Premium
Using imaging spectroscopy to predict above‐ground plant biomass in alpine grasslands grazed by large ungulates
Author(s) -
Schweiger Anna K.,
Risch Anita C.,
Damm Alexander,
Kneubühler Mathias,
Haller Rudolf,
Schaepman Michael E.,
Schütz Martin
Publication year - 2015
Publication title -
journal of vegetation science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.1
H-Index - 115
eISSN - 1654-1103
pISSN - 1100-9233
DOI - 10.1111/jvs.12214
Subject(s) - biomass (ecology) , grassland , vegetation (pathology) , environmental science , normalized difference vegetation index , ungulate , plant community , physical geography , ecology , scale (ratio) , leaf area index , geography , ecological succession , habitat , biology , cartography , medicine , pathology
Aims Imaging spectroscopy enables measurement of vegetation optical properties to predict vegetation characteristics that are important for a wide range of ecological applications. Our aim was to predict fresh above‐ground biomass of heterogeneous alpine grasslands in two areas and at two ecological scales. We assessed model plausibility for an intensively studied alpine grassland site (plant community scale) having distinct biomass and ungulate grazing patterns. Location Alpine grasslands in the S wiss N ational P ark. Methods Biomass data were collected in 51 plots and combined with imaging spectroscopy data to establish simple ratio models. We analysed the predictive power and transferability of models developed in two areas (Val Trupchun, Il Fuorn) and at two ecological scales (regional, local). In a next step, we compared our results to the broadband normalized difference vegetation index ( NDVI ). Finally, we assessed the correlations between model predictions and plant biomass distribution at the plant community scale. Results The best local simple ratio models yielded a model fit of R 2 = 0.60 and R 2 = 0.30, respectively, the best regional model a fit of R 2 = 0.44. NDVI model performance was weaker for the regional and one local area, but slightly better for the other local area. However, at the plant community scale only the local model showed a significant positive correlation ( R S = 0.39) with the known biomass distribution. Further, predictive power decreased when models were transferred from one local area to another or from one ecological scale to another. Conclusions Our study demonstrated that imaging spectroscopy is generally useful to predict above‐ground plant biomass in alpine grasslands with distinct grazing patterns. Site‐specific local models based on simple ratio indices performed better than the NDVI or regional models, suggesting that standardized approaches might not be adequate, particularly in heterogeneous grasslands inhabited by large ungulates. We emphasize the importance of collecting ground reference data covering the expected range of productivity and plant species composition. Moreover, plant community‐scale data from a previous study proved to be extremely valuable to test model plausibility.