The Effect of Urbanization on Butterfly Assemblages in Melbourne, Australia

The effect of urbanization on butterfly assemblages in Melbourne, Australia

By Jessica S. Kurylo

orcid.org/0000-0001-5864-7198

Submitted in total fulfilment of the requirements of the degree of Doctor of Philosophy

School of Ecosystem and Forest Sciences Faculty of Science The University of Melbourne

December 2017

Cover Photo*:

1 – Orange Palm Dart (Cephrenes augiades) 1 2 4 6 2 – Australian Painted Lady (Vanessa kershawi) 3 – Common Brown (Heteronympha merope) 4 – Common Grass Blue (Zizina otis labradus) 7 5 – Cabbage White (Pieris rapae) 5 6 – Yellow Admiral (Vanessa itea) 3 7 – Green Grass Dart (Ocybadistes walker) 8 8 – Imperial Hairstreak (Jalmenus evagoras)

*All photos by J. Kurylo

Abstract

Urbanization is the fastest expanding land-use conversion in the world. Alteration or loss of habitats in urban areas often results in overall biodiversity decline. Insects are integral components of ecosystems, providing important services such as decomposition and pollination. Despite this, we know little about the ecology of insects within urban settings.

Butterflies are well suited for study in urban areas because they are conspicuous components of any landscape and play an important role within the ecosystem. To date, there has been a lack of understanding of the interplay between the ability of urban areas to provide necessary butterfly habitat resources (e.g., adult and larval food plants) and how the presence of those resources affects the butterfly community.

Urban ecology studies are often entrenched in patch-matrix theory, but the use of a gradient paradigm allows for a more mechanistic understanding of the response of species and higher taxa to increasing urbanization. Therefore, this thesis uses a continuous gradient of imperviousness (percent cover of impervious surfaces) to assess how the urban landscape may support the local butterfly community through the availability of their required resources.

This PhD first aimed to quantify the spatial variation in butterfly community structure across a gradient of imperviousness using two measures of habitat quality for adult butterflies and larval stages: native and non-native floral abundance and the availability of remnant vegetation. Approximately 21% of the region’s butterfly species were detected within the study area. Butterfly species richness and abundance and floral abundance declined with increasing imperviousness. The effects of total and native floral abundance measures on butterfly richness and abundance were small, but positive. The availability of remnant vegetation had a small, but positive influence on butterfly species richness.

The second aim was to assess the response of larval host plants (LHPs) over the same imperviousness gradient. Laval host plant richness and cover decreased with increasing imperviousness, but LHP richness in particular had a significant, positive influence on butterfly richness and abundance. Larval host plant cover explained 57% of the variation

i within part of the butterfly community. Common butterflies within the matrix had a significant relationship with some of their specific LHPs.

The third aim was to determine if the local butterfly community derived any benefit from wildlife gardening (resource supplementation or management decisions to benefit wildlife) within urban areas. Comparisons of the butterfly communities across gardens managed for wildlife versus those that are not, revealed that garden type did not influence butterfly responses, but instead led to increased native floral resources in wildlife gardens. This study suggests the landscape context of wildlife gardens may play a role in their effectiveness.

This work demonstrates that the landscape-scale alterations imposed by urbanization have a negative impact upon butterfly communities, and the adult and larval food resources they require. Increases in these resources had some positive effects, but these were mostly small and in some cases only benefitted common butterflies. Management programs aiming to increase habitat for butterflies in urban areas need to consider species-specific requirements and urban context in order to support urban butterfly communities.

Declaration

This is to certify that:

1) This thesis is my original work.

2) Due acknowledgement has been made in the text to all other material used.

3) This thesis is less than 80,000 words in length, exclusive of tables, figures, bibliographies, and appendices.

______Jessica S. Kurylo December 2017

iii

Preface

This PhD thesis consists of five chapters, one of which has been submitted for publication in an international peer-reviewed journal*. Jessica S. Kurylo designed the experiments, collected the data in the field, analysed the data, and wrote the thesis and manuscripts. Co-authors supervised or assisted with various stages of this PhD project and contributed in revising the following manuscript:

Chapter 2 Kurylo, J.S., C.G. Threlfall, K.M. Parris, A. Ossola, N.S.G. Williams, and K.L. Evans. 20XX. Colourful wings in grey landscapes: Butterfly community structure along a gradient of imperviousness. Submitted to Landscape Ecology late Sept 2017.

Results from this PhD have been presented at the following conferences: Kurylo, J.S., C. Threlfall, K. Evans, N.S. Williams. 2015. Do Floral Traits Explain Butterfly Use of Gardens in Urban Landscapes? Oral presentation at the Ecological Society of America Conference in Adelaide, SA. Kurylo, J.S., C. Threlfall, K. Evans, N.S. Williams. 2015. Colourful wings in a grey landscape: Butterfly species richness and abundance along an urban imperviousness gradient. Poster session at the Ecological Society of America Conference in Baltimore, Maryland, USA. Kurylo, J.S., C. Threlfall, K. Evans, N.S. Williams. 2015. Effect of Wildlife Gardens on Butterflies at Landscape and Garden Scales. Oral presentation at the International Association of Landscape Ecologists Conference in Portland, Oregon, USA.

Additional conferences where data were presented, but not directly included in the PhD: Kurylo, J.S., C. Threlfall, K. Evans, N.S. Williams, K. Parris. 2017. Counting flowers in the ‘burbs: how floral richness and abundance changes along an imperviousness gradient. Oral presentation at the Victorian Biodiversity Conference, Melbourne, VIC.

Kurylo, J.S., C. Threlfall, K. Evans, N.S. Williams, K. Parris. 2016. Counting flowers in the ‘burbs: how floral richness and abundance changes along an imperviousness gradient. Poster session at the Ecological Society of America Conference in Fremantle, WA.

*Chapter 2 was submitted to an American journal; for consistency, the thesis has been written in American English.

Acknowledgements

This PhD project was funded through several sources: Melbourne School of Land and Environment Student Awards 2014-2016; Frank Keenan Trust Fund Scholarships 2014 and 2015; and Dr. Nick Williams generously provided some funding in support of Chapter 3. I won a John Hodgson Scholarship in partial support of a lab visit with Dr. Karl Evans at the University of Sheffield, and an ESA Student Travel Grant in 2015; all conference travel was self-funded. I am grateful to have been supported during my candidature through the Melbourne International Research Scholarship (MIRS), Melbourne International Fee Remission Scholarship (MIFRS), and a studentship from the School of Ecosystem and Forest Sciences.

I am grateful to my supervisors Drs. Kirsten Parris, Karl Evans, Nick Williams, and particularly Caragh Threlfall for whom I don’t have enough words to express the extent of my gratitude for her kindness, tutelage, and generosity of time over the last 4 years. Thanks to my panel, Drs. Chris Walsh and Tim New for your guidance and advice. Thanks also to Parks Victoria and to the many golf courses whom allowed me access. Many thanks to the Boroondara City Council, particularly Andrea Lomdahl, coordinator of the Backyard Biodiversity Program, for coordinating access to wildlife gardens participating in their program and to all the home owners whom allowed me access to their gardens. A great multitude of thanks to the Burnley Campus horticultural staff for tireless help with plant ID (John Rayner, Leanne Hanrahan, Dr. Susan Murphy, Jill Kellow, John Delpratt, Jenny Bear, Glenys Rose, and, in particular, Sascha Andrusiak for hours spent sitting through countless pictures of flowers). Helen Vickers, Linda Parker, and Jessica Baumann for sharing their respective plant knowledges. I thank Dr. Samantha Imberger for allowing me to use her lab and the administrative staff (Ross Payne, Vickie Mimis, and all the receptionists) for all the odds and ends they help us with. Thank-you to my amazing field assistants, John Delpratt and Linda Worland, for a job well done and field time well spent.

To my friends and fellow students here in Melbourne, thank-you! Thank-you for conversations, smiles, support, meals, and knowledge shared. Most especially thank-you Sascha, Anne, Carola, Helen, Kate, and Zheng. To the academic staff members (Drs. Claire

Farrell, Chris Szota, and Steve Livesley, in particular) whose ears I’ve bent on frustrating subjects, thank-you for your time and advice. I thank Hayley Lambert and Dr. Craig Nitschke for doing your jobs and helping me navigate a difficult situation. Thank-you to Dr. Dieter Houchuli and his Sydney University lab group for extending your warm hospitality to include me. To my former partner, thank-you. I would not have made it this far without the many conversations, knowledge, encouragement, ideas, adventures, meals, wine, movies, and walks we shared; your culinary support during my first field season was especially invaluable. This past year you reminded me just how strong, determined, and resilient a woman I am. To my family half way around the world, I love you and thank-you for supporting me on this crazy journey.

vii

Table of Contents

Abstract ...... i

Declaration ...... iii

Preface ...... iv

Acknowledgements ...... vi

Table of Contents ...... 8

List of Tables ...... 9

List of Figures ...... 16

Chapter 1 - Introduction ...... 20

Chapter 2 - Colorful wings in grey landscapes: Butterfly community structure along a

gradient of imperviousness ...... 42

Chapter 3 - Butterfly larval host plants within the urban matrix ...... 68

Chapter 4 - Wildlife gardening increases floral resources but not butterfly species

richness or abundance ...... 95

Chapter 5 - Synthesis ...... 125

Appendices Chapter 2 ...... 133

Appendices Chapter 3 ...... 157

Appendices Chapter 4 ...... 164

List of Tables

Table 2-1 – Model averaging table for Question 1 assessing the impact of imperviousness and its scale on the species richness and abundance of butterflies. Abundance-GBSW is the abundance of butterflies that are not Grass Blues and Small Whites. Local imperviousness is the percent imperviousness in the sampling cell, Landscape Imperviousness is the percent imperviousness within a 750 m buffer from the center of the sampling cell. Sampling conditions (temperature, date, and time of day) are reported as a combined value. Sites were sampled twice; spring to mid- summer and mid-summer to fall to account for seasonality of the butterfly species. The parameter estimate and standard error are reported for each predictor term within each model in the top model set. The weighted model average model (Wt. Model Ave.) is highlighted with grey shading, which is also used to highlight response variables for which no models had ∆AICc < 2 relative to the model with the smallest AICc value. The partial r2 for Richness in round 2, and Abundance-GBSW in both sampling rounds is based on the calculation of pseudo r2 (1-(residual / null deviance)). ……………………………………………………………………………………………………………………….………….54

Table 2-2 – Model averaging table for Question 2 assessing the additional impact of floral resources on the species richness and abundance of butterflies. Abundance- GBSW is the abundance of butterflies that are not Grass Blues and Small Whites. Local imperviousness is the percent imperviousness in the sampling cell, Landscape Imperviousness is the percent imperviousness within a 750 m buffer from the center of the sampling cell. Sampling conditions (temperature, date, and time of day) are reported as a combined value. Total floral abundance is the sum of nectar producing flowers along the transect. Native and exotic floral abundance are the respective components of the total. Sites were sampled twice; spring to mid-summer and mid- summer to fall to account for seasonality of the butterfly species. The parameter estimate and standard error are reported for each predictor term within each model in the top model set. The weighted model average model (Wt. Model Ave.) is highlighted with grey shading, which is also used to highlight response variables for which no models had ∆AICc < 2 relative to the model with the smallest AICc value. The partial r2

9 for Richness in round 2, and Abundance-GBSW in both sampling rounds is based on the calculation of pseudo r2 (1-(residual / null deviance)). …..………………………………………….55

Table 3-1 – Larval host plant genera and species surveyed along transects (Field 2013 and Braby 2012). Species with * were found during surveys. Species with ** had <5 occurrences. …………………………………………………………………………………………………………….74

Table 3-2 - Butterfly species identified within the study area across Melbourne, Australia’s eastern suburbs between November 2014 and March 2015 and their total abundance across the both sampling rounds. Known larval host plants from Victoria (Field 2013) and Australia-wide, but not including Victoria (underlined species; Braby 2012) are also listed. Species with an * was targeted for larval host plant surveys and a † indicates 5 or more occurrence records in the 2015/2016 sampling season. …………75

Table 3-3 – Best supported models of the change in larval host plant richness and cover across a gradient of imperviousness. Imperviousness is the percent impervious surfaces within a sampling cell. Date is the calendar day of the vegetation survey. The r2 for richness models is based on the calculation of pseudo r2 (1-(residual / null deviance)). ……………………………………………………………………………………………………………….79

Table 3-4 – Top models for understanding the relationship between the butterfly community and larval host plant richness and % cover. Imperviousness is the % impervious surfaces within a sampling cell. The r2 for total richness is based on the calculation of pseudo r2 (1-(residual / null deviance)). ………………………………………………80

Table 3-5 – The eigenvalues and cumulative variance percentage of RDA for butterflies with >10 occurrences and their larval host plants with >5 occurrences in Melbourne’s south east suburbs. The species scores for the butterflies and biplot scores for the explanatory variables used in this RDA are also presented. ………………………………………83

10 is the % impervious surfaces within a sampling cell. The r2 for richness models is based on the calculation of pseudo r2 (1-(residual / null deviance)). …………………………………..84

Table 4-1 –Mean tree cover (%), butterfly and floral species richness and abundance for each garden type in sampling rounds 1 and 2. There are 54 traditional gardens within areas untargeted for wildlife gardening (CUT), and 27 each wildlife gardens (WG) and control gardens within areas targeted for wildlife gardening (CT). Abundance or richness -GB, B, SW is the abundance or richness of less common butterflies (without the three most common butterflies (Common Grass Blues, ‘blues’, and Small Whites)). …………………………………………………………………………………………………………………106

Table 4-2 - Parameter estimates of the mixed modelling results to assess the difference in floral abundance and richness measures between garden types (WG – wildlife gardens and control gardens in areas targeted (CT) and untargeted for wildlife gardening (CUT)). Intercept of the model of the relationship with wildlife gardens (the reference category) and the other listed predictor variables is reported. Two models were run, first to determine if there was a difference between garden types, and the second to assess if the sampling date and percent imperviousness within a 500 m radius buffer (Imperv) had any effect on the results. ………………………………………………108

Table 4-3 - Parameter estimates of the mixed modelling results for assessing the difference in percent tree cover in a 500 m radius buffer centered on the garden between garden types (WG – wildlife gardens and control gardens in areas targeted (CT) and untargeted for wildlife gardening (CUT)). Intercept of the model of the relationship with wildlife gardens (the reference category) and the other listed predictor variables is reported. The first model was used to determine if there was a difference between garden types and the second to assess if the sampling date and percent imperviousness within a 500 m radius buffer (Imperv) had any effect on the results. ……………………………………………………………………………………………………………………109

The significance of an interaction between garden type and imperviousness within a 500 m radius buffer (Imperv) was first assessed, if not significant the model was rerun sans the interaction. Outcomes for the second model are not presented as they were not different from the first model. Richness and abundance – GB, B, and SW is the total richness or abundance minus abundance of Grass Blues, ‘blues’, and Small White butterflies. Intercept of the model of the relationship with wildlife gardens (the reference category) and the other listed predictor variables is reported. Sampling conditions (Temp -temperature, date, and time of day) were included in both models.

…………………………………………………………………………………………………………………………………111

Appendix 2 Table 1 - – Butterfly flight periods coinciding with our sampling period, plus an additional month on either end, for the butterflies identified during our surveys from November 2014 through March 2015. Figures represent occurrence records indicating the percentage of adults sampled in each month from Field (2013). Gray shading indicates the most likely period to detect the species ‘on the wing’, with dark gray indicating the peak flight period(s) of each species. ……………………………..…133

Appendix 2 Table 2 - All floral resources identified along transects and whether they produce nectar, based on an extensive literature search in Web of Science and Google Scholar (August/September 2016). If no information was found for a species, we assigned a value of yes or no if another species in that genus was known to produce nectar or not, respectively. Because some genera have been poorly studied and no empirical information could be obtained, these have been categorized as no. ………134

Appendix 2 Table 3 – Plots and histograms of the data for three response variables used for modelling for both sampling rounds. Total abundance was logarithmically transformed before use in models. Abundance-GBSW is the abundance of butterflies that are not Grass Blues and Small Whites. ……………………………………………………………..144

Appendix 2 Table 4 - Supplementary Table S3 - Model sets used for each study question. Sampling conditions (SC) are calendar date, temperature, and time of day

(minutes after civil dawn). Base model refers to the top models (∆AICc <2) for each response variable from Question 1. ………………………………………………………………………..142

Appendix 2 Table 5 - Validation plots for the first or zero ranked top model for each question and response variable in Chapter 2. The table number is also provided to reference the predictor variables associated with each of the models represented by each set of plots. …………………………………………………………………………………………………….143

Appendix 2 Table 6 - List of likely butterfly species in Melbourne’s regional species pool. The list was initially drawn from the Atlas of Living Australia (ALA, website, accessed May 2017) covering an extent larger than the study area, generally incorporating the Gippsland Plain Bioregion, but excluding the Dandenong Ranges. The full list was further cross referenced for accuracy (both distribution and habitat) with Field (2013) to develop the list below. Species marked with an asterisk (*) were identified during the round 1 or round 2 surveys (November 2014 - March 2015). While at the study sites we also found two additional species (Imperial Jezebel Delias harpalyce, and an individual that was either a Blotched or Varied Dusky-blue Candalides acasta or C. hyacinthinus). Below this list, we included the map downloaded from the ALA defining the area used to compose the regional species pool. ……………………………………………………………………………………………………………………….150

Appendix 2 Table 7 - Model averaging table for Question 3 assessing the influence of available remnant vegetation on the species richness and abundance of butterflies. Abundance-GBSW is the abundance of butterflies that are not Grass Blues and Small Whites. Local imperviousness is the percent imperviousness in the sampling cell, Landscape Imperviousness is the percent imperviousness within a 750 m buffer from the center of the sampling cell. Sampling conditions are temperature, date, and time of day. Cell remnant is the presence or absence of remnant vegetation within the sampling cell, while buffer remnant is the presence or absence of remnant vegetation within a 750 m buffer from the center of the sampling cell. Sites were sampled twice; spring to mid-summer and mid-summer to fall to account for seasonality of the butterfly species. The parameter estimate and standard error are reported for each predictor term within each model in the top model set. The weighted model average model (Wt. Model Ave.) is highlighted with grey shading, which is also used to highlight response variables for which no models had ∆AICc < 2 relative to the model with the smallest AICc value. The partial r2 for Richness in round 2, and Abundance-GBSW in

13 both sampling rounds is based on the calculation of pseudo r2 (1-(residual / null deviance)). ……………………………………………………………………………………………………………..153

Appendix 2 Table 8 - Table of butterfly species identified within the study area across Melbourne, Australia’s eastern suburbs between November 2014 and March 2015 and their known larval host plants identified from observation in Victoria (Field 2013) or elsewhere in Australia (underlined species; Braby 2012). ……………………………………….155

Appendix 3 Table 1 - Model set to investigate the relationship between the butterfly community and larval host plant (LHP) richness and % cover. Imperviousness is the % impervious surfaces within a sampling cell. ……………………………………………………………161

Appendix 3 Table 2 – Plots and histograms of the data for the response variables. Total abundance was square root transformed, while Abundance-GBSW (the abundance of butterflies that are not Grass Blues and Small Whites) was logarithmically transformed before use in models. ………………………………………………………………………………………………162

Appendix 3 Table 3 - Model set to investigate the relationship between individual butterfly species abundance and their larval host plant (LHP) % cover. A variable combining each of a butterfly species’ LHP, Combined, was included in the modelling to understand if the cumulative effect of the LHP was more important than the individual LHP species. Imperviousness is the % impervious surfaces within a sampling cell. For each butterfly species, models 4-6 were repeated for each individual LHP. Model numbers in Table 5 will not correspond with the numbers in this example. ..163

Appendix 4 Table 1 - All floral resources identified within the gardens and whether or not they produce nectar, based on an extensive literature search in Web of Science and Google Scholar (August/September 2016). Information from the gray literature on nectar production was not used as it is often more anecdotal than empirical. If no information could be found for a species, we assigned a value of yes or no if another species in that genus was known to produce nectar or not, respectively. Some genera have been poorly studied and no empirical information could be obtained, thus we categorized these as a no. Species with an * are known or potential larval host plants for butterflies that occur in the study area (Chapters 1 and 2). ……………………………….168

Appendix 4 Table 2 - Table showing the proportion of gardens (out of 108) where butterfly species were present. Garden types: wildlife gardens (WG) and control gardens in areas targeted (CT) and untargeted for wildlife gardening (CUT). ………….174

Appendix 4 Table 3 - – Parameter estimates of the mixed modelling results for Question 1 assessing the difference in butterfly richness and abundance between garden types (WG – wildlife and control gardens in areas targeted (CT) and untargeted for wildlife gardening (CUT)) in reference to tree cover (%). Wildlife gardens were used as the intercept. The significance of an interaction between garden type and tree cover (%) within a 500 m radius buffer (Tree) was first assessed, if not significant the model was rerun sans the interaction. The last set of results were to assess if the inclusion of tree cover and imperviousness (Imperv, cover (%) within a 500 m radius buffer) within the same model improved model fit. Richness and abundance – GB, B, and SW is the total richness or abundance minus abundance of Common Grass Blues, ‘blues’, and Small White butterflies. Sampling conditions (Temp -temperature, date, and time of day) were included in both models. Intercept of the model of the relationship with wildlife gardens (the reference category) and the other listed predictor variables is reported. Rounds 1 and 2 indicate the 2 sampling rounds for data collection. ……….172

List of Figures

……………………………………………………………………………………………………………………………………………….46

Figure 2-2 – Accumulative butterfly species richness curves for each sampling round. Richness is assessed within each of the five impervious surface categories (Cat 1-5; 100 total transects; 20 per category) used to help evenly stratify the sampling sites across the gradient. …………………………………………………………………………………………..………………..51

Figure 2-3 – Abundance of nectar providing floral species by sampling round and further broken down by native and exotic species. …………………………………………………..52

Figure 2-4 – Predicted relationship between imperviousness and butterfly species richness and butterfly abundance during spring to early-summer (round 1) (A and B) and early-summer to early fall (round 2) (C and D). Three data points from panels A and B (abundance of 288 at 9.44% imperviousness; abundance of 108 at 43.24% imperviousness; abundance of 84 at 25.25% imperviousness) and two points from panels C and D (richness of 8 at 32.93% and 6 at 11.38% imperviousness; abundance of 110 at 10.42% and 96 at 50.95% imperviousness) have been excluded to improve figure clarity. …………………………………………………………………………………………………………….53

Figure 2-5 – Predicted relationship between butterfly abundance and imperviousness when floral resources are included in the model during the spring to mid-summer (round 1) (A and B) and mid-summer to fall (round 2) (C and D). Floral abundance was the sum of mean number of binned (<25, 25-50, 51-100, 101-200, 201-500, 501-2000, and 2001-4000) floral units (raceme, umbel, capitulum, etc.) per cell. Due to unusually high butterfly abundances three data points have been excluded from panels A and B (abundance of 288 at 9.44% imperviousness, 2237 exotic floral abundance (EFA); abundance of 108 at 43.24% imperviousness, 6379 EFA; abundance of 84 at 25.25% imperviousness, 6127 EFA) and two points from panels C and D (abundance of 110 at 9.44% imperviousness, 4719 total floral abundance; abundance (TFA) of 96 at 44.49% imperviousness, 3694 TFA) to improve figure clarity. ……………………………………………….57 16

Figure 3-1 – Location of the sampling cells (500 x 500 m) and the percentage cover of impervious surfaces across the 28 km diameter study area in Melbourne, Australia. A GIS dataset supplied by Melbourne Water was used to measure the impervious- surface cover (Grace 2012, but see Chapter 2 for details). ……………………………………….72

Figure 3-2: Predictive relationships with 95% interval for the response of larval host plant (LHP) cover (% Cover LHP) and species richness (LHP Richness) within a 500 x 500 m sampling cell across the imperviousness gradient. ……………………………………………….78

Figure 3-3: The multiplicative effect on butterfly total richness of the four predictive variables (mean and 95% confidence interval), as predicted by the models they are contained within (see Appendix 3 Table 1 for model set). The horizontal line is at 1. The four predicitve variables are larval host plant (LHP) species richness (a) , LHP cover is the % cover of LHP within the sampled area (b), imperviousness is the % impervious surface cover in the sampling cell (c), and imperviousness^2 is its quadratic form (d).

………………………………………………………………………………………………………………………..…………81

Figure 3-4 - The multiplicative effect on A) total butterfly abundance and B) abundance of less common butterflies (excluding Grass Blues and Small Whites) of the four predictive variables (mean and 95% confidence interval), as predicted by the models they are contained within (see Appendix 3 Table 1 for model sets). The horizontal line is at 1. The four predicitve variables are larval host plant (LHP) species richness (a), LHP cover is the % cover of LHP within the sampled area (b), imperviousness is the % impervious surface cover in the sampling cell (c), and imperviousness^2 is its quadratic form (d). …………………………………………………………………………………………………………………..82

Figure 4-1 – Location of sampling cells (500 x 500 m) and gardens within the 28 km diameter study area in Melbourne, Australia. Grid area represents the Gippsland Plain Bioregion. Inset is one of the cells targeted for wildlife gardening with approximate location of both wildlife and control gardens. ………………………………………………..………100

17 abundance and richness, outcomes were similar in the first sampling round. To improve presentation three data points (Total Floral Abundance: 1894 for CUT and 2134 for WG; Native floral Abundance: 2134 for WG) were excluded. ……………………107

Figure 4-3 – Butterfly species richness and abundance by garden type (control gardens in sampling cells targeted (CT) and untargeted (CUT) for wildlife gardening and wildlife gardens (WG)) from the first sampling round. There was no observable difference in these measures between garden types across either sampling round. Abundance and richness of less common butterflies are the total abundance and richness without the three most common butterflies (Common Grass Blues, ‘Blues’, and Small Whites). 110

Figure 4-4 – Predictive relationships with 95% confidence interval of the interaction between garden type (control gardens in sampling cells targeted (CT) and untargeted (CUT) for wildlife gardening and wildlife gardens (WG)) and imperviousness within the 500 m buffer (%) for the abundance of the less common butterflies from the first sampling round. ……………………………………………………………………………………………………..110

Figure 4-5 – Nonmetric multidimensional scaling (NMDS) ordination plot of the butterfly communities from round 1 (A) and round 2 (B) within each garden type: wildlife gardens (WG, black triangles and gray polygon), control gardens in areas targeted (CT, cyan circles and polygon) and not targeted (CUT, yellow squares and polygon) for wildlife gardening. Identified butterfly species are labelled in red. …….112

Appendix 3 Figure 1 - Diagram depicting subsampling design for larval host plants within sampling cells. ……………………………………………………………………………………………..157

Appendix 3 Figure 2 – Top model validation plots for question 1 about the response of larval host plant richness (A) and cover (%; B) along a gradient of increasing impervious surface cover. ………………………………………………………………………………………………………158

Appendix 3 Figure 3 – Validation plots for question 2 about the response of butterfly species richness (A), total butterfly abundance (B), and the abundance of less common butterflies (total abundance minus Grass Blues and Small Whites; C) to the richness and cover of larval host plants within the study area. In the case of multiple top

18 models, only the first top model’s validation plots are presented. Please see Table 3-4 for more information. ……………………………………………………………………………………….……159

Appendix 3 Figure 4 – Validation plots for question 3 about the response of Grass Blues (A), Small Whites (B), Common Browns (C), and Green Grass Darts to the cover (%) of their larval host plants. In the case of multiple top models, only the first top model’s validation plots are presented. Please see Table 3-6 for more information. 160

Appendix 4 Figure 1 – Validation plots for models presented in Table 4-2 and 4-3 assessing the difference in available resources across different garden types. For brevity, only round 1 models are presented: total floral abundance (A), exotic floral abundance (B), native floral abundance (C), total floral richness (D), exotic floral richness (E), native floral richness (F), and tree cover (%; G). …………………………………..164

Appendix 4 Figure 2 – Validation plots for models presented in Table 4-4 assessing the difference in butterflies across different garden types. For brevity, only round 1 models are presented (total butterfly richness (A), total butterfly abundance (B)). Richness (C) and abundance (D) – GB, B, and SW are the total richness or abundance minus abundance of Grass Blues, ‘blues’, and Small White butterflies. ……………………………..166

Appendix 4 Figure 3 - Nonmetric multidimensional scaling (NMDS) ordination plot of the butterfly communities from round 1 (A) and round 2 (B) within each garden type: wildlife gardens (WG, black triangles and gray polygon), control gardens in areas targeted (CT, cyan circles and polygon) and not targeted (CUT, yellow squares and polygon) for wildlife gardening. Identified butterfly species are labelled in red. …….172

Chapter 1

Impacts of Urbanization on Butterfly Communities

URBANIZATION:

Urban areas house the majority of the world’s human population, and urbanization, the process by which natural areas become part of the built environment, is one of the fastest expanding land uses worldwide (Seto et al. 2013). The effects of urbanization on the local environment are widespread, and often permanent, leading to declines in native wildlife diversity and decreased ecological processes (Pimm and Raven 2000; Pereira and Daily 2006; McKinney 2008; Yen 2011; McDonald et al. 2013). Conceptually, the urbanization process is a combination of human disturbance, primary and secondary biophysical processes, and biological introductions (Parris 2016). In reality, this process is quite complex and dynamic, playing out over space and time at multiple scales (Grimm et al. 2008). Its effects extend beyond the physical border of the built environment through changes in regional and global biogeochemical cycles, altered hydrologic cycles, and demands for natural resources to support continued urban growth (Grimm 2008, Seto et a. 2013).

The biophysical processes that result from urbanization include loss or fragmentation of habitat, climatic changes, environmental pollution, and alteration of hydrologic, light and noise regimes (Parris 2016). This habitat loss and alteration leads to the loss of the dependant organisms, leaving little to no alternative habitat for non-mobile species (Johnson and Klemens 2005; Parris 2016). Habitat fragmentation is less detrimental, but it can lead to habitat degradation and does alter an organism’s ability to move between habitat patches, especially where drastic changes, such as the addition of roads, are made (Johnson and Klemens 2005; Parris 2016). Not only do urban areas create their own climate, the subsequent urban heat island affects the phenology of flowering plants and animals (Parris and Hazzell 2005; Parlow 2011; Villalobos-Jimenez and Hassall 2017). Chemical pollution affects reproduction, growth rates, and community composition (Kagley et al. 2014; Morrissey et al. 2014). Urbanization has far reaching impacts on the local hydrologic cycle, and the hydroperiod changes that result from increased water runoff have a myriad of impacts

20 on aquatic flora and fauna (Walsh et al. 2001; Walsh et al. 2004; Roodsari and Chandler 2017). Lastly, altered light and noise regimes have a broad array of effects on acoustic communication (Da Silva et al. 2015; Caorsi et al. 2017), reproduction (Dominoni et al. 2013; Potvin and MacDougall-Shackleton 2015; McLay et al. 2017), and plant phenology (Ffrench-Constant et al. 2017). Ultimately, these biophysical processes alter the spatial and temporal availability of the resources all forms of life need to survive (Shochat et al. 2006 - trends; Parris 2016; Vaugoyeau et al. 2016).

IMPACTS ON BIODIVERSITY:

Temporally, long term studies have shown species richness decreases over time with increasing urbanization (Stefanescu et al. 2004; Fattorini 2011; Pavez et al. 2010; Strohbach et al. 2014). Spatially, the effect is similar; total species richness decreases with increasing urbanization intensity, but it is also apparent that this trend is taxon, species, and scale dependant (Lepczyk et al. 2008; McKinney 2008; Luck and Smallbone 2010; Nielsen et al. 2014). While the overall community response is typically negative, different species within a community can have neutral or positive responses to increasing urbanization. For example, richness of neither leaf-mining Lepidoptera using remnants (Rickman and Conner 2003), nor crab spiders using open green spaces (Argañaraz and Gleiser 2017) showed a response to surrounding urbanization level. Similarly, total bird species richness was not significantly different between urban and peri-urban areas of three Chilean cities (Paz Silva et al. 2016), while ground dwelling spider richness was highest in urban versus suburban or rural sites (Magura et al. 2010). Some taxa have higher richness at intermediate levels of disturbance (carabid beetles (Tóthmérész et al. 2011; but see Magura et al. 2013); birds (Tratalos et al. 2007; Luck and Smallbone 2010; Conole 2014); millipedes (Bogyo et al. 2015); and vegetation (Zerbe et al. 2003)), in accordance with the intermediate disturbance hypothesis (Connell 1978). This response is thought to be driven by higher resource availability at the interface of rural and suburban areas (Lepczyk et al. 2008; Miserendino et al. 2011).

The response of the abundance or density of organisms is not as well described in the urban ecology literature, but similar to species richness, there are positive, negative, and neutral responses depending on the taxa. For example, abundance of 83 out of 21

132 native bird species responded in some way to increasing urbanization: 33 negatively, 5 positively, and 6 had a responded to intermediate disturbance levels, with the rest having a mixed response to chosen urbanization intensity measures (Lepczyk et al. 2008). Raptor and larger sized native mammal abundance decreased, while small and medium sized mammal abundance did not vary significantly over 30 years with increasing urbanization and vegetation changes near Santiago, Chile (Pavez et al. 2010). Carabid beetles had higher abundance in urban versus rural areas in Romania (Tóthmérész et al. 2011), but, carrion beetles in Poland showed similar abundances across urban and rural sites (Ulrich et al. 2007). These disparate responses of abundance to urbanization require further investigation, as a species abundance is a useful measure of how it responds to disturbance, and may shed light on the ecological impacts of urbanization on biota more broadly.

Native species richness and abundance often decrease at sharper rates than exotic species with increasing urbanization (Luck and Smallbone 2010; Kotze et al. 2011). This trend has been found for vegetation (Kent et al. 1999; Luck and Smallbone 2010; Schmidt et al. 2014; Vakhlamova et al. 2014), birds (Lepczyk et al. 2008, Paz Silva et al. 2016; van Heezik and Adams 2016), fish (Cervantes-Yoshida et al. 2015) and beetles (Kotze et al. 2011). Contrary to this trend, Aronson et al. (2014) found the majority of urban bird and vegetation species are native, but this study excluded planted vegetation and the plant species lists included from the 1950s may no longer be valid. Further, the study was binary, either urban or rural (Aronson et al. 2014), which is why studies with finer spatial resolution along gradients are more informative to examine overall species trends in response to urbanization (McDonnell and Pickett 1990).

Urbanization can impact on the structure of ecological communities, most often a simplification through reductions in species richness. Fish communities in urbanized stream sections changed over 16 years, while the upstream communities in low impact areas were not different (Cervantes-Yoshida et al. 2015). Bird communities and their guild structures differed between urban and peri-urban sites (Paz Silva et al. 2016). Spider communities changed from forest spiders in more rural location to open-habitat spiders in urban areas (Magura et al. 2010). Often these urban communities are considered simplified (Ulrich et al. 2007; Gagné and Fahrig 2011) and dominated by

22 generalist as opposed to specialized species (Gagné and Fahrig 2011; Tóthmérész et al. 2011; Paz Silva et al. 2016). This happens through filtering where the urban environment is suggested to select for species with particular life history or functional traits that make them better adapted to the urban environment (Jokimaki et al. 2014; McDonnell and Hahs 2015; Aronson et al. 2017). Habitat resources made available through the physical layout of an urban area and those supplemented by humans, act as filters within the urban environment (Shochat et al. 2006; Paz Silva et al. 2016; Aronson et al. 2017), resulting in simplified urban communities.

Responses to the availability of a species’ required resources features prominently in the urban ecology literature. For example, no correlation was found between prey and Golden-orb weaving spider (Nephila plumipes) abundances within urban green spaces along an urban rural gradient in Sydney, Australia (Lowe et al. 2017), but spider persistence at sites was closely linked to prey abundance (Lowe et al. 2015). The diet of raptors changed from native rodents to exotic rabbits as scrublands were fragmented and reduced by urbanization near Santiago, Chile (Pavez et al. 2010). Birds responded positively to increased street tree richness, particularly native tree species across three southern Chilean cities (de Castro Pena et al. 2017) and increased nectar availability in Sydney, Australia (Davis et al. 2016). These studies shed light on the types of resources important to many taxa in urban areas, however the literature to date has been biased towards studies of birds and plants, where much less is known about the response of invertebrates, reptiles or fungi.

IMPACTS ON BUTTERFLIES:

Butterflies are promoted as bioindicators for a range factors associated with healthy ecosystems such as pollination services and overall ecosystem functioning (Pe’er and Settele 2008), monitoring the success of restorations (Lomov et al. 2006), and sustainable forest management practices (Maleque et al. 2009). This is due to most species’ close association at all life stages (egg, larvae, adult) with plants as food resources and because they are relatively easy to sample and identify (Dennis 2010; Gerlach et al. 2012). Regardless of landscape setting, butterflies within a community respond to biotic and abiotic landscape elements based on their ecological and life history traits (Dennis 2010; Lizée et al. 2011; Bates et al. 2014). These attributes, close 23 association with plants and response based on life history traits, also make them well suited as a target species to look at community and species responses to increasing urbanisation.

Globally, many studies have assessed how butterflies respond to habitat fragmentation (Kobayashi et al. 2009; Flick et al. 2012; Bossart and Antwi 2013; Soga and Koike 2013), patch size, shape, and quality (Clausen et al. 2001; Collinge et al. 2003; Bergman et al. 2004; Krauss et al. 2005; Haaland and Gyllin 2009; Soga and Koike 2012) or how butterfly abundance or species richness change along a disturbance gradient (Kitahara and Fujii 1994; Hogsden and Hutchinson 2004; Koh and Sodhi 2004; Knapp et al. 2008; Bergerot et al. 2011). Often these studies only consider one or two species and/or take place in more rural settings with low housing and human population densities (Clausen et al. 2001; Flick et al. 2012; Collinge et al. 2003; Bergman et al. 2004; Krauss et al. 2005; Haaland and Gyllin 2009; Bossart and Antwi 2013). While these studies are informative, they do not shed light on the response of butterflies in highly urbanised landscapes.

Studies assessing the effects of urbanisation on butterflies have found that there is a greater species richness and abundance in rural versus urban areas (Blair and Launer 1997; Knapp et al. 2008; Öckinger et al. 2009; Fattorini 2011; Lizée et al. 2011; Dennis et al. 2017). Impervious-surface cover, specifically, has a generally negative effect on butterfly richness and/or abundance (Blair and Launer 1997; Hardy and Dennis 1999; Hogsden and Hutchinson 2004; Stefanescu et al. 2004; Clark et al. 2007; Lizée et al. 2011; Concepción et al. 2015; Olivier et al. 2016). However, studies with better resolution or replication along the gradient have shown a peak in butterfly species richness and abundance at the moderate (or suburban) level of disturbance (Blair and Launer 1997; Blair 1999; Hogsden and Hutchinson 2004; Van Dyck et al. 2009). Some butterflies, such as Small Whites (Cabbage White; Pieris rapae), are more common in urban areas than natural or rural areas (Hardy and Dennis 1999, Kocher and Williams 2000). This is likely related to the wide availability of key resources, particularly their larval host plants, Brassicaceae, in many urban green spaces (Yahner 2001, Leston and Koper 2017). These studies however have mostly used comparisons of the butterfly community across land uses to infer urban impacts. While this is informative, is does

24 not allow for investigation of which resources within these landscapes are driving patterns, nor which resources are important for different life stages of butterflies (larval stages versus adults). Hence, to better understand the response of butterflies to urban land uses, studies incorporating resource gradients are needed.

Resource availability within urban landscapes, particularly residential areas outside of greenspace habitats such as remnants or parks, is likely to influence the butterfly community. Bigger and/or more heterogenous urban green spaces have higher butterfly species richness (Giuliano et al. 2004; Koh and Sodhi 2004; Knapp et al. 2008; Williams 2011; Jarošík et al. 2011; Soga and Koike 2013; Bates et al. 2014; Barranco- León et al. 2016). Floral resources vary in availability over time, and in urban areas tend to persist longer either due to human preferences (Alexio et al. 2014) or other abiotic factors, such as the urban heat island effect, higher soil moisture from domestic watering, or greater nutrient availability (Davis et al. 2016). While exotic plant species richness is greater in urban versus rural areas (Dunn and Heneghan 2011), most butterflies are opportunistic nectar feeders (Hardy and Dennis 1999; Dennis 2010) using a variety of native and non-native floral resources in urban garden studies (Di Mauro et al. 2007; Matteson and Langellotto 2011; Shwartz et. 2013; Olivier et al. 2016). Further, many butterflies use both native and exotic larval host plants (Shapiro 2002; Field 2013). Total vegetation cover, though, tends to decrease with increasing urbanization (Clarkson et al. 2007; Cilliers and Siebert 2011), which in turn reduces available plant resources (either as potential adult or larval plant resources) across the urban landscape. While this research has started to investigate some aspects of habitat quality (e.g. impacts of mowing, patch size, patch heterogeneity) within urban areas, local variables such as adult and larval food resource availability across the urban landscape and their potential synergies have not been fully considered to date.

Wildlife gardening is a way for homeowners within urban and suburban areas to help improve its suitability for wildlife. A wildlife garden is any garden that includes elements to benefit wildlife and can range in form and commitment from planting an entire garden with native plants or planting pollinator-friendly flowers to maintaining a bird feeder or providing logs and rocks for fauna to shelter under (Thompson 2006; Goddard et al. 2010; Salisbury et al. 2015). While these types of gardens are designed

25 to be beneficial to various forms of wildlife, often targeting pollinators, butterflies, or birds, their efficacy has rarely been evaluated in Australia or elsewhere (Smith et al. 2006b; Widows and Drake 2014). Despite this, they are being promoted by local governments and conservation organisations (e.g., Gardens for Wildlife programs sponsored by Whitehorse and Knox City Councils, Victoria, Australia and The Wildlife Trusts in the United Kingdom to the North American Butterfly Association’s region specific butterfly gardening guides). Empirically, there is little known about how butterflies use wildlife gardens. The general perception of a butterflies use of gardens or wildlife gardens is as a nectar source rather than as breeding habitat for most species (Vickery 1995), but little work has been done in this area (Gaston et al. 2005; Di Mauro et al. 2007; Burghardt et al. 2009). While we know that butterflies use gardens (wildlife and non-wildlife gardens), no systematic study of gardens has been conducted to compare the efficacy of one garden type to another, or to a control.

Hence, in this thesis I examine the response of butterflies to urbanization by comparing assemblages across resource gradients within urban and suburban areas, with a specific focus on testing the efficacy of wildlife gardening. The gradient approach used here has not been employed widely to date, and is proposed as a new way of studying responses of wildlife to urbanization.

APPROACHES TO THE STUDY OF URBANIZATION AND ITS IMPACTS ON BIODIVERSITY:

The ecological effects of urbanization have been studied in two ways: firstly, via a focus on the natural areas and green spaces within the metropolitan area, and secondly via a focus on the metropolitan area as a whole (McDonnell and Pickett 1990). The former is a binary landscape view entrenched in patch-matrix theory (Forman 1995; e.g. urban parks and natural areas are considered to be habitat and everything outside these is matrix (non-habitat), more specifically, the urban matrix), while the latter can be explored using a gradient paradigm (Whittaker 1967; McDonnell and Pickett 1990). Taking a binary view of the whole landscape discounts potential available resources within the urban matrix. Within an urban setting, gradients can be broadly (proportion of impervious-surface cover, road or human population density) or more specifically defined (temperature, rainfall, soil moisture) (McDonnell and Hahs 2008; Parris 2016). The idea that environmental variation is 26 ordered spatially underpins the gradient paradigm in ecology, thus the spatial distributions of populations, communities and species are considered the result of particular environmental gradients (Whittaker 1967; McDonnell and Pickett 1990; Parris 2016).

Impervious-surfaces, including paved surfaces and building cover, could be considered one of the more prominent and harshest elements within the urban matrix (Pauleit and Breuste 2011). This is often measured indirectly using population or housing density data readily available from local municipalities. However, as GIS technology has advanced so has the opportunity to directly measure this variable in the landscape (McDonnell and Hahs 2008; Pauleit and Breuste 2011). Yet this opportunity has not yet been widely capitalized upon in urban ecological research.

Many urban ecology studies, including those on butterflies, focus on biodiversity in urban parks and greenspaces, while others consider impervious-surface cover as an explanatory variable, either directly or indirectly measured. However, this is rarely the focus (Bolger et al. 2000; Gibb and Hochuli 2002; Rickman and Conner 2003; Jellinek et al. 2004; Sandström et al. 2006; Garden et al. 2007; Matteson et al. 2008; Jarošík et al. 2011; Williams 2011; Lizée et al. 2012; Tóthmérész et al. 2011; Soga and Koike 2012; Soga et al. 2015; Lizée et al. 2016; Threlfall et al. 2016). Parks and greenspaces may be chosen a priori along an urban-rural gradient, but by default, the sampling locations within the matrix are not evenly spaced across a gradient of impervious-surface cover (see site selection details: Hogsden and Hutchinson 2004; Clark et al. 2007; Bergerot et al. 2011; Konvicka and Kadlec 2011). The use of continuous or more complete gradients would further our mechanistic understanding of the effect of impervious- surface cover (Brady et al. 2009). To my knowledge, no study to date has explicitly set out to examine a continuous and complete gradient of impervious-surface cover (from 0 – 100%) across the urban matrix to understand its impact on urban fauna.

More research is needed on how resources in the urban matrix aid or hinder urban biodiversity. Garden studies such as the Biodiversity in Urban Gardens (BUGS; http://www.bugs.group.shef.ac.uk/) and Biodiversity and Ecosystem Services in Multifunctional Landscapes (BESS; Urban BESS; http://bess-urban.group.shef.ac.uk/) projects in Britain have increased our knowledge of the patterns and drivers for plants 27

(Thompson et al. 2003; Thompson et al. 2004) and insects (Smith et al. 2006a, c) in urban gardens specifically. Other studies also offer insights into the processes and specific habitat values within the urban matrix for insects (Di Mauro et al. 2007; Lerman and Milan 2016; Olivier et al. 2016), shrews (Vergnes et al. 2013), and birds (Lerman and Warren 2011; Davis et al. 2017). Similar to studies of urban greenspace, most garden studies have an uneven distribution and/or limited number of sites along a gradient of urbanisation (Di Mauro et al. 2007; Smith et al. 2006a, c). Residential garden areas alone comprise around a third of the urban matrix (between 22 % in the United Kingdom (Loram et al. 2007) to 36 % in Dunedin, New Zealand (Mathieu et al. 2007), therefore greater understanding of habitat quality values and urban wildlife response to the urban matrix is needed.

LOCAL CONTEXT – THE BUTTERFLIES OF MELBOURNE:

The deleterious effects of urbanization on butterfly populations in Australia were documented as far back as 1897, when Waterhouse (1897) noted the once plentiful Bank’s Brown (Heteronympha banksii) had become rare upon settlement of Mosman Bay in Sydney. The Action Plan for Australian Butterflies identified urbanization and its related processes as concerns or threats for 40 of Australia’s butterfly species (Sands and New 2002, New and Sands 2002). This list includes Eltham copper (Paralucia pyrodiscus lucida) from Melbourne, which will require management in perpetuity to survive in its small, isolated urban habitats (New and Sands 2002). Relatively few studies have investigated the effects of urbanization on Australia’s invertebrates (Yen 2011), and there has been no systematic assessment of how butterfly faunas have responded to urbanization in Australia.

The greater Melbourne metropolitan area covers six diverse biogeographic regions (Hahs et al. 2009; Conn 1993) and supports 95 butterfly species (Field 2013). The western portion of the Gippsland Plain Bioregion covers the south-eastern Melbourne metropolitan area and supports approximately 66 butterfly species (see Chapter 2; Appendix 2 Table 4). Many of the local governments across this part of Melbourne have biodiversity strategies (Bundoora, Manningham, Whitehorse), and two, the cities of Boroondara and Knox, have active wildlife gardening programs to help them achieve a goal of increasing urban biodiversity. The programs attract local residents with an 28 interest in making their gardens more wildlife friendly by offering free advice and plants. Both programs have been successful in recruiting many local residents, but neither has evaluated their effectiveness in reaching the local governments’ biodiversity goals.

THIS THESIS:

The overall aim of my PhD was to evaluate the response of the butterfly community to biotic and abiotic elements within the urban matrix of the greater Melbourne area. Despite much research on the conservation of Australia’s butterflies, there has been limited examination of urban butterfly communities. We know little of their composition and which ecological or landscape factors are either encouraging persistence in the urban matrix or driving decline. There is also a general lack of information about the factors that influence the suitability of the matrix in urban areas not just for Australian butterflies, but for most urban butterflies worldwide. To date, few studies have included the urban matrix in their experimental design, despite making suggestions about how it should be managed (Sweany et al. 2014).

My PhD therefore specifically aims to:

1) Quantify the spatial variation in butterfly community structure along an impervious-surface gradient (as a proxy for urbanization) as a function of: a) the availability of native and non-native floral resources; and b) the distribution and cover of butterfly larval host plants. 2) Assess the distribution and cover of butterfly larval host plants along an impervious-surface gradient. 3) Determine if the local butterfly community benefits from wildlife gardening within the urban matrix.

The outcomes of my PhD will not only contribute to the theoretical development of urban ecology, it will also assist land managers, conservationists, and homeowners to conserve butterfly communities in urban environments.

THESIS OUTLINE:

Chapter 2 reports on a landscape-scale study assessing the richness and abundance of the butterfly community in south-eastern Melbourne. For this study, I surveyed 100 sampling cells twice during the flight season, counting butterflies and flowers along the same transect. These sites were evenly distributed along a continuous gradient of impervious-surface cover, to develop a more mechanistic understanding of the impact of impervious-surface cover on the butterfly community. Here, I tested the prediction that impervious-surface cover reduced the richness and abundance of butterflies, but that certain habitat resources could potentially mitigate those impacts. Thus, this study also includes an examination of the relative importance of floral abundance (as adult food resources) and mapped remnant areas (primarily as a proxy for available native larval host plants). Outcomes address Aim 1 of the thesis.

Chapter 3 describes a finer-scale, targeted survey for larval host plants within the urban matrix. The aim of this study was to better understand how larval host plants were distributed across the urban matrix, and how their distribution and cover affected the species richness and abundance of butterflies. I selected 60 of my 100 cells based on those with the highest and lowest butterfly abundances from Chapter 2. I then surveyed along the same transect for the larval host plants of the butterfly species identified during fieldwork for Chapter 2. Butterfly surveys were also conducted within these cells using the same protocols as in Chapter 2. Here, I tested the prediction that larval host plant presence or absence within the urban matrix influences the butterfly community structure. Outcomes of this chapter address Aim 2 of the thesis.

Chapter 4 was designed to assess the benefit of wildlife gardening within the urban matrix for butterflies. The study focused on a local council’s wildlife gardening program that encouraged residents, within a certain proximity to a stream-restoration project completed by the council, to create wildlife gardens on their property with the intention of creating wider green corridors than the available council right-of-way. One hundred and eight gardens, including 27 wildlife gardens and 81 control gardens (27 within areas targeted for wildlife gardening and 54 from areas not targeted for wildlife gardening), were recruited for survey. I conducted point counts for butterflies in each 30 of the gardens twice over the flight season. The species richness and abundance of butterflies was then compared between the garden types and between areas targeted and not targeted by the wildlife gardening program. Here, I tested the assumption that wildlife gardening improves the availability of habitat for butterflies in the urban matrix. Outcomes of this chapter address Aim 3 of the thesis.

Chapter 5 synthesizes this research, including its aims and insights. It also includes suggestions for further research and management.

REFERENCES:

Aleixo KP, de Faria LB, Groppo M, Castro MMdN, da Silva CI (2014) Spatiotemporal distribution of floral resources in a Brazilian city: Implications for the maintenance of pollinators, especially bees. Urban Forestry & Urban Greening 13:689-696 Argañaraz C, Gleiser RM (2017) Does urbanization have positive or negative effects on Crab spider (Araneae: Thomisidae) diversity? Zoologia 34:1-8 Aronson MF et al. (2014) A global analysis of the impacts of urbanization on bird and plant diversity reveals key anthropogenic drivers. Proceedings: Biological Sciences 281:20133330 Aronson MFJ et al. (2017) Biodiversity in the city: key challenges for urban green space management. Frontiers in Ecology and the Environment 15:189-196 Bates AJ et al. (2014) Garden and landscape scale correlates of moths of differeing conservation status: signifcant effects of urbanization and habitat diversity. PLoS One 9:e86925 Bergerot B, Fontaine B, Julliard R, Baguette M (2010) Landscape variables impact the structure and composition of butterfly assemblages along an urbanization gradient. Landscape Ecology 26:83-94 Bergerot B, Fontaine B, Julliard R, Baguette M (2011) Landscape variables impact the structure and composition of butterfly assemblages along an urbanization gradient. Landscape Ecology 26:83-94 Bergman K-O, Askling J, Ekberg O, Ingnell H, Wahlman H, Milberg P (2004) Landscape effects of butterfly assemblages in an agriculutral region. Ecography 27:619-628 Blair RB (1999) Birds and butterflies along an urban gradient: surrogate taxa for assessing biodiversity. Ecological Applications 9:164-170 Blair RB, Launer AE (1997) Butterfly diversity and human land use: species assemblages along an urban gradient. Biological Conservation 80:113-125

Bogyó D, Magura T, Simon E, Tóthmérész B (2015) Millipede (Diplopoda) assemblages alter drastically by urbanisation. Landscape and Urban Planning 133:118-126 Bolger DT, Suarez AV, Crooks KR, Morrison SA, Case TJ (2000) Arthropods in urban habitat fragments in Southern California: Area, Age, and Edge Effects. Ecological Applications 10:1230-1248 Bossart JL, Antwi JB (2012) Species-specific traits predict genetic structure but not genetic diversity of three fragmented Afrotropical forest butterfly species. Conservation Genetics 14:511-528 Brady MJ, McAlpine CA, Miller CJ, Possingham HP, Baxter GS (2009) Habitat attributes of landscape mosaics along a gradient of matrix development intensity: matrix management matters. Landscape Ecology 24:879-891 Burghardt KT, Tallamy DW, Shriver WG (2009) Impact of native plants on bird and butterfly biodiversity in suburban landscapes. Conservation Biology 23:219-224 Caorsi VZ, Both C, Cechin S, Antunes R, Borges-Martins M (2017) Effects of traffic noise on the calling behavior of two Neotropical hylid frogs. PLoS One 12:e0183342 Cervantes-Yoshida K, Leidy RA, Carison SM (2015) Contemporary land changes alters fish communities in a San Francisco Bay watershed, California, USA. PLoS One 10:e0141707 Cilliers SS, Siebert SJ (2011) Urban flora and vegetation: patterns and processes. In: Niemelä J (ed) Urban Ecology: patterns, processess, and applications. Oxford University Press, New York, pp 148-158. Clarkson BD, Wehi PM, Brabyn LK (2007) A spatial analysis of indigenous cover patterns and implications for ecological restoration in urban centres, New Zealand. Urban Ecosystems 10:441-457 Clausen HD, Holbeck HB, Reddersen J (2001a) Factors influencing abundance of butterflie and burnet moths in uncultivated habitats of an organic farm in Denmark. Biological Conservation 98:167-178 Clausen HD, Holbeck HB, Reddersen J (2001b) Factors influencing abundance of butterflies and burnet moths in the uncultivated habitatshabitats of an organic farm in Denmark. Biological Conservation 98:167-178 Collinge SK, Prudic KL, Oliver JC (2003) Effects of local habitat characteristics and landscape context on grassland butterfly diviersity. Conservation Biology 17:178-187 Concepción ED, Moretti M, Altermatt F, Nobis MP, Obrist MK (2015) Impacts of urbanisation on biodiversity: the role of species mobility, degree of specialisation and spatial scale. Oikos 124:1571-1582

Conn BJ (1993) Natural regions and vegetation of Victoria vol 1. Inkata Press, Melbourne, Australia. Connell JH (1978 ) Diversity in tropical rain forests and coral reefs. Science 199:1302-1310 Conole LE (2014) Degree of adaptive response in urban tolerant birds shows influence of habitat-of-origin. PeerJ 2:e306 Da Silva A, Valcu M, Kempenaers B (2015) Light pollution alters the phenology of dawn and dusk singing in common European songbirds. Philosophical transactions of the Royal Society of London Series B, Biological sciences 370 Davis A, Major RE, Taylor CE (2016) Do trees flower longer in the city? A comparison of flowering eucalyptus trees in streets, remnants and continuous forest and their association with nectarivorous birds. Urban Ecosystems 19:735-747 De Castro Pena JC, Martello F, Ribeiro MC, Armitage RA, Young RJ, Rodrigues M (2017) Street trees reduce the negative effects of urbanization on birds. PLoS One 12:e0174484 de las Nieves Barranco-Leon M, Luna-Castellanos F, Vergara CH, Badano EI (2016) Butterfly conservation within cities: a landsaccpe scale approach integrating natural habitats and abandonded fields in central Mexico. Tropical Conservation Science 9:607-628 Dennis EB, Morgan BJT, Roy DB, Brereton TM (2017) Urban indicators for UK butterflies. Ecological Indicators 76:184-193 Dennis RLH (2010) A resource-based habitat view for conservation: butterflies in the Brittish landscape. 1 edn. John Wiley & Sons Ltd., United Kingdom Di Mauro D, Dietz T, Rockwood L (2007) Determining the effect of urbanization on generalist butterfly species diversity in butterfly gardens. Urban Ecosystems 10:427-439 Dominoni D, Quetting M, Partecke J (2013) Artificial light at night advances avian reproductive physiology. Proceedings: Biological Sciences 280:1-8 Dunn CP, Heneghan L (2011) Composition and diversity of urban vegetation. In: Niemela J (ed) Urban Ecology Patterns, Processess, and Applications. Oxford University Press, New York, pp 103-114 Fattorini S (2011a) Insect extinction by urbanization: A long term study in Rome. Biological Conservation 144:370-375 Fattorini S (2011b) Insect rarity, extinction and conservation in urban Rome (Italy): a 120-year- long study of tenebrionid beetles. Insect Conservation and Diversity 4:307-315 Ffrench-Constant RH, Somers-Yeates R, Bennie J, Economou T, Hodgson D, Spalding A, McGregor PK (2016) Light pollution is associated with earlier tree budburst across the United Kingdom. Proceedings: Biological Sciences 283

Field R (2013) Butterflies: Indentification and life history. Museum Victoria Publishing, Melbourne, Victoria Flick T, Feagan S, Fahrig L (2012) Effects of landscape structure on butterfly species richness and abundance in agricultural landscapes in eastern Ontario, Canada. Agriculture, Ecosystems & Environment 156:123-133 Forman RT (1995) Land mosaics: the ecology of landscapes and regions. Cambridge University Press, New York Gagné SA, Fahrig L (2011) Do birds and beetles show similar responses to urbanization? Ecological Applications 21:2297-2312 Garden JG, McAlpine CA, Possingham HP, Jones DN (2007) Habitat structure is more important than vegetation composition for local-level management of native terrestrial reptile and small mammal species living in urban remnants: A case study from Brisbane, Australia. Austral Ecology 32:669-685 Gaston KJ, Smith RM, Thompson K, Warren PH (2005) Urban Domestic Gardens (II): experimental tests of methods for increasing biodiversity. Biodiversity and Conservation 14:395-413 Gerlach J, Samways M, Pryke J (2013) Terrestrial invertebrates as bioindicators: an overview of available taxonomic groups. Journal of Insect Conservation 17:831-850 Gibb H, Hochuli DF (2002) Habitat fragmentation in an urban environment: large and small fragments support different arthropod assemblages. Biological Conservation 106:91- 100 Giuliano WM, Accamando AK, McAdaams EJ (2004) Lepidoptera-habitat relationships in urban parks. Urban Ecosystems 7:361-370 Goddard MA, Dougill AJ, Benton TG (2010) Scaling up from gardens: biodiversity conservation in urban environments. Trends Ecol Evol 25:90-98 Grimm NB, Faeth SH, Golubiewski NE, Redman CL, Wu J, Xuemei B, Briggs JM (2008) Global change and the ecoogy of cities. Science 319:756-760 Haaland C, Gyllin M (2009) Butterflies and bumblebees in greenways and sown wildflower strips in southern Sweden. Journal of Insect Conservation 14:125-132 Hahs A, McDonnell MJ, Holland K, Caryl F (2009) Biodiversity of Metropolitan Melbourne. Australian Research Centre for Urban Ecology, Melbourne, Australia. Hardy PB, Dennis RLH (1999) The impact of urban development on butterflies within a city region. Biodiversity and Conservation 8:1261-1279 Hogsden KL, Hutchinson TC (2004) Butterfly aeemblages along a human disturbance gradients in Ontario, Cannada. Canadian Journal of Zoology 82:739-748

Jarošík V, Konvička M, Pyšek P, Kadlec T, Beneš J (2011) Conservation in a city: Do the same principles apply to different taxa? Biological Conservation 144:490-499 Jellinek S, Drixcoll DA, Kirkpatrick JB (2004) Environmental and vegetation variables have a greater influence thatn habitat fragmentation in structuring lizard communities in remnant urban bushland Austral Ecology 29:294-304 Johnson EA, Klemens MW (2005) The impacts of sprawl on biodiversity. In: Johnson EA, Klemens MW (eds) Nature in fragments: the legacy of sprawl. Columbia University Press, New York, pp 18-53 Jokimäki J, Suhonen J, Jokimäki-Kaisanlahti ML, Carbó-Ramírez P (2014) Effects of urbanization on breeding birds in European towns: Impacts of species traits. Urban Ecosystems 19:1565-1577 Kagley AN, Kardong KE, Snider RG, Casillas E (2014) Effects of chemical contaminants on growth, age-structure, and reproduction of Mytilus edulis complex from Puget sound, Washington. Bulletin of EnvironmentalCcontamination and Toxicology 93:7-12 Kent M, Stevens RA, Zhang L (1999) Urban plant ecolgy patterns and processes: a case study of the flora of the City of Plymouth, Devon, U.K. . Journal of Biogeography 26:1281-1298 Kitahara M, Fujii K (1994) Biodiverity and community structure of temerate butterfly species within a gradient of human disturbance: an analysis based on the concept of generalist vs. specialist strategies. Population Ecology 36:187-199 Knapp S, Kühn I, Mosbrugger V, Klotz S (2008) Do protected areas in urban and rural landscapes differ in species diversity? Biodiversity and Conservation 17:1595-1612 Kobayashi T, Nakashizuka T, Kitahara M (2009) Effects of fragmentation of secondary broadleaf deciduous forests on populations of the near-threatened butterfly, Sasakia charonda (Lepidoptera, Nymphalidae), in central Japan. Ecological Research 24:57-64 Kocher SD, Williams EH (2000) The diversity and abundance of North American butterflies vary with habitat disturbance and geography. Journal of Biogeography 27:789-794 Koh LP, Sodhi NS (2004) Importance of reserves, fragments, and parks for butterfly conservation in a tropical urban landscape. Ecological Applications 14:1695-1708 Konvicka M, Kadlec T (2011) How to increase value of urban areas for butterfly conservation? A lesson from Prague nature reserved and parks. European Journal of Entomology 108:219-229 Kotze J, Venn S, Niemela J, Spence J (2011) Effects of urbanization on the ecology and evolution of arthropods. In: Niemela J (ed) Urban Ecology Patterns, Processes, and Applications. Oxford University Press, New York, pp 159-166.

Krauss J, Steffan-Dewenter I, Muller CB, Tscharntke T (2005) Relative importance of resource quantity, isolation and habitat quality for landscape distribution of a monophagous butterfly. Ecography 28:465-474 Lepczyk CA, Flather CH, Radeloff VC, Pidgeon AM, Hammer RB, Liu J (2008) Human impacts on regional avian diversity and abundance. Conserv Biol 22:405-416 Lerman SB, Milam J (2016) Bee fauna and floral abundance within lawn-dominated suburban yards in Springfield, MA. Annals of the Entomological Society of America 109:713-723 Lerman SB, Warren PS (2011) The conservation value of residential yards: linking birds and people. Ecological Applications 21:1327-1339 Leston L, Koper N (2017) Urban rights-of-way as extensive butterfly habitats: A case study from Winnipeg, Canada. Landscape and Urban Planning 157:56-62 Lizée M-H, Manel S, Mauffrey J-F, Tatoni T, Deschamps-Cottin M (2012) Matrix configuration and patch isolation influences override the species–area relationship for urban butterfly communities. Landscape Ecology 27:159-169 Lizée M-H, Mauffrey J-F, Tatoni T, Deschamps-Cottin M (2011a) Monitoring urban environments on the basis of biological traits. Ecological Indicators 11:353-361 Lizée M-H, Tatoni T, Deschamps-Cottin M (2016) Nested patterns in urban butterfly species assemblages: respective roles of plot management, park layout and landscape features. Urban Ecosystems 19:205-224 Lizée MH, Bonardo R, Mauffrey JF, Bertaudiere-Montes V, Tatoni T, Deschamps-Cottin M (2011b) Relative importance of habitat and landscape scales on butterfly communities of urbanizing areas. Comptes Rendus Biologies 334:74-84 Lomov B, Keith DA, Britton DR, Hochuli DF (2006) Are butterflies and moths useful indicators for restoration monitoring? A pilot study in Sydney's Cumberland Plain Woodland. Ecological Management & Restoration 7:204-210 Loram A, Tratalos J, Warren PH, Gaston KJ (2007) Urban domestic gardens (X): the extent & structure of the resource in five major cities. Landscape Ecology 22:601-615 Lowe EC, Wilder SM, Hochuli DF (2016) Persistence and survival of the spider Nepila plumippes in cities: do increased prey resources drive the success of an urban exploiter. Urban Ecosystems 19:705-720 Lowe EC, Wilder SM, Hochuli DF (2017) Life history of an urban-tolerant spider shows resilience to anthropogenic habitat disturbance. Journal of Urban Ecology 3:1-10 Luck GW, Smallbone LT (2010) Species diversity and urbanisation: patterns, drivers, and implications. In: Gaston KJ (ed) Urban Ecology. Cambridge University Press, New York, pp 88-119

Magura T, Horváth R, Tóthmérész B (2010) Effects of urbanization on ground-dwelling spiders in forest patches, in Hungary. Landscape Ecology 25:621-629 Magura T, Nagy D, Tóthmérész B (2013) Rove beetles respond heterogeneously to urbanization. Journal of Insect Conservation 17:715-724 Maleque MA, Maeto K, Ishii HT (2009) Arthropods as bioindicators of sustainable forest management, with a focus on plantation forests. Applied Entomology and Zoology 44:1-11 Mathieu R, Freeman C, Aryal J (2007) Mapping private gardens in urban areas using object- oriented techniques and very high-resolution satellite imagery. Landscape and Urban Planning 81:179-192 Matteson KC, Ascher JS, Langellotto GA (2008) Bee richness and abundance in New York City urban gardens. Annals of the Entomological Society of America 101:140-150 Matteson KC, Langellotto GA (2011) Small scale additions of native plants fail to increase beneficial insect richness in urban gardens. Insect Conservation and Diversity 4:89-98 McDonald RI, Marcotuillio PJ, Güneralp B (2013) Urbanization and global trends in biodiveristy and ecosystem services. In: Elmqvist T (ed) Urbanization, Biodiversity and Ecosystem Services: challenges and opportunities. Springer, New York, pp 31-52. McDonnell MJ, Hahs AK (2008) The use of gradient analysis studies in advancing our understanding of the ecology of urbanizing landscapes: current status and future directions. Landscape Ecology 23:1143-1155 McDonnell MJ, Hahs AK (2015) Adaptation and Adaptedness of Organisms to Urban Environments. Annual Review of Ecology, Evolution, and Systematics 46:261-280 McDonnell MJ, Pickett STA (1990) Ecosystem structure and function along urban-rural gradients: an unexpected opportunity for ecology. Ecology 71:1232-1237 McKinney ML (2008) Effects of urbanization on species richness: A review of plants and animals. Urban Ecosystems 11:161-176 McLay LK, Green MP, Jones TM (2017) Chronic exposure to dim artificial light at night decreases fecundity and adult survival in Drosophila melanogaster. Journal of Insect Physiology 100:15-20 Miserendino ML, Casaux R, Archangelsky M, Di Prinzio CY, Brand C, Kutschker AM (2011) Assessing land-use effects on water quality, in-stream habitat, riparian ecosystems and biodiversity in Patagonian northwest streams. The Science of the Total Environment 409:612-624

Morrissey CA, Stanton DW, Tyler CR, Pereira MG, Newton J, Durance I, Ormerod SJ (2014) Developmental impairment in eurasian dipper nestlings exposed to urban stream pollutants. Environ Toxicol Chem 33:1315-1323 New T, Sands DPA (2002) Conservation concerns for butterflies in urban areas of Australia. Journal of Insect Conservation 6:207-215 Nielsen AB, van den Bosch M, Maruthaveeran S, van den Bosch CK (2013) Species richness in urban parks and its drivers: A review of empirical evidence. Urban Ecosystems 17:305- 327 Öckinger E, Dannestam Å, Smith HG (2009) The importance of fragmentation and habitat quality of urban grasslands for butterfly diversity. Landscape and Urban Planning 93:31-37 Olivier T, Schmucki R, Fontaine B, Villemey A, Archaux F (2016) Butterfly assemblages in residential gardens are driven by species’ habitat preference and mobility. Landscape Ecology 31:865-876 Parlow E (2011) Urban Climate. In: Niemela J (ed) Urban Ecology: patterns, processes, and applications. Oxford University Press, New York, pp 31-45 Parris KM (2016) Ecology of Urban Environments. John Wiley & Sons, Chichester, UK Parris KM, Hazell DL (2005) Biotic effects of climate change in urban environments: The case of the grey-headed flying-fox (Pteropus poliocephalus) in Melbourne, Australia. Biological Conservation 124:267-276 Pauleit S, Breuste JH (2011) Land-use and surface-cover as urban ecological indicators. In: Niemelä J (ed) Urban Ecology: patterns, processess, and applications. Oxford University Press, New York, pp 19-29 Pavez EF, Lobos GA, Jaksic FM (2010) Long term changes in landscape and in small mammal and raptor assemblages in Central Chile. Revista Chilena De Historia Natural 83:99-111 Paz Silva C, Sepulveda RD, Barbosa O (2016) Nonrandom filtering effect on birds: species and guilds response to urbanization. Ecology and Evolution 6:3711-3720 Pe'er G, Settele J (2008) The rare butterfly Tomares nesimachus (Lycaenidae) as a bioindicator for pollination services and ecosystem functioning in northern Israel. Israel Journal of Ecology & Evolution 54:111-136 Pereira HM, Daily GC (2006) Modeling biodiversity dynamics in countryside landscape. Ecology 87:1877-1885 Pimm SL, Raven P (2000) Extinctin by the numbers. Nature 403:843-845

Potvin DA, MacDougall-Shackleton SA (2015) Traffic noise affects embryo mortality and nestling growth rates in captive zebra finches. Journal of Experimental Zoology Part A, Ecological genetics and physiology Rickman JK, Connor EF (2003) The effect of urbanization on the quality of remnant habitats for leaf-mining lepidoptera on Quercus agrifolia. Ecography 26:777-787 Roodsari BK, Chandler DG (2017) Distribution of surface imperviousness in small urban catchments predicts runoff peak flows and stream flashiness. Hydrological Processes 31:2990-3002 Salisbury A, Armitage J, Bostock H, Perry J, Tatchell M, Thompson K, Diamond S (2015) EDITOR'S CHOICE: Enhancing gardens as habitats for flower-visiting aerial insects (pollinators): should we plant native or exotic species? Journal of Applied Ecology 52:1156-1164 Sands DPA, New TR (2002) The Action Plant for Australian Butterflies. Environment Australia, Canberra Sandström UG, Angelstam P, Mikusiński G (2006) Ecological diversity of birds in relation to the structure of urban green space. Landscape and Urban Planning 77:39-53 Schmidt KJ, Poppendieck H-H, Jensen K (2014) Effects of urban structure on plant species richness in a large European city. Urban Ecosystems 17:427-444 Seto KC, Parnell S, Elmqvist T (2013) A global outlook on urbanization. In: Elmqvist T (ed) Urbanization, biodiversity, and ecosystem services: challenges and opportunities. Springer, New York, pp 1-12 Shapiro AM (2002) The California urban butterfly fauna is dependant on alien plants. Diversity and Distributions 8:31-40 Shochat E, Warren PS, Faeth SH, McIntyre NE, Hope D (2006) From patterns to emerging processes in mechanistic urban ecology. Trends in Ecology & Evolution 21:186-191 Shwartz A, Muratet A, Simon L, Julliard R (2013) Local and management variables outweigh landscape effects in enhancing the diversity of different taxa in a big metropolis. Biological Conservation 157:285-292 Smith RM, Gaston KJ, Warren PH, Thompson K (2006a) Urban domestic gardens (VIII) : environmental correlates of invertebrate abundance. Biodiversity and Conservation 15:2515-2545 Smith RM, Thompson K, Hodgson JG, Warren PH, Gaston KJ (2006b) Urban domestic gardens (IX): Composition and richness of the vascular plant flora, and implications for native biodiversity. Biological Conservation 129:312-322

Smith RM, Warren PH, Thompson K, Gaston KJ (2006c) Urban domestic gardens (VI): environmental correlates of invertebrate species richness. Biodiversity and Conservation 15:2415-2438 Soga M et al. (2015) Landscape versus local factors shaping butterfly communities in fragmented landscapes: does host plant diversity matter? Journal of Insect Conservation 19:781-790 Soga M, Koike S (2012) Life-history traits affect vulnerability of butterflies to habitat fragmentation in urban remnant forests. Ecoscience 19:11-20 Soga M, Koike S (2013) Patch isolation only matters for specialist butterflies but patch area affects both specialist and generalist species. Journal of Forest Research 18:270-278 Stefanescu C, Herrando S, Pάramo F (2004) Butterfly species richness in the north-west Mediterranean Basin: the role of natural and human-induced factors. Journal of Biogeography 31:905-915 Strohbach MW, Hrycyna A, Warren PS (2014) 150 years of changes in bird life in Cambridge, Massachusetts from 1860 to 2012. The Wilson Journal of Ornithology 126:192-206 Sweaney N, Lindenmayer DB, Driscoll DA (2014) Is the matrix important to butterflies in fragmented landscapes? Journal of Insect Conservation 18:283-294 Thompson K (2006) No nettles required: the truth about wildlife gardening. Transworld Publishers, London Thompson K, Austin KC, Smith RM, Warren PH, Angold PG, Gaston KJ (2003) Urban domestic gardens (I): putting small-scale plant diveristy in context. Journal of Vegetation Science 14:71-78 Thompson K, Hodgson JG, Smith RM, Warren PH, Gaston KJ (2004) Urban domestic gardens (III): Composition and diversity of lawn floras. Journal of Vegetation Science 15:373- 378 Threlfall CG, Williams NSG, Hahs AK, Livesley SJ (2016) Approaches to urban vegetation management and the impacts on urban bird and bat assemblages. Landscape and Urban Planning 153:28-39 Tóthmérész B, Máthé I, Balázs E, Magura T (2011) Responses of carabid beetles to urbanization in Transylvania (Romania). Landscape and Urban Planning 101:330-337 Tratalos J, Fuller RA, Warren PH, Davies RG, Gaston KJ (2007) Urban form, biodiversity potential and ecosystem services. Landscape and Urban Planning 83:308-317 Ulrich W, Zalewski M, Komosiński K (2007) Diversity of carrion visiting beetles at rural and urban sites. Coomunity Ecology 8:171-181

Vakhlamova T, Rusterholz H-P, Kanibolotskaya Y, Baur B (2014) Changes in plant diversity along an urban–rural gradient in an expanding city in Kazakhstan, Western Siberia. Landscape and Urban Planning 132:111-120 Van Dyck H, Van Strien AJ, Maes D, Van Swaay CAM (2009) Declines in Common, Widespread Butterflies in a Landscape under Intense Human Use. Conservation Biology 23:957-965 van Heezik Y, Adams AL (2014) Vulnerability of native and exotic urban birds to housing densification and changing gardening and landscaping trends. Urban Ecosystems 19:1551-1563 Vaugoyeau M et al. (2016) Interspecific variation in the relationship between clutch size, laying date and intensity of urbanization in four species of hole-nesting birds. Ecology and Evolution 6:5907-5920 Vergnes A, Kerbiriou C, Clergeau P (2013) Ecological corridors also operate in an urban matrix: A test case with garden shrews. Urban Ecosystems 16:511-525 Villalobos-Jimenez G, Hassall C (2017) Effects of the urban heat island on the phenology of Odonata in London, UK. International Journal of Biometeorology 61:1337-1346 Walsh CJ, Sharpe AK, Breen PF, Sonneman JA (2001) Effects of urbanization on streams of the Melbourne region, Victoria, Australia. I. benthic macroinvertebrate communities. Freshwater Biology 46:535-551 Waterhouse GA (1897) The genus Heteronympha in New South Wales. Proceedings of the Linnean Society of New South Wales 22:240-243 Whittaker RH (1967) Gradient analysis of vegetation. Biological Reviews 42:207-264 Widows SA, Drake D (2014) Evaluating the National Wildlife Federation's Certified Wildlife Habitat™ program. Landscape and Urban Planning 129:32-43 Williams MR (2011) Habitat resources, remnant vegetation condition and area determine distribution patterns and abundance of butterflies and day-flying moths in a fragmented urban landscape, south-west Western Australia. Journal of Insect Conservation 15:37-54 Yahner RH (2001) Butterfly communities in residential landscape of Central Pennsylvania. Northeastern Naturalist 8:113-118 Yen A (2011) Melbourne's terrestrial invert biodiveristy; losses, gains, and the new perspective. The Victorian Naturalist 128:201-208

Chapter 2

Colorful wings in grey landscapes: Butterfly community structure along a gradient of imperviousness

ABSTRACT:

Heterogeneity in the quantity and quality of resources provided in the urban matrix may moderate relationships between urbanization intensity and the structure of biotic communities. To assess this I quantified the spatial variation in butterfly community structure along an impervious surface gradient using two measures of urban matrix quality: native and non-native floral resource availability and the occurrence of remnant habitat patches. Butterfly communities were surveyed in 100 cells (500 x 500 m), selected using a random-stratified sampling design, across a continuous gradient of imperviousness in Melbourne, Australia. Sampling occurred twice during the butterfly flight season and data were analysed using generalized linear models at local and landscape scales. My results showed a decline in butterfly species richness and abundance with increasing impervious surface cover. Two-thirds of the region’s butterfly species were absent or scarce from within the study area. The effects of floral resources and remnant vegetation were small, but positive. Total butterfly abundance responded positively to total floral abundance early in the sampling season and negatively to exotic floral abundance later in the sampling season. In summary, landscape imperviousness reduced both species richness and abundance of butterfly communities. However, the provision of floral resources and conservation of remnant vegetation within urban landscapes may mitigate some of the negative effects of urbanization on butterfly communities by improving habitat quality of the urban matrix.

INTRODUCTION:

Urban areas are amongst the fastest expanding land cover types worldwide (Seto et al. 2013), decreasing ecological integrity and driving local extinctions and population decline across many species groups (Aronson et al. 2017). Native species richness and abundance generally decrease with increasing urbanization, but this trend varies

42 across species (depending on their ecological and life history requirements; Dennis 2010, Driscoll et al. 2013), location (due to variation in the precise nature of the urban gradient and city level characteristics; Norton et al. 2016) and spatial scale of analysis (Savard et al. 2000; Luck and Smallbone 2010).

When applying simple ecological frameworks such as the patch-matrix model (Forman 1995) to urban areas, large areas of green-space (e.g., parks) are often viewed as habitat patches embedded in an inhospitable environment dominated by impervious surfaces, the ‘urban matrix’. Percent cover of impervious surface, from here termed ‘imperviousness’, is becoming a common measure of urbanization intensity within the matrix because it better reflects permanent land change unlike other proxies such as distance to city center, population density, road density, etc. (McDonnell and Hahs 2008). It is increasingly recognised that the urban matrix, when assessed at fine spatial scales, is highly heterogeneous in terms of its vegetation composition, structure, and management, and thus its ability to support biodiversity (Thompson et al. 2004; Threlfall et al. 2016; Norton et al. 2016). Urban ecological studies to date, however, have largely focused on understanding factors determining the quality of large patches of green spaces with far less attention given to factors underlying fine scale variation in the habitat quality within the matrix (Sadler et al. 2010).

Butterflies can exploit small disjunct patches of habitat due to their high mobility and small body size, but are also sensitive to spatial and temporal variation in resource availability (Lütolf et al. 2009; Pohl et al. 2011; Ibbe et al. 2011). Butterflies thus provide a suitable taxon for investigating urbanization impacts on biodiversity, particularly their responses to features of the urban matrix and spatial scales (Concepción et al. 2015). Recent studies suggest that butterflies respond negatively to increased urban development (Olivier et al. 2016; Ramirez-Restrepo and MacGregor- Fors 2017). While these and other previous studies have provided useful information, most have focused on butterfly community responses within large patches of green space (Williams 2009, Lizée et al. 2011; Chong et al. 2014; Sing et al. 2016). As such, they provide incomplete information on changes in butterfly communities across the entirety of the urban matrix, or which features of the matrix influence these responses.

Nectar availability is a key resource that can determine butterfly abundance (Dennis 2010; Curtis et al. 2015). Urban areas contain a wide range of non-native plant species, which can be locally abundant (Loram et al. 2008; Threlfall et al. 2016). While butterflies readily use both native and exotic flowers (Bergerot et al. 2010; Dennis 2010; Jain et al. 2016), I am not aware of any comprehensive assessment of how flowering plant origin influences urban butterfly communities, although such effects have been documented for some bee guilds (Hanley et al. 2014; Pardee and Philpott 2014; Threlfall et al. 2015). Larval host plants are another key resource that can structure butterfly communities and their responses to environmental change (Curtis et al. 2015; Soga et al. 2015). Remnant native vegetation patches, which may contain abundant larval host plants and native floral resources, may thus provide patches of higher quality habitat in urban areas – the benefits of which could spill over to influence butterfly communities within the urban matrix. Further, most butterflies have known habitat affinities (e.g., grassland or woodland; Dennis 2010) and will use a matrix that is structurally similar (i.e., woodland butterflies key-in on trees), particularly if food plants are available (Lütolf et al. 2009; Ibbe et al. 2011; Öckinger et al. 2012; Soga and Koike 2012). Therefore it is important to understand the potential mitigating effects of the availability of these features within the urban matrix when assessing the impacts of urbanization on the butterfly community.

Here, I quantify how butterfly species richness and abundance vary along a gradient of imperviousness, in Melbourne, Australia. My aim was to provide a mechanistic understanding of the effects of imperviousness on the butterfly community and to investigate whether matrix quality attributes mitigate those effects. Specifically, I asked: 1) Does the butterfly community respond more strongly at the local or landscape scale to a gradient of imperviousness? 2) Does the provision of native and non-native floral resources mediate the effect of imperviousness on the butterfly community? 3) Does the occurrence of remnant native habitat influence the butterfly community? I expect the butterfly community to be negatively impacted by impervious-surface cover with a greater impact at the local scale. I expect the provision of floral resources to have a positive impact on butterfly abundance in particular. I also expect that the presence of native remnant vegetation will provide a

44 wide range of resources (including larval host plants), thus providing a positive impact on the butterfly community, particularly species richness.

METHODS:

Study Area:

I conducted this study within the eastern suburbs of Melbourne, Australia’s second largest city with approximately 4.5 million residents. Melbourne’s greater metropolitan area lies across four bioregions. To minimize variation in biophysical properties (e.g., soil type, climate) and vegetation communities, the study area was restricted to the Gippsland Plain Bioregion which is dominated by a variety of grassy woodland and heathland vegetation types (Hahs et al. 2009). The 28 km-diameter study area, centered in the Boroondara local government area (latitude = -37.829967° S, longitude = 145.071481° E), contains a representative mosaic of residential areas with small to large residential parcels, several local urban centers with higher human population densities, intensively managed sports fields, small pocket parks, and mixed-use woodland reserves and parklands along the Yarra River.

Site Selection:

Using ArcMap 10.2 (ESRI, Redlands, CA), a grid of 500 m x 500 m cells was generated over the study area. Grid cell imperviousness was calculated using the total impervious surface cover data from a GIS dataset supplied by Melbourne Water (Grace 2012). This dataset maps all the impervious surfaces (e.g., roads, roofs, sidewalks) within Melbourne’s greater metropolitan area classified by using infra-red aerial imagery at a 0.5 m resolution. Imperviousness within the grid cells ranged from 2 – 97% across the study area. Twenty cells for each of five imperviousness categories: (0-20%, 20-40%, 40-60%, 60-80%, and 80-100%) were randomly selected giving a total of 100 cells (Fig. 2-1), in which imperviousness ranged from 2 – 94%.

Butterfly Sampling:

I surveyed butterfly abundance and species richness within each cell along a 1 km

Figure 2-1 – Percentage cover of impervious surfaces and the location of the sampling cells across the 28 km diameter study area in Melbourne, Australia. Impervious-surface cover was measured using a GIS dataset supplied by Melbourne Water (Grace 2012). transect using a modified Pollard Walk, a standard butterfly surveying technique (5 x 5 m sampling box, 50 m/min pace; Pollard 1977; Collier et al. 2006). Transect routes were selected along accessible streets, trails or footpaths and were randomly selected to cover all major land uses within each cell (e.g., industrial, residential, and greenspace) in relation to their relative abundance. Transects were kept as continuous as possible within cells, though in a few instances transects were segmented due to access restrictions. All butterflies seen within the sampling box along each transect were recorded and identified to species when possible (using photographs or capture and release with a hand net). All identifications followed Field (2013) and were conducted by myself to allow consistency in identifications and avoid double counts. In limited cases (0.03 %), butterflies crossed the transect too quickly to be correctly identified to species and were classified to family level (i.e., blues (Lycaenidae), darts (Hesperiidae), whites (Pieridae), or browns (Nymphalidae)). These individuals were included in abundance calculations, but only contributed to the species richness counts when no other species of that family were identified on that transect.

Butterfly surveys were conducted between 09:00 and 17:30 when weather conditions were most favourable for butterfly activity, i.e., air temperatures between 13 and

34°C, wind speed <10 km/hr, and cloud cover <60%. Surveys were conducted twice during the austral butterfly flight season to account for seasonal variation in the butterfly community (Appendix 2 Table 1; round 1: 8 November 2014 to 22 January 2015 (late spring into mid-summer); round 2: 23 January to 22 March 2015 (mid- summer into early fall); Williams 2011; Field 2013). At each transect, the first and second survey rounds were conducted at least 50 days apart to avoid counting the same individuals (most individual butterflies live for less than 30 days (Pyle 1992; Orr and Kitching 2010)).

Environmental variables:

Three of my four environmental variables of interest were calculated at the local scale (500 x 500 m cell, i.e., the spatial resolution butterflies were sampled) and larger landscape scale (750 m circular buffer around the centre point of each sampling cell) to assess how butterfly response varied between scales. Daily movement data for Australian butterflies is largely unknown. The 750 m radius threshold was arbitrarily chosen, but it is larger than the daily movement distance of Small White (Pieris rapae) (250-600 m; Jones et al. 1980), a non-native member of the local butterfly community. Landscape scale variables are thus likely to influence the structure of butterfly communities by influencing the quality of the habitat through which butterflies could move, while local scale variables are more likely to influence butterfly community structure by determining local resource availability.

The amount (ha) of remnant natural vegetation was calculated from a native vegetation GIS layer (‘Native Vegetation – Modelled Extent 2005’ supplied by Department of Environment, Land, Water & Planning) mapped at a resolution of 12.5 x 12.5 m based on existing maps, ground truthing, and expert validation. Despite being 10 years old, this is the most recent map of remnant native vegetation of the study area. During fieldwork it became apparent that a number of these mapped remnants had been lost to development. Thus, we conducted additional ground truthing and validation using Google Earth aerial imagery taken within 5 years of my sampling to subsequently drop remnant vegetation polygons that had been lost to development.

Most studies sample only a small portion of a site for floral abundance (median area of site assessed = 0.69%; Szigeti et al. 2016), but unlike more rural or natural settings, the spatial variability of vegetation within the urban matrix is large (Thompson et al. 2004). Thus, to address relationships between floral abundance and butterflies the former was measured along the entire transect within the same butterfly sampling box (covering approximately 2% of the grid sampling cell plus the vertical extent up to 5 m). Floral surveys were completed within two days of the butterfly surveys, with 90% of them being completed on the same day. Flowering forbs, shrubs, and trees along the transect route were identified to species where possible (otherwise to genus or family); we did not try to identify cultivars or varieties (Thompson et al. 2004). Grasses were excluded a priori since they do not produce nectar. Of the 546 taxa of flowering forbs, shrubs, and trees 404 were identified to species and 142 to genus.

The number of floral units on each flowering plant, i.e., raceme, umbel, capitulum, etc., were recorded in seven categories (<25, 25-50, 51-100, 101-200, 201-500, 501- 2000, and 2001-4000 floral units) adapting methods from Feber et al. (1996) and Carvell et al. (2006). These data were used to calculate floral resources available to foraging butterflies by using the mid-values of each category to calculate the total number of floral units using only data from species that produce nectar or possess nectaries. We did this as nectar is the dominant food source for all the butterfly species detected during my surveys (Orr and Kitching 2010; Field 2013). Species were classified as producing nectar based on data for that species or genus obtained through extensive literature searches in Web of Science and Google Scholar (conducted in August and September 2016; Appendix 2 Table 2). We then calculated total floral abundance and that of native and exotic species, defining exotic species as those with known origins outside Australia and its islands.

Data analysis:

Butterfly species richness and abundance for each cell was compared between rounds 1 and 2 using a matched paired t-test. Butterfly species richness and abundance was then modelled as a function of environmental variables by constructing separate models for the two sampling rounds. Two butterfly species – Common Grass Blue (Zizina otis) (GB) and the non-native Small White (a.k.a., Cabbage White, Pieris rapae) 48

(SW) – were very widespread (occurring in 95% of all cells across both rounds) and were, often by an order of magnitude, the two most common butterfly species in a cell. We thus calculated total species richness and abundance with and without these two species. Species richness including and excluding GB and SW were highly correlated with each other (round 1: Spearman’s r = 0.89; round 2: Spearman’s r = 0.82) so we constructed models for three response variables: total species richness, total abundance, and abundance excluding GB and SW. Prior to modelling, all data were checked for spatial autocorrelation using the package ‘ape’ (Paradis et al. 2016) in R 3.2.1 (R Core Development Team 2015). For most response variables there was no evidence for spatial autocorrelation, and in all other cases Moran’s I values were extremely small and negative (richness round 2: Moran’s I = -0.042; abundance of less common species round 1: Moran’s I = -0.038 and round 2 = -0.057), and as such had little effect on my models.

Butterfly species richness in round 1 was normally distributed, but richness in round 2 had a non-Gaussian distribution and was modelled as a Poisson distribution using a generalised linear model with a log link. Total butterfly abundance, from both sampling rounds, had a Gaussian distribution following logarithmic transformation. Abundance of the less common species (excluding GB and SW) is highly skewed and therefore modelled as a negative binomial distribution using a generalised linear model with a log link. Plots and histograms of the response variable data are provided in Appendix 2 Table 3. All modelling was run in R, and the ‘MASS’ package (Venables and Ripley 2002) was used for negative binomial models.

My set of predictor variables were: i) sampling conditions, i.e., sampling date (with 21 June (austral winter equinox) as day one), time of day (minutes after sun-rise - defined as civil dawn) and temperature (°C), ii) floral abundance (i.e., total, native, and exotic floral abundance), all of which were square root transformed to improve normality, iii) impervious surface (percentage cover at the local (500 m cell) and landscape scales (750 m buffer), and iv) presence/absence of native remnant vegetation (at local and landscape scales); we used presence/absence due to insufficient variation in the amount of remnant vegetation. In addition, we quantified variation in tree cover at both scales, however this variable was highly correlated with imperviousness

(Spearman’s r = -0.80 (local scale) and -0.76 (landscape scale)) and was subsequently removed from the analysis. Given the disparity in ranges, all predictor variables except the presence of remnant vegetation, were standardised prior to analysis by centering and scaling using the ‘scale’ function in R.

I used an information theoretic approach to enable multi-model inference (Burnham and Anderson 2002). All possible models were constructed for each of my three key research questions (Appendix 2 Table 4). Due to strong collinearity between imperviousness at the local and landscape scales (Spearman’s r = 0.82) I did not include both variables in the same model. Similarly, total floral abundance and exotic abundance were highly correlated in both rounds (round 1 = 0.89, round 2 = 0.97); total floral abundance was also correlated with native abundance in round 1 = 0.79 and hence, these were not included in the same models. Collinearity among other variables were less than Spearman’s r = 0.65.

Linear and quadratic terms for each of my key predictor variables were used to account for non-linear relationships (Appendix 2 Table 4). Question 1 assessed the butterfly community response to imperviousness; models included imperviousness and sampling conditions. When addressing the second and third questions, all variables retained in the top models from Question 1 (defined as ∆AICc <2, and referred to as the base model) were included to assess whether floral abundance and presence/absence of remnant vegetation improved the fit of the models relative to models that only included the amount of imperviousness. For each question I ranked models using the Akaike Information Criterion corrected for small sample size (AICc) generated using the ‘AICcmodavg’ (Mazerolle 2016) R package. In cases where many

‘top’ models resulted, I conducted model averaging over all models within two ∆AICc points of the model with the smallest AICc to calculate model-averaged parameter estimates, their associated unconditional standard errors, and model-averaged partial r2 values. Model averaging was conducted by setting a parameter estimate and partial r2 for a predictor as zero if it was not present in a given model. Validation plots were run in R and visually inspected for all top models. The plots for the first top models (e.g., the zero model from the AIC ranking) are provided in Appendix 2 Table 5.

RESULTS: 50

Overall, 14 butterfly species were detected in the 100 cells, 10 were found in both sampling rounds, with only one exotic species (Small White, SW) being detected. Detected species represent 21.21% (14 of 66 species) of the total non-vagrant butterfly species within the regional species pool (Appendix 2 Table 6). Mean total species richness per cell (± standard error) was 2.67 ± 0.12 (round 1) and 2.7 ± 0.17 (round 2).

These differences were not statistically significant (t= -0.18 1, 99, p = 0.86). The most abundant species were GB and SW which comprised 91.00% of the 3037 individual butterflies counted in round 1, and 91.80% of the 1834 individuals counted in round 2. Butterflies were significantly more abundant in round 1 than round 2 (mean abundance round 1: 30.37 ± 3.26 individuals, round 2: 18.34 ± 1.86, t = 4.54 1, 99, P < 0.001; mean abundance excluding GB and SW round 1: 2.72 ± 0.50, round 2: 1.94 ± 0.30, P = 0.06). My sampling efforts across the study area were adequate in capturing the majority of the urban butterfly community as indicated by the species accumulation curves (Fig. 2-2). Total floral abundance was 40% higher in round 1 (Fig.2-3). Round 1 Round 2 12 12

10 10

8 8

6 6

4 4

2 2

Accumulative Richness Accumulative Accumulative Richness Accumulative 0 0 0 5 10 15 20 0 5 10 15 20

Question 1: Effects of impervious surfaces and scale dependency

I found consistent evidence that increased imperviousness at local and landscape scales reduced butterfly species richness and abundance in both sampling rounds. These effects had a greater explanatory power than the sampling conditions, i.e., date,

time of day, and temperature (Table 2-1). The Floral Abundance relative strength of local and landscape scale 800000 600000 effects was not consistent between rounds, except 400000 with regard to abundance excluding GB and SW in 200000 which case local scale effects consistently had 0 Round 1 Round 2 greater explanatory capacity (Table 2-1). Species Native 294755 56486 Exotic 508351 259880 richness declined linearly as local or landscape Figure 2-3 – Abundance of nectar scale impervious surface increased (Table 2-1; Fig. providing floral species by sampling round and further broken down by 2-4). There was stronger evidence for a quadratic native and exotic species. relationship between abundance and percentage impervious cover, especially in round 1, with limited variance in abundance when impervious cover was less than 25%, but strong declines in abundance above this threshold (Table 2-1; Fig. 2-4).

Question 2: Effects of floral resources

Incorporating floral abundance measures improved the fit of species richness and abundance models compared to that achieved when only using impervious surface cover, especially when modelling total abundance (cf. Tables 2-1 & 2-2). Butterfly species richness increased with total and native floral abundances in both rounds (Table 2-2). In round 1, exotic floral abundance was also positively associated with species richness, but explanatory power was consistently limited (Table 2-2).

Total butterfly abundance was negatively associated with exotic floral abundance in round 1 (Fig. 2-5, panel B), and positively associated with total floral abundance in round 2 (Table 2-2; Fig. 2-5, panel D). Abundance of butterflies excluding GB and SW were negatively correlated with exotic floral abundance in round 1, but all floral abundance metrics had negligible influence on this abundance measure in round 2 (Table 2-2). Effects of impervious surface cover on butterfly richness and abundance when taking floral abundance into account remained similar to those that arose when not taking floral abundance into account, although there were some small reductions in explanatory capacity (cf. Table 2-1 & 2-2; Fig. 2-5, panels A, C).

Question 3: Effects of remnant vegetation

The presence of remnant vegetation at the local scale increased butterfly species richness in round 2, but had little or no effect on species richness in round 1 or the butterfly abundance measures (Appendix 2 Table 7).

A 5 B 70 60 4 50 3 40

2 30 20

Species Richness Species 1 10

0 Abundance Butterfly Total 0 0 20 40 60 80 100 0 20 40 60 80 100 Local Imperviousness (%) Landscape Imperviousness (%)

C 5 D 70 60 4 50 3 40

2 30 20

Species Richness Species 1 10

0 Abundance Butterfly Total 0 0 50 100 0 20 40 60 80 100 Landscape Imperviousness (%) Local Imperviousness (%)

Figure 2-4 – Predicted relationship between imperviousness and butterfly species richness and butterfly abundance during spring to early-summer (round 1) (A and B) and early-summer to early fall (round 2) (C and D). Three data points from panels A and B (abundance of 288 at 9.44% imperviousness; abundance of 108 at 43.24% imperviousness; abundance of 84 at 25.25% imperviousness) and two points from panels C and D richness of 8 at 32.93% and 6 at 11.38% imperviousness; abundance of 110 at 10.42% and 96 at 50.95% imperviousness) have been excluded to improve figure clarity.

Table 2-1 – Model averaging table for Question 1 assessing the impact of imperviousness and its scale on the species richness and abundance of butterflies. Abundance- GBSW is the abundance of butterflies that are not Grass Blues and Small Whites. Local imperviousness is the percent imperviousness in the sampling cell, Landscape Imperviousness is the percent imperviousness within a 750 m buffer from the center of the sampling cell. Sampling conditions (temperature, date, and time of day) are reported as a combined value. Sites were sampled twice; spring to mid-summer and mid-summer to fall to account for seasonality of the butterfly species. The parameter estimate and standard error are reported for each predictor term within each model in the top model set. The weighted model average model (Wt. Model Ave.) is highlighted with grey shading, which is also used to highlight response variables for which no models had ∆AICc < 2 relative to the model with the smallest AICc value. The partial r2 for Richness in round 2, and Abundance-GBSW in both sampling rounds is based on the calculation of pseudo r2 (1-(residual / null deviance)).

Response Sampling Model Local Imperviousness Landscape Imperviousness Variable ∆AICc Conditions r2 2 2 Linear term Quadratic term Partial r Linear term Quadratic term Partial r Partial r2 Richness 0 0.219 -0.471 ± 0.088 -0.156 ± 0.092 0.219 0.775 0.219 -0.455 ± 0.089 0.219

0.798 0.204 -0.455 ± 0.086 0.204

Wt. Model Ave. 0.211 -0.332 ± 0.032 -0.067 ± 0.004 0.152 -0.130 ± 0.013 0.006

Abundance 0 0.239 -0.443 ± 0.086 -0.207 ± 0.074 0.239 Round1 Abundance 0 0.275 -0.906 ± 0.157 0.275 - GBSW 1.531 0.339 -0.886 ± 0.154 0.254 0.064 Wt. Model Ave. 0.319 -0.892 ± 0.137 0.261 0.043 Richness 0 0.171 -0.223 ± 0.066 0.171 0.751 0.275 -0.215 ± 0.066 0.275 1.964 0.173 -0.215 ± 0.068 0.023 ± 0.056 0.173 Wt. Model Ave. 0.167 -0.072 ± 0.053 0.053 -0.147 ± 0.010 0.004 ± 0.000 0.114

Abundance 0 0.250 -0.415 ± 0.088 -0.285 ± 0.094 0.208 0.104 0.634 0.255 -0.438 ± 0.088 -0.181 ± 0.077 0.212 0.093

Wt. Model Ave. 0.252 -0.240 ± 0.030 -0.165 ± 0.017 0.120 -0.185 ± 0.022 -0.076 ± 0.005 0.090 0.099 Round2 Abundance 0 0.350 -0.995 ± 0.158 0.350 - GBSW 0.426 0.347 -0.977 ± 0.160 0.347 0.816 0.361 -0.903 ± 0.168 0.193 ± 0.159 0.361 1.946 0.388 -0.996 ± 0.154 0.350 0.037 Wt. Model Ave. 0.357 -0.692 ± 0.163 0.045 ± 0.003 0.253 -0.283 ± 0.059 0.098 0.005

Table 2-2 – Model averaging table for Question 2 assessing the additional impact of floral resources on the species richness and abundance of butterflies. Abundance-GBSW is the abundance of butterflies that are not Grass Blues and Small Whites. Local imperviousness is the percent imperviousness in the sampling cell, Landscape Imperviousness is the percent imperviousness within a 750 m buffer from the center of the sampling cell. Sampling conditions (temperature, date, and time of day) are reported as a combined value. Total floral abundance is the sum of nectar producing flowers along the transect. Native and exotic floral abundance are the respective components of the total. Sites were sampled twice; spring to mid-summer and mid-summer to fall to account for seasonality of the butterfly species. The parameter estimate and standard error are reported for each predictor term within each model in the top model set. The weighted model average model (Wt. Model Ave.) is highlighted with grey shading, which is also used to highlight response variables for which no models had ∆AICc < 2 relative to the model with the smallest AICc value. The partial r2 for Richness in round 2, and Abundance-GBSW in both sampling rounds is based on the calculation of pseudo r2 (1-(residual / null deviance)).

Response Landscape Total Floral Local Imperviousness Native Floral Abundance Exotic Floral Abundance Sampling Variable Model Imperviousness Abundance ∆AICc Conditions r2 Linear Quadratic Partial Linear Quadratic Partial Linear Quadratic Partial Linear Quadratic Partial Linear Quadratic Partial Partial r2 term term r2 term term r2 term term r2 term term r2 term term r2 Richness -0.457 ± 0.188 -0.171 0 0.275 0.211 0.071 0.085 ± 0.087 ± 0.057 -0.528 ± 0.032 -0.204 1.120 0.267 0.263 0.062 0.088 ± 0.088 ± 0.068 -0.425 ± 0.306 -0.168 1.291 0.265 0.178 0.061 0.086 ± 0.103 ± 0.064

-0.468 ± 0.090 -0.082 0.077 -0.042 0.009 -0.056 Wt. Model Ave. 0.270 0.217 0.034 0.017 0.015 0.052 ± 0.006 ± 0.005 ± 0.005 ± 0.002 ± 0.000 ± 0.003 Abundance -0.469 ± -0.004 -0.420 -0.196

Round1 0 0.484 0.212 0.245 0.073 ± 0.069 ± 0.083 ± 0.056 Abundance -0.890 ± -0.357 -0.178 0.087 0 0.394 0.353 0.055 -GBSW 0.151 ± 0.155 ± 0.120 -0.852± -0.357 0.281 0.374 0.333 0.035 0.080 0.148 ± 0.143 -0.872 ± -0.357 -0.095 Wt. Model Ave. 0.384 0.344 0.046 0.084 0.146 ± 0.059 ± 0.010

Richness -0.215 ± 0.118 0 0.219 0.149 0.048 0.068 ± 0.065 -0.211 ± 0.127

Round2 0.241 0.215 0.145 0.056 0.068 ± 0.065

-0.208 ± 0.119 0.189 0.308 0.247 0.222 0.074 0.068 ± 0.070 ± 0.086 -0.238 ± 0.101 0.916 0.205 0.180 0.034 0.070 ± 0.067 -0.191 ± 0.067 0.144 0.942 0.237 0.167 0.064 0.069 ± 0.060 ± 0.069 -0.223 ± 1.074 0.171 0.066 -0.235 ± 0.109 1.306 0.199 0.174 0.040 0.070 ± 0.067 -0.215 ± 1.825 0.159 0.066 -0.072 ± -0.144 0.026 0.051 0.058 Wt. Model Ave. 0.212 0.051 0.119 0.019 0.025 0.004 ± 0.010 ± 0.001 ± 0.002 ± 0.018 Abundance -0.418 ± 0.115 0.533 -0.161 0 0.508 0.187 0.216 0.061 0.074 ± 0.082 ± 0.097 ± 0.069 -0.394 ± 0.181 0.593 -0.172 1.902 0.499 0.178 0.211 0.061 0.075 ± 0.112 ± 0.110 ± 0.070 -0.110 ± 0.050 -0.301 ± 0.083 0.550 -0.164 Wt. Model Ave. 0.506 0.050 0.135 0.215 0.061 0.010 ± 0.003 0.038 ± 0.008 ± 0.075 ± 0.022 Abundance -0.996 ± 0 0.350 0.350 -GBSW 0.159 -0.997 ± 0.426 0.347 0.347 0.160 -0.923 ± -0.803 0.239 0.558 0.383 0.289 0.036 0.158 ± 0.141 ± 0.129 -0.903 ± 0.193 ± 0.816 0.361 0.361 0.168 0.159 -0.946 ± -0.141 1.501 0.356 0.298 0.009 0.163 ± 0.150 -0.956 ± -0.096 1.673 0.354 0.296 0.004 0.165 ± 0.150 -0.996 ± 1.946 0.388 0.350 0.037 0.154 -0.529 ± 0.028 ± -0.433 ± -0.038 0.040 Wt. Model Ave. 0.361 0.160 0.142 0.007 0.031 0.080 0.001 0.109 ± 0.003 ± 0.002

A B 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0

Total Butterfly Abundance Butterfly Total 0 50 100 0 50 100 Total Butterfly Abundance Butterfly Total Landscape Imperviousness (%) Local Imperviousness (%)

C C D 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0

0 5000 10000 15000 20000 0 10000 20000 30000

Total Butterfly Abundance Butterfly Total Total Butterfly Abundance Butterfly Total Exotic Floral Abundance Total Floral Abundance

Figure 2-5 – Predicted relationship between butterfly abundance and imperviousness when floral resources are included in the model during the spring to mid-summer (round 1) (A and B) and mid- summer to fall (round 2) (C and D). Floral abundance was the sum of mean number of binned (<25, 25- 50, 51-100, 101-200, 201-500, 501-2000, and 2001-4000) floral units (raceme, umbel, capitulum, etc.) per cell. Due to unusually high butterfly abundances three data points have been excluded from panels A and B (abundance of 288 at 9.44% imperviousness, 2237 exotic floral abundance (EFA); abundance of 108 at 43.24% imperviousness, 6379 EFA; abundance of 84 at 25.25% imperviousness, 6127 EFA) and two points from panels C and D (abundance of 110 at 9.44% imperviousness, 4719 total floral abundance; abundance (TFA) of 96 at 44.49% imperviousness, 3694 TFA) to improve figure clarity.

DISCUSSION:

Approximately 21% of butterfly species in the regional species pool were detected during my transect surveys. Excluding the five species in the regional pool found mostly on tree tops, which would likely be poorly sampled through the Pollard walk, and accounting for additional species documented during floral surveys within my transects (one imperial jezebel (Delias harpalyce, a tree top dwelling species) and an

57 individual that was either a blotched or varied dusky-blue (Candalides acasta or C. hyacinthinus)) my detection rate was closer to 26% (16 of 61 species). Remaining undetected species are either rare or locally absent, strongly suggesting that many species have not survived as the study area has become urbanised. Studies of urban parks found from 5% of regional species (60 species from 10 parks in Kuala Lumpur over 1 year of about 793 species from peninsular Malaysia; Sing et al. 2016), to 49% of species from Adelaide, Australia (surveyed 4 parks over 2 years; Collier et al. 2006) to 89% (35 of 39 species detected in 46 urban and peri-urban remnants over 5 years in Perth, Australia; Williams 2009). Additionally, a study encompassing a broader landscape view found at least 45% (27/60) of butterfly species in Rome have been lost to urbanization since 1900 (Fattorini 2011). The proportion of regional species that I detected in my study area is at the low end of the range reported in these studies. Increased sampling intensity or additional field seasons in future studies may increase detection rates (Westphal et al. 2008; Hughes et al. 2017). Regardless, my study still suggests there has likely been a significant loss of butterfly species from areas of Melbourne that are now urbanised.

Response to Imperviousness

My results demonstrate that some butterfly species, including native (GB) and non- native (SW) species, can occur at relatively high densities in the urban matrix, despite most species occurring at very low densities. Urban butterfly studies typically, but not invariably, find that species richness and total abundance decline with increasing urbanization intensity (Ramirez-Restrepo and MacGregor-Fors 2017). Results from this study are similar, but extend much of this earlier work by considering whether butterflies respond to urbanization intensity in a non-linear manner at both local and landscape scales. No evidence was found for a strong unimodal pattern in which either species richness or abundance peaked at intermediate levels of urbanization. This has been documented for various taxa, especially birds, and is thought to be driven by greater habitat diversity in suburban areas as compared to more developed areas (Marzluff 2005). Within the urban matrix, butterfly species richness did not decline until impervious surface cover increased above 25%. This should not be considered evidence that low levels of urban development do not adversely influence butterfly

58 communities, but it does suggest that there is a threshold of development intensity above which urban butterfly richness may substantially decline. The apparent lack of a unimodal response along the urbanization gradient also suggests that enhanced habitat diversity or other factors associated with suburban habitats provide limited benefits to butterflies.

Species richness and total abundance were negatively associated with both local and landscape scale urbanization intensity, but the most influential spatial scale varied between sampling rounds for richness and total abundance. In contrast, the abundance of the less common species (excluding GB and CW) was most strongly and consistently influenced by local-scale urbanization intensity. Concepción et al. (2015) found a similar negative response of butterfly richness at multiple scales of urbanization intensity, where specialist butterflies (determined by the number of known larval host plants) responded at multiple scales and generalist butterflies only at intermediate scales (2 and 3 km radius). When further assessing species by mobility, highly mobile species responded to multiple scales of urbanization intensity whereas poorly mobile species responded only at the smallest spatial scales (Concepción et al. 2015).

Urbanization and habitat fragmentation affects rare and specialist species more than generalist species (Kitahara and Fujii 1994; Clark et al. 2007; Lizée et al. 2011). Specialized species not only tend to be less abundant than generalist species across the landscape, but also spatially restricted (Kitahara and Fujii 1994). Given that there may be more specialized butterfly species (either restricted by habitat, mobility, or food resource use) within the subset of less common species their consistent response to local-scale urbanization intensity may not by surprising.

The abundance of floral resources influenced butterfly species richness and abundance, with the strongest effects being on total butterfly abundance. While much of the work assessing impacts of floral abundance on urban pollinators has focused on taxa other than butterflies (Blackmore et al. 2014; Pardee and Philpott 2014; Lerman and Milam 2016), there is a small literature that similarly highlights the importance of floral abundance to butterflies in both urban (Fontaine et al. 2016) and non-urban settings (Clausen et al. 2001; Pywell et al. 2004; Kitahara et al. 2008; Curtis et al. 2015). 59

However, it is important to note that, after taking into account the urban gradient and floral abundance, much of the spatial variation in butterfly species richness and abundance remains unexplained by the models. This suggests that other factors, such as availability of larval host plants (see Chapter 3), variation in climate factors, etc., would also contribute to the spatial patterning of butterfly community structure along urbanization gradients.

While most butterflies are nectar generalists, some species show greater specialization in their floral nectar selection (Stefanescu and Traveset 2009, Dennis 2010), which could limit survival of some butterfly species unable to adapt to new food sources in urban areas (Jain et al. 2016). Native floral abundance had a small positive influence on butterfly richness, but exotic floral abundance did not influence butterfly abundance or did so negatively. One must be cautious to not interpret this outcome as compelling evidence that, at least in my study region, exotic plants are unlikely to act as useful nectar resources for butterflies. My results are correlative, without behavioural data on which flowers butterflies actually use as nectar sources it would be inappropriate to suggest my results stand in full contrast to literature suggesting exotic flowers are good supplementary nectar sources within the urban matrix (Bergerot et al. 2010; Dennis 2010; Jain et al. 2016). For instance, there could be more subtle explanations. Perhaps, this result is just a matter of preferential feeding on natives versus an avoidance of exotic floral resources, or, perhaps, it is due to floral resource availability and timing. The latter is evidenced by the shifting influence of exotic to total floral abundance on total butterfly abundance between sampling rounds as total floral abundance dropped by 60% in round 2, when there were far fewer nectar providing flowers available. Specifically, the abundance of native floral resources dropped by 80% over the season, but only dropped by 50% in the case of exotic flowers.

One plausible explanation for the apparent negative relationship between butterfly abundance and exotic floral abundance, is that exotic flowers may be particularly abundant in areas where local residents have invested heavily in landscaping or gardening activities. This would likely increase the abundance of exotic ornamental plants while also reducing the relative abundance of native plants, including larval host plants. However, the three most common and abundant nectar producing exotic floral

60 species in my study were yard weeds (Medicago polymorpha, Taraxacum spp., and Trifolium repens), which another study found to be important food sources within the urban matrix for other pollinators (Lerman and Milam 2016). A French study found that some floral specialist butterflies prefer exotic flowers, but other life history traits limit the ability of these butterflies to disperse into urban areas (Bergerot et al. 2010). Other pollinators readily used plants from different biogeographic regions that are similar to the study region but do not use plants from very dissimilar regions (e.g., temperate versus tropical regions) (Hanley et al. 2014; Salisbury et al. 2015). Thus, classification of flowers as native or exotic might not be reflective of how floral resources are perceived by the butterfly community. My results do suggest that butterfly species richness would benefit from planting native floral nectar species within the urban matrix. However, more work needs to be done to tease apart the mechanisms driving urban butterfly responses to floral resource availability.

The response of the butterfly community to imperviousness was largely unaltered when the presence of native vegetation remnants was included as an additional predictor, suggesting that presence of remnant vegetation does not mediate butterfly responses to urban development. This is surprising as I expected remnant native vegetation to have a strong influence on butterfly communities, particularly their richness (Burghardt et al. 2009; Chong et al. 2014). A number of factors likely contribute to the limited influence of remnant vegetation on butterfly community structure in my study. First, the remnant vegetation patches within my survey area vary greatly in their size (from a few m2 to several thousand m2) and habitat quality (from restored to degraded), and thus only some remnant patches may benefit butterflies. Second, and more specifically, most of the butterfly species detected in my surveys use a wide range of larval host plants (Field 2013), many of which can be readily found, as planted or spontaneous vegetation, within the urban matrix, perhaps limiting the importance of remnant habitats because some of the resources they provide occur elsewhere within the urban environment. Thirdly, the composition of the vegetation within those patches was unknown. These patches may not contain any larval host plants, or none relevant to the butterflies who can still access them.

Only two of the 14 butterfly species identified are known to utilize only a few larval host plants within this region of Australia (Splendid Ochre (Trapezites symmomus) uses one; Ringed Xenica (Geitoneura acanthi) uses two); the 12 other butterfly species use many more species as larval host plants (Appendix 2 Table 8; Field 2013). Further, including plant species known to be used by butterflies across all of Australia, no butterfly species I identified uses fewer than four larval host plants (Braby 2012; Field 2013). Nine of my 14 identified butterfly species are known to use exotic larval host plants, many of which are common weeds in the region. Of note, white clover (Trifolium repens), an exotic lawn weed and one of the three most abundant species in my floral abundance surveys, is a favored larval host plant of the most abundant butterfly in my study, GB. Given the flexibility that many butterflies within the community exhibit in their use of multiple native and exotic host plants, perhaps the premise of a close relationship between remnant vegetation patches and butterfly species richness is overstated.

CONCLUSIONS:

I found that impervious surfaces have a negative influence on the butterfly community regardless of the spatial scale of analysis. The vast majority of butterfly species occurring within the regional species pool appear to be excluded from the urban matrix. The presence of urban remnant habitats does little to bolster butterfly richness or abundance. Only one native butterfly species is abundant within the urban matrix and the community as a whole exhibits further declines when impervious surface cover exceeds approximately 25%. The less common butterfly species within my community (i.e. all species except GB and SW) are more responsive to local scale environmental variables than those at the larger landscape scale, as suggested for numerous other taxa (Beninde et al. 2015). My study provides evidence that some features of the urban matrix, for example provision of native nectar sources, can be managed to enhance butterfly communities.

REFERENCES: Aronson MFJ et al. (2017) Biodiversity in the city: key challenges for urban green space management. Frontiers in Ecology and the Environment 15:189-196

Beninde J, Veith M, Hochkirch A (2015) Biodiversity in cities needs space: a meta-analysis of factors determining intra-urban biodiversity variation. Ecology Letters 18:581-592 Bergerot B, Fontaine B, Renard M, Cadi A, Julliard R (2010) Preferences for exotic flowers do not promote urban life in butterflies. Landscape and Urban Planning 96:98-107 Blackmore LM, Goulson D, Stewart A, Bezemer M (2014) Evaluating the effectiveness of wildflower seed mixes for boosting floral diversity and bumblebee and hoverfly abundance in urban areas. Insect Conservation and Diversity 7:480-484 Braby M (2004) The Complete Field Guide to Butterflies of Australia. CSIRO Publishing, Collingwood, Victoria. Burghardt KT, Tallamy DW, Shriver WG (2009) Impact of native plants on bird and butterfly biodiversity in suburban landscapes. Conservation Biology 23:219-224 Burnham KP, Anderson DR (2002) Model Selection and Multi-Model Inference : A Practical Information-Theoretic Approach (2nd Edition). Springer, Secaucus, NJ, USA Carvell C, Westrich P, Meek WR, Pywell RF, Nowakowski M (2006) Assessing the value of annual and perennial forage mixtures for bumblebees by direct observation and pollen analysis. Apidologie 37:326-340 Chong KY, Teo S, Kurukulasuriya B, Chung YF, Rajathurai S, Tan HTW (2014) Not all green is as good: Different effects of the natural and cultivated components of urban vegetation on bird and butterfly diversity. Biological Conservation 171:299-309 Clark PJ, Reed JM, Chew FS (2007) Effects of urbanization on butterfly species richness, guild structure, and rarity. Urban Ecosystems 10:321-337 Clausen HD, Holbeck HB, Reddersen J (2001) Factors influencing abundance of butterflie and burnet moths in uncultivated habitats of an organic farm in Denmark. Biological Conservation 98:167-178 Collier N, Mackay DA, Benkendorff K, Austin AD, Carthew SM (2006) Butterfly communities in South Australian urban reserves: Estimating abundance and diversity using the Pollard walk. Austral Ecology 31:282-290 Concepción ED, Moretti M, Altermatt F, Nobis MP, Obrist MK (2015) Impacts of urbanisation on biodiversity: the role of species mobility, degree of specialisation and spatial scale. Oikos 124:1571-1582 Curtis RJ, Brereton TM, Dennis RLH, Carbone C, Isaac NJB, Diamond S (2015) Butterfly abundance is determined by food availability and is mediated by species traits. Journal of Applied Ecology 62:1676-1684 Dennis RLH (2010) A resource-based habitat view for conservation: butterflies in the Brittish landscape. 1 edn. John Wiley & Sons Ltd., United Kingdom

Driscoll DA, Banks SC, Barton PS, Lindenmayer DB, Smith AL (2013) Conceptual domain of the matrix in fragmented landscapes. Trends Ecol Evol 28:605-613 Fattorini S (2011) Insect rarity, extinction and conservation in urban Rome (Italy): a 120-year- long study of tenebrionid beetles. Insect Conservation and Diversity 4:307-315 Feber RE, Smith H, Macdonald DW (1996) The effects on butterfly abundance of the management of uncropped edges of arable fields. Journal of Applied Ecology 33:1191- 1205 Field R (2013) Butterflies: Indentification and life history. Museum Victoria Publishing, Melbourne, Victoria Fontaine B, Bergerot B, Le Viol I, Julliard R (2016) Impact of urbanization and gardening practices on common butterfly communities in France. Ecology and Evolution 6:8174- 8180 Forman RT (1995) Land mosaics: the ecology of landscapes and regions. Cambridge University Press, New York Grace Detailed-GIS Services (2012) Directly connected imperviousness compilation for Melbourne Water selected catchments. Melbourne, Australia Hahs A, McDonnell MJ, Holland K, Caryl F (2009) Biodiversity of Metropolitan Melbourne. Australian Research Centre for Urban Ecology, Melbourne, Australia Hanley ME, Awbi AJ, Franco M (2014) Going native? Flower use by bumblebees in English urban gardens. Annals of Botany 113:799-806 Hughes BB et al. (2017) Long-Term Studies Contribute Disproportionately to Ecology and Policy. BioScience 67:271-281 Ibbe M, Milberg P, Tunér A, Bergman K-O (2011) History matters: Impact of historical land use on butterfly diversity in clear-cuts in a boreal landscape. Forest Ecology and Management 261:1885-1891 Jain A, Kunte K, Webb EL (2016) Flower specialization of butterflies and impacts of non-native flower use in a transformed tropical landscape. Biological Conservation 201:184-191 Jones RE, Gilbert N, Guppy M, Nealis V (1980) Long distance movement of Pieris rapae. Journal of Animal Ecology 49:629-642 Kitahara M, Fujii K (1994) Biodiverity and community structure of temerate butterfly species within a gradient of human disturbance: an analysis based on the concept of generalist vs. specialist strategies. Population Ecology 36:187-199 Kitahara M, Yumoto M, Kobayashi T (2008) Relationship of butterfly diversity with nectar plant species richness in and around the Aokigahara primary woodland of Mount Fuji, central Japan. Biodiversity and Conservation 17:2713-2734

Lerman SB, Milam J (2016) Bee fauna and floral abundance within lawn-dominated suburban yards in Springfield, MA. Annals of the Entomological Society of America 109:713-723 Lizée M-H, Mauffrey J-F, Tatoni T, Deschamps-Cottin M (2011) Monitoring urban environments on the basis of biological traits. Ecological Indicators 11:353-361 Loram A, Thompson K, Warren PH, Gaston KJ (2008) Urban doestic gardens (XII): The richness and composition of the flora in five UK cities. Journal of Vegetation Science 19:321-330 Luck GW, Smallbone LT (2010) Species diversity and urbanisation: patterns, drivers, and implications. In: Gaston KJ (ed) Urban Ecology. Cambridge University Press, New York, pp 88-119 Lütolf M, Guisan A, Kienast F (2009) History matters: relating land-use change to butterfly species occurrence. Environ Manage 43:436-446 Marzluff JM (2005) Island biogeography for an urbanizing world: how extinction and colonization may determine biological diversity in human-dominated landscape. Urban Ecosystems 8:157-177 Mazerolle MJ (2016) AICcmodavg: Model Selection and Multimodel Inference Based on (Q)AIC(c). McDonnell MJ, Hahs AK (2008) The use of gradient analysis studies in advancing our understanding of the ecology of urbanizing landscapes: current status and future directions. Landscape Ecology 23:1143-1155 Norton BA, Evans KL, Warren PH (2016) Urban Biodiversity and Landscape Ecology: Patterns, Processes and Planning. Current Landscape Ecology Reports 1:178-192 Öckinger E et al. (2012) The landscape matrix modifies the effect of habitat fragmentation in grassland butterflies. Landscape Ecology 27:121-131 Olivier T, Schmucki R, Fontaine B, Villemey A, Archaux F (2016) Butterfly assemblages in residential gardens are driven by species’ habitat preference and mobility. Landscape Ecology 31:865-876 Orr A, Kitching R (2010) The Butterflies of Australia. Allen & Unwin, Sydney, New South Wales, Australia Paradis E, Claude J, Strimmer K (2004) APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20:289-290 Pardee GL, Philpott SM (2014) Native plants are the bee’s knees: local and landscape predictors of bee richness and abundance in backyard gardens. Urban Ecosystems 17:641-659 Pohl NB, Van Wyk J, Campbell DR (2011) Butterflies show flower colour preferences but not constancy in foraging at four plant species. Ecological Entomology 36:290-300

Pollard E (1977) A method for assesing changes in the abundance of butterflies. Biological Conservation 12:115-134 Pyle RM (1992) Handbook for Butterfly Watchers. Houghton Mifflin Company, New York. Pywell RF et al. (2004) Assessing habitat quality for butterflies on intensively managed arable farmland. Biological Conservation 118:313-325 R Development Core Team (2015) R: A language and environment for statistical computing. R Foundataion for Statistical Computing, Vienna, Austria Ramírez-Restrepo L, MacGregor-Fors I (2017) Butterflies in the city: a review of urban diurnal Lepidoptera. Urban Ecosystems 20:171-182 Sadler JP, Bates AJ, Hale J, James P (2010) Brining cities alive: the importance of urban green spaces for people and biodiversity. In: Gaston KJ (ed) Urban Ecology. Cambridge University Press, New York, pp 230-260 Salisbury A, Armitage J, Bostock H, Perry J, Tatchell M, Thompson K, Diamond S (2015) EDITOR'S CHOICE: Enhancing gardens as habitats for flower-visiting aerial insects (pollinators): should we plant native or exotic species? Journal of Applied Ecology 52:1156-1164 Savard J-PL, Clergeau P, Mennechez G (2000) Biodiversity concepts and urban ecosystems. Landscape and Urban Planning 48:131-142 Seto KC, Parnell S, Elmqvist T (2013) A global outlook on urbanization. In: Elmqvist T (ed) Urbanization, biodiversity, and ecosystem services: challenges and opportunities. Springer, New York, pp 1-12 Sing K-W, Jusoh WFA, Hashim NR, Wilson J-J (2016) Urban parks: refuges for tropical butterflies in Southeast Asia? Urban Ecosystems 19:1131-1147 Soga M et al. (2015) Landscape versus local factors shaping butterfly communities in fragmented landscapes: does host plant diversity matter? Journal of Insect Conservation 19:781-790 Soga M, Koike S (2012) Life-history traits affect vulnerability of butterflies to habitat fragmentation in urban remnant forests. Ecoscience 19:11-20 Stefanescu C, Traveset A (2009) Factors influencing the degree of generalization in flower use by Mediterranean butterflies. Oikos 118:1109-1117 Szigeti V, Kőrösi Á, Harnos A, Nagy J, Kis J (2016) Measuring floral resource availability for insect pollinators in temperate grasslands - a review. Ecological Entomology 41:231- 240

Thompson K, Hodgson JG, Smith RM, Warren PH, Gaston KJ (2004) Urban domestic gardens (III): Composition and diversity of lawn floras. Journal of Vegetation Science 15:373- 378 Threlfall CG, Ossola A, Hahs AK, Williams NSG, Wilson L, Livesley SJ (2016) Variation in vegetation structure and composition across urban green space types. Frontiers in Ecology and Evolution 4:66 Threlfall CG, Walker K, Williams NSG, Hahs AK, Mata L, Stork N, Livesley SJ (2015) The conservation value of urban green space habitats for Australian native bee communities. Biological Conservation 187:240-248 Venables B, Ripley B (2002) Modern Applied Statistics with S. Fourth edn. Springer, New York. Westphal C et al. (2008) Measuring bee diversity in different European habitats and biogeographical regions. Ecological Monographs 78:653-671 Williams MR (2009) Butterflies and day-flying moths in a fragmented urban landscape, south- west Western Australia: patterns of species richness. Pacific Conservation Biology 15:32-46 Williams MR (2011) Habitat resources, remnant vegetation condition and area determine distribution patterns and abundance of butterflies and day-flying moths in a fragmented urban landscape, south-west Western Australia. Journal of Insect Conservation 15:37-54 Yen A (2011) Melbourne's terrestrial invert biodiveristy; losses, gains, and the new perspective. The Victorian Naturalist 128:201-208

Chapter 3 Butterfly Larval Host Plants within the Urban Matrix

ABSTRACT:

Variability in urbanisation intensity across the landscape leads to spatial heterogeneity in urban plant communities. This variation has implications for fauna that depend on specific resources provided by plants, including butterflies and their larval host plants. Despite this, the variation in plant communities and subsequent impacts on fauna in urban landscapes are still understudied. This study evaluates the response of butterfly larval host plant richness and cover along a gradient of increasing impervious surface cover within the urban matrix and understand the role of plants in structuring the urban butterfly community. Sixty transects were sampled for larval host plants (LHP) and butterflies. LHP for 11 of the more common butterflies from the local community were targeted to ensure there would be enough data to assess potential relationships. Laval host plants were sampled once and butterflies twice during the flight season. Generalized linear models were used to evaluate the response of LHPs along the imperviousness gradient, the response of the butterfly community to LHP richness and cover, and to understand if there was a significant relationship between individual butterfly species and their LHP cover. A redundancy analysis was also used to understand how much of the variation within the butterfly community was explained by variation in LHP distribution. I found LHP richness and cover declined with increasing impervious surface cover. LHP richness demonstrated a stronger relationship than cover to the butterfly community. While impervious surface cover had a negative effect on the butterfly community, particularly the less common species. Variation in LHP cover explained 57% of the variation in the abundance of the four most common butterflies. Among these, three butterflies had a positive, significant relationship with the cover of one or two of their specific LHPs. In conclusion, increased impervious surface cover reduced both the richness and cover of LHPs for butterfly richness and abundance within the urban matrix. Exotic species were the most commonly encountered LHPs. The local butterfly community was positively impacted by LHPs within Melbourne’s urban matrix, with LHP richness having more of an effect than either cover or impervious surface cover. The structure of the local

68 butterfly community was in part influenced by certain exotic LHPs more than natives within the matrix. Thus, LHPs are providing an important resource the butterflies need to complete their life cycle and maintain populations in the altered landscape of a city. This suggests that increases in certain LHPs may mitigate the negative impacts of urbanization on urban butterflies.

INTRODUCTION:

The availability, density, and spatial configuration of required resources within the landscape determines its usefulness for any species (Dennis 2010). Urban areas are not typically considered to contain potential habitat for most taxa and species because of an inferred lack of available resources resulting from the spatial heterogeneity of both land-use and vegetation in addition to high levels of regular disturbance (Grimm et al. 2008; Zhao et al. 2010; Pauleit and Breuste 2011; Thompson et al. 2004). Despite these suggestions, research demonstrates that urban areas can support a variety of taxa, including bats and birds (Davis et al. 2015; Threlfall et al. 2016; Narango et al. 2017), insects (Pardee and Philpott 2014; Olivier et al. 2016; Narango et al. 2017), small mammals (Brady et al. 2009; Vergnes et al. 2013), and frogs (Parris 2006). These studies suggest specific resources are required to support these taxa within urban environments, but also highlight a need to better understand habitat requirements for other taxa where research is still lacking.

The urban matrix is comprised of formal (e.g., parks, remnants) and informal (e.g., wastelands, residential areas) green spaces; with most urban ecology studies focused on the more formal greenspace, particularly large parks and remnants, embedded within the urban matrix (Bolger et al. 2000; Parris 2006; Williams 2011; Threlfall et al. 2016). This focus limits our knowledge to what is found within their boundaries and often fails to account for how less formal greenspaces, such as residential gardens, may also be influencing urban biodiversity (Goddard et al. 2010; Lin and Fuller 2013). Residential garden areas alone comprise between 22 (Loram et al. 2007) and 36 % (Mathieu et al. 2007) of the urban matrix, thus there is much potential for the provision of required resources for a variety of taxa within the urban matrix. However, the biodiversity values of the urban matrix are understudied (Lin and Fuller 2013; Sweany et al. 2014). 69

Butterflies, with their two-stage life cycle, make a good taxa to study the importance of plants within the matrix. They require larval food plants (i.e., larval host plants), adult food resources (e.g., rotting fruit, flowering plants for nectar or pollen, etc.), and shelter or resting places (Dennis 2010). The availability of adult food resources is the most commonly measured resource for butterflies (Chapter 2; Krauss et al. 2003; Matter et al. 2009; Soga and Koike 2012). However, larval host plants play an equally important role in the life cycle of butterflies, and thus, are also an important resource to assess. Availability of known larval host plants has been demonstrated to positively influence butterflies within urban remnants (Koh and Sodhi 2004; Williams 2011; Soga et al. 2015), and urban wastelands (Öckinger et al. 2009, Strausz et al. 2012). Despite knowledge of LHP availability and use in urban greenspaces, we do not know how widely available LHP’s are within the urban matrix outside of urban greenspaces.

Literature to date has not focussed on butterflies within the urban matrix or the influence of LHPs in influencing matrix quality. Instead, large green spaces and less densely populated settings, such as agricultural landscapes dominate the urban butterfly larval host plant literature. There is a subset of the literature focussing on the urban matrix itself, but it gives no clear indication of the availability of LHP across the matrix nor their value for the butterfly community as a whole therein. Plant species richness and cover decreases with increasing urbanization (Kent et al. 1999; Clarkson et al. 2007; Cilliers and Siebert 2011), but it is not known if similar patterns exist among LHPs richness and cover. While some butterflies are known to use LHPs within the urban matrix, particularly within the residential matrix, (Graves and Shapiro 2003; Levy and Connor 2004; Strausz et al. 2012), they often contain less eggs than plants in more natural settings (Vickery 1995; Levy and Connor 2004; Gaston et al. 2005). Despite examples of butterfly species range expansion following the introduction of their LHPs into new parts of Australia (e.g., Dainty Swallowtail (Papilio anactus) on citrus and Orange Palm Darts (Cephrenes augiades) on palms; Field 2013), LHP and butterfly distributions in other urban areas are not well correlated (Hardy and Dennis 1999). This result may not be unexpected given that within a butterfly community there are a range of life history traits that influence how strictly a butterfly species is associated with its LHP; generalists typically have wider habitat ranges and use more

LHP compared to specialist butterflies (Dennis 2010). Much of the literature around the distribution of butterflies and their associated larval host plants is limited to the more specialized rare or threatened butterfly species or to studies limited to just a butterfly species or two (Braschler and Hill 2007; Čelik 2012; Pennekamp et al. 2013; Wilson et al. 2015). A broader perspective of the ability of LHPs to influence local butterfly community structure is currently needed.

The availability of LHPs within the landscape may influence the structure of the local urban butterfly community. Hardy and Dennis (1999) found significant, but mostly small correlations between the presence of butterflies and their LHPs across the urban matrix. While in gardens, Fontaine et al. (2015) found a significant positive correlation between the occurrence of specific butterfly species and species groups and their LHPs. Konvicka and Kadlec (2011) suggested the occurrence of certain butterfly species in urban parks had to do with their association with shrubs rather than highly managed grasses. The importance of exotic LHPs has been suggested, where Shapiro (2002) noted 40% of California’s urban butterfly species have no known native larval plant hosts within the urban matrix. In Australia, Dainty Swallowtails and Orange Palm Darts have been able to establish in residential gardens due to the planting of exotic LHP (introduced citrus trees (e.g., grapefruit, lemons, kumquats) and ornamental palms) (Field 2013). Despite these associations, no landscape scale studies have been conducted that investigate the impact of urbanization on LHP’s within the urban matrix, and the resulting impact on the butterfly community.

This chapter therefore focuses on the importance of LHP availability within the urban matrix, and the role of LHP’s in structuring the local butterfly community. I first ask how does butterfly LHP richness and cover changes along an imperviousness gradient. I secondly ask if there is a relationship between LHP richness and cover and butterfly richness and abundance; and finally, if there are discernible relationships between particular butterfly species abundance and their specific LHPs.

METHODS:

Site selection:

Sixty sampling cells selected from the original 100 500 x 500 m survey cells described in Chapter 2 were further investigated (Fig. 3-1). An even distribution along a gradient of impervious surface cover was maintained by choosing 12 sampling cells from each imperviousness category (0-20%, 20-40%, 40-60%, 60-80%, 80-100%). The six cells with the highest and lowest total abundance of butterflies summed across both sampling rounds within each imperviousness category were selected for use in this study.

Figure 3-1 – Location of the sampling cells (500 x 500 m) and the percentage cover of impervious surfaces across the 28 km diameter study area in Melbourne, Australia. A GIS dataset supplied by Melbourne Water was used to measure the impervious-surface cover (Grace 2012, but see Chapter 2 for details).

Vegetation surveys:

In December 2015 and January 2016, I conducted targeted surveys for larval host plants (LHP; Table 3-1) for butterflies found in our earlier surveys (Table 3-2). Except for Imperial Hairstreaks (Jalmenus evagoras), LHPs for butterfly species with an abundance of three or more individuals were included in my plant surveys. This was limited to plant species known occur in greater Melbourne area. Given the large number of entire plant families listed as LHP for Meadow Argus (Junonia villida calybe), I only surveyed for Plantago spp., and Plantago lanceolata in particular, as they are

72 considered common larval host plants for the species (Braby 2012; Field 2013). Further, LHPs for Marbled Xenica (Geitoneura klugii) were included as I had expected to find more of this species in the initial surveys.

LHPs were subsampled along previously established transects in each cell, using 20 50 m segments in each transect. Each segment was subdivided into 5 strips measuring 2 m x 2.5 m on either side of the centreline (Appendix 3 Figure 1; thus, 5 x 2 m total). This equates to 20% coverage of the transect area but only 0.004% of the sampling cell area. Previous studies were used to guide the sampling design and level of effort, however, no consistent size or approach has been used (Clausen et al. 2001; Williams 2011; Schneider et al. 2003; Koh and Sodhi 2004; Öckinger et al. 2009; Soga and Koike 2012; Soga et al. 2015).

Within each strip an LHP’s percent cover was recorded as: 0.1 – 1%, 2- 5%, 6-25%, 26- 50%, 51-75%, and 76-100% (Daubenmire 1959). Percent cover within the sampling unit (a 5 x 2 m strip), not the percent cover within the vegetated area of the sampling unit was used to determine an LHPs percent cover. This kept sampling effort the same across all cells. This distinction was required because within the sampling units there are impervious surfaces (sidewalks, trails, brick fences, driveways, etc.) that also occupy variable area.

Butterfly surveys:

Butterfly surveys were conducted using the same methods as described in Chapter 2, except with an earlier cut-off time of 4:00 pm. Cells were surveyed twice, once in late December 2015/early January 2016 and then late January/early February 2016 using a modified Pollard walk (5 x 5 x 5 m box at a 1 min/ 50 m segment pace; Pollard 1977).

Data analysis:

All modelling was run in R 3.2.1 (R Core Development Team 2015). The data were checked prior to modelling for spatial autocorrelation using the ‘ape’ package (Paradis et al. 2016). There was no evidence for spatial autocorrelation for most response variables, and in all other cases Moran’s I values were small (percent cover of

Medicago spp. Moran’s I = 0.048 and Ehrharta erecta Moran’s I = <0.001, both larval host plants).

To address question 1, larval host plant species richness (total number of targeted LHP identified along the transect) and cover (%, summed across species) were modelled as a function of imperviousness (%) and sampling date (with 21 June (austral winter

Table 3-1 – Larval host plant genera and species surveyed along transects (Field 2013 and Braby 2012 (underlined)). Species with * were found during surveys. Species with ** had <5 occurrences.

List of target larval host plants Arctotheca calendula* Lolium spp.*

Artemisia spp.** Lomandra longifolia* Australina pusilla Lomandra spp.* Austrostipa flavescens Medicago spp.* Austrostipa spp. Microlaena stipoides* Bachypodium distachyon* Nasturtium*

Brachypodium spp.* Panicum spp.* Brassica spp.* Parietaria debilis Bromus catharticus* Parietaria Judaica*

Bromus spp.* Paspalum spp.* Chrysocephalum spp.* Pennisetum spp.* Citrus spp.* Plantago lanceolate* Cleome Plantago spp.* crucifer veggies Poa spp.*

Cullen spp. Poa tenera Cynodon dactylon* Reseda* Daviesia brevifolia Rhodanthe chlorocephala ssp. roseum

Ehrharta erecta* Rhodanthe spp. Ehrharta longiflora* Rytidosperma spp.* Ehrharta spp.* Sisybrium officinale Euchiton involucratus* Soleirolia soleirolia* garden peas and beans Themeda triandra**

Geijera parviflora Trifolium spp.* Gnaphalium spp.* Urtica incisa Hirschfeldia incana** Urtica urens*

Imperta cylindrical** Vulpia spp. Lepidium spp.** Xerochrysum bracteata Lepidium africanum** Xerochrysum spp.*

Species Total Known Larval Host Plants Abundance Common Grass Blue*† 3570 Daviesia brevifolia, low growing Fabaceae, and non-natives Zizina otis labradus including garden peas and beans, Cullen spp., Medicago spp., Trifolium spp.

Cabbage White*† 835 Brassicaceae, Cleome, in particular crucifer veggies, Pieris rapae Hirschfeldia incana, Lepidium africanum, Nasturtium, Reseda, and Sisybrium officinale Common Brown*† 211 Microlaena stipoides, Poa tenera, P. poiformis, Themeda Heteronympha merope triandra, and non-native Brachypodium distachyon, Bromus catharticus, Cynodon dactylon, Ehrharta erecta Green Grass Dart*† 45 Imperta cylindrica, Thuarea involute, and non-natives Ocybadistes walkeri Brachypodium spp., Bromus spp., Cynodon dactylon, Ehrharta spp., Lolium spp., Panicum spp., Paspalum spp., Pennisetum spp. Australian 19 Ammobium alatum, Ammobium spp., Euchiton involucratus, Painted Lady* Onopordum acantheum, Rhodanthe roseum, Rhodanthe Vanessa kershawi spp., Xerochrysum bracteata, Xerochrysum spp., and Artemisia spp., Chrysocephalum spp., Gnaphalium spp., and Arctotheca calendula, Arctotheca spp. Dainty Swallowtail*† 16 Citrus spp. (native and non-native), Geijera parviflora, Papilio anactus Limonia acidissima, and Poncirus trifoliata Yellow Admiral*† 16 Australina pusilla, Parietaria debilis, Pipturis argenteus, Vanessa itea Urtica incisa, and non-native Pipturis argenteus, Parietaria judaica, Soleirolia, U. urens Ringed Xenica* 10 Microlaena stipoides, Poa sieberiana, Poa tenera, Themeda Geitoneura acantha triandra Imperial Hairstreak 7 Acacia spp., but preferential to A. mearnsii, A. decurrens, A. Jalmenus evagoras dealbata, and A. melanoxylon, A. harpophylla, A. irroranta, and Amyema pendula Meadow Argus* 3 Acanthaceae, Asteraceae, Convolvulaceae, Dipsacaceae, Junonia villida Goodeniaceae, Gentianaceae, Plantaginaceae, Portulacaceae, Scrophulariaceae, Centaurium spp., Epaltes spp., Evolvulus spp., Goodenia spp., Hygrophila spp., Hyptis spp., Plantago spp., Portulaca spp., Ruellia spp., Scaevola spp., Stemoidia spp., Verbena spp., Veronica spp., and non- native introduced plantains, particularly Plantago lanceolata, Antirrhinum spp., Lantana spp., Russelia spp., Phyla spp., Scabiosa spp., Stachytarpheta spp., Verbena spp. Splendid Ochre*† 3 Lomandra longifolia, L. filiformis, L. hystrix, L. obliqua, L. Trapezites symmomus spicata Orange Palm Dart 1 ornamental and exotic palms Cephrenes augiades Marbled Xenica* 1 Austrostipa flavescens, Joycea pallida, Rytidosperma spp, Geitoneura klugii Poa labillardieri, P. morrisii, P. tenere, Themeda triandra, and non-natives Bachypodium distachyon, Ehrharta longiflora, E. calycina, and Vulpia spp.

Species Total Known Larval Host Plants Abundance Saltbush Blue 1 Atalaya hemiglauca, Atriplex spp., Chenopodium spp. Theclinesthes serpentatus (native and non-native), Einadia spp., Halosarcia halocnemoides, and Rhagodia spp.

equinox) as day one). Two competing models were run with linear and quadratic terms to account for possible non-linear relationship. LHP richness models were run with a Poisson distribution using a generalised linear model with a log link, while cover (%) models were run with a Gaussian distribution. A difference in the Akaike information criterion corrected for small sample size equal to 2 (∆AICc=2) was used to discern support for different models (Burnham and Anderson 2002). Overdispersion was checked for all Poisson models (all models were <1.5). Validation plots were visually inspected after running them in R. Plots for the top model are provided in Appendix 3 Figure 2.

Common Grass Glues (Zizina otis) (GB) and the non-native Small Whites (a.k.a., Cabbage White, Pieris rapae) (SW) were very widespread (occurring in 93% of all cells) within the sampling cells. Due to this, total species richness and abundance with and without these two species was calculated. Total richness was highly correlated with richness excluding GB and SW (Spearman’s r = 0.94). Therefore, models were constructed for three response variables: total species richness, total abundance, and abundance excluding GB and SW. Only the 11 targeted butterfly species were included in these response variables (Table 3-2), not all butterfly species found during sampling. I restricted analyses to these 11 because only their LHPs were surveyed. Butterfly species richness was taken as the total number of individual species across the sampling season, but abundance measures were an average of the two sampling rounds.

Butterfly species richness and abundance were modelled as a function of LHP richness, cover, their combination, impervious surface cover (%) within the sampling cell and its quadratic form (Appendix 3 Table 1). Competing models of all combinations were included and AIC was used to discern the top model(s). Total abundance was square root transformed and abundance excluding GB and SW was logarithmically transformed to reduce their skew. Plots and histograms of the response variable data

76 are provided in Appendix 3 Table 2. Using a generalised linear model, the richness data fit a Poisson distribution with a log link, while abundance data were best described by a Gaussian distribution. Validation plots for the top models were visually inspected and the plots for first top model (e.g., the zero model from the AIC ranking) are provided in Appendix 3 Figure 3.

To further discern the relative importance of the predictor variables on the butterfly community, I calculated the multiplicative effect (with 95% confidence interval) of each predictor across its range (e.g., multiplicative effect of predictor variable = exp (regression coefficient x range) for each response variable (see Parris 2006). For example, a multiplicative effect of 3 corresponds to a predicted three-fold increase in butterfly richness or abundance across the range (from lowest to highest) of a given predictor variable when all other predictors are held constant. No change in butterfly richness or abundance would correspond to an effect of 1. Thus, any predictor variable with a multiplicative effect substantially different from 1 would likely have a biologically important effect on the butterfly community. If one were using a null- hypothesis significance testing framework, if the confidence interval fails to encompass 1, then a null hypothesis of no effect can be rejected at α=0.05 (Cumming and Finch 2001).

To understand if the variation in butterfly abundance could be explained by the variation in LHP abundance I ran a redundancy analysis (RDA). Prior to analysis, a Hellinger transformation was implemented to standardize the data (Legendre and Gallagher 2001). Two RDAs were performed: a less restricted one for butterflies with at least five occurrences and one restricted to butterflies with at least 10 occurrences out of 60 sampling cells. I ran two versions because the first one contained many zeros which could make relationships hard to detect. Only associated larval host plants with at least 5 occurrences out of 60 sampling cells were included in either RDA. A low total inertia of 0.08 was found for the less restricted RDA indicating little of the variation could be explained by the variables within this model (7 butterflies, 23 LHP). Thus, only the restricted RDA is presented. This analysis was run in the ‘vegan’ R package (Oksanen et al. 2007).

To understand if there were discernible relationships between butterflies and their LHP in question 3, species specific relationships were examined using butterfly abundance and cover of their specific LHPs. I again employed a competitive modelling approach with all combinations of four predictors: each LHP’s cover, a variable that combines the cover of all the LHPs for a butterfly species, imperviousness, and its quadratic form (Appendix 3 Table 3). A combined LHP cover variable was created to understand if the cumulative effect of LHP’s was more important than their individual effects. General linear models with Gaussian distributions were used. Validation plots were visually insepceted for all top models after being run in R. The plots for the first top model (e.g., the zero model from the AIC ranking) and are provided in Appendix 3 Figure 4.

RESULTS:

Only 42 of the targeted 59 larval host plant species were located during vegetation surveys with the three most common being Cynodon datcylon in 57/60 cells and Ehrharta erecta and Pennisetum spp. each in 56/60 cells. Pennisetum spp., Cynodon datcylon, followed by Paspalum spp. were the top three species by cover. Marbled Xenica was the only targeted butterfly species not found during surveys.

Imperviousness (%) Date Linear Quadratic Model ∆AICc r2 Parameter Estimate Parameter Estimate Parameter Estimate (95% CI) (95% CI) (95% CI) Richness 2 0 0.268 -0.138 (-0.225 to -0.052) -0.154 (-0.226 to -0.083) -0.007 (-0.089 to 0.076) 1 16.30 0.097 -0.127 ( -0.208 to -0.046) -0.046 (-0.126 to 0.035) Cover (%) 1 0 0.392 -0.040 (-0.055 to -0.025) -0.008 (-0.023 to 0.007) 2 1.45 0.391 -0.040 (-0.055 to -0.025) -0.006 (-0.018 to 0.007) -0.006 (-0.022 to 0.009)

Larval host plant (LHP) richness and cover were negatively affected by imperviousness (Fig. 3-2; Table 3-3). LHP richness had a quadratic relationship with imperviousness, but there was not clear support for a quadratic relationship for LHP cover. All measures of butterfly richness and abundance had a positive relationship with LHP richness (Table 3-4). LHP cover had a positive relationship with total butterfly abundance. Imperviousness had a negative effect particularly on the abundance of the less common butterflies. The calculation of multiplicative effects confirmed the abovepatterns, demonstrating the importance of LHP richness and cover above that of imperviousness (Figs. 3-3 and 3-4). These calculations also indicate that LHP richness had a greater effect on the butterfly community, particularly total butterfly abundance, than LHP cover or imperviousness. Imperviousness had a small, mostly negative to no effect on the butterfly community across these models.

Only four butterflies had >10 occurrences (Grass Blues, Small Whites, Green Grass Darts (Ocybadistes walker), and Common Browns (Heteronympha merope)) and were thus used in the RDA. Larval host plants explained 50.9% of the variation in the distribution of these butterfly species (Table 3-5). Grass Blues had a particularly high species score on RDA1 indicating that this butterfly and its LHP comprised a large portion of the explained variation within the community.

Larval Host Plant Imperviousness (%)

Model ∆AICc r2 Richness Cover Linear Parameter Estimate Parameter Estimate Parameter Estimate (95% CI) (95% CI) (95% CI) Total butterfly richness 1 0 0.329 0.372 (0.177 to 0.585) 2 1.572 0.343 0.348 (0.138 to 0.570) 0.071 (-0.102 to 0.243) 4 1.757 0.339 0.359 (0.158 to 0.576) -0.057 (-0.220 to 0.108) Total butterfly abundance 2 0 0.392 0.675 (0.380 to 0.971) 0.224 (-0.072 to 0.520) 1 0.068 0.378 0.787 (0.528 to 1.050) 5 1.361 0.392 0.683 (0.386 to 0.980) 0.323 (-0.036 to 0.682) 0.163 (-0.169 to 0.495) Abundance of less common butterflies 1 0 0.072 0.184 (0.462 to 0.339) 4 0.538 0.083 0.146 (-0.019 to 0.311) -0.107 (-0.272 to 0.058) 10 1.462 0.049 -0.158 (-0.316 to -0.001) 2 1.783 0.064 0.152 (-0.028 to 0.332) 0.063 (-0.117 to 0.243)

With the exception of Small Whites, each of the butterflies was significantly correlated with one or two of their LHPs (Table 3-6). Only Small Whites had a significant, although negative, relationship. The low occurrence within our sampling of the LHP for this species may have influenced this result.

DISCUSSION:

Larval host plant (LHP) richness and cover experienced strong declines with increasing imperviousness. These changes in LHP distribution were also important in structuring the urban butterfly community over and above the negative impact of increasing impervious surface cover. This highlights the significant role of LHPs in determining the distribution of adult butterflies within the urban matrix.

a c

b d

Response of LHPs across the urban matrix

I found a significant negative decline with increasing imperviousness in larval host plant richness and cover within Melbourne’s urban matrix. Larval host plants had a similar decrease with increasing urban cover in Manchester, UK (Hardy and Dennis 1999). Several studies show that plant richness and cover generally decline with increasing urbanization (Sukopp et al. 1979; Kent et al. 1999; Clarkson et al. 2007; Cilliers and Siebert 2011), thus, LHPs follow this established trend. LHP richness did not experience as sharp a decline as their cover along the gradient, indicating a trade-off whereby there is a similar richness of available LHP species, but their combined cover decreases. Interestingly, plant cover had a linear inverse relationship with increasing imperviousness, while richness started its decline above 40% impervious cover. The

a c

A b d

a c

b d

Figure 3-4: The multiplicative effect on A) total butterfly abundance and B) abundance of less common butterflies (excluding Grass Blues and Small Whites) of the four predictive variables (mean and 95% confidence interval), as predicted by the models they are contained within (see Appendix 3 Table 1 for model sets). The horizontal line is at 1. The four predicitve variables are larval host plant (LHP) species richness (a), LHP cover is the % cover of LHP within the sampled area (b), imperviousness is the % impervious surface cover in the sampling cell (c), and imperviousness^2 is its quadratic form (d).

Axes RDA1 RDA2 Eigenvalues 24.912 2.497 Cumulative Variance Explained 0.509 0.560 Species Scores Grass Blues (Zizina otis labradus) 5.147 0.288 Cabbage Whites (Pieris rapae) 0.894 -1.434 Common Browns (Heteronympha merope ) 0.216 -0.698 Green Grass Darts (Ocybadistes walkeri) 0.149 -0.341 Biplot scores for explanatory variables Imperviousness (%) -0.14067 0.40917 Brachypodium distachyon -0.02985 0.00583 Brassica spp. -0.06955 -0.05477 Bromus catharticus 0.33302 0.12836 Bromus spp. 0.18680 -0.10793 Cynodon dactylon -0.04385 -0.34066 Ehrharta erecta -0.01248 -0.54988 Ehrharta longiflora -0.16881 0.13085 Ehrharta spp. -0.20555 -0.12901 Lolium spp. 0.01077 0.21454 Medicago spp. 0.58365 0.17172 Microlaena stipoides 0.09254 -0.73266 Nasturtium 0.13637 0.14155 Panicum spp. 0.02436 -0.03884 Paspalum spp. 0.45735 -0.08367 Pennisetum spp. 0.45276 -0.12394 Poa spp. 0.03738 -0.28031 Trifolium spp. 0.56819 0.11300

higher plant species richness displayed by LHPs between 15 and 40% imperviousness appears to follow a similar pattern found in other studies with higher species in the suburban portions of rural – urban gradients (Kent et al. 1999; Zerbe et al. 2003; Zhang et al. 2016). The response recorded in the current study differs in magnitude to studies using a distance from city center measure, however, the similar pattern is interesting and warrants further investigation. Results from this and a previous study (Chapter 2) indicate that plant richness and cover are responding directly to impervious- surface

Table 3-6 –Top models for understanding the relationship between individual butterfly species abundance and their larval host plant cover (%). A variable combining each of a species LHP, Combined, was included in the modelling to understand if the cumulative effect of the LHP was more important than the individual LHP species. Imperviousness is the % impervious surfaces within a sampling cell. The r2 for richness models is based on the calculation of pseudo r2 (1-(residual / null deviance)).

Larval Host Plant Imperviousness (%) Model ∆AICc r2 Cover Linear Quadratic Species Parameter Estimate (95% CI) Parameter Estimate (95% CI) Parameter Estimate (95% CI) Grass Blues (Zizina otis labradus) 1 0 0.305 Combined 0.412 (0.253 to 0.572) 4 0.583 0.312 Combined 0.422 (0.263 to 0.581) -0.163 (-0.417 to 0.091) Small Whites (Pieris rapae) 11 0 0.108 -0.191 (-0.331 to -0.050) -0.124 (-0.270 to 0.023) 10 0.651 0.079 -0.173 (-0.314 to -0.032) Common Browns (Heteronympha merope) 6 0 0.181 Ehrharta erecta 0.206 (0.096 to 0.315) 7 0.856 0.170 Microlaena stipoides 0.199 (0.089 to 0.310) Green Grass Darts (Ocybadistes walkeri) 7 0 0.094 Pennisetum spp. 0.136 (0.034 to 0.238)

84 cover in the landscape (potentially via a decrease in available growing space) regardless of distance from city center. At the more impervious end of the gradient, many of the LHPs present were common perennial, exotic species that are lawn weeds in Melbourne (e.g., Trifolium spp., Paspalum spp., Plantago lanceolata; Richardson et al. 2011), which is a pattern observed elsewhere (Kent et al. 1999; Schmidt et al. 2014; Vakhlamova et al. 2014).

Weedy species are more tolerant of the disturbance and drier soil conditions that often define the growing conditions within more built-up environments (Vakhlamova et al. 2014; Kalusová et al. 2017). Two commonly planted turf species, Pennisetum spp. (comprised mainly of P. clandestinum) and Cynodon datcylon, provided the highest proportions of cover at this end of the gradient. Whether through plant traits (e.g. disturbance tolerant weed species), or human preference (e.g. planted turf grasses), these LHPs have persisted through different filters (Aronson et al. 2016) present in greater Melbourne’s most urbanized areas.

This study was potentially limited by issues surrounding plant surveys in urban areas. However, despite restrictions in access, potential under-sampling in areas recently mown, and differences in growth rates and flowering between species, I am confident these issues would not change the strong pattern of declines in LHPs with increasing urban cover I revealed. More work such as this needs to be undertaken not only to refine effective methods for vegetation surveys in urban environments, but because the data garnered was informative and practical.

Butterfly community response to LHPs

LHP richness and cover were important in influencing the structure of the urban butterfly community in my study. LHP richness, in particular, had a significant effect on the community. The literature is not definitive on this matter with previous studies showing mixed results. Grassland butterflies lacked a correlation with LHP richness while woodland butterflies were positively influenced by it in Japan’s urban green spaces (Soga et al. 2015). Meanwhile, Hardy and Dennis (1999) found a low correlation between butterfly richness and LHP presence across two scales spanning the Manchester, England area.

The abundance of less common butterflies was not positively affected by LHP richness or cover. An earlier study within the same area indicated these less abundant butterflies had a strong negative response to local level imperviousness and the abundance of exotic flowers (Chapter 2). It would seem that while the less abundant species can persist in the urban matrix, there are additional factors such as life history traits (Wood and Pullin 2002; Börschig et al. 2013; Curtis et al. 2015; Olivier et al. 2016), microclimate needs (Levy and Connor 2004; Serruys and Van Dyck 2014; Curtis et al. 2015), and larval host plant quality (Levy and Connor 2004; Serruys and Van Dyck 2014) limiting their population sizes.

Butterfly species specific response to their LHPs

I was not able to find strong correlations between most of the targeted butterfly species and their larval host plants. Variation in LHP cover did explain 57% of the variation in abundance for the four most common butterfly species, but this was driven by one butterfly species (Grass Blues). Among these common butterfly species, three had a significant relationship with one or two of their LHPs found during my surveys. Amongst these three butterflies, only Common Browns had a significant relationship with a native LHP. The other important LHP were weedy (Ehrharta erecta, Medicago spp., Trifolium spp.) or planted (Cynodon dactylon and Pennisetum spp., but mostly P. clandestinum) species, most of which are exotic. Each of these LHPs were nearly ubiquitous in my sampling, occurring in at least 85% (51/60) of sampling cells. Perhaps it is the frequency of these species within the urban matrix that allows them to have more of an impact on the butterfly species, even in highly impervious and urban parts of the landscape. Regardless, breeding records would be needed to confirm if these three butterflies are indeed reproducing within the urban matrix as is indicated by the relationship between them and some of their LHPs.

There are other instances within the literature of an implied influence of LHPs structuring the butterfly community within urban settings. Abundance of four widespread butterfly species/species groups showed positive correlations with the presence of their larval host plant families in French gardens (Fontaine et al. 2016). In Melbourne, both Dainty Swallowtails (Papilio anactus) and Orange Palm Darts (Cephrenes augiades sperthias) have become part of the local community due to their 86 larval host plants being planted more widely (Field 2013), suggesting that these species also have a close association with their LHP. Orange Palm Darts use horticultural palms and Dainty Swallowtails use a variety of citrus trees in Melbourne the area, none of which are native to the local flora (Field 2013; Kitching et al. 1999). These two butterfly species are more commonly found in town than the surrounding rural landscape (Field 2013), implying they must be using the urban matrix, particularly the residential matrix, as breeding habitat (Levy and Connor 2004; Strausz et al. 2012; Serruys and Van Dyck 2014). Despite these examples of seemingly close relationships between some butterflies and their LHP, my data indicates that butterfly distribution appears more limited than the distribution of larval host plants (Botha and Botha 2006, Dennis 2010), emphasizing that it is difficult to predict butterfly distributions for most species based on solely on larval host plant distribution (Dennis et al. 2010) or only when using butterfly sighting versus breeding records (Hardy and Dennis 1999).

Life history traits such as voltinism (number of broods per season), mobility, and number of LHPs used are commonly used to classify butterflies as either generalists or specialists. Each of the four most common butterflies would be classified as generalists according to Field (2013) as they have more than one brood per season and are polyphagous, meaning all are known to use LHPs from multiple genera. High mobility is also often used to categorize generalists, but we lack detailed species mobility data in Australia to make such classifications. Being a generalist is often thought to separate urban and non-urban butterflies (Clark et al. 2007; Börschig et al. 2013; Deguines et al. 2016), however some generalists are also uncommon in urban areas. For example, Meadow Argus (Junonia villida calybe) are multivoltine and use more larval host plants than Grass Blues, the most abundant butterfly in this study, yet few Meadow Argus were observed. Yellow Admirals (Vanessa itea) are also multivoltine and polyphagous (using about eight LHPs, compared to the dozens used by Meadow Argus), however this species was found across the whole impervious gradient, albeit in very low numbers within the highest imperviousness category (80-100%). Its LHP, Parietaria judaica, was found in low cover within half the sites in this imperviousness category. As such, it would appear Yellow Admirals can persist in low densities at higher levels of

87 urbanization unlike Meadow Argus, but neither would be considered ‘common’ despite being generalists.

Management practices used within urban green spaces may be just as important as LHP richness in structuring the urban butterfly community. Konvicka and Kadlec (2011) suggest less mowing could lead to more or better habitat for butterflies. For instance, mowing within greenspaces reduced the number of eggs found on sites and at multiple scales across Vienna, Austria (Strausz et al. 2012). Taller vegetation (from 5 to 40 cm tall) in power line right-of-ways in Winnipeg, Canada had a positive impact on butterfly abundance (Leston and Koper 2017). This study was not able to address management practices, but anecdotally from Chapter 4, residential gardens with less mowing or weeding appeared to have more butterflies than well maintained gardens. Aside from reduced mowing, conducting fewer general garden maintenance activities (pruning, weeding, pesticide application) have shown to be beneficial for butterflies. For example, pesticide use in French gardens had a negative impact on butterflies (Muratet and Fontaine 2015). Leaving a litter layer may help buffer pupae from winter’s temperature fluxes for butterflies that overwinter on or near the ground (Stuhldreher and Fartmann 2014). Continued work in this area would not only be useful in teasing out the effects of management practices on the urban butterfly community’s use of their LHP, but also in establishing better informed guidelines for both home owners and urban greenspace managers on the friendliest management practices for the urban butterfly community.

The results of this study emphasize the negative impact of urbanization on LHP plants, but shows that LHP richness and cover have differential responses to increasing imperviousness. The most common LHPs within the matrix are either weedy or planted exotic species. The urban butterfly community was positively influenced by LHPs within the urban matrix; with LHP richness having more of an effect on the butterfly community than did LHP cover or imperviousness. The availability of certain LHPs within the urban matrix appears to be influencing the local butterfly community’s structure. As such, they deserve more research attention than they currently receive in the literature. More mechanistic research, similar to this study, is needed to understand what other biotic or abiotic forces are at work in allowing some butterfly

88 species to be more common than others within the urban matrix. Additionally, more research is needed around potential opportunities to increase habitat suitability and quality measures within the urban matrix.

REFERENCES:

Aronson MF et al. (2016) Hierarchical filters determine community assembly of urban species pools. Ecology 97:2952-2963 Bolger DT, Suarez AV, Crooks KR, Morrison SA, Case TJ (2000) Arthropods in urban habitat fragments in Southern California: Area, Age, and Edge Effects. Ecological Applications 10:1230-1248 Börschig C, Klein A-M, von Wehrden H, Krauss J (2013) Traits of butterfly communities change from specialist to generalist characteristics with increasing land-use intensity. Basic and Applied Ecology 14:547-554 Botha C, Botha J (2006) Bring butterflies back to your garden. KwaZulu-Natal Coastal Branch of the Botanical Society of South Africa, Mayville, South Africa Braby MF (2012) Further notes on the butterfly fauna of La Trobe Wildlife Sanctuary and adjacent nature conservation reserves, Victoria, and its conservation significance. The Victorian Naturalist 129:86-96 Brady MJ, McAlpine CA, Miller CJ, Possingham HP, Baxter GS (2009) Habitat attributes of landscape mosaics along a gradient of matrix development intensity: matrix management matters. Landscape Ecology 24:879-891 Braschler B, Hill JK (2007) Role of larval host plants in the climate-driven range expansion of the butterfly Polygonia c-album. Journal of Animal Ecology 76:415-423 Burnham KP, Anderson DR (2002) Model Selection and Multi-Model Inference : A Practical Information-Theoretic Approach (2nd Edition). Springer, Secaucus, NJ, USA Čelik T (2012) Adult demography, spatial distribution and movements of Zerynthia polyxena (Lepidoptera: Papilionidae) in a dense network of permanent habitats. European Journal of Entomology 109:217-227 Cilliers SS, Siebert SJ (2011) Urban flora and vegetation: patterns and processes. In: Niemelä J (ed) Urban Ecology: patterns, processess, and applications. Oxford University Press, New York, pp 148-158. Clark PJ, Reed JM, Chew FS (2007) Effects of urbanization on butterfly species richness, guild structure, and rarity. Urban Ecosystems 10:321-337

Clarkson BD, Wehi PM, Brabyn LK (2007) A spatial analysis of indigenous cover patterns and implications for ecological restoration in urban centres, New Zealand. Urban Ecosystems 10:441-457 Clausen HD, Holbeck HB, Reddersen J (2001) Factors influencing abundance of butterflie and burnet moths in uncultivated habitats of an organic farm in Denmark. Biological Conservation 98:167-178 Cumming G, Finch S (2001) A primer on the understanding, use, and calculation of condidence intervals that are based on central and noncentral distributions. Educational and Psychological Measurements 61:532-574 Curtis RJ, Brereton TM, Dennis RLH, Carbone C, Isaac NJB, Diamond S (2015) Butterfly abundance is determined by food availability and is mediated by species traits. Journal of Applied Ecology 62:1676-1684 Daubenmire R (1959) A canopy-cover method of vegetation analysis. Northwest Science 33:43- 46 Davis A, Major RE, Taylor CE (2015) The association between nectar availability and nectarivore density in urban and natural environments. Urban Ecosystems 18:503-515 Deguines N, Julliard R, de Flores M, Fontaine C (2016) Functional homogenization of flower visitor communities with urbanization. Ecology and Evolution 6:1967-1976 Dennis RLH (2010) A resource-based habitat view for conservation: butterflies in the Brittish landscape. 1 edn. John Wiley & Sons Ltd., United Kingdom Field R (2013) Butterflies: Indentification and life history. Museum Victoria Publishing, Melbourne, Victoria Fontaine B, Bergerot B, Le Viol I, Julliard R (2016) Impact of urbanization and gardening practices on common butterfly communities in France. Ecology and Evolution 6:8174- 8180 Gaston KJ, Smith RM, Thompson K, Warren PH (2005) Urban Domestic Gardens (II): experimental tests of methods for increasing biodiversity. Biodiversity and Conservation 14:395-413 Goddard MA, Dougill AJ, Benton TG (2010) Scaling up from gardens: biodiversity conservation in urban environments. Trends Ecol Evol 25:90-98 Graves SD, Shapiro AM (2003) Exotics as host plants of California butterfly fauna. Biological Conservation 110:413-433 Grimm NB, Faeth SH, Golubiewski NE, Redman CL, Wu J, Xuemei B, Briggs JM (2008) Global change and the ecoogy of cities. Science 319:756-760

Hardy PB, Dennis RLH (1999) The impact of urban development on butterflies within a city region. Biodiversity and Conservation 8:1261-1279 Kalusová V, Čeplová N, Lososová Z (2017) Which traits influence the frequency of plant species occurrence in urban habitat types? Urban Ecosystems 20:65-75 Kent M, Stevens RA, Zhang L (1999) Urban plant ecolgy patterns and processes: a case study of the flora of the City of Plymouth, Devon, U.K. . Journal of Biogeography 26:1281-1298 Kitching RL, Scheermeyer E, Jones RE, Pierce NE (1999) Biology of Australian butterflies vol 6. Monographs on Australian lepidoptera. CSIRO Publishing, Collingwood, VIC, Australia Koh LP, Sodhi NS (2004) Importance of reserves, fragments, and parks for butterfly conservation in a tropical urban landscape. Ecological Applications 14:1695-1708 Konvicka M, Kadlec T (2011) How to increase value of urban areas for butterfly conservation? A lesson from Prague nature reserved and parks. European Journal of Entomology 108:219-229 Krauss J, Steffan-Dewenter I, Tscharntke T (2003) How does landscape context contribute to effects of habitat fragmentation on diversity and population density of butterflies. Journal of Biogeography 30:889-900 Legendre P, Gallagher ED (2001) Ecologically meaningful transformations for ordination of species data. Oecologia 129:271-280 Leston L, Koper N (2017) Urban rights-of-way as extensive butterfly habitats: A case study from Winnipeg, Canada. Landscape and Urban Planning 157:56-62 Levy JM, Connor EF (2004) Are gardens effective in butterfly conservation a case study with the pipevine swallowtail, Battus philenor. Journal of Insect Conservation 8:323-330 Lin BB, Fuller RA (2013) 2013 Sharing or sparing? How should we grow the world's cities? Journal of Applied Ecology 50:1161-1168 Loram A, Tratalos J, Warren PH, Gaston KJ (2007) Urban domestic gardens (X): the extent & structure of the resource in five major cities. Landscape Ecology 22:601-615 Mathieu R, Freeman C, Aryal J (2007) Mapping private gardens in urban areas using object- oriented techniques and very high-resolution satellite imagery. Landscape and Urban Planning 81:179-192 Matter SF, Ezzeddine M, Duermit E, Mashburn J, Hamilton R, Lucas T, Roland J (2009) Interactions between habitat quality and connectivity affect immigration but not abundance or population growth of the butterfly, Parnassius smintheus. Oikos 118:1461-1470 Muratet A, Fontaine B (2015) Contrasting impacts of pesticides on butterflies and bumblebees in private gardens in France. Biological Conservation 182:148-154

Narango DL, Tallamy DW, Marra PP (2017) Native plants improve breeding and foraging habitat for an insectivorous bird. Biological Conservation 213:42-50 Öckinger E, Dannestam Å, Smith HG (2009) The importance of fragmentation and habitat quality of urban grasslands for butterfly diversity. Landscape and Urban Planning 93:31-37 Oksanen J et al. (2017) Vegan: Community Ecology Package. The R Project Olivier T, Schmucki R, Fontaine B, Villemey A, Archaux F (2016) Butterfly assemblages in residential gardens are driven by species’ habitat preference and mobility. Landscape Ecology 31:865-876 Paradis E, Claude J, Strimmer K (2004) APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20:289-290 Pardee GL, Philpott SM (2014) Native plants are the bee’s knees: local and landscape predictors of bee richness and abundance in backyard gardens. Urban Ecosystems 17:641-659 Parris KM (2006) Urban amphibian assemblages as metacommunities. Journal of Animal Ecology 75:757-764 Pauleit S, Breuste JH (2011) Land-use and surface-cover as urban ecological indicators. In: Niemelä J (ed) Urban Ecology: patterns, processess, and applications. Oxford University Press, New York, pp 19-29 Pennekamp F, Monteiro E, Schmitt T (2012) The larval ecology of the butterfly Euphydryas desfontainii (Lepidoptera: Nymphalidae) in SW-Portugal: food plant quantity and quality as main predictors of habitat quality. Journal of Insect Conservation 17:195-206 Pollard E (1977) A method for assesing changes in the abundance of butterflies. Biological Conservation 12:115-134 R Development Core Team (2012) R: A language and environment for statistical computing. R Foundataion for Statistical Computing, Vienna, Austria. Richardson FJ, Richardson RG, Shepherd RCH (2011) Weeds of the south-east : an identification guide for Australia. 2nd edn. R.G. and F.J. Richardson, Meredith, Australia Schmidt KJ, Poppendieck H-H, Jensen K (2014) Effects of urban structure on plant species richness in a large European city. Urban Ecosystems 17:427-444 Schneider C, Dover J, Fry GLA (2003) Movement of two grassland butterflies in the same habitat network: the role of adult recources and size of the study area. Ecological Entomology 28:219-227

Serruys M, Van Dyck H (2014) Development, survival, and phenotypic plasticity in anthropogenic landscapes: trade-offs between offspring quantity and quality in the nettle-feeding peacock butterfly. Oecologia 176:379-387 Shapiro AM (2002) The California urban butterfly fauna is dependant on alien plants. Diversity and Distributions 8:31-40 Soga M et al. (2015) Landscape versus local factors shaping butterfly communities in fragmented landscapes: does host plant diversity matter? Journal of Insect Conservation 19:781-790 Soga M, Koike S (2012) Relative importance of quantity, quality and isolation of patches for butterfly diversity in fragmented urban forests. Ecological Research 27:265-271 Strausz M, Fiedler K, Franzén M, Wiemers M (2012) Habitat and host plant use of the Large Copper Butterfly Lycaena dispar in an urban environment. Journal of Insect Conservation 16:709-721 Stuhldreher G, Fartmann T (2014) When habitat management can be a bad thing: effects of habitat quality, isolation and climate on a declining grassland butterfly. Journal of Insect Conservation 18:965-979 Sukopp H, Blume H-P, Kunick W (1979) The soil, flora, and vegetation of Berlin's waste lands. In: Laurie IC (ed) Nature in Cities. Wiley, Chichester, New York, p 428 Sweaney N, Lindenmayer DB, Driscoll DA (2014) Is the matrix important to butterflies in fragmented landscapes? Journal of Insect Conservation 18:283-294 Thompson K, Hodgson JG, Smith RM, Warren PH, Gaston KJ (2004) Urban domestic gardens (III): Composition and diversity of lawn floras. Journal of Vegetation Science 15:373- 378 Threlfall CG, Ossola A, Hahs AK, Williams NSG, Wilson L, Livesley SJ (2016) Variation in vegetation structure and composition across urban green space types. Frontiers in Ecology and Evolution 4:66 Vakhlamova T, Rusterholz H-P, Kanibolotskaya Y, Baur B (2014) Changes in plant diversity along an urban–rural gradient in an expanding city in Kazakhstan, Western Siberia. Landscape and Urban Planning 132:111-120 Vergnes A, Kerbiriou C, Clergeau P (2013) Ecological corridors also operate in an urban matrix: A test case with garden shrews. Urban Ecosystems 16:511-525 Vickery ML (1995) Gardens: the neglected habitat. In: Pullin AS (ed) Ecology and Conservation. Chapman & Hall, London, pp 123-134 Williams MR (2011) Habitat resources, remnant vegetation condition and area determine distribution patterns and abundance of butterflies and day-flying moths in a

fragmented urban landscape, south-west Western Australia. Journal of Insect Conservation 15:37-54 Wilson RJ, Bennie J, Lawson CR, Pearson D, Ortúzar-Ugarte G, Gutiérrez D (2015) Population turnover, habitat use and microclimate at the contracting range margin of a butterfly. Journal of Insect Conservation 19:205-216 Wood BC, Pullin AS (2002) Persistence of species in a fragmented urban landscape: the importance of dispersal ability and habitat availability for grassland butterflies. Biodiversity and Conservation 11:1451-1468 Zerbe S, Maurer U, Schmitz S, Sukopp H (2003) Biodiversity in berlin and its potential for nature conservation. Landscape and Urban Planning 63:139-148 Zhang D et al. (2016) Effects of forest type and urbanization on species composition and diversity of urban forest in Changchun, Northeast China. Urban Ecosystems 19:455-473 Zhao J, Ouyang Z, Xu W, Zheng H, Meng X (2010) Sampling adequacy estimation for plant species composition by accumulation curves - A case study of urban vegetation in Beijing, China. Landscape and Urban Planning 95:113-121

Chapter 4

Wildlife gardening increases native floral resources but not butterfly species richness or abundance

ABSTRACT:

Wildlife gardening is a popular activity undertaken within residential areas. It is broadly promoted as a means of encouraging residents to make their gardens more ‘wildlife friendly’. While theory and anecdotal evidence suggest these schemes should be effective, quantitative evaluation of wildlife gardening practices is lacking across most taxa they target. My overall objective was to determine if there was a difference in butterfly richness or abundance between gardens managed to benefit wildlife and those that are not. To test this, I conducted point counts for butterflies in residential gardens participating in a wildlife gardening program (n=27) and compared these to 1) non-participating control gardens within the neighborhood (n=27), and 2) gardens from areas not targeted by the gardening program (n=54). I measured garden floral resources and calculated tree cover (%) within the surrounding landscape as both are important potential resources for butterflies. Imperviousness (percent cover of impervious surfaces) within the landscape surrounding each garden was also measured as it is known to have a negative effect on the butterfly community. I found no effect of garden type on butterfly species richness or abundance. Higher native floral species richness and native floral abundance were found in wildlife gardens than controls, while exotic species richness and floral abundance were highest in control gardens from areas not targeted for wildlife gardening. Imperviousness and percent cover of trees had no observable influence on the butterfly community within the gardens. While I found no benefit to butterflies of the wildlife gardens in my study, they did support higher native floral richness and abundance than control gardens. Thus, wildlife gardens may be providing some benefits to other taxa by increasing native plant availability. I suggest there should be greater consideration of the taxa to be targeted in wildlife gardening programs. I further recommend that additional robust and quantitative evaluation of wildlife gardening programs and their specific activities

95 is needed in order to design schemes that can deliver greater benefits, as current evidence suggests there is much room for improvement.

INTRODUCTION:

Wildlife gardening is one approach that may maintain and enhance wildlife populations within the urban matrix (Thompson 2006; Gaston et al. 2007; Goddard et al. 2010; Mumaw and Bekessy 2017). Most broadly defined as any resource supplementation or management decision to benefit wildlife, wildlife gardening can take many forms from regularly maintaining a bird feeder to stopping the use of pesticides within a garden. Wildlife gardens, in addition to providing resources, may act as islands of habitat or stepping-stones that aid dispersal through the urban matrix between patches of habitat (Goddard et al. 2010; Beumer and Martens 2016), although this has rarely been tested empirically. Local governments and various conservation organizations have promoted wildlife gardening as a means of increasing household exposure to wildlife along with reaching general and specific wildlife conservation goals in urban and peri-urban areas (e.g., Gardens for Wildlife programs sponsored by Whitehorse and Knox City Councils, Victoria, Australia, The Wildlife Trusts (United Kingdom) promoting wildlife gardening for many taxa, or the North American Butterfly Association’s region specific butterfly gardening guides). Through these efforts, the concept and practice of wildlife gardening has become widespread in residential areas (Shaw et al. 2013).

The ecological value of residential yards in urban areas for various wildlife taxa, most notably insects (Owens 1991; Lerman and Milam 2016; Smith et al. 2006a, c) and birds (Owens 1991; Lerman and Warren 2011; Narango et al. 2017), has been well documented, with a smaller body of literature focusing on wildlife gardens (Gaston et al. 2005; Di Mauro et al. 2007; Fontaine et al. 2016). Despite the extent of this literature evaluating urban wildlife within residential yards, there is currently little empirical evidence that broad-scale wildlife gardening programs and activities work as intended, particularly in terms of conservation for uncommon species or species of concern within the urban matrix. Garden bird feeding is an exception to this, as there is research assessing its benefits and disadvantages to the urban bird community (Chamberlain et al. 2005; Kummer and Bayne 2015: Galbraith et al. 2017). While 96 thorough evaluation is lacking for other taxa, the literature does indicate that increasing native plant richness or occurrence within the urban matrix, has a positive impact on insects (Burghardt et al. 2010; Smith et al. 2015; Threlfall et al. 2015; Threlfall et al. 2017), birds (White et al. 2005; Threlfall et al. 2017), and bats (Threlfall et al. 2016; Threlfall et al. 2017).

Butterflies are well documented, fairly conspicuous, and well-liked by the public, making them a suitable taxon for public engagement with nature (New 1997; Thompson 2006). It has been suggested that gardens with nectar sources could help stem butterfly decline in the face of habitat loss (Levy and Connor 2004), and butterflies are known to respond quickly to make use of floral resource as it becomes available across the landscape (Ezzeddine and Matter 2008; Ibbe et al. 2011; Jonason et al. 2011; Curtis et al. 2015). Supplementing nectar resources in gardens may either allow for an expanded habitat area outside of habitat patches or easier passage through the urban matrix between habitat patches (Dennis 2010). Conservation management practices such as mulching (rather than having bare soil) and encouraging a diversity of habitats within small urban parks benefits urban butterfly abundance (Shwartz et al. 2013), possibly as these practices improve site conditions for wild food plants for butterflies. Residential garden management practices such as providing floral nectar sources, leaving certain weedy plants (nettles and Brassicaceae) to grow (Fontaine et al. 2016), and using less insecticides in gardens (Fontaine et al. 2016; Muratet and Fontaine 2015) increases butterfly richness and abundance. However, it is not known if butterflies benefit from general or broadly scoped wildlife gardening programs and how landscape context affects those benefits.

My overall objective was to assess if gardens participating in a wildlife gardening program within metropolitan Melbourne, Australia differ in their butterfly community compared to control gardens that do not participate. First I asked if there was a difference between the garden types in their available floral resources or surrounding tree cover. I further assessed if there was a difference in butterfly richness, abundance, and composition between these garden types and if their landscapes’ context (i.e., surrounding tree and impervious surface cover), influenced garden butterfly communities.

I predicted that gardens managed to benefit wildlife would have higher butterfly richness and abundance and native floral resources. This was expected because the program from which I recruited wildlife gardens promoted and encouraged the planting of native species. I also predicted surrounding tree cover would positively influence the garden butterfly community. This was expected because butterflies are known to use a matrix that is structurally similar to their natural habitat (i.e., woodland butterflies key-in on tree-like structure) (Lütolf et al. 2009; Ibbe et al. 2011; Öckinger et al. 2012) and previous landscape level butterfly surveys showed many members of the local butterfly community prefer open woodland and savannah-like habitats (Chapter 2; Field 2013).

METHODS:

Study area:

This study was conducted within the eastern suburbs of Melbourne, Australia. With a population of approximately 4.5 million residents, it is Australia’s second largest city. The study area was restricted to the Gippsland Plain Bioregion, an area dominated by a variety of heathland and grassy woodland vegetation types (Hahs et al. 2009).

The wildlife gardening program I recruited gardens from is within the Boroondara local government area (LGA). The 28 km2 diameter study area centered on this primarily residential inner east suburb with a population of about 177,000 located approximately 5 km east of Melbourne’s city center. This LGA is bordered on three sides by the Yarra River and Gardiners’ Creek, which provide a green, albeit mostly thin, border. Beyond these borders and to the east, the urban landscape continues with mostly similar human population densities interspersed with a variety of green spaces from playing fields to parks containing areas of restored native vegetation.

The wildlife gardening scheme I focused on encourages residents to implement wildlife friendly gardening practices. These included planting local native plant species, putting in ponds, or eliminating certain types of regular maintenance such as mowing and herbicide application. The program started in 2010, and its primary aim was to expand the habitat corridor along sections of streams the local government has restored and extensively revegetated with native species. This program’s design means that there

98 were specific areas within the LGA targeted by the program, allowing comparison between areas targeted and not targeted for wildlife gardening. For gardens entering this program an inception date was difficult to establish as few (n=6, 16%) wildlife gardens were started entirely from scratch and most owners had made some changes (such as replacing, adding, or removing plant(s)) since entering the program.

Site selection:

A grid of 500 m x 500 m cells was generated over the study area using ArcMap 10.2 (ESRI , Redlands, CA). I used an impervious surface cover GIS data layer supplied by Melbourne Water (Grace 2012) to calculate impervious surface cover (%) in each grid cell. All the impervious surfaces (e.g., roads, roofs, footpaths) within Melbourne’s greater metropolitan area are mapped by this data layer at a 0.5m resolution. Grid cells ranged from 2 – 97% imperviousness across the entire study area, but had a narrower range in areas targeted for wildlife gardening (20-60%).

Grid cells in areas targeted by the wildlife gardening program contained between one and 11 wildlife gardens. However, not all gardens listed as part of the program were still in existence, I contacted about 60% of the listed program gardens. Gardens were only recruited if they had been participating in the program for at least 6 months (range 6 – 48 months, 16.50 ± 2.25 month average duration of participation) through direct contact from the program’s coordinator to homeowners. I included a grid cell in my study if three wildlife gardens, at least 50 m apart, agreed to participate. Three control gardens within that same cell were recruited either by letter dropping or door knocking at homes near random points generated by ArcMap. These gardens also had to be at least 50 m from other study gardens. I selected an equal number of grid cells in areas targeted and not targeted for wildlife gardening by this program to be able to assess if the wildlife gardens were having a spill over effect within the local area (e.g. if butterflies attracted by the wildlife gardens were also found in the surrounding neighbourhood). Six control gardens within these untargeted grid cells were similarly recruited from near random points. To control for imperviousness, I ‘paired’ targeted and not targeted grid cells within 2% imperviousness (Fig. 1). Pairs were sampled within seven days of each other, but were not otherwise compared. In

Cells in areas not targeted for wildlife gardening Cells in areas targeted for wildlife gardening wildlife gardens control gardens

total, 108 gardens were recruited from a total of 18 grid cells: 27 wildlife (WG), 27 control gardens in targeted grid cells (CT), and 54 control gardens in untargeted grid cells (CUT).

Butterfly surveys:

Butterflies were identified to species and counted during 10 min point counts (5 m diameter). In preliminary testing, 10 minutes yielded higher species richness results than 5 minutes, but counts >10 minutes made it difficult to avoid double counting individuals. Point counts were centered on areas within a garden to maximize the amount of planted area, but in many cases also included mown grass areas. In the case of wildlife gardens point counts were centered upon the part of the garden where the most effort towards wildlife gardening had been made (determined by talking to the homeowner). In cases where permission could not be obtained for a traditional garden within reasonable distance from that point, a garden with a low or no fence was

100 chosen, the point was visually located near the property line, and the count done from the footpath.

Surveys were conducted between 9 am and 5 pm and under appropriate weather conditions: temperature between 13°C and 33°C, average wind <10 km/hr, and cloud cover <60%. All attempts were made to identify butterflies ‘on-the-wing’, but when this was not possible the clock was stopped and the butterfly netted or a picture taken for identification (Collier et al. 2006; Williams 2008). In 30 % of cases (87% of which were ‘blues’ early in the sampling season), butterflies crossed the point count area too quickly to be correctly identified to species and were classified to family level (i.e., blues (Lycaenidae), darts/skippers (Hesperiidae), or browns (Nymphalidae)). These individuals contributed to species richness counts for the garden only when no other species of that family were identified, and they were included in abundance calculations.

While there is a single mid-summer peak in butterfly richness in Melbourne (Kitching et al. 1999), activity within the flight season varies between species, particularly multivoltine species (Appendix 2 Table 1; Field 2013). Most butterflies live less than 30 days (Pyle 1992; Orr and Kitching 2010), therefore, to avoid counting the same individuals twice effort was made to survey gardens at least 30 days apart (actual minimum was 55 days) during the flight season. The first sampling round was 11 Nov 2014 – 6 Jan 2015 (late spring to early / mid-summer) and the second sampling round was 23 Jan 2015 - 8 Mar 2015 (mid-summer to early fall). All point counts were conducted by one observer (JK) and identifications followed Field (2013).

Environmental variables:

There is a paucity of daily movement data for native Australian butterflies, but Jones et al. (1980) have reported that Small Whites (Peris rapae), an exotic member of the local butterfly community, typically move between 250 and 600 m in a day. Therefore, a 500 m buffer should capture the majority of the local butterfly community’s average daily foraging movements. High-resolution geospatial data were used to calculate two predictor variables within this 500 m circular buffer around each garden. Imperviousness (%) within the buffer was calculated from the same GIS layer used for

101 site selection. Tree cover (%) was obtained from a GIS layer which mapped tree cover at a 0.5 m x 0.5 m pixel resolution across the focal region using LiDar data obtained in 2009 (‘High_Res_Landcover_2009’ supplied by Grace GIS Services, Melbourne, Australia (unpublished layer)). I assessed these two variables because imperviousness has known negative effects on butterflies (Chapter 2) and tree cover could have a positive impact on the butterfly species within the local community with open woodland habitat associations.

Floral surveys were completed immediately before or after the butterfly surveys, within the same 5 m diameter point count area. Where possible, flowering forbs, shrubs, and trees were identified to species (otherwise to genus or family); following Thompson et al. (2004) I did not try to identify cultivars or varieties. Nectar is the primary food source for the butterflies found in my surveys (Orr and Kitching 2010; Field 2013), so only plants known to produce nectar or possessing nectaries were included. Larval host plants that do not produce nectar were not included in these surveys. Extensive literature searches in Web of Science and Google Scholar (Appendix 4 Table 1) were used to determine which flower species are considered to produce nectar, based on information for the species or genus. Since grasses do not produce nectar, they were excluded a priori. Of the 206 taxa of flowering forbs, shrubs, and trees, 134 were identified to species and 72 to genus level within the gardens. Exotic species were defined as those with known origins outside Australia and its islands.

Floral resources were recorded using a scale with seven categories modified from Feber et al. (1996) and Carvell et al. (2006): <25, 25-50, 51-100, 101-200, 201-500, 501- 2000, and 2001-4000 floral units. Floral units are the number of racemes, umbels, capitula, etc. on each flowering plant. Floral abundance was then summed across taxa using the mid-value of each category to determine the total number of floral units in each garden. From these data, I calculated total floral abundance, total floral species richness and their native and exotic components.

Data analysis:

I used linear mixed models to quantify relationships between the species richness and abundance of butterflies and floral resources in different garden types, to account for

102 possible non-independence of the six gardens within each sampling cell, cell number was used as a random effect in all models (Zuur et al. 2009). Wildlife gardens were used as the reference category for all modelling. Common Grass Blue (Zizina otis Godart) (GB), the unknown ‘blues’ group (B), and the non-native Small White (i.e., Cabbage White, Pieris rapae Linnaeus, SW) were very widespread (found in 92% of gardens across both rounds) and were the most common butterfly taxa in my sampling. To account for their disproportionate abundance I calculated total species richness and abundance with and without these three taxa. Therefore, I constructed models for four response variables: total species richness and abundance of butterflies and richness and abundance excluding GB, B, and SW. Each sampling round was separately assessed. Total floral abundance and total floral richness were correlated in both rounds (round 1 = 0.69, round 2 = 0.79); exotic floral abundance was correlated with exotic richness (round 1 = 0.75, round 2= 0.85) and total floral species richness (round 1 =0.72, round 2=0.67), thus, no floral richness and abundance variables were included in the same models. Prior to modelling, I checked all data for spatial autocorrelation using the R package ‘ape’ (Paradis et al. 2016). I only constructed models that did not take spatial autocorrelation into account as there was no evidence for significant spatial autocorrelation (Moran’s I values were consistently small, negative, and with P > 0.05). All modelling was run in R 3.2.1 (R Core Development Team 2015) with the ‘lme4’ package (Bates et al. 2017) and the ‘lmerTest’ package (Kuznetsova et al. 2016) was used to produce P values to report for the LMMs.

The fixed effects included garden type (wildlife (WG), control gardens in targeted (CT) or untargeted (CUT) areas), sampling conditions (i.e. sampling date (day one = 21 June (winter equinox)), time of day (calculated as minutes after sun-rise) and temperature (°C)), percent cover of impervious surfaces within a 500 m radius buffer centered on each garden, and percent tree cover within the same buffer area. Given the disparity in ranges, all non-categorical predictor variabless were standardized using the ‘scale’ function in R, all continuous variables prior to analysis by centering and scaling.

Two separate models were used to assess if there was a difference between garden types in their available floral resources and surrounding tree cover. The first model investigated if there was a difference in the aforementioned response variables (floral

103 resources or tree cover) between garden types, while the second model built upon the first to assess whether other factors (sampling date and surrounding imperviousness) influenced those differences. When running mixed-models within the lme4 package, data fitting a Gaussian distribution can only be run using linear mixed models (LMM), and a general linear mixed model is used to fit other distributions. Therefore, floral abundance measures and tree cover were modelled using LMM, while floral richness was modelled with a GLMM using a negative binomial distribution. To reduce skew in their distributions, native floral abundance was natural log transformed in both rounds, total floral abundance required a square root transformation in round 1 and a natural log transformation in round 2, while exotic floral abundance required a natural log transformation in round 1 and square root transformation in round 2. Validation plots were visually inspected for all models, but only those for round 1 and inclusive of date and impervious are provided in Appendix 4 Figure 1.

Three separate models were used to assess differences in the butterfly community between the garden types. To assess if landscape context had an effect on butterfly richness or abundance measures I ran a model (garden type + sampling conditions) inclusive of garden type*imperviousness (%) or garden type*tree cover (%). If the interaction was not significant, I ran the same model excluding the interaction. A third model was run with garden type + sampling conditions + tree cover (%) + imperviousness (%) to understand if the two landscape measures (tree cover and imperviousness) have a combined effect of the butterfly community with the garden types. Floral variables were not included in the third model set due to the high correlation between garden type of floral resources. Richness measures were run as GLMM using a Poisson distribution with log link. Total abundance was log transformed prior to modelling to approximate a Gaussian distribution and run as an LMM. Abundance of the less common species (excluding GB, B, and SW) was modelled with a negative binomial GLMM using a log link. In round 2 when assessing the interaction between gardens the negative binomial GLMM would not converge, thus a GLMM with log link was used. Imperviousness had a negative but non-significant effect on the butterfly community, indicating I was able to essentially control for it by pairing the

104 targeted and untargeted sampling cells. Validation plots were visually inspected for all models, but only those for round 1 are provided in Appendix 4 Figure 2.

To identify potential differences in the butterfly community between the garden types, a non-metric multidimensional scaling (NMDS) ordination was performed. Bray-Curtis dissimilarities were used on the identified butterfly species’ abundance (i.e., excludes abundance of unknown individuals grouped by family) and garden type as the only explanatory variable. Similar results were obtained when this analysis was repeated using a dataset inclusive of all individuals (with unknown individuals grouped by family; Appendix 4 Figure 3). I performed an ANISOM to determine if there was a difference in the community structure between the garden types (Clarke and Green 1988; Clarke and Warwick 2001). The ANOSIM’s R statistic is a measure of distinctiveness between groups. It can range between -1 and 1, with 0 being no difference and 1 being a distinct difference between categories (Clarke and Warwick 2001). These analyses were run within the R package ‘vegan’ (Oksanen et al. 2017).

RESULTS:

I identified eight butterfly species in the gardens, one of which was only found in round 1 and two of which were only found in round 2 (Appendix 4 Table 2). These eight species represent 12% (8/66 species) of the total non-vagrant butterfly species within the regional species pool, but 57% (8/14 species) of species in the local urban butterfly community (see Appendix 4 Table 3, Chapter 2). Mean total butterfly species richness per garden was significantly higher in the first sampling round within the control gardens from areas targeted (CT) and untargeted (CUT) for wildlife gardening (Table 4-1). Across all gardens the most abundant species were Common Grass Blues (GB), unknown ‘blues’ (B), and Small White (SW) which comprised 93% of the 683 individual butterflies counted in round 1, and 89% of the 472 individuals counted in round 2.

Difference in resources between garden types

Native floral abundance was highest in WG gardens (ranging from 6 (round 1) to 24 (round 2) times higher; Fig. 4-2, Table 4-1) and differed significantly between all garden types (Table 4-2). Exotic floral abundance was highest in CUT gardens (Table 4-1), but

105 was only significantly different between CUT and WG (Table 4-2). Total floral abundance in the second sampling round was significantly different between CT and WG, and almost significantly different between CT and WG (Fig. 4-2, Table 4-2), where WG had higher total floral abundance.

Garden Round 1 Round 2 Type Mean ± Std Err Mean ± Std Err Butterflies Richness CUT 2.69 ± 0.17 1.41 ± 0.14 CT 2.56 ± 0.27 1.07 ± 0.19 WG 1.52 ± 0.27 1.07 ± 0.18 Richness CUT 0.30 ± 0.07 0.35 ± 0.08 -GB, B, SW CT 0.12 ± 0.07 0.30 ± 0.09 WG 0.26 ± 0.10 0.22 ± 0.08 Total Abundance CUT 6.31 ± 0.69 4.81 ± 0.78 CT 6.48 ± 1.14 3.59 ± 0.93 WG 6.19 ± 2.00 4.26 ± 1.28 Abundance CUT 0.39 ± 0.10 0.56 ± 0.13 -GB, B, SW CT 0.30 ± 0.17 0.33 ± 0.11 WG 0.70 ± 0.34 0.48 ± 0.25 Floral Resources Total Abundance CUT 278.87 ± 68.45 132.52 ± 27.25 CT 229.04 ± 59.17 90.37 ± 19.04 WG 361.96 ± 83.17 278.19 ± 85.86 Native Abundance CUT 41.39 ± 36.00 7.80 ± 3.99 CT 37.30 ± 19.11 31.81 ± 16.71 WG 255.85 ± 87.29 195.26 ± 71.46 Exotic Abundance CUT 237.48 ± 59.02 124.72 ± 26.84 CT 191.74 ± 59.14 58.56 ± 12.55 WG 106.11 ± 26.49 82.93 ± 46.44 Total Richness CUT 3.80 ± 0.41 3.13 ± 0.38 CT 3.41 ± 0.39 2.52 ± 0.34 WG 5.78 ± 0.74 4.37 ± 0.53 Native Richness CUT 0.24 ± 0.10 0.26 ± 0.11 CT 0.41 ± 0.13 0.48 ± 0.19 WG 3.15 ± 0.76 2.63 ± 0.50 Exotic Richness CUT 3.56 ± 0.38 2.87 ± 0.36 CT 3.00 ± 0.36 2.04 ± 0.29 WG 2.63 ± 0.46 1.74 ± 0.39 Tree Cover % CUT 35.71 ± 0.75 CT 47.74 ± 0.27 WG 45.00 ± 0.73

106

Figure 4-2 – Boxplots of floral abundance and species richness across the different garden types (control garden in grid cells targeted (CT) and untargeted (CUT) for wildlife gardening and wildlife gardens (WG)) from the first sampling round. For both abundance and richness, outcomes were similar in the first sampling round. To improve presentation three data points (Total Floral Abundance: 1894 for CUT and 2134 for WG; Native floral Abundance: 2134 for WG) were excluded.

Native floral richness was highest in WG gardens (ranging from 6 (round 1) to 13 (round 2) times higher; Fig. 4-2, Table 4-1) and differed significantly between garden types (Table 4-2). Exotic floral richness was highest in CUT gardens (Table 4-1), but was only significantly different between CUT and WG, except in round 1 when accounting for date and imperviousness (Table 4-2). Total floral species richness was highest in WG (Table 4-1) and differed between WG and both CT and CUT gardens with particularly large difference in native species (Fig. 4-2, Table 4-2) across both sampling rounds.

Tree cover differed between gardens in areas targeted for wildlife gardening (WG, CT) and gardens in untargeted areas (CUT). Tree cover was 11% higher in areas targeted for wildlife gardening compared to untargeted areas (Table 4-3).

3.2 Difference in butterfly richness and abundance between garden types

With one exception, no difference in butterfly richness and abundance was found between garden types in either sampling round (Fig. 4-3, Table 4-4). In round 2, there

107

Total Floral Abundance Native Floral Abundance Exotic Floral Abundance Total Floral Richness Native Floral Richness Exotic Floral Richness Est ± Std Err Pvalue Est ± Std Err Pvalue Est ± Std Err Pvalue Est ± Std Err Pvalue Est ± Std Err Pvalue Est ± Std Err Pvalue Round 1 Intercept 16.79 ± 1.78 3.87 ± 0.40 3.65 ± 0.34 1.75 ± 0.11 1.15 ± 0.22 0.97 ± 0.13 CT -3.98 ± 2.30 0.087 -2.68 ± 0.55 < 0.001 0.68 ± 0.41 0.099 -0.53 ± 0.16 0.001 -2.05 ± 0.42 < 0.001 0.13 ± 0.18 0.471 CUT -2.41 ± 2.23 0.289 -3.01 ± 0.50 < 0.001 1.13 ± 0.44 0.016 -0.42 ± 0.14 0.002 -2.57 ± 0.38 < 0.001 0.30 ± 0.16 0.054 Round 2 Intercept 4.92 ± 0.28 3.86 ± 0.35 6.02 ± 1.10 1.48 ± 0.11 0.93 ± 0.23 0.55 ± 0.16 CT -1.33 ± 0.39 < 0.001 -2.82 ± 0.48 < 0.001 0.55 ± 1.55 0.723 -0.55 ± 0.18 0.002 -1.72 ± 0.38 < 0.001 0.16 ± 0.22 0.481 CUT -0.66 ± 0.34 0.054 -3.17 ± 0.43 < 0.001 3.69 ± 1.35 0.007 -0.33 ± 0.14 0.018 -2.31 ± 0.37 < 0.001 0.50 ± 0.19 0.008 Round 1 Intercept 16.52 ± 1.85 3.87 ± 0.42 3.65 ± 0.36 1.76 ± 0.11 1.06 ± 0.24 0.98 ± 0.14 CT -4.03 ± 2.31 0.084 -2.68 ± 0.55 < 0.001 0.68 ± 0.41 0.099 -0.53 ± 0.16 0.001 -2.01 ± 0.41 < 0.001 0.14 ± 0.18 0.461 CUT -1.85 ± 2.43 0.454 -3.03 ± 0.55 < 0.001 1.12 ± 0.48 0.026 -0.43 ± 0.15 0.003 -2.45 ± 0.41 < 0.001 0.27 ± 0.11 0.110 Date -0.54 ± 0.10 0.598 -0.09 ± 0.22 0.678 -0.06 ± 0.20 0.783 -0.08 ± 0.06 0.192 -0.21 ± 0.16 0.180 0.0021 ± 0.06 0.973 Imperv -0.53 ± 0.98 0.586 0.05 ± 0.23 0.818 0.03 ± 0.18 0.879 0.04 ± 0.07 0.564 -0.03 ± 0.21 0.892 0.041 ± 0.07 0.549 Round 2 Intercept 4.92 ± 0.28 3.91 ± 0.36 5.92 ± 1.14 1.48 ± 0.11 0.94 ± 0.23 0.57 ± 0.17 CT -1.33 ± 0.39 0.001 -2.80 ± 0.48 < 0.001 0.52 ± 1.57 0.7432 -0.55 ± 0.17 0.002 -1.70 ± 0.38 < 0.001 0.16 ± 0.22 0.475 CUT -0.66 ± 0.36 0.072 -3.28 ± 0.46 < 0.001 3.93 ± 1.46 0.0082 -0.35 ± 0.15 0.022 -2.32 ± 0.40 < 0.001 0.48 ± 0.20 0.017 Date -0.11 ± 0.14 0.440 -0.18 ± 0.18 0.353 -0.004 ± 0.56 0.9942 -0.08 ± 0.06 0.179 -0.15 ± 0.16 0.351 -0.05 ± 0.07 0.509 Imperv 0.03 ± 0.16 0.874 0.16 ± 0.20 0.427 -0.28 ± 0.63 0.6538 0.032 ± 0.07 0.664 0.04 ± 0.20 0.833 0.04 ± 0.08 0.653

108

Tree Cover (%) Est ± Std Error Pvalue Intercept 45.25 ± 2.03 CT -0.17 ± 1.66 0.918 CUT -8.34 ± 2.75 0.007 Round 1 Intercept 45.55 ± 2.11 CT -0.04 ± 1.64 0.981 CUT -8.00 ± 2.92 0.006 Date -1.35 ± 1.37 0.339 Imperv 1.23 ± 0.80 0.126 Round 2 Intercept 46.12 ± 1.94 CT -1.04 ± 1.80 0.567 CUT -9.64 ± 2.65 0.002 Date -0.54 ± 1.21 0.664 Imperv 0.91 ± 0.81 0.268

was a significant interaction between imperviousness and garden type (p < 0.05, Table 4-4) for the abundance of the less common butterflies (without GB, B, and SW). Their abundance in WG was significantly greater than their abundance in CT and CUT gardens at lower imperviousness levels (Fig. 4-4).

Tree cover did not have an observable influence on the butterfly community between the garden types, nor was there an interaction between garden type and tree cover (Appendix 4 Table 3). The inclusion of both tree cover and imperviousness within the same model had no observable effect (Appendix 4 Table 3).

The garden types did not contain distinctive butterfly communities in either sampling round as indicated by the ANISOM R statistic values being close to zero (Fig. 4-5, Appendix 4 Table 2; round 1: R = 0.03, P value = 0.002, round 2: R = <-0.01, P value =0.006; Clark and Warwick 2001).

109

110

Table 4-4 – Parameter estimates of the mixed modelling results to assess the difference in butterfly richness and abundance between garden types (WG – wildlife gardens and control gardens in areas targeted (CT) and untargeted for wildlife gardening (CUT). Wildlife gardens were used as the reference category for modelling. The significance of an interaction between garden type and imperviousness within a 500 m radius buffer (Imperv) was first assessed, if not significant the model was rerun sans the interaction. Outcomes for the second model are not presented as they were not different from the first model. Richness and abundance – GB, B, and SW is the total richness or abundance minus abundance of Grass Blues, ‘blues’, and Small White butterflies. Intercept of the model of the relationship with wildlife gardens (the reference category) and the other listed predictor variables is reported. Sampling conditions (Temp -temperature, date, and time of day) were included in both models.

Richness Richness– GB, B, and SW Total Abundance Abundance – GB, B, and SW Est ± Std Err Pvalue Est ± Std Err Pvalue Est ± Std Err Pvalue Est ± Std Err Pvalue Round 1 Intercept 0.15 ± 0.21 -1.16 ± 0.39 1.29 ± 0.21 -0.58 ± 0.49 CT 0.39 ± 0.27 0.139 -0.70 ± 0.67 0.297 0.35 ± 0.27 0.207 -0.54 ± 0.76 0.481 CUT 0.33 ± 0.24 0.182 -0.61 ± 0.56 0.275 0.37 ± 0.27 0.177 -0.66 ± 0.66 0.320 Temp 0.20 ± 0.08 0.011 0.32 ± 0.20 0.112 0.34 ± 0.10 0.003 0.23 ± 0.25 0.367 Time -0.08 ± 0.08 0.320 -0.23 ± 0.23 0.299 -0.05 ± 0.10 0.627 -0.16 ± 0.26 0.554 Date 0.03 ± 0.08 0.672 0.10 ± 0.19 0.597 0.09 ± 0.10 0.389 -0.17 ± 0.24 0.493 Imperv -0.02 ± 0.23 0.919 0.66 ± 0.54 0.222 -0.07 ± 0.23 0.754 -0.26 ± 0.47 0.576 Imperv*CT 0.11 ± 0.26 0.679 -0.54 ± 0.69 0.437 0.02 ± 0.28 0.946 0.37 ± 0.70 0.591 Imperv*CUT 0.10 ± 0.26 0.693 -0.05 ±0.64 0.934 0.08 ± 0.29 0.788 0.62 ± 0.62 0.313 Round 2 Intercept 0.03 ± 0.25 -1.80 ± 0.57 1.18 ± 0.23 -1.48 ± 0.53 CT 0.06 ± 0.31 0.843 0.65 ± 0.68 0.335 -0.12 ± 0.27 0.658 0.14 ± 0.59 0.810 CUT 0.10 ± 0.31 0.749 0.64 ± 0.64 0.317 0.02 ± 0.30 0.943 0.57 ± 0.61 0.349 Temp 0.21 ± 0.12 0.074 0.04 ± 0.19 0.848 0.42 ± 0.12 0.003 0.08 ± 0.21 0.721 Time 0.09 ± 0.11 0.422 0.21 ± 0.20 0.308 0.04 ± 0.12 0.716 0.26 ± 0.20 0.199 Date <0.01 ± 0.12 0.964 0.15 ± 0.19 0.432 -0.07 ± 0.13 0.583 -0.03 ± 0.22 0.909 Imperv -0.16 ± 0.29 0.597 -0.60 ± 0.56 0.284 -0.07 ± 0.28 0.801 -1.00 ± 0.46 0.031 Imperv*CT 0.19 ± 0.35 0.574 1.17 ± 0.65 0.073 -0.07 ± 0.32 0.842 1.71 ± 0.62 0.006 Imperv*CUT 0.34 ± 0.33 0.309 0.72 ± 0.62 0.249 0.10 ± 0.32 0.753 1.26 ± 0.54 0.020

111

Stress = 0.057

B Stress = 0.067

DISCUSSION:

My results show that wildlife gardening by its broadest definition, i.e. any garden managed to benefit wildlife, does not appear to have had a beneficial effect on the butterfly community within the study area. Wildlife gardening activities in the study

112 area increased native plant species richness and floral abundance within the urban matrix. This was an expected outcome given the program from which the gardens were recruited heavily promoted the planting of native species and provides 20 free local native plant seedlings per participating household. Despite these increased food resources, I found no substantial impact on the butterfly community, neither at the scale of the wildlife garden, nor in the nearby gardens surrounding groups of wildlife gardens. Further, I found no significant influence from surrounding tree cover or impervious surface cover.

It was somewhat unexpected to find no influence from surrounding tree cover on butterfly richness or abundance as many members of the local community are associated with open woodland habitat types (Chapter 2; Field 2013) and butterflies use matrices with similar vegetation structure to their preferred habitats (Lütolf et al. 2009; Ibbe et al. 2011; Öckinger et al. 2012). While Yahner (2001) showed a positive correlation between trees and the butterfly community in residential neighborhoods central Pennsylvania, USA, my results are more similar to Shwartz et al. (2013), who found no impact of tree cover on butterfly richness and abundance in small urban parks in Paris, France.

Wildlife gardens had higher total floral richness, but similar total floral abundance to control gardens. Butterflies have been shown to respond positively to higher floral richness and abundance in urban greenspaces (see Chapter 2; Matteson and Langellotto 2012; Soga and Koike 2012; Leston and Koper 2017). A concurrent landscape level study from Melbourne (Chapter 2) showed a small, but positive impact of native floral abundance on butterfly richness. Therefore, it was expected that wildlife gardens would have had a higher proportion of butterflies, particularly the less abundant butterflies, compared to control gardens. While gardens did contain some flowering plants known to be used as larval host plants for some butterflies in this region, I did not sample for all larval host plants as many are grasses, and hence were excluded from my surveys of floral resources. I did expect however that the higher native floral abundance, and by default nectar availability, in wildlife gardens would have influenced adult butterfly abundance or richness, but this was not found to be the case compared to control gardens. Most butterflies are nectar generalists (Dennis

113

2010) and do not have strong constancy in floral selection (Stefanescu and Traveset 2009; Pohl et al. 2011). Nectar flower selection is based on floral cues rather than floral species origin (Corbet 2000). A previous study suggested that diet specialists (e.g., butterflies with specific relationships to particular flowers) appeared to avoid urban areas (Bergerot et al. 2010). Therefore, it may not be surprising that neither the butterflies in this study nor in New York City community gardens (Matteson and Langellotto 2011) responded positively to the availability of native floral resources. Perhaps future research focusing on floral resource relationships such as determining which floral resources butterflies actually use as adult food plants (Corbet 2000; Tiple et al. 2009; Shackleton and Ratnieks 2015) and further investigating their nectar production quantity and quality (Comba et al. 1999; Corbet et al. 2001; Gomez et al. 2008) would be more useful than associative correlations in understanding the actual effects of floral resources on butterflies.

While this study found no effect of increased floral resources on the butterfly community, increased native vegetation in urban areas has been shown to benefit other taxa. Bird and bat richness both increased with increasing proportion of native vegetation in Melbourne, Australia (Threlfall et al. 2016). Native plants contained more caterpillars, and five times more caterpillars of larval host plant specialist species, than neighbouring ornamental plants in Delaware, USA residential gardens (Burghardt et al. 2010). Gardens with more native vegetation had higher caterpillar biomass leading to flow-on effects by improving habitat quality for insectivorous birds in Delaware, USA (Burghardt et al. 2010) and Washington, DC, USA (Narango et al. 2017) residential gardens. The abundance or occurrence of certain bee species and guilds was positively associated with native plant occurrence in Ohio, USA (Pardee and Philpott 2014) and Melbourne, Australia (Threlfall et al. 2015). Therefore, while the wildlife gardens in my study were not found to be benefiting the local butterfly community, they could be benefiting other urban wildlife.

Wildlife gardens in my study appear to have no distinct advantage over control gardens in supporting butterflies. Given butterflies are nectar generalists and the butterfly communities were similar across the garden types, it is likely only generalist species will be supported. Butterfly species, such as the Grass Blue or Australian

114

Painted Lady found in the gardens, would be considered generalists if they exhibit one of more of the following traits: multiple larval host plants, occur across multiple habitat types, have good mobility or dispersal ability, wide adult diet preferences, or are multivoltine. Other species known to be in the region, but not found in this study, may be limited by mobility or habitat affinity (Bergerot et al. 2010; Olivier et al. 2016), rather than floral resource availability as studied here.

The areas targeted for wildlife gardening by this program are near or along established greenways and revegetated creek lines which were assumed to harbour higher species richness than the adjacent urban matrix (Vergnes et al. 2012, 2013). Despite this proximity, only a small portion, 12%, of the regional butterfly community was found in any of the gardens in this study, and only accounts for 57% of the known local urban butterfly community (see Chapter 2). This difference is likely contextual as results from Chapter 2 show a decline in butterfly richness and abundance above 25% impervious cover. Impervious surface cover ranged from 27-54% for the individual gardens within a 500 m buffer (30-60% in the sampling cells). Thus, these gardens are at a point in the landscape where a lower portion of the butterfly community is present and more specialized members of the community are not likely to be found regardless of whether wildlife gardens provide additional resources (Kuschel 1990; Thompson 2006; Bergerot et al. 2010).

Goddard et al. (2010) argues that wildlife gardening efforts are more likely to be effective if spatially consolidated. While I was not able to directly address this idea in this study, I would suggest the landscape context of wildlife gardening programs and activities should also be considered when designing a garden or establishing a program, in addition to the spatial aggregation of such activities, as suggested by Goddard et al. (2010). In the case of wildlife gardens used in this study, their establishment in areas with greater than 30% impervious surface cover effectively means they may not be particularly useful for butterfly conservation for two reasons: first, butterfly richness and abundance starts to decrease above 25% imperviousness (Chapter 2), and second, more specialized species (either habitat or larval host plant specialists) are not likely to be found within residential urban areas (Thompson 2006; Bergerot et a. 2010). Thus, encouraging residents to install larval host plants as part of

115 their wildlife friendly gardening activities for a butterfly not likely to occur in the local area does not make sense ecologically (Botha and Botha 2006; Hunt et al. 2007) or financially (assuming the program provides or pays for the plants).

Wildlife gardening programs and activities may have greater impact on biodiversity if they focus on expanding particular local habitat types or features, limit their target taxa (e.g., birds, butterflies, lizards), or concentrate on a few locally important species. A narrower focus would allow programs to be more ecologically thoughtful in their recommendations potentially leading to greater impact, but should also make evaluating their impact easier to ascertain. Wildlife gardening participants in this study were incentivised to plant local native species (20 free seedlings) and had a consultation with a landscape designer, but were otherwise free to establish or manage gardens of their choosing and interest. Thus, the wildlife gardens ranged from a homeowner who merely stopped maintaining part of the garden, to those who installed ponds, or completely changed the composition of their gardens to native plants. Since these varied activities benefit different aspects of urban biodiversity they may not have a collective impact as habitat within the urban matrix for certain taxa groups. Disparate or one off efforts may not act as stepping stones or provide adequate habitat patches to maintain a population as intended. The lack of a difference between wildlife and both types of control gardens in this study demonstrates that these wildlife gardens have not collectively benefited the local butterfly community. This outcome is not surprising given most of the wildlife gardens in this study were not created to target butterflies, they were created to generally encourage urban wildlife. Narrowing the focus of a general wildlife gardening program could also lend itself to the spatial consolidation of resources and efforts advocated by Goddard et al. (2010). For instance, native birds respond positively to collective management of required resources within yards and streetscapes (White et al. 2005; Belaire et al. 2014), but research is lacking for other taxa.

To date there has not been a robust evaluation of either in situ wildlife gardening activities or programs in terms of how successful they are at increasing biodiversity or benefiting target species or taxa. Although, there have been a few formal studies assessing the potential of yards to provide wildlife habitat based on plants and other

116 habitat quality features (Smith et al. 2006b; Widows and Drake 2014), no studies exist specifically looking at the impact of wildlife gardens compared to other garden types for butterflies. While some studies have assessed the impact of management practices on butterflies (Shwartz et al. 2013; Fontaine et al. 2016; Muratet and Fontaine 2015) and of factors affecting garden use by butterflies within the landscape (Di Mauro et al. 2007; Matteson and Longelletto 2012; Olivier et al. 2016) none have directly compared gardens managed and not managed for wildlife. Without direct comparison between gardens managed or not managed for wildlife in relation to their target species or taxa I cannot determine the true benefit of wildlife friendly gardening practices.

CONCLUSIONS:

In conclusion, wildlife gardens in this study did not provide added benefit to the local butterfly community. Butterfly richness, abundance, and community composition were not different across the garden types, further indicating that these wildlife gardens did not provide a local spill-over effect of butterflies into the local landscape. They did provide more native floral resources than control gardens which may benefit other taxa within the urban matrix. To better understand the impact of wildlife gardening activities and determine success of locally promoted wildlife gardening programs, robust evaluation methods, similar to the study presented here, are needed.

REFERENCES:

Bates D, Maechler M, Bolker B, Walker S (2017) Linear mixed-effects models using 'Eigen' and S4. R-Project Belaire JA, Whelan CJ, Minor ES (2014) Having our yards and sharing them too: the collective effects of yards on native bird species in an urban landscape. Ecological Applications 24:2132-2143 Bergerot B, Fontaine B, Renard M, Cadi A, Julliard R (2010) Preferences for exotic flowers do not promote urban life in butterflies. Landscape and Urban Planning 96:98-107 Beumer C, Martens P (2015) BIMBY’s first steps: a pilot study on the contribution of residential front-yards in Phoenix and Maastricht to biodiversity, ecosystem services and urban sustainability. Urban Ecosystems 19:45-76

117

Botha C, Botha J (2006) Bring butterflies back to your garden. KwaZulu-Natal Coastal Branch of the Botanical Society of South Africa, Mayville, South Africa. Burghardt KT, Tallamy DW, Philips C, Shropshire KJ (2010) Non-native plants reduce abundance, richness, and host specialization in lepidopteran communities. Ecosphere 1:art11 Carvell C, Westrich P, Meek WR, Pywell RF, Nowakowski M (2006) Assessing the value of annual and perennial forage mixtures for bumblebees by direct observation and pollen analysis. Apidologie 37:326-340 Chamberlain DE, Vickery JA, Glue DE, Robinson RA, Conway G, J., Woodburn RJW, Cannon AR (2005) Annual and seasonal trends in the use of garden feeders by birds in winter. Ibis 147:563-575 Clarke KR, Green RH (1988) Statistical design and analysis for a 'biological effects' study. Marine Ecology - Progress Series 46:213-226 Clarke KR, Warwick RM (2001) Change in Marine Communities: an approach to statistical analysis and interpretation. 2nd edn. PRIMER-E Ltd, Plymouth, UK. Collier N, Mackay DA, Benkendorff K, Austin AD, Carthew SM (2006) Butterfly communities in South Australian urban reserves: Estimating abundance and diversity using the Pollard walk. Austral Ecology 31:282-290 Comba L, Corbet SA, Barron A, Bird A, Collinge S, Miyazaki M, Powell M (1999) Garden flowers: insect visits and the floral reward of horticulturally-modified variants. Annals of Botany 83:73-86 Corbet SA (2000) Butterfly nectaring flowers: butterfly morphology and flower form. Entomologia Experimentalis et Applicata 96:289-298 Corbet SA et al. (2001) Native or Exotic? Double or Single? Evaluating plants for pollinator friendly gardens. Annals of Botany 87:219-232 Curtis RJ, Brereton TM, Dennis RLH, Carbone C, Isaac NJB, Diamond S (2015) Butterfly abundance is determined by food availability and is mediated by species traits. Journal of Applied Ecology 62:1676-1684 Dennis RLH (2010) A resource-based habitat view for conservation: butterflies in the Brittish landscape. 1 edn. John Wiley & Sons Ltd., United Kingdom.

118

Di Mauro D, Dietz T, Rockwood L (2007) Determining the effect of urbanization on generalist butterfly species diversity in butterfly gardens. Urban Ecosystems 10:427-439 ESRI (2011) ArcGIS Desktop: Release 10. Environmental Systems Research Institute, Redlands, CA. Ezzeddine M, Matter SF (2008) Nectar flower use and electivity by butterflies in sub- alpine meadows. Journal of the Lepidopterists' Society 62:138-142 Feber RE, Smith H, Macdonald DW (1996) The effects on butterfly abundance of the management of uncropped edges of arable fields. Journal of Applied Ecology 33:1191-1205 Field R (2013) Butterflies: Indentification and life history. Museum Victoria Publishing, Melbourne, Victoria Fontaine B, Bergerot B, Le Viol I, Julliard R (2016) Impact of urbanization and gardening practices on common butterfly communities in France. Ecology and Evolution 6:8174-8180 Galbraith JA, Stanley MC, Jones DN, Beggs JR (2017) Experimental feeding regime influences urban bird disease dynamics. Journal of Avian Biology 48:700-713 Gaston KJ, Fuller RA, Loram A, MacDonald C, Power S, Dempsey N (2007) Urban domestic gardens (XI): variation in urban wildlife gardening in the United Kingdom. Biodiversity and Conservation 16:3227-3238 Gaston KJ, Smith RM, Thompson K, Warren PH (2005) Urban Domestic Gardens (II): experimental tests of methods for increasing biodiversity. Biodiversity and Conservation 14:395-413 Goddard MA, Dougill AJ, Benton TG (2010) Scaling up from gardens: biodiversity conservation in urban environments. Trends Ecol Evol 25:90-98 Gomez JM, Bosch J, Perfectti F, Fernandez JD, Abdelaziz M, Camacho JP (2008) Association between floral traits and rewards in Erysimum mediohispanicum (Brassicaceae). Annals of Botany 101:1413-1420 Grace Detailed-GIS Services (2012) Directly connected imperviousness compilation for Melbourne Water selected catchments. Melbourne, Australia.

119

Hahs A, McDonnell MJ, Holland K, Caryl F (2009) Biodiversity of Metropolitan Melbourne. Australian Research Centre for Urban Ecology, Melbourne, Australia. Hunt L, Grund R, Keane D, Forrest J (2007) Attracting butterflies to your garden: what to grow and conserve in the Adelaide Region. Butterfly Conservation Australia, Australia Ibbe M, Milberg P, Tunér A, Bergman K-O (2011) History matters: Impact of historical land use on butterfly diversity in clear-cuts in a boreal landscape. Forest Ecology and Management 261:1885-1891 Jonason D, Andersson GK, Ockinger E, Rundlof M, Smith HG, Bengtsson J (2011) Assessing the effect of the time since transition to organic farming on plants and butterflies. The Journal of Applied Ecology 48:543-550 Jones RE, Gilbert N, Guppy M, Nealis V (1980) Long distance movement of Pieris rapae. Journal of Animal Ecology 49:629-642 Kitching RL, Scheermeyer E, Jones RE, Pierce NE (1999) Biology of Australian Butterflies vol 6. Monographs on Australian lepidoptera. CSIRO Publishing, Collingwood, VIC, Australia Kummer JA, Bayne EM (2015) Bird feeders and their effects on bird-window collisions at residential houses. Avian Conservation and Ecology 10:6 Kuschel G (1990) Beetles in a suburban environment: a New Zealand case study. New Zealand Department of Scientific and Industrial Research, Aukland, New Zealand Kuznetsova A, Brockhoff PB, Christensen RHB (2016) Tests in linear mixed effects models. The R-Project Lerman SB, Milam J (2016) Bee fauna and floral abundance within lawn-dominated suburban yards in Springfield, MA. Annals of the Entomological Society of America 109:713-723 Lerman SB, Warren PS (2011) The conservation value of residential yards: linking birds and people. Ecological Applications 21:1327-1339 Leston L, Koper N (2017) Urban rights-of-way as extensive butterfly habitats: A case study from Winnipeg, Canada. Landscape and Urban Planning 157:56-62

120

Levy JM, Connor EF (2004) Are gardens effective in butterfly conservation a case study with the pipevine swallowtail, Battus philenor. Journal of Insect Conservation 8:323-330 Lütolf M, Guisan A, Kienast F (2009) History matters: relating land-use change to butterfly species occurrence. Environmental Management 43:436-446 Matteson KC, Langellotto GA (2011) Small scale additions of native plants fail to increase beneficial insect richness in urban gardens. Insect Conservation and Diversity 4:89-98 Matteson KC, Langellotto GA (2012) Evaluating community gardens as habitat for an urban butterfly. Cities and the Environment 5:Article 10 Mumaw L, Bekessy S (2017) Wildlife gardening for collaborative public–private biodiversity conservation. Australasian Journal of Environmental Management 24:242-260 Muratet A, Fontaine B (2015) Contrasting impacts of pesticides on butterflies and bumblebees in private gardens in France. Biological Conservation 182:148-154 Narango DL, Tallamy DW, Marra PP (2017) Native plants improve breeding and foraging habitat for an insectivorous bird. Biological Conservation 213:42-50 New TR (1997) Are Lepidoptera an effective 'umbrella group' for biodiversity conservation? Journal of Insect Conservation 1:5-12 Öckinger E et al. (2012) The landscape matrix modifies the effect of habitat fragmentation in grassland butterflies. Landscape Ecology 27:121-131 Oksanen J et al. (2017) Vegan: Community Ecology Package. R package version 2.0-2. The R Project Olivier T, Schmucki R, Fontaine B, Villemey A, Archaux F (2016) Butterfly assemblages in residential gardens are driven by species’ habitat preference and mobility. Landscape Ecology 31:865-876 Orr A, Kitching R (2010) The Butterflies of Australia. Allen & Unwin, Sydney, New South Wales, Australia Owens J (1991) The Ecology of a Garden: The First Fifteen Years. University Press, Cambridge, Great Britian Paradis E, Claude J, Strimmer K (2004) APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20:289-290

121

Pardee GL, Philpott SM (2014) Native plants are the bee’s knees: local and landscape predictors of bee richness and abundance in backyard gardens. Urban Ecosystems 17:641-659 Pohl NB, Van Wyk J, Campbell DR (2011) Butterflies show flower colour preferences but not constancy in foraging at four plant species. Ecological Entomology 36:290-300 Pyle RM (1992) Handbook for Butterfly Watchers. Houghton Mifflin Company, New York R Development Core Team (2015) R: A language and environment for statistical computing. R Foundataion for Statistical Computing, Vienna, Austria Shackleton K, Ratnieks FLW (2015) Garden varieties: how attractive are recommended garden plants to butterflies? Journal of Insect Conservation 20:141-148 Shaw A, Miller K, Wescott G (2013) Wildlife gardening and connectedness to nature: engaging the unengaged. Environmental Values 22:483-502 Shwartz A, Muratet A, Simon L, Julliard R (2013) Local and management variables outweigh landscape effects in enhancing the diversity of different taxa in a big metropolis. Biological Conservation 157:285-292 Smith LS, Broyles MEJ, Larzleer HK, Fellowes MDE (2015) Adding ecological value to the urban lawnscape. Insect abundance and diversity in grass-free lawns. Biodiversity and Conservation 24:47-62 Smith RM, Gaston KJ, Warren PH, Thompson K (2006a) Urban domestic gardens (VIII) : environmental correlates of invertebrate abundance. Biodiversity and Conservation 15:2515-2545 Smith RM, Thompson K, Hodgson JG, Warren PH, Gaston KJ (2006b) Urban domestic gardens (IX): Composition and richness of the vascular plant flora, and implications for native biodiversity. Biological Conservation 129:312-322 Smith RM, Warren PH, Thompson K, Gaston KJ (2006c) Urban domestic gardens (VI): environmental correlates of invertebrate species richness. Biodiversity and Conservation 15:2415-2438 Soga M, Koike S (2012) Relative importance of quantity, quality and isolation of patches for butterfly diversity in fragmented urban forests. Ecological Research 27:265-271

122

Stefanescu C, Traveset A (2009) Factors influencing the degree of generalization in flower use by Mediterranean butterflies. Oikos 118:1109-1117 Thompson K (2006) No nettles required: the truth about wildlife gardening. Transworld Publishers, London Thompson K, Hodgson JG, Smith RM, Warren PH, Gaston KJ (2004) Urban domestic gardens (III): Composition and diversity of lawn floras. Journal of Vegetation Science 15:373-378 Threlfall CG et al. (2017) Increasing biodiversity in urban green spaces through simple vegetation interventions. Journal of Applied Ecology 54:1874-1883 Threlfall CG, Walker K, Williams NSG, Hahs AK, Mata L, Stork N, Livesley SJ (2015) The conservation value of urban green space habitats for Australian native bee communities. Biological Conservation 187:240-248 Threlfall CG, Williams NSG, Hahs AK, Livesley SJ (2016) Approaches to urban vegetation management and the impacts on urban bird and bat assemblages. Landscape and Urban Planning 153:28-39 Tiple AD, Deshmukh VP, Dennis RL (2005) Factors influencing nectar plant resource visits by butterflies on a university campus: implications for conservation. Nota lepidopterologica 28:213-224 Vergnes A, Kerbiriou C, Clergeau P (2013) Ecological corridors also operate in an urban matrix: A test case with garden shrews. Urban Ecosystems 16:511-525 Vergnes A, Viol IL, Clergeau P (2012) Green corridors in urban landscapes affect the arthropod communities of domestic gardens. Biological Conservation 145:171- 178 White JG, Antos MJ, Fitzsimons JA, Palmer GC (2005) Non-uniform bird assemblages in urban environments: the influence of streetscape vegetation. Landscape and Urban Planning 71:123-135 Widows SA, Drake D (2014) Evaluating the National Wildlife Federation's Certified Wildlife Habitat™ program. Landscape and Urban Planning 129:32-43 Williams MR (2008) Asssesing diversity of diurnal lepidoptera in habitat fragments: testing the efficiency of strip transects. Environmental Entomology 37:1313- 1322

123

Yahner RH (2001) Butterfly communities in residential landscape of Central Pennsylvania. Northeastern Naturalist 8:113-118 Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith G (2009) Mixed Effects Models and

Extensions in Ecology with R. Springer Science+Business Media, LLC, New York

124

Chapter 5

Synthesis

In this thesis, I assessed the response of the butterfly community in south-eastern Melbourne, Australia, to impervious-surface cover and the availability of key resources in the landscape. In this synthesis, I draw together themes that emerged from these investigations, highlight the implications for future research, and discuss ways landscape managers can improve the urban matrix to support populations of butterflies in cities.

LANDSCAPE-SCALE PROCESSES:

I found the species richness and abundance of butterflies responded differentially through space and time to elements within the urban landscape. Regardless of the scale studied, increases in the percent cover of impervious surfaces (i.e., imperviousness) within the urban matrix had a negative effect on the butterfly community (Chapters 2-4), the percent cover and species richness of butterfly larval host plants (Chapter 3), and the abundance of adult food plants (Chapter 2). While my findings further confirm the literature showing the negative effects of urbanization upon both the butterfly community and vegetation, my use of an impervious-surface gradient to stratify my site selection sets my study apart from most. Much of the urban ecology literature is entrenched in patch matrix theory (Forman 1995) and uses parks embedded within the urban matrix to study the effects of urbanization on wildlife.

My application of the gradient paradigm (Whittaker 1967; McDonnell and Pickett 1990) across the urban matrix allowed me to assess the spatial distribution of the butterfly community resulting from variation in the availability of resources for butterflies along this gradient. Further, an effort was made to distribute my sites evenly along the gradient allowing for a more mechanistic and detailed understanding of the response of the butterfly community and vegetation across the urban matrix compared to studies with few sites or sites grouped at a few points along an urban gradient. My gradient sampling design showed a decline in both butterfly species richness and abundance starting at around 25% imperviousness, whereas larval host plant (LHP) species richness started to decline above 40% and LHP cover started to

125 decline above 1%. Thus, the use of a continuous impervious gradient yielded insight into the greatest point of impact along an urban gradient and provides more detailed response curves.

Using this approach, my work demonstrates the importance of landscape context for efforts to increase urban wildlife populations. The wildlife gardening program included in my study sits within the landscape where imperviousness is greater than 25%. In Chapter 2, I found that this was the point in the landscape where butterfly species richness and abundance start to decline. Thus, it is hard to discern if the lack of a difference in the butterfly communities between garden types I found (Chapter 4) is due to a lack of resources targeted to butterflies or pollinators, or because the gardens were created at a point in the landscape where butterfly communities were already diminished. It is only through the use on a continuous gradient of imperviousness that this was able to be revealed, and this suggests that future research on the effectiveness of wildlife gardening needs to account for landscape context and landscape-scale processes.

FINE-SCALE PROCESSES:

The availability of required resources within the matrix had an overall positive influence on the butterfly community despite the negative impact of imperviousness. The effect of these resources, though, were mostly small and in some cases only benefitted common butterflies. An interesting outcome from this thesis, was that at larger scales, imperviousness had a stronger effect than floral abundance (Chapter 2), but larval host plants (LHPs) had a stronger effect than imperviousness (Chapter 3). However, at fine scales within individual gardens, none of these variables appeared to influence the butterfly community (Chapter 4). Ideally both required resources would have been studied and subsequently interrogated concurrently, but the level of effort required to achieve that in one study was not possible given the scale at which I was working (entire urban gradient, 24 km diameter). Within community ecology it is understood that the most limiting factor is often the most important factor (Levin 1970). Thus, my data indicates that the limited availability of LHPs within the matrix appears to limit the butterfly community more than adult food resources within the matrix. 126

The unique approach taken in this thesis to evaluate the composition (native and exotic) of available plant resources, allows for far greater insight than the current urban ecology literature can offer. Temporal variation in floral resource availability is not often assessed in urban butterfly studies, but it obviously influences the butterfly community through variation in different components of vegetation (total, native, and exotic) over the season, as demonstrated in this thesis. Somewhat surprisingly, the presence of native vegetation did not have a strong positive impact on the butterfly community as expected. Native floral resources had a small positive influence on butterfly species richness across the urban matrix (Chapter 2), but higher native floral species richness and abundance in wildlife gardens did not translate to higher butterfly species richness, abundance, or different community composition in those gardens compared to control gardens (Chapter 4). Similarly, native larval host plants (LHP) were not as important to the butterfly community as expected, but they were also not as common within the urban matrix as exotic LHPs (Chapter 2). Interestingly, while exotic LHPs were important to the butterfly community in general, or to certain species specifically (Chapter 3), exotic floral resources had a mostly negative influence, particularly on adult butterfly abundance (Chapter 2). Therefore, while exotic plants are important, they have both negative and positive impacts on the butterfly community depending on which part of butterfly lifecycle was being assessed. Thus, plant composition within the urban matrix is extremely important to the butterfly community. While the addition of targeted native plant resources may have a positive impact on butterfly species richness, the positive benefits provided by exotic vegetation, especially for larval stages, should not be over-looked.

The response of the less common butterflies was mostly, as expected, at the local scale, but did show a small positive response to landscape-scale factors (Chapter 2). Increased impervious surface cover and/or the lack of unmanaged greenspace, also appeared to influence their abundance more than the two most common species (Chapters 2-4). This indicates variety of responses within this group, where some of the more common species (e.g. Common Browns (Heteronympha merope), Green Grass Darts (Ocybadistes walker), and Australian Painted Ladies (Vanessa kershawi) are able to survive in this urban setting, but not thrive in the same way as Grass Blues

127

(Zizina otis labradus) and Small Whites (Vanessa kershawi). While there were sufficient records of Common Browns and Green Grass Darts surveyed to discern a significant relationship with one or two of their LHPs, no such relationship was found for the Australian Painted Ladies, who use Arctotheca calendula, a common weed within the sampling area as a larval host plant. It is possible that regular mowing discounts A. calendula’s widespread use as an LHP, because mowing likely reduces the quality of this plant and/or restricts how long it may be able to support larvae as they develop. The disparities in abundances among these less common butterflies (Chapters 2 and 4) raise valid questions about how abiotic factors such as routine management practices affect the butterfly community.

LIMITATIONS OF THIS THESIS:

In this thesis, I took a landscape-scale approach, which while informative, has also limited some of the finer-scale inferences that could be drawn. My study assessed the butterfly community in the southeast suburbs of Melbourne, Australia. The butterfly community sampled is likely only a subset of the potential butterfly community across the greater metropolitan area. For example, the western portions of metropolitan Melbourne would likely support different butterfly species given the extensive volcanic plain grasslands that existed pre-European settlement (Hahs et al. 2009). I did record five additional species incidentally in the study region, and there are potentially other species that went undetected, despite the level of survey effort. Thus, the urban butterfly community species list for southeast Melbourne presented herein is likely not complete.

Australia’s butterfly community is relatively understudied compared to that of places such as the United Kingdom. There is a general lack of natural history information and insight about the traits and preferences of Australian butterfly species. Additionally, given the spatial scale of my study, it was not feasible to obtain information about direct interactions between each butterfly species and plants (e.g. larval host plants (LHP) and adult food plants) recorded in my surveys. The ecological interaction literature outside of urban areas is extensive, but only a handful of studies exist for these types of interactions within urban areas. Whilst relationships between some butterfly species and their LHPs were indicated in this thesis (Chapter 3), breeding 128 records are needed to confirm if the urban matrix is providing suitable reproductive habitat to those butterfly species (Hardy and Dennis 1999; Dennis 2010).

Similarly, I was not able to record which flowers were used as nectar sources by butterflies (Chapters 2 and 4), as direct observations of feeding were not made. I made the most ecologically relevant decisions I could in restricting my floral species list, but certainly the list of nectar providing species I used may also include plants that are not that well utilised by butterflies. Greater information on the natural history, nectar preferences and foraging behaviours of species in this study would be useful to gather in the future. For example, I have incidentally observed one member of the urban butterfly community regularly probing flowers that are thought not to provide nectar. It could be this butterfly species also consumes pollen or that these floral species actually do provide nectar. Either way, the paucity of natural history and species trait information for Australian butterflies (and for many plants in urban environments) limits the conclusions that can be drawn from my data.

FUTURE DIRECTIONS FOR CONTINUED AND COMPLEMENTARY RESEARCH:

Building a truly mechanistic understanding of the impact of the urbanization matrix on the butterfly community will require a combination of landscape-scale studies (as conducted here) and finer-scale studies centered on understanding a butterflies’ ecological interactions with their resources. Behavioural studies of urban butterflies, similar to those conducted in rural settings, would provide valuable information on how to better manage the urban matrix for either continued population persistence or potential species conservation. Larval host plant (LHP) selection, use, and overall quality (e.g., amount of nitrogen, plant tissue pollution levels, plant age, etc.) may change with increasing urbanization, which may affect survival and fitness of larvae and subsequently adult butterflies. Similarly, the quality of the nectar provided by plants, particularly ornamentals, within the urban matrix can affect butterfly fitness (Geister et al. 2008; Cahenzli and Erhardt 2013). These would provide fruitful avenues for future research.

Dispersal ability may also differ in butterflies from urban, suburban, and rural settings. Given higher floral species richness in residential areas, it is possible that urban

129 butterflies may be more flexible in their floral selection than rural butterflies. The butterfly species found in urban areas may be better able to disperse and move between available floral patches, including those that provide adult and larval resources. Plants flower over a longer period in urban settings compared to rural settings (Parlow 2011), which could also potentially lead to phenological changes in butterflies such as living longer, shifted flight periods, or more broods per year. Future research should investigate these shifts in floral quality imposed by urbanization and their interaction with species traits, given the rapid pace of urbanization globally and its potentially serious, negative impacts on urban biota.

Future research should take a longer time frame to investigate the strong declines seen in butterfly species richness and abundance, that started at 25% imperviousness (Chapter 2). More years of repeated sampling would be needed to determine if this point is the threshold for the butterfly community’s tolerance of imperviousness within the landscape, the variation in tolerance between species or as the place where required resources shift in availability. Continued examination of floral resources for both adults and larvae would also provide interesting insights not only into their temporal and spatial variation, but their value and ability to mitigate the negative impacts of urbanization, on the butterfly community.

MANAGEMENT LESSONS FOR THE URBAN MATRIX:

My results show that the availability of required resources, floral nectar and LHPs, within the urban matrix can mitigate the negative effects of urbanization. While my overall results measured those of the butterfly community, each species therein has its own response to urbanization. Therefore, amendments, such as additions of floral plants or appropriate LHPs, and management strategies, such as reduced mowing or pesticide use, within the urban matrix need to be specifically targeted to the needs of species of concern or more thoughtfully considered for specific taxa of interest (e.g., not all nectar sources serve pollinators equally). Further, the landscape context of those actions is also important and should be considered before implementation. For example, increased general nectar resources within residential gardens did not have the intended positive impact on the butterfly community, and instead other actions should be considered, such as increasing specific LHPs or changing the location of 130 management activities in the landscape. These are considerations that both wildlife gardening programs and urban greenspace managers should take into account before implementing programs or management practices broadly aimed at benefiting urban wildlife.

CONCLUSION:

In summary, this thesis was undertaken with the aim of examining the response of butterflies to urbanization by comparing assemblages across resource gradients within the urban matrix. I specifically focussed on understanding the role of adult food plants, larval host plants, and the efficacy of wildlife gardening in supporting butterfly communities. The gradient approach used here has not been employed widely to date, and is proposed as a new way of studying responses of wildlife to urbanization impacts. Importantly, this thesis showed that at landscape scales, increases in the species richness of LHPs positively influenced the community over and above the negative impacts of urbanization. I also showed that the effectiveness of WG’s needs to be considered in light of their position along an urban gradient. Collectively this thesis improves knowledge of urban ecosystems and contributes to expanding urban ecology theory, particularly via the use of a sophisticated landscape gradient. This approach can be used in future studies to improve knowledge of urban impacts to biodiversity, and suggest practical ways to reduce these impacts.

REFERENCES:

Cahenzli F, Erhardt A (2013) Nectar amino acids enhance reproduction in male butterflies. Oecologia 171:197-205 Dennis RLH (2010) A resource-based habitat view for conservation: butterflies in the Brittish landscape. 1 edn. John Wiley & Sons Ltd., United Kingdom Geister TL, Lorenz MW, Hoffmann KH, Fischer K (2008) Adult nutrition and butterfly fitness: effects of diet quality on reproductive output, egg composition, and egg hatching success. Frontiers in Zoology 5:10 Hahs A, McDonnell MJ, Holland K, Caryl F (2009) Biodiversity of Metropolitan Melbourne. Australian Research Centre for Urban Ecology, Melbourne, Australia

131

Hardy PB, Dennis RLH (1999) The impact of urban development on butterflies within a city region. Biodiversity and Conservation 8:1261-1279 Levin SA (1970) Community equlibria and stability, and an extension of the competative exclusion principle. American Society of Naturalists 104:413-423 McDonnell MJ, Pickett STA (1990) Ecosystem structure and function along urban-rural gradients: an unexpected opportunity for ecology. Ecology 71:1232-1237 Parlow E (2011) Urban Climate. In: Niemela J (ed) Urban Ecology: patterns, processes, and applications. Oxford University Press, New York, pp 31-45 Whittaker RH (1967) Gradient analysis of vegetation. Biological Reviews 42:207-264

132

Appendices Chapter 2

Appendix 2 Table 1 – Butterfly flight periods coinciding with our sampling period, plus an additional month on either end, for the butterflies identified during our surveys from November 2014 through March 2015. Figures represent occurrence records indicating the percentage of adults sampled in each month from Field (2013). Gray shading indicates the most likely period to detect the species ‘on the wing’, with dark gray indicating the peak flight period(s) of each species.

Species Oct Nov Dec Jan Feb Mar April Green Grass Dart 7 20 15 7 19 18 11 Ocybadistes walkeri Australian Painted Lady 18 17 20 15 7 5 3 Vanessa kershawi Yellow Admiral 14 16 22 11 8 7 4

Vanessa itea Common Brown 1 14 32 22 12 10 7 Heteronympha merope Imperial Hairstreak 4 27 30 20 18 1 Jalmenus evagoras Ringed Xenica 1 7 43 28 14 6 Geitoneura acantha

Marbled Xenica 1 9 19 32 25 11 3 Geitoneura klugii Meadow Argus 5 7 15 21 15 14 10 Junonia villida Dainty Swallowtail 2 7 13 13 20 29 13 Papilio anactus Cabbage White 8 7 12 12 15 13 15 Pieris rapae Common Grass Blue 8 8 15 16 21 15 9 Zizina otis labradus Splendid Ochre 3 19 59 19 Trapezites symmomus

Orange Palm-Dart 10 30 30 Cephrenes augiades Salt-bush Blue 6 13 9 9 16 25 9 Theclinesthes serpentatus

133

Appendix 2 Table 2 – All floral resources identified along transects and whether they produce nectar, based on an extensive literature search in Web of Science and Google Scholar (August/September 2016). If no information was found for a species, we assigned a value of yes or no if another species in that genus was known to produce nectar or not, respectively. Because some genera have been poorly studied and no empirical information could be obtained, these have been categorized as no.

Genus or species Nectar? Genus or species Nectar? Abelia spp. Yes Justicia brandegeeana Yes Abutilon x hybridium Yes Justicia carnea Yes Acacia lonigifolia Yes Kalanchoe blossfeldiana Yes Acacia spp. Yes Kennedia nigricans Yes Acanthus mollis Yes Kniphofia spp. Yes Achillea millefolium Yes Koelreuteria paniculata No Acokanthera oblongifolia Yes Kunzea ericoides Yes Actinotus hellianthi Yes Lagerstroemia indica Yes Aechmea spp. Yes Lagunaria patersonii Yes Aeonium arboreum Yes Lampranthus deltoides Yes Aeonium haworthii Yes Lampranthus multiradiatus Yes Agapanthus praecox Yes Lampranthus spp. Yes Ageratum spp. Yes Lantana camara Yes Ajuga reptans Yes Lantana montevidensis Yes Alcea spp. Yes Lantana spp. Yes Alisma plantago-aquatica Yes Lapsana communis Yes Allium spp. Yes Lathyrus odoratus Yes Aloe spp. Yes Lavandula pinnata Yes Aloysia triphylla No Lavandula spp. Yes Alstroemeria spp. Yes Lavatera maritima Yes Alyogyne hakeifolia Yes Lavatera x clementii Yes Alyogyne huegleii Yes Leonotis leonurus Yes Alyogyne huegleii 'blue heeler' Yes Leptospermum petersonii Yes Amaranthus cruentus No Leptospermum spp. Yes Amaryllis belladonna Yes Leucanthemum spp. Yes Anenome spp. Yes Leucanthemum x superbum Yes cultivar Anigozanthos spp. Yes Leucospermum glabrum Yes Anthemis tinctoria Yes Leucospermum spp. Yes Antirrhinum spp. cultivar Yes Ligustrum japonicum Yes Aptenia cordifolia Yes Ligustrum lucidum Yes Aquilegia spp. Yes Ligustrum spp. Yes Araujia sericifera Yes Lilium longiflorum Yes Arbutus unedo Yes Limonium perezii Yes Arbutus x andrachnoides Yes Linum marginale Yes

134

Genus or species Nectar? Genus or species Nectar? Arctotheca calendula No Linum trigynum Yes Arctotis x hybrida No Liriope muscari No Argyranthemum frutescens Yes Lithodora diffusa Yes Aristea ecklonii No Lobelia cardinalis Yes Armeria maritima Yes Lobelia spp. (annual) Yes Arthropodium cirratum Yes Lobularia maritima Yes Arthropodium minus Yes Lonicera spp. Yes Asparagus aethiopicus Yes Lophostermon confertus Yes Aster x frikartii Yes Loropetalum chinense No Azaleas spp. Yes Luma apiculata No Backhousia citriodora Yes Lychnis coronaria Yes Baeckea linifolia No Lycianthes rantonnetii No Banksia spp. Yes Lysiosepalum involucratum No Begonia fuchsioides No Lythrum hyssopifolia Yes Begonia spp. No lythrum junceum Yes Bellis perennis Yes Lythrum salicaria Yes Bergenia x schmidtii Yes Mackaya bella No Bidens ferulifolia Yes Magnolia spp. No Bidens pilosa Yes Malus domestica Yes Bidens spp. cultivar Yes Malus spp. Yes Billardiera heterophylla No Malva neglecta Yes Borage officinalis Yes Malva spp. Yes Bougainvillea spp. Yes Malvaviscus arboreus Yes Bouvardia spp. Yes Mandevilla bolivinensis Yes Brachychition acerifolius Yes Mandevilla sanderi cultivar Yes Brachychiton populneus Yes Matricaria chamomilla Yes Brachychiton x belladonna carabelle Yes Matthiola incana Yes Brachyscome spp. Yes Maurandya barclayana Yes Brachysome multifida cultivar Yes Medicago lupulina Yes Brassica spp. Yes Medicago polymorpha Yes Brugmansia x candida Yes Medicago sativa Yes Brunsfelsia latifolia Yes Melaleuca armillaris Yes Buddleja spp. Yes Melaleuca decussata Yes Bulbine semiparbata No Melaleuca linariifolia Yes Burchardia umbellata Yes Melaleuca nesophila Yes Busaria spinosa Yes Melaleuca styphelioides Yes Bystropogon canariensis No Melaleuca thymifolia Yes Calendula spp. Yes Melaleucca spp. Yes Calicotome spinosa No Melia azedarach Yes Callistemon spp. Yes Melissa officianalis Yes Calluna vulgaris Yes Mentha aquatica Yes Calodendrum capense No Mentha spicata Yes Calystegia sepium Yes Mentha spp. Yes Calystegia silvatica Yes Metrosideros excelsa Yes

135

Genus or species Nectar? Genus or species Nectar? Camellia spp. Yes Minuria leptophylla Yes Campanula persicifolia Yes Mirabilis jalapa Yes Campanula poscharskyana Yes Modiola caroliniana No Campsis radicans Yes Muehlenbeckia complexa Yes Canna spp. Yes Myoporum insulare Yes Capparis spinosa Yes Myoporum spp. Yes Capsicum spp. Yes Myosotis sylvatica Yes Carpbrotus aequilaterus Yes Myrtus communis Yes Carpbrotus spp. Yes Nandina domestica No Cassia spp. Yes Nemesia spp. Yes Cassinia spp. No Nepeta cateria Yes Castanospermum australe Yes Nerium oleander Yes Catalpa bignonioides Yes Nigella damascena Yes Ceanothus thyrsiflorus Yes Ocimum basilicum Yes Celosia cristata Yes Oenothera biennis Yes Celosia plumosa Yes Oenothera lindheimeri Yes Cenia turbinata No Oenothera speciosa Yes Centaurea gymnocarpa Yes Olea europaea No Centaurium erythraea No Olearia ramulosa Yes Centranthus ruber Yes Origanum spp. Yes Cerastium tomentosum Yes Osteospermum fruticosum No Cerastium vulgare Yes Oxalis corniculata complex Yes Ceratopetalum gummiferum Yes Oxalis latifolia Yes Ceratostigma griffithii Yes Ozothamnus ferrugineus No Ceratostigma willmottianum Yes Ozothamnus spp. No Cestrum parqui Yes Pachystachys lutea Yes Chaenostoma cordatum No Pandorea spp. Yes Chamelaucium spp. Yes Papaver spp. Yes Chamelaucium uncinatum Yes Passiflora spp. Yes Chlorophytum comosum No Pelargonium citronellum Yes Choisya ternata Yes Pelargonium spp. Yes Chrysocephalum apiculatum Yes Penstemon serrulatus Yes Chrysocephalum semipapposum Yes Penstemon spp. Yes Cirsium vulgare Yes Pericallis × hybrida No Cistus spp. Yes Persicaria capitata Yes Citrus spp. Yes Petroselinum crispum Yes Citrus × limon Yes Petunia x hybrida Yes Clivia caulescens Yes Phaseolus vulgaris Yes Clivia spp. Yes Philadelphus coronarius Yes Clytostoma callistegioides No Philotheca myoporoides Yes Coleonema spp. Yes Phlomis fruticosa Yes Consolida ajacis Yes Phlox paniculata Yes Convolvulus angustissimus Yes Phlox spp. Yes Convolvulus arvensis Yes Photinia spp. Yes

136

Genus or species Nectar? Genus or species Nectar? Convolvulus cneorum Yes Phyla canescens Yes Convolvulus mauritanicus Yes Pieris japonica Yes Cordyline australis Yes Pimelea spp. Yes Coreopsis lanceolata Yes Pittosporum spp. Yes Correa alba Yes Plectranthus argentatus Yes Correa glabra Yes Plectranthus ecklonii Yes Correa reflexa Yes Plectranthus fruitcosus Yes Correa spp. Yes Plectranthus neochilus Yes Corymbia ficifolia Yes Plectranthus x mona lavender Yes Cosmos spp. Yes Plumbago spp. Yes Cotoneaster dammeri Yes Polygala myrtifolia Yes Cotoneaster horizontalis Yes Polygonum aviculare Yes Cotoneaster spp. Yes Pomaderris aspera Yes Cotula australis No Portulaca grandiflora Yes Cotyledon orbiculata Yes Portulaca oleracea Yes Crambe strigosa Yes Pratia pedunculata Yes Crassula coccinea Yes Primula spp. Yes Crassula falcotla Yes Prostanthera lasianthos Yes Crassula multicava Yes Prostanthera scutellarioides Yes Crassula spp. Yes Prostanthera spp. Yes Crassula tetragona Yes Protea neriifolia Yes Crocosmia spp. Yes Prunella vulgaris Yes Crowea exalata No Prunus spp. Yes Cucurbita spp. Yes Psidium spp. Yes Cuphea hyssopifolia Yes Psoralea pinnata Yes Cuphea ignea Yes Ptilotus exaltatus No Cuphea spp. cultivar Yes Punica granatum Yes Cydonia oblonga Yes Pyracantha angustifolia Yes Cymbalaria muralis Yes Pyracantha coccenia Yes Dahlia spp. Yes Ranunculus repens Yes Dahlia spp. - double bloom No Rhaphiolepis indica Yes Dampiera diversifolia Yes Rhododendron spp. Yes Daphne spp. Yes Rhus spp. Yes Darwinia citriodora Yes Ricinocarpos pinifolius Yes Daucus carota Yes Roldana petasitis No Delesperma cooperi Yes Rosa spp. Yes Delosperma spp. Yes Rosmarinus officinalis Yes Delphiunium spp. cultivar Yes Rotheca myricoides Yes Dianella spp No Rubus spp. Yes Dianthus caryophyllus Yes Rudbekia laciniata var. hortensia Yes Dianthus spp. Yes Russelia equisetiformis Yes Diascia spp. No Salpichroa origanifolia Yes Dicliptera sericea Yes Salvia 'Celestial Blue' Yes Dietes spp. No Salvia chamelaeagnea Yes

137

Genus or species Nectar? Genus or species Nectar? Digitalis spp. Yes Salvia chiapensis cultivar Yes Dissotis rotundifolia No Salvia farinacea Yes Duranta erecta Yes Salvia greggii Yes Echeveria spp. Yes Salvia involucrata Yes Echinacea purpurea Yes Salvia leucantha Yes Echinops spp. Yes Salvia mexicana Yes Echium candicans Yes Salvia microphylla Yes Echium plantagineum Yes Salvia nemorosa Yes Elaeocarpus reticulatus Yes Salvia officinalis Yes Epacris impressa Yes Salvia sclarea var. Turkestanica Yes Epidendrum radicans x secundum No Salvia splendens Yes Epilobium billardiereanum Yes Salvia spp. Yes Epilobium ciliatum Yes Salvia uliginosa Yes Epilobium spp. Yes Salvia 'Waverly' Yes Eremaphila spp. Yes Salvia 'wendy's wish' Yes Eremophila alternifolia Yes Sannantha pluriflora No Eremophila iaanii Yes Sannantha spp. No Eremophila nivea Yes Sannantha virgata No Erica cerinthoides Yes Santolina chamaecyparissus No Erica cinerea Yes Santolina pinnata subs. No neapolitana Erica glauca var. elegans Yes Satureja thymbra Yes Erica gracillis cultivar Yes Scabiosa japonica var alpina Yes Erica spp. Yes Scabiosa spp. Yes Erigeron glaucus Yes Scaevola spp. Yes Erigeron karvinskianus Yes Schinus molle Yes Erigeron spp. Yes Sedum album Yes Eriostemon spp. No Sedum mexicanum Yes Erodium cicutarium Yes Sedum pachyphyllum Yes Erysimum cheiri Yes Sedum reflexum Yes Erythrina crista-galli Yes Sedum telephium Yes Erythrina x sykesii Yes Senecio cineraria Yes Escallonia iveyi Yes Senna artemisioides Yes Escallonia macranthra 'rubra' Yes Serissa spp. No Escallonia rubra Yes Silene pendula Yes Escholtzia californica Yes Silene vulgaris Yes Eucalyptus tree Yes Sisyrinchium bellum No Eucomis spp. Yes Sisyrinchium iridifolium No Euonymus japonica Yes Solanum aviculare Yes Euphorbia characias Yes Solanum jasminoides No Euphorbia milii Yes Solanum lycopersicum No Euryomyrtus ramosissima No Solanum mauritianum No Euryops spp. No Solanum melongena No Fagopyrum esculentum Yes Solanum pseudocapsicum No

138

Genus or species Nectar? Genus or species Nectar? Feijoa sellowiana No Solanum spp. No Felicia amelloides No Sonchus oleracus Yes Foeniculum vulgare Yes Spergularia tasmanica No Fragaria × ananassa Yes Spiraea cantoniensis No Francoa ramosa Yes Spiraea japonica Yes Freesia laxa subsp laxa Yes Spiraea thunbergii Yes Fuchsia 'koralle' Yes Stachys arvensis Yes Fuchsia spp. Yes Stachys byzantina Yes Fumaria spp. Yes Strelitzia spp. Yes Galenia pubescens No Streptosolen jamesonii Yes Gardenia brighamii Yes Syringa wolfii Yes Gardenia spp. Yes Syzygium spp. Yes Gazania rigens cultivar Yes Tagetes spp. Yes Genista monsperssulana No Tanacetum parthenium Yes Genista spp. No Taraxacum spp. Yes Genista x spechiaria No Tecomaria capensis Yes Geranium maderense Yes Tetratheca spp. No Geranium solanderi Yes Thalictrum delavayi No Geranium viscosissimum Yes Thryptomene saxicola Yes Gladiolus spp. Yes Thryptomene spp. Yes Glottiphyllum spp. No Thunbergia alata Yes Goodenia spp. Yes Thymus vulgaris Yes Grevillea rhyolitica x juniperina Yes Tibouchina grandiflora Yes Grevillea spp. cultivar Yes Tilia x europa Yes Grevillea victoriae Yes Toxicodendron succedaneum Yes Grewia occidentalis Yes Tradescantia fluminensis No Guizotia abyssinica Yes Tradescantia virginiana No Halgania preissiana No Tragopogon porrifolius Yes Halimium spp. No Tribulus terrestris Yes Hebe brachysiphon Yes Trifolium repens Yes Hebe spp. Yes Tristaniopsis spp. Yes Hedera helix Yes Tritonia spp. Yes Hedychium gardnerianum Yes Tropaeolum spp. Yes Helianthus spp. Yes Tulbaghia violacea Yes Heliochrysum petiolare Yes Ugni molinae No Heliotropium arborescens Yes Ulex europaeus Yes Heliotropium europaeum Yes Verbascum spp. No Helleborus spp. Yes Verbena litoralis Yes Hemerocallis spp. Yes Verbena rigida Yes Hibbertia scandens Yes Verbena spp. Yes Hibbertia spp. Yes Verbena spp. cultivar herb Yes Hibiscus diversifolius Yes Veronica persica Yes Hibiscus spp. Yes Veronica spicata cultivar Yes Homalocladium platycladum No Viburnum odoratissimum Yes

139

Genus or species Nectar? Genus or species Nectar? Hydrangea spp. Yes Viburnum plicatum Yes Hydrangea tardiva Yes Viburnum spp. Yes Hymenosporum flavum Yes Viminaria juncea No Hypericum calycinum No Vinca spp. cultivar Yes Hypericum gramineum No Viola bankssi Yes Hypericum tetrapterum No Viola spp. Yes Iberis sempervirens Yes Viola tricolor var. hortensis Yes Impatiens soldeii Yes virgilia oroboides Yes Impatiens spp. Yes Vriesea spp. hybrids Yes Indigofera decora Yes Wahlenbergia stricta Yes Ipomoea alba Yes Wahlenbergia stricta - double Yes bloom Ipomoea purpurea Yes Wedelia acapulcensis Yes Iris spp. Yes Westringia longifolia Yes Isotoma fluviatilis Yes Westringia spp. Yes Ixia polystachya No Wisteria spp. Yes Ixia spp. Yes Xerochrysum spp. Yes Jacaranda mimosifolia Yes Yucca spp. Yes Jasminum mesnyi Yes Zantedeschia aethiopica No Jasminum spp. Yes Zinnia elgans cultivar Yes

140

Appendix 2 Table 3 – Plots and histograms of the data for three response variables used for modelling for both sampling rounds. Total abundance was logarithmically transformed before use in models. Abundance-GBSW is the abundance of butterflies that are not Grass Blues and Small Whites. Response Variable Sampling Round 1 Sampling Round 2 Richness

Total Abundance

Abundance - GBSW

141

Appendix 2 Table 4 - Model sets used for each study question. Sampling conditions (SC) are calendar date, temperature, and time of day (minutes after civil dawn). Base model refers to the top models (∆AICc <2) for each response variable from Question 1.

Question Model Set Y = SC Y = SC + %cell impervious 1. Effect of Y = SC + %cell impervious + (%cell impervious)2 imperviousness Y = SC + %buffer impervious and at which Y = SC + %buffer impervious + (%buffer impervious)2 scale? Y = %cell impervious Y = %cell impervious + %cell impervious2 Y = %buffer impervious Y = %buffer impervious + (%buffer impervious)2 Y = SC Y = SC + total floral abundance 2. Effect of Y = SC + total floral abundance + (total floral abundance)2 floral resource Y = SC + native floral abundance provision? Y = SC + native floral abundance + (native floral abundance)2 Y = SC + non-native floral abundance Y = SC + non-native floral abundance + (non-native floral abundance)2 Y = total floral abundance Y = total floral abundance + (total floral abundance)2 Y = native floral abundance Y = native floral abundance + (native floral abundance)2 Y = non-native floral abundance Y = non-native floral abundance + (non-native floral abundance)2 Y = base model(s) from Question 1 Y = base model + total floral abundance Y = base model + total floral abundance + (total floral abundance)2 Y = base model + native floral abundance Y = base model + native floral abundance + (native floral abundance)2 Y = base model + non-native floral abundance Y = base model + non-native floral abundance + (non-native floral abundance)2 Y = SC Y = SC + cell remnant presence/absence 3. Effect of Y = SC + buffer remnant presence/absence available Y =cell remnant presence/absence remnant Y = buffer remnant presence/absence vegetation? Y = base model(s) from Question 1 Y = base model + cell remnant presence/absence Y = base model + buffer remnant presence/absence

142

Appendix 2 Table 5 - Validation plots for the first or zero ranked top model for each question and response variable in Chapter 2. The table number is also provided to reference the predictor variables associated with each of the models represented by each set of plots. Abundance-GBSW is the abundance of butterflies that are not Grass Blues and Small Whites.

Round Question Associated Response Validation Plots Table Variable 1 1 Table 2-1 Richness

Abundance

143

Abundance -GBSW

2 Table 2-2 Richness

Abundance

144

Abundance -GBSW

3 Appendix Richness 2 Table 7

Abundance

145

Abundance -GBSW

2 1 Table 2-1 Richness

Abundance

146

Abundance -GBSW

2 Table 2-2 Richness

Abundance

147

Abundance -GBSW

3 Appendix Richness 2 Table 7

Abundance

148

Abundance -GBSW

149

Appendix 2 Table 6 – List of likely butterfly species in Melbourne’s regional species pool. The list was initially drawn from the Atlas of Living Australia (ALA, website, accessed May 2017) covering an extent larger than the study area, generally incorporating the Gippsland Plain Bioregion, but excluding the Dandenong Ranges. The full list was further cross referenced for accuracy (both distribution and habitat) with Field (2013) to develop the list below. Species marked with an asterisk (*) were identified during the round 1 or round 2 surveys (November 2014 - March 2015). While at the study sites we also found two additional species (Imperial Jezebel Delias harpalyce, and an individual that was either a Blotched or Varied Dusky-blue Candalides acasta or C. hyacinthinus). Below this list, we included the map downloaded from the ALA defining the area used to compose the regional species pool.

Species Common Name 1 Acrodipsas brisbanensis Bronze Ant-blue 2 Acrodipsas myrmecophila Small Ant-blue 3 Antipodia chaostola Heath Sand Skipper 4 Argynnina cyrila Forest Brown 5 Belenois java Casper White 6 Candalides acasta Blotched Dusky Blue 7 Candalides hyacinthinus Varied Dusky Blue 8 Charaxes sempronius Tailored Emperor 9 Danaus petilia Lesser Wanderer 10 Dispar compacta Barred Skipper 11 Euploea corinna Common Crow 12 Eurema smilax Small Grass Yellow 13 Geitoneura acantha* Ringed Xenica 14 Geitoneura klugii* Marbled Xenica 15 Hesperilla chrysotricha Golden-haired Sedge-skipper 16 Hesperilla donnysa Varied Sedge-skipper 17 Hesperilla idothea Flame Sedge-skipper 18 Hesperilla ornata Spotted Sedge-skipper 19 Heteronympha banksii Bank's Brown 20 Heteronympha cordace Bright-eyed Brown 21 Heteronympha merope* Common Brown 22 Heteronympha paradelpha Spotted Brown 23 Heteronympha penelope Shouldered Brown 24 Hypochrysops delicia Moonlight Jewel 25 Hypochrysops ignitus Fiery Jewel 26 Jalmenus evagoras* Imperial Hairstreak 27 Jalmenus icilius Amethyst Hairstreak

150

Species Common Name 28 Jalmenus ictinus Stencilled Haristreak 29 Junonia villida* Meadow Argus 30 Lampides boeticus Long-tailed Pea-blue 31 Lucia limbaria Chequered Copper 32 Mesodina halyzia Eastern Iris-skipper 33 Nacaduba biocellata Two-line Spotted-blue 34 Neolucia agricola Fringed Heath-blue 35 Ocybadistes walkeri* Green Grass-Dart 36 Oreixenica kershawi Striped Xenica 37 Oreixenica lathoniella Silver Xenica 38 Papilio aegeus Orchard Swallowtail 39 Papilio demoleus Chequered Swallowtail 40 Paralucia aurifer Bright Copper 41 Paralucia pyrodiscus Firey Copper 42 Pasma tasmanica Two-spotted Grass-skipper 43 Pseudalmenus chlorinda Silky Hairstreak 44 Signeta flammeata Bright Shield-skipper 45 Suniana lascivia Dark Grass-dart 46 Theclinesthes serpentatus* Saltbush Blue 47 Tisiphone abeona Varied Sword-grass Brown 48 Toxidia andersoni Southern Grass-skipper 49 Toxidia doubledayi Lilac Grass-skipper 50 Toxidia parvula Handed Grass-skipper 51 Trapezites eliena Orange Ochre 52 Trapezites luteus Yellow Ochre 53 Trapezites phigalia Heath Ochre 54 Trapezites phigalioides Montane Ochre 55 Trapezites symmomus* Splendid Ochre 56 Vanessa itea* Yellow Admiral 57 Vanessa kershawi* Australian Painted Lady 58 Zizina otis* Common Grass-Blue Species included, they are established, but not native to Victoria, Australia 59 Cephrenes augiades* Orange Palm Dart 60 Papilio anactus* Dainty swallowtail 61 Pieris rapae* Cabbage White Species included, but typically occur in tree tops and would have been poorly sampled by our sampling design 62 Candalides absimilis Common Pencilled Blue 63 Delias aganippe Spotted Jezebel

151

Species Common Name 64 Delias harpalyce Imperial Jezebel 65 Ogyris abrota Dark Purple Azure 66 Ogyris olane Broad Winged Azure

Map downloaded from the ALA defining the area used to compose the regional butterfly species pool.

Legend: Lepidoptera records from the Atlas of Living Australia Area used to generate the potential regional butterfly species pool

152

Appendix 2 Table 7 - Table of butterfly species identified within the study area across Melbourne, Australia’s eastern suburbs between November 2014 and March 2015 and their known larval host plants identified from observation in Victoria (Field 2013) or elsewhere in Australia (underlined species; Braby 2012).

Species Known Larval Host Plants Common Grass low growing Fabaceae, in particular Daviesia brevifolia, Cullen Blue spp., medic, clovers, garden peas and beans Zizina otis labradus Cabbage White Brassicaceae, in particular crucifer veggies, Sisybrium officinale, Pieris rapae Hirschfeldia incana, Lepidium africanum, Cleome, Reseda, and Nasturtium Common Brown Microlaena stipoides, Poa tenera, P. poiformis, Themeda triandra, Heteronympha Cynodon dactylon, and non-native Ehrharta erecta, merope Brachypodium distachyon, and Bromus catharticus Green Grass Dart Imperta cylindrica, Cynodon dactylon, Thuarea involute, and non- Ocybadistes natives Bromus spp. and Ehrharta spp., Brachypodium spp., walkeri Lolium spp., Panicum spp., Paspalum spp., and Pennisetum spp. Australian Bracteantha bracteata, Bracteantha spp., Rhodanthe roseum, Painted Lady Rhodanthe spp., Ammobium alatum, Ammobium spp., Vanessa kershawi Onopordum acantheum, Euchiton involucratus, and Artemisia spp, Chrysocephalum spp., Gnaphalium spp., and Arctotheca calendula, Arctotheca spp. Dainty Swallowtail Citrus spp. (native and non-native), Geijera parviflora, Limonia Papilio anactus acidissima, and Poncirus trifoliata Yellow Admiral Parietaria debilis, Australina pusilla, Pipturis argenteus, Urtica Vanessa itea incisa, and non-native U. urens, Pipturis argenteus, Parietaria judaica, Soleirolia soleirolia Ringed Xenica Microlaena stipoides, Poa sieberiana, Poa tenera, Themeda Geitoneura triandra acantha Imperial Hairstreak Acacia spp., but preferential to A. mearnsii, A. decurrens, A. Jalmenus evagoras dealbata, and A. melanoxylon, A. harpophylla, A. irroranta, and Amyema pendula Meadow Argus Acanthaceae, Goodeniaceae, Asteraceae, Gentianaceae, Junonia villida Plantaginaceae, Portulacaceae, Convolvulaceae, Dipsacaceae, Scrophulariaceae, Hygrophila spp., Ruellia spp., Epaltes spp., Evolvulus spp., Centaurium spp., Goodenia spp., Scaevola spp., Stemoidia spp., Hyptis spp., Plantago spp., Portulaca spp., Veronica spp., Verbena spp., and non-native introduced plantains, particularly Plantago lanceolata, Scabiosa spp., Antirrhinum spp., Russelia spp., Stachytarpheta spp., Lantana spp., Phyla spp., Verbena spp. Splendid Ochre Lomandra longifolia, L. hystrix, L. filiformis, L. obliqua, L spicata Trapezites symmomus 153

Orange Palm Dart ornamental and exotic palms Cephrenes augiades Marbled Xenica Austrodanthonia spp, Austrostipa flavescens, Themeda triandra, Geitoneura klugii Poa labillardieri, P. morrisii, P. tenere, Joycea pallida, and non- natives Bachypodium distachyon, Ehrharta longiflora, E. calycina, and Vulpia spp.

Saltbush Blue Atriplex spp, Einadia spp, Rhagodia spp, Chenopodium spp Theclinesthes (native and non-native), Atalaya hemiglauca and Halosarcia serpentatus halocnemoides

154

Appendix 2 Table 8 – Model averaging table for Question 3 assessing the influence of available remnant vegetation on the species richness and abundance of butterflies. Abundance-GBSW is the abundance of butterflies that are not Grass Blues and Small Whites. Local imperviousness is the percent imperviousness in the sampling cell, Landscape Imperviousness is the percent imperviousness within a 750 m buffer from the center of the sampling cell. Sampling conditions (temperature, date, and time of day) are reported as a combined value. Cell remnant is the presence or absence of remnant vegetation within the sampling cell, while buffer remnant is the presence or absence of remnant vegetation within a 750 m buffer from the center of the sampling cell. Sites were sampled twice; spring to mid-summer and mid-summer to fall to account for seasonality of the butterfly species. The parameter estimate and standard error are reported for each predictor term within each model in the top model set. The weighted model average model (Wt. Model Ave.) is highlighted with grey shading, which is also used to highlight response variables for which no models had ∆AICc < 2 relative to the model with the smallest AICc value. The partial r2 for Richness in round 2, and Abundance-GBSW in both sampling rounds is based on the calculation of pseudo r2 (1-(residual / null deviance)).

Response Local Imperviousness Landscape Imperviousness Cell Remnant Buffer Remnant Sampling Variable Model ∆AICc 2 Partial Partial Partial Partial Conditions r Linear term Quadratic term Linear term Quadratic term Linear term Linear term 2 r2 r2 r2 r2 Partial r Richness 0 0.219 -0.471 ± 0.088 -0.156 ± 0.092 0.219 0.775 0.204 -0.455 ± 0.089 0.204 0.798 0.204 -0.455 ± 0.089 0.204 Wt. Model Ave. 0.211 -0.332 ± 0.041 -0.067 ± 0.004 0.152 -0.130 ± 0.013 0.058 Abundance 0 0.239 -0.443 ± 0.086 -0.207 ± 0.074 0.239 0.329 0.245 -0.311 ± 0.130 -0.236 ± 0.077 0.126 0.371 ± 0.274 0.006

Round1 Wt. Model Ave. 0.242 -0.383 ± 0.0.54 -0.220 ± 0.028 0.187 0.170 ± 0.019 0.003 Abundance 0 0.339 -0.886 ± 0.154 0.254 0.064 – GBSW 1.531 0.275 -0.906 ± 0.157 1.666 0.343 -0.960 ± 0.177 0.216 -0.281 ± 0.351 0.004 0.004 Wt. Model Ave. 0.324 -0.908 ± 0.162 0.250 -0.064 ± 0.005 0.001 0.048

155

Richness 0 0.182 0.463 ± 0.132 0.182 0.786 0.170 -0.223 ± 0.067 0.170 0.848 0.201 -0.114 ± 0.101 0.019 0.290 ± 0.202 0.031 1.303 0.195 -0.094 ± 0.104 0.012 0.317 ± 0.207 0.036 1.537 0.159 -0.215 ± 0.066 0.159 Wt. Model Ave. 0.182 -0.045 ± 0.006 0.024 -0.068 ± 0.004 0.039 0.247 ± 0.028 0.067 Abundance 0 0.255 -0.438 ± 0.088 -0.181 ± 0.077 0.212 0.067 0.634 0.250 -0.415 ± 0.088 -0.285 ± 0.094 0.208 0.060 1.306 0.255 -0.284 ± 0.136 -0.311 ± 0.095 0.113 0.361 ± 0.287 0.005 0.052 1.480 0.254 -0.364 ± 0.098 -0.293 ± 0.094 0.159 0.242 ± 0.203 0.003 0.066 1.771 0.251 -0.363 ± 0.134 -0.196 ± 0.080 0.110 0.210 ± 0.282 <0.00 0.061

Round2 Wt. Model Ave. 0.253 -0.199 ± 0.025 -0.162 ± 0.018 0.091 -0.187 ± 0.022 -0.083 ± 0.006 0.082 0.088 ± 0.007 0.000 0.037 ± 0.001 0.001 0.062 Abundance 0 0.350 -0.995 ± 0.158 0.350 – GBSW 0.046 0.367 -0.726 ± 0.232 0.083 0.689 ± 0.432 0.020 0.126 0.367 -0.739 ± 0.231 0.016 0.646 ± 0.432 0.082 0.426 0.347 -0.997 ± 0.160 0.816 0.361 -0.903 ± 0.168 0.193 ± 0.159 0.361 0.962 0.377 -0.649 ± 0.232 0.190 ± 0.155 0.093 0.646 ± 0.437 0.017 1.843 0.353 -0.951 ± 0.176 0.250 0.200 ± 0.364 0.002 1.946 0.388 -0.996 ± 0.154 0.350 0.037 Wt. Model Ave. 0.362 -0.531 ± 0.134 0.042 ± 0.003 0.166 -0.327 ± 0.0.32 0.062 0.290 ± 0.014 0.008 0.014 ± 0.000 0.000 0.002

156

Appendices Chapter 3

Appendix 3 Figure 1: Diagram depicting subsampling design for larval host plants within sampling cells. Transect centerline

1 km long transect

20 segments 50 m long x 5 m wide

Strips: 5 – 2 m wide every 5 m 2 Segment = 250 m 2 Strips = 50 m , effectively 20% of segment Strips were be at 0 m, 10 m, 20 m, 30 m, and 40 m

Not to scale

157

Appendix 3 Figure 2 – Top model validation plots for question 1 about the response of larval host plant richness (A) and cover (%; B) along a gradient of increasing impervious surface cover.

158

Appendix 3 Figure 3 –Validation plots for question 2 about the response of butterfly species richness (A), total butterfly abundance (B), and the abundance of less common butterflies (total abundance minus Grass Blues and Small Whites; C) to the richness and cover of larval host plants within the study area. In the case of multiple top models, only the first top model’s validation plots are presented. Please see Table 3-4 for more information. A B C

159

A B

C D

160

Appendix 3 Table 1: Model set to investigate the relationship between the butterfly community and larval host plant (LHP) richness and % cover. Imperviousness is the % impervious surfaces within a sampling cell.

Model # Model 1 LHP richness 2 LHP richness + LHP cover 3 LHP cover 4 LHP richness + Imperviousness 5 LHP richness + LHP cover + Imperviousness 6 LHP cover + Imperviousness 7 LHP richness + Imperviousness + Imperviousness^2 8 LHP richness + LHP cover + Imperviousness + Imperviousness^2 9 LHP cover + Imperviousness + Imperviousness^2 10 Imperviousness 11 Imperviousness + Imperviousness^2

161

Response Variable Larval Host Plant Richness

Larval Host Plant Cover

Butterfly Richness

Total Butterfly Abundance

Butterfly Abundance - GBSW

162

Appendix 3 Table 3: Model set to investigate the relationship between individual butterfly species abundance and their larval host plant (LHP) % cover. A variable combining each of a butterfly species’ LHP, Combined, was included in the modelling to understand if the cumulative effect of the LHP was more important than the individual LHP species. Imperviousness is the % impervious surfaces within a sampling cell. For each butterfly species, models 4-6 were repeated for each individual LHP. Model numbers in Table 5 will not correspond with the numbers in this example.

Model # Model 1 Combined 2 Combined + imperviousness 3 Combined + imperviousness + imperviousness^2 4 LHP 5 LHP + imperviousness 6 LHP + imperviousness + imperviousness^2 7 imperviousness 8 imperviousness + imperviousness^2

163

Appendices Chapter 4

Residuals vs Fitted

164

D Residuals vs Fitted

E Residuals vs Fitted

F Residuals vs Fitted

G Residuals vs Fitted

165

A B

Residuals vs Fitted Residuals vs Fitted

C D

Residuals vs Fitted Residuals vs Fitted

166

Appendix 4 Figure 3 - Nonmetric multidimensional scaling (NMDS) ordination plot of the butterfly communities from round 1 (A) and round 2 (B) within each garden type: wildlife gardens (WG, black triangles and gray polygon), control gardens in areas targeted (CT, cyan circles and polygon) and not targeted (CUT, yellow squares and polygon) for wildlife gardening. Taxa are labelled in red. Unidentified species are grouped by family blues (Lycaenidae), browns (Nymphalidae), and darts (Hesperiidae).

Stress = 0.088

Stress = 0.051

167

Genus or species Nectar? Genus or species Nectar? Abelia spp. yes Jacaranda mimosifolia yes Acanthus mollis yes Jasminum spp. yes Achillea millefolium yes Kalanchoe blossfeldiana yes Agapanthus praecox yes Lagerstroemia indica yes Alcea spp. yes Lantana camara yes Allium spp. yes Lantana montevidensis yes Alyogyne huegleii yes Lathyrus odoratus yes Amaryllis belladonna yes Lavendula spp. yes Anigozanthos spp. yes Lechenaultia biloba yes Antirrhinum spp. cultivar yes Leptospermum petersonii yes Aptenia cordifolia yes Leptospermum spp. yes Arctotheca calendula* no Leucanthemum x superbum yes Argyranthemum frutescens yes Leucochrysum albicans yes Arthropodium cirratum yes Ligustrum spp. yes Asparagus aethiopicus yes Limonium perezii yes Austromyrtus dulcis no Liriope muscari no Azalea spp. yes Lobularia maritima yes Banksia spp. yes Lycianthes rantonnetii no Begonia spp. no Magnolia spp. yes Bella perennis yes Matthiola incana yes Bidens coronata yes Medicago polymorpha* yes Brachyscome spp. yes Melaleuca fulgens yes Brassica spp. yes Melaleuca linariifolia yes Brunsfelsia latifolia yes Melaleuca thymifolia yes Bulbine bulbosa no Mentha spp. yes Calendula spp. yes Modiola caroliniana no Callistemon spp. yes Myoporum floribunda yes Calocephalus citreus yes Myoporum parvifolium yes Calodendrum capense no Myoporum spp. yes Genus or species Nectar? Genus or species Nectar? Campanula poscharskyana yes Nerium oleander yes 168

Carbrotus spp. yes Nigella damascena yes Carpobrotus aequilaterus yes Oenothera lindheimeri yes Cassinia spp. no Olearia ramulosa yes Ornithogalum Celosia spp. yes longibracteatum no Centaurium erythraea no Osteospermum fruticosum no Centranthus ruber yes Oxalis corniculata complex yes Cerastium tomentosum yes Oxalis latifolia yes Chamelaucium axillare yes Oxalis stricta yes Chamelaucium spp. yes Ozothamnus diosmifolius no Chamelaucium x verticordia yes Papaver spp. yes Chlorophytum comosum no Pelargonium spp. yes Choysia ternata yes Pelargonium australe yes Chrysocephalum apiculatum* yes Petroselinum crispum yes Chrysocephalum semipapposum* yes Petunia x hybrida yes Coleonema spp. yes Philadelphus cornaria yes Convolvulus cnerorum* yes Phlomis fruticosa yes Convolvulus mauritanicus* yes Phlox spp. yes Cordyline australis yes Pimelea spp. yes Correa spp. yes Plectranthus argentatus yes Cotoneaster spp. yes plectranthus graveolens yes Crassula multicava yes Plectranthus neochilus yes Crassula tetragona yes Plumbago spp. yes Crocosmia spp. yes Polygala myrtifolia yes Crowea exalta no Pomaderris aspera yes Cymbalania muralis yes Portulaca grandiflora* yes Dampiera linearis yes Prostanthera incana yes Darwinia citriodora yes Prostanthera spp yes Delosperma spp. yes Ptilotus exaltatus no Dianella spp. no Punica granatum yes Dianthus spp. yes Ranunculus repens yes Dicliptera sericea yes Rhodanthe anthemoides* no Dietes spp. no Ricinocarpus pinifolius yes Echeveria spp. yes Rosa spp. yes Eremaphila spp. yes Rosmarinus officinalis yes Eremophila calorhabdos yes Salvia farinacea cultivar yes Eremophila maculata yes Salvia leucantha yes Erica carnea cultivar yes Salvia microphylla yes Erigeron karvinskianus yes Salvia nemorosa yes Erigeron spp. yes Salvia officinalis yes Erysimum cheiri yes Salvia uliginosa yes Escallonia rubra yes Santolina chamaecyparissus no Escholtzia californica yes Scabiosa spp. yes Eucalyptus spp. yes Scaevola spp. yes Euphorbia characias yes Sedum spp. yes

169

Genus or species Nectar? Genus or species Nectar? Eurymurtus ramosissima no Senecio cineraria yes Euryops spp. no Senna artemisioides yes Feijoa sellowiana no Solanum aviculare yes Felicia amelloides no Solanum lycopersicum no Fragaria × ananassa yes Stachys byzantina yes Fuschia spp. yes Streptosolen jamesii yes Gazania rigens cultivar yes Tanacetum parthenium yes Geranium solanderi yes Taraxacum spp. yes Gladiolus spp. yes Tecomaria capensis yes Goodenia ovata* yes Thryptomene saxicola yes Goodenia spp.* yes Thryptomene spp. yes Grevelia rosmarinifolia yes Thunbergia alata yes Grevelia spp yes Thymus vulgaris yes Halgania cyanea no Toxicodendron succedaneum yes Hebe spp. yes Trifolium repens* yes Helleborus spp. yes Tritonia spp. yes Hemerocallis spp. yes Tropaeolum spp.* yes Hibbertia scandens yes Verbena spp. cultivated yes Hibiscus spp. yes Viburnum spp. yes Hibiscus syriacus yes Vinca spp. yes Homoranthus papillatus no Viola spp. yes Hydranga spp. yes Viola tricolor var. hortensis yes Impatiens balfourii yes Wahlenbergia stricta yes Impatiens spp. yes Westringia fruiticosa yes Iris spp. yes Westringia spp. yes Isotoma axillaris yes Xerochrysum bracteatum* yes Isotoma fluviatilis yes Xerochrysum spp.* yes Ixia spp. yes Zantedeschia aethiopica no

170

Appendix 4 Table 2 – Table showing the proportion of gardens (out of 108) where butterfly species were present. Garden types: wildlife gardens (WG) and control gardens in areas targeted (CT) and untargeted for wildlife gardening (CUT).

Round 1 Round 2 CUT CT WG CUT CT WG Identified species Australian Painted Lady 1.85 0.00 1.85 0.00 0.00 0.00 Vanessa kershawi Cabbage White 34.26 19.44 12.96 21.30 12.04 13.89 Pieris rapae Common Brown 1.85 0.93 1.85 3.70 0.93 0.00 Heteronympha merope Dainty Swallowtail 2.78 0.00 0.00 2.78 0.00 0.93 Papilio anactus Grass Blue 12.96 5.56 2.78 31.48 9.26 8.33 Zizina otis labradus Green Grass Dart 0.00 0.93 0.93 2.78 0.00 1.85 Ocybadistes walkeri Meadow Argus 0.00 0.00 0.00 0.93 0.00 0.00 Junonia villida Yellow Admiral 0.93 0.00 0.93 0.00 0.00 0.00 Vanessa itea Unidentified individuals by family Blues (Lycaenidae) 32.41 16.67 6.48 1.85 0.00 0.00 Browns (Nymphalidae) 0.93 1.85 0.00 1.85 1.85 0.00 Darts/Skippers (Hesperiidae) 6.48 0.00 0.93 3.70 2.78 1.85

171

Appendix 4 Table 3 – Parameter estimates of the mixed modelling results for Question 1 assessing the difference in butterfly richness and abundance between garden types (WG – wildlife and control gardens in areas targeted (CT) and untargeted for wildlife gardening (CUT)) in reference to tree cover (%). Wildlife gardens were used as the intercept. The significance of an interaction between garden type and tree cover (%) within a 500 m radius buffer (Tree) was first assessed, if not significant the model was rerun sans the interaction. The last set of results were to assess if the inclusion of tree cover and imperviousness (Imperv, cover (%) within a 500 m radius buffer) within the same model improved model fit. Richness and abundance – GB, B, and SW is the total richness or abundance minus abundance of Common Grass Blues, ‘blues’, and Small White butterflies. Sampling conditions (Temp -temperature, date, and time of day) were included in both models. Intercept of the model of the relationship with wildlife gardens (the reference category) and the other listed predictor variables is reported. Rounds 1 and 2 indicate the 2 sampling rounds for data collection.

Richness Richness– GB, B, and SW Total Abundance Abundance – GB, B, and SW Est ± Std Err Pvalue Est ± Std Err Pvalue Est ± Std Err Pvalue Est ± Std Err Pvalue Round 1 Intercept 0.15 ± 0.22 -1.62 ± 0.56 1.21 ± 0.22 -0.91 ± 0.65 CT 0.30 ± 0.30 0.32 -0.33 ± 0.86 0.70 0.48 ± 0.30 0.11 -0.10 ± 0.91 0.91 CUT 0.42 ± 0.25 0.09 0.33 ± 0.62 0.59 0.48 ± 0.27 0.09 -0.10 ± 0.75 0.90 Temp 0.20 ± 0.08 < 0.01 0.34 ± 0.20 0.10 0.35 ± 0.10 < 0.01 0.22 ± 0.26 0.40 Time -0.07 ± 0.09 0.44 -0.16 ± 0.22 0.46 -0.05 ± 0.10 0.64 -0.15 ± 0.26 0.57 Date 0.05 ± 0.08 0.57 0.10 ± 0.20 0.62 0.12 ± 0.11 0.28 -0.05 ± 0.26 0.85 Tree 0.05 ± 0.25 0.85 0.45 ± 0.57 0.43 0.23 ± 0.24 0.35 0.75± 0.78 0.34 Tree*CT 0.06 ± 0.34 0.85 -0.45 ± 0.98 0.65 -0.30 ± 0.36 0.41 -1.21 ± 1.17 0.30 Tree*CUT 0.04 ± 0.27 0.87 -0.36± 0.62 0.56 -0.16 ± 0.28 0.57 -0.70 ± 0.84 0.41 Round 2 Intercept -0.11 ± 0.28 -1.63 ± 0.57 1.07 ± 0.24 -1.28 ± 0.54 CT -0.02 ± 0.39 0.97 0.35 ± 0.73 0.63 0.07 ± 0.29 0.80 0.10 ± 0.62 0.87

172

CUT 0.36 ± 0.32 0.27 0.54 ± 0.63 0.39 0.22 ± 0.30 0.47 0.50 ± 0.61 0.41 Temp 0.19 ± 0.11 0.09 0.03 ± 0.18 0.86 0.42 ± 0.12 < 0.01 0.08 ± 0.23 0.71 Time 0.08 ± 0.11 0.46 0.14 ± 0.19 0.46 0.08 ± 0.11 0.46 0.23 ± 0.20 0.27 Date < 0.001 ± 0.11 0.10 0.17 ± 0.19 0.36 -0.06 ± 0.12 0.64 0.06 ± 0.25 0.81 Tree 0.33 ± 0.27 0.22 0.26 ± 0.57 0.65 0.28 ± 0.23 0.22 0.47 ± 0.45 0.30 Tree*CT 0.05 ± 0.44 0.91 -0.15 ± 0.83 0.86 -0.31 ± 0.36 0.40 -1.04 ± 0.80 0.20 Tree*CUT -0.31 ± 0.30 0.30 -0.28 ± 0.62 0.66 -0.14 ± 0.26 0.60 -0.32 ± 0.51 0.52 Round 1 Intercept 0.12 ± 0.19 -1.42 ± 0.40 1.29 ± 0.19 -0.49 ± 0.47 CT 0.33 ± 0.24 0.16 -0.62 ± 0.64 0.33 0.34 ± 0.26 0.17 -0.71 ± 0.70 0.31 CUT 0.44 ± 0.22 0.05 0.15 ± 0.50 0.76 0.41 ± 0.26 0.11 -0.51 ± 0.64 0.42 Temp 0.20 ± 0.08 < 0.01 0.35 ± 0.20 0.08 0.35 ± 0.10 < 0.01 0.27 ± 0.26 0.29 Time -0.07 ± 0.08 0.43 -0.14 ± 0.21 0.51 -0.05 ± 0.10 0.62 -0.16 ± 0.26 0.54 Date 0.05 ± 0.08 0.54 0.08 ± 0.20 0.69 0.11 ± 0.10 0.31 -0.15 ± 0.24 0.53 Tree 0.09 ± 0.09 0.32 0.14 ± 0.22 0.53 0.08 ± 0.10 0.46 0.07 ± 0.28 0.79 Round 2 Intercept 0.03 ± 0.22 -1.49 ± 0.43 1.13 ± 0.21 -0.91 ± 0.48 CT -0.02 ± 0.27 0.96 0.24 ± 0.55 0.66 -0.07 ± 0.24 0.77 -0.34 ± 0.59 0.54 WG 0.24 ± 0.28 0.39 0.42 ± 0.52 0.42 0.15 ± 0.28 0.59 0.24 ± 0.56 0.67 Temp 0.21 ± 0.11 0.07 0.04 ± 0.18 0.83 0.41 ± 0.12 < 0.01 0.10 ± 0.23 0.68 Time 0.08 ± 0.11 0.45 0.14 ± 0.19 0.47 0.07 ± 0.11 0.51 0.30 ± 0.22 0.17 Date 0.02 ± 0.12 0.89 0.18 ± 0.18 0.33 -0.06 ± 0.12 0.62 0.08 ± 0.23 0.73 Tree 0.11 ± 0.11 0.30 0.04 ± 0.25 0.85 0.15 ± 0.10 0.16 0.19 ± 0.23 0.41 Round 1 Intercept 0.15 ± 0.19 -1.29 ± 0.41 1.27 ± 0.20 -0.43 ± 0.50 CT 0.34 ± 0.24 0.15 -0.61 ± 0.64 0.34 0.34 ± 0.25 0.18 -0.70 ± 0.70 0.32 WG 0.37 ± 0.24 0.12 -0.31 ± 0.55 0.57 0.45 ± 0.27 0.11 -0.64 ± 0.74 0.38 Temp 0.20 ± 0.08 < 0.01 0.35 ± 0.20 0.08 0.35 ± 0.10 < 0.01 0.27 ± 0.26 0.29 Time -0.08± 0.09 0.37 -0.21 ± 0.23 0.35 -0.04 ± 0.10 0.66 -0.19 ± 0.27 0.49

173

Date 0.05 ± 0.08 0.51 0.12 ± 0.20 0.54 0.10 ± 0.10 0.33 -0.14 ± 0.24 0.55 Imperv 0.07 ± 0.09 0.43 0.51 ± 0.25 0.04 -0.04 ± 0.10 0.70 0.10 ± 0.27 0.73 Tree 0.09 ± 0.09 0.33 0.18 ± 0.22 0.40 0.08 ± 0.11 0.45 0.07 ± 0.28 0.81 Round 2 Intercept 0.07 ± 0.22 -1.42 ± 0.43 1.10 ± 0.21 -0.91 ± 0.39 CT -0.01 ± 0.27 0.96 0.23 ± 0.55 0.68 -0.07 ± 0.24 0.76 -0.54 ± 0.46 0.24 WG 0.17 ± 0.29 0.57 0.24 ± 0.56 0.67 0.22 ± 0.30 0.47 0.10 ± 0.53 0.86 Temp 0.21 ± 0.11 0.07 0.03 ± 0.19 0.86 0.42 ± 0.12 < 0.01 0.06 ± 0.21 0.77 Time 0.10 ± 0.11 0.36 0.19 ± 0.20 0.33 0.06 ± 0.11 0.62 0.24 ± 0.20 0.23 Date 0.02 ± 0.12 0.86 0.19 ± 0.18 0.31 -0.06 ± 0.12 0.61 0.06 ± 0.23 0.79 Imperv 0.08 ± 0.11 0.48 0.21 ± 0.21 0.31 0.15 ± 0.11 0.15 0.15 ± 0.20 0.44 Tree 0.11 ± 0.11 0.30 0.04 ± 0.20 0.83 -0.07 ± 0.10 0.51 0.13 ± 0.18 0.47

174

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: Kurylo, Jessica

Title: The effect of urbanization on butterfly assemblages in Melbourne, Australia

Date: 2017

Persistent Link: http://hdl.handle.net/11343/213488

File Description: Final thesis

Terms and Conditions: Terms and Conditions: Copyright in works deposited in Minerva Access is retained by the copyright owner. The work may not be altered without permission from the copyright owner. Readers may only download, print and save electronic copies of whole works for their own personal non-commercial use. Any use that exceeds these limits requires permission from the copyright owner. Attribution is essential when quoting or paraphrasing from these works.