Predicting the loss of forests, carbon stocks and biodiversity driven by a neotropical 'gold rush'.

The loss of tropical forests represents a major threat to biodiversity. With accelerating deforestation in large parts of the Amazon, the Guiana Shield region, with its large expanse of closed forest cover, has the potential to play a crucial role in both climate change mitigation and biodiversity conservation. However, the region is now facing increasing deforestation pressures, primarily from artisanal gold mining activities concentrated in the nation of Guyana. To identify areas of Guyana at the highest risk of deforestation over the next 25 years, we employed a spatio-temporal modelling approach that accounted for the stochastic and contagious nature of deforestation. Our model predicted a 9 % net decrease in total forest cover by 2043. While the primary drivers of deforestation were mining and human settlements, protected areas were shown to reduce the probability of deforestation. Therefore, we assessed the potential impact of a proposed expansion of the protected area network in Guyana, on forest loss, carbon stocks and habitat loss for the country's most threatened forest vertebrates. Establishing the proposed protected areas would reduce forest loss by 17 %, predicted habitat losses by an average of 1.9 % per vertebrate group, and aboveground carbon emissions by 466,968 t over the next 25 years. These findings highlight the utility of using predictive models to identify areas at risk of future deforestation, which can contribute to the development of effective strategies against tropical forest loss, biodiversity loss and climate change. • Used spatio-temporal modelling to predict 9% decrease in forest in Guyana by 2043, driven by mining and human settlements. • Proposed protected areas can cut forest loss by 17%, habitat losses by 1.9% per taxa group & carbon emissions by 466,968 t. • Findings highlight how predictive models can aid in strategies against forest loss, biodiversity decline, and climate change. [ABSTRACT FROM AUTHOR]