Bombus Spp.) in Semi-Natural Meadows

INFLUENCE OF ABIOTIC AND BIOTIC FACTORS AT PATCH AND LANDSCAPE SCALE ON BUMBLEBEES (BOMBUS SPP.) IN SEMI-NATURAL MEADOWS

ABIOOTILISTE JA BIOOTILISTE FAKTORITE MÕJU KIMALASTE POPULATSIOONIDELE POOLLOODUSLIKEL KOOSLUSTEL: MAASTIKULINE ANALÜÜS

ISABEL DIAZ FORERO

A Thesis for applying for the degree of Doctor of Philosophy in Environmental Protection

Väitekiri filosoofiadoktori kraadi taotlemiseks keskkonnakaitse erialal

Tartu 2011 Eesti Maaülikool Estonian university of life sciences

INFLUENCE OF ABIOTIC AND BIOTIC FACTORS AT PATCH AND LANDSCAPE SCALE ON BUMBLEBEES (BOMBUS SPP.) IN SEMI-NATURAL MEADOWS

ABIOOTILISTE JA BIOOTILISTE FAKTORITE MÕJU KIMALASTE POPULATSIOONIDELE POOLLOODUSLIKEL KOOSLUSTEL: MAASTIKULINE ANALÜÜS

ISABEL DIAZ FORERO

A Thesis for applying for the degree of Doctor of Philosophy in Environmental Protection

Väitekiri filosoofiadoktori kraadi taotlemiseks keskkonnakaitse erialal

Tartu 2011 Institute of Agricultural and Environmental Sciences Estonian University of Life Sciences

According to the verdict No 85 of October 31, 2011, the Doctoral Committee of Ag- ricultural and Natural Sciences of the Estonian University of Life Sciences has accepted the thesis for the defence of the degree of Doctor of Philosophy in Environmental Protection.

Opponent: Prof. Dave Goulson Institute of Biological and Environmental Sciences University of Stirling Stirling, Scotland

Supervisors: Prof. Valdo Kuusemets Institute of Agricultural and Environmental Sciences Estonian University of Life Sciences Tartu, Estonia

Prof. Marika Mänd Institute of Agricultural and Environmental Sciences Estonian University of Life Sciences Tartu, Estonia

Defence of the thesis: Estonian University of Life Sciences, room 1A5, Kreutzwaldi 5, Tartu. December 16, 2011, at 10:00.

The English language was edited by Ingrid H. Williams, and the Estonian language by Kadri Kask.

Publication of this thesis is supported by the Estonian University of Life Sciences and by the Doctoral School of Earth Sciences and Ecology created under the auspices of European Social Fund.

Contents

LIST OF ORIGINAL PUBLICATIONS 9 ABBREVIATIONS 10 1. INTRODUCTION 11 2. REVIEW OF THE LITERATURE 13 2.1. The importance and decline of bumblebees 13 2.2. The use of patch and landscape factors in ecological studies 16 2.3. Habitat quality: definition and importance 17 3. AIMS OF THE STUDY 19 4. MATERIALS AND METHODS 20 4.1 Study region 20 4.2. Bumblebee survey 22 4.3. Patch-scale factors 23 4.3.1. Vegetation structure 23 4.3.2. Spatial characteristics 23 4.4. Landscape-scale factors 24 4.4.1. Landscape composition 24 4.4.2. Landscape configuration 24 4.5. Data analysis 25 5. RESULTS 28 5.1. Total bumblebee species richness and abundance 28 5.1.1. Bumblebee species richness and abundance in northeast Estonia 28 5.1.2. Relations between bumblebees and patch-scale factors (Paper II) 30 5.1.3. Relations between bumblebees and landscape-scale factors (Paper II) 30 5.1.4. Connectivity patterns between bumblebees and the factors at patch and landscape scale (Paper II) 31 5.1.5. Models to predict bumblebee species richness and abundance (Paper II) 32 5.2. Relations between long-tongued bumblebees and the factors at patch and landscape scale 34 5.3. Relations between the local abundance of bumblebee species and forest habitats 35 5.3.1. Proportion of forest (Paper IV) 35 5.3.2. Proportion of brushwood (Paper IV) 36 5.3.3. Landscape indices (Paper IV) 37 5.3.4. Joint effects of landscape factors related to forest (Paper IV) 37 6. DISCUSSION 39 6.1. Influence of patch-scale factors on bumblebees 39 6.2. Influence of landscape composition on bumblebees 41 6.3. Influence of landscape configuration on bumblebees 43 6.4. Influence of forest habitats on the local abundance of bumblebee species 45 6.5. Bumblebees as potential indicators of habitat quality 48 7. CONCLUSIONS AND IMPLICATIONS FOR CONSERVATION 50 REFERENCES 53 Appendix 1 61 Appendix 2 63 SUMMARY IN ESTONIAN 66 ACKNOWLEDGEMENTS 68 Publications 69 CURRICULUM VITAE 130 ELULOOKIRJELDUS 132 LIST OF PUBLICATIONS 134 LIST OF ORIGINAL PUBLICATIONS

The present thesis is based on the following research papers, which are referred to by their Roman numerals in the text.

I Diaz-Forero, I., Liivamägi, A., Kuusemets, V. and Luig J. 2010. Pollinator richness and abundance in Northeast Estonia: bumblebees, butterflies and day-flying moths. Forestry Studies | Met- sanduslikud Uurimused 53, 5–14.

II Diaz-Forero, I., Kuusemets, V., Mänd, M., Liivamägi, A., Kaart, T. and Luig, J. 2011. Influence of local and landscape factors on bumblebees in semi-natural meadows: a multiple-scale study in a forested landscape. Journal of Insect Conservation. (Submitted).

III Diaz-Forero, I., Kuusemets, V., Mänd, M. and Luig, J. 2011. Bumblebees as potential indicators for the evaluation of habitat quality. Sustainable Development and Planning V. WIT Trans- actions on Ecology and the Environment. WIT Press. Vol 150, 409-417.

IV Diaz-Forero, I., Kuusemets, V., Mänd, M., Liivamägi, A., Kaart, T. and Luig, J. 2011. Effects of forest habitats on the local abundance of bumblebee species: a landscape-scale study. Baltic Forestry 17(2), ISSN 1392-1355. (In press).

The papers are reproduced by kind permission of the corresponding journal or publisher.

Authors’ contributions to the papers: Paper Idea and study Data Analysis of data Manuscript design collection preparation I ID-F, VK AL, ID-F, JL ID-F, AL ID-F, AL, VK II ID-F, VK, MM AL, ID-F, JL ID-F, AL, MM, TK ID-F, MM, VK, TK III ID-F, VK, MM ID-F, JL ID-F, MM ID-F, MM, VK IV MM, ID-F, VK AL, ID-F, JL ID-F, AL, MM, TK ID-F, MM, VK, TK

AL – Ave Liivamägi; ID-F – Isabel Diaz-Forero; JL – Jaan Luig; MM – Marika Mänd; TK – Tanel Kaart; VK – Valdo Kuusemets

9 ABBREVIATIONS

AREA patch area AREA_MN-Forest mean patch area of forest ED edge density at patch level ED_Forest edge density of forest ED_LAND edge density at landscape level FRAC fractal dimension index IJI interspersion and juxtaposition index PERIM patch perimeter PLS Partial Least Squares PRD patch richness density SHAPE shape index SHDI Shannon’s diversity index

10 1. INTRODUCTION

Bumblebees (Bombus spp.) and other bees are considered a vital element of global biodiversity and an important group of pollinators in agroecosystems. They are valuable pollinators of cultivated crops as well as wild plants (Sepp et al. 2004, Goulson et al. 2006, Rundlöf et al. 2008, Knight et al. 2009). Compared with the majority of bees, bumblebees have particularly large and hairy bodies; due to these characteristics, they are well adapted to places with low temperatures (Goulson 2010). That is why bumblebees are mostly found in temperate, alpine and arctic zones.

Mainly due to the intensification of farming practices in agriculture, bumblebees and other pollinators have declined (Mänd et al. 2002, Goulson et al. 2006, Holzschuh et al. 2008, Xie et al. 2008). Agricul- tural intensification is characterised by the use of fertilisers and pesticides and the reduction of flowering plants; causing the fragmentation of landscapes and the loss of suitable habitats for insects (Krewenka et al. 2011). Agri-environmental schemes are being applied in many European countries to mitigate the negative consequences of the intensification of farming practices on biodiversity. The development of more effective agri-environmental measures has become an issue of great concern among decision makers, mainly due to the growing interest of politicians, farmers and consumers in more environmentally-friendly farming practices (Kleijn et al. 2006, Holzschuh et al. 2008).

According to Goulson et al. (2011), bumblebees have been well-studied in modern agricultural landscapes of Western Europe, the United King- dom, Japan and North America. However, very little is known about the conservation and ecology of most of the bumblebee species elsewhere (Goulson et al. 2011). Those areas that have been well-studied generally consist of large monoculture fields separated by field margins and small patches of woodland. In contrast, the landscape in Estonia has a very mosaic pattern, where approximately 32% of the whole territory is constituted by agricultural land (Mander and Palang 1994), but only a small proportion is cultivated and it contains many patches of natural habitats (Mänd et al. 2002). Moreover, there is evidence that the proportion of forest increased substantially during the 20th century (from 14% to 42%) (Palang et al. 1998). Understanding the associations between bumblebees and the surrounding landscape is relevant for the conservation of this group of pollinators, particularly in countries that have frag-

11 mented landscapes with high proportions of forest and natural habitats. The species abundance and distribution are influenced by processes that occur at multiple spatial scales. Some drivers of biodiversity loss, such as habitat destruction and fragmentation, are likely to operate at large spatial scales (Jones 2011); that is why multiple scale studies are currently needed. Even more, conservation strategies may be enhanced when properly defined landscape-scale variables are included (Mazerolle and Villard 1999).

Considering that Estonia has a very patchy landscape pattern with a significant presence of natural and semi-natural habitats (Mänd et al. 2002), the aim of this thesis is to analyse the influence of biotic and abiotic factors, at patch and landscape scale, on bumblebees. At the patch scale, factors that describe the vegetation structure and spatial characteristics of the study sites, such as area, perimeter and shape, were considered. At the landscape scale, a set of factors describing the composition and configuration of the landscape matrix were calculated at various spatial scales. This study was developed to determine which variables should be considered when designing biodiversity conservation strategies and agri- environmental measures for this type of landscape mosaic (i.e., which factors have a greater influence on bumblebees in forested landscapes). Finally, a discussion about the potential use of bumblebees as indicators for the evaluation of habitat quality in semi-natural meadows is presented.

12 2. REVIEW OF THE LITERATURE

2.1. The importance and decline of bumblebees

From an ecological perspective, insects are important because they perform numerous ecosystem services, e.g., pollination, seed dispersal, pest control, waste decomposition, food for other invertebrates and verte- brates, etc. (Samways 2005). Among insects, bees are considered an important group of pollinators, as they play a key role in supporting the diversity of natural and semi-natural vegetation (Sepp et al. 2004, Goul- son et al. 2006, Rundlöf et al. 2008, Ahrné et al. 2009, Potts et al. 2010, Goulson et al. 2011) as well as the survival of other organisms (Goulson et al. 2006, Goulson et al. 2011). Economically, bees are important in agriculture because a wide variety of crops depend on insect pollination (Goulson 2010).

Bumblebees (any member of the bee genus Bombus) are social insects that belong to the family Apidae and the order Hymenoptera. Approxi- mately 250 species of bumblebees have been identified worldwide (Goulson 2010, Goulson et al. 2011). They live in organized colonies and are mainly found in temperate regions, alpine and arctic zones (Goulson 2010). Bumblebees are characterized by their particularly large and hairy bodies and for being moderately slow flying insects. Most bumblebee species have an annual life cycle (Goulson 2010) (Fig- ure 1). After hibernation, queens emerge in late winter or spring and start looking for foraging resources (i.e., nectar and pollen) and suitable places to build their nests. Then they prepare wax pots in which to store the food, and wax cells in which to lay their eggs. The queen has to forage frequently, besides incubating the brood, to provide enough pollen supply (Goulson 2010). At this stage, the bumblebee queen and her colony seem to be very vulnerable, as a reduction in food resources in the habitat or an extreme change in the weather conditions could have fatal consequences (Goulson 2010). After 4–5 weeks the bumblebee workers emerge and start foraging, whilst the queen continues laying eggs. Towards the end of the summer males and new queens are produced. The males and the virgin queens mate; later the males die. In autumn the old queen stops laying eggs and the colony gradually dies out. The new queens, already fertilized, leave the nest and search for a sheltered place to hibernate during winter. The following spring the cycle is repeated once again.

13 Figure 1. Summary of the annual life cycle of a bumblebee colony (Image taken from: www.bumblebee.org/images/lifecycle.jpg)

Cuckoo bumblebees (subgenus Psithyrus) are social parasites that depend on other bumblebee species (host species); they have an annual life cycle similar to other bumblebee species, but they emerge from hibernation later in the season and take over the nests of ‘true’ bumblebee queens (Goulson 2010).

During the life cycle of bumblebees, semi-natural habitats constitute areas of important value as they provide essential resources, such as flowering plants and nesting sites (Svensson et al. 2000, Kells and Goulson 2003). In Europe, it has been found that species richness of bumblebees tends to be particularly high in flower-rich meadows (Goulson 2010).

Among bees, bumblebees are considered the best documented group in the existing literature (Potts et al. 2010). According to Goulson et al.

14 (2011), bumblebees have been well-studied in different parts of Western Europe, the United Kingdom, Asia and North America, but very little is known about the distribution and ecology of most of the bumblebee species that live in other regions. In addition, there is insufficient information about the habitat requirements and characteristics of many species of bumblebees (Goulson et al. 2011). In Europe and North America, many bumblebee species have declined and become extinct at local levels, whereas other species are still common and widely distributed (Ah- rné et al. 2009). The causes of these differences in response are not clear, but they appear to involve particular characteristics of single species, such as diet and foraging distances (Ahrné et al. 2009). Williams (2005) suggested that some species have a more specific habitat, but Goulson et al. (2006) argued that bumblebees are generally not habitat specialists because they are found in more than one biotope. However, data from monitoring in Finland have shown that some species prefer particular types of habitat; e.g., the bumblebee B. schrencki is a species said to prefer moist forests and forest margins, whereas the species B. soroeensis prefers dry and open landscapes (Söderman 1999).

The decline of bees, particularly bumblebees, has been widely recognised in recent decades. Evidence of significant range declines has been documented across the world and in several European countries (Westphal et al. 2003, Samways 2005, Goulson et al. 2006, Williams and Osborne 2009, Potts et al. 2010): in the United Kingdom, for example, seven of the 27 bumblebee species have acquired the status of endangered species (Goulson 2010), and six of the 16 non-parasitic bumblebees have declined significantly, including one species that has become extinct (i.e., B. subterraneus) (Potts et al. 2010). Moreover, extinctions of 13 species in at least one country between 1950 and 2000 were reported in a review on bumblebees that included 11 central and western European countries; and four species were declared extinct in the whole region (Kosior et al. 2007, Goulson 2010). The decline of bumblebees has been widely associated with agricultural intensification which is characterized by rigorous use of fertilizers and pesticides, and the reduction of flower resources and suitable habitats (Carvell 2002, Mänd et al. 2002, Goulson et al. 2006, Holzschuh et al. 2008, Xie et al. 2008, Ahrné et al. 2009). Other causes that have been mentioned in the literature are commercial beekeeping and the introduction of non-native bumblebees for pollination of greenhouse crops, which have threatened the existence of some native species, through competition, the proliferation of new pathogens,

15 etc. (Goulson 2010). Some bumblebee species (such as B. terrestris) have been commercialized in large numbers: approximately one million or more colonies have been shipped all over the world (Goulson 2010).

Bumblebees have different tongue lengths depending on the species. Long-tongued bumblebees are particularly important because they are pollinators of deep perennial flowers (they prefer deep-corolla flowers). A perennial is a plant that lives for more than two years; they are commonly found in natural and semi-natural ecosystems. Longer-tongued species have been found to be more vulnerable than bumblebees with a short-length tongue (Goulson et al. 2008). Narrower food-plant specialization and a larger tongue length have been suggested as characteristics that confer greater susceptibility to decline on some bumblebee species (Williams and Osborne 2009); e.g., Goulson et al. (2005) found that some long-tongued bumblebee species that gather pollen from the plant family Fabaceae have declined. Specialist species depend on specific plant families or plant species, thus their decline may be related to the reduction of suitable food resources (Williams and Osborne 2009). However, many studies agreed that further research is required on the ecology of rare species and the role of diet specialisation in bumblebee decline (e.g., Goulson et al. 2005, Goulson et al. 2006, Williams and Osborne 2009).

2.2. The use of patch and landscape factors in ecological studies

Patch can be defined as “a relatively homogeneous area that differs from its surroundings” (Freemark et al. 2002). For the purposes of this thesis, patch-scale refers to local-scale; so patch-scale variables mean the characteristics or factors at the scale of the habitat or meadow under study. Morrison et al. (1998) defined habitat as an area with the resources (e.g., food, water, shelter) and environmental conditions (e.g., climate, presence or absence of predators) that permits the occupancy, reproduction and survival of a particular species.

The term “landscape” from a human perspective can be described as an area that is composed of multiple elements (or “patches”) of different types of land cover, and the variety of these elements creates heterogeneity within an area (Wiens 2002). From an ecological perspective, what comprises a landscape will depend on the scales over which a given species moves and its perception of the surroundings (Manning et al. 2004),

16 then the relevant scale of a landscape will be connected with the organism or the ecological process under study and the questions considered (Freemark et al. 2002).

Patch and landscape indices have been widely used in numerous studies as predictors of species presence and abundance of invertebrates (Hanski et al. 1995, Bäckman and Tiainen 2002, Steffan-Dewenter 2003, Billeter et al. 2008, Pocewicz et al. 2009). However, there is not enough information about the influence of landscape factors on the diversity and abundance of bumblebees (Hatfield and LeBuhn, 2007), as most conservation strategies have focused on habitat characteristics and requirements of species at patch scale, but not on the potential effects of the surrounding matrix (Steffan-Dewenter et al. 2002, Cozzi et al. 2008). Steffan- Dewenter et al. (2002) found that only few studies provide some insight into how the surrounding landscape context is affecting the community structure and diversity of bee species. The effectiveness of biodiversity conservation usually depends on knowledge of the influence of different factors at multiple scales on the distribution of organisms and the conditions that allow them to move across the landscape (Gutzwiller 2002). Even more, some authors have recognized the importance of landscape structure on the shaping of insect assemblages and communities (e.g. Steffan-Dewenter et al. 2002, Sepp et al. 2004, Samways 2005, Holzs- chuh et al. 2008, Rundlöf et al. 2008).

2.3. Habitat quality: definition and importance

Many definitions have been given to habitat quality. As early as 1983, Van Horne defined habitat quality as “the product of density, mean individual survival probability, and mean expectation of future offspring, for residents in one area as compared to other areas” (Van Horne 1983). In our study, we adopted the one presented by Hall et al. (1997); they defined habitat quality as the capacity of the environment to supply appropriate conditions for individual and population persistence. In a habitat with low quality, a species may decline, lower its abundance, or may have problems to reproduce (Mortelliti et al. 2010).

The main processes that can cause landscape change are habitat fragmentation, decline and degradation; these are also regarded as important threats to biodiversity (Fischer and Lindenmayer 2007). Habitat degradation is a process that involves the gradual deterioration in quality of an

17 area of habitat (Fischer and Lindenmayer 2007, Mortelliti et al. 2010). Habitat quality has been recognized as an important element in metapopulation models and in landscape metrics (Thomas et al. 2001, Mortel- liti et al. 2010), many studies have reported that the inclusion of “patch quality” variables enhances the explanatory power of models (Mortelliti et al. 2010). There is evidence that habitat quality plays an important role in shaping the patterns of species’ distribution and in controlling the spatial dynamics in fragmented landscapes (Thomas et al. 2001); this means that habitat quality could influence whether species occupy a particular patch and how they move across the landscape.

18 3. AIMS OF THE STUDY

The main objective of this thesis is to show how abiotic and biotic factors, at patch and landscape scale, influence the abundance and species richness of bumblebees in semi-natural meadows. In addition, the relations between local abundance of bumblebee species and the main land cover type in our study region (i.e., forest) were analysed. In this study, we considered patch and landscape characteristics that may have potential relevance to the ecology of bumblebee species richness and abundance; some of them have been used in previous studies on insects (e.g., Mazerolle and Villard 1999, Krauss et al. 2003, Kumar et al. 2009).

We hypothesised that patch and landscape factors influence bumblebee richness and abundance differently; some factors may have a positive effect whereas others may have a negative effect. The relations between bumblebees and the landscape variables may depend on the foraging distances and landscape preferences of the species. Bumblebees have been well studied in regions with warmer climates (compared to Estonia) and open landscapes, mostly in agricultural landscapes (e.g., Steffan-Dewen- ter, 2003; Hatfield and LeBuhn, 2007; Krewenka et al., 2011). To our knowledge, very few studies have been developed on the influence of local and landscape factors on bumblebees in forested landscapes. The specific goals of this thesis were:

• To estimate the species richness and abundance of bumblebees in semi-natural meadows across Northeast Estonia (Paper I). • To evaluate the influence of a set of patch-scale and landscape-scale variables on the species richness and abundance of bumblebees (Pa- per II), and specifically on the species richness and abundance of long-tongued bumblebees (Paper III). • To analyse how the presence of forest habitats (i.e., forest and brushwood) at different spatial scales influence the local abundance of bumblebee species (Paper IV). • To investigate the potential use of bumblebees as indicators for the evaluation of habitat quality (Paper III).

19 4. MATERIALS AND METHODS

4.1 Study region

This study was carried out in Ida-Virumaa County, which is located in northeastern Estonia (Figure 2). The county of Ida-Virumaa is bordered to the north by the Finnish Gulf, to the east by the Narva River, and to the south by the Lake Peipsi. On the west and south-west, it is surrounded by Lääne-Viru and Jõgeva counties, respectively. Ida-Virumaa is a region affected by mining activities as it contains large deposits of oil shale, a mineral used for power generation in Estonia. Ida-Virumaa is characterised by a very patchy landscape pattern with a variety of land cover types, predominantly forest (195,245 ha, approximately 58.0% of the territory in the region), arable land (41,671 ha, approximately 12.4%), brushwood (21,701 ha, approximately 6.5%) and meadows (19,031 ha, approximately 5.7%), and small proportions of human settlements. The total area of our study region is 336,400 ha, which represents about 7.5% of the total area of Estonia (the territory of Estonia occupies 45,000 km2, Peterson and Aunap 1998).

Figure 2. Map of the study region, Ida-Virumaa, with the location of the 22 study sites.

20 Table 1. Number of species and individuals of bumblebees at each study site (modified from Paper I). # Study site Geographic Area Bumblebees coordinates (ha) Number of Number of species individuals 2008 2009 2008 2009 1 Smolnitsa 59º00’38”N 0.12 8 10 16 21 27º36’52”E 2 Kuningaküla 59º07’35”N 0.69 7 12 13 36 27º48’10”E 3 Raadna Oja 58º58’53”N 3.38 1 2 3 4 27º07’31”E 4 Lemmaku 59º00’20”N 3.61 5 7 7 15 27º09’36”E 5 Mäetaguse 59º13’40”N 3.65 5 8 6 15 27º19’44”E 6 Atsalama 59º14’31”N 1.47 3 9 9 22 27º23’12”E 7 Kurtna 59º14’59”N 1.12 4 6 6 9 27º33’58”E 8 Pannjärve 59º17’18”N 0.48 8 11 12 24 27º33’24”E 9 Tagajõgi Roostoja 59º04’54”N 0.23 6 7 11 11 27º04’37”E 10 Tagajõgi Oonurme 59º07’12”N 1.02 5 7 10 12 26º59’26”E 11 Piilse 59º14’09”N 3.87 4 5 9 6 26º59’31”E 12 Kalvi 59º28’52”N 3.32 5 16 11 29 26º46’36”E 13 Kalvi Oru 59º27’34”N 1.23 3 8 3 15 26º48’24”E 14 Liimala 59º26’06”N 1.42 8 11 15 22 26º59’01”E 15 Moldova 59º25’59”N 1.91 7 9 11 17 27º04’11”E 16 Valaste 59º26’18”N 0.89 1 1 1 3 27º25’20”E 17 Päite 59º24’16”N 1.06 3 10 4 16 27º42’02”E 18 Udria 59º23’36”N 0.82 5 9 10 23 27º54’07”E 19 Soldina 59º23’01”N 0.26 9 14 14 31 28º04’43”E 20 Auvere 59º20’36”N 3.59 6 7 9 14 27º58’20”E 21 Narva Karjäär 59º15’52”N 0.91 6 11 12 29 27º49’15”E 22 Meriküla 59º24’46”N 1.37 8 9 15 16 27º57’12”E

21 This research was conducted in 22 semi-natural meadows (Figure 2, Ta- ble 1) that differ in vegetation characteristics, size and shape and in their surrounding landscape composition and configuration. Some sites were situated in coastal areas, others were surrounded mainly by forests or grasslands and few study sites were located close to urban areas. The areas of the selected meadows ranged from 0.10 to 3.83 ha (Table 1).

4.2. Bumblebee survey

The fieldwork was carried out during the summers of 2008 and 2009. Every year, we sampled each meadow twice: first during June, and second between the end of July and the beginning of August. The recordings were made via systematic walking surveys at a slow pace. Bumble- bee counts were conducted during the warmer time of the day, between 10:00 h and 16:00 h, in suitable weather conditions, when the temperature was above 18 ºC and the wind speed was less than five on the Beau- fort scale. Our fieldwork method was based on Goulson et al. (2006). Each visit was done during a period of about one hour, or until the observer was satisfied that most of the species on the site were recorded. The species richness and abundance of bumblebees was determined by counting individuals and identifying species by sight, mainly when they were visiting flowers. When the observer could not identify the bumblebee species on site, some individuals were caught with an insect net for later identification in the laboratory. The nomenclature used was based on Fauna Europaea (2011).

The weather conditions in Estonia are continuously measured by the Estonian Meteorological and Hydrological Institute. We obtained data from the Jõhvi weather station, which is situated in the middle of the study area. In the summer of 2008 (i.e., June, July and August), average air temperature ranged from 13.8 to 15.9 ºC (minimum temperature was between 1.6 and 3.7, and maximum temperature was between 24.1 and 27.7), whereas, in 2009, average air temperature was between 13.4 to 16.6 ºC (minimum temperature was between 2.2 and 3.6, and maximum temperature was between 24.4 and 27.7). The average monthly air temperature was relatively consistent from 2008 to 2009; however in 2009, it was higher in May (by 1.1°C), July (by 0.7ºC) and August (by 0.2ºC), and lower in June (by 0.4°C). Larger differences were found in the total precipitation from year to year. In June 2008 the total precipitation was 127.9 mm (112.1 mm in 2009), in July the total precipitation was higher

22 in 2009 than in 2008 (139 mm and 54.5 mm, respectively), and in Au- gust 2008 there was a period of heavy rain and the total precipitation reached 242.2 mm (96.1 mm in 2009). In general, the summer of 2009 had more sunny and favourable weather conditions for bumblebees.

4.3. Patch-scale factors

4.3.1. Vegetation structure

At each study site, we recorded variables that described the structure of the vegetation. Specifically, we registered the number of species of flowering plants (or potential food plants for bumblebees), the percent cover of flowering plants and grass height. The percent cover of flowering plants was determined through a visual estimation of the overall coverage and was performed by the same person at all study sites. We used a direct method to measure grass height (Stewart et al. 2001). The “direct measurement method” consists of placing a hand lightly on the vegetation at the level below which about 80% of the vegetation is estimated by eye to be growing, ignoring occasional tall stalks, and then reading the height with the help of a rule (Stewart et al. 2001). In the statistical analyses, we used the total number of species of flowering plants recorded in the first year, and the arithmetic mean of the four observations (one per visit) of percent cover of flowering plants and grass height. The cover of flowering plants was measured in percentages to an accuracy of 10%.

4.3.2. Spatial characteristics

In order to describe some spatial characteristics at the patch scale, five indices were calculated at each study site using Fragstats (Version 3.3): patch area (AREA), patch perimeter (PERIM), shape index (SHAPE), fractal dimension index (FRAC) and edge density (ED) (Appendix 1).

SHAPE characterises the complexity of a patch shape compared to a standard shape. In this study, the raster version of Fragstats was used, which evaluates patch shape with a square standard; this means that shape index is minimum for square patches and increases as patches become increasingly non-square (McGarigal and Marks 1995). Shape index is commonly applicable and has been widely used in landscape ecological research (Forman and Godron 1986). FRAC is another measure of shape complexity (McGarigal and Marks 1995), but it is calculated based on

23 patch size. The concept of fractal was introduced by Mandelbrot (1977); it is described as a geometric form that displays structure at all spatial scales. To calculate the fractal dimension of natural planar shapes, he proposed a perimeter-area method. This method quantifies the degree of complexity of the planar shapes (McGarigal and Marks 1995). On the other hand, ED (or alternatively Perimeter Area Ratio – PAR) at the patch level is a function of the patch perimeter and it takes into account the shape and the complexity of the patch (McGarigal and Marks 1995).

4.4. Landscape-scale factors

4.4.1. Landscape composition

To analyse the composition of the landscape, we calculated the proportion of different land cover types in the area surrounding each study site using ArcGIS 9.3, excluding the area of the study site itself. We used a digital Estonian Basic Map provided by the Estonian Land Board at a scale of 1:10,000. The original maps included more than 30 different land cover types that were organised into 11 categories: meadows, forests, brushwood, mires, arable land, abandoned peatland, bodies of fresh water, sea, green areas, human settlements and others. All of them were calculated at four spatial scales: 250 m, 500 m, 1000 m and 2000 m radius. In our analyses, we used five of these land cover types, only those that had an important presence in our study region (the land cover types than occupy more than 5% of the territory) (Appendix 1): forests, brushwood, meadows, human settlements and arable land. The forest cover in our study region is mainly composed of managed mixed forest; birches, pines and spruces are among the dominant trees. The brushwood cover is characterized by the presence of deciduous trees, woody seedlings, shrubs and young trees, primarily willows, maples, birches, among others, which have become established on abandoned agricultural land, in overgrown meadows or in forest clearings.

4.4.2. Landscape configuration

Four landscape indices were used to describe the configuration of the landscape, calculated with Fragstats (Version 3.3) (Appendix 1): patch richness density (PRD), interspersion and juxtaposition index (IJI), edge density at landscape level (ED_LAND) and Shannon’s diversity index (SHDI). PRD was used to standardize patch richness to a per area basis

24 (McGarigal and Marks 1995). We used IJI to measure the extent to which patch types are interspersed (not necessarily dispersed); higher values are given to landscapes in which the patch types are well interspersed (or equally adjacent to each other), whereas lower values are given to landscapes in which the patch types are poorly interspersed (or the distribution of patch type adjacencies is disproportionate) (McGarigal and Marks 1995). Interspersion can be defined as “the degree to which a given patch or landscape type is scattered rather than aggregated or clumped”, and juxtaposition is the “adjacency of different patch or landscape types” (Freemark et al. 2002). ED_LAND equals the length of all borders between different patch types (classes) in a reference area divided by the total area of the reference unit; in contrast to patch density, edge density takes the shape and the complexity of the patches into account (Eiden et al. 2000). Edge density measures the complexity of the shapes of patches and, similar to patch density, is an expression of the spatial heterogeneity of a landscape mosaic (Eiden et al. 2000). Additionally, SHDI was used to measure the diversity of the landscape based on two components: richness, defined as the number of different patch types, and evenness in the distribution of areas among patch types (Eiden et al. 2000).

In addition, we used two extra indices in order to describe the distribution of patches of the most predominant land cover type in the study region, i.e., forest, across the landscape mosaic: mean patch area of forest (AREA_ MN-Forest) and edge density of forest (ED_Forest). AREA_MN-Forest equals the sum, across all patches of the corresponding patch type (here, forest) of the area of the patches, divided by the total number of patches of the same type (McGarigal et al. 2002). We used “mean patch size” because it gives information about the size of the patches and the number of patches at the same time. Edge density equals the sum of the lengths of all edge segments involving the corresponding patch type, in this case forest, divided by the total landscape area and multiplied by 10,000 (to convert to hectares) (McGarigal et al. 2002). All landscape variables were estimated at four spatial scales (i.e., 250 m, 500 m, 1000 m and 2000 m radius), measured as circles around the centre of each study site.

4.5. Data analysis

We develop our analyses in various steps. First, we conducted dependent samples (paired) t-tests to evaluate differences between bumblebee species richness in 2008 and 2009, and between bumblebee abundance

25 in the same years. We used Spearman rank order correlations to analyse the relationships between bumblebees (i.e., total abundance, calculated as the total number of individuals found in 2008 and 2009; and total species richness, calculated as the total number of species found during the two years) and: the patch-scale factors (i.e., number of species of flowering plants, average percent cover of flowering plants, average grass height, AREA, PERIM, SHAPE, FRAC and ED) and the landscape- scale factors (i.e., the proportion of different land cover types, PRD, IJI, ED_LAND, SHDI and AREA_MN-Forest; calculated at four spatial scales: 250 m, 500 m, 1000 m and 2000 m radius). When the correlation coefficient (rs) was between 0.0 and ±0.3, the correlation was considered weak; when rs was between ±0.3 and ±0.6, the correlation was medium; and when rs was between ±0.6 and ±1, the correlation was strong; in all cases, the correlation was statistically significant if the p value was less than 0.05.

In addition, to simultaneously examine the connectivity patterns (also called latent factors) of the patch and landscape characteristics, and the overall bumblebee species richness and abundance the Partial Least Squares (PLS) analysis was applied. PLS is the multivariate statistical technique particularly well suited for situations where multicollinearity exists in the dataset and the number of variables is high compared to the number of observations (about PLS in ecological studies, see Carrascal et al. 2009, for example; for a detailed description of PLS, see Krishnan et al. 2011). To test the statistical significance of latent factors the permutation test with 10,000 permutation samples was applied. This permutation also served to assess the singular vectors, giving a threshold to decide which variables were contributing the most of the latent factor.

Second, we calculated the total number of species and individuals of long-tongued bumblebees found in our study sites. Then, we examined the relations between long-tongued bumblebee species richness and abundance and the local and landscape factors using also Spearman rank order correlations.

Third, we analysed the relations between the local abundance of each bumblebee species with the proportion of forest and brushwood, and other forest-related variables (i.e., AREA_MN-Forest and ED_Forest); these variables were chosen due to their importance in our study region and their potential impact on bumblebees. To analyse these relations we

26 used Spearman rank order correlations. For each landscape variable and species combination the scale corresponding to the strongest relationship was selected and the statistical significance of the selected relationships was tested, considering 24 tests (one for every species) performed with a single landscape variable (at the most suitable scale) and the species abundance as one experiment, and applying the Benjamini-Hochberg correction to the p values. Additionally, multiple regression analysis was performed to study the joint effect of the four landscape variables at different scales on the local abundance of individual species, using only the bumblebee species with more than 20 individuals.

We used STATISTICA 9 software to perform the t-tests, correlation analyses and multiple regression analysis. The PLS analysis was performed with SAS 9.1 software.

27 5. RESULTS

5.1. Total bumblebee species richness and abundance

5.1.1. Bumblebee species richness and abundance in northeast Estonia

We counted 597 individuals of bumblebees in total in our study sites: 207 in 2008 and 390 in 2009 (Paper I). From the total number of individuals, we found 363 workers, 150 males and 84 queens (Paper I). A total of 24 species of bumblebees (genus Bombus), including 5 species of cuckoo bumblebees (subgenus Psithyrus), were recorded in the study region (Paper II). They represent approximately 83% of the total bumblebee species found in Estonia. In the country, 29 species of bumblebees have been recorded, including 7 species of cuckoo bumblebees. An average of 5.32 species (SD = 2.25) and 9.41 individuals (SD = 4.22) of bumblebees per study site were found in 2008, and 8.59 species (SD = 3.45) and 17.73 individuals (SD = 8.78) in 2009 (Paper II).

The places with the highest overall species richness of bumblebees were Kalvi with 18 species in total, and Soldina, Narva Karjäär and Pannjärve with 15 species each; and the places with the lowest overall number of species were Valaste and Raadna Oja (Paper I) (Table 1).

The number of species and individuals of bumblebees were significantly higher in 2009 compared with the previous year (t = 6.0, df = 21, p < 0.001; t = 5.7, df = 21, p < 0.001, respectively) (Paper I). However, the species richness adjusted to the common number of individuals was not statistically significantly different in 2008 and 2009 (t = 1.58, df = 42, p = 0.121). Also, total bumblebee species richness was strongly positively correlated with total bumblebee abundance (rs = 0.94, p < 0.001); but, after rarefaction was applied to adjust total bumblebee species richness, this relationship was weak and not significant (rs = 0.27, p = 0.233) (Pa- per II).

The bumblebee species with the highest number of individuals were B. pascuorum, B. lucorum and B. ruderarius with 140, 70 and 58 individuals, respectively (Paper I and II) (Figure 3). These three species, together with B. cryptarum, were also the most common bumblebees in the area, as they were found in most of the study sites. On the other hand, the two

28 Figure 3. Total number of individuals (Log10 transformed) per bumblebee species (Modified from Paper I).

rarest bumblebee species with the lowest abundance were B. muscorum and B. distinguendus (Paper I and II) (Figure 3). Three of the bumblebee species found (i.e., B. distinguendus, B. muscorum and B. soroeensis) are in the Estonian Red List of Threatened Species (http://elurikkus.ut.ee) (Paper I).

Among the species that we found in our study sites, five were long- tongued bumblebees: B. distinguendus, B. hortorum, B. ruderarius, B. sylvarum and B. pascuorum (Paper I and III); and eight were short- tongued bumblebees: B. cryptarum, B. lapidarius, B. lucorum, B. terrestris, B. hypnorum, B. jonellus, B. pratorum and B. soroeensis (http:// www.nhm.ac.uk). The other species of bumblebees found had a mid- length tongue. Total species richness and abundance of long-tongued bumblebees ranged from 1 to 4 species and from 2 to 29 individuals, respectively (Paper III).

29 5.1.2. Relations between bumblebees and patch-scale factors (Paper II)

A total of 133 species of flowering plants were found in our study sites (Appendix 2). Flowering plant species richness ranged from 7 to 43 species per study site. We found that bumblebee abundance was strongly positively correlated with flowering plant species richness (rs = 0.65, p < 0.001).

Concerning the relations between the spatial characteristics of the study sites and bumblebees, we found that bumblebee species richness was strongly negatively correlated with shape index (SHAPE) (rs = -0.60, p = 0.003) and medium negatively correlated with fractal dimension index

(FRAC) (rs = -0.57, p = 0.004). There were not significant relationships either between bumblebees and other spatial characteristics, or between bumblebees and average grass height or average percent cover of flowering plants (p > 0.05).

5.1.3. Relations between bumblebees and landscape-scale factors (Paper II)

We found that the proportion of human settlements in the areas surrounding the study sites was positively correlated with bumblebee abundance at 250 m and 1000 m (rs = 0.48, p = 0.024; rs = 0.51, p = 0.014, respectively). Additionally, bumblebee species richness was positively correlated with the proportion of meadows at the largest spatial scale, i.e., 2000 m (rs = 0.51, p = 0.015).

Concerning the relations between bumblebees and landscape indices, we found positive correlations between bumblebee abundance and Shan- non’s diversity index (SHDI) at 2000 m and edge density (ED_LAND) at 1000 m (rs = 0.44, p = 0.039; rs = 0.50, p = 0.018, respectively).

In contrast, we found that the proportion of forest was negatively correlated with bumblebee species richness at the spatial scales of 1000 m and 2000 m (rs = -0.45, p = 0.036; rs = -0.47, p = 0.025, respectively). Also, negative correlations were detected between proportion of brushwood and bumblebee species richness at 250 m and 500 m (rs = -0.57, p = 0.005; rs = -0.44, p = 0.040, respectively). Mean patch area of forest (AREA_MN-Forest) at the largest spatial scale was also negatively correlated with bumblebee species richness (rs = -0.51, p = 0.015).

30 5.1.4. Connectivity patterns between bumblebees and the factors at patch and landscape scale (Paper II)

Two connectivity patterns were identified with Partial Least Squares (PLS) analysis, which together accounted for 100% and 31.5% of bumblebee richness and abundance variance, and patch and landscape characteristics variance, respectively (in Figure 4, the percentages are presented separately for the two connectivity patterns).

The main result of PLS is the first connectivity pattern that connects mainly the overall number of species and individuals of bumblebees with

Figure 4. Results of the Partial Least Squares (PLS) analysis. The dots mark the location of the patch and landscape characteristics (X) and the squares with arrows mark the location of the bumblebee species richness (adjusted) and abundance (Y) in relation to the two connectivity patterns. The dotted lines denote the approximate cut-off for statistical significance of the right singular vectors (patch and landscape characteristics vectors) as assessed through permutation tests (p = 0.05); for clarity only the patch and landscape characteristics with p < 0.1 are shown with the variable name. R250, R500, R1000 and R2000, denote the different spatial scales at which the landscape factors were calculated. (For a description of the variables, see the Appendix 1)

31 the patch and landscape characteristics (First singular vector, Figure 4). According to the permutation test, the overall bumblebee richness and abundance were significantly positively related with the proportion of human settlements, especially at the smallest spatial scale (p < 0.05). In contrast, the proportion of arable land at the scale of 250 m and mean patch area of forests (AREA_MN-forest) (especially at larger spatial scales) showed negative relations (p < 0.05) with the overall bumblebee richness and abundance pattern, indicating that the larger the values of these variables, the smaller the number of species and individuals of bumblebees.

As an additional result of PLS, a second connectivity pattern (Second singular vector, Figure 4) was found. It reflects the differences in relative bumblebee species richness (i.e., how heterogeneous or homogeneous are the study sites in relation to the number of individuals) and its relation with the patch and landscape characteristics. Statistically significant were only the second singular vector values corresponding to the proportion of arable land at the spatial scale of 2000 m, the proportion of meadows at the same spatial scale, and the proportion of human settlements at 500 m (p < 0.05). The second singular vector values of the proportion of arable land and the proportion of meadows were positive, which mean that the relative species richness of bumblebees may be higher (for the same number of individuals) in places with a higher proportion of arable land and meadows (particularly at the largest spatial scale). The negative second singular vector values of the proportion of human settlements indicate that the relative bumblebee species richness may be lower (for the same number of individuals) in places with a higher proportion of human settlement (especially at the smallest spatial scale). These results are also true in the case of bumblebee abundance, but vice versa: the number of individuals of bumblebees may be higher (for the same number of species) in places with a higher proportion of human settlements and lower proportions of arable land and meadows.

5.1.5. Models to predict bumblebee species richness and abundance (Paper II)

The regression models based on the patch and landscape factors tested here explained 83% and 73% of the variation in total bumblebee abundance and species richness, respectively (Table 2). Both models were highly statistically significant. The model for total bumblebee abundance

32 included four variables: one patch-scale factor and three landscape-scale factors. Species richness of flowering plants was significantly positively related with bumblebee abundance and emerged as the most important predictor in our model. In contrast, two landscape variables, i.e., the proportion of arable land and the mean patch area of forest (AREA_ MN-forest), were negatively related with bumblebee abundance.

Five variables were included in the model for total bumblebee species richness: two patch-scale factors and three landscape-scale factors (Table 2). The most important predictor of bumblebee abundance was shape index (SHAPE). Patch area (AREA) was significantly positively related with bumblebee richness, whereas SHAPE, the proportion of arable land and the AREA_MN-forest were all negatively related with the dependent variable.

Table 2. Regression models for total bumblebee abundance and total bumblebee species richness (adjusted to the common number of individuals)

Dependent R2 Variable included in the model Regression p value variable coefficient Total bumblebee 0.83* Species richness of flowering 0.44 < 0.001 abundance plants Proportion of arable land at -0.08 0.008 250 m Mean patch area of forest -0.12 0.048 (AREA_MN-forest) at 1000 m Edge density at landscape level 0.09 0.120 (ED_Land) at 1000 m Total bumblebee 0.73* Patch area (AREA) 0.18 0.028 species richness Shape index (SHAPE) -0.88 0.003 Proportion of arable land at -0.05 0.025 2000 m Mean patch area of forest -0.16 <0.001 (AREA_MN-forest) at 1000 m Patch richness density (PRD) at -0.13 0.103 500 m

* Significant at p < 0.001

33 5.2. Relations between long-tongued bumblebees and the factors at patch and landscape scale

We found that both species richness and abundance of long-tongued bumblebees had strong positive correlations with flowering plant species richness and percent cover of flowering plants (Figure 5). At landscape scale, we found that total species richness of long-tongued bumblebees correlated positively with the proportion of meadows at 1000 m and

2000 m (rs = 0.55, p = 0.008; rs = 0.58, p = 0.004, respectively). In addition, long-tongued bumblebee abundance was positively correlated with edge density at landscape level (ED_LAND) at 500 m and 1000 m (rs =

0.44, p = 0.040; rs = 0.42, p = 0.040, respectively). Positive relationships were also found between species richness of long-tongued bumblebees and Shannon’s diversity index (SHDI) at 2000 m (rs = 0.54, p = 0.009). (Paper III)

Figure 5. Spearman rank correlations between the species richness of flowering plants and: (a) long-tongued bumblebee species richness and (b) long-tongued bumblebee abundance; and between the average percent cover of flowering plants and: (c) long- tongued bumblebee species richness and (d) long-tongued bumblebee abundance.

34 In contrast, negative correlations were found between species richness of long-tongued bumblebees and proportion of forest at 500 m and 1000 m (rs = -0.45, p = 0.040; rs = -0.42, p = 0.040, respectively) (Paper III), and also with mean patch area of forest (AREA_MN-Forest) at 500 m and 2000 m (rs = -0.46, p = 0.035; rs = -0.50, p = 0.025, respectively). Other variables at patch scale (i.e., AREA, PERIM, SHAPE, FRAC and ED) and landscape scale (i.e., proportion of arable land, brushwood and human settlements, PRD and IJI) do not appear to be important for long-tongued bumblebee species richness and abundance.

5.3. Relations between the local abundance of bumblebee species and forest habitats

5.3.1. Proportion of forest (Paper IV)

We found that two species of cuckoo bumblebees have medium positive correlations with the proportion of forest: P. bohemicus and P. norvegicus at 250 m (rs = 0.54, p = 0.040; rs = 0.57, p = 0.040, respectively). Among the bumblebee species known to prefer forest and forest margins, we found that B. schrencki was positively associated with the proportion of forest at 500 m (rs = 0.58, p = 0.040). In addition, the most abundant species in the study area, B. pascuorum, was medium positively correlated with this variable at 250 m, but the relationship was nearly statistically significant (rs = 0.49, p = 0.070).

In contrast, negative correlations were found between some species of bumblebees and the proportion of forest. B. veteranus and B. terrestris, both species have negative relationships with the proportion of forest in a similar progressive trend, i.e., the larger the spatial scale, the stronger the relationship between the variables (Figure 6); the strongest relationship was found at the spatial scale of 2000 m (in the case of B. veteranus, rs = -0.63, p = 0.032; and in the case of B. terrestris, rs = -0.55, p = 0.040). On the contrary, B. ruderarius and B. lapidarius had nearly significant negative correlations with the proportion of forest at the smallest spatial scale, i.e., 250 m (rs = -0.46, p = 0.091; rs = -0.50, p = 0.061, respectively).

35 Figure 6. Relationships between the local abundance of bumblebee species and the studied landscape characteristics at various spatial scales based on the Spearman rank correlation coefficients (rs). The width of the circle indicates the strength of the relationship (the bigger the circle, the stronger the relationship between the variables) and the colour determines the direction of the relationship (black circles correspond to positive relationships and white circles to negative relationships). The stars (*) inside the circles indicate the statistically significant correlations, after Benjamini-Hochberg correction (p < 0.05), for the spatial scale with the strongest relationship (Other correlations, that were statistically significant before correction, are described in the text). (Paper IV).

5.3.2. Proportion of brushwood (Paper IV)

We found a medium positive correlation between B. schrencki and the proportion of brushwood at a large spatial scale (at 1000 m, rs = 0.60, p = 0.031). B. pascuorum was also positively correlated with this variable at 1000 m, but this correlation did not remain statistically significant after Benjamini-Hochberg correction (rs = 0.44, p = 0.120). In addition, we found that the subspecies B. s. soroeensis and B. s. proteus showed opposing relationships with the proportion of brushwood (Figure 6): B. s.

36 soroeensis was positively correlated at 500 m, whereas B. s. proteus was negatively correlated at 250 m; however, these correlations were not statistically significant after the correction (rs = 0.44, p = 0.120; rs = -0.45, p = 0.120, respectively). Other negative correlations were detected between the proportion of brushwood and some species of bumblebees, i.e., B. terrestris at 500 m (rs = -0.56, p = 0.040), B. veteranus at 1000 m

(rs = -0.63, p = 0.031), and B. lapidarius with a nearly significant correlation at 500 m (rs = -0.54, p = 0.050). Also, P. bohemicus appeared to have a nearly significant negative correlation with the proportion of brushwood at 250 m (rs = -0.52, p = 0.060); in contrast, this species was found to be positively correlated with the proportion of forest at the same spatial scale (see previous section) (Figure 6).

5.3.3. Landscape indices (Paper IV)

Some bumblebee species showed significant positive relationships with edge density of forest (ED_Forest), i.e., B. pascuorum, B. pratorum and P. sylvestris at 2000 m; however, none of these correlations remained statistically significant after Benjamini-Hochberg correction (rs = 0.45, p =

0.150; rs = 0.43, p = 0.150; rs = 0.46, p = 0.150, respectively). In addition, medium negative relationships were detected between the ED_For- est and some bumblebee species (but these correlations were not statistically significant after the correction): B. sylvarum at 250 m (rs = -0.49, p

= 0.150), B. s. proteus at 1000 m (rs = -0.46, p = 0.150) and B. veteranus at 2000 m (rs = -0.46, p = 0.150). We found medium negative relationships between the mean patch area of forest (AREA_MN-Forest) and some bumblebee species: B. terrestris and B. veteranus at 2000 m (rs = -0.59, p = 0.040; rs = -0.65, p = 0.020, respectively). There were other nearly significant negative correlations between AREA_MN-Forest and two bumblebee species at 500 m, i.e.,

B. ruderarius (rs = -0.51, p = 0.090) and B. lapidarius (rs = -0.52, p = 0.090).

5.3.4. Joint effects of landscape factors related to forest (Paper IV)

The joint effects of landscape variables on the abundance of individual species had the best fit at different spatial scales (Table 3). The highest association for two of the most abundant species, B. pascuorum and B. ruderarius, was found at the scale of 2000 m; for both species the models were statistically significant and explained 51% and 43% of the variation

37 in their abundance, respectively. Other species that showed the best fit at the largest spatial scale were B. cryptarum and B. veteranus, and their models explained 50% and 30%, respectively; however, only the regression model for B. cryptarum was statistically significant (Table 3). In contrast, P. bohemicus was mostly influenced by the landscape variables at the smallest spatial scale (250 m); this model explained 62% of the variation and it was statistically significant (Table 3).

Previously, the bumblebee species B. cryptarum was not significantly associated with any of the single landscape variables (Figure 6). However, this species showed strong associations at the largest spatial scale when the joint effects of these landscape variables were analysed, as it was mentioned above. Moreover, B. cryptarum showed significant associations inside the regression model: a positive relationship with the proportion of forest, and negative relationships with edge density of forest (ED_Forest) and mean patch area of forest (AREA_MN-Forest) (Table 3).

Table 3. Results of multiple regression analyses. For each bumblebee species with over 20 individuals, four models corresponding to the different spatial scales (i.e., 250 m, 500m, 1000 m and 2000 m) were fitted. The regression coefficients and model fit characteristics of the best model found (at the spatial scale with the highest R2 and the smallest p-value) for each bumblebee species are presented here (Modified from Paper IV). (For a description of the variables, see Appendix 1).

Species Scalea Regression coefficients R2 Model Pforest Pbrushw ED_Forest AREA_MN- p value forest B. cryptarum 2000 0.081* 0.049 -0.066* -0.061* 0.50 0.014 B. lapidarius 500 -0.070 -0.103 0.019 0.030 0.37 0.081 B. lucorum 250 0.028 -0.084 -0.001 -0.533 0.29 0.189 B. pascuorum 2000 0.061 0.359 0.040 -0.109* 0.51 0.013 B. ruderarius 2000 0.202* 0.321 -0.210* -0.181* 0.43 0.041 B. s. soroeensis 1000 -0.023 0.208 0.002 -0.019 0.30 0.179 x proteus B. veteranus 2000 -0.005 -0.058 -0.005 -0.028 0.30 0.177 P. bohemicus 250 0.087* -0.103* 0.007 -0.471* 0.62 0.002 a Spatial scale with the best fit * Regression coefficients significant at p < 0.05

38 6. DISCUSSION

6.1. Influence of patch-scale factors on bumblebees

Differences between years were found for the total number of species and individuals of bumblebees that were recorded in 2008 and 2009. A possible explanation for that difference may be that the weather conditions in the second year were more suitable for bumblebees than in the previous year, and that is why higher values were found in 2009. Overall, the weather in 2008 was slightly colder with more rain, which prob- ably affected bumblebee activity. However, after rarefaction was applied to the species richness of bumblebees, the number of species in 2008 and 2009 were not significantly different anymore, indicating that the weather conditions may influence more bumblebee abundance than species richness.

This study shows that the diversity of flowering plants is a very important factor for total bumblebee abundance. Similarly, Rundlöf et al. (2008) found that local abundance of forage resources was significantly positively associated with bumblebee abundance. Also, they found that higher abundance of flowering plants was associated with higher abundance of bumblebees from large colonies (Rundlöf et al. 2008). The number of individuals of bumblebees may depend on the availability of flowering resources, because generally the most common and abundant species tend to be those that have a broad diet and emerge early in the season (Goulson et al. 2005); for example, Bäckman and Tiainen (2002) found that bumblebees with short tongues were the most abundant species in their study. In general, our results indicate that enhancing the presence of flowering plant species in semi-natural meadows may increase the overall abundance of bumblebees. This result is consistent with previous studies, which have suggested that the species richness of flowering plants is an important local factor for bumblebee communities (e.g., Bäckman and Tiainen 2002, Mänd et al. 2002, Kells and Goulson 2003, Rundlöf et al. 2008, Ahrné et al. 2009).

In the case of long-tongued bumblebees, we found that both total species richness and abundance were positively influenced by the diversity of flowering plants and with percent cover of flowering plants. This makes sense, as the higher the availability and diversity of food resources in the habitat, the higher the bumblebees’ chances of finding the plant species they require. In addition, the foraging distance of some species

39 of bumblebees is very restricted, depending on the resources within their central habitat (Walther-Hellwig and Frankl 2000). Some rare long- tongued species seem to have a rather small foraging range (e.g., B. distinguendus) (Charman et al. 2010), and others, such as B. ruderarius and B. sylvarum, are considered ‘doorstep foragers’ because they appear to forage within 500 m around their nests (Goulson 2010). On the other hand, we found that contrary to overall bumblebee species richness and abundance, long-tongued bumblebees were strongly positively associated with percent cover of flowering plants.

Patch area seems to have a positive influence on bumblebee species richness. This also makes sense as the larger the area of the habitat, the higher the chances of finding the suitable food resources and nesting sites that bumblebee species require. Also, in patches of smaller size, habitat-specialist plants may have a higher probability of extinction (Quintana-As- cencio and Menges 1996); this may influence the bumblebee species that depend on these types of plants. Previous studies on insects have also found significant positive relationships between habitat area and species richness (e.g., Steffan-Dewenter and Tscharntke 2000; Krauss et al. 2003; Steffan-Dewenter 2003; Öckinger and Smith 2006).

Other patch-scale factors, specifically shape index and fractal dimension index, showed negative relationships with total bumblebee species richness. Shape index was also one of the main predictors of bumblebee species richness. This index describes the complexity of the patch shape; this means that the more irregular the shape of the habitat, the lower may be the number of species in that habitat. The importance of patch shape on organisms can be described using the “interior-to-edge ratio” (Forman and Godron 1986): a circular or square patch consists mostly of an interior area with a surrounding band of edge. A square patch has a higher “interior-to-edge ratio” compared to a patch with a more complex or irregular shape (with the same area), as the latest has proportionally less interior area. Forman and Godron (1986) suggested that patches with higher “interior-to-edge ratio” may have higher species diversity, less probability of presence of barriers within the patch, and more foraging efficiency for animals inside the patch. However, the effect of patch shape on the foraging efficiency has not been well studied and further research is needed (Forman and Godron 1986). The fractal dimension index is also related with the shape of the patch; it is another measure of shape complexity, but it is calculated based on the patch size.

40 6.2. Influence of landscape composition on bumblebees

We found that total bumblebee species richness and abundance were positively influenced by the proportion of human settlements at various spatial scales. These positive associations may be explained by the presence of gardens in residential areas, which may support a high diversity of flowering plants and thus provide suitable nesting sites, shelter and alternative foraging resources for bumblebees. This has been found also in previous studies in the case of bees (i.e., honeybees and bumblebees) (Cussans et al. 2010), and bumblebees (Goulson et al. 2002, McFreder- ick and LeBuhn 2006, Goulson et al. 2010). Some abundant bumblebee species, such as B. ruderarius, seem to prefer plant communities close to human settlements (Söderman 1999). Generally, people like to have plants with blossoms in their gardens, so the percentage of nectar-rich flowers might be high in human-inhabited areas. In addition, gardens seem to support extraordinarily high densities of nests of bumblebees (Osborne et al. 2008, Goulson 2010).

In contrast, Ahrné et al. (2009) found that the proportion of urban areas had a negative effect on bumblebee richness, as the increased presence of urban structures such as roads and buildings decreases the proportion of suitable habitat patches for bumblebees, such as field boundaries and pastures. However, most of our study sites were located relatively far from large towns, which mean that the density of roads, especially main roads, is very low, and the presence of buildings and houses is not very evident. Also, the roadsides and field boundaries in Estonia are commonly covered by lush herbaceous flora (Mänd et al. 2002), which may favour bumblebees. Our measurements of the proportion of human settlements included also the presence of abandoned buildings (ruins) or places with ruderal plants; these areas are very common in Estonia and may offer a high diversity of flowering plants. In addition, Winfree et al. (2007) suggested that bee species richness may be higher when the proportion of natural habitats in the landscape is high, even though the level of human disturbance is intermediate; that is, the negative effects of human disturbance may occur only when the proportion of natural land cover is very low. Our study region is covered by high proportions of forests, brushwood and meadows; this means that the presence of these land cover types may mask the effect of human settlements on bumblebees.

41 The results from this study show that the overall number of species of bumblebees (and also long-tongued bumblebee species) may increase with the presence of meadows in the landscape at the largest spatial scales. Similarly, Hatfield and LeBuhn (2007) found that the most consistent positive influence on species richness of bumblebees was the proportion of meadows in the surrounding landscape, at a 2-km buffer from the edge of the focal habitat. In addition, Le Féon et al. (2010) found that the species richness, abundance and diversity of bees were negatively affected by agricultural intensification, whereas bee species richness was positively influenced by the amount of semi-natural habitats in the landscape. On the other hand, it has been found that, in general, bumblebees have large foraging ranges (Steffan-Dewenter et al. 2002, Westphal et al. 2006, Hatfield and LeBuhn 2007); some species are able to fly distances of more than 2000 m (e.g., B. pascuorum and B. terrestris) (Chapman et al. 2003, Zurbuchen et al. 2010). Dispersal abilities of bumblebees allow them to retrieve floral resources in adjacent meadows, increasing the probability of finding the flowering plants that bumblebee species require (Hatfield and LeBuhn 2007).

Total bumblebee species richness seems to be negatively influenced by the presence of forest in the surrounding landscape at the largest spatial scales. This may happen because some bumblebee species may not be able to find suitable nesting sites in the forest and also, they may have different preferences in terms of the landscape context. Goulson (2010) suggested that the sites chosen for nesting vary between species, depending on the habitat type and the place where this habitat is located. Also, overall bumblebee species richness appears to be negatively influenced by the proportion of brushwood. Brushwood areas are commonly dominated by willows that often grow in wetlands and along forest edges (Sepp et al. 2004). These habitats are rich in blooming flowers and are important for bumblebees in early spring (i.e., April and May), particularly for the species that emerge early in the season. However, areas dominated by willows may also represent an ecological trap for bumblebees: early emerging species might tend to build their nests near the forest, where later in the season food would become scarce and these areas would no longer be able to provide enough forage resources for bumblebees (R. Karise, personal communication).

In the case of long-tongued bumblebee species richness, they also appear to be negatively affected by the proportion of forest. These results suggest

42 that some species of long-tongued bumblebees may prefer open areas. In general, most long-tongued bumblebee species have specialised diets and are expected to visit a particular type of flowering plant; those flowers are more likely to be found in open areas than in patches of forest. Also, it has been suggested that early-emerging bumblebee species are associated with forests while late-emerging species are associated with grasslands; most late-emerging species are medium or long-tongued bumblebees (Goulson et al. 2005).

On the other hand, overall bumblebee species richness and abundance were negatively associated with the proportion of arable land. The negative effect of arable land on bumblebees may be explained by the openness of the landscape in those areas, which could make the bumblebees more vulnerable to wind and other climatic factors, as there are fewer places that may offer shelter and protection. Also, foraging resources are sometimes scarce in agricultural land and this may result in the decline of bumblebees (Goulson et al. 2005), while the presence of semi-natural grasslands in the landscape context may increase the presence of bumblebees (Öckinger and Smith 2007). Grasslands are more likely to contain a higher availability of nesting sites for bumblebees than the surrounding cultivated land (Öckinger and Smith 2007). Similarly, Le Féon et al. (2010) found that bee species richness and abundance were negatively affected by agricultural intensification. Overall pollinator diversity may be enhanced by the presence of semi-natural habitats in the landscape context (Billeter et al. 2008, Jauker et al. 2009, Le Féon et al. 2010).

6.3. Influence of landscape configuration on bumblebees

Overall bumblebee abundance (and long-tongued bumblebee abundance, in particular) seems to be positively influenced by edge density at landscape level. This positive relation may occur because there is a strong dependency of bumblebee abundance on the availability of flowering plants (as mentioned before). Kumar et al. (2009) explained that habitat edges contain a great abundance and diversity of floral resources, making them suitable places for flower visitors. The presence of edges and other compensating areas near to the main habitat is very important for the survival of bumblebees, especially in patchy landscapes with diverse land cover types, as they may find complementary food resources and nesting places there. Furthermore, bumblebee queens are more frequently observed along forest boundaries and field boundaries (Svensson

43 et al. 2000). Similarly, Sepp et al. (2004) found that the distribution of bumblebees was positively related with the length of ecotones between cultivated land and different types of forest. A study of bumblebees in Estonia suggested that edges are particularly important in spring, when bumblebee queens mostly forage on the flowering willows that are commonly found in the forest edges (Sepp et al. 2004). Positive effects of linear elements, such as edges, on bumblebees have been found before (Osborne et al. 2008).

We found that Shannon’s diversity index seems to be an important landscape metric for bumblebee abundance at the largest spatial scale. This index also seems to be important for long-tongued bumblebee species richness. Shannon’s diversity index indicates the level of complexity of the landscape matrix, and increases as the number of different patch types increases and the distribution of patch types becomes more equitable (Eiden et al. 2000). This means that our study sites are surrounded by different patch types that might be suitable habitat fragments for bumblebees, increasing the availability of food resources in the landscape and thus, their likelihood of survival. Williams and Osborne (2009) suggested that the ability of bumblebees to fly long distances from the colony makes them less susceptible to the fragmentation and patchiness of the landscape, as they become more flexible in the utilisation of food resources. In the case of long-tongued bumblebees, as they have very special needs in terms of food resources, a diverse landscape matrix may increase their survival possibilities. Other authors have found similar positive relationships between insects and the diversity of the landscape matrix (e.g., Steffan-Dewenter 2003, Kivinen et al. 2006). Kivinen et al. (2006) argued that in boreal agricultural landscapes, the presence of patches of semi-natural grasslands and other non-crop biotopes in adjacent open areas may have a positive effect on the species richness of some insects (such as butterflies), as movement of species between different habitat types can increase overall species richness in the landscape context.

In contrast, overall bumblebee species richness and abundance appear to be negatively influenced by mean patch area of forest. Also, this landscape factor seems to affect in a negative way the species richness of long-tongued bumblebees. A possible explanation for the negative associations may be that a high number of patches of forest could be seen as potential obstacles in the landscape by some species of foraging bum-

44 blebees (Kreyer et al. 2004, Goulson et al. 2010), particularly for those species that have large foraging distances, such as B. lapidarius and B. terrestris (Walther-Hellwig and Frankl 2000a, Walther-Hellwig and Frankl 2000b). In the same way, Winfree et al. (2007) found that bee species richness and abundance were negatively associated with the extent of forest cover, suggesting that the number of bees decreased as forest cover increased in the surrounding landscape.

6.4. Influence of forest habitats on the local abundance of bumblebee species

In the literature, different kinds of relationships have been found between forest cover and bees: e.g., Taki et al. (2007) found that bee abundance and species richness were positively related to forest cover (at a radius of 750 m), whereas Winfree et al. (2007) found negative relationships between similar variables (at a radius of 1600 m). Other authors have found differences in behaviour between some species of bumblebees. There is the case of B. pascuorum and B. terrestris that were studied by Kreyer et al. (2004); they found that, although forest cover did not represent a barrier for either species, B. terrestris seems to prefer open landscapes. Previously in our study, we found that overall bumblebee species richness and abundance seems to be negatively influenced by forest; however, when we analysed the influence of forest at single-species level, we found different types of relationships between bumblebee species and forest habitats. In general, our results show that the presence of forest in the surrounding landscape is an important factor for some species of bumblebees.

In this study, B. pascuorum (the most abundant and widely distributed species in the study area), B. schrencki and particularly two species of cuckoo bumblebees (i.e., P. bohemicus and P. norvegicus) seemed to prefer landscapes with high proportions of forest in the surrounding areas. This finding is consistent with a study on bumblebees in Finland that recognised P. bohemicus and P. norvegicus as species preferring forest habitats (Bäckman and Tiainen 2002). Some abundant species, such as B. pascuorum, are known to emerge early in the season from hibernation (Goulson and Darvill 2004); in general, early-emerging bumblebee species have been associated with forests (Goulson et al. 2005). Another possible explanation for the positive associations between bumblebees and forest may be that some species (e.g., B. pascuorum) tend to nest

45 above the ground in areas of leaf litter and thickets, and woodland areas are likely to offer these types of nesting sites (Goulson et al. 2010). On the other hand, our study confirms that the species B. schrencki is positively influenced by the presence of forest. This species is known to prefer forest and forest margins (Söderman 1999). According to Söder- man (1999), the expansion of this bumblebee in the Baltic countries was promoted by the rapid afforestation of open fields.

One of the most abundant species in our study sites (i.e., B. ruderarius, which is a long-tongued species) and three other species (i.e., B. terrestris, B. veteranus and B. lapidarius) showed negative trends with the proportion of forest. They seem to prefer open areas. Mänd et al. (2002) found that the species B. lapidarius and B. veteranus were particularly numerous in agricultural habitats, which are open areas. The bumblebee B. lapidarius belongs to a group of specialists on Fabaceae, a large family of flowering plants that are commonly found in grasslands (Goulson et al. 2005). Also, species such as B. terrestris and B. lapidarius are considered spatial generalists because they have large foraging distances (Walther-Hellwig and Frankl 2000a, Walther-Hellwig and Frankl 2000b). In a recent study, Hagen et al. (2011) found that some bumblebee species are able to fly long distances (maximum distances of 1.3 – 2.5 km) and to use large areas (0.25 – 43.53 ha); e.g., they found that B. terrestris can flight a maximum distance of 2.5 km. These bumblebee species may prefer an open landscape to have more freedom for their long-distance flights. Bäckman and Tiainen (2002) also classified B. ruderarius, B. lapidarius and B. veteranus as species preferring open habitats.

Concerning the relations between the proportion of brushwood and bumblebees, we found some positive and some negative relationships. Some species seem to be positively influenced by this variable (i.e., B. schrencki and B. pascuorum), whereas others seem to be negatively influenced (i.e., B. terrestris, B. veteranus, B. lapidarius and P. bohemicus). In addition, we found that the subspecies B. s. soroeensis and B. s. proteus seem to be ecologically different: B. s. soroeensis appears to prefer brushwoods, whereas B. s. proteus does not. Positive relationships may occur because many patches of brushwood have grown in areas that were former meadows. The soil in these areas is rich in calcium and can therefore support a great amount of flowering plant species. Ad- ditionally, as already mentioned (see Section 6.2), brushwood areas are

46 dominated by willows, which are very important flowering plants in early spring for bumblebees; but these areas may represent an ecological trap for some bumblebee species, as food resources become scarce in summertime.

Relationships involving edge density of forest were positive for some species of bumblebees, i.e., B. pascuorum, B. pratorum and P. sylvestris, whereas negative relationships were found for B. sylvarum, B. s. proteus and B. veteranus. Positive relationships between edge density of forest and some species, such as B. pascuorum, may occur because these bumblebees prefer bell-shaped flowers (or flowers that hang in a downward position). Flowers having this shape occur commonly in berry-bearing plants, and these plants often grow close to forests or in forest margins. In general, edges may support a greater abundance and diversity of flowering plants. The species B. pratorum was also positively associated with forest edges. This finding is consistent with Goulson et al. (2005); they suggested that early-emerging species like B. pratorum are related to woodland and woodland edges. Sepp et al. (2004) argued that edges are particularly important in April and May as bumblebee queens forage from the flowering willows that are commonly found in the forest edges of Estonia. In agricultural areas, linear elements of the landscape such as woodland edges, fence lines and hedgerows are likely to have more bumblebee nests compared with non-linear elements, such as woodland or grassland (Osborne et al. 2008, Goulson 2010).

Only negative associations were found between mean patch area of forest and bumblebees. Four species of bumblebees (i.e., B. terrestris, B. veteranus, B. lapidarius and B. ruderarius) were associated with this landscape index in a negative way. Mean patch area of forest appears to be important for the bumblebee species that seem to prefer open areas, judging from the negative relationships found between some species and the proportion of forest. Similarly, it was found on a study on B. terrestris that even though forests do not represent a barrier for this bumblebee, this species seems to prefer open areas (Kreyer et al. 2004). A possible explanation is that a high number of patches of forest may be seen as potential obstacles in the landscape for some species of foraging bumblebees (Kreyer et al. 2004, Goulson et al. 2010), particularly for those that have large foraging distances, such as B. lapidarius and B. terrestris (Walther-Hellwig and Frankl 2000a, Walther-Hellwig and Frankl 2000b).

47 The joint effects of the set of landscape variables related with forests appear to be important for some bumblebee species (i.e., B. pascuorum, B. ruderarius, B. cryptarum and P. bohemicus), specially, but not only, at the largest spatial scale (i.e., 2000 m).

6.5. Bumblebees as potential indicators of habitat quality

Ecological indicators can be defined as factors that communicate important information about ecosystems and the impact of human activities on them. Ecosystems are complex and the use of ecological indicators is needed in order to describe them in simpler terms that can be understood and used by scientists and non-scientists alike to make management decisions (Girardin et al. 1999).

Insects are considered key indicators of environmental change due to their diversity of habitat characteristics and requirements. The role of insects as ecological indicators has been extensively tested and studied (e.g., Sepp et al. 2004, Billeter et al. 2008). Among insects, bumblebees (and bees in general) are seen as a vital element of global biodiversity and an important group of pollinators. As mentioned before, they play a key role in supporting not only crops, but also the diversity of natural and semi-natural vegetation (Rundlöf et al. 2008, Goulson et al. 2011) and the survival of other organisms (Goulson et al. 2006, Goulson et al. 2011). Bumblebees are known to be sensitive to environmental changes and thus, they may serve as good indicators of habitat quality (Sepp et al. 2004, Haaland and Gyllin 2010). Also, bumblebees are easy to find and to identify when doing monitoring compared with other insects: they have fairly large bodies and they are moderately slow flying insects.

In Estonia, bumblebees are regarded as significant indicators of habitat and landscape diversity (Mänd et al. 2002), and have been proposed as biodiversity indicators at the landscape level of the agri-environmental programme (Sepp et al. 2004). As already mentioned, bumblebees and other pollinators are at risk. Thus, there is a current need for the protection of endangered species as well as the conservation of their habitats. Semi-natural habitats, such as meadows, are areas of important value for bumblebees, as they provide essential resources like food and nesting sites (Svensson et al. 2000, Kells and Goulson 2003). Some conservationists’ studies of endangered species have emphasised the role and importance of large-scale dynamics (e.g., Goulson et al. 2011); it therefore appears

48 relevant to consider interactions between species and landscape elements when developing biodiversity conservation strategies.

Hatfield and LeBuhn (2007) suggested that bumblebee communities provide an excellent model for evaluating the importance of factors at patch and landscape scale. Even though bumblebees are known to have large foraging distances (Steffan-Dewenter et al. 2002, Westphal et al. 2006, Hatfield and LeBuhn 2007), they also appear to display a high dependency on their central foraging place (Osborne and Williams 2001, Hatfield and LeBuhn 2007). Our results show that bumblebees are related with variables at patch scale (e.g., species richness of flowering plants, percent cover of flowering plants) as well as variables at landscape scale (e.g., proportion of meadows, proportion of forest, edge density (ED_LAND), Shannon’s diversity index (SHDI)) in different ways.

Habitat quality may be assessed by its suitability for insects (Fahrig and Jonsen 1998), particularly bumblebees, using important ecological differences between generalists and specialist species (e.g., some long- tongued bumblebees). Specialist species are more susceptible to degradation and decrease of suitable habitats than generalist species, because specialist species are dependent on specific types of habitats or flowering plants; therefore, if the amount of suitable habitat decreases, it may be more difficult for these species to find the foraging resources they need. A greater tongue length in bumblebees has been suggested as a trait that confers a greater susceptibility to decline on some bumblebee species (Williams and Osborne 2009).

49 7. CONCLUSIONS AND IMPLICATIONS FOR CONSERVATION

This thesis showed the influence of a set of important biotic and abiotic factors, considered at patch and landscape scales, on the species richness and abundance of bumblebees (and long-tongued bumblebees, in particular) in the semi-natural meadows of northeast Estonia. At single- species level, this study provided information about the effects of forest habitats on the local abundance of bumblebee species. According to the results of this study, the following conclusions can be drawn:

• Even though northeast Estonia (i.e., Ida-Virumaa County) has been environmentally affected by mining activities and by the presence of power plants, this region could be considered an important area for conservation of bumblebees, as it supports a high number of species across its territory (compared with other regions of Estonia, Sepp et al. 2004): more than 80% of the bumblebee species known in Estonia have been found in this region. The mosaic landscape of northeast Estonia with forests, brushwood, human settlements and meadows seem to be favourable for bumblebees.

• This study confirms that the presence of a high diversity and abundance of flowering plants may benefit bumblebees in semi-natural meadows, especially long-tongued species richness and abundance. Also, the size of the central habitat seems to positively influence the number of species of bumblebees: bigger areas appear to be better.

• Human settlements in rural areas may favour bumblebee species richness and abundance, particularly when it includes gardens and other places with a high diversity of flowering plants, and when the percentage of natural and semi-natural habitats in the landscape is high.

• There is evidence that the presence of meadows may benefit overall bumblebee species richness (and particularly long-tongued species); whereas arable land may have a negative effect on overall species richness and abundance of bumblebees.

• The existence of edges (especially at landscape level) may promote overall bumblebee abundance (and also long-tongued bumblebee

50 abundance), as these are considered compensating areas that may offer shelter, food and protection for bumblebees.

• Some bumblebee species may benefit from a heterogeneous landscape with a high proportion of forest habitats (e.g., B. schrencki), whereas others seem to prefer open landscapes (e.g., B. veteranus). Also, bumblebee species that have large foraging distances (e.g., B. terrestris) may prefer open landscapes because the presence of many patches of forest in the surrounding landscape could narrow their foraging area, affecting their long-distance flights.

• In general, bumblebees benefit from a rich and diverse landscape matrix with an important presence of patches of natural and semi- natural habitats. It appears that the ability of some bumblebee species to fly long distances makes them less vulnerable to the level of fragmentation or patchiness in a given landscape.

• In countries with patchy landscapes, like Estonia, it is important to consider ecological indicators that are strongly associated with both patch and landscape variables. Bumblebees, because of their reliance on these variables, have the potential to serve as indicators of habitat quality.

Overall, we found that not only the availability of flowering plants at patch level, but also the quality and diversity of the surrounding landscape (i.e., the presence of patches of natural and semi-natural habitats), are important factors affecting bumblebees. Landscapes with high percentages of meadows, with a strong presence of edges and a diverse matrix, may support a higher diversity and abundance of bumblebees. With the presence of adjacent patches of meadow and habitat edges in the surrounding landscape, it seems that there is an increased probability that bumblebees will encounter floral resources during their life cycle.

Policies supporting agri-environmental measures should be improved: financial resources should target not only one farmer or changes at the local level, but also changes at the landscape level. Changes at the level of one farm are not sufficient to support the entire system that also incorporates the surrounding landscape. To maintain biodiversity, heterogeneous landscapes including patches of natural and semi-natural

51 habitats (particularly forests, brushwood and meadows) need to be preserved. We should consider not only variables at the local level but also the landscape context around targeted areas at large spatial scales when designing conservation strategies for bumblebees and agri-environmental measures.

52 REFERENCES

Ahrné, K., Bengtsson, J., Elmqvist, T. 2009. Bumble bees (Bombus spp) along a gradient of increasing urbanization. PLoS One 4, e5574. Bäckman, J.P., Tianen, J. 2002. Habitat quality of field margins in a Finnish farmland area for bumblebees (Hymenoptera: Bombus and Psithyrus). Agriculture, Ecosystems & Environment 89: 53–68. Billeter, R., Liira, J., Bailey, D., Bugter, R., Arens, P., Augenstein, I., Avi- ron, S., Baudry, J., Bukacek, R., Burel, F., Cerny, M., De Blust, G., de Cock, R., Diekoetter, T., Dietz, H., Dirksen, J., Dormann, C., Dur- ka, W., Frenzel, M., Hamersky, R., Hendrickx, F., Herzog, F., Klotz, S., Koolstra, B., Lausch, A., Le Coeur, D., Maelfait, J. P., Opdam, P., Roubalova, M., Schermann, A., Schermann, N., Schmidt, T., Sch- weiger, O., Smulders, M.J.M., Speelmans, M., Simova, P., Verboom, J., van Wingerden, W.K.R.E., Zobel, M., Edwards, P.J. 2008. Indica- tors for biodiversity in agricultural landscapes: a pan-European study. Journal of Applied Ecology 45(1): 141–150. Carvell, C. 2002. Habitat use and conservation of bumblebees (Bombus spp.) under different grassland management regimes. Biological Con- servation 103: 33–49. Carrascal, LM., Galván, I., Gordo, O. 2009. Partial least squares regression as an alternative to current regression methods used in ecology. Oikos 118: 681-690. Chapman, R.E., Wang, J., Bourke, A.F.G. 2003. Genetic analysis of spatial foraging patterns and resource sharing in bumble bee pollinators. Molecular Ecology 12: 2801–2808. Charman, T.G., Sears, J., Green, R.E., Bourke, A.F.G. 2010. Con- servation genetics, foraging distance and nest density of the scarce Great Yellow Bumblebee (Bombus distinguendus). Molecular Ecology 19(13): 2661–2674. Cozzi, G., Müller, C.B., Krauss, J. 2008. How do local habitat management and landscape structure at different spatial scales affect fritillary butterfly distribution on fragmented wetlands? Landscape Ecology 23: 269–283. Crist, T.O., Pradhan-Devare, S.V., Summerville, K.S. 2006. Spatial variation in insect community and species responses to habitat loss and plant community composition. Oecologia 147: 510–521.

53 Cussans, J., Goulson, D., Sanderson, R., Goffe, L., Darvill, B., Osborne, J.L. 2010. Two bee-pollinated plant species show higher seed production when grown in gardens compared to arable farmland. PLoS One 5, e11753. Drakare, S., Lennon, J.J., Hillebrand, H. 2006. The imprint of the geographical, evolutionary and ecological context on species-area relationships. Ecology Letters 9: 215–227. Eiden, G., Kayadjanian, M., Vidal, C. 2000. From land cover to landscape diversity in the European Union, Chapter 1. Capturing landscape structures: Tools. http://ec.europa.eu/agriculture/publi/landscape/ch1.htm. Fahrig, L., Jonsen, J. 1998. Effect of habitat patch characteristics on abundance and diversity of insects in an agricultural landscape. Eco- systems 1(2): 197–205. Fauna Europaea (2011) Fauna Europaea version 2.4. http://www.faunaeur.org. Fischer, J., Lindenmayer, D.B. 2007. Landscape modification and habitat fragmentation: a synthesis. Global Ecology and Biogeography 16(3): 265–280. Forman, R.T.T., Godron, M. 1986. Landscape ecology. John Wiley & Sons, Inc. USA. 619 p. Freemark K., Bert D., Villard MA. 2002. Patch-, landscape-, and regional-scale effects on biota. In: Applying landscape ecology in biological conservation, Gutzwiller, K.J. (ed.). Springer, New York. p. 58–83. Girardin, P., Bockstaller, C., Van der Werf, H. 1999. Indicators: Tools to evaluate the environmental impacts of farming systems. Journal of Sustainable Agriculture 13(4): 6–21. Goulson, D. 2010. Bumblebees: behaviour, ecology and conservation. Second Edition. Oxford University Press, USA. 317 p. Goulson, D., Darvill, B. 2004. Niche overlap and diet breadth in bumblebees; are rare species more specialized in their choice of flowers? Apidologie 35: 55–64. Goulson, D., Hanley, M.E., Darvill, B., Ellis, J.S. 2006. Biotope associations and the decline of bumblebees (Bombus spp.). Journal of Insect Conservation 10: 95–103. Goulson, D., Hanley, M.E., Darvill, B., Ellis, J.S., Knight, M.E. 2005. Causes of rarity in bumblebees. Biological Conservation 122(1): 1–8.

54 Goulson, D., Hughes, W.H.O., Derwent, L.C., Stout, J.C. 2002. Colony growth of the bumblebee, Bombus terrestris, in improved and conventional agricultural and suburban habitats. Oecologia 130: 267–273. Goulson, D., Lepais, O., O’Connor, S., Osborne, J.L., Sanderson, R.A., Cussans, J., Goffe, L. and Darvill, B. 2010. Effects of land use at a landscape scale on bumblebee nest density and survival. Journal of Applied Ecology 47(6): 1207–1215. Goulson, D., Lye, G.C., Darvill, B. 2008. Decline and conservation of bumblebees. Annual Review of Entomology 53: 191–208. Goulson, D., Rayner, P., Dawson, B., Darvill, B. 2011. Translating research into action; bumblebee conservation as a case study. Journal of Applied Ecology 48: 3–8. Gutzwiller, K.J. 2002. Spatial factors affecting organism occurrence, movement, and conservation: introduction to section II. In: Apply- ing landscape ecology in biological conservation, Gutzwiller, K.J. (ed.). Springer, New York. p. 55 – 57. Haaland, C., Gyllin, M. 2010. Butterflies and bumblebees in greenways and sown wildflower strips in southern Sweeden. Journal of Insect Conservation 14: 125–132. Hagen, M., Wikelski, M., Kissling, W.D. 2011. Space use of bumblebees (Bombus spp.) revealed by radio tracking. PLoS ONE 6 (5), e19997. Hall, L.S., Krausman, P.R., Morrison, M.L. 1997. The habitat concept and a plea for standard terminology. Wildlife Society Bulletin 25: 173–182. Hanski, I., Pakkala, T., Kuussaari, M., Lei, G. 1995. Metapopulation persistence of an endangered butterfly in a fragmented landscape. Oikos 72: 21–28. Hatfield, R.G., LeBuhn, G. 2007. Patch and landscape factors shape community assemblage of bumble bees, Bombus spp. (Hymenoptera: Apidae), in montane meadows. Biological Conservation 139: 150– 158. Holzschuh, A., Steffan-Dewenter, I., Tscharntke, T. 2008. Agricultural landscapes with organic crops support higher pollinator diversity. Oikos 117: 354–361. Jauker, F., Diekötter, T., Schwarzbach, F., Wolters, V. 2009. Pollinator dispersal in an agricultural matrix: opposing responses of wild bees

55 and hoverflies to landscape structure and distance from main habitat. Landscape Ecology 24: 547–555. Jones, J.P.G. 2011. Monitoring species abundance and distribution at the landscape scale. Journal of Applied Ecology 48: 9–13. Kells, A.R., Goulson, D. 2003. Preferred nesting sites of bumblebee queens (Hymenoptera: Apidae) in agroecosystems in the UK. Bio- logical Conservation 109: 165–174. Kivinen, S., Luoto, M., Kuussaari, M., Helenius, J. 2006. Multi-species richness of boreal agricultural landscapes: effects of climate, biotope, soil and geographical location. Journal of Biogeography 33: 862–875. Kleijn, D., Baquero, R.A., Clough, Y., Díaz, M., Esteban, J., Fernández, F., Gabriel, D., Herzog, F., Holzschuh, A., Jöhl, R., Knop, E., Kruess, A., Marshall, E. J. P., Steffan-Dewenter, I., Tscharntke, T., Verhulst, J., West, T. M., Yela, J. L. 2006. Mixed biodiversity benefits of agri- environment schemes in five European countries. Ecology Letters 9: 243–254. Knight, M.E., Osborne, J.L., Sanderson, R.A., Hale, R.J., Martin, A.P., Goulson, D. 2009. Bumblebee nest density and the scale of available forage in arable landscapes. Insect Conservation and Diversity 2: 116–124. Kosior, A, Celary, W, Olejniczak, P, Fijal, J, Krol, W, Solarz, W., Plonka, P. 2007. The decline of the bumble bees and cuckoo bees (Hymenop- tera: Apidae: Bombini) of Western and Central Europe. Oryx 41: 79–88. Krauss J., Steffan-Dewenter I., Tscharntke T. 2003. How does landscape context contribute to effects of habitat fragmentation on diversity and population density of butterflies? Journal of Biogeography 30: 889–900. Krewenka, K.M., Holzschuh, A., Tscharntke, T., Dormann, C.F. 2011. Landscape elements as potential barriers and corridors for bees, wasps and parasitoids. Biological Conservation 144: 1816–1825. Kreyer, D., Oed, A., Walther-Hellwig, K., Frankl, R. 2004. Are forests potential landscape barriers for foraging bumblebees? Landscape scale experiments with Bombus terrestris agg. and Bombus pascuorum (Hymenoptera, Apidae). Biological Conservation 116(1): 111–118.

56 Krishnan, A., Williams, LJ., McIntosh, AR., Abdi, H. 2011. Partial Least Squares (PLS) methods for neuroimaging: A tutorial and review. NeuroImage 56: 455-475. Kumar, S., Simonson, S.E., Stohlgren, T.J. 2009. Effects of spatial heterogeneity on butterfly species richness in Rocky Mountain National Park, CO, USA. Biodiversity Conservation 18: 739–763. Le Féon, V., Schermann-Legionneta, A., Delettrea, Y., Avironb, S., Bil- leterc, R., Bugterd, R., Hendrickxe, F., Burel, f. 2010. Intensification of agriculture, landscape composition and wild bee communities: A large scale study in four European countries. Agriculture, Ecosystems & Environment 137: 143–150. Mänd, M., Mänd, R., Williams, I.H. 2002. Bumblebees in the agricultural landscape of Estonia. Agriculture, Ecosystems & Environment 89: 69–76. Mander, U., Palang, H. 1994. Changes of landscape structure in Estonia during Soviet period. GeoJournal 33: 45–54. Manning, A.D., Lindenmayer, D.B., Nix, H.A. 2004. Continua and Umwelt: alternative ways of viewing landscapes. Oikos 104: 621–628. Mazerolle, M.J., Villard, M.A. 1999. Patch characteristics and landscape context as predictors of species presence and abundance: a review. Ecoscience 6: 117–124. McFrederick, Q.S., LeBuhn, G. 2006. Are urban parks refuges for bumble bees Bombus spp. (Hymenoptera: Apidae)? Biological Conserva- tion 129: 372–382. McGarigal, K., Cushman, S.A., Neel, M.C., Ene, E. 2002. FRAG- STATS, Spatial Pattern Analysis Program for Categorical Maps. Computer software program produced by the authors at the Uni- versity of Massachusetts, Amherst. http://www.umass.edu/landeco/ research/fragstats/fragstats.html. McGarigal, K., Marks, B.J. 1995. FRAGSTATS: spatial pattern analysis program for quantifying landscape structure. Gen. Tech. Rep. PNW- GTR-351. Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. 122 p. Morrison, M.L., Marcot, B.G., Mannan, R.W. 1998. Wildlife-Habitat Relationships: Concepts and Applications. University of Wisconsin Press. 435 pp.

57 Mortelliti, A., Amori, G., Boitani, L. 2010. The role of habitat quality in fragmented landscapes: a conceptual overview and prospectus for future research. Oecologia 163(2): 535–547. Öckinger, E., Smith, H.G. 2006. Landscape composition and habitat area affects butterfly species richness in semi-natural grasslands. Oe- cologia 149: 526–534. Öckinger, E., Smith, H.G. 2007. Semi-natural grasslands as population sources for pollinating insects in agricultural landscapes. Journal of Applied Ecology 44: 50–59. Osborne, J.L., Martin, A.P., Shortall, C.R., Todd, A.D., Goulson, D., Knight, M.E., Hale, R.J., Sanderson, R.A. 2008. Quantifying and comparing bumblebee nest densities in gardens and countryside habitats. Journal of Applied Ecology 45: 784–792. Osborne, J.L., Williams, I.H. 2001. Site constancy of bumble bees in an experimentally patchy habitat. Agriculture Ecosystems & Environ- ment 83: 129–141. Palang, H., Mander, Ü., Luud, A. 1998. Landscape diversity changes in Estonia. Landscape and Urban Planning 41: 163–169. Peterson, U., Aunap, R. 1998. Changes in agricultural land use in Esto- nia in the 1990s detected with multitemporal Landsat MSS imagery. Landscape and Urban Planning 41: 193–201. Pocewicz, A., Morgan, P., Eigenbrode, S.D. 2009. Local and landscape effects on butterfly density in northern Idaho grasslands and forests. Journal of Insect Conservation 13: 593–601. Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O., Kunin, W.E. 2010. Global pollinator declines: trends, impacts and drivers. Trends in Ecology & Evolution 25: 345–353. Pöyry, J., Paukkunen, J., Heliölä, J., Kuussaari, M. 2009. Relative contributions of local and regional factors to species richness and total density of butterflies and moths in semi-natural grasslands. Oecologia 160: 577–587. Quintana-Ascencio, P.F., Menges, E.S. 1996. Inferring metapopulation dynamics from patch-level incidence of florida of scrub plants. Con- servation Biology 10: 1210–1219. Rundlöf, M., Nilsson, H., Smith, H.G. 2008. Interacting effects of farming practice and landscape context on bumble bees. Biological Conservation 141: 417–426.

58 Samways, M.J. 2005. Insect Diversity Conservation. Cambridge Uni- versity Press, USA. 356 p. Sepp, K., Mikk, M., Mand, M., Truu, J. 2004. Bumblebee communities as an indicator for landscape monitoring in the agri-environmental programme. Landscape and Urban Planning 67: 173–183. Söderman, G. 1999. Diversity of pollinator communities in Eastern Fennoscandia and Eastern Baltics. Results from pilot monitoring with yellow traps in 1997 – 1998. Finnish Environment Institute 355, Nature and Natural Resources. Edita Ltd. Helsinki, Finland. 69 p. Steffan-Dewenter, I. 2003. Importance of habitat area and landscape context for species richness of bees and wasps in fragmented orchard meadows. Conservation Biology 17: 1036–1044. Steffan-Dewenter, I., Münzenberg, U., Bürger, C., Thies, C., Tscharn- tke, T. 2002. Scale-dependent effects of landscape context on three pollinator guilds. Ecology 83: 1421–1432. Steffan-Dewenter, I., Tscharntke, T. 2000. Butterfly community structure in fragmented habitats. Ecology Letters 3: 449–456. Stewart, K.E.J., Bourn, N.A.D., Thomas, J.A. 2001. An evaluation of three quick methods commonly used to assess sward height in ecology. Journal of Applied Ecology 38: 1148–1154. Svensson, B., Langerlöf, J., Svensson, B.G. 2000. Habitat preferences of nest-seeking bumble bees (Hymenoptera: Apidae) in an agricultural landscape. Agriculture Ecosystems & Environment 77: 247–255. Taki, H., Kevan, P.G., Ascher, J.S. 2007. Landscape effects of forest loss in a pollination system. Landscape Ecology 22: 1575–1587. Thomas, J.A., Bourn, N.A.D., Clarke, R.T., Stewart, K.E., Simcox, D.J., Pearman, G.S., Curtis, R., Goodger, B. 2001. The quality and isola- tion of habitat patches both determine where butterflies persist in fragmented landscapes. Proceedings of the Royal Society B Biological Sciences 268: 1791–1796. Van Horne, B. 1983. Density as a misleading indicator of habitat quality. The Journal of Wildlife Management 47(4): 893–901. Walther-Hellwig, K., Frankl, R. 2000a. Foraging habitats and foraging distances of bumblebees, Bombus spp. (Hym., apidae), in an agricultural landscape. Journal of Applied Entomology 124: 299–306.

59 Walther-Hellwig, K., Frankl, R. 2000b. Foraging distances of Bombus muscorum, Bombus lapidarius, and Bombus terrestris (Hymenoptera, Apidae). Journal of Insect Behavior 13(2): 239–246. Westphal, C., Steffan-Dewenter, I., Tscharntke, T. 2003. Mass flowering crops enhance pollinator densities at a landscape scale. Ecology Let- ters 6: 961–965. Westphal, C., Steffan-Dewenter, I., Tscharntke, T. 2006. Bumblebees experience landscapes at different spatial scales: possible implications for coexistence. Oecologia 149(2): 289–300. Wiens, J.A. 2002. Central concepts and issues of landscape ecology. In: Applying landscape ecology in biological conservation, Gutzwiller, K.J. (ed.). Springer, New York. p. 3–21. Williams, P.H. 2005. Does specialization explain rarity and decline among British bumblebees? A response to Goulson et al. Biological Conservation 122: 33–43. Williams, P.H., Osborne, J.L. 2009. Bumblebee vulnerability and conservation world-wide. Apidologie 40: 367–387. Winfree, R., Griswold, T., Kremen, C. 2007. Effect of human disturbance on bee communities in a forested ecosystem. Conservation Bi- ology 21(1): 213–223. Xie, Z., Williams, P.H., Tang, Y. 2008. The effect of grazing on bumblebees in the high rangelands of the eastern Tibetan Plateau of Sichuan. Journal of Insect Conservation 12: 695–703. Zurbuchen, A., Landert, L., Klaiber, J. Muller, A., Hein, S., Dorn, S. 2010. Maximum foraging ranges in solitary bees: only a few individuals have the capability to cover long foraging distances. Biological Conservation, 143(3): 669–676.

60 Appendix 1

Variables at patch and landscape scale. Variable Description Unit

Bumblebees

SRBumb Bumblebee species richness; number of spe- - cies. NIBumb Bumblebee abundance; number of individu- - als. Vegetation structure at the patch scale SRFlowPlants Species richness of flowering plants; number - of species. AvCoverFP Average percent cover of flowering plants. % Percentage of the meadow that is covered by flowering plants. AvGrassH Average grass height. cm

Spatial characteristics at the patch scale AREA Patch area; size of the patch. ha

PERIM Perimeter of the patch. m

SHAPE Shape index; SHAPE equals patch perimeter - (m) divided by the square root of patch area (m2), adjusted by a constant to adjust for a square standard.a FRAC Fractal dimension index; FRAC equals 2 - times the logarithm of patch perimeter (m) divided by the logarithm of patch area (m2).a ED Edge density; sum of the length (m) of the m/ha edge segment of the patch per unit area.a

Landscape composition Pforest Proportion of patches that are forest. %

Pmeadows Proportion of patches that are meadows. %

PArLand Proportion of patches that are arable land. %

PHumSet Proportion of patches that are human set- % tlements; including residential areas, buildings, cattle sheds, roads, ruins (or buildings’ remains) and green houses. Pbrushw Proportion of patches that are brushwood. %

61 Landscape configuration PRD Patch richness density; PRD equals the num- No/100 ber of patch types per 100 ha.a ha IJI Interspersion and juxtaposition index; meas- % ure of distribution of patch adjacencies.a ED_LAND Edge density at landscape level; total length of m/ha all edge segments per unit area of landscape.a ED_Forest Edge density of forest; equals the sum, across all patches of the corresponding patch type (here, forest) of the area of the patches, divided by the total number of patches of the same type.a SHDI Shannon’s diversity index; SHDI equals - minus the sum, across all patch types, of the proportional abundance of each patch type multiplied by that proportion.a AREA_MN-Forest Mean patch area of forests; AREA_MN ha equals the sum, across all patches of the corresponding patch type (i.e., forest) of the area of the patches, divided by the total number of patches of the same type.a a Source: McGarigal et al. (2002)

62 Appendix 2

List of plant species (the nomenclature was based on http://elurikkus.ut.ee/).

Apiaceae Boraginaceae Aegopodium podagraria Anchusa officinalis Angelica sylvestris Echium vulgare Anthriscus sylvestris Myosotis arvensis Pastinaca sativa Peucedanum palustre Campanulaceae Campanula glomerata Asteraceae Campanula patula Achillea millefolium Campanula persicifolia Achillea ptarmica Campanula rapunculoides Achillea salicifolia Anthemis tinctoria Campanula rotundifolia Carduus crispus Campanula trachelium Centaurea jacea Jasione montana Centaurea scabiosa Cirsium arvense Caryophyllaceae Cirsium oleraceum Dianthus deltoides Cirsium vulgare Honkenya peploides Crepis spp Lychnis flos-cuculi Eupatorium cannabinum Lychnis viscaria Hieracium spp Sagina nodosa Inula salicifolia Silene dioica Leontodon autumnalis Silene nutans Leontodon hispidus Silene pratensis Leontodon spp Silene vulgaris Leucanthemum vulgare Stellaria graminea Pilosella officinarum Pilosella spp Stellaria holostea Solidago virgaurea Stellaria palustris Tanacetum vulgare Stellaria spp Taraxacum officinale Tragopogon pratensis Crassulaceae Sedum acre Brassicaceae Berteroa incana Dipsacaceae Bunias orientalis Knautia arvensis Sinapis arvensis Succisa pratensis

63 Lythraceae Euphorbiaceae Lythrum salicaria Euphorbia helioscopia

Onagraceae Fabaceae Epilobium angustifolium Anthyllis baltica Epilobium spp Lathyrus japonicus Lathyrus pratensis Papaveraceae Lotus corniculatus Fumaria officinalis Medicago falcata Medicago lupulina Plantaginaceae Melilotus albus Plantago lanceolata Trifolium arvense Plantago media Trifolium medium Trifolium montanum Polygonaceae Trifolium pratense Rumex acetosa Trifolium repens Rumex acetosella Vicia cracca Rumex thyrsiflorus Vicia sepium Primulaceae Geraniaceae Lysimachia vulgaris Geranium palustre Geranium pratense Geranium sylvaticum Ranunculaceae Ranunculus acris Hypericaceae Ranunculus cassubicus Hypericum maculatum Ranunculus flammula Hypericum perforatum Trollius europaeus

Iridaceae Rosaceae Iris pseudacorus Agrimonia eupatoria Alchemilla vulgaris Lamiaceae Comarum palustre Acinos arvensis Filipendula ulmaria Galeopsis tetrahit Filipendula vulgaris Lamium album Fragaria vesca Mentha arvensis Geum rivale Origanum vulgare Geum urbanum Prunella vulgaris Potentilla anserina Scutellaria galericulata Potentilla argenta Thymus serpyllum Potentilla erecta

64 Rubiaceae Melampyrum sylvaticum Galium album Rhinanthus minor Galium boreale Rhinanthus serotinus Galium molluga Verbascum nigrum Galium palustre Veronica chamaedrys Galium Spurium Veronica longifolia Galium uliginosum Veronica officinalis Galium verum Veronica spicata

Scrophulariaceae Solanaceae Euphrasia officinalis Solanum dulcamara Linaria vulgaris Melampyrum nemorosum Valerianaceae Melampyrum pratense Valeriana officinalis

65 SUMMARY IN ESTONIAN

Abiootiliste ja biootiliste faktorite mõju kimalaste populatsioonidele poollooduslikes kooslustes: maastikuline analüüs

Kimalasi peetakse agroökosüsteemide oluliseks tolmeldajarühmaks ning elurikkuse kujundajateks. Nende tegevus suurendab põllukultuuride saagikust ja looduslike taimekoosluste mitmekesisust. Samas on viimas- tel aastakümnetel märgatud kimalaste arvukuse vähenemist ja seda just intensiivse põllumajandusega aladel. Põhjuseid on mitmeid, kuid enim tuuakse välja põllumajanduslike kemikaalide liigset kasutamist, kimalaste toidubaasi kadumist ja elupaikade hävinemist.

Kimalaste kui olulise tolmeldajarühma efektiivse kaitse korraldamiseks on vaja teada nende seoseid elupaiga ja seda ümbritseva maastikuga. Kõige rohkem on kimalasi uuritud Lääne-Euroopa põllumajandusmaas- tikes, mida iseloomustavad monokultuuride kasvatamine, maastike ava- tus ja homogeensus. Märksa vähem on tähelepanu pööratud kimalaste ökoloogia, leviku ja maastiku mõju uurimisele mitmekesistes maastikes, mis võivad bioloogilise mitmekesisuse säilitamise seisukohast osutuda olulisemaks intensiivpõllumajanduslikest aladest. Eesti maastikud on mosaiiksed, koosnedes põllumajandusaladest, poollooduslikest ja loo- duslikest aladest ning metsamassiividest, mistõttu on sellistes looduslikes tingimustes võimalik saada kimalaste kohta uut olulist informatsiooni. Sellest tulenevalt püstitati antud uurimuse põhieesmärgiks selgitada välja abiootiliste ja biootiliste faktorite mõju kimalaste liigirikkusele ja arvukusele nii elupaiga kui ka elupaika ümbritseva maastiku tasandil. Uurimisalaks valiti suure metsa ja looduslike elupaikade osatähtsusega Ida-Virumaa, kus 2008.-2009. aastal määrati 22-l poollooduslikul niidul kimalaste liigiline koosseis ja arvukus. Lisaks analüüsiti elupaiga ning ümbritseva maastiku peamisi parameetreid ja struktuuri.

Kuigi Ida-Virumaa loodust mõjutavad oluliselt tööstusettevõtted, kae- vandused ning elektrijaamad, on kimalaste liigirikkus antud piirkonnas suur ja tööstuse mõju kimalaste populatsioonidele ei täheldatud. Kokku leiti 22 liiki kimalasi, sealhulgas 5 liiki kägukimalasi, mis on 80% Eestis teadaolevatest kimalaseliikidest. Kõige arvukamad liigid olid põldkima- lane (Bombus pascuorum), maakimalane (B. lucorum) ja tumekimalane (B. ruderarius). Haruldasematest liikidest leiti ristikukimalast (B. distinguendus) ja samblakimalast (B. muscorum).

66 Maastikulise analüüsi põhjal võib teha järgmised järeldused:

• elupaiga ehk niidu tasandil määrab kimalaste liigirikkust ja arvukust õitsvate toidutaimede liigirikkus ja arvukus;

• maastiku tasandil mõjutab kimalaste populatsioone positiivselt poollooduslike niitude osatähtsus, mis suurendab kimalaste, eriti spetsia- listliikide nagu pikasuiseliste liikide, arvukust ja liigirikkust;

• elupaika ümbritsevas maastikus suurendab kimalaste liigirikkust ja arvukust õiterohkete aedadega inimasustuse lähedus. Negatiivset mõju avaldab suurte avatud põllumaade lähedus;

• oluliseks kimalaste liigirikkuse ja arvukuse suurendajaks on metsa- servad ja muud koosluste piirialad, mis pakuvad kimalastele mitme- kesisemaid olusid, rohkem toitu, varjepaiku ja pesitsusalasid;

• selgus, et erinevad kimalaseliigid eelistavad erinevaid maastikuele- mente. Näiteks Schrencki kimalased (B. schrencki) eelistavad suure- ma metsa osatähtsusega maastikke, hallkimalased (B. veteranus) ja karukimalased (B. terrestris) aga seevastu rohkem avatud maastikke.

Antud uurimus näitas, et kimalaste liigirikkust ja arvukust suurendab mitmekesine maastik koos erinevate maastikuelementidega, sobilike elupaikade, poollooduslike koosluste ja servalade rohkus.

Uurimus näitas, et kimalaste arvukus ja liigirikkus ei sõltu ainult bio- loogilistest teguritest nagu toidu olemasolu, vaid sõltub ka abiootilis- test teguritest ja maastikku ning elupaika iseloomustavatest näitajatest. Seetõttu sobivad kimalased hästi elupaiga kvaliteeti iseloomustavateks indikaatorliikideks. Kimalaste liigirikkuse ja arvukuse hoidmiseks tuleb säilitada maastikulist mitmekesisust metsakoosluste, poollooduslike ja looduslike niitude ja neid ühendavate looduslike koridoridega ning seda mitte ainult piiratud elupaiga tasandil, vaid laiemal, elupaika ümbritse- val maastikulisel tasandil.

67 ACKNOWLEDGEMENTS

First and foremost, thank you God for having made everything possible by giving me strength and courage to finish this thesis.

I would like to thank my beloved husband, Mario Alberto Luna del Risco, without your love and comprehension this would not have been possible. Special thanks to my parents, sisters and brother who are always there for me, thank you for always giving me your love and support! Thanks to my grandmother for her motivational words.

My deepest gratitude to Prof. Valdo Kuusemets, my main supervisor, for his encouragement, guidance and patience he demonstrated during my studies. I am very grateful to Prof. Marika Mänd, my co-supervisor, for her invaluable help and guidance throughout the process, and for being always very supportive and encouraging.

I am very thankful to Dr. Reet Karise and Dr. Jaan Liira for their useful comments that greatly improved my thesis. Special thanks to Ave Lii- vamägi for being a great partner in the fieldwork and during the whole process, and for her continuous support. I would like to thank Tanel Kaart for his contribution and advice on statistics and data analysis. I am also grateful to Eda Tursk for her advice during my studies. Thanks to In- grid H. Williams, Kadri Kask and Monika Suškevics for linguistic help.

I would like to express my gratitude to all those who, somehow, gave me courage and support during my studies, especially to my friends: Laura and Immu (and little Otto), Clotilde and Miguel (and little Diego), Miguelito, Merje and Gregory, Floor and Jeroen (and the girls, Sirle and Hadewig), Manuel and Yianna, Alejandra, Shirley, Claudia and all my other friends and family in Colombia who were always there for me, even from the distance.

Doctoral studies were supported by Ministry of Education and Research and (or in cooperation with) Archimedes Foundation. This research was funded by targeted financing from the Estonian Ministry of Education and Research (SF1090050s07, SF0170057s09), Estonian Science Foun- dation Grant No. 7391 and by an applied research project of the Esto- nian Ministry of Agriculture (T8014PKPK). I

Publications Diaz-Forero, I., Liivamägi, A., Kuusemets, V. and Luig J. 2010

Pollinator richness and abundance in Northeast Estonia: bumblebees, butterflies and day-flying moths

Forestry Studies | Metsanduslikud Uurimused 53, 5–14 Forestry Studies|Metsanduslikud Uurimused 53, 5–14, 2010 DOI: 10.2478/v10132-011-0085-7 Pollinator richness and abundance in Northeast Estonia: bumblebees, butterﬂies and day-ﬂying moths

Isabel Diaz-Forero*, Ave Liivamägi, Valdo Kuusemets and Jaan Luig

Diaz-Forero, I., Liivamägi, A., Kuusemets, V., Luig, J. 2010. Pollinator richness and abundance in Northeast Estonia: bumblebees, butterﬂies and day- ﬂying moths. – Forestry Studies | Metsanduslikud Uurimused 53, 5–14. ISSN 1406-9954.

Abstract. We studied diversity and abundance of three groups of pollinators: bumblebees, butterflies and day-flying moths, in Ida-Viru County, Estonia. The field work was done during the summers of 2008 and 2009, in 22 semi-natural meadows located across Northeast Estonia. In total, we found 22 species of bumblebees (gen. Bombus), including 5 species of cuckoo bumblebees, 56 species of butterflies and 42 species of day-flying moths. We recorded 597 individuals of bumblebees, 768 individuals of butterflies and 330 individuals of day-flying moths in our study sites. We analysed differences between years (2008 and 2009) for the species richness and abundance of bumblebees, butterflies and day-flying moths; the relationships between insect species richness and area of meadow, the total number of species per meadow and the local abundance of each species separately. We found significant differences between bumblebee richness and abundance in 2008 and 2009, and between butterfly abundance at the same years. These differences may be due to more favourable conditions: warmer and dryer weather in the second year. We found no significant relationships between area of meadow and species richness of bumblebees, butterflies and day-flying moths; however, when we analysed the abundance of each species separately, we found that two species of bumblebees, i.e. B. pascuorum and B. schrencki, and one species of day-flying moths, i.e. Chiasmia clathrata, were negatively related with area of meadow. Although, Northeast Estonia is a region that has been environmentally affected by mining activities and the presence of power plants, it could be considered an important area that supports a significant richness and abundance of pollinators across its territory. Key words: insects, species richness, semi-natural meadows, Bombus, Lepi- doptera, Ida-Virumaa. Authors’ address: Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Kreutzwaldi, 5, 51014 Tartu, Estonia, *e-mail: [email protected]

Introduction

Grasslands are the most species-rich habitats in European landscapes. Insects constitute an important part of the biodiversity of semi-natural grasslands (Öckinger & Smith, 2007), as they provide unique ecosystem services in the form of nutrient cycling and pollination. Insects are also important environmental indicators because they respond to climatic and management changes faster than plants, which they

5 71 I. Diaz-Forero et al. need for food and reproduction. Pollinator diversity in semi-natural grasslands is at risk mainly because of the intensification of farming practices (i.e. increased use of fertilisers and pesticides) (Carvell, 2002; Mänd et al., 2002; Goulson et al., 2006; Holzschuh et al., 2008 Xie et al., 2008), abandonment of traditional agricultural land use practices (e.g. mowing, grazing, etc.) and successive transformation of remnants into forest. The remaining semi-natural grasslands become more fragmented in the landscape context. All those factors are significantly affecting the diversity of insect communities inhabiting semi-natural grasslands (Cozzi et al., 2008; Bergman et al., 2008; Sjödin et al., 2008). The loss of pollinators has been an important topic during recent years. There are a lot of articles describing that many bumblebee and butterfly species have under- gone significant range declines in different European countries (Mänd et al., 2002; Kells & Goulson, 2003; Goulson et al., 2006; Williams & Osborne, 2009). In contrary to the most of Europe, there are relatively few studies that have reported increases of butterflies in their abundance and distribution (Kuussaari et al., 2007). They observed that increasing trends and expansions of butterflies are generally associated with climate change (Oliver et al., 2009). There is little information on density changes of different species of bumblebees. Some bumblebee species have decreased drastically, but some other species have increased (Goulson et al., 2006). We studied diversity and abundance of three groups of pollinators: bumblebees, butterflies and day-flying moths. The main objective of our research was to determine the species richness and abundance of these three groups of insects in 22 meadows located in Northeast Estonia. We examined the differences between years (2008 and 2009) for the species richness and abundance of bumblebees, butterflies and day-flying moths; and the relationship between insect species richness and area of meadow, analysing first, the total number of species per meadow, and then, the abundance of each species separately.

Materials and methods Study region Ida-Viru is a county located in the Northeastern part of Estonia. The total area of our study region is 3364 km2, which represents 7.4% of the total area of Estonia. The areas of the selected meadows ranged from 0.10 to 3.83 ha. Ida-Virumaa is a region affected mostly by mining activities as it contains large deposits of oil shale, a mineral used for power generation in Estonia. The landscape in the region is mainly dominated by forests, grasslands and arable land, and in a lower proportion, by mires and fresh water bodies. Study sites were chosen in Northeast Estonia with grasslands situated in coastal areas, in forested landscapes and in ﬂooded meadows.

Field work We visited 22 semi-natural meadows in 2008 and 2009, located in Ida-Viru county. In both years, we sampled each meadow two times. Field works took place in June, July and August, which are the warmest months of the year. Insect counts were done during approximately 45 minutes systematic walking surveys (Kumar et al., 2009); during the warmer time of the day, between 11:00 h and 16:00 h; and when the weather conditions were suitable, i.e. temperature was above 18 ºC and wind speed was less than 5 by Beaufort scale. The number of species and individuals of the three insect

6 72 Pollinator richness and abundance in Northeast Estonia: bumblebees, butterflies and day-flying moths groups were determined by sight at each meadow. When the observer could not identify the species, the individual was caught with an insect net for later identification. The nomenclature of the insects follows Fauna Europaea Web Service (2004).

Weather conditions In our study area, the closest weather station is Jõhvi that is situated in the middle of the region. The average air temperature per month was higher in 2009, compared with the previous year. We found larger differences between the sums of precipitation: in June, it was higher in 2008 than in 2009 (128 mm and 112 mm, respectively); the same trend was observed in August, with a heavy rain period in 2008 that reached 242 mm (96 mm in 2009); in May the sum of precipitation was similar in both years (28 mm in 2008 and 33 mm in 2009); and in July the sum of precipitation was higher in 2009 than in 2008 (139 mm and 55 mm, respectively). In general, the second year of ﬁeld work, 2009, had more sunny days and favourable weather conditions for insects.

Statistical analyses In our study, Pearson’s correlation tests were performed to analyse the relationship between area of meadow and insect species richness, using first, the total number of species per meadow, and then, the local abundance of each species separately. In addition, we examine the non-parametric relations between the same variables using the Spearman Rank Order correlations. We conducted dependent samples (paired) t-tests in order to evaluate the differences between number of individuals and species in 2008 and 2009 for the three groups of insects separately: bumblebees, butterflies and day-flying moths. We used the software STATISTICA 9 for all the statistical analyses.

Results Bumblebee richness and abundance We recorded a total of 22 species of bumblebees (gen. Bombus), including 5 species of cuckoo bumblebees (subgen. Psithyrus). In Estonia, there are 22 species of bumblebees and 7 species of cuckoo bumblebees. An average of 10.7 species and 27.1 individuals of bumblebees per study site were found (Figs. 1a–1b). We counted a total of 597 individuals in our study sites: 207 in 2008 and 390 in 2009. From the total number of individuals of bumblebees found, 363 were workers, 150 males and 84 females. The bumblebee species with the highest number of individuals were B. pascuorum, B. lucorum and B. ruderarius with 140, 70 and 58 individuals, respectively (Fig. 2a). Together with B. cryptarum, these species were also the most common bumblebees in the area, as they were found in most of the study sites. On the other hand, the two rarest bumblebee species with the lowest abundance were B. muscorum and B. distinguendus. We found ﬁve species of long-tongued bumblebees: B. distinguendus, B. hortorum, B. ruderarius, B. sylvarum, and B. pascuorum; and eight species of short-tongued bumblebees: B. cryptarum, B. lapidarius, B. lucorum, B. terrestris, B. hypnorum, B. jonellus, B. pratorum and B. soroeensis (http://www.nhm.ac.uk). The other species of bumblebees found have a mid-length tongue. Three species: B. distinguendus, B. muscorum and B. soroeensis are included in the Estonian Red List of Threatened Species (http:// elurikkus.ut.ee).

7 73 I. Diaz-Forero et al.

��

Figure 1. Box-plots showing total number of (a) species and (b) individuals of bumblebees, butter- ﬂies and moths found in the study area. Joonis 1. Karpdiagrammil (a) liikide ja (b) isendite koguarvu jaotumine erinevates putukarühmades.

The places with the highest richness of bumblebees were Kalvi with 18 species, and Soldina, Narva Karjäär and Pannjärve with 15 species each; and the places with the lowest number of species were Valaste and Raadna Oja, with 2 and 3 species of bumblebees, respectively (Table 1). We found significant differences between species richness and abundance of bumblebees in 2008 and 2009 (t = 6.0, df = 21, p = 0.000006; t = 5.7, df = 21, p = 0.00001, respectively): the number of species and individuals was significantly higher in the second year. Concerning the relations between bumblebee species richness and area of meadow, we found no significant correlations between the variables (p > 0.05). However, when the local abundance of each species and area of meadow were analysed, using non- parametric (Spearman rank order) correlations, we found that two species of bumblebees, i.e. B. pascuorum and B. schrencki, were both negatively correlated with area of meadow (r = -0.5, n = 22, p < 0.05).

Butterfly richness and abundance A total of 768 individuals of butterflies belonging to 56 species were found in our study sites: 333 individuals in 2008 and 435 individuals in 2009. We found an average of 15.6 species and 34 individuals of butterflies per study site (Figs. 1a–1b). The most abundant species found were Coenonympha glycerion with 65 individuals, Aphantopus hyperantus with 62 individuals and Thymelicus lineola with 61 individuals (Fig. 2b). The most common species found in our study sites were Thymelicus lineola, Aphantopus hyperantus and Pieris napi. The species with the lowest abundance were Vanessa atalanta, Carterocephalus palaemon, Lycaena virgaureae, Pieris rapae, Polygonia c-album, Papilio machaon, Lycaena alciphron and Heteropterus morpheus. In our study region, we found 5 species of butterflies that are protected under the EU Habitat Directive Natura 2000 (http://elurikkus.ut.ee): Parnassius mnemosyne, Lycaena dispar, Euphydryas maturna, Euphydryas aurinia and Coenonympha hero.

8 74 Pollinator richness and abundance in Northeast Estonia: bumblebees, butterﬂies and day-ﬂying moths

�� Figure 2. Total number of individuals per species �� of (a) bumblebees, (b) butterﬂies and �� (c) day-ﬂying moths. �� Joonis 2. Uurimisalalt püütud isendite arvu �� jaotumine liigiti kimalastel (a), päe- � �� valiblikatel (b) ja päeval aktiivsetel �� hämarikuliblikatel (c).

9 75 I. Diaz-Forero et al.

Table 1. Number of species and individuals of bumblebees, butterﬂies and day-ﬂying moths. Nota- tions: SR – total species richness, IND – total number of individuals. Tabel 1. Kimalaste, päevaliblikate ja päeval aktiivsete hämarikuliblikate isendite (IND) ja liikide (SR) arv uurimisaladel.

Bumblebees Butterflies Day-flying moths Kimalased Päevaliblikad Päeval aktiivsed hämarikuliblikad # Study site Area Geographic SR IND SR IND SR IND Uurimisalad (ha) coordinates Pindala Geograafilised (ha) koordinaadid 1 Smolnitsa 0.12 59º00’38”N 14 37 24 46 6 9 27º36’52”E 2 Kuningaküla 0.69 59º07’35”N 12 49 14 31 6 7 27º48’10”E 3 Raadna Oja 3.38 58º58’53”N 3 7 15 30 8 11 27º07’31”E 4 Lemmaku 3.61 59º00’20”N 9 22 16 25 6 10 27º09’36”E 5 Mäetaguse 3.65 59º13’40”N 12 21 17 27 12 29 27º19’44”E 6 Atsalama 1.47 59º14’31”N 9 31 22 52 14 32 27º23’12”E 7 Kurtna 1.12 59º14’59”N 8 15 16 45 13 32 27º33’58”E 8 Pannjärve 0.48 59º17’18”N 15 36 29 53 12 21 27º33’24”E 9 Tagajõgi_R 0.23 59º04’54”N 11 22 11 24 11 16 27º04’37”E 10 Tagajõgi_O 1.02 59º07’12”N 9 22 18 51 5 14 26º59’26”E 11 Piilse 3.87 59º14’09”N 7 15 8 16 11 23 26º59’31”E 12 Kalvi 3.32 59º28’52”N 18 40 10 17 7 9 26º46’36”E 13 Kalvi_Oru 1.23 59º27’34”N 10 18 6 10 11 14 26º48’24”E 14 Liimala 1.42 59º26’06”N 12 37 11 27 4 5 26º59’01”E 15 Moldova 1.91 59º25’59”N 13 28 24 65 6 8 27º04’11”E 16 Valaste 0.89 59º26’18”N 2 4 8 11 3 6 27º25’20”E 17 Päite 1.06 59º24’16”N 10 20 11 21 7 13 27º42’02”E 18 Udria 0.82 59º23’36”N 10 33 15 39 9 17 27º54’07”E 19 Soldina 0.26 59º23’01”N 15 45 14 37 7 18 28º04’43”E 20 Auvere 3.59 59º20’36”N 9 23 15 40 7 13 27º58’20”E 21 Narva_K 0.91 59º15’52”N 15 41 20 45 8 12 27º49’15”E 22 Meriküla 1.37 59º24’46”N 13 31 18 56 7 11 27º57’12”E

10 76 Pollinator richness and abundance in Northeast Estonia: bumblebees, butterﬂies and day-ﬂying moths

Pannjärve, Moldova and Narva Karjäär were sites with the highest number of species of butterflies with 29, 24 and 20 species, respectively. On the contrary, the sites with the lowest butterfly diversity were Kalvi Oru with 6 species, and Valaste and Piilse with 8 species each (Table 1). According to the t-tests results, we found significant differences between number of individuals of butterflies in 2008 and 2009 (t = 3.0, df = 21, p = 0.007): it was higher in the second year. However, there was no significant difference between years in the case of butterfly species richness (p > 0.05). From the analysis of the relationship between butterfly species richness and area of meadow, we found no significant correlations (p > 0.05). The same results were obtained when we analysed the relations between local abundance of each species and area of meadow.

Day-flying moth richness and abundance We found a total of 42 species and 330 individuals of day-flying moths in our study sites. 179 individuals were recorded in 2008 and 151 in 2009. In average, we found 8.2 species and 13 individuals of moths per study site (Figs. 1a–1b). The most abundant species found in Ida-Virumaa were Siona lineata and Scopula immorata, with 48 and 45 individuals, respectively (Fig. 2c). These two were also the most common species found in the majority of our study sites, along with Euclidia glyphica. Among the species of day-flying moths, we found that Rheumaptera hastata is considered a species of least concerned in the Estonian Red List of Threatened Species (http://elurikkus.ut.ee). The places with the highest diversity of day-flying moths were: Atsalama hoiuala with 14 species, Kurtna Särgjärv with 13, and Pannjärve and Mäetaguse, both with 12 species (Table 1). Our results from t-tests show that there were no significant differences between moth diversity and abundance in 2008 and 2009 (p > 0.05). We also found that species richness of moths was not significantly correlated with area of meadow (p > 0.05). In the case of local abundance of each species of moths, we found that there was a negative non-parametric (Spearman rank order) correlation between the species Chiasmia clathrata and area of meadow (r = -0.4, n = 22, p < 0.05).

Discussion Semi-natural habitats constitute areas of great value for pollinators, particularly bumblebees and butterﬂies, mainly due to the availability of a broad range of ﬂowering resources and nesting places (Svensson et al., 2000; Kells & Goulson, 2003; Öckinger & Smith, 2007). Estonia has a very mosaic landscape with a mixture of forest, agricultural land and semi-natural areas (Mänd et al., 2002). Although Ida-Viru is a region mainly dominated by forest (which constitutes 58% of the whole area), semi-natural meadows represent important targeting areas for biodiversity conservation. Among the bumblebee species found in Northeast Estonia, B. pascuorum was the most common and abundant species. This is a very common bumblebee species, particularly in Europe and northern Asia (http://www.nhm.ac.uk). In a study done in Estonia by Mänd et al. (2002), they also found that B. pascuorum and B. lucorum were some of the most dominant bumblebee species in semi-natural habitats, and B. muscorum was one of the rarest species. The extremely rare and declining bumblebees

11 77 I. Diaz-Forero et al.

B. muscorum and B. distinguendus (mid-tongued and long-tongued species, respectively), are late-emerging species that are associated with unimproved grasslands (Goulson et al., 2005). In our study, these species were found in places located close to fresh waterbodies. This is consistent with the results obtained by Goulson et al. (2006), in which several declining bumblebee species were found mostly in coastal areas. They suggested that a possible explanation for the distribution of these rare species was that coastal areas are less impacted by agriculture and intensive farming (Goulson et al., 2006). Bumblebee species richness and abundance was higher in 2009, compared with 2008; it may be due to more favourable weather conditions in the second year. The same could apply in the case of butterfly abundance, as the weather in the summer of 2008 was colder and more windy than in 2009. The butterfly species Coenonympha glycerion was the most abundant species found in our study area. However, in countries like Finland, it has shown a consistent decline in semi-natural grasslands (Kuussaari et al., 2007; Pöyry et al., 2009). Pöyry et al. (2009) explained that this species has shown a declining trend in Finland mainly due to decreasing availability of habitats. This species has been also commonly found in bog habitats (Kulfan et al., 1997). Other butterfly species that we found to be dominant in our study sites (i.e. Aphantopus hyperantus, Thymelicus lineola and Pieris napi) are also considered common species in Finish agricultural landscapes (Pöyry et al., 2005; Kuussaari et al., 2007). Contrary to our results, in Finland Pieris napi has been commonly found in arable field margins (Kuussaari et al., 2007). The most abundant species of day-flying moths in our study sites were Siona lineata and Scopula immorata. According to Pöyry et al. (2005), Siona lineata is commonly found in old and abandoned pastures, while Scopula immorata seems to prefer the second type of habitat. Among our study sites, we found that Pannjärve was the place with one of the highest species richness of bumblebees, butterflies and day-flying moths. The study site Narva Karjäär was also found to have one of the highest numbers of species of bumblebees and butterflies. On the other hand, Valaste was one of the places with the lowest number of species of bumblebees and butterflies. This place is located in the northern part of Ida-Viru county, very close to the Gulf of Finland in the Baltic sea. Although area of meadow did not seems to be an important factor for the total species richness of our three groups of pollinators, there seems to be a negative influence on some species of bumblebees (i.e. B. pascuorum and B. schrencki) and day-flying moths (i.e. Chiasmia clathrata).

Conclusion Even though, Northeast Estonia is a region that has been environmentally affected by mining activities and the presence of power plants, it could be considered an important area for conservation of some species of pollinators, as it supports a significant richness and abundance of bumblebees, butterflies and day-flying moths across its territory. The mosaic landscape with forests, grasslands and arable areas, and the presence of semi-natural meadows in the region seems to be favourable for pollinators.

Acknowledgements. The study was funded by targeted ﬁnancing of the Estonian Ministry of Education and Research (SF1090050s07) and by the applied research

12 78 Pollinator richness and abundance in Northeast Estonia: bumblebees, butterﬂies and day-ﬂying moths project of the Estonian Ministry of Agriculture (T8014PKPK). We greatly thank Prof. Marika Mänd for her continuous support and useful comments, and the anonymous reviewers for their helpful comments on the manuscript.

References Bergman, K-O., Ask, L., Askling, J., Ignell, H., Wahlman, H., Milberg, P. 2008. Importance of boreal grasslands in Sweden for butterfly diversity and effects of local and landscape habitat factors. – Biodiversity and Conservation, 17, 139–153. Carvell, C. 2002. Habitat use and conservation of bumblebees (Bombus spp.) under different grassland management regimes. – Biological Conservation, 103, 33–49. Commission for Nature Conservation of the Estonian Academy of Sciences. 2001-2002. Estonian Red List of Threatened Species. [WWW document]. – URL http://elurikkus.ut.ee. Cozzi, G., Müller, C. B., Krauss, J. 2008. How do local habitat management and landscape structure at different spatial scales affect fritillary butterfly distribution on fragmented wetlands? – Landscape Ecology, 23, 269–283. European Commission. Habitats Directive (92/43/EEC) European Commission (1992/1995). Council Directive 92/43/EEC of 21 May 1992 on the conservation of natural habitats and of wild fauna and flora. [WWW document]. – URL http://europa.eu. Fauna Europaea Web Service. 2004. Fauna Europaea version 1.1. [WWW document]. – URL http://www. faunaeur.org. Goulson, D., Hanley, M.E., Darvill, B., Ellis, J.S. 2006. Biotope associations and the decline of bumblebees (Bombus spp.). – Journal of Insect Conservation, 10, 95–103. Goulson, D., Hanley, M.E., Darvill, B., Ellis, J.S., Knight, M.E. 2005. Causes of rarity in bumblebees. – Bio- logical Conservation, 122, 1–8. Holzschuh, A., Steffan-Dewenter, I., Tscharntke, T. 2008. Agricultural landscapes with organic crops support higher pollinator diversity. – Oikos, 117, 354–361. Kells, A.R., Goulson, D. 2003. Preferred nesting sites of bumblebee queens (Hymenoptera: Apidae) in agroecosystems in the UK. – Biological Conservation, 109, 165–174. Kulfan, M., Degma, P., Kalivoda, H. 1997. Lepidoptera of different grassland types across the Morava floodplain. – Journal of Research on the Lepidoptera, 34, 39–47. Kumar, S., Simonson, S.E., Stohlgren, T.J. 2009. Effects of spatial heterogeneity on butterfly species richness in Rocky Mountain National Park, CO, USA. – Biodiversity and Conservation, 18, 739–763. Kuussaari, M., Heliölä, J., Pöyry, J., Saarinen, K. 2007. Contrasting trends of butterfly species preferring semi-natural grasslands, field margins and forest edges in northern Europe. – Journal of Insect Conservation 11, 351–366. Mänd, M., Mänd, R., Williams, I.H. 2002. Bumblebees in the agricultural landscape of Estonia. – Agriculture, Ecosystems & Environment, 89, 69–76. Natural History Museum. 2011. Research and curation. [WWW document]. – URL http://www.nhm. ac.uk. Oliver, T., Hill, J.K., Thomas, C.D., Brereton, T., Roy, D.B. 2009. Changes in habitat specificity of species at their climatic range boundaries. – Ecology Letters, 12, 1091–1102. Öckinger, E., Smith, H.G. 2007. Semi-natural grasslands as population sources for pollinating insects in agricultural landscapes. – Journal of Applied Ecology, 44, 50–59. Pöyry, J., Lindgren, S., Salminen, J., Kuussaari, M. 2005. Responses of butterfly and moth species to restored cattle grazing in semi-natural grasslands. – Biological Conservation, 122, 465–478. Pöyry, J., Luoto, M., Heikkinen, R.K., Kuussaari, M., Saarinen, K. 2009. Species traits explain recent range shifts of Finnish butterflies. – Global Change Biology, 15, 732–743. Sjödin, N. E., Bengtsson, J., Ekbom, B. 2008. The influence of grazing intensity and landscape composition on the diversity and abundance of flower-visiting insects. – Journal of Applied Ecology, 45, 763–772. Svensson, B., Langerlöf, J., Svensson, B.G. 2000. Habitat preferences of nest-seeking bumble bees (Hy- menoptera: Apidae) in an agricultural landscape. – Agriculture, Ecosystems & Environment, 77, 247–255. Williams, P.H., Osborne, J.L. 2009. Bumblebee vulnerability and conservation world-wide. – Apidologie, 40, 367–387. Xie, Z., Williams, P.H., Tang, Y. 2008. The effect of grazing on bumblebees in the high rangelands of the eastern Tibetan Plateau of Sichuan. – Journal of Insect Conservation, 12, 695–703.

13 79 I. Diaz-Forero et al.

Kolme tolmeldajaterühma: kimalased, päevaliblikad ja päeval aktiivsed hämarikuliblikad, liigirikkus ja arvukus Ida-Virumaal Isabel Diaz-Forero*, Ave Liivamägi, Valdo Kuusemets ja Jaan Luig

Kokkuvõte

Töö eesmärgiks oli uurida kolme tolmeldajate rühma: kimalased, päevaliblikad ja päeval aktiivsed hämarikuliblikad, liigirikkust ja arvukust Ida-Virumaal. Välitööd tehti aastatel 2008 ja 2009 22-l poollooduslikul niidul. Kimalasi leiti 22 liiki (gen. Bombus), sealhulgas 5 liiki kägukimalasi (subgen. Psithyrus), 597 isendit. Kõige arvukamad liigid olid B. pascuorum, B. lucorum and B. ruderarius. Päevaliblikaid leiti 56 liiki ja 768 isendit, kõige arvukamad liigid olid Coenonympha glycerion, Aphantopus hyperantus ja Thymelicus lineola. Päeval aktiivseid hämarikuliblikaid leidsime 42 liiki ja 330 isendit, kõige arvukamad liigid olid Siona lineata ja Scopula immorata. Kimalaste liigirikkus ja arvukus ning liblikate arvukus erinesid statistiliselt oluliselt aastatel 2008 ja 2009. See võis olla põhjustatud erinevate aastate ilmastikutingi- muste erinevustest, kuivõrd 2009. aasta oli kuivem ja soojem. Elupaiga suurus ei avaldanud statistiliselt usaldusväärset mõju uuritud tolmeldajate rühmade liigirikkusele. Samas omas elupaiga suurenemine negatiivset mõju mõningate liikide, nagu B. pascuorum, B. schrencki ja Chiasmia clathrata arvukusele. Received December 3, 2010, revised January 01, 2011, accepted January 28, 2011

14 80 II Diaz-Forero, I., Kuusemets, V., Mänd, M., Liivamägi, A., Kaart, T. and Luig, J. 2011

Influence of local and landscape factors on bumblebees in semi-natural meadows: a multiple-scale study in a forested landscape

Journal of Insect Conservation. (Submitted) Journal of Insec t Conservation Influence of local and landscape factors on bumblebees in semi-natural meadows: a multiple-scale study in a forested landscape --Manuscript Draft--

Manuscript Number: Full Title: Influence of local and landscape factors on bumblebees in semi-natural meadows: a multiple-scale study in a forested landscape

Article Type: Manuscript Keywords: Bombus; land cover types; landscape indices; Fragstats; Partial Least Squares (PLS); landscape matrix

Corresponding Author: Isabel Diaz-Forero, MSc. Estonian University of Life Sciences Tartu, ESTONIA

Corresponding Author Secondary Information: Corresponding Author's Institution: Estonian University of Life Sciences Corresponding Author's Secondary Institution: First Author: Isabel Diaz-Forero, MSc. First Author Secondary Information: All Authors: Isabel Diaz-Forero, MSc. Valdo Kuusemets, PhD. Marika Mänd, PhD. Ave Liivamägi, MSc. Tanel Kaart, PhD. Jaan Luig, MSc.

All Authors Secondary Information: Abstract: Understanding the effects of local and landscape factors on bumblebees is relevant for the conservation of this group of pollinators. Bumblebees have been well-studied in agricultural landscapes of Western Europe, Asia and North America, but few studies have been developed on bumblebees in forest-dominated landscapes of Eastern Europe. We developed this study in 22 semi-natural meadows located in a patchy forested landscape of Estonia. We investigated the influence of habitat characteristics and landscape factors (calculated at four spatial scales: 250 m, 500 m, 1000 m and 2000 m radius) on the total species richness and abundance of bumblebees. Correlation analysis, Partial Least Squares (PLS) and stepwise forward-selection multiple regression analysis were applied in this study. The presence of a high diversity of flowering plants in semi-natural meadows may benefit the abundance of bumblebees. At the local level, patch area and shape seem to have positive and negative influences, respectively, on bumblebee species richness. At the landscape level, human settlements with the presence of gardens may favour bumblebee richness and abundance. Also, bumblebee species may increase with a high presence of meadows in the landscape, and may decrease with high percentages of forest and brushwood. Overall, forested landscapes with a strong presence of edges and a diverse matrix may support a higher diversity and abundance of bumblebees. Both local and landscape factors should be considered when designing conservation strategies and agri-environmental measures.

83 Manuscript Common.Links.ClickHereToDownload Common.Links.ClickHereToViewLinkedReferences

1 1 2 3 4 Influence of local and landscape factors on bumblebees in semi-natural meadows: a multiple-scale study in a 5 forested landscape 6 7 Isabel Diaz-Forero, Valdo Kuusemets, Marika Mänd, Ave Liivamägi, Tanel Kaart, Jaan Luig 8 9 10 11 Abstract Understanding the effects of local and landscape factors on bumblebees is relevant for the conservation of 12 this group of pollinators. Bumblebees have been well-studied in agricultural landscapes of Western Europe, Asia 13 and North America, but few studies have been developed on bumblebees in forest-dominated landscapes of Eastern 14 Europe. We developed this study in 22 semi-natural meadows located in a patchy forested landscape of Estonia. We 15 16 investigated the influence of habitat characteristics and landscape factors (calculated at four spatial scales: 250 m, 17 500 m, 1000 m and 2000 m radius) on the total species richness and abundance of bumblebees. Correlation analysis, 18 Partial Least Squares (PLS) and stepwise forward-selection multiple regression analysis were applied in this study. 19 The presence of a high diversity of flowering plants in semi-natural meadows may benefit the abundance of bum- 20 blebees. At the local level, patch area and shape seem to have positive and negative influences, respectively, on 21 bumblebee species richness. At the landscape level, human settlements with the presence of gardens may favour 22 23 bumblebee richness and abundance. Also, bumblebee species may increase with a high presence of meadows in the 24 landscape, and may decrease with high percentages of forest and brushwood. Overall, forested landscapes with a 25 strong presence of edges and a diverse matrix may support a higher diversity and abundance of bumblebees. Both 26 local and landscape factors should be considered when designing conservation strategies and agri-environmental 27 measures. 28 29 30 Keywords Bombus, land cover types, landscape indices, Fragstats, Partial Least Squares (PLS), landscape matrix 31 32 33 I. Diaz-Forero (*), V. Kuusemets, M. Mänd, A. Liivamägi, J. Luig 34 Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Kreutzwaldi 5, 51014 35 Tartu, Estonia 36 37 E-mail addresses: [email protected], [email protected] 38 39 T. Kaart 40 Institute of Veterinary Medicine and Animal Sciences, Estonian University of Life Sciences, Kreutzwaldi 62, Tartu, 41 51014, Estonia 42 43 44 45 Introduction 46 47 In recent decades, the decline of insect pollinators, particularly bees, has been widely recognised. Evidence of this 48 decline has been documented across the world and in several European countries (Kells and Goulson 2003; Sam- 49 ways 2005; Williams and Osborne 2009; Potts et al. 2010). Bumblebees and other insects are a vital component of 50 51 global biodiversity as they play a key role in supporting not only crops, but also the diversity of natural and semi- 52 natural vegetation (Sepp et al. 2004; Goulson et al. 2006; Rundlöf et al. 2008; Ahrné et al. 2009; Knight et al. 2009; 53 Potts et al. 2010; Goulson et al. 2011). Their decline has been mainly associated with agricultural intensification that 54 is characterised by the rigorous use of fertilisers and pesticides and the reduction of flower resources (Carvell 2002; 55 Mänd et al. 2002; Goulson et al. 2006; Holzschuh et al. 2008; Xie et al. 2008; Ahrné et al. 2009); causing the frag- 56 mentation of landscapes and the loss of suitable habitats for insects (Krewenka et al. 2011). Agri-environmental 57 58 schemes are being applied in many European countries to alleviate the negative consequences of the intensification 59 of farming practices on biodiversity. The development of more effective agri-environmental measures has become 60 61 62 63 64 65

84 1 2 2 3 4 an issue of great concern among decision makers, mainly due to the growing interest of politicians, farmers and con- 5 sumers in more environmentally-friendly farming practices (Kleijn et al. 2006; Holzschuh et al. 2008). 6 Bumblebees have been well-studied in modern agricultural landscapes of Western Europe, the United Kingdom, 7 Japan and North America (Goulson et al. 2011). However, very little is known about the distribution, conservation 8 9 and ecology of bumblebees elsewhere (Goulson et al. 2011). Those areas that have been well-studied usually consist 10 of large monoculture fields separated by field margins and few patches of woodland. In contrast, the landscape in 11 Estonia has a very mosaic pattern, where 32% of the whole territory is agricultural land, but only a small proportion 12 is cultivated, and it contains many patches of natural habitat (Mänd et al. 2002). Moreover, there is evidence that the 13 proportion of forest in Estonia increased substantially during the 20th century (from 14% to 42%) (Palang et al. 14 1998). Understanding the associations between bumblebees and the surrounding landscape is relevant for the con- 15 16 servation of this group of pollinators, particularly in areas that have fragmented landscapes with high proportions of 17 forest and natural habitats. Generally, bumblebees have been studied in regions with warmer climates (compared to 18 Estonia) and open landscapes, mostly in agricultural landscapes (e.g., Steffan-Dewenter 2003; Hatfield and LeBuhn 19 2007; Krewenka et al. 2011). Therefore, research on bumblebee populations conducted on the northern regions and 20 in more forested landscapes are of great interest. To our knowledge, very few studies have been developed on the 21 influence of landscape factors on bumblebees in forested landscapes. 22 23 Most conservation strategies for pollinators have focused on habitat characteristics and requirements of species 24 at the local level, excluding the potential effects of the surrounding landscape mosaic (Steffan-Dewenter et al. 2002; 25 Cozzi et al. 2008). However, the effectiveness of biodiversity conservation usually depends on the knowledge about 26 the influence of different factors at multiple scales on the distribution of organisms and the conditions that allow 27 them to move across the landscape (Gutzwiller 2002). 28 In our study, we considered patch and landscape factors that may have potential relevance to the ecology of 29 30 bumblebee species richness and abundance; some of them have been used in previous studies on insects (e.g., Maze- 31 rolle and Villard 1999; Krauss et al. 2003; Kumar et al. 2009). Patch can be defined as “a relatively homogeneous 32 area that differs from its surroundings” (Freemark et al. 2002). For the purposes of this study, patch-scale refers to 33 local-scale: so when talking about patch-scale variables, it means the characteristics or factors at the scale of the 34 habitat or meadow under study. Our landscape variables were chosen considering a set of principles for applying 35 landscape ecology to biological conservation, suggested by Freemark et al. (2002). The first one is to “maintain 36 37 landscape mosaics that are more permeable” (Freemark et al. 2002); in this case we chose variables related with 38 connectivity or with aspects that may influence the dispersal patterns of species (e.g., interspersion and juxtaposition 39 index, edge density). The second one is to “maintain landscape mosaics with sufficient proportion of suitable habi- 40 tat” (Freemark et al. 2002); in our study we considered the proportion of important land cover types in the surround- 41 ing landscape (e.g., meadows, forest, brushwood). The third principle is: “with sufficient suitable habitat, patch size 42 distribution is of secondary importance” (Freemark et al. 2002); concerning this principle, we included factors re- 43 44 lated with the distribution and configuration of individual patches (e.g., mean patch area of forest), and with the con- 45 figuration of the landscape matrix (e.g., Shannon’s diversity index). From an ecological perspective, what comprises 46 a landscape will generally depend on the scales over which a given species moves and its perception of the sur- 47 roundings (Manning et al. 2004); that is, the relevant scale of a landscape depends on the organism, or the ecological 48 process under study (Freemark et al. 2002). In the case of bumblebees, we considered four spatial scales (i.e., 250 m, 49 500 m, 1000 m and 2000 m radius) taking into account the ranges of flight distances of different bumblebee species: 50 51 as it was reported in a recent study, some species can flight more than 2 km (Hagen et al. 2011). 52 Considering that Estonia has a patchy landscape with a relevant presence of natural and semi-natural land cover 53 types (mainly forest), the aim of this study was to investigate the influence of patch-scale and landscape-scale fac- 54 tors on the species richness and abundance of bumblebees. At the local level (or patch scale), we considered vari- 55 ables describing the vegetation structure and other spatial characteristics of the study sites (e.g., patch area, perime- 56 ter and shape). At the landscape level, we used the most important land cover types in the study region (i.e., forests, 57 58 meadows, brushwood, arable land and human settlements) and a set of landscape indices that were calculated at 59 multiple spatial scales. 60 61 62 63 64 65

85 1 3 2 3 4 5 Materials and methods 6 7 Study area 8 9 10 The study was carried out in Ida-Virumaa, which is located in northeast Estonia (Fig. 1). This region is affected 11 mainly by mining activities as it contains large deposits of oil shale, the mineral used for power generation in Esto- 12 nia. The study was conducted in 22 semi-natural meadows that differ in vegetation characteristics, size and shape, 13 and in their surrounding landscape composition and configuration. The areas of the selected meadows ranged from 14 0.10 to 3.83 ha. 15 16 The study region is characterised by a patchy landscape structure with a variety of land cover types, predomi- 17 nantly forest (195,245 ha, approximately 58.0% of the territory in the region), arable land (41,671 ha, approximately 18 12.4%), brushwood (21,701 ha, approximately 6.5%) and meadows (19,031 ha, approximately 5.7%), and smaller 19 proportions of human settlements, mires, green areas, etc. The forest cover in the region is mainly composed of 20 managed mixed forest; birches, pines and spruces are among the dominant trees. The total area of our study region is 21 336,400 ha, which represents about 7.4% of the total area of Estonia (the territory of Estonia occupies 45,000 km2, 22 23 Peterson and Aunap 1998). 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 Fig. 1 Map of the study region, Ida-Virumaa County, with the location of the 22 study sites 51

52 53 54 Bumblebee survey 55 56 The fieldwork was carried out during the summers of 2008 and 2009. Every year, we sampled each meadow twice: 57 first during June, and second between the end of July and the beginning of August. The recordings were made via 58 systematic walking surveys. Bumblebee counts were conducted during the warmer time of the day, between 10:00 h 59 60 and 16:00 h, under suitable weather conditions, in which the temperature was above 18 ºC and the wind speed was 61 62 63 64 65

86 1 4 2 3 4 less than five on the Beaufort scale. Our fieldwork method was based on Goulson et al. (2006). Each visit was done 5 during a period of about 45 min to one hour, or until the observer was satisfied that all the species on the site were 6 recorded. The species richness and abundance of bumblebees was determined by counting individuals and identify- 7 ing species by sight, mainly on the wing or when they were standing on the flowers. When the observer could not 8 9 identify the bumblebee species on site, some individuals were caught with an insect net for later identification in the 10 laboratory. The nomenclature used was based on Fauna Europaea (2004). 11 12 Variables at the patch scale 13 14 At each study site, we recorded variables that described the structure of the vegetation. Specifically, we registered 15 16 the number of species of flowering plants, the percent cover of flowering plants and the grass height at the end of 17 each visit. The percent cover of flowering plants was determined through a visual estimation of the overall coverage 18 and was performed by the same person at all study sites. We used a direct method to measure grass height (Stewart 19 et al. 2001). The “direct measurement method” consists of placing a hand lightly on the vegetation at the level below 20 which about 80% of the vegetation is estimated by eye to be growing, ignoring occasional tall stalks, and then read- 21 ing the height with the help of a rule (Stewart et al. 2001). In the statistical analyses, we used the total number of 22 23 species of flowering plants recorded in the first year, and the arithmetic mean of the four observations (one per visit) 24 of percent cover of flowering plants and grass height. The cover of flowering plants was measured in percentages 25 with an accuracy of 10%. 26 In addition, five indices were calculated at all study sites with Fragstats (Version 3.3): area (AREA), perimeter 27 (PERIM), shape index (SHAPE), fractal dimension index (FRAC) and edge density (ED). SHAPE characterises the 28 complexity of a patch shape compared to a standard shape. In this study, the raster version of Fragstats was used, 29 30 which evaluates patch shape with a square standard; this means that shape index is minimum for square patches and 31 increases as patches become increasingly non-square (or more irregular) (McGarigal and Marks 1995). Shape index 32 is commonly applicable and has been widely used in landscape ecological research (Forman and Godron 1986). 33 FRAC is another measure of shape complexity (McGarigal and Marks 1995), but it is calculated based on the patch 34 size. Fractal is defined as a geometric form that displays structure at all spatial scales (McGarigal and Marks 1995). 35 To calculate the fractal dimension of natural planar shapes, he proposed a perimeter-area method. This method 36 37 quantifies the degree of complexity of the planar shapes (McGarigal and Marks 1995). On the other hand, ED (or 38 alternatively Perimeter Area Ratio – PAR) at the patch level is a function of the patch perimeter and it takes into 39 account the shape and the complexity of the patch (McGarigal and Marks 1995). 40 41 Variables at the landscape scale 42 43 44 To analyse the composition of the landscape, the proportion of the main land cover types in the area surrounding 45 each study site was calculated using ArcGIS 9.3. We used a digital Estonian Basic Map provided by the Estonian 46 Land Board at the scale of 1:10,000. The original maps included more than 30 different land cover types that were 47 organised into 11 categories: meadows, forests, brushwoods (bushes, woody seedlings and young trees), mires, ar- 48 able land, abandoned peatland, bodies of fresh water, sea, green areas, human settlements (including residential ar- 49 eas, private areas, buildings, cattle sheds, roads, ruins and green houses) and others. After calculating the percentage 50 51 of each land cover type at four spatial scales (i.e., 250 m, 500 m, 1000 m and 2000 m radius), we chose only the 52 most relevant land cover types, those that had the strongest presence in the study region (i.e., the land cover types 53 than occupy more than 5% of the territory): forest, arable land, brushwood, meadows and human settlements. 54 We used five Fragstats indices to describe the configuration of the landscape (Fragtats, version 3.3) (Appendix 55 1): patch richness density (PRD), interspersion and juxtaposition index (IJI), edge density (ED), Shannon’s diversity 56 index (SHDI) and mean patch area of forest (AREA_MN). These indices were chosen because they describe impor- 57 58 tant aspects of the landscape structure and may influence bumblebee species richness and abundance. PRD was 59 used to standardize patch richness to a per area basis (McGarigal and Marks 1995). We used IJI to measure the ex- 60 tent to which patch types are interspersed (not necessarily dispersed); higher values are given to landscapes in which 61 62 63 64 65

87 1 5 2 3 4 the patch types are well interspersed (or equally adjacent to each other), whereas lower values are given to land- 5 scapes in which the patch types are poorly interspersed (or the distribution of patch type adjacencies is dispropor- 6 tionate) (McGarigal and Marks 1995). Interspersion can be defined as “the degree to which a given patch or land- 7 scape type is scattered rather than aggregated or crumpled”, and juxtaposition is the “adjacency of different patch or 8 9 landscape types” (Freemark et al. 2002). ED_LAND equals the length of all borders between different patch types 10 (classes) in a reference area divided by the total area of the reference unit; in contrast to patch density, edge density 11 takes the shape and the complexity of the patches into account (Eiden et al. 2000). Edge density measures the com- 12 plexity of the shapes of patches and, similar to patch density is an expression of the spatial heterogeneity of a land- 13 scape mosaic (Eiden et al. 2000). Additionally, SHDI was used to measure the diversity of the landscape based on 14 two components: richness, defined as the number of different patch types, and evenness in the distribution of areas 15 16 among patch types (Eiden et al. 2000). Finally, we used AREA_MN to describe the distribution and configuration of 17 patches of the most predominant land cover type in the study region: forest. AREA_MN equals the sum, across all 18 patches of the corresponding patch type (here, forest) of the area of the patches, divided by the total number of 19 patches of the same type (McGarigal et al. 2002). We used “mean patch size” because it gives information about the 20 size of the patches and the number of patches at the same time. 21 22 23 Statistical analyses 24 25 In the statistical analyses, we used the total bumblebee abundance calculated as the total number of individuals 26 found in 2008 and 2009, and the total bumblebee species richness calculated as the total number of species found 27 during the two years at each study site. Before performing the analyses, the logarithm and square root transforma- 28 tions were applied to the bumblebee variables (bumblebee species richness and abundance) and landscape parame- 29 30 ters, respectively, to normalise the data. As the total bumblebee species richness and abundance were strongly corre- 31 lated, we applied the rarefaction to adjust the species richness at different sites to the common number of individu- 32 als. 33 The statistical analyses were developed in various steps. First, we conducted Spearman rank order correlations to 34 analyse the relationships between the total species richness and abundance of bumblebees, and the patch-scale and 35 landscape-scale factors (for a description of the variables, see Appendix 1). When the correlation coefficient (r ) was 36 s 37 between 0.0 and ±0.3, the correlation was considered weak; when rs was between ±0.3 and ±0.6, the correlation was 38 medium; and when rs was between ±0.6 and ±1, the correlation was strong; in all cases, the correlation was statisti- 39 cally significant if the p value was less than 0.05. Second, to simultaneously examine the connectivity patterns of the 40 patch and landscape characteristics, and the overall bumblebee species richness and abundance the Partial Least 41 Squares (PLS) analysis was applied. PLS is the multivariate statistical technique particularly well suited for situa- 42 tions where multicollinearity exists in the dataset and the number of variables is high compared to the number of 43 44 observations (about PLS in ecological studies, see Carrascal et al. 2009, for example). In this study, the connectivity 45 patterns (called also latent factors) of two centered and normalized bumblebee variables stored in matrix Y (i.e., 46 adjusted total bumblebee species richness and total number of individuals), and 50 centered and normalized patch 47 and landscape characteristics (i.e., 10 patch-scale variables and 40 landscape-scale variables) stored in matrix X 48 were evaluated by singular value composition of the form Y’X=USV’ (apostrophe denotes the matrix transposition). 49 Matrixes U and V are the matrixes of the left and right singular vectors (representing the bumblebee richness and 50 51 abundance profiles and landscape profiles, respectively), best characterizing the correlation between X and Y; ma- 52 trix S contains the singular values measuring the quality of latent factors (for a detailed description of PLS, see 53 Krishnan et al. 2011). The percent of variation accounted for by partial least squares factor i was evaluated as the 2 2 54 ratio of the sums of squares of latent variables and initial variables: R Xi=SS(XVi)/SS(X) and R Yi=SS(YUi)/SS(Y) 55 for landscape characteristics and bumblebee richness and abundance, respectively; Vi and Ui denote the ith column 56 of the matrix. To test the statistical significance of latent factors the permutation test with 10,000 permutation sam- 57 58 ples was applied (to correct for the axis rotations and reflections the Procrustes rotation was used). This permutation 59 also served to assess the singular vectors, giving a threshold to decide which variables were contributing the most of 60 the latent factor. 61 62 63 64 65

88 1 6 2 3 4 Finally, we performed the stepwise forward-selection multiple regression analysis to determine the combinations 5 of the most important predictors for total bumblebee species richness and abundance. For each landscape variable, 6 the spatial scale with the strongest relationship was selected. In multiple regression analyses the significance level 7 0.15 was the limit to enter the argument into the model. 8 9 We used STATISTICA 9 software to perform the correlation analyses and multiple regression analysis. The PLS 10 analysis was performed with SAS 9.1 software. 11 12 13 Results 14 15 16 Bumblebee species richness and abundance 17 18 We identified 19 species of bumblebees and 5 species of cuckoo bumblebees (gen. Bombus) in the study area (Ap- 19 pendix 2). They represent approximately 83% of the total bumblebee species found in Estonia. Currently in the 20 country there are 29 species of bumblebees, including 7 species of cuckoo bumblebees. An average of 10.7 species 21 and 27.1 individuals of bumblebees were found per study site. The total number of individuals recorded was 597, 22 23 including 150 males, 84 queens and 363 workers. The most abundant species were B. pascuorum, B. lucorum and B. 24 ruderarius with 140, 70 and 58 individuals in total, respectively. In contrast, B. muscorum and B. distinguendus 25 were the species with the lowest abundance. 26 As it was mentioned before, total bumblebee species richness was strongly positively correlated with total bum- 27 blebee abundance (rs = 0.94, p < 0.001); however, after rarefaction was applied to adjust bumblebee species richness 28 to the common number of individuals, this relationship was weak and not significant (r = 0.27, p = 0.233). 29 s 30 31 Relations between patch-scale factors and bumblebees 32 33 A total of 133 species of flowering plants were found in our study sites. Flowering plant species richness ranged 34 from 7 to 43 species per study site. We found that bumblebee abundance was strongly positively correlated with 35 flowering plant species richness (r = 0.65, p < 0.001). 36 s 37 Concerning the relations between the spatial characteristics of the meadows and bumblebees, we found that 38 bumblebee species richness was strongly negatively correlated with shape index (SHAPE) (rs = -0.60, p = 0.003) 39 and medium negatively correlated with fractal dimension index (FRAC) (rs = -0.57, p = 0.004). There were not sig- 40 nificant relationships neither between bumblebees and other spatial characteristics, nor between bumblebees and 41 average grass height or average percent cover of flowering plants (p > 0.05). 42 43 44 Relations between landscape-scale factors and bumblebees 45 46 We found that the proportion of human settlements in the areas surrounding our study sites was positively correlated 47 with bumblebee abundance at 250 m and 1000 m (rs = 0.48, p = 0.024; rs = 0.51, p = 0.014, respectively). Addition- 48 ally, bumblebee species richness was positively correlated with the proportion of meadows at the largest spatial 49 scale, i.e., 2000 m (r = 0.51, p = 0.015). Concerning the relations between bumblebees and landscape indices, we 50 s 51 found positive correlations between bumblebee abundance and Shannon’s diversity index (SHDI) at 2000 m and 52 edge density (ED_LAND) at 1000 m (rs = 0.44, p = 0.039; rs = 0.50, p = 0.018, respectively). 53 In contrast, we found that proportion of forest was negatively correlated with bumblebee species richness at the 54 spatial scales of 1000 m and 2000 m (rs = -0.45, p = 0.036; rs = -0.47, p = 0.025, respectively). Also, negative corre- 55 lations were detected between proportion of brushwood and bumblebee species richness at 250 m and 500 m (rs = - 56 0.57, p = 0.005; r = -0.44, p = 0.040, respectively). Mean patch area of forest at the largest spatial scale was also 57 s 58 negatively correlated with bumblebee species richness (rs = -0.51, p = 0.015). 59 60 Connectivity patterns between bumblebees and the local and landscape factors 61 62 63 64 65

89 1 7 2 3 4 5 Two connectivity patterns were identified with Partial Least Squares (PLS) analysis, which together accounted for 6 100% and 31.5% of bumblebee richness and abundance variance, and patch and landscape characteristics variance, 7 respectively (in Fig. 2, the percentages are presented separately for the two connectivity patterns). 8 9 The first connectivity pattern connects mainly the overall number of species and individuals of bumblebees with 10 the patch and landscape characteristics (First singular vector, Fig. 2). According to the permutation test, the overall 11 bumblebee richness and abundance were significantly positively related with the proportion of human settlements, 12 especially at the smallest spatial scale (p < 0.05). In contrast, the proportion of arable land at the scale of 250 m, the 13 proportion of brushwood also at 250 m and mean patch area of forests (AREA_MN) (especially at larger spatial 14 scales) showed negative relations (p < 0.05) with the bumblebee richness and abundance pattern, indicating that the 15 16 larger the values of these variables, the smaller the number of species and individuals of bumblebees. 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 Fig. 2 Results of the Partial Least Squares (PLS) analysis. The dots mark the location of the patch and landscape 44 45 characteristics (X) and the squares with arrows mark the location of the bumblebee species richness (adjusted) and 46 abundance (Y) in relation to the two connectivity patterns. The dotted lines denote the approximate cut-off for statis- 47 tical significance of the right singular vectors (patch and landscape characteristics vectors) as assessed through per- 48 mutation tests (p = 0.05); for clearness only the patch and landscape characteristics with p < 0.1 are shown with the 49 variable name. R250, R500, R1000 and R2000, denote the different spatial scales at which the landscape factors 50 were calculated. (For a description of the variables, see the Appendix 1) 51 52 53 54 The second connectivity pattern (Second singular vector, Fig. 2) reflects the changes in relative bumblebee spe- 55 cies richness (i.e., how heterogeneous or homogeneous are the study sites in relation to the number of individuals). 56 Statistically significant were only the second singular vector values corresponding to the proportion of arable land at 57 the spatial scale of 2000 m, proportion of meadows also at the largest spatial scale, and proportion of human settle- 58 59 ments at 1000 m (p < 0.05). The second singular vector values of proportion of arable land and proportion of mead- 60 ows were positive, which mean that the relative species richness of bumblebees may be higher (compared with the 61 62 63 64 65

90 1 8 2 3 4 number of individuals) if the proportion of arable land and meadows is high (particularly at the largest spatial scale). 5 The negative second singular vector values of proportion of human settlements indicate that the relative species 6 richness may be lower (compared with the number of individuals) if the proportion of human settlement is high (es- 7 pecially at the smallest spatial scale). 8 9 10 Models to predict bumblebee species richness and abundance 11 12 The regression models based on the patch and landscape factors tested here explained 83% and 73% of the variation 13 in total bumblebee abundance and species richness, respectively (Table 1). Both models were highly statistically 14 significant. The model for total bumblebee abundance included four variables: one patch-scale factor and three land- 15 16 scape-scale factors. Species richness of flowering plants was significantly positively related with bumblebee abun- 17 dance and emerged as the most important predictor in our model. In contrast, two landscape variables, proportion of 18 arable land and mean patch area of forest (AREA_MN), were negatively related with bumblebee abundance. 19 Five variables were included in the model for total bumblebee species richness: two patch-scale factors and three 20 landscape-scale factors (Table 1). The most important predictor of bumblebee abundance was shape index 21 (SHAPE). Patch area (AREA) was significantly positively related with bumblebee richness, whereas SHAPE, pro- 22 23 portion of arable land and AREA_MN were all negatively related with the dependant variable. 24 25 26 27 Table 1 Regression models for total bumblebee abundance and total bumblebee species richness (adjusted to the 28 common number of individuals) 29 2 30 Dependent variable R Variable included in the model Regression p value 31 coefficient 32 Total bumblebee 0.83* Species richness of flowering plants 0.44 < 0.001 33 abundance Proportion of arable land at 250 m -0.08 0.008 34 Mean patch area of forest (AREA_MN) at 1000 m -0.12 0.048 35 36 Edge density at landscape level (ED_Land) at 1000 0.09 0.120 37 m 38 Total bumblebee spe- 0.73* Patch area (AREA) 0.18 0.028 39 cies richness (ad- Shape index (SHAPE) -0.88 0.003 40 justed) Proportion of arable land at 2000 m -0.05 0.025 41 Mean patch area of forest (AREA_MN) at 1000 m -0.16 <0.001 42 43 Patch richness density (PRD) at 500 m -0.13 0.103 44 * Significant at p < 0.001 45 46 47 Discussion 48

49 50 Effects of patch-scale factors on bumblebees 51 52 This study shows that the diversity of flowering plants is a very important factor for total bumblebee abundance. 53 Similarly, Rundlöf et al. (2008) found that local abundance of forage resources was significantly positively asso- 54 ciated with bumblebee abundance, but not with bumblebee species richness. Also, they found that higher abundance 55 of flowering plants was associated with higher abundance of bumblebees from large colonies (Rundlöf et al. 2008). 56 57 The number of individuals of bumblebees may depend on the availability of flowering resources, because generally 58 the most common and abundant species tend to be those that have a broad diet and emerge early in the season 59 (Goulson et al. 2005), whereas specialist species (e.g., some long-tongued bumblebees) are less abundant as they 60 61 62 63 64 65

91 1 9 2 3 4 often depend on a small number of plant species (Goulson and Darvill 2004; Goulson et al. 2005; Goulson 2010). In 5 general, our results indicate that enhancing the presence of flowering plant species in semi-natural meadows may 6 increase the abundance of bumblebees. This result is consistent with previous studies, which have suggested that the 7 species richness of flowering plants is an important local factor for bumblebee communities (e.g., Bäckman and 8 9 Tiainen 2002; Mänd et al. 2002; Kells and Goulson 2003; Hatfield and LeBuhn 2007; Rundlöf et al. 2008; Ahrné et 10 al. 2009). 11 Patch area seems to have a positive influence on bumblebee species richness. This makes sense as the larger the 12 area of the habitat, the higher the chances of finding the suitable food resources and nesting sites that bumblebee 13 species require. Also, in patches of smaller size, habitat-specialist plants may have a higher probability of extinction 14 (Quintana-Ascencio and Menges 1996); this may influence some bumblebee species that depend on these types of 15 16 plants. Previous studies on insects have also found significant positive relationships between habitat area and species 17 richness (Steffan-Dewenter and Tscharntke 2000; Krauss et al. 2003; Steffan-Dewenter 2003; Öckinger and Smith 18 2006). 19 Other patch-scale factors, specifically shape index and fractal dimension index, showed negative relationships 20 with bumblebee species richness. Shape index was also one of the main predictors of bumblebee species richness. 21 This index describes the complexity of the patch shape; this means that the more irregular the shape of the habitat, 22 23 the lower may be the number of species in that habitat. The importance of patch shape on organisms can be de- 24 scribed using the “interior-to-edge ratio” (Forman and Godron 1986): a circular or square patch consists mostly of 25 an interior area with a surrounding band of edge. A square patch has a higher “interior-to-edge ratio” compared to a 26 patch with a more complex or irregular shape (with the same area), as the latest has proportionally less interior area. 27 Forman and Godron (1986) suggested that patches with higher “interior-to-edge ratio” may have higher species di- 28 versity, less probability of presence of barriers within the patch, and more foraging efficiency for animals inside the 29 30 patch. However, the effect of patch shape on the foraging efficiency has not been well studied and further research is 31 needed. The fractal dimension index is also related with the shape of the patch; it is another measure of shape com- 32 plexity, but it is calculated based on the patch size. 33 34 Effects of landscape composition on bumblebees 35 36 37 We found that bumblebee abundance was positively influenced by the proportion of human settlements at various 38 spatial scales. Overall bumblebee richness and abundance was also positively related with the proportion of human 39 settlements (in the first connectivity pattern, in PLS), particularly at the smallest spatial scale. This may be explained 40 by the presence of gardens in residential areas, which may support a high diversity of flowering plants and thus pro- 41 vide suitable nesting sites, shelter and alternative foraging resources for bumblebees. This has been found also in 42 previous studies in the case of bees in general (Cussans et al. 2010), and particularly bumblebees (Goulson et al. 43 44 2002; McFrederick and LeBuhn 2006; Goulson et al. 2010). Some abundant bumblebee species, such as B. ruder- 45 arius, seem to prefer plant communities close to human settlements (Söderman 1999). Generally, people like to have 46 plants with blossoms in their gardens, so the percentage of nectar-rich flowers might be high in human-inhabited 47 areas. Gardens seem to support extraordinarily high densities of nests for bumblebees (Osborne et al. 2008; Goulson 48 2010). 49 In contrast, Ahrné et al. (2009) found that the proportion of urban areas had a negative effect on bumblebee rich- 50 51 ness, as the increased presence of urban structures such as roads and buildings decreases the proportion of suitable 52 habitat patches for bumblebees, such as field boundaries and pastures. However, most of our study sites were lo- 53 cated relatively far from large towns, which mean that the density of roads, especially main roads, is very low, and 54 the presence of buildings and houses is not very evident. Also, the roadsides and field boundaries in Estonia are 55 commonly covered by lush herbaceous flora (Mänd et al. 2002), which may favour bumblebees. Our measurements 56 of the proportion of human settlements included also the presence of abandoned buildings (ruins) or gardens with 57 58 ruderal plants; these areas are very common in Estonia and may offer places with a high diversity of flowering 59 plants. In addition, Winfree et al. (2007) suggested that bee species richness may be higher when the proportion of 60 natural habitats in the landscape is high, even though the level of human disturbance is intermediate; that is, the 61 62 63 64 65

92 1 10 2 3 4 negative effects of human disturbance may occur only when the proportion of natural land cover is very low. Our 5 study region is cover by high proportions of forests, brushwood and meadows; this means that the presence of these 6 land cover types may mask the effect of human settlements on bumblebees. 7 Our results show that bumblebee species richness may increase with the presence of meadows in the landscape at 8 9 the largest spatial scale. Similarly, Hatfield and LeBuhn (2007) found that the most consistent positive influence on 10 species richness of bumblebees was the proportion of meadows in the surrounding landscape, at a 2-km buffer from 11 the edge of the focal habitat. In addition, Le Féon et al. (2010) found that the species richness, abundance and diver- 12 sity of bees were negatively affected by agricultural intensification, whereas bee species richness was positively 13 affected by the amount of semi-natural habitats in the landscape. On the other hand, it has been found that, in gen- 14 eral, bumblebees have large foraging ranges (Steffan-Dewenter et al. 2002; Westphal et al. 2006; Hatfield and Le- 15 16 Buhn 2007); some species are known to fly more than 2000 m (e.g., B. pascuorum and B. terrestris) (Chapman et al. 17 2003; Zurbuchen et al. 2010). Dispersal abilities of bumblebees allow them to retrieve floral resources in adjacent 18 meadows; increasing the probability of finding the flowering plants that some species require (Hatfield and LeBuhn 19 2007). 20 Bumblebee species richness seems to be negatively influenced by the presence of forest in the surrounding land- 21 scape at the largest spatial scales. This may happen because some bumblebee species may not be able to find their 22 23 suitable nesting sites in the forest and also, they may have different preferences in terms of the landscape context. 24 Goulson (2010) suggested that the sites chosen for nesting vary between species, depending on the habitat type and 25 the place where this habitat is located. Also, overall bumblebee species richness appears to be negatively influenced 26 by the proportion of brushwood. Brushwood areas are commonly dominated by willows that often grow in wetlands 27 and along the forest edges (Sepp et al. 2004). These habitats are rich in blooming flowers and are important for 28 bumblebees in early spring (i.e., April and May), particularly for some species that emerge early in the season. Many 29 30 patches of brushwood have grown in areas that were former meadows; the soil in these areas is rich in calcium and 31 can therefore support a great amount of flowering plant species. However, areas dominated by willows may also 32 represent an ecological trap for bumblebees: early emerging species might tend to build their nests near the forest, 33 where later in the season food would become scarce and these areas would no longer be able to provide enough for- 34 age resources for bumblebees. 35 On the other hand, bumblebee species richness and abundance were negatively associated with the proportion of 36 37 arable land. The negative effect of arable land on bumblebees may be explained by the openness of the landscape in 38 those areas, which could make the bumblebees more vulnerable to wind and other climatic factors, as there are 39 fewer places that may offer shelter and protection. Also, foraging resources are sometimes scarce in agricultural land 40 and this may result in the decline of bumblebees (Goulson et al. 2005), while the presence of semi-natural grasslands 41 in the landscape context may increase the presence of bumblebees (Öckinger and Smith 2007). Grasslands are more 42 likely to contain a higher availability of nests for bumblebees than the surrounding cultivated land (Öckinger and 43 44 Smith 2007). Similarly, Le Féon et al. (2010) found in a recent study that bee species richness and abundance were 45 negatively affected by agricultural intensification. Overall pollinator diversity may be enhanced by the presence of 46 semi-natural habitats in the landscape context (Billeter et al. 2008; Jauker et al. 2009; Le Féon et al. 2010). 47 48 Effects of landscape configuration on bumblebees 49 50 51 Bumblebee abundance seems to be positively influenced by edge density at landscape level. This positive relation 52 may occur because there is a strong dependency of bumblebee abundance on the availability of flowering plants (as 53 it was mentioned above). Kumar et al. (2009) explained that habitat edges contain a great abundance and diversity of 54 floral resources, making them suitable places for flower visitors. The presence of edges and other compensating ar- 55 eas nearby the main habitat is very important to bumblebees’ survival, as they may find complementary food re- 56 sources and nesting places there. Furthermore, bumblebee queens are more frequently observed along forest bounda- 57 58 ries and field boundaries (Svensson et al. 2000). Similarly, Sepp et al. (2004) found that the distribution of bumble- 59 bees was positively related with the length of ecotones between cultivated land and different types of forest. A study 60 on bumblebees in Estonia suggested that edges are particularly important in April and May, when bumblebee queens 61 62 63 64 65

93 1 11 2 3 4 mostly forage the flowering willows that are commonly found in the forest edges (Sepp et al. 2004). Positive effects 5 of linear elements, such as edges, on bumblebees have been found before (Osborne et al. 2008). 6 We found that Shannon’s diversity index seems to be an important landscape metric for bumblebee abundance at 7 the largest spatial scale. This index indicates the complexity of the surrounding landscape matrix, and increases as 8 9 the number of different patch types increases and the distribution of patch types becomes more equitable (Eiden et 10 al. 2000). This means that our study sites are surrounded by different patch types that might be suitable habitat 11 fragments for bumblebees, increasing the availability of food resources in the landscape and thus, their likelihood of 12 survival. Other authors have found similar positive relationships between insects and the diversity of the landscape 13 matrix (e.g., Steffan-Dewenter 2003; Kivinen et al. 2006). Kivinen et al. (2006) argued that in boreal agricultural 14 landscapes, the presence of patches of semi-natural grasslands and other non-crop biotopes in adjacent open areas 15 16 may have a positive effect on the species richness of some insects (such as butterflies), as movement of species be- 17 tween different habitat types can increase overall species richness in the landscape context. 18 In contrast, overall bumblebee species richness and abundance appear to be negatively influenced by mean patch 19 area of forest. A possible explanation for this negative association may be that a high number of patches of forest 20 could be seen as potential obstacles in the landscape by some species of foraging bumblebees (Kreyer et al. 2004; 21 Goulson et al. 2010), particularly for those species that have large foraging distances, such as B. lapidarius and B. 22 23 terrestris (Walther-Hellwig and Frankl 2000a; Walther-Hellwig and Frankl 2000b). Similarly, Winfree et al. (2007) 24 found that bee species richness and abundance were negatively associated with the extent of forest cover, suggesting 25 that the number of bees decreased as forest cover increased in the surrounding landscape. 26 27 28 Conclusions and implications for conservation 29 30 31 We found that not only the availability of food resources at patch level, but also the quality and diversity of the sur- 32 rounding landscape, are important factors affecting bumblebee species richness and abundance. 33 The results from this study have important implications for the conservation of bumblebees and for the develop- 34 ment of agri-environmental measures in patchy forested landscapes. First, the presence of a high diversity and abun- 35 dance of flowering plants may benefit bumblebee abundance in semi-natural meadows, but considering only local 36 37 factors may not be sufficient. Second, the existence of edges in patchy forested landscapes may support bumblebees, 38 as these are considered compensating areas that may offer shelter, food and protection for them. Third, the presence 39 of human settlements in the landscape matrix may favour bumblebees, particularly when these areas include gardens 40 and other places with a high diversity of flowering plants, and when the percentage of natural and semi-natural habi- 41 tats in the landscape is high, particularly meadows. Finally, bumblebees benefit from a rich and diverse landscape 42 matrix with an important presence of patches of natural and semi-natural habitats. 43 44 Policies supporting agri-environmental measures should be improved, because if financial resources target only 45 one farmer or only changes at the local level, these measures are not likely to be very effective for biodiversity con- 46 servation. Changes at the level of one farm are not sufficient to support the entire system that also incorporates the 47 surrounding landscape. To maintain biodiversity, heterogeneous landscapes including patches of semi-natural habi- 48 tats need to be preserved. In conclusion, we should consider not only variables at the local level but also the land- 49 scape context around targeted areas at large spatial scales when designing conservation strategies for bumblebees 50 51 and agri-environmental measures. 52 53 54 Acknowledgements This research was funded by targeted financing from the Estonian Ministry of Education and 55 Research (SF1090050s07, SF0170057s09), Estonian Science Foundation Grant No. 7391 and by an applied research 56 project of the Estonian Ministry of Agriculture (T8014PKPK). We would like to thank Prof. Jaan Liira for his useful 57 58 comments that greatly improved the manuscript. 59 60 61 62 63 64 65

94 1 12 2 3 4 5 6 Appendix 1 7 8 9 Local and landscape factors considered in this study. 10 11 Variable Description Unit 12 13 Bumblebees 14 15 SRBumb Bumblebee species richness; number of species - 16 17 NIBumb Bumblebee abundance; number of individuals - 18 19 Vegetation structure at local level 20 21 SRFlowPlants Species richness of flowering plants; number of species. - 22 23 AvCoverFP Average percent cover of flowering plants. Percentage % 24 of the meadow that is covered by flowering plants. 25 AvGrassH Average grass height. cm 26 27 Spatial characteristics at local 28 level (Fragstats indices) 29 30 AREA Patch area; size of the patch. ha 31 32 PERIM Perimeter of the patch. m 33 SHAPE Shape index; SHAPE equals patch perimeter (m) di- - 34 35 vided by the square root of patch area (m2), adjusted by 36 a constant to adjust for a circular standard (vector) or 37 square standard (raster)a. 38 FRAC Fractal dimension index; FRAC equals 2 times the loga- - 39 rithm of patch perimeter (m) divided by the logarithm of 40 patch area (m2)a. 41 42 ED Edge density; sum of the length (m) of the edge segment m/ha a 43 of the patch per unit area . 44 Landscape composition 45 46 Pforest Proportion of patches that are forest. % 47 48 Pmeadows Proportion of patches that are meadows. % 49 50 PArLand Proportion of patches that are arable land. % 51 52 PHumSet Proportion of patches that are human settlements; in- % 53 cluding residential areas, buildings, cattle sheds, roads, 54 ruins (or buildings’ remains) and green houses. 55 Pbrushw Proportion of patches that are brushwood; including % 56 57 bushes, woody seedlings and young trees. 58 Landscape configuration 59 (Fragstats indices) 60 61 62 63 64 65

95 1 13 2 3 4 PRD Patch richness density; PRD equals the number of patch No/100 ha 5 types per 100 haa. 6 IJI Interspersion and juxtaposition index; measure of distri- % 7 a bution of patch adjacencies . 8 9 ED_LAND Edge density at landscape level; total length of all edge m/ha a 10 segments per unit area of landscape . 11 SHDI Shannon’s diversity index; SHDI equals minus the sum, - 12 across all patch types, of the proportional abundance of 13 each patch type multiplied by that proportiona. 14 AREA_MN Mean patch area of forests; AREA_MN equals the sum, ha 15 16 across all patches of the corresponding patch type, in our 17 case forest, of the area of the patches, divided by the 18 total number of patches of the same typea. 19 a Source: McGarigal et al. (2002) 20 21 22 23 Appendix 2 24 25 List of bumblebee species observed. 26 27 Bombus cryptarum (Fabr.) 28 29 Bombus distinguendus Morawitz 30 Bombus hortorum (L.) 31 Bombus hypnorum (L.) 32 Bombus jonellus (Kirby) 33 Bombus lapidarius (L.) 34 Bombus lucorum (L.) 35 36 Bombus muscorum (L.) 37 Bombus pascuorum (Scopoli) 38 Bombus pratorum (L.) 39 Bombus ruderarius (Müller) 40 Bombus semenoviellus Skorikov 41 Bombus schrencki Morawitz 42 43 Bombus soroeensis ssp. Soroeensis (Fabr.) 44 Bombus soroeensis ssp. Proteus (Fabr.) 45 Bombus soroeensis ssp. soroeensis x proteus (Fabr.) 46 Bombus sylvarum (L.) 47 Bombus terrestris (L.) 48 Bombus veteranus (Fabr.) 49 50 Psithyrus bohemicus Seidl. 51 Psithyrus campestris (Panzer) 52 Psithyrus norvegicus (Sparre-Schneider) 53 Psithyrus rupestris (Fabr.) 54 Psithyrus sylvestris (Lep.) 55 56 57 58 59 60 61 62 63 64 65

96 1 14 2 3 4 References 5 6 Ahrné K, Bengtsson J, Elmqvist T (2009) Bumble bees (bombus spp) along a gradient of increasing urbanization. 7 PLoS One 4:e5574 8 9 Bäckman JP, Tiainen J (2002) Habitat quality of filed margins in a Finnish farmland area for bumblebees (Hymen- 10 optera: Bombus and Psithyrus). Agr Ecosyst Environ 89:53–68 11 Carrascal LM, Galván I, Gordo O (2009) Partial least squares regression as an alternative to current regression me- 12 thods used in ecology. Oikos 118:681–690 13 Carvell C (2002) Habitat use and conservation of bumblebees (Bombus spp) under different grassland management 14 regimes. Biol Conserv 103:33–49 15 16 Chapman RE, Wang J, Bourke AFG (2003) Genetic analysis of spatial foraging patterns and resource sharing in 17 bumble bee pollinators. Mol Ecol 12: 2801–2808 18 Cozzi G, Müller CB, Krauss J (2008) How do local habitat management and landscape structure at different spatial 19 scales affect fritillary butterfly distribution on fragmented wetlands? Landscape Ecol 23:269–283 20 Crist TO, Pradhan-Devare SV, Summerville KS (2006) Spatial variation in insect community and species responses 21 to habitat loss and plant community composition. Oecologia 147:510–521 22 23 Cussans J, Goulson D, Sanderson R, Goffe L, Darvill B, Osborne JL (2010) Two bee-pollinated plant species show 24 higher seed production when grown in gardens compared to arable farmland. PLoS One 5:e11753 25 Drakare S, Lennon JJ, Hillebrand H (2006) The imprint of the geographical, evolutionary and ecological context on 26 species-area relationships. Ecol Lett 9:215–227 27 Eiden G, Kayadjanian M, Vidal C (2000) Capturing landscape structures: Tools. The European Commission. 28 http://ec.europa.eu/agriculture/publi/landscape/ch1.htm. Accessed 10 October 2011 29 30 Fauna Europaea (2011) Fauna Europaea version 2.4. http://www.faunaeur.org. Accessed 15 June 2011 31 Forman RTT, Godron M (1986) Landscape ecology. John Wiley & Sons, New York 32 Freemark K, Bert D and Villard MA (2002) Patch-, landscape-, and regional-scale effects on biota. In: Gutzwiller 33 KJ (ed) Applying landscape ecology in biological conservation. Springer, New York, pp 58–83 34 Goulson D (2010) Bumblebees: behaviour, ecology and conservation. Oxford University Press, Oxford 35 Goulson D, Darvill B (2004) Niche overlap and diet breadth in bumblebees; are rare species more specialized in 36 37 their choice of flowers? Apidologie 35: 55–64 38 Goulson D, Hanley ME, Darvill B, Ellis JS (2006) Biotope associations and the decline of bumblebees (Bombus 39 spp). J Insect Conserv 10:95–103 40 Goulson D, Hanley ME, Darvill B, Ellis JS, Knight ME (2005) Causes of rarity in bumblebees. Biol Conserv 122:1– 41 8 42 Goulson D, Hughes WHO, Derwent LC, Stout JC (2002) Colony growth of the bumblebee, Bombus terrestris, in 43 44 improved and conventional agricultural and suburban habitats. Oecologia 130:267–273 45 Goulson D, Lepais O, O’Connor S, Osborne JL, Sanderson LA, Cussans J, Goffe L, Darvill B (2010) Effects of land 46 use at a landscape scale on bumblebee nest density and survival. J Appl Ecol 47:1207–1215 47 Goulson D, Rayner P, Dawson B, Darvill B (2011) Translating research into action; bumblebee conservation as a 48 case study. J Appl Ecol 48:3–8 49 Gutzwiller KJ (2002) Spatial factors affecting organism occurrence, movement, and conservation: introduction to 50 51 section II. In: Gutzwiller KJ (ed) Applying landscape ecology in biological conservation, Springer, New 52 York, pp 55–57 53 Hagen M, Wikelski M, Kissling WD (2011) Space use of bumblebees (Bombus spp) revealed by radio tracking. 54 PLoS ONE 6(5):e19997 55 Hatfield RG, LeBuhn G (2007) Patch and landscape factors shape community assemblage of bumble bees, Bombus 56 spp (Hymenoptera: Apidae), in montane meadows. Biol Conserv 139:150–158 57 58 Holzschuh A, Steffan-Dewenter I, Tscharntke T (2008) Agricultural landscapes with organic crops support higher 59 pollinator diversity. Oikos 117:354–361 60 61 62 63 64 65

97 1 15 2 3 4 Jauker F, Diekötter T, Schwarzbach F, Wolters V (2009) Pollinator dispersal in an agricultural matrix: opposing 5 responses of wild bees and hoverflies to landscape structure and distance from main habitat. Landscape 6 Ecol 24:547–555 7 Kells AR, Goulson D (2003) Preferred nesting sites of bumblebee queens (Hymenoptera: Apidae) in agroecosystems 8 9 in the UK. Biol Conserv 109,165–174 10 Kivinen S, Luoto M, Kuussaari M, Helenius J (2006) Multi-species richness of boreal agricultural landscapes: ef- 11 fects of climate, biotope, soil and geographical location. J Biogeogr 33:862–875 12 Kleijn D, Baquero RA, Clough Y et al (2006) Mixed biodiversity benefits of agri-environment schemes in five Eu- 13 ropean countries. Ecology Letters 9: 243–254 14 Knight ME, Osborne JL, Sanderson RA, Hale RJ, Martin AP, Goulson D (2009) Bumblebee nest density and the 15 16 scale of available forage in arable landscapes. Insect Conserv and Divers 2:116–124 17 Krauss J, Steffan-Dewenter I, Tscharntke T (2003) How does landscape context contribute to effects of habitat 18 fragmentation on diversity and population density of butterflies? J Biogeogr 30: 889–900 19 Krewenka KM, Holzschuh A, Tscharntke T, Dormann CF (2011) Landscape elements as potential barriers and cor- 20 ridors for bees, wasps and parasitoids. Biol Conserv 144:1816–1825 21 Kreyer D, Oed A, Walther-Hellwig K, Frankl R (2004) Are forests potential landscape barriers for foraging bumble- 22 23 bees? Landscape scale experiments with Bombus terrestris agg and Bombus pascuorum (Hymenoptera, 24 Apidae). Biol Conserv 116(1): 111–118 25 Krishnan A, Williams LJ, McIntosh AR, Abdi H (2011) Partial Least Squares (PLS) methods for neuroimaging: A 26 tutorial and review. NeuroImage 56:455–475 27 Kumar S, Simonson SE, Stohlgren TJ (2009) Effects of spatial heterogeneity on butterfly species richness in Rocky 28 Mountain National Park, CO, USA. Biodivers Conserv 18:739–763 29 30 Le Féon V, Schermann-Legionneta A, Delettrea Y et al (2010) Intensification of agriculture, landscape composition 31 and wild bee communities: A large scale study in four European countries. Agr Ecosyst Environ 137:143– 32 150 33 Mänd M, Mänd R, Williams IH (2002) Bumblebees in the agricultural landscape of Estonia. Agr Ecosyst Environ 34 89:69–76 35 Manning AD, Lindenmayer DB, Nix HA (2004) Continua and Umwelt: alternative ways of viewing landscapes. 36 37 Oikos 104:621–628 38 Mazerolle MJ, Villard MA (1999) Patch characteristics and landscape context as predictors of species presence and 39 abundance: a review. Ecoscience 6:117–124 40 McFrederick QS, LeBuhn G (2006) Are urban parks refuges for bumble bees Bombus spp (Hymenoptera: Apidae)? 41 Biol Conserv 129:372–382 42 McGarigal K, Cushman SA, Neel MC, Ene E (2002) FRAGSTATS, Spatial Pattern Analysis Program for Categori- 43 44 cal Maps. University of Massachusetts, Amherst 45 McGarigal K, Marks BJ (1995) FRAGSTATS: spatial pattern analysis program for quantifying landscape structure. 46 Gen. Tech. Rep. PNW-GTR-351 Portland, OR: US Department of Agriculture, Forest Service, Pacific 47 Northwest Research Station 48 Öckinger E, Smith HG (2006) Landscape composition and habitat area affects butterfly species richness in semi- 49 natural grasslands. Oecologia 149:526–534 50 51 Öckinger E, Smith HG (2007) Semi-natural grasslands as population sources for pollinating insects in agricultural 52 landscapes. J Appl Ecol 44:50–59 53 Osborne JL, Martin AP, Shortall CR, Todd AD, Goulson D, Knight ME, Hale RJ, Sanderson RA (2008) Quantify- 54 ing and comparing bumblebee nest densities in gardens and countryside habitats. J Appl Ecol 45: 784–792 55 Palang H, Mander Ü, Luud A (1998) Landscape diversity changes in Estonia. Landscape Urban Plan 41:163–169 56 Peterson U, Aunap R (1998) Changes in agricultural land use in Estonia in the 1990s detected with multitemporal 57 58 Landsat MSS imagery. Landscape Urban Plan 41: 193–201 59 Pöyry J, Paukkunen J, Heliölä J, Kuussaari M (2009) Relative contributions of local and regional factors to species 60 richness and total density of butterflies and moths in semi-natural grasslands. Oecologia 160:577–587 61 62 63 64 65

98 1 16 2 3 4 Potts SG, Biesmeijer JC, Kremen C, Neumann P, Schweiger O, Kunin WE (2010) Global pollinator declines: trends, 5 impacts and drivers. Trends Ecol Evol 25:345–353 6 Quintana-Ascencio PF, Menges ES (1996) Inferring Metapopulation Dynamics from Patch-Level Incidence of Flo- 7 rida of Scrub Plants. Conserv Biol 10 (4):1210–1219 8 9 Rundlöf M, Nilsson H, Smith HG (2008) Interacting effects of farming practice and landscape context on bumble 10 bees. Biol Conserv 141:417–426 11 Samways MJ (2005) Insect Diversity Conservation. Cambridge University Press, New York 12 Sepp K, Mikk M, Mänd M, Truu J (2004) Bumblebee communities as an indicator for landscape monitoring in the 13 agri-environmental programme. Landscape Urban Plan 67:173–183 14 Söderman G (1999) Diversity of pollinator communities in Eastern Fennoscandia and Eastern Baltics Results from 15 16 pilot monitoring with yellow traps in 1997 – 1998. Finnish Environment Institute. Edita Ltd, Helsinki 17 Steffan-Dewenter I (2003) Importance of habitat area and landscape context for species richness of bees and wasps 18 in fragmented orchard meadows. Conserv Biol 17:1036–1044 19 Steffan-Dewenter I, Münzenberg U, Bürger C, Thies C, Tscharntke T (2002) Scale-dependent effects of landscape 20 context on three pollinator guilds. Ecology 83:1421–1432 21 Steffan-Dewenter I, Tscharntke T (2000) Butterfly community structure in fragmented habitats. Ecol Lett 3:449–456 22 23 Stewart KEJ, Bourn NAD, Thomas JA (2001) An evaluation of three quick methods commonly used to assess sward 24 height in ecology. J Appl Ecol 38:1148–1154 25 Svensson B, Langerlöf J, Svensson BG (2000) Habitat preferences of nest-seeking bumble bees (Hymenoptera: 26 Apidae) in an agricultural landscape. Agr Ecosyst Environ 77:247–255 27 Walther-Hellwig K, Frankl R (2000a) Foraging habitats and foraging distances of bumblebees, Bombus spp (Hym, 28 apidae), in an agricultural landscape. J Appl Entomol 124: 299–306 29 30 Walther-Hellwig, K, Frankl, R (2000b) Foraging distances of Bombus muscorum, Bombus lapidarius, and Bombus 31 terrestris (Hymenoptera, Apidae) J Insect Behav 13(2): 239–246 32 Westphal C, Steffan-Dewenter I, Tscharntke T (2006) Bumblebees experience landscapes at different spatial scales: 33 possible implications for coexistence. Oecologia 149:289–300 34 Williams PH, Osborne JL (2009) Bumblebee vulnerability and conservation world-wide. Apidologie 40:367–387 35 Winfree R, Griswold T, Kremen C (2007) Effect of human disturbance on bee communities in a forested ecosystem 36 37 Conserv Biol 21: 213–223 38 Xie Z, Williams PH, Tang Y (2008) The effect of grazing on bumblebees in the high rangelands of the eastern Ti- 39 betan Plateau of Sichuan. J Insect Conserv 12:695–703 40 Zurbuchen A, Landert L, Klaiber J, Muller A, Hein S, Dorn S (2010) Maximum foraging ranges in solitary bees: 41 only a few individuals have the capability to cover long foraging distances. Biol Conserv 143:669–676 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

III Diaz-Forero, I., Kuusemets, V., Mänd, M. and Luig, J. 2011

Bumblebees as potential indicators for the evaluation of habitat quality

Sustainable Development and Planning V. WIT Transactions on Ecology and the Environment. WIT Press. Vol 150, 409-417 Sustainable Development and Planning V 409

Bumblebees as potential indicators for the evaluation of habitat quality

I. Diaz-Forero, V. Kuusemets, M. Mänd & J. Luig Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Estonia

Abstract

Habitat fragmentation, decline and degradation are considered important threats to biodiversity and the principal processes that contribute to landscape change. It is fundamental to understand the quality of habitats (and the location of suitable ones) to develop appropriate biodiversity conservation strategies. Insects are considered key indicators of environmental change due to their diversity of habitat characteristics and requirements. Habitat quality may be assessed by its suitability for insects using important ecological differences between generalist and specialist species. Specialist species are more severely affected by the degradation and decrease of suitable habitats than generalists, as they are dependent on specific types of flowering plants or local environments. In our study, we collected data on five species of long-tongued bumblebees (gen. Bombus) including B. pascuorum, B. hortorum, B. ruderarius, B. sylvarum and B. distinguedus. The richness and abundance of long-tongued bumblebee species were recorded in 22 semi-natural meadows in Northeast Estonia. We identified abiotic and biotic factors, at both patch and landscape scale, which significantly impacted total species richness and abundance of long-tongued bumblebees. Overall, we found that besides the availability of food resources in the habitat, it is important to bear in mind the quality and diversity of the surrounding matrix when designing biodiversity conservation strategies. In countries with patchy landscapes, like Estonia, it is important to consider ecological indicators that are strongly associated with both patch and landscape variables. Therefore, bumblebees have the potential to serve as good indicator species for habitat quality. Keywords: long-tongued bumblebees, species richness, flowering plants, landscape structure.

103 410 Sustainable Development and Planning V

1 Introduction

The main processes that contribute to landscape change are habitat fragmentation, decline and degradation; these are also important threats to biodiversity [1]. Fischer and Lindenmayer [1] considered habitat degradation to be the gradual deterioration in quality of an area of habitat for a given species. Many definitions have been given to habitat quality [2]. In our study, we adopted the one presented by Hall et al. [3]; they defined habitat quality as “the ability of the environment to provide conditions appropriate for individual and population persistence”. It is well known and widely accepted in many countries that bumblebees and other important pollinators are declining [4–7]. The intensification of agriculture has led to the reduction of suitable habitats and decrease of food resources for pollinators [8, 9]. It is fundamental to become acquainted with the quality of the remaining habitats and the location of suitable ones in order to develop appropriate biodiversity conservation strategies. To do that, we must know what indicators can be used to evaluate habitat quality. Long-tongued bumblebees are important pollinators of deep perennial flowers. Longer-tongued species have shown increasing vulnerability in the United Kingdom. Goulson et al. [10] found that "the most severely affected species tend to be those with long tongues associated with deep perennial flowers". In contrast, Williams and Osborne [7] found that long tongues and food-plant specialisation were not associated with declines of bumblebee species. However, some studies agreed that further research is required on the ecology of rare species and the role of diet specialisation in bumblebee decline (e.g. [4, 7, 11]). In our study, we collected field data on five species of long-tongued bumblebees: B. pascuorum, B. hortorum, B. ruderarius, B. sylvarum and B. distinguendus. We identified biotic and abiotic factors, at both patch and landscape scale, which may be significant indicators for species richness and abundance of long-tongued bumblebees. In addition, we discuss why bumblebees could be used as potential indicators of habitat quality.

2 Materials and methods

We completed field work during the summers of 2008 and 2009 in Ida-Virumaa, a county in northeast Estonia. This region has a very patchy landscape mosaic with a variety of land cover types, predominantly forest, arable land and meadow. Even though northeastern Estonia has been impacted by mining activities, it is still considered a region that supports a significant number of species and abundance of pollinators, including bumblebees [12]. In both years, we visited 22 semi-natural meadows and sampled each meadow two times. Bumblebee counts took place in June, July and August, and were completed within approximately 45 minutes. We performed systematic walking surveys [13] during the warmer part of the day, between 11:00 h and 16:00 h, when weather conditions were suitable (i.e., temperature was above 18ºC and wind

104 Sustainable Development and Planning V 411 speed was less than 5 as measured by the Beaufort scale). The identification of species and counting of individual bumblebees was done by sight. When the observer could not identify the species, the bumblebee was caught with an insect net for later identification. The nomenclature of the insects follows that used in the Fauna Europaea Web Service [14]. In this study, we used our data on total species richness and abundance of long-tongued (LT) bumblebees (i.e., B. pascuorum, B. hortorum, B. ruderarius, B sylvarum and B. distinguendus). At patch scale, we measured the following variables: species richness of flowering plants, percent cover of flowering plants, patch area, shape, fractal dimension index and edge density. We identified the flowering plants and counted the number of species on site. Percent cover of flowering plants was recorded via a visual estimation of the overall coverage at each study site. We used the total number of species of flowering plants and the arithmetic means of the four observations of percent cover of flowering plants. At landscape scale, we considered the following indices: patch richness density (PRD), interspersion and juxtaposition index (IJI), edge density at landscape level (ED_Land) and Shannon’s diversity index (SHDI). In addition, we calculated the proportion of different land cover types around the study areas (i.e., arable land, meadow, forest and human settlements). Calculations were performed with ArcGIS 9.3 software using the digital Estonian Basic Map provided by the Estonian Land Board at a scale of 1:10,000. All landscape variables were estimated at four spatial scales (250, 500, 1000 and 2000 m radius). Fragstats software (Version 3.3) was used to compute the patch and landscape indices. In this study, Spearman rank order correlations were performed to analyse the relationships between species richness or abundance of long-tongued bumblebees and our variables at patch and landscape scale.

3 Results and discussion

3.1 Long-tongued (LT) bumblebees and their relationships with patch and landscape variables

We found five species of LT bumblebees: B. pascuorum, B. hortorum, B. ruderarius, B. sylvarum and B. distinguendus. Total species richness and abundance of LT bumblebees ranged from 1 to 4 species and from 2 to 29 individuals, respectively. The most abundant species were B. pascuorum and B. ruderarius, whereas B. distinguendus was extremely rare. The results from correlation analyses are presented in Table 1. We found that both species richness and abundance of LT bumblebees had positive relationships with species richness of flowering plants and percent cover of flowering plants. The higher the availability and diversity of food resources in the habitat, the better the bumblebees’ chances of finding the flowering plant species they require. In addition, the dispersal distances of some rare species of bumblebees are very restricted, depending on resources within the habitat or in compensating areas, such as edges. Some rare, long-tongued species have a rather small foraging range (e.g., B. distinguendus) [15]. In general, increasing

105 412 Sustainable Development and Planning V the presence of flowering plant species in semi-natural meadows enhances both the species richness and abundance of bumblebees by providing better quality habitats. This is consistent with previous studies on bumblebees (e.g. [5, 11, 16]).

Table 1: Relationships between long-tongued bumblebees and variables at patch and landscape scale.

Variable Spatial scale Long-tongued bumblebees (m radius) Species richnessa Abundancea Flowering - 0.67*** 0.76*** plant species richness Percent cover - 0.61** 0.58** of flowering plants Proportion of 500 -0.45* -0.10 forest (%) 1000 -0.45* -0.01 Proportion of 1000 0.55** 0.09 meadows (%) 2000 0.58** 0.10 Edge density 500 0.27 0.44* at landscape 1000 0.39 0.42* level (ED_Land) (m/ha) Shannon’s 2000 0.54** 0.36 diversity index (SHDI) a Spearman rank correlation coefficients (rs) are shown * Correlations significant at P < 0.05 ** Correlations significant at P < 0.01 *** Correlations significant at P < 0.001

At landscape scale, we found that species richness of LT bumblebees correlated positively with proportion of meadows at 1000 m and 2000 m. Similarly, Hatfield and LeBuhn [17] found that the most consistent positive influence on species richness and abundance of bumblebees was the proportion of meadows in the surrounding landscape, at a 2-km buffer from the edge of the focal habitat. In general, bumblebees have large foraging ranges [17–19]. Even though some LT bumblebee species have short foraging distances, as mentioned above, other species are known to fly more than 2000 m (e.g., B. pascuorum) [20]. Dispersal abilities of bumblebees allow them to retrieve floral resources in adjacent meadows, increasing the probability of individuals finding flowering plants [17]. In addition, LT bumblebee abundance correlated positively with ED_Land at 500 m and 1000 m. There is a strong dependency of bumblebee abundance on the availability of flowering plants. Kumar et al. [13] explained that habitat

106 Sustainable Development and Planning V 413 edges contain a great abundance and diversity of floral resources, making them suitable places for flower visitors. Also, the presence of edges and other compensating areas is very important to bumblebee survival, especially in patchy landscapes with diverse land cover types. Sepp et al. [21] explained that forest edges are particularly important in April and May, when bumblebee queens mostly forage flowering willows that are commonly found in the forest edges of Estonia. Positive relationships were found between species richness of LT bumblebees and SHDI at 2000 m. This landscape index indicates the level of complexity of the surrounding matrix, as the higher the value of SHDI, the higher the number of patch types and the more equitable the distribution of those patch types across the landscape [22]. Other authors have found similar positive relationships between insects and the diversity of the landscape matrix [23, 24]. Williams and Osborne [7] suggested that the ability of bumblebees to fly long distances from the colony makes them less susceptible to the fragmentation and patchiness of the landscape, as they become more flexible in the utilisation of food resources. The presence of different patch types in the surrounding landscape of their habitats increases the probabilities of finding suitable habitat fragments with the needed flowering plant species; concurrently, this enhances the survival possibility of bumblebee species. In contrast, negative correlations were found between species richness of LT bumblebees and proportion of forest at 500 m and 1000 m. These results suggest that some species of LT bumblebees prefer open areas. In general, LT bumblebee species have specialised diets and are expected to visit a particular type of flowering plants; those flowers are more likely to be found in open areas than in patches of forest. Similarly, Bäckman and Tiainen [16] found that the long- tongued species B. ruderarius prefers open habitats. Additionally, it has been suggested that early-emerging bumblebee species are associated with forests while late-emerging species are associated with grasslands; most late-emerging species are medium or long-tongued bumblebees [11]. Other variables at patch scale (i.e., patch area, shape, fractal dimension index and edge density) and landscape scale (i.e., proportion of arable land, proportion of human settlements, PRD and IJI) do not appear to be important for LT bumblebee species richness and abundance.

3.2 Bumblebees as potential indicators of habitat quality

Ecological indicators can be defined as factors that communicate important information about ecosystems and the impact of human activities on them. Ecosystems are complex and the use of ecological indicators is needed in order to describe them in simpler terms that can be understood and used by scientists and non-scientists alike to make management decisions [25]. Insects are considered key indicators of environmental change due to their diversity of habitat characteristics and requirements. The role of insects as ecological indicators has been tested and studied extensively (e.g. [21, 26]). Bees are a vital element of global biodiversity and an important group of pollinators, as they play a key role in supporting not only crops, but also the diversity of

107 414 Sustainable Development and Planning V natural and semi-natural vegetation [27, 28] and the survival of other organisms [4, 27]. Among bees, bumblebees are considered to be the best-documented group [6]. Bumblebees are known to be sensitive to environmental changes and serve as good indicators of habitat quality [21, 29]. In Estonia, bumblebees are considered significant indicators of habitat and landscape diversity [5], and have been proposed as biodiversity indicators at the landscape level of the agri-environmental programme [21]. As we mentioned above, bumblebees and other pollinators are at risk. Thus, there is a current need for the protection of endangered species as well as the conservation of their habitats. Semi-natural habitats, such as meadows, are areas of important value for bumblebees, as they provide essential resources like food and nesting sites [30, 31]. In a recent study, Le Féon et al. [32] found that bees were negatively associated with agricultural intensification, while they were affected positively by the amount of semi-natural habitats in the surrounding landscape. Some conservationists’ studies of endangered species have emphasised the role and importance of large-scale dynamics (e.g. [27]); it therefore appears relevant to consider interactions between species and landscape elements when developing biodiversity conservation strategies. Hatfield and LeBuhn [17] suggested that bumblebee communities provide an excellent model for evaluating the importance of factors at patch and landscape scale. Even though bumblebees are known to have large foraging distances [17– 19], they appear to display a high dependency on their central foraging place [17, 33]. Our results show that bumblebees are related with variables at patch scale (i.e., species richness of flowering plants and percent cover of flowering plants) as well as variables at landscape scale (i.e., proportion of meadows, proportion of forest, ED_Land and SHDI) in different ways. Habitat quality may be assessed by its suitability for insects [34] using important ecological differences between generalists and specialist species. Specialist species are more susceptible to degradation and decrease of suitable habitats than generalists because they are dependent on specific types of habitats or flowering plants. A greater tongue length in bumblebees has been suggested as one factor that confers greater susceptibility to decline on some bumblebee species [7].

4 Conclusions

Overall, we found that not only the availability of food resources at patch level, but also the quality and diversity of the surrounding matrix, are important factors affecting the species richness and abundance of long-tongued bumblebees. Landscapes with high percentages of meadows, with a strong presence of edges and a diverse matrix, may support a higher diversity and abundance of long- tongued bumblebees. With the presence of adjacent patches of meadow and habitat edges in the surrounding landscape, there is an increased probability that bumblebees will encounter floral resources during their life cycle. In addition, it appears that the ability of bumblebees to fly long distances makes them less vulnerable to the level of fragmentation and patchiness in a given landscape.

108 Sustainable Development and Planning V 415

In countries with patchy landscapes, like Estonia, it is important to consider ecological indicators that are strongly associated with both patch and landscape variables. Bumblebees, because of their reliance on these variables, have the potential to serve as accurate indicators of habitat quality.

References

[1] Fischer, J. & Lindenmayer, D.B., Landscape modification and habitat fragmentation: a synthesis. Global Ecology and Biogeography, 16(3), pp. 265–280, 2007. [2] Mortelliti, A., Amori, G. & Boitani, L., The role of habitat quality in fragmented landscapes: a conceptual overview and prospectus for future research. Oecologia, 163(2), pp. 535–547, 2010. [3] Hall, L.S., Krausman, P.R. & Morrison, M.L., The habitat concept and a plea for standard terminology. Wildlife Society Bulletin, 25, pp. 173–182, 1997. [4] Goulson, D., Hanley, M.E., Darvill, B. & Ellis, J.S., Biotope associations and the decline of bumblebees (Bombus spp.). Journal of Insect Conservation, 10(2), pp. 95–103, 2006. [5] Mänd, M., Mänd, R. & Williams, I.H., Bumblebees in the agricultural landscape of Estonia. Agriculture, Ecosystems & Environment, 89(1-2), pp. 69–76, 2002. [6] Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O. & Kunin, W.E., Global pollinator declines: trends, impacts and drivers. Trends in Ecology and Evolution, 25(6), pp. 345–353, 2010. [7] Williams, P.H. & Osborne, J.L., Bumblebee vulnerability and conservation world-wide. Apidologie, 40(3), pp. 367–387, 2009. [8] Carvell, C., Habitat use and conservation of bumblebees (Bombus spp.) under different grassland management regimes. Biological Conservation, 103(1), pp. 33–49, 2002. [9] Xie, Z., Williams, P.H. & Tang, Y., The effect of grazing on bumblebees in the high rangelands of the eastern Tibetan Plateau of Sichuan. Journal of Insect Conservation, 12(6), pp. 695–703, 2008. [10] Goulson, D., Lye, G.C. & Darvill, B., Decline and conservation of bumblebees. Annual Review of Entomology, 53, pp. 191–208, 2008. [11] Goulson, D., Hanley, M.E., Darvill, B., Ellis, J.S. & Knight, M.E., Causes of rarity in bumblebees. Biological Conservation, 122(1), pp. 1–8, 2005. [12] Diaz-Forero, I., Liivamägi, A., Kuusemets, V. & Luig, J., Pollinator richness and abundance in Northeast Estonia: bumblebees, butterflies and dayflying moths. Forestry Studies | Metsanduslikud Uurimused, 53, pp. 5– 14, 2010. [13] Kumar, S., Simonson, S.E. & Stohlgren, T.J., Effects of spatial heterogeneity on butterfly species richness in Rocky Mountain National Park, CO, USA. Biodiversity and Conservation, 18(3), pp. 739–763, 2009. [14] Fauna Europaea Web Service. www.faunaeur.org

109 416 Sustainable Development and Planning V

[15] Charman, T.G., Sears, J., Green, R.E. & Bourke, A.F.G., Conservation genetics, foraging distance and nest density of the scarce Great Yellow Bumblebee (Bombus distinguendus). Molecular Ecology, 19(13), pp. 2661– 2674, 2010. [16] Bäckman, J.P. & Tianen, J., Habitat quality of field margins in a Finnish farmland area for bumblebees (Hymenoptera: Bombus and Psithyrus). Agriculture, Ecosystems & Environment, 89(1-2), pp. 53–68, 2002. [17] Hatfield, R.G. & LeBuhn, G., Patch and landscape factors shape community assemblage of bumble bees, Bombus spp. (Hymenoptera: Apidae), in montane meadows. Biological Conservation, 139(1-2), pp. 150–158, 2007. [18] Steffan-Dewenter, I., Münzenberg, U., Bürger, C., Thies, C. & Tscharntke, T., Scale-dependent effects of landscape context on three pollinator guilds. Ecology, 83(5), pp. 1421–1432, 2002. [19] Westphal, C., Steffan-Dewenter, I. & Tscharntke, T., Bumblebees experience landscapes at different spatial scales: possible implications for coexistence. Oecologia, 149(2), pp. 289–300, 2006. [20] Zurbuchen, A., Landert, L., Klaiber, J. Muller, A., Hein, S. & Dorn, S., Maximum foraging ranges in solitary bees: only a few individuals have the capability to cover long foraging distances. Biological Conservation, 143(3), pp. 669–676, 2010. [21] Sepp, K., Mikk, M., Mänd, M. & Truu, J. Bumblebee communities as an indicator for landscape monitoring in the agri-environmental programme. Landscape and Urban Planning, 67(1-4), pp. 173–183, 2004. [22] Eiden, G., Kayadjanian, M. & Vidal, C., From land cover to landscape diversity in the European Union, Chapter 1. Capturing landscape structures: Tools. ec.europa.eu/agriculture/publi/landscape/ch1.htm [23] Kivinen, S., Luoto, M., Kuussaari, M. & Helenius, J., Multi-species richness of boreal agricultural landscapes: effects of climate, biotope, soil and geographical location. Journal of Biogeography, 33(5), pp. 862–875, 2006. [24] Steffan-Dewenter, I., Importance of habitat area and landscape context for species richness of bees and wasps in fragmented orchard meadows. Conservation biology, 17(4), pp. 1036–1044, 2003. [25] Girardin, P., Bockstaller, C. & Van der Werf, H., Indicators: Tools to evaluate the environmental impacts of farming systems. Journal of Sustainable Agriculture, 13(4), pp. 6–21, 1999. [26] Billeter, R., Liira, J., Bailey, D., Bugter, R., Arens, P., Augenstein, I., Aviron, S., Baudry, J., Bukacek, R., Burel, F., Cerny, M., De Blust, G., De Cock, R., Diekotter, T., Dietz, H., Dirksen, J., Dormann, C., Durka, W., Frenzel, M., Hamersky, R., Hendrickx, F., Herzog, F., Klotz, S., Koolstra, B., Lausch, A., Le Coeur, D., Maelfait, J.P., Opdam, P., Roubalova, M., Schermann, A., Schermann, N., Schmidt, T., Schweiger, O., Smulders, M.J.M., Speelmans, M., Simova, P., Verboom, J., van Wingerden, W.K.R.E., Zobel, M. & Edwards, P.J., Indicators for biodiversity in

110 Sustainable Development and Planning V 417

agricultural landscapes: a pan-European study. Journal of Applied Ecology, 45(1), pp. 141–150, 2008. [27] Goulson, D., Rayner, P., Dawson, B. & Darvill, B., Translating research into action; bumblebee conservation as a case study. Journal of Applied Ecology, 48(1), pp. 3–8, 2011. [28] Rundlöf, M., Nilsson, H. & Smith, H.G., Interacting effects of farming practice and landscape context on bumble bees. Biological Conservation, 141(2), pp. 417–426, 2008. [29] Haaland, C. & Gyllin, M., Butterflies and bumblebees in greenways and sown wildflower strips in southern Sweden. Journal of Insect Conservation, 14(2), pp. 125–132, 2010. [30] Kells, A.R. & Goulson, D., Preferred nesting sites of bumblebee queens (Hymenoptera: Apidae) in agroecosystems in the UK. Biological Conservation, 109(2), pp. 165–174, 2003. [31] Svensson, B., Langerlöf, J. & Svensson, B.G., Habitat preferences of nest- seeking bumble bees (Hymenoptera: Apidae) in an agricultural landscape. Agriculture Ecosystems & Environment, 77(3), pp. 247–255, 2000. [32] Le Féon, V., Schermann-Legionneta, A., Delettrea, Y., Avironb, S., Billeterc, R., Bugterd, R., Hendrickxe, F. & Burel, F., Intensification of agriculture, landscape composition and wild bee communities: A large scale study in four European countries. Agriculture, Ecosystems & Environment, 137(1-2), pp. 143–150, 2010. [33] Osborne, J.L. & Williams, I.H., Site constancy of bumble bees in an experimentally patchy habitat. Agriculture Ecosystems & Environment, 83 (1-2), pp. 129–141, 2001. [34] Fahrig, L. & Jonsen, J., Effect of habitat patch characteristics on abundance and diversity of insects in an agricultural landscape. Ecosystems, 1(2), pp. 197–205, 1998.

111

IV Diaz-Forero, I., Kuusemets, V., Mänd, M., Liivamägi, A., Kaart, T. and Luig, J. 2011

Effects of forest habitats on the local abundance of bumblebee species: a landscape-scale study

Baltic Forestry 17(2), ISSN 1392-1355. (In press) Effects of forest habitats on the local abundance of bumblebee species: a landscape-scale study

ISABEL DIAZ-FORERO1*, VALDO KUUSEMETS1, MARIKA MÄND1, AVE LIIVAMÄGI1, TANEL KAART2 AND JAAN LUIG1

1Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Kreutzwaldi 5, Tartu, 51014, Estonia; *e-mail: [email protected], [email protected]

2Institute of Veterinary Medicine and Animal Sciences, Estonian University of Life Sciences, Kreutzwaldi 62, Tartu, 51014, Estonia

Abstract

The main objective of our study was to analyse how different bumblebee species are influenced by the presence of forest habitats in the surrounding landscape. The local abundance of bumblebee species was studied in 22 semi-natural meadows located in Northeast Estonia. The proportion of forest cover and brushwood cover, the mean patch area of forest and the edge density of forest were calculated at four spatial scales (i.e., radii of 250 m, 500 m, 1000 m and 2000 m). In total, we found 597 individuals of bumblebees belonging to 24 species (gen. Bombus), including five species of cuckoo bumblebees (subgen. Psithyrus). Our study shows that some bumblebee species seem to have different preferences in terms of the structure of the landscape. Some species may benefit from a heterogeneous landscape with a high proportion of forest habitats (e.g., B. schrencki), whereas others seem to prefer open areas (e.g., B. veteranus). We also found that some bumblebee species that have large foraging distances (e.g., B. terrestris) seem to be negatively affected by forest, as the presence of a high quantity of forest patches in the surrounding landscape could narrow their foraging area. The joint effects of the set of landscape variables related with forests appear to be important for some bumblebee species, specially, but not only, at the largest spatial scale. Overall, our results indicate that the presence of forest is very important for bumblebees, even for some species that prefer open areas, as forest habitats and edges may offer overwintering sites and nesting places for them. In countries like Estonia, where forests are widely distributed and represent a relevant part of the landscape mosaic, this type of habitat should be preserved if conservation of biodiversity is desired. In addition, conservation efforts targeting particular species of bumblebees should consider the landscape preferences of the species under study, and these efforts should aim to maintain the habitat types that are suitable for most bumblebee species.

Key words: Bombus, forest cover, brushwood, edge density, mean patch area, landscape indices 1

115 Introduction

Bees are considered a vital element of global biodiversity and an important pollinator group in agro-ecosystems. Their activities support both crops and the diversity of wild plants (Sepp et al. 2004, Goulson et al. 2006, Rundlöf et al. 2008, Knight et al. 2009, Potts et al. 2010). However, mainly due to the intensification of farming practices in agriculture, bumblebees and other pollinators are at risk (Mänd et al. 2002, Goulson et al. 2006, Holzschuh et al. 2008, Xie et al. 2008).

The decline of pollinators, particularly bees, has been widely recognised based on evidence from many countries worldwide (Mänd et al. 2002, Kells and Goulson 2003, Goulson et al. 2006, Williams and Osborne 2009, Potts et al. 2010). Bumblebees are considered the best-documented group in the existing literature on the topic (Potts et al. 2010). However, according to Goulson et al. (2006), very little is known about the habitat requirements of bumblebees. Many bumblebee species in Europe and North America have declined and become extinct at local levels, whereas other species are still common and widely distributed (Ahrné et al. 2009). The causes of these differences in response are not clear, but they appear to involve particular characteristics of single species such as diet and foraging distances (Ahrné et al. 2009). Williams (2005) suggested that some species have more specific habitat, but Goulson et al. (2006) argued that bumblebees are generally not habitat specialists because all the bumblebee species that they studied were found in more than one biotope; moreover, most species were found across a broad range of biotopes. However, data from monitoring in Finland have shown that some species prefer particular types of habitats (Söderman 1999). According to Goulson et al. (2011), bumblebees have been well-studied in modern agricultural landscapes of Western Europe, the United Kingdom, Asia and North America. These areas usually consist of large monoculture fields separated by field margins and patches of woodlands. However, very little is known about the distribution and ecology of bumblebees elsewhere. It seems important to study the associations between bumblebee species and landscape-scale factors, particularly in areas that have mosaic landscapes with high proportions of forest and natural habitats. In addition, Jones (2011) argues that the abundance and distribution of species are impacted by processes that occur at multiple spatial scales. For this reason, multiple-scale studies are currently needed (Jones, 2011). Generally, bumblebees are studied in regions having warmer climates and open landscapes. Therefore, studies on bumblebee populations conducted on the northern areas and in more forested landscapes are of great interest.

Different kinds of relationships have been found between forest cover and bees: e.g., Taki et al. (2007) found that bee abundance and species richness were positively related to forest cover (at a radius of 750 m), whereas Winfree et al. (2007) found negative relationships between similar

116 variables (at a radius of 1600 m). Other authors have found differences in the behaviour between some species of bumblebees. There is the case of B. pascuorum and B. terrestris that were studied by Kreyer et al. (2004); they found that although forest cover did not represent a barrier for both species, B. terrestris seems to prefer open landscapes. However, there is still little knowledge about the influence of forest on bumblebees and species-specific studies are currently needed (Kreyer et al. 2004). Estonia is a country with a landscape dominated by forest; for this reason, it is important to know how this land cover type influences bumblebees and whether forest could serve as a potential habitat for some bumblebee species.

We analysed how the local abundance of different bumblebee species is influenced by the presence of forest cover in the surrounding landscape. Our study was carried out in Northeast Estonia, where forests dominate the landscape. Our main objective was to study the relationships of every single species of bumblebees to a set of landscape variables related to forest and calculated at various spatial scales. In addition, we performed multiple regression analysis for each bumblebee species with more than 20 individuals using the landscape variables at different spatial scales as predictors; this was done to explain the joint effects of the landscape variables on the abundance of individual species.

Materials and methods

Study area

Our research was carried out in Ida-Virumaa County, Northeast Estonia. The total area of the county is 336,400 ha, approximately 7.4% of the total area of Estonia. The landscape in the region is generally dominated by forests, which occupy an area of 195,245 ha (approximately 58% of the total area of Ida-Virumaa). The region is also covered to a lesser extent by brushwoods having a total area of 21,701 ha (approximately 6.5% of the area of the county). The most common types of forests in the study area are coniferous and mixed and are dominated by pines and spruce trees. We selected 22 semi-natural meadows in the region as study sites. The areas of these meadows range from 0.10 to 3.83 ha.

Survey

Field work was done in 2008 and 2009, during June, July and August, the warmest months of the year. The visits were made twice each year. The first visit was made in early summer in June and the second in late summer, during the end of July or at the beginning of August. We used a systematic walking survey to count bumblebees (Kumar et al. 2009). The counting of individuals and identification of bumblebee species were done by sight, mainly on the wing or by the colour of the bumblebee when they were standing on the flowers. When the observer could not identify the 3

117 bumblebee species on site, some individuals were captured with an insect net for later identification in the laboratory. The surveys took approximately 45 min or more per visit per study site, until all the species present in the site were recorded. Our method was based on Goulson et al. (2006). Individuals and species of bumblebees were counted between 10:00 h and 16:00 h when the weather conditions were appropriate (i.e., temperature more than 18ºC and wind speed less than 5 by the Beaufort scale). Our source for the nomenclature of bumblebees was the Fauna Europaea Web Service (2004). The average air temperature in Ida-Virumaa County during summer (including June, July and August) was a bit higher in 2009 than in 2008 (15.1ºC and 14.9ºC, respectively).

Proportion of forest and brushwood

We determined the percentage of forest cover and brushwood cover within a radius of 250 m, 500 m, 1000 m and 2000 m around the centre of each study site. The forest cover in our study region is mainly composed of managed mixed forest; birches, pines and spruces are among the dominant trees. The brushwood cover is characterized by the presence of deciduous trees, woody seedlings, shrubs and young trees, primarily willows, maples, birches, among others, which have become established on abandoned agricultural land, in overgrown meadows or in forest clearings. Calculations were performed using the digital Estonian Basic Map at a scale of 1:10,000 provided by the Estonian Land Board. We used the software ArcGIS 9.3.

Landscape indices

In addition to the proportion of forest and brushwood, we used two Fragstats indices in our study: mean patch area of forest (AREA_MN) and edge density of forest (ED). AREA_MN equals the sum, across all patches of the corresponding patch type (here, forest) of the area of the patches, divided by the total number of patches of the same type (McGarigal et al. 2002). We used “mean patch size” because it gives information about the size of the patches and the number of patches at the same time. ED equals the sum of the lengths of all edge segments involving the corresponding patch type, divided by the total landscape area and multiplied by 10,000 (to convert to hectares) (McGarigal et al. 2002). These variables were also calculated at four spatial scales, i.e., radii of 250 m, 500 m, 1000 m and 2000 m. We used the software Fragstats, version 3.3.

Statistical analysis

Data on bumblebees were collected during two years, but we used the total numbers of species and individuals in our analyses. Spearman rank order correlation analysis was applied to describe the relationships between the landscape variables (i.e., proportion of forest, proportion of brushwood, mean patch area of forest and edge density of forest) at various spatial scales (at radii of 250 m, 4

118 500 m, 1000 m and 2000 m) and local abundance of different bumblebee species. For each landscape variable and species combination the scale corresponding to the strongest relationship was selected and the statistical significance of the selected relationships was tested, considering 24 tests (one for every species) performed with a single landscape variable (at the most suitable scale) and the species abundance as one experiment, and applying the Benjamini-Hochberg correction to the p values. The relationships with corrected p value < 0.05 were considered statistically significant. When the correlation coefficient (rs) was between 0.0 and ±0.3, the correlation was considered to be weak; when rs was between ±0.3 and ±0.6, the correlation was medium; and when rs was between ±0.6 and ±1, the correlation was strong. Additionally, multiple regression analysis was performed to study the joint effect of the four landscape variables at different scales on the local abundance of individual species, using only the bumblebee species with more than 20 individuals. STATISTICA 9 software was used for all the statistical analyses.

Results

Local abundance of bumblebee species

A total of 24 species of bumblebees (gen. Bombus), including five species of cuckoo bumblebees (subgen. Psithyrus), were found in Ida-Virumaa. Currently, 29 species of bumblebees, including seven species of cuckoo bumblebees, occur in Estonia. The total number of individuals recorded was 597, including 150 males, 84 queens and 363 workers. The bumblebee species with the highest number of individuals were B. pascuorum, B. lucorum and B. ruderarius with 140, 70 and 58 individuals, respectively. In contrast, B. muscorum and B. distinguendus were the bumblebee species having the lowest abundance (Table 1).

Relations between bumblebees and proportion of forest

We found that two species of cuckoo bumblebees have medium positive correlations with the proportion of forest: P. bohemicus and P. norvegicus at 250 m (rs = 0.54, p = 0.04; rs = 0.57, p = 0.04, respectively). Among the bumblebee species known to prefer forest and forest margins, we found that B. schrencki was positively associated with the proportion of forest at 500 m (rs = 0.58, p = 0.04). In addition, the most abundant species in the study area, B. pascuorum, was medium positively correlated with this variable at 250 m, but the relationship was nearly statistically significant (rs = 0.49, p = 0.07).

119 scale of 2000 m (in the case of B. veteranus, rs = -0.63, p = 0.03; and in the case of B. terrestris, rs = -0.55, p = 0.04). On the contrary, B. ruderarius and B. lapidarius had nearly significant negative correlations with the proportion of forest at the smallest spatial scale, i.e., 250 m (rs = -0.46, p =

0.09; rs = -0.50, p = 0.06, respectively).

Relations between bumblebees and proportion of brushwood

We found a medium positive correlation between B. schrencki and the proportion of brushwood at a large spatial scale (at 1000 m, rs = 0.60, p = 0.03). B. pascuorum was also positively correlated with this variable at 1000 m, but this correlation did not remain statistically significant after

Benjamini-Hochberg correction (rs = 0.44, p = 0.12). In addition, we found that the subspecies B. s. soroeensis and B. s. proteus showed opposing relationships with the proportion of brushwood (Figure 1): B. s. soroeensis was positively correlated at 500 m, whereas B. s. proteus was negatively correlated at 250 m; however, these correlations were not statistically significant after the correction (rs = 0.44, p = 0.12; rs = -0.45, p = 0.12, respectively). Other negative correlations were detected between the proportion of brushwood and some species of bumblebees, i.e., B. terrestris at 500 m (rs = -0.56, p = 0.04), B. veteranus at 1000 m (rs = -0.63, p = 0.03), and B. lapidarius with a nearly significant correlation at 500 m (rs = -0.54, p = 0.05). Also, P. bohemicus appeared to have a nearly significant negative correlation with the proportion of brushwood at 250 m (rs = -0.52, p = 0.06); in contrast, this species was found to be positively correlated with the proportion of forest at the same spatial scale (see previous section) (Figure 1).

Relations between bumblebees and landscape indices

Some bumblebee species showed significant positive relationships with edge density, i.e., B. pascuorum, B. pratorum and P. sylvestris at 2000 m; however, none of these correlations remained statistically significant after Benjamini-Hochberg correction (rs = 0.45, p = 0.15; rs = 0.43, p =

0.15; rs = 0.46, p = 0.15, respectively). In addition, medium negative relationships were detected between the edge density of forest and some bumblebee species (but these correlations were not statistically significant after the correction): B. sylvarum at 250 m (rs = -0.49, p = 0.15), B. s. proteus at 1000 m (rs = -0.46, p = 0.15) and B. veteranus at 2000 m (rs = -0.46, p = 0.15).

We found medium negative relationships between the mean patch area of forest and some bumblebee species: B. terrestris and B. veteranus at 2000 m (rs = -0.59, p = 0.04; rs = -0.65, p = 0.02, respectively). There were other nearly significant negative correlations between this variable and two bumblebee species at 500 m, i.e., B. ruderarius (rs = -0.51, p = 0.09) and B. lapidarius (rs = -0.52, p = 0.09).

Joint effects of landscape variables on individual species 6

120 The joint effects of landscape variables on the abundance of individual species had the best fit at different spatial scales (Table 2). The highest association for two of the most abundant species, B. pascuorum and B. ruderarius, was found at the scale of 2000 m; for both species the models were statistically significant and explained 51% and 43% of the variation in their abundance, respectively (Table 2). Other species that showed the best fit at the largest spatial scale were B. cryptarum and B. veteranus, and their models explained 50% and 30%, respectively; however, only the regression model for B. cryptarum was statistically significant (Table 2). In contrast, P. bohemicus was mostly influenced by the landscape variables at the smallest spatial scale (250 m); this model explained 62% of the variation and it was statistically significant (Table 2).

Previously, the bumblebee species B. cryptarum was not significantly associated with any of the single landscape variables (Figure 1). However, this species showed strong associations with the combination of landscape variables at the largest spatial scale, as it was mentioned above. In addition, B. cryptarum showed some significant associations inside the regression model: a positive relationship with the proportion of forest, and negative relationships with edge density and mean patch area of forest (Table 2).

Discussion

Our results showed that the presence of forest in the surrounding landscape of the habitat is an important factor for some species of bumblebees. In our study, B. pascuorum (the most abundant and widely distributed species in the study area), B. schrencki and particularly two species of cuckoo bumblebees (i.e., P. bohemicus and P. norvegicus) seemed to prefer landscapes having high proportions of forest in the surrounding areas. This finding is consistent with a study on bumblebees in Finland that recognised these species of cuckoo bumblebees as forest species (Bäckman and Tiainen 2002). On the other hand, B. pascuorum seemed to be positively influenced by the presence of forest cover at a small spatial scale. A possible explanation for this positive association was given by Goulson et al. (2010). They argued that B. pascuorum tends to nest above the ground in areas of leaf litter and thickets, and woodland areas are likely to offer these types of nesting sites (Goulson et al. 2010). In addition, we found clear indications that B. schrencki was positively influenced by the presence of forest. This species is said to prefer forest and forest margins (Söderman 1999). According to Söderman (1999), the expansion of this bumblebee in the Baltic countries was promoted by the rapid afforestation of open fields.

Other bumblebee species, particularly B. terrestris, B. veteranus, B. lapidarius and B. ruderarius, showed negative trends with the proportion of forest. These species seemed to prefer open areas. Mänd et al. (2002) found that the species B. lapidarius, B. veteranus and B. lucorum were particularly numerous in agricultural habitats, which are open areas. The bumblebee B. lapidarius 7

121 belongs to a group of specialists on Fabaceae, a large family of flowering plants that are commonly found in grasslands (Goulson et al. 2005). Also, species such as B. terrestris and B. lapidarius are considered spatial generalists because they have large foraging distances (Walther- Hellwig and Frankl 2000a, Walther-Hellwig and Frankl 2000b). In a recent study, Hagen et al. (2011) found that some bumblebee species are able to flight long distances (maximum distances of 1.3 – 2.5 km) and to use large areas (0.25 – 43.53 ha); e.g., they found that B. terrestris can flight a maximum distance of 2.5 km. These bumblebee species may prefer an open landscape to have more freedom for their long-distance flights. Similarly, Bäckman and Tiainen (2002) classified B. ruderarius, B. lapidarius and B. veteranus as species preferring open habitats. Other species, such as B. pascuorum, are considered ubiquitous as they can be found in different types of habitats (Bäckman and Tiainen 2002, Goulson et al. 2006).

Medium and strong relationships involving brushwood were found for some bumblebees. Some species were positively related with the proportion of brushwood, i.e., B. schrencki and B. pascuorum, whereas others were negatively related, i.e., B. terrestris, B. veteranus, B. lapidarius and P. bohemicus. In addition, we found that the subspecies B. s. soroeensis and B. s. proteus seem to be ecologically different: B. s. soroeensis appears to prefer brushwoods, whereas B. s. proteus does not. Positive relationships may occur because many patches of brushwood have grown in areas that were former meadows. The soil of these areas is rich in calcium and can therefore support more flowering plant species. Additionally, brushwood areas are dominated by willows that offer food during their spring flowering period. However, brushwood areas dominated by willows might also represent an ecological trap for bumblebees. Food resources would become scarce in summer and these areas would no longer be able to provide good forage places for bumblebees.

Relationships involving edge density of forest seem to be positive for some species of bumblebees, i.e., B. pascuorum, B. pratorum and P. sylvestris, whereas negative relationships were found for B. sylvarum, B. s. proteus and B. veteranus. A positive relationship between B. pascuorum and edge density may occur because this species prefers bell-shaped flowers. Flowers having this shape occur commonly in berry-bearing plants, and these plants often grow close to forests or in forest margins. In general, edges may support a greater abundance and diversity of flowering plants. The species B. pratorum was also positively associated with forest edges. This finding is consistent with Goulson et al. (2005); they suggested that early-emerging species like B. pratorum are related to woodland and woodland edges. Sepp et al. (2004) suggested that edges are particularly important in April and May as bumblebee queens forage from the flowering willows that are commonly found in the forest edges of Estonia.

122 Negative associations were found between the mean patch area of forest and some species of bumblebees (i.e., B. terrestris, B. veteranus, B. lapidarius and B. ruderarius). Kreyer et al. (2004) have found that even though forest did not seem to represent a barrier for the bumblebee B. terrestris, this species seems to prefer open areas. Overall, mean patch area of forest showed a negative pattern of relationships with the local abundance of the bumblebee species that were significantly associated with this variable, even when the joint effects of the landscape variables were analysed in multiple regression. Moreover, this landscape index appears to be important for the bumblebee species that seem to prefer open areas, judging from the negative relationships found with the proportion of forest. One possible explanation is that a high number of patches of forest may be seen as potential obstacles in the landscape for some species of foraging bumblebees (Kreyer et al. 2004, Goulson et al. 2010), particularly for those like B. lapidarius and B. terrestris that have large foraging distances (Walther-Hellwig and Frankl 2000a, Walther-Hellwig and Frankl 2000b).

The joint effects of the set of landscape variables related with forests appear to be important for some bumblebee species (i.e., B. pascuorum, B. ruderarius, B. cryptarum and P. bohemicus), specially, but not only, at the largest spatial scale.

Conclusions

Our study shows that some bumblebee species seem to have preferences related to the structure of the landscape. Some species may benefit from a heterogeneous landscape with a high proportion of forest habitats (e.g., B. schrencki), whereas others seem to prefer open landscapes (e.g., B. veteranus). Some bumblebee species that have large foraging distances (e.g., B. terrestris) also seem to prefer open landscapes because the presence of many forest patches in the surrounding landscape could narrow their foraging area, affecting their long-distance flights. Other studies have also found differences in behaviour and preferences between some bumblebee species (e.g., Kreyer et al. 2004).

In general, our results indicate that the presence of forest is very important for bumblebees, even for those species that seem to prefer open areas, because forest habitats may provide overwintering sites and nesting places. Similarly, Taki et al. (2007) concluded in their study on potential pollinators that forest loss at the landscape scale may cause negative impacts on bee communities. In countries like Estonia, where forests are widely distributed and represent a relevant part of the landscape mosaic, forest habitats seems to be very important for some species of bumblebees and therefore should be preserved if conservation of biodiversity is desired. In addition, conservation efforts intended to protect particular species of bumblebees should consider their preferences for

123 surrounding landscapes. These efforts should aim to maintain the habitat types that are suitable for the target species.

Acknowledgements

This research was funded by targeted financing of the Estonian Ministry of Education and Research (SF1090050s07, SF0170057s09), Estonian Science Foundation Grant No. 7391 and by an applied research project of the Estonian Ministry of Agriculture (T8014PKPK). We greatly thank the two anonymous referees whose comments greatly improved the manuscript.

References

Ahrné, K., Bengtsson, J. and Elmqvist, T. (2009) Bumble Bees (Bombus spp) along a Gradient of Increasing Urbanization. PLoS ONE 4 (5): e5574.

Bäckman, J.C. and Tiainen, J. 2002. Habitat quality of field margins in a Finish farmland area for bumblebees (Hymenoptera: Bombus and Psithyrus). Agriculture, Ecosystems and Environment 89: 53–68.

Fauna Europaea Web Service. 2004. Fauna Europaea version 1.1. http://www.faunaeur.org.

Goulson, D., Hanley, M.E., Darvill, B. and Ellis, J.S. 2006. Biotope associations and the decline of bumblebees (Bombus spp.). Journal of Insect Conservation 10 (2): 95–103.

Goulson, D., Hanley, M.E., Darvill, B., Ellis, J.S. and Knight, M.E. 2005. Causes of rarity in bumblebees. Biological Conservation 122: 1–8.

Goulson, D., Lepais, O., O’Connor, S., Osborne, J.L., Sanderson, R.A., Cussans, J., Goffe, L. and Darvill, B. 2010. Effects of land use at a landscape scale on bumblebee nest density and survival. Journal of Applied Ecology 47 (6): 1207–1215.

Goulson, D., Rayner, P., Dawson, B. and Darvill, B. 2011. Translating research into action; bumblebee conservation as a case study. Journal of Applied Ecology 48 (1): 3–8.

Hagen, M., Wikelski, M. and Kissling, W.D. 2011. Space use of bumblebees (Bombus spp.) revealed by radio tracking. PLoS ONE 6 (5): e19997.

Holzschuh, A., Steffan-Dewenter, I. and Tscharntke, T. 2008. Agricultural landscapes with organic crops support higher pollinator diversity. Oikos 117: 354–361.

Jones, J.P.G. 2011. Monitoring species abundance and distribution at the landscape scale. Journal of Applied Ecology 48 (1): 9-13. 10

124 Kells, A.R. and Goulson, D. 2003. Preferred nesting sites of bumblebee queens (Hymenoptera: Apidae) in agroecosystems in the UK. Biological Conservation 109: 165–174.

Knight, M.E., Osborne, J.L., Sanderson, R.A., Hale, R.J., Martin, A.P. and Goulson, D. 2009. Bumblebee nest density and the scale of available forage in arable landscapes. Insect Conservation and Diversity 2: 116–124.

Kreyer, D., Oed, A., Walther-Hellwig, K. and Frankl, R. 2004. Are forests potential landscape barriers for foraging bumblebees? Landscape scale experiments with Bombus terrestris agg. and Bombus pascuorum (Hymenoptera, Apidae). Biological Conservation 116 (1): 111–118.

Kumar, S., Simonson, S.E. and Stohlgren, T.J. 2009. Effects of spatial heterogeneity on butterfly species richness in Rocky Mountain National Park, CO, USA. Biodiversity and Conservation 18 (3): 739–763.

Mänd, M., Mänd, R. and Williams, I.H. 2002. Bumblebees in the agricultural landscape of Estonia. Agriculture, Ecosystems and Environment 89: 69–76.

McGarigal, K., Cushman, S.A., Neel, M.C. and Ene, E. 2002. FRAGSTATS, Spatial Pattern Analysis Program for Categorical Maps. Computer software program produced by the authors at the University of Massachusetts, Amherst. Available at the website: http://www.umass.edu/landeco/research/fragstats/fragstats.html.

Natural History Museum. 2011. Research and curation. Available at the website: http://www.nhm.ac.uk.

Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O. and Kunin, W.E. 2010. Global pollinator declines: trends, impacts and drivers. Trends in Ecology and Evolution 25 (6): 345–353.

Rundlöf, M., Nilsson, H. and Smith, H.G. 2008. Interacting effects of farming practice and landscape context on bumble bees. Biological Conservation 141 (2): 417–426.

Sepp, K., Mikk, M., Mand, M. and Truu, J. 2004. Bumblebee communities as an indicator for landscape monitoring in the agri-environmental programme. Landscape and Urban Planning 67: 173–183.

Söderman, G. 1999. Diversity of pollinator communities in Eastern Fennoscandia and Eastern Baltics. Results from pilot monitoring with yellow traps in 1997 - 1998. Finnish Environment Institute 355, Nature and Natural Resources, Edita Ltd, Helsinki, Finland, 69 pp.

125 Taki, H., Kevan, P.G. and Ascher, J.S. 2007. Landscape effects of forest loss in a pollination system. Landscape Ecology 22: 1575–1587.

Walther-Hellwig, K. and Frankl, R. 2000a. Foraging habitats and foraging distances of bumblebees, Bombus spp. (Hym., Apidae), in an agricultural landscape. Journal of Applied Entomology 124 (7-8): 299–306.

Walther-Hellwig, K. and Frankl, R. 2000b. Foraging distances of Bombus muscorum, Bombus lapidarius, and Bombus terrestris (Hymenoptera, Apidae). Journal of Insect Behavior 13 (2): 239– 246.

Winfree, R., Griswold, T. and Kremen, C. 2007. Effect of human disturbance on bee communities in a forested ecosystem. Conservation Biology 21 (1): 213–223.

Williams, P.H. 2005. Does specialization explain rarity and decline among British bumblebees? A response to Goulson et al. Biological Conservation 122: 33–43.

Williams, P.H. and Osborne, J.L. 2009. Bumblebee vulnerability and conservation world-wide. Apidologie 40: 367–387.

Xie, Z., Williams, P.H. and Tang, Y. 2008. The effect of grazing on bumblebees in the high rangelands of the eastern Tibetan Plateau of Sichuan. Journal of Insect Conservation 12: 695–703.

126 Table 1. Total number of individuals of bumblebees per species found in 2008 and 2009

Species Number of individuals 2008 2009 Bombus cryptarum (Fabr.) 13 11 Bombus distinguendus Morawitz 1 0 Bombus hortorum (L.) 5 6 Bombus hypnorum (L.) 3 6 Bombus jonellus (Kirby) 2 16 Bombus lapidarius (L.) 5 21 Bombus lucorum (L.) 18 52 Bombus muscorum (L.) 1 0 Bombus pascuorum (Scopoli) 70 70 Bombus pratorum (L.) 6 9 Bombus ruderarius (Müller) 26 32 Bombus semenoviellus Skorikov 0 7 Bombus schrencki Morawitz 6 10 Bombus soroeensis ssp. Soroeensis (Fabr.) 9 5 Bombus soroeensis ssp. Proteus (Fabr.) 1 15 Bombus soroeensis ssp. soroeensis x proteus (Fabr.) 8 13 Bombus sylvarum (L.) 5 13 Bombus terrestris (L.) 2 16 Bombus veteranus (Fabr.) 3 24 Psithyrus bohemicus Seidl. 6 27 Psithyrus campestris (Panzer) 3 13 Psithyrus norvegicus (Sparre-Schneider) 3 8 Psithyrus rupestris (Fabr.) 9 9 Psithyrus sylvestris (Lep.) 2 7 Total 207 390 Average per meadow 9.41 17.73 Standard error 0.90 1.87

127

Figure 1. Relationships between the local abundance of bumblebee species and the studied landscape characteristics at various spatial scales based on the Spearman rank correlation coefficients (rs). The width of the circle indicates the strength of the relationship (the bigger the circle, the stronger the relationship between the variables) and the colour determines the direction of the relationship (black circles correspond to positive relationships and white circles to negative relationships). The stars (*) inside the circles indicate the statistically significant correlations, after Benjamini-Hochberg correction (p < 0.05), for the spatial scale with the strongest relationship

128 Table 2. Results of the multiple regression analyses. For each bumblebee species with over 20 individuals, four models corresponding to the different spatial scales were fitted; regression coefficients and model fit characteristics of the best model (with the highest R2 and the smallest p- value) are presented

Species Scale with Model Regression coefficients R2 Model the best fit intercept Proportion Proportion of EDa AREA_MNb p value of forest brushwood (m/ha) (ha) (%) (%) B. cryptarum 2000 3.231 0.081* 0.049 -0.066* -0.061* 0.500 0.014 B. lapidarius 500 2.354 -0.070 -0.103 0.019 0.030 0.370 0.081 B. lucorum 250 4.247 0.028 -0.084 -0.001 -0.533 0.290 0.189 B. pascuorum 2000 2.032 0.061 0.359 0.040 -0.109* 0.508 0.013 B. ruderarius 2000 11.368 0.202* 0.321 -0.210* -0.181* 0.427 0.041 B. s. soroeensis 1000 1.505 -0.023 0.208 0.002 -0.019 0.295 0.179 x proteus B. veteranus 2000 2.555 -0.005 -0.058 -0.005 -0.028 0.297 0.177 P. bohemicus 250 -1.215 0.087* -0.103* 0.007 -0.471* 0.618 0.002 a Edge density of forest; b Mean patch area of forest * Regression coefficients significant at p < 0.05

129 CURRICULUM VITAE

Personal information Name: Isabel Diaz Forero Nationality: Colombian Date of birth: September 28th, 1981 Place of birth: Bucaramanga, Colombia Address: Estonian University of Life Sciences. Kreutzwaldi 5, Tartu 51014, Estonia E-mail: [email protected], [email protected]

Education 2007-2011 PhD studies in Environmental Sciences and Applied Biology. Estonian University of Life Sciences. Tartu, Estonia 2005-2006 MSc studies in Environment and Resource Management. Vrije Universiteit. Amsterdam, The Netherlands 1999-2004 BSc studies in Environmental Engineering. Universidad Pon- tificia Bolivariana. Bucaramanga, Colombia

Additional training 2009 NOVA PhD course: Insect Conservation. Oscarsborg, Nor- way 2008 BioLandMan course: Landscape Ecology and Analyses. Jel- gava, Latvia

Academic degree MSc studies in Environment and Resource Management

Languages Spanish (Mother tongue), English (High level), French (Inter- mediate level)

Professional employment Since 2007 Estonian University of Life Sciences, Institute of Agricultural and Environmental Sciences. Tartu, Estonia. Assistant 2006 UNESCO-IHE, Institute for Water Education. Delft, The Netherlands. Internship / Researcher 2004-2005 Electrificadora de Santander SA ESP, Energy Company. Buca- ramanga, Colombia. Environmental Adviser 2003-2005 Hidroriente Ltda, Engineering and Consulting firm. Bucara- manga, Colombia. Assistant Engineer

130 Research interests Insect conservation; bumblebees (Bombus spp.); bees (Hyme- noptera: Apidae); landscape ecology; habitat quality

Awards and grants 2011 Award on Intellectual Distinction UPB – Universidad Pontifi- cia Bolivariana, Colombia 2011 ESF DoRa T8 “Participation of young researchers in international exchange of knowledge” – 5th International Conference on Sustainable Development and Planning, England 2011 Mobility Support Grant from the Doctoral School of Earth Sciences and Ecology – 10th International ICPBR Pollination Symposium, Mexico 2011 IUBS Young Scientist Funding – 10th International ICPBR Pollination Symposium, Mexico 2009 ESF DoRa T8 “Participation of young researchers in international exchange of knowledge” – 2nd European Congress of Conservation Biology, Czech Republic 2009 ESF DoRa T8 “Participation of young researchers in international exchange of knowledge” – NOVA PhD course on Insect Conservation, Norway 2004 Bachelor’s Degree with Honours – Universidad Pontificia Bolivariana, Colombia

Memberships 2011 International Commission for Plant-Bee Relationships (IC- PBR), member 2008 European Association for the promotion of Science and Tech- nology (EUROSCIENCE), member

131 ELULOOKIRJELDUS

Isikuandmed Eesnimi: Isabel Perekonnanimi: Diaz Forero Rahvus: Kolombia Sünniaeg: 28. september 1981 Sünnikoht: Bucaramanga, Kolombia Töökoht: Põllumajandus- ja Keskkonnainstituut, Eesti Maaülikool, Kreutzwaldi 5, Tartu 51014, Eesti E-mail: [email protected], [email protected]

Haridustee 2007-2011 Doktoriõpe keskkonnateaduse ja rakendusbioloogia erialal. Eesti Maaülikool, Tartu, Eesti 2005-2006 Magistriõpe keskkonna- ja ressursimajanduse erialal (in Environment and Resource Management), Vrije Universiteit. Amsterdam, Holland 1999-2004 Bakalaureuseõpe keskkonnatehnoloogia erialal, Universidad Pontificia Bolivariana. Bucaramanga, Kolombia

Erialane täiendamine 2009 NOVA doktorikursus „Insect Conservation”. Oscarsborg, Norra 2008 BioLandMan’i kursus „Landscape Ecology and Analyses”. Jelgava, Läti

Teaduskraad MSc keskkonna- ja ressursimajanduse erialal

Keelteoskus Hispaania (emakeel), Inglise (kõrgtase), Prantsuse (kesktase)

Teenistuskäik Alates 2007 Põllumajandus- ja keskkonnainstituut, Eesti Maaülikool, laboriassistent 2006 UNESCO-IHE, Institute for Water Education. Delft, Hol- land, praktika / uurija 2004-2005 Electrificadora de Santander SA ESP, Energy Company. Buca- ramanga, Kolombia, keskkonnanõunik 2003-2005 Hidroriente Ltda, Engineering and Consulting firm. Bucara- manga, Kolombia, tehnoloog-assistent

132 Uurimistöö põhisuunad Putukate kaitse; kimalased (Bombus spp.); mesilased (Hyme- noptera: Apidae); maastikuökoloogia; elupaigakvaliteet

Teaduspreemiad ja stipendiumid 2011 Award on Intellectual Distinction UPB – Universidad Pontificia Bolivariana, Kolombia 2011 ESF DoRa T8 reisistipendium “Participation of young researchers in international exchange of knowledge” konverentsil osalemiseks: 5th International Conference on Sustainable Development and Planning, Inglismaa 2011 Reisistipendium maateaduste ja ökoloogia doktorikoolist sümpoosionil osalemiseks: 10th International ICPBR Pollination Symposium, Mehhiko 2011 IUBS noorteadlase toetus sümpoosionil osalemiseks: 10th Inter- national ICPBR Pollination Symposium, Mehhiko 2009 ESF DoRa T8 reisistipendium “Participation of young researchers in international exchange of knowledge” konverentsil osalemiseks: 2nd European Congress of Conservation Biology, Tšehhi 2009 ESF DoRa T8 reisistipendium doktorikoolis osalemiseks: NOVA doktorikursus: Insect Conservation, Norra 2004 bakalaureusekraad Degree with Honours – Universidad Ponti- ficia Bolivariana, Kolombia

Teadusorganisatsiooniline tegevus: 2011 International Commission for Plant-Bee Relationships (ICPBR), liige 2008 European Association for the promotion of Science and Technology (EUROSCIENCE), liige

133 LIST OF PUBLICATIONS

Publications indexed in the ISI Web of Science database

Diaz-Forero, I., Kuusemets, V., Mänd, M., Liivamägi, A., Kaart, T. and Luig, J. 2011. Influence of local and landscape factors on bumblebees in semi-natural meadows: a multiple-scale study in a forested landscape. Journal of Insect Conservation (Submitted). Diaz-Forero, I., Kuusemets, V., Mänd, M., Liivamägi, A., Kaart, T. and Luig, J. 2011. Effects of forest habitats on the local abundance of bumblebee species: a landscape-scale study. Baltic Forestry 17(2), ISSN 1392-1355. Diaz-Forero, I., Kuusemets, V., Mänd, M. and Luig, J. 2011. Bumble- bees as potential indicators for the evaluation of habitat quality. Sus- tainable Development and Planning V. WIT Transactions on Ecology and the Environment. WIT Press. Vol 150, 409-417.

Peer-reviewed articles in other international research journals

Diaz-Forero, I., Liivamägi, A., Kuusemets, V. and Luig J. 2010. Pol- linator richness and abundance in Northeast Estonia: bumblebees, butterflies and day-flying moths. Forestry Studies | Metsanduslikud Uurimused 53, 5–14.

Conference abstracts

Diaz-Forero, I., Kuusemets, V. and Mänd M. 2011. Effects of human settlements and green areas at multiple spatial scales on the diversity and abundance of bumblebees. 10th International Pollination Sympo- sium, June 27–29, Cholula, Mexico. Diaz Forero, I., Liivamägi, A., Kuusemets V. and Luig, J. 2010. How are habitat and landscape factors influencing the diversity and abundance of bumblebees? Conference “Nature Conservation beyond 2010”, May 27–29, Tallinn, Estonia.

134 Diaz Forero, I., Liivamägi, A., Luig, J. and Kuusemets V. 2009. Re- lationships between landscape structure, human impacts and insect diversity. Symposium: Promoting grassland insect conservation and diversity. 2nd European Congress of Conservation Biology ‘Conserva- tion biology and beyond: from science to practice’, September 1–5, Prague, Czech Republic. Diaz Forero, I., Liivamägi, A., Luig, J. and Kuusemets V. 2009. Insect diversity in semi-natural habitats of Estonia: its relation with habitat and landscape structure. 5th International Conference “Research And Conservation Of Biological Diversity In Baltic Region”, April 22–24, Daugavpils, Latvia.

135

VIIS VIIMAST KAITSMIST

MARIT KOMENDANT ANTENNAL CONTACT CHEMORECEPTION IN GROUND BEETLES (COLEOPTERA: CARABIDAE) JOOKSIKLASTE (COLEOPTERA: CARABIDAE) ANTENNAALNE KEMORETSEPTSIOON Dr. Enno Merivee Prof. Anne Luik May 6, 2011

VAHUR PÕDER COMPATIBILITY OF ENERGY CONSUMPTION WITH THE CAPACITY OF WIND GENERATORS ENERGIA TARBIMISE SOBIVUS TUULEGENERAATORITE VÕIMSUSEGA Prof. Andres Annuk June 21, 2011

MERIKE LILLENBERG RESIDUES OF SOME PHARMACEUTICALS IN SEWAGE SLUDGE IN ESTONIA, THEIR STABILITY IN THE ENVIRONMENT AND ACCUMULATION INTO FOOD PLANTS VIA FERTILIZING MÕNEDE RAVIMIJÄÄKIDE SISALDUS EESTI REOVEESETTES, NENDE STABIILSUS KESKKONNAS JA AKUMULEERUMINE KOMPOSTVÄETISEST TOIDUTAIMEDESSE Prof. Lembit Nei Prof. Kalev Sepp September 16, 2011

MARGE MALBE THE ROLE OF SELENIUM IN UDDER HEALTH OF DAIRY COWS SELEENI TOIME LÜPSILEHMADE UDARA TERVISELE Prof. emer. Hannu Saloniemi Dots. Andres Aland. October 7, 2011

EHA ŠVILPONIS FACTORS INFLUENCING THE DISTRIBUTION AND OVERWINTERING SURVIVAL OF THE POTATO ROT NEMATODE (DITYLENCHUS DESTRUCTOR THORNE 1945) KARTULIINGERJA (DITYLENCHUS DESTRUCTOR THORNE 1945) LEVIKUT JA TALVIST ELLUJÄÄMIST MÕJUTAVAD FAKTORID Prof. Anne Luik December 9, 2011

ISBN 978-9949-484-09-6