Sonoma Mountain Institute

Summary:

At Sonoma Mountain Institute, we collected plant species data on 72 stratified fixed plots, some paired with exclosures, on six properties to measure the effect of our grazing management on plant biodiversity and on native plant biodiversity. Total plant biodiversity rose on all properties, ranging between 50% and 300% relative increases. Native plant cover rose between 50% and 250% percent on three properties. On two properties native plant cover rose initially by 25% and 50% peaked around 2016 or 2017 before declining to no change. We suspect that changes in vegetation height and thatch depth drove some of these changes. We believe this information can help managers improve biodiversity results on the land, help us better understand prehistoric herbivory regimes, and guide our philosophical orientation towards conservation in the Anthropocene. 

Introduction:

Biodiversity conservation is probably the most important job of any land steward. Yet we have very little information regarding the most basic factors that drive biodiversity and native plant persistence on working landscapes. The biodiversity information that we do have largely focuses on individual species that are already at risk. We wanted to learn more about what factors drove vegetative biodiversity and native plant persistence so that those lessons could be applied to the vast majority of rangeland in California. 

In 2012, Sonoma Mountain Institute started a grazing trial on property in Sonoma County. We grazed that property using electric fence to create higher livestock densities and to control livestock graze/ recovery periods. As the years went on, we added more properties to our management and collected data from more plots on those properties as we added them. After grazing periods that ranged between several hours and several weeks, depending on the property, we removed cattle from a given paddock until the average grass plant in that paddock had regrown several leaves. Our recovery periods ranged from several weeks when vegetation was growing quickly, to six months or more during the summer/fall dry season in California. Than we would regraze that paddock. Every year we collected vegetation data from fixed, stratified plots on these properties. We were interested in monitoring changes in plant biodiversity and percent native cover with our management, as well as other ecological proxies, such as thatch depth, annual vs. perennial cover, vegetation height, etc. On one previously ungrazed property, this included the creation of grazing exclosures and monitoring vegetation in those exclosures, versus adjacent grazed plots. 

Hypothesis:

We hypothesized that by mimicking the movement of large herds of mammals, such as those that inhabited this part of North America and most other continental environments over the evolutionary history of those plant genera, we would increase plant biodiversity. Less clear to us was how this would influence native plant cover, as both North America and Eurasia (the native territory of most non-native plants in California annual grasslands)  have an equally long history of large animal herbivory. We hypothesized that plants from both continents should be equally adapted to herbivory. We wanted to collect data so that we could trace the different ecological mechanisms that created these changes.

Methods:

Sonoma Mountain Institute contracted with a botanist to setup an experimental design and conduct annual vegetation monitoring. That botanist was not part of management on the different properties.

The experimental design that we settled on was a series of fixed plots, ten meters by ten  meters. We used a stratified sample so as to better quantify changes in areas that showed certain characteristics (areas with invasive species, areas with a high native component, open areas vs. areas with higher tree canopy coverage, etc). In these plots, the botanist cataloged all the different plant species that could be identified during peak phenology, which grazing events sometimes delayed. The botanist estimated percent cover for each plant species in the plot. In addition, thatch depth was measured, mid and high vegetation height and phenology were recorded. On one property (ungrazed in the decade prior to our management) we established five small grazing exclosures paired with a neighboring grazed plot. In addition, on that property one acre was excluded from grazing and a monitoring plot located in that exclosure. 

Results:

Biodiversity:

We were surprised by our biodiversity results. All the grazed plots added plant species. With between three and eight years of data collection, depending on the property, the number of plant species per plot has increased between 61% and 243% and all of our plots are still adding plant species. Most of these species came in the form of new forbs, both native and non-native, at low levels, though native grass species also increased substantially. Non-native grasses, both perennial and annual show declines. For example, on our Petaluma property, over half of the plots show substantial increases in native and non-native forbs and native grasses, while at the same time showing substantial decreases in non-native grass cover. This pattern holds on most plots on most properties. One exception being a property that had very high thistle density in the baseline (Walsh). Over a few years of management, non-native thistle densities dropped by an order of magnitude on several plots, confounding the grass/forb patterns that we see on other properties. The Petaluma property has a much higher increase in diversity. We are thinking about possible sampling explanations for that.

Native Cover: 

Percentage of native cover increased dramatically on three of the six properties and increased to a lesser degree on two properties. However, on two properties where native plant cover increased by 25%-50% until 2017 and 2016, when native cover decreased until it settled at +2% (Mitsui) and +8% (Glen Ellen). Two of the three properties that have increased native cover seem to have plateued around 2016 (Petaluma after adding 60% native cover and Pangea after adding 250% native cover.) There was only a modest increase in percent native cover on one property (Walsh at 5%). Cayetanna percent native cover is 145% greater in 2019 than at the baseline and continuing to rise. There were some plots that showed declines in native grasses, though many increased.

Percent Relative Change

DIVERSITY
Petaluma	Glen Ellen	Cayetanna	Pangea	Mitsui	Walsh
243	61	75	63	93	88
NATIVE COVER
Petaluma	Glen Ellen	Cayetanna	Pangea	Mitsui	Walsh
44	8	145	144	2	5

Thatch:

Thatch levels have decreased on most plots and on all properties, even on properties that were much more heavily grazed when we took over management. Much of that can probably be attributed to the fact that we rested all heavily grazed ranches, sometimes for an entire season, before grazing, so our monitoring process didn’t capture the lowest thatch levels. In general, in the first year of grazing thatch depths dropped very substantially and by the second year of grazing they were between one and two centimeters and where they hold steady. 

Perennial and Annual:

In general, when looking at the properties as a whole, there has been no change in the percentage of annual and perennial species on all the plots. However this conceals the fact that plant species have changed dramatically. Many plots dominated by annual grasses in the baseline see a big increase in perennial forbs. However, many plots that were dominated by perennial non-native grasses in the baseline see large decreases in those species, with annual grasses and forbs increasing. 

Plant Height:

Plant heights on the Petaluma property are confounded by the fact that the sampling period (2013-2019) included some of the driest and some of the wettest years on record for Sonoma County. Our baseline data was collected in 2013, after the driest water year on record. Despite that, the mid level height for our baseline (pre-grazing) was tied the for highest vegetation heights. “Mid Level Vegetation” heights were highest in 2016, when an average water year coincided with a low stocking rate. “High vegetation” heights do not seem to be nearly as impacted by grazing and track closely with precipitation. The more rainfall, the higher the highest plant in the plot grows, regardless of grazing intensity. 

Another confounding factor is the sampling date. Growth rates at peak phenology are very high, so sampling a week earlier or later in the season can have a huge impact on veg heights measurements. It seems likely that sampling bias affects our height measurements to a certain degree as this was not something we considered at the outset of the trial. Anecdotally vegetation heights over the last three years have been much higher on most of the properties, as high rainfall and management factors coincided to create much higher vegetation heights.

Even with our sampling problems, increases in species diversity native cover track closely with changes in vegetation height, as vegetation heights increase fewer species are added. This pattern is most noticeable on the properties where we have the highest sampling densities, Petaluma and Glen Ellen.

The impacts of vegetation height on species factors seems to be greatest as the vegetation is taller. Patterns were most distinct at Glen Ellen, where vegetation heights are much higher than on the other five properties. We would expect this pattern to continue until bare ground started to increase, offsetting increases in stem count that come from shorter vegetation heights. 

Blue- Vegetation Height, Red- Plant Species Numbers, Yellow-Native Plant Percent Cover

Exclosures:

We setup exclosures on a ranch that had not been grazed for many years before our grazing management. Some of those exclosures were paired plots where we had exclosures immediately adjacent to grazed plots. We have thrown out the data on one of those exclosures for reasons that we will discuss. The grazed plots added over 60% more plant species over the course of the study than the ungrazed plots. It is important to note however that the ungrazed plots still added plant species, but at a much lower rate. The ungrazed exclosures had no change in percent native cover while the grazed plots had a 125% relative increase in native cover. 

However, our plot located inside a one acre grazing exclosure added species and native species in line with many of the other plots, a perplexing result.

One of the exclosures was accidentally grazed on the first year of the study (2013). At the time we thought this was a big mistake but now it is one of the more interesting data points. In the first year after the baseline, biodiversity and native cover were the same on the two plots, as we would expect since management was the same, both increasing in-line with what we saw in all the grazed plots. But species were added at a lower rate in the exclosure in the next two years as grazing was discontinued. By the third year, nonnative grass levels jumped back to pregrazing levels and by the end of the period (2019) the exclosure is starting to lose plant species and native cover, reversing gains after grazing was no longer allowed. No other plot has demonstrated this pattern. 

Discussion:

The overall trend that we see from our research is one where plant species, including native species, increase through our management. The increase comes from native and non-native forbs (many species, all having quite low total cover percentages) and native grasses. These gains come at the expense of non-native grasses, both annual and perennial. 

Native percent cover did increase on all properties initially, but on two properties (Glen Ellen and Mitsui) it has declined back to no change since around 2016/2017. We are looking into possible causes for this. In addition, some plots did lose native grass cover over time. We are looking into what could cause this. One theory is that the old moribund material of some of the species was creating a large footprint, though only a small percentage of that material was living. This is supported by several of the plots that showed this tendency at the start of the trial have since regained their native grass cover.  

We think that increases in biodiversity and percent native cover on these properties occurred because we created herbivory regimes that more closely resemble the herbivory regimes that these plant genera evolved under. We hope to use this information to help land managers refine their biodiversity management. In addition, at the risk of being tautological, we hope to use this information to advance our understanding of prehistoric herbivory regimes. 

One of the first takeaways that struck us in our analysis of this data is how consistent our results were. Over five different properties all with different geologies and aspects, we achieved broadly similar results. These properties ranged from being very heavily utilized/overutilized by livestock at set stocking rates, to being completely ungrazed for a decade or more. When we started managing these properties, the management realities dictated that we use a wide range of livestock densities. Some of the properties were grazed using high stock density, with multiple livestock moves per day, others were grazed much more extensively, often with grazing periods that were several weeks or a month. The year-to-year results were also broadly similar, with similar numbers of species being added over the course of three or five years, despite the fact that our dataset spans both some of the driest and wettest years on record for Sonoma County. Recovery periods during the grazing season were relatively constant on all properties as was a summer/fall rest period, roughly from the first of July to mid December. 

This gives managers an interesting data point about how to allocate scarce management resources. We also think this suggests that there was a wide range of possible herbivory regimes over evolutionary significant periods. But we also think this suggests that it is easy for managers to create herbivory regimes that lie outside what was common over evolutionary history; decades of ‘set stocking’ and decades of herbivore exclusion being perhaps equally uncommon over evolutionary history. 

The plots with the highest biodiversity were under tree canopies, and the plots with highest biodiversity had high native plant diversity. Nowhere did we find very high biodiversity strictly with non-native plants. It is hard to know what is driving higher biodiversity under tree canopies. Trees might increase diversity by creating a range of sunlight conditions, through nutrient impacts, or some other means. But treed sites also tend to have less or no historic agricultural disturbance. Historic agricultural disturbance could have removed individual plants that have failed to recolonize the site, or could have altered soil properties in some important way. On our properties, areas with higher tree densities also tends to be on slopes (due to historic agricultural practices), tend to have lower herbaceous heights, etc..  

We think that this data also points to some of the possible mechanisms involved with species recruitment, often involving competition. First, thatch effectively competes for light with other species, particularly during the seedling stage. In addition, grazing reduces the height of the vegetation. During the drought of 2012-2013 when our baseline data was collected for our Petaluma property, precipitation levels were an amazing 20% of some of the subsequent years. Yet vegetation heights at the baseline were still higher in the ungrazed dry year than in the grazed wet years. Since herbaceous plants need to be rooted to the ground, we hypothesize that the “3:2 Thinning Rule” would suggest that there will be the potential for many more stems per plot when vegetation is short. With more stems per plot there is the possibility to have more individual plants per plot and with more individual plants there is the possibility to have more species. We could hypothesize that herbaceous plant biodiversity would increase with decreasing veg height until bare ground started to increase. At this point the manager would have the most possible stems, therefore the most possible species. Which is a hypothesis that needs to be tested.

In addition, we suspect that vegetation height could provide insights into the mechanisms driving the increase in percent native cover that we detected. Since the native plants in our area are much shorter on average than the non native plants, they would be disproportionately disadvantaged by very high vegetation heights and we could expect them to increase with lower vegetation heights. Percent native cover quickly declined with higher veg heights, while total species numbers were less sensitive. It took longer to remove every last species of a plant than it did to reduce populations. 

Another factor affecting our outcome might involve our use of relative cover estimates rather than absolute cover. Because we chose to use relative cover estimates, any increase in percent cover from some species will decrease percent cover for others. Our anecdotal observations suggest that a certain amount of the change in percent cover comes from an increase in forbs and natives grasses, rather than a decrease non-native grasses.

One of the burning questions to come out of our data so far is why exclosure plots have continued to add species, though at a much slower rate than control plots in the paired study. This is particularly perplexing in our one acre exclosure, which added more species than in the paired plots and is not very different from the average results on grazed plots. If these plots had been adding species at their current rate for any amount of time, there would be hundreds of plant species on these plots. However, some exclosures species counts are still in the single digits. One likely explanation is that the exclosures in our paired plots are too small and sunlight, or other impacts from grazed areas are affecting species composition inside the exclosure. This makes sense in the paired plots, however the plot in our one acre exclosure has added even more species than the paired plots. We feel like we have ruled out sampling bias because our research team has not encountered the same problem in other contexts using the same methods. This is one of many questions that we are in the process of creating experiments and hypotheses to test. 

Another observation from the exclosures is that vegetation patterns seem to track with climate in exclosures, with different components of the vegetation rising and falling with different precipitation regimes. However in the grazed plots, the effect of rainfall on species composition is much more muted. We think this is because management is having a strong effect on vegetation and the vegetation has not yet reached a new equilibrium around the management regime.

Conclusion:

We are still working to understand the data that we are seeing. It is possible that we are measuring the birth of a new plant community. A lot of attention in conservation is devoted to native species. Often in conservation circles we encounter the paradigm that the native/non-native divide is a zero sum game; gains in one category are offset by loses in the other. Our evidence does not support this view. From a native plant perspective, even if we increase native plant species by only one percent and at the same time increase non-native species by fifty percent, we should consider that a success. We would like to see more native species on the properties we manage and we are pleased at the slow, steady increases that we are achieving and we are interested to see where native plant cover will go from here on our properties. At the same time it seems unlikely that our management is going to serve as a native plant filter. Maybe it is unrealistic at this point to think that is possible in California annual rangelands without herculean efforts that cannot be sustained over a landscape level and may have ecological costs that outweigh their advantages. 

At the same time adding floristic biodiversity, even non-native biodiversity, has huge benefits to conservation. It seems possible that low levels of many non-native plant species in close association throughout the landscape will encourage the naturalization process, as predators and pathogens that are adapted to one species are exposed to other possible prey and hosts. More interesting is the fact that an increase in non-native plant biodiversity probably encourages an increase in native animal biodiversity. For example, the huge increase in forb species, both native and non-native, that we have seen and the percent of the landscape occupied by forbs, should have a major positive effect on pollinator species, most of which would be native. What the impact would be on other components of native biodiversity, that are much harder to study than plants, such as birds or soil biology are all questions we are interested in exploring.