Predicted climate conditions and cover crop composition modify weed communities in semiarid agroecosystems

The US Northern Great Plains is one of the largest expanses of small grain agriculture, but excessive reliance on off-farms inputs and predicted warmer and drier conditions hinder its agricultural sustainability. In this region, the use of cover crops represents a promising approach to increase biodiversity and reduce external inputs; however little information exists about how cover crop mixture composition, predicted climate and management systems could impact the performance of cover crops and weed communities. In the 4th cycle of a cover crop-wheat rotation, we experimentally increased temperature and reduced moisture as expected to occur with climate change, and assessed impacts on the presence and composition of cover crop mixtures and termination methods on weed communities. Under ambient climate conditions, mean total cover crop biomass was 43%– 53% greater in a five species early-season cover crop mixture compared with a seven species mid-season mixture, and differences were less pronounced in warmer and drier conditions (19%– 24%). We observed a total of 18 weed species; 13 occurring in the early-season mixture, 13 in the mid-season mixtures and 14 in the fallow treatments. Weed species richness and diversity was lower in warmer and drier treatments, and we observed a shift in weed communities due to the presence and composition of cover crop mixtures as well as climate manipulations. Overall, results suggest that adoption of cover crop mixtures in semiarid agroecosys - tems requires jointly addressing weed management and soil moisture retention goals, a challenge further complicated by predicted climate conditions.


| INTRODUC TI ON
In the dryland sections of the Northern Great Plains of US and Canada, a region characterised by long cold winters, short hot summers, large ranges in temperatures and unpredictable precipitation, cultivation and off-farm inputs in the form of pesticides and fertilisers are used to grow large expanses of wheat (Triticum aestivum L.).These monocultures are often grown in rotation with 12-to 16month long fallow period, a strategy used to conserve soil moisture (Padbury et al., 2002).Despite success of this approach in terms of crop yield and quality, numerous environmental costs associated with the concentration of high input wheat production has motivated the search for alternative practices that take advantage of ecological processes to secure crop yield, nutrient cycling and pest regulation (Peterson et al., 2020).
Cover crops, a suite of non-marketable plants that do not provide direct income to farmers and are grown for their ecosystem services, have been studied as one ecologically-based method to enhance the overall sustainability of agricultural production (Kumar et al., 2020).
Given the weed suppression potential of cover crops, their use has been proposed as a tool for managing herbicide resistance (Menalled et al., 2016).
Ecological research in non-cropping systems indicated that increased plant species diversity could result in enhanced productivity, system stability, pest suppression and nutrient retention (Tilman et al., 2014).These findings spawned interest among farmers and agroecologists in replacing single-species cover crops with multi-species mixtures as an ecologically-based method to increase agroecosystem's diversity (Isbell et al., 2017).Despite their popularity, recent studies indicated that increasing the number of species in cover crop mixtures may not necessarily provide enhanced agroecosystem services (Smith et al., 2014), including improved weed suppression (Florence and McGuire, 2020;Smith et al., 2020).
In semiarid regions, where moisture is a limiting factor for crop production, farmers must balance cover crop benefits with possible reduction of water availability needed for subsequent crops (Garland et al., 2021).Across sections of the Northern Great Plains, this challenge is further exacerbated by predicted climate change.
Mean temperature and precipitation are expected to increase over the next 30 years, while summer rainfall is projected to decrease by 10%-20% (Conant et al., 2018), resulting in greater evapotranspiration rates that will likely reduce effective soil moisture (Whitlock et al., 2017).A common approach to manage soil water content usage by cover crops in water-limited agroecosystems is terminating them during anthesis (O'Dea et al., 2013), most commonly with herbicides, an approach that increases the selective pressure for herbicide resistance weeds and unintentionally impacts beneficial biodiversity (Adhikari et al., 2019;Menalled et al., 2016).Alternative approaches for cover crop termination, like livestock-grazing and haying, may be more economically and ecologically sustainable by providing alternative revenues to farmers and enhancing the provision of ecosystem services (Thiessen Martens and Entz, 2011).
To our knowledge, no study conducted in the Northern Great Plains has evaluated the impacts of predicted warmer and drier climate conditions on cover crops and weed communities, and little research has been conducted about the agronomic and ecological impacts of alternative methods for cover crop termination (but see McKenzie et al., 2017).To address these knowledge gaps, we took advantage of a crop-wheat rotation study that comprises cover crops with different species composition to evaluate the impact that warmer and drier conditions and the methods used for cover crop termination have on cover crop biomass production and weed communities.We expected that warmer and drier conditions would decrease cover crop biomass accumulation, and that this impact would differ across cover crop mixtures.In turn, we expected a shift in weed communities as a function of cover crop management strategies interacting with temperature and moisture conditions.

| Study site and experimental design
The study was located at the Montana State University Northern Agricultural Research Center, MT (48°29′48.8″N,109°48′10.4″W).
The site receives an average annual precipitation of 285 mm and has an average high and low temperatures of 13.4°C and 0.0°C respectively (Western Regional Climate Center, 2020).The experiment was designed to evaluate different cover crop mixtures in rotation with winter and spring wheat (Triticum aestivum L.), and terminated with herbicides, by grazing or by haying.The experiment was established as two adjacent replicated trials, one for each phase of the rotation, and designed as a restricted-randomisation strip plot design with three replicates, and termination treatments assigned perpendicular to the fallow and cover crop treatments (Figure 1).This study was conducted in the cover crop phase during the 4th cycle of the cover crop-wheat rotation and included three levels: a summer fal- were sprayed with 2240 kg ha -1 of a.i.glyphosate.Cover crop mixtures were selected to include a variety of functional groups and growth habits.The early-season mixture had five species: radish (Raphanus raphanistrum L.), field pea (Pisum sativum L.), oat (Avena sativa L.), turnip (Brassica rapa L.) and hairy vetch (Vicia villosa Roth).
In 2018, the mid-season mixture had seven species: oat, radish, field pea, turnip, lentil (Lens culinaris Medikus) and sorghum-sudangrass  1) using a 3.7-m Conservapack no-till disk planter.At planting, a total of 112 kg ha -1 fertiliser with a 20-20-20 dosage of N-P-K was applied to the cover crops.
Cover crops were terminated shortly after anthesis in oat with either herbicide, haying and baling, or cattle grazing.The herbicide application (glyphosate, applied at 2,500 kg ai/ha) and haying treatments occurred on 9 July 2018 and 8 July 2019.The cattle grazing treatment was designed to be high intensity short duration using 8-10 yearling bulls from 11 July to 13 July 2018 and 9 July to 11 July 2019.In addition, 2 kg ai/ha of glyphosate plus 0.34 kg of ai/ha dicamba were used to terminate any regrowth of cover crops and weeds in the grazed and hayed termination treatments, approximately four weeks after the initial termination.
To create warmer and drier conditions (climate manipulation, hereafter), open-top chambers and rain-out shelters were used.
To increase the temperature by ~2°C, open-top chambers had a basal diameter of 1.5 m and a top diameter of 1 m and were built out of Sun-Lite HP (1 mm thick) (Solar Components Corporation, Manchester, New Hampshire, USA) fiberglass with high transmittance of visible wavelengths (Marion et al., 1997).Rain-out shelters were constructed following Yahdjian and Sala (2002) with wooden frames that supported gutters made out of corrugated clear polycarbonate material (Palram Americas, Kutztown, Pennsylvania, USA) to reduce precipitation by approximately 55%.

| Statistical analysis
All analysis were implemented in R version 3.5.2(R Core Team, 2020).To assess the effects of the presence (fallow or cover crop) and composition (early-season and mid-season) of cover crops, climate manipulations (ambient, and warmer and drier), and year on cover crop biomass and temperature and soil moisture measurements, we fit linear mixed effects models using the 'lmer' function in the 'lme4' package (Bates et al., 2015).Identity of individual sensors was included as a random effect to account repeated measure from the same moisture and temperature sensors.
Response variables for weed communities included biomass, species richness, species diversity and species composition.We fit linear mixed-effects models to assess the effects of the presence and composition of cover crops, termination, climate manipulation and year on cover crop and weed biomass.We calculated species richness as the total number of species present per 0.56 m 2 area each year, and diversity using the Shannon diversity index with the 'diversity' function in the 'vegan' package (Oksanen et al., 2019).For all analyses, cover crop, climate manipulation, termination treatments, year and their interactions were treated as fixed effects.An identifier of each individual cover crop replication was included as a random effect to account for the nesting of climate manipulation levels within cover crop and termination treatments.Assumptions of normality and equal variance were assessed visually using diagnostic plots and the weed biomass response was log-transformed prior to the analysis.For the cover crop and weed biomass responses, we used Type III analysis of variance (ANOVA) and examined pairwise differences using Tukey tests implemented through the 'emmeans' package.Results were back-transformed when necessary and graphics were produced using the 'ggplot2' package.
We calculated dissimilarity among samples using weed biomass and the Bray-Curtis metric applying the'vegdist' function also in the 'vegan' package (Oksanen et al., 2019).We used the permutational TA B L E 1 Seeding rates kg ha -1 and biological and ecological characteristics of cover crop species  Soil moisture varied in response to the presence and composition of cover crops (F 2, 85.55 = 30.49,p < 0.001), climate manipulation (F 1, 166.91 = 6.21, p = 0.014), and year (F 1, 87.17 = 5.64, p = 0.019) (Figure S1).The lowest mean soil moisture was observed in the early-season mixture under warmer and drier conditions (−6.9 and −4.8 bar in 2018 and 2019, respectively).Soil moisture under ambient conditions was similar to warmer and drier conditions (−6.1 bar in 2018 and 2019).In contrast, we observed the greatest mean soil moisture under warmer and drier conditions in the fallow treatments (−2.9 and −2.4 bar in 2018 and 2019, respectively), which differed from fallow treatments under ambient conditions (−0.8 and −1.9 bar in 2018 and 2019, respectively).

| Cover crop biomass
Total biomass varied in response to cover crop composition (F 1, 57 = 87.91,p < 0.001) and year (F 1, 31 = 44.15,p < 0.001), the later probably due to year to year differences in precipitation and temperature, but not across termination methods (F 2, 31 = 0.93, p = 0.41) or climate manipulations (F 1, 32 = 0.76, p = 0.39).However, there was an interaction between cover crop composition and climate manipulations (F   2).Mean biomass of radish, turnip and the legumes were not affected by climate manipulations in the early-season mixture, however, mean biomass of radish and turnip in the mid-season mixture was larger under warmer and drier conditions than ambient conditions (Table 2 and Table S1).The two warm-season C4 grass species planted -sorghum-sudan grass (in 2018), and millet (in 2019) -contributed only 1 and 2%, respectively, to the total mid-season crop biomass, and mean biomass of these species was further reduced under warmer and drier conditions.

| Weed communities
Our expectation was that crop management strategies and climate manipulations would result in distinctive weed communities.In agreement, weed biomass differed in response to the presence and composition of cover crops (F 2, 88 = 8.92, p < 0.001), climate manipulations (F 1, 88 = 12.7, p < 0.001) and the interaction between these two variables (F 2, 88 = 3.29, p = 0.041).Neither termination method (F 2, 97 = 0.97, p = 0.379) nor year (F 1, 88 = 0.049, p = 0.825) impacted weed biomass.In 2018, mean weed biomass was lower in the earlyseason mixture under warmer and drier conditions (7.6 kg ha -1 ), compared with the biomass sampled in the mid-season cover crop under ambient conditions (117.1 kg ha -1 ).In 2019, mean weed biomass was also lower in the early-season mixture under warmer and drier conditions (7.7 kg ha -1 ), when compared with the mid-season mixture (146.7 kg ha -1 ) and fallow (112.6 kg ha -1 ) under ambient conditions (Figure 3).

| DISCUSS ION
Predicted climate conditions could impact the distribution, composition, demography and competitiveness of weed species (Kumar et al., 2020) and impact the efficacy of herbicide-based weed management strategies (Ziska, 2020).Yet, there is little information about the impact that climate change could have on the performance of cover crop mixtures as a weed management tool.In our study, located in a semiarid and cold section of the Northern Great Plains, an increase of summer temperatures coupled with a reduction in moisture availability, differentially impacted the performance of early-season and mid-season cover crop mixtures, and resulted in a shift in weed biomass production and community structure.
In regions experiencing low precipitation, such as the semiarid sections of the Northern Great Plains, managers must balance potential weed suppressive ability of cover crops with moisture retention for the subsequent crop (Kumar et al., 2020).Our results indicated that cover crop mixtures and fallow treatments differently influenced the temperature and soil moisture conditions, as well as their associated weed communities.In agreement with Blanco-Canqui et al. (2015), the five species early-season cover crop mixture produced the most biomass, and maintained the coolest temperatures under both ambient, and warmer and drier conditions.In contrast to observed temperature patterns, the largest reduction in soil moisture during the growing period was observed in the early-season cover crop mixture, suggesting early-season mixtures could reduce moisture storage and compromise subsequent crop yields (Krueger et al., 2011), a challenge that could be increased under the predicted warmer climate conditions of the region (Whitlock et al., 2017).
The suitability of cover crops as weed management tools depends on their ability to effectively establish and capture resources during the growing season.In cold and semiarid environments, niche differentiation may allow cover crop mixtures to enhance resource use efficiency (Elhakeem et al., 2019); however, the selection of species can be challenging as their success is heavily dependent on interspecific differences, environmental conditions and management strategies (Lithourgidis et al., 2011).In accordance with previous research conducted in this region (Carr et al., 2004), cover crop species composition and planting date were important drivers of cover crop productivity.Specifically, the five species cover crop mixture planted in early spring gained more biomass compared with mixtures with seven cover crop species   Resource availability and accumulation of growing degree days can impact the relative abundance of species in cover crop mixtures (Baraibar et al., 2020); however little is known about the extent to which various plant functional groups respond to predicted warmer and drier conditions.We observed a reduction in the total cover crop biomass in the early-season mixture under warmer and drier conditions.In particular, oat was more sensitive to higher temperatures low (control), a five-species early-season cover crop mixture, which was planted in early spring, and a seven-species mid-season cover crop mixture, which was planted approximately two weeks later.The location of each treatment was randomised in the first year of the F I G U R E 1 Simplified experimental design to assess the impact of summer fallow, an early-season cover crop mixture, and a mid-season cover crop mixture; three methods of cover crop termination (Herbicide, Grazed and Hayed); and two climate manipulations (Ambient, A, and Warmer and Drier, W/D) on weed communities study (2012-2013) and maintained through time.Within each cover crop level and termination method, we established two temperature and moisture manipulation treatments (ambient, and warmer and drier, see below).Summer fallow began after winter wheat harvest on 12 July 2017 and 26 July 2018, and ended on 21 September 2018 and 13 September 2019.Before planting the cover crop mixtures, all plots Soil and air temperatures were measured under ambient, and warmer and drier treatments in 3-h intervals using Maxim Integrated Thermochron iButtons (DS1921) placed 10 cm below and 20 cm above the soil surface.Temperature was measured from 10 May to 1 July 2018, and 9 May to 1 July 2019 and 10 May to 22 July 2018.Soil moisture was measured once weekly from 15 May to 1 July 2018 and 23 May to 29 July 2019, using the Delmorst soil moisture measuring systems (model KS-D1) with sensors installed at a depth of 10 cm.To assess changes to cover crop biomass in response to cover crop composition, termination and climate manipulations, we collected aboveground biomass from two 0.56 m crop rows located in each treatment on 1 July 2018 and 2019 (n = 36 per year).Also, on 1 July 2018 and 2019, we sampled weed biomass throughout 0.56 m 2 frames established in the centre of each treatment.Cover crop and weed biomass were separated by species, dried, and the weight of each species was recorded.Because different legume cover crop species were planted within our two-year study period, they were combined to represent 'legumes' as a functional group in our analysis.

|
multivariate ANOVA (perMANOVA) and algorithm 'adonis' in the 'vegan' package to evaluate whether a significant a proportion of the variation in weed species composition was accounted by the presence and composition of cover crops, climate manipulations and year.To visualise differences in weed species composition across treatments, we performed a nonmetric multidimensional scaling (NMDS) analysis and used the ordination to calculate potential change in species composition.The analysis was conducted using the 'metaMDS' algorithm in the 'vegan' package.Finally, we conducted an indicator species analysis to assess the existence of key species associated with the presence and composition of cover crops and climate conditions.Temperature and moisture Air temperature, measured 20 cm above ground level, varied depending on the presence and composition of cover crops (F 2, 14 = 10.75, p = 0.002), climate manipulations (F 1, 2076 = 91.33,p < 0.001) and year (F 1, 14 = 998.41,p < 0.001) (FigureS1).Mean air temperature under ambient condition was coolest in the early-season cover crop mixture (18.9 and 15.5°C in 2018 and 2019, respectively), compared with the fallow(19.3and 15.9°C in 2018 and 2019, respectively)    and the mid-season mixture (19.5 and 16.1°C in 2018 and 2019, respectively).Similarly, under warmer and drier conditions, mean air temperatures in the summer fallow and the mid-season mixture treatments were warmer(20.2and 20.4,respectively, in 2018; 16.7   and 16.9, respectively, in 2019)  than in the early-season mixture treatment(19.8°C in 2018 and 16.3°C in 2019).Soil temperature, measured 10 cm belowground, varied in response to the presence and composition of cover crops (F 2, 14 = 7.42, p = 0.006), climate manipulation (F 1, 19681 = 25.07,p < 0.001) and year (F 1, 13 = 322.33,p < 0.001) (FigureS1).However, the effect was dependent on the presence and composition of cover crops (Cover crop × Climate; F 2, 18506 = 22.6, p < 0.001).Mean soil temperature under ambient climate conditions was lowest in the early-season cover crop mixture (18.9 and 15.5°C in 2018 and 2019, respectively), compared with the fallow (19.3 and 15.9°C in 2018 and 2019, respectively) and mid-season mixture(19.5 and 16.1°C in 2018 and   2019, respectively).Under warmer and drier conditions, the earlyseason mixture was cooler(19.8and 16.3°C in 2018 and 2019, respectively)  than summer fallow(20.2and 16.7°C in 2018 and 2019,   respectively)  and the mid-season mixture(20.4 and 16.9°C in 2018   and 2019, respectively).
Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/wre.12514by Montana State University Library, Wiley Online Library on [29/12/2022].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License and drier conditions (833 and 835 kg ha -1 in 2018 and 2019, respectively; Table

TA B L E 2
Mean (SE) cover crop species biomass sampled in (A) 2018, and (B) 2019 in the early-season mixture (Early) and mid-season mixture (Mid) under ambient and warmer and drier (W/D) conditions A. 2018 Cover crop species biomass kg ha -

F I G U R E 3
Weed biomass kg ha -1 sampled in 2018 (A) and 2019 (B) as a function of presence and composition of cover crops (i.e.fallow; early-season, Early; or mid-season, Mid), and temperature and soil moisture (i.e.ambient or warmer and drier) conditions.Error bars indicate the standard errors of the mean, while letters above bars indicate significant differences among treatments (p < 0.05)

TA B L E 4
Mean (SE) biomass of weed species biomass (kg/ha) sampled in the in the fallow, early-season cover crop mixture and midseason cover crop mixture in 2018 and 2019.Samples were obtained under ambient (A) and warmer and drier (W/D) climate conditions.When a weed species only occurred in one plot, standard error is represented as(-)