| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Agronomy, Iowa State Univ., Ames, IA 50011
b Dep. of Statistics, Iowa State Univ., Ames, IA 50011
c Dep. of Ecology, Evolution, and Organismal Biology, Iowa State Univ., Ames, IA 50011
d Dep. of Crop and Soil Sciences, Center for Applied Genetic Technologies, Univ. of Georgia, Athens, GA 30602. This journal paper of the Iowa Agric. and Home Econ. Exp. Stn., Ames, IA, Project No. 6631, was supported by the Hatch Act and State of Iowa funds
* Corresponding author (vpicasso{at}iastate.edu).
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: A1, Ames–one harvest • A3, Ames–three harvests • AIC, Akaike's Information Criterion • B1, Boone–one harvest • B3, Boone–three harvests • ISU, Iowa State University
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Received for publication April 23, 2007.
a Dep. of Agronomy, Iowa State Univ., Ames, IA 50011
b Dep. of Statistics, Iowa State Univ., Ames, IA 50011
c Dep. of Ecology, Evolution, and Organismal Biology, Iowa State Univ., Ames, IA 50011
d Dep. of Crop and Soil Sciences, Center for Applied Genetic Technologies, Univ. of Georgia, Athens, GA 30602. This journal paper of the Iowa Agric. and Home Econ. Exp. Stn., Ames, IA, Project No. 6631, was supported by the Hatch Act and State of Iowa funds
* Corresponding author (vpicasso{at}iastate.edu).
Species diversity can increase natural grasslands productivity but the effect of diversity in agricultural systems is not well understood. Our objective was to measure the effects of species composition, species richness, and harvest management on crop and weed biomass in perennial herbaceous polycultures. In 2003, 49 combinations of seven species (legumes, C3 and C4 grasses) including all monocultures and selected two to six species polycultures were sown in small plots at two Iowa, USA, locations in a replicated field design. Plots were split in half and managed with either one or three harvests in each of 2004 and 2005. Biomass increased log-linearly with species richness in all location-management environments and the response was not different between managements. Polycultures outyielded monocultures on average by 73%. The most productive species in monoculture for each management best explained the variation in biomass productivity. The biomass of plots containing this species did not increase with richness in most environments but biomass of plots without this species increased log-linearly in all cases. Weed biomass decreased exponentially with richness in all environments. On average, increasing species richness in perennial herbaceous polycultures increased productivity and weed suppression, but well-adapted species produced high biomass yield regardless of richness.
Abbreviations: A1, Ames–one harvest • A3, Ames–three harvests • AIC, Akaike's Information Criterion • B1, Boone–one harvest • B3, Boone–three harvests • ISU, Iowa State University
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The contribution of plant species diversity to productivity and other ecosystem functions is a controversial issue in ecology (Loreau et al., 2001). Species diversity may refer to the number of species present in an area (i.e., richness) and/or their relative abundance (i.e., evenness), but most studies use species richness as a proxy for diversity. The current consensus among ecologists is that ecosystem functions are influenced by individual species' traits, complementarity among them, and environmental factors (Hooper et al., 2005). Although increasing diversity can have a range of effects on ecological processes depending on species composition and environmental context, diversity can increase productivity because of (i) the major effect of one or few very productive species (Huston et al., 2000); (ii) positive interactions among species due to complementarity or facilitation (Tilman et al., 2006b); or (iii) a combination of both (Loreau and Hector, 2001). In most experiments with randomly assembled plant communities from natural grasslands, plots with greater species diversity tend to outyield plots with low diversity (Hector et al., 1999; Tilman et al., 2006b). However, diversity may not change ecosystem functions that are controlled primarily by abiotic factors or by the dominance of a single species (Hooper et al., 2005). When environmental factors are held constant, increasing species richness generally decreases community susceptibility to invasion by weeds or other pests, because fewer resources are available to invaders and because species that are strongly competitive or that offer biotic control of a prospective invader are more likely to be included (Knops et al., 1999). Diversity may also increase ecosystem stability by reducing variability in response to environmental fluctuations and increasing resistance and resilience to perturbations (Loreau et al., 2002).
In agriculture, research on forage crop mixtures and on intercrops has focused on the role of diversity in productivity and other agronomic variables. The effect of diversity on biomass productivity depends on the particular system, species, and processes considered (Sanderson et al., 2004; Trenbath, 1974). Mixtures of forages have received considerable attention because domesticated pastures and hay fields often are seeded with multiple species and natural grazing lands contain a diversity of species. A review of the literature on forage mixture experiments shows that productivity can be maximized at either low or high diversity, but complex mixtures (i.e., mixtures with more than two species) can maximize more ecosystem functions at the same time, e.g., temporal distribution of production, persistence, resistance to invasion, tolerance to fluctuating environmental conditions, and positive impacts on water quality (Sanderson et al., 2004). Experiments using cattle grazing indicate that complex mixtures comprising grasses, legumes, and composites are more productive in dry years, and have less weed invasion than simple mixtures (Sanderson et al., 2005).
Most ecological research modeling the relationship between diversity and ecosystem function cannot be extrapolated directly to agriculture. First, ecologists typically use a relatively large number of native species in randomly assembled communities, without including "true replications" (sensu Huston and McBride, 2002) of the same species combinations in their experimental design. Second, the research is often conducted in low fertility soils and with a single management scheme, typically a one-time biomass harvest (see for example, Tilman et al., 1996). Finally, the experiments usually include just a few dominant species, which comprise most species in agriculture and which can have major effects on ecosystem functions (Hooper, 1997). In contrast, most agricultural research considers only a few mixtures of well-adapted species (Soder et al., 2007), and consequently does not provide a range of species richness to fit a productivity-richness model.
Our objective was to measure the effect of species composition, species richness, and harvest management on aboveground crop and weed biomass in perennial herbaceous polycultures in fertile agricultural soils in central Iowa. We tested the hypotheses that (i) crop biomass productivity increases with increasing species richness; (ii) weed biomass is reduced with increasing species richness; (iii) the presence or absence of certain species changes the slope of the regression of biomass production on species richness; and (iv) harvest management changes the slope of the regression of biomass production on species richness. Our intent was to bridge the gap between agronomic and ecological experiments by conducting the study under agricultural conditions (i.e., well-adapted species grown on fertile soils) using true replication and a set of mixtures large enough to model the relationship between diversity and ecosystem function.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
Plots were mowed on 18 June and 12 Sept. 2003 in Boone, and on 20 June and 13 Sept. 2003 in Ames, to a 0.15-m height to control weeds; biomass was not measured or removed. In October 2003, we measured plant establishment by counting the number of plants of each species present along a 1-m transect per plot. Eastern gamagrass was reseeded by hand in April 2004, because no plants were found in 2003.
In 2004, each plot was split in half to form two 2 by 3 m subplots that were allocated to either a three-harvest management, simulating a hay system with removal of all biomass, or to a one-harvest management, simulating a perennial grain system with only seed biomass of selected species being removed. The same allocations were also used in 2005. These two contrasting managements allow us to compare a forage system where all biomass is removed (three harvests) with a perennial grain crop system, where most nutrients remain on the plots (one harvest). Plots were machine clipped with a flail-type harvester (Carter Mfg., Brookston, IN) equipped with an electronic weigh system. A single 1-m-wide by 3-m-long strip was harvested for biomass through the center of each small plot. The adjacent 50-cm strips on either side of the measured area were cut immediately after data collection so that all forage was clipped to ground level.
Plots in the hay management were harvested on 26 May, 13 July, and 13 Sept. 2004 and 25 May, 8 July, and 22 Aug. 2005 in Boone, and on 28 May, 15 July, and 16 Sept. 2004 and 3 June, 27 July, and 15 Sept. 2005 in Ames. In the grain management, the reproductive structures of the four species with the highest seed yield and most easily harvested seeds (Illinois bundleflower, orchardgrass, intermediate wheatgrass, and switchgrass) were hand-harvested from the entire plot area as each species matured, dried, weighed, and added to the total plot biomass. Although some seed was produced by alfalfa, pollinators were infrequent in the plots and the seed yield was minimal. Biomass on these plots was harvested once at the end of the growing season, on 8 Nov. 2004 and 4 Oct. 2005 in Boone, and 9 Nov. 2004 and 27 Oct. 2005 in Ames.
Before each harvest, biomass was sampled by clipping two 0.09-m2 quadrats per plot. Samples were weighed fresh, separated into component species in the laboratory, dried, and weighed dry. The dry matter percentage of the samples and the fresh weight of the machine harvested strips were used to calculate total biomass dry weight per square meter. The relative proportion of each species was calculated and used to differentiate biomass productivity of the seeded species versus weeds. Due to field labor constraints, samples were not collected immediately before the third harvest in the three-harvest management in 2004 at Ames although machine harvest was conducted. Therefore, because data on the relative proportion of each species and of weeds were missing for this harvest, the data were dropped from the analysis and the total yield for the hay management in Ames in 2004 comprised only the first two harvests.
The dominant species in each plot was defined as the seeded species with maximum biomass production in the plot. The Berger–Parker dominance index, which represents the proportion of whole plot biomass produced by the dominant species, was calculated for each plot (Wilsey and Polley, 2004).Values of the dominance index close to one indicate communities that are dominated by a single species.
In this paper, "biomass of seeded species" refers to aboveground plant biomass of the species deliberately seeded into the plot; "biomass of cultivated weeds" refers to seeded species that occurred in plots where they were not intended, either because seeds were in the soil seed-bank or carried from adjacent plots by wind, rodents, birds, or the planter; "biomass of wild weeds" refers to species outside the seeded list. We define "biomass of cultivated species" as the sum of the biomass of seeded species plus the biomass of cultivated weeds; "weed biomass" is the sum of biomass of cultivated and wild weeds; and "total biomass" of a plot comprises biomass of seeded species plus weed biomass. Although crop species diversity comprises species richness and evenness, in this experiment we manipulated only crop species richness as a proxy for crop diversity.
Data Analysis
To determine differences among entries and relevant interactions, an overall analysis of variance for biomass was performed including location, management, year, and species combinations as fixed effects and replication and block within replications as random effects. If interaction effects were significant, further analyses by location, management, or year were conducted. The yield of monoculture plots was separated using Fisher's protected least significant difference (LSD) with 
= 0.05.
A series of nine linear models of biomass of seeded species as the dependent variable and seeded species richness as the independent variable and transformations of each variable using natural logarithm and inverse functions were compared using R2 and Akaike's Information Criterion (AIC) statistics (data not shown). Biomass of seeded species as a function of the natural logarithm of seeded species richness was the best model using these criteria and was used in all further analyses. The linear effect of seeded species richness on seeded biomass, cultivated biomass, weed biomass, and total biomass was tested with orthogonal contrasts.
To determine the relative importance of the presence or absence of a single species and the importance of seeded species richness on the biomass of seeded species, we used a model selection procedure as described by Deutschman (2001). Several mixed models were constructed using biomass of seeded species as the dependent variable, with entries nested within the following explanatory variables, which were considered to be fixed effects: (i) the presence or absence of each species individually, (ii) the natural logarithm of seeded species richness, (iii) both the natural logarithm of seeded species richness and the presence or absence of the single species that individually best explained the variation in biomass, and (iv) both variables included above together with their interaction. Models were compared using AIC, where models with smaller AIC values indicated a better fit. The single species that best explained the variation in biomass within each environment was denoted the "driver species" for that environment. We used orthogonal contrasts to test the linear effects of logarithm of seeded species richness on biomass of seeded species in plots with and without the driver species. If the linear contrasts were significant, we calculated the regression coefficients, standard error, and their 95% confidence intervals from the individual plot data. Two slopes were considered different from each other if their 95% confidence intervals did not overlap. To test for functional group composition effects, orthogonal contrasts in each environment were performed among plots with different functional group composition.
For the weed biomass data, a similar set of analyses was conducted to that described for biomass of seeded species. The best model for total weed biomass as a function of seeded species richness across all environments was an exponential function (i.e., natural logarithm of weed biomass as a linear function of seeded species richness). The same model selection procedure, based on the AIC values, was conducted to determine the relative importance of the presence or absence of single species and of seeded species richness on total weed biomass. Linear effects of seeded richness were then tested with orthogonal contrasts on entries with the most weed suppressive species and on those without the most weed suppressive species.
All analyses of variance and contrasts were performed using PROC MIXED, and regression coefficients were calculated using PROC REG, in the SAS statistical software package (SAS Institute, 2003).
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The observed number of seeded species increased linearly as actual-seeded richness increased (R2 = 0.69 overall, 0.76 for B1, 0.73 for B3, 0.68 for A1, and 0.59 for A3; P < 0.0001 in all cases). Therefore, seeded species richness was a good estimator of observed seeded species richness. The species which were most often missing were eastern gamagrass and switchgrass, particularly at higher seeded richness levels. Because the analyses with seeded and observed richness produced very similar results, we report the results using seeded richness to be able to make planned comparisons based on the original experimental design. Evenness was not controlled experimentally and all mixture plots showed a high level of dominance (Table 3 ). The species that dominated most plots based on the Berger–Parker dominance index were alfalfa for B3, orchardgrass for A1, and both alfalfa and orchardgrass for B1 and A3 (Table 3).
|
|
|
|
|
Cultivated weed biomass, wild weed biomass, and total weed biomass decreased exponentially with seeded richness in all environments (Fig. 1 and 3 ). Models of natural logarithm–transformed weed biomass as a function of seeded species richness had higher AIC values than models including only the most weed suppressive species in each environment (Table 5). Therefore, the impact of particular species on weed suppression was greater than richness per se. As with biomass, of the 10 models compared using AIC (Table 5), the best was a factorial model of seeded species richness and the most weed suppressive species. For B1, the best model also included the interaction between species richness and the presence or absence of the most weed suppressive species, but for the other three environments, the best model did not include the interaction.
|
A strong negative exponential relationship between total weed biomass and biomass of seeded species was observed in each environment. All linear regressions of natural logarithms of total weed biomass as a function of biomass of seeded species were significant at P < 0.0001. Equations for total weed biomass (y) as a function of biomass of seeded species (x) were: y = 70.8(0.997)x (adj. R2 = 0.17) for B1; y = 213.0(0.997)x (adj. R2 = 0.34) for B3; y = 174.4(0.995)x (adj. R2 = 0.30) for A1; and y = 304.7(0.995)x (adj. R2 = 0.31) for A3.
Functional Composition Effects on Biomass of Seeded Species and Weeds
Plots including both C3 grasses and legumes outyielded plots with only legumes or only C3 grasses in almost all environments (Table 6
). The two exceptions were in B3, where legume-only plots outyielded legume–C3 grass mixtures, and in A1, where C3 grasses were no different from legume–C3 grass mixtures. As a group, legumes outyielded C3 grasses in B3, and C3 grasses outyielded legumes in the one-harvest management at both locations. The contribution of C4 grasses to biomass production was marginal overall during the period of measurements of this study, as expected for newly establishing warm-season grasses. Adding a C4 grass to a plot with legumes and C3 grasses did not change the productivity in any environment.
|
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
However, a deeper analysis of the data revealed that the highest yields were from plots including a single driver species, and that plots at any species richness (even monocultures) that included this species had nearly equivalent yields and were the highest of those observed for a given environment. This is also consistent with a recent meta-analysis of ecological research: yields of diverse polycultures do not differ from the best yielding monoculture (Cardinale et al., 2006).
The diversity–productivity relationship can be explained by complementarity among species or by selection effects (Loreau and Hector, 2001). Different species with complementary traits (e.g., rooting depth) can use different resources or niches, providing the community as a whole access to more resources and making it more productive than its constituent species individually. Positive interactions among species can also increase the performance of the community, a process called facilitation. Selection effects are apparent when highly productive species dominate a mixture due to processes such as interspecific competition (Loreau and Hector, 2001). Selection effects can result from sampling species for inclusion in randomly assembled communities (Huston and McBride, 2002), because the probability of including a highly productive species in the mixture increases as the number of species in the mixture increases.
In our experiment, species combinations were not assembled at random, but rather designed so that each level of species richness had plots with and without each species. This design does not avoid the sampling effect, because the proportion of plots with each single species has to increase with increasing species richness. However, we were able to separate the individual species effects from richness effects. We showed that individual species had major effects on productivity: the presence or absence of alfalfa in the three-harvest management and intermediate wheatgrass in one-harvest management better explained biomass productivity than did species richness. This suggests that the relationship between biomass productivity and species richness may change depending on the presence or absence of certain "driver" species. These driver species are well adapted to the agronomic environment (soils, climate, and management) and are highly productive. The driver species can dominate the plant community when it has a higher competitive ability than other species (e.g., alfalfa in B3) but in other situations, other species that are less productive but more competitive can dominate instead (e.g., orchardgrass in A1).
The inclusion of a driver species in the polyculture is the main factor that explained the increase in average productivity of the more species-rich communities. Polycultures where the driver species was present did not show an increase in biomass productivity with species richness. However, biomass productivity increased with species richness in the absence of the driver species. Seeding a monoculture or a binary mixture of the best adapted, highest yielding species is easier to manage and may be a better option than seeding a complex polyculture in some circumstances (e.g., Tracy and Sanderson, 2004b). Nevertheless, polyculture yields were typically as high as the best monoculture, and in some environments were even higher (e.g., A3). Because increasing species richness may offer other benefits than simply biomass productivity, as we discuss below, mixture planting may be a better option.
The results of this study also support our second hypothesis that weed biomass is reduced with increasing species richness. These findings are consistent with the hypothesis that resident diversity increases the competitive environment and makes invasions by other species more difficult (Elton, 1958). Our results are limited to the context of the species, soil types, and high weed pressure of the agricultural conditions studied. Nevertheless, these results add to the experimental evidence supporting the contention that plant species richness increases resistance of ecosystems to weeds and other pests (Dukes, 2002; Knops et al., 1999; Tracy and Sanderson, 2004a). Highly weed-suppressive species had strong effects on weed invasion; some monocultures (alfalfa in Boone, orchardgrass in Ames) were as effective at suppressing weeds as polycultures. However, the same species may not maximize both biomass productivity and weed suppression simultaneously. In our experiment, the alfalfa monoculture in B3 maximized both productivity and weed suppression, but in the other three environments, different species maximized each function (combinations of intermediate wheatgrass, alfalfa, and orchardgrass). This suggests that as more agronomic and ecological functions are considered, species-rich polycultures may simultaneously optimize more functions, and may be more beneficial than simple monocultures (Sanderson et al., 2004).
In this paper we reported results from the second and third year after establishment of perennial plant communities. Although for these 2 yr the trends of productivity vs. richness were the same, we expect that succession in further years may change species richness, the evenness of the community, the identity of driver species, and the slope of the productivity vs. richness regression. The potential of polycultures to maintain stable productivity over time under agricultural conditions is a relevant issue to be explored in further years of this experiment.
Our final hypothesis that harvest management changes the relationship between productivity and richness can be rejected because the overall trends were similar for both management strategies (i.e., they had equal slopes [Fig. 1 and 2]). This means that species-rich plots had higher yields and lower weed invasion under both one- and three-harvest managements, and that the increase in yield with richness was similar for both managements. These findings suggest that species-rich polycultures should be considered even in intensively harvested agroecosystems, such as herbaceous perennial mixtures harvested as biofuel feedstocks (Tilman et al., 2006a). However, the identity of the driver species changed for each management regime, suggesting that harvest management needs to be considered when choosing the species to include in a polyculture. Furthermore, in farming systems where management practices may change due to unplanned conditions such as weather or market fluctuations, polycultures may offer the advantage of more flexibility in management options, whereas monocultures are more rigid in their requirements.
Differences in biomass productivity among plots with different functional group composition can be explained by differences in species composition. For instance, in environments where alfalfa was the driver species, plots with legumes outyielded plots with C3 grasses, but where intermediate wheatgrass was the driver, plots with C3 grasses outyielded legume plots. Plots with a combination of legumes and C3 grasses in most cases were the highest yielding plots, as is well known in forage production. The marginal contribution of the C4 grasses is explained because C4 grasses tend to take longer to establish, and this is particularly the case in our experiment with Eastern gamagrass, which did not establish at all in the first year. We expect that the contribution of C4 grasses will increase in future years of the experiment.
In a sustainable agriculture context, crop communities need to simultaneously optimize various ecosystem functions, including, but not limited to, productivity. Although dominance effects explained the majority of the relationship between diversity and productivity in our study, the productivity of perennial monoculture and polyculture plots over longer time frames needs to be assessed. Because single species may have great effects on the relationship between diversity and ecosystem function, experiments where species effects cannot be discriminated from richness effects should be considered with caution. Controlling experimentally both richness and species composition increases the number of plots exponentially, necessitating the judicious choice of species adapted to the environments considered.
This experiment yielded two main conclusions. First, increasing species richness in perennial herbaceous polycultures has measurable benefits in terms of productivity and weed suppression. This finding, well acknowledged by ecologists, should be applied to the design of agriculture systems. Second, well-adapted species have major effects on the relationship between diversity and ecosystem function. This finding, more widely understood by the agricultural community, should inform more ecological theory. More research bridging these two scientific traditions is needed to advance toward a more sustainable agriculture.
This research was cofunded by grants from Raymond Baker Center for Plant Breeding at Iowa State University, Ames, IA, to E.C. Brummer; Fulbright-Uruguay Graduate Fellowship, NCR-SARE Graduate Student Project No. GNC05-055, and Natural Systems Agriculture Graduate Fellowship from The Land Institute, Salina, KS, to V. Picasso. The authors would like to thank Mark Smith and several students in the ISU forage breeding lab for field and lab assistance, and the Graduate Program in Sustainable Agriculture at Iowa State University.
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.
Received for publication April 23, 2007.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. W. Singer, K. A. Kohler, K. J. Moore, and D. W. Meek Living Mulch Forage Yield and Botanical Composition in a Corn-Soybean-Forage Rotation Agron. J., August 31, 2009; 101(5): 1249 - 1257. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. E. Frankow-Lindberg, C. Brophy, R. P. Collins, and J. Connolly Biodiversity effects on yield and unsown species invasion in a temperate forage ecosystem Ann. Bot., April 1, 2009; 103(6): 913 - 921. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M. Tobias A Genome May Reduce Your Carbon Footprint The Plant Genome, March 1, 2009; 2(1): 5 - 8. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
