Final Report for GW06-029
In this project we tested the effectiveness of livestock grazing as a management tool for vernal pool lands. Vernal pools are unique seasonal wetlands that provide essential habitat for many rare and endangered plant and animal species. Recent anecdotal evidence has suggested that livestock grazing on vernal pool grasslands in California may help maintain native plant diversity by suppressing the non-native annual grasses that typically surround these pools. In collaboration with local ranchers, we tested this idea by introducing sheep grazing to a site with vernal pools in Sonoma County, California. Exclusionary fencing on half of each pool allowed us to compare the plant community composition of grazed and ungrazed plots each spring from 2005 (before grazing was introduced) through 2008. We found that spring sheep grazing at the experimental site did not significantly increase native plant cover or richness. Sheep grazed heavily in pool bottoms once pools dried, contrary to management goals. However, sheep grazing did reduce cover of an invasive late-season forb, and increased cover of an early-season native aquatic forb. Sheep grazing may be effective at enhancing natives if done while pools are still inundated, or after natives have senesced. A companion study showed that high-productivity vernal pools grazed by cattle did have greater native richness and cover than ungrazed pools. Further study will help refine these conclusions, but land managers in Sonoma County are already looking for local ranchers with whom to partner to steward vernal pool lands. We have presented our results in a number of forums, we hosted a highly successful workshop bringing local ranchers and land managers together to discuss vernal pool grazing, and we are working on a manuscript for publication in a scientific journal.
Vernal pools are temporary wetlands that once occurred throughout California grasslands and provided habitat for many rare and endemic organisms. Today, less than 10% of pool habitat remains, and native pool species across California are in decline. Anecdotal evidence indicates that pools on active ranchlands often have healthier native plant populations than do those from which livestock grazing was removed in recent decades. Land managers in Sonoma County and elsewhere in California are poised to reintroduce managed grazing to pools in hopes of encouraging natives, but have hesitated for lack of careful research on which to base their management plans.
To address this gap in knowledge, we carried out a sheep-grazing experiment in collaboration with local ranchers. At a Sonoma County site with vernal pools, half of each pool was fenced to exclude grazing; the other halves were unfenced and readily accessible to sheep. We monitored and compared the hydrology and vegetation of grazed and ungrazed plots, and presented our results in a number of forums. We organized a field workshop to share our results with local ranchers and land managers. We are in the process of preparing a manuscript describing our results for submission to a peer-reviewed scientific journal focusing on applied ecology.
Our objectives for the first four months of the grant period (fall 2006) were to analyze preliminary data, present preliminary results at a grasslands management conference, and begin monitoring vernal pool inundation. Our objectives for 2007 were to continue experimental sheep grazing, monitor vegetation of grazed and ungrazed plots, monitor vernal pool hydrology, analyze the resulting new data, begin manuscript preparation, and plan for an educational field session. Our objectives for 2008 were to continue experimental sheep grazing and vegetation monitoring for a final season, to host a field session sharing our results and informing ranchers and land managers about vernal pool grazing, and to complete and submit a manuscript to an ecological journal.
We established a field experiment in collaboration with Kerry Williams, a local sheep rancher and director of the Sotoyome Resource Conservation District; and Denise Cadman, City of Santa Rosa Natural Resource Specialist.
Our study site is a set of 7 pools located on a 40-acre preserve at the edge of a city-owned farm. In February 2005, we designed a fencing plan to exclude sheep from half of each pool. Fencing was installed by the City of Santa Rosa. We mapped the pools and, because depth has a strong influence on species composition, measured the maximum depth of each. Ewes grazed the 5-acre area from mid-May to mid-August, at an initial density of 3.4 animals per acre. In April, we placed 16 permanently marked quadrats in each pool (8 on each side, with 2 in each of four topographic zones—bottom, mid-depth, edge, and upland). We then determined plant community composition (percent cover of each species, and species richness) in each quadrat. In September, we sampled residual dry matter (RDM; 8 samples per pool) to help quantify grazing intensity. This regime of grazing and plant community sampling was repeated annually through summer 2008 (RDM monitoring was done only once). Native and exotic species richness and life form cover were analyzed statistically to assess the influence of grazing.
In conjunction with the sheep grazing experiment (but not funded by SARE), we conducted a second, non-experimental study to compare the plant communities of vernal pools grazed by cattle year-round to those of vernal pools where no grazing had occurred for at least ten years.
All statistical analyses for both the experimental and comparative parts of this study were done with SAS 8.2, JMP 5.1 (SAS Institute, Cary, North Carolina), and PC-ORD 4.2 (MjM Software, Gleneden Beach, Oregon).
For the sheep grazing experiment, we analyzed data on aboveground dry plant biomass using a multi-way ANOVA, with grazing treatment (sheep present or excluded), vernal pool (1-7) treated as a random effect, and pool zone (bottom, mid-depth and edge) nested within pool as the three grouping factors.
To visualize the community-level effects of grazing in the sheep exclosure experiment, we used non-metric multidimensional scaling (NMS) to reduce the dimensionality of the data on plant species cover. The predictor variables we used for this analysis were grazing treatment, pool, and zone and the response variables were relative cover of each species in 2006. we used Sorensen (Bray-Curtis) distance measure in the NMS analyses. To quantitatively assess the effects of grazing on community composition, we used multi-response permutation procedure (MRPP), a nonparametric randomization method of testing for differences between two or more groups. MRPP analysis produces a weighted average (A) of within-group mean distances. Higher values of the statistic indicate more closely-related groups: A values of 1 indicate all items within groups are identical, whereas A values of 0 indicate that variation within groups is equal to that expected by chance alone. In this analysis, we used a blocked version of the procedure (multi-response blocked permutation procedure; MRBP), with grazing treatment and pool as predictor variables and relative plant cover of each species in 2006 as the response variables. We used Euclidean distance measures in these analyses.
We used a multi-way MANOVA to assess the effect of sheep grazing on the percent cover of eight plant groups: native and exotic annual grasses, native and exotic perennial grasses, native and exotic annual forbs, exotic perennial forbs, and native non-grass monocots. As before, grouping factors for this MANOVA were grazing treatment, vernal pool (a random effect), and zone (nested in pool; a random effect). The response variables for this analysis were relative change in cover of each of the eight plant groups from 2005 to 2006. We calculated relative change as the natural log of the ratio of 2006 cover to 2005 cover. Data from paired quadrats were averaged for cover. If the MANOVA indicated a significant grazing effect, I proceeded with univariate ‘protected’ ANOVAs (sensu Scheiner 2001) on the individual cover variables using the same model. As discussed by Scheiner (2001), this approach is an effective method for dealing with potential correlations among multiple dependent variables. In addition to the percent cover analyses, we also assessed the effect of sheep grazing on the richness of native and exotic species using multi-way ANOVAs. Grouping factors for this analysis were grazing treatment, pool (random effect), zone nested in pool (random effect), and plant origin (exotic or native) and the response variable was relative change in plant species richness.
To evaluate the relationships of relative change of native cover and exotic cover to pool depth and residual dry biomass in the sheep exclosure experiment, we used four linear regression analyses. For regressions involving maximum pool depth, plant cover was averaged for each pool.
We are currently completing additional analyses using repeated measures tests.
In this section we present an overview of results; for details including statistical analyses, see below.
We found that native plant cover on this site was greatest at pool bottoms and lowest at pool edges and uplands. This gradient is typical of vernal pools surrounded by non-native grassland. Therefore, from a native plant management perspective, a primary goal might be to reduce biomass at pool edges. However, our results showed that sheep grazing reduced biomass most in pool bottoms and least at pool edges.
Cover and richness of native plant species, when considered as a group, did not differ significantly between grazed and ungrazed plots. No strong differences were apparent in the response of different plant functional groups (e.g. annual or perennial, forbs or grasses) to grazing. Sheep grazing did reduce the cover of one late-season invasive species, pennyroyal (Mentha pulegium) and appeared to increase the cover of one rare early-season native aquatic forb, Lobb’s buttercup (Ranunculus lobbii). Pennyroyal, an increasingly problematic invader of vernal pools, was abundant in pool bottoms, where sheep grazed most heavily. Pennyroyal has strong aromatic compounds and was not expected to be heavily grazed by sheep, but was in this case. The sheep did not appear to suffer any ill effects from ingesting pennyroyal. The uplands at this site are dominated by tall fescue, which appeared to be less palatable to the sheep, especially late in the season; this may have influenced the sheep’s forage choices and encouraged them to graze the pennyroyal.
Lobb’s buttercup is a small aquatic vernal pool species, which flowers and sets seed in late winter to early spring while pools are still inundated. Because this very fine-stemmed, ephemeral native species is not visible in mid-spring when vegetation sampling occurred, we made informal ocular estimates of its cover in February. These informal observations indicated that the species had greater cover in grazed plots than in ungrazed plots. The species may have benefited from the reduction of biomass (i.e., thatch) from pool bottoms, while avoiding consumption by the sheep because of its phenology. Since the pools were inundated at the time, the sheep could not reach the Lobb’s buttercup during the aquatic species’ active growth and reproductive phases.
We found no relationship between change in native or exotic species cover and pool depth or aboveground biomass.
While it is encouraging that spring sheep grazing did not have an overall negative effect on natives at the site, we had hypothesized that the grazing would actually increase natives. Our results indicate that altering the timing of grazing, to occur before or after most native species’ active growth and reproductive phases (i.e. before March and after May), may be more effective. This will not always be feasible from a livestock management perspective, but may be in some cases. Another possible factor influencing our results is that the native seedbank at our experimental site may be very limited. This situation is probably common to many vernal pools in Sonoma County, especially those from which grazing has been removed in recent decades. At these sites, our results suggest that supplemental seeding or inoculation of the pools with native seed may be necessary to increase native cover and diversity.
Our cattle-grazing study, a companion to the SARE-funded experiment, had significantly different results. Plant community composition was significantly different at grazed pools compared to ungrazed pools, and differences were greatest at pool edges. Grazed pool edges had greater cover of native forbs, native annual grasses and native perennial grasses, and lower cover of native non-grass monocots and of non-native perennial grasses, than ungrazed pools. Richness of native species was 19% higher in grazed pools, and richness of exotics was 21% lower, compared to ungrazed pools.
The differences in foraging preferences between cattle and sheep help explain the differing results from the two components of our study. Cattle generally select grasses over forbs for forage. Since the non-native plant species in this system are primarily grasses, and many of the native vernal pool species are forbs, managed cattle grazing appears to be a useful tool for maintaining native plant diversity in these pools. Sheep more readily consume forbs, and thus may be more effective when vernal pool forbs are inaccessible or have already reproduced, or where invasive forbs are encroaching on vernal pool bottom vegetation.
Results from a multi-way ANOVA indicated that plant biomass was influenced by sheep grazing (F1,20=7.93, p=0.011) and zone (F14,20=4.06, p=0.002). Sheep grazing reduced aboveground plant biomass by 37% in the bottom zone, 24% in the mid zone, and 8% in the edge zone (Figure 3).
The NMS and MRBP analyses of relative cover of each species in 2006 indicated that the composition of plant assemblages was not significantly affected by sheep grazing (Figure 4; A=-0.007, p=0.771).
Results from a MANOVA revealed that sheep grazing did not significantly affect relative change in cover of plant groups (F8,13=1.24, p=0.351). This result did not allow me to proceed with “protected ANOVAs” to test for treatment effects on each plant group individually. However, clear trends were apparent for three of the eight plant groups. Exotic annual forbs decreased more in grazed plots than in ungrazed plots. Exotic perennial forbs and grasses remained unchanged or slightly increased in grazed plots while they declined in ungrazed plots. No grazing-related trends were apparent in cover of any native plant groups.
Results from an ANOVA indicated that there was no significant effect of grazing on relative change of total species richness (F1,60=0.00, p=0.982), and exotics and natives did not differ significantly in how they responded to grazing (F1,60= 0.13, p=0.716). Relative change of species richness did vary with origin (F1,60=23.06, p<0.0001) and zone (F14,60=2.03, p=0.0301) but not with pool (F6,14=1.65, p=0.206). Exotic species richness declined in all zones, while native species richness showed smaller declines in the bottom and mid zones, and no change in the edge zone (where native richness was lowest).
Regression analyses indicated that neither relative change in cover of natives nor relative change in cover of exotics varied significantly with aboveground plant biomass (natives, F1,54=0.97, p=0.329, R2=0.018; exotics, F1,54=0.15, p=0.703, R2= 0.003). Similarly, neither relative change in cover of natives nor relative change in cover of exotics varied significantly with maximum pool depth (natives, F1,12=1.19, p=0.298, R2=0.090; exotics, F1,12=0.17, p=0.687, R2=0.014).
Results from a multi-way ANOVA indicated that vegetation height was influenced significantly by cattle grazing (F1,53=4.52, p=0.038) and block (F1,42=58.08, p<0.0001), but not by pool (F25,39=1.32, p=0.210) or by zone (F54,53=0.348, p=0.999). Mean vegetation height was 7% lower in grazed pools than in ungrazed pools.
Results of NMS analysis of relative cover by species indicated that, overall, the composition of plant assemblages differed between grazed and ungrazed pools (Figure 7). The MRPP analysis indicated that these differences in species composition between grazed and ungrazed pools were significant (A=0.06, p<0.0001).
Results of a MANOVA also indicated that relative cover of plant groups differed significantly between grazed and ungrazed pools (F1,53=7.93, p<0.0001). Subsequent protected ANOVAs on individual plant groups indicated that the difference between grazed and ungrazed pools was due to significant differences in relative cover of four native plant groups and one exotic plant group (Table 1). These differences were most apparent in the edge zone, where grazed pools had greater cover of native forbs, native annual grasses and native perennial grasses, and lower cover of native non-grass monocots (Figure 8c). Relative cover of exotic perennial grasses also differed slightly between grazed and ungrazed pools; no species in this group were observed at grazed sites, while a single species, velvetgrass (Holcus lanatus), was present in low abundance at pool edges of one ungrazed site. In addition, trends toward increased cover of exotic forbs and decreased cover of exotic grasses are evident in the data for pool edges. In the mid zone, three differences were evident: native perennial grasses and native annual forbs were more abundant in grazed pools, and native non-grass monocots were less abundant (Figure 8b). A trend toward decreased cover of exotic annual grasses was visible in the mid zone as well. In the bottom zone, only one group differed in cover between grazed and ungrazed pools: native perennial grasses were more abundant in grazed pools (Figure 8a). All exotic groups were very rare in pool bottoms, regardless of grazing status. Relative cover of plant groups varied significantly with pool for four of the nine plant groups: native and exotic annual forbs, native perennial forbs, native perennial grasses and native non-grass monocots (Table 1). Relative cover of plant groups varied significantly with zone for only two of the nine plant groups: exotic annual grasses and exotic perennial grasses (Table 1). Relative cover of plant groups did not vary significantly with block for any plant groups (Table 1).
An ANOVA indicated that although grazing did not affect total species richness (F1,186=0.02, p=0.9025), richness did vary with origin (F1,186=35.86, p<0.0001) and native and exotic species richness did respond differently to grazing (F1,186=7.30, p=0.008). Richness of native species was 19% higher in grazed pools, and richness of exotics was 21% lower (Figure 9). Species richness did not vary with zone (F54,186=1.34, p=0.077), pool (F25,55=1.47, p=0.118) or block (F1,56=1.12, p=0.294).
Education and Outreach
In February 2007, I presented my work in a public thesis defense at Sonoma State University, as well as submitting it in written form. In April 2007, I was invited to and spoke at the State of the Laguna conference held by a local non-profit, the Laguna Foundation. In August 2007, I spoke at the joint annual meeting of the Ecological Society of America and the Society for Ecological Restoration.
In February 2008, we held an all-day workshop, “Livestock Grazing on Vernal Pool Landscapes.” The event was a collaborative effort; Joan Schwan was the lead organizer, with SARE funding; local sheep rancher Kerry Williams helped plan and recruit for the session; the Sotoyome Resource Conservation District helped organize and host the session. The City of Santa Rosa provided a venue for the meeting free of charge, as well as access to their lands for a field trip. Western United Dairymen assisted us with publicity. A broad array of 13 experts spoke at the meeting, including two range management specialists, a Nature Conservancy ecologist, a California Department of Fish and Game biologist responsible for managing numerous local vernal pool preserves, and several highly respected local ranchers and dairymen with experience on vernal pool landscapes, including the President of the Sonoma County Farm Bureau. The afternoon portion of the workshop entailed field trips to three local vernal pool sites which are grazed by either sheep or cattle; at each site, both the rancher and land manager or biologist involved in the site guided the visit and described their experiences. Over 70 people attended the workshop and field trips, almost 50% more than the number for which we initially planned. Twenty-five attendees completed workshop evaluations; of these, all agreed that the workshop met its stated goals; 17 would like to learn more about developing a livestock grazing plan; 16 felt the workshop could be repeated annually (8 others felt it should be repeated every two or three years). Despite skepticism from some quarters that biologists and ranchers could come together for a grazing workshop, the workshop was well-attended, fostered lively discussion, and received positive evaluations from participants.
Our partnership with rancher Kerry Williams on this project was featured in a front-page article in the Sonoma West Times, a local newspaper. Link: http://www.sonomawest.com/articles/2006/11/16/sonomawest/news/news2.txt