Drivers of wild bee abundance and diversity in social-ecological landscapes
Drivers of wild bee abundance and diversity in social-ecological landscapes
Keywords
Functional dispersion
functional divergence
functional evenness
functional richness
honeybees
Mediterranean
pollinators
1. Introduction
Human disturbances, particularly changes in land use and management, have been linked to the decline in pollinator diversity and abundance (Dicks et al., 2021, Millard et al., 2021) and have attracted public and policy attention. Substantial efforts are underway to respond, through action at global (Convention on Biological Diversity, 2018), regional and national scales, with countries adopting pollinator strategies and action plans (European Commission, 2023). The recently adopted European Union (EU) Nature Restoration Law requires the EU Member States to take appropriate measures to improve pollinator diversity and reverse the decline of pollinator populations by 2030 and thereafter achieve an increasing trend in pollinator populations. Nonetheless, several challenges are anticipated in its implementation, including the need for well-defined and practical strategies, reliable data, resource and finance mobilisation and effective community engagement, are anticipated, and may compromise the success of the Nature Restoration Law (Hering et al., 2023, Stoffers et al., 2024).
Habitat modification, at both at local micro-habitat and landscape scales, affects the abundance and diversity of pollinators (Duchenne et al., 2020, Gaspar et al., 2022, Kleijn et al., 2018, Scheper et al., 2015). The persistence of pollinator species at local scale is a consequence of the interplay between the local micro-habitat conditions and landscape management practices (Cohen et al., 2022, Coutinho et al., 2021, Gaspar et al., 2022, Marcacci et al., 2022, Rodríguez et al., 2021). Increased availability and diversity of food resources and nesting sites at the local scale are positively related with pollinator richness and abundance. Similarly, proximity to semi-/natural habitats is associated with diverse and abundant pollinator communities (Gaspar et al., 2022, Marcacci et al., 2022).
Landscape-scale variables may exert stronger influence on pollinator abundance and diversity than local variables, as positive landscape-level effects can offset smaller negative impacts at the local level (Perović et al., 2015, Steckel et al., 2014). For example, wild bee abundance, richness and functional diversity increased with vegetation diversity in road verges, but these results were influenced by landscape diversity, distance from the city and pollinator traits (Dietzel et al., 2024). There is a growing body of research addressing the impact of local and landscape variables on wild bee communities but most of the studies focus on agricultural and natural systems, and we know less about the impact of environmental variables on bee communities in urban systems (Buchholz and Egerer, 2020, Cohen et al., 2022, Fauviau et al., 2024, Liang et al., 2023, Wenzel et al., 2020), and even less on the variation in wild bee communities across landscapes having different habitats combined with gradients of environmental conditions in social-ecological landscapes (Biella et al., 2022, Gillespie et al., 2024, Neokosmidis et al., 2016).
Social-ecological factors impacting on wild bee communities include abiotic factors, such as soil structure, slope, aspect and temperature (Antoine and Forrest, 2021, Kammerer et al., 2021; Maher et al., 2019), but also biotic factors, including interspecific interactions with managed bees (Baldock, 2020; Iwasaki and Hogendoorn, 2022). These social-ecological filters often interact, resulting in filtered functional trait combinations at the local scale, and therefore impacting on ecosystem functions and services (Grilo et al., 2022). For example, urbanisation causes drastic and often irreversible habitat changes leading to overall negative impacts on pollinator richness and abundance (Liang et al., 2023) but can also have neutral or positive effects on pollinators depending on the availability of specific urban features such as botanical gardens, allotments, residential gardens, and vacant lots which may serve as important habitats and improve habitat connectivity for pollinator communities (Baldock, 2020; Biella et al., 2022; Theodorou et al., 2020). Such refuges often support high wild bee richness, particularly in landscapes with well-connected urban green areas (Baldock et al., 2019; Biella et al., 2022; Wenzel et al., 2020). The abundance and diversity of bee communities has also been shown to be correlated with socio-economic drivers, with income inequality affecting bee abundance and richness (Baldock et al., 2019; Reynolds et al., 2025). Since the responses of pollinators often are trait-specific (Buchholz and Egerer, 2020, Liang et al., 2023), these complex and interacting social-ecological variables can also affect the functional traits of bee communities through filtering, hence affecting fitness and survival. For example, ground nesting and solitary bees are more strongly negatively affected by urbanisation (Liang et al., 2023) while those with polylectic diets (i.e. generalist), cavity-nesting behaviour and later emergence are also expected to be more favoured compared to other groups by urban development (Ayers and Rehan, 2021). More dense urban areas, with highly fragmented green spaces and characterised by and urban heat island effect favour large, social, polylectic and/or above-ground nesting species (Ferrari and Polidori, 2022).
The interactions within these social-ecological systems impact on wild bee abundance and functional diversity are developed here into a conceptual framework that has been used to identify the following research questions (Fig. 1):
- a)How do local habitat factors (i.e. local habitat type and plant species richness) impact wild bee abundance, species richness and functional diversity?
- b)How do landscape scale factors (i.e. altitude, slope, aspect, proximity to urban areas, Normalised Difference Vegetation Index (NDVI) and Landscape Shannon Diversity Index (LSDI)) impact wild bee abundance, species richness and functional diversity?
- c)Does honeybee abundance modify the impact of local and landscape factors on wild bee abundance, species richness and functional diversity?
Fig. 1. Conceptual framework to analyse the impact of local and landscape habitat variables on wild bee communities.
2. Methodology
2.1. Study area and sampling design
The study was conducted in the archipelago of Malta, characterised by Mediterranean climatic conditions, and namely dry and hot summers and mild and wet winters. The islands consist of heterogeneous landscapes shaped by both natural processes and centuries of human activity, and including steppe, garrigue, shrubland and woodland habitats within an agricultural and urban and built-up area matrix. The archipelago predominantly consists of limestone, influencing soil composition and structure, and the topography which consists of low-lying hills and plateaus with valleys supporting riparian vegetation (Balzan et al., 2018). Centuries of human activities have altered soil and vegetation cover, converted land for cultivation, led to the construction of terraces, and driven intensive urban sprawl. Current drivers are related to rapid urban development leading to increased densification and urban sprawl across the main islands, Malta and Gozo, with an observed rural-urban gradient in the reduction of green space availability and ecosystem service capacities, together with the contemporary agricultural abandonment of less productive land and agricultural intensification in other areas (Balzan et al., 2022).
2.2. Bee transects and functional traits
The data analysed in this study were derived from a national monitoring dataset of plant-bee interactions in the Maltese Islands (Balzan et al., 2023, Balzan et al., 2017, Balzan et al., 2016) with data collection conducted annually using standardised transects across six habitat types: agricultural, coastal, garden, road verge, steppe, and shrubland habitats (Table A.1, Supplementary Information). A total of 204 sites were visited between April and May from 2016 to 2019, with each site visited once each year. The transects were carried out under dry, cloudless, and low-wind conditions to minimise weather-related variability. At each site, standardised, geolocated belt transects measuring 2 ×25 m were surveyed through timed walks of 20 minutes by a single observer. While some transects were revisited in multiple years, others were sampled once to maximise spatial and habitat representativeness across the Maltese Islands. Given the objective of this study to assess the wild bee functional diversity, we only considered the transects with wild bee abundance (Fig. 2; Table A.1). During these transects, flower-visiting bees were captured using a hand net, with both the plant and bee species identities recorded for each interaction. Bees that were identified in the transect were released immediately, while collected bees were identified in the laboratory to genus level (Michez et al., 2019) and then sent to a specialist for determination at species level (see Acknowledgment section). Bees that were not possible to identify to species level were identified to the genus level.
Fig. 2. Sampling points location within the study region (LULC map: Copernicus Coastal Zones Land Cover/Land Use 2018).
Functional traits of wild bees, presented in Supplementary Information Table C.1, were considered as they determine how species respond to habitat change. For example, body size, diet breadth, sociality, and nesting behaviour influence the species response to urbanisation (Brasil et al., 2024). Body size was measured quantitatively using a digital calliper for the intertegular distance (ITD) and body length (in mm), while the dry weight was measured by dehydrating the specimens at 70°C for 24 hours to remove any residual moisture, then weighing them using an analytical balance (detection limit: 0.001 g). Additional traits data were extracted from the European Pollinator Traits Database (Miličić and Vujić, 2023).
Using the functional trait data, Functional Diversity (FD) indices were calculated using the FD-package (version 1.0–12.1; (Laliberté and Legendre, 2010). Functional Richness (FRic) measures the extent of multi-dimensional functional space occupied by a community. The FRic increases with the addition of functionally unique species but remains the same with the addition of a redundant species. Functional Evenness (FEve) assesses how species traits are distributed across trait space and are weighted by the relative abundance of each species. Functional Divergence (FDiv) captures the extent to which species differ in their functional traits relative to the centre of the functional space while considering the abundance of these species. If a newly added species is functionally similar, then an increase in its abundance will lead to a decrease in FDiv. Functional Dispersion (FDis) is the mean distance in multidimensional trait spaces of individuals species from its abundance-weighted community centroid. FDis is used to identify environmental filtering processes that shape communities (Laliberté and Legendre, 2010, Mason et al., 2005, Villeger et al., 2008). A low dispersion is recorded when bee species with similar traits accumulate at sites and high dispersion is recorded with higher trait diversity and niche partitioning (Buchholz et al., 2020, Coutinho et al., 2021, Dietzel et al., 2024).
2.3. Local and Landscape Drivers
We considered the “local level” to be the sampling plot as a representative of the habitat in which it is found. For the local scale metrics, we considered the local habitat type category (described in Table A.1, Supplementary Information) and flowering plant species richness in each transect. When including wild bee abundance and diversity, or the functional diversity indices as a response variable, we also considered honeybee abundance as an independent variable. For the landscape scale metrics, we used national datasets and remotely sensed data, which is presented and described in Table 1, to characterise the habitat conditions surrounding the transects within circular buffers with varying radii (r=100, 200, 500, 1000, 2000m). Terrain data was obtained from a Digital Terrain Model (DTM) acquired in 2012 for the Maltese Islands from airborne LIDAR survey carried out in 2012 at a resolution of 1 m. Land use land cover data was obtained from (Balzan et al., 2018), who had developed a land use land cover (LULC) map using Sentinel 2 satellite images provided by Copernicus. Within the context of increasing impact of urbanisation on biodiversity and ecosystem services (Balzan et al., 2018, Balzan and Santis, 2023), we considered distance of each transect from the nearest urban land cover and Shannon Diversity Index. Sentinel-2 imagery from the dry and wet seasons between 2018 and 2023 was used to assess vegetation coverage while we obtained Level-2 Land Surface Temperature (LST) data from the Sentinel-3 satellite images spanning multiple dates and covering both the dry and wet seasons during the years 2019-2023, to ensure comprehensive temporal coverage. To reduce the bias arising from using only one satellite image/product, we used the median during the sampling year while cloudy images were discarded. Honeybee hive density data at 1Km2 resolution was digitised using anonymised records of hive density as provided by the Ministry for Agriculture, Fisheries and Animal Rights (MAFA), Malta in 2020.
대화 참여하기