ii
Acknowledgements
To my advisor, Dr. Michael Draney, I cannot thank you enough for your guidance, encouragement, and mentorship. Your lessons have helped me the past when writing this thesis and leading me through my studies, are helping me in the present for helping to get my results published, and will help me in the future for wherever my ecology career takes me.
To my committee members, Dr. Matthew Dornbush and Dr. Patrick Robinson, thank you for your valuable input for making my thesis the best that it could be, especially in regards to interpreting sparse data and expanding my conclusions to the greater picture.
To the Heirloom Plant fund and the Office of Graduate Studies at the University of Wisconsin - Green Bay, thank you for the financial support necessary to fund both this thesis project and my travel to Golden, CO to present preliminary data at the 20th International Congress of Arachnology.
To statistician Dr. Megan Olson-Hunt, thank you for helping me navigate the sometimes complicated world of statistical analysis. Your suggestions and viewpoints helped me make sense of some of my more difficult analyses and made me a convert to R.
To Kathryn Corio of the Fewless Herbarium and Kate Hau of Bay Beach Wildlife Sanctuary, thank you for helping me identify the trickier plant species and saplings that I encountered in my field sites.
To Mike Reed of Bay Beach Wildlife Sanctuary, Mark Konlock of the Green Bay Botanical Garden, and the Friends of Peninsula State Park, thank you for helping me locate a suitable field site to research garlic mustard.
To all of my colleagues who helped me with various tasks at my field site, from vegetative sampling, to setting up and breaking down plots, to taking plant measurements, thank you for your kindness and helping hands. Mia Spaid, Holly Harpster, Candace Kraft, Claire Rebman, Ellie Roark, and Wilson Gaul, there was no way I could finish this project alone.
To my family and friends, thank you for your emotional support during some of the more difficult periods of my graduate studies. Also, thank you for your advice on topics such as beating writer's block and designing figures that are easier for the reader to see.
To anybody else that I did not mention by name, but helped me finish this thesis and earn my Master's degree, thank you. No matter how great or small your contribution, this thesis could not be done without you.
iii
Abstract
Spider diversity response to garlic mustard (Alliaria petiolata) invasion in a Wisconsin forest understory
Megan Sprovach
The invasive herb garlic mustard is considered a major contributor to biodiversity loss in forest understories throughout the United States, including in northeastern Wisconsin. Research on the effect of garlic mustard invasion on spiders is limited. The purpose of this project is examine the effect of garlic mustard invasion on the spider community structure at Bay Beach Wildlife Sanctuary in Green Bay, Wisconsin. Four pitfall traps per each of thirty 5 X 5-meter plots, representing a gradient of garlic mustard cover, were sampled monthly from June to September in 2015. Time-controlled vegetative sampling was conducted in each plot at the beginning and end of the collection season and all vegetation within each plot was identified to species. Spider species richness, spider Shannon diversity, and the ratio of web-building to wandering spider species were quantified at the plot level. The spider richness, diversity and guild ratios were modeled against garlic mustard cover. Individual spider species were examined for correlation with garlic mustard cover.
When accounting for plant height deviation, the relationship between Shannon diversity and garlic mustard cover is inconclusive, but spider species richness increased with greater garlic mustard cover. The ratio of web-building to wandering spiders was not correlated with garlic mustard cover. June vegetative samples of the native linyphiid Ceraticelus fissiceps were also positively correlated with garlic mustard, but not strongly enough to be used as an indicator for plot-level ecological change. The opportunistic native linyphiid Diplostyla concolor and the native salticid Pelegrina proterva were also commonly found in the plots. These results suggest that the presence of an invasive species in a habitat could benefit spider populations under unusually low biodiversity conditions, but caution must be exercised when considering these results for policy implications.
iv
Table of Contents
Acknowledgements ...... ii
Abstract ...... iii
Table of Contents ...... iv
List of Tables ...... v
List of Figures ...... vi
Introduction ...... 1
Methods ...... 8 Research Site ...... 8 Sampling Methods ...... 8 Plant Identification ...... 10 Statistical Analyses ...... 11
Results ...... 14 Summary of Sampling Data ...... 14 Pitfall Trap Data Analysis ...... 22 Vegetative Sample Data Analysis ...... 24 Plant Height as a Fixed Effect ...... 30
Discussion ...... 32
Policy Implications ...... 38
References ...... 42
v
List of Tables
Table 1 List of plant species found at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin ...... 15
Table 2 List of all identified spider families and species from pitfall and vegetative samples at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin . 17
Table 3 Correlation matrices for pitfall trap and vegetative sample variables of interest ...... 21
Table 4 Ranked abundance of spider species (adult specimens only) sampled from pitfall traps at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin ...... 23
Table 5 Ranked abundance of spider species (adult specimens only) sampled from vegetation at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin ...... 25
vi
List of Figures
Figure 1 Number of garlic mustard stems versus number of active spider webs. Adapted from Smith and Schmitz (2015), Appendix A1 ...... 5
Figure 2 Diagram of the quadrat used for vegetative sampling ...... 11
Figure 3 Frequency of Alliaria petiolata percent cover values in the Bay Beach Wildlife Sanctuary research site by number of plots ...... 14
Figure 4 Species accumulation curves for pitfall and vegetative sampling data from June, July, August, and September at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin ...... 20
Figure 5 Relationship between garlic mustard cover and the Shannon diversity of adult spiders sampled from pitfall traps placed in June, July, August, and September ...... 24
Figure 6 Relationship between garlic mustard cover and the Shannon diversity of adult spiders sampled from vegetation in June and September ...... 26
Figure 7 Relationship between garlic mustard cover and adult spider species richness sampled from vegetation in June and September ...... 27
Figure 8 Square-transformed Shannon diversity of spider species sampled from June vegetation according to stratified categories of garlic mustard percent cover ...... 29
Figure 9 Relationship between number of Ceraticelus fissiceps sampled in June vegetative samples and garlic mustard percent cover. Points indicate individual plot values ...... 30
1
Introduction
Invasive species are a driver of ecological change (Vitousek et.al. 1996). They are present in almost all modern ecosystems, negatively impacting the biodiversity of an ever-increasing number of habitats, modifying ecological processes, and increasing the costs of land management. One such invasive, garlic mustard (Alliaria petiolata), is an obligate biennial herb native to Europe, from where it was brought to North America for use as a medicine and a foodstuff (Grieve 1959). Since its introduction in 1868, it has spread across the United States and Canada at an alarming rate, outcompeting native understory species in the forests that it invades (Nuzzo 2000). Forests in the midwestern and northeastern United States are especially affected (Anderson et.al. 1996; Nuzzo
2000).
The life cycle of garlic mustard provides insight into how the herb is able to successfully establish itself in forest understories. Immature plants form basal rosettes and remain in this form throughout the first growing season and while they overwinter
(Nuzzo 2000). Plants reach adulthood during the second year and produce flowering stalks, dying just after seed production. Stands of garlic mustard are known to be composed of even- to near even-aged individuals, especially in the early stages of invasion. The plant’s gametes can be fertilized either via generalist pollinators or autogamously, and it produces a large number of viable seeds (Anderson et.al. 1996). The herb shows a high degree of phenotypic plasticity, as factors such as light level can affect the plant’s structure at both the individual and population level (Rodgers et.al. 2008). In addition, garlic mustard is allelopathic, producing a variety of chemical compounds,
2 including glucosinolates, which not only protect the plant against herbivory, but can leach into the soil via the roots and inhibit the growth of competing native plants. Some of these leachates can disrupt mycorrhizal fungi, limiting the types of plant species that can survive in garlic mustard dominated habitat (Roberts and Anderson 2001).
Several studies have assessed the impact of garlic mustard on herbivorous arthropods, particularly butterflies, weevils, and flea beetles (Haribal and Renwick 1998;
Renwick and Lopez 1999; Blossey et.al. 2001; Renwick et.al. 2001; Davis et.al. 2006;
Keeler and Chew 2008). However, comparatively few studies have examined the effect of garlic mustard on higher trophic levels. Dávalos and Blossey (2004) collected predatory ground beetles from underneath coverboards placed in 89 sampling sites within adjacent 64 x 64-m plots in two central New York forest parks. One plot represented a garlic mustard-invaded habitat, and the other plot represented uninvaded habitat. They compared ground beetle (Coleoptera: Carabidae) count and biomass with species richness from understory plants sampled from the same sites. The researchers found little influence from garlic mustard on Carabidae. However, a single taxon of arthropods may not represent all invertebrate species, especially those with different trophic levels and life histories. For instance, a similar study was conducted by Alerding and Hunter (2013), sampling springtails (Collembola) from epigeal cores within garlic mustard invaded and non-invaded plots. Soil and leaf litter samples in garlic mustard-invaded areas contained two to three times more springtails as control samples. Neither study examined invasion gradients, focusing on simple garlic mustard presence or absence.
Given the varying results, more studies are needed on relationships between garlic mustard and invertebrates to better understand which taxa are most affected by garlic
3 mustard invasion. Arthropods are important components of all ecosystems as their presence affects all ecological processes. A large portion of animal decomposition is mediated by insects such as carrion beetles (Coleoptera:Silphidae) (Gibbs and Stanton
2001). Burrowing in soils by termites (Isoptera) and ants (Formicidae) alter rates of water infiltration into the soil (Lavelle 1997). Insects, especially bees (Anthophila), account for the majority of animal-mediated plant pollination (Buchmann and Nabhan 1996). In addition, some of these arthropods might serve as effective ecological indicators for monitoring habitat quality in forest understories. Such tools are highly sought-after by land managers wishing to monitor habitats, as attempting to monitor every single species in a habitat is unfeasible (Lambeck 1997).
When selecting indicator species, researchers must keep generation time, taxon diversity, and habitat specificity in mind (Carignan and Villard, 2002). A longer generation time may lead to an extinction debt, a phenomenon where species experience a lag in their extirpation after habitat destruction, making their decline much less apparent than for short-lived species (Tilman et.al. 1994). The lack of rapid response to ecological change would limit such species’ use as an early warning system for habitat degradation
(Temple and Wiens 1989). In addition, the ecological needs of indicator species do not completely overlap the ecological needs needs of the non-indicators that it is meant to represent (Lindenmayer 1999). Therefore, any monitoring of indicator species should encompass multiple taxa (Carignan and Villard, 2002). Potentially, species populations that are limited by habitat type would make better indicators than generalists, because specialist populations are expected to be more susceptible to the effects of habitat change.
4
Hutto (1998) identified multiple landbird species that were specific to both different habitats and specific disturbance regimes within the same habitat.
Spiders (Order Araneae) may be ideal invertebrates for use as biological indicators, as they are diverse and abundant in virtually all terrestrial habitats worldwide
(Wise 1993). Many spiders exhibit effective dispersal behavior by ballooning, which allows them to rapidly reach and colonize suitable habitat via wind dispersal (Crawford et.al. 1995). Many species of spiders are also sensitive to the structure of their habitat
(Wise 1993). Web-builders and cursorial spiders rely on the presence of ideal web locations and foraging substrate respectively. These locations often correlate to specific vegetation structures (Souza et.al. 2015, Wise 1993). Most spiders have a short lifespan of one to two years, which would make their populations responsive to environmental change and unlikely to exhibit extinction debt (Foelix 1982). Several studies have identified spiders as ecological indicators in multiple environments (Pétillon et.al. 2006,
Hore and Uniyal 2008, Steffen and Draney 2009, Buchholz 2010).
The only study found in a literature search that examines the relationship between spider diversity and garlic mustard was conducted by Smith and Schmitz (2015). They designed a model attempting to estimate how plant invasions impacted the herbivorous insect community by serving as ideal habitat for spider foraging. Their model incorporates field data obtained from plots representing a garlic mustard gradient. They concluded from a regression analysis that mature garlic mustard stems supported about five times as many spiders as native plant vegetation, although the authors did not verify the statistical significance of this finding or assess the model's fit (Figure 1). This field data only addressed web-building spiders that used garlic mustard as web substrate, with
5 an emphasis on the families Linyphiidae and Theridiidae for use in their model.
Therefore, the data is not representative of the entire spider community.
Figure 1. Number of garlic mustard stems versus number of active spider webs. Adapted from Smith and Schmitz (2015), Appendix A1.
The literature search failed to reveal any other studies pertaining to spiders and garlic mustard, but other invasive plant species have been shown to affect spider community structures. Bultman and DeWitt (2007) used pitfall traps to sample cursorial spiders from a Michigan dune succession forest invaded by Vinca minor. They found that the invasion transformed the local spider community from a vagrant web-builder- dominated community to a hunter-dominated assemblage. They hypothesize that this transformation was due to V. minor’s growth as a tangled mat of dense foliage,
6 dramatically changing the structure of the microhabitat. Web-building spider abundance was also disproportionately reduced in Phragmites australis-invaded Spartina salt marshes, relative to non-invaded marshes (Gratton and Denno 2005). Although spiders were not identified to species, Gratton and Denno (2005) reported a shift in spider family composition. Invasive plants have also been shown to affect the presence of native and invasive spider species, with either potentially benefiting from the invasion. Populations of the opportunistic sheet weaver Diplostyla concolor drastically increase in oak savannah invaded by common buckthorn (Steffen and Draney 2009). The wolf spider
Trochosa ruricola and the underleaf web-building spider Enoplognatha ovata, both from
Eurasia, also commonly dominate degraded forests in the Midwest (Draney pers. obs.).
Conversely, Centaurea maculosa provided a substrate for 38 times more web-building native spiders, especially Dictyna, compared to native grassland vegetation (Pearson
2009).
The objective of this thesis was to address a major knowledge gap by identifying the effects of garlic mustard dominance of eastern deciduous forest on understory spider communities. We asked two fundamental questions. First, how does the forest understory spider community structure differ across levels of garlic mustard abundance in the forest understory? To answer this question, I evaluated how garlic mustard invasion affects spider species richness and diversity. I hypothesized that spider species richness and diversity would be negatively correlated with the severity of garlic mustard invasion in any given habitat due to the loss of cover and substrate heterogeneity caused by a decrease in plant species richness. I also hypothesized that the abundance of web- building spiders would increase in more heavily-invaded habitats, while the abundance of
7 hunting spiders would decrease. Second, I asked if the abundance of any specific spider species correlated, positively or negatively, with the presence or density of garlic mustard? If so, what habitat characteristics might the species indicate, and was this species native or non-native? I hypothesized that, based on the majority of research on plant invasions, the populations of non-native invasive spider species would increase and the populations of native species would decrease in garlic mustard-dominated habitats.
8
Methods
Research Site
I sampled spider communities at Bay Beach Wildlife Sanctuary in Green Bay,
Brown County, Wisconsin. Located just south of the Bay of Green Bay, it was first established as an urban waterfowl refuge before expanding into a 700-acre property featuring animal exhibits, a wildlife rehabilitation program, and a trail system. Primary habitats within the sanctuary include manmade lagoons, wetlands, and southern mesic forest. I selected sampling locations from forest understories dominated by boxelder
(Acer negundo) and eastern cottonwood (Populus deltoides). Also present was green ash
(Fraxinus pennsylvanica), which was less frequent due to the presence of invasion by emerald ash borer (Coleoptera, Buprestidae: Agrilus planipennis). The forest understory was composed of a variety of forbs, along with saplings of the canopy tree species. The understory includes several populations of garlic mustard of varying size and age, with first-year rosettes being more populous and denser than second-year bolting plants.
Sampling Methods
In May 2015, I established thirty 5 x 5 meter plots in suitable mesic forest habitat.
The first plot was selected in a location that included a sub-population of garlic mustard.
We measured 20 meters starting from the center of the first plot. The location of the 20- meter mark served as the center of a new plot. We repeated this process until thirty plots were established, making sure that each plot center was placed 20 meters away from all others in an attempt to increase sample independence. We used flags to mark the center and corners of each plot. To ensure a spread of garlic mustard percent cover values,
9 twenty of these plots were located near the northeastern entrance at Danz Avenue, where garlic mustard was abundant, and ten plots were located near the northeastern entrance at
East Shore Drive, where garlic mustard was less abundant. In both sub-locations, we also placed the edge of each plot at least two meters away from any trails to reduce the impact of disturbance by humans using the trails.
I divided each plot into four quadrants and installed a pitfall trap in the center of each quadrant using a tulip bulb planter every month from June to September. I installed a total of 120 for one week each month, yielding 3,360 trap-days of sampling effort. The pitfall traps were composed of 120 mL plastic sample cups containing about one inch of
50% propylene glycol and a drop of dish detergent. The traps were placed so that the lip of the cup was at ground level. To protect the traps, as well as limit the size of animals caught, I roofed the traps with 15 x 15 cm square of plywood suspended by nails about
1.5 cm from the ground.
To account for vegetation-dwelling spider species less likely to be represented in pitfall samples, I sampled the ground-accessible vegetation in each plot in June and
September. For fifteen total minutes (0.25 person-hours of effort), we gently but thoroughly sampled each plot's vegetation for all spider-like invertebrates using visual search and by brushing specimens onto beating trays using horsehair brushes. Spiders were collected using aspirators and collecting jars and killed and preserved in 70% ethanol
I sorted the captured spiders and transferred any propylene glycol-preserved samples to 70% ethanol. Then, I identified all adult spiders to the species level. The first
10 two males and females of each species were selected as voucher specimens to be deposited in the Richter Museum of Natural History at UW-Green Bay.
Plant Identification
To assess plant biodiversity, I conducted vegetative surveys within each site once per plot during the growing season. At the plot level, I identified understory species within the bounds of the plot, as well as tree canopy species within viewing distance, using a visual search. At the quadrant scale, I placed a 1.0 x 0.5 -meter quadrat constructed from PVC piping one meter north of each pitfall trap and identified every plant species within the quadrat. I also measured the percent cover of each plant species and calculated the plant species richness. Because of their differing structures, I treated first-year and second-year garlic mustard as separate species, along with saplings of canopy trees.
To assess the structural diversity of the vegetation and leaf litter in each plot, I selected three equidistant points along the center of each quadrat (Figure 2). I used a meter stick to measure the height of the tallest plant at each point, as well as the depth of the leaf litter. I identified each plant species using Voss and Reznicek (2012) and sent any problematic specimens to be confirmed by the staff at the Fewless Herbarium at the
University of Wisconsin – Green Bay and at Bay Beach Wildlife Sanctuary.
11
0.25 m 0.25 m 0.25 m 0.25 m
0.5 m
1.0m
Figure 2. Diagram of the quadrat used for vegetative sampling. Stars mark the locations of leaf-litter depth and plant height samples, indicated by electrical tape placed on the quadrat's rim.
Statistical Analyses
I ran all statistical analyses using R version 3.3.0. To verify the soundness of my sampling design, I plotted first- and second-year garlic mustard in a histogram and assessed the spread of percent cover values. I created species accumulation curves to assess sampling effort sufficiency in a way that controls for different levels of spider diversity in each plot.
To assess spider biodiversity, I calculated the adult spider species richness, as well as the Shannon Diversity Index (H), at the plot level for each sampling method.
Shannon's H is calculated as
)
𝐻 = 𝑃$𝑙𝑛(𝑃$) $*+ where S is the number of species and 𝑃$ is the fraction of a population composed of species i (Shannon, 1948).
12
My predictor variables were the percent cover of first- and second-year garlic mustard, the standard deviation of plant height and leaf litter depth for each plot, and plant species richness. My response variables were Shannon diversity index values, spider species richness, the number of spiders for each species, and the ratio of web- building versus hunting spiders. I assessed possible relationships between my variables of interest using several methods. First, I constructed one of two correlation coefficient matrices to search for meaningful correlations between overall plot measurements or diversity indices for each sampling method. If the data were parametric, I used Pearson's coefficients and if non-parametric, I used Spearman's coefficients. If two variables showed no possible correlation in neither the pitfall trap data nor the vegetative sample data, I excluded those relationships from further analysis. Second, I identified the spider species accounting for the largest proportion of spider individuals per sampling method.
In both cases, I further assessed any possible relationships by plotting scatterplots, both for individual sample periods, and for the sampling methods as a whole. If any scatterplots showed a possible relationship, I attempted to fit a regression model to the relationship in question. If a variable could be corrected for normality, I transformed the variable in question. To account for unequal variances and missing data, I fitted mixed effect models to all relationships with repeated measures. I selected the best-fitting models using a combination of methods depending on the family of model, including assessing plots of fitted values versus residual values and comparing the AIC values of multiple plots. I calculated Satterthwaite approximations to obtain reliable p-values for any mixed effect models.
13
If the results of my analyses showed a statistically significant relationship between garlic mustard cover and another variable selected through the previous methods, I used similar methods to analyze any variables that might explain this relationship. If I needed to assess the nature of these relationships in greater detail, I stratified garlic mustard into four categories: "very low", "low", "high", and "very high", and used those categories to run an ANOVA.
14
Results
Summary of Sampling Data
Spearman's correlation coefficients revealed no potential relationship between first-year garlic mustard cover and any of my response variables for either sampling method. Therefore, all analyses were conducted on second-year garlic mustard cover. The plots I selected within the site represented a wide spread of second-year garlic mustard percent covers, although there was a very large number of zero or near-zero values
(Figure 3). Including garlic mustard, forty species of plants were found throughout the plots (Table 1). One species of Helianthus and one species of Digitaria were sampled without inflorescence. Attempts to collect flowering specimens of these species failed, and they consequently could not be identified to species. 14 12 10 8 6 Number of Plots 4 2 0
0 10 20 30 40 50 60 70 80 90
A. petiolata Percent Cover
Figure 3. Frequency of Alliaria petiolata percent cover values in the Bay Beach Wildlife Sanctuary research site by number of plots.
15
Table 1. List of plant species found at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin. Non-native or otherwise invasive species are denoted with asterisks.
Family Species Common Name
Apocynaceae Asclepias syriaca Common Milkweed Asteraceae Ageratina altissima White Snakeroot Asteraceae Arctium minus* Lesser Burdock Asteraceae Cirsium arvense* Canada Thistle Asteraceae Cirsium vulgare* Bull Thistle Asteraceae Helianthus sp. Sunflower Balsaminaceae Impatiens capensis Jewelweed Berberidaceae Berberis thunbergii* Japanese Barberry Boraginaceae Cynoglossum officinale* Gypsyflower Boraginaceae Hackelia virginiana Beggar's Lice Brassicaceae Alliaria petiolata* Garlic Mustard Brassicaceae Hesperis matronalis* Dame's Rocket Cyperaceae Carex stricta Upright Sedge Cyperaceae Carex bebbii Bebb's Sedge Fagaceae Quercus bicolor Swamp White Oak Geraniaceae Geranium maculatum Spotted Geranium Lamiaceae Galeopsis tetrahit* Brittlestem Hempnettle Lamiaceae Glechoma hederacea* Creeping Charlie Lamiaceae Leonurus cardiaca* Common Motherwort Oleaceae Fraxinus pennsylvanica Green Ash Onagraceae Circaea lutetiana canadensis Enchanter's Nightshade Oxalidaceae Oxalis stricta Yellow Woodsorel Poaceae Digitaria sp. Crabgrass Poaceae Elymus virginicus Virginia Wildrye Poaceae Phalaris arundinacea* Reed Canary Grass Poaceae Poa pratensis* Kentucky Bluegrass Ranunculaceae Thalictrum dioicum Early Meadow-Rue Rhamnaceae Frangula alnus* Glossy Buckthorn Rhamnaceae Rhamnus cathartica* Common Buckthorn Rosaceae Geum canadense White Avens Rubiaceae Galium triflorum Fragrant Bedstraw Salicaceae Populus deltoides Eastern Cottonwood Salicaceae Populus tremuloides Quaking Aspen Sapindaceae Acer negundo Boxelder Solanaceae Solanum dulcamara* Climbing Nightshade Urticaceae Boehmeria cylindrica Smallspike False Nettle Urticaceae Urtica dioica Stinging Nettle Violaceae Viola sororia Common Blue Violet Vitaceae Parthenocissus quinquefolia Virginia Creeper Vitaceae Vitis riparia River-Bank Grape
16
A total of 415 adult spiders were identified, representing 13 families and 44 species (Table 2). The genus Linyphiidae was the best represented, accounting for 12 species. Of note is that one species of Clubionidae, Clubiona opeongo, is noted to exist only in Canada, making this the first documented occurrence of the species in the United
States (Dondale and Redner 1982).
17
Table 2. List of all identified spider families and species from pitfall and vegetative samples at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin.
Family Species
Araneidae Araniella displicata (Hentz, 1847) Larinioides patagiatus (Clerck, 1757) Mangora placida (Hentz, 1847)
Clubionidae Clubiona abboti L. Koch, 1866 Clubiona opeongo Edwards, 1958
Dictynidae Emblyna sublata (Hentz, 1850)
Linyphiidae Agyneta fabra (Keyserling, 1886) Bathyphantes pallidus (Banks, 1892) Centromerus sylvaticus (Blackwall, 1841) Ceraticelus fissiceps (O. Pickard-Cambridge, 1874) Collinsia plumosa (Emerton, 1882) Diplostyla concolor (Wider, 1834) Glyphesis scopulifer (Emerton, 1882) Gonatium crassipalpum Bryant, 1933 Hypselistes florens (O. Pickard-Cambridge, 1875) Microneta viaria (Blackwall, 1841) Neriene clathrata (Sundevall, 1830) Pityohyphantes costatus (Hentz, 1850)
Lycosidae Pirata insularis (Emerton, 1885) Schizocosa ocreata (Hentz, 1844) Trochosa terricola Thorell, 1856
Philodromidae Philodromus rufus vibrans Dondale, 1964 Philodromus vulgaris (Hentz, 1847)
Phrurolithidae Phrurotimpus borealis (Emerton, 1911) Scotinella minnetonka (Chamberlin & Gertsch, 1930)
Salticidae Hentzia mitrata (Hentz, 1846) Naphrys pulex (Hentz, 1846) Neon nelli Peckham & Peckham, 1888 Pelegrina proterva (Walckenaer, 1837)
Tetragnathidae Tetragnatha dearmata Thorell, 1873 Tetragnatha guatemalensis O. Pickard-Cambridge, 1889 Tetragnatha straminea Emerton, 1884 Tetragnatha versicolor Walckenaer, 1841
Theridiidae Dipoena nigra (Emerton, 1882) Enoplognatha ovata (Clerck, 1757) Theridion murarium Emerton, 1882 Theridula emertoni Levi, 1954 Thymoites unimaculatus (Emerton, 1882) Yunohamella lyrica (Walckenaer, 1841)
Theridiosomatidae Theridiosoma gemmosum (L. Koch, 1877)
Thomisidae Ozyptila conspurcata Thorell, 1877 Xysticus ferox (Hentz, 1847)
Uloboridae Hyptiotes cavatus (Hentz, 1847) Uloborus glomosus (Walckenaer, 1841)
18
Species accumulation curves reveal that an adequate portion of spider species was accounted for in most pitfall trap sample months and all vegetative sample months
(Figure 4). The asymptote of each curve represents the number of species at which no more can be theoretically sampled. Both sampling methods revealed a low amount of spider species diversity. Pitfall trap species accumulation started to plateau at around 10 species in June and July, while vegetative sample accumulation started to plateau at around 30 species in June. Pitfall traps sampled in June and July show a large degree of overlap, with August and September pitfall trap samples showing similar overlap.
Some pitfall trap data was lost over the course of the study. Over the field season,
I lost 22.5% of traps in June, 5% of traps in July, 7.5% of traps in August, and 2.5% of traps in September. The high percentage of trap loss in June was caused by a major flooding event that occurred days after installing my traps, filling the collection jars and diluting the preservative. This was especially frequent in my plots near the East Shore
Drive entrance. Other factors for trap loss included damage to the traps and predation by raccoons (Procyon lotor).
Because the majority of my data were non-parametric, I used Spearman's coefficients to calculate the correlation matrices (Table 3). The pitfall trap correlation matrix revealed no possible correlations between garlic mustard cover and any other variable, but the vegetative sample correlation matrix revealed possible correlations between garlic mustard and plant height standard deviation, spider species richness, and the number of hunting spiders. In both sample types, the number of web-building spiders showed a possible correlation with spider richness and Shannon diversity. The number of hunting spiders from pitfall traps also showed a possible correlation with spider richness,
19 but at a much smaller degree than the number of web-building spiders. For the vegetative spider samples, plant height standard deviation showed a possible correlation with spider species richness, Shannon diversity, and the number of web-building spiders. There was no possible correlation between plant species richness or litter depth standard deviation and garlic mustard cover, spider species richness or Shannon diversity for either sampling method. Therefore, we decided not to examine these two variables any further.
20
Pitfall Traps 12 10 8 6 4 2 Accumulated Species Richness Accumulated 0
0 20 40 60 80 100 120
Sites
Vegetative Samples 35 30 25 20 15 10 5 Accumulated Species Richness Accumulated 0
0 5 10 15 20 25 30
Sites
Figure 4. Species accumulation curves for pitfall and vegetative sampling data from June (solid line), July (dashed line), August (dotdashed line) and September (dotted line) at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin. Shaded areas indicate 95% confidence intervals. Sites were added in a random order without replacement.
21 Hunting Spiders Web Builders Shannon Diversity Adult Spider Richness Plant Height SD Litter Depth SD Plant Richness Year 2 Garlic Mustard Cover Vegetative Samples Hunting Spiders Web Builders Shannon Diversity Adult Spider Richness Plant Height SD Litter Depth SD Plant Richness Year 2 Garlic Mustard Cover Pitfall Traps possible significant interaction (CC ≥ ± 0.30 ) Table 3. Correlation matrices for pitfall trap and vegetative sample variables of interest. Bold values indicate a
Mustard Cover Year 2 Garlic Mustard Cover Year 2 Garlic
- 0.16 - 0.03 - 0.06 0.41 0.23 0.26 0.44 0.33 0.00 1.00 0.15 0.11 0.09 0.27 0.08 1.00
Richness Plant Richness Plant - 0.09 - 0.04 - 0.04 0.17 0.13 0.16 0.26 0.07 1.00 0.29 0.03 0.03 0.35 1.00
SD Depth Litter SD Depth Litter - 0.06 - 0.27 - 0.04 - 0.17 0.05 0.02 0.09 1.00 0.05 0.05 0.05 1.00
SD Height Plant SD Height Plant - 0.03 - 0.10 - 0.09 - 0.09 0.10 0.42 0.31 0.54 1.00 1.00
Richness Adult Spider Richness Adult Spider 0.20 0.86 0.72 1.00 0.34 0.96 0.99 1.00
Diversity Shannon Diversity Shannon 0.06 0.76 1.00 0.25 0.97 1.00
Builders Web Builders Web - 0.26 1.00 0.10 1.00
Spiders Hunting Spiders Hunting 1.00 1.00
22
Pitfall Trap Data Analysis
At the trap level, adult spider counts were so low that I could not make any meaningful statistical analyses. To remedy this, I pooled the pitfall traps in each plot together. To ensure equal sampling effort between plots, I removed any plots with trap- level missing data from the analyses for any given month, leaving 19 plots in June, 25 plots in July, 24 plots in August, and 28 plots in September.
Pitfall traps were overwhelmingly dominated by the opportunistic native linyphiid
Diplostyla concolor, with 83.9% of the total adult specimens (Table 4). The remaining most abundant species were Trochosa terricola, Phrurotimpus borealis, Bathyphantes pallidus, and Centromerus sylvaticus, each within 1.6 - 4.3% of the total (3 - 8 specimens).
23
Table 4. Ranked abundance of spider species (adult specimens only) sampled from pitfall traps at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin. All samples pooled. Samples from plots with missing pitfall trap data were not analyzed.
Family Species Abundance Proportion (%)
Linyphiidae Diplostyla concolor 156 83.9 Lycosidae Trochosa terricola 8 4.3 Phrurolithidae Phrurotimpus borealis 5 2.7 Linyphiidae Bathyphantes pallidus 3 1.6 Linyphiidae Centromerus sylvaticus 3 1.6 Lycosidae Pirata insularis 3 1.6 Thomisidae Xysticus ferox 2 1.1 Linyphiidae Ceraticelus fissiceps 1 0.5 Linyphiidae Microneta viaria 1 0.5 Linyphiidae Neriene clathrata 1 0.5 Salticidae Neon nelli 1 0.5 Lycosidae Schizochosa ocreata 1 0.5 Tetragnathidae Tetragnatha versicolor 1 0.5
Total: 6 Families 13 Species 186 99.8
When all four sampling periods were combined and analyzed together for pitfall data, the Shannon diversity of adult spiders was not correlated with garlic mustard cover
(Figure 5). However, when analyzed separately, Shannon diversity from spiders sampled in July may indicate a possible correlation. A linear regression analysis confirmed this possibility (p = 0.039). There was also no significant relationship between garlic mustard cover and spider species richness (p = 0.842). The abundance of D. concolor was not significantly correlated with garlic mustard cover (p = 0.9464).
24 2
●
● ● 1.5 ● ● ●
● yJuly = 0.01x + 0.58 ●
1 ● ●
● ● Shannon Diversity 0.5
●
0 ● ● ●
0 20 40 60 80 100
A. petiolata Percent Cover
Figure 5. Relationship between garlic mustard cover and the Shannon diversity of adult spiders sampled from pitfall traps placed in June (circles and solid line), July (squares and dashed line), August (triangles and dotdashed line) and September (crosses and dotted line). Points indicate individual plot values.
Vegetative Sample Data Analysis
The most abundant spider species sampled from the plot vegetation was the small native linyphiid Ceraticelus fissiceps with 27% of the total adult specimens (Table 5).
The remaining four most abundant species were Pelegrina proterva, Emblyna sublata,
Enoplognatha ovata, and Pityohyphantes costatus, each with 5.0 - 23.8% of the total (10
- 48 specimens).
25
Table 5. Ranked abundance of spider species (adult specimens only) sampled from vegetation at Bay Beach Wildlife Sanctuary in Brown County, Wisconsin.
Proportion Family Species Abundance (%)
Linyphiidae Ceraticelus fissiceps 55 27.2 Salticidae Pellegrina proterva 48 23.8 Dictynidae Emblyna sublata 16 7.9 Theridiidae Enoplognatha ovata 11 5.4 Linyphiidae Pityohyphantes costatus 10 5.0 Tetragnathidae Tetragnatha versicolor 9 4.5 Linyphiidae Hypselistes florens 5 2.5 Salticidae Naphrys pulex 5 2.5 Linyphiidae Neriene clathrata 3 1.5 Uloboridae Uloborus glomosus 3 1.5 Linyphiidae Agyneta fabra 2 1.0 Theridiidae Dipoena nigra 2 1.0 Linyphiidae Glyphesis scopulifer 2 1.0 Uloboridae Hyptiotes cavatus 2 1.0 Salticidae Hentzia mitrata 2 1.0 Araneidae Larinioides patagiatus 2 1.0 Philodromidae Philodromus rufus vibrans 2 1.0 Philodromidae Philodromus vulgaris 2 1.0 Theridiosomatidae Theridiosoma gemmosum 2 1.0 Theridiidae Yunohamella lyrica 2 1.0 Tetragnathidae Tetragnatha guatemalensis 2 1.0 Theridiidae Thymoites unimaculatus 2 1.0 Araneidae Araniella displicata 1 0.5 Clubionidae Clubiona opeongo 1 0.5 Linyphiidae Centromerus sylvaticus 1 0.5 Clubionidae Clubiona abbotii 1 0.5 Linyphiidae Diplostyla concolor 1 0.5 Linyphiidae Gonatium crassipalpum 1 0.5 Linyphiidae Collinsia plumosa 1 0.5 Araneidae Mangora placida 1 0.5 Tetragnathidae Tetragnatha dearmata 1 0.5 Lycosidae Trochosa terricola 1 0.5 Theridiidae Theridion murarium 1 0.5 Tetragnathidae Tetragnatha straminea 1 0.5 Theridiidae Theridula emertoni 1 0.5
Total: 11 Families 35 Species 202 100.3
26
Plotting vegetative sample Shannon diversity against garlic mustard cover suggested that samples from both June and September were positively correlated with garlic mustard cover (Figure 6). I modeled this relationship using a linear mixed-effect regression, which revealed this correlation to be statistically significant (p = 0.018).
Individually, simple linear regressions showed a significant positive relationship between
June Shannon diversity and garlic mustard cover (p = 0.030), and a borderline non- significant relationship between September Shannon diversity and garlic mustard cover
(p = 0.085). 1.0
●
0.8 ● ● ● ● ● ● ● ● yJune = 0.004x + 0.376
0.6 ● ● ●● ● ● ● ● ● ● ● ● 0.4
● ● Shannon Diversity ySep = 0.002x + 0.069 0.2 ●
● ● 0.0
0 20 40 60 80 100
A. petiolata Percent Cover
Figure 6. Relationship between garlic mustard cover and the Shannon diversity of adult spiders sampled from vegetation in June (circles and solid line) and September (crosses and dotted line). Points indicate individual plot values.
There was also a possible positive correlation between adult spider species richness and garlic mustard cover, although this relationship may not have been linear
(Figure 7). Comparing model fitness confirmed that these data fit a negative-binomial
27 distribution. A multiple negative-binomial mixed-effect regression with sampling date and the standard deviation of plant height as fixed effects showed a statistically significant correlation between spider species richness and garlic mustard cover (p <
0.001). Individually, simple linear regressions showed a significant positive correlation between June spider species richness and garlic mustard cover (p < 0.001) as well as between September spider species richness and garlic mustard cover (p = 0.032).
8 ●
6 ● ● ●
yJune = 0.04x + 2.21 ● ● ● ● ● ●
4 ●
●● ● ● ● ● ●
2 ● ● ● ySep = 0.01x + 0.75
Adult Spider Species Richness ● ●
0 ●
0 20 40 60 80 100
A. petiolata Percent Cover
Figure 7. Relationship between garlic mustard cover and adult spider species richness sampled from vegetation in June (circles and solid line) and September (crosses and dotted line). Points indicate individual plot values.
In some cases, plotting the relationship between vegetative sample Shannon diversity and garlic mustard cover resulted in a plot that appeared to have highly unequal variance. To explore this phenomenon further, I analyzed the Shannon diversity of spiders sampled from June vegetation against stratified garlic mustard cover categories
28
(Figure 8). June samples were chosen over September samples because there was not a large enough sample yield in September to meaningfully analyze those data. An ANOVA revealed that, when square-transformed, Shannon diversity significantly differed between at least two strata (p = 0.011). A Tukey's comparison of means identified this difference as being between "low" (22.6 - 45.0%) and "very high" (67.6 - 90.1%) garlic mustard percent cover categories (p = 0.017).
I also used the stratified garlic mustard cover categories to assess their relationship with the ratio of web-building and wandering spiders in the June vegetative samples and the July pitfall samples. The ratio of web-building to wandering spiders in the vegetative samples was 36/9 in "very low" cover plots, 44/4 in "low" cover plots, 17/7 in "high" cover plots, and 25/5 in "very high" cover plots. A Fisher's Exact Test for count data revealed that no significant changes in the ratio between web-building and wandering vegetative spiders occur between differing garlic mustard cover levels (p =
0.135). The ratio of web-building to wandering spiders in the pitfall samples was 65/11 in
"very low" cover plots, 11/0 in "low" cover plots, 11/1 in "high" cover plots, and 5/1 in
"very high" cover plots. A second Fisher's Exact also revealed that no significant changes in the ratio between web-building and wandering pitfall spiders occur between differing garlic mustard cover levels (p = 0.675).
29
0.8
0.6
0.4
0.2 Squared Shannon Diversity
0 Low High Very Low Very Very High Very A. petiolata Percent Cover
Figure 8. Square-transformed Shannon diversity of spider species sampled from June vegetation according to stratified categories of garlic mustard percent cover. Very Low = 0 - 22.5%. Low = 22.6 - 45.0%. High = 45.1 - 67.5%. Very High = 67.6 - 90.1%.
In June, the number of C. fissiceps, the most abundant spider species in the vegetative samples, showed a possible small positive non-linear correlation with garlic mustard cover (Figure 9). Comparing several models revealed that this relationship best fit a negative-binomial distribution, which can be expected from overdispersed count data. A negative-binomial regression model confirmed a significant but small positive correlation with garlic mustard cover (p = 0.003). There was an outlier present in the data where ten C. fissiceps were found in one plot. I conducted a sensitivity anaylsys by removing this plot point and reanalyzing the resulting dataset. The modified data were
30 better modeled by the negative binomial distribution (AIC = 88.6 < 101.8), but the significance of the relationship between garlic mustard cover and D. concolor count did not significantly change (p = 0.003). No relationship could be seen between garlic mustard cover and June P. proterva count. As a result, less abundant spider species are unlikely to show statistical significance themselves and thus were not analyzed.
● 10 8 n t
6 ● s C o u p
e ● c i s
4 ● s i f
. ● C
2 ● ● ● ● ● ●
●● ● ● ●
0 ● ● ● ● ●
0 20 40 60 80 100
A. petiolata Percent Cover
Figure 9. Relationship between number of Ceraticelus fissiceps sampled in June vegetative samples and garlic mustard percent cover. Points indicate individual plot values.
Plant Height as a Fixed Effect
To determine whether a significant degree of change in spider diversity and richness was explained by the effect of garlic mustard presence on the forest understory structural diversity, I re-ran each of my most integrative models with the standard
31 deviation of plant height as a predictor variable in simple regressions and as a fixed effect in mixed effect regressions. The linear regression model of July pitfall trap sample
Shannon diversity did not show any relationship between Shannon diversity and plant height standard deviation (p = 0.735), and the correlation between Shannon diversity and garlic mustard cover became borderline non-significant (p = 0.058). The linear mixed- effect regression model of vegetative sample Shannon diversity showed a significant relationship between Shannon diversity and plant height standard deviation (p = 0.022), while the correlation between Shannon diversity and garlic mustard cover became borderline non-significant (p = 0.076). The negative binomial mixed-effect regression of vegetative sample spider richness showed a significant relationship between Shannon diversity and plant height standard deviation (p = 0.009), while the correlation between
Shannon diversity and garlic mustard cover increased in significance (p = 0.003). The negative binomial regression of the June C. fissiceps data showed no relationship between the standard deviation of plant height and the number of C. fissiceps (p = 0.781), while garlic mustard cover remained significantly correlated with C. fissiceps count (p =
0.006).
32
Discussion
The Shannon diversity of spiders sampled from pitfall samples in July, as well as those sampled manually from vegetation, displayed linear positive correlations with garlic mustard cover, but only when the standard deviation of plant height was not included in the model. This leaves my first hypothesis, that spider species diversity would decrease with increasing garlic mustard dominance in forest understory habitats, inconclusive. However, the species richness of spiders sampled manually from vegetation displayed a positive negative binomial trend against garlic mustard cover even when the standard deviation of plant height is factored in. Not only do these results contradict my hypothesis that spider richness would decrease with increasing garlic mustard dominance, but it also contradicts much of the literature which states that invasive plant species negatively impact biodiversity, at least for this guild (Litt et al., 2014; Vitousek et al.
1996).
I believe that the poor overall habitat quality of my site may help explain these results. The species accumulation curves indicated that these sites had very low spider diversity, especially for spiders sampled from pitfall traps, which yielded only 13 species in 6 families. Extrapolating the species accumulation curves would not yield a much higher species number. A study on optimal pitfall trap size for various litter-dwelling arthropods estimated the maximum richness of spiders caught in 6.5 cm diameter traps to be over 20 (Work et al. 2002). Studies from deciduous forests in South Carolina yielded spiders representing 13 families in the same number of trap-days as this study (Vickers and Culin 2014).
33
The Bay Beach sites themselves were dominated by many other invasive species besides garlic mustard. Non-native or otherwise invasive plant species accounted for
37.5% of all species identified from the research site (Table 1). This percentage is very high compared to other cases in the literature. Out of 138 plant species surveyed from 50 sites within Maryland deciduous forests undergoing restoration, around 20% were non- native (Parker et al. 2010). The percent of non-native understory plant species in
Californian redwood forests was at most around 23% (Loya and Jules 2008). In addition, pitfall traps were often found filled with slugs. Although I did not identify the slugs that I caught, previous studies at Bay Beach Wildlife Sanctuary mentioned that the invasive slug Deroceras reticulatum was responsible for damaging many understory plants, while avoiding garlic mustard (Hahn et al. 2011). A very low density of tree saplings indicated heavy deer browse damage, which may reduce spider abundance (Robertson et al. 2016).
In some of my plots, the total cover of understory plants was low. One plot in particular only had a total of 9% plant cover. Some other plots were composed of near-monotypic stands of dense, low-lying plant species such as first-year garlic mustard or creeping
Charlie (Glechoma hederacea). The combined factors may result in a situation where, left with no sufficient options for native plants, the presence of second-year garlic mustard may offer at least some relief.
Exactly how second-year garlic mustard cover, but not first-year garlic mustard cover, increased spider richness should be examined. My initial idea was that garlic mustard alters the structural diversity of the habitat by displacing native vegetation.
However, the effect of plant height standard deviation did not appear to explain spider species richness, while second-year garlic mustard cover did. Additionally, plant species
34 diversity did not correlate with spider species richness or Shannon diversity in either sampling method. It is possible that the standard deviation of plant height is not the best way to assess the structural diversity. It is also possible that the second-year form of the plant itself might be of a complex-enough structure that it could provide web-building substrates, habitats, or ambush points for a wide variety of spider species, which we are not detecting from measuring plant height alone.
An alternative to taking plant height measurements may be found using fractal geometry. Studies have already attempted to apply fractal geometry to plant species growth. The fractal dimensions of three species of brown algae become more diverse as they develop (Corbit and Garbary, 1995). While not perfect, a multiscale approach to calculating the fractal dimensions of leaves showed potential for differentiating between tree species (Bruno et al. 2008). The fractal nature of woody plants can predict the body size of arthropods collected from said plants (Morse et al. 1985). The application of this method could be used to more accurately represent plant structural diversity, even in sensitive habitats, as all that are needed to calculate fractal dimensions are photographs.
The idea of using two-dimensional photographs to measure the structural complexity of a three-dimensional object may sound counterintuitive. Warfe and colleagues (2008) rectified this by taking multiple photographs of the same plant with narrow depths of field. Each photo was focused on parts of the plant at different distances from the camera.
Use of such techniques will guarantee that the computed fractal dimension is a true representation of the three-dimensional plant.
I had hypothesized that as garlic mustard cover increased, the ratio of web- building to wandering spiders would change in favor of web-builders. However, my data
35 revealed no significant correlation between the ratio of web-building to wandering spider species and second-year garlic mustard cover. This could indicate that in low-diversity habitats, hunting strategies are not as crucial drivers of survival as other factors when examining the spider community as a whole.
There is a positive negative-binomial correlation between C. fissiceps sampled from vegetation in June and second-year garlic mustard cover, regardless of whether plant height standard deviation is included in the model. The fact that C. fissiceps is a native linyphiid would indicate that the presence of garlic mustard is not necessarily detrimental to all native spider species, again contrary to my hypothesis that garlic mustard dominance would most benefit non-native spider species. However, the degree of increase in C. fissiceps count is not large enough to reliably use as an indicator for garlic mustard invasion alone. In addition, the model itself may not be precise enough to yield any predictions. In the scatterplot of second-year garlic mustard cover versus C. fissceps count (Figure 9), the number of C. fissiceps had a wide range of values in plots with mid-range cover of garlic mustard, while the counts were much closer together in plots with either high or low garlic mustard cover. Still, this does not necessarily invalidate the potential for C. fissiceps to be used as an indicator for overall habitat quality.
Surprisingly, a similar correlation with garlic mustard cover did not exist with P. proterva, despite being only 4% less abundant than C. fissiceps. The cause of this variation requires further study, but it might be found by examining their life histories.
Both C. fissiceps and P. proterva are ballooning native generalist predators, but they have different hunting strategies. Linyphiids such as C. fissiceps capture prey in sheet webs,
36 while salticids such as P. proterva stalk and pounce at prey. It has been shown that linyphiids seem to prefer placing webs on garlic mustard compared to native vegetation
(Smith and Schmitz 2015). Perhaps this is why C. fissiceps appears to benefit from increased garlic mustard cover while P. proterva does not.
The dominance of D. concolor in my pitfall trap samples deserves special mention. While D. concolor presence was not significantly correlated with second-year garlic mustard percent cover, it was still almost ubiquitous within the site. Other studies have described D. concolor as a species tolerant of a wide range of habitats, although it appears to thrive after disturbances such as plant invasions or forest fires (Koponen 2005;
Steffen and Draney 2009). The heavy presence of non-native species in Bay Beach
Wildlife Sanctuary may be a reason why D. concolor is thriving so well.
While flooding was responsible for a significant loss of data, it may also provide insight into the nature of the spider community at the site. My plots were located in forest habitats dominated by boxelder, eastern cottonwood, and to an extent green ash, all of which are often found in riparian forests in addition to mesic forests (Bendix and Hupp
2000). Bay Beach Wildlife Sanctuary contains wetland habitats in close proximity with the forest habitats. It is possible that some of my plots may have been placed within the riparian zones. In addition, in riparian zones with increased flooding disturbance, the spider community shifts toward species capable of ballooning (Lambeets et al. 2008). As such, these frequently flooding areas could serve as sources for ballooning spider species to spread throughout the site into the drier mesic forest habitats. The dominance of ballooning spider species within my plots was apparent, as 76.9% of pitfall trap spiders and 68.6% of vegetative spiders were either confirmed to exhibit ballooning behavior or
37 were within a genus containing known ballooning species (Bell et al. 2005). Of the three dominant species in my plots, D. concolor and P. proterva are known to exhibit ballooning behavior. It is also possible that C. fissiceps is a ballooning species, as six other species of Ceraticelus have been documented to exhibit ballooning behavior. A wind chamber experiment such as the one described by Weyman (1995) could confirm whether C. fissiceps exhibits ballooning behavior. Individuals could be collected from the field and tested immediately without needing to be reared in the lab, as small linyphiids such as C. fissiceps are often able to balloon as adults (Greenstone 1982).
38
Policy Implications
The possible benefit of second-year garlic mustard to spider species richness brings up a paradigm shift currently happening in restoration ecology. For a long time, restoration has has aimed to return disturbed habitats to as close to their original states as possible, which is often difficult to impossible given the extent of their changes, the loss of historically native species, and permanent shifts in ecosystem function equilibriums
(Hobbs et al. 2009). Now, land managers are beginning to accept that exotic species may not always be harmful to an ecosystem, and in some ways may be beneficial if a habitat is not likely to be restored to its original state (D'Antonio and Meyerson 2002). That said, the results of this thesis should not be taken as sufficient evidence that garlic mustard invasion is always good thing for low-diversity environments. The negative impact of garlic mustard presence on biodiversity is well documented (Anderson et al. 1996;
McCarthy 1997; White et al. 1993), and the drawbacks of garlic mustard presence far outweigh the benefits. Land managers should instead use this study as a reminder that there are certain ecological niches that garlic mustard fills in certain low-diversity habitats. Removal of garlic mustard should therefore be accompanied by revegetation with a non-invasive species that would take over garlic mustard's ecological function.
Future studies of a similar scale should be designed to address the issues with this study.
Because the garlic mustard at my site did not grow in thick stands, assessing the plots at different scales could result in much more variable garlic mustard covers. I designed my study to examine two scales: one at the plot level and one at the pitfall trap level.
Unfortunately, the pitfall trap data was so sparse that I was forced to pool the data,
39 reducing my study to one level of scale. The size of my plots, 5 x 5 meters, was necessary to ensure a wide enough range of garlic mustard percent cover values. It may be that increasing the plot size would allow for a greater number of possible scales for study, but the research site would have to be carefully chosen to properly represent a gradient of garlic mustard cover.
Larger scales should also be considered. When evaluating individual sites, it would appear that spiders such as D. concolor, C. fissiceps, and P. proterva would not be as useful indicators of habitat quality as the presence of invasive plant species, the number of tree saplings, or the amount of leaf litter because the latter are much quicker and easier to assess on the field. However, using these spiders as indicators may be of greater relevance at the landscape scale. Logistics prevented me from examining more than one research site, and Bay Beach Wildlife Sanctuary may not be a representative sample of all low-diversity habitats. Future studies should be done to identify more forest habitats with very low species diversity. The species compositions of these forests could be compared to Bay Beach Wildlife Sanctuary to determine whether there is an overlap in these spider species. Such is already the case with D. concolor, as this species has been found in abundance in low-quality mesic forests in Chicago, Illinois (Steffen & Draney
2009). A larger study, possibly a collaborative one, could then compare multiple research sites serving as a gradient of low- to high-diversity habitats. By comparing multiple examples of habitat quality, researchers could identify the point at which spider species common in low-quality habitats are no longer dominant. The absence of these spider species would function as a litmus test to show whether forest restoration efforts are
40 working. Similar efforts have been made in riparian forest restoration projects using plants and dung beetles (Gollan et al. 2011; Moffatt and McLachlan 2004),
It may be possible that the spider species dominant in my sites may not overlap easily with other habitats, even similar ones, in the same ecosystem (Lambeck 1997). In this case, categorizing them by their life histories and using these categories as indicators themselves may be a more employable alternative to relying on individual spider species.
However, it became apparent during my literature reviews that we know little about many spider species found in North America, including D. concolor, C. fissiceps, and P. proterva. Linyphiids in particular are poorly understood, despite being a common spider family in northern temperate regions, often found in leaf litter layers (Draney and Buckle,
2005). Studying abundant spider species in greater detail may allow us to identify their ecological niches and thus determine whether these niches are being filled in any given habitat.
The overall results of my study serve as a reminder that habitat diversity responses to plant invasion are much more complicated than land managers may be led to believe. Habitats with very low ecological diversity may not follow the same trends that more diverse habitats exhibit. It is tempting to focus one's attention on healthier habitats, possibly because they are easier to study, but restoration activities often start with low- diversity habitats, and these novel ecosystems will increase in abundance as climate, land use, and species composition changes continue (Hobbs et al. 2009). Political and financial support for habitat restoration will only happen if land managers can succeed in improving biodiversity in the long term. We will not truly know the degree of our success
41 if we cannot accurately model the network of organism interactions, both native and exotic.
42
References
Alerding, A. B. & R. M. Hunter. 2013. Increased Springtail Abundance in a Garlic Mustard-Invaded Forest. Northeastern Naturalist 20(2):275-288.
Anderson, R. C., S. S. Dhillion, & T.M. Kelley. 1996. Aspects of the Ecology of an Invasive Plant, Garlic Mustard (Alliaria petiolata), in Central Illinois. Restoration Ecology 4(2):181-191.
Bell, J. R., D. A. Bohan, E. M. Shaw, & G. S. Weyman. 2005. Ballooning dispersal using silk, world fauna, phylogenies, genetics and models. Bulletin of Entomological Research 95(2): 69-114.
Bendix, J. & C. R. Hupp. 2000. Hydrological and geomorphological impacts on riparian plant communities. Hydrological Processes 14(16-17):2977-2990.
Blossey, B., V. Nuzzo, H. L. Hinz, & E. Gerber. 2001. Developing biological control of Alliaria petiolata (M. Bieb.) Cavara and Grande (garlic mustard). Natural Areas Journal 21(4):357-367.
Bruno, O. M., R. de Oliveira Plotze, M. Falvo, & M. de Castro. 2008. Fractal dimension applied to plant identification. Information Sciences 178(12):2722-2733.
Buchholz, S. 2010. Ground spider assemblages as indicators for habitat structure in inland sand ecosystems. Biodiversity and Conservation 19(9):2565-2595.
Buchmann, S. L. & G. P. Nabhan. 1996. The Forgotten Pollinators. Island Press, Washington, DC.
Bultman, T. L. & D. J. DeWitt. 2008. Effect of an invasive ground cover plant on the abundance and diversity of a forest floor spider assemblage. Biological Invasions 10:749-756.
Carignan, V. & M. Villard. 2002. Selecting indicator species to monitor ecological integrity: a review. Environmental Monitoring and Assessment 78(1):45-61.
Corbit, J. D. & D. J. Garbary. 1995. Fractal Dimension as a Quantitative Measure of Complexity in Plant Development. Proceedings: Biological Sciences 262(1363):1-6.
Crawford, R. I., P. M. Sugg, & J. S. Edwards. 1995. Spider Arrival and Primary Establishment on Terrain Depopulated by Volcanic Eruption at Mount St. Helens, Washington. American Midland Naturalist 133(1):60-75.
D'Antonio, C & L. A. Meyerson. 2002. Exotic Plant Species as Problems and Solutions in Ecological Restoration: A Synthesis. Restoration Ecology 10(4):703-713.
43
Dávalos, A. & B. Blossey. 2004. Influence of the Invasive Herb Garlic Mustard (Alliaria petiolata) on Ground Beetle (Coleoptera: Carabidae) Assemblages. Environmental Entomology 33(3):564-576.
Davis, A. S., D. A. Landis, V. Nuzzo, B. Blossey, E. Gerber, & H. L. Hinz. 2006. Demographic models inform selection of biocontrol agents for garlic mustard (Alliaria petiolata). Ecological Applications 16(6):2399-2410.
Dondale, C. D. & J. H. Redner. 1982. The insects and arachnids of Canada, Part 9. The sac spiders of Canada and Alaska: Clubionidae and Anyphaenidae. Research Branch Aggriculture Canada Publication 1724:1-194.
Draney, M. L. & D. J. Buckle. 2005. Linyphiidae. Pp. 124-161. In Spiders of North America: an identification manual (D. Ubick, P. Paquin, P. E. Cushing, & V. Roth eds.). American Arachnological Society, Stockton, California.
Foelix. R. F. 1982. Biology of Spiders. Harvard University Press, Cambridge, London.
Gibbs, J. P. & E. J. Stanton. 2001. Habitat fragmentation and arthropod community change: carrion beetles, phoretic mites, and flies. Ecological Applications 11(1):79-85.
Gollan, J. R., C. A. M. Reid, P. B. Barnes, & L. Wilkie. 2011. The ratio of exotic-to- native dung beetles can indicate habitat quality in riparian restoration. Insect Conservation and Diversity 4(2):123-131.
Gratton, C. & R. F. Denno. 2005. Restoration of Arthropod Assemblages in a Spartina Salt Marsh following Removal of the Invasive Plant Phragmites australis. Restoration Ecology 13(3):358-372.
Greenstone, M. H. 1982. Ballooning Frequency and Habitat Predictability in Two Wolf Spider Species (Lycosidae: Pardosa). The Florida Entomologist 65(1):83-89.
Grieve, M. 1959. A modern herbal, vol 2. Hafner Publishing Co., New York.
Hahn, P. G., M. L. Draney, & M. E. Dornbush. 2011. Exotic Slugs Pose a Previously Unrecognized Threat to the Herbaceous Layer in a Midwestern Woodland. Restoration Ecology 19(6):786-794.
Haribal, M. & J. A. A. Renwick. 1998. Isovitexin 6 “-O-beta-D-glucopyranoside: A feeding deterrent to Pieris napi oleracea from Alliaria petiolata. Phytochemistry 47(6):1237-1240.
Hobbs, R. J., E. Higgs, & J. A. Harris. 2009. Novel ecosystems: implications for conservation and restoration. Trends in Ecology and Evolution 24(11):599-605.
44
Hore, U. & V. P. Uniyal. 2008. Use of Spiders (Araneae) as Indicator for Monitoring of Habitat Conditions in Tarai Conservation Area, India. Indian Forester 134(10):1371-1380.
Hutto, R.L. 1998. Using landbirds as an indicator species group. Pp. 75-92. In Avian conservation: Research and Management. (J.M. Marzluff & R. Sallabanks eds.). Island Press, Covelo.
Keeler, M. S. & F. S. Chew. 2008. Escaping an evolutionary trap: preference and performance of a native insect on an exotic invasive host. Oecologia 156(3):559- 568.
Koponen, S. 2005. Early succession of a boreal spider community after forest fire. Journal of Arachnology 33(2):230-235.
Lambeck, R. J. 1997. Focal species: A Multi-Species Umbrella for Nature Conservation. Conservation Biology 11(4):849-856.
Lambeets, K., M. J. Vandegehuchte, J. Maelfait, & Dries Bonte. 2008. Understanding the impact of flooding on trait-displacements and shifts in assemblage structure of predatory arthropods on river banks. Journal of Animal Ecology 77(6):1162-1174.
Lavelle, P. 1997. Faunal activities and soil processes: adaptive strategies that determine ecosystem function. Pp. 93-132. In Advances in Ecological Research (M. Begon & A. H. Fitter eds.). Academic Press, New York.
Lindenmayer, D. B. 1999. Future directions for biodiversity conservation in managed forests: indicator species, impact studies and monitoring programs. Forest Ecology and Management 115(2-3):277-287.
Litt, A. R., E. E. Cord, T. E. Fulbright, & G. L. Schuster. 2014. Effects of Invasive Plants on Arthropods. Conservation Biology 28(6):1532-1549.
Loya, D. T. & E. S. Jules. 2008. Use of species richness estimators to improve evaluation of understory plant response to logging: a study of redwood forests. Plant Ecology 194:179-194.
McCarthy, B. 1997. Response of a forest understory community to experimental removal of an invasive nonindigenous plant (Alliaria petiolata, Brassicaceae). Pp. 117- 130. In Assessment and management of plant invasions (J. O. Luken & J. W. Thieret eds.). Springer-Verlat. New York.
Moffatt, S. F. & S. M. McLachlan. 2004. Understory indicators of disturbance for riparian forests along an urban-rural gradient in Manitoba. Ecological Indicators 4(1):1-16.
45
Morse, D. R., J. H. Lawton, M. M. Dodson, & M. H. Williamson. 1985. Fractal dimension of vegetation and the distribution of arthropod body lengths. Nature 314:731-733.
Nuzzo, V. 2000. ELEMENT STEWARDSHIP ABSTRACT for Alliaria petiolata (Alliaria officinalis) Garlic Mustard. The Nature Conservancy, Arlington.
Parker, J. D., L. J. Richie, E. M. Lind, & K. O. Maloney. 2010. Land use history alters the relationship between native and exotic plants: the rich don't always get richer. Biological Invasions 23(6):1557-1571.
Pearson, D. E. 2009. Invasive plant architecture alters trophic interactions by changing predator abundance and behavior. Oecologica 159(3):549-558.
Pétillon, J., A. Canard, & F Ysnel. 2006. Spiders as indicators of microhabitat changes after a grass invasion in salt-marshes: synthetic results from a case study in the Mont-Saint-Michel Bay. Cahiers de Biologie Marine 47(1):11-18.
Renwick, J. A. A. & K. Lopez. 1999. Experience-based food consumption by larvae of Pieris rapae: addiction to glucosinolates? Entomologia Experimentalis et Applicata 91(1):51-58.
Renwick, J. A. A., W. Zhang, M. Haribal, A. B. Attygalle, & K. D. Lopez. 2001. Dual Chemical Barriers Protect a Plant Against Different Larval Stages of an Insect. Journal of Chemical Ecology 27(8):1575-1583.
Roberts, K. J. & R. C. Anderson. 2001. Effect of Garlic mustard [Alliaria petiolata (Bieb. Cavara & Grande)] Extracts on Plants and Arbuscular Mycorrhizal (AM) Fungi. American Midland Naturalist 146(1):146-152.
Robertson, E. J., M. J. Chips, W. P. Carson, & T. P. Rooney. 2016. Deer herbivory reduces web-building spider abundance by simplifying forest vegetation structure. PeerJ 4: e2538.
Rodgers, V. L., K. A. Stinson, & A. C. Finzi. 2008. Ready or Not, Garlic Mustard Is Moving In: Alliaria petiolata as a Member of Eastern North American Forests. Bioscience 58(5):426-436.
Shannon, C. E. 1948. A Mathematical Theory of Communication. Bell System Technical Journal 27(3):379 - 423.
Smith, L. M. & O. J. Schmitz. 2015. Invasive plants may promote predator-mediated feedback that inhibits further invasion. Ecology and Evolution 5(12):2411-2419.
46
Souza, H. C., Y. F. Messas, M. O. Gonzaga, & J. Vasconcellos-Neto. 2015. Substrate selection and spatial segregation by two congeneric species of Eustala (Araneae: Araneidae) in southeastern Brazil. Journal of Arachnology 43(1):59-66.
Steffen, J. F. & M. L. Draney. 2009. Diversity and Activity of Ground-Dwelling Spiders (Araneae) in Four Sub-Communities in a Degraded Oak Woodland at the Chicago Botanic Garden, Cook County, Illinois. Great Lakes Entomologist 43(3-4):185- 203.
Temple, S. A. & J. A. Wiens. 1989. Bird populations and environmental changes: can birds be bio-indicators? American Birds 43:260-270.
Tilman, D., R. M. May, C. L. Lehman, & M. A. Nowak. 1994. Habitat destruction and the extinction debt. Nature 371:65-66.
Vickers, M. E. & J. D. Culin. 2014. Spider (O:Araneae) Responses to Fire and Fire Surrogate Fuel Reduction in a Piedmont Forest in Upstate South Carolina. Southeastern Naturalist 13(2):396-406.
Vitousek, P. M., C. M. D’Antomion, L. L. Loope, & R. Westbrooks. 1996. Biological Invasions as Global Environmental Change. American Scientist 84(5):468-478.
Voss, E. G. & A. A. Reznicek. 2012. Field Manual of Michigan Flora. University of Michigan Press, Ann Arbor.
Warfe, D. M., L. A. Barmuta, & S. Wotherspoon. 2008. Quantifying habitat structure: surface convolution and living space for species in complex environments. Oikos 117:1764-1773.
Weyman, G. S. 1995. Laboratory Studies of the Factors Stimulating Ballooning Behavior by Linyphiid Spiders (Araneae, Linyphiidae). Journal of Arachnology 23(2): 75- 84.
White, D. J., E. Haber, & C. Keddy. 1993. Invasive plants of natural habitats in Canada. Canadian Museum of Nature. Ottawa, Canada.
Wise, D. H. 1993. Spiders in Ecological Webs. Cambridge University Press, Cambridge.
Work, T. T., C. M. Buddle, L. M. Korinus, & J. R. Spence. 2002. Pitfall Trap Size and Capture of Three Taxa of Litter-Dwelling Arthropods: Implications for Biodiversity Studies. Environmental Entomology 31(3):438-448.