Outstanding Challenges in the Transferability of Ecological Models.

Predictive models are central to many scientific disciplines and vital for informing management in a rapidly changing world. However, limited understanding of the accuracy and precision of models transferred to novel conditions (their 'transferability') undermines confidence in their predictions. Here, 50 experts identified priority knowledge gaps which, if filled, will most improve model transfers. These are summarized into six technical and six fundamental challenges, which underlie the combined need to intensify research on the determinants of ecological predictability, including species traits and data quality, and develop best practices for transferring models. Of high importance is the identification of a widely applicable set of transferability metrics, with appropriate tools to quantify the sources and impacts of prediction uncertainty under novel conditions.

Profound and pervasive degradation of Madagascar's freshwater wetlands and links with biodiversity.

Reflecting a global trend, freshwater wetlands in Madagascar have received little conservation or research attention. Madagascar is a global conservation priority due to its high level of species endemism but most work has focused on protecting forests. For the first time, we investigated the state of wetlands across the country to determine the effects of human disturbance. We conducted a rapid survey of 37 wetlands, using waterbirds and benthic invertebrates as ecological indicators. We recorded nine variables relating to human disturbance, revealing widespread wetland destruction. Principal Components Analysis reduced the nine variables to a single Principal Component (PC) that explained 50% of the dataset variance, demonstrating that different forms of human disturbance are ubiquitous and inseparable. The disturbance PC provides an index of how pristine a lake is and in Generalized Linear Models (GLMs) was significantly inversely related to the number of waterbird species present and the density of Chironomidae. The disturbance PC was estimated for every wetland in a GIS-derived dataset of wetland locations in Madagascar, giving a country-wide frequency distribution of disturbance. To validate the estimated PC values, we used the GLMs to predict the number of endemic bird species at an independent sample of 22 lakes. The predicted values correlated with the observed number of species, demonstrating that our procedure can identify lakes with high biodiversity value. The disturbance PC provides a convenient method for ranking sites, and a country-wide ranking demonstrates that the only near-pristine lakes in Madagascar are small sites that have been preserved by remoteness from human activity and not conservation management. The strategy of conserving high biodiversity remnants is insufficient because existing remnants suffer some degree of degradation and only support small populations of threatened species. Large-scale restoration of degraded wetlands is required for the long-term conservation of Madagascar's freshwater biodiversity.