Journal list menu
Plant community responses to integrating livestock into a reduced-till organic cropping system
Corresponding Editor: Troy W. Ocheltree.
Abstract
The problems with herbicide- and tillage-based weed management in agriculture are well documented and have precipitated research into finding alternatives. Integrating livestock grazing into organic agroecosystems has benefits and is a viable method for terminating cover crops, yet its impacts on weed communities are largely unknown. This lack of knowledge is particularly true in semi-arid environments, including the Northern Great Plains, where we conducted our research. We compared weed community responses (biomass, species richness, Simpson’s diversity, composition) of a sheep-grazed organic cropping system with those of two contrasting cropping systems (high input conventional no-till, tilled organic) across a five-year crop rotation (safflower, sweet clover, winter wheat, lentils, winter wheat). We found that the conventional no-till and tilled organic systems suppressed weed biomass and reduced species richness and diversity, while the grazed organic resulted in higher weed biomass, species richness, and diversity. During the first two years of the study, the composition of the two organic communities were distinct from the conventional no-till communities but were indistinguishable from one another. Over the final three years of the study, grazed organic communities were tightly grouped and became distinct from both the tilled and conventional communities. We found that weed biomass and diversity were highest in the sweet clover and lowest in the winter wheat. The spring annual crops, safflower and lentil, demonstrated similar weed biomass, species richness, and composition. Our findings indicate that integrating livestock into cropping systems alters plant communities and increases the agroecosystem plant biodiversity of semi-arid organic farming and that specific crops interact with cropping systems to alter agroecosystem plant communities. However, the increase in weed biomass associated with our grazing treatment makes this approach impractical as the sole weed management strategy and necessitates that integrating livestock into semi-arid organic cropping systems must be part of a larger integrated weed management program.
Introduction
Supplying agricultural yields of food, fuel, fiber, and feed through high input agricultural systems has consequences for soil health, ecosystem function, and cropland biodiversity (Vitousek et al. 1997, Stoate et al. 2001, Newbold et al. 2015, de Castro Marcato et al. 2017). This includes substantially reducing cropland weed abundance and diversity (Menalled et al. 2001). Weeds are merely plant species’ other than those intentionally sown in agroecosystems (Liebman et al. 2001) and, while abundant weed communities could reduce crop production and yields, diverse weed communities can benefit agroecosystems by increasing soil fertility, providing habitat for natural predators of crop pests, benefiting pollinators and increasing pollination, shaping microbial communities, and facilitating nutrient cycling (Jordan and Vatovec 2004, Ishaq et al. 2020). A significant reduction in weed diversity can have multitrophic consequences, impacting everything from belowground ecosystems (Culman et al. 2010), pollinator abundance and activity (Adhikari et al. 2019), to avian diversity and abundance (Marshall et al. 2003). The negative ecological effects of high input agricultural systems have amplified interest in organic cropping systems. Organic cropping systems promote crop production and soil fertility through ecosystem processes such as nutrient cycling, crop–weed competition, pollination, and by biologically based pest regulation (Altieri 1999, Tilman et al. 2002, Isbell et al. 2017, Islam and Ashilenje 2018). Maintaining agroecosystem biodiversity is vital to sustaining these processes (Altieri 1999, Jordan and Vatovec 2004, Robertson et al. 2014, Dainese et al. 2019) and is shaped by agronomic management (Booth and Swanton 2002, Smith and Mortensen 2017); therefore, it is imperative to understand the unique effects of specific organic management practices.
Organic cropping systems primarily rely on tillage, an agricultural practice that mechanically disturbs the soil in preparation for planting and for weed control. However, tillage decreases soil organic matter and soil water holding capacity (Matson et al. 1997) and facilitates wind and water erosion (Seitz et al. 2019). There is a growing interest in reducing tillage intensity in organic systems (Mirsky et al. 2012, Wallace et al. 2017). One of the principal methods used for weed control in minimum tillage organic system is cover cropping. Cover cropping utilizes non-marketable plants to increase soil organic matter, reduce erosion, improve soil nutrient retention, and suppress weeds, and is one of the most common practices in ecologically based cropping (Tonitto et al. 2006). However, there are multiple obstacles associated with depending on this method and other non-tillage methods for weed control in organic systems of water-limited regions (Carr et al. 2013). First, sufficient cover crop growth is required to effectively compete with weeds, which is difficult to obtain without depleting the soil moisture for subsequent cash crop growth (Townley-Smith et al. 1993, Tanaka et al. 1997). Second, no-till methods used to terminate cover crops in areas with sufficient moisture (i.e., roller crimping) are ineffective in arid and semi-arid environments because they require mature crops and delayed termination (Izard 2007, Mirsky et al. 2009, Carr et al. 2012), which reduces soil moisture. Third, the annual cover crops used in arid and semi-arid systems do not adequately suppress perennial weeds (Carr et al. 2013, Briar et al. 2019). Finally, in systems with no synthetic fertilizer inputs, cover crops are needed to improve soil health and fertility (Carr et al. 2012); however, the nutrients supplied by cover crops may require tillage to become bioavailable (Vaisman et al. 2011, Burgess et al. 2014).
Integrating livestock into cropping systems is when the two systems (livestock production and crop production) utilize the products of the other for production (Hilimire 2011). Integrating livestock into cropping systems is utilized in organic farming to increase agroecosystem biodiversity, reduce tillage frequency and intensity, and manage weeds and nutrients (Hilimire 2011, MacLaren et al. 2019, Adhikari and Menalled 2020, Peterson et al. 2020). In the semi-arid Northern Great Plains (NGP), an important agricultural region of central North America, research has demonstrated that livestock grazing can be an effective method for terminating cover crops (Hatfield et al. 2007a, b), controlling insect pests (Hatfield et al. 2007a), providing alternative revenue for farmers (McKenzie et al. 2017), and shaping ground beetle communities, while increasing their diversity and activity (Adhikari and Menalled 2020). However, research within the region on the effects of livestock grazing on cropland plant communities is limited and has produced contrasting results. In conventional small grain-based systems, Miller et al. (2015) and Barroso et al. (2015) found that incorporating livestock grazing resulted in higher weed cover and altered community composition. However, Miller et al. (2015) found grazing resulted in higher species richness, while Barroso et al. (2015) found that grazing did not affect species richness and had minimal effects on community diversity. In contrast to both these studies, another study from the same region found that sheep grazing did not impact weed abundance or community composition when integrated into organic vegetable production (McKenzie et al. 2016).
To evaluate the impact that integrating grazing into semi-arid organic cropping systems has on weed communities, we utilized sheep grazing for weed and nutrient management, and as a cover crop termination method over a five-year crop rotation. To get perspective on the relative effects of our grazing treatment, across the same five-year rotation we compared weed community responses with those of two common and contrasting cropping systems, a “conventional no-till” system that utilized herbicides and synthetic fertilizer and a tilled organic cropping system. Specifically, our objectives were to (1) assess how cropping systems affect weed communities (biomass, species richness, Simpson’s Diversity Index, and composition) over a five-year period; and (2) evaluate how crop phase (i.e., phase of the five-year rotation) interacted with cropping system to affect weed community response variables. We hypothesized that weed community response variables (biomass, species richness, Simpson’s diversity, and composition) would vary between the three cropping systems and through time. We further hypothesized that the crop phases would interact with cropping systems to affect the weed community response variables.
Materials and Methods
Site description and management history
A five-year cropping system experiment was conducted from 2013 to 2017 at the Montana State University Fort Ellis Research Farm (45°40′ N, 111°4′ W), located 6 km east of the Montana State University. The Fort Ellis site is a Blackmore silt loam soil type (a fine-silty, mixed-effects, superactive, frigid Typic Arguistoll). The climate of the study site is continental, with an average annual precipitation of 464 mm and a mean annual temperature of 6.2°C (WRCC 2020). The site, composed of the three blocks of our study, was used for pasture between 1994 and 2004 and largely consisted of perennial grasses Bromus inermis, Thinopyrum intermedium, and Poa compressa. In 2004, plots in each block were randomly assigned a wheat rotation (continuous spring wheat, spring wheat-fallow, winter wheat-fallow) and a management practice (sheep grazing, herbicide treatment, and tillage); these rotations were maintained until 2009. In 2009, plots were then randomly designated continuous spring wheat or alfalfa, under one of three management practices (sheep grazing, herbicide treatment, or tillage), with treatments being maintained until the beginning of our experiment in fall of 2012. For more on the history of our site prior to our study, see Barroso et al. (2015).
Experimental design, management practices, and data collection
Our experiment, beginning in fall 2012, had three replicate blocks, each 225 × 90 m. Within each block, one of the three cropping systems was randomly assigned to a 75 × 90 m plot. Each 75 × 90 m plot was divided into five split plots (13 × 90 m), each randomly assigned to a different starting phase of a five-year crop rotation, making all phases present in all years. To mitigate the effects of neighboring cropping systems, plots and split plots were separated by two-meter buffer zones, or areas maintained weed free with no sown crops.
The three crop management systems included a conventional no-till system, a tilled organic system, and a reduced-till sheep grazed organic system. The conventional no-till system used herbicides for cover crop termination and weed management and synthetic fertilizer for nutrient management. The tilled organic system relied on tillage for weed control, cover crop termination, and seedbed preparation. The grazed organic plots were grazed with sheep to terminate cover crops before crop seeding and after harvest to remove crop residue and weed biomass. The goal of the grazed organic system was to reduce tillage; during the first three years, the five-year rotation tillage was eliminated, and it was significantly reduced in the other two years of the study. Specific management details can be found in Appendix S1.
Crop phases of the rotation were safflower (Carthamus tinctorius) undersown with the biennial sweet clover (Melilotus officinalis) (year 1), the sweet clover was then grown as a cover crop in the next phase (year 2), followed by winter wheat (Triticum aestivum cv. Yellowstone) (year 3), lentil (Lens culinaris) (year 4), and winter wheat (year 5). During the first year of the study, it was not possible to have the biennial sweet clover and pea (Pisum sativum) was planted in the sweet clover split plot and terminated as a cover crop.
Each year, composition of weed communities was assessed to species and harvested in late June before termination of the cover crop split plots, and in July before the harvest of the crops. In each split plot, aboveground biomass was harvested in eight 0.5 × 1 m frames; two frames were randomly located in each quartile relative to the length of the split plot. Frames were oriented perpendicular with the crop rows, ensuring that each frame sampled four crop rows and the associated inter-row spaces. The sampled biomass was dried for at least one week at 45°C and weighed to the nearest 0.01 g. The biomass was then averaged across the eight frames and converted to g/m2.
Statistical analysis
We estimated species richness at the split-plot level as the list of species observed in at least one quadrat/frame. Using weed biomass, we calculated Simpson’s Diversity Index for each split plot using the R-Package vegan (Oksanen et al. 2019). As we were comparing plant communities, for this analysis, we removed those split plots that had either one or zero species (n = 7).
A linear mixed-effects model (lmer in the R-package lmerTest; Kuznetsova et al. 2017) with a Gaussian error distribution was fit with weed biomass as a response variable, while a generalized linear mixed-effects model with a Poisson distribution (glmer in the R-package lme4; Bates et al. 2015) was fit with species richness as a response variable, and a Beta Distribution generalized linear mixed-effects model (glmmTMB in the R-package glmmTMB; Brooks et al. 2017) was fit with Simpson’s Diversity Index as a response variable. Explanatory variables included cropping system, year (factor), crop phase, and all interactions as fixed effects. To account for repeated measures through time, split plot was treated as the random effect. We began with maximal models that included all random effects and dropped terms when singularities occurred (Barr et al. 2013). All models were simplified by eliminating fixed-effects terms when they were non-significant or when models were rank deficient, beginning with the 3-way interaction between cropping system, year, and crop phase. Estimated marginal means, R-package emmeans (Lenth 2020), for weed biomass, species richness, and Simpson’s Diversity Index were derived from the mixed-effects model output and post hoc pairwise comparisons were completed using the Tukey method. There were no differences in weed biomass, species richness, or community composition between the two winter wheat phases, and models were improved by collating data from the two phases.
The assumption of normality was assessed and weed biomass was natural log transformed to account for non-normal residuals. Homoscedasticity was assessed and ensured using a visual assessment of the residuals and a Levene test. For the generalized linear mixed-effects models, overdispersion was assessed by calculating the sum of squared Pearson residuals and comparing it to the residual degrees of freedom. The linear mixed-effects model differences between explanatory variables and response variables at the P < 0.05 level were calculated from F statistics of type III sum of squares based on Satterthwaite’s approximations of degrees of freedom. For Generalized linear mixed-effects models, differences between explanatory variables and response variables at the P < 0.05 level were calculated from a type III Wald chi-square test from the R-package car (Fox and Weisberg 2019).
To compare weed communities, we created a Bray-Curtis community dissimilarity matrix using the vegan package (Oksanen et al. 2019). Split plots with no weed biomass were removed from the community dissimilarity analysis. Non-parametric multivariate analysis of variance (NP-MANOVA) using the Bray-Curtis dissimilarity matrix (adonis in the R-package vegan; Oksanen et al. 2019) was used to compare weed communities of each cropping system in each year, and of each crop phase in each cropping system. In this analysis, we accounted for the nested structure of the study by stratifying across split plot. In addition, we represented the weed communities of the cropping system for each year and of the crop phases for each cropping system using the Bray-Curtis dissimilarity matrix by implementing a Kruskal non-metric multidimensional scaling ordination (NMDS; k = 6) from the R-package MASS (Venables and Ripley 2002). We also compared communities using species relative abundance in cropping system for each year and in crop phase for each cropping system. Within the specified levels, we calculated species relative abundance by dividing species biomass by total weed biomass.
All statistics and graphics were performed and produced in the statistical software R, version 3.6.0 (R Core Team 2019), and graphics were produced using R-packages ggplot2 (Wickham 2016) and cowplot (Wilke 2019) packages.
Results
Cropping system (P = 0.01), crop phase (P < 0.001), year (P < 0.001), the interaction between crop phase and cropping system (P < 0.001), and the interaction between cropping system and year (P < 0.001; Table 1, Fig. 1) affected total weed biomass. In the model, the variance of the random effect (split-plot) was 0.02 ± 0.45. The conventional no-till and the tilled organic cropping systems had similarly low total weed biomass and low variation across the five-year study (Fig 1A). The grazed organic cropping system had the overall highest weed biomass because it increased over the five-year period. During the first two years of the study (2013–2014), weed biomass did not differ from the other two systems, but over the final three years of the study (2015–2017) it increased significantly and differed from both other systems (Fig 1A). Within the conventional no-till system, mean weed biomass was lowest in the winter wheat crop phase, followed by lentil, safflower, and sweet clover (Fig. 1B). Weed biomass in the tilled organic system demonstrated the same trend; however, there was an increase during the sweet clover phase, making it greater than the other crop phases (Fig. 1B). In the grazed organic system, the only significant pairwise difference in mean weed biomass was between lentil and winter wheat (Fig. 1B).
| Explanatory | Sum Sq | Mean Sq | NumDF | DenDF | F value | P |
|---|---|---|---|---|---|---|
| Crop phase | 74.05 | 24.68 | 3 | 146.49 | 35.70 | <0.001*** |
| Cropping system | 6.93 | 3.46 | 2 | 43.71 | 5.01 | 0.01* |
| Year | 13.76 | 3.44 | 4 | 145.24 | 4.98 | <0.001*** |
| System : year | 33.38 | 4.17 | 8 | 144.98 | 6.03 | <0.001*** |
| System : crop | 26.05 | 4.34 | 6 | 146.35 | 6.28 | <0.001*** |
Note
- Explanatory variables include crop phase, cropping system, year, and interactions between cropping system (system) and year, and between cropping system (system) and crop phase (crop). *P < 0.05, **P < 0.01, ***P < 0.001.
Cropping system (P = 0.01), crop phase (P < 0.001), year (P < 0.001), the interaction between cropping system and year (P < 0.001), and the interaction between cropping system and crop phase affected species richness (P = 0.02; Table 2, Fig. 2). The variance of the random effect (split-plot) was 0.007 ± 0.08. In 2013, species richness was greater in the conventional no-till than the grazed organic system (Fig. 2A). There were no significant differences in 2014, but over the final three years of the study (2015–2017), species richness decreased in the conventional no-till and increased in the grazed organic. Specifically, in 2015 and 2017 more species were found in the grazed organic system than both other management systems, and in 2016, both the grazed and tilled organic systems had higher species richness compared with the conventional no-till system (Fig. 2A). Among all cropping systems, species richness was highest in the sweet clover phase and, in the conventional no-till and the tilled organic, was lowest in the winter wheat phase, while it was the lowest in the lentil phase of the grazed organic system (Fig. 2B). The interaction between crop phase and cropping system resulted from differences in species richness of the winter wheat phase among the three cropping systems: Mean species richness was higher in the grazed organic (8.08 ± 0.57) than both the tilled organic (6.33 ± 0.49; P = 0.05), and the conventional no-till (4.64 ± 0.41; P < 0.001), and the difference between tilled organic and conventional no-till was also significant (P = 0.02).
| Response | Explanatory | Chisq | df | P |
|---|---|---|---|---|
| Species richness | Intercept | 5789.69 | 1 | <0.001*** |
| Crop phase | 148.78 | 3 | <0.001*** | |
| Cropping system | 8.61 | 2 | 0.01* | |
| Year | 23.1 | 4 | <0.001*** | |
| System : year | 44.7 | 8 | <0.001*** | |
| System : crop | 15 | 6 | 0.02* | |
| Simpson’s Diversity Index | Intercept | 14.85 | 1 | <0.001*** |
| Crop phase | 63.86 | 3 | <0.001*** | |
| Cropping system | 1.70 | 2 | 0.43 | |
| Year | 3.55 | 4 | 0.47 | |
| System : year | 43.77 | 8 | <0.001*** | |
| System : crop | 25.12 | 6 | <0.001*** |
Note
- Explanatory variables include crop phase, cropping system, year, and interactions between cropping system (system) and year, and between cropping system (system) and crop phase (crop). *P < 0.05, **P < 0.01, ***P < 0.001.
Overall, neither cropping system not year affected Simpson’s Diversity Index (P = 0.43, P = 0.47, respectively), but it was affected by crop phase (P < 0.001), and the interactions between cropping system and year (P < 0.001) and crop phase (P < 0.001; Table 2, Fig. 3). The variance of the random effect (split-plot) was <0.001 ± <0.001. Over the course of the study, there was a trend that Simpson’s diversity decreased in the conventional no-till system and increased in the grazed organic (Fig. 3A). The interaction between crop phase and cropping system was the result of differences among the winter wheat diversity values of the three cropping systems: Grazed organic (0.53 ± 0.03) and tilled organic (0.42 ± 0.03) were both more diverse than conventional no-till (0.26 ± 0.03; P < 0.001, P = 0.002, respectively; Fig. 3B).
Weed community composition varied in response to cropping system (P = 0.001), crop phase (P = 0.001), year (P = 0.001), and by the interactions between crop and cropping system (P = 0.001), and between cropping system and year (P = 0.001; Table 3; Fig. 4). The 2013 conventional no-till community differed from the other conventional no-till communities but otherwise the conventional no-till communities were closely grouped, and the tilled organic communities were all tightly grouped (Fig. 4A). The cropping system, year interaction was the result of a substantial community shift in the grazed organic system after 2014, resulting in two distinct groups 2013–2014 and 2015–2017 (Fig. 4A). The weed communities of the wheat phase displayed the most variation and differed from the communities of the other crop phases (Fig. 4B). The weed communities of the other phases were more tightly grouped than the wheat phase and, while the sweet clover communities were distinct from the other communities, NP-MANOVA pairwise analysis of the safflower and lentil communities demonstrated no difference between the two (P = 0.10; Fig. 4B).
| Explanatory | df | Sum Sq | Mean Sq | F. model | R 2 | P |
|---|---|---|---|---|---|---|
| Crop phase | 3 | 6.53 | 2.18 | 8.18 | 0.09 | 0.001** |
| Cropping system | 2 | 4.62 | 2.31 | 8.68 | 0.06 | 0.001** |
| Year | 4 | 4.25 | 1.06 | 3.99 | 0.06 | 0.001** |
| System : year | 8 | 3.91 | 0.49 | 1.84 | 0.05 | 0.001** |
| System : crop | 6 | 3.39 | 0.56 | 2.12 | 0.05 | 0.001** |
| Residuals | 183 | 48.7 | 0.27 | NA | 0.68 | NA |
| Total | 206 | 71.4 | NA | NA | 1 | NA |
- *P < 0.05, **P < 0.01.
Overall, the most abundant species of the cropping systems across the years of the study was Thlaspi arvense (Fig. 5). Another consistent trend across systems was that both Lactuca serriola and Capsella bursa-pastoris increased over the course of the study, although the magnitude of increase differed between systems (Fig. 5). There were other trends and shifts in species relative abundance across systems and years; however, one of the most noteworthy was the increase of Bromus tectorum within the grazed organic from 0.02% of total abundance in 2013 to 38.5% in 2017 (Fig. 5A). For the relative abundance of species comprising 95% of total abundance for the cropping systems in each year of the study (2013–2017), see reference Appendix S2.
Thlaspi arvense was abundant in all crop phases and systems. Capsella bursa-pastoris was abundant in all phases except lentil and safflower of the two organic cropping systems, while L. serriola was abundant in just the sweet clover phase of all three phases. The safflower and lentil phases in all systems were discernible by abundant Chenopodium album and Malva neglecta. One other noteworthy trend was the B. tectorum abundance in the winter wheat of the conventional no-till and grazed organic systems. For the relative abundance of species comprising 95% of total abundance for the crop phases in each of the cropping systems, see reference Appendix S2.
Discussion
In agriculture, cropping system is an important driver of plant community assembly because it modifies both abiotic and biotic conditions (Booth and Swanton 2002, Smith and Mortensen 2017). In our study, over a five-year period, we investigated how weed communities responded to three contrasting cropping systems, conventional high input no-till, tilled organic, and sheep grazed organic. In support of our first hypothesis, our cropping systems affected all weed community response variables (biomass, species richness, Simpson’s diversity, and composition), and the weed community responses varied through time. The second aspect of the study was a comparison of how different crop phases interacted with cropping system to affect weed communities. In support of our second hypothesis, we found that crop phase interacted with cropping system to affect all weed community response variables.
Impact of cropping system on weed communities
In conventional, high input systems synthetic herbicides provide a strong influence on weed communities by selecting against susceptible species (Smith and Mortensen 2017), and studies have consistently found lower weed biomass, species richness, and diversity in conventional systems compared with organic systems (Menalled et al. 2001, Pollnac et al. 2008, 2009, Benaragama et al. 2019). Consistent with these findings, our conventional no-till cropping system had the lowest weed biomass, species richness, and diversity of the three cropping systems.
The soil disturbance associated with tillage affects all aspects of weed communities, from the distribution of weed seeds in the soil, to the germination, establishment, and persistence of emergent weed communities (Nichols et al. 2015, Smith and Mortensen 2017). Consequently, standard tillage cropping systems exert a substantial selection pressure on weed communities resulting in less abundant, diverse, and different communities compared with reduced-till communities (Légère et al. 2008, Santín-Montanyá et al. 2013, Bajwa 2014, Nichols et al. 2015, Smith and Mortensen 2017). Consistent with this, the weed communities of our tilled organic cropping system had low weed biomass over the course of the study, intermediate species richness and diversity levels, and differed in composition from the other two cropping systems. Interestingly, over the course of the study, while the conventional no-till communities became less diverse and the grazed organic communities more diverse, both decreased in similarity with the tilled organic communities and increased in similarity with one another. For example, over the course of the study, B. tectorum became one of the most abundant weeds in reduced-till systems (grazed organic and conventional no-till) but was nearly absent from the tilled organic system. These results underscore that differences among cropping systems are not necessarily driven by organic vs. conventional practices but by the type, intensity, and frequency of disturbance to the system (Smith and Mortensen 2017).
Research on integration of livestock into cropping systems has produced a wide range of weed community responses (Tracy and Davis 2009, Hilimire 2011, McKenzie et al. 2016, Schuster et al. 2018, MacLaren et al. 2019, Rosa-Schleich et al. 2019). These discrepancies stem from differences in duration of study, and location and system studied. Several short-term studies found that integrating livestock into cropping systems had no effect on weed biomass or community composition (Tracy and Davis 2009, McKenzie et al. 2016), and previous short-term work at our study site found minimal effects on diversity (Barroso et al. 2015), and species richness (Miller et al. 2015). Consistent with these studies, the grazed organic communities of the first two years were indistinguishable from the tilled organic communities, demonstrating that during this period the details of organic management other than weed management were more important in shaping weed communities. However, consistent with previous work (Hilimire 2011, Rosa-Schleich et al. 2019), as our study progressed, the grazing treatment exerted a unique influence on weed communities, with the 2015–2017 communities being tightly grouped and distinct from weed communities of the other cropping systems. The 2015–2017 grazed organic weed communities were abundant but also had higher species richness and diversity than the other two cropping systems. This increase in weed diversity coupled with the intrinsic increase in multitrophic diversity (e.g., sheep) make integrating sheep grazing into semi-arid organic communities an option for increasing cropland biodiversity. However, the significant increase in weed biomass beginning in the third year of the study demonstrates that using sheep grazing as the sole weed management method is not practical and must be integrated into a larger integrated weed management strategy.
Impact of crop phase on weed communities
Cultural weed management practices, such as crop rotations and cover cropping, are effective at controlling weeds (Smith and Gross 2007, Nichols et al. 2015, Smith et al. 2015, Gaba et al. 2018) and shape weed community assembly (Booth and Swanton 2002, Smith et al. 2015, Smith and Mortensen 2017). There are two primary ways that crop rotations shape weed communities. First, different crops and rotations necessitate different management methods (Nichols et al. 2015). In our study, this was evident in the high weed biomass within the sweet clover phase of the tilled organic system, which was likely the result of our inability to utilize tillage to manage weeds while the biennial crop established. It was also evident in the low weed biomass and diversity in the conventional no-till winter wheat compared with the broad leaf phases. This was due to the ability to use broad spectrum herbicides in winter wheat, while having to use more selective and less effective herbicides in the broad leaf phases.
Crop–weed competition is the second mechanism through which crops and crop rotations shape weed communities (Zimdahl 2007, Nichols et al. 2015). This can be direct competition for resources between crops and weeds. For example, research has shown that winter wheat is a competitive crop (Smith and Gross 2007, Gaba et al. 2018), which is consistent with the low weed abundance and diversity in all our winter wheat phases. Crop–weed competition can also be engineered by farmers through the use of crops with different phenologies (Zimdahl 2007). In our study, phenological filtering resulted in similarity of weed communities in wheat and sweet clover, due to both being in the field over the winter and having higher proportions of winter annuals compared with spring annuals. The communities of the spring planted lentil and safflower crops had an even more substantial overlap, with very similar weed biomass, species richness, diversity, and composition. The overlap is due to similar phenological filtering and because spring planted crop have similar critical periods of weed control before planting.
Conclusion
Semi-arid cropping systems exert strong and differing effects on cropland weed communities. Our conventional, herbicide-based, cropping system and our tillage-based organic cropping system decreased weed biomass, but at the cost of cropland biodiversity. There was a two-year lag phase after we initiated the integration of sheep grazing into our organic system before the effects of this novel cropping system became apparent. After this lag phase, the weed communities of the grazed organic system demonstrated a significant shift in composition and changes in species richness and weed biomass. The significant increase in weed biomass in our grazed organic cropping system beginning in the third year of our study makes sheep grazing impractical as the sole weed management method in semi-arid organic farming. In order to effectively integrate sheep grazing into semi-arid cropping systems, it will need to be part of a larger integrated weed management program. Furthermore, we found that crop phase substantially and uniquely affects weed communities and interacts with cropping systems to shape weed communities and directly impact agroecosystem biodiversity. Our findings coupled with the reduced off farm inputs implicit with integrating sheep grazing into cropping systems (compared with herbicide- and tillage-based cropping systems) can help farmers enhance agroecosystem biodiversity and may have implications for the sustainability of semi-arid organic farming.
Acknowledgments
We would like to extend our deepest gratitude to Dr. Perry Miller, Jeff Holmes, and Dr. Pat Hatfield for accommodating and facilitating our research at Montana State University’s Fort Ellis Research Farm. Many thanks to Dr. Judit Barroso, Dr. Zachariah Miller, Dr. Subodh Adhikari, Stephen Johnson, Devon Regan, Tessa Scott, Clare Dittemore, and the countless undergraduate technicians for their invaluable contributions. Finally, we are grateful to the United States Department of Agriculture (USDA) Organic Agriculture Research and Extension Initiative (OREI) (grant 2012-51300-20004) and the USDA Organic Transitions (ORG) (grant 2011-51106-31006) for funding this research.
