Final report for GNE18-165
Project Information
Our aim with this project was to understand and quantify how mixtures of different crop species grown together could provide benefits such as reducing arthropod pests, suppressing weeds and increasing crop production. We expected to see an overall increase in arthropod predators – both generalist predators and weed seed predators. We expected these increases to then help suppress herbivore arthropods, therefore reducing the need for chemical insecticides and potentially increasing yields. We also expected certain ground-dwelling beetles to consume more weed seeds and potentially show preference for specific weeds seeds, therefor reducing weeds in the crop mixtures or altering the weed community.
We conducted field and greenhouse experiments over a two-year period with several forage crops as monocultures, 2-species mixtures and 4-species mixtures. Additionally, we considered crops as one of three different functional groups – broadleaf crops, grass crops and a mixture of both. We sampled the weed community, arthropod community and monitored weed seed predation as well as sampled our crops for a yield and calculated a partial economic analysis based on inputs and yield.
We found that although there was not a significant increase in mixtures compared to corn alone, sorghum had a similar partial return over variable costs because of it’s lower seed cost. Sorghum and corn grown together appeared to have a detrimental effect on yield – a lower production compared to each grown alone. However, corn grown with a broadleaf (soybeans), had similar production to corn alone, showing that some species better complement each other and some species appear to compete. Additionally, we saw an increase in yield in soybeans and sunflower grown together compared to each monoculture, indicating broadleaf crops like these may be better complementary crops in a mixture compared to growing multiple high-yielding grasses together.
We did not show data that supported our hypotheses regarding the arthropod community. Predators were greatest in abundance in monocultures than in mixtures, although total arthropod numbers were higher in crop mixtures. We observed higher rates of broadleaf weed seed predation and higher suppression of broadleaf weeds in mixtures, which supports our hypothesis and shows a potential shift in the weed seed bank and weed community in crop mixtures. Our greenhouse experiments also support this, showing a higher proportion of soybean crops had a lower abundance and lower biomass of broadleaf weeds. The same was not true for grass crops and grass weed suppression, mostly because overall grass weeds were relatively low.
Overall, we showed some benefits in the form of complementarity for crop mixtures. More research should be done to tailor exact mixtures based on species and proportion of each species. Our field data show some trends related to functional group and competition between crops compared to those grown with broadleaf crops and the ability of broadleaf crops to complement other crops when grown in mixture. We also demonstrated that seed costs were the greatest input cost and had a large effect on partial return over variable costs. Therefore, crop mixtures may have additional economic benefits for growers. Although the arthropod community was greater in abundance with increased crop diversity, we did not observe the same trend for predator arthropods as we expected. This could be due to our smaller plot sizes, environmental effects not accounted for or simply our timing of data sampling along with later planting of forages crops and that our sampling events did not allow enough time for predator numbers to catch up to the number of prey available. This could be verified with additional sampling, including repeating experiments in the same field over several years rather than rotating fields as we did. Finally, although we were not able to link additional weed suppression directly to arthropod weed predators, we did show that crop species may suppress certain weeds and shift the weed community. In both field and greenhouse experiments, the incorporation of broadleaf crops in mixture did suppress both abundances and biomass of broadleaf weeds. Although we were not able to show the same for grass weeds, this could be due to the overall lower grass weed abundances and biomass that we observed.
Overall, crop mixtures do provide some benefits to agronomic systems. With further research, we could tailor crop mixtures to provide yield and economic benefits in addition to environmental benefits of natural weed and arthropod pest suppression.
Objective 1: Determine how species diversity within crop mixtures can influence arthropod communities, particularly weed seed predators. Research has established that increased plant species diversity maintains a more stable community, is more productive, and provides additional habitats for animals. These additional niches allow greater diversity of arthropod species to inhabit crop mixtures compared to monocultures. I expect arthropod abundance and diversity, specifically weed-seed predators, to increase as crop mixture diversity increases, and increase weed suppression. If weed-seed predators show no preference for broadleaf and grass weeds, I expect weed functional groups to be suppressed equally and increased crop species diversity in mixtures will only increase the rate of consumption, but not alter the composition.
Objective 2: Determine how functional groups within crop mixtures can shift or suppress weed composition. I will test functional groups (grasses, broadleaves) in mixtures of forage crops to determine their influence on the weed community. I will measure both composition, the proportion of grass and broadleaves, and suppression, the total number of weeds. Because of planting density, herbicides and optimal timing and depth for rapid germination, I expect the forage crops to outcompete, and therefore suppress, weed populations of the same functional group. Consistent with my preliminary results, I expect grass forage crops to better suppress grass weeds, shifting the weed community to comprise more broadleaf weeds and the opposite for broadleaf forage crops.
Objective 3: Determine how an increase in weed seed predators and selective suppression of functional groups ultimately influences the number and composition of weeds in forage crop mixtures. I aim to quantify interactions of the first two objectives to determine the ultimate result of weed community suppression and composition. The dominant functional group of the forage mixture, regardless of the number of species, is likely to shift the weed composition. However, increased number of forage species is expected to increase the rate of consumption by weed-seed predators and therefore suppress overall weed abundance. By evaluating these interactions, I will identify which pressure, the functional group or the number of crop species, drives the weed community.
Objective 4: Quantify productivity of forage mixtures and tradeoffs among input costs, yield and quality. I will quantify forage yields to compare biomass, quality and determine gross profit. Input costs, including herbicides, will be used to calculate net return, quantify economic value of weed control and compare input and net return for each forage mixture.
Modern agriculture specializes by growing a single monoculture, but by doing so suffers from intensification practices that expand crop land area or require higher inputs, higher costs and can cause environmental degradation9, 10, 11. These practices decrease biodiversity and destabilize ecosystem interactions with unintended consequences, such as exacerbating insect pest populations7 or selecting for herbicide resistant weeds. An alternative to intensive monoculture production is to incorporate multiple crops in a polyculture system. Forage crops are unique because they can incorporate a mixture of species harvested with moisture, fermented and used for cattle consumption. By diversifying crop fields, it is likely we can overcome some of negative effects of ongoing agricultural intensification and use ecosystem services to stabilize the farmers’ fields to mitigate environmental degradation.
Many studies have quantified ecological effects of greater species diversity in natural systems, including grasslands, but few have applied these same concepts to agriculture1,2 or have not focused on silage systems3,4. The species-diversity hypothesis supports the expectation that mixtures of species will be more efficient using available resources, stable and facilitate additional habitats12. This hypothesis predicts that plant communities with greater levels of diversity will have less severe pest invasions and disease5,1 because of higher natural enemy populations, fewer resources available to competing weed species and decreased host abundance for pathogens6,1. On the other hand, however, the mechanism behind many of the measurable benefits of species diversity or mixtures has been argued to be due to “driver species” in the mixture rather than the increased number of species1,2. However, both mechanisms have shown greater plant species diversity to better suppress weeds or unwanted plants and facilitate greater arthropod diversity. Applying these concepts to forage mixtures will allow us to quantify ecosystem services, yields and net return in different forage mixtures applicable to Pennsylvania farmers.
Research
Experimental design, planting and management
To quantify the value of forage mixtures under field conditions, we conducted a two-year experiment in no-till fields at the Pennsylvania State University’s Russell E. Larson Agricultural Research Center at Rock Springs, Pennsylvania (40.7118 N, 77.9693 W, 370m above sea level). We used a complete block design with eight treatments that were replicated in five blocks. To understand the influence of weed seed predators compared to herbicides, we split each plot and treated half with an herbicide, leaving the other half untreated as a control. Congruent with practices on local commercial farms, we varied the location of the experimental field annually and preceding crops each year differed. Due to varying field dimensions, we adjusted plot size to match the space available. Plots were 12.2 x 15.2 m (split at 7.6 m) in 2019 and 12.2 x 12.2 m (split at 6.1 m) in 2020. For tractor maneuvering, alleys (12.2 wide) were planted between blocks, not within blocks. We planted forages on June 12, 2019 and June 8, 2020.
Our eight forage crop treatments were designed to test species diversity (i.e., the number of species in a cropping system) and functional group diversity (i.e., if monocots, dicots or the combination of the two) for weed suppression. More specifically, we wanted to test if the number of crop species alone was able to suppress weeds due to diversity or if the amount of diversity, that is, functional group diversity, also played a role on weed suppression and on which weeds were suppressed. Therefore, our five replicates of forage crop mixtures in the field were organized in a nested experimental design with species diversity variables nested within functional group variables. Functional group variables were “grass”, “broadleaf” and “both” - a mixture of broadleaf and grass forages. Crop species diversity was nested within grass and broadleaf variables and included two monocultures each (broadleaf: soybean, sunflower; grass: corn, sorghum) a simple mixture of two species (broadleaf: soybean/sunflower, grass: corn/sorghum) and a complex mixture of all four species (soybean/sunflower/corn/sorghum). To be consistent in regards to population density, all mixtures were planted at a 1:1 ratio of seed rates and at the same density of 40,000 plants per acre. Cultivars were the same in both years and were selected based on availability and commercially available cultivars from a local supplier (Seedway, Mifflinburg, Pennsylvania, USA).
Forages were planted in 11.8cm row spacings along with 224 kg ha-1 starter fertilizer (10-20-20). Additional nitrogen (50.4 kg on N ha-1) was added within four weeks of planting. We used a corn planter to plant all corn and an interseeder to plant all other crop species. Mixtures that consisted of both corn and another species had two passes – one from each piece of equipment in order to plant most precisely. For our herbicide split plots, we applied S-Metolachlor (1367.7 ml a.i. ha-1), Atrazine (638.8 ml a.i. ha-1), Mesotrione (152.0 ml a.i. ha-1) and Bicyclopyrone (38.0 ml a.i. ha-1) (Trade name: Acuron®) on maize only plots in 2019 and S-Metolachlor (1538.9 ml a.i. ha-1) (Trade name: Dual II Magnum®; 21 g. a.i. ha-1) on all other plots in 2019 and all plots including maize in 2020. Herbicides were applied within 3 days post-planting.
Weed sampling
At four weeks and eight weeks post-planting, we sampled a 0.25m2 area by quantifying the number of broadleaf weeds and grass weeds as a total and further evaluated the most common weeds: marestail, velvetleaf, lambsquarters, pigweed, cheat and foxtail to get a better idea of weed species diversity. In 2019, we harvested all weeds, dried and weighed to assess weed biomass. In 2020, weed populations were so low at both 4 and 8 weeks, we did not assess biomass.
Arthropod community
To measure arthropod communities, pitfall traps were set-up over 48-hours to measure relative ground arthropod communities in June and July. The number of weed seed predators were quantified from these pitfall samples by focusing on Carabidae and Staphylinidae ground beetles. Furthermore, we grouped predators as Carabid and Rove beetles, spiders, harvestmen, millipedes and ants to analyze the overall number of ground predators in plots.
In addition, we measured arthropod weed seed preference in 2020 by using weed seed sentinel prey cards with redroot pigweed and giant foxtail seeds (White et al. 2008) and monitored over 48 hours. The number of total seeds consumed and proportion of grass and broadleaf seeds consumed was quantified to measure rate of consumption and preference. In 2019, we did a single assessment of foliar arthropod community by sweep net sampling of 25 sweeps per split plot to compliment ground dwelling arthropod community.
Yield and Economics
Yield was assessed for each forage treatment. At the end of the growing season, forages were harvested by hand from plots to determine final yields. We harvested 0.91m of row space for each split-plot. We then dried down harvested biomass to 0% moisture and weighed for dry matter. Finally, we used these values to calculate economic values as well as cost of inputs and used these to determine partial return of variable costs. Input costs comprised of seed costs, herbicide costs, fertilizer machinery and fuel costs (Table 1). Gross return was calculated as $90 for every MT dry matter of forage, regardless of species or quality. Partial return over variable costs was the difference of inputs from gross return.
Greenhouse experiments
Greenhouse experiments were designed to complement field experiments by assessing a larger range of functional diversity (proportion of monocot crop species to dicot crop species). We set up trays with 10 plants in each of either soybean or maize. Proportions ranged from 0 soybeans/10 maize plants to 10 soybeans/0 maize plants. We also had a control tray with no crop plants. Each tray was then seeded with 20 foxtail seeds, 10 velvetleaf and 10 pigweed seeds. Trays were watered twice weekly and emerged weeds were recorded weekly for 6 weeks. At the end of six weeks, both crops and weeds were harvested. This was replicated three times in 2019, but foxtail germination was almost 0% throughout the experiment, so this was replicated an additional four times with new foxtail seed in 2020. Experiments from 2019 were excluded from analysis.
Statistical Analysis
We analyzed data using JMP Pro 15 (SAS Institute Inc., 2020). For statistical significance, we used α = 0.05.
Yield, nor input costs or partial return over variable costs did not fit a normal distribution (W = 0.9605, P = 0.0002; W = 0.9717, P = 0.0032; W = 0.9660, P = 0.0008; respectively) even after a transformation. Therefore, these three variables were analyzed as a non-parametric analysis with a Wilcoxon/ Kruskal-Wallis Test. Means were compared using Tukey’s HSD test (α = 0.05).
Weed counts and biomass and four and eight weeks also did not fit a normal distribution. Because of the higher number of zeros in the data (i.e., no visible weeds present) and the data being count data, weed counts and biomass were analyzed with a generalized linear model (GLM) with a Poisson distribution. Factors considered year, block nested within year, function group, and number of crop species nested within functional group. Analysis was done separately for week 4 and week 8. Weed seed predation was also not normally distributed and was therefore analyzed with a GLM with a Poisson distribution. Factors tested for weed seed predation were also block, functional group, number of crop species nested within functional group and herbicide application. Similarly, for weed seed predators caught in pitfall traps, we analyzed the total number of ground beetles – focused on weed seed predators Carabidae and Staphylinidae with a GLM and a Poisson distribution do to violations of normality. Factors tested in the model included block, functional group, number of crop species nested within functional group and herbicide application.
Both overall pitfall arthropod abundances and pitfall predator abundances were not normally distributed and were therefore fitted to a Poisson distribution and analyzed with a GLM with factors year, block nested within year, functional group, number of crop species nested within functional group, and herbicide application.
Sweep net arthropod abundances were cubic-root transformed to achieve normality. An analysis of variance (ANOVA) was used to test the factors block, functional group, number of crop species nested within functional group and herbicide application.
Finally, for greenhouse experiments, we analyzed both broadleaf and grass weeds biomass at six weeks with the factors replication, number of corn plants and soybean plants as well as their interaction and used corn and broadleaf biomass at six weeks as covariates in the model. We used a GLM with a Poisson distribution because data did not fit the assumption of normality.
Yields, variable costs and partial returns over variable costs
Yield was not affected by year (X2 = 0.0766, P = 0.7820), by number of crop species (X2 = 2.3980, P = 0.3015), nor by herbicide application (X2 = 0.0766, P = 0.7820). However, yield was affected by treatment (X2 = 67.7881, P < 0.0001), and by functional group type (X2 = 55.7480, P < 0.0001). Yield is shown in Figure 1 as a scatter plot for all yields regardless of functional type for the number of species planted together on the horizontal axis (one, two or four species). Data shows an average monoculture (one species) yield of about 8,030 kg ha-1 compared to 8,760 kg ha-1 for a two-species mixture. The four-species mixture had lower yields of only 7,440 kg ha-1. Furthermore, standard deviation was greatest in monocultures, followed by two-species polycultures, and four-species mixtures having the smallest standard deviation (Figure 1). Comparing standard errors, however, indicated the opposite trend considering monocultures were the most abundant crop tested and four-species mixtures had the fewest sampling events. Yield response varied for the different functional diversities tested. For broadleaf crops, increasing species diversity from a single species to two species increased yield. However, for grass crops, increasing the number of crop species grown together from a single species to two species decreased yields. Finally, when broadleaf and grass crop species were grown together, the two-species polyculture yielded better than the four-species polyculture. Because the two species polyculture consisted of a soybean/corn mixture only, it’s difficult to determine if the two-species polyculture did better than the four-species polyculture because of the specific species combined or because it had fewer species in it. We’ll discuss this further with the greenhouse results section.
Variable costs associated with economic analysis were not affected by year (X2 = 0.1079, P = 0.7426). However, variable costs were affected by crop treatment (X2 = 129.4483, P < 0.0001) ranging from $246.55 - $486.82 ha-1 (Table 2). Variable costs were also affected by functional group (X2 = 25.7402, P < 0.0001) with broadleaf crops costing the least ($321.43 ha-1), followed by grass crops ($374.22 ha-1) and the mixture of both costing the most ($414.76 ha-1). This was largely due to seed costs (corn having the highest seed cost) followed by machinery costs associated with multiple passes for some of the mixtures. The number of crop species also affected variable costs (X2 = 12.9467, P = 0.0015) with two-species mixtures having the greatest variable costs ($404.86 ha-1) compared to monocultures ($345.74 ha-1) and four-species mixtures ($368.51 ha-1). Finally, as we expected, herbicide-sprayed plots also had significantly higher costs at $403.56 ha-1 compared to control plots with only $334.36 ha-1 (X2 = 16.3112, P < 0.0001).
Partial return over variable costs was not significantly different for years (X2 = 0.0276, P = 0.8681), number of crop species (X2 = 1.0616, P = 0.5881), nor herbicide application (X2 = 0.4748, P = 0.4908). Partial return did vary by crop treatment (X2 = 65.0138, P < 0.0001) and crop functional type (X2 = 49.6581, P < 0.0001). Sorghum alone had the greatest return ($725.09 ha-1) considering each crop treatment, followed by corn ($576.33 ha-1). In comparison, soybeans were at a break-even amount of only $38.91 ha-1. The various mixtures retuned amounts over variable costs between the low-profiting sunflower and soybean monocultures and the higher-profiting sorghum and corn (Table 2). Consider crop functional groups, grass-type crops had the overall greatest return over variable costs at $580.07 ha-1, followed by a mixture of both grass and broadleaves ($390.66 ha-1) and finally only broadleaf crops ($138.55 ha-1).
Yields were highest with corn, followed closely by sorghum (Table 2). The mixture of corn and sorghum, however, was significantly lower than corn monoculture yields and comparably lower than sorghum monoculture yields. In comparison, both sunflower and soybean monoculture yields yielded poorly. Although not significantly higher than the constituent monocultures, the mixture of these two yielded nearly 2,000 kg ha-1 higher than either one alone. Furthermore, looking at the other mixtures comprised of both broadleaf and grass crop functional groups, yields were statistically similar to the corn monoculture in the corn/soybean mixtures and average in the four-species mixture. These results indicate that overall, the grass crop species, corn and sorghum, yield well alone and when mixed with a different functional group, such as we would expect with complementarity. However, when mixed with a similar functional group, i.e. the four-species mixture or corn/sorghum, yields seemed to be inhibited, indicating some potential competition between these crops due to their similarities. Interesting, however, is that we didn’t see this same pattern with broadleaf crop species. It could be that planting populations were lower than is typical for soybeans, but nevertheless, the same total population was sowed in soybean/sunflower, indicating that broadleaf crops may have higher complementarity abilities than grass crops. The mechanisms behind this could be many. As our broadleaf crops didn’t yield as high as the grass crop species, it could be that there were different limiting factors among the different crop species (i.e., sunlight, nutrients, water, or another environmental limitation in grass crops and genetics in broadleaf crops).
Variable costs were higher for herbicides and for corn seed compared to the other crop seed costs. Although yield was similar for herbicide sprayed (8,327 kg ha-1) compared to control plots (8,077 kg ha-1), there was a $69 ha-1difference in variable costs with herbicide sprayed plots being higher. Therefore, return over variable costs was greater in non-sprayed plots. Additionally, considering corn had higher costs, the return over variable costs was actually $148 ha-1lower than sorghum (Table 2), even though corn yielded 906 kg ha-1 more dry biomass. Although soybean, when mixed with corn, also yielded high, soybean seeds were not reduced enough to offset return over variable costs enough and returned less than corn grown alone. Soybeans and sunflower alone did not do well considering costs and their overall low yield in monoculture. Soybeans returned only enough to break even and sunflower returned only $168 ha-1, a very small amount compared to sorghum at $725 ha-1.
Considering both yields, input costs and the typical return over those variable costs, sorghum may be a viable option for many growers. Unfortunately, it appears the competition between sorghum and corn as we tested in our experiments, may need to be better tailored to become more profitable for growers. However, rotations between corn and sorghum may be an option and allow for more diversity in cropping systems over time and may help with some pests and diseases. Comparatively, corn/soybeans, a mixture becoming more common with growers, is shown to yield well here, with returns similar, although slightly lower than monocultures. This mixture could be a good option for farmers looking to diversify cropping systems for other reasons, such as pests, soil health or for additional nitrogen. Finally, our four-species mixture, although yields and return over variable costs were significantly lower than either corn or sorghum as a monoculture, did show some promise. Considering corn and soybeans worked well together, and soybeans and sunflower worked well together, but corn and sorghum did not, it could be that a single grass crop mixed with multiple broadleaf crops could provide the ideal diversity and complementarity needed to thrive and yield well in a cropping system. We would recommend future work evaluate in more detail, the potential of tailoring cropping mixtures considering broadleaf crops could be mixed in greater diversity compared to grass crops.
Weed abundances and biomass
Grass type weed abundances at four weeks was affected by year (X2 = 103.8446, P < 0.0001), block(year) (X2 = 401.4038, P < 0.0001), functional group (X2 = 11.8779, P = 0.0026), number of crop species(functional group) (X2 = 222.7186, P < 0.0001) and by herbicide application (X2 = 239.5483, P < 0.0001). Grass type weed abundances at eight weeks was also affected by year (X2 = 29.1196, P < 0.0001), block(year) (X2 = 664.8587, P < 0.0001), functional group (X2 = 21.9819, P < 0.0001), number of crop species(functional group) (X2 = 253.2010, P < 0.0001) and by herbicide application (X2 = 474.8592, P < 0.0001). Overall, there were more weeds in 2019 (6.93 weeds per m2) compared to 2020 (1.13 weeds per m2). Contrary to our expectations, the number of weeds also increased as the number of crop species increased at both 4 and 8 weeks (Figure 3).
Similar to grass weeds, broadleaf weed abundances at four weeks was affected by year (X2 = 565.3983, P < 0.0001), block(year) (X2 = 416.6797, P < 0.0001), functional group (X2 = 48.4144, P = 0.0026), number of crop species(functional group) (X2 = 97.5839, P < 0.0001) and by herbicide application (X2 = 542.6166, P < 0.0001). Broadleaf weed abundances at eight weeks was also affected by year (X2 = 440.7721, P < 0.0001), block(year) (X2 = 314.8137, P < 0.0001), functional group (X2 = 17.0943, P = 0.0002), number of crop species(functional group) (X2 = 13.1857, P = 0.0043) and by herbicide application (X2 = 465.8260, P < 0.0001). Similar to grass weeds, 2019 had significantly more broadleaf weed in 2019 (14.1 weed per m2) than in 2020 (1.8 weeds per m2). there were more weeds in 2019 (6.93 weeds per m2) compared to 2020 (1.13 weeds per m2). Contrary to our expectations, the number of weeds also increased as the number of crop species increased at both 4 and 8 weeks (Figure 3).
The dry biomass for grass weeds at week 4 was affected by block (X2 = 32.3720 P < 0.0001), number of crop species(functional group) (X2 = 37.2270, P < 0.0001) and by the herbicide application (X2 = 22.2775, P < 0.0001), but not by functional group alone (X2 = 0.2868, P = 0.8664). By week 8, grass dry biomass was still affected by block (X2 = 1609.9042, P < 0.0001), functional group (X2 = 542.6166, P < 0.0001), number of species(functional group) (X2 = 23.0501, P < 0.0001) and by herbicides (X2 = 4.6364, P = 0.0313). Results were nearly the same for the broadleaf weeds at both 4 and 8 weeks. At 4 weeks, broadleaf weed biomass was affected by block (X2 = 199.4248, P < 0.0001), functional group (X2 = 15.8663, P = 0.0004), number of crop species(functional group) (X2 = 62.5565, P < 0.0001) and by herbicide application (X2 = 12.9406, P = 0.0003). At 8 weeks, broadleaf weed biomass was affected by block (X2 = 400.5189, P < 0.0001), functional group (X2 = 598.1765, P < 0.0001), number of crop species(functional group) (X2 =345.9913, P < 0.0001) and by herbicide application (X2 = 1478.6061, P < 0.0001).
Overall, Broadleaf weeds were more abundant than grass weed, at both 4 and 8 weeks (Figure 3). Interestingly, broadleaf weeds decreased in the four-species mixture at 8 weeks compared to 4 weeks. The overall dry biomass was relatively low at 4 weeks. However, at 8 weeks, broadleaf weeds in control plots were much greater than in herbicide plots. In comparison, grass weed biomass was similar in both herbicide and control plots (Figure 4). However, at both 4 and 8 weeks, the biomass of both broadleaf and grass weeds increased in 2-species mixtures compared to monocultures, but grass weed biomass was lower in the 4-species mixture than the 2-species mixtures, and broadleaf weed biomass was lowest in the 4-species mixture than either of the others (Figure 5). Although herbicides did suppress weeds overall, herbicides were best at suppressing broadleaf weeds. In comparison, weeds were similar at 4 weeks, but both grass and broadleaves were somewhat suppressed in both abundance and dry biomass by the 4-species mixture at 8 weeks. The 2-species mixture had no apparent suppression of biomass or abundance of either grass or broadleaf weeds.
Weed seed predation and predator abundances
Weed seed predation for pigweed was only affected by number of crop species (functional group) (X2 =10.9347, P = 0.0121) and herbicide application (X2 =11.6059, P = 0.0007), but not from block or functional group alone (P > 0.05). In comparison, only herbicide application affected foxtail consumption (X2 =13.6619, P = 0.0002). Block, functional group and number of crop species (functional group) did not have an effect on foxtail seed consumption (P > 0.05). There was a significant increase in both pigweed and foxtail seed predation in control plots compared to herbicide sprayed plots indicating the herbicide had a negative effect on weed seed predators, thereby reducing natural predators (Figure 6). The overall impact was nearly half the consumption of either pigweed or foxtail in herbicide-sprayed plots. Additionally, pigweed consumption was reduced by an increase in the number of crop species(functional group). Foxtail consumption decreased in either mixture compared to monocultures (Figure 7). These data show some benefit of additional weed seed consumption, especially considering the detrimental effect from herbicide application.
Weed seed predators were affected by block (X2 =13.0936, P = 0.0108) and number of crop species (functional group) (X2 =3.9530, P = 0.1386), but not by functional group alone (X2 =12.3000, P = 0.0064) or herbicide application) (X2 =0.2137, P = 0.6439). Monocultures had the highest number of weed seed predators overall – 1.825 per plot per 48hrs, followed by 4-species mixture with 1.2 weed seed predators per plot per 48hrs and finally the 2-spcies mixture with only 1.067 predators per plot per 48hrs. Because ground beetles varied throughout the field (block) and by the number of crop species within a plot, it is difficult to understand if beetles preferred difference in the field because of something related to the environment (i.e., slope, organic matter, etc), were directly attracted to specific crop species or mixtures or prevalence of weed seeds within plots.
Arthropod abundances
Total number of arthropods collected in pitfall traps were affected by year (X2 =282.2031, P < 0.0001), block(year) (X2 =197.2345, P < 0.0001), functional group (X2 =133.3364, P < 0.0001), number of crops species(functional group) (X2 =221.6660, P < 0.0001) and by herbicide application (X2 =63.2831, P < 0.0001). Similarly, arthropod predators collected from pitfall traps were affected by block(year) (X2 =107.1255, P < 0.0001), functional group (X2 =38.4158, P < 0.0001), number of crop species (functional group) (X2 =102.8788, P < 0.0001) and by herbicide application (X2 =14.4665, P = 0.0001), but not by year (X2 =0.9083, P = 0.3406). In 2019, overall predator abundances per plot was 16.2 compared to 34.0 total arthropods caught in pitfall traps (Table 4). In comparison, there was 16.3 predators and 19.3 total arthropods in 2020, indicating a much greater predator ratio in 2020 compared to 2019. Predators abundances were highest in grass type crop species compared to either broadleaf crops or combinations. However, total number of arthropods was greatest in both crop species, followed by grass, then broadleaf crops. Similarly, we saw an increase in total arthropods in the 4-species mixture and a decrease in 2 species mixture Compared to the monocultures. Of these, predators were greatest in the monocultures and fewest in the 4-species mixture. Finally, there was a decrease in both total arthropods and predators in herbicide applied plots compared to control plots.
Overall, we did not observe what we hypothesized. Although we expected an increase in arthropod abundances, we expected this to also result in greater predator abundances. We observed the opposite, with the greatest number of ground dwelling predators in monocultures. We are unable to determine the mechanism behind this - perhaps these predators were attracted to something else, such as a specific crop or something we did not control for in field experiments, such as soil structure, soil organic matter or something else. Or possibly, other higher predators such as bird or rodents were able to consumer these ground predators and reduce their numbers in crop species mixtures compared to monocultures. It could also be that the predators did not have enough time to catch up to the higher number of other arthropods in monocultures, specifically in 2020, in which samples were analyzed for July 1 compared to 2019 in which pitfalls were sampled July 31.
In comparison, although sweep netting of foliar arthropods was only done in 2019, foliar arthropods collected were affected by both block (F4,69 = 2.9586, P = 0.0268) and number of crop species (F3,69 = 4.9234, P = 0.0105). Functional group and herbicide application did not affect foliar arthropods (P > 0.05). Again, we see a significant affect from block in the arthropod community. Overall, it is likely that much of the arthropod variability and differences in abundances was due to field and environmental conditions that we did not control for. The repeated block affects may also indicate that plot sizes were too small to properly control for variability within the field and may have facilitated movement of arthropods between plots – similar to what one would observe as edge effects in a larger production field
Greenhouse Experiments
Broadleaf weed biomass was affected by replication (X2 =43.4031, P < 0.0001), the number of soybean plants (X2 =4.1120 7, P = 0.0401) and corn plants (X2 = 8.1956, P = 0.0042), but not their interaction (X2 = 0.0025, P = 0.9604). Additionally, the broadleaf weed biomass was affected by the covariant corn biomass (X2 = 5.2661, P = 0.0217) and soybean biomass (X2 =5.2661, P = 0.0401). In comparison, grass weed biomass was not affected by any factor or either soybean or corn plant biomass (P > 0.05). Broadleaf weed abundances were affected by replication (F3,47 =8.0935, P = 0.0003) only. All other factors had no effect on broadleaf weed abundances (P > 0.05). Similarly, grass weed abundances were affected by replication only (X2 =26.2907, P < 0.0001), not by any other factors (P > 0.05). In general, grass weed biomass and abundances were both low. Grass weed biomass was lower in almost treatments compared to the control, with only treatments with 6 and 8 corn plants having more grass weed biomass than the control. All treatments had lower broadleaf biomass than the control, indicating crops do suppress weed biomass to some extent. Although broadleaf biomass was highly variable, there was a general increase in broadleaf weed biomass with an increase in corn proportions as a crop mixture. This is in agreement with our hypothesis, that broadleaf plants (i.e. soybean) would better suppress grass weeds. Although grass weed biomass was generally low and not significant, there was slightly lower grass weed biomass with higher proportions of corn.
Table 1: Economic values ($USD) associated with inputs and yield for each treatment.
|
|
|
|
Value ($USD ha-1) |
Value ($USD MT-1) |
|
Input Costs |
Seed |
|
|
|
|
|
|
Corn |
247.11 |
|
|
|
|
Sorghum |
24.71 |
|
|
|
|
Sunflower |
39.54 |
|
|
|
|
Soybean |
|
|
|
|
Herbicide |
|
|
|
|
|
|
Dual |
67.06 |
|
|
|
|
Acuron |
88.96 |
|
|
|
Machinery |
|
|
|
|
|
|
Single pass |
26.19 |
|
|
|
Fuel |
|
|
|
|
|
|
Cost per MT |
|
3.93 |
|
Gross Return |
|
|
|
|
|
|
Forage |
|
|
|
|
|
|
Dry matter |
|
90.00 |
|
|
|
|
|
|
Table 2: Variable costs ($USD ha-1) are shown for each crop treatment as means. Different letters indicated significant difference between crop treatment input costs (Tukey’s HSD, P < 0.05).
|
Crop Treatment |
Variable costs ($USD ha-1) |
Return over variable costs ($USD ha-1) |
Yield (kg ha-1) |
|
Corn |
$486.82 a |
$576.33 ab |
11,812.78 a |
|
Sorghum |
$256.48 d |
$725.09 a |
10,906.34 ab |
|
Corn/Sorghum |
$385.01 b |
$393.40 bcd |
8,648.93 bcd |
|
Soybean |
$393.11 b |
$38.91 e |
4,800.18 e |
|
Sunflower |
$246.55 d |
$167.94 de |
4,605.50 e |
|
Soybean/Sunflower |
$326.78 c |
$255.63 cde |
6,471.31 de |
|
Corn/Soybean |
$471.56 a |
$449.88 bc |
10,238.24 abc |
|
Corn/Sorghum/Soybean/Sunflower |
$368.51 bc |
$301.07 cd |
7,439.85 cde |
Table 3: Means ± SE of ground beetles caught in 48hr pitfall traps.
|
No. Crop Species |
Pterostichus melanarius |
Poecilus chalcites |
Pterostichus spp. |
Amara arena |
Chlaenius tricolor |
Notrophilus acquaticus |
Amara spp |
Pterostichus mutus |
Poecilus lucublodus |
Harpalus affinis |
Harpalus spp |
Other/ Unknown |
Total |
|
1 |
0.925 ±0.26 |
0.025 ±0.025 |
0.450 ±0.19 |
0.025 ±0.03 |
0.075 ±0.06 |
0.175 ±0.08 |
0 |
0.025 ±0.03 |
0.025 ±0.03 |
0 |
0 |
0.100 ±0.05 |
1.825 ±0.33 |
|
2 |
0.633 ±0.16 |
0.033±0.033 |
0.133 ±0.13 |
0 |
0.033 ±0.03 |
0.100 ±0.06 |
0.033 ±0.03 |
0.033 ±0.03 |
0 |
0.033 ±0.03 |
0 |
0.033 ±0.03 |
1.067 ±0.21 |
|
4 |
0.700 ±0.50 |
0±0 |
0.200 ±0.13 |
0 |
0 |
0.200 ±0.13 |
0 |
0 |
0 |
0 |
0.100 ±0.10 |
0 |
1.200 ±0.49 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Both |
0.75 ±0.23 |
0 |
0.083 ±0.06 |
0 |
0.042 ±0.04 |
0.167 ±0.08 |
0.042 ±0.04 |
0 |
0 |
0 |
0.042 ±0.04 |
0 |
1.125 ±0.25 |
|
Broad |
0.5 ±0.16 |
0 |
0.600 ±0.27 |
0 |
0 |
0.200 ±0.09 |
0 |
0.033 ±0.03 |
0 |
0.033 ±0.04 |
0 |
0.133 ±0.06 |
1.500 ±0.32 |
|
Grass |
1.154 ±0.39 |
0.077 ±0.05 |
0.154 ±0.11 |
0.038 ±0.04 |
0.115 ±0.08 |
0.077 ±0.08 |
0 |
0.038 ±0.04 |
0.038 ±0.04 |
0 |
0 |
0.038 ±0.04 |
1.731 ±0.41 |
Table 4: Means ± SE are shown for total number of arthropods and number of predators collected from pitfall traps after 48hrs.
|
|
Total Number of Arthropods |
Total Number of Predators |
|
Year |
|
|
|
2019 |
34.0 |
16.2 |
|
2020 |
19.3 |
16.3 |
|
Number Crop Species |
|
|
|
1 |
28.7 |
18.4 |
|
2 |
22.4 |
14.6 |
|
4 |
31.2 |
12.7 |
|
Functional Group |
|
|
|
Both |
30.1 |
15.8 |
|
Broad |
24.1 |
16.3 |
|
Grass |
26.4 |
16.7 |
|
Herbicide |
|
|
|
Control |
29.9 |
17.5 |
|
Applied |
23.4 |
15.1 |
Overall, results did not strongly support our hypotheses; that higher diversity in mixtures would suppress weeds, increase arthropod predators or increase yields and economic value. Although each aspect tested in these experiments do show some promise, that with further research, could improve cropping systems to be more sustainable with crops species mixtures. For example, we saw that overcall corn and sorghum were better competitors than either of the broadleaf crops used in our experiments. Grown together, these species were competing with each other, but grown with species that complemented them, such as corn and soybean mixed, or soybean and sunflower, there were benefits in yield. Furthermore, economic value may have benefits such as reduced seed costs that we were able to show with sorghum compared to corn.
Mixtures did increase overall arthropod abundances, as was expected, but not predator abundances. Our data showed that there were more predators in monoculture compared to our mixtures, although mixtures with higher functional diversity or more species did have a higher number of total arthropods.
Finally, weed suppression from corn may the greatest benefit. We were able to show a greater suppression of broadleaf weeds in greenhouse experiments due to higher proportions of corn. However, there was no difference in abundances of either weed type related to functional group as was shown in greenhouse experiments. Similarly, and likely linked to the higher number of ground dwelling predators, there was also a higher consumption of both pigweed and foxtail seeds in field monocultures compared to either mixture.
Education & Outreach Activities and Participation Summary
Participation Summary:
In both 2019 and 2020, education and outreach related to this project related primarily to working with undergraduate students. We had 3 undergraduate students helping from May-September 2019, and 2 students working with us on this project in 2020. I was able to use the project to help show weed identification to the students and we facilitated several open discussions about the hypotheses and ideas behind the project. Furthermore, using pitfall traps and sweep netting, the students were able to see and help identify several different species of insects down to family. This project allowed for direct training of both weed and arthropod ID, followed by several opportunities to practice field-based ID.
Additionally, one of the undergraduate students, Hayden Bock, had an interest in exploring soil-related components (compaction) among different crop types and arthropod weed seed predators. Both John Tooker and I worked with Hayden to help him develop an independent research project to explore his interests in soil compaction, weed seed arthropod predators and forage species mixtures. We used this SARE project as a starting point to help Hayden design his own objectives.
Project Outcomes
This project contributes to agricultural sustainability because we’ve shown some promising benefits of incorporating crops with different functional groups in mixture – specifically, the benefit of incorporating broadleaf crops. The combination of broadleaf crops shows complementarity to high-yielding grass crops, and potential for increased yield with lower input costs, thereby potential for higher economic return for farmers as well. Additionally, it was specifically broadleaf crops that were better able to suppress broadleaf weeds – the most problematic weeds in our study.
We’ve developed an additional research project related to this to further evaluate weed seed predators and their preference for specific weed seeds.
Additional research should be conducted to find the exact proportions of specific crops that work best together. Population of crops and additional research on which inputs (i.e., how much fertilizer) would also be beneficial. Furthermore, expanding this work into areas beyond central Pennsylvania may show different results and would provide further insight into some of the mechanisms responsible for the results we observed.