|Aim We conduct the first assessment of likely future climate change impacts
for biodiversity across the West African protected area (PA) network using climate
projections that capture important climate regimes (e.g. West African
Monsoon) and mesoscale processes that are often poorly simulated in general
circulation models (GCMs).
Location West Africa.
Methods We use correlative species distribution models to relate species
(amphibians, birds, mammals) distributions to modelled contemporary climates,
and projected future distributions across the PA network. Climate data
were simulated using a physically based regional climate model to dynamically
downscale GCMs. GCMs were selected because they accurately reproduce
important regional climate regimes and generate a range of regional climate
change responses. We quantify uncertainty arising from projected climate
change, modelling methodology and spatial dependency, and assess the spatial
and temporal patterns of climate change impacts for biodiversity across the PA
Results Substantial species turnover across the network is projected for all
three taxonomic groups by 2100 (amphibians = 42.5% (median);
birds = 35.2%; mammals = 37.9%), although uncertainty is high, particularly
for amphibians and mammals, and, importantly, increases across the century.
However, consistent patterns of impacts across taxa emerge by early to midcentury,
suggesting high impacts across the Lower Guinea forest.
Main conclusions Reducing (e.g. using appropriate climate projections) and
quantifying uncertainty in climate change impact assessments helps clarify likely
impacts. Consistent patterns of high biodiversity impacts emerge in the early
and mid-century projections, while end-of-century projections are too uncertain
for reliable assessments. We recommend that climate change adaptation
should focus on earlier projections, where we have most confidence in species
responses, rather than on end-of-century projections that are frequently used.
In addition, our work suggests climate impact should consider a broad range
of species, as we simulate divergent responses across taxonomic groups.|