First results in modelling distribution patterns of anecic earthworms on catchment scale: a Boosted Regres-sion Tree model approach

Palm, Juliane und van Schaik, Loes und Schröder, Boris (2010) First results in modelling distribution patterns of anecic earthworms on catchment scale: a Boosted Regres-sion Tree model approach. In: Boden und Standortqualität - Bioindikation mit Regenwürmern, 25.-26.02.2010, Osnabrück.

Vorschau

PDF Herunterladen (197kB)

Kurzfassung

Earthworms can serve as indicator species for various important soil processes. Therefore we can link distribution patterns of earthworms to the incidence of induced soil processes. Species distribution models can help us to predict earthworm distributions on different scales. Studies on larger scales are rare due to large efforts in data acquisition, especially if the focus is on process understanding. The current study focuses on anecic earthworms using a Boosted regression tree model approach for predicting the earthworm distribution in an agricultural area in Baden- Württemberg (Germany). We surveyed management, topography, soil parameters and presence-absence, abundance and fresh weight of anecic earthworms (i.e. Lumbricus terrestris) at 75 locations within the Weiherbach catchment. Our final model has acceptable performance (AUC=0.76 after validation). Topographic indices (wetness and Beer’s index) as well as soil parameters such as moisture, texture and tillage are the most relevant environmental predictors.

Eintragstyp:	Konferenz- oder Workshop-Beitrag ("Berichte der DBG")
Stichwörter:	Verbreitungsmodelle, Regenwürmer, Bodenhydrologie, Boosted Regression Trees (BRT)
Bereiche:	Kommissionen > Kommission III: Bodenbiologie und Bodenökologie
Benutzer:	Heinz-Christian Fründ
Hinterlegungsdatum:	05 Apr 2010 11:18
Letzte Änderung:	13 Dez 2015 16:25
URI:	https://eprints.dbges.de/id/eprint/496

Aktionen (Anmeldung erforderlich)

Eintrag anzeigen