Predicting biodiversity loss in mountain rivers due to glacier retreat

Glaciers are shrinking worldwide and many will disappear by 2100. Although physical/ chemical processes in glacier-fed rivers are well understood, there are major uncertainties about biodiversity responses and species extinction risks. These uncertainties seriously hinder conservation decisions in mountain biodiversity hotspots. Problems are compounded by (1) sparse/biased invertebrate/algae/microbe data; (2) ambiguities from multiple algorithms used in species distribution models (SDM) to predict current/future biodiversity, and; (3) uncertainty propagated from climate-glacier-river hydrology models that drive SDMs. For example, we evaluated SDM utility for the entire alpine zone of the Alps but can only provide robust predictions for 13 species (cf. 100s of invertebrate and thousands of algae/microbe species). This project aims to improve predictions by reducing SDM uncertainties with improved datasets and modelling, to better understand the risks of mountain biodiversity loss.

The project incorporates multiple elements of uncertainty reduction, spanning biological data collation/collection, hydrological modelling, and statistical analytics. Key challenges are likely to include:

1. Evaluating the major gaps/spatial biases in available biological datasets with a comprehensive review/meta-analysis.

2. Collection of new datasets to address current problems with SDMS. This student could choose to focus on invertebrates, algae, bacteria or fungi. Data collection could be undertaken in the Alps to advance existing models, or alongside a new NERC-funded project in the Himalayas beginning in 2025 to compare predictions among regions.

3. Evaluation of uncertainties in SDMs. New biological datasets will allow us to evaluate the 12 algorithms in the R package BIOMOD2 to understand and identify the extent to which they introduce uncertainty into river biodiversity predictions.

4. Reducing uncertainty in SDM predictions – existing Alps SDMs are driven by data describing glacier extent, catchment area, slope, and soil pH. Additional spatial data (land cover/land use, geology and lake distribution) is expected to enhance model predictions. Applications of SDMs to regions beyond the Alps is possible with new predictions of river network evolution based on DEM datasets.

5. Additional uncertainty in our current SDM predictions arises from glacier models being driven by ensembles of 13 global climate models with only one ‘middle of the road’ Shared Socio-economic Pathway (SSP2-4.5). A full understanding of likely ecosystem responses to glacier loss needs to be based on a range of likely climate pathways.

6. Identifying where limited conservation resources should be targeted. For example, SDMs show areas where species could find refuge if glaciers persist beyond 2100. Decision-making, incorporating remaining uncertainties, can then be informed by the extent to which these areas need protection (national parks) or are at risk of development (e.g. hydropower).