Substituting vetch and chicory for rye in a cover crop mixture enhanced nutrient release

The study was conducted at the Centre for Sustainable Food Systems at the UBC Farm from 13 May 2014 to 26 Aug. 2014. The soil is a coarse-textured, sandy loam Bose Humo-Ferric Podzol. The cover crops used in this study were rye S. cereale L. ‘common’, Lana vetch (referred to as vetch) Vicia villosa Roth subsp. dasycarpa (Ten). Cavill. ‘Lana’, and chicory C. intybus L. ‘Puna’. Chicory was chosen rather than other forbs within Brassicaceae due to the heavy use of brassicas as a cash crop in the site’s crop rotations. Six residue combinations were used: rye monoculture, chicory monoculture, vetch monoculture, rye:vetch biculture, rye:vetch:chicory polyculture, and a soil only control. Treatment weights were based on regional yields and site data for the monoculture of the cover crops: ∼5000 kg·ha⁻¹ for rye, ∼2000 kg·ha⁻¹ for vetch (Lawson et al. 2013), and 927 kg·ha⁻¹ for chicory. Polyculture proportions of dry weight were 80% rye and 20% vetch in the biculture and 70% rye, 10% vetch, and 20% chicory in the polyculture. Proportions were designed following a separate greenhouse study (data not shown). The total weights of the polycultures were consistent with the rye monoculture at 23 g or ∼5000 kg·ha⁻¹. The final dry weights for each treatment were vetch 9.03 g; chicory 4.24 g; rye 22.90 g; rye:vetch 18.25 g rye and 4.59 g vetch; and rye:vetch:chicory 15.92 g rye, 2.33 g vetch, and 4.31 g chicory. The study was executed through harvesting cover crop tissue from a source field and placing them in a separate field in litter bags. Prior to harvest, rye had stems extended but without seed heads visible (∼feekes stage 10), vetch had visible flowers, and chicory had leaf tissue but no stems. Aboveground tissues were randomly sampled from established plots on 6 May 2014 and stored at 4 °C prior to burial on 13 May 2014.

To measure nutrient release, tissues were placed inside fiberglass screen and buried. Fiberglass screen, 1 mm mesh size, was cut to 0.2 m × 0.228 m, folded in half, and sealed along three edges using epoxy. Fresh residues were clipped into 5-cm-long pieces and placed in mesh bags based on dry-weight targets. Four Plant Root Simulator (PRS^®) Probes (Western Ag, Saskatoon, SK, Canada), two made of anion-exchange resins and two cation-exchange resins, were placed alternating in each bag and then filled with the appropriate plant tissue. About 60 g of soil was then added to ensure residue contact with the probe membranes. A sample of each species was dried at 60 °C for 72 h to calculate percent dry weight. A commercial laboratory, Pacific Soil Analysis, measured initial residue nutrient concentrations using a LECO Total N Analyzer for total C and the Kjeldahl method for total N.

Prior to the incubation, the experimental field was grown in an organic rotation featuring carrots, beets, beans, and spinach. Soil cores were taken at a 15 cm depth and analyzed by Pacific Soil Analysis, for pH using a 1:1 ratio of soil to distilled water with a pH meter, total C as above, organic matter using the Walkley–Black wet oxidation method, total N using a Technicon Autoanalyser on a semi-micro Kjeldahl digest, and available phosphorus (P) colormetrically using the ascorbic acid color development method on a 1:10 soil to Bray (NH₄F) extract. Soil in the experimental plots had pH = 6.0, total C = 4%, organic matter = 7.0%, total N = 0.2%, and available P = 69.3 ppm. The incubation was set up as 3 (2 m × 3 m) blocks and blocked lengthwise across the experimental field. In each experimental area, a trench of 15 cm depth was dug, and the bags were spaced 15 cm apart with probes on the bottom side with six bags arranged in each block. A mesh sheet made of the same material of the bags was placed over each block and backfilled with soil level to the ground. The mesh sheet covering each block was used to allow access to the litter bags without destructive harvesting of each bag. Every second week after the start of the experiment, the mesh cover was lifted up, and new probes were replaced through removing the existing probes and placing new probes under the litter. Soil was then backfilled to cover the mesh and continue the incubation. At the start of the experiment, the field was irrigated to field capacity to simulate a strong spring rain as is typical in this region and to ensure contact between the resin membranes and the tissue. No additional irrigation was applied throughout the experiment. All plots were weeded bi-weekly. The field was not planted but remained bare, in line with the goals of the incubation to assess the release of plant available nutrients in the absence of root uptake.

Western Ag analyzed the probes for nitrate and ammonium colorimetrically using an automated flow injection analysis system and all remaining ions with inductively coupled plasma spectrometry.

Data analyses were conducted using R version 3.5.0. For cumulative available N, a linear mixed-effects model was used with package nlme. A full model was tested with a crossed interaction between treatment and time with block as a main effect with a random intercept for each litter bag. Two models were tested with time as either a categorical variable, to allow comparison of means, and as a continuous variable as an orthogonal quadratic polynomial. Nutrient release in the first week was analyzed using an analysis of variance including a block effect followed by a Tukey’s post-hoc test if P < 0.05 using the false discovery rate correction. Cd, Mn, Pb, and Cu were below levels of detection.

The quantity and time of release of available N depended on the residue treatment combinations (Fig. 1). A linear mixed-effects model demonstrated a significant interaction between time as a quadratic polynomial and residue treatments (P < 0.001). The release of N in the vetch monoculture was linear, whereas the release from the other treatments increased markedly in the second half of the growing season. To compare between treatment means in each time period, a second linear mixed-effects model was tested and had a significant interaction between time as a categorical variable and residue treatments (P < 0.001). In the first two sampling periods, there were no treatment differences. In the third and fourth sampling periods, the vetch monoculture was greater than those without vetch (rye, chicory, and control) but similar to the polycultures (rye:vetch and rye:vetch:chicory). All treatments with vetch were equivalent until the final sampling period when the rye:vetch:chicory polyculture was greater than the rye:vetch biculture. The rye, chicory, and control treatments did not differ, except for during the seventh sampling period when chicory was greater than rye. It was during the seventh sampling period when the rye:vetch:chicory polyculture began to increase relative to the other treatments suggesting it is N released from chicory that caused the rye:vetch:chicory treatment to release the most N at the final time point (see inset in Fig. 1).

How can we explain why substituting chicory for rye increased the quantity of N released? C/N ratios for the tissue were 10.4 for vetch, 21.8 for chicory, 41.3 for rye, 35.1 for rye:vetch, and 34.4 for rye:vetch:chicory. The low C/N ratio for vetch helps explain why it released more N than rye. This is a similar result to a litter bag study showing high N release from vetch mulch (Halde and Entz 2016). Rye was harvested with stems present containing structural tissue that may be slow to decompose as grasses can have a higher proportion of structural carbohydrates that impede N mineralization (Ranells and Wagger 1996). The C in chicory may occur in more labile forms than in grasses which are more accessible to microbial degradation promoting N mineralization. However, although tissue chemistry offers one explanation, the N release for chicory likely depends on initial soil N conditions and the proportion of legume biomass in the mixture.

The results for nutrient release after 1 wk were mixed, demonstrating differences in release for ammonium, P, zinc (Zn), and sulfur (S) (Table 1). There was a treatment effect on ammonium release (F_[5,11] = 10.1, P < 0.01) but not for nitrate (F_[5,11] = 2.7, P = 0.08). Post hoc comparisons showed that vetch was greater than all other treatments for ammonium (Table 1). The ratio of ammonium to nitrate across all treatments was 5.4 in week 1 compared with 0.06 across the time series. The total available N previously described (Fig. 1) was mainly nitrate, and therefore, ammonium was not separately analyzed (data not shown). There was a treatment effect on phosphate release (F_[5,11] = 6.1, P = 0.03). Both vetch and rye:vetch:chicory had greater P release than the soil control, and vetch was also higher than rye. There was a treatment effect on Zn (F_[5,11] = 9.8, P < 0.01) with vetch being higher than all treatments except rye:vetch:chicory. There was a treatment effect on S (F_[5,11] = 19.4, P < 0.01). Chicory had a greater S release compared with all other treatments. The release of nutrients likely correlates with initial tissue chemistry that was not measured in this study. Certain cover crops may mobilize nutrients from inaccessible forms such as mineral P and then release those nutrients through decomposition into plant available forms. These preliminary results motivate future studies examining how functional diversity can be used to promote the cycling of nutrients in systems with few external inputs such as with organic rotations.

One strategy to enhance the ecosystem functioning of agroecosystems is to increase the functional diversity of cover crops (Finney and Kaye 2017). Increasing the functional diversity of cover crops provided comparable N and P release to the best monoculture in the legume vetch. These results support previous research on grass:legume bicultures that show increased nutrient release relative to grass monocultures (Ranells and Wagger 1996; Odhiambo and Bomke 2000; Rosecrance et al. 2000), while also adding a new understanding of the potential to use forbs in a cover crop polyculture. We did not observe that the polyculture mixture released greater nutrients than the legume monoculture. However, reducing the quantity of rye tissue by 30% to include 10% chicory and 20% vetch increased the total amount of N released by 257%. Our results suggest legumes may release N early in the growing season, whereas forbs may release N later in the growing season compared with grasses with a high C/N ratio that reduce the overall quantity of N released. Modulating the ratio of grass:forb:legume in a cover crop polyculture offers a management strategy to fine-tune nutrient release to synchronize with the uptake of a given cash crop.