The Restoration of Plant-Pollinator Mutualsims on a Reclaimed Strip-Mine

Thesis

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University

By

Sarah Cusser, B.A.

Graduate Programs in Evolution Ecology and Organismal Biology

The Ohio State University

2012

Thesis Committee:

Karen Goodell, Advisor

Allison Snow

Elizabeth Marschell

Mary Gardiner

Copyright by

Sarah Cusser

2012

2

Abstract

Plant-pollinator mutualisms are one of several functional relationships that must be reinstated to ensure the long-term success of habitat restoration projects. However, these mutualisms are not likely to reinstate themselves until the very particular resource requirements of pollinators have been met. By giving special attention to these requirements, habitat restoration projects are more likely to be successful in the long- term. I used network analysis to determine how aspects of the restoration effort itself influence the reestablishment and organization of plant-pollinator communities at an experimentally restored site in Central Ohio. Specifically, I investigate the influence of landscapes factors, floral diversity and the role of non-native plants on the structure and stability of plant-pollinator networks. I found that the diversity and distribution of floral resources affect the organization and stability of plant pollinator networks. Plots with high floral diversity far from remnant habitat compensated for loses in pollinator diversity by attracting generalized pollinators, which increase network redundancy and robustness. I also found that non-native plants play a central role in the structure of networks. I conclude that aspects of the restoration effort itself can be successfully tailored to incorporate the restoration of pollinators.

ii 3

Dedication

Dedicated to my beautiful bunnies Moose and Edie

ii4i

Acknowledgments

Special thanks to Ryan Martyn, Reem Najjar, and Carol Brown for help in the field. To

Chia Hau Lin, Amy McKinney, and Jessie Lanterman, as well as Joseph Pipia for technical advice and support. Thanks to Jenise Bauman for help coding in R and to Nicole

Cavender, Shana Byrd, The Wilds staff for logistical help.

iv5

Vita

2002…………………………………………………….…………….……..Pembroke Hill School

2006…………………………………………………….……………….…..B.A. Biology, Pomona College

2009- present……………………………………………………….……Graduate Teaching and Research

Associate, Department of

Evolution Ecology and Organismal

Biology, The Ohio State University

Fields of Study

Major Field: Evolution Ecology and Organismal Biology

v6

Table of Contents

Abstract……………………………………………………………………………………..…iii

Dedication………………………………………………………………………………….…iv

Acknowledgements………..………………………………………………….…………v

Vita……………………………………………………………………………………….………vi

List of Tables…………………………………………………………………………….…..vii

List of Figures…………….…………………………………………………….……………viii

Chapter 1: The influence of floral distribution and richness on the restoration of robust

plant-pollinator networks on a reclaimed strip mine. ……..…1

Chapter 2: The role of non-native plants in the restoration of plant-pollinator

mutualisms in a constructed prairie. …………………………………29

References………………………………………………………………………………..…..49

Appendix A: Figures and Tables………………………………………………….…56

vi7

List of Tables

Table 1.Comparison of Pollinator Communities……………………………..57

Table 2.Pollinator Generalization…………………………………………………..61

Table 3.Comparison Network Structure and Stability………………..…..62

Table 4. Floral Community Constituents……………………………………..…64-65

Table 5. Comparison of Centrality Indices…………………………………..…67

Table 6.Seed Mixtures………………………………………………………………..…70

Table 7. Centrality Indices……………………………………………………………..71-72

Table 8. Centrality Indices Continued…………………………………………….72-73

vii8

List of Figures

Figure 1. Map of the Wilds……………………………………………………………56

Figure 2. Comparison of Pollinator Community …………………………….58

Figure 3. NMDS of High Floral Diversity Plots………………………………...59

Figure 4.NMDS of Low Floral Diversity Plots………………………………….60

Figure 5. Comparison of Network Structure and Stability……………...63

Figure 6. Schematic of Centrality Indices.…………………………………..…65

Figure 7.Comparison of Floral Community…………………………...…..…..66

Figure 8.Comparison of Centrality Indices……………………………………..68

Figure 9. Network Diagram……………………………………………………………69

vi9ii

Chapter 1: The influence of floral distribution and richness on the restoration of

robust plant-pollinator networks on a reclaimed strip mine.

Introduction

The primary goal of restoration is to return degraded ecosystems to pre- disturbance composition, structure, and function (Jordan et al. 1999). In principal, the composition and structure of a plant community can be restored relatively simply with the addition of native plants and the removal or control of invasive ones (Hobbs and

Norton 1996, Ried 2009). However, how to restore functional relationships between plants and the other organisms they interact with in a community is less well established. Pollination is one of several functional relationships that must be reinstated for ecological restoration to be successful (Dixon 2009). Animal pollinators play a fundamental role in the maintenance of plant communities and contribute to the reproduction of somewhere between 70% and 90% of flowering plant species

(Buchmann and Nabhan 1996, Kearns et al. 1998). Pollination mutualisms are not likely to reinstate themselves in degraded plant communities until the floral, nesting, and over-wintering resource requirements of pollinators have been met (Exeler et al. 2009,

Roulston and Goodell 2011). By giving special attention to these habitat requirements, restoration is more likely to succeed in reinstating pollination services in plant communities (Forup and Memmott 2005, Forup et al. 2008, Menz et al. 2010). 110

When considering how to restore pollination service in plant communities, special care should be directed towards promoting pollinator richness (Murray et al.

2009, Potts et al. 2009). Pollinator richness has been shown to promote the reproductive output (Gomez et al. 2007, Hoehn et al. 2008), genetic resilience (Bowles et al. 1994, Linhart 1995, Wilcock and Neiland 2002), and community stability (Steffan-

Dewenter et al. 2005, Steffan-Dewenter and Westphal 2008) of a variety of ecosystems.

Closely linked to pollinator richness is the distribution and richness of floral resources

(Steffan-Dewenter et al. 2001, Potts et al. 2003, Steffan-Dewenter 2003, Kremen et al.

2007). Potts et al. (2003) found that aspects of the floral community directly shape the composition of the pollinator communities they support. Consequently, to restore pollinator richness, ecologists must consider the distribution and richness of floral resources in a community.

As pollinator communities are taxonomically and ecologically diverse, the distribution of floral resources can affect subsets of the community differently. Bees, an important group of pollinators, are central-place foragers and require suitable floral resources within the flight range of their nesting habitat (Michener 2000). Flight range varies with species and is correlated with body size (Greenleaf et al. 2007). In the case of some of the smallest bees, foraging is limited to within a couple of hundred meters of the nesting habitat (Zurbuchen et al. 2011). Thus, even relatively short distances can serve as a filter, eliminating potential pollinators from a community (Klein 2009, Vergara and Badano 2009, Carvalheiro et al. 2010). Situations where floral and nesting resources 11 do not overlap in space are considered “partial habitats” (Westrich 1996). Partial habitats contain only some of the needed resources of bees and function either as a nesting area or a foraging area, but not both. As a result, these habitats are not likely to support diverse bee communities (Matheson et al. 1996). Lepidopteran and Dipteran pollinator species on the other hand are not central place foragers, but rather they require specific ovipositing and larval substrates to be distributed throughout the habitat (Scott 1986, Schweiger et al. 2007). Most butterflies are specialised in there requirements, each species having particular needs in respect to temperature, humidity, larval food plants and adult food sources. Thus, butterfly and hoverfly communities respond to the distribution of floral resources very differently than bee communities

(Steffan-Dewenter and Tscharntke 2000, Bergman et al. 2004, Krauss et al. 2005,

Biesmeijer and Schaffers 2006, Ghazoul 2006, Schweiger et al. 2007, Jauker et al. 2009).

Along with the distribution of floral resources in a community, the richness of resources can also directly influence pollinator richness. For example, Steffan-Dewenter et al. (1999) found a positive relationship between floral richness and abundance and pollinator richness. Similarly, Hegland et al. (2006) found that floral composition, in part, determines the richness and abundance of pollinators and the carrying capacity of the community.

Ecologists are increasingly interested in understanding plant-pollinator interactions in a community context (Memmott 1999). Using network analysis allows

12 ecologists to transcend the narrow habitat and taxonomic boundaries of species-specific approaches, and allows for the development of overarching descriptions of diverse pollinator assemblages. The approach provides powerful quantitative tools to explore the influence of ecological factors on pollinator community structure and function

(Dunne et al. 2002a, Bascompte 2007). In restoration, network statistics provide the

“meter stick” to which the success of a restoration project can be compared. By using plant-pollinator network structure as a description of the functional target community that restoration efforts hope to reinstate (Henson et al. 2009, Williams 2010), ecologists can compare the structure of networks from restored sites to those of ancient intact sites, and form accurate conclusions about the relative success of projects (Forup and

Memmott 2005, Forup et al. 2008, Kaiser-Bunbury et al. 2010, Williams 2010). Network analysis is also useful to restoration in its ability to model the initial assembly of plant- pollinator networks (Guimaraes et al. 2007, Bascompte and Stouffer 2009) and to model the stability and resilience of completed restoration projects in the face of potential future disturbance (Kaiser-Bunbury et al. , Memmott et al. 2004, Memmott et al. 2006).

In this study, we evaluated the influence of floral distribution and richness on the composition of pollinator communities and the structure and stability of plant-pollinator networks in constructed patches of native prairie. Located within a grassland landscape that was formerly surface mined for coal in central Ohio, our experimental patches varied in distance from a remnant hardwood forest and in the richness of plants seeded.

Our objectives were to first identify how pollinator communities changed with distance 13 from the remnant habitat and then to determine how these changes affected plant- pollinator network structure and stability. Further, we wanted to investigate how patches of high floral richness compared to those of low floral richness in terms of how they experienced increased distance from the remnant habitat. To answer these questions, we described the composition of the pollinator community, as well as the structure and stability of the plant-pollinator network for each constructed patch. We hypothesized that if the remnant forest is acting as either a source population or nesting habitat for pollinators, then increased distance from the forest would negatively affect pollinator richness. In turn, losses of pollinator richness would negatively affect plant- pollinator network structure and stability. We further hypothesized that patches with high floral richness, would attract a high richness of pollinators even in distant plots.

Thus high floral richness plots would respond to the effects of increased distance less severely than patches with low floral richness as seen in network structure and stability.

Methods

Study System

This study was conducted at The Wilds in Muskingum County, Ohio. The site was surface mined for coal until the mid 1980s. Soon after it was recontoured and seeded with a low diversity of non-native grasses and forbes as mandated by the 1977 Mining and Reclamation Act (Day et al. 1978). Along one edge of the site remains a remnant hardwood forest (Figure 1), left undisturbed by the strip mining. In 2009, 48 experimental restoration plots were established. Plots were 10 m radius and separated 14 by 70-100 m within a 1 km2 area of the grassland. 16 of the plots were seeded with a low diversity of prairie plant species (Appendix 1). The remaining 32 plots were seeded with the same low richness mixture but also supplemented with additional seeds of a high diversity mixture (Appendix 1). Since seeding, the plots have been invaded by native adventives and non-native plant species common to the Midwest (Appendix 1).

Prairie plants were chosen for the restoration effort as they were likely to establish and succeed despite the low soil organic matter and low nutrients that are characteristic of highly disturbed reclaimed mine sites (Rodrigue and Burger 2004). As network analysis requires a large amount of data, we chose to focus our sampling effort on a subset of the 48 plots: eight low floral richness and 16 high floral richness plots (Figure 1).

Distance from the center of each plot to the nearest edge of remnant forest was measured with maps made with a Trimble Geo-XT GPS with post-processing in ArcGIS.

Plots within 250 m of the remnant habitat were classified as “near” to the forest (N=12).

Plots farther than 300 m from the remnant habitat were classified as “far” from the remnant forest (N=12). Near plots and far plots were separated by a small lake (Figure

1).

Floral Sampling

To quantify the floral community, flower surveys were carried out every other week from late June until mid August 2010. Thus, each plot was surveyed four times over the course of the season. Surveys were timed seasonally to encompass main pollinator flight activity. During each survey we counted the number of floral units in 12 randomly

15 placed one-meter square quadrats. Floral units were defined from the pollinator’s perspective rather than by flower anatomy (Appendix 1). Thus one floral unit was separated from another by the distance that a small pollinator would have to fly rather than walk (Saville 1993). We refer to floral abundance as the total number of floral units in 12 one-meter square quadrats and floral richness as the total species diveristy of open flowers in the entire 10-meter radius plot.

Pollinator Sampling

To quantify the pollinator community we used the same randomly placed 12 one- meter square quadrats used for floral surveys. Pollinator surveys were performed on warm, dry days (22-38 degrees C) with moderate wind speed (0.1-3.4 m/s), between

8:00 and 16:00 h. We observed all open flowers in each one-meter square quadrat for

10 minutes. Insects were identified as pollinators if they were observed actively probing floral reproductive parts. Insects observed perched on flowers but not probing floral reproductive parts were not considered pollinators. Pollinators observed foraging within the quadrat were collected with nets and put into individual clean waxed paper envelopes. Pollinators were then placed into killing vials that contained fume of ethyl acetate. A clean envelope was used for every individual to minimize the contamination of pollen load between specimens. To ensure that we collected insect specimens foraging on less abundant floral species, we performed a 40-minute sampling of the entire plot in addition to the 12 randomly placed quadrats. In the 40-minute sample we collected all insects seen foraging anywhere within the 10 m radius plot. Thus each plot

16 was sampled for a total of 160 minutes/survey. Plots were surveyed every other week, four times over the course of the season.

Processing Specimens

To extend the record of pollinator visitation to include flowers visited in the recent past, we analyzed pollen found on all collected insect specimens. Previous studies have established that the inclusion of pollen load data in a network gives a more complete visitation history (Bosch et al. 2009). To determine pollen load composition, we rubbed a small cube of fuchsin-stained gelatin (~2 cubic mm) over each specimen. We specifically targeted pollen held in specialized pollen-carrying structures such as scopa, as our intention was to estimate an inclusive visitation history, rather than identify pollen available for pollination, as in other studies (Bosch 2009, Hinners 2009). The gelatin was then mounted onto glass slides and examined with a compound microscope.

To avoid contamination, laboratory utensils and work surfaces were cleaned after each specimen. Pollen grains were identified to the lowest possible taxonomic level with the aid of a reference collection from the study site. In most cases pollen was identified to species. However, in the cases of Trifolium and Melilotus spp. we recorded pollen to the genus level because those species could not be clearly distinguished by pollen morphology. We only recorded the presence of pollen with more than 15 grains per slide. Slides with fewer than 15 grains we removed from analysis as fewer than this amount was likely the result of contamination rather than a true visitation record.

Contamination could have either resulted from a pollinator picking up heterospecific

17 pollen at a flower left by another pollinator that had previously visited another plant species (Bosch 2009) or from accidental contamination during netting or processing.

Contamination was minimal and 97.8% of slides with pollen contained more than 15 grains for each species. Pollinators were then identified to the lowest possible taxonomic level. Of the total 2086 specimens collected, 1,982 were identified to species.

The remaining 104 specimens, primarily Diptera of the families Sarcophaghidae and

Tachinidae, were identified to morphospecies; these specimens accounted for less than

5% of the total collected specimens.

Pollinator Community Analysis

To determine how pollinator communities varied with distance from the remnant habitat in plots of different floral richness, we compared pollinator abundance, species richness, and evenness in plots near and far from the remnant habitat for both low and high floral richness treatments. We tested evenness using H/Hmax, where H is the

Shannon richness index and Hmax is its maximum value (log [species richness]). Because we compared plots near to the forest with plots far from the forest for each of the richness treatments, we used two sided T tests. To determine if there were differences in pollinator community composition between near and far plots we used non-metric multidimensional scaling (NMDS). NMDS is a non-parametric, permutational technique that is similar to ANOVA but makes no distributional assumptions and can be applied to multivariate data (Anderson 2001, McArdle and Anderson 2001). Using R statistical program (Dormann 2008, Jari Oksanen 2009, Team 2009), we determined Bray-Curtis

18 distances from differences in the relative abundance of pollinators for the most abundant 11 genera. These 11 genera accounted for 87% of the total specimens. The remaining specimens were grouped into four categories based on order: uncommon

Hymenoptera, uncommon Diptera, uncommon Lepidoptera, and uncommon

Coleoptera. The term “uncommon” is used without reference to abundance of those species outside the plots, but rather refers to the frequency at which they were collected at our site. We performed ordination on the genus level rather than by species because it is not possible to execute ordination with more explanatory variables than samples. Ordination values were determined using 50 runs of 200 iterations with random starting positions, accepted stress less than 10 and instability less than 0.0003.

We used two dimensions to maintain interpretability and Adonis tests to determine if the differences we found in pollinator community composition were statistically significant. Adonis uses permutation tests with pseudo-f ratios to determine if the ordination values fall out into significantly different space by group (Anderson 2001). In our case, we grouped plots by distance from the forest (near vs. far) for both floral richness treatments (high and low). As plots were close enough for pollinators to fly between them, we checked for spatial autocorrelation. Chao-Sorrenson similarity values were calculated for pollinator assemblages in all site pairs using R and the geographical distance between all site pairs were measured using ArcGIS. The similarity distance matrix produced was tested for spatial autocorrelation using a Mantel test (999

19 permutations in R). The Pearson correlation method was used. No significant autocorrelation was found (Mantel statistic r= 0.01154, P= 0.45).

Plant-Pollinator Network Analysis

To determine how plant-pollinator network structure and stability changed with distance from the remnant habitat for plots of different floral richness, we built quantitative plant-pollinator interaction networks for each plot. Networks were assembled by combining data from visitation and pollen load networks. Thus, 24 networks were constructed in which the cell values indicate the number of times that individual pollinators of species “a” were collected foraging on flowers of plant species

“p” in addition to the number of times that pollen of plant species “p” was found on pollinator species “a”. By combining visitation and pollen load data in the same network, we avoided the dangers of under-sampling interactions and could be confident in our estimation of network specialization (Bluthgen and Menzel 2006, Bosch et al. 2009). By making individual networks for each plot we also avoided including “forbidden links” between plants and pollinators that did not overlap in space. For each network we calculated connectance, plant niche overlap and network robustness. Both network connectance and plant niche overlap describe network architecture. Network connectance, the proportion of possible links that are realized, ranges from 0 to 1 and provides a measure of network generalization adjusted for network size, with higher values indicating greater generalization (Jordano 1987, Lundgren and Olesen 2005).

Thus, in a maximally connected network, where nearly all pollinators interact with

20 nearly all plants, connectance would approach 1. Conversely, in a minimally connected network, where few pollinators interact with few plants, connectance would be close to

0. Niche overlap quantifies the extent to which resources are shared by species of the same trophic level. In a network with a high degree of niche overlap, many pollinator species visit a single plant species, or many plants share a single pollinator species.

Niche overlap can also be thought of as a measure of the redundancy in a network, quantifying how many species fill a similar functional role. We determined plant niche overlap in particular using Horn’s index in which values of 0 indicate no common use of pollinators between plants, and values near 1 indicate a high degree of shared pollinators between plants (Mueller and Altenberg 1985). To determine how network architecture affects network stability, we calculated the robustness of each network.

Rather than describing network architecture itself, robustness quantifies a network’s ability to retain its structure following disturbance, in particular, how stable a network is to the removal, or extinction, of species (Dunne et al. 2002b). We simulated extinction by removing plant species and observing which pollinators were left without forage resources. Pollinator species were considered to go “extinct” when all of their plant hosts had been removed from the network. We used the bipartite package in r to execute a removal algorithm in which plant species were removed at random without replacement. Simulations were repeated 100 times for each network. We used the technique developed by Burgos et al. (2007) to report a quantitative measure of robustness with a single parameter r, ranging from 0 to 1. A network in which r=0 would

21 be considered fragile, with even a very few plants eliminated, most pollinators would go extinct. Likewise networks with r approaching 1 would be considered robust. In a robust network most pollinators survive even if a large fraction of the plant species are eliminated. Our simulation assumes that all plant species are equally effective as forage resources, so that a pollinator must lose all of its plant hosts before it goes extinct. We also assume that pollinators remaining after plant removal do not expand their floral diet, a shift that could rescue some insect species from going extinct as their preferred plants are eliminated. To quantify how generalized pollinator species were within networks, we used the metric d’, which gives the specialization of each pollinator based on its discrimination from random selection of plants (Bluthgen and Menzel 2006). d’ accounts for the frequency of interactions, giving more weight to frequent interactions.

Data were first tested for and met assumptions of normality and equal variances. Data were compared using two-sided T tests and comparisons were deemed significant at the alpha = 0.05 level.

Results

Floral Communities

Floral richness was significantly higher in plots seeded with high richness seed mixtures (two-sided T test, df =22, T= 2.93, P= 0.007). Low richness plots averaged 14 plant species per plot (S.E.= 0.67) while high richness plots averaged 19 species per plot

(S.E.= 0.625). There was no significant effect of the seeding treatment on floral abundance (two-sided T test, df =22, T= .611, P= 0.555). Low richness plots averaged

22

998 floral units per plot (S.E.= 42.75) and high richness plots averaged 936 floral units per plot (S.E.= 61.5). We found no relationship between distance from the remnant forest and floral abundance (two-sided T test, df =22, T= 0.915, P= 0.34) or floral richness (two-sided T test, df =22, T = 0.28, P= 0.77).

Pollinator Community Analysis

We captured a total of 2086 pollinator specimens of 103 species. The specimens were of four orders: Hymenoptera (56 species), Diptera (34 species), Lepidoptera (10 species) and Coleoptera (3 species). 65% of the specimens were of 6 species: Bombus impatiens, Apis mellifera, Toxomerus marginatus, Eristalis transversa, Halictus ligatus, and Lasioglossum mitchelli. Of the remaining species, 33 were singletons representing the only specimen caught of that species and 26 were doubletons.

Looking at high floral richness plots, we found no difference in pollinator abundance or community evenness between plots near and far from the remnant habitat (Table 1,

Figure 2a,c). Pollinator richness, however, was significantly greater near the remnant forest than far from the forest (Table 1, Figure 2b). High floral richness plots far from the remnant habitat harbored significantly different pollinator communities than those close to the forest edge (Adonis, df=15, F = 3.11, P= 0.024, Figure 3). Distant plots lacked a significant proportion of uncommon Hymenoptera species, including Megachile,

Augochlora, Augochloropsis, Andrena, Heriades, and Hoplitis spp. (two-sided T test, df=14, T = 2.78, P= 0.014), and uncommon Diptera species including Allograpta, Syritta,

Tropida, and Hedriodiscus sp. (two-sided T test, df =14, T= 2.31, P= 0.03). Distant plots

23 also lacked a significant proportion of the abundant Dipteran genus Toxomerus (two- sided T test, df =14, T= 3.73, P= 0.002). As a result, common Hymenoptera species comprised proportionally more of the pollinator community of distant high floral richness plots (two-sided T test, df =14, T= 3.55, P= 0.003) including Bombus, Halictus,

Apis, Ceratina, and Xylocopa spp. than in plots of similar floral richness close to the forest edge. Floral visitor specialization (d’) in our networks ranged from very generalized (d’= 0.005) to relatively specialized (d’= 0.791). Those species that increased in relative abundance in distant, high richness plots (Bombus, Halictus, Apis, Ceratina, and Xylocopa spp.) were among the most generalized floral visitors (Table 2).

Looking at low floral richness plots, we found no change in pollinator abundance or community evenness with distance from the remnant habitat (Table 1, Figure 2a,c).

Similar to high floral richness plots, pollinator richness was significantly higher in plots near the remnant forest that in those far from the forest (Table 1, Figure 2b). Unlike high richness plots, however, low richness plots near and far from the remnant forest shared similar pollinator communities (Adonis, df= 7, F= 1.32, P= 0.2786, Figure 4).

Plant-Pollinator Network Analysis:

We observed pollinators foraging on 47 flower species and found 45 species of pollen on collected specimens. While we observed pollinators foraging on Linum perenne, Iberis umbellata, and Verbascum blattaria, we found no pollen of these flowers on any insect. Similarly, while we never observed pollinators foraging on Brassica, we found the pollen of an unknown Brassica sp. on two specimens. Otherwise, visitation

24 and pollen load networks were similar in composition, and in the 24 networks we observed 728 unique plant-pollinator interactions.

High floral richness plots far from the forest were significantly more connected and had greater plant niche overlap than plots of the same floral treatment close to the forest edge (Table 3, Figure 5a,b). Distant plots had similar levels of network robustness as in those plots near (Table 3, Figure 5c). Low richness plots near and far from the remnant habitat shared similar levels of connectance and plant niche overlap (Table 3,

Figure 5a,b), but far from the forest were significantly less robust than plots of the same floral treatment close to the forest edge (Table 3, Figure 5c).

Discussion

Studies have shown that pollinator richness increases with floral richness (Ghazoul

2006) and decreases with distance from high quality habitat (Carvalheiro et al. 2010).

What has not been demonstrated until now is how these factors interact or how they affect plant-pollinator network structure and stability. The combined effect of distance and floral richness on pollinator communities was not as straightforward as we had initially predicted. While we did find that increased distance from the remnant habitat negatively affected pollinator richness, in turn, decreasing connectance, niche overlap and robustness, we were surprised by the influence of floral richness. Our findings suggest that high floral richness in areas far from remnant habitat did promote network stability, but not by attracting a diverse set of pollinators as we had predicted. Rather, high floral richness stabilized network structure in distant plots by attracting a

25 generalized set of pollinators. Generalized pollinators work to increase network connectance and niche overlap to create more redundant and robust networks far from the remnant habitat despite decreases in pollinator richness. In this section we describe the mechanism by which floral richness promotes network stability in plots far from remnant habitat and offer suggestions as to how our results can be applied in future restoration projects.

Both low and high floral richness plots had lower pollinator richness far from the remnant forest. What differed between the two treatments however was the composition of the pollinator community in distant plots. On average, pollinator species in distant, high floral richness plots were more generalized, visiting a greater diversity of flowers than pollinators in plots of similar floral richness close to the forests edge. Thus when distant, high richness plots experienced our simulated species extinction, the increased connectance and niche overlap created by generalized pollinators, buffered those networks against further species loss, rendering them robust. Distant low floral richness plots, on the other hand, maintained a similar composition of pollinators across the study site. As a result there was no structural compensation for the loss of pollinator richness by generalized pollinators, and distant low richness networks were relatively more fragile to our simulated extinctions.

Our results support the positive relationship between connectance, generalization and robustness found in other studies (Dunne et al. 2002a, Estrada 2007, Gilbert 2009,

Gonzalez et al. 2009a). Generalized participants, working to increase network

26 connectance, have been found to be important to the overall structure and stability of networks, and play a more important role in the cohesiveness of communities than predicted by their relative abundance. In particular, Gonzalez et al. (2009) found that generalized participants helped to buffer systems from the ill effects of species loss and that networks rapidly became fragmented when generalized species were theoretically removed. Here, we further demonstrate that while pollinator richness improves network stability (Murray et al. 2009, Potts 2009), the generalization of individual pollinators can have equally positive effects on network structure. As a result, we suggest that if the goal of restoration is to promote functionally effective and stable communities, attraction of, not only a diverse set of pollinators, but also a generalized pollinator community is important. We suggest that promoting generalized pollinator communities will encourage the reinstatement of functional mutualisms, and consequently the long-term success of restoration projects.

Researchers realize that, beyond generalization and connectance, the manner in which participants are connected can also affect robustness. Estrada (2007) showed that networks with the same connectance and identical degree distributions could display different degrees of resilience to species loss. Estrada concluded that the investigation of structural organization beyond connectance in networks is important in understanding why networks vary in robustness. Here we show that both connectance and plant niche overlap are associated with increased network robustness. With expanded overlap in pollinators used by plants, our networks become structurally 27 reinforced, increasing robustness. These results agree with the findings of Bluthgen

(2010) who notes the stabilizing role of redundant species, suggesting that systems with functional redundancy may be more resilient for long periods of time following local extinctions.

Our research also supports that the location of floral resources in relation to undisturbed habitats can be very important in restoration projects. Considering the small scale of our field site, our results show that even distances as small as one kilometer can have major effects on the composition of pollinator communities and on the structure and stability of plant-pollinator networks. While restoration can promote the formation of effective pollinator communities (Forup et al. 2008) as well as communities that function similarly to undisturbed remnant communities (Williams

2010), we show that aspects of the restoration effort itself may promote or discourage the probability that these functional communities form. In projects confined by limited resources, special care should be directed towards strengthening communities far from remnant habitats. Our results suggest that the planting of highly diverse floral communities are most beneficial to pollinators far from source populations and indicate that high richness plantings can effectively promote network stability at distances relatively far (>300m) from remnant habitat.

28

Chapter 2: The role of non-native plants in the restoration of plant-pollinator

mutualisms in a constructed prairie.

Introduction

In the last twenty years, non-native invasive plants have become recognized as both a serious threat to native plant populations as well as a primary cause of habitat degradation (Wilcove et al. 1998). Land managers acknowledge that non-native invasive plants threaten the health and persistence of native plant assemblages through the transfer of diseases (Manchester and Bullock 2000), competition for limited resources

(Daehler 2003), and the alteration of ecosystem functions (Gordon 1998). As a result, the removal of non-native invasive plants has become the focus of many restoration projects.

Despite the often-documented negative effects of non-native species on native plant assemblages, we suggest that it would be overhasty to conclude on the basis of those documented effects that non-native species serve no productive functional role in the process of habitat restoration. If the goal of restoration is to assemble functioning and stable plant communities, we suggest that land managers look beyond the origin of a plant to determine its role in the community before choosing the appropriate course of action. The immediate removal of non-native plants from restoration sites may not 29 generate optimal conditions for the achievement of restoration goals, especially in cases where the complete removal or control of non-native plants is not possible or cost effective or in cases where non-native plants now constitute the majority of vegetation.

In such cases, it may be beneficial to consider how restoration may proceed with non- native species present or even whether non-native plants may provide valuable contributions during the transition to a more diverse, predominantly native plant community. Optimistically, non-native invasive plants may provide essential functions that promote the long term success of habitat restoration projects.

It has been well established that the reinstatement of pollination function is an important step towards the complete restoration of plant communities (Menz et al.

2010). As somewhere between 70% and 90% of flowering plants require pollination for successful sexual reproduction (Buchmann and Nabhan 1996, Kearns et al. 1998), pollination services underpin the success and long-term sustainability of restored ecosystems (Dixon 1999). Work investigating the influence of non-native plants on the sexual reproduction of native plants has yielded varied results. It has been suggested that non-native plants can have negative (competitive), neutral, or positive (facilitative) effects on native plant pollination (Rathcke 1983). Competition can occur when native plants suffer from pollen limitation as a result of sharing insect visitors with non-native plants (Knight et al, 2005) either from reduced pollen quantity or reduced pollen quality if the pollen transferred is contaminated with the conspecific pollen of non-native plants

(Campbell and Motten 1985). Facilitation of pollination service by non-native plants, on 30 the other hand, may take place if non-native plants increase the rate of visitation by pollinators, thereby fostering the reproductive success of native plant species (Ghazoul

2006). Because a large portion of invasive plants have been introduced as ornamental species and produce an abundance of large, brightly colored flowers, non-native plants are often very attractive to pollinators (Richardson et al. 2000, Memmott and Waser

2002, Stout and Morales 2009). Many non-native plants are pollination generalists that provide an accessible source of nectar and pollen to a broad range of visitors. Studies have shown that pollinators can quickly learn to exploit these new resources (Morales and Traveset 2009). Non-native plants may facilitate natives by attracting pollinators to native plant patches through increased floral display or floral diversity (Ghazoul 2006) or by supporting the pollinator community while floral resources are otherwise scarce.

When the nectar and pollen of native plants are either spatially or temporally not available, non-native plants may provide these important resources to pollinators

(Chittka and Schurkens 2001, Shapiro 2002). Non-native plants can extend the geographic range of pollinator populations by providing resources at the spatial fringe of existing populations (Shapiro 2002, Williams et al. 2006) or extend pollinator flight season providing resources during times when native plants are insufficient (Bjerknes et al. 2007). It has also been suggested that non-native plants may increase the carrying capacity of an ecosystem by providing abundant resources to foraging insects (Shapiro

2002). Many claims have been made about the influence of non-native plants on pollinator populations, but surprisingly little research has been done investigating the

31 effect of non-native plants on pollinators in a restoration context. While it’s clear that pollinators use non-native plant resources (Waser et al. 1996), and that non-native plants are well integrated into plant-pollinator networks (Memmott and Waser 2002,

Valdovinos et al. 2009), the role of non-native plants in the assembly and restoration of functional pollination mutualisms is less understood.

If the goal of a restoration is to assemble a functional and stable community, special care should be taken to identify and encourage the development of those plants and pollinators that contribute most to the structure and stability of that community.

Historically though, the importance of a species within a community has not been easy to define and has proven even more difficult to quantify. However, borrowing techniques used in social and economic disciplines, restoration ecologists can quantify the functional importance of plants in restoration projects using centrality indices.

Originally deployed in social and information sharing networks, centrality indices measure the contribution of individual participants to network structure. Centrality indices at their most basic, measure how close an individual is to the center of a network of interactions and how well that individual connects members of that community to each other through its interactions with others. A highly central individual, sometimes called a “hub” or “connector”, interacts closely with the entire network. The concept of network centrality was first introduced in 1948 when Bavelas applied centrality to describe human communication networks in an effort to evaluate the relationship between network structure and the efficiency of communication in 32 group processes (Bavelas 1948). In the next decades the concept grew in popularity, being used to describe a wide range of phenomena including the importance of Moscow as a central hub in the 12th century network of river transportation in central Russia

(Pitts 1965), in explaining the diffusion patterns of technological innovations in the steel industry in the early 1900’s (Czepiel 1974), and in the dissemination of ideas in the modern democratic decision making process (Tsadiras et al. 2005).

Only recently however has the concept of network centrality been used to investigate ecological networks. In 2009 Gonzalez et al. used the technique to measure the importance of generalist pollinators in plant-pollinator networks. The study showed that not all participants were equally important to network structure, but rather that participants with high centrality scores were disproportionately more essential to the structure and stability of networks. The study was able to illustrate that the theoretical removal of central species from a community destabilized networks more quickly than the removal of more peripheral species, even when those peripheral species were more abundant. Because central species are important to the structure and stability of networks, the identification and encouragement of those species should be given precedence in restoration. By quantifying the relative importance of species within a community using centrality scores, managers can approach restoration and conservation projects with an informed and unbiased perspective.

Our objective was to determine the role of non-native plants in plant-pollinator networks under going restoration. Located within a grassland landscape that was 33 formerly surface mined for coal in central Ohio, our restoration patches were seeded with a mix of prairie plant species and subsequently invaded by native adventives and non-native plant species. To determine the relative importance of non-native plants in plant-pollinator networks, we calculated centrality scores for all plant species within the experimental patches. Because of the high relative abundance of non-native plant species, we predicted that non-native plants would be more central to networks than either native adventives or restoration plants seeded into patches.

Methods

Study Location

This study was conducted at The Wilds in Muskingum County, Ohio. The site was surface mined for coal until the mid 1980s. Soon after, the site was recontoured and seeded with a low diversity of non-native grasses and forbs as mandated by the 1977

Mining and Reclamation Act (Day et al. 1978). In 2009, 48 experimental plots of 10 m radius and separated by 70-100 m were established within a 1 km2 area of the grassland. Plots were seeded with one of three prairie plant seed mixes (appendix 1) and subsequently invaded by native adventives and non-native plant species (appendix

2). Prairie plants were chosen for restoration as they were likely to establish and succeed despite the characteristically low soil organic matter and degraded soil nutrient profile of highly disturbed reclaimed mine sites (Rodrigue and Burger 2004). Of the original 48 plots, 24 were used for this study. To minimize potential confounding

34 variables, 8 plots of each seed mixture were chosen from across the 1 km2 area (Figure

1).

Floral Sampling

To quantify floral abundance at the site, flower surveys were carried out every other week from late June until mid August 2010 at each of the 24 plots. Thus each plot was surveyed four times over the course of the season. Surveys were timed seasonally to encompass the main pollinator flight activity. We counted the number of floral units in 12 randomly placed one-meter square quadrats. Floral units were defined from the pollinator’s perspective rather than by flower anatomy (Appendix 2). Thus one floral unit was separated from another by the distance that a small pollinator would have to fly rather than walk (Saville 1993). We refer to per plot floral abundance as the total number of floral units in 12 one-meter square quadrats in each plot. Plant species were categorized by origin for analysis. Species seeded as part of the restoration effort are categorized as “restoration plants”. Native adventive species, not planted in the restoration effort but native to the area, are categorized as “native plants”. Finally, plant species that were neither planted in the restoration effort and are not native to the area are categorized as “non-native plants” (Appendix 2). Plants were categorized using the native status codes provided by the USDA Plants Database (USDA 2011). To avoid confusion between the plants we introduced as part of the restoration process and non- native plants, we choose to use the term non-native instead of “introduced”, as the

USDA refers to them. The term non-native avoids the connotations of other, more

35 negative words (i.e. “invasive”, “alien”, “weed”, “exotic”, etc.). Thus our distinction of a plant as non-native makes no allusion to that plant’s invasiveness; it denotes that the plant did not originate in central Ohio and was not planted as part of our restoration effort.

Pollinator Sampling

To quantify pollinator richness and abundance we used the same randomly placed 12 one-meter square quadrats used to measure the floral community. Pollinator surveys were performed on warm, dry days (22 – 38 degrees C) with moderate wind speed (0.1-3.4 m/s), between 0800 and 1600. We observed all open flowers in each one-meter square quadrat for 10 min. Insects were classified as pollinators only if they were observed actively probing flowers and contacting reproductive parts.

Consequently, insects observed perched on flowers but not probing floral reproductive parts were not considered pollinators. Pollinators observed foraging within the quadrat were collected into individual clean waxed-paper envelopes and placed into killing vials.

Clean envelopes were used to minimize the contamination of insect pollen load by pollens carried by other individuals. To ensure that we collected insect specimens foraging on less abundant floral species, we performed a 40-minute sampling of the entire plot in addition to the 12 randomly placed quadrats. In the 40-minute sample we collected all insects seen foraging anywhere within the 10 m radius plot. Thus, each plot was sampled for a total of 160 minutes/survey. Plots were surveyed every other week, four times over the course of the season.

36

Processing of Pollinator Specimens

To extend the record of pollinator visitation to include flowers visited in the recent past, we analyzed pollen found on all collected insect specimens. Previous studies have established that the inclusion of pollen load data in a network gives a more complete visitation history (Bosch et al. 2009). To determine pollen load composition, we rubbed a small cube of fuchsin-stained gelatin (~2 cubic mm) over each specimen.

We specifically targeted pollen held specialized pollen carrying structures such as scopa, because our intention was to estimate an inclusive visitation history, rather than identify pollen available for pollination as in other studies (Bosch 2009, Hinners 2009). The gelatin was then mounted onto glass slides. To avoid contamination, laboratory utensils and work surfaces were cleaned after each specimen. Pollen grains were identified to the lowest possible taxonomic level with the aid of a reference collection from the study site. In most cases pollen was identified to species. However, in the cases of Trifolium and Melilotus spp. we recorded pollen to the genus level because reliable species identification was beyond our capability. We only recorded only those taxa of pollen present in quantities greater than 15 grains per slide. Slides with fewer than 15 grains were removed from analysis as slide with fewer than this amount were likely the result of contamination rather than a true visitation record. Contamination could have either resulted from a pollinator picking up heterospecific pollen at a flower left by another pollinator that had previously visited another plant species (Bosch 2009) or from accidental contamination during netting or processing. Contamination was minimal and

37

97.8% of slides with pollen contained more than 15 grains for each recorded taxa.

Pollinators were then identified to the lowest possible taxonomic level. Of the total

2086 specimens collected, 1,982 were identified to species. The remaining 104 specimens, primarily Diptera of the families Sarcophaghidae and Tachinidae, were identified to morphospecies; these specimens accounted for less than 5% of the total collected specimens.

Network Analysis

To determine the importance of different plant categories in plant-pollinator networks, we determined link abundance and centrality scores for species within each category. Quantitative plant-pollinator networks were assembled for each of the 24 patches by combining data from visitation and pollen load networks. Thus, 24 networks were constructed in which the cell values indicate the number of times that individual pollinators of species “a” was collected foraging on flowers of plant species “p” in addition to the number of times that pollen of plant species “p” was found on pollinator species “a”. By combining visitation and pollen load data in the same network, we avoid the dangers of under-sampling interactions (Bluthgen and Menzel 2006, Bosch et al.

2009). By making individual networks for each plot we also avoid including “forbidden links” between plants and pollinators that do not overlap in space.

Once constructed, we used networks to determine link abundance for each plant category. Link abundance measures how often a plant participates in a network and is determined by combining the number of times that a plant was visited by any pollinator

38 and the number of times that the pollen of that plant was found on any pollinator. Thus, for a given plant, the measure is correlated to both the physical abundance of the plant, and the plant’s popularity among pollinators (Jordano et al. 2003).

For each network we also determined four indices of plant species centrality: betweenness, weighted betweenness, closeness, and weighted closeness. Both betweenness and closeness are calculated based on the presence or absence of links between plants and their pollinators, thus neither take account of the frequency of interactions. Using the bipartite package in R statistical program, we determined betweenness and closeness for individual plant species in our networks using a technique developed by Freeman (1978). Betweenness ranges from 0 to 1 and measures the degree to which a node connects compartments within a network

(Freeman 1978). One characteristic feature of plant-pollinator networks is the presence of compartments within a network. Compartmentalization describes the extent to which a network can be divided into subsections, or compartments, such that organisms within a compartment interact more strongly with one another than with species in other compartments (Dicks et al. 2002, Olesen et al. 2007). Thus participants that have high levels of betweenness act as connectors between compartments, bridging otherwise distinct regions of the network. For this reason, participants with high betweenness are also called “connectors” (Estrada and Bodin 2008). For our purposes, a plant with a high betweenness, approaching 1, connects compartments of plants and pollinators in the network that would otherwise be sparsely connected or not

39 connected at all. Conversely, plants with very low values of betweenness, approaching

0, contribute minimally to the connection of compartments within a network. The presence of connectors is important to the overall cohesiveness of a network (Freeman

1978). Another measure of centrality is closeness. Closeness, which also ranges from 0 to 1, measures the proximity of a node to all other nodes in the network (Freeman

1978). Put another way, closeness describes how close a plant is to the center of a compartment. Often called “hubs”, individual plants with high closeness values, approaching 1, interact with many species in the compartment either directly or indirectly. Likewise a plant with low closeness would be relatively unconnected (Martin

Gonzalez et al. 2009). Figure 2 shows a schematic highlighting the distinction between closeness and betweenness.

Weighted betweenness and weighted closeness, unlike their unweighted counterparts, take into account the frequency of interactions between plants and their pollinators. Adjusting for link abundance gives frequent interactions a greater weight

(Gonzalez et al. 2009b). Outside of this distinction, weighted betweenness and weighted closeness are identical to their unweighted counterparts. For our purposes, plants involved in frequent interactions exhibit higher values of weighted betweenness and weighted closeness than plants with less infrequent interactions (Freeman 1978).

Indices were first calculated for each plant and then grouped by category (restoration plants, native adventive plants, and non-native plants) for comparison. We used T tests to compare indices between non-native plants and restoration and native plant groups.

40

Statistics were tested for and met assumptions of normality and equal variances and results were deemed significant at the alpha = 0. 05 level.

Results

Forty-one plant species participated in plant-pollinator networks. In addition to those 41 species, we found Stellaria and Cardamine species growing in restoration plots.

However, because neither visitation observation nor pollen analysis evidenced pollinator visitation to these plants,, we did not include these plants in our analysis.

There also existed six plant species that occurred in no more than one or two plots.

Because we were interested in average centrality scores, we eliminated these plants from analysis and included only plants that occurred in three or more of the 24 plots. Of the 41 total plant species that participated in networks, five were native adventive plant species, 17 were restoration plant species, and 19 were non-native. Non-native plants accounted for 51% of the total floral abundance in plots. Restoration plants accounted for 43% and the remaining 6% of flowers were native adventives (Figure 3). There were a total of 5286 links in our 24 networks. Non-native plants were associated with 36% of those links. Restoration plants accounted for 56% and the remaining 8% of links involved native adventives (Figure 3).

Centrality Indices

Across plant species, betweenness averaged 0.053 (SE= 0.0035) and ranged from 0 for many plants to 0.145 in the case of restoration plant Rudbeckia hirta. Weighted betweenness averaged 0.052 (SE= 0.0062) and ranged from 0 for many species to 0.063

41

(Rudbeckia triloba). Closeness averaged 0.053 (SE= 0.0007) and ranged from a minimum

0.028 (Calystegia sp.) to a maximum of 0.063 (Rudbeckia hirta). Weighted closeness averaged 0.03 (SE= 0.0012) and ranged from 0.006 (Securigera varia) to a maximum of

0.19 in the case of the restoration plant species Coreopsis tinctoria (Appendix 2).

Non-native plants as a group averaged a betweenness of 0.051 (SE= 0.004), a weighted betweenness of 0.032 (SE= 0.007), a closeness of 0.054 (SE= 0.0009) and weighted closeness of 0.029 (SE= 0.0016) (Figure 4). When compared to native adventive plants, non-native plants had significantly higher values of betweenness and weighted betweenness. Non-native plants had similar values of betweenness as restoration plants and significantly lower values of weighted betweenness than restoration plants. Non-native plants had significantly higher values for both closeness and weighted closeness than native plants. When compared to restoration plants, non- natives had similar values of closeness and significantly lower values of weighted closeness than restoration plants. (Table 1, figure 4).

Discussion

Our results show not only that pollinators at our restoration site are using non- native plants, but also that non-native plants are central to the plant-pollinator networks we described. Based on unweighted metrics, non-native plants are as central to network organization as those planted in our restoration effort. Adjusted by link abundance, non-native plants decrease in centrality, though are still more central than

42 native volunteers. In this section we discuss the implications and limitations of our findings, and outline potential applications of our findings to habitat restoration.

In our plant-pollinator networks, measures of betweenness and closeness followed similar patterns between groups. This is likely a result of the fact that many of our networks, which were composed primarily of generalist plant and pollinator species, lacked distinct compartments (Figure 5). When networks lack distinct compartments, hubs and connectors become functionally similar (Freeman 1978). In other words, in a network comprising a single compartment, the most centralized participant, which has the shortest path to all other participants, is also the best connector of participants. The metrics of closeness and betweenness become redundant in the analysis of networks without that lack distinct compartments. As a result, we discuss centrality in our networks without distinguishing between closeness and betweenness.

Compared to other studies we found our participants to have relatively low centrality scores. Our networks averaged a betweenness of 0.053 and closeness of

0.053, considerably lower than values recorded in other networks (Gonzalez et al.

2009a). Using similar techniques, Gonzalez et al. (2009) found centrality scores that averaged across 0.4 across networks. We suppose that our low values are a result of the small size of our networks and the generalized nature of pollinators at our restoration site. In generalized networks many plant species share the role of hub, which gives the network a more circular shape, and dilutes the importance of any one plant as a central figure (Wasserman 1997). We speculate that in the nearly 25 years that our site lay

43 fallow between recountouring and restoration, the pollinator community lost whatever specialized pollinators had been in place before the disturbance, which stripped the site of the resources such specialized pollinators would have depended upon. The remaining pollinator community, supported in the interim by non-native and native adventive species, were likely the most generalized pollinators of the initial community. While our restoration efforts aim to eventually reinstate those pollinators lost in the disturbance, our communities at this early stage of the process are likely composed of those generalized pollinators that first recolonized the site. Given that centrality has not been determined for other constructed communities, our low centrality values may prove to be representative of communities early in the process of recovery.

Our comparisons of the centrality of plant groups based on the presence or absence of interactions indicate that we correctly predicted that non-native plants play a central role in network structure. Our results show that restoration and non-native plant species instantiated comparable values of centrality. By contrast, native adventive plants had significantly lower values of centrality than either of the other categories, implying that they contributed less to the structure and stability of networks. These patterns are likely the reflection of differences in relative floral abundance between the plant groups. Restoration and non-native plants constituted similar shares of the total available flowers at the site: comprising 43% and 51% respectively. Native adventive flowers, by contrast, were relatively scares: comprising only 6% of the total floral abundance. It is likely that pollinators did not encounter native adventive plants as often

44 while foraging and consequently did not visit them, reducing their unweighted centrality scores. As restoration and non-native plants shared a similar level of abundance, pollinators were equally likely to encounter them in plots, probably accounting for their comparable unweighted centrality scores.

Indices adjusted for interaction frequency, however, resulted in different patterns. While pollinators may be equally likely to encounter restoration and non- native plants during foraging, the higher weighted centrality values for restoration plants suggest that insect foragers preferred foraging on restoration plants to foraging on non-native ones. This result is especially interesting given that restoration plants had

10% lower richness and were 8% less abundant than non-native plants. Restoration plants accounted for the majority of links in the networks: 56%, as compared to the 36% and 8% of non-native and native adventive plants respectively. This finding agrees with other studies that have shown pollinators to prefer native flowers over non-native ones

(Thomson 1978, Moragues and Traveset 2005) and lends support to the concept that non-native plants do not necessarily compete with natives for pollination services

(Ghazoul 2006).

While our study supports the hypothesis that non-native plants play an important role in plant-pollinator network structure in our constructed prairie communities, some limitations apply to our conclusions. Our study took place over the course of a single season and our results represent a snapshot of a pollinator community in flux as it assembles over time. Plant-pollinator networks have been shown

45 to be very dynamic over time (Olesen et al. 2008) and our findings may not represent the long-term condition of the site. This is especially important to consider as these communities will likely change as the restoration process proceeds. As succession occurs and the relative abundance of plants and pollinators change at our study site, it is likely that the centrality of the plants will fluctuate as well.

Our findings also may reflect particular features of our study system. The reclaimed mine site where we worked was highly disturbed and has low soil organic matter and nutrients, conditions which may favor the establishment of weedy non- native plants over natives that may be at a relative disadvantage in such inhospitable conditions (Rodrigue and Burger 2004). In sites where native adventives have a greater chance to colonize, they may play more of a central role in the structure of plant- pollinator networks than we found here.

Our results give further support to the importance of active restoration of degraded habitats. In the short time since restoration began, the plants that we targeted in the restoration effort became the core of our plant-pollinator networks, being preferentially exploited by the pollinator community. Nvertheless, it is important to remember that for nearly 25 years native adventives and non-native plant species did and continue to help support the pollinator community at our study site. Because of this we suggest that in the incipient phases of habitat restoration, before either native volunteers or restoration plants dominate, non-native plants may help pollinator communities transition into restored habitats. Non-native plants support the pollinator

46 community, potentially facilitating visitation to new native species that occur in low density by sustaining pollinator populations during periods of resource dearth and by attracting pollinators from a distance. A growing body of evidence supports the conclusion that non-native plants may be important in the restoration process. Carroll

(2011) found that methods that incorporate beneficial non-native species into ecosystem management, may promote the persistence of ecosystem services, and increase the likelihood of sustainable outcomes in restoration projects. Similarly,

Schlaepfer et al. (2011) suggested that given non-native species have the ability to persist and flourish in the face of land-use and climatic change, they may contribute to the success of conservation goals in highly disturbed and dynamic habitats. They suggest that if managed correctly, non-native plants may be able to fill functional roles in degraded ecosystems. While both of the aforementioned papers offer some support from primary literature, both are highly theoretical. Our research, while limited in the scope of its applications, is one the first to offer evidence of the supporting role of non- native plants in the restoration of functionally stable plant-pollinator communities. By using network analysis, land managers can determine which plants are especially important to the structure and stability of plant-pollinator networks, which may even allow them to make better informed decisions in future restoration projects concerning which plants to seed for the encouragement robust plant-pollinator networks. Before native plants become established in the restoration process, and other resources become available, non-native plants may be able to provide important resources to

47 burgeoning pollinator communities and contribute important structural components to plant-pollinator networks. In some situations, it may prove to be the case that non- native plants provide important forage resources to pollinators and should be left in place for a limited period of time. We suggest that by encouraging the establishment of robust pollinator communities from the onset, restoration projects are more likely to be successful in restoring pollination function in a community in the long term.

48

References

Anderson, M. J. 2001. A new method for non-parametric multivariate analysis of variance. Austral Ecology 26:32-46. Bascompte, J. 2007. Networks in ecology. Basic and Applied Ecology 8:485-490. Bascompte, J. and D. B. Stouffer. 2009. The assembly and disassembly of ecological networks. Philos. Trans. R. Soc. B Biol. Sci. Philosophical Transactions of the Royal Society B: Biological Sciences 364:1781-1787. Bavelas, A. 1948. A Mathematical Model for Group Structures. Human Organization 7:16-30. Bergman, K.-O., J. Askling, O. Ekberg, H. Ignell, H. Wahlman, and P. Milberg. 2004. Landscape effects on butterfly assemblages in an agricultural region. Ecography 27:619-628. Biesmeijer, J., Roberts, S., Reemer, M., Ohlemuller, R., Edwards, M., Peeters, T., and A. Schaffers, Potts, S., Kleukers, R., Thomas, C., Settele, J., Kunin, W. 2006. Parallel declines in pollinators and insect pollinated plants in Britain and the Netherlands. Science 313. Bjerknes, A. L., O. Totland, S. J. Hegland, and A. Nielsen. 2007. Do alien plant invasions really affect pollination success in native plant species? Biological Conservation 138:1-12. Bluthgen, N. and F. Menzel. 2006. Measuring specialization in species interaction networks. BMC ecology 6. Bosch, J., A. M. M. Gonzalez, A. Rodrigo, and D. Navarro. 2009. Plant-pollinator networks: adding the pollinator's perspective. ecology letters 12:409-419. Bosch, J., Martin-Gonzalez, A., Rodrigo A., Navarro D. . 2009. Plant-pollinator networks: adding the pollinator’s perspective . ecology letters 12:409-419. Bowles, M. L., C. J. Whelan, C. Society for Ecological Restoration, R. Symposium on the, P. Restoration of Endangered, and Animals, editors. 1994. Restoration of endangered species : conceptual issues, planning, and implementation. Cambridge University Press, Cambridge [England]; New York. Buchmann, S. L. and G. P. Nabhan. 1996. The forgotten pollinators. Island Press [for] Shearwater Books, Washington, D.C. Carroll, S. P. 2011. Conciliation biology: the eco-evolutionary management of permanently invaded biotic systems. Evolutionary Applications 4:184-199.

49

Carvalheiro, L. G., C. L. Seymour, R. Veldtman, and S. W. Nicolson. 2010. Pollination services decline with distance from natural habitat even in biodiversity-rich areas. Journal of Applied Ecology 47:810-820. Chittka, L. and S. Schurkens. 2001. Metrology: Successful invasion of a floral market. Nature 411:653. Czepiel, J. A. 1974. Word-of-Mouth Processes in the Diffusion of a Major Technological Innovation. Journal of Marketing Research 11:172-180. Daehler, C. C. 2003. Performance Comparisons of Co-Occurring Native and Alien Invasive Plants: Implications for Conservation and Restoration. Annual Review of Ecology, Evolution, and Systematics 34:183-211. Day, J. M., I. Practising Law, and S. United. 1978. Federal regulation of mine safety, health, and reclamation. Practising Law Institute, New York. Dicks, L. V., S. A. Corbet, and R. F. Pywell. 2002. Compartmentalization in plant–insect flower visitor webs. Journal of Animal Ecology 71:32-43. Dixon, K. 2009. Pollination and Restoration. Science 325. Dormann, C. F., Gruber, B., Frund, J. 2008. Introducing the bipartite package: Analyzing Ecological Networks. R Development Core Team. Dunne, J. A., R. J. Williams, and N. D. Martinez. 2002a. Food-Web Structure and Network Theory: The Role of Connectance and Size. Proceedings of the National Academy of Sciences of the United States of America 99:12917-12922. Dunne, J. A., R. J. Williams, and N. D. Martinez. 2002b. Network structure and biodiversity loss in food webs: robustness increases with connectance. ecology letters 5. Estrada, E. 2007. Food webs robustness to biodiversity loss: the roles of connectance, expansibility and degree distribution. Journal of Theoretical Biology 244:296-307. Estrada, E. and ñ. Bodin. 2008. USING NETWORK CENTRALITY MEASURES TO MANAGE LANDSCAPE CONNECTIVITY. Ecological Applications 18:1810-1825. Exeler, N., A. Kratochwil, and A. Hochkirch. 2009. Restoration of riverine inland sand dune complexes: implications for the conservation of wild bees. Journal of Applied Ecology 46:1097-1105. Forup, M. L., K. S. E. Henson, P. G. Craze, and J. Memmott. 2008. The restoration of ecological interactions: plant-pollinator networks on ancient and restored heathlands. Journal of Applied Ecology 45:742-752. Forup, M. L. and J. Memmott. 2005. The restoration of plant-pollinator interactions in hay meadows. Restoration Ecology 13:265-274. Freeman, L. C. 1978. Centrality in social networks conceptual clarification. Social Networks 1:215-239. Ghazoul, J. 2006. Floral diversity and the facilitation of pollination. Journal of Ecology 94:295-304. Gilbert, A. J. 2009. Connectance indicates the robustness of food webs when subjected to species loss. Ecological Indicators 9:72-80.

50

Gomez, J. M., J. Bosch, F. Perfectti, J. Fernandez, and M. Abdelaziz. 2007. Pollinator diversity affects plant reproduction and recruitment: the tradeoffs of generalization. Oecologia 153:597-605. Gonzalez, A. M., B. Dalsgaard, and J. M. Olesen. 2009a. Centrality measures and the importance of generalist species in pollination networks. Ecological Complexity 7:36-43. Gonzalez, A. M. M., B. Dalsgaard, and J. M. Olesen. 2009b. Centrality measures and the importance of generalist species in pollination networks. Ecological Complexity 7:36-43. Gordon, D. R. 1998. Effects of Invasive, Non-Indigenous Plant Species on Ecosystem Processes: Lessons from Florida. Ecological Applications 8:975-989. Greenleaf, S., N. Williams, R. Winfree, and C. Kremen. 2007. Bee foraging ranges and their relationship to body size. Oecologia 153:589-596. Guimaraes, P. R., G. Machado, M. A. M. de Aguiar, P. Jordano, J. Bascompte, A. Pinheiro, and S. F. dos Reis. 2007. Build-up mechanisms determining the topology of mutualistic networks. Journal of Theoretical Biology 249:181-189. Henson, K. S. E., P. G. Craze, and J. Memmott. 2009. The restoration of parasites, parasitoids, and pathogens to heathland communities. Ecology 90:1840-1851. Hinners, S., Hjelmroos-koski, M. . 2009. Receptiveness of foraging wild bees to exotic landscape elements. American Midland Naturalist 162:253-265. Hobbs, R. J. and D. A. Norton. 1996. Towards a Conceptual Framework for Restoration Ecology. Restoration Ecology 4:93-110. Hoehn, P., T. Tscharntke, J. M. Tylianakis, and I. Steffan-Dewenter. 2008. Functional group diversity of bee pollinators increases crop yield. Proc. R. Soc. B Biol. Sci. Proceedings of the Royal Society B: Biological Sciences 275:2283-2291. Jari Oksanen, R. K., Pierre Legendre, Bon O'Hara, Gavin L. Simpson, Peter Solovmos, M. Henry H. Stevens, Helene Wanger. 2009. Vegan: Community Ecology Package.in R. Package, editor. Jauker, F., T. DiekËotter, F. Schwarzbach, and V. Wolters. 2009. Pollinator dispersal in an agricultural matrix: opposing responses of wild bees and hoverflies to landscape structure and distance from main habitat. Landscape Ecology 24:547-555. Jordan, W. R., M. E. Gilpin, and J. D. Aber. 1999. Restoration ecology : a synthetic approach to ecological research. Cambridge University Press, Cambridge, Angleterre. Jordano, P. 1987. PATTERNS OF MUTUALISTIC INTERACTIONS IN POLLINATION AND SEED DISPERSAL - CONNECTANCE, DEPENDENCE ASYMMETRIES, AND COEVOLUTION. American Naturalist 129:657-677. Jordano, P., J. Bascompte, and J. M. Olesen. 2003. Invariant properties in coevolutionary networks of plant–animal interactions. ecology letters 6:69-81. Kaiser-Bunbury, C. N., S. Muff, J. Memmott, C. B. Muller, and A. Caflisch. The robustness of pollination networks to the loss of species and interactions: a quantitative approach incorporating pollinator behaviour. ecology letters 13:442-452. 51

Kaiser-Bunbury, C. N., A. Traveset, and D. M. Hansen. 2010. Conservation and restoration of plantanimal mutualisms on oceanic islands. Perspectives in Plant Ecology, Evolution & Systematics 12. Kearns, C. A., D. W. Inouye, and N. M. Waser. 1998. Endangered mutualisms: The conservation of plant-pollinator interactions. Annual Review of Ecology and Systematics 29:83-112. Klein, A.-M. 2009. Nearby rainforest promotes coffee pollination by increasing spatio- temporal stability in bee species richness. Forest Ecology and Management 258. Krauss, J., I. Steffan-Dewenter, C. B. Muller, and T. Tscharntke. 2005. Relative importance of resource quantity, isolation and habitat quality for landscape distribution of a monophagous butterfly. Ecography 28:465-474. Kremen, C., N. M. Williams, M. A. Aizen, B. Gemmill-Herren, G. LeBuhn, R. Minckley, L. Packer, S. G. Potts, T. Roulston, I. Steffan-Dewenter, D. P. Vazquez, R. Winfree, L. Adams, E. E. Crone, S. S. Greenleaf, T. H. Keitt, A. M. Klein, J. Regetz, and T. H. Ricketts. 2007. Pollination and other ecosystem services produced by mobile organisms: a conceptual framework for the effects of land-use change. ecology letters 10:299-314. Linhart. 1995. Restoration, Revegetation, and the Importance of Genetic and Evolutionary Perspectives.in Proceedings: Wild Land Shrub and Arid Land Restoration Las Vegas. Lundgren, R. and J. M. Olesen. 2005. The dense and highly connected world of Greenland's plants and their pollinators. Arctic Antarctic and Alpine Research 37:514-520. Manchester, S. J. and J. M. Bullock. 2000. The impacts of non-native species on UK biodiversity and the effectiveness of control. Journal of Applied Ecology 37:845- 864. Martin Gonzalez, A. M., B. Dalsgaard, and J. M. Olesen. 2009. Centrality measures and the importance of generalist species in pollination networks. Ecological Complexity 7:36-43. Matheson, A., L. Linnean Society of, and A. International Bee Research. 1996. The conservation of bees. Academic Press, London; San Diego, CA. McArdle, B. H. and M. J. Anderson. 2001. Fitting Multivariate Models to Community Data: A Comment on Distance-Based Redundancy Analysis. Ecology 82:290-297. Memmott, J. 1999. The structure of a plant-pollinator food web. ecology letters 2:276- 280. Memmott, J., D. Alonso, E. L. Berlow, A. Dobson, J. A. Dunne, R. Sole, and J. Weitz. 2006. Biodiversity loss and ecological network structure. Pages 325-347 in M. Pascual and J. A. Dunne, editors. Ecological Networks: Linking Structure to Dynamics in Food Webs. Memmott, J. and N. M. Waser. 2002. Integration of alien plants into a native flower- pollinator visitation web. Proceedings of the Royal Society of London Series B- Biological Sciences 269:2395-2399. 52

Memmott, J., N. M. Waser, and M. V. Price. 2004. Tolerance of pollination networks to species extinctions. Proceedings of the Royal Society B-Biological Sciences 271:2605-2611. Menz, M. H. M., R. D. Phillips, R. Winfree, C. Kremen, M. A. Aizen, S. D. Johnson, and K. W. Dixon. 2010. Reconnecting plants and pollinators: challenges in the restoration of pollination mutualisms. Trends in Plant Science 16. Michener, C. D. 2000. The bees of the world. Johns Hopkins University Press, Baltimore, Md. Moragues, E. and A. Traveset. 2005. Effect of Carpobrotus spp. on the pollination success of native plant species of the Balearic Islands. Biological Conservation 122:611-619. Morales, C. L. and A. Traveset. 2009. A meta-analysis of impacts of alien vs. native plants on pollinator visitation and reproductive success of co-flowering native plants. ecology letters 12:716-728. Mueller, L. D. and L. Altenberg. 1985. Statistical Inference on Measures of Niche Overlap. Ecology 66:1204-1210. Murray, T. E., M. Kuhlmann, and S. G. Potts. 2009. Conservation ecology of bees: populations, species and communities. Apidologie 40:211-236. Olesen, J. M., J. Bascompte, Y. L. Dupont, and P. Jordano. 2007. The modularity of pollination networks. Proceedings of the National Academy of Sciences of the United States of America 104:19891-19896. Olesen, J. M., J. Bascompte, H. Elberling, and P. Jordano. 2008. Temporal dynamics in a pollination network. Ecology 89:1573-1582. Pitts, F. R. 1965. A GRAPH THEORETIC APPROACH TO HISTORICAL GEOGRAPHY. The Professional Geographer 17:15-20. Potts, S. 2009. Enhancing pollinator biodiversity in intensive grasslands. Journal of Applied Ecology 46:369-379. Potts, S. G., B. Vulliamy, A. Dafni, G. Ne'eman, and P. Willmer. 2003. Linking bees and flowers: How do floral communities structure pollinator communities? Ecology 84:2628-2642. Potts, S. G., B. A. Woodcock, S. P. M. Roberts, T. Tscheulin, E. S. Pilgrim, V. K. Brown, and J. R. Tallowin. 2009. Enhancing pollinator biodiversity in intensive grasslands. Journal of Applied Ecology 46:369-379. Richardson, D. M., N. Allsopp, C. M. D'Antonio, S. J. Milton, and M. Rejmanek. 2000. Plant invasions - the role of mutualisms. Biological Reviews 75:65-93. Ried, A., Morin, L., Downey, P., French, K., Virtue, J. 2009. Does invasive plant management aid the restoration of natural system. Biological Conservation 142:2342-2349. Rodrigue, J. A. and J. A. Burger. 2004. Forest Soil Productivity of Mined Land in the Midwestern and Eastern Coalfield Regions. Soil Sci. Soc. Am. J. 68:833-844. Roulston, T. H. and K. Goodell. 2011. The role of resources and risks in regulating wild bee populations. Annual Review of Entomology 56:293-312. 53

Saville, N. M. 1993. Bumblebee ecology in Woodlands and Arable farmland. University of Cambridge. Schlaepfer, M. A., D. F. Sax, and J. D. Olden. 2011. The Potential Conservation Value of Non-Native Species El Valor de Conservación Potencial de Especies No Nativas. Conservation Biology 25:428- 437. Schweiger, O., M. Musche, D. Bailey, R. Billeter, T. Diekˆtter, F. Hendrickx, F. Herzog, J. Liira, J.-P. Maelfait, M. Speelmans, and F. Dziock. 2007. Functional richness of local hoverfly communities (Diptera, Syrphidae) in response to land use across temperate Europe. Oikos 116:461-472. Scott, J. A. 1986. The butterflies of North America : a natural history and field guide. Stanford University Press, Stanford, Calif. Shapiro, A. M. 2002. The Californian urban butterfly fauna is dependent on alien plants. Diversity and Distributions 8:31-40. Steffan-Dewenter, I. 2003. Importance of habitat area and landscape context for species richness of bees and wasps in fragmented orchard meadows. Conservation Biology 17:1036-1044. Steffan-Dewenter, I., U. Munzenberg, and T. Tscharntke. 2001. Pollination, seed set and seed predation on a landscape scale. Proceedings of the Royal Society of London Series B-Biological Sciences 268:1685-1690. Steffan-Dewenter, I., S. G. Potts, and L. Packer. 2005. Pollinator diversity and crop pollination services are at risk. Trends in Ecology & Evolution 20:651-652. Steffan-Dewenter, I. and T. Tscharntke. 2000. Butterfly community structure in fragmented habitats. ecology letters 3:449-456. Steffan-Dewenter, I. and C. Westphal. 2008. The interplay of pollinator diversity, pollination services and landscape change. Journal of Applied Ecology 45:737- 741. Stout, J. C. and C. L. Morales. 2009. Ecological impacts of invasive alien species on bees. Apidologie 40:388-409. Team, R. D. C. 2009. R: A language and environment for statistical computing.in R. F. f. S. Computing, editor., Vienna, Austria. Thomson, J. D. 1978. Effects of Stand Composition on Insect Visitation in Two-Species Mixtures of Hieracium. American Midland Naturalist 100:431-440. Tsadiras, A., I. Kouskouvelis, P. Bozanis, and E. Houstis. 2005. Using Fuzzy Cognitive Maps as a Decision Support System for Political Decisions: The Case of Turkey‚Äôs Integration into the European Union Advances in Informatics. Pages 371-381. Springer Berlin / Heidelberg. USDA, N. 2011. The PLANTS Database. National Plant Data Team, Greensboro, NC. Valdovinos, F. S., R. Ramos-Jiliberto, J. D. Flores, C. Espinoza, and G. Lopez. 2009. Structure and dynamics of pollination networks: the role of alien plants. Oikos 118:1190-1200.

54

Vergara, C. H. and E. I. Badano. 2009. Pollinator diversity increases fruit production in Mexican coffee plantations: The importance of rustic management systems. Agric. Ecosyst. Environ. Agriculture, Ecosystems and Environment 129:117-123. Waser, N. M., L. Chittka, M. V. Price, N. M. Williams, and J. Ollerton. 1996. Generalization in pollination systems, and why it matters. Ecology 77:1043-1060. Westrich, P. 1996. Habitat requirements of central European bees and the problem of partial habitats.in S. L. B. A. Matheson, C. O'Toole, P. Westrich & I.H. Williams, editor. In The Conservation of Bees. Academic Press, London, UK. Wilcock, C. and R. Neiland. 2002. Pollination failure in plants: why it happens and when it matters. Trends in Plant Science 7:270-277. Wilcove, #160, D. S., ROTHSTEIN, D., DUBOW, J., PHILLIPS, A., LOSOS, and E. 1998. QUANTIFYING THREATS TO IMPERILED SPECIES IN THE UNITED STATES. American Institute of Biological Sciences, Washington, DC, ETATS-UNIS. Williams, N. M. 2010. Restoration of Nontarget Species: Bee Communities and Pollination Function in Riparian Forests. Restoration Ecology 19. Williams, N. S. G., M. J. McDonnell, G. K. Phelan, L. D. Keim, and R. Van Der Ree. 2006. Range expansion due to urbanization: Increased food resources attract Grey- headed Flying-foxes (Pteropus poliocephalus) to Melbourne. Austral Ecology 31:190-198. Zurbuchen, A., L. Landert, J. Klaiber, A. Muller, S. Hein, and S. Dorn. 2011. Maximum foraging ranges in solitary bees: only few individuals have the capability to cover long foraging distances. Biol. Conserv. Biological Conservation 143:669-676.

55

Appendix A: Figures and Tables

Figure 1: Map of the study site at the Wilds Conservation Park in Muskingum County, Ohio. Remnant forest is shown in green and the lake in blue. Green dots represent high floral diversity plots. Yellow dots represent low floral diversity plots. The large red ovals distinguish “near plots” (within 250 m of the remnant forest) and “far plots” (more than 300m from the remnant forest). 56

Table 1: Shows the results t tests comparing pollinator abundance, pollinator richness and community evenness between near and far plots of low and high floral diversity treatments. *indicates a significant difference at the = .05 level.

Pollinator Community Analysis df t p value Low Floral Pollinator Abundance 6 1.59 .161 Diversity Pollinator Richness 6 2.50 .046* Community Evenness 6 .073 .941

High Floral Pollinator Abundance 14 .477 .64 Diversity Pollinator Richness 14 2.11 .05* Community Evenness 14 .84 .41

57

120 35 a) b) a 30 100 a 25 b 80 b 20 60 15

40 10

Pollinator Richness Pollinator Abundance 20 5

0 0 Low Diveristy High Diveristy Low Diveristy High Diveristy c)

1 0.9 Near 0.8 Far 0.7

0.6

0.5

Evenness 0.4 0.3

0.2

0.1

0

Low Diveristy High Diveristy

Figure 2: Compares a) pollinator abundance, b) richness and c) evenness between near plots (<250m from remnant forest, n=8) and far plots (>300m from remnant habitat, n=8) for the low and high floral diversity treatments. Letters indicate significant differences at the = .05 level.

58

Figure 3: NMDS ordination of high diversity plots. Plots are indicated with bold crosses, pollinator groups are indicated with small dots. The solid line surrounds plots near to the remnant forest edge (<250 m, n=8). The dashed line surrounds plot far from the forest edge (>300m, n=8). Adonis reveals that pollinator communities near and far from the remnant forest are significantly different.

59

Figure 4: NMDS ordination of low diversity plots. Plots are indicated with bold crosses. Pollinator groups are indicated with species with small dots. The solid line surrounds plots near to the remnant forest edge (<250 m, n=4). The dashed line surrounds plot far from the forest edge (>300m, n=4). Adonis reveals that pollinator communities near and far from the remnant forest are not significantly different.

60

Table 2: Gives the value of d’, a measure of pollinator specialization, for

some of the pollinator species. d’ close to 0 indicates pollinator generalization, likewise d’ = 1 indicates extreme specialization. Bolded species are those that increase in relative abundance in distant high diversity plots.

Generalized Species d' Bombus perplexus 0.0996 Bombus auricomus 0.1412 Halictus confusus 0.2048 Toxomeris marginatus 0.2171

Ceratina strenua 0.2227

Apis mellifera 0.2441 Xylocopa virginica 0.2494 Halictus ligatus 0.2562 Ceratina calcarata 0.2613

Bombus bimaculatus 0.2788

Ceratina dupla 0.2996 Bombus impatiens 0.3011 Augochlorella aurata 0.3014 Augochloropsis metallica 0.3099

Augochlorella persimilis 0.3102

Bombus griseocollis 0.3155

Phyciodes tharos 0.3590 Cupido cocyta 0.3717 Hyleaus affinis 0.3792

Megachile mendica 0.3819 Hyleaus mesillae 0.3866 Specialized Hoplitis spoliata 0.4033 Andrena brevipalpis 0.4334 Megachile montivaga 0.5920 Heriades leavitti 0.5979 Melissodes bidentis 0.7968

61

Table 3: Shows the results of two sided t tests comparing connectance, pollinator niche overlap, and network robustness between near and far plots for both low and high diversity treatments. *indicates a significant difference at the = .05 level.

Plant- Pollinator Network Analysis df t p value

Low Floral Connectance 6 1.01 .15 Diversity Pollinator Niche 6 .823 .44 Overlap Network Robustness 6 2.445 .05*

High Floral Connectance 14 2.35 .03* Diversity Pollinator Niche 14 2.54 .02* Overlap Network Robustness 14 .098 .92

62

a) b) 0.4 a 0.3 0.35 a ab ab b 0.25 0.3 c a b

0.25 0.2

0.2 b 0.15 0.15

0.1 0.1

Connectance Network Pollinator Overlap Niche 0.05 0.05

0 0

Low Diveristy High Diveristy Low Diveristy High Diveristy

c) 0.6 a c 0.5 b c

0.4

0.3 Near 0.2 Far

Network Robustness 0.1

0 Low Diveristy High Diveristy

Figure 5: Compares a) network connectance, b) pollinator niche overlap and c) network robustness between near (<250m from remnant forest, N=8) and far plots (>300m from remnant habitat, N=8) of the low and high floral diversity treatments. Letters indicate significant differences at the = .05 level.

63

Table 4: Lists the plant species, family and floral unit of plants seeded in low and high floral diversity plots as well as the native adventives and non-native plants that participated in plant-pollinator networks. .

Low Floral Diversity Seed Mix Family Floral Unit Base Mix- Coreopsis tinctoria Asteraceae Capitula in all plots (n=24) Bidens aristosa Asteraceae Capitula Rudbeckia hirta Asteraceae Capitula Rudbeckia triloba Asteraceae Capitula Rudbeckia subtomentosa Asteraceae Capitula Ibiris umbellata Brassicaceae Umbel Gaillardia pulchella Asteraceae Capitula Chrysanthemum leucanthemum Asteraceae Capitula Linum perenne Linaceae Flower High Floral Diversity Seed Mixes In half of Aster novae-angilae Asteraceae Capitula High Diversity Plots (n=8) Echinacea purpurea Asteraceae Capitula Eryngium yuccifolium Apiaceae Umbel Eupatorium maculatum Asteraceae Capitula Allium cernuum Liliaceae Flower Helianthus mollis Asteraceae Capitula Heliopsis helianthoides Asteraceae Capitula Liatris pycnostachya (Cormb) Asteraceae Capitula Solidago rigida Asteraceae Capitula Verbena stricta Lamiaceae Flower Zizia aurea Apiaceae Umbel In half of Astragulus canadensis Fabaceae Inflorescence High Diversity Plots (n=8) Chamaechrista fasciculata Fabaceae Flower Cleome glabra Cleomaceae Flower Dalea candida Fabaceae Inflorescence Agastache foeniculum Lamiaceae Flower Monarda citrodora Lamiaceae Flower Monarda fistulosa Lamiaceae Flower Penstemon digitalis Plantaginaceae Flower Vernonia fasciculata Asteraceae Capitula Salvia azurea Lamiaceae Flower

Native Adventives Achillea millefolium Asteraceae Umbel Asclepias syriaca Asclepiadoidea Inflorescence Erigeron annus Asteraceae Capitula Phytolacca americana Phytolaccaceae Flower Apocynum cannabinum Apocynaceae Flower Satureja vulgaris Lamiaceae Flower

Continued

64

Table 4 continued

Non-Natives Calystegia sp. Convolvulaceae Flower Cichorium intybus Asteraceae Capitula Cirsium arvense Asteraceae Capitula Cirsium vulgare Asteraceae Capitula Daucus carota Apiaceae Umbel Dianthus sp. Caryophyllaceae Flower Dipsacus fullonum Dipsacaceae Inflorescence Galium sp. Rubiaceae Cluster Hypericum perforatum Hypericaceae Flower Lactuca serriola Asteraceae Capitula Lotus corniculatus Fabaceae Flower Medicago sativa Fabaceae Inflorescence Melilotus sp. Fabaceae Inflorescence Potentilla recta Rosaceae Flower

Securigera varia Fabaceae Flower Trifolium pratense Fabaceae Inflorescence Trifolium sp. Fabaceae Inflorescence Verbascum blattaria Scrophulariaceae Flower Verbascum thapsus Scrophulariaceae Flower

C B C Plants

Pollinator s Figure 6: Shows a simple schematic of a bi-modal plant-pollinator network. Plants are depicted as dark circles, pollinators as open circles and lines indicate a link between the two. Boxes indicate compartments. Plants labeled the letter ‘C’ are well connected within each compartment and would be considered compartment hubs, with high values of closeness. The plant labeled with the letter ‘B’ connects the two compartments and would have a high value of betweenness.

65

Continued

a) Native Restoration Non-native

b) Native Restoration Non-native

c) Native Restoration Non-native

Figure 7: Shows a comparison of the floral community. Panel a) compares floral richness between native adventives, restoration and non-native plants. Panel b) compares floral abundance and c) link abundance.

66

Table 5: Shows the results for T-Tests comparing network betweenness, weighted betweenness, closeness and weighted closeness between non-native plants and both native and restoration plant species in restoration plots at the Wilds in Muskingum County Ohio. The weighting takes into account the frequency of interactions between plant and pollinator. *indicates a significant difference at the = .05 level.

df T P Betweenness Non-native * Native 150 -4.18 <0.001* Non-native * Restoration 323 2.23 0.027* Weighted Betweenness Non-native * Native 132 -1.57 0.120 Non-native * Restoration 248 4.35 <0.001* Closeness Non-Native * Native 78 -2.19 0.032* Non-Native * Restoration 336 1.08 0.280 Weighted Closeness Non-Native * Native 156 -4.50 <0.001* Non-Native * Restoration 286 3.87 <0.001*

67

Figure 8: Compares network betweenness, weighted betweenness, closeness and weighted closeness between native adventive, restoration and non-native plants in plots at the Wilds in Muskingum County, Ohio. Weighted measures account for the frequency of interactions between plant and pollinator. Letters indicate significant differences at the = .05 level.

68

PLANTS RESTORATION NATIVE NON-NATIVE

POLLINATORS

Figure 9: Shows the plant pollinator network across plots. Non-native plants and links with non native plants are shown in the lightest grey.

69

Table 6: List of seed mixtures

Base mix- In all 24 plots Coreopsis tinctoria Bidens aristosa Rudbeckia hirta Rudbeckia triloba Rudbeckia subtomentosa Ibiris umbellata Gaillardia pulchella Chrysanthemum leucanthemum Linum perenne High Floral Diversity Seed Mix- In 8 plots Aster novae-angilae Echinacea purpurea Eryngium yuccifolium Eupatorium maculatum Allium cernuum Helianthus mollis Heliopsis helianthoides Liatris pycnostachya (Cormb) Solidago rigida Zizia aurea High Floral Morphology Seed Mix- In 8 plots Astragulus canadensis Chamaechrista fasciculata Cleome glabra Dalea candida Agastache foeniculum Monarda citrodora Monarda fistulosa Penstemon digitalis Vernonia fasciculata Salvia azurea

70

Table 7: Average closeness, betweenness, weight closeness and weighted betweenness and standard deviation for each plant species. Status indicates how plants were defined. Status was determined using USDA website looking specifically at the status in Central Ohio.

Plant Floral Unit Status Number Average (SE) of Plots Betweenness Achillea Umbel Native 20 0.0115 0.0057 millefolium Agastache Inflorescence Restoration 5 0.0222 0.0152 foeniculum Asclepias syriaca Inflorescence Native 3 0.0000 0.0000 Astragalus Inflorescence Restoration 3 0.0035 0.0035 canadensis Calystegia sp. Flower Non-native 4 0.0038 0.0038 Chamaecrista Flower Restoration 3 0.0000 0.0000 fasciculata Chrysanthemum Capitula Restoration 21 0.0353 0.0095 leucanthemum Cichorium intybus Capitula Introduced 5 0.0034 0.0021 Cirsium arvense Capitula Introduced 24 0.0890 0.0129 Cirsium vulgare Capitula Introduced 6 0.0210 0.0150 Coreopsis tinctoria Capitula Restoration 24 0.0918 0.0150 Daucus carota Umbel Introduced 24 0.1096 0.0259 Dianthus sp. Flower Introduced 3 0.0000 0.0000 Dipsacus Inflorescence Introduced 16 0.0233 0.0116

fullonum Echinacea Capitula Restoration 10 0.0412 0.0143 purpurea Erigeron annus Capitula Native 21 0.0379 0.0139 Gaillardia pulchella Capitula Restoration 13 0.0247 0.0082 Galium sp. Cluster Introduced 5 0.0003 0.0003 Heliopsis Capitula Restoration 12 0.0350 0.0138 helianthoides Hypericum Flower Introduced 16 0.0211 0.0078 perforatum Lactuca serriola Capitula Introduced 4 0.0150 0.0150 Liatris Capitula Restoration 5 0.0152 0.0139 pycnostachya Linum perenne Flower Restoration 6 0.0000 0.0000 Lotus corniculatus Flower Introduced 24 0.1335 0.0172 Medicago sativa Inflorescence Introduced 3 0.0010 0.0010 Melilotus sp. Inflorescence Introduced 23 0.0416 0.0105 Monarda fistulosa Flower Restoration 19 0.0523 0.0136 Monarda Flower Restoration 13 0.0298 0.0134

citriodora

71 Continued

Table 7 continued Phytolacca Flower Native 3 0.0000 0.0000

americana Potentilla recta Flower Introduced 3 0.0204 0.0167 Rudbeckia hirta Capitula Restoration 24 0.1456 0.0231 Rudbeckia triloba Capitula Restoration 24 0.1066 0.0151 Satureja vulgaris Inflorescence Restoration 6 0.0123 0.0080 Securigera varia Flower Introduced 3 0.0000 0.0000 Solidago rigida Inflorescence Native 3 0.0171 0.0171 Trifolium Inflorescence Introduced 14 0.0185 0.0077

pratense Trifolium sp. Inflorescence Introduced 21 0.0415 0.0088 Verbascum Flower Introduced 7 0.0035 0.0035 blattaria Verbascum Flower Introduced 3 0.0035 0.0054 thapsus Verbena stricta Inflorescence Native 3 0.0192 0.0054 Vernonia Capitula Restoration 3 0.0000 0.0000 fasciculata

Table 8: Average closeness, betweenness, weight closeness and weighted betweenness and standard deviation for each plant species. Status indicates how plants were defined. Status was determined using USDA website looking specifically at the status in Central Ohio.

Plant Average (SE) Average (SE) Average (SE) Weighted Closeness Weighted Betweenness Closeness Achillea 0.0000 0.0000 0.0518 0.0006 0.0018 0.0081 millefolium Agastache 0.0421 0.0188 0.0447 0.0032 0.0113 0.0252 foeniculum Asclepias syriaca 0.0000 0.0000 0.0336 0.0007 0.0017 0.0030 Astragalus 0.0000 0.0000 0.0405 0.0042 0.0052 0.0090 canadensis Calystegia sp. 0.0000 0.0000 0.0289 0.0116 0.0003 0.0006 Chamaecrista 0.0000 0.0000 0.0408 0.0037 0.0029 0.0050 fasciculata Chrysanthemum 0.0149 0.0033 0.0555 0.0021 0.0023 0.0103 leucanthemum Cichorium intybus 0.0000 0.0000 0.0575 0.0097 0.0032 0.0072 Cirsium arvense 0.1193 0.0244 0.0606 0.0024 0.0031 0.0152 Cirsium vulgare 0.0083 0.0034 0.0432 0.0058 0.0051 0.0125 Coreopsis tinctoria 0.2160 0.0441 0.0623 0.0028 0.0081 0.0398 Daucus carota 0.2053 0.0419 0.0625 0.0035 0.0046 0.0225

72 Continued

DianthusTable 8 continued sp. 0.0000 0.0000 0.0688 0.0129 0.0128 0.0222 Dipsacus 0.0334 0.0084 0.0481 0.0024 0.0025 0.0100 fullonum Echinacea 0.0366 0.0116 0.0514 0.0059 0.0060 0.0190 purpurea Erigeron annus 0.1132 0.0247 0.0537 0.0042 0.0036 0.0165 Gaillardia pulchella 0.0280 0.0078 0.0519 0.0034 0.0039 0.0139 Galium sp. 0.0000 0.0000 0.0434 0.0029 0.0023 0.0050 Heliopsis 0.0201 0.0058 0.0530 0.0029 0.0029 0.0100 helianthoides Hypericum 0.0470 0.0118 0.0484 0.0030 0.0031 0.0125 perforatum Lactuca serriola 0.0000 0.0000 0.0451 0.0068 0.0046 0.0092 Liatris 0.0000 0.0000 0.0428 0.0063 0.0050 0.0112 pycnostachya Linum perenne 0.0000 0.0000 0.0454 0.0022 0.0009 0.0022 Lotus corniculatus 0.1836 0.0375 0.0629 0.0023 0.0072 0.0353 Medicago sativa 0.0000 0.0000 0.0419 0.0019 0.0024 0.0041 Melilotus sp. 0.0364 0.0076 0.0541 0.0023 0.0061 0.0291 Monarda fistulosa 0.2036 0.0467 0.0513 0.0020 0.0062 0.0269 Monarda 0.0622 0.0172 0.0481 0.0023 0.0049 0.0177 citriodora Phytolacca 0.0000 0.0000 0.0490 0.0146 0.0022 0.0039 americana Potentilla recta 0.0000 0.0000 0.0391 0.0063 0.0049 0.0084 Rudbeckia hirta 0.1654 0.0338 0.0632 0.0028 0.0058 0.0283 Rudbeckia triloba 0.2014 0.0411 0.0628 0.0029 0.0051 0.0252 Satureja vulgaris 0.0000 0.0000 0.0393 0.0032 0.0017 0.0042 Securigera varia 0.0000 0.0000 0.0384 0.0017 0.0004 0.0007 Solidago rigida 0.0000 0.0000 0.0479 0.0006 0.0056 0.0098 Trifolium 0.0006 0.0002 0.0462 0.0030 0.0026 0.0098 pratense Trifolium sp. 0.0026 0.0006 0.0544 0.0020 0.0037 0.0168 Verbascum 0.0000 0.0000 0.0376 0.0067 0.0014 0.0038 blattaria Verbascum 0.0000 0.0000 0.0323 0.0102 0.0022 0.0038 thapsus Verbena stricta 0.0000 0.0000 0.0412 0.0102 0.0022 0.0038 Vernonia 0.0000 0.0000 0.0336 0.0078 0.0003 0.0004 fasciculata

73