Assessing the distribution of bats in southern Africa to highlight conservation priorities

Abstract

Approximately 25% of bats globally are threatened, but limited data on African bats, which account for 20% of bat species, hinders our understanding of their conservation status across this ecologically diverse continent. This study combined: modelling techniques, to predict current species distributions for 58 southern African bat species and project past, current and future distributions of 22 endemic and near-endemic species; bat acoustic surveys, to assess landscape features influencing bat activity in arid and semi-arid regions; and conservation planning software to design a large-scale monitoring network for bats across this subcontinent.

Species distribution models were employed using a robust and well established presence-only modelling technique (Maximum Entropy – Maxent) to model the current distributions of 58 species in southern Africa. Although the important eco-geographical variables were species- or in some cases family-specific, overall water availability (both temporary and permanent), seasonal precipitation, vegetation and karst (caves/limestone) areas were the most important factors associated with distribution patterns. These species distributions were then used to identify range-restricted and narrow niche breadth species, alongside other life-history strategies considered to put species at risk, such as Old World pteropodids and cave-dwelling bats to identify species most at risk. Nine of the 58 species in this study were identified as ‘at risk’. Considering range-restriction and endemism separately, the results showed that range-restricted species were a higher proportion (50%) of ‘at risk’ species than endemics (41%) but six of the nine identified species were endemic and range restricted (67%). If only areas of high species richness are prioritised, important areas with low species richness but rare, ‘at risk’ or endemic species would be excluded.

Species distributions are not fixed but may shift due to changes in environmental conditions. Accurately predicting changes in species’ distributions due to anthropogenic climate change remains a fundamental challenge for conservation biologists, and this is amplified when dealing with taxa such as bats that are inherently difficult to study and in areas, such as Africa, with sparse ecological data. To better understand endemic bat species risk to climate change in southern Africa and to highlight historical and future likely refugia, Maxent was employed to forecast range-shifts for 22 southern African endemic or near-endemic species. Species distributions were projected during the Last Glacial Maximum (LGM ~22,000 BP), present (1950-2000) and future (2070: averaged 2061-2080, using IPCC5 scenarios) climatic conditions. Climate change was predicted to change species composition extensively within a relatively short timescale (within 60 years). By 2070, 86% of species modelled are predicted to have range contractions and six species were highlighted to be most at risk, with range contractions of more than 20%.

The majority of southern Africa is composed or arid or semi-arid regions. Generally arid and semi-arid areas are overlooked and understudied due to low species richness, yet these areas are known to have a high proportion of endemic species. As part of this study, driven transects were carried out across arid and semi-arid areas to assess bat activity in these areas. Bat activity was recorded at 94% of the acoustic surveys, demonstrating that driven transects are an effective method of surveying bats in southern Africa. Bat activity increased at lower altitudes and higher latitudes, which characteristically have more rainfall, permanent water and vegetation. Although water has been shown in other studies to be important for bats, temporary water was not shown to influence bat activity and permanent water was positively correlated with bat activity for hipposiderids and rhinolophids and FM bats, which may reflect the fact that water features important for bats at smaller scale. The same two vegetation types that were consistently negatively correlated with bat activity were drier vegetation types (Karoo-Namib shrubland) and high salinity halophytic vegetation.

Finally, a systematic conservation planning software tool (Marxan) was used to design multi-species monitoring networks that incorporated all 58 target species across the 11 ecoregions found in southern Africa. To ensure rare, endemic and range-restricted species were monitored at the same level as widespread species, species distributions (mapped using Maxent) were extracted by ecoregion. Monitoring targets (i.e. a percentage of species distribution across ecoregions) were standardised to ensure the same percentage of predicted distribution was included across all species (rare and widespread). To account for different resources and capacity, three optimal monitoring networks (minimum monitoring stations to achieve the monitoring targets) were proposed to survey 1, 5 or 10% of all species distributions within each ecoregion. The optimal solution for monitoring 1% of species distributions within ecoregions was found by monitoring 1,699 stations (survey sites), or for 5% 8,486 stations and finally for 10% 17,867 stations would be needed.

In conclusion, the findings presented in this thesis have important conservation implications and have the potential to inform the practical steps required towards the introduction of a bat monitoring programme in southern Africa. While this study has highlighted challenges to African bat conservation, it has also demonstrated that an integrated and multi-disciplinary approach, using emerging techniques and conservation tools (e.g. conservation planning and automated call analysis software) can be used to fill knowledge gaps and inform conservation priorities in the absence of systematically collected data.