Search for: All records

The Endangered Species Act (ESA) was a landmark protection for rare organisms in the United States. Although the ESA is known for its protection of wildlife, a majority of listed species are actually plants and lichen. Climate change will impact species populations globally. Already-rare species, like those listed in the ESA, are at an even higher risk due to climate change. Despite this, the risk climate change poses to endangered plants has not been systematically evaluated in over a decade. To address this gap, we modified previously existing qualitative assessment toolkits used to examine the threat of climate change in federal documentation on listed wildlife. These modified toolkits were then applied to the 771 ESA listed plants. First, we evaluated how sensitive ESA listed plants and lichens were to climate change based on nine quantitative sensitivity factors. Then, we evaluated if climate change was recognized as a threat for a species, and if actions were being taken to address the threats of climate change. We found that all ESA listed plant and lichen species are at least slightly (score of 1) sensitive to climate change, and therefore all listed plants and lichens are threatened by climate change. While a majority of ESA listing and recovery documents recognized climate change as a threat, very few had actions being taken in their recovery plans to address climate change directly. While acknowledging the threat that climate change poses to rare plants is an important first step, direct action will need to be taken to ensure the recovery of many of these species. 

Abstract. We investigate techniques for using deep neural networks to produce surrogatemodels for short-term climate forecasts. A convolutional neural network istrained on 97 years of monthly precipitation output from the 1pctCO2 run (theCO2 concentration increases by 1 % per year) simulated by the second-generation Canadian Earth System Model (CanESM2). The neural network clearly outperforms a persistence forecast anddoes not show substantially degraded performance even when the forecast lengthis extended to 120 months. The model is prone to underpredicting precipitationin areas characterized by intense precipitation events. Scheduled sampling(forcing the model to gradually use its own past predictions rather than groundtruth) is essential for avoiding amplification of early forecasting errors.However, the use of scheduled sampling also necessitates preforecasting(generating forecasts prior to the first forecast date) to obtain adequateperformance for the first few prediction time steps. We document the trainingprocedures and hyperparameter optimization process for researchers who wish toextend the use of neural networks in developing surrogate models.