Introduction
Growing concerns about the need to balance food production with biodiversity and natural resources conservation has led to an increased interest among agroecologists, producers, and the public in reducing the reliance on off-farm synthetic inputs to provide the conditions necessary for sustained agronomic production (Altieri, Reference Altieri1999; Foley et al., Reference Foley, Ramankutty, Brauman, Cassidy, Gerber, Johnston, Mueller, O'Connell, Ray, West, Balzer, Bennett, Carpenter, Hill, Monfreda, Polasky, Rockström, Sheehan, Siebert, Tilman and Zaks2011). To be successful, this approach to farming requires that the agronomic, economic and ecological conditions remain favorable for crop production (Robertson and Swinton, Reference Robertson and Swinton2005). The need to reconcile these apparently disparate goals has precipitated a call for the development of ecologically based management systems (Matson et al., Reference Matson, Parton, Power and Swift1997; Robertson et al., Reference Robertson, Allen, Boody, Boose, Creamer, Drinkwater, Gosz, Lynch, Havlin, Jackson, Pickett, Pitelka, Randall, Reed, Seastedt, Waide and Wall2008; Reganold et al., Reference Reganold, Jackson-Smith, Batie, Harwood, Kornegay, Bucks, Flora, Hanson, Jury and Meyer2011). Such systems are those that augment ecological processes or community structure to support crop production and pest management (Magdoff, Reference Magdoff2007; Altieri et al., Reference Altieri, Ponti, Nicholls, Gurr, Wratten and Snyder2012). One such practice is the use of cover crops, a suite of non-marketable plants grown to improve soil quality (Dabney et al., Reference Dabney, Delgado and Reeves2001).
Cover crops can improve physical, chemical and biological soil properties through several mechanisms. For example, cover crop root growth can increase soil macroporosity, thereby increasing saturated hydraulic conductivity, and plant growth can increase evapotranspiration, thereby increasing water storage capacity. Both processes improve water infiltration and may help prevent soil surface sealing (Dabney, Reference Dabney1998; Dabney et al., Reference Dabney, Delgado and Reeves2001). Humification of cover crop residues can increase soil organic matter (Reeves, Reference Reeves, Hatfield and Stewart1994; Hartwig and Ammon, Reference Hartwig and Ammon2002; Lal, Reference Lal2004), thus increasing cation exchange capacity (CEC), enhance mobilization of recalcitrant macronutrients (Kamh et al., Reference Kamh, Horst, Amer, Mostafa and Maier1999) and provide food for soil microorganisms (Wild, Reference Wild1993; Hu et al., Reference Hu, van Bruggen and Grünwald1999). If legumes are included in cover crop mixtures, biological nitrogen (N) fixation and subsequent mineralization can further increase plant-available N (Snapp et al., Reference Snapp, Swinton, Labarta, Mutch, Black, Leep, Nyiraneza and O'Neil2005). Cover crops can also be an important component of integrated pest management (IPM) strategies by reducing the per capita weed seed production via competitive exclusion (Gallandt et al., Reference Gallandt, Liebman and Huggins1999), decreasing weed seed survival through microbially mediated seed decay (Dabney et al., Reference Dabney, Schreiber, Rothrock and Johnson1996; Liebman and Davis, Reference Liebman and Davis2000), and minimizing weed emergence either by preventing light transmittance or through allelopathy (Teasdale et al., Reference Teasdale, Beste and Potts1991; Gallandt et al., Reference Gallandt, Liebman and Huggins1999). Finally, cover crops can provide habitat for beneficial organisms such as parasitoids of phytophagous insects and weed seed predators, enhancing the effectiveness of conservation biological control in highly disturbed agroecosystems (Barbosa, Reference Barbosa and Barbosa1998; Landis and Menalled, Reference Landis, Menalled and Barbosa1998; Altieri and Nicholls, Reference Altieri, Nicholls, Collins and Qualset1999).
A major drawback to the use of cover crops is that producers do not generate direct revenue during the season in which they are grown (Sulc and Tracy, Reference Sulc and Tracy2007; Thiessen Martens and Entz, Reference Thiessen Martens and Entz2011). One approach to overcome this challenge may be integration of livestock grazing for cover crop termination. Integrating livestock may offer producers several benefits besides a method of terminating cover crops. First, livestock production may provide alternative sources of revenue through the production of food (e.g., meat and dairy products) and fiber (e.g., wool) or through grazing leases (Gardner et al., Reference Gardner, Faulkner, Hargrove and Hargrove1991; Franzluebbers, Reference Franzluebbers2007; Thiessen Martens and Entz, Reference Thiessen Martens and Entz2011). Secondly, livestock grazing could be a component of an integrated weed management program because it may reduce the number of flowering ramets producing seeds (Popay and Field, Reference Popay and Field1996; Meissner and Facelli, Reference Meissner and Facelli1999). Finally, livestock grazing may aid in nutrient cycling through inputs of urine and feces (Bakker, Reference Bakker, WallisDeVries, Wieren and Bakker1998; Thiessen Martens and Entz, Reference Thiessen Martens and Entz2011).
Despite these agronomic benefits, at least two potential adverse outcomes could arise from the integration of livestock grazing. First, trampling by livestock could alter soil physical properties such as macroporosity and compaction (Franzluebbers, Reference Franzluebbers2007). However, if grazing occurs when soils are dry, the compaction is usually limited to the top 10 cm of soil, and naturally attenuated by freeze–thaw and wetting–drying cycles (Bell et al., Reference Bell, Kirkegaard, Swan, Hunt, Huth and Fettell2011). Second, grazing represents a potential mass export of nutrients as the animals leave the fields. However, Thiessen Martens and Entz (Reference Thiessen Martens and Entz2011) noted that while grazing ruminants can retain up to 25% of the N they consume, the remaining N, deposited as urine and feces, is often more labile than plant detritus.
Demand for produce from horticultural market-gardens has grown substantially in recent years (Low and Vogel, Reference Low and Vogel2011). Consumer preferences, costs and environmental concerns often prohibit intensive use of off-farm synthetic inputs in these systems. Thus, the development of ecologically based management practices is especially important for market-garden vegetable production systems. While there has been substantial research on the agronomic consequences of cover crops (Dabney, Reference Dabney1998; Dabney et al., Reference Dabney, Delgado and Reeves2001; Hartwig and Ammon, Reference Hartwig and Ammon2002; Snapp et al., Reference Snapp, Swinton, Labarta, Mutch, Black, Leep, Nyiraneza and O'Neil2005; Tillman et al., Reference Tillman, Smith, Holland, Gurr, Wratten and Snyder2012), most of this research has focused on their use in large-scale commodity production. In contrast, there has been substantially less investigation into cover crop use in market-garden vegetable production. Furthermore, there is a paucity of research on the effects of integrating livestock grazing in market-garden vegetable production (but see Franzluebbers, Reference Franzluebbers2007; Kahimba et al., Reference Kahimba, Sri Ranjan, Froese, Entz and Nason2008; Thiessen Martens and Entz, Reference Thiessen Martens and Entz2011: McKenzie et al., Reference McKenzie, Goosey, O'Neill and Menalled2016).
To fill this knowledge gap, we compared the agronomic effects of sheep-grazing for cover crop termination with those of mowing over two consecutive years in a vegetable production farm. This on-farm study consisted of two phases. In the first phase, we assessed the impacts of sheep grazing as a method of termination on weed pressure and soil quality, as well as the forage quality of the cover crop. In the second phase, we estimated yield in subsequently grown cash crops. We had two trials of our experiment, one trial beginning its first phase in 2012 and the second trial beginning its first phase in 2013. Our research was designed to address four main questions: (1) Does sheep grazing for terminating cover crops remove as much plant biomass as mowing, a common approach to cover crop termination? (2) Does sheep grazing affect soil physical and chemical properties differently than mowing? (3) Can cover crops serve as viable forage for sheep? (4) Do cash crop yields differ between the two management strategies in the subsequent year?
Materials and Methods
Study site
Our study was conducted at Townes Harvest Farm (THF), a 1.2 ha certified organic, irrigated, diversified vegetable farm on the campus of Montana State University—Bozeman (45°40′N, 111°4′W). The farm is divided into six 40 × 35 m2 units, each following a 6-year rotation beginning with a cover crop season (Year 1) and followed by cash crops in the subsequent five growing seasons (Years 2–6). In a single growing season, each year of the rotation is represented by one unit. THF has a 4-year yield of marketable crops of 10,200 kg (C. Holt, Farm Manager, Personal communication) and has Turner loam (fine loamy over sandy, mixed, superactive, frigid and typic Agriustoll) soil with a clay loam texture, based on our particle size analysis showing 25% sand, 44% silt and 30% clay. THF receives approximately 380–480 mm of annual precipitation and has a mean annual air temperature ranging from 3.9 to 7.2°C (NRCS, 2013). Growing season mean monthly temperatures, total monthly precipitation and 30-year means from US Climate Reference Network Station USC00241044 are in Table 1 (Diamond et al., Reference Diamond, Karl, Palecki, Baker, Bell, Leeper, Easterling, Lawrimore, Meyers, Helfert, Goodge and Thorne2013).
Table 1. Mean monthly temperatures and total precipitation observations for 2012–2014 growing seasons and 30-year normal mean monthly temperature and total precipitation for Bozeman, MT, USA.
Data are reported from US Climate Reference Network Station USC00241044 on the Strand Union Building of Montana State University, Bozeman, MT, USA (Diamond et al., Reference Diamond, Karl, Palecki, Baker, Bell, Leeper, Easterling, Lawrimore, Meyers, Helfert, Goodge and Thorne2013).
Cover crop phase experimental design
The cover crop phase followed a single factor, completely randomized design with two treatment-levels (sheep-grazed or mowed for cover crop termination) and three replicates per treatment-level. Each replicate consisted of a 10 × 15 m2 rectangular plot. On June 8, 2012 and on June 25, 2013, the farm manager cultivated the soil in all plots and seeded a cover crop consisting of 56 kg ha−1 buckwheat (Fagopyrum esculentum Moench), 23 kg ha−1 beets (Beta vulgaris L.), 11 kg ha−1 sweetclover [Melilotus officinalis (L.) Lam.] and 68 kg ha−1 pea (Pisum sativum L).
Between August 3 and 7, 2012, we terminated the cover crops at anthesis by either tractor mowing or sheep grazing. Similar treatments were imposed between August 7 and 11, 2013. For the sheep-grazing plots, we set up temporary electrical fences charged between 3500 and 6000 V, stocked each plot with 6–11 Rambouilet yearling rams and allowed them to graze ad libitum until the cover crop appeared >90% removed. In each grazed plot, we placed large watering troughs to provide the sheep with supplemental water. For the mowed plots, the farm manager cut all plant material using a Flex Hitch Rotary Kutter (King Kutter, Winfield, AL) mowing deck with a 1.22 m effective swath width, leaving all plant biomass in situ.
Cash crop phase experimental design
In 2013 and 2014, three 1 m wide seedbeds were tilled to a depth of 25–35 cm with a 1.07 m-diameter spader (Celli Co., Flori, Italy) through the previously grazed or previously mowed cover crop plots. The farm manager planted each seedbed with either kohlrabi (Brassica oleracea L. var. gongyloides L., 15 plants m−2), spinach (Spinacia oleracae L., 21 plants m−2), or lettuce (Lactuca sativa L., 15 plants m−2). This component of our study followed a split-plot design with one subplot per factor-level within each whole plot, with cover crop termination method as the whole-plot factor and cash crop as the subplot factor. Planting and harvesting dates for cash crops are listed in Table 2.
Table 2. Crop planting and harvest dates for 2013 and 2014 cash crop phases at Towne's Harvest Farm, Bozeman, MT, USA.
Soil physical and chemical characteristics
We assessed the impact of cover crop termination strategies on soil quality by measuring several soil physical and chemical parameters. Post-termination soil moisture was measured from September 13 to November 20, 2012 and from August 13 to October 9, 2013 using 7 cm soil moisture data probes (HOBO U30 Station, Onset Computer Corp., Bourne, MA) placed near the center of each plot to a depth of approximately 7 cm. We measured soil temperature by placing four temperature loggers (iButton® model DS1921G, Maxim Integrated, San Jose, CA) approximately 10 cm below the soil surface near the center of each plot. Soil temperatures were measured from June 13 to August 15, 2012 and from June 28 to October 9, 2013. All probes were removed from August 5 to 12, 2013 to accommodate cover crop termination, and were reinstalled on August 13, 2013. Temperatures within a plot from all recovered data loggers were averaged to obtain mean plot temperature through time.
We assessed foliar N content, an estimate of resource availability, of the aboveground plant biomass at anthesis in the cover crop phase. Foliar N is a sensitive index of total N because plants uptake N over a larger spatial scale than would be captured by a soil sample (Hausenbuiller, Reference Hausenbuiller1985). Furthermore, usually <1% of soil N is NO3 – at any point in time, whereas immobilized foliar N in detritus and living plant tissue comprises the largest pool of N in terrestrial ecosystems (Foth and Ellis, Reference Foth and Ellis1997). Foliar N was estimated as part of the crude protein assay during the cover crop phase (see ‘Forage quality’ section for details). Samples were initially ground in a plant matter grinder with a 2 mm screen (Wiley® Mill, Thomas Scientific Inc., Swedesboro, NJ) and further ground in a laboratory grinding mill (Cyclone Mill, UDY Corporation, Fort Collins, CO).
To assess soil conditions prior to seeding the cash crop phase, soil compaction and soil nutrient content were measured on April 28, 2013 and on May 12, 2014. Compaction was assessed using a soil penetrometer at four locations per plot and at four depth ranges per sampling location: 0–15, 15–30, 30–45 and 45–60 cm. Values from the four sampling sites were averaged for each depth within each plot. Soil nutrient content was evaluated with two randomly located soil core samples per plot using a hydraulic auger with a 5 cm diameter. For each sample, we separated the soil by depths as above and combined them into one sample per depth and plot. All soil samples were sent to a third party soil analytics laboratory (Agvise Laboratories, Northwood, ND) to be assayed for Olsen P, potassium, calcium, electrical conductivity, CEC, organic matter and pH.
Forage quality
Prior to terminating the cover crop, we measured its forage quality on July 31, 2012 and on August 2, 2013 by collecting all above ground plant material from four 0.33 m2 quadrats randomly placed within each plot and combining biomass from all quadrats. Samples were dried at 60°C for 1 week, weighed to the nearest 0.1 g, and ground to pass a 2 mm screen (Wiley® Mill, Thomas Scientific Inc., Swedesboro, NJ). Foliar N samples were further ground in a laboratory grinding mill (Cyclone Mill, UDY Corporation, Fort Collins, CO). Acid and neutral detergent fiber was assayed in an ANKOM 200 Fiber Analyzer (ANKOM Technology, Macedon, NY) following manufacturer protocols (ANKOM Technology, 2011a, b). Crude protein and foliar-N were assayed using a LECO LP528 Nitrogen/Protein Analyzer (LECO®, St. Joseph, MI), correcting for dry matter. To obtain estimates of dry matter, two 1.000 ± 0.010 g samples of ground plant material were placed into tins. The mass of the empty tin as well as the mass of the sample and tin combined were recorded. Samples were dried in a drying oven at 100°C for 72 h, placed in a desiccator, and weighed. Dry matter (DM) was calculated as:
$${\rm DM} = \displaystyle{{m_{\rm f}} \over {m_0}} \times 100{\rm \% }$$
where m f is the final mass of the sample after drying, and m 0 is the initial mass of the sample. These metrics were used to obtain dry digestible matter (DDM), dry matter intake (DMI) and relative feed value (RFV) (Undersander and Moore, Reference Undersander and Moore2002), which were calculated as:
$${\rm DDM} = 88.9 - \left( {0.779 \times {\rm ADF}} \right)$$
$${\rm DMI} = \displaystyle{{120} \over {{\rm NDF}}}$$
and
$${\rm RFV} = \displaystyle{{{\rm DDM} \times {\rm DMI}} \over {1.29}}$$
Cover crop and weed biomass
We took biomass samples of all cover crop and weed species on July 25, 2012 and on August 1, 2013 at anthesis, but prior to cover crop termination. Post-cover crop termination biomass samples were collected on September 7, 2012 and on August 30, 2013. For biomass data, we again randomly placed four 0.33 m2 quadrats in each plot, cut all plant material flush to the soil surface within quadrats and separated it by species. For each species, we combined the biomass collected in the four quadrats within a plot. We dried all samples at 60°C to constant mass and weighed them to the nearest 0.1 g.
Cash crop yield
We estimated cash-crop yield by weighing the fresh marketable biomass of all harvest plants within each 5 m crop sampling zone. Due to heavy rain and subsequent soil crusting in the spring of 2013 (Table 1), the spinach crop failed in half of our experimental plots, and those data were excluded from analysis.
Data analysis
We compared daily minimum, mean, and maximum soil temperature between grazed and mowed plots, as well as prior to and after cover crop termination using the conditional F-test of a generalized least squares fitted linear mixed effects model with an autoregressive moving average (ARMA) residual covariance structure based on the Julian Date following Zuur et al. (Reference Zuur, Ieno, Walker, Saveliev and Smith2009). ARMA residual covariance structures were parameterized by comparing Akaike's Information Criterion (AIC) values among candidate models with ARMA structures for P = 0, 1, 2 and q = 0, 1, 2, and plotting normalized residuals versus their autocorrelation function values. For these models, we treated cover crop termination method and period of the growing season (pre- versus post-termination) as fixed effects. Each trial year and each experimental plot was modeled with a random intercept with cover crop termination method nested within trial year, plots nested within cover crop termination method and period of the growing season nested within plots. Post-termination mean daily soil moisture was analyzed similarly, except that period was excluded as a fixed effect.
We compared soil chemical properties and compaction between treatments and among soil depths using a nested ANOVA where depth was nested within treatment and treatment was blocked by year. Foliar N at anthesis during the cover crop phase was compared between mowed and grazed plots using a one-way ANOVA with trial year as a blocking factor. Forage quality metrics were analyzed similarly.
We compared the total plant biomass in the cover crop phase at anthesis between grazed and mowed plots as well as between trials using a one way ANOVA blocked by trial year. Once cover crops were terminated, we tested treatment effects on biomass reduction using ANCOVA with pretreatment Malva neglecta (Wallr.) biomass as the covariate. ANCOVA was used due to high M. neglecta biomass before and after cover crop termination. Prior to the analysis, a Box-Cox power transformation analysis revealed that a log-transformation of biomass was warranted. The log-response (LR) ratio was calculated as:
$${\rm LR} = {\rm ln}\left( {\displaystyle{{M_{\rm f}} \over {M_0}}} \right) = {\rm ln}(M_{\rm f} ) - {\rm ln}(M_0 )$$
where M 0 is the pretreatment biomass and M f is the post-treatment biomass.
Cash crop yields were compared with a split-plot ANOVA with cover crop termination strategy as the main plot factor and cash crop rows as the subplot factor. As mentioned previously (see “Cash crop yield estimation” section), the spinach crop failed in half of our experimental plots in 2013 and was excluded from analysis. Thus, data from 2013 and 2014 were analyzed separately.
All analyses were conducted in R version 3.0.2 (R Development Core Team, 2013). Autoregressive mixed models were conducted in the nlme package of R (Pinheiro et al., Reference Pinheiro, Bates, DebRoy and Sarkar2015). Mean separations for significant interactions were performed using Tukey's HSD in the TukeyC package of R (Faria et al., Reference Faria, Jelihovschi and Allaman2012). Graphics were constructed in the sciplot (Morales et al., Reference Morales2012) and ggplot2 (Wickham, Reference Wickham2009) packages of R.
Results
Soil physical and chemical characteristics
Cover crop termination method did not influence maximum [F (1,9) = 0.09; P = 0.761], mean [F (1,9) < 0.01; P = 0.99] or minimum daily soil temperature [F (1,9) = 0.09; P = 0.77, Fig. 1A, C]. Similarly, cover crop termination method did not affect maximum [F (1,9) < 0.01; P = 0.99], mean [F (1,9) < 0.01; P = 0.99] or minimum post-termination soil moisture [F (1,9) < 0.01; P = 0.99; Fig 1B, D]. Maximum [F (1,922) = 31.62; P < 0.001] and mean [F (1,922) = 19.72; P < 0.001] but not minimum [F (1,922) < 0.01; P = 0.98] daily soil temperature differed between the pre-terminated and post-terminated periods. However, differences between pre-termination and post-termination maximum [F (1,922) = 0.14; P = 0.71], mean [F (1,922) = 0.40; P = 0.53] and minimum [F (1,922) = 0.14; P = 0.71] daily soil temperatures did not vary between grazed and mowed plots.
Figure 1. Effects of sheep grazing and mowing on (A) soil temperature in 2012, (B) soil moisture in 2012, (C) soil temperature in 2013 and (D) soil moisture in 2013 during the cover crop phase at Townes Harvest Farm, Bozeman, MT, USA. The legend in the top right panel (A) applies to all other panels (B–D).
Overall soil compaction differed between previously grazed [1.26 ± 0.14 MPa, mean ± standard error (SE)] and previously mowed (1.59 ± 0.17 MPa, mean ± SE) plots the spring following cover crop termination [F (1,5) = 8.18; P = 0.035]. Soil compaction also increased with depth [F (3,30) = 40.02; P < 0.001], with the greatest compaction between 45 and 60 cm below the soil surface (P < 0.001) and the least compaction between 0 and 15 cm below the soil surface (P < 0.001) (Fig. 2). There was no difference in soil compaction resistance between depths of 15–30 cm and 30–45 cm. However, the general increase in soil compaction with depth did not vary between previously grazed and previously mowed plots [F (3,30) = 1.52; P = 0.23].
Figure 2. Effects of sheep grazing and mowing on soil compaction at Townes Harvest Farm in (A) 2013 and (B) 2014, Bozeman, MT, USA. Values are reported as mean ± SE. The legend in left panel (A) also applies to the right panel (B).
With the exception of soil pH, there was no difference in any of the measured soil chemical properties between previously grazed and previously mowed plots (Table 3). While we found that soil in previously mowed plots was more basic than in previously grazed plots, the difference in pH (0.067) was less than the sensitivity of the test (0.1). All measured soil chemical properties differed between topsoil and subsoil. However, none of those differences in soil chemistry between strata varied between previously grazed and previously mowed plots.
Table 3. Impacts of sheep grazing and mowing on soil chemistry in the cash crop phase of 2013 and 2014 at Townes Harvest Farm, Bozeman, MT, USA.
Mowed and grazed plots refer to treatment applied in cover crop phase in 2012 and 2013, respectively. Values are reported as mean ± SE. OM, organic matter; EC, electrical conductivity; CEC, cation exchange capacity.
Significance levels: 0.05 ≥ P ≥ 0.01*; 0.01 > P ≥ 0.001**; 0.001 > P***.
Forage quality
We found no difference in the RFV of the cover crop between grazed and mowed treatments (Table 4). Similarly, there was no interactive effect of treatment and year on RFV. There were no treatments or interactive effects on any of the other forage quality parameters measured.
Table 4. Forage quality of cover crops at Townes Harvest Farm, Bozeman, MT, USA.
Values are reported as mean ± SE.
Cover crop and weed biomass
Overall total biomass in the cover crop phase did not differ between mowed and grazed plots pooled across both trials of our experiment [F (1,5) = 0.01; P = 0.92; Fig. 3A, B]. In 2012, grazing reduced total biomass more than mowing [F (1,7) = 13.33; P = 0.008]. Total biomass in grazed plots was reduced 88.6 ± 6.07% compared with 75.8 ± 10.8% in mowed plots. Cover crop biomass declined more than weed biomass [F (1,7) = 42.98; P < 0.001]. However, there was no interaction between termination method and plant class [F (1,7) = 3.33; P = 0.11]. Grazing reduced cover crop biomass 97.8 ± 1.1% and weed biomass 56.5 ± 18.8%, while mowing reduced cover crop biomass 89.1 ± 5.1% and weed biomass 34.2 ± 25.4%.
Figure 3. Impacts of termination approach (sheep grazing or mowing) on cover crop and weed biomass in (A) 2012 prior to cover crop termination, (B) 2012 after cover crop termination, (C) 2013 prior to cover crop termination and (D) 2013 after cover crop termination at Townes Harvest Farm, Bozeman, MT, USA. Values are reported as mean ± SE. The legend in the top right panel (A) applies to all other panels (B–D).
In 2013, cover crop biomass was reduced more than weed biomass [F (1,7) = 16.89; P = 0.005], but as in 2012, there was no interaction between termination method and plant class [F (1,7) = 1.56; P = 0.25]. Grazing reduced cover crop biomass 98.3 ± 1.2% compared with 83.0 ± 7.1% in mowed plots. While grazing reduced weed biomass 65.8 ± 9.4%, mowing increased weed biomass 36.7 ± 76.9% (Fig. 3C, D), principally due to an increase in the biomass of two weed species in one mowed plot. Common mallow (M. neglecta Wallr.) biomass in one mowed plot increased from 26.8 g m−2 at anthesis to 71.5 g m−2 after cover crop termination. Similarly, while we did not detect redroot pigweed (Amaranthus retroflexus L.) in that plot at anthesis; its biomass was 23.6 g m−2 following cover crop termination.
Cash-crop yields
In 2013, the spinach crop failed in three of the six plots, probably due to heavier than normal rain during spring and subsequent soil crusting (Table 1; C. Holt, Personal communication). Two of these plots had their cover crops terminated by mowing and one plot had its cover crop terminated by grazing. Thus, to avoid introducing bias, we excluded spinach yields from our analysis. Overall in 2013, we found marginally higher yields in mowed plots than in grazed plots [F (1,2) = 9.98; P = 0.087; Fig. 4A]. There was a marginally significant interactive effect of cash crop species and termination method on yield [F (1,4) = 6.34; P = 0.065], but no main effect of cash crop species on yields [F (1,4) = 1.81; P = 0.25]. This is likely the result of a marginal difference between lettuce and kohlrabi yields in mowed plots (P = 0.055). In 2014, we found no overall difference in cash-crop yields between previously grazed and previously mowed plots [F (1,2) = 0.729; P = 0.48; Fig. 4B]. Yields varied among crop species [F (2,8) = 19.33; P < 0.001], with lower yields of spinach than either lettuce (P = 0.002) or kohlrabi (P = 0.002). However, these differences in yield by crop species did not vary between previously grazed and previously mowed plots [F (2,8) = 0.141; P = 0.87].
Figure 4. Impacts of sheep grazing and mowing on subsequent cash-crop yields (A) in 2013 and (B) in 2014 at Townes Harvest Farm, Bozeman, MT, USA. Values are reported as mean ± SE. The legend in left panel (A) also applies to the right panel (B).
Discussion
Soil physical and chemical characteristics
Previous research suggests that, if properly applied, soil compaction by livestock grazing is limited to the top 10 cm of soil, ephemeral, and similar to the compaction caused by farm machinery (Greenwood and McKenzie, Reference Greenwood and McKenzie2001; Franzluebbers and Stuedemann, Reference Franzluebbers and Stuedemann2008; Tracy and Zhang, Reference Tracy and Zhang2008; Bell et al., Reference Bell, Kirkegaard, Swan, Hunt, Huth and Fettell2011). Grazing may increase soil compaction when applied at higher soil moisture content because soil load capacity declines with increasing soil moisture content (Hamza and Anderson, Reference Hamza and Anderson2005). In our study, we imposed sheep grazing and mowing treatments during a relatively dry period. Thus, we predicted soil compaction would be similar between treatments, but found that soil compaction was lower in previously grazed plots than in previously mowed plots. This may be a consequence of the relatively low stocking rate and short duration of grazing.
During the cover crop phase of our study, there was no difference in soil temperature between grazed and mowed plots. Our study involved a relatively short duration of grazing and in comparison with mowing it removed comparable amounts of living vegetation. As a consequence, while not directly measured, it is possible that both mowed and grazed plots had similar rates of evapotranspiration after cover crop termination. Unsurprisingly, we detected differences in maximum and mean daily soil temperature between the pre-termination and post-termination periods, reflecting an increase in solar radiation reaching the soil surface following termination of the cover crop. However, these differences between periods of the growing season were not modified by cover crop termination treatments. Similarly, we did not detect any differences in soil moisture content between grazed and mowed plots. Our soil probes allowed us to estimate soil moisture in the top 7 cm of the soil, but plants can uptake soil water much deeper than 10 cm (Wild, Reference Wild1993). Therefore, although plots were grazed during the driest part of the year for Southwestern Montana, our methods may not necessarily reflect the effects of sheep grazing on total plant available water.
Despite the expected differences in soil chemistry between soil strata independent of treatments, and in accordance with previous studies (Marrs et al., Reference Marrs, Rizand and Harrison1989; Franzluebbers, Reference Franzluebbers2007), we did not observe differences between grazed and mowed plots with the exception of pH. As noted above, the difference in pH between cover crop termination strategies was less than the sensitivity of the test, and we suggest that readers exercise caution when interpreting this result. The lack of differences in soil nutrient concentrations between grazed and mowed plots agrees with Collins (Reference Collins2003) who noted that for most nutrients, ruminants return up to 90% of what they consume in their excreta. Because crops require a greater quantity of N than any other soil macronutrient, and N is more easily lost from soils than other soil macronutrients (Thiessen Martens and Entz, Reference Thiessen Martens and Entz2011), its changes may be of the greatest concern for producers. Thiessen Martens and Entz (Reference Thiessen Martens and Entz2011) reported that sheep can retain up to 25% of the N they consume, but the N deposited in their feces and urine is often more labile than mineralized plant detritus. Thus, repeated grazing may result in declines in total N over longer temporal scales. The short-term nature of our study precludes us from evaluating the mid- and long-term consequences of sheep grazing on the movement of N and other nutrients.
Forage quality
We did not detect any differences in forage quality of cover crops, as measured by RFV, between treatment plots. Additionally, we did not find an effect of year or an interactive effect of year and treatment on forage quality. These results suggest that the forage quality of our cover crop was spatially homogeneous and temporally consistent.
The values we obtained for acid detergent fiber, a metric of indigestible fiber, were lower than a maximum value of 290 g kg−1 (29%) for premium quality alfalfa (Medicago sativa L.) forage (Bath and Marble, Reference Bath and Marble1989; Buxton, Reference Buxton1996). Forbes (Reference Forbes2007) notes that optimal forage crude protein concentrations for sheep nutrition range between 130 and 160 g kg−1 (13–16%). In 2012, crude protein concentrations were within this range for both treatments. However, in 2013, crude protein values were greater than this optimal range. Excess protein has a few potential adverse health effects on sheep including increased risk of heat stress, pizzle rot in rams and urolithiasis (Pugh and Baird, Reference Pugh and Baird2012). However, Kyriazakis and Oldham (Reference Kyriazakis and Oldham1993) found that sheep can discriminate among available forages to optimize their crude protein intake. Thus, it is unlikely that sheep grazing cover crops would suffer the deleterious effects of excess crude protein intake, especially if they graze the cover crop for less than 1 month.
Both years the forage quality, as measured by RFV, fit within the top two categories as defined by the American Forage and Grassland Council (Hopper et al., Reference Hopper, Peterson and Burton2004). Alfalfa forage of such standards commanded US$77.76 and 69.37 Mg−1, respectively (Hopper et al., Reference Hopper, Peterson and Burton2004). Based on these values, the cover crop in the first trial represents US$435.59 ± 74.60 and 409.69 ± 76.79 ha−1 worth of potential forage for mowed and grazed plots, respectively. In the second trial, the cover crop represented US$276.18 ± 29.57 and 255.30 ± 27.81 ha−1 worth of potential forage for mowed and graze plots, respectively. Thus, a producer who terminated his/her cover crop by grazing with his/her own livestock grazing may save a substantial amount of money on high-quality fodder that he/she would otherwise have to buy if he/she terminated the cover crop by mowing.
In a study conducted in the Imperial Valley of California, alfalfa growers granted grazing leases to sheep ranchers in lieu of harvesting hay for US$0.06–0.11 head−1 day−1 (Bell and Guerrero, Reference Bell and Guerrero1997). At these rates and a stocking rate of 400–730 head ha−1, a grazing lease would be worth US$24.00–80.30 ha−1 day−1. Assuming a 6-day lease—the duration of grazing in our study—a producer could potentially sell a grazing lease for US$144.00–481.80 ha−1. Hence, a grazing lease could at least partially off-set the cost of cover crop husbandry. These estimates do not represent a complete economic analysis, but this would be a valuable avenue for future research.
Cover crop and weed biomass
The effect of cover crop termination method on cover crop and weed biomass was not uniform. In 2012, sheep grazing reduced total plant biomass more than did mowing. However, in 2013 we found no differences between mowed and grazed plots. Despite these differences, in both years cover crop biomass declined more than weed biomass. In the 2012 trial, weed biomass was greater than cover crop biomass prior to termination with A. retroflexus and M. neglecta as the dominant species. Malva neglecta has a prostrate growth form and thus may have avoided termination through either grazing or mowing (Sean McKenzie, personal observation). Despite the erect stems of A. retroflexus, many stems of this species were able to re-sprout following termination (Sean McKenzie, personal observation). The dominant cover crop species was F. esculentum, which has erect stems. In contrast to A. retroflexus, F. esculentum stem did not re-sprout after being mowed or grazed. These results suggest that grazing and mowing are equally effective at terminating a cover crop, but the efficacy of these cover crop termination methods may be temporally variable and may depend on the species composition of the weed community and the cover crop, as well as the species of livestock used for cover crop termination. In addition, some cover crop and weed species—including M. officinalis and A. retroflexus—are considered toxic to livestock, which may preclude grazing for cover crop termination when such species comprise a substantial proportion of the total biomass. For a detailed discussion of the effects of these cover crop termination methods on plant community dynamics see our companion study (McKenzie et al., Reference McKenzie, Goosey, O'Neill and Menalled2016).
Cash-crop yields
Similar to previous studies (Franzluebbers, Reference Franzluebbers2007; Bell et al., Reference Bell, Kirkegaard, Swan, Hunt, Huth and Fettell2011; Hilimire, Reference Hilimire2011; Thiessen Martens and Entz, Reference Thiessen Martens and Entz2011), we did not find any detriments to crop yields from integrating livestock into our cropping system. In addition, our results concur with the observation of Franzluebbers (Reference Franzluebbers2007), that using livestock to terminate cover crops does not impact subsequent crop yields. To our knowledge, this is the first study to quantify the legacy effects of integrating livestock for cover crop termination on cash-crop yield in horticultural vegetable market-gardens, and thus precludes any direct comparisons with previous studies in such systems.
Conclusion
Our study indicated that integrating livestock for cover crop termination can enhance the economic feasibility of cover crop use without negatively impacting growing conditions or agronomic outputs. This may be especially important in horticultural vegetable market-gardens, where consumer preferences, costs and environmental concerns prohibit or at least discourage the use of off-farm inputs for nutrient and pest management. Our study is one of a very few studies (e.g., Lowy, Reference Lowy2009; McKenzie et al., Reference McKenzie, Goosey, O'Neill and Menalled2016) to investigate the impacts of integrating livestock into horticultural market-gardens. While we found no negative legacy impacts of using sheep grazing to terminate cover crops, future research should investigate longer-term legacy effects, as well as the impacts of repeated grazing.
Acknowledgements
We would like to extend our deepest gratitude to Charles Holt, farm manager at Townes Harvest Farm (THF) and David Baumbauer, Director of Horticulture at Montana State University (MSU), for allowing us to conduct our research at THF and accommodating our research needs. Many thanks to the other members of the Weed Ecology Laboratory at MSU, Subodh Adhikari, Dr. Judit Barraso, Erin Burns, Krista Ehlert, Stephen Johnson, Dr. Zachariah Miller and Nar Ranabadht, as well as our undergraduate technicians Torrin Daniels, Christian Larsen, Madison Nixon, Wyatt Holmes, Janki Patel, Lori Saulsbury, Alexandra Thornton and Emmett Wester for their assistance with data collection and processing. Thanks to Hilary Parkinson, plant identification diagnostician for Schutter Diagnostic Laboratory at MSU, for her assistance with plant identification. Finally, we are grateful to the Montana Agricultural Experiment Station, the Western Sustainable Agriculture Research and Extension Program (grant no. SW11–086) the NIFA Organic Research and Extension Initiative (grant no. MONB00314) and the NIFA Organic Transitions program (grant no. MONB00128) of the USDA for funding this research.
Open access
