Within each quadrat, we determined the identity and percent cover of all species present. We also recorded the percent cover of bare ground, water, and thatch . In addition, we estimated the number and percent cover of germinating seedlings for native species. Because lowgrowing graminoids and forbs were often overlaid by taller species, the total percent cover could exceed 100% in each quadrat. To measure the pool area, we used a Trimble GPS to map out the perimeters of each pool. We used a laser level to measure the depth of each pool. We obtained climate data from the National Oceanic and Atmospheric Administration Daily Summaries dataset for the Santa Barbara Municipal Airport weather station to calculate the average annual rainfall each pool experienced after it was restored .The 69 pools surveyed in this study were restored between 1986 and 2017. The pools all shared similar attributes in terms of past and restored abiotic and biotic conditions, so we constructed a chronosequence that used a space-for-time substitution to examine the effect of time since restoration on native and exotic cover and richness. Past restoration actions included grading and berm enhancement to attain basin topography with an area ranging from 66 to 1,367 m2 and a maximum depth ranging from 53.5 to 80 cm, planting of locally-sourced native plant species via seeding and transplanting, and hand-weeding and herbicide treatments of exotic species during a 2- to 5-year implementation phase . In the spring of 2019, we conducted vegetation surveys in each pool when the majority of the native species were at peak biomass. For each pool, we laid out 2 transects bisecting the pool along its elliptical major and minor axes . Every other meter along each transect, grow room we laid down a 1 m2 quadrat with 1% subdivisions. We identified every plant species present and estimated its percent cover in each quadrat. We also estimated the percent cover of bare ground and thatch. Because low growing graminoids and forbs were overlaid with taller species, the total percent cover could exceed 100% in each quadrat.

We also categorized each quadrat as being in the central, transition, or upland zone of the pool. To measure relative elevation, we used a laser level to calculate the elevation of each quadrat above the deepest point of the pool. To determine pool hydroperiod, we installed 0.8 m rulers in the deepest part of each pool in January 2019 and recorded the depth of the water in each pool every week beginning 11 January until all the pools dried up by 5 July. To measure the site and pool area, we used a Trimble GPS to map out the perimeters of the sites and the pools. We also used these data to calculate each pool’s perimeter-to-area ratio and the distance of each pool from the edge of the restoration site. We obtained climate data from the National Oceanic and Atmospheric Administration Daily Summaries dataset for the Santa Barbara Municipal Airport weather station to calculate the precipitation each pool experienced the year before restoration began, the precipitation each pool experienced the year that restoration began, the precipitation each pool experienced the year after restoration began, and the average annual precipitation each pool experienced after restoration began .For each quadrat in each sampling year, we calculated the maximum monthly exotic plant species percent cover, total exotic plant species richness, maximum monthly native plant species percent cover, and total native plant species richness. The exotic species cover distribution was skewed right as determined by histogram and Q–Q plot analyses, so we used raw data to construct a generalized linear mixed effects model with a gamma distribution, using a logarithmic link function. The exotic species richness and native species richness distributions were not normally distributed as determined by histogram and Q–Q plot analyses, so we used raw data to construct a generalized linear mixed effects model with a Poisson distribution. The native species cover distribution was normally distributed according to histogram and Q–Q plot analyses, so we used raw data to construct a linear mixed effects model.

All four models were predicted by the age of the pool during each sampling year and the zone , and the interaction thereof, as fixed effects, with sampling year, quadrat name , pool depth , pool area , and average annual precipitation included as random effects.The increase in exotic cover and richness in our multiyear monitoring study suggests that short-term restoration efforts do not guarantee long-term success in the transition and upland zones of restored pools. The pools in this study were created and planted with native species within a grassland landscape. Intensive exotic species weeding continued for about 2–5 years after each pool was created, but then the pools entered the maintenance phase and were only periodically hand-weeded or cleared with a weed-whacker. Although the initial intensive weeding kept exotic cover low, exotic cover increased in the transition and upland zones over time. This suggests that the initial weeding successfully reduced exotic species, which is why exoticcover remained low for several years after the implementation phase. However, without continual removal, recruitment from exotic populations adjacent to the restored pools allowed for eventual recolonization of the site. Previous studies have shown that restored native populations can subsequently decline and even go extinct due to low growth rates that are negatively affected by interannual environmental variability and competition by invasive species . Indeed, other long-term monitoring studies in other ecosystems, such as grasslands and forests, have also shown that restored plant communities never reach the species diversity of natural reference ecosystems . Our study adds to a growing body of evidence that short-term restoration projects do not guarantee the long-term persistence of diverse native assemblages. Our results indicated that exotic plants invaded pool transition and upland zones, but not central zones, suggesting that invasion into the pool edges comes from the surrounding invaded grassland matrix. Invasive exotic species are often unsuccessful in the central zones because of their inability to tolerate prolonged inundation . However, increased drought due to climate change may result in drier conditions even in the deepest parts of pools, perhaps making the zone less hospitable for vernal pool specialists and more susceptible to natural recruitment by invasive species .

Although restoration efforts may plant and establish native populations within a vernal pool, the surrounding landscape often consists of unrestored grassland invaded by exotic grasses, which may contribute many propagules to pool edges. In addition, once propagules establish in the pool, positive feedbacks such as litter build-up can cause exotic populations to invade and persist . These edge effects are common throughout restored ecosystems . Small-scale restoration projects, which typically occur amidst fragmented habitat in the form of patches, can be susceptible to edge effects due to stressful environmental conditions and disturbances originating outside of the habitat patch . For example reinvasion of Phragmites australis from the surrounding landscape into wetlands is common, as is the encroachment of trees from forests into adjacent meadows . Several studies have shown that exotic species abundance increases closer to forest edges, where disturbance and exotic propagule supply is high . It is, therefore, critical to evaluate and manage edges of restoration projects as they face unique pressures that can jeopardize native assemblages.Our results highlight the importance of both sustained inundation of central zones and active management of transition and upland zones of vernal pools to reduce invasion. Collinge et al. have similarly emphasized the role of both abiotic and biotic filters in creating and sustaining restored native communities that are resistant to exotic invasion . Biotic filters that can decrease susceptibility to reinvasion include adaptive management strategies, such as planting with competitive native species and active control of exotic competitors through an array of long-term weed management techniques . In vernal pools, drying cannabis strategically planting suites of species at different elevation zones within pools can also increase native establishment and persistence. For example, in our studies, E. macrostachya, J. mexicanus, and J. phaeocephalus were able to dominate the central zone, while Carex praegracilis, E. macrostachya, Distichlis spicata, J. mexicanus, and E. triticoides performed well in the transition zone, and Stipa pulchra, Cyperus eragrostis, and Hordeum brachyantherum were able to establish and persist in the upland zone despite exotic invasion, so these species can be the foci of zonal planting palettes for future local restoration projects. Although intensive hand-weeding did not create resistance in the edges of the pools and may not be sustainable in the long run due to time and resource constraints, feasible long-term weeding strategies may focus more on large-scale contexts. For example, the upland and surrounding unrestored grassland matrix probably accounted for the exotic invasion of the transition and upland zones of the pools, so large-scale grassland management techniques such as grazing and prescribed fire disturbance may reduce exotic species dominance in both the grassland and the edges of the vernal pools . Even periodic reductions of exotic species could help to sustain greater native abundance in the edge zones. Overall, our studies evaluating the trajectories of plant assemblages post-implementation suggest that active management of restored habitats should persist beyond the implementation phase, which means projects need to be budgeted with long-term monitoring and adaptive management plans. Although 5 years of intensive restoration efforts can successfully reestablish native assemblages, our studies showed that native cover and richness decreased significantly in older pools. Other studies of restored wetlands similarly showed that restored wetlands initially achieving high native plant diversity can subsequently experience a decline in native diversity and an increase in exotic diversity 5–11 years post-implementation . Our long-term monitoring dataset provides unique insight into plant community trajectories over time by showing that, even when central zones of restored vernal pools can remain native-dominated, the drier pool edges exposed to the surrounding exotic grassland matrix can experience reinvasion over time, much like how forest edges and other edge habitats can experience reinvasion when not actively managed . Short-term success can be misleading, and long-term monitoring is important to evaluate the success of restoration and guide adaptive management over time. Identifying drivers of reinvasion can be particularly useful for guiding adaptive management. In our study, the main abiotic variables that correlated with increased exotic diversity and/or decreased native diversity were the amount of edge area, relative elevation, and precipitation. For example, less precipitation during restoration implementation can correlate with higher exotic richness, although a wet year before restoration may promote higher exotic cover and lower native cover in the upland zone, perhaps due to competition from exotics taking advantage of higher winter water resources . Although the precipitation that a restoration site experiences cannot be manipulated, knowing whether it is a particularly wet or dry year at a restoration site can inform management decisions, e.g., resources should be allocated to weeding exotic species out of pool edges during wet years. In addition, the invasion front of vernal pools may be reduced by creating circular pools with less edge area exposed to the surrounding exotic grassland matrix and associated edge effects. Because surrounding invasive grassland populations contribute propagules that invade pool edges, restoration efforts can also prioritize creating or restoring vernal pools in smaller grassland sites with fewer invasive species. For example, vernal pools may be constructed in smaller green spaces within urban areas that are traditionally deemed too small for other habitat restoration projects. However, manipulation of these abiotic environmental variables alone cannot be relied upon to maintain high native cover and low exotic cover, especially in the higher-elevation transition and upland zones that are more hospitable to generalist species. These edge zones experienced an increase in exotic diversity and/or a decrease in native diversity over time, possibly due to the overwhelming propagule pressure from the surrounding unrestored grassland. These propagules likely take advantage of the higher-elevation edge zones of the vernal pools that, when not seeded with native species, provide hospitable open niche space for generalist grasses and forbs to inhabit . Other studies have shown that abiotic manipulation can lead to incomplete restoration, especially in hospitable environments that are easily colonized by exotic species . Sengl et al. showed that retired farmland passively restored to grassland did not achieve the same native species richness as reference sites and were instead colonized by invasive grasses.