Expert knowledge is increasingly used in conservation science to classify dynamic and complex ecosystems, while overcoming typical data-limitations. Based on the expertise of field operators from the local Environment Agency, we classified 161 stream sites in Trentino according to the presence of known hydrological, morphological and chemical alterations, while also identifying sites in reference conditions. We then used machine learning approaches to examine the degree to which a-priori expert classification matched data-driven classification based on the taxonomic and functional composition of benthic macroinvertebrates. The majority of stream sites were a-priori considered impacted by either one or a combination of anthropogenic alterations (~80%), with only 16% of sites in reference conditions. Random Forest a-posteriori classification of stream sites according to taxonomic and functional macroinvertebrate data matched the expert-based classification only partially. While stream sites considered in reference conditions were correctly classified, discrimination among hydro-morphological and chemical alterations was often poor. This suggests that indicators used to assess the ecological status of streams in mountain areas based on macroinvertebrates taxonomic and functional classification can assess the overall stress of a waterbody, but they show poor sensitivity to specific stressors, with relevant outcomes for the water management of Alpine running waters.
Vallefuoco, F.; Larsen, S.; Zolezzi, G.; Bruno, M.C. (2021). Classification of streams ecological status: comparing expert-knowledge and data-driven approaches. In: SEFS 12: Symposium for European Freshwater Sciences, online, 25-30 July 2021: 264. handle: http://hdl.handle.net/10449/69544
Classification of streams ecological status: comparing expert-knowledge and data-driven approaches
Vallefuoco, F.;Larsen, S;Bruno, M. C.
2021-01-01
Abstract
Expert knowledge is increasingly used in conservation science to classify dynamic and complex ecosystems, while overcoming typical data-limitations. Based on the expertise of field operators from the local Environment Agency, we classified 161 stream sites in Trentino according to the presence of known hydrological, morphological and chemical alterations, while also identifying sites in reference conditions. We then used machine learning approaches to examine the degree to which a-priori expert classification matched data-driven classification based on the taxonomic and functional composition of benthic macroinvertebrates. The majority of stream sites were a-priori considered impacted by either one or a combination of anthropogenic alterations (~80%), with only 16% of sites in reference conditions. Random Forest a-posteriori classification of stream sites according to taxonomic and functional macroinvertebrate data matched the expert-based classification only partially. While stream sites considered in reference conditions were correctly classified, discrimination among hydro-morphological and chemical alterations was often poor. This suggests that indicators used to assess the ecological status of streams in mountain areas based on macroinvertebrates taxonomic and functional classification can assess the overall stress of a waterbody, but they show poor sensitivity to specific stressors, with relevant outcomes for the water management of Alpine running waters.File | Dimensione | Formato | |
---|---|---|---|
abstract Vallefuoco oral.pdf
accesso aperto
Tipologia:
Versione editoriale (Publisher’s layout)
Licenza:
Tutti i diritti riservati (All rights reserved)
Dimensione
583.37 kB
Formato
Adobe PDF
|
583.37 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.