Recipient Organization
UNIVERSITY OF CALIFORNIA, DAVIS
410 MRAK HALL
DAVIS,CA 95616-8671
Performing Department
Plant Sciences
Non Technical Summary
California grasslands provide economically and environmentally important ecosystem services including forage for cattle, soil stabilization, carbon storage, nutrient cycling, and habitat for rare and endangered species. These systems have been subjected to drastic human manipulation from over-grazing and development, both of which have contributed to changes in species composition and the dominance of these areas by invasive annual plants. While some of the more naturalized invaders provide forage for cattle, other invaders cause widespread problems for rangeland sustainability by limiting forage production, outcompeting desirable species, increasing fire frequency, and reducing water availability. Limiting invader establishment and eradicating harmful invaders is thus a crucial component of rangeland management and overall grassland health.The proposed research seeks to understand seed banking strategies in invasive annual plants in order to more effectively manage their control. Many plants use seed banks, or pools of viable dormant seeds in the soil to minimize their risk of germinating into a poor environment. While a healthy native seed bank may increase a system's resilience to change, a dense invasive seed bank would have the same effect, making it more difficult to eradicate an invader. These seed bankers present an entirely different problem from species without seed banks, as eradication aboveground does not necessarily indicate successful control. My project uses a combination of experimentally-collected data, existing observational data, and novel modeling techniques to test how seed banking and non-seed banking invasive annual species respond to climate and to predict time to eradication for these invaders. It then uses an existing management dataset to generalize these findings to invasive species not included in the experiment. Together this project will help researchers better understand seed banking strategies of plants and will help land managers better control invasive species. Further, it will provide the primary director with critical training as a scientist and educator.
Animal Health Component
50%
Research Effort Categories
Basic
50%
Applied
50%
Developmental
(N/A)
Goals / Objectives
The overarching goal of this project is toimprovethe health of grassland systems by managing invasive species through an increased understanding of seed banks. Many invaders are thought to maintain seed banks, but we know very littleabout what types of species generally maintain seed banks, and even less about how these species are going to respond to future climates, all of whichcomplicatestheir control. For effective management, managers needto make decisions on which invaders to prioritize and when to prioritize them.This project has three objectives: (1) understand the population dynamics of seed banking and non-seed banking invasives under different rainfall regimes, (2) project time to eradication of these different invaders, and (3) generalize this information to a broader range of species.
Project Methods
How does the seed banking strategy of a species affect its invasiveness under different rainfall regimes?To answer this question, I will use a manipulative watering experiment to monitor germination and fecundity in a set of annual invaders that vary in seed banking strategy. I will conduct this field experiment in a valley grassland at the Sierra Foothill Research and Extension Center (SFREC) near Browns Valley, CA. This site is actively grazed and is a characteristic rangeland of California with a Mediterranean climate (cool, wet winters and dry, hot summers) and high levels of interannual rainfall variability. Rangelands in California are dominated by annual plants that typically germinate in the fall (October-December) and then set seed the following spring/summer (April-June) before finishing their life cycle.Dr. Hallett, the Primary Mentor, has an existing watering experiment at SFREC where rainfall is manipulated to mimic high and low rainfall years, with 8 sheltered plots (low rainfall), 8 water-addition plots (high rainfall), and 8 control plots (ambient rainfall). Sheltered plots consist of a 6.4 × 5.2 m metal frame with clear polyethylene roofs that are erected shortly before rainfall events and are subsequently removed to limit the effects of solar radiation. This method successfully reduced soil moisture by 50% throughout the growing season in a previous project using these shelters. Watered plots are outfitted with a microsprinkler in the center of the plot. Amount of water added will increase hourly rainfall in treatment plots by 50% throughout the growing season (October through April). Soil moisture will be measured using five soil moisture probes in each plot.For this study, I chose a set of 14 invasive annual grasses and forbs that vary in invasiveness according to the California Invasive Plant Council Inventory. Although seed bank information on many of these species is limited, literature suggests a large range in seed bank persistence.I will collect seeds of each species from SFREC and I will sow 25 seeds eachinto 10x10 cm subplots, with three subplots per species to investigate the effects of density on fecundity. I will record germination, thin to the desired density of the subplot (low = 2 individuals, medium = 5, and high = 10), and mark individuals with toothpicks. To limit recruitment from the natural seed bank, herbicide has been applied to these plots before seed production for the past two years and will be applied againprior to the experimental set up. Fecundity estimates will be assessedby counting the number of surviving individuals along with the number of flowers in each plot to calculate a per individual estimate of flowers. This number will be multiplied by the average seed production of up to five individuals per subplot.To assess seed banking strategy, I will measure annual seed survival. I will bury seed bags of each species 5 cm below the soil surface into each plot in early fall, and dig them up the following fall. Seeds will be cut open and tested for viability using tetrazolium staining techniques. Species with higher annual seed survival will represent those that are likely to form seed banks, while those with lower seed survival will be representative of non-seed banking species.To test the effects of watering treatments on germination and seed production of species with varying seed banking strategies, I will use generalized linear models in R with watering treatment, seed survival, seeding density and their interactions as my predictors and a random intercept for species within subplot within plot.What duration of control is needed to effectively eradicate seed banking and non-seed banking invaders? To answer this question, I will parameterize an annual plant population growth rate model and introduce environmental stochasticity with a two-state markov chain to predict time to extinction for the various invaders given recruitment, seed set, and seed survival in the different rainfall treatments as well as some level of eradication each year. Specifically, I will use the demographic data gathered in my field experiment to parameterize a density dependent annual plant population growth rate model modified to include eradication effort.?The future distribution of wet and dry years will be essential to understanding how populations of invaders change over time. I will include environmental stochasticity in the model by allowing germination rates and fecundity to differ between wet years (i.e. vital rates obtained under watering treatments) versus dry years (vital rates obtained under shelter treatments). First, I will determine the future probability of wet vs. dry years based on the average of 34 CMIP5 global climate models (GCMs) for California then incorporate these probabilities into a two-state markov chain model that describes environmental transition probabilities. With this stochastic growth rate model, I will use numerical techniques to calculate the conditions under which extinction occurs as well as the time to extinction for each invasive species. I will then look for patterns of extinction behavior based on each species' tendency to form seed banks. To investigate whether seed banking strategy can be used to predict time to extinction, I will group species by seed banking strategy using k-means clustering on seed survival data and compare model behavior between the species-level and strategy-level models.How can this information be used to manage existing rangelands? To generalize and apply what I learn from the experimental study, I will use an existing 7-year monitoring dataset from the East Bay Regional Park District (EBRPD), an actively grazed rangeland with high species overlap from my experiment, and use HMMs to predict seed survival and assign seed banking strategy to the species in the dataset. I will then forecast the duration of control for species in this dataset using the previously built stochastic annual plant model.The EBRPD dataset consists of three sites (Vasco Caves Regional Preserve, Pleasanton Ridge Regional Park, and Sunol Regional Wilderness), each with six circular 900-m2 plots where presence/absence of aboveground species was monitored from 2005-2012. Both the EBRPD sites and SFREC are considered annual grassland/hardwood systems. They share similar species composition, annual rainfall, and soils and have been paired together in a previous study.I will use HMMs to estimate seed survival of each invader in the EBRPD dataset. These models take an observed state, in this case the presence/absence of each species aboveground, to predict the presence of the hidden state, the seed bank (Figure 1). I will then use maximum likelihood to estimate transition rates between these two states for the species within the dataset: seed survival rates in the seed bank (s), colonization rates (c), and germination rates (g). For the species that occur in my experimental plots as well as within the EBRPD dataset (12 of the 14 species), I will use Root Mean Square Error to compare the predicted seed survival output from the HMMs to my experimentally obtained rates to test how well the models predict seed survival. I will then use the annual plant model with environmental stochasticity to investigate the duration and eradication dynamics of the species occurring in both sites. Finally, if I find that seed survival predicts response to rainfall in Question 1 and HMMs do an appropriate job of predicting seed survival in 12 of my study species, I will also generate control recommendations for species within the EBRPD dataset that were not a part of my experimental study. To do this, I will use bootstrapping to predict germination and seed set rates under high and low rainfall years given new levels of seed survival as predicted by the HMM, and use the annual plant model with stochasticity to generate new control scenarios.