relative to plants established from small seeds. The negative direct effect of panicles m–2 on kernels spike–1 in the large seed size class was more than offset by a positive indirect path associated with wheat spikes m–2. A similar response was observed for wheat established from small seed, but was not of sufficient magnitude to override the negative effect of panicles m–2 on wheat kernels spike–1. The other instances where kernels spike–1 were affected in plants derived from large seed were relatively minor and were associated with negative indirect effects involving seeds panicle–1 and grain yield and with seeds per panicle and kernel weight.
While wild oat competition had significant effects on wheat spikes and kernels spike–1, kernel weight was largely unaffected. However, panicles m–2 and seeds panicle–1 both had direct effects on kernel weight, with the response being greatest for plants derived from small seed. While the coefficients differed between seed size classes, neither direct effect was significant within the respective seed size class, suggesting that the response was of minor importance.
Overall, wild oat panicles were largely responsible for wheat grain yield reductions with both seed size classes. These effects occurred indirectly through a reduction in wheat spikes m–2 and directly through a reduction in kernels spike–1. However, the overall effects were less for plants derived from large seed, which demonstrated a greater ability to counter wild oat interference via compensatory mechanisms. In contrast, wheat derived from small seed was more vulnerable to wild oat interference, with the negative direct effects of panicles m–2 and seeds panicle–1 on kernels spike–1 being major pathways for plants established from small seed.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Regardless of the seed size class, inflorescences area–1 accounted for the major route of interference in the derived models (Fig. 3 and 4). Previous research investigating the competitive interactions between cereals and wild oat has shown strong negative reciprocal effects associated with inflorescence area–1 (Peters, 1985; Morishita and Thill, 1988; Cudney et al., 1989). Our results also demonstrated that the first-formed yield component was largely responsible for interference effects. However, our findings showed that the reduction in inflorescences initiated a series of compensatory mechanisms causing competitive processes to operate late in the development of both species.
In the current study, wheat spikes and wild oat panicles were considered as exogenous variables and their reciprocal effects could not be directly tested. However, the correlation values were –0.65 and –0.69 (P < 0.01) with the large and small seed size classes, respectively. As such, it is somewhat surprising that the direct negative effect of panicles and spikes on the yield of the competing species was minor and instead occurred as a result of indirect pathways. In fact, spikes and panicles each accounted for the major pathways of interference but principally as a result of competitive effects on later-formed yield components. This response indicates that each species was able to partially compensate for the reduction in inflorescences by allocating resources to the later-formed yield components, which in turn were negatively affected by competitive processes.
In general, the direct negative effect of spring wheat on wild oat occurred as a result of reductions in seed weight, while the direct negative effect of wild oat on spring wheat occurred as a result of reductions in kernels spike–1. Paradoxically, wild oat seeds panicle–1 were largely unaffected by wheat interference, and wheat kernel weight was immune to the effects of wild oat competition. This apparent contradiction may reflect the survival mechanisms utilized by each species. As resources become limiting, wild oat appears to allocate resources to ensure seed number is maximized, while compensatory mechanisms in wheat preferentially favor seed weight. In short, wild oat and wheat utilize quantitative and qualitative reproductive strategies, respectively. In the case of wheat, this mechanism is probably the indirect result of selection procedures aimed at increasing grain yield.
This differential response also may have its basis in endogenous variables measured. Although the models generally explained a large portion of the variance in each endogenous variable, such was not the case for kernel weight and seeds per panicle. The small values for the coefficients of determination, and the large residuals for these variables, suggests that other factors were affecting these yield components. In a study on crop–weed interactions in rice (Oryza sativa L.) using path analysis, Pantone et al. (1992) speculated that elements in the path with large residuals might be largely dependent of factors outside the framework considered. Which is to say, other latent variables that were not measured in the current study may have had an effect on kernel weight and seeds panicle–1.
Further research is warranted to ascertain the causal factors affecting these yield components. Seeds panicle–1 are chronologically determined by development of the first, second, and higher order branches of the panicle, spikelet initiation, floret initiation, and the number of fertile florets spikelet–1 (Landes and Porter, 1990). Quantifying these additional components may provide further insights as to the importance of seeds panicle–1 as a yield determinant and its sensitivity to interspecific competition. With respect to wheat kernel weight, our results corroborate previous research that has determined this yield component to be highly stable across a wide range of environments, plant densities, and spatial arrangements (Lebsock and Amaya, 1969; Bush and Kofoid, 1982; Joseph et al., 1985; Garcia del Moral et al., 1991). Nonetheless, numerous factors potentially affect kernel weight and a more detailed assessment of these variables would be of benefit.
It is of interest to note that wild oat seed weight was highly sensitive to wheat interference. Moreover, the fact that the first-formed wheat yield component negatively affected the last-formed wild oat yield component at first seems conflicting. However, seed weight reflects the cumulative effects of all previously formed yield components and the biotic and abiotic factors affecting them. Consequently, both pre- and postanthesis conditions can affect the expression of this trait.
While the factors affecting wild oat seed weight have not been investigated to the same degree as in cereal crops, similar processes are likely operative. Seed weight is largely determined at the postanthesis period and is a function of the grain filling rate and duration (Nass and Reiser, 1975; Simmons et al., 1982; Wych et al., 1982; Bauer et al., 1985; Coventry et al., 2003). Competitive processes that delay wild oat development, in particular heading date, could shorten the grain filling period and correspondingly reduce seed weight. Sharma et al. (1977) found that low light intensities severely delayed wild oat heading. As such, enhanced crop canopy development attributed to wheat spikes in the large seed size class could delay wild oat heading and reduce the grain filling duration, negatively affecting seed weight.
Concurrently, any delay in phenological development could cause wild oat grain filling to occur under less favorable environmental conditions, accentuating the effect. Postanthesis development coincides with the highest annual temperatures, increasing the potential for heat and water stress, both of which have been shown to negatively impact seed weight in a number of cereals (Chowdhury and Wardlaw, 1978; Wiegand and Cuellar, 1981; Wych et al., 1982; Bruckner and Frohberg, 1987). These same environmental conditions affect wild oat seed weight. Adkins et al. (1987) observed greater reductions in wild oat seed numbers and seed weights when plants were grown under high temperature conditions compared to low, while O'Donnell and Adkins (2001) reported that moisture stress significantly reduced wild oat seed production.
The effects of postanthesis stress on seed weight can partially be compensated for by the remobilization of stored assimilates (Evans and Wardlaw, 1976). This compensatory translocation is an important mechanism, as the reported contribution of stored assimilates to cereal grain yield ranges from 20 to 70% (Gallagher et al., 1975). Any factor that reduces photosynthesis will correspondingly limit vegetative biomass and the amount of carbohydrates available for remobilization, negatively affecting seed weight and yield. Toward that end, the effect of light intensity has received considerable attention (Judel and Mengel, 1982; Thorne and Wood, 1987). Sharma et al. (1977) found that low light intensities severely reduced wild oat shoot dry weights. Reductions in leaf and tiller numbers each contributed to this response, with the latter being affected the most. As such, seed weight also could be reduced if early season competitive effects caused a reduction in the storage of assimilates, principally because of reductions in wild oat secondary tillers.
Delayed phenological development and the lack of preanthesis assimilate accumulation are largely associated with competition for light. Therefore, it seems plausible that enhanced crop canopy development attributed to wheat spikes from plants established from large seed more effectively shaded wild oat and negatively affected these processes as compared to plants established from small seed. Hence, the strong negative association between wheat spikes and wild oat seed weight.
Overall, our results suggest that the impact of wheat competition on wild oat seed weight is a major route of interference. This effect is potentially exploitable, having practical applications from a weed management perspective. In greenhouse experiments, Sharma et al. (1977) found that wild oat plants produced from small seed had lower dry weights than plants produced from large seed. Similarly, Peters (1985) found that wild oat plants established from small seed had delayed emergence, fewer main stem leaves, fewer tillers, produced less seed, and resulted in reduced dry weights compared to plants derived from large seed. When sown at a depth of 25 mm, wild oat plants established from small seed reduced barley dry weight by 26%, compared with 44% for plants established from large seed.
In short, wild oat populations in which the majority of seedlings arise from small seed would potentially be less competitive than populations established from large seed. This could have long-term repercussions in terms of the degree of yield loss and the associated management intensity required to minimize wild oat interference. While the development of competitive cropping systems could foster this goal, additional field research is needed to quantify the effects of wild oat seed size on seed bank demographic processes and competitive ability.
Received for publication August 22, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
