Cover crop mixtures were designed to fulfill different ecological goals, and we evaluated their impact on weed population density and species communities across a wide geographical area in central California. We hypothesize that both functionally uniform and diverse cover crops can effectively provide orchard ground cover that displaces weeds, but that functionally diverse cover crops will emerge and compete for resources more consistently across growing seasons and locations and are therefore better able to impact weed community composition. To evaluate these hypotheses, we examined indicators of cover crop function: 1) elimination of bare ground compared to ground cover provided by weedy resident vegetation, 2) relative incidence of cover crops and weeds, 3) stability of cover crop incidence over space and time, and 4) downstream impacts on weed communities.Experimental design and soil cover treatments. We compared the effects of several ground cover treatments in replicated large-plot experiments in commercial almond orchards in Tehama, Merced, and Kern Counties in California. These locations span nearly 600 km in the Central Valley of California, including a range of environmental variables, cannabis racks especially rainfall. This region produces virtually all of the almonds in the United States .
The experiment used a randomized complete block design with four replications of three or four soil cover treatments at each site, and practices were implemented for two years on the same plots, beginning in the fall of 2017 and ending in the late summer of 2019. Plots were about 25.5 m wide at each site, encompassing four orchard alleys , and the entire length of the orchard management units . Two different winter cover crop mixes were planted in orchard alleyways. The “uniform” mix consisted of five functionally similar species that are designed to provide diverse floral resources for honeybees to support almond tree pollination. This mix is used commercially in California and distributed as ‘PAm Mustard Mix’ by the Project Apis m. Seeds For Bees program. The mix was comprised of 35% canola , 15% ‘Bracco’ white mustard , 15% ‘Nemfix’ yellow mustard Czern 20% daikon radish , and 15% common yellow mustard . The uniform mix was planted at 9 kg per planted ha. The “diverse” mix consisted of five species from the grass, brassica, and legume groups that are commonly used together in functionally-diverse cover crop mixes . This mix was comprised of 10% ‘Bracco’ white mustard, 10% daikon radish, 30% ‘Merced’ rye , 20% ‘PK’ berseem clover , and 30% common vetch . The diverse mix was planted at 56 kg per planted ha. Two control treatments were also implemented which reflected mainstream orchard management practices of winter vegetation. The “resident” vegetation treatment involved winter vegetation management with mowing and seasonal herbicide applications which allowed resident vegetation growth. The “bare” treatment involved multiple herbicide applications, as determined by grower cooperators, to eliminate winter vegetation.
The Tehama site included only the resident treatment to better reflect standard practices in this region of California which has more abundant winter rainfall. The Merced and Kern sites featured both the resident and bare treatments to better reflect high intensity production systems in these regions. Sites and horticultural management. The study was designed to use commercially relevant spatial and temporal scales, and orchard management was determined by grower cooperators for agronomic relevance. All orchards were equipped with microsprinkler irrigation, and irrigation schedules were determined based on almond evapotranspiration models in accordance with local weather conditions and recommendations. Conventional irrigation, insecticide, fungicide, and fertilizer treatments and rates were determined by each grower and applied to the tree rows only. Tree rows were maintained with conventional herbicide programs to create vegetation-free zones at the base of trees. Each of the sites was subjected to regular traffic from machinery and farmworkers to complete these orchard management operations throughout the cover crop growing season. The Tehama County orchard was located in the northern Sacramento Valley on Kimball loam soils . Average precipitation at the site is 645 mm annually. The site was planted in 2016 with almond varieties ‘Nonpareil’ and ‘Monterey’ in alternating rows. Cover crops were drill seeded in a 3.6 m wide swath down the alleyways on November 6, 2017 and November 9, 2018 and mowed for termination on March 30, 2018 and May 25, 2019. The young trees were pruned in February 2018, and every other alley was subsequently mowed to mulch tree prunings. No data described in this paper were collected from those mowed alleys.
The whole orchard was mowed on January 29, 2019 to destroy unharvested nuts for navel orange worm sanitation; data were collected from cover crop regrowth after this mowing event. Frost during almond bloom was a concern at this site, and irrigation was applied in 12-hour long sets to mitigate forecasted frosts in February or March of each year, which is outside of typical almond irrigation timings. The Merced County orchard, planted in 2008, was located in the northern San Joaquin Valley on Alamo clay soils . Average precipitation at the site is 325 mm annually. The site had 50% ‘Nonpareil’ and 12.5% each ‘Monterey’, ‘Fritz’, ‘Carmel’, and ‘Wood Colony’ almond varieties, with ‘Nonpareil’ in every other row and the remaining varieties mixed evenly in the alternate rows. Cover crops were direct seeded on November 2, 2017 in a 3.6 m wide swath with a seed drill and mowed for termination on April 9, 2018. In year two, cover crops were broadcast planted on December 21, 2018 with a rotary spreader and mowed on March 19, 2019 for navel orange worm sanitation following data collection and again on April 12, 2019 for final cover crop termination. The first replicate of the uniform mix was not planted at this site in 2017, and data from that plot was not included in the analysis.The Kern County orchard, planted in 2006, was located in the southern San Joaquin Valley on primarily Hesperia sandy loam . Average precipitation at the site is 180 mm annually. The site had 50% ‘Nonpareil’ and 25% each ‘Monterey’ and ‘Fritz’ almond varieties, with ‘Nonpareil’ in alternate rows and the other two varieties evenly mixed in every other row. A 4.8 m wide swath was planted down the center of each orchard alley. Cover crops were direct seeded on October 30, 2017 and mowed for termination on April 2, 2018. In year two, cover crops were planted on November 1, 2018 and mowed on April 5, 2019. Immediately prior to both planting dates, indoor grow rack alleyways across the whole orchard were disked for seedbed preparation and ground leveling. Supplemental irrigation was applied across the orchard in 20-hour long sets throughout the winter of 2017-2018 to support the cover crop a45hapiro45gate frost concerns. At this site, the bare ground cover treatment only involved a deep ripping tillage operation to address soil compaction. Data collection. Orchard alley plant communities were evaluated with point-intercept transects. Each plot was surveyed with a single 50 m long transect with points observed evenly at each meter along the transect. Each transect was placed beginning 75 m from the end of the second tree row over from the edge of each plot. The transect extended diagonally across a single orchard alley, starting and ending on opposite edges of the planted swath. Plant incidence was observed for the top layer of vegetation, with occurrence of one actively growing plant or bare ground recorded at each point along the transect. Therefore, incidence is a relative measure of how much ground cover is associated with each vegetation type. Plants were identified to species visually, except in the case of the white and yellow mustards in the uniform mix which were identified as one operational taxonomic unit due to morphological similarities. Transects were surveyed on March 29, 2018 and March 22, 2019 at the Tehama site, March 30, 2018 and March 15, 2019 at the Merced site, and March 27, 2018 and March 16, 2019 at the Kern site. These timings coincide with cover crop flowering for most species as well as winter weed flowering for many endemic species in the study area. Statistical analysis. Analyses were performed in R 4.0.3 . Comparisons of bare ground among treatments were made with ANOVA. ANOVA assumptions were inspected visually with qqPlot from the car package , and subsequently the response variable was arcsine square root transformed to deal with a heavy tailed distribution.
One outlier was identified with the Bonferroni outlier test using outlierTest. This outlier value was excluded from further analyses because it was collected in the same plot at the Merced site that had been previously excluded because it had not been planted in 2017 . Finally, normality of the transformed, outlier-free model was formally assessed with a Shapiro-Wilk test usi46hapiroiro.test. Models with combinations of possible predictor variables , modeled as fixed effects due to the number of sites and years in this study) were compared with Aikake information criterion using the aictab function from the AICcmodavg package . The best model included treatment, site, and their 2-way interactions as predictors, and neither year nor block were included in the final model. The resulting ANOVA analysis was performed with Anova from the car package, and contrasts were made with least-squares means using the emmeans package . Associations between cover crop and weed incidence were analyzed using linear models. Linear models were created with the lm function in base R. Weed incidence was the responsevariable, and we created models with cover crop incidence and cover crop treatment , both with and without their interaction terms, as well as a model which only included cover crop incidence. Linear models were compared with the anova function from base R. Linear regression including only cover crop incidence as a predictor for weed incidence was statistically similar to linear regression that additionally used cover crop mix or cover crop mix and the two-way interaction as predictors. Therefore, we considered the most parsimonious model with only cover crop incidence as a predictor of weed incidence. Cover crop stability was assessed by comparing coefficients of variation for incidence of each cover crop mix as pooled across sites and years in this study. Pooled coefficients of variation were compared with the modified signed-likelihood ratio test as implemented in the cvequality package . Weed communities in the different cover crop treatments were analyzed with non-metric multidimensional scaling . NMDS was based on Bray-Curtis dissimilarity and was calculated using the metaMDS function in the vegan package . Cover crops influenced weed communities but to different extents depending on the site and year. Throughout the springtime evaluations in this study, we observed five weed species at the Kern site, six weed species at the Merced site, and 22 weed species at the Tehama site. The Kern site primarily included annual bluegrass and common chickweed Vill with lesser populations of little mallow , shepherd’s purse Medik and Italian ryegrass Husnot. The Merced site also had large populations of annual bluegrass and common chickweed, as well as little mallow, whitestem filaree L’Hér. California burclover , and wild oat . The Tehama site had significant populations of annual bluegrass, common chickweed, shepherd’s purse, whitestem filaree, buckhorn plantain , chicory , annual sowthistle , field bindweed , and bermudagrass Pers. The remainder of the species at the Tehama site were primarily dicotyledonous, winter annual species, with lesser populations of some grasses and summer annual or perennial dicotyledonous species. Weed communities clustered by cover crop treatments , though sites also predicted weed communities and effect sizes were generally small . While no fixed factors significantly explained weed communities in Merced, year was a significant factor in Tehama and cover crop treatment was a significant grouping factor for weed communities in Kern .Orchard cover crop mixes, as implemented in this study, were effective at establishing, reducing bare soil, and suppressing weeds. However, these effects were highly variable, and there is little evidence that the cover crop mixes we used had fixed impacts on the composition of orchard weed communities. Differences in management and climate at each site year, especially as related to cover crop planting, spring mowing, and weather conditions during cover crop establishment, likely contributed to this variability.