Resource management requires relevant and up to date science to inform the decision-making process, yet for many access to balanced scientific information is limited. Further, rapid increase in publications, makes it difficult to stay abreast of the science related to a particular management issue. The National Park Service (NPS) manages a myriad of natural and cultural resources, including natural sounds and night skies, all requiring relevant and up to date science to effectively manage. Peer-reviewed scientific publications addressing wildlife responses to noise and artificial lights are more numerous each year; more than 50% of all papers have been published in the past five years on noise effects and over 65% pertaining to the effects of artificial lights. To obtain a repeatable measure of progress in these fields, we developed standardized and validated queries for Web of Science (WOS) to capture relevant peer-reviewed papers. To cope with the challenge of extracting a comprehensive and balanced summary of these papers, we are developing automated analytical tools to summarize the effects over a range of organisms, ecological contexts, noise and light sources, and biological response variables. This research provides periodic updates to resource managers about the evolving knowledge related to these resources and the aspects of park management to which this knowledge pertains. A parallel effort to summarize the effects of noise and artificial lights on human health and visitor experience is underway.
Results/Conclusions
Our WOS search resulted in 359 artificial light and 2753 noise related articles from 1978-2018. To determine if the searches adequately captured the literature, we compared the WOS search output to citations in highly cited review papers in both research areas, excluding all gray literature papers. The WOS searches were refined to insure we were not missing papers of interest. The final search criteria identified 90% of the noise effect papers and 92% of the artificial light papers. To eliminate papers that were not relevant to the topic, based on our expert knowledge, each paper was labeled as relevant (Wildlife, Human, Review-Wildlife, Review-Human) or irrelevant (Off-topic, Generic, Measurement). Implementing a machine learning algorithm we successfully automated classification of the current literature search. Additional algorithms are underdevelopment to further classify papers (e.g. noise or light sources, taxonomic group, etc). Application of these algorithms to subsequent literature searches will result in efficient and systematic classification of the relevant science on the effects of noise and light on humans and wildlife.