Mining supplies critical minerals, supports millions of livelihoods, and can help fuel economic development, particularly in low- and middle-income countries. Yet, these contributions can entail substantial trade-offs for biodiversity and ecosystems, through habitat loss, deforestation, and pollution. These impacts are particularly concerning when mining occurs in places which are also hotspots for biodiversity, such as Madagascar. In this thesis, I explore the challenges and opportunities for reconciling mining and biodiversity conservation in Madagascar. Using spatial data analysis and counterfactual methods for impact evaluation I evaluate the real and potential impacts of mining on the forests and, by proxy, biodiversity of Madagascar, and the effectiveness of policies to mitigate that impact.

First, I evaluate the effectiveness of a key policy mechanism for mitigating the impacts of infrastructure development on biodiversity: biodiversity offsetting. Using statistical methods for counterfactual impact evaluation (statistical matching, difference-in-differences, and fixed effects panel regressions), I show that Madagascar’s largest mine, Ambatovy, has likely prevented enough deforestation within its four biodiversity offsets to compensate for the 2,000 hectares of forest cleared at the mine site. As such, the mine has likely achieved No Net Loss of forest. This shows that biodiversity offsetting can contribute towards mitigating the impact of mining, but it requires considerable investment and there are important caveats.

Next, I switch focus from industrial to artisanal and small-scale mining (ASM). ASM is widespread across Madagascar and has occurred within several protected areas. Using geological data to define and map the potential distribution of primary ruby, sapphire, and emerald deposits, I find that 11-14% of the most important area for biodiversity in Madagascar (defined using four measures) could host primary gem deposits and be impacted by gem mining in future. However, this approach also revealed a vast area of potentially prospective land outside these areas where decentralised, community managed zones for ASM could be established, minimising environmental trade-offs.

Finally, I evaluate what can happen when an artisanal mining rush involving tens of thousands of people occurs within a protected area: focussing on the 2016 sapphire rush at Bemainty in the eastern rainforests of Madagascar. Using the synthetic control approach for impact evaluation and drawing on additional field data (from a lemur census and semi-structured interviews) I find that the mining rush did not significantly increase deforestation or degradation, relative to counterfactual forest loss from other causes. Field data indicate that lemur populations appeared to remain healthy three years after the rush. These results emphasize the heterogeneity of impacts of ASM and the need for more robust, case-study evaluations to ensure policy responses are evidence-based and appropriate.

Overall, I find that while there is potential for biodiversity to be impacted by artisanal gem mining across Madagascar, the impacts of ASM are highly heterogenous and, in some cases, may be lower than alternative livelihood activities driving land cover change. Furthermore, there are opportunities for policies, including pragmatic alternative approaches to licensing and zonation, to mitigate the environmental trade-offs of mining in Madagascar, but these require effective governance.