Addressing risks to biodiversity arising from a changing climate: The need for ecosystem restoration in the Tana River Basin, Kenya

with less sensitivity to the changing climate which can be used in ecosystem restoration proj- ects is vital. In terms of restoration, there are still gaps in knowledge of how to effectively restore most types of ecosystems [87]. Some researchers have shown the importance of intact faunal communities, particularly those that disperse seeds, for the successful restoration of for- est areas [88]. Implications for tourism These changes to the biodiversity of the Tana River Basin will have implications for tourism. Kenya is known internationally as a hotspot of wildlife tourism, with the majority of tourists visiting to watch wildlife. Nearly 10% of Kenya’s GDP comes from tourism. Most visitors focus on the more well-known PAs in the south of the country. Visitor numbers in selected PAs in South Africa, Namibia and Botswana were higher in areas where there was high preda- tor species richness and a presence of locally rare ungulate species [89]. A high diversity of large mammals was also found to be important in attracting high numbers of wildlife tourists. The PAs within the Tana River Basin generally saw a decrease in suitability for many of the predator species examined here, such as the wild dog, cheetah and African lion. Similarly, a reduction in the number of species in the PAs with climate change could impact tourist num- bers. This could be particularly significant for the Tsavo East National Park, which could expe- rience reductions in the number of species present in the future. Other than the Masai Mara,

the Tsavo ecosystem is the most popular with tourists [77]. In order to maintain high visitor numbers, additional conservation spending may be necessary. Limitations Although this analysis provides important information about the impacts of climate change and development plans on the terrestrial biodiversity of the Tana River Basin, a number of caveats need to be considered. There are several relevant factors that have not been included in this analysis due to the spatial scale used in the Wallace Initiative database [47]. These factors include the potential spread of disease pathogens and pests, interactions between species (food availability, predator-prey relationships and competitive interactions) [90], the effects of extreme climatic events [91] and the direct biotic effects of increases in CO2 concentrations on plants [92]. In addition, many species, especially narrow-ranged endemic species, are not included in the Wallace Initiative database. Narrow-ranged species are generally sensitive to climate change [93] so many of the endemic species not included in this analysis will also be vulnerable to the effects of global temperature rise. Species with fewer than 10 data points (occupied grid cells) were excluded from the Wallace Initiative database in order to maintain a robust analysis [47]. Despite these limitations, the Wallace Initiative fulfils many of the criteria recently set out as contributing to best practice in species distribution modelling [94]. GBIF data combines many datasets worldwide and the data was further cleaned (checked for locational consistency

and outliers) before use. Commission errors were accounted for by clipping species to their biogeographic zones and omission errors were taken into account by generous buffers around species distributions. Furthermore, uncertainty is taken into account by considering a range of GCMs and dispersal rates. There are some general limitations with species distribution models which should also be considered. In some cases, there is not sufficient data available to fully inform the model as to the true distribution of a species [95, 96]. Datasets used to drive the SDM are often biased because of an unequal sampling effort across the study area [97]. SDMs can be overfitted, which can lead to flawed outputs by limiting the model’s capacity to generalise. During the PLOS ONE Risks to Kenya’s biodiversity from climate change PLOS ONE | https://doi.org/10.1371/journal.pone.0254879 July 21, 2021 18 / 25