Plant Diversity Effects on Plant-Pollinator Interactions in Urban and Agricultural Settings

Research Collection

Doctoral Thesis

Plant diversity effects on plant-pollinator interactions in urban and agricultural settings

Author(s): Hennig, Ernest Ireneusz

Publication Date: 2011

Permanent Link: https://doi.org/10.3929/ethz-a-006689739

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library Diss. ETH No. 19624

Plant Diversity Effects on Plant-Pollinator Interactions in Urban and Agricultural Settings

A dissertation submitted to the ETHZURICH¨

for the degree of DOCTOROF SCIENCES

presented by ERNEST IRENEUSZ HENNIG Degree in Environmental Science (Comparable to Msc (Master of Science)) University Duisburg-Essen born 09th February 1977 in Swiebodzice´ (Poland)

accepted on the recommendation of Prof. Dr. Jaboury Ghazoul, examiner Prof. Dr. Felix Kienast, co-examiner Dr. Simon Leather, co-examiner Prof. Dr. Alex Widmer, co-examiner

2011

You can never make a horse out of a donkey

my father Andrzej Zbigniew Hennig

Young Man Intrigued by the Flight of a Non-Euclidian Fly (Max Ernst, 1944)

Contents

Abstract

Zusammenfassung

1 Introduction 9 1.1 Competition and facilitation in plant-plant interactions for pollinator services .9 1.2 Pollination in the urban environment ...... 11 1.3 Objectives ...... 12 1.4 References ...... 12

2 Does plant diversity enhance pollinator facilitation? An experimental approach 19 2.1 Introduction ...... 20 2.2 Materials & Methods ...... 21 2.2.1 Study Design ...... 21 2.2.2 Data Collection ...... 22 2.2.3 Analysis ...... 22 2.3 Results ...... 23 2.3.1 Pollinator Species and Visits ...... 23 2.3.2 Plant species and experimental treatments ...... 24 2.3.3 Visitor Diversity of Bees and Syrphids between Plant Species . . . . . 24 2.3.4 Visits of Bees and Syrphids between Plant Species ...... 25 2.3.5 Plant interactions ...... 25 2.4 Discussion ...... 26 2.4.1 Diversity and visits of ﬂower visitors ...... 26 2.4.2 Plant interactions – facilitation, competition or no interactions? . . . . 27 2.5 Conclusion ...... 28 2.6 References ...... 29 2.7 Tables and Figures ...... 34

3 Linking competition and facilitation processes in pollination systems to plant species richness 39 3.1 Introduction ...... 40 3.2 Materials & Methods ...... 41 3.2.1 Field Sites ...... 41 3.2.2 Study Species ...... 41 3.2.3 Field work ...... 42 3.2.4 Statistical Analysis ...... 43 3.3 Results ...... 45 3.3.1 Treatment characteristics ...... 45 3.3.2 Pollinators ...... 45 3.3.3 Seed set in the focal plant species ...... 47 3.4 Discussion ...... 48 3.4.1 Visitor community richness and visitation frequency and the surrounding plant community ...... 48 3.4.2 Seed set in the focal plants ...... 49 3.4.3 Conclusion ...... 51 3.5 Acknowledgement ...... 51 3.6 References ...... 51 3.7 Tables and Figures ...... 57

4 Plant-Pollinator Interactions within the Urban Environment 67 4.1 Introduction ...... 68 4.2 Materials & Methods ...... 70 4.2.1 Study Site ...... 70 4.2.2 Data Collection ...... 70 4.2.3 Analysis ...... 71 4.3 Results ...... 74 4.3.1 Plant Species ...... 74 4.3.2 Flower Visitor of Trifolium pratense ...... 74 4.4 Discussion ...... 77 4.4.1 Local Community Effects ...... 77 4.4.2 Landscape Scale Effects ...... 78 4.4.3 Interaction between Local and Landscape Scales ...... 79 4.4.4 Conclusion ...... 80 4.5 Acknowledgements ...... 80 4.6 References ...... 81 4.7 Tables & Figures ...... 89 4.8 Appendices ...... 97 4.A Number of occurrences and ﬂoral abundance of plant species found in the urban study...... 97 4.B Pollinator species ...... 100 4.C Landscape Structures at different scales ...... 102 4.D Green Area Structures ...... 103

5 Pollinating Animals in the Urban Environment 105 5.1 Introduction ...... 106 5.2 Materials & Methods ...... 108 Study Site ...... 108 Data Collection ...... 108 Analysis ...... 110 5.3 Results ...... 111 Pollinator Species ...... 111 Plant Species ...... 112 Landscape Metrics ...... 112 5.4 Bees and Syrphid Flies ...... 113 Bees...... 113 Syrphids ...... 114 5.5 Discussion ...... 114 Scale effects of landscape metrics ...... 114 Local and large scale factors ...... 117 Conclusion ...... 119 5.6 References ...... 119 5.7 Tables and Figures ...... 126

6 Synthesis 135 6.1 Summary ...... 135 6.2 Discussion ...... 136 6.3 Outlook ...... 138 6.4 References ...... 139 List of Tables

2.1 Plant species combinations used in the experimental design and the number of observations (n). The plant species are: Ao = Anchusa officinalis L., At = Anthemis tinctoria L., Bs = Buphthalum salicifolium L., Cj = Centaurea jacea L.s.l., Cc = Centaurea cyanus L., Cm = Chelidonium majus L., Cp = Chrysanthemum praecox Horvatic,´ Hn = Helianthemum nummularium (L.) Mill. s.l., Ka = Knautia arvensis (L.) Coult. s.str., Mm = Malva moschata L., Sl = Silene latifolia Poir., So = Scabiosa ochroleuca L., Sv = Silene vulgaris (Moench) Garcke s.l., Tc = Tanacetum corymbosum (L.) Schultz-Bip...... 34 2.2 Species richness, plot visits, and floral visits of each order. The contribution of the most common species of the Hymenoptera and Diptera is given...... 35 2.3 t-values from the Analysis of Variance for bee and syrphid diversity, and bee and syrphid log-transformed visits among Anchusa officinalis, Centaurea cyanus, Centaurea jacea, Chrysanthemum praecox. Asteriks indicate significant results (∗ = < 0.05, ∗∗ = < 0.01, ∗∗∗ = < 0.001)...... 35 2.4 Test for the importance of correlation structures using repeated observations at the plot level (”Plot Number”) and on each day (”Date”) for each plant species and pollinator family and the response variables ”diversity” (upper table section) and ”floral visits per flower density” of the focal plant species (lower table section). We used always models with lowest Akaike‘s Information Criterion, which provide better fits...... 36 2.5 Estimates and F-values from the generalized least square models with diversity of bees and syrphids as a function of plant species treatment (PS), floral density of heterospecifics (FAD) and the interaction in each plant species (Anchusa officinalis, Centaurea cyanus, Centaurea jacea, Chrysanthemum praecox). Asteriks indicate significant results (∗ = < 0.05, ∗∗ = < 0.01, ∗∗∗ = < 0.001)...... 36 2.6 Estimates and F-values from the generalized least square models with bee and syrphid visits per flower as a function of plant species treatment (PS), floral density of heterospecifics (FAD) and the interaction in each plant species (An- chusa officinalis, Centaurea cyanus, Centaurea jacea, Chrysanthemum praecox). Asteriks indicate significant results (∗ = < 0.05, ∗∗ = < 0.01, ∗∗∗ = < 0.001). . . 37

3.1 The experimental design at Grandcour. The number in the cells present the field sites...... 57 3.2 Plant species and treatment order in the experimental set up at Grandcour. The number represent the field number...... 57 3.3 The differences among field sites in metre obtained by the command ”pairdist()” using the package spatstat 1.9-6 (Baddeley and Turner, 2005) in R 2.7.2 (R Development Core Team, 2009). The Swiss coordinates were obtained from the web site: www.ecogis.admin.ch...... 58 3.4 Results from the generalized least square model after stepwise exclusion of non-significant terms for visitor species richness and the logarithmized values of visitation frequency at patch level with pooled data (both experiments) and non-pooled data. Predictors were centered...... 59 3.5 Visits and percentages of visits to the two focal plants Borago officinalis and Sinapis alba. The percentage visits of each species is related to the visitation frequency of the next higher taxonomic level. Visits percentages are referred to total visits. ∆BO and ∆SA are the percentages of plant to treatment visits. BO1 and BO2 are flower visitor records of B. officinalis from the first and second experiment, respectively...... 60 3.6 Mean and standard errors of the explanatory used in the models for both plant species and for the first and second experimental period for B. officinalis.... 61 3.7 Results from the generalized least square models for visitor diversity and logarithmized flower visitation frequency in the two studied plant species Borago officinalis and Sinapis alba. All predictors were centered...... 62 3.8 Results of the generalized least square model for seed set in Borago officinalis and Sinapis alba. For Borago officinalis results from the models with pooled data and for each of the two experiments at Grandcour (non-pooled data) are shown. All predictors were centered...... 63 4.1 The table shows the plant species visited more than once by Bombus pascuorum. COUNT = no. of plots the plant species was visited, COUNT.PLANT = no. of plots the plant species occurred, FLORAL VISITS/FLOWER/COUNT = average of floral visits/flower over all plots where the species occurred...... 89 4.2 Results from linear regression for correlation coefficients of the response variables (RESP) visits of all visitor species of Trifolium pratense, all bee species, bees wihtout the main visitor, and the main visitor of T. pratense Bombus pascuorum with the landscape metrics (LM) green area size (GA) and green area edge density (ED) with scale from 20 - 200 m radius...... 90 4.3 Results from linear regression for visits of all species (Model I), bees (Model II), bees excluding the main visitor (Model III) and B. pascuorum (Model IV) to T. pratense...... 91 4.4 Statistics and residual autocorrelation in the regression models for visits of all species, bees, bees without the main visitor B. pascuorum, and the main visitor alone to T. pratense, together with the transformed explanatory variables (EXP), i.e., floral abundance of the plant species L. corniculatus (LC), T. repens (TR), and C. jacea (CJ)...... 92

5.1 Pollinator species, their number of plot visits (sum) and number of plots, where hey have been recorded during the study in the city Zurich¨ in 2008 (count). . . 126 5.2 Plant species and their families found in the urban study...... 129 5.3 Mean and standard deviations of landscape metrics for ﬁeld sites with syrphid species and without syrphid species. Test statistics is derived from Wilcoxon exact test...... 131

List of Figures

2.1 The boxplots are showing differences in mean diversity (upper left) and visitation frequency per floral density (upper right) for bees (a) and syrphid flies (s) for each plant species (Ao = Anchusa officinalis, Cc = Centaurea cyanus, Cj = Centaurea jacea, and Cp = Chrysanthemum praecox)...... 37

3.1 Sinapis alba L. being destroyed by beetles in June 2008 before the start of the first experiment in the experimental fields at Grandcour, canton Fribourg. . . . 64 3.2 The difference in plant diversity expressed as Margalef Index among the treatments. 64 3.3 Margalef plant species richness and log-transformed floral density between experiments (I, II), treatments (2, 6, 12, 20), and treatments within each experiment. 65 3.4 Differences between fields in plant species richness (Margalef Index) and floral density (log-transformed)...... 66

4.1 Map of the study region with the 89 studied locations...... 93 4.2 Boxplot with the mean diversity in the four most common plant species and their species accumulation curves (TR = Trifolium repens, CJ = Centaurea jacea, LC = Lotus corniculatus, TP = Trifolium pratense)...... 94 4.3 Correlation results between visitation variables and landscape metrics. Solid lines show p-values at 0.05 and dashed lines at 0.1. Note the different scaling of the y-axis. There were no significant correlations between bees and the extent of green area as well as edge density, all visitors and edge density, and bees w/o main visitor and extent of green area...... 95 4.4 Number of flower visits by all species, bees, and B. pascuorum for T. pratense in species plots with T. pratense alone and in combinations with Centaurea jacea (Cj), Trifolium repens (Tr), and Lotus corniculatus (Lc) (mean and standard error). Plot frequencies for each combination are shown above bars. Results are presented in the text...... 96 5.1 The map shows the city of Zurich¨ and the different landscape elements forest, green area, paved area, building and water. Red dots represent the 89 study locations...... 131 5.2 Species accumulation curves for pollinator, plant, bee and syrphid diversity with 95% confidence intervals using the method ”random” in the statistical program R.132 5.3 The ten most frequent plant species and their plot frequencies Tr.pr = Trifolium pratense, Tr.re = Trifolium repens, Ga.al = Galium album, Be.pe = Bellis perennis, Pl.la = Plantago lanceolata, Tr.ca = Trifolium campestre, Lo.co = Lotus corniculatus, Hy.ra = Hypochoeris radicata, Ce.ja = Centaurea jacea, Ra.re = Ranunculus repens...... 132 5.4 Correlation between of extent of green area and green area edge density with bee diversity, bee visits, and bee visits excluding honeybees across the 89 sampled locations. Note the differences in scale at the y-axis...... 133 5.5 Correlation between the extent of green area and edge density with syrphid diversity and syrphid visitation frequency across the 89 sampled locations. Note the differences in scale at the y-axis...... 134 Abstract

Most flowering plants require insects as agents of pollen transfer for successful reproduction by seed. The efficiency of the pollination process at an individual plant level may be affected by the richness and density of the surrounding plant community, as well as the structure of the wider landscape, all of which might affect the abundance of pollinators and the pattern of flower visitation among species and individuals. Plants can interact in competitive or facilitative manner in attracting flower visitors, and this might be reflected in seed set. This thesis aims to investigate questions in the field of pollination ecology. I thought to first examine how facilitative interactions differ among plant species in terms of attracting the two flower visiting invertebrate families, Apidae and Syrphidae. A following study then explored the effect of plant species richness and flower density on seed set of two plant species, which differed in terms of their bee and syrphid pollinator communities. I moved thereafter from semi-natural landscape to urban landscapes and investigated how a plant species interacts with sympatrics in the urban environment in terms of attracting flower visitors generally, as well as bees and its main visitor specifically. The final chapter investigates the impact of the urban matrix on the diversity and abundance of flower visitors. Considering plant interactions for visits of pollinating insects (Chapter 2), there were only competitive interactions with increasing plant species richness and heterospecific floral density for one of the four studied plant species Anchusa officinalis (L.), Centaurea cyanus (L.), Centaurea jacea (L.s.l.), and Chrysanthemum praecox (HORVATIC´). Bee visits were negatively influenced by heterospecific floral density only in C. praecox, and syrphid visits declined with increasing heterospecific floral density and plant species richness only in C. praecox. Significant differences in visitor diversity and visitation frequency between the four plant species can explain the results. Indeed, the three plant species that were unaffected by plant species richness and heterospecific floral density were mainly pollinated by honeybees and bumble bees. Plants being visited by social bee species may therefore interact less with co-occurring plant species, while plant species visited by syrphids and bees to a similar extent are more likely subjected to competition with other plant species. As visitor species richness differs among plant species, competitive or facilitative interactions among plant species for pollination, and the outcome of such interactions in terms of seed set, might depend on the degree of generalization of pollinators’ visitation to plant species. Thus plants which flowers are pollinated by a wide range of insects might be more likely involved in interactions (competitive or facilitatory) than those pollinated by few specialist pollinators. I explored this idea by comparing seed set and pollinator visitation in two plant species, Borago officinalis, pollinated by bees (mainly by bumblebees), and Sinapis alba, a generalist species pollinated by a wide variety of bees and syrphids. Community plant species richness positively affected seed set of the more specialized plant species, Borago officinalis, but had no impact on the species richness of its visitor community. Floral density of all plant species had a positive effect on seed set in Borago officinalis, but only when floral visits increased. The generalist plant, Sinapis alba, had reduced seed set when visitor species richness increased in species-poor plant treatments, while floral density positively affected seed set when visitor species richness increased. The results indicated that plant species richness and floral density had different and often interacting impacts on seed set in two plant species that contrast in their degree of pollinator specialization. The limited number of species compared precludes broader generalization, although the results suggest that interactions among plants for pollinator services depend on the degree of generalization in the visitor community. In urban environments plant and animal communities are expected to be altered in comparison to (semi-)natural environments due to fragmentation and alteration of habitats, pollution, disturbance, and introduction of non-native plant and animal species. I considered plant species interactions in the context of two landscape metrics: extent and edge density of green area in an urban ecosystem. Correlations of flower visits per flower of the focal plant, Trifolium pratense, with the two landscape metrics showed that extent of green area affect pollinator visitation more than edge density of green area. The correlation coefficients increased with geographic scale. I also showed that the extent of green area affected plant-plant interactions for floral visits by bees in that with increasing extent of green area co-occurring plant species seemed to have less negative impact on visits by bees to T. pratense. This suggests that competitive effects among plant species for pollinator visits are relaxed at larger extent of green areas. This effect, however, was influenced by the frequency of the main visitor Bombus pascuorum. Excluding B. pascuorum only heterospecific flower abundance interacted with the extent of green area negatively on visits by the remaining bees. Effects of plant communities on visitation patterns of plants thus need to take into account landscape factor and visitor composition to explain their impact. Diversity of and visits by flower visitors might also be shaped by the landscape structure interacting with local plant community diversity. The influence of extent and edge density of green areas on the diversity of bees and syrphids and on the local plot visits by bees, bees without the most common visitor species, the honeybee Apis mellifera, and syrphids were investigated separately, in relation to local plant diversity and abundance. Correlation coefficients between the response variables and the extent of green area increased with scale (except for plot visits by all bees), while edge density varied over the scale from 20 to 200 m radius considerably except for syrphid diversity as well as bee and syrphid plot visits. Variation of bee and syrphid diversity as well as syrphid plot visits are thus better explained by patterns at larger scales, while the extent of green area was more important than green area edge density. The different impact of the urban landscape on foraging of bees and syrphids with regards to plant diversity and floral abundance implies that management activities in the urban environment need to take into account the urban structure.

Zusammenfassung

Die meisten Blutenpflanzen¨ benotigen¨ Insekten als Vektoren fur¨ Pollentransfer und eine erfol- greiche Reproduktion. Die Effizienz des Bestaubungsprozesses¨ auf der Ebene einer einzelnen Pflanze kann durch den Reichtum an Arten in der Pflanzengesellschaft, deren Blutendichte¨ und der Landschaftsstruktur beeintrachtigt¨ werden, welche ebenso die Vielfalt der Bestauber¨ und die Abundanz der Blutenbesuche¨ beeinflußen. Pflanzen konnen¨ Betreff ihrer Blutenbesuche¨ miteinander konkurrieren oder einander unterstutzend¨ interagieren, das sich im reproduktiven Erfolg widerspiegeln kann. Die vorliegende Dissertation untersucht Fragen bezuglich¨ der Interak- tion von Pflanzen um Bestauberdienste¨ in (semi-)naturlichen¨ und urbanen Gebieten sowie den Effekt der urbanen Landschaft auf Pflanze-Bestauber-Interaktionen.¨ Im ersten Schritt wurden positive Interaktionen zwischen Pflanzen hinsichtlich des Einflußes auf Blutenbesuche¨ durch die Wirbellosenfamilien Bienen (Apidae) und Schwebfliegen (Syrphidae) untersucht. Eine nachfolgende Studie untersuchte dann die Wirkung des Artenreichtums und der Blutendichte¨ auf die Samenproduktion in zwei Pflanzenarten, die sich im Bezug auf ihre Bestaubergemeinschaften¨ unterscheiden. Im Anschluß daran wurde der Blick von naturnahen Landschaften auf urbane Landschaften geschwenkt und untersucht, wie eine Pflanzenart in einem urbanen Ambiente mit sympatrischen Arten im Hinblick auf Blutenbesuche¨ von allen potentiellen Bestauberarten,¨ Bienen und ihrem wichtigsten Besucher interagiert. Das letzte Kapitel analysiert die Auswirkungen der urbanen Matrix auf die Vielfalt sowie Haufigkeit¨ von blutenbesuchenden¨ Insekten. Bei der Untersuchung von Pflanzen-Interaktionen bezuglich¨ der Besuche von Bestaubern¨ (Kapitel 2), liessen sich nur kompetitive Wechselwirkungen mit zunehmender Artenzahl und Blutendichte¨ von heterospezifischen Pflanzen bei einer der vier untersuchten Pflanzenarten Anchusa officinalis (L.), Centaurea cyanus (L.), Centaurea jacea (L.s.l.) und Chrysanthemum praecox (HORVATIC´) feststellen. Die Blutenbesuche¨ durch Bienen nahmen mit steigender Blutendichte¨ heterospezifischer Arten nur bei der Art C. praecox ab, wahrend¨ Blutenbesuche¨ durch Schwebfliegen mit zunehmender Blutendichte¨ wie auch Artenzahl an Pflanzen ebenso nur in C. praecox abnahmen. Dieses Ergebnis kann durch die signifikanten Unterschiede in der Vielfalt und Haufigkeiten¨ an Blutenbesuchen¨ zwischen den vier Pflanzenarten erklart¨ werden. So wurden die drei Pflanzenarten, deren Butenbesuche¨ nicht durch den Artenreichtum und der Blutendichte¨ heterospezifischer Arten beeinflußt waren vor allem von den sozialen Honigbienen und Hummeln besucht. Pflanzen, welche demnach vor allem von sozialen Bienenarten besucht werden, konnten¨ daher weniger mit sympatrischen Pflanzenarten interagieren, wahrend¨ jene Pflanzenarten, die sowohl von Schwebfliegen wie Bienen in annahernd¨ gleichem Umfang besucht werden eher einem Wettbewerb um Blutenbesuche¨ mit anderen Pflanzenarten unterliegen. Da der Artenreichtum an Besuchern sich zwischen Pflanzenarten unterscheidet, konnen¨ kompetitive sowie unterstutzende¨ Wechselwirkungen zwischen Pflanzenarten fur¨ die Bestaubung¨ und damit deren Ausgang auf die Samenproduktion vom Ausmaß der Zusammensetzung der Bestaubergemeinschaft¨ abhangen.¨ Pflanzen, deren Bluten¨ von einer Vielzahl an Insekten bestaubt¨ werden sind moglicherweise¨ eher in Interaktionen (Wettbewerb oder mutualistische bzw. kom- mensalistische Interaktionen) involviert als jene, die von wenigen Insektenarten bestaubt¨ werden. Dem liegt zugrunde, daß bei einem großeren¨ Besucherpool eine hohere¨ Wahrscheinlichkeit besteht, daß es Arten gibt, die mehrere Pflanzen besuchen. Ich untersuchte diese Idee durch den Vergleich von Samenproduktion und Bestauberbesuchen¨ in den zwei Pflanzenarten, Borago officinalis, die von Bienen (hauptsachlich¨ von Hummeln) bestaubt¨ wurde, und Sinapis alba, eine Pflanzenart mit einem großerem¨ Besucherspektrum bestehend aus Bienen und Schwebfliegen. Eine Zunahem der Pflanzendiversitat¨ hatte eine positiven Effekt auf die Samenproduktion von Borago officinalis, jedoch nicht auf das Artenreichtum ihrer Besucher. Im Gegensatz dazu produzierte Sinapis alba weniger Samen mit zunehmender Pflanzendiversitat,¨ wenn zugleich der Artenreichtum an Besucher sich erhohte.¨ Die Blutendichte¨ aller Pflanzenarten hatte einen positiven Effekt auf die Samenproduktion von Borago officinalis, jedoch nur, wenn die Zahl an Blutenbesuchen¨ ebenso zunahm. Im Vergleich wirkte sich die Blutendichte¨ zwar auch positiv auf die Samenproduktion in S. alba aus, doch nur wenn die Artenvielfalt an Besuchern sich zugleich erhohte.¨ Diese Ergebnisse zeigen, dass die Pflanzendiversitat¨ und die Blutendichte¨ unterschiedlich und oft miteinander wechselwirkende Auswirkungen auf die Samenproduktion in zwei Pflanzenarten haben, die sich in ihren Bestaubergemeinschaften¨ unterscheiden. Die begrenzte Zahl von Pflanzenarten, die in dem Experiment genutzt wurde, lassen jedoch eine Verallgemeinerung der Ergebnisse in Bezug auf andere Pflanzen weniger zu, gleichwohl lassen die Ergebnisse die Vermutung zu, daß Wechselwirkungen zwischen Pflanzen um Bestauberdienste¨ von der Zusammensetzung und damit vom potentiellen Grad der Generalisierung innerhalb der Besuchergemeinschaft abhangen.¨ Es kann angenommen werden, daß die Zusammensetzung der Pflanzen- und Tiergemeinschaften in der urbanen Umwelt anders ist als in einer (semi-)naturlichen¨ Umwelt, was auf Fragmen- tierung, Veranderung¨ von Lebensraumen,¨ Verschmutzung und die Einfuhrung¨ von gebietsfrem- den Pflanzen- und Tierarten zuruckzuf¨ uhren¨ ist. Ich betrachtete die Interaktionen zwischen Pflanzenarten in einem urbanen Okosystem¨ mit Bezug auf die zwei Landschaftsmetriken Umfang und Grenzlinienanteil (Randbiotope) von Grunfl¨ achen.¨ Die Korrelationen der Blutenbesuche¨ pro Blute¨ bei der Zielpflanze Trifolium pratense mit den beiden Landschaftsmetriken ergaben, daß das Ausmaß der Grunfl¨ achen¨ die Besuche der Bestauber¨ mehr beeinflußte als der Gren- zlinienanteil. Die Korrelationskoeffizienten erhohten¨ sich bei beiden Landschaftsmetriken mit geographischer Skala, das auf zunehmende Bedeutung der Landschaftsmetriken mit Zunahme der betrachtenden Flache¨ hinweist. Ich zeigte auch, daß das Ausmaß der Grunfl¨ achen¨ die Pflanze- Pflanze-Interaktionen bezuglich¨ der Blutenbesuche¨ beeinflußte. So nahmen mit zunehmender Große¨ die negativen Auswirkungen von anderen Pflanzenarten auf die Bienenbesuche zu T. pratense ab. Dies deutet darauf hin, dass kompetitive Interaktionen zwischen Pflanzenarten um Besuche von Bestaubern¨ mit zunehmenden Ausmaß an Grunfl¨ achen¨ abnehmen. Jedoch hing dieser Zusammenhang maßgeblich vom haufigsten¨ Besucher, der Hummel Bombus pascuorum, ab. Bei der Analyse der Besuchsdaten ohne die Besuche von B. pascuorum hatte die Pflanzendiversitat¨ keinen Effekt. Allerdings nahm die Zahl der Besuche der verbleibenden Bienen mit zunehmender Abundanz heterospezifischer Bluten¨ und Grunfl¨ achen¨ ab. Die unterschiedlichen Ergebnisse zeigen, daß bei der Betrachtung von Pflanzenbesuchen durch Insekten neben der Pflanzengemeinschaft ebenso die Landschaftsstruktur und die Zusammensetzung der Bestaubergemeinschaft¨ berucksichtigt¨ werden muß. Die Vielfalt und Zahl an blutenbesuchenden¨ Insekten kann damit durch die Interaktion zwischen Landschaftsstruktur und der lokalen Pflanzengesellschaft beeinflußt werden. Der Einfluß des Umfangs und der Grenzliniendichte (Randbiotope) von Grunfl¨ achen¨ auf die Vielfalt von Bienen und Schwebfliegen und auf die lokalen Besuche der Untersuchungsflachen¨ (nicht der Bluten)¨ aller Bienen, Bienen mit Ausschluß des haufigsten¨ Besuchers, der Honigbiene Apis mellifera, und Schwebfliegen wurden in Bezug zu der lokal untersuchten Pflanzendiversitat¨ und Blutendichte¨ analysiert. Die Korrelationskoeffizienten zwischen den Antwortvariablen und dem Ausmaß an Grunfl¨ achen¨ nahmen mit der Skala zu (ausgenommen Besuche durch alle Bienen), wahrend¨ die Grenzliniendichte uber¨ die ganze analysierte Skala, d.h. 20 bis 200 m Radius, mit Ausnahme fur¨ die Schwebfliegenvielfalt sowie die Besuche durch Bienen und Schwebfliegen erheblich variierte. Die Variation in der Vielfalt an Bienen und Schwebfliegen sowie der lokalen Besuche durch Schwebfliegen sind somit besser durch Muster auf großeren¨ Skalen erklart,¨ wahrend¨ der Umfang an Grunfl¨ achen¨ wichtiger war als deren Grenzliniendichte (Randbiotope). Die unterschiedlichen Auswirkungen der urbanen Landschaft auf die Nahrungssuche von Bienen und Schwebfliegen in Wechselwirkung mit der Vielfalt und Blutenabundanz¨ zeigt auf, daß Management-Aktivitaten¨ im stadtischen¨ Umfeld die urbane Struktur und ihren unterschiedlichen Einfluß auf die jeweiligen Bestauberfamilien¨ berucksichtigen¨ mussen.¨

Chapter 1

Introduction

Pollination by animals is an essential ecological process, which ensures plant reproduction in 60-90 % of angiosperms (Richards, 1986; Renner, 2006; Kremen et al., 2007), and provides products and food for human mankind and animals (Kremen et al., 2007). Entomophilous pollination describes a mutualistic interaction between plants and insect pollinators. Pollinators forage on flowers for resources such as nectar and pollen, and shelter (Sakai, 2002), in the process they distribute pollen among flowers and thereby fertilize them. The quantity and quality of transferred pollen contributes to plant reproductive success. There are, however, factors outside (extrinsic) and inside (intrinsic) the system of interactions between plants, and between plants and pollinators, which influence pollen exchange and deposition in many plants (≈ 62%, Burd, 1994; Kwak et al., 1998; Ashman et al., 2004).

1.1 Competition and facilitation in plant-plant interactions for pollinator services

Plants often compete with each other for pollination services (Rathcke, 1983; Sargent and Ackerly, 2008). Competition for pollinators implies that some plants are enhancing services of pollinating animal species by the expense of co-occurring plants. Plants can compete by displaying more attractive floral morphologies (Kevan and Baker, 1983; Giurfa et al., 1999; Oberrath and Bohning-¨ Gaese, 1999; Dicks et al., 2002) or providing more floral resources (Potts et al., 2003, 2004). In competitive situations plants receive fewer pollinator visits and less pollen is deposited on flowers, 10 Chapter 1

resulting in lower seed set (pollen limitation, Ashman et al., 2004), provided of course that seed set is pollen limited and not resource limited (Knight et al., 2005; Yang et al., 2005). Competition can therefore drive evolution of plant traits and thereby enhance diversity of flowering plants (Bawa, 1995; Fenster et al., 2004). In contrast, facilitative interactions among plants are, in the context of pollination, those that mutually enhance pollen deposition among co-flowering plants (Schemske, 1981; Rathcke, 1983). As deposition of larger pollen quantities is more likely when more pollinators are visiting flowers (Engel and Irwin, 2003), plants can facilitate each other by increasing pollinator diversity and visit frequency. For example, a single highly attractive plant may benefit co-occurring heterospecifics by enhancing their visitation frequency (magnet species effect, Laverty, 1992; Johnson et al., 2003; Juillet et al., 2007; Molina-Montenegro et al., 2008; Jakobsson et al., 2009). Higher species richness of plants at a local level can also increase pollinator visits to individual plants by supporting a broad spectrum of pollinators (joint or complementary attraction, Sih and Baltus, 1987; Moeller, 2004; Ghazoul, 2006; Pontin et al., 2006). However, a larger diversity of pollinators can cause interference among flower visitors, displacing some pollinator species and possibly affecting the quality of the pollination services (Fontaine et al., 2008). Diverse or dense plant communities can counterbalance competition among pollinators by providing visitation alternatives, which are not or less visited by the dominating pollinator (competitor-free-space, Ghazoul, 2006). Furthermore, a diversity of plant species at a local scale can offer alternative or complementary resources (complementary resource hypothesis, Ghazoul, 2006; Geber and Moeller, 2006), which supports a wider array of pollinators by which pollination processes are buffered from population fluctuations of particular pollinating species. In addition to local scales, facilitative interactions can act over time scales. Diverse plant patches are more likely to supply flower visitors with resources over longer periods which maintain the pollinator community to the benefit of late flowering species (sequential mutualism, Waser and Real, 1979). Co-coccurring flowering heterospecifics have increased risks of heterospecific pollen transfer which may reduce seed set (McLernon et al., 1996; Morales and Traveset, 2008; Flanagan et al., 2009; Mitchell et al., 2009), although in high species densities can reduce the impact of improper pollen transfer (positive frequency dependence, Lamont et al., 1993; Groom, 1998; Anderson and Johnson, 2006). Conversely, overloaded pollen deposition as a consequence of high floral densities can cause also reduced fitness in terms of seed set due to pollen interactions (negative frequency dependence, Young and Young, 1992). In summary, several mechanisms, both negative and positive, interact in determining the effects of plant species on each other in species-rich communities. The outcome of such interactions is likely to be as much due to the mix of plant species as it is the mix of pollinator species attracted to the flowering community. 1.2 Pollination in the urban environment 11

1.2 Pollination in the urban environment

Competition and facilitation processes in plant pollination are also affected by extrinsic factors that shape the composition and behavior of the pollinator community. Facilitation among plants in other systems than pollination has been shown to increase with environmental severity, while competition dominates interactions among plants in benign environments (stress-gradient hypothesis, Choler et al., 2001; Callaway et al., 2002; Brooker, 2006). Humans have a large impact on ecosystems, notably by impoverishing biodiversity by habitat destruction and alteration, introduction of non-native species, climate change, and pollution. With increasing human population, the pressure on ecosystems will increase. A proper understanding of ecosystem processes is therefore essential for appropriate conservation management. Urban environments are among the most heavily impacted landscapes, which represent a heterogeneous environment, consisting of different landscape elements and microhabitats (Rebele, 1994). In urban environments semi-natural habitats occur as fragments surrounded by a more-or-less hostile and artificial matrix (island biogeography theory, metapopulation theory MacArthur and Wilson, 1963; Hanski, 1998), which can impede the movement of flower visiting animal species (Bhattacharya et al., 2003; Zanette et al., 2005) and reduce the local diversity of flower visitors (Gibb and Hochuli, 2002; Harris and Johnson, 2004; Ahrne´ et al., 2009). Reduced abundance and diversity of flower visitors implies possible pollen limitation in plant species within isolated urban patches (Cheptou and Avendano,˜ 2006; Spigler and Chang, 2009). Though plant species can overcome failed pollination success by self-fertilization (Kalisz and Vogler, 2003; Morgan et al., 2005; Charlesworth, 2006; Eckert et al., 2009), plant species in urban settings may benefit from facilitative interactions among co-occurring heterospecifics. Urban studies on flower visitor diversity in gardens, park areas and road verges have reported contrasting results (Saure, 1996; Stuke, 1998; Zanette et al., 2005; McFrederick and LeBuhn, 2006; Fetridge et al., 2008; Hopwood, 2008; Matteson et al., 2008). With increasing distance from the inner urban core to rural areas, landscape heterogeneity changes (urban-rural-gradient, Collins et al., 2000; McKinney, 2002). In semi-natural landscapes variation of flower visitor diversity and abundance in scale is significantly related to landscape heterogeneity (see e.g., Steffan-Dewenter and Tscharntke, 2002). Consequently, the impact of urban matrix on the diversity and abundance of pollinating animal species can vary with scale, and this may explain contrasting results among current studies. 12 Chapter 1

1.3 Objectives

This thesis seeks to address two main themes. The first relates to the extent to which facilitation among heterospecific co-flowering plants enhances pollinator attraction and seed set and has the following specific objectives:

1. Do interactions among plants in attracting pollinators shaped by the composition of the pollinator communities?

2. How is seed set of plant species affected by plant diversity and ﬂoral density?

The second theme extends this concept into urban areas where we anticipate that isolated urban flower patches will be particularly vulnerable to reduced pollinator numbers. We seek to find answers with the following specific objectives:

3. Do plant species benefit from co-occurring flowering plant species in attracting flower visitors in urban environments and how is this interaction affected by the structure of the urban environment (using the landscape metrics extent and edge density of green area)?

4. How are ﬂower visitor diversity and abundance inﬂuenced by the urban environment, based on two landscape metrics: extent and edge density of green area?

1.4 References

Ahrne,´ K., Bengtsson, J., and Elmqvist, T. (2009). Bumble Bees (Bombus spp.) along a Gradient of Increasing Urbanization. PLoS One (Public Library of Science Biology), 4(5):e5574.

Anderson, B. and Johnson, S. D. (2006). The effects of ﬂoral mimics and models on each others’ ﬁtness. Proceedings of the Royal Society B-Biological Sciences, 273(1589):969–974.

Ashman, T., Knight, T., Steets, J., Amarasekare, P., Burd, M., Campbell, D., Dudash, M., Johnston, M., Mazer, S., Mitchell, R., Morgan, M., and Wilson, W. (2004). Pollen limitation of plant reproduction: Ecological and evolutionary causes and consequences. Ecology, 85(9):2408– 2421. 1.4 References 13

Bawa, K. S. (1995). Pollination, seed dispersal and diversiﬁcation of angiosperms. Trends in Ecology & Evolution, 10(8):311–312.

Bhattacharya, M., Primack, R. B., and Gerwein, J. (2003). Are roads and railroads barriers to bumblebee movement in a temperate suburban conservation area? Biological Conservation, 109(1):37–45.

Brooker, R. W. (2006). Plant-plant interactions and environmental change. New Phytologist, 171(2):271–284.

Burd, M. (1994). Bateman Principle and plant reproduction - The role of pollen limitation and fruit and seed set. Botanical Review, 60(1):83–139.

Callaway, R. M., Brooker, R. W., Choler, P., Kikvidze, Z., Lortie, C. J., Michalet, R., Paolini, L., Pugnaire, F. I., Newingham, B., Aschehoug, E. T., Armas, C., Kikodze, D., and Cook, B. J. (2002). Positive interactions among alpine plants increase with stress. Nature, 417(6891):844– 848.

Charlesworth, D. (2006). Evolution of plant breeding systems. Current Biology, 16(17):R726– R735.

Cheptou, P.-O. and Avendano,˜ L. G. (2006). Pollination processes and the allee effect in highly fragmented populations: consequences for the mating system in urban environments. New Phytologist, 172(4):774–783.

Choler, P., Michalet, R., and Callaway, R. M. (2001). Facilitation and competition on gradients in alpine plant communities. Ecology, 82(12):3295–3308.

Collins, J. P., Kinzig, A., Grimm, N. B., Fagan, W. F., Hope, D., Wu, J. G., and Borer, E. T. (2000). A new urban ecology. American Scientist, 88(5):416–425.

Dicks, L. V., Corbet, S. A., and Pywell, R. F. (2002). Compartmentalization in plant-insect ﬂower visitor webs. Journal of Animal Ecology, 71(1):32–43.

Eckert, C. G., Kalisz, S., Geber, M. A., Sargent, R., Elle, E., Cheptou, P.-O., Goodwillie, C., Johnston, M. O., Kelly, J. K., Moeller, D. A., Porcher, E., Ree, R. H., Vallejo-Mar´ın, M., and Winn, A. A. (2009). Plant mating systems in a changing world. Trends in Ecology & Evolution, in press, corrected proof.

Engel, E. C. and Irwin, R. E. (2003). Linking pollinator visitation rate and pollen receipt. American Journal of Botany, 90(11):1612–1618. 14 Chapter 1

Fenster, C., Armbruster, W., Wilson, P., Dudash, M., and Thomson, J. (2004). Pollination syndromes and ﬂoral specialization. Annual Review of Ecology, Evolution and Systematics, 35:375–403.

Fetridge, E. D., Ascher, J. S., and Langellotto, G. A. (2008). The Bee Fauna of Residential Gardens in a Suburb of New York City (Hymenoptera: Apoidea). Annals of the Entomological Society of America, 101(6):1067–1077.

Flanagan, R. J., Mitchell, R. J., Knutowski, D., and Karron, J. D. (2009). Interspeciﬁc pollinator movements reduce pollen deposition and seed production in Mimulus ringens (Phrymaceae). American Journal of Botany, 96(4):809–815.

Fontaine, C., Collin, C. L., and Dajoz, I. (2008). Generalist foraging of pollinators: diet expansion at high density. Journal of Ecology, 96(5):1002–1010.

Geber, M. A. and Moeller, D. A. (2006). Pollinator responses to plant communities and implications for reproductive character evolution. In Harder, L. D. and Barrett, S. C., editors, Ecology and Evolution of Flowers, chapter 6, pages 102–119. Oxford University Press.

Ghazoul, J. (2006). Floral diversity and the facilitation of pollination. Journal of Ecology, 94:295–304.

Gibb, H. and Hochuli, D. F. (2002). Habitat fragmentation in an urban environment: large and small fragments support different arthropod assemblages. Biological Conservation, 106(1):91– 100.

Giurfa, M., Dafni, A., and Neal, P. R. (1999). Floral symmetry and its role in plant-pollinator systems. International Journal of Plant Sciences, 160(6):S41–S50.

Groom, M. J. (1998). Allee effects limit population viability of an annual plant. American Naturalist, 151(6):487–496.

Hanski, I. (1998). Metapopulation dynamics. Nature, 396(6706):41–49.

Harris, L. F. and Johnson, S. D. (2004). The consequences of habitat fragmentation for plant– pollinator mutualisms. International Journal of Tropical Insect Science, 24(1):29–43.

Hopwood, J. L. (2008). The contribution of roadside grassland restorations to native bee conservation. Biological Conservation, 141(10):2632–2640.

Jakobsson, A., Lazaro,´ A., and Totland, O. (2009). Relationships between the ﬂoral neighbor- hood and individual pollen limitation in two self-incompatible herbs. Oecologia, 160(4):707– 719. 1.4 References 15

Johnson, S. D., Peter, C. I., Nilsson, L. A., and Agren,˚ J. (2003). Pollination success in a deceptive orchid is enhanced by co-occurring rewarding magnet plants. Ecology, 84(11):2919– 2927.

Juillet, N., Gonzalez, M. A., Page, P. A., and Gigord, L. D. B. (2007). Pollination of the European food-deceptive Traunsteinera globosa (Orchidaceae): the importance of nectar- producing neighbouring plants. Plant Systematics and Evolution, 265(1-2):123–129.

Kalisz, S. and Vogler, D. W. (2003). Beneﬁts of autonomous selﬁng under unpredictable pollinator environments. Ecology, 84(11):2928–2942.

Kevan, P. G. and Baker, H. G. (1983). Insects as ﬂower visitors and pollinators. Annual Review of Entomology, 28:407–453.

Knight, T. M., Steets, J. A., Vamosi, J. C., Mazer, S. J., Burd, M., Campbell, D. R., Dudash, M. R., Johnston, M. O., Mitchell, R. J., and Ashman, T. L. (2005). Pollen limitation of plant reproduction: Pattern and process. Annual Review of Ecology, Evolution and Systematics, 36:467–497.

Kremen, C., Williams, N. M., Aizen, M. A., Gemmill-Herren, B., LeBuhn, G., Minckley, R., Packer, L., Potts, S. G., Roulston, T., Steffan-Dewenter, I., Vazquez, D. P., Winfree, R., Adams, L., Crone, E. E., Greenleaf, S. S., Keitt, T. H., Klein, A. M., Regetz, J., and Ricketts, T. H. (2007). Pollination and other ecosystem services produced by mobile organisms: a conceptual framework for the effects of land-use change. Ecology Letters, 10(4):299–314.

Kwak, M. M., Velterop, O., and van Andel, J. (1998). Pollen and gene ﬂow in fragmented habitats. Applied Vegetation Science, 1:37–54.

Lamont, B. B., Klinkhamer, P. G. L., and Witkowski, E. T. F. (1993). Population fragmentation may reduce fertility to zero in Banksia goodii - A demonstration of the Allee effect. Oecologia, 94(3):446–450.

Laverty, T. M. (1992). Plant interactions for pollinator visits: a test of the magnet species effect. Oecologia, 89(4):1432–1939.

MacArthur, R. H. and Wilson, E. O. (1963). An equilibrium theory of insular zoogeography. Evolution, 17(4):373–387.

Matteson, K. C., Ascher, J. S., and Langellotto, G. A. (2008). Bee richness and abundance in New York city urban gardens. Annals of the Entomological Society of America, 101(1):140–150.

McFrederick, Q. S. and LeBuhn, G. (2006). Are urban parks refuges for bumble bees Bombus spp. (Hymenoptera : Apidae)? Biological Conservation, 129(3):372–382. 16 Chapter 1

McKinney, M. L. (2002). Urbanization, biodiversity, and conservation. Bioscience, 52(10):883– 890.

McLernon, S. M., Murphy, S. D., and Aarssen, L. W. (1996). Heterospeciﬁc pollen transfer between sympatric species in a midsuccessional old-ﬁeld community. American Journal of Botany, 83(9):1168–1174.

Mitchell, R. J., Irwin, R. E., Flanagan, R. J., and Karron, J. D. (2009). Ecology and evolution of plant-pollinator interactions. Annals of Botany, 103(9):1355–1363.

Moeller, D. A. (2004). Facilitative interactions among plants via shared pollinators. Ecology, 85(12):3289–3301.

Molina-Montenegro, M. A., Badano, E. I., and Cavieres, L. A. (2008). Positive interactions among plant species for pollinator service: assessing the ’magnet species’ concept with invasive species. Oikos, 117(12):1833–1839.

Morales, C. L. and Traveset, A. (2008). Interspeciﬁc pollen transfer: Magnitude, prevalence and consequences for plant ﬁtness. Critical Reviews in Plant Sciences, 27(4):221–238.

Morgan, M. T., Wilson, W. G., and Knight, T. M. (2005). Plant population dynamics, pollinator foraging, and the selection of self-fertilization. American Naturalist, 166(2):169–183.

Oberrath, R. and Bohning-Gaese,¨ K. (1999). Floral color change and the attraction of insect pollinators in lungwort (pulmonaria collina). Oecologia, 121(3):383–391.

Pontin, D. R., Wade, M. R., Kehrli, P., and Wratten, S. D. (2006). Attractiveness of single and multiple species ﬂower patches to beneﬁcial insects in agroecosystems. Annals of Applied Biology, 148(1):39–47.

Potts, S. G., Vulliamy, B., Dafni, A., Ne’eman, G., and Willmer, P. (2003). Linking bees and ﬂowers: How do ﬂoral communities structure pollinator communities? Ecology, 84(10):2628– 2642.

Potts, S. G., Vulliamy, B., Roberts, S., O’Toole, C., Dafni, A., Ne’eman, G., and Willmer, P. G. (2004). Nectar resource diversity organises ﬂower-visitor community structure. Entomologia Experimentalis et Applicata, 113(2):103–107.

Rathcke, B. (1983). Competition and facilitation among plants for pollination. In Real, L., editor, Pollination Biology, pages 305–329. Academic Press.

Rebele, F. (1994). Urban ecology and special features of urban ecosystems. Global Ecology and Biogeography Letters, 4(6):173–187. 1.4 References 17

Renner, S. S. (2006). Rewardless ﬂowers in the angiosperms and the role of insect cognition in their evolution. In Waser, N. M. and Ollerton, J., editors, Plant-pollinator interactions: from specialization to generalization, chapter 6, pages 123–144. The University of Chicago Press.

Richards, A. J. (1986). Plant Breeding Systems. George Allen & Unwin.

Sakai, S. (2002). A review of brood-site pollination mutualism: plants providing breeding sites for their pollinators. Journal of Plant Research, 115(1119):161–168.

Sargent, R. D. and Ackerly, D. D. (2008). Plant-pollinator interactions and the assembly of plant communities. Trends in Ecolog & Evolution, 23(3):123–130.

Saure, C. (1996). Urban habitats for bees: the example of the city of Berlin. In Matheson, A., Buchmann, S. L., O‘Toole, C., Westrich, P., and Williams, I. H., editors, The Conservation of Bees, number 18 in Linnean Society Symposium, chapter 4, pages 47–53. Academic Press.

Schemske, D. W. (1981). Floral convergence and pollinator sharing in 2 bee-pollinated tropical herbs. Ecology, 62(4):946–954.

Sih, A. and Baltus, M. S. (1987). Patch size, pollinator behavior, and pollinator limitation in Catnip. Ecology, 68(6):1679–1690.

Spigler, R. B. and Chang, S. M. (2009). Pollen limitation and reproduction varies with population size in experimental populations of Sabatia angularis (Gentianaceae). Botany-Botanique, 87(3):330–338.

Steffan-Dewenter, I. and Tscharntke, T. (2002). Insect communities and biotic interactions on fragmented calcareous grasslands - a mini review. Biological Conservation, 104(3):275–284.

Stuke, J. (1998). Die Bedeutung einer stadtischen¨ Grunanlage¨ fur¨ die Schwebﬂiegenfauna (Diptera: Syrphidae) dargestellt am Beispiel des Bremer ”Stadtwaldes”. Abhandlungen. Natur- wissenschaftlicher Verein zu Bremen, 44(1):93–114.

Waser, N. M. and Real, L. A. (1979). Effective mutualism between sequentially ﬂowering plant–species. Nature, 281(5733):670–672.

Yang, C. F., Sun, S. G., and Guo, Y. H. (2005). Resource limitation and pollen source (self and outcross) affecting seed production in two louseworts, Pedicularis siphonantha and P. longiﬂora (Orobanchaceae). Botanical Journal of the Linnean Society, 147(1):83–89.

Young, H. J. and Young, T. P. (1992). Alternative outcomes of natural and experimental high pollen loads. Ecology, 73(2):639–647. 18 Chapter 1

Zanette, L. R. S., Martins, R. P., and Ribeiro, S. P. (2005). Effects of urbanization on Neotropical wasp and bee assemblages in a Brazilian metropolis. Landscape and Urban Planning, 71(2- 4):105–121. Chapter 2

Does plant diversity enhance pollinator facilitation? An experimental approach

Hennig, Ernest Ireneusz & Ghazoul, Jaboury

ETH Zurich,¨ Ecosystem Management, Institute of Terrestrial Ecosystems, Universitatstrasse¨ 16, 8092 Zurich,¨ Switzerland

Abstract The composition of a pollinator community visiting any single plant species can be affected by the density of floral cues. In this study we investigated possible differences in the attraction of bees and syrphids to plant species within a range of experimental plant communities that differ in species richness and floral abundance. Further, we examined how different responses of bees and syrphids might affect plant-plant interactions with focus on facilitation among plants for pollinator attraction. We used an experimental design with treatments of 1, 2, 4, and 6 plant species of different floral traits and recorded pollinator species and visits to all plant species. We investigated differences in two pollinator families, bees and syrphids, and the effect of plant species richness and heterospecific floral density on diversity and visitation frequency of these two families on Anchusa officinalis, Centaurea cyanus, C. jacea, and Chrysanthemum praecox. Our analysis revealed no effect of plant species richness treatment and heterospecific floral density on bee and syrphid diversity in each plant species, while bee visits were negatively influenced by heterospecific floral density only in C. praecox, likewise syrphid visits declined with increasing heterospecific floral density and plant species richness only in C. praecox. The three plant species unaffected by plant species richness and heterospecific floral density were mainly visited by honeybees and bumble bees. We conclude that plants being visited by social bee species may interact less with co-occurring plant species, while plant species visited mainly by syrphids compete 20 Chapter 2

with other plant species. Keywords: facilitation, competition, plants, bees, syrphids, diversity, visitation frequency, experimental design, subsets, unbalanced data

2.1 Introduction

In pollination ecology plants can facilitate each other by increasing floral visitation and hence reproductive success in two ways. First, where both plant species benefit from the presence of each other (mutualism), and second, where only one plant species is benefitting from the presence of the other (comensalism) (Callaway, 2007; Bronstein, 2009). Facilitation can be an important driver of plant community processes and has been shown to increase species diversity (Lortie, 2007; Valiente-Banuet and Verdu, 2007; Gross, 2008). In recent years several studies have shown facilitative effects in plant communities including the attraction of flower visiting animals (Johnson et al., 2003; Moeller, 2004; Ghazoul, 2006; Juillet et al., 2007; Ebeling et al., 2008). Understanding facilitation of pollinator attraction is important in the context of current anthropogenic threats, which may cause the disruption of mutualistic relationships, specifically in plant-pollinator interactions by altering community structure. Thus, habitat alteration and fragmentation can cause pollinator species decline (Harris and Johnson, 2004; Aguilar et al., 2006; Morandin et al., 2007; Pauw, 2007), climate change is assumed to disrupt plant-pollinator interactions (Memmott et al., 2007; Tylianakis et al., 2008; Hegland et al., 2009), and invasive or introduced species can disrupt native plant-pollinator interactions (Goulson, 2003; Ghazoul and Shaanker, 2004; Traveset and Richardson, 2006; Nienhuis et al., 2009). Several mechanisms of plant facilitation in pollination ecology have been proposed (Ghazoul, 2006). Plant patches with many different flowering species attract a variety of pollinator species due to differential attraction by plants of different pollinators (i.e., complementary attraction), and, additionally, due to the provision of different qualities and quantities of resources (i.e., resource complementarity, Manning, 2001; Klinkhamer et al., 2001; Muller¨ et al., 2006). In diverse plant patches more pollinators may also be encountered because floral refuges are available for pollinators that are excluded from the most resource-rich flowers (e.g., Reader et al., 2005). Less attractive plant species thus benefit in terms of floral visits from displaced pollinators. Similarly, plant patches with many flowers could have increased floral visits, because a patch with high abundance of flowers may be more attractive than a patch with few flowers. Thus, plant species which are flowering in multi-species patches may be buffered from anthropogenic threats by virtue of attraction of more pollinators, and a greater variety of pollinators, the latter providing 2.2 Materials & Methods 21

insurance to the loss of some pollinating species. There are also several negative impacts of flowering plant species within a diverse floral setting including improper pollen transfer (e.g., Bell et al., 2005), and competition for floral visits. Thus plants that have a smaller floral display (Grindeland et al., 2005; Ishii, 2006; Makino et al., 2007), or offer fewer floral resources (e.g., Sutherland et al., 1999), may be subjected to competition if pollinators limit seed set. Pollinator community composition and visitation pattern also depend on certain floral traits such as floral structure (Lunau and Maier, 1995; Giurfa et al., 1999; Gumbert et al., 1999; Fontaine et al., 2006). Syrphid flies and some solitary bee species may be less able to exploit flowers with deep hidden nectar, while open accessible flowers may be less visited by long-tongued bee species. Yellow and white coloured flowers (as seen by humans) may be more preferred by dipterans than hymenopterans, the latter tending to visit red and blue coloured flowers (Lunau and Maier, 1995; Sutherland et al., 1999). Therefore, we assume that the composition of multispecies patches is of key importance in determining the strength of facilitative and competitive interactions among plants for pollination services. Here, we aim to investigate if – there are any differences among pollinator families in the degree to which they are attracted to flower patches with different diversity and if – facilitative effects are widespread among entomophilous plants.

2.2 Materials & Methods

2.2.1 Study Design

The study was located at the research station Eschikon (SUI1 693851/SUI2 256288) in the canton Zurich,¨ North Switzerland. Thirty-eight plots of 1 m2 were established in a rectangular field of 17 x 38 m, which was surrounded by grass stripes and maize crop. Seventeen insect-pollinated plant species from different families and with different floral structures and colours were chosen from the BIOFLOR database (Klotz et al., 2002) (Tab. 2.1). Plants were ordered from ”UFA Samen” and ”Wildstaudengartnerei¨ Patricia Willi” and transplanted in May 2007 into plastic pots of 19 cm diameter and 25 cm height using soil from Klasmann ”Container Substrat” (pH 5.2, 210 mg N l−1, 240 mg P l−1, 270 mg K l−1). Plant species were randomly combined together in treatments with one, two, four, and six plant species and established in plots within the field. Each plot was located 2 m from the next and 22 Chapter 2

contained 24 individual plants. In single species plots, all 24 individuals were of the same plant species, in two-species plots, 12 individuals from each species were taken (Tab. 2.1). In a 4-species plot, each species had 6 individuals, and ﬁnally four individuals from six species were placed in six-species plots (Tab. 2.1). We randomly assigned treatments to each plot in the ﬁeld. To avoid selection and complementarity effects, each species was used additionally in a single species plot and in different combinations with the other study species.

2.2.2 Data Collection

One hundred and seventy-four observations were conducted on warm, unclouded days from 1000 to 1800 between May 21st until July 7th 2007. The order of observations at the plots was chosen randomly. Each plot was observed for 20 minutes and the position of observation was changed every five minutes to ensure pollinator access to treatments from different directions. Within each plot the pollinator species, the number of plot visits, and the number of flower visits by each pollinator were recorded. Pollinators were identified to species level by morphological structures such as colour patterns, wing structures, and foraging behaviour (e.g., syrphids typically hover in flowering patches). We recorded floral visits when pollinator landed and rummaged on flowers, while plot visits were recorded as pollinators entered the plot and landed on a flower. We did not count floral and plot visits for each individual separately, but pooled plot and floral visits for each pollinator species which ease the observation of several pollinator species at the same time. We could not therefore follow pollinators leaving and repeatedly visiting the same plot. As repeated plot and floral visits by the same pollinator were expected for all plots (and treatments) and plots were repeatedly observed during the duration of the experiment, we expected the possible overestimation of plot and floral visits to have minor effect in the analyses. The total number of flowers in each plot was also recorded after every observation period. Unknown pollinator species were caught for later identification.

2.2.3 Analysis

Visit number by each pollinator species to each plot was used to calculate a Shannon diversity index of flower visitors using vegan-package in R (Oksanen et al., 2009). The number of floral visits to each plant was divided by the number of flowers of each plant species to analyse differences in floral visits to the plants within each treatment. We used these response variables to analyze the relationship of bee and syrphid diversity as well as (log-transformed) floral visits 2.3 Results 23

in each plant species with plant species number (centered, i.e., values were substracted from the mean), floral density of heterospecifics (centered), and the interaction between these two explanatory variables in the frame of generalized least square models in the R-package nlme (Pinheiro et al., 2009). We used subsets of the data for each plant species and each of the pollinator families, Apidae and Syrphidae, due to the unbalanced data set by number of observations for plants, treatments and pollinator families, which can give false F-values and provide for wrong estimated coefficients (Quinn and Keough 2002, p.187-191; Gelman and Hill 2007, p.199-233). Differences in bee and syrphid diversity as well as log-transformed floral visits between plants and the two pollinator families we analyzed using Analysis of Variance after averaging the data at the plot level. For analyzing the relation of diversity and visits to plant species richness and floral density, we applied generalized least square models without correlation structure. However, as we conducted repeated observations on plots at different dates we analyzed two additional models accounting for correlation among observations on plots and on dates by applying compound symmetries (Zuur et al., 2009). We compared the three models using the log-Likelihood ratio and Akaike‘s Information Criterion (AIC) and selected the model with the lower AIC. The chosen model was investigated further for interaction among plant species number and floral density of heterospecifics using log-Likelihood ratio tests and maximum likelihood. Finally, we examined if accounting for different variances among treatments, plots, dates, and a combination of these variables improves model fit according to the AIC value. All models were investigated for violating the normal and homodascity assumptions. We did not use the model for syrphid visits in Centaurea jacea as the strong heterodascity could not be removed due to the few observations.

2.3 Results

2.3.1 Pollinator Species and Visits

A total of 105 flower visiting species were observed from 5,188 visits to the treatment plots with a total of 37,797 visited flowers in 174 20-min observations on 22 days. Most species visiting the experimental treatments were from the orders Diptera and Hymenoptera, as were most floral visits (Tab. 2.2). Coleoptera, Hemoptera, and Lepidoptera were of minor importance in species contribution and in visit frequency to treatments. Almost all hymenopteran flower visitors belonged to the family Apidae, while among Diptera, species from the families Syrphidae and Muscidae were most frequently observed (Tab. 2.2). 24 Chapter 2

The most frequently recorded species in the family Apidae were Apis mellifera, Bombus lapidarius, Bombus hortorum, and one unknown Lasioglossum species, while the most dominant syrphid species were Eristalis tenax, Episyrphus balteatus, and Sphaerophoria scripta (Tab. 2.2). The abundance of Apis mellifera may be artificially high owing to some hives located 300 m from the field site. The most common pollinator species, however, changed among the plant species. Visitation frequency to Anchusa officinalis, Centaurea cyanus, and C. jacea was dominated by bees, which contributed 97.7%, 94.6%, and 81.6% of flora visits, respectively (Tab. 2.2). In comparison, only 14.8% of floral visits in Chrysanthemum praecox were accounted by bees. The reverse can be found for syrphids. Visits by syrphids to A. officinalis, C. cyanus and C. jacea were of minor importance with 0.6%, 4.4%, and 10.6%, respectively, while in C. praecox half of the floral visits were by syrphids (48.2%, Tab. 2.2). As the plants were observed different number of times, we used below an ANOVA approach to compare mean visitation frequency.

2.3.2 Plant species and experimental treatments

The four treatments were observed a different number of times. Nine plots with each plant species alone were observed 84 times, ten plots with two plant species 61 times, four plots with four plant species 21 times, and two plots with six species just eight times. Flower density for one plant species in a single treatment ranged from maximal 1,725 flowers in Anchusa officinalis in a two-species treatment to just three flowers in Knautia arvensis and Centaurea jacea in a four-species treatment. Floral density of entire treatments ranged from six to 2,070 flowers and was significantly different between treatments (χ2 = 15.6, df = 3, p = 0.001). Considering the impact of treatments and floral density on the diversity of syrphid flies and bees, we focused on the four plant species Anchusa officinalis, Centaurea cyanus, Centaurea jacea, and Chrysanthemum praecox, which occur in almost all treatments (Tab. 2.1).

2.3.3 Visitor Diversity of Bees and Syrphids between Plant Species

F p < The ANOVA showed significant differences in bee ( 3, 20 = 12.34, 0.001) and syrphid F p Anchusa officinalis diversity ( 3, 19 = 4.09, = 0.02) among plant species. had the largest bee diversity, followed by Chrysanthemum praecox, Centaurea cyanus, and Centaurea jacea (Fig. 2.1). Other than Anchusa officinalis, there were not significant differences among plant species in terms of bee diversity (Fig, 2.1, Tab. 2.3). Syrphid diversity was lowest in A. officinalis and 2.3 Results 25

highest diversity was found in C. praecox. Among the two Centaurea species, C. cyanus had larger syrphid diversity than C. jacea (Fig. 2.1, Tab. 2.3). Comparing the two families in each plant species, we found a clear difference between bee and syrphid diversity only for A. officinalis F p < C. praecox ( 1,11 = 42.32, 0.001) in that syrphids were less diverse than bees. For syrphids F p were more diverse than bees but the significance was marginal ( 1,12 = 4.68, = 0.05). Number of visits by bees and syrphids also differed significantly among plant species (bees: F p < F p < 3, 20 = 11.19, 0.001; syrphids: 3, 19 = 11.51, 0.001). We observed most bee visits in A. officinalis (Tab. 2.3), though this was not significantly different from visits to C. cyanus. Chrysanthemum praecox was visited significantly least by bees, but it did not differ from bee visits in C. jacea. This contrasts visitation patterns observed for syrphids. Syrphids mostly visited C. praecox, though it did not differ significantly from visitation frequencies to C. jacea (Tab. 2.3). Centaurea cyanus was only visited significantly less than C. praecox, while A. officinalis was significantly least visited by syrphids (Fig, 2.1, Tab. 2.3).

2.3.4 Visits of Bees and Syrphids between Plant Species

Visitation frequency between the two pollinator families differed significantly in A. officinalis F p < C. cyanus F p < ( 1,11 = 188.4, 0.001) and ( 1,11 = 84.26, 0.001). In both plant species C. jacea F p fewer syrphid than bee visits were recorded (Fig. 2.1). In ( 1,4 = 0.4, = 0.56) and Chrysanthemum praecox F p ( 1,12 = 2.62, = 0.13) visits of bees and syrphids were not significantly different (Fig. 2.1). These results indicate that the four plant species were different attractive to the two flower visiting groups.

2.3.5 Plant interactions

Accounting for repeated observations at plot levels improved models by terms of Akaike‘s Infor- mation Criterion only for bee diversity in Chrysanthemum praecox, syrphid diversity in Centaurea cyanus and C. praecox, and bee visits in C. cyanus and C. praecox (Tab. 2.4). Considering the date of observation as possible correlation structure improved Akaike‘s Information Criterion for the models of bee diversity and bee visits in Centaurea cyanus (Tab. 2.4). We decided to take the results of these models for the discussion. Neither the diversity of bees, or syrphids was affected in any of the four plant species by the number of co-occurring plant species or floral density of co-occurring heterospecific plant species (Tab. 2.5). We found only marginally significant influence of plant species richness on syrphid 26 Chapter 2

Anchusa officinalis F p Chrysanthemum diversity in ( 1,17 = 3.55, = 0.08) and on bee diversity in praecox F p ( 1,46 = 3.54, = 0.07). Visitation frequencies of bees were influenced by floral density in Chrysanthemum praecox. Larger floral densities of conspecifics significantly reduced bee visits to C. praecox (βˆ = – 0.07, F p C. praecox 1,45 = 8.97, = 0.005). Syrphid visits to were significantly influenced by plant species richness and heterospecific floral density. Though syrphid visits increased with plant species richness, increasing floral density of co-occurring heterospecifics reduced syrphid vis- C. praecox ˆ F p itation frequency to (β = – 0.05, 1,59 = 4.09, = 0.05). Bee and syrphid visits in A. officinalis, C. cyanus, and C. jacea were neither influenced by plant species number, nor by heterospecific floral density or the interaction between these two explanatory variables (Tab. 2.6).

2.4 Discussion

2.4.1 Diversity and visits of ﬂower visitors

Most plant species are assumed to be generalists in terms of their pollinators (Waser et al., 1996), though the difference in diversity among bees and syrphids at each plant species and among plant species might be attributed to functional floral traits, including colour and aspects of morphology (Fenster et al., 2004; Glover, 2007). Anchusa officinalis, for example, has long-tubular flowers which prevents access to floral resources by pollinators that lack long tongues, including almost all syrphids (except Rhingia campestris). Notwithstanding that many small bees also lack long tongues to access these flowers, this morphologically limiting access to floral resources likely explains the significant difference between bee and syrphid diversity at this plant species. The flowers of Centaurea cyanus, C. jacea and Chrysanthemum praecox consist of ray and disc florets, which are more accessible to short-tongued pollinators, hence explaining the lack of differences between syrphids and bees in these species. Floral colour preferences might also explain visitation patterns among plant species. In A. officinalis, bumblebees were the most common pollinators (Bombus hortorum 4,416 visits, and B. pascuorum 4,224 visits) which contributed 56.4% of all visits (15,311) at A. officinalis. Bumblebees prefer blue coloured flowers (Lunau and Maier, 1995), which can explain the large visitation frequency of bumblebees to A. officinalis. Likewise the large visitation frequency of the syrphid species Eristalis tenax to Chrysanthemum praecox can be attributed to their innate preference for yellow-flowering plants (Lunau and Maier, 1995). However, innate preference alone might not explain the preference to particular plant species. 2.4 Discussion 27

Honeybees have a lesser innate preference for particular floral colours (Lunau and Maier, 1995). Although hives in the proximity of the study area may have distorted the visitation patterns of honeybees, we observed the largest contribution of honeybees to visitation frequencies in the plant species C. cyanus and C. jacea. The fact that other plant species were offered as potential resources the same time in the experimental design, which received far less honeybee visits, allows the assumption that floral cues other than floral colours were attracting honeybees to these two focal plants. Honeybees showed floral constancy to plant species with high reward amount (Gegear and Laverty, 2004). With regards to the communication abilities of honeybees in terms of location and reward in plant species, C. cyanus and C. jacea could have offered large reward amounts and were therefore preferentially visited by honeybees.

2.4.2 Plant interactions – facilitation, competition or no interactions?

The study showed that pollinator families differed in diversity and visitation frequency among plant species. There were, however, no significant positive effects of plant species richness or floral density on diversity of bees and syrphids in the focal plants. Instead, we found heterospecific floral density to negatively influence bee visits and in interaction with plant species richness negatively syrphid visits to Chrysanthemum praecox. Thus, these results provide no evidence for facilitation among plant species for pollinator diversity or pollinator floral visits, but rather evidence of competition among plant species. Despite some studies reporting significant positive interactions among plant species that serve to enhance pollinator attraction (Moeller, 2004; Ghazoul, 2006; Hegland et al., 2009; Lazaro´ et al., 2009), there are number of other studies where facilitative effects have not been found (Feldman et al., 2004; Kirchner et al., 2005; Feldman, 2008). The strength of interplant interactions through pollinator sharing may depend on the identity of the pollinating taxon. Thus, Hegland et al. (2009) found 14.1% of pairwise comparisons among preferentially bee-visited plants to be positively significant, while only 6.25% of all pairwise comparisons were significantly positive between preferentially fly-visited plants. In our study only Chrysanthemum praecox competed for bee and syrphid visits with co-occurring plant species and floral density of heterospecific plant species. The absence of facilitation in terms of increasing bee and syrphid visits with plant species richness or floral density in all four plant species might simply be due to the fidelity of pollinator species (Lazaro´ et al., 2009). Thus, other co-occurring plant species and the abundance of heterospecific flowers might have little effect on pollinator diversity or visitation patterns of the focal plant when pollinators prefer to visit one plant species. Indeed, preference of particular flowering species by bumblebees has been previously demonstrated (Gegear and Laverty, 2005), which even in our study accounted for 92% of floral visits to A. officinalis, and this may explain the absence of 28 Chapter 2

beneficial interactions with co-occurring heterospecific plant species for A. officinalis. Honeybees were the primary bee visitors to flowers of Centaurea cyanus and C. jacea (86.8% and 75.7%, respectively) again suggesting large preference for these plant species which renders facilitatory interactions in relatively diverse swards unlikely. Conversely, Chrysanthemum praecox was preferentially visited by syrphids, which can also show preferences to certain plant species (Goulson and Wright, 1998; Nakano and Washitani, 2003; Pontin et al., 2006; Amaya-Marquez, 2009). The preference of the syrphid flies Eristalis tenax and Episyrphus balteatus for flowers of Chrysanthemum praecox makes it thus likely that the absence of facilitative interactions among plant species can be referred to fidelity of syrphids to plant species. Chrysanthemum praecox has a more generalized visitor community than the other focal plants. As a generalist-pollinated plant it is more likely that Chrysanthemum praecox is vulnerable to competition with other plant species for shared pollinators (e.g., van der Muren and Kwak, 2003; Lazaro´ et al., 2009). As the number of plant species increases there is an increased likelihood for competition among plants that share the same pollinators. Therefore generalist plant species might be more likely to suffer competition as plant species richness or floral density of co-occurring generalist plant species increases (Lazaro´ et al., 2009). This might explain the negative effect of plant species richness and floral abundance on visitation frequency in Chrysanthemum praecox.

2.5 Conclusion

This study provides evidence that interactions among plant species for attracting pollinators is shaped by the particular pollinating taxa that visits the plants. In particular, plant species that are preferentially visited by social bees, i.e. honeybees and bumble bees, were not affected by surrounding plant species richness, nor by heterospecific floral density. It can be therefore assumed that plant species visited most frequently by social bees are effectively removed from facilitatory interactions with other heterospecifics. In the total absence of such plants it is possible that social bees might switch to other more open flowers, but these are already well visited by a variety of generalist pollinators, hence facilitation might be irrelevant in these experimental systems. 2.6 References 29

2.6 References

Aguilar, R., Ashworth, L., Galetto, L., and Aizen, M. A. (2006). Plant reproductive suscepti- bility to habitat fragmentation: review and synthesis through a meta-analysis. Ecology Letters, 9(8):968–980.

Amaya-Marquez, M. (2009). Floral constancy in bees: a revision of theories and a comparison with other pollinators. Revista Colombiana de Entomologia, 35(2):206–216.

Bell, J. M., Karron, J. D., and Mitchell, R. J. (2005). Interspeciﬁc competition for pollination lowers seed production and outcrossing in Mimulus ringens. Ecology, 86(3):762–771.

Bronstein, J. L. (2009). The evolution of facilitation and mutualism. Journal of Ecology, 97(6):1160–1170.

Callaway, R. M. (2007). Positive interactions and interdependence in plant communities. Springer.

Ebeling, A., Klein, A. M., Schumacher, J., Weisser, W. W., and Tscharntke, T. (2008). How does plant richness affect pollinator richness and temporal stability of ﬂower visits? Oikos, 117(12):1808–1815.

Feldman, T. S. (2008). The plot thickens: does low density affect visitation and reproductive success in a perennial herb, and are these effects altered in the presence of a co-ﬂowering species? Oecologia, 156(4):807–817.

Feldman, T. S., Morris, W. F., and Wilson, W. G. (2004). When can two plant species facilitate each other’s pollination? Oikos, 105(1):197–207.

Fenster, C., Armbruster, W., Wilson, P., Dudash, M., and Thomson, J. (2004). Pollination syndromes and ﬂoral specialization. Annual Review of Ecology, Evolution and Systematics, 35:375–403.

Fontaine, C., Dajoz, I., Meriguet, J., and Loreau, M. (2006). Functional diversity of plant- pollinator interaction webs enhances the persistence of plant communities. PLOS Biology, 4(1):129–135.

Gegear, R. J. and Laverty, T. M. (2004). Effect of a colour dimorphism on the ﬂower constancy of honey bees and bumble bees. Canadian Journal of Zoology - Revue Canadienne de Zoologie, 82(4):587–593. 30 Chapter 2

Gegear, R. J. and Laverty, T. M. (2005). Flower constancy in bumblebees: a test of the trait variability hypothesis. Animal Behaviour, 69:939–949.

Ghazoul, J. (2006). Floral diversity and the facilitation of pollination. Journal of Ecology, 94:295–304.

Ghazoul, J. and Shaanker, R. U. (2004). Sex in space: Pollination among spatially isolated plants. Biotropica, 36(2):128–130.

Giurfa, M., Dafni, A., and Neal, P. R. (1999). Floral symmetry and its role in plant-pollinator systems. International Journal of Plant Sciences, 160(6):S41–S50.

Glover, B. (2007). Understanding Flower & Flowering. An Integrated Approach. Oxford University Press.

Goulson, D. (2003). Effects of introduced bees on native ecosystems. Annual Review of Ecology, Evolution and Systematics, 34:1–26.

Goulson, D. and Wright, N. P. (1998). Flower constancy in the hoverﬂies Episyrphus balteatus (Degeer) and Syrphus ribesii (L.) (Syrphidae). Behavioral Ecology, 9(3):213–219.

Grindeland, J. M., Sletvold, N., and Ims, R. A. (2005). Effects of ﬂoral display size and plant density on pollinator visitation rate in a natural population of Digitalis purpurea. Functional Ecology, 19(3):383–390.

Gross, K. (2008). Positive interactions among competitors can produce species-rich communities. Ecology Letters, 11(9):929–936.

Gumbert, A., Kunze, J., and Chittka, L. (1999). Floral colour diversity in plant communities, bee colour space and a null model. Proceedings of the Royal Society of London Series B - Biological Sciences, 266(1429):1711–1716.

Harris, L. F. and Johnson, S. D. (2004). The consequences of habitat fragmentation for plant– pollinator mutualisms. International Journal of Tropical Insect Science, 24(1):29–43.

Hegland, S. J., Nielsen, A., Lazaro, A., Bjerknes, A. L., and Totland, O. (2009). How does climate warming affect plant-pollinator interactions? Ecology Letters, 12(2):184–195.

Ishii, H. (2006). Floral display size inﬂuences subsequent plant choice by bumble bees. Func- tional Ecology, 20(2):233–238.

Johnson, S. D., Peter, C. I., Nilsson, L. A., and Agren,˚ J. (2003). Pollination success in a deceptive orchid is enhanced by co-occurring rewarding magnet plants. Ecology, 84(11):2919– 2927. 2.6 References 31

Kirchner, F., Luijten, S., Imbert, E., Riba, M., Mayol, M., Gonzalez-Martinez, S., Mignot, A., and Colas, B. (2005). Effects of local density on insect visitation and fertilization success in the narrow-endemic Centaurea corymbosa (Asteraceae). Oikos, 111(1):130–142.

Klinkhamer, P. G. L., de Jong, T. J., and Linnebank, L. A. (2001). Small-scale spatial patterns determine ecological relationships: an experimental example using nectar production rates. Ecology Letters, 4(6):559–567.

Klotz, S., Kuhn,¨ I., and Durka, W. (2002). BIOFLOR - Eine Datenbank mit biologisch- okologischen¨ Merkmalen zur Flora von Deutschland, volume 38 of Schriftenreihe fur¨ Veg- etationskunde. Bundesamt fur¨ Naturschutz.

Lazaro,´ A., Lundgren, R., and Totland, O. (2009). Co-ﬂowering neighbors inﬂuence the diversity and identity of pollinator groups visiting plant species. Oikos, 118(5):691–702.

Lortie, C. J. (2007). An ecological tardis: the implications of facilitation through evolutionary time. Trends in Ecology & Evolution, 22(12):627–630.

Lunau, K. and Maier, E. J. (1995). Innate color preferences of ﬂower visitors. Journal of Comparative Physiology A – Sensory Neural and Behavioral Physiology, 177(1):1–19.

Makino, T. T., Ohashi, K., and Sakai, S. (2007). How do floral display size and the density of surrounding flowers influence the likelihood of bumble bee revisitation to a plant? Functional Ecology, 21(1):87–95.

Manning, R. (2001). Fatty acids in pollen: a review of their importance for honey bees. Bee World, 82(2):60–75.

Memmott, J., Craze, P. G., Waser, N. M., and Price, M. V. (2007). Global warming and the disruption of plant-pollinator interactions. Ecology Letters, 10(8):710–717.

Moeller, D. A. (2004). Facilitative interactions among plants via shared pollinators. Ecology, 85(12):3289–3301.

Morandin, L. A., Winston, M. L., Abbott, V. A., and Franklin, M. T. (2007). Can pastureland increase wild bee abundance in agriculturally intense areas? Basic and Applied Ecology, 8(2):117–124. 32 Chapter 2

Muller,¨ A., Diener, S., Schnyder, S., Stutz, K., Sedivy, C., and Dorn, S. (2006). Quantitative pollen requirements of solitary bees: Implications for bee conservation and the evolution of bee-ﬂower relationships. Biological Conservation, 130(4):604–615.

Nakano, C. and Washitani, I. (2003). Variability and specialization of plant-pollinator systems in a northern maritime grassland. Ecological Research, 18(3):221–246.

Nienhuis, C. M., Dietzsch, A. C., and Stout, J. C. (2009). The impacts of an invasive alien plant and its removal on native bees. Apidologie, 40(4):450–463.

Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Simpson, G. L., Solymos, P., Stevens, M. H. H., and Wagner, H. (2009). vegan: Community Ecology Package. R Cran – The R Project for Statistical Computing. R package version 1.15-4.

Pauw, A. (2007). Collapse of a pollination web in small conservation areas. Ecology, 88(7):1759– 1769.

Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., and the R Core team (2009). nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1-92.

Reader, T., MacLeod, I., Elliott, P. T., Robinson, O. J., and Manica, A. (2005). Inter-order interactions between flower-visiting insects: Foraging bees avoid flowers previously visited by hoverflies. Journal of Insect Behavior, 18(1):51–57.

Sutherland, J. P., Sullivan, M. S., and Poppy, G. M. (1999). The influence of floral character on the foraging behaviour of the hoverfly, Episyrphus balteatus. Entomologia Experimentalis et Applicata, 93(2):157–164.

Traveset, A. and Richardson, D. M. (2006). Biological invasions as disruptors of plant reproductive mutualisms. Trends in Ecology & Evolution, 21(4):208–216.

Tylianakis, J. M., Didham, R. K., Bascompte, J., and Wardle, D. A. (2008). Global change and species interactions in terrestrial ecosystems. Ecology Letters, 11(12):1351–1363.

Valiente-Banuet, A. and Verdu, M. (2007). Facilitation can increase the phylogenetic diversity of plant communities. Ecology Letters, 10(11):1029–1036. 2.6 References 33

van der Muren, C. and Kwak, M. M. (2003). Insect diversity on yellow Asteraceae in road verges in the Netherlands. Proceedings of the Section Experimental and Applied Entomology of the Netherlands Entomological Society, 14:115–118.

Waser, N. M., Chittka, L., Price, M. V., Williams, N. M., and Ollerton, J. (1996). Generalization in pollination systems, and why it matters. Ecology, 77(4):1043–1060.

Zuur, A. F., Ieno, E. N., Walker, N. J., Saveliev, A. A., and Smith, G. M. (2009). Mixed Effects Models and Extensions in Ecology with R. Springer Verlag. 34 Chapter 2

2.7 Tables and Figures

Table 2.1: Plant species combinations used in the experimental design and the number of observations (n). The plant species are: Ao = Anchusa ofﬁcinalis L., At = Anthemis tinctoria L., Bs = Buphthalum salicifolium L., Cj = Centaurea jacea L.s.l., Cc = Centaurea cyanus L., Cm = Chelidonium majus L., Cp = Chrysanthemum praecox Horvatic,´ Hn = Helianthemum nummularium (L.) Mill. s.l., Ka = Knautia arvensis (L.) Coult. s.str., Mm = Malva moschata L., Sl = Silene latifolia Poir., So = Scabiosa ochroleuca L., Sv = Silene vulgaris (Moench) Garcke s.l., Tc = Tanacetum corymbosum (L.) Schultz-Bip.

1-spec. n 2-spec. n

Ao 14 Ao-Cc 1 At 2 Ao-Cp 15 Cc 15 Ao-Mm 2 Cj 12 At-Mm 1 Cm 11 At-Tc 3 Cp 18 Cc-Cp 13 Hn 4 Cc-Hn 7 Mm 2 Cm-Bs 8 Sl 6 Cp-Hn 8 So-Sv 3 4-spec. n 6-spec. n

Ao-Cc-Cj-Ka 7 Ao-At-Bs-Cp-Mm-Sv 4 At-Bs-Cp-Tc 9 Cc-Cj-Mm-So-Sv-Tc 4 Ao-At-Cm-Mm 4 At-Cm-Sv-Tc 1 TOTAL 174 2.7 Tables and Figures 35

Table 2.2: Species richness, plot visits, and ﬂoral visits of each order. The contribution of the most common species of the Hymenoptera and Diptera is given.

all Ao Cc Cj Cp taxa spp. plot visits floral visits floral visits floral visits floral visits floral visits

Hymenoptera 25 2,549 (49.2%) 28,151 (74.5%) 14,955 (53.1%) 8,861 (31.5%) 905 (3.2%) 1,018 (3.6%) Apidae 22 2,512 (98.5%) 28,072 (99.7%) 14,954 (99.9%) 8,861 (100%) 905 (100%) 953 (93.6%) Apis mellifera 1,336 (53.2%) 10,509 (37.4%) 825 (5.5%) 7,692 (86.8%) 685 (75.7%) 27 (2.8%) Bombus lapidarius 4,827 (17.1%) 4,827 (17.2%) 3,504 (23.4%) 963 (10.9%) 152 (16.8%) 6 (0.6%) Bombus hortorum 160 (6.4%) 4,496 (16%) 4,416 (29.5%) 67 (0.8%) − 3 (0.3%) Lasioglossum spec. 222 (8.8%) 653 (2.3%) 9 (0.06%) 27 (0.3%) 9 (1%) 224 (23.5%)

Diptera 56 2,537 (49%) 9,285 (24.5%) 125 (1.3%) 495 (5.3%) 202 (2.2%) 5,343 (57.5%) Syrphidae 26 1,326 (52.3%) 6,209 (66.9%) 92 (73.6%) 414 (83.6%) 118 (58.4%) 3,107 (58.2%) Eristalis tenax 604 (45.6%) 4,536 (73.1%) 36 (39.1%) 104 (25.1%) 77 (65.3%) 2,386 (76.8%) Episyrphus balteatus 183 (13.8%) 495 (8%) 20 (21.7%) 194 (46.9%) 23 (19.5%) 134 (4.3%) Sphaerophoria scripta 146 (11%) 363 (5.8%) 9 (9.8%) 15 (3.6%) 5 (0.4%) 189 (6.1%) Muscidae 14 771 (30.4%) 2,081 (22.4%) 15 (12%) 30 (6.1%) 20 (9.9%) 1,648 (30.8%)

Coleoptera 12 51 (1%) 68 (0.2%) − 3 (4.4%) − 49 (72.1%)

Hemiptera 9 26 (0.5%) 34 (0.1%) − − − 32 (94.1%)

Lepidoptera 3 20 (0.4%) 259 (0.7%) 231 (89.2%) 3 (1.2%) 2 (0.8%) 5 (1.9%) TOTAL 105 5,183 (100%) 37,797 (100%) 15,311 (40.5%) 9,362 (24.8%) 1,109 (2.9%) 6,447 (17.1%)

Table 2.3: t-values from the Analysis of Variance for bee and syrphid diversity, and bee and syrphid log-transformed visits among Anchusa ofﬁcinalis, Centaurea cyanus, Centaurea jacea, Chrysanthemum praecox. Asteriks indicate signiﬁcant results (∗ = < 0.05, ∗∗ = < 0.01, ∗∗∗ = < 0.001).

Bees Syrphids Ao Cc Cj Cp Ao Cc Cj Cp Ao −5.016∗∗∗ −4.502∗∗∗ −4.801∗∗∗ 2.533∗ 0.487 3.123∗∗ Shannon Diversity Cc −0.616 0.215 −1.543 0.613 Cj 0.783 2.018

Ao −0.309 −2.839∗ −4.955∗∗∗ 2.525∗ 2.886∗∗ 5.795∗∗∗ Visitation Frequency Cc −2.599∗ −4.646∗∗∗ 0.921 3.404∗∗ Cj −0.999 −1.715 36 Chapter 2

Table 2.4: Test for the importance of correlation structures using repeated observations at the plot level (”Plot Number”) and on each day (”Date”) for each plant species and pollinator family and the response variables ”diversity” (upper table section) and ”floral visits per flower density” of the focal plant species (lower table section). We used always models with lowest Akaike‘s Information Criterion, which provide better fits.

Original Model Plot Number Date AIC AIC L.Ratio p-value AIC L.Ratio p-value Apidae 47.24 49.23 0.008 0.93 49.21 0.029 0.86 A. ofﬁcinalis Syrphidae 32.5 34.34 0.158 0.69 32.71 1.795 0.18 Apidae 57.8 56.79 0.0013 0.97 54.42 2.373 0.124 C. cyanus Syrphidae 68.86 66 4.859 0.028 70.63 0.229 0.633 Apidae 43.09 45.09 5.9−05 0.99 43.45 1.639 0.2 C. jacea Syrphidae 25.7 27.7 1.24−14 1 26.002 5.329−15 1 Apidae 78.21 77.11 3.092 0.08 75.77 4.439 0.035 C. praecox Syrphidae 86.61 85.06 3.544 0.06 88.6 0.011 0.92

Apidae 100.27 102.27 0.0004 0.98 101.78 0.492 0.483 A. ofﬁcinalis Syrphidae −53.79 −52.33 0.533 0.465 −52.295 0.503 0.478 Apidae 123.94 124.77 1.172 0.279 122.175 3.767 0.052 C. cyanus Syrphidae 19.86 20.058 1.804 0.179 21.85 0.012 0.914 Apidae 85.02 87.02 3.25−09 1 85.73 1.283 0.257 C. jacea Syrphidae − − − − − − − Apidae –1.48 –3.73 4.25 0.039 −0.023 0.541 0.462 C. praecox Syrphidae 108.48 109.86 0.616 0.433 110.42 0.058 0.81

Table 2.5: Estimates and F-values from the generalized least square models with diversity of bees and syrphids as a function of plant species treatment (PS), floral density of heterospecifics (FAD) and the interaction in each plant species (Anchusa officinalis, Centaurea cyanus, Centaurea jacea, Chrysanthemum praecox). Asteriks indicate significant results (∗ = < 0.05, ∗∗ = < 0.01, ∗∗∗ = < 0.001).

Bees Syrphids Plant Species EXP numDF denDF βˆ F numDF denDF βˆ F

PS 1 38 6.89−05 <0.001 1 17 −0.091 3.547 Anchusa ofﬁcinalis FAD 1 38 −9.37−05 <0.001 1 17 −0.0005 0.389 PS:FAD − − − − − − − −

PS 1 41 0.046 1.082 1 34 −0.038 0.152 Centaurea cyanus FAD 1 41 0.0053 0.035 1 34 −0.047 0.325 PS:FAD − − − − − − − −

PS 1 17 0.193 0.644 1 8 3.256 1.294 Centaurea jacea FAD 1 17 −0.138 0.766 1 8 −2.327 1.484 PS:FAD − − − − − − − −

PS 1 46 0.602 3.538 1 43 0.001 0.0002 Chrysanthemum praecox FAD 1 46 −0.009 0.709 1 43 0.036 0.196 PS:FAD − − − − − − − − 2.7 Tables and Figures 37

Table 2.6: Estimates and F-values from the generalized least square models with bee and syrphid visits per flower as a function of plant species treatment (PS), floral density of heterospecifics (FAD) and the interaction in each plant species (Anchusa officinalis, Centaurea cyanus, Centaurea jacea, Chrysanthemum praecox). Asteriks indicate significant results (∗ = < 0.05, ∗∗ = < 0.01, ∗∗∗ = < 0.001).

Bees Syrphids Plant Species EXP numDF denDF βˆ F numDF denDF βˆ F

PS 1 39 0.0724 0.803 1 16 0.0074 0.807 Anchusa ofﬁcinalis FAD 1 39 −0.012 0.041 1 16 −0.0023 0.292 PS:FAD − − − − 1 16 −0.0045 1.837

PS 1 41 −0.039 0.149 1 35 −0.0062 0.271 Centaurea cyanus FAD 1 41 −0.088 1.787 1 35 0.009 0.698 PS:FAD − − − − − − − −

PS 1 17 0.109 0.996 − − − − Centaurea jacea FAD 1 17 −0.160 0.465 − − − − PS:FAD − − − − − − − −

PS 1 45 0.0769 2.754 1 59 0.109 2.335 Chrysanthemum praecox FAD 1 45 –0.0684 8.965∗∗ 1 59 −0.0807 3.665 PS:FAD 1 45 −0.0255 1.821 1 59 –0.0526 4.092∗

● 2

1.5

● 0

● 1.0 −2

Diversity −4 0.5

● ● Log (Visits/Floral Density) Log (Visits/Floral ● ● −6 ●

● 0.0 ● ● −8 a s a s a s a s a s a s a s a s Ao Cc Cj Cp Ao Cc Cj Cp

Figure 2.1: The boxplots are showing differences in mean diversity (upper left) and visitation frequency per floral density (upper right) for bees (a) and syrphid flies (s) for each plant species (Ao = Anchusa officinalis, Cc = Centaurea cyanus, Cj = Centaurea jacea, and Cp = Chrysanthemum praecox).

Chapter 3

Linking competition and facilitation processes in pollination systems to plant species richness

Hennig, Ernest Ireneusz1, Arnold, Philipp2 and Jaboury Ghazoul1

1 ETH Zurich,¨ Ecosystem Management, Institute of Terrestrial Ecosystems, Universitatstrasse¨ 16, 8092 Zurich,¨ Switzerland 2 University of Fribourg, Department of Biology, Unit of Ecology & Evolution, Chemin du Musee´ 10, 1700 Fribourg, Switzerland

Abstract Plant species can benefit or compete with co-occurring sympatrics for pollinator services. The outcome of this interaction on seed set may depend on the degree of pollinator generalization. We used an experimental block design and two plant species with different degrees of pollinator generalization to investigate the impact of plant species richness and heterospecific floral density on the pollinator community and seed set of each plant species. We recorded in two experiments plant species richness, floral density, visitor species richness and visitation frequency at the treatment and plant level. Pollinator community and visitation frequency differed significantly between the two plant species. Further, a positive effect of plant species richness on seed set was observed in the more specialized plant species, but plant species richness had no impact on the plant insect visitor community. Species richness of flower visitors reduced seed set. Overall, flower density had a positive effect on seed set in the specialized plant species, but this depends on 40 Chapter 3

the number of floral visits. In contrast, the generalist plant species had reduced seed set when visitor species richness increased in species-poor plant treatments, while floral density positively affected seed set when visitation frequency increased. The results indicated that the effect of plant species richness and floral density on seed set is mediated by the type of pollinator community at the plant species in question. Only one specialist plant species appear to benefit from plant species richness, while only one generalist plant species compete for pollination services. Such plant interactions for pollinator services also varied over time, highlighting the potential of temporal variability in plant interactions. Keywords: seed set, facilitation, competition, plant, visitor, species richness, floral density, visitation frequency, Borago officinalis, Sinapis alba, experiment

3.1 Introduction

In ecological facilitation among plants the performance or fitness of one species benefits from the presence of another, which itself is not negatively impacted (Callaway, 2007; Bronstein, 2009). Facilitation has been advanced as a mechanism by which species diversity can be maintained or even enhanced (Hacker and Gaines, 1997; Fontaine et al., 2006; Valiente-Banuet and Verdu, 2007; Gross, 2008). In the context of plant-pollinator interactions, higher plant species diversity has been shown to enhance pollinator visitation rate and seed set among the component plants (Kunin and Shmida, 1997; Johnson et al., 2003; Moeller, 2004; Ghazoul, 2006; Hegland and Boeke, 2006; Juillet et al., 2007; Pellegrino et al., 2008; Ebeling et al., 2008). Conversely, plants might compete for pollinators, or the quality of pollen transfer in species rich plots may decline owing to the deposition of non-compatible heterospecific pollen (Bell et al., 2005; Morales and Traveset, 2008; Mitchell et al., 2009), which potentially reduces species diversity (Grabas and Laverty, 1999). The balance between facilitation and competition among plant species will therefore contribute to the outcome of plant-pollinator interactions and, by inference, might potentially affect seed set of plant species within a community. Some plant species escape this balance by limiting pollinator access to flowers, through a variety of means, most notably floral morphology (Fenster et al., 2004). This reduces pollen contamination and competition by enhancing pollinator constancy, albeit at the expense of limiting the potential number of pollinating species. Most plants are, however, generalists in the range of pollinators they attract (Waser et al., 1996). Yet the variety of pollinators visiting a plant depends on the ability of the surrounding environment to sustain a species-rich pollinator community, and patch factors such as species diversity and evenness of the plant community, as well as species traits such as floral structure (Giurfa et al., 1999; Fenster et al., 2004) or colour (Gumbert et al., 1999), constrain the types of pollinators that 3.2 Materials & Methods 41

are attracted to, or are able to access, the ﬂowers. Thus for both generalist and specialist plants the relative importance of facilitation and competition may have a role in determining pollination, seed set and ultimately community diversity. By establishing experimental sites along arable ﬁelds in a relatively homogeneous landscape and using two plant species with different degrees of pollinator generalization, we sought to answer the following questions:

1. Does plant diversity or ﬂoral abundance affect pollinator diversity, visitation rate, and seed set in focal plants?

2. Does pollinator visitation rate among generalist and specialist plant species differ across a plant diversity gradient?

3.2 Materials & Methods

3.2.1 Field Sites

Eight field sites around the village of Grandcour (SUI1 561107/SUI2 191320), canton Fribourg, in Switzerland, were established in May-June 2006. The landscape consisted almost entirely of agricultural fields with different crop species (in particular cereals). Each field site was situated along a field margin and consisted of four adjacent treatments, each of 6x9m (54 m2), sown in mixtures of 2, 6, 12, and 20 native, wild plant species (Tab. 3.1). Treatments were assigned in random order to each field site (Tab. 3.2). Distances between field sites varied from 172 m to 3,270 m, with the average nearest neighbour distance being 1,784 m (Tab. 3.3).

3.2.2 Study Species

We used two experimental plant species Borago officinalis L. (Boraginaceae) and Sinapis alba L. (Brassicaceae). Both are annuals, self-incompatible, insect-pollinated, xenogamic and offer nectar as a reward (Klotz et al., 2002). Borago officinalis produces blue-coloured, down-ward facing, short-lived (< 48h) flowers and is pollinated by bees, especially bumblebees (Williams, 1998; Osborne, 1999; Pontin et al., 2006). The flowers consist of 5 petals, which are fused at the base and arranged in a star-shaped form. Self-incompatibility is assured by protandry as the 42 Chapter 3

anthers release pollen immediately after anthesis, while the stigma is initially covered by the anthers and starts to elongate over time (Osborne, 1999). The yellow flowers in Sinapis alba plants are arranged loosely in a raceme and have four, non- overlapping, free petals at right angles to each other. The male parts consist of four long and two short stamens surrounding the style and releasing the pollen immediately after anthesis. Thus, self-pollination can occur but Sinapis alba has been shown to express a progressive type of sporophytic incompatibility, turning pollen tubes from the style away within a few hours (Snie´ zko˙ and Winiarczyk, 1996). Sinapis alba attracts a wide variety of bees, butterflies and flies (Glover, 2007). The experimental plants were grown from seed in a greenhouse and subsequently transplanted into pots of 10 cm diameter and 10 cm height with Klasmann ”Containersubstrat” soil (pH 5.2, 210 mg N/l, 240 mg P/l, 270 mg K/l), and then, 8 weeks after germination, transplanted into plastic pots of 19.5 cm height and 20 cm diameter in soil ”Okohum¨ Universalerde” (pH 6.2, 280 mg N/l, 260 mg P/l, 720 mg K/l). They were then placed outside the greenhouses for acclimatisation to field conditions. Sinapis alba L. was also grown from seed. Plants were germinated on wet filter paper in petri- dishes. After two weeks plants were transplanted into 10 cm diameter and 10 cm high pots in garden soil from the company Pouget Solami No. 21 (pH 6.4, 182 mg N/l, 208 mg P/l, 234 mg K/l).

3.2.3 Field work

Between 29th May and 2nd June 2008 five individuals of each plant species were placed in a circle of 1m radius within each treatment at all sites. From 18th to 27th June (except 23rd June), we recorded the number of pollinator visits to the focal plants and to all flowering plants within the circle at the center of each plot for 1 h between 0900 and 1600. To account for possible differences in the daily activity of pollinator species and their effect on the treatment, the observation time at each plot was split into two daily periods, before and after midday, and the order of observations for the treatments within each site was chosen randomly. The observations were restricted to one field site and four treatments per day. The number of plant species and flowers on the focal plants and all other plant species occurring within the 1m circle were counted after every daily observation period. All experimental plants were marked and collected between 27th - 29th June 2008 and returned to the greenhouse, where seeds and fruits were counted between 1st to 3rd July 2008. Unfortunately, thirty-eight percent (61 out of 160) of Sinapis alba individuals were destroyed by the pollen beetle Meligethes aeneus F. (Fig. 3.1) during the first experiment, including all 3.2 Materials & Methods 43

individuals at one field site (site 10) and 50% or more at three other sites (sites 4 (50%), 11 (87.5%), 12 (50%)). The remaining individuals produced very few seeds. We assumed therefore an impact of beetle herbivory on seed set in Sinapis alba and excluded the results for this plant species from the first experiment. Therefore, four additional randomly chosen sites with each of the four treatments were observed in a second experiment. Exactly the same experimental protocol was implemented with Borago officinalis L. and Sinapis alba on 21th July 2008. Observations were conducted from 24th to 27th July as before. Plants were collected on 29th July 2008 and brought into the greenhouse, where seeds were counted between 1st - 4th August 2008.

3.2.4 Statistical Analysis

Number of flowering plants in treatments were compared with their expected number using χ2-test. Owing to different floral composition between treatments we calculated the estimated χˆ 2 statistics separately for each treatment to investigate the deviation of expected plant species number. Yates‘ correction for continuity was applied when there was one degree of freedom as suggested by Fowler et al. (1998, , p. 116). Statistical significance on the treatment and experimental level was calculated by comparing the summed estimated χˆ 2 statistics of treatments and the experiment with values from χ2 distribution tables (p. 484, Sachs and Hedderich, 2006). Plant species richness within the 1m circle were calculated using the Margalef index (Magurran, 2004), S − 1 (3.1) ln(N) where S is the number of species and N the number of individuals. We used the Menhinick index, S √ (3.2) N to express pollinator species richness, because it takes account of single pollinator visits. Treatments were observed twice daily and nested within field sites. To account for temporal pseudoreplication, all explanatory variables were averaged across the two observations for each treatment at each day. We account for possible spatial pseudoreplication due to nestedness of treatments using Generalized Least Square Models (GLS), which account possible correlation of errors among grouped observations using correlation structures (Piepho et al., 2003; Bolker et al., 2009; Zuur et al., 2009). Differences between fitness, plant diversity, floral density, visitor diversity, and visit frequency at the field level among the two plant species and the two experiments were analyzed using Wilcoxon and in occurrences of ties Wilcoxon Exact Test (Sachs and Hedderich, 2006). 44 Chapter 3

We analyzed the influence of plant and visitor diversity, and floral density and visit frequency on fitness using separate analyses for each plant species as data for Sinapis alba were only available from the second experiment. We conducted three models:

1. In the first model, we analyzed the effect of plant diversity, the logarithm of floral density, and their interactions on the logarithm of flower visits divided by the number of focal plant flowers, and pollinator diversity at the patch level.

2. In the second model, we investigated the effect of plant diversity, the logarithm of floral density, and their interactions on pollinator diversity and log-transformed visitation frequency of B. officinalis and S. alba. The analysis for Borago officinalis was conducted including experiment as an explanatory variable to account for temporal differences.

3. In the third model, we investigated the effect of plant diversity, the logarithm of ﬂoral density, pollinator diversity, the logarithm of pollinator visits, and their interactions on seed set in both studied plant species. Again, for Borago ofﬁcinalis potential differences between the two experiments were investigated by including the experiments as an explanatory variable.

In all models we log-transformed visitation frequency (expressed as visits divided by flower number of the focal plants) and floral density to remove skewness and centered (i.e., values were substracted from the mean) each variable. All models were examined for normality by investigating the distribution of the normalized residuals, and for homodascity by plotting the normalized residuals against the fitted values. Furthermore, the normalized residuals were plotted against the explanatory variables to reveal patterns indicating violation of independence (Zuur et al., 2009). In case the assumptions were violated, variance functions were applied for corrections (Pinheiro and Bates, 2004; Zuur et al., 2009), as transformations of the response variable may affect the interpretability (McArdle and Anderson, 2004). The correlation structure of each model was investigated and highly correlated variables centered. We reduced the full models using stepwise exclusion of non-significant terms to achieve the most parsimonious model. All analyses were conducted using the statistical program R (R Development Core Team, 2009) by applying the packages nlme (Pinheiro et al., 2009), AED (Zuur, 2008), MASS (Venables and Ripley, 2002), and coin (Hothorn et al., 2008). 3.3 Results 45

3.3 Results

3.3.1 Treatment characteristics

The observed number of flowering plant species did not correspond to expectations based on 2 2 treatment design (χˆ = 252.2 < 64.47 = χ 47;0.95 , p < 0.001), but the Margalef index of plant F p species richness correlated with the treatment levels ( 3,40 = 3.92, = 0.02, Fig. 3.2). We therefore used the Margalef Index for plant species richness as one explanatory variable in the models. Mean plant species richness of field sites did not differ between the first and second experiment, or among treatment levels within the experiments (interaction experiments and treatments), but between treatments (treatments from both experiments pooled together, Fig. 3.3). There was no difference in plant species richness among field sites (Fig. 3.4). Mean (log-transformed) floral density did not differ between the experiments, or among treatment levels within the experiments (interaction experiment and treatments, or among treatments (treatments from both experiments pooled together, Fig. 3.3), or among field sites (Fig. 3.4).

3.3.2 Pollinators

Pollinator-plant community relation

Pollinator visitation and its relation to plant species richness and density differed between the two experiments. In the first experiment (18th – 27th June) pollinator species richness of plants within treatments was positively correlated with plant diversity but negatively with floral density (Tab. 3.4). Pollinator visitation frequency within treatments was not correlated with plant diversity, but positively correlated with floral density (Tab. 3.4). In the second experiment (24th – 27th July) pollinator species richness was not correlated with either plant species richness or floral density (Tab. 3.4). Flower visitation frequency within treatments was positively correlated with floral density, but the correlation depended on plant species richness (i.e., interaction plant species richness and floral density). Thus, when floral density and plant species richness increased, the number of floral visits declined (Tab. 3.4). 46 Chapter 3

Pollinator community of focal plant species

A total of 112 flower visiting invertebrate species were recorded, including 12,565 individual visits, 31 visitor species from 1,451 individual visits in the two experiments for Borago officinalis (Tab. 3.5). Most flower visitors to B. officinalis were the bee species Apis mellifera, Bombus pratorum, and B. lucorum/terrestris. Bombus pratorum was the most frequent visitor of B. officinalis (77%, Tab. 3.5). Among Syrphidae, only Sphaerophoria scripta was a frequent visitor of Borago officinalis. Bombus pratorum was not recorded in the second experiment. Sinapis alba received 315 individuals visits from 23 species during the second experiment. Flower visitors were almost equally divided between several species of Syrphidae (Sphaerophoria scripta and Eristalis tenax) and Apidae (Lasioglossum species), with very small numbers of Coleoptera, Diptera, and Lepidoptera (Tab. 3.5). Among bees, an unidentified larger Lasioglossum species was a frequent visitor.

Pollinator diversity and visitation frequency of focal plant species and experiments

The average Menhinick species richness of visitors differed significantly between the two focal plant species, but it is not different when taking only the values of the second experiment for both plant species (Tab. 3.6). For B. officinalis, visitor species richness did not differ between the experiments (Tab. 3.6), but visitor species richness among field sites differed (Kruskal Wallis Hˆ = 21.71, df = 11, p = 0.03). There was no difference in S. alba among field sites (Kruskal Wallis Hˆ = 5.67, df = 3, p = 0.13). Flowers of B. officinalis were more frequently visited than flowers of S. alba, even when taking only the values for the second experiment (Tab. 3.6). There were also differences across experiments, with B. officinalis receiving almost three times as many visits per flower in the first experiment (Tab. 3.6). The difference remained significant after log-transformation of the ratio (W = 290, p = 0.04). Again there was significant variation in visits per flower (Kruskal Wallis Hˆ = 37.34, df = 11, p < 0.001) and the log-transformed ratio of flower visits per flower (Kruskal Wallis Hˆ = 35.1, p < 0.001) among field sites, while in S. alba neither visitation frequency per flowers nor its log-transformation differed (for both variables Kruskal Wallis Hˆ = 4.81, df = 3, p = 0.19). 3.3 Results 47

Inﬂuence of plant species richness and ﬂoral density on visitor diversity and visitation frequency of each focal plant

In B. officinalis, plant species richness was not correlated with pollinator species richness or pollinator visits in both experiments (Tab. 3.7). Floral density differed in the effect on visitor species richness between experiments. There was only a significant negative effect of floral density on visitor species richness in the second experiment (Tab. 3.7). In S. alba, plant species richness and floral density were not significantly influencing visitor species richness and visits (Tab. 3.7).

3.3.3 Seed set in the focal plant species

Borago ofﬁcinalis

Seed production of Borago officinalis plants was higher in the second experiment than in the first (Tab. 3.8). The effects of plant species richness, visitor species richness, and the interaction of floral density and visitation frequency also differed between the two experiments (Tab. 3.8). There were no such significant relationships in the first experiment. In the second experiment plant species richness increased seed production, while visitor species richness had a negative effect. More seeds were produced when both floral densities and floral visits increased (Tab. 3.8), while floral density or floral visits alone had no influence on seed set (Tab. 3.8).

Sinapis alba

Sinapis alba showed a significant interaction effect of visitor species richness with floral density, and with plant species richness, while the explanatory variables alone had no effect on seed set in S. alba (Tab. 3.8). Thus, disproportionately fewer seeds were produced by the combined interaction of increasing plant species richness and visitor species richness. Conversely, seed set increased disproportionately when more visitor species richness were visiting the focal plant at denser floral patches. 48 Chapter 3

3.4 Discussion

Our study indicates that species-rich vegetation attracts a more diverse visitor community, while high floral density causes a decline in visitor species richness. The first result is not unexpected as several previous studies have demonstrated that more diverse plant communities attract a wider variety of flower visitor species (Ghazoul, 2006; Hegland and Boeke, 2006; Pontin et al., 2006), presumably on account of the wider variety of resources and floral attractants (Westerkamp, 1997; Potts et al., 2003, 2004; Fontaine et al., 2006). Densely flowering patches are often associated with high resource availability and are therefore also expected to attract a more species rich flower visiting community (Sih and Baltus, 1987; Willmer, 1983; Kunin and Shmida, 1997; Totland and Matthews, 1998; Bosch and Waser, 2001; Hegland and Boeke, 2006; Ebeling et al., 2008). Indeed, floral density was associated with an increase in the number of floral visits but, contrary to expectations, not visitor species richness. A decline in visitor species richness with floral density may be explained by depletion of nectar/pollen resulting from rapid and disproportionately high visitation frequencies to dense flowering patches by social bees (Rathcke, 1983; Reader et al., 2005; Geber and Moeller, 2006). Bumblebees and honeybees, which were the most common visitors in our study sites (55.1% and 33.5%, respectively), have been shown to deplete reward amounts faster than co-occurring flower visitors, and thereby competitively exclude other flower visiting animal species from the depleted plant species (p. 164 – 166, and references therein Goulson, 2003). We have no data on competition among flower visitors or floral resource amount to support this hypothesis. The relationships for visitor species richness with plant species richness and floral density were significant only in the first experiment despite there being no difference in the plant community or in floral density between the two experiments. Even so, the visitor community was significantly richer in species in the second experiment, while more flower visits were recorded in the first experiment. Such patterns therefore appear to be temporally variable emphasizing the need to attach caution to studies undertaken within a single limited time period.

3.4.1 Visitor community richness and visitation frequency and the surrounding plant community

The degree of generalization of the flower visitor community can explain the difference in visitation frequencies between different plant species. Plant species with a more specialized flower visitor community might be less influenced by the surrounding flowering plant community than generalist species, because specialized pollinators are expected to switch less between plant 3.4 Discussion 49

species (Lazaro´ et al., 2009). Moving between plant species requires learning of skills to exploit different floral forms. Initial bouts on different flowers would therefore result in poor resource acquisition due to poor exploitation abilities. In our study, neither the generalist plant species or the relative specialists showed a clear response, in terms of visitor species richness, to increasing plant species richness. The lack of any relationship with plant species richness, and with floral density, in the generalist plant species can be explained by the role of field margins as habitats for foraging pollinators in extensively used agricultural landscapes (Thomas and Marshall, 1999; Sutherland et al., 2001; Marshall and Moonen, 2002; Blomqvist et al., 2003; Kohler et al., 2008). Pollinators were probably foraging indiscriminately between treatments within plots, thereby neutralizing any treatment effect. Increasing floral density of heterospecifics reduced visitor species richness in B. officinalis, but only in the second experiment. Although there was no significant difference in floral density or pollinator species richness between the experiments, the most abundant flower visitor to this focal plant, Bombus pratorum, did not occur in the second experiment (Tab. 3.5). Specific pollinator species are less affected by surrounding plant species richness (Lazaro´ et al., 2009). The absence of the most specific and most abundant flower visitor make it therefore impossible to identify all interactions among plant species.

3.4.2 Seed set in the focal plants

Borago ofﬁcinalis

In our study, fitness in B. officinalis increased with plant species richness, indicating benefits secured through facilitation which are sufficient to offset any costs associated with heterospecific pollen transfer (e.g., Bell et al., 2005; Flanagan et al., 2009). Plant species richness did not, however, affect B. officinalis visitor species richness, but heterospecific floral density reduced visitor species richness (Tab. 3.5). B. officinalis fitness was not, however, significantly affected by heterospecific floral density, and significantly declined with visitor species richness. As generalist pollinator species reduce seed set by transfer of heterospecific pollen (Larsson, 2005), we suggest that generalist flower visitor species may be attracted by heterospecific floral density. Enhanced seed set in B. officinalis is therefore attributed to the specific pollinator species, which provide a more pure pollen load. The positive effect of heterospecific floral density and floral visits on seed set can be attributed to the same process. An alternative explanation is that plant species richness might increase plant seed set if visits to flowers on the same individual plant are reduced, which would increase outcrossing and seed set 50 Chapter 3

by reducing geitonogamy (de Jong et al., 1993). Data on visitor movement on single B. officinalis individuals would be necessary to confirm this. Finally, switching between plant species may demand the learning of new flower handling skills, which is why bees may increase selectivity with reward amount (Gegear and Laverty, 2001). Thus, with increasing floral density bees may have tended to visit B. officinalis more frequently (i.e., positive interaction between floral density and floral visitation frequency on fitness (Tab. 3.8)). Interestingly, there was only a very weak positive effect of visits in the first experiment. Frequent visits enhance fitness in terms of seed set in plants (Engel and Irwin, 2003), in particular for specialist flower visitors (Larsson, 2005; Perfectti et al., 2009). The lack of visits by the most abundant and most specific B. offcinalis flower visitor B. pratorum in the second experiment can thus explain the lack of relations between fitness and the predictor variables (Franzen and Larsson, 2009; Perfectti et al., 2009).

Sinapis alba

The effect of flower visitor species richness on seed set in S. alba depends on species richness of co-occurring plants and their floral density. Seed set in S. alba declines with increasing visitor species richness when plant species richness is high, suggesting that deposition of heterospecific pollen impacts seed set. S. alba had significantly greater species-rich visitor communities than B. officinalis, which increases the probability of heterospecific pollen transfer (Lazaro´ et al., 2009). Thus, plant species with generalist visitor communities can be expected to suffer reduced seed set with increasing plant species richness due to pollen contamination (Flanagan et al., 2009). In contrast, visitor species richness and heterospecific floral density increased fitness in S. alba, but there was no significant correlation between the predictor variables (Tab. 3.7). Fifty percent of the visits to S. alba were by flies, which are less effective pollinators than bees (Kandori, 2002). The remaining visits were almost entirely accounted for by a single Lasioglossum bee species. Species-rich flower visitor communities can be beneficial for reproduction in plants, when flower visitors differ in their effectiveness (Perfectti et al., 2009). Moreover, specific flower visitors provide for purer pollen loads, which reduce probability of fertilization and seed set (Young and Young, 1992). Plant species can therefore benefit from heterospecific flowers by increased seed set. The effects of heterospecific pollen transfer can be neutralized by less efficient flower visitors and pollen transfer. 3.5 Acknowledgement 51

3.4.3 Conclusion

We have shown that processes may act differently upon seed set on two plant species with different flower visitor community compositions. We found positive effects of plant species richness and floral density on seed set in the plant species with a specialist visitor community, while the plant species with a generalist visitor community competed with co-occurring plant species. However, there was no evidence of direct effects of plant species richness and floral density on seed set in terms of the effects on visitor communities. We attributed the lack of clear relationships to the small variability in plant species richness between treatments and to the temporal variability in visitation frequency and visitor community composition. This highlights the need to consider the effect of temporal variation on plant-plant interactions and pollinator community when designing field experiments to find answers on pollination ecology.

3.5 Acknowledgement

We would like to thank Prof. Louis Bersier from the University of Fribourg, Switzerland, for his cooperation, Eveline Thevoz for ﬁeld assistance, and Dr. Chris Kettle from ETH Zurich¨ for reviewing this paper.

3.6 References

Baddeley, A. and Turner, R. (2005). Spatstat: an R package for analyzing spatial point patterns. Journal of Statistical Software, 12(6):1–42. ISSN 1548-7660.

Bell, J. M., Karron, J. D., and Mitchell, R. J. (2005). Interspeciﬁc competition for pollination lowers seed production and outcrossing in Mimulus ringens. Ecology, 86(3):762–771.

Blomqvist, M. M., Vos, P., Klinkhamer, P. G. L., and ter Keurs, W. J. (2003). Declining plant species richness of grassland ditch banks - a problem of colonisation or extinction? Biological Conservation, 109(3):391–406.

Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., and White, J.-S. S. (2009). Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution, 24(3):127–135. 52 Chapter 3

Bosch, M. and Waser, N. M. (2001). Experimental manipulation of plant density and its effect on pollination and reproduction of two confamilial montane herbs. Oecologia, 126(1):76–83.

Bronstein, J. L. (2009). The evolution of facilitation and mutualism. Journal of Ecology, 97(6):1160–1170.

Callaway, R. M. (2007). Positive interactions and interdependence in plant communities. Springer. de Jong, T. J., Waser, N. M., and Klinkhamer, P. G. L. (1993). Geitonogamy - The neglected side of selﬁng. Trends in Ecology & Evolution, 8(9):321–325.

Engel, E. C. and Irwin, R. E. (2003). Linking pollinator visitation rate and pollen receipt. American Journal of Botany, 90(11):1612–1618.

Fenster, C., Armbruster, W., Wilson, P., Dudash, M., and Thomson, J. (2004). Pollination syndromes and ﬂoral specialization. Annual Review of Ecology, Evolution and Systematics, 35:375–403.

Fontaine, C., Dajoz, I., Meriguet, J., and Loreau, M. (2006). Functional diversity of plant- pollinator interaction webs enhances the persistence of plant communities. PLOS Biology, 4(1):129–135.

Fowler, J., Cohen, L., and Jarvis, P. (1998). Practical Statistics for Field Biology. John Wiley & Sons, 2nd edition.

Franzen, M. and Larsson, M. (2009). Seed set differs in relation to pollen and nectar foraging ﬂower visitors in an insect-pollinated herb. Nordic Journal of Botany, 27(4):274–283.

Gegear, R. J. and Laverty, T. M. (2001). The effect of variation among ﬂoral traits on the ﬂower constancy of pollinators. In Chittka, L. and Thomson, J. D., editors, Cognitive Ecology of Pollination, chapter 1, pages 1–20. Cambridge University Press.

Ghazoul, J. (2006). Floral diversity and the facilitation of pollination. Journal of Ecology, 94:295–304.

Giurfa, M., Dafni, A., and Neal, P. R. (1999). Floral symmetry and its role in plant-pollinator systems. International Journal of Plant Sciences, 160(6):S41–S50.

Glover, B. (2007). Understanding Flower & Flowering. An Integrated Approach. Oxford University Press.

Goulson, D. (2003). Effects of introduced bees on native ecosystems. Annual Review of Ecology, Evolution and Systematics, 34:1–26.

Grabas, G. P. and Laverty, T. M. (1999). The effect of purple loosestrife (Lythrum salicaria L, Lythraceae) on the pollination and reproductive success of sympatric co-ﬂowering wetland plants. Ecoscience, 6(2):230–242.

Gross, K. (2008). Positive interactions among competitors can produce species-rich communities. Ecology Letters, 11(9):929–936.

Hacker, S. D. and Gaines, S. D. (1997). Some implications of direct positive interactions for community species diversity. Ecology, 78(7):1990–2003.

Hegland, S. J. and Boeke, L. (2006). Relationships between the density and diversity of ﬂoral resources and ﬂower visitor activity in a temperate grassland community. Ecological Entomology, 31(5):532–538.

Hothorn, T., Hornik, K., van de Wiel, M. A., and Zeileis, A. (2008). Implementing a class of permutation tests: The coin Package. Journal of Statistical Software, 28(8):1–23.

Kandori, I. (2002). Diverse visitors with various pollinator importance andtemporal change in the important pollinators of Geranium thunbergii (Geraniaceae). Ecological Research, 17(3):283– 294.

Kohler, F., Verhulst, J., van Klink, R., and Kleijn, D. (2008). At what spatial scale do high-quality habitats enhance the diversity of forbs and pollinators in intensively farmed landscapes? Journal of Applied Ecology, 45(3):753–762.

Kunin, W. and Shmida, A. (1997). Plant reproductive traits as a function of local, regional, and global abundance. Conservation Biology, 11(1):183–192.

Larsson, M. (2005). Higher pollinator effectiveness by specialist than generalist ﬂower-visitors of unspecialized Knautia arvensis (Dipsacaceae). Oecologica, 146(3):394–403.

Lazaro,´ A., Lundgren, R., and Totland, O. (2009). Co-ﬂowering neighbors inﬂuence the diversity and identity of pollinator groups visiting plant species. Oikos, 118(5):691–702.

Magurran, A. E. (2004). Measuring Biological Diversity. Blackwell Publishing.

Marshall, E. J. R. and Moonen, A. C. (2002). Field margins in northern Europe: their functions and interactions with agriculture. Agriculture Ecosystems & Environment, 89(1-2):5–21.

McArdle, B. H. and Anderson, M. J. (2004). Variance heterogeneity, transformations, and models of species abundance: a cautionary tale. Canadian Journal of Fisheries and Aquatic Sciences, 61(7):1294–1302.

Mitchell, R. J., Irwin, R. E., Flanagan, R. J., and Karron, J. D. (2009). Ecology and evolution of plant-pollinator interactions. Annals of Botany, 103(9):1355–1363.

Moeller, D. A. (2004). Facilitative interactions among plants via shared pollinators. Ecology, 85(12):3289–3301.

Morales, C. L. and Traveset, A. (2008). Interspeciﬁc pollen transfer: Magnitude, prevalence and consequences for plant ﬁtness. Critical Reviews in Plant Sciences, 27(4):221–238. 3.6 References 55

Osborne, J. L. (1999). Borage. Bee World, 80(1):33–36.

Pellegrino, G., Bellusci, F., and Musacchio, A. (2008). Double ﬂoral mimicry and the magnet species effect in dimorphic co-ﬂowering species, the deceptive orchid Dactylorhiza sambucina and rewarding Viola aethnensis. Preslia, 80(4):411–422.

Perfectti, F., Gomez, J. M., and Bosch, J. (2009). The functional consequences of diversity in plant-pollinator interactions. Oikos, 118(9):1430–1440.

Piepho, H. P., Buchse, A., and Emrich, K. (2003). A hitchhiker’s guide to mixed models for randomized experiments. Journal of Agronomy and Crop Science, 189(5):310–322.

Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., and the R Core team (2009). nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1-92.

Pinheiro, J. C. and Bates, D. M. (2004). Mixed-Effects Models in S and S-PLUS. Statistics and Computing. Springer.

Potts, S. G., Vulliamy, B., Dafni, A., Ne’eman, G., and Willmer, P. (2003). Linking bees and ﬂowers: How do ﬂoral communities structure pollinator communities? Ecology, 84(10):2628– 2642.

R Development Core Team (2009). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0.

Rathcke, B. (1983). Competition and facilitation among plants for pollination. In Real, L., editor, Pollination Biology, pages 305–329. Academic Press.

Sachs, L. and Hedderich, J. (2006). Angewandte Statistik. Methodensammlung mit R. Springer, 12 edition. 56 Chapter 3

Sih, A. and Baltus, M. S. (1987). Patch size, pollinator behavior, and pollinator limitation in Catnip. Ecology, 68(6):1679–1690.

Snie´ zko,˙ R. and Winiarczyk, K. (1996). Pollen tube incompatibility reaction on the stigma in self-pollinated Sinapis alba L. Acta Societatis Botanicorum Poloniae, 65(1-2):101–105.

Sutherland, J. P., Sullivan, M. S., and Poppy, G. M. (2001). Distribution and abundance of aphidophagous hoverflies (Diptera: Syrphidae) in wildflower patches and field margin habitats. Agricultural and Forest Entomology, 3(1):57–64.

Thomas, C. F. G. and Marshall, E. J. P. (1999). Arthropod abundance and diversity in differently vegetated margins of arable ﬁelds. Agriculture Ecosystems & Environment, 72(2):131–144.

Totland, O. and Matthews, I. (1998). Determinants of pollinator activity and ﬂower preference in the early spring blooming Crocus vernus. Acta Oecologica-International Journal of Ecology, 19(2):155–165.

Valiente-Banuet, A. and Verdu, M. (2007). Facilitation can increase the phylogenetic diversity of plant communities. Ecology Letters, 10(11):1029–1036.

Venables, W. N. and Ripley, B. D. (2002). Modern Applied Statistics with S. Springer, New York, 4th edition. ISBN 0-387-95457-0.

Waser, N. M., Chittka, L., Price, M. V., Williams, N. M., and Ollerton, J. (1996). Generalization in pollination systems, and why it matters. Ecology, 77(4):1043–1060.

Westerkamp, C. (1997). Bees and ﬂowers are competitors, not partners. Mitteilungen der Deutschen Gesellschaft fur¨ Allgemeine und Angewandte Entomologie, 11(1-6):781–784.

Williams, C. S. (1998). The identity of the previous visitor inﬂuences ﬂower rejection by nectar-collecting bees. Animal Behaviour, 56:673–681.

Willmer, P. G. (1983). Thermal constraints on activity patterns in nectar-feeding insects. Ecolog- ical Entomology, 8(4):455–469.

Young, H. J. and Young, T. P. (1992). Alternative outcomes of natural and experimental high pollen loads. Ecology, 73(2):639–647.

Zuur, A. (2008). AED: Data sets and routines for book. R package version 1.0.

Zuur, A. F., Ieno, E. N., Walker, N. J., Saveliev, A. A., and Smith, G. M. (2009). Mixed Effects Models and Extensions in Ecology with R. Springer Verlag. 3.7 Tables and Figures 57

3.7 Tables and Figures

Table 3.1: The experimental design at Grandcour. The number in the cells present the ﬁeld sites.

Plant Species Family Treatment

2 6 12 20

Achillea millefolium Asteraceae 2 3, 6, 10 1, 5, 7, 11, 12 all Agrostemma githago Caryophyllaceae 7 2, 11, 12 1, 3, 5, 6, 10 all Anthemis tinctoria Asteraceae 1, 5, 11, 12 2, 3, 6, 7, 10 all Centaurea cyanus Asteraceae 6 3, 11, 12 1, 2, 5, 7, 10 all Centaurea jacea Asteraceae 1 5, 10, 12 2, 3, 6, 7, 11 all Cichorium intybus Asteraceae 1 5, 6 2, 3, 7, 10, 11, 12 all Daucus carota Apiaceae 3, 11 2, 7 1, 5, 6, 10, 12 all Dipsacus fullonum Dipsacaceae 2 12 1, 3, 5, 6, 7, 10, 11 all Echium vulgare Boraginaceae 5 7, 11 1, 2, 3, 6, 10, 12 all Hypericum perforatum Hypericaceae 2, 3, 10 1, 5, 6, 7, 10, 11, 12 all Leucanthemum vulgare Asteraceae 5 1, 3, 10 2, 6, 7, 11, 12 all Malva moschata Malvaceae 11, 12 1, 7 2, 3, 5, 6, 10 all Malva sylvestris Malvaceae 7 3, 6, 11 1, 2, 5, 10, 12 all Origanum vulgare Lamiaceae 1, 5, 6 2, 3, 7, 10, 11, 12 all Papaver rhoeas Papaveraceae 3, 7, 10, 11 1, 2, 5, 6, 12 all Pastinaca sativa Apiaceae 12 2, 5, 10 1, 3, 6, 7, 11 all Silene pratensis Caryophyllaceae 6 2, 5, 7 1, 3, 10, 12 all Tanacetum vulgare Asteraceae 10 1, 6, 12 2, 3, 5, 7, 11 all Verbascum lychnitis Scrophulariaceae 3 7 1, 2, 5, 6, 10, 11, 12 all Verbascum thapsus Scrophulariaceae 10 1, 2, 6 3, 5, 7, 11, 12 all

Table 3.2: Plant species and treatment order in the experimental set up at Grandcour. The number represent the ﬁeld number.

Field Site Treatment Order

1 6–12–2–20 2 20–2–12–6 3 2–6–20–12 5 2–12–6–20 6 6–2–20–12 7 12–20–6–2 10 6–2–20–12 11 20–6–12–2 12 12–6–20–2 58 Chapter 3

Table 3.3: The differences among ﬁeld sites in metre obtained by the command ”pairdist()” using the package spatstat 1.9-6 (Baddeley and Turner, 2005) in R 2.7.2 (R Development Core Team, 2009). The Swiss coordinates were obtained from the web site: www.ecogis.admin.ch.

Field no. SUI1 SUI2 1 2 3 5 6 7 10 11 12 1 560022 191268 0 608.4 1310.4 1942.2 1896.8 1809.1 2373.5 2089.1 2640.4 2 559536 190902 608.4 0 1345.2 2225.7 2223.6 2297.3 2943.5 2678 3245.5 3 560616 190100 1310.4 1345.2 0 1042.0 1116.2 1573.3 2442.5 2337.5 3267.2 5 561652 190212 1942.2 2225.7 1042.0 0 172 912.4 1766.8 1816.7 2891.7 6 561697 190378 1896.8 2223.6 1116.2 172 0 741 1596.3 1645.4 2722.3 7 561824 191108 1809.1 2297.3 1573.3 912.4 741 0 891.4 904.4 1985 10 562324 191846 2373.5 2943.5 2442.5 1766.8 1596.3 891.4 0 377.4 1304 11 561979 191999 2089.1 2678 2337.5 1816.7 1645.4 904.4 377.4 0 1092 12 561933 193090 2640.4 3245.5 3267.2 2891.7 2722.3 1985 1304 1092 0 Mean 1630 1951.9 1603.8 1418.8 1345.9 1234.9 1521.7 1437.8 2127.6 3.7 Tables and Figures 59

Table 3.4: Results from the generalized least square model after stepwise exclusion of non-signiﬁcant terms for visitor species richness and the logarithmized values of visitation frequency at patch level with pooled data (both experiments) and non-pooled data. Predictors were centered.

Visitor Species Richness log(Visitation Frequency)

Value Std.Error t-value p-value Value Std.Error t-value p-value

Pooled data

(Intercept) 1.11 0.09 11.89 <0.00 4.64 0.07 66.00 <0.00 plant spec. rich. 0.28 0.10 2.78 0.01 − − − − ﬂoral dens. −0.11 0.04 −2.64 0.01 0.46 0.06 7.77 <0.00

Non-pooled data

(Intercept) 1.00 0.08 12.91 <0.001 4.73 0.06 82.16 <0.001 1. experiment 0.64 0.17 3.71 <0.001 −0.27 0.09 −2.94 0.01 1. experiment : plant spec. rich. 0.22 0.11 2.00 0.05 −0.01 0.15 −0.04 0.97 1. experiment : floral dens. −0.11 0.04 −2.48 0.02 0.47 0.08 5.77 <0.001 1. experiment : plant spec. rich. : floral dens. –––– 0.15 0.16 0.95 0.35 (Intercept) 1.64 0.15 10.7 <0.001 4.46 0.07 62.79 <0.001 2. experiment −0.64 0.17 −3.71 <0.001 0.27 0.09 2.94 0.01 2. experiment : plant spec. rich. 0.37 0.26 1.40 0.17 −0.19 0.23 −0.82 0.41 2. experiment : floral dens. 0.15 0.19 0.8 0.43 0.44 0.14 3.09 0.004 2. experiment : plant spec. rich. : floral dens. –––– −0.94 0.24 −3.86 <0.001 60 Chapter 3

Table 3.5: Visits and percentages of visits to the two focal plants Borago ofﬁcinalis and Sinapis alba. The percentage visits of each species is related to the visitation frequency of the next higher taxonomic level.

Visits percentages are referred to total visits. ∆BO and ∆SA are the percentages of plant to treatment visits. BO1 and BO2 are flower visitor records of B. officinalis from the first and second experiment, respectively.

Visitor Species ALL Borago ofﬁcinalis ∆BO Sinapis alba ∆SA BO1 BO2 Hymenoptera 10,673; 84.9% 1,412; 97.3% 13.2% 156; 49.5% 1.5% 1,123;98% 289;94.8% Apis mellifera 4,782; 44.9% 486; 34.4% 10.2% –– 326; 28.4% 160; 52.5% Bombus pratorum 565; 5.3% 435; 30.8% 77% –– 435; 38% – Bombus lucorum/terrestris 1,527; 14.3% 186; 13.2% 12.2% –– 153; 13.4% 33; 10.8% Lasioglossum spec. 1,971; 18.5% 101; 7.2% 5.1% 115; 73.7% 5.8% 65; 5.7% 36; 11.8%

Syrphidae 1,105; 8.8% 33; 2.3% 3% 152; 48.3% 13.8% 20;1.8% 13;4.3% Episyrphus balteatus 33; 3.0% 4; 12.1% 12.1% 11; 7.2% 33.3% 3;0.3% 1;0.3% Eristalis tenax 179; 16.2 % –– 50; 15.9% 27% – – Platycheirus cf. albimanus 15; 1.4% 6; 18.2% 40% 8; 2.5% 53.3% 2; 0.2% 4; 1.3% Sphaerophoria scripta 517; 46.8 % 14; 42.4% 2.7% 54; 35.5% 10.4% 9; 0.8% 5; 1.6%

Coleoptera 43; 0.4% 2; 0.1% 4.7% 2; 0.6% 4.7% 2; 0.2% –

Diptera 708; 5. 6% 2; 0.1% 0.3% 4; 1.3% 0.6% 1; 0.1% 1; 0.3%

Lepidoptera 28; 0.2% 2; 0.1% 7.1% 1; 0.3% 3.6% – 2; 0.7%

Hemiptera 8; 0.1% –– –– – –

TOTAL 12,565 1,451 11.5% 315 2.5% 1,146 305 3.7 Tables and Figures 61

Table 3.6: Mean and standard errors of the explanatory used in the models for both plant species and for the ﬁrst and second experimental period for B. ofﬁcinalis.

Borago ofﬁcinalis n Sinapis alba n W p

vis. spp. rich. (pooled) 0.94 ± 0.04 45 1.18 ± 0.07 16 191.5 0.005 vis. spp. rich. (2nd exp.) 1.06 ± 0.09 13 1.18 ± 0.07 16 83.5 0.38 vis. visits (pooled) 5.69 ± 0.86 48 0.71 ± 0.1 16 687 < 0.001 vis. visits (2nd exp.) 2.77 ± 0.69 16 1.18 ± 0.07 16 187 0.03

B. ofﬁcinalis

1. experiment n 2. experiment n W p

vis. spp. rich. 0.98 ± 0.04 32 1.06 ± 0.09 13 144 0.11 vis. visits 7.15 ± 1.17 32 2.77 ± 0.69 16 386 0.004 62 Chapter 3

Table 3.7: Results from the generalized least square models for visitor diversity and logarithmized ﬂower visitation frequency in the two studied plant species Borago ofﬁcinalis and Sinapis alba. All predictors were centered.

Visitor Species Richness log(Visitation Frequency)

Borago ofﬁcinalis – pooled data

Value Std.Error t-value p-value Value Std.Error t-value p-value

(Intercept) 0.92 0.05 18.15 0.00 1.44 0.24 5.94 0.00 plant spec. rich. −0.00 0.06 −0.01 0.99 −0.16 0.18 −0.89 0.38 ﬂoral dens. 0.00 0.03 0.09 0.93 −0.11 0.07 −1.52 0.14

Borago ofﬁcinalis – non-pooled data

Value Std.Error t-value p-value Value Std.Error t-value p-value

(Intercept) 0.89 0.05 16.78 <0.001 1.61 0.27 5.93 <0.001 1. experiment −0.03 0.12 −0.22 0.83 −0.64 0.50 −1.3 0.20 1. experiment : plant spec. rich. −0.02 0.07 −0.28 0.78 −0.22 0.20 −1.07 0.29 1. experiment : floral dens. −0.02 0.03 −0.59 0.56 −0.13 0.08 −1.65 0.11 1. experiment : plant spec. rich. : floral dens. –––– –––– (Intercept) 0.86 0.11 7.98 <0.001 0.95 0.41 2.3 0.03 2. experiment 0.03 0.12 0.22 0.83 0.64 0.49 1.3 0.20 2. experiment : plant spec. rich. 0.22 0.2 1.14 0.26 0.60 0.51 1.19 0.24 2. experiment : floral dens. −0.45 0.14 −3.17 0.003 0.4 0.36 1.12 0.27 2. experiment : plant spec. rich. : floral dens. –––– ––––

Sinapis alba

Value Std.Error t-value p-value Value Std.Error t-value p-value

(Intercept) 1.18 0.11 10.28 0.00 −0.51 0.18 −2.79 0.02 plant spec. rich. −0.08 0.17 −0.51 0.62 0.11 0.46 0.25 0.81 ﬂoral dens. 0.03 0.11 0.23 0.82 −0.13 0.30 −0.44 0.67 plant spec. rich. : ﬂoral dens. −0.06 0.22 −0.30 0.77 −0.19 0.56 −0.34 0.74 3.7 Tables and Figures 63

Table 3.8: Results of the generalized least square model for seed set in Borago ofﬁcinalis and Sinapis alba. For Borago ofﬁcinalis results from the models with pooled data and for each of the two experiments at Grandcour (non-pooled data) are shown. All predictors were centered.

Value Std.Error t-value p-value

Borago ofﬁcinalis – pooled data

(Intercept) 1.73 0.07 24.13 0.00 plant. spec. rich. 0.09 0.14 0.63 0.53 ﬂoral dens. −0.01 0.06 −0.23 0.82 vis. spec. rich. 0.31 0.32 0.99 0.32 vis. visits −0.01 0.09 −0.09 0.93 plant spec. rich.: vis. spec. rich. −0.21 0.59 −0.36 0.72 plant spec. rich. : vis. visits 0.01 0.15 0.05 0.96 vis. visits : vis. spec. rich. −0.05 0.29 −0.18 0.85 vis. visits : ﬂoral dens. 0.02 0.08 0.22 0.83

Borago ofﬁcinalis – non-pooled data

(Intercept) 1.54 0.06 26.25 <0.001 1. experiment 0.57 0.14 3.98 <0.001 1. experiment : plant spec. rich. 0.06 0.10 0.57 0.57 1. experiment : floral dens. 0.07 0.04 1.66 0.10 1. experiment : vis. spec. rich. 0.22 0.25 0.88 0.38 1. experiment : vis. visits 0.13 0.07 1.80 0.07 1. experiment : plant spec. rich. : vis. spec. rich. −0.67 0.46 −1.46 0.15 1. experiment : floral dens. : vis. visits −0.04 0.05 −0.72 0.47 (Intercept) 2.11 0.13 15.98 <0.001 2. experiment −0.57 0.14 −3.98 <0.001 2. experiment : plant spec. rich. 0.78 0.35 2.27 0.02 2. experiment : floral dens. −0.33 0.25 −1.35 0.18 2. experiment : vis. spec. rich. −1.4 0.46 −3.02 0.003 2. experiment : vis. visits 0.01 0.14 0.10 0.92 2. experiment : plant spec. rich. : vis. spec. rich. −0.67 0.46 −1.46 0.15 2. experiment : floral dens. : vis. visits 0.64 0.31 2.1 0.04

Sinapis alba

(Intercept) 3.06 0.07 43.86 <0.001 plant spec. rich. 0.34 0.19 1.72 0.09 floral dens. −0.12 0.11 −1.10 0.28 vis. spec. rich. 0.33 0.26 1.29 0.20 vis. visits 0.01 0.13 0.05 0.96 plant spec. rich. : vis. spec. rich. −1.66 0.76 −2.20 0.03 plant spec. rich. : vis. visits −1.80 0.93 −1.94 0.06 floral dens. : vis. spec. rich. 1.55 0.64 2.43 0.02 floral dens. : vis. visits 0.47 0.37 1.29 0.20 64 Chapter 3

Figure 3.1: Sinapis alba L. being destroyed by beetles in June 2008 before the start of the ﬁrst experiment in the experimental ﬁelds at Grandcour, canton Fribourg.

2.0

1.5

1.0

Margalef Index of Plant Diversity Margalef Index 0.5

2 6 12 20 Treatments

Figure 3.2: The difference in plant diversity expressed as Margalef Index among the treatments. 3.7 Tables and Figures 65

Margalef Index of Plant Spp. Richness Log (Floral Density)

F1,40 = 1.35, p = 0.25 8 2.0

1.5 6

5 1.0

0.5 3 F1,40 = 1.76, p = 0.19

I II I II

F = 6.35, p = 0.001 ● 3,40 8 F3,40 = 0.41, p=0.74 2.0

● 7

1.5 6

5 1.0

0.5 ● 3

2 6 12 20 2 6 12 20

F3,40 = 0.91, p = 0.44 8 F3,40 = 1.08, p = 0.37 2.0

1.5 ● 6

5 1.0

0.5 ● 3

I.2 I.6 I.12 I.20 II.2 II.6 II.12 II.20 I.2 I.6 I.12 I.20 II.2 II.6 II.12 II.20

Figure 3.3: Margalef plant species richness and log-transformed ﬂoral density between experiments (I, II), treatments (2, 6, 12, 20), and treatments within each experiment. 66 Chapter 3

Margalef Index of Plant Spp. Richness Log (Floral Density)

^ H = 9.56, df = 11, p = 0.57 8 2.0

1.5 6

5 1.0

0.5 3 ^ H = 10.53, df = 11, p = 0.48

A 1 A 2 A 5 A 6 A 7 A 10 A 11 A 12 B 2 B 3 B 6 B 11 A 1 A 2 A 5 A 6 A 7 A 10 A 11 A 12 B 2 B 3 B 6 B 11

Figure 3.4: Differences between ﬁelds in plant species richness (Margalef Index) and ﬂoral density (log-transformed). Chapter 4

Plant-Pollinator Interactions within the Urban Environment

Hennig, Ernest Ireneusz & Ghazoul, Jaboury

ETH Zurich,¨ Ecosystem Management, Institute of Terrestrial Ecosystems, Universitatstrasse¨ 16, 8092 Zurich,¨ Switzerland

Abstract Urban ecosystems pose different environmental constraints on plant and animal communities than natural ecosystems, and this in turn might affect ecological interactions. We investigated plant-pollinator interactions in urban vegetation communities in the context of local community structure and two landscape metrics describing the surrounding urban matrix. We recorded plant species diversity, floral density, pollinator species and floral visits at 89 flowering patches within the urban matrix, and used Trifolium pratense as a focal plant species. We correlated visits by all species, by bees, by bees when excluding the main visitor Bombus pascuorum, and by the main visitor alone Trifolium pratense with the landscape metrics ”green area”, representing the total extent of vegetated areas, and ”edge density of green areas”, representing the degree of habitat fragmentation, at scales from 20 to 200 meters. Extent of green area was positively correlated with visits by all species, by bees, and by Bombus pascuorum, but negatively with bees excluding the main visitor. The correlation increased with scale for all four variables, and the strongest correlations were found for B. pascuorum. The direction of correlation between edge density and the visitation variables was similar as for green area, but showed considerable variation at scales below 100 m radius. Published in ”Perspectives in Plant Ecology, Evolution and Systematics” (2011) 68 Chapter 4

Visits by all bees and all visitors to Trifolium pratense were negatively affected by flowering plant species diversity. There was a positive interaction between extent of green area and plant diversity on bee visitation, and a negative interaction between the extent of green area and heterospecific flowers for bees (excluding B. pascuorum). Increasing floral abundance and plant diversity reduced pollinator visitation to T. pratense, suggesting competitive effects through both quantity and diversity of resources. Competition for flower visitors was mainly from Lotus corniculatus and Trifolium repens, both of which frequently co-occurred with T. pratense. We conclude that T. pratense competes for pollinators with other co-occurring plants, but that the nature of the surrounding urban matrix mediates interactions among plant species for pollinator services at different scales, particularly with respect to the proportion of green areas. Keywords: Trifolium pratense, Bombus pascuorum, flower visitors, plant-plant interactions, local community effects, landscape scale effects

4.1 Introduction

In recent years the field of urban ecology was growing substantially (Alberti, 2008; Marzluff et al., 2008; Mayer, 2010). The European urban population is expected to increase in the next 40 years in Europe (United Nations, 2008), which is presumed to have impact on species diversity, ecosystem function and biogeochemical cycles (Alberti, 2005; Marzluff et al., 2008). In the scope of the debate about declining biodiversity, urban areas are assumed to contain the potential to promote diversity (Zerbe et al., 2003; Miller, 2005). In particular, fragments of semi-natural habitats represented within the urban environment by parks, gardens and other green patches, can maintain diverse plant compositions and fulfill several ecosystem services for the urban population (Townsend, 2008, see p. 7-10). Urban areas are considered as the most pervasively changed landscapes. Plant communities within urban areas occur in small fragmentated and more or less isolated patches (Alberti, 2005). The persistence of species within such patches depends, at least in part, on the effective functioning of pollination for plant reproduction, a process that has been shown to be vulnerable to population fragmentation in a variety of plant species (Kearns et al., 1998; Ghazoul, 2005; Aguilar et al., 2006; Pauw, 2007). Smaller plant patches tend to receive fewer pollinator visits and suffer pollen limitation and reduced genetic exchange (Barrett and Kohn, 1991; Kwak et al., 1998; Ashman et al., 2004). These outcomes may change community composition by favouring selfed or abiotically-pollinated plants at the expense of outcrossed plants (Williams et al., 2005; Lososova´ et al., 2006; Lindborg, 2007), or plants that are pollinated by a wide suite of generalist pollinators (Memmott et al., 2004). This in turn could ultimately reduce flowering plant diversity and the 4.2 Materials & Methods 69

diversity of plant functional types. Additionally, small patches with low plant diversity may support fewer pollinator species, which may reinforce pollen limitation by reducing pollination services. Thus, species poor and generalized (homogenized) flowering plant communities may be expected within urban habitats (Rathcke and Jules, 1993; Kitahara and Fujii, 1994; Marvier et al., 2004; Van der Veken et al., 2004). Currently, however, such expectations are little more than conjecture and the few published studies show equivocal support to disturbance impacts on pollinators, (Morgan, 1999; Steffan-Dewenter et al., 2001; Donaldson et al., 2002; Quesada et al., 2004; Ramos and Santos, 2006; Diekotter¨ et al., 2007), and these studies do not address the urban system. Certainly, plant community composition is also determined by other factors such as urban land cover, soil conditions and microclimate (Godefroid and Koedam, 2007). The pattern of land cover might influence plant- pollinator interactions through the abundance and connectivity of green patches. Low building density might allow space for green ’corridors’ that facilitate movement of pollinators, and hence gene flow, between plant patches (Gilbert et al., 1998; Kwak et al., 1998; Tewksbury et al., 2002; Townsend and Levey, 2005). Studies conducted in urban environments have generally addressed impacts on plant and insect diversity (Zanette et al., 2005; Godefroid and Koedam, 2007; Kearns and Oliveras, 2009), but rarely considered ecological processes such as pollination. Recently there is growing consideration on plant-pollinator interactions in urban environments (Cheptou and Avendano,˜ 2006; Van Rossum and Triest, 2010). This study therefore seeks to contribute to the field of urban ecology by investigating the influence of urban heterogeneity on plant-pollination interactions. In particular, we examined the urban plant and pollinator community to determine:

a.) how two landscape metrics, edge density and the extent of green area, inﬂuence visits to a focal plant species in the urban environment

b.) whether and how patch plant diversity and ﬂoral density affect the frequency of pollinator visits to a focal plant species in the urban environment.

4.2 Materials & Methods

4.2.1 Study Site

The study was conducted in the largest city of Switzerland, Zurich¨ (380,499 (2008) inhabitants on 8,774 ha), located in its north-eastern part. Most of the city‘s surface is paved (buildings, 70 Chapter 4

streets) (60%), while 37.5% consist of forests, parks and agricultural land (Statistisches Amt Zurich,¨ 2008). Following recent predictions an increase of the population in Switzerland over the next 50 years of about 20% is expected (Giannakouris, 2008). Our study sites were distributed over the entire city (Fig. 4.1) with 27 (out of a total of 89) located within the urban centre and the rest located in the more peripheral regions of the city.

4.2.2 Data Collection

Plants and Pollinators

Between 26th May and 10th September 2008 non-recurring observations in randomly chosen green areas within the city such as parks, stripes along roads, and gardens on flower visitors and plant species were conducted. At each of the 89 sites the number and identity of each plant species in bloom, and flower visitation frequency and identity of each pollinator species visiting the site was recorded within a 2×2 m plots for 40 minutes between 0900 – 1700 on sunny days. The position of the observer was changed every 10 minutes to avoid bias due to shading. Flower visitors were only recorded when entering the plot and landing on flowers and actively searching for rewards. Flower visitors that revisited the plot were not recorded if they moved less than 2 m away from its edge. Familiarity with flower visitor taxa had been acquired through extensive field observations and collections in the city in the preceding two years. Nonetheless, unknown pollinators, when encountered, were caught and later identified using relevant keys (Amiet, 1996; van Veen, 2004; Oosterbroek, 2006). Some species pairs or groups could not be readily differentiated in the field (e.g., B. terrestris and B. lucorum), so these were treated as single species (B. terrestris), as were the Lasioglossum species group (scored as Lasioglossum spp.) and the Halictus species group (Halictus spp.). At each plot we determined plant species (following Ba¨ßler et al., 1996) and counted the number of flowers of all plants after each observation period. Densely clustered floral heads as Asteraceae, Apiaceae or some flowers of Fabacaea (e.g., Trifolium spp.) were considered as one single flower. Observational sites with single plant species were chosen at locations where other heterospecifics did not occur at least within a distance of 20 meters from the selected site. At each study site plots were chosen (based on a subjective assessment) to reflect the surrounding flowering community. 4.2 Materials & Methods 71

Landscape Metrics

We used two landscape metrics to quantify spatial structure around each of the 89 patches within the urban environment of Zurich¨ using ArcGIS 9.2 (Environmental Systems Research Institute (ESRI) Inc., 2006) and QGIS 1.3 Mimas (?): (a) ”green area” was calculated as the ratio of the area (m2) of any open vegetated space (e.g., meadows, grasslands, gardens, parks, and strips along roads etc., but excluding forest which, as a heavily shaded environment, is not open) over the total land area considered and (b) ”edge density” (m ha−1), deﬁned as the ratio of edge length of open vegetated space over its total area (Elkie et al., 1999), which represents the degree of habitat fragmentation. At each site values for the two landscape metrics were calculated for multiple layers from 20 to 100 m radius in 10-meter-steps, and from 100 to 200 meters in 25-meter-steps. We used maximum distance value of 200 m from each observational plot for layers for two reasons: (1) successful pollen transfer among plant species have been shown to be roughly within the same range in urban habitats (Van Rossum and Triest, 2010), and (2) surveyed areas centered on observation plots began to overlap.

4.2.3 Analysis

Species Diversity

0 0 Diversity was calculated using the Shannon (H ) index deﬁned as: H = −∑ pi ln pi, with pi as the proportion of individuals found in the ith species (Magurran, 2004). Species accumulation curves were used to illustrate the rate at which new species were found. Levelling off of curves indicates that fewer species will be discovered with increasing sampling size. For the calculation of the species accumulation curves we applied the method random, which adds sites in random order, and permutated the procedure 1000 times. Permutations prevent that the overall shape of the curve is inﬂuenced by an especially speciose (e.g., species-rich) sample.

Landscape Metrics

Relationships between landscape metrics and ﬂower visits to the most common plant species at each distant class were tested using Spearman correlation. Plots at the edge of the map were 72 Chapter 4

excluded from the correlation analysis, because of incomplete information about the extent and edge density of green area. Correlation coefficients of each analysis were plotted to show the relation with increasing scale, and the direction of the relation within each distance class. We showed significant results at p-values 0.05 and 0.1 due to the large variation, though we stress that care should be taken in interpreting results as Type I error increases when setting p-values at 0.1. Linear regression was used to analyze the significance of the trend along scales.

Local and Large Scale Effects

The flower visits/floral abundance of the focal plant ratio was related to plant diversity and floral abundance of heterospecific plant species. As there might be differences among the responses of different pollinator taxa, the analysis was divided into four groups: all flower visitors, bees, bees without the most common visitor species, and the most common visitor species alone. To reveal the impact of plant diversity, floral abundance, and landscape metrics on visitation rate of a focal plant, we used the most common plant species, Trifolium pratense L. in this study. Using multiple linear regression we explored the relation of plant diversity, floral density, and the landscape metrics green area and edge density at highest R2-values with ratio of flower visits to T. pratense divided by the number of flowers within the patch.

Plant Interactions

We used linear regression analysis to examine the effect of co-occurring plant species on visits to the focal plant. As each pair of species occurred in different samples (see the sample sizes in Fig. 4.4), we analyzed plant-plant interactions in separate regression models. Because visitation to the focal plant can be influenced by the floral abundance of other species or the entire plant community, we correlated floral abundance of each non-focal plant species with the remaining plant community usingESpearmanˆ rank correlation. We applied Analysis of Variance (ANOVA) to test for differences in Shannon diversity of flower visitors between the plant species, as well among Trifolium pratense floral visits when the focal plant was flowering alone, with Lotus corniculatus, and, with Centaurea jacea. Differences between pairs of plant species were analyzed using t-tests when the overall test was significant. 4.2 Materials & Methods 73

Model Diagnostics

All models were investigated visually for violating the assumption of normal distribution and homoscedasticity of residuals. Explanatory and response variables were transformed when heteroscedasticity or non-normality occurs or in case of outliers. We thus square rooted visits to the focal plant, log-transformed floral densities, and applied an arcsine square root transformation to proportions. In the multiple regression analysis for local and landscape effects explanatory variables were centered around their mean (i.e., each observation substracted from the mean of the variable under consideration) to avoid multicollinearity. With increasing scale the extent of overlap between study areas increased. We therefore used the global Moran‘s I to determine the degree of spatial autocorrelation among residuals following the procedure described in the appendix of Dormann et al. (2007). Global Moran‘s I is generally defined as n n n ∑i=1 ∑ j=1 wi j(xi − x¯)(x j − x¯) I = n n n 2 ∑i=1 ∑ j=1 wi j ∑i=1(xi − x¯) with the spatial weights matrix wi j and xi and x j as the values of the residuals at sites i and j (Legendre and Legendre, 1998; Bivand et al., 2009b). We applied a semiparametric spatial eigenvector approach as described by Tiefelsdorf and Griffith (2007). Eigenvectors describe the spatial pattern with associated autocorrelation levels and were derived for each linear regression model separately. To find suitable eigenvectors for each model we added them sequentially and checked the spatial residual autocorrelation using Moran‘s I each time. We no longer included eigenvectors once there was no significant spatial autocorrelation. For the regression models we presented the coefficients with their standard errors and p-values. We also show Moran‘s I,R2 and p-values of linear regression models without and with spatial eigenvectors to show model improvement. The statistical analyses were conducted with the program R (R Development Core Team, 2009) using the packages vegan (Oksanen et al., 2009) to calculate the diversity of plants, AED (Zuur, 2008), faraway (Faraway, 2009) for investigating multicollinearity, and the packages ncf (Bjornstad, 2009) and spdep (Bivand et al., 2009) for calculation of Moran‘s I and eigenvectors. 74 Chapter 4

4.3 Results

4.3.1 Plant Species

Eighty-nine plant patches were observed between 26th May and 10th September 2008. The average distance between patches was 3.5 km with a maximum distance between two patches of 11.17 km and a minimum distance of 40 m. We recorded 67 plant species from 19 plant families (see 4.A). The most species-rich plant families were Asteraceae (17 spp.), Fabaceae (11 spp.), and Lamiaceae (8 spp.). Most flowers were counted for Fabaceae (9,965), Asteraceae (4,180), and Lamiaceae (1,133). On average 5.02 ± 2.3 (mean and standard deviation) plant species and 212.2 ± 221.2 flowers were recorded. Maximum plant species number was 12 plant species and maximum floral abundance was 950 flowers in a single plot. Minimum values were one plant species and 18 flowers. The most common plant family was Fabaceae (94% of all patches), contributing most of the flowers (53%). The most frequently recorded plant species was Trifolium pratense (Fabaceae) (61 out of 89 plots), which had the largest floral abundance (3,164 flowers across all plots). We used Trifolium pratense as a focal species.

4.3.2 Flower Visitor of Trifolium pratense

Flower Visitor Community

Flowers of Trifolium pratense were visited 1,533 times by 34 visitor species in 61 plots, which averages to 25 floral visits per patch (see 4.B). Most flower visits were by hymenopterans (1,470 visits, 95.9%), which almost entirely consisted of species from the family Apidae (1,469 visits), with the bumblebee Bombus pascuorum as the main visitor (1,037 visits, 67.6% of all visits). Diptera including Syrphidae contributed 3.3% (50 visits) of visits, whereas Lepidoptera (12 visits, 0.8 %) and Hemiptera (1 visit, 0.1 %) had minor relevance as Trifolium pratense visitors. In view of this our analyses of the effect of plant diversity, floral abundance and landscape metrics were conducted on all visitor, bee visit, and the main visitor data. The main visitor Bombus pascuorum visited 27 plant species in our study, although 14 of these species were visited by this bumblebee only once across the entire study. The remaining 13 plant species together with the frequency of their visitation by B. pascuorum are listed in Tab. 4.7. Across all plots most visits by B. pascuorum were to Trifolium pratense, largely due to 4.3 Results 75

the ubiquity of T. pratense in most plots. Centaurea jacea was, however, the most visited plant species when only the plots in which particular plants occurred were considered (Tab. 4.7). With regards to the three most common sympatrics Trifolium repens, Lotus corniculatus, and Centaurea cyanus, there was no significant correlation between floral abundance of the plant community with the floral abundance of each heterospecific plant species. There was no significant T. pratense F p F difference in visits to by all species ( 2,41 = 0.044, = 0.96), by all bees ( 2,40 = p F B. pascuorum F 0.052, = 0.95), bees excluding the main visitor ( 2,36 = 0.13, p = 0.88), and ( 2,36 = 0.51, p = 0.6) between field sites with T. pratense, T. pratense together with L. corniculatus, and T. pratense with T. repens (comparisons with C. jacea and with combinations of three and four plant species were excluded due to low number of replicates, Fig. 4.4). In comparison, Shannon diversity of flower visitors differed significantly between the four plant F p C. jacea species ( 3,118 = 4.23, = 0.007), but only had a significantly larger flower visitor diversity than T. pratense (t = 2.41, p = 0.017, Fig. 4.2). Taking into account the different number of sites each plant species occurred the expected mean species richness of flower visitors between C. jacea and T. pratense differed significantly as indicated by the confidence intervals, which do not overlap (Fig. 4.2).

Green Area Size

Mean proportion of green area size declined with increasing scale, with highest values encountered at the lowest scale (20 m or ≈ 1,250 m2) (see 4.C). The contribution of different landscape elements representing green areas also varied across scales (see 4.4). Parks, gardens, and green strips covered proportionately less area with increasing scale, while contribution of forests to the green area metric was greatest at the 70 - 80 m radius scale. Green area size correlated positively with visits of all species, all bees and B. pascuorum (the main visitor of T. pratense), but negatively with bees when excluding the main visitor. Correlation values were largest for visits of B. pascuorum (at 200 m scale) and lowest for visits of ”bees without the main visitor” (at 70 m scale or ≈ 15,500 m2) (Fig. 4.3). Maximum correlation coefficient values occurred at above 150 m scale (≈ 70,500 m2) for all four response variables. At scales smaller than 90 m radius (≈ 25,000 m2) correlation coefficients were variable and increased thereafter (Fig. 4.3). As the scale of spatial analysis increased (from 20 to 200 m radius) the correlation coefficient between green area and floral visits to T. pratense (after accounting for the number of T. pratense flowers in each plot) by all bees, bees without the main visitor, and also by B. pascuorum increased (Fig. 4.3). 76 Chapter 4

Edge Density

Maximum and mean edge density declined with scale. Largest values were observed at the lowest scale (i.e., 20 m). Edge density was positively correlated with visits of all species, all bees and B. pascuorum (Fig. 4.3). As with green area, correlation coefficients were highly variable at scales smaller than 100 m radius (≈ 31,500 m2). The highest correlation value was found in bee visits when excluding the main visitor B. pascuorum (Spearman‘s Correlation Coefficient of −0.30 at 100 m radius or ≈ 31,500 m2), while lowest correlation was in visits of all bee species at 90 m scale (Spearman‘s Correlation Coefficient = 0.002). Largest correlation coefficients were found for B. pascuorum visits and bee visits with the main visitor excluded, but none of the variables showed an effect of increasing scale on correlation (Tab. 4.4).

Effects of Local and Large Scale Factors

There was no spatial residual autocorrelation in the models for the effects of local and landscape variables on visits to the focal plant, though a few models concerning the focal-heterospecific plant interactions did show some spatial residual pattern (Tab. 4.3-4.4). In all affected models the autocorrelation was removed after the inclusion of the first eigenvector (Tab. 4.4). Visits to T. pratense by all visitor species declined with increasing floral abundance of heterospecific plants (Tab. 4.3), a result that remained consistent as the diversity of plant species increased. Plant diversity and heterospecific floral density interacted to cause a disproportionate decline in visits by bees excluding B. pascuorum (Tab. 4.3). Heterospecific floral density interacted with green area in a similar manner with respect to visits by bees excluding B. pascuorum. Further, fewer bees visited the focal plant with increasing edges of green landscape elements. Trifolium pratense was also visited less frequently by bees (including B. pascuorum) as the diversity of plant species increased, but this was dependent on the extent of green area (at high values for green area plant diversity was positively correlated with the number of bee visits (Tab. 4.3)). Visits by Bombus pascuorum alone to T. pratense were influenced negatively by floral density of heterospecific sympatrics, but there was no effect of plant diversity or landscape metrics. 4.4 Discussion 77

Plant Interactions

Centaurea jacea, Lotus corniculatus, and Trifolium repens were the three most commonly co- occurring plants with T. pratense. Floral density of L. corniculatus significantly decreased floral visits to T. pratense by all species and by bees (Tab. 4.4). Increasing floral density of Trifolium repens reduced floral visits to T. pratense by all bees. There was no effect of heterospecific floral abundance relative to T. pratense flowers on visits by all species, bees and by B. pascuorum.

4.4 Discussion

4.4.1 Local Community Effects

Plant diversity within the observation patch negatively affected visits by all species and by all bees to T. pratense, suggesting competition among flowering species for bee visits. Additionally, the main visitor of T. pratense, Bombus pascuorum, visited the focal plant less frequently when floral density of heterospecific plant species increased. This contrasts with results from semi-natural and natural environments, where plant diversity and abundance was frequently reported to enhance pollinator diversity and visitation frequency (Potts et al., 2003; Moeller, 2004; Fontaine et al., 2006; Ghazoul, 2006; Hegland and Boeke, 2006; Ebeling et al., 2008; Molina-Montenegro et al., 2008). Interactions among plants for flower visitors are, however, influenced by the degree of generalization (i.e., here the capability of using different partners) (Lazaro´ et al., 2009). In urban environment plants that are hosts for specialist pollinators might be less abundant (Cane, 2005), which suggests that such specialist pollinators may be particularly vulnerable to the distribution and composition of urban plant communities. Thus urban plant communities may tend to favour plant species pollinated by a wide variety of flower visitors, and therefore the potential for competition exists if pollinators are limiting (Lazaro´ et al., 2009). Although T. pratense was visited by few pollinator species, these pollinators also frequently visited other co-occurring plant species, including white-flowered Trifolium repens, which was the second most abundant plant in the plots. Flowers of Trifolium repens, however, received only 0.1 visits/plot by B. pascuorum compared to 0.4 for the red-flowered T. pratense, suggesting that it is unlikely to be a competitor should pollinators be limited. Taking into the account visits to all plant species and the number of plant occurrences in plots B. pascuorum appears to prefer another common plant Centaurea jacea (1.1 visits per flower) over T. pratense (0.4 visits per flower). Thus potential competitive interactions among plant species for generalist pollinators 78 Chapter 4

may differ substantially depending on the particular composition of the local plant community.

4.4.2 Landscape Scale Effects

Green Area Size

There was positive correlation of all visitors, bees, and Bombus pascuorum with green area over the entire scale. Numerical studies emphasized the importance of the extent of green areas for animals in urban environments (Saure, 1996; Stuke, 1998; Angold et al., 2006; McFrederick and LeBuhn, 2006; Smith et al., 2006; Colding, 2007). For example, urban green areas such as parks, gardens and green roofs provide floral resources and nesting sites for flower visitors (Saure, 1996; Oberndorfer et al., 2007; Matteson et al., 2008; Frankie et al., 2009), while flowering lawns can be used as ”stepping stones” and corridors (Dearborn and Kark, 2010). Trifolium pratense therefore benefitted from increased extent of green areas, which facilitated foraging between ”green” patches as is suggested by the positive correlation of extent of green area with visits to T. pratense by all visitors, bees, and B. pascuorum. Additionally, with increasing scale of analysis, the absolute amount of green area increases and as it does the number and perhaps variety of resources and nesting sites is likely to increase (Band et al., 2005; Colding, 2007). This also implies that more bee species will tend to visit the focal plant when extent of green area increases at all increments from 20 - 200 m radius. The negative correlation between green area and visits by bees excluding B. pascuorum (Fig. 4.3) might be attributed to species foraging ranges (Greenleaf et al., 2007). Small bees, such as Lasioglossum spp., have small foraging ranges (Gathmann and Tscharntke, 2002; Greenleaf et al., 2007) and are more likely to be prone to a decline in the extent of green area in the surrounding habitat matrix than larger bees such as B. pascuorum (Westphal et al., 2006; Williams and Kremen, 2007; Zurbuchen et al., 2010a,b). Large-bodied bees such as B. pascuorum have a capacity to forage over large distances, and consequently are expected to be less vulnerable to habitat fragmentation (Steffan-Dewenter and Tscharntke, 1999; Westphal et al., 2006). Additionally, green areas such as parks or large gardens offer suitable nesting possibilities for bumblebees in urban environments (McFrederick and LeBuhn, 2006; Ahrne´ et al., 2009; Matteson and Langelotto, 2009). 4.4 Discussion 79

Edge Density of Green Area

The positive correlation between green area edge density with floral visits to T. pratense over the entire scale might be explained by the role of green area edges as foraging routes and habitats. In rural environments, pollinators tend to forage along linear landscape elements (Van Geert et al., 2010), indicating that flower visitors might use edges as foraging routes (Tewksbury et al., 2002). A higher edge density might therefore facilitate pollinator movement and enhance flower visits. Moreover, pollinators may respond positively to edges owing to many flowering shrubs and trees that grow along garden and park edges, and even along road margins (Hopwood, 2008). The correlation coefficients for all visitors and all bees were lower than for B. pascuorum and negative for bees excluding B. pascuorum, suggesting that there is species-specific response to edge density. Flower visitors perceive the landscape differently depending on their foraging capability, which depends on body size (Greenleaf et al., 2007; Zurbuchen et al., 2010b). Smaller flower visitors tend to forage at smaller scales, and their vulnerability to habitat fragmentation is reflected by the negative effect of edge density on bee visits when the large-bodied B. pascuorum was excluded.

4.4.3 Interaction between Local and Landscape Scales

The proportion of green area interacted positively with plant diversity with respect to floral visits by bees to T. pratense and negatively with heterospecific floral density. As the proportion of green area in urban environments increases we expect the urban landscape to support a greater number of plant and pollinator species due to an increase in resources (i.e., plant species for pollinators) and niches (i.e., microhabitats for pollinators and plants) (Saure, 1996; Bastin and Thomas, 1999; Cornelis and Hermy, 2004; Thompson et al., 2004; McFrederick and LeBuhn, 2006; Meyer et al., 2007). The extent to which changing plant resource across scales is able to support urban pollinator populations may lead to competition among plants for increasingly limited pollinators or, conversely, may allow larger pollinator populations to be sustained to the mutual (facilitative) benefit of the plants species. Thus in small urban green areas less diverse plant communities and smaller plant populations offer fewer resources for pollinators (Sih and Baltus, 1987). Flowering patches in urban environments are, arguably, more fragmented than in semi-natural or natural areas, which imposes longer foraging distances and hence higher energy requirements (Zurbuchen et al., 2010a). Pollinators might therefore be more attracted to diverse plant patches, which increases the likelihood of visits to the focal plant (i.e., facilitation). Conversely, the greater abundance of floral resources in larger green areas allows for a higher 80 Chapter 4

degree of ﬂoral constancy and, potentially, competition among plants for pollinators.

4.4.4 Conclusion

Correlations of flower visits in T. pratense with the landscape metrics showed that the extent of green areas is the most important landscape metric for the most common pollinators of T. pratense. This corroborates results from previous studies, which have shown that the size of parks or gardens was the most important variable for urban fauna conservation (McFrederick and LeBuhn, 2006; Smith et al., 2006). The continuous increase of the correlation coefficients suggests that green area is more important at larger scales, which confirms results from non-urban environments (Steffan-Dewenter and Tscharntke, 2002; Meyer et al., 2007, 2009). We have shown that increasing green area increases visitation frequency to the focal plant, but plant diversity independently of green area reduces visitation frequency suggesting possible plant competition for pollinators. Conversely, heterospecific floral abundance alone had no effect on bee visitation frequency, but fewer bees visited the focal plant when heterospecific floral abundance and green area increased. These results emphasize the need to consider green areas in plant-plant interactions for attracting flower visitors owing to their potential of providing pollination functions.

4.5 Acknowledgements

We like to thank Pedro Jordano and three anonymous reviewers for helpful comments and critic on a former draft of the manuscript. 4.6 References 81

4.6 References

Ahrne,´ K., Bengtsson, J., and Elmqvist, T. (2009). Bumble Bees (Bombus spp.) along a Gradient of Increasing Urbanization. PLoS One (Public Library of Science Biology), 4(5):e5574.

Alberti, M. (2005). The effects of urban patterns on ecosystem function. International Regional Science Review, 28(2):168–192.

Alberti, M. (2008). Advances in Urban Ecology. Springer.

Amiet, F. (1996). HYMENOPTERA. APIDAE, 1. Teil, volume 12 of INSECTA HELVETICA. Schweizerische Entomologische Gesellschaft.

Angold, P. G., Sadler, J. P., Hill, M. O., Pullin, A., Rushton, S., Austin, K., Small, E., Wood, B., Wadsworth, R., Sanderson, R., and Thompson, K. (2006). Biodiversity in urban habitat patches. Science of the Total Environment, 360(1-3):196–204.

Band, L. E., Cadenasso, M. L., Grimmond, C. S., Grove, J. M., and Pickett, S. T. A. (2005). Heterogeneity in urban ecosystems: Patterns and process. Ecosystem Function in Heterogeneous Landscapes, pages 257–278.

Barrett, S. C. and Kohn, J. R. (1991). Genetic and Evolutionary Consequences of Small Popula- tion Size in Plants: Implications for Conservation. In Falk, D. A. and Holsinger, K. E., editors, Genetics and Conservation of Rare Plants, chapter 1, pages 3–30. Oxford University Press.

Baßler,¨ M., Jager,¨ E., and Werner, K. (1996). ROTHMALER Exkursionsﬂora von Deutschland. Gefaßpﬂanzen:¨ Grundband, volume 2. Gustav Fischer, 16 edition.

Bastin, L. and Thomas, C. D. (1999). The distribution of plant species in urban vegetation fragments. Landscape Ecology, 14(5):493–507.

Bivand, R., with contributions by Luc Anselin, ao, R. A., Berke, O., Bernat, A., Carvalho, M., Chun, Y., Christensen, B., Dormann, C., Dray, S., Halbersma, R., Krainski, E., Lewin-Koh, 82 Chapter 4

N., Li, H., Ma, J., Millo, G., Mueller, W., Ono, H., Peres-Neto, P., Reder, M., Tiefelsdorf, M., and Yu, D. (2009a). spdep: Spatial dependence: weighting schemes, statistics and models.R package version 0.4-52.

Bivand, R. S., Pebesma, E. J., and Gomez-Rubio,´ V. (2009b). Applied Spatial Data Analysis with R (Use R!). Springer, 1 edition.

Bjornstad, O. N. (2009). ncf: spatial nonparametric covariance functions. R package version 1.1-3.

Cane, J. H. (2005). Bees, pollination, and the challenges of sprawl. In Johnson, E. A. and Klemens, M. W., editors, Nature in Fragments, chapter 5, pages 109–124. Columbia University Press.

Colding, J. (2007). ’Ecological land-use complementation’ for building resilience in urban ecosystems. Landscape and Urban Planning, 81(1-2):46–55.

Cornelis, J. and Hermy, M. (2004). Biodiversity relationships in urban and suburban parks in Flanders. Landscape and Urban Planning, 69(4):385–401.

Dearborn, D. C. and Kark, S. (2010). Motivations for Conserving Urban Biodiversity. Conserva- tion Biology, 24(2):432–440.

Diekotter,¨ T., Haynes, K. J., Mazeffa, D., and Crist, T. O. (2007). Direct and indirect effects of habitat area and matrix composition on species interactions among ﬂower-visiting insects. Oikos, 116(9):1588–1598.

Donaldson, J., Nanni, I., Zachariades, C., Kemper, J., and Thompson, J. D. (2002). Effects of habitat fragmentation on pollinator diversity and plant reproductive success in renosterveld shrublands of South Africa. Conservation Biology, 16(5):1267–1276.

Dormann, C. F., McPherson, J. M., Araujo, M. B., Bivand, R., Bolliger, J., Carl, G., Davies, R. G., Hirzel, A., Jetz, W., Kissling, W. D., Kuehn, I., Ohlemueller, R., Peres-Neto, P. R., Reineking, B., Schroeder, B., Schurr, F. M., and Wilson, R. (2007). Methods to account for spatial autocorrelation in the analysis of species distributional data: a review. Ecography, 30(5):609–628. 4.6 References 83

Elkie, P. C., Rempel, R. S., and Carr, A. P. (1999). Patch Analyst User‘s Manual. A Tool for Quantifying Landscape Structure. Ontario Ministry of Natural Resources. Northwest Science & Technology Thunder Bay, Ontario TM-002.

Environmental Systems Research Institute (ESRI) Inc. (2006). ArcGIS.

Faraway, J. (2009). faraway: Functions and datasets for books by Julian Faraway. R package version 1.0.4.

Fontaine, C., Collin, C. L., and Dajoz, I. (2008). Generalist foraging of pollinators: diet expansion at high density. Journal of Ecology, 96(5):1002–1010.

Fontaine, C., Dajoz, I., Meriguet, J., and Loreau, M. (2006). Functional diversity of plant- pollinator interaction webs enhances the persistence of plant communities. PLOS Biology, 4(1):129–135.

Frankie, G. W., Thorp, R., Hernandez, J., Rizzardi, M., Ertter, B., Pawelek, J. C., Witt, S. L., Schindler, M., Coville, R., and Wojcik, V. (2009). Native bees are a rich natural resource in urban California gardens. California Agriculture, 63(3):113–120.

Gathmann, A. and Tscharntke, T. (2002). Foraging ranges of solitary bees. Journal of Animal Ecology, 71(5):757–764.

Ghazoul, J. (2005). Pollen and seed dispersal among dispersed plants. Biological Reviews, 80(3):413–443.

Ghazoul, J. (2006). Floral diversity and the facilitation of pollination. Journal of Ecology, 94:295–304.

Giannakouris, K. (2008). Population and social conditions. Technical Report 72, EUROSTAT. European Comission.

Gilbert, F., Gonzalez, A., and Evans-Freke, I. (1998). Corridors maintain species richness in the fragmented landscapes of a microecosystem. Proceedings of the Royal Society of London Series B - Biological Sciences, 265(1396):577–582.

Godefroid, S. and Koedam, N. (2007). Urban plant species patterns are highly driven by density and function of built-up areas. Landscape Ecology, 22(8):1227–1239. 84 Chapter 4

Greenleaf, S. S., Williams, N. M., Winfree, R., and Kremen, C. (2007). Bee foraging ranges and their relationship to body size. Oecologia, 153(3):589–596.

Hopwood, J. L. (2008). The contribution of roadside grassland restorations to native bee conservation. Biological Conservation, 141(10):2632–2640.

Kearns, C. A., Inouye, D. W., and Waser, N. M. (1998). Endangered mutualisms: The conservation of plant-pollinator interactions. Annual Review of Ecology and Systematics, 29:83–112.

Kearns, C. A. and Oliveras, D. M. (2009). Environmental factors affecting bee diversity in urban and remote grassland plots in Boulder, Colorado. Journal of Insect Conservation, 13(6):655–665.

Kitahara, M. and Fujii, K. (1994). Biodiversity and community structure of temperate butterﬂy species within a gradient of human disturbance - An analysis based on the concept of generalist vs. specialist strategies. Researches on Population Ecology, 36(2):187–199.

Kwak, M. M., Velterop, O., and van Andel, J. (1998). Pollen and gene ﬂow in fragmented habitats. Applied Vegetation Science, 1:37–54.

Lazaro,´ A., Lundgren, R., and Totland, O. (2009). Co-ﬂowering neighbors inﬂuence the diversity and identity of pollinator groups visiting plant species. Oikos, 118(5):691–702.

Legendre, P. and Legendre, L. (1998). Numerical Ecology (Developments in Environmental Modelling). Elsevier Science Ltd, 2 ed edition.

Lindborg, R. (2007). Evaluating the distribution of plant life-history traits in relation to current and historical landscape conﬁgurations. Journal of Ecology, 95(3):555–564.

Lososova,´ Z., Chytry,´ M., Kuhn,¨ I., Hajek,´ O., Horakov´ a,´ V., Pysek,ˇ P., and Tichy,´ L. (2006). Patterns of plant traits in annual vegetation of man-made habitats in central Europe. Perspectives in Plant Ecology, Evolution and Systematics, 8(2):69–81.

Magurran, A. E. (2004). Measuring Biological Diversity. Blackwell Publishing.

Marvier, M., Kareiva, P., and Neubert, M. G. (2004). Habitat destruction, fragmentation, and disturbance promote invasion by habitat generalists in a multispecies metapopulation. Risk Analysis, 24(4):869–878. 4.6 References 85

Marzluff, J. M., Shulenberger, E., Endlicher, W., Alberti, M., Bradley, G., Ryan, C., Simon, U., and ZumBrunnen, C. (2008). Urban Ecology: An International Perspective on the Interaction Between Humans and Nature. Springer.

Matteson, K. C., Ascher, J. S., and Langellotto, G. A. (2008). Bee richness and abundance in New York city urban gardens. Annals of the Entomological Society of America, 101(1):140–150.

Matteson, K. C. and Langelotto, G. A. (2009). Bumble Bee Abundance in New York City Com- munity Gardens: Implications for Urban Agriculture. Cities and the Environment, 2(1):article 5, 12 pp.

Mayer, P. (2010). Urban ecosystems research joins mainstream ecology. Nature, 467(7312):153– 153.

McFrederick, Q. S. and LeBuhn, G. (2006). Are urban parks refuges for bumble bees Bombus spp. (Hymenoptera : Apidae)? Biological Conservation, 129(3):372–382.

Memmott, J., Waser, N., and Price, M. (2004). Tolerance of pollination networks to species extinctions. Proceedings of the Royal Society of London Series B - Biological Sciences, 271(1557):2605–2611.

Meyer, B., Gaebele, V., and Steffan-Dewenter, I. D. (2007). Patch size and landscape effects on pollinators and seed set of the horseshoe vetch, Hippocrepis comosa, in an agricultural landscape of central Europe. Entomologia Generalis, 30(2):173–185.

Meyer, B., Jauker, F., and Steffan-Dewenter, I. (2009). Contrasting resource-dependent responses of hoverﬂy richness and density to landscape structure. Basic and Applied Ecology, 10(2):178– 186.

Miller, J. R. (2005). Biodiversity conservation and the extinction of experience. Trends in Ecology & Evolution, 20(8):430–434.

Moeller, D. A. (2004). Facilitative interactions among plants via shared pollinators. Ecology, 85(12):3289–3301.

Morgan, J. W. (1999). Effects of population size on seed production and germinability in an endangered, fragmented grassland plant. Conservation Biology, 13(2):266–273. 86 Chapter 4

Oberndorfer, E., Lundholm, J., Bass, B., Coffman, R. R., Doshi, H., Dunnett, N., Gafﬁn, S., Kohler, M., Liu, K. K. Y., and Rowe, B. (2007). Green roofs as urban ecosystems: Ecological structures, functions, and services. Bioscience, 57(10):823–833.

Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Simpson, G. L., Solymos, P., Stevens, M. H. H., and Wagner, H. (2009). vegan: Community Ecology Package. R package version 1.15-4.

Oosterbroek, P. (2006). The European Families of the Diptera. KNNV Publishing.

Pauw, A. (2007). Collapse of a pollination web in small conservation areas. Ecology, 88(7):1759– 1769.

Potts, S. G., Vulliamy, B., Dafni, A., Ne’eman, G., and Willmer, P. (2003). Linking bees and ﬂowers: How do ﬂoral communities structure pollinator communities? Ecology, 84(10):2628– 2642.

Quesada, M., Stoner, K. E., Lobo, J. A., Herrerias-Diego, Y., Palacios-Guevara, C., Munguia- Rosas, M. A., Salazar, K. A. O., and Rosas-Guerrero, V. (2004). Effects of forest fragmentation on pollinator activity and consequences for plant reproductive success and mating patterns in bat-pollinated bombacaceous trees. Biotropica, 36(2):131–138.

R Development Core Team (2009). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0.

Ramos, F. N. and Santos, F. A. M. (2006). Floral visitors and pollination of Psychotria ten- uinervis (Rubiaceae): Distance from the anthropogenic and natural edges of an Atlantic forest fragment. Biotropica, 38(3):383–389.

Rathcke, B. and Jules, E. (1993). Habitat fragmentation and plant pollinator interactions. Current Science, 65(3):273–277.

Sabatino, M., Maceira, N., and Aizen, M. A. (2010). Direct effects of habitat area on interaction diversity in pollination webs; ecological applications. Ecological Applications, 20(6):1491.

Sih, A. and Baltus, M. S. (1987). Patch size, pollinator behavior, and pollinator limitation in Catnip. Ecology, 68(6):1679–1690. 4.6 References 87

Smith, R. M., Warren, P. H., Thompson, K., and Gaston, K. J. (2006). Urban domestic gardens (VI): environmental correlates of invertebrate species richness. Biodiversity and Conservation, 15(8):2415–2438.

Statistisches Amt Zurich¨ (2008). Medienmitteilung des Statistischen Amtes. Rekordhohes Bevolkerungswachstum¨ im Kanton Zurich.¨ Technical report, Kanton Zurich.¨

Steffan-Dewenter, I., Munzenberg, U., and Tscharntke, T. (2001). Pollination, seed set and seed predation on a landscape scale. Proceedings of the Royal Society of London Series B - Biological Sciences, 268(1477):1685–1690.

Steffan-Dewenter, I. and Tscharntke, T. (1999). Effects of habitat isolation on pollinator communities and seed set. Oecologia, 121(3):432–440.

Steffan-Dewenter, I. and Tscharntke, T. (2002). Insect communities and biotic interactions on fragmented calcareous grasslands - a mini review. Biological Conservation, 104(3):275–284.

Tewksbury, J. J., Levey, D. J., Haddad, N. M., Sargent, S., Orrock, J. L., Weldon, A., Danielson, B. J., Brinkerhoff, J., Damschen, E. I., and Townsend, P. (2002). Corridors affect plants, animals, and their interactions in fragmented landscapes. Proceedings of the National Academy of Sciences of the United States of America, 99(20):12923–12926.

Thompson, K., Hodgson, J. G., Smith, R. M., Warren, P. H., and Gaston, K. J. (2004). Urban domestic gardens (III): Composition and diversity of lawn ﬂoras. Journal of Vegetation Science, 15(3):373–378.

Tiefelsdorf, M. and Grifﬁth, D. A. (2007). Semiparametric ﬁltering of spatial auto correlation: the eigenvector approach. Environment and Planning A, 39(5):1193–1221.

Townsend, C. R. (2008). Ecological Applications: toward a sustainable world. Blackwell Publishing.

Townsend, P. A. and Levey, D. J. (2005). An experimental test of whether habitat corridors affect pollen transfer. Ecology, 86(2):466–475.

United Nations (2008). World urbanization prospects: The 2007 revision population database. Technical report, United Nations. 88 Chapter 4

Van der Veken, S., Verheyen, K., and Hermy, M. (2004). Plant species loss in an urban area (Turnhout, Belgium) from 1880 to 1999 and its environmental determinants. Flora, 199(6):516– 523.

Van Geert, A., Van Rossum, F., and Triest, L. (2010). Do linear landscape elements in farmland act as biological corridors for pollen dispersal? Journal of Ecology, 98(1):178–187.

Van Rossum, F. and Triest, L. (2010). Pollen dispersal in an insect-pollinated wet meadow herb along an urban river. Landscape and Urban Planning, 95(4):201–208. van Veen, M. D. (2004). Hoverﬂies of Northwest Europe. KNNV Publishing.

Walther-Hellwig, K., Fokul, G., Frankl, R., Buechler, R., Ekschmitt, K., and Wolters, V. (2006). Increased density of honeybee colonies affects foraging bumblebees. Apidologie, 37(5):517–532.

Westphal, C., Steffan-Dewenter, I., and Tscharntke, T. (2006). Bumblebees experience landscapes at different spatial scales: possible implications for coexistence. Oecologia, 149(2):289– 300.

Williams, N. M. and Kremen, C. (2007). Resource distributions among habitats determine solitary bee offspring production in a mosaic landscape. Ecological Applications, 17(3):910– 921.

Williams, N. S. G., Morgan, J. W., Mcdonnell, M. J., and Mccarthy, M. A. (2005). Plant traits and local extinctions in natural grasslands along an urban-rural gradient. Journal of Ecology, 93(6):1203–1213.

Zerbe, S., Maurer, U., Schmitz, S., and Sukopp, H. (2003). Biodiversity in Berlin and its potential for nature conservation. Landscape and Urban Planning, 62(3):139–148.

Zurbuchen, A., Cheesman, S., Klaiber, J., Mueller, A., Hein, S., and Dorn, S. (2010a). Long foraging distances impose high costs on offspring production in solitary bees. Journal of Animal Ecology, 79(3):674–681.

Zurbuchen, A., Landert, L., Klaiber, J., Mueller, A., Hein, S., and Dorn, S. (2010b). Maximum foraging ranges in solitary bees: only few individuals have the capability to cover long foraging distances. Biological Conservation, 143(3):669–676.

Zuur, A. (2008). AED: Data sets and routines for book. R package version 1.0. 4.7 Tables & Figures 89

4.7 Tables & Figures . COUNT = no. 39 10 07 16 44 54 34 54 03 02 29 12 2 ...... 0 0 1 0 0 2 0 0 0 0 3 1 FLORAL VISITS/FLOWER/COUNT Bombus pascuorum 3 2 2 8 037 76 17 16 68 23 , 126 168 254 1 FLORAL VISITS 4 163 426 91 38 14 16 , , 383 954 430 378 167 127 3 2 SUM FLOWERS 5 6 8 2 3 2 3 5 61 43 19 20 19 COUNT.PLANT 9 4 3 3 2 2 2 2 2 2 49 13 12 The table shows the plant species visited more than once by PLANTTrifolium pratense COUNT Trifolium repens Centaurea jacea Lotus corniculatus Prunella vulgaris Knautia arvensis Prunella grandiflora Cichorium intybus Hypochaeris radicata Medicago sativa Scabiosa columbaria Thymus polytrichus Vicia cracca of plots the plant species was visited, COUNT.PLANT = no. of plots the plant species occurred, FLORAL Table 4.1: VISITS/FLOWER/COUNT = average of floral visits/flower over all plots where the species occurred. 90 Chapter 4

Table 4.2: Results from linear regression for correlation coefﬁcients of the response variables (RESP) visits of all visitor species of Trifolium pratense, all bee species, bees wihtout the main visitor, and the main visitor of T. pratense Bombus pascuorum with the landscape metrics (LM) green area size (GA) and green area edge density (ED) with scale from 20 - 200 m radius.

RESP LM F1,11 t p-value

GA 5.2 2.28 0.04 all ssp. ED 0.25 0.50 0.63 GA 12.2 3.49 0.005 all bees ED 0.89 0.95 0.36 GA 51.03 7.14 < 0.001 B. pascuorum ED 0.28 0.53 0.61 GA 32.75 −5.72 < 0.001 bees w/o B. pascuorum ED 1.34 −1.16 0.27 4.7 Tables & Figures 91

Table 4.3: Results from linear regression for visits of all species (Model I), bees (Model II), bees excluding the main visitor (Model III) and B. pascuorum (Model IV) to T. pratense.

Model I Model II Model III Model IV

Constant 0.555∗∗∗ 0.512∗∗∗ 0.113∗∗∗ 0.364∗∗∗ (0.057)(0.054)(0.021)(0.051) Plant Diversity (PD) −0.142 −0.276∗ −0.099 (0.136)(0.128)(0.052) Het. Floral Density (HFD) −0.123∗ −0.006 −0.119∗∗ (0.051)(0.024)(0.040) GA 0.090 −0.106 (0.292)(0.118) ED −2.498∗ (0.936) PD × HFD −0.211∗ −0.167∗∗∗ (0.084)(0.047) PD × GA 1.470∗ (0.597) HFD × GA −0.395∗∗ (0.117)

R2 0.185 0.166 0.372 0.139 F 4.016∗ 3.506∗ 4.941∗∗∗ 8.881∗∗ N 57 57 57 57 Moran‘s I −0.027 −0.035 −0.029 −0.027 Table 4.4: Statistics and residual autocorrelation in the regression models for visits of all species, bees, bees without the main visitor B. pascuorum, and the main visitor alone to T. pratense, together with the transformed explanatory variables (EXP), i.e., ﬂoral abundance of the plant species L. corniculatus (LC), T. repens (TR), and C. jacea (CJ).

(a) log-transformed ﬂoral density

Non-spatial model Spatial model visits EXP βˆ R2 M(I) AIC βˆ R2 M(I) AIC

CJ −0.014 0.003 0.012 − − − − − all LC −0.123∗∗ 0.415 0.124∗ 2.731 −0.106∗∗ 0.606 −0.192 −1.591 TR −0.031 0.041 0.055∗ 13.048 −0.061 0.346 −0.092 5.096

CJ −0.013 0.002 0.024 − − − − − bees LC −0.12∗∗ 0.418 0.17∗ 1.663 −0.102∗∗ 0.62 −0.15 −3.14 TR −0.029 0.034 0.054∗ 13.639 −0.06∗ 0.35 −0.1 5.369

CJ 0.028 −0.069 −0.205 − − − − − bees w/o Bp LC −0.006 −0.07 −0.015 − − − − − TR 0.033 0.011 −0.049 − − − − −

CJ −0.064 0.035 0.076 − − − − − Bp LC −0.133∗ 0.339 0.122∗ 10.284 −0.078 0.642 −0.187 4.454 TR −0.065 0.133 −0.008 − − − − −

(b) arcsine square root transformed proportion heterospeciﬁc/focal plant ﬂowers

Non-spatial model Spatial model visits EXP βˆ R2 M(I) AIC βˆ R2 M(I) AIC

CJ 0.345 0.071 −0.063 − − − − − all LC −0.345 0.174 0.123∗ 8.244 −0.273 0.42 −0.16 4.591 TR −0.141 0.035 0.051 − − − − −

CJ 0.314 0.07 −0.043 − − − − − bees LC −0.339 0.179 0.154∗ 7.174 −0.267 0.436 −0.135 3.161 TR −0.139 0.033 0.053 − − − − −

CJ 0.156 −0.054 −0.192 − − − − − bees w/o Bp LC −0.166 −0.003 −0.034 − − − − − TR 0.116 −0.019 −0.047 − − − − −

CJ 0.225 0.021 −0.017 − − − − − Bp LC −0.227 0.053 0.129∗ 16.030 −0.142 0.551 −0.165 6.096 TR −0.241 0.075 −0.028 − − − − − 4.7 Tables & Figures 93

Figure 4.1: Map of the study region with the 89 studied locations. 94 Chapter 4

● 50

1.5 40

30 1.0

Shannon Diversity 0.5 CJ 10 LC TP TR Expected Mean Number of Species 0.0 0 TP CJ LC TR 0 10 20 30 40 50 60 Plant Species Sites

Figure 4.2: Boxplot with the mean diversity in the four most common plant species and their species accumulation curves (TR = Trifolium repens, CJ = Centaurea jacea, LC = Lotus corniculatus, TP = Trifolium pratense). Green Area Edge Density

0.22 ● ●

● 0.12 ●

●

● ● ● ● ● ● ● 0.08 ● ● ● ● ●

0.17 ●

● ● ● ● ●

All Visitors 0.04

0.12 ● 0.00 ● 0 20000 60000 100000 0 20000 60000 100000

● 0.15 ●

●

0.20 ●

●

● ● ● ● ● 0.10 ● ● 0.16 ● ● ● ● ●

●

● ●

Bees ● ● 0.05 ● 0.12 ●

0.08 ● 0.00 ● 0 20000 60000 100000 0 20000 60000 100000

● ● 0.35 ● 0.22

● ●

●

0.30 ● ● ● ● ● ● 0.18 ● ● ● ● ● ● 0.25 ● ● Spearman‘s Correlation Coefficient (rho)

● ● ●

B. pascuorum 0.14 ● 0.20

● ● 0.10 0 20000 60000 100000 0 20000 40000 60000 80000 100000 120000

● −0.1 ● 0.00 ●

● ●

● ● ● ●

● ● −0.05 ● ●

● ● −0.2

−0.10 ●

● ● ●

●

● ● −0.15 ● ● Bees w/o B. pascuorum ● −0.3 ● 0 20000 60000 100000 0 20000 60000 100000 Area (m2)

Figure 4.3: Correlation results between visitation variables and landscape metrics. Solid lines show p-values at 0.05 and dashed lines at 0.1. Note the different scaling of the y-axis. There were no signiﬁcant correlations between bees and the extent of green area as well as edge density, all visitors and edge density, and bees w/o main visitor and extent of green area. 96 Chapter 4

1.0 n = 3 1.0 n = 3 n = 4 0.8 0.8 n = 4 n = 6 n = 6 n = 21 n = 20 0.6 n = 17 0.6 n = 17

n = 3 n = 3 0.4 n = 4 0.4

bee visits/flower n = 4 visits by all spp./flower visits by 0.2 n = 3 0.2 n = 3

0.0 0.0 Tp Tp TpCj TpLc TpTr TpCj TpLc TpTr TpCjLc TpCjTr TpLcTr TpCjLc TpCjTr TpLcTr TpTrLcCj TpTrLcCj

1.0 1.0

0.8 n = 3 0.8 n = 4 B. pascuorum

n = 17 0.6 n = 6 0.6 visits/flower

n = 16

0.4 0.4 n = 2 n = 3 n = 6 n = 3

B. pascuorum 0.2 0.2 n = 3 n = 16 n = 2 n = 17 n = 4 n = 2 n = 2 visits/flower by bees w/o 0.0 0.0 Tp Tp TpCj TpLc TpTr TpCj TpLc TpTr TpCjLc TpCjTr TpLcTr TpCjLc TpCjTr TpLcTr TpTrLcCj TpTrLcCj

Figure 4.4: Number of ﬂower visits by all species, bees, and B. pascuorum for T. pratense in species plots with T. pratense alone and in combinations with Centaurea jacea (Cj), Trifolium repens (Tr), and Lotus corniculatus (Lc) (mean and standard error). Plot frequencies for each combination are shown above bars. Results are presented in the text. 4.8 Appendices 97

4.8 Appendices

4.A Number of occurrences and ﬂoral abundance of plant species found in the urban study.

Family Species Count Floral Ab. Achillea millefolium L. 5 208 Anthemis tinctoria L. 1 82 Bellis perennis L. 25 1,716 Centaurea jacea L. s. l. 19 383 Cichorium intybus L. 2 14 Cirsium arvense (L.) Scop. 2 33 Crepis biennis L. 7 36 Crepis praemorsa (L.) Walther 2 9 Asteraceae Crepis tectorum L. 1 4 Erigeron annuus (L.) Pers. 9 144 Hieracium murorum L. 2 20 Hieracium pilosella L. 8 354 Hypericum perforatum L. 1 5 Hypochaeris radicata L. 19 378 Leucanthemum vulgare Lam. s. str. 9 771 Senecio jacobaea L. 1 5 Taraxacum sect. Ruderalia Wiggers 1 18

SUBTOTAL 17 113 4,175

Daucus carota L. 8 208 Apiaceae Heracleum sphondylium L. 2 17

Anchusa ofﬁcinalis L. 1 120 Boraginaceae Echium vulgare L. 1 198 Myosotis arvensis (L.) Hill 1 51

Campanula cervicaria L. 1 13 Campanulaceae Campanula rapunculoides L. 4 61

Cerastium fontanum Baumg. s. str. 12 262 Dianthus carthusianorum L. 1 7 Caryophyllaceae Silene latifolia Poir. 1 28 Silene vulgaris (Moench) Garcke s. l. 2 36

Convolvulaceae Convolvulus arvensis L. 1 1

Knautia arvensis (L.) Coult. s. str. 6 38 Dipsacaceae Scabiosa columbaria L. 2 4

continued on the next page ... 98 Chapter 4

... continued Anthyllis vulneraria L. s. l. 3 64 Galega ofﬁcinalis L. 1 2 Lathyrus latifolius L. 1 11 Lotus corniculatus L. 20 954 Medicago falcata L. s. str. 3 167 Fabaceae Medicago lupulina L. 1 54 Trifolium campestre Schreb. 20 2,400 Trifolium micranthum Viv. 4 711 Trifolium pratense L. 61 3,164 Trifolium repens L. 42 2,422 Vicia cracca L. s. str. 5 16

SUBTOTAL 11 161 9,965

Geranium molle L. 6 92 Geraniaceae Geranium pyrenaicum Burm. f. 1 5

Galeopsis ladanum L. 1 14 Glechoma hederacea L. 1 4 Lamium purpureum L. s. l. p. p. 3 47 Prunella grandiﬂora (L.) Scholler 8 430 Lamiaceae Prunella vulgaris L. 5 91 Salvia pratensis L. 4 222 Stachys recta L. 1 198 Thymus praecox Opiz s. l. 3 127

SUBTOTAL 8 26 1,133

Linaceae Linum tenuifolium L. 1 1

Malvaceae Malva moschata L. 2 30

Epilobium hirsutum L. 1 99 Onagraceae Epilobium parviﬂorum Schreb. 1 3

Plantago lanceolata L. 23 215 Plantaginaceae Plantago media L. 1 1

Ranunculus acris L. 11 86 Ranunculaceae Ranunculus repens L. 14 550

Geum urbanum L. 3 40 Rosaceae Potentilla reptans L. 1 7

Galium album Mill. 25 333 Rubiaceae Galium verum L. s. str. 1 3

Rhinanthus alectorolophus (Scop.) Pollich s. l. 4 983 Scrophulariaceae Veronica agrestis L. 7 86 continued on the next page ... 4.A Number of occurrences and ﬂoral abundance of plant species found in the urban study.99

... continued

Verbenaceae Verbena ofﬁcinalis L. 1 26

19 67 447 18,882 100 Chapter 4

4.B Pollinator species

Species Centaurea jacea Lotus corniculatus Trifolium pratense Trifolium repens

Hymenoptera Andrena ﬂaviceps 2 0 1 0 Andrena haemorrhoa 0 0 2 3 Anthidium manicatum 0 2 0 0 Anthidium strigatum 0 2 0 0 Anthophora plumipes 0 2 10 0 Anthophora species 0 0 0 1 Apis mellifera 1,405 52 103 1,359 Bombus hortorum 0 0 108 0 Bombus humilis 15 0 68 3 Bombus lapidarius 48 142 8 42 Bombus lapidarius bright 5 0 0 0 Bombus lucorum/terrestris 12 21 42 3 Bombus pascuorum 124 168 1,037 76 Bombus pratorum 42 0 0 0 Bombus ruderarius 20 0 0 0 Bombus subterraneus 0 0 1 0 Bombus sylvarum 1 0 43 6 Cerceris rybyensis 0 0 1 1 Chelostoma ﬂorisomne 1 0 0 0 Colletes similis 1 0 0 0 Dasypoda hirtipes 1 0 0 0 Eumeninae 1 0 0 0 Halictus scabiosae 77 0 0 0 Halictus sextinctus 19 0 0 0 Halictus subauratus 11 0 1 0 Hylaeus communis 0 10 1 1 Lasioglossum species 46 1 32 9 Megachile centuncularis 2 0 0 0 Megachile ericetorum 6 137 1 1 Megachile 0206.1522 0 31 0 0 Osmia caerulescens 3 34 11 14 Osmia spec1406.1208 17 0 0 0 Polistes galicus 1 0 0 0 Stelis signata 0 1 0 0

SUBTOTAL 1,860 603 1,470 1,519

Diptera Bibio marco 1 0 0 0 Calliphora 0906.1453 1 0 0 0 Chrysotoxum bicinctum 0 1 0 0 Diptera 1306.1250 0 0 0 1 Empis tessellata 0 0 1 0 Episyrphus balteatus 3 1 3 3 Eristalis arbustorum 1 0 0 0 Eristalis tenax 36 3 29 0 Eristalis pertinax 0 0 1 0 Heliophilus hybridus 11 0 1 0 Heliophilus pendulus 1 0 3 0 Heliophilus trivittatus 0 0 2 0 continued on the next page ... 4.B Pollinator species 101

Lucilia big 0 3 0 0 Myotropa ﬂorea 1 0 1 0 Pipizella pennina 0 0 0 2 Platycheirus albimanus 0 0 1 0 Rhingia campestris 1 0 0 0 Sarcophaga carnaria 2 0 0 4 Scaevia selenitica 0 0 1 0 Sicus ferrugineus 5 0 0 1 Sphaerophoria scripta 2 0 6 0 Syritta pipiens 1 1 0 0 Syrphus ribesii 9 0 1 0 Syrphus small 1 1 0 0

SUBTOTAL 76 10 50 11

Lepidoptera Autographa gamma 6 0 6 0 Cerapteryx graminis 22 0 0 0 Maniola jurtina 1 0 0 0 Pieris rapae 2 0 0 0 Polyommatus icarus 3 12 4 0 Pyrausta aurata 0 0 1 0 Thymelicus sylvestris 0 0 1 0 Zygaena ﬁlipendula 2 0 0 0

SUBTOTAL 36 12 12 0

Coleoptera Leptura rubra 1 0 0 0 Oedemera virescens 3 0 0 0 Trichius fasciatus 3 0 0 0

SUBTOTAL 7 0 0 0

Heteroptera Adelphocoris seticornis 0 0 1 0 SUM 1,979 625 1,533 1,530 102 Chapter 4

4.C Landscape Structures at different scales

Percentage of overlap for the total and green area, and the mean, mininum and maximum values for the proportion and absolute area of green landscape structures as well as edge density at each scale (radius [m]) in the study area.

Overlap Green Area (absolute) Green Area (proportion) Edge Density Scale %Total %GA x min max sd x min max sd x min max sd

20 0 0 848.51 174.77 1236.07 306.26 0.68 0.14 0.98 0.24 0.145 0 0.32 0.051 30 1.16 1.46 1754.02 262.14 2781.15 676.29 0.62 0.09 0.98 0.24 0.123 0 0.25 0.042 40 2.44 3.04 2915.11 350.32 4863.11 1132.9 0.58 0.07 0.97 0.23 0.11 0 0.21 0.036 50 3.76 4.51 4395.55 433.1 7576.06 1734.15 0.56 0.06 0.96 0.22 0.103 0 0.18 0.031 60 5.49 6.51 6167.11 470.37 10812.16 2511.49 0.55 0.04 0.96 0.22 0.10 0.02 0.16 0.028 70 7.7 8.35 8223.07 545.28 14565.67 3404.32 0.53 0.04 0.95 0.22 0.096 0.02 0.14 0.026 80 10.41 10.65 10576.36 1062.54 18688.71 4369.05 0.53 0.05 0.93 0.22 0.096 0.02 0.14 0.026 90 13.21 13.07 13245.01 1890.05 23746.62 5483.48 0.52 0.07 0.93 0.22 0.093 0.02 0.14 0.026 100 15.69 15.41 16192.55 2514.60 29085.64 6718.04 0.52 0.08 0.93 0.21 0.092 0.02 0.14 0.027 125 21.24 21.43 24817.66 3368.81 45521.23 10340.7 0.51 0.07 0.93 0.21 0.09 0.02 0.13 0.026 150 26.41 26.96 35257.30 6938.34 65740.12 14661.89 0.5 0.1 0.93 0.21 0.09 0.01 0.13 0.026 175 32.03 32.79 47313.1 8498.21 90141.9 19835.58 0.49 0.09 0.94 0.21 0.089 0.01 0.13 0.026 200 37.55 38.18 61052.84 12093.15 118258.28 25550.25 0.49 0.1 0.94 0.2 0.088 0.01 0.13 0.026 4.D Green Area Structures 103

4.D Green Area Structures

0.7 0.6 Forest Seminatural 0.5 Park Garden 0.4 Green Strip 0.3 0.2 0.1 0.0 Average Proportional Coverage (%) Proportional Coverage Average 0 40000 80000 120000 Study Area (m2)

Mean percentage of coverage of each landscape element considered as green area across the studied scale (i.e., 20-200 m radius).

Chapter 5

Pollinating Animals in the Urban Environment

Hennig, Ernest Ireneusz & Ghazoul, Jaboury

ETH Zurich,¨ Ecosystem Management, Institute of Terrestrial Ecosystems, Universitatstrasse¨ 16, 8092 Zurich,¨ Switzerland

Abstract The expansion of urban environments raises concerns about the potential of urban environments to support biodiversity and ecosystem functions. In particular, while urban environments contain habitats for flowering plants and their pollinating animal species, it is unclear how the plant-pollinator processes may be influenced by the urban matrix. We investigated the effect of the urban matrix on the diversity and visits of flower visiting animals within the city of Zurich.¨ We recorded plant diversity, floral abundance, flower visitor diversity and plot visits at 89 plant patches. The urban matrix surrounding each site was analyzed for the landscape metrics edge density and the extent of green area at 10 m distance intervals from 20 – 100 m and 25 m distance intervals from 100 – 200 m, respectively. Hymenoptera and Diptera were the most common flower visiting groups, with Apidae and Syrphidae the most common families in each of the two groups. Honeybees contributed 50.5% of all bee visits. The correlation between edge density and bee diversity and visitation frequency varied substantially over the entire spatial range, while the correlation for syrphid diversity and visitation frequency levelled off at 80 m radius (≈ 20,000 m2). In contrast, the correlation between green area and the diversity and visitation frequency of both groups was more consistent,

Published in ”Urban Ecosystems” (2011) 106 Chapter 5

increasing up to 100 m scale (≈ 31,500 m2) and levelling off thereafter in bee diversity, while continuing to increase for syrphid diversity and visits. The variation in the correlation of bee visits was partly accounted for by the large contribution of honeybees. Excluding honeybees from the analysis resulted in correlation coefficients levelling off at 80 m radius, while there was still considerable variation in edge density. Multiple linear regression showed significant effects of plant diversity on bee diversity and visits, and on syrphid visits, but not on syrphid diversity. Floral abundance had a positive effect on bee visits and weakly on bee diversity, while for syrphid diversity there was a negative interaction of floral abundance and green area. The extent of green area positively affected bee diversity and visits, and syrphid visits, while edge density affected negatively bee visits. The study showed that plant diversity and floral abundance in urban environments promote pollinating flower visitors. Moreover, the extent of green area and edge density are important urban mosaic variables that affect pollinator abundance and visitation frequency at multiple scales. Keywords: urban matrix, spatial scale, flower visitors, pollinator, bees, syrphids, diversity, visitation frequency, edge density, green area, ArcGIS

5.1 Introduction

Human population continues to increase, particularly in urban areas which now contain more than half the world’s population (Cohen, 2003; Crane and Kinzig, 2005; Grimm et al., 2008). For Europe, despite decreasing population overall, approximately 80% of the population will be expected to live in cities in the next twenty years (United Nations, 2008). As population density in cities increases, and as cities themselves continue to grow, there are likely to be impacts on the semi-natural elements of urban environments through habitat degradation or loss. This has social implications in that exposure to nature for most of the world’s population (i.e., those living in urban areas) may be increasingly restricted (Crane and Kinzig, 2005). Given such concerns, we should be aware of the potential for urban areas to provide opportunities for human interactions with nature. Indeed, urban areas do have potential to support surprising amounts of biodiversity (Rebele, 1994; Niemela,¨ 1999), but this potential can be enhanced by the structure and composition of urban areas, in other words the urban habitat matrix. Habitat heterogeneity, which has been demonstrated to support species diversity (Di Gulio and Nobis, 2008), can be high in urban environments which may include a variety of forms of suitable habitats for plants and insects. For example, green roofs provide new space to plant and animal species (Brenneisen, 2006; Kadas, 2006; Oberndorfer et al., 2007) while derelict ’brown-ﬁeld’ sites such as inoperative railways, and green areas such as parks and gardens, harbour many animal species (McFrederick and LeBuhn, 2006; Saure, 1996). Building complexes may even 5.1 Introduction 107

change the abiotic conditions locally, which affect animal species abundances, diversity and distribution (Godefroid and Koedam, 2007). Sustainable ecosystems need not only species, but also effectively functioning ecological processes. Pollination is a supporting ecosystem service for both wild plants and many planted crops. Within the urban environment the maintenance of viable populations of flowering plants may therefore depend on insect pollinators that the urban environment supports. Recently there has been growing concern about many pollinator groups which are thought to be declining in intensively managed and transformed landscapes (Biesmeijer et al., 2006; Vamosi et al., 2006). A decline in pollinator abundance and diversity may result in pollen limitation and reduced reproductive success in plants. At the same time, low plant diversity may sustain fewer pollinator species. Urban environments can provide a range of resources for pollinating animals, including nesting sites and nutritional resources (Saure, 1996; Cane et al., 2006; Hopwood, 2008), within, for example, gardens (McFrederick and LeBuhn, 2006; Osborne et al., 2008) and along roadsides (Hopwood, 2008). In addition to the extent of suitable habitats to support pollinators, the foraging ability of insects and hence their potential role as pollinator service providers, depends on the connectivity among suitable habitats in the urban environment (Gilbert et al., 1998; Donaldson et al., 2002; Tewksbury et al., 2002; Townsend and Levey, 2005). Roads and railroads, for example, have been shown to reduce the movement of pollinating animals (Bhattacharya et al., 2003). The urban matrix can therefore play a major role in determining pollinator movement, diversity, and density. To analyze how pollinator communities and processes respond to the urban environment we investigated pollinator diversity and floral visitation with respect to the area of green space and density of habitat edges at different scales on bee and syrphid fly diversity and their patch visit frequency at 89 locations within an urban matrix. Green areas constitute any patch of vegetation within the urban matrix. This may include relatively large areas such as parks, which themselves might consist of several habitat types, or smaller patches such as urban gardens, vegetated strips along roads, vegetated derelict plots, or even green roofs. These green areas often provide breeding, nesting, and foraging sites for bees and syrphids (e.g., Stuke, 1998; Saure, 1996; McFrederick and LeBuhn, 2006). Small green areas might be used as ”stepping stones” on foraging bouts between larger areas (Dover and Settele, 2009; Dearborn and Kark, 2010). Edges are the boundaries between urban land use types, and their quantification is thus subject to the number of land uses recognised and mapped. Linear features and habitat edges in farmland serve as foraging routes for bees and syrphids (Van Geert et al., 2010) and few bee species such as bumblebees choose edges for nest sites (Osborne et al., 2008). There is also evidence that linear features and habitat edges in rural and suburban landscapes might act as a barrier to the 108 Chapter 5

movement of bees and syrphids (Bhattacharya et al., 2003; Wratten et al., 2003). Given this uncertainty, we explore how density of habitat edges at different scales affect the abundance and diversity of the urban pollinator community.

5.2 Materials & Methods

Study Site

The city of Zurich¨ has 380,499 inhabitants (2008) and is the largest city of Switzerland, located in its north-eastern part. Population growth in the city from 1998 to 2008 was 8.2% (Statistisches Amt Zurich,¨ 2008), and is expected to increase as the entire Swiss population is projected to grow by up to 20% in the next 50 years (Giannakouris, 2008). Zurich¨ covers an area of 8,774 ha, of which about 60% is paved (buildings, streets), and 37.5 % consists of forests, parks and agricultural land (Statistisches Amt Zurich,¨ 2008). Few study sites (27 out of 89) were located within the Zurich¨ urban centre, while the rest were located towards the edge of the centre where there are more extensive green areas (Fig. 5.1).

Data Collection

Plant diversity and pollinator visits

Green areas within the city such as parks, green strips along roads, and gardens, were randomly chosen for observations on plants and pollinators between 26th May and 10th September 2008. At each of the 89 plots observations on flower visitors and plant species were conducted once within 2×2 m plots. Most (40.4%) flowering plant patches within which these plots were located were relatively small (up to 1,500 m2) with only few plots located in the midst of a wider expanse of flowering plants. The flowering plant community within each plot was not obviously different in composition to the surrounding patch, though this was not quantitatively assessed. The number and identity of each flowering plant species, and flower visitation frequency and identity of each pollinator visiting the site were recorded. Records were made only for flower visitors landing on flowers and actively searching for rewards. We counted every insect entering the plot as a new individual. While this may overestimate the actual number of individuals visiting the plot, we 5.2 Materials & Methods 109

assume that the consistent application of this rule across all plots allowed unbiased comparison across plots. Observations of pollinator visits lasted 40 minutes between 0900 – 1700 on sunny days. The observer spent 10 minutes at the corner of each plot before moving to the next corner so as to avoid bias due to shading. Familiarization with the identity of flower visiting insects in this region was obtained through extensive field observations and collection as part of earlier studies. Even so, some insects remained difficult to identify to species (including, for example, some Lasioglossum and Bombus bees). Where possible these were caught for later identification. We were able to identify 71% visitors to species, and 92.6% to genus. Specimens were determined using relevant keys (Amiet, 1996; van Veen, 2004; Oosterbroek, 2006). Plant species at each plot were identified (following Ba¨ßler et al., 1996) and the number of flowers of all plants were counted after each observation period. Densely clustered floral heads of Asteraceae, Apiaceae and some flowers of Fabaceae (e.g., Trifolium spp.) were considered as single flowers. Sites that contained only a single flowering plant species were selected at locations, where there were no other heterospecifics at least within 20 meters of the plot, so as to avoid site misinterpretation. Similarly, and so far as casual visual assessment allowed, the 2×2 m plots were positioned to reflect the surrounding flowering community mix.

Landscape Metrics

Two landscape metrics, the extent of green area and edge density were used to quantify spatial structure around each of the 89 patches within the urban environment of Zurich¨ using ArcGIS 9.2 (Environmental Systems Research Institute (ESRI) Inc., 2006). This was done at a variety of spatial scales, as described below. Green area included meadows, grassland, gardens, and parks (but not forests), and was calculated as the proportion of green area (m2) to the total area (m2). Edge density (m ha−1) was defined as the edge length of green areas divided by its total area. We do not distinguish between the types of land use adjacent to green area patches. For each site, values for the two landscape metrics were calculated for multiple scales from 20 to 100 m radius (i.e., 1,260 to 31,500 m2) in 10-meter-steps, and from 100 to 200 meters (i.e., 125,500 m2) in 25-meter-steps. We used a maximum 200 m radius scale from each observational plot, because successful pollen transfer among insect-pollinated plant species has been previously shown to be less than 200 m (mean 130 m) in urban habitats (Van Rossum and Triest, 2010). While it is well known that several bees, particularly honeybees and bumblebees, can cover far greater distances within single foraging bouts, it seems far less probable that pollen is successfully transferred between conspecific plants over such distances, particularly in urban matrices, although beyond the abovementioned study there is little direct evidence to support this contention. Nevertheless, other reasons for limiting the scale to 200 m radius are that, first, 110 Chapter 5

beyond this scale changes in the two landscape metrics were marginal and, second, surveyed areas centered on observation plots began to overlap.

Analysis

Landscape metrics

We performed Spearman‘s rank correlation of the response variables ”diversity” of bees and syrphids, ”patch visits by bees” and ”patch visits by syrphid ﬂies” with the landscape metrics at each distance. The extent of green area as a proportion of the total area was arcsine square root transformed before the analysis. The correlation coefﬁcient ρ of each analysis was plotted against the area analyzed to show the degree of relationship at each scale (i.e., each of the 10–25 m radius increments).

Statistical Analysis

We calculated Shannon diversity index for plants and pollinators. Shannon diversity index accounts for the number of species and their abundance. We used flower number to represent plant abundance. Floral abundance was used independently in the analyses, because it describes resource amount. Though some frequent plant species may influence the attractiveness of patches to flower visitors, we considered the effect of particular plant species as part of the error. We analyzed the diversity and the ratio visitation frequency/floral density of bees and syrphid flies against the landscape metrics, plant diversity, and floral density using multiple linear regression. Interactions between plant diversity and floral abundance with the landscape metrics were used to account for large and local scale effects. For landscape metrics, we chose only the scales with highest R2-values after squaring the correlation coefficients (Dunn-Rankin et al., 2004). As the diversity and the ratio visitation frequency/floral density of syrphids consist of many zero values (54% and 27%, respectively), we performed for these two variables Wilcoxon (exact) tests on edge density and green area at highest R2-values. For significant differences between the means, multiple linear regressions were applied on non-zero values. Explanatory variables were investigated for multicollinearity using variance inflation factors. Variance inflation factors larger than five indicate multicollinearity (Sheater, 2009; Zuur et al., 2009). We removed multicollinearity by centering all explanatory variables around their mean (i.e., substracting each value from the mean). Models were investigated for violation of homodascity 5.3 Results 111

and normality. Transformations of explanatory and response variables were applied to achieve assumptions. We tested for spatial independency of residuals using Moran‘s I following the procedure described in Dormann et al. (2007). Stepwise reduction of non-significant explanatory variables using F-test was applied to find the most parsimonous model. For all calculations the statistical program R was used with the packages vegan (Oksanen et al., 2009) to calculate bee, syrphid, and plant diversity, the package exactRankTests (Hothorn and Hornik, 2006) for the Wilcoxon exact tests, the package faraway (Faraway, 2009) for the investigation of the variance inflation factors, and the packages ncf (Bjornstad, 2009) and spdep (Bivand et al., 2009) to calculate the Moran‘s I.

5.3 Results

Pollinator Species

A total of 2,862 visits by 148 insect species were recorded, averaging to 32 visits per patch. Species accumulation curves for species richness of syrphids and bees approached but did not reach an asymptote at 89 sites observations, the same being the case for plants (Fig. 5.2). Thus while we are confident that we have comprehensively ”sampled” most members of the pollinator and plant communities within the city of Zurich,¨ new rare, transient or occasional flower visitor species are likely to be found. Most plot visits were by Hymenoptera (79.5%) and Diptera (17.5%). The most common visits within each of the two groups were made by Apoidea (98%) and Syrphidae (82.2%), respectively. Apoidea were mostly represented by honeybees (Apis mellifera) (48.7%) and bumblebees (28.6%), among which the most frequent observed species was Bombus pascuorum (18.5%) (for full species list see Tab. 5.1). Among the syrphid flies, the four most common species were Eristalis cf. tenax (23.4%), Episyrphus balteatus (13.1%), Cheilosia scutellata (12.2%), and Sphaerophoria scripta (11.7%). Hymenoptera were the most species-rich group (65 spp., 44%), although the number of Diptera species is approximately equal (61 spp., 41.2%). Apoidea (55 spp., 84.6%) and Syrphidae (34 spp., 55.7%) were the most species rich families within each group. For this reason, both families were independently analysed for the effects of plant diversity, floral abundance and landscape metrics. As honeybees (Apis mellifera L.) accounted for 50.5% of all bee plot visits (1,086 out of 2,150 plot visits) we conducted additionally analyses on bee plot visits excluding honeybees. 112 Chapter 5

Plant Species

We recorded 67 plant species from 19 plant families (Tab. 5.2). The most species-rich plant families were Asteraceae (17 spp.), Fabaceae (11 spp.), and Lamiaceae (8 spp.). Most flowers were counted for Fabaceae (9,965), Asteraceae (4,180), and Lamiaceae (1,133). On average 5.02 ± 2.3 (mean and standard deviation) plant species and 212.2 ± 221.2 flowers were recorded. Maximum plant species number was 12 plant species and maximum floral abundance was 950 flowers in a single plot. Minimum values were one plant species and 18 flowers. The most frequently recorded plant species was Trifolium pratense L. (Fabaceae) (61 out of 89 plots, Fig. 5.3), which had the largest floral abundance (3,164). When considering the number of plots a plant species was observed, Rhinanthus alectophorus (Scop.) Pollich s.l. (Scrophulariaceae) was the species with the largest floral abundance in the study (245.75 flowers/plot).

Landscape Metrics

Edge Density

Edge density explained considerably more of the variation in syrphid diversity and syrphid visit frequency (maximum value 10.7% and 9.7%, respectively, at 80 m radius or ≈ 20,000 m2) than that of bee diversity (maximum value 1.9% at 200 m or ≈ 125,500 m2), bee visit frequency (maximum value 5.2% at 150 m radius or 70,000 m2), and bee visits excluding honeybees (maximum value 0.64% at 60 m or ≈ 11,500 m2). Bee diversity was negatively associated with edge density at the lowest scales (≈ 1,250 m2), and positively thereafter, while bee visits were almost negatively correlated and showed considerable variation in correlation. Similarly bee plot visits excluding honeybees showed considerable variation in correlation with edge density (maximum value of 0.64% at 60 m radius), but were more weakly correlated with edge density than plot visits of all bees. Correlation coefﬁcients of bees were negative below 50 m radius (≈ 7,500 m2), and between 80 - 175 m (i.e., ≈ 20,000 - 95,000 m2), while otherwise positive (Fig. 5.4). By contrast, edge density positively inﬂuenced syrphid diversity and visitation frequency at all scales, reaching an asymptote at 100 m radius (≈ 31,500 m2) for diversity and visitation frequency (Fig. 5.5). 5.4 Bees and Syrphid Flies 113

Green Area

As with edge density, the extent of green area explains far more of the variation in syrphid diversity (maximum 16.5% at 200 m radius) than for bee diversity (maximum 3.4% at 100 m radius), and similarly for syrphid visit frequency (maximum 15.7% at 200 m radius) than in bee visit frequency (maximum 0.2% at 200 m radius). There was always a positive effect of green area on bee and syrphid diversity, and syrphid visits (Fig. 5.4 – 5.5). Bee visits were negatively correlated with green area at radii below 30 m (≈ 2,500 m) and above 100 m (≈ 31,500 m2). Excluding honeybee visits correlation was only negative at the lowest scale (20 m radius) and increased thereafter with scale, reaching an asymptote at 60 m radius (≈ 11,000 m2, Fig. 5.4). Almost all curves followed an asymptotic form with the exception of bee visits, which showed considerable variation at scales below 100 m radius. For bee diversity there was a levelling off at 100 m radius scale (≈ 31,500 m2), while this occurs more gradually and at larger spatial scales for syrphid ﬂies (Fig. 5.5).

5.4 Bees and Syrphid Flies

Bees

Bee diversity increased significantly with plant diversity (t = 4.3, p < 0.001) and green area size (t = 2.5, p = 0.01). We found bee abundance to increase also significantly with plant diversity (t = 3.93, p < 0.001), green area size (t = – 2.14, p = 0.04) and floral abundance (t = 4.0, p < 0.001), while edge density affected negatively bee visits (t = – 2.4, p = 0.02). Moran‘s I values for the model residuals of bee diversity (Moran‘s I = – 0.51, p = 0.7) and bee visits (Moran‘s I = – 1.01, p = 0.84) showed no significant spatial pattern of residuals, which indicated no effect of spatial autocorrelation on parameter estimation. Removing honeybees, bee plot visits increased with plant diversity (t = 4.2, p < 0.001), floral abundance (t = 3.3, p = 0.001), and green area size (t = 2.15, p = 0.03). There was no spatial autocorrelation of model residuals (Moran‘s I = – 0.65, p = 0.74). 114 Chapter 5

Syrphids

We found highly significant differences between sites with and without syrphid diversity when accounting for the size of green area at 200 m radius (W = 1254, p < 0.001) and edge density at 80 m radius (W = 1206.5, p = 0.002). Similarly, there was a significant difference between field sites with and without syrphid visits when considering the size of green area at 200 m radius (W = 960, p = 0.02). There was no significant difference between field sites with and without syrphid visits considering edge density at 80 m radius (W = 893, p = 0.08). For both syrphid diversity and plot visits there were larger mean values for edge density and green area size when syrphids were observed (Tab. 5.3). Syrphid diversity increased with floral abundance and green area (t = 2.22, p = 0.03), but there was no effect of floral abundance (t = 0.51, p = 0.61) and green area (t = 1.29, p = 0.21) alone on syrphid diversity. More syrphids visited plots with larger plant diversity (t = 2.11, p = 0.04) and with increasing size of the surrounding green area (t = 2.3, p = 0.03). With increasing number of flowers, however, fewer syrphids visited the plots (t = – 2.24, p = 0.03). As with bees, models for syrphid diversity (Moran‘s I = 1.36, p = 0.09) and syrphid visits (Moran‘s I = 1.14, p = 0.13) showed no significant spatial residual pattern.

5.5 Discussion

Scale effects of landscape metrics

Landscape factors influenced diversity and visits of bees and syrphids at flowering patches at different spatial scales. In almost all cases, the correlation between diversity or visit frequency of both pollinator groups with the landscape metrics of extent of green area and edge density became stronger at larger scales. In semi-natural and natural landscapes correlation coefficients of bee species richness as well as visits with proportion of seminatural habitats also increased with scale (Steffan-Dewenter and Tscharntke, 2002). This reflects that the variation in bee diversity and visits can be explained by the extent of green area and edge density at larger scales than 200 m radius even in urban environments, although other studies reported reduced foraging ranges of bees in cities (Lopez-Uribe et al., 2008; Van Rossum and Triest, 2010). Given, however, the large foraging ranges of bees (Gathmann and Tscharntke, 2002; Darvill et al., 2004; Knight et al., 2005), green areas can function as habitats for foraging and as ”stepping stones” on long-distant foraging bouts (Dover and Settele, 2009; Dearborn and Kark, 2010), while edges of green areas 5.5 Discussion 115

can be expected to serve as foraging routes (Osborne et al., 2008). Our results provide also evidence that the variation of syrphid density and diversity in urban environments is more explained at larger (i.e., above 100 m radius or ≈ 31,500 m2) than smaller scales. These results corroborate earlier studies (Sommaggio, 1999; Meyer et al., 2009) and can be referred to the response of the five most common syrphid species Episyrphus balteatus, Eristalis tenax, Sphaerophoria scripta, Syrphus ribesii, and Syritta pipiens in our study (which account for 59.6% of syrphid floral visits and 58.1% of recorded syrphid plot visits (Tab. 5.1)). These species travel distances of several kilometres while migrating in natural environments (Gatter and Schmid, 1990) and are known synanthropic species (Bankowska,´ 1980). Although there is little information on foraging ranges of these syrphids (and syrphids in general) in urban environments, the potential for long-distance movement exists and that explains larger correlation coefficients with increasing scale in our study. The stronger positive response of syrphid diversity (compared to bee diversity) to landscape metrics over the entire scale (Fig. 5.4 - 5.5) can be also explained by the different biology of syrphids and bees. In contrast to bees parental care is not known from syrphids (van Veen, 2004). Additionally, syrphid larvae depend on different resources than adults and live in different habitats (Sommaggio, 1999; van Veen, 2004). Areas with increased mosaic richness of habitats and landscape structures promote therefore syrphid species richness and abundance (Haenke et al., 2009), which is more likely to be found in larger green areas and with a larger density of green area edges.

Edge Density of Green Area

For bees (unlike syrphids), as scale increases the correlation of species diversity against edge density continues to increase. This suggests that as habitats are increasingly dissected at larger scales a wider range of bee species are observed at the local patch scale. Presumably, this is due to the larger ranges of many bees (in contrast to syrphids) such that larger scales better reflect the habitat structure that supports wide ranging bees. Thus, bee diversity is a function of large scale habitat structure. On the other hand, green edges provide nesting sites for bees (Osborne et al., 2008). An increased edge density offers therefore opportunities of breeding for many bee species, which can explain the positive correlation. For bee visits, the scale relationships expressed by the results are apparently explained by the different spatial area over which different bees forage (Gathmann and Tscharntke, 2002; Greenleaf et al., 2007). At smaller scales different responses of large and small foraging bees to edge density are reflected by larger variation in the correlation coefficient. Removing honeybees (densities of which are strongly dependent on urban honeybee-hives) reduced the variation and made the 116 Chapter 5

correlation positive at scales between 50 - 70 m radius. Although the correlation coefficients still varied considerably, we conjecture that the variation is due to species specific responses of bees to edge density, although there is little information on such specific responses of bee species to urban green edges. An increase in the edge density/syrphid visitation rate correlation coefficient with scale is obviously explained by the importance of edges for syrphids. Syrphids require different microhabitats for larval and adult growth (Sommaggio, 1999; van Veen, 2004). Flowering trees and shrubs provide larval food and resources. In urban environments trees and shrubs often grow or are transplanted along edges of green areas (Lundholm and Marlin, 2006; Hopwood, 2008), while the green area core is most often a sword used for recreational activities. The increasing correlation thus reflected larger resource amount with increasing edge density of green areas. Edges also facilitate movement (hence visits) of syrphids that forage over larger areas, at least up to 20,000 m2 after which there is no further improvement in the correlation of edge density with increasing scale. Syrphids have been shown to be sensitive to different barriers (Wratten et al., 2003), which may differ in the extent to which they influence syrphid movements. Responses of edges on movement of insects across the landscape can be, however, species specific. As no direct comparison of edge types or their effects on insect activity has been made within urban environments, we know little about the specific responses of syrphids to these habitat elements. In contrast to the abovementioned explanations, which refer to the biology of bees and syrphids, the shape of areas can also determine movement patterns (Ims, 1995, and references therein). Green areas with higher edge densities have also an irregular shape with concave and convex boundaries. Curved boundaries can act as channels for emigration and immigration (in Ims (1995): Hanski and Peltonen, 1988; Hardt and Forman, 1989), promoting bee and syrphid movement. This is obviously supported by our results as there was positive correlation between green area edge density and syrphid visits. In contrast, bee visits were negatively correlated over the entire scale (i.e., from 20 - 200 m radius) with the exception that honeybees were removed. Presumably honeybees prefer open spaces, which location may be easier to communicate in the hive (Steffan-Dewenter and Kuhn, 2003).

Green Area

The correlation for bee and syrphid diversity with green area was positive over all scales indicating a positive relationship of green area with diversity of these groups, although the strength of the relation levels off at a radius of 100 m (≈ 31,500 m2) for bee diversity, while for syrphid diversity it continues to improve. Thus, syrphid diversity is more strongly related to the extent of green area than bee diversity. 5.5 Discussion 117

The correlation pattern between the visit frequency of syrphids and green area resembles the results for syrphid diversity with a maximum value at the largest scale (≈ 125,500 m2), though the curve continues to rise more steeply than in syrphid diversity. In other words, as the extent of green area increases at larger scales both syrphid diversity and visits increases. These responses may be related in that higher syrphid diversity may itself increase visitation frequency as different ﬂowers partition resources. In contrast, bee visit frequency showed much variability at scales between 1,250 - 50,000 m2 (i.e., 20 - 125 m radius), but tended to level off thereafter. The variability refers to visits by honeybees, because the correlation of bee plot visits were similar to results for bee diversity after removing honeybees. For honeybees presumably other factors are more important in explaining the variability, including ﬂower composition or the management of the green area (Brown and Freitas, 2002; Smith et al., 2006). Moreover, honeybees can forage over large distances (Goulson, 2003), which is why they can be affected by other factors not accounted for by the spatial scale of our study. Finally, honeybees are usually kept in hives and might be therefore buffered against negative impacts of urbanity such as fragmentation and habitat alteration, which could affect the search for adequate nesting sites. For wild bees, however, a positive correlation with green area indicates the importance of green patches as foraging and nesting places. In areas with large support of green area structures such as parks and gardens, more wild bee species can be therefore expected.

Local and large scale factors

We found that plant diversity and floral abundance are positively correlated on bee diversity and abundance, and syrphid visits. More diverse plant patches attract a wider range of floral visitors on account of the wider range of resources they offer (Potts et al., 2004; Ghazoul, 2006). In rural or natural settings it is not unexpected that flower visitor diversity is correlated with plant diversity (e.g., Biesmeijer et al., 2006; Hegland and Boeke, 2006; Ebeling et al., 2008), but it is perhaps more surprising to note that this correlation also holds true in a highly transformed urban setting, where floral resource patches are more limited in both number and extent. More limited flowering patch availability might be expected to undermine patterns of resource selection and differentiation by pollinators (i.e., pollinators are forced to make do with what they get), leading to a breakdown in the correlation between resource variety and pollinator diversity. This does not, however, appear to be the case for bees and syrphids, which suggests that the promotion of pollinator diversity within urban settings needs to consider not only the number of green patches at large scales, but also their floral composition at patch scales. In contrast to floral diversity, floral abundance (independently of diversity) had no effect on bee 118 Chapter 5

diversity. The lack of any relation between bees and floral abundance can be attributed to the foraging ranges of honeybees and bumblebees which can exceed several hundred meters at least in rural settings (Gathmann and Tscharntke, 2002). This species account for the majority of floral visits by bees. While foraging ranges are less clear in urban centers, it is feasible that honeybees briefly use green areas such as lawns, gardens or meadows as stepping stones between the nest and more rewarding sites which might even lie beyond the periphery of a city such as Zurich.¨ In such a case, floral visitation by honeybees is less a function of patch quality but rather whether it lies along the path between the nest and a larger resource area (Angold et al., 2006; Dover and Settele, 2009; Dearborn and Kark, 2010). In contrast to bee species, syrphid species were influenced by the number of flowers in the plots as might be expected for insects that have smaller foraging ranges and therefore are more dependent on patches within a smaller locality. A greater number of syrphid species in plots with larger floral abundance can be simply attributed to greater availability of resources (Sih and Baltus, 1987; MacLeod, 1999). Syrphids also have a range of habitat requirements for breeding and foraging (Sommaggio, 1999; van Veen, 2004): larvae of some species occur in tree holes, in muddy ponds, or hunt on aphids, while most adult syrphids forage on flowers or suck plant sap. We expect that larger green areas will provide this diversity of breeding requirements and with increasing flower number would have attracted more syprhids. We conjecture that the negative effect of increasing floral density and green area on syrphid visits is referred to larger resource availability with increasing patch size. The likelihood for finding more rewarding floral spots increases with patch size and fewer syrphid visits could have been then observed at a single floral spot. This study emphasizes the importance of green area for diversity and visits of bees and syrphids in our urban environment. Our results showed that green area was the only significant variable which describes the urban matrix and affects bees and syrphids. In particular, bee diversity and visits were responding positively on green area size. This is in line with other urban studies reporting an increase of bee species with size of green structures such as parks (McFrederick and LeBuhn, 2006; Ahrne´ et al., 2009). The positive effect of green area on bees and syrphids can be attributed to the provision of food in terms of floral resources and nesting sites. Edge density of green areas was the second important variable describing our urban environment affecting bee visits. Although edges along green areas contain many flowering shrubs and trees which potentially provide nesting and foraging sites for bees, we found no evidence to support this. On the contrary, increasing extent of edges was associated with reduced plot visits by bees. This may be largely due to the negative affect of edges on visitation by honeybees because excluding honeybees from the analysis resulted in no significant effect of edge density on bee visitation frequency. This suggests that honeybee presumably prefer more open areas for foraging. 5.6 References 119

Conclusion

The spatial factors (i.e., landscape metrics) of edge density and the extent of green area showed a positive impact on bee and syrphid fly diversity, determining presumably their foraging behavior and potential services to plant species within the urban environment. In addition, the importance of these metrics varied depending on the scale of analysis with the most appropriate scales being between 80 and 100 m radius (≈ 20,000 m2) and above 150 m (≈ 70,000 m2) for diversity as well as visitation frequency of bees and syrphids, respectively. The local factors plant diversity and flower abundance were almost positively affecting bee diversity and visits, while we found positive effects of plant diversity and floral abundance on syrphid diversity, while less syrphid visits were recorded with increasing floral abundance. The different effects on these two flower visitor groups emphasize the need to consider conservation schedule independently for each. The variation of the correlation between the landscape metrics and visits as well as diversity of bees and syrphids with scale suggests that finding appropriate schedules for realizing conservation of pollinator species within the urbanity needs to consider the spatial scale. These results are derived from a single study of a small city. Generalizing from these results to other urban areas needs to be undertaken with caution as there is no replicability at the level of the city. It is possible that these responses differ among cities depending on the overall size (and therefore proximity to semi-natural and rural areas), levels of pollution, intensity of human activity or other factors (Schwarz, 2010). There remains a need for a more comprehensive study conducted over several cities that vary in size before broader conclusions can be made with confidence.

5.6 References

Ahrne,´ K., Bengtsson, J., and Elmqvist, T. (2009). Bumble Bees (Bombus spp.) along a Gradient of Increasing Urbanization. PLoS One (Public Library of Science Biology), 4(5):e5574.

Amiet, F. (1996). HYMENOPTERA. APIDAE, 1. Teil, volume 12 of INSECTA HELVETICA. Schweizerische Entomologische Gesellschaft.

Bankowska,´ R. (1980). Fly communities of the family Syrphidae in natural and anthropogenic habitats of Poland. Memorabilia Zoologica, 33:3–93.

Ba¨ßler, M., Jager,¨ E., and Werner, K. (1996). ROTHMALER Exkursionsﬂora von Deutschland. Gefa¨ßpﬂanzen: Grundband, volume 2. Gustav Fischer, 16 edition.

Bhattacharya, M., Primack, R. B., and Gerwein, J. (2003). Are roads and railroads barriers to bumblebee movement in a temperate suburban conservation area? Biological Conservation, 109(1):37–45.

Biesmeijer, J. C., Roberts, S. P. M., Reemer, M., Ohlemuller, R., Edwards, M., Peeters, T., Schaffers, A. P., Potts, S. G., Kleukers, R., Thomas, C. D., Settele, J., and Kunin, W. E. (2006). Parallel declines in pollinators and insect-pollinated plants in Britain and the Netherlands. Science, 313(5785):351–354.

Bivand, R., with contributions by Luc Anselin, ao, R. A., Berke, O., Bernat, A., Carvalho, M., Chun, Y., Christensen, B., Dormann, C., Dray, S., Halbersma, R., Krainski, E., Lewin-Koh, N., Li, H., Ma, J., Millo, G., Mueller, W., Ono, H., Peres-Neto, P., Reder, M., Tiefelsdorf, M., and Yu, D. (2009). spdep: Spatial dependence: weighting schemes, statistics and models. R package version 0.4-52.

Bjornstad, O. N. (2009). ncf: spatial nonparametric covariance functions. R Cran – The R Project for Statistical Computing. R package version 1.1-3.

Brenneisen, S. (2006). Space for Urban Wildlife: Designing Green Roofs as Habitats in Switzer- land. Urban Habitats, 4(1):27–36.

Brown, K. S. and Freitas, A. V. L. (2002). Butterﬂy communities of urban forest fragments in Campinas, Sao˜ Paulo, Brazil: Structure, instability, environmental correlates, and conservation. Journal of Insect Conservation, 6(4):217–231.

Cane, J. H., Minckley, R. L., Kervin, L. J., Roulston, T. H., and Williams, N. M. (2006). Complex responses within a desert bee guild (Hymenoptera : Apiformes) to urban habitat fragmentation. Ecological Applications, 16(2):632–644.

Cohen, J. E. (2003). Human population: The next half century. Science, 302(5648):1172–1175.

Crane, P. and Kinzig, A. (2005). Nature in the metropolis. Science, 308(5726):1225–1225.

Darvill, B., Knight, M. E., and Goulson, D. (2004). Use of genetic markers to quantify bumblebee foraging range and nest density. Oikos, 107(3):471–478. 5.6 References 121

Dearborn, D. C. and Kark, S. (2010). Motivations for Conserving Urban Biodiversity. Conserva- tion Biology, 24(2):432–440.

Di Gulio, M. and Nobis, M. (2008). Landschaftszerschneidung und Biodiversitat:¨ Barrieren oder Ausbreitungswege? Forum fur¨ Wissen, pages 23–30.

Dormann, C. F., Schweiger, O., Augenstein, I., Bailey, D., Billeter, R., de Blust, G., DeFilippi, R., Frenzel, M., Hendrickx, F., Herzog, F., Klotz, S., Liira, J., Maelfait, J. P., Schmidt, T., Speelmans, M., van Wingerden, W. K. R. E., and Zobel, M. (2007). Effects of landscape structure and land-use intensity on similarity of plant and animal communities. Global Ecology and Biogeography, 16(6):774–787.

Dover, J. and Settele, J. (2009). The inﬂuences of landscape structure on butterﬂy distribution and movement: a review. Journal of Insect Conservation, 13(1):3–27.

Dunn-Rankin, P., Knezek, G. A., Wallace, S., and Zhang, S. (2004). Scaling methods. Lawrence Erlbaum Associates, 2nd edition edition.

Environmental Systems Research Institute (ESRI) Inc. (2006). ArcGIS.

Faraway, J. (2009). faraway: Functions and datasets for books by Julian Faraway. R package version 1.0.4.

Gathmann, A. and Tscharntke, T. (2002). Foraging ranges of solitary bees. Journal of Animal Ecology, 71(5):757–764.

Gatter, W. and Schmid, U. (1990). Die Wanderungen der Schwebﬂiegen (Diptera, Syrphidae) am Randecker Maar. Spixiana, 15:9–100.

Gelman, A. and Hill, J. (2007). Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press.

Ghazoul, J. (2006). Floral diversity and the facilitation of pollination. Journal of Ecology, 94:295–304. 122 Chapter 5

Giannakouris, K. (2008). Population and social conditions. Technical Report 72, EUROSTAT. European Comission.

Godefroid, S. and Koedam, N. (2007). Urban plant species patterns are highly driven by density and function of built-up areas. Landscape Ecology, 22(8):1227–1239.

Goulson, D. (2003). Effects of introduced bees on native ecosystems. Annual Review of Ecology, Evolution and Systematics, 34:1–26.

Greenleaf, S. S., Williams, N. M., Winfree, R., and Kremen, C. (2007). Bee foraging ranges and their relationship to body size. Oecologia, 153(3):589–596.

Grimm, N. B., Foster, D., Groffman, P., Grove, J. M., Hopkinson, C. S., Nadelhoffer, K. J., Pataki, D. E., and Peters, D. P. C. (2008). The changing landscape: ecosystem responses to urbanization and pollution across climatic and societal gradients. Frontiers in Ecology and the Environment, 6(5):264–272.

Haenke, S., Scheid, B., Schaefer, M., Tscharntke, T., and Thies, C. (2009). Increasing syrphid ﬂy diversity and density in sown ﬂower strips within simple vs. complex landscapes. Journal of Applied Ecology, 46(5):1106–1114.

Hanski, I. and Peltonen, A. (1988). Island colonization and peninsulas. Oikos, 51(1):105–106.

Hardt, R. A. and Forman, R. T. T. (1989). Boundary form effects on woody colonization of reclaimed surface mines. Ecology, 70(5):1252–1260.

Hopwood, J. L. (2008). The contribution of roadside grassland restorations to native bee conservation. Biological Conservation, 141(10):2632–2640.

Hothorn, T. and Hornik, K. (2006). exactRankTests: Exact Distributions for Rank and Permuta- tion Tests. R package version 0.8-18.

Ims, R. A. (1995). Movement patterns related to spatial structures. In Hannsson, L., Fahrig, L., and Merriam, G., editors, Mosaic Landscappes and Ecological Processes, volume 2 of IALE Studies in Landscape Ecology, chapter 4, pages 85–109. Chapman & Hall. 5.6 References 123

Kadas, G. (2006). Rare Invertebrates Colonizing Green Roofs in London. Urban Habitats, 4(1):66–86.

Lopez-Uribe, M. M., Oi, C. A., and Del Lama, M. A. (2008). Nectar-foraging behavior of Euglossine bees (Hymenoptera : Apidae) in urban areas. Apidologie, 39(4):410–418.

Lundholm, J. T. and Marlin, A. (2006). Habitat origins and microhabitat preferences of urban plant species. Urban Ecosystems, 9:139–159. 10.1007/s11252-006-8587-4.

MacLeod, A. (1999). Attraction and retention of Episyrphus balteatus DeGeer (Diptera : Syrphidae) at an arable ﬁeld margin with rich and poor ﬂoral resources. Agriculture Ecosystems & Environment, 73(3):237–244.

McFrederick, Q. S. and LeBuhn, G. (2006). Are urban parks refuges for bumble bees Bombus spp. (Hymenoptera : Apidae)? Biological Conservation, 129(3):372–382.

Meyer, B., Jauker, F., and Steffan-Dewenter, I. (2009). Contrasting resource-dependent responses of hoverﬂy richness and density to landscape structure. Basic and Applied Ecology, 10(2):178– 186.

Niemela,¨ J. (1999). Ecology and urban planning. Biodiversity and Conservation, 8(1):119–131.

Oosterbroek, P. (2006). The European Families of the Diptera. KNNV Publishing.

Osborne, J. L., Martin, A. P., Shortall, C. R., Todd, A. D., Goulson, D., Knight, M. E., Hale, R. J., and Sanderson, R. A. (2008). Quantifying and comparing bumblebee nest densities in gardens and countryside habitats. Journal of Applied Ecology, 45(3):784–792. 124 Chapter 5

Quinn, G. P. and Keough, M. J. (2002). Experimental Design and Data Analysis for Biologists. Cambridge University Press.

Rebele, F. (1994). Urban ecology and special features of urban ecosystems. Global Ecology and Biogeography Letters, 4(6):173–187.

Schwarz, N. (2010). Urban form revisited–Selecting indicators for characterising European cities. Landscape and Urban Planning, 96(1):29 – 47.

Sheater, S. J. (2009). A Modern Approach to Regression with R. SPRINGER VERLAG.

Sih, A. and Baltus, M. S. (1987). Patch size, pollinator behavior, and pollinator limitation in Catnip. Ecology, 68(6):1679–1690.

Sommaggio, D. (1999). Syrphidae: can they be used as environmental bioindicators? Agriculture Ecosystems & Environment, 74(1-3):343–356.

Statistisches Amt Zurich¨ (2008). Medienmitteilung des Statistischen Amtes. Rekordhohes Bevolkerungswachstum¨ im Kanton Zurich.¨ Technical report, Kanton Zurich.¨

Steffan-Dewenter, I. and Kuhn, A. (2003). Honeybee foraging in differentially structured landscapes. Proceedings of the Royal Society of London Series B - Biological Sciences, 270(1515):569–575.

Steffan-Dewenter, I. and Tscharntke, T. (2002). Insect communities and biotic interactions on fragmented calcareous grasslands - a mini review. Biological Conservation, 104(3):275–284.

Townsend, P. A. and Levey, D. J. (2005). An experimental test of whether habitat corridors affect pollen transfer. Ecology, 86(2):466–475.

United Nations (2008). World urbanization prospects: The 2007 revision population database. Technical report, United Nations.

Vamosi, J. C., Knight, T. M., Steets, J. A., Mazer, S. J., Burd, M., and Ashman, T. L. (2006). Pollination decays in biodiversity hotspots. Proceedings of the National Academy of Sciences of the United States of America, 103(4):956–961.

Van Geert, A., Van Rossum, F., and Triest, L. (2010). Do linear landscape elements in farmland act as biological corridors for pollen dispersal? Journal of Ecology, 98(1):178–187.

Wratten, S. D., Bowie, M. H., Hickman, J. M., Evans, A. M., Sedcole, J. R., and Tylianakis, J. M. (2003). Field boundaries as barriers to movement of hover ﬂies (Diptera : Syrphidae) in cultivated land. Oecologia, 134(4):605–611.

Zuur, A. F., Ieno, E. N., Walker, N. J., Saveliev, A. A., and Smith, G. M. (2009). Mixed Effects Models and Extensions in Ecology with R. Springer Verlag. 126 Chapter 5

5.7 Tables and Figures

Table 5.1: Pollinator species, their number of plot visits (sum) and number of plots, where hey have been recorded during the study in the city Zurich¨ in 2008 (count).

Species group family sum count Andrena flaviceps Hymenoptera Apidae 19 4 Andrena flava Hymenoptera Apidae 1 1 Andrena haemorrhoa Hymenoptera Apidae 16 4 Andrena minutula Hymenoptera Apidae 1 1 Anthidium species Hymenoptera Apidae 1 1 Anthidium florentinum Hymenoptera Apidae 4 1 Anthidium manicatum Hymenoptera Apidae 3 2 Antidium septendentatum Hymenoptera Apidae 1 1 Anthidium strigatum Hymenoptera Apidae 2 1 Anthophora species Hymenoptera Apidae 4 4 Anthophora plumipes Hymenoptera Apidae 6 3 Apis mellifera Hymenoptera Apidae 1,086 51 Bombus hortorum Hymenoptera Apidae 11 3 Bombus humilis Hymenoptera Apidae 44 18 Bombus lapidarius1 Hymenoptera Apidae 5 1 Bombus lapidarius2 Hymenoptera Apidae 54 20 Bombus lucorum Hymenoptera Apidae 45 11 Bombus pascuorum Hymenoptera Apidae 412 62 Bombus pratorum Hymenoptera Apidae 45 5 Bombus ruderarius Hymenoptera Apidae 9 4 Bombus subterraneus Hymenoptera Apidae 1 1 Bombus sylvarum Hymenoptera Apidae 11 6 Chelostoma florisomne Hymenoptera Apidae 16 6 Colletes similis Hymenoptera Apidae 1 1 Halictus species Hymenoptera Apidae 1 1 Halictus scabiosae Hymenoptera Apidae 47 5 Halictus subauratus Hymenoptera Apidae 17 7 Heriades truncorum Hymenoptera Apidae 2 1 Hylaeus communis Hymenoptera Apidae 35 10 Hylaeus nigritus Hymenoptera Apidae 1 1 Hylaeus signatus Hymenoptera Apidae 1 1 Hylaeus sinuatus Hymenoptera Apidae 1 1 Lasioglossum cf. albipes Hymenoptera Apidae 7 2 L. calceatum Hymenoptera Apidae 5 5 L. clypeare Hymenoptera Apidae 1 1 L. cf. morio Hymenoptera Apidae 32 9 L. cf. nitidulum Hymenoptera Apidae 13 1 L. pauxillum Hymenoptera Apidae 2 1 L. species1 Hymenoptera Apidae 140 30 L. species2 Hymenoptera Apidae 8 8 L. species3 Hymenoptera Apidae 2 1 L. species4 Hymenoptera Apidae 1 1 L. smaragdulus Hymenoptera Apidae 3 2 L. xanthopus Hymenoptera Apidae 3 1 Megachile ericetorum Hymenoptera Apidae 59 10 Megachile centuncularis Hymenoptera Apidae 4 2 Megachile nigriventris Hymenoptera Apidae 2 1 5.7 Tables and Figures 127

Megachile species1 Hymenoptera Apidae 4 1 Megachile species2 Hymenoptera Apidae 2 1 Osmia species1 Hymenoptera Apidae 2 1 Osmia species2 Hymenoptera Apidae 5 1 Osmia truncorum Hymenoptera Apidae 4 3 Osmia caerulescens Hymenoptera Apidae 26 10 Sphecodes species Hymenoptera Apidae 2 2 Stelis signata Hymenoptera Apidae 1 1 Aulacus species1 Hymenoptera Aulacidae 29 3 Aulacus species2 Hymenoptera Aulacidae 1 1 Braconinae species Hymenoptera Braconidae 2 2 Cephus pygmaeus Hymenoptera Cephidae 1 1 Cerceris rybyensis Hymenoptera Crabronidae 6 4 Eumeninae species Hymenoptera Euminidae 1 1 Ephialtes manifestator Hymenoptera Ichneumonidae 1 1 Ichneumon suspiciosus Hymenoptera Ichneumonidae 1 1 Athalia rosae Hymenoptera Tenthredinidae 2 2 Polistes galicus Hymenoptera Vespidae 1 1 Hylemyza species Diptera Anthomyiidae 4 3 Bibio marci Diptera Bibionidae 7 3 Dilophus febrilis Diptera Bibionidae 1 1 Calliphorida species Diptera Calliphoridae 2 2 Lucilia species Diptera Calliphoridae 5 3 Lucilia small Diptera Calliphoridae 2 2 Lucilia big Diptera Calliphoridae 3 2 Lucilia bronze Diptera Calliphoridae 8 1 Pollenia rudis Diptera Calliphoridae 2 1 Chrysoperla carnea Diptera Chrysopidae 1 1 Physocephala vittata Diptera Conopidae 1 1 Sicus ferrugineus Diptera Conopidae 11 4 Empis tessellata Diptera Empididae 1 1 Empis species Diptera Empididae 6 5 Hybobatidae species Diptera Hybobatidae 1 1 Musca species Diptera Muscidae 1 1 Sarcophaga carnaria Diptera Sarcophagidae 19 12 Cheilosia caerulescens Diptera Syrphidae 3 2 Cheilosia impressa Diptera Syrphidae 1 1 Cheilosia longula Diptera Syrphidae 2 1 Cheilosia scutellata Diptera Syrphidae 50 5 Cheilosia species Diptera Syrphidae 3 2 Cheilosia soror Diptera Syrphidae 11 1 Chrysotoxum bicintum Diptera Syrphidae 2 2 Criorhina asilica Diptera Syrphidae 1 1 Episyrphus balteatus Diptera Syrphidae 54 25 Eristalis arbustorum Diptera Syrphidae 5 4 Eristalis pertinax Diptera Syrphidae 3 2 Eristalis tenax Diptera Syrphidae 96 27 Eupeodes lapponicus Diptera Syrphidae 4 3 Eupeodes species Diptera Syrphidae 1 1 Ferdinandea cuprea Diptera Syrphidae 1 1 Fischeria bicolor Diptera Syrphidae 2 1 Heliophilus hybridum Diptera Syrphidae 4 1 Heliophilus pendulus Diptera Syrphidae 3 3 Heliophilus trivittatus Diptera Syrphidae 9 3 Melanostoma scalare Diptera Syrphidae 12 8 128 Chapter 5

Merodon equestris Diptera Syrphidae 2 2 Myatropa ﬂorea Diptera Syrphidae 7 6 Parasyrphus species Diptera Syrphidae 1 1 Pipizella pennina Diptera Syrphidae 26 7 Platycheirus albimanus Diptera Syrphidae 5 5 Rhingis campestris Diptera Syrphidae 1 1 Scaevia selenitica Diptera Syrphidae 1 1 Sphaerohoria scripta Diptera Syrphidae 48 23 Syrphid species Diptera Syrphidae 1 1 Syrphus species Diptera Syrphidae 3 2 Syrphus ribesii Diptera Syrphidae 27 10 Syritta pipiens Diptera Syrphidae 20 12 Syritta species Diptera Syrphidae 1 1 Volucella bombylans Diptera Syrphidae 1 1 Dinera species Diptera Tachinidae 2 2 Phasia obesa Diptera Tachinidae 1 1 Phasia pusilla Diptera Tachinidae 1 1 Siphona geniculata Diptera Tachinidae 2 1 Tachina fera Diptera Tachinidae 3 2 Tachinidae Species1 Diptera Tachinidae 1 1 Tachinidae Species2 Diptera Tachinidae 1 1 Diptera Species1 Diptera 1 1 Diptera Species2 Diptera 1 1 Diptera Species3 Diptera 1 1

Nemophora metallica Lepidoptera Adelidae 4 1 Pyrausta aurata Lepidoptera Crambidae 2 2 Thymelicus sylvestris Lepidoptera Hesperiidae 1 1 Polyommatus icarus Lepidoptera Lycaenidae 19 9 Autographa gamma Lepidoptera Noctuidae 2 1 Cerapteryx graminis Lepidoptera Noctuidae 8 1 Hypaena proboscidalis Lepidoptera Noctuidae 1 1 Noctuidae species Lepidoptera Noctuidae 1 1 Maniola jurtina Lepidoptera Nymphalidae 1 1 Pieris rapae Lepidoptera Pieridae 11 5 Zygaena ﬁlipendulae Lepidoptera Zygaenidae 2 1

Anthaxia helvetica Coleoptera Buprestidae 1 1 Anthaxia millefolii Coleoptera Buprestidae 7 1 Chrysomelidae Species Coleoptera Chrysomelidae 3 1 Leptura rubra Coleoptera Cerambycidae 2 1 Scymnus species Coleoptera Coccinellidae 1 1 Variimorda villosa Coleoptera Mordellidae 1 1 Oedemera species Coleoptera Oedemeridae 1 1 Oedemera virescens Coleoptera Oedemeridae 11 6 Trichius fasciatus Coleoptera Scarabaeidae 4 2

Adelphocoris seticornis Heteroptera Miridae 1 1 Lygus pratensis Heteroptera Miridae 2 2

TOTAL 5 36 2,862 148 5.7 Tables and Figures 129

Table 5.2: Plant species and their families found in the urban study.

Species Family Count Floral Ab.

Achillea millefolium L. Asteraceae 5 208 Anchusa officinalis L. Boraginaceae 1 120 Anthemis tinctoria L. Asteraceae 1 82 Anthyllis vulneraria L. s. l. Fabaceae 3 64 Bellis perennis L. Asteraceae 25 1,716 Campanula cervicaria L. Campanulaceae 1 13 Campanula rapunculoides L. Campanulaceae 4 61 Centaurea jacea L. s. l. Asteraceae 19 383 Cerastium fontanum Baumg. s. str. Caryophyllaceae 12 262 Cichorium intybus L. Asteraceae 2 14 Cirsium arvense (L.) Scop. Asteraceae 2 33 Convolvulus arvensis L. Convolvulaceae 1 1 Crepis biennis L. Asteraceae 7 36 Crepis praemorsa (L.) Walther Asteraceae 2 9 Crepis tectorum L. Asteraceae 1 4 Daucus carota L. Apiaceae 8 208 Dianthus carthusianorum L. Caryophyllaceae 1 7 Echium vulgare L. Boraginaceae 1 198 Epilobium hirsutum L. Onagraceae 1 99 Epilobium parviflorum Schreb. Onagraceae 1 3 Erigeron annuus (L.) Pers. Asteraceae 9 144 Galega officinalis L. Fabaceae 1 2 Galeopsis ladanum L. Lamiaceae 1 14 Galium album Mill. Rubiaceae 25 333 Galium verum L. s. str. Rubiaceae 1 3 Geranium molle L. Geraniaceae 6 92 Geranium pyrenaicum Burm. f. Geraniaceae 1 5 Geum urbanum L. Rosaceae 3 40 Glechoma hederacea L. Lamiaceae 1 4 Heracleum sphondylium L. Apiaceae 2 17 Hieracium murorum L. Asteraceae 2 20 Hieracium pilosella L. Asteraceae 8 354 Hypericum perforatum L. Asteraceae 1 5 Hypochaeris radicata L. Asteraceae 19 378 Knautia arvensis (L.) Coult. s. str. Dipsacaceae 6 38 Lamium purpureum L. s. l. p. p. Lamiaceae 3 47 Lathyrus latifolius L. Fabaceae 1 11 Leucanthemum vulgare Lam. s. str. Asteraceae 9 771 Linum tenuifolium L. Linaceae 1 1 Lotus corniculatus L. Fabaceae 20 954 Malva moschata L. Malvaceae 2 30 Medicago falcata L. s. str. Fabaceae 3 167 Medicago lupulina L. Fabaceae 1 54 Myosotis arvensis (L.) Hill Boraginaceae 1 51 Plantago lanceolata L. Plantaginaceae 23 215 Plantago media L. Plantaginaceae 1 1 Potentilla reptans L. Rosaceae 1 7 Prunella grandiflora (L.) Scholler Lamiaceae 8 430 Prunella vulgaris L. Lamiaceae 5 91 Ranunculus acris L. Ranunculaceae 11 86 Ranunculus repens L. Ranunculaceae 14 550 130 Chapter 5

Rhinanthus alectorolophus (Scop.) Pollich s. l. Scrophulariaceae 4 983 Salvia pratensis L. Lamiaceae 4 222 Scabiosa columbaria L. Dipsacaceae 2 4 Senecio jacobaea L. Asteraceae 1 5 Silene latifolia Poir. Caryophyllaceae 1 28 Silene vulgaris (Moench) Garcke s. l. Caryophyllaceae 2 36 Stachys recta L. Lamiaceae 1 198 Taraxacum sect. Ruderalia Wiggers Asteraceae 1 18 Thymus praecox Opiz s. l. Lamiaceae 3 127 Trifolium campestre Schreb. Fabaceae 20 2,400 Trifolium micranthum Viv. Fabaceae 4 711 Trifolium pratense L. Fabaceae 61 3,164 Trifolium repens L. Fabaceae 42 2,422 Verbena ofﬁcinalis L. Verbenaceae 1 26 Veronica agrestis L. Scrophulariaceae 7 86 Vicia cracca L. s. str. Fabaceae 5 16

67 19 447 18,882 5.7 Tables and Figures 131

Table 5.3: Mean and standard deviations of landscape metrics for ﬁeld sites with syrphid species and without syrphid species. Test statistics is derived from Wilcoxon exact test.

no syrphids syrphids W p-value

edge density 0.09 ± 0.03 0.1 ± 0.02 1,162.5 0.008 diversity green area 0.71 ± 0.23 0.86 ± 0.19 1,254 0.0005

edge density 0.088 ± 0.029 0.1 ± 0.025 893 0.08 visits green area 0.69 ± 0.3 0.81 ± 0.2 960 0.02

Figure 5.1: The map shows the city of Zurich¨ and the different landscape elements forest, green area, paved area, building and water. Red dots represent the 89 study locations. 132 Chapter 5

140 Pollinators Plants 120 Bees Syrphids 100

40 Accumulated Species Richness Accumulated 20

0 20 40 60 80 Sites

Figure 5.2: Species accumulation curves for pollinator, plant, bee and syrphid diversity with 95% conﬁdence intervals using the method ”random” in the statistical program R.

Plot Frequencies 20

Tr.pr Tr.re Pl.la Tr.ca Ga.alBe.pe Lo.co Hy.ra Ce.ja Ra.re

Figure 5.3: The ten most frequent plant species and their plot frequencies Tr.pr = Trifolium pratense, Tr.re = Trifolium repens, Ga.al = Galium album, Be.pe = Bellis perennis, Pl.la = Plantago lanceolata, Tr.ca = Trifolium campestre, Lo.co = Lotus corniculatus, Hy.ra = Hypochoeris radicata, Ce.ja = Centaurea jacea, Ra.re = Ranunculus repens. 5.7 Tables and Figures 133

Green Area Edge Density

● ●

● ● ● ● ● ● ● ● ● 0.10 ● ● ● ● ● ● 0.15 ● ●

0.05 ●

● ● 0.00 ● 0.10 ● Bee Diversity −0.05

● ● 0 20000 60000 100000 0 20000 60000 100000

0.02 ● ● ● −0.16 ● ● ● ● ● ● ● ● ● 0.00 ● −0.18 ●

● ●

● −0.20 ● ● Bee Visits −0.02

● ● −0.22 ● ● −0.04 ● ● ● 0 20000 60000 100000 0 20000 60000 100000

● ● ● ●

Spearman‘s Correlation Coefficient (rho) ● ● ● ●

● ● ● 0.05

0.05 ● ●

● ●

0.00 ● ● ● ● ● 0.00 ● ●

● ● Bee Visits w/o Honeybees −0.05 ● ● 0 20000 60000 100000 0 20000 60000 100000 Area (km2)

Figure 5.4: Correlation between of extent of green area and green area edge density with bee diversity, bee visits, and bee visits excluding honeybees across the 89 sampled locations. Note the differences in scale at the y-axis. 134 Chapter 5

Green Area Edge Density

● ● 0.40 ● ● ●

● 0.3

● ● ● ● ● ● ● ● ● 0.35 ●

● 0.2 ● ●

● ● 0.30 ● Syrphid Diversity

● 0.1

0.25 ●

● ●

0 20000 40000 60000 80000 100000 0 20000 40000 60000 80000 100000

0.40 ● ● ● 0.3 ● ● ● ● ● ● ● ● 0.35 ●

● ● ● 0.2 0.30 ● ● ● Spearman‘s Correlation Coefficient (rho)

●

0.25 ● 0.1 Syrphid Visits ●

0.20 ●

● 0.0 ● 0.15 ● ●

0 20000 40000 60000 80000 100000 0 20000 40000 60000 80000 100000

Area (m2)

Figure 5.5: Correlation between the extent of green area and edge density with syrphid diversity and syrphid visitation frequency across the 89 sampled locations. Note the differences in scale at the y-axis. Chapter 6

Synthesis

6.1 Summary

In the first objective of the thesis interactions among plants in attracting pollinators were suggested to be shaped by the composition of the pollinator community. In the second chapter we found evidence that preferentially bee-visited plant species did not compete or benefit from co-occurring flowering plant species, while plant species with a generalized pollinator community, i.e., visited to an equal extent by bees and syrphids, were competing for floral visits. The second objective challenged the reproductive success and in the third chapter we investigated how seed set of plant species is affected by plant diversity and floral density. Seed set in plant species indeed differed in terms of the visitor community composition. Seed set in Borago officinalis, which was preferentially bumblebee visited, was positively affected by plant diversity or floral density of plant species. Another plant species, Sinapis alba, which was visited by bees and syrphids to an equal extant, was negatively affected by floral density of sympatrics. Plant species with a more specialized visitor community were less negatively affected by increased number of co-occurring plant species, while plant species with more generalized visitor community appears to have a disadvantage in terms of seed set. In urban areas we anticipated that isolated flower patches will be particularly vulnerable to reduced pollinator numbers, and that interactions among plant species and their pollinators are affected by the structure of the urban environment, particularly with regards to the extent and edge density of green areas. In Chapter 4, we have shown that in an urban environment the plant species Trifolium pratense received fewer floral visits when ocurring with two other plant species. 136 Chapter 6 Synthesis

Such plant-plant interactions were evidently influenced by the extent of green area, in that with increasing extent of green area there was increased bee flower visitation frequency. Excluding the main visitor, however, the extent of green area interacted negatively only with heterospecific floral abundance on bee visitation frequency to the focal plant. The last question dealt with the impact of the urban landscape, described by the two landscape metrics extent and edge density of green area, on bees and syrphids at a scale from 20 – 200 m radius. In Chapter 5, we found evidence that both the extent and edge density of green areas influenced the diversity and visit frequency of flower visiting bees and syrphids at several scales. Edge density had a positive effect on bee and syrphid diversity, and syrphid plot visits with increasing scale, while bee plot visits were negatively affected at scales above 100 m radius. We found the same patterns of correlation for the extent of green area, although correlation coefficients of edge density and extent of green area with syrphid diversity and syrphid plot visits levelled off at scales above 100 m radius.

6.2 Discussion

Many studies have shown that plant diversity positively affects flower visitor diversity and floral visitation frequency (Chapter 2, Ghazoul, 2006; Pontin et al., 2006; Ebeling et al., 2008; Hegland et al., 2009; Lazaro´ et al., 2009). Plant species in species-rich flowering patches can benefit from growing with other plant species by achieving higher visitation frequencies, which can result in increased reproductive success. Plants can, however, also compete for pollinator species and/or experience improper pollen transfer with increasing heterospecific plant diversity (e.g., Waser, 1978; Rathcke, 1988; Bell et al., 2005; Morales and Traveset, 2008). The occurrence of either competition or facilitation among plant species for pollinator visits can be mediated by different floral traits (‘phenotypic generalization/specialization‘, Westerkamp, 1997; Ollerton et al., 2007; Campbell et al., 2010). For example, certain floral colours attract specific groups of pollinators (Lunau and Maier, 1995; Campbell et al., 2010; Ortigosa and Gomez,´ 2010), while morphology reduces access to resources for pollinators with inappropriate mouth structures (Westerkamp, 1997; Neal et al., 1998; Glover, 2007). These floral preferences can be also the result of perceptual capability (Chittka and Raine, 2006), detectability against the background (Lunau et al., 1996), innate preference (e.g., Lunau and Maier, 1995), or learned preference (e.g., Goulson and Cory, 1993), which can differ among bees and flies (e.g., Fenster et al., 2004). Facilitation among plant species for attracting pollinators can thus be shaped by the pollinator composition. In this study, plant species that are preferentially visited by social bees, i.e. 6.2 Discussion 137

honeybees and bumble bees, were not affected by surrounding plant species richness, nor by heterospecific floral density, while plant species visited by syrphids and bees to an equal extent were receiving fewer flower visits. It can be therefore assumed that plant species visited most frequently by social bees are subjected less to interactions with other heterospecifics for flower visits (Chapter 2). As the community composition of pollinators determined plant-plant interactions for visitation frequency it is likely that plant-plant interactions in terms of reproductive success are also influenced by the pollinator community (Gomez´ et al., 2007; Franzen and Larsson, 2009). Plant species visited by few pollinator species can experience larger seed set, because the likelihood of interactions of conspecific with heterospecific pollen is reduced (Chapter 3, Larsson, 2005; Morales and Traveset, 2008; Mitchell et al., 2009). Conversely, plant species with a species-rich pollinator community are more likely subjected to interspecific pollen transfer and hence reduced seed set, because there is higher probability of having pollinator species moving between plant species (Chapter 3, Gomez´ et al., 2007; Hegland and Totland, 2008). For every plant species pollinator community composition can, however, vary over time and space (Price et al., 2005; Lazaro´ et al., 2010), and thereby influence interactions among plants for pollination services (Herrera, 1988, 2005), which renders often impossible to find general patterns. It is very likely that urban natural plant communities are less diverse than natural or semi-natural plant communities, because urban constraints, in particular habitat loss, fragmentation and disturbance (Tratalos et al., 2007; Knapp et al., 2008a; Chapman and Underwood, 2009), allow the persistence of fewer plant species and more homogenous plant communities among green urban patches (Lososova´ et al., 2006; Knapp et al., 2008b; Vallet et al., 2008). Although theory claims that facilitative interactions among plants are more likely under stressful conditions (Bertness and Callaway, 1994; Callaway and Walker, 1997; Holmgren et al., 1997; Brooker and Callaghan, 1998; Pugnaire and Luque, 2001), pollinator communities can also become depauperate due to habitat loss and fragmentation (Cane et al., 2006), which implies that plant species might compete for visits if pollinators become limited. Our study shows that with an increase in plant diversity and floral abundance of heterospecific plant species there was a decrease in pollinator visits in our focal plant species, which indicates that competition among urban plant species is also likely (Chapter 4). Although there was no comparison of visitation frequency between natural or semi-natural habitats, and urban environments of the same studied plant species, natural habitats might support a larger community of flower-visiting insects due to enhanced quality and quantity of resources and nesting sites. An increasing extent of green area in urban environments can be therefore expected to render urban conditions similar to natural environments (McFrederick and LeBuhn, 2006; Smith et al., 2006) and mediate interactions among plant species for pollinator visits (Chapter 4). This emphasizes the need to consider landscape properties such as the extent of green areas 138 Chapter 6 Synthesis

in plant-plant interactions for attracting flower visitors owing to their potential of supporting pollinator communities. Landscape properties can, however, have different impacts on the diversity and visitation frequency of bees and flies (Chapter 5), which can be attributed to the biology of the pollinator family. For example, adult and larval syrphid flies have different resource and habitat requirements and adults are decoupled from their larvae (Sommaggio, 1999; van Veen, 2004). Bees, on the other hand, are spatially constrained by having to provision a nest that is fixed in space (i.e., central place foraging). Thus it can be expected that both groups forage differently in the landscape in terms of distance, and can be differently affected by the landscape structure at altering scales (Chapter 5). Additionally, many bees demand appropriate soil conditions for nesting sites (e.g., Potts and Willmer, 1997), and syrphids need species-dependent different structures such as plants sensitive to aphid-infection, holes in rotten wood, or ponds (Sommaggio, 1999; van Veen, 2004). The extent of green area, which is expected to correlate positively with the diversity of semi-natural and resources, shows a strong and positive impact on bee and syrphid fly diversity as well as bee abundance (Chapter 5), which in turn affects facilitative and competitive interactions among plants for floral visits and hence potential services to plant species within the urban environment.

6.3 Outlook

There are two major issues missing, that would otherwise provide a more complete picture of facilitation among plant-pollinator systems. The effect of pollinator and plant community composition on reproductive success was investigated in (semi-)natural environments (Chapter 2, 3), but there is no information about these processes in urban environments. In the scope of growing interest in urban ecology and the potential of urban environments supporting biodiversity and their value for conservation (Dearborn and Kark, 2010), more information is required on how urban plants interact for pollination services. Second, this thesis did not account for the landscape structure at different scales in the (semi-)natural environment. It can be argued that (semi-)natural, agricultural environments are more uniform than urban environments in terms of habitats for plants and pollinators. Nonetheless few studies have shown that in (semi-)natural environments landscape structure and spatial scale of consideration inﬂuence diversity and abundance of pollinators (Steffan-Dewenter et al., 2002; Meyer et al., 2007) and highlight the potential of landscape metrics to inﬂuence facilitative and competitive interactions among plants for pollinators in (semi-)natural environments. Finally, there is need to stress the importance of long-term studies at different locations with different plant species to make reliable generalizations 6.4 References 139

on plant-plant and plant-pollinator interactions. Large daily, seasonal and annual variation in pollinator community composition can mitigate plant-plant interactions (Chapter 3, Herrera, 2005; Lazaro´ et al., 2010), while the geographic mosaic concept makes plant interactions for pollination services in different locations less predictable (Johnson, 2006). Thus studies on different short time-scales through and over few years on visitation patterns and ﬁtness components such as seed set would provide information about the temporary forces driving the dynamic of plant-plant interactions for pollinator services, while experiments at different locations would allow for spatial generalization in interactive processes among plants.

6.4 References

Bell, J. M., Karron, J. D., and Mitchell, R. J. (2005). Interspeciﬁc competition for pollination lowers seed production and outcrossing in Mimulus ringens. Ecology, 86(3):762–771.

Bertness, M. D. and Callaway, R. (1994). Positive interactions in communities. Trends in Ecology & Evolution, 9(5):191–193.

Brooker, R. W. and Callaghan, T. V. (1998). The balance between positive and negative plant interactions and its relationship to environmental gradients: a model. Oikos, 81(1):196–207.

Callaway, R. M. and Walker, L. R. (1997). Competition and facilitation: A synthetic approach to interactions in plant communities. Ecology, 78(7):1958–1965.

Campbell, D. R., Bischoff, M., Lord, J. M., and Robertson, A. W. (2010). Flower color inﬂuences insect visitation in alpine New Zealand. Ecology, 91(9):2638–2649.

Chapman, M. and Underwood, A. (2009). Comparative effects of urbanisation in marine and terrestrial habitats. In McDonnell, M. J., Hahs, A. K., and Breuste, J. H., editors, Ecology of Cities and Towns, chapter 4, pages 51–89. Cambridge University Press.

Chittka, L. and Raine, N. E. (2006). Recognition of ﬂowers by pollinators. Current Opinion in Plant Biology, 9(4):428–435.

Dearborn, D. C. and Kark, S. (2010). Motivations for Conserving Urban Biodiversity. Conserva- tion Biology, 24(2):432–440. 140 Chapter 6 Synthesis

Fenster, C., Armbruster, W., Wilson, P., Dudash, M., and Thomson, J. (2004). Pollination syndromes and ﬂoral specialization. Annual Review of Ecology, Evolution and Systematics, 35:375–403.

Franzen, M. and Larsson, M. (2009). Seed set differs in relation to pollen and nectar foraging ﬂower visitors in an insect-pollinated herb. Nordic Journal of Botany, 27(4):274–283.

Ghazoul, J. (2006). Floral diversity and the facilitation of pollination. Journal of Ecology, 94:295–304.

Glover, B. (2007). Understanding Flower & Flowering. An Integrated Approach. Oxford University Press.

Gomez,´ J. M., Bosch, J., Perfectti, F., Fernandez,´ J., and Abdelaziz, M. (2007). Pollinator diversity affects plant reproduction and recruitment: the tradeoffs of generalization. Oecologia, 153(3):597–605.

Goulson, D. and Cory, J. S. (1993). Flower constancy and learning in foraging preferences of the green-veined white butterﬂy Pieris napi. Ecological Entomology, 18(4):315–320.

Hegland, S. J., Nielsen, A., Lazaro, A., Bjerknes, A. L., and Totland, O. (2009). How does climate warming affect plant-pollinator interactions? Ecology Letters, 12(2):184–195.

Hegland, S. J. and Totland, O. (2008). Is the magnitude of pollen limitation in a plant community affected by pollinator visitation and plant species specialisation levels? Oikos, 117:883–891.

Herrera, C. M. (1988). Variation in mutualisms - The spatio-temporal mosaic of a pollinator assemblage. Biological Journal of the Linnean Society, 35(2):95–125.

Herrera, C. M. (2005). Plant generalization on pollinators: Species property or local phe- nomenon? American Journal of Botany, 92(1):13–20.

Holmgren, M., Scheffer, M., and Huston, M. A. (1997). The interplay of facilitation and competition in plant communities. Ecology, 78(7):1966–1975.

Johnson, S. D. (2006). Pollinator-driven speciation in plants. In Harder, L. D. and Barrett, S. C., editors, Ecology and Evolution of Flowers, chapter 16, pages 295–310. Oxford University Press. 6.4 References 141

Knapp, S., Kuhn,¨ I., Schweiger, O., and Klotz, S. (2008a). Challenging urban species diversity: contrasting phylogenetic patterns across plant functional groups in Germany. Ecology Letters, 11(10):1054–1064.

Knapp, S., Kuhn,¨ I., Wittig, R., Ozinga, W. A., Poschlod, P., and Klotz, S. (2008b). Urbanization causes shifts in species’ trait state frequencies. Preslia, 80(4):375–388.

Larsson, M. (2005). Higher pollinator effectiveness by specialist than generalist ﬂower-visitors of unspecialized Knautia arvensis (Dipsacaceae). Oecologica, 146(3):394–403.

Lazaro,´ A., Lundgren, R., and Totland, O. (2009). Co-ﬂowering neighbors inﬂuence the diversity and identity of pollinator groups visiting plant species. Oikos, 118(5):691–702.

Lazaro,´ A., Nielsen, A., and Totland, O. (2010). Factors related to the inter-annual variation in plants’ pollination generalization levels within a community. Oikos, 119(5):825–834.

Lunau, K. and Maier, E. J. (1995). Innate color preferences of ﬂower visitors. Journal of Comparative Physiology A – Sensory Neural and Behavioral Physiology, 177(1):1–19.

Lunau, K., Wacht, S., and Chittka, L. (1996). Colour choices of naive bumble bees and their implications for colour perception. Journal of Comparative Physiology A – Sensory Neural and Behavioral Physiology, 178(4):477–489.

McFrederick, Q. S. and LeBuhn, G. (2006). Are urban parks refuges for bumble bees Bombus spp. (Hymenoptera : Apidae)? Biological Conservation, 129(3):372–382.

Mitchell, R. J., Flanagan, R. J., Brown, B. J., Waser, N. M., and Karron, J. D. (2009). New frontiers in competition for pollination. Annals of Botany, 103(9):1403–1413.

Morales, C. L. and Traveset, A. (2008). Interspeciﬁc pollen transfer: Magnitude, prevalence and consequences for plant ﬁtness. Critical Reviews in Plant Sciences, 27(4):221–238.

Neal, P. R., Dafni, A., and Giurfa, M. (1998). Floral symmetry and its role in plant-pollinator systems: Terminology, distribution, and hypotheses. Annual Review of Ecology and Systematics, 29:345–373. 142 Chapter 6 Synthesis

Ollerton, J., Killick, A., Lamborn, E., Watts, S., and Whiston, M. (2007). Multiple meanings and modes: on the many ways to be a generalist ﬂower. Taxon, 56(3):717–728.

Ortigosa, A. L. and Gomez,´ J. M. (2010). Differences in the diversity and composition of the pollinator assemblage of two co-ﬂowering congeneric alpine wallﬂowers, Erysimum nevadense and E. baeticum. Flora, 205(4):266–275.

Potts, S. G. and Willmer, P. (1997). Abiotic and biotic factors inﬂuencing nest-site selection by Halictus rubicundus, a ground-nesting halictine bee. Ecological Entomology, 22(3):319–328.

Price, M. V., Waser, N. M., Irwin, R. E., Campbell, D. R., and Brody, A. K. (2005). Temporal and spatial variation in pollination of a montane herb: A seven-year study. Ecology, 86(8):2106– 2116.

Pugnaire, F. I. and Luque, M. T. (2001). Changes in plant interactions along a gradient of environmental stress. Oikos, 93(1):42–49.

Rathcke, B. (1988). Interactions for pollination among coﬂowering shrubs. Ecology, 69(2):446– 457.

Sommaggio, D. (1999). Syrphidae: can they be used as environmental bioindicators? Agriculture Ecosystems & Environment, 74(1-3):343–356.

Steffan-Dewenter, I., Munzenberg,¨ U., Burger,¨ C., Thies, C., and Tscharntke, T. (2002). Scale- dependent effects of landscape context on three pollinator guilds. Ecology, 83(5):1421–1432.

Tratalos, J., Fuller, R. A., Warren, P. H., Davies, R. G., and Gaston, K. J. (2007). Urban form, biodiversity potential and ecosystem services. Landscape and Urban Planning, 83(4):308–317.

Vallet, J., Daniel, H., Beaujouan, V., and Roze, F. (2008). Plant species response to urbanization: comparison of isolated woodland patches in two cities of North-Western France. Landscape Ecology, 23(10):1205–1217. van Veen, M. D. (2004). Hoverﬂies of Northwest Europe. KNNV Publishing. 6.4 References 143

Waser, N. M. (1978). Interspeciﬁc pollen transfer and competition between co-occurring plant species. Oecologia, 36(2):223–236.

Westerkamp, C. (1997). Keel blossoms: Bee ﬂowers with adaptations against bees. Flora, 192(2):125–132.

Acknowledgements

There are many people I want to thank for their tremendous support and many more, to realize and ﬁnish this PhD thesis. Among these I would like to mention in particular,

Prof. Jaboury Ghazoul for giving me the opportunity to study the relationship between pollinators and plants, for his enormous help and support in many stages of developing my PhD thesis, and in teaching me how to write proper English.

Prof. Felix Kienast, Dr. Simon Leather, and Prof. Alex Widmer for their time grappling with my thesis.

ETH Zurich¨ rectorat/scholarship ofﬁce for supporting me with some additional funding.

Prof. Felix Bersier´ and his students Philipp Arnold and Eveline Thevoz from the University of Fribourg for their collaboration in the Grandcour Project.

Prof. Michael Scherer-Lorenzen, University of Freiburg, Germany, for very helpful advices in establishing a study design at Eschikon.

Dr. Andrea Pluss¨ for her enormous help in ﬁnding answers on tricky statistical questions and the opportunity to contribute to different research projects in the group.

Dr. Chris Kettle for his enormous help in writing the Grandcour chapter and the synthesis, and the many helpful advices.

Dr. Chris Kaiser-Bunbury especially for his help in the initial phase of my PhD thesis.

Charlotte Klank for statistical discussions and new insights, her help in the green houses and the ﬁeld work at Eschikon and Grandcour, and for coordinating the logistics of my PhD in the year I had my accident. 146 Acknowledgements

Smitha Krishnan for statistical discussions and new insights, and her tremendous help in writing up some sections.

Thomas Hahn for his technical support in Geographical Information Systems (GIS)

Philippe Matter for his help in the green houses and the ﬁeld work at Eschikon and Grandcour.

Rebecca Kittel and Patrik Peyer for their excellent support in continuing my observations at Eschikon 2007 after my accident, and Isabelle Peter for her help in transplanting several hundred plants in few days at Eschikon.

Brigitte Pichler and Maya Frei from the Research Station Eschikon, and Jorg¨ Leuenberger from ETH Zurich¨ for taking care of my plants and many helpful horticultural and technical advices.

Other people eased my work though these activities are not directly related to the PhD thesis. My special thanks to Ankara Chen and Julia Born for managing adminstrative issues and helping with such matters, Florian Knaus for the ornithological break offs and for organizing interesting journal club meetings, Thomas Hahn for enjoyable spare alpine hikes, Nadine Ruhr¨ for her support and hints to structure my PhD thesis, and Werner and Brigitte Pfahler, and Karin Ruhr¨ for their initial support. My deepest thanks to my parents Jolanta Krystyna and Andreas Zbigniew Hennig, and my

grandmother Irena Kedzierska, for their conﬁdence and support not only during critical stages in my PhD life. Z głebi, serca chce, wam za wszystko dziekowa, c.´ Bez waszej pomocy nie zrealizowałbym moja, prace., Et bien entendu, j’aimerais remercier Geraldine´ Zosso et ses parents Denise et Jean-Bernard pour leur soutien pendant toute le preparation´ de mon doctorat. Ernest IreneuszHennig

Education 03/2006– PhD, Institute of Terrestrial Ecosystems, ETH Zürich, Switzerland. 04/2011 Research topic: Faciliation among plants for pollinator services. 05–08/2004 Diploma in Environmental Science, Department of Botany, University Duisburg- Essen, Essen, Germany. Specialisation in Plant Ecology and Geology. Diploma thesis: Epiphyte diversity of three diﬀerent forest types in Bolivia.

Jobs 07/2010– Graduate Research Assistant, Ecosystem Management, ETH Zürich, Zürich, Switzer- 12/2010 land. Coordination of students during evaluation of invertebrate fauna. 08/2008 Course Tutor, Ecosystem Management, ETH Zürich, Zürich, Switzerland. Coordination of students during evaluation of invertebrate fauna. 2003–2004 Student Research Project, Department of Environmental Geology, University Duisburg-Essen, Essen, Germany. Platinum in limnic depositions. 10/2003– Undergraduate Research Assistant, Department of Hydrobiology, University 01/2004 Duisburg-Essen, Essen, Germany. Classiﬁcation of limnic invertebrates in the frame of limnic habitat evaluation in North Rhine- Westphalia. 03–12/2002 Undergraduate Research Assistant, Department of Environmental Geology, Univer- sity Duisburg-Essen, Essen, Germany. Laboratory assistance and organisation. 10–11/2001 Volunteer, Fundacion Jatun Sacha, Quito, Ecuador. Reforestation and establishment of nature trails.

Scholarship 07–09/2009 ETH Zürich Scholarship for Doctoral Students.

Publications

E. I. Hennig and J. Ghazoul, “Pollinating animals within the urban environment,” Urban Ecosys- tems, 2011: in press. E. I. Hennig and J. Ghazoul, “Plant-pollinator interactions within the urban environment,” Perspectives in Plant Ecology Evolution and Systematics, vol. 13, pp. 137–150, 2011. R. Linares-Palomino, V. Cardona, E. I. Hennig, I. Hensen, D. Hoﬀmann, J. Lendzion, D. Soto, S. K. Herzog, and M. Kessler, “Non-woody life-form contribution to vascular plant species richness in a tropical American forest,” Plant Ecology, vol. 201, pp. 87–99, Mar 2009.

Presentations Poster Hennig, E.I. & J. Ghazoul. Relationship between plant community traits and pollinator species richness in urban and sub-urban areas. biology07. The Annual Meeting of the Swiss Zoological, Botanical and Mycological Societies. February 15–16, 2007. ETH Zürich, Switzerland