Community-based long-term management to address reinvasion of restored grassland vernal wetlands

Exotic annual grasses can dominate ecosystems by producing a thick layer of dead plant litter, hereafter “thatch”, that promotes the regeneration of exotic grasses and inhibits native plants. Vernal pool wetlands within a grassland matrix are threatened by these exotic annual grasses, meriting the need for long-term management.  We utilized the investment of local community members to test the efficacy of long-term thatch management on urban vernal pool plant assemblages. We recruited over 40 undergraduate students to perform manual annual summer thatch removal around the edges of 15 urban vernal pools for four years.  We coupled thatch removal with annual native seed addition because our analysis of environmental DNA (“eDNA”) in the soil seed bank revealed a lack of native plant species and an abundance of exotic plant species. We measured vegetation composition in a set of 180 permanent monitoring quadrats (12 per experimental pool) over five years to quantify the effect of annual thatch removal and native seed addition on thatch, bare ground, native plant species cover and richness, and exotic plant species cover and richness. Our annual thatch removal treatment successfully reduced thatch accumulation and increased bare ground, but it did not result in a consistent decrease in exotic plant cover or increase in native plant cover. Instead, the effects of thatch manipulation on plant composition were modulated by annual precipitation, with exotic species increasing during dry years and native species increasing during wet years. The addition of native plant seed caused an increase in native plant species richness, but only after three years of annual thatch removal and native seed addition. Our results indicate that the restoration of native vernal pool plants can be limited by invasive species, native seed availability, and annual precipitation. Our findings show how engaging the local community in the long-term restoration of urban ecosystems can address the persistent threat of invasion and build up capacity for native plant populations to increase over time.

Methods
Species Composition Surveys

Plant species cover and richness were monitored annually starting in August 2019 before the first treatment year to determine baseline vegetation composition (all species were identifiable even after senescence), and then in June 2020 to 2023.  Percent cover of each native and exotic species was measured in 12 permanent 1-m2 quadrats in each pool (including the six quadrats that were seeded in the experiment) using a quadrat with 1% subdivisions.  Total percent cover of each quadrat often exceeded 100% due to multiple canopy layers of graminoids and forbs.

Additionally, we measured the richness of native and exotic plant species in the seed bank prior to experimental manipulations.  We collected and mixed three 8 cm-deep soil cores from each of the permanent monitoring quadrats using a 4 cm-diameter auger in July 2019.  We spread 50 g of each homogenized sample over PRO-MIX© BX BiofungicideTM potting soil in germination trays.  Trays were set up outside in November 2019 and hand-watered weekly.  Mauchamp and colleagues (2002) concluded that only identifying species that germinated from field soil cores in a controlled environment in a traditional “grow-out” analysis was inadequate in capturing the total seed bank diversity in wetlands; therefore, we assessed seed bank diversity via soil DNA sequencing according to protocols developed by Stephanie Ma Lucero and colleagues (pers. comm.).  We sieved each homogenized soil sample through a 4 mm sieve, then mixed 5 g of the homogenized soil with liquid nitrogen and extracted DNA using the Qiagen DNeasy® PowerSoil® Pro Kit.  Extracted DNA was amplified using standard polymerase chain reaction (PCR) protocols, using the rbcL forward and reverse primers.  Samples underwent two PCR amplification cycles to attain sufficient DNA for sequencing, and amplicons were then cleaned using AMPure beads.  Amplicons were sequenced using an Illumina 600-cycle MiSeq Reagent Kit V3.  Sequences were cleaned using the “dada2” coding package and then matched with family, genus, and species using the National Center for Biotechnology Information’s Basic Local Alignment Search Tool (BLAST), BLAST+, and the “annotate” coding package (Gentry 2024; Callahan et al. 2016).  Sequences were matched to taxa based on the BLAST Maximum Score metric.

Data Analysis

All analyses and visualization were performed in Rstudio version 1.4.1106 (R Core Team 2023).  All graphs were generated using the functions in the package “ggplot2”.

We compared the cumulative effects of treatments on diversity metrics using an analysis of variance via the aov and anova functions from the “stats” package.  We summed the percent cover of every exotic or native plant species to obtain the total exotic or native plant cover for each quadrat at each sampling time.  We also used the “vegan” package to calculate the Simpson’s Index including all native and exotic species for each quadrat at each time to assess how treatments affected species evenness and richness (Oksanen et al. 2022).

Most of the datasets were not normally distributed based on diagnostic residual tests, which thus required the use of models that accounted for the particular distributions that best fit the datasets.  We constructed repeated measures generalized linear mixed-effects models (GLMMs) with sampling year, thatch manipulation treatment, and seeding treatment, and their interaction effects, included as fixed effects, and quadrat included as a random effect to account for repeated measures.  The thatch percent cover dataset followed a beta distribution based on diagnostic residual tests.  The bare ground percent cover and total native plant percent cover datasets were zero-inflated, so we constructed hurdle models with gamma distributions.  Moreover, the percent bare ground dataset exhibited unequal variances, so percent bare ground was log-transformed using the appropriate Box-Cox transformation (Osborne 2019).  Total exotic plant percent cover among quadrats also followed a gamma distribution and exhibited unequal variances, so total exotic plant percent cover was square-root-transformed using the appropriate Box-Cox transformation.  Native and exotic plant species richness datasets followed a Poisson distribution.  GLMMs were generated using the “glmmTMB” and “lme4” packages (Brooks et al. 2017; Bates et al. 2014).  Post-hoc Tukey’s Honestly Significant Difference tests were performed on GLMM outputs using the emmeans function from the “emmeans” package to determine significant interactions among year, thatch treatment, and seeding addition on dependent variables (Lenth 2023).

Additionally, we determined differences in plant community composition among thatch and seeding treatments by using a permutational multivariate analysis of variance (PERMANOVA) via the adonis2 function of the  “vegan” package (Oksanen et al. 2022).  Post-hoc pairwise comparisons between two-way factorial thatch manipulation and seeding treatments were evaluated using the pairwise_adonis function developed by Maurice Goodman (O’Leary et al. 2021).  We also conducted a Similarity Percentages analysis on community matrices using the simper function of the “vegan” package (Oksanen et al. 2022).

Literature Cited

Bates, Douglas, Martin Mächler, Ben Bolker, and Steve Walker. “Fitting linear mixed-effects models using lme4.” Journal of Statistical Software 67, no. 1 (2014).

Brooks, Mollie E., Kasper Kristensen, Koen J. Van Benthem, Arni Magnusson, Casper W. Berg, Anders Nielsen, Hans J. Skaug, Martin Machler, and Benjamin M. Bolker. “glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling.” The R journal 9, no. 2 (2017): 378-400.

Callahan, B. J., Paul J. McMurdie, Michael J. Rosen, Andrew W. Han, Amy Jo A. Johnson, and Susan P. Holmes. (2016). DADA2: High-resolution sample inference from Illumina amplicon data. Nature Methods, 13(7), 581–583.

Gentry J (2024). Annotate: Annotation for microarrays. R package version 1.82.0.

Lenth R (2023). emmeans: Estimated Marginal Means, aka Least-Squares Means. R package version 1.8.9, https://CRAN.R-project.org/package=emmeans.

Mauchamp, André, Philippe Chauvelon, and Patrick Grillas. "Restoration of floodplain wetlands: opening polders along a coastal river in Mediterranean France, Vistre marshes." Ecological Engineering 18, no. 5 (2002): 619-632.

Oksanen J, Simpson G, Blanchet F, Kindt R, Legendre P, Minchin P, O'Hara R, Solymos P, Stevens M, Szoecs E, Wagner H, Barbour M, Bedward M, Bolker B, Borcard D, Carvalho G, Chirico M, De Caceres M, Durand S, Evangelista H, FitzJohn R, Friendly M, Furneaux B, Hannigan G, Hill M, Lahti L, McGlinn D, Ouellette M, Ribeiro Cunha E, Smith T, Stier A, Ter Braak C, Weedon J (2022). vegan: Community Ecology Package. R package version 2.6-4. https://CRAN.R-project.org/package=vegan.

O’Leary, Jennifer K., Maurice C. Goodman, Ryan K. Walter, Karissa Willits, Daniel J. Pondella, and John Stephens. "Effects of estuary-wide seagrass loss on fish populations." Estuaries and Coasts 44, no. 8 (2021): 2250-2264.

Osborne, Jason. "Improving your data transformations: Applying the Box-Cox transformation." Practical Assessment, Research, and Evaluation 15, no. 1 (2019): 12.

R Core Team (2023). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/>.