Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
6292707 | Ecological Indicators | 2017 | 9 Pages |
â¢Optical traits determine the spectral signal of vegetation stands.â¢We ask how optical traits link plant species composition to canopy reflectance.â¢We analyze seasonal variations of these links with remotely-sensed trait indicators.â¢Results show the contribution of biochemistry and structure to floristic mapping.
Most plant species feature similar biochemical compositions and thus similar spectral signals. Still, empirical evidence suggests that the spectral discrimination of species and plant assemblages is possible. Success depends on the presence or absence of faint but detectable differences in biochemical (e.g., pigments, leaf water and dry matter content) and structural properties (e.g., leaf area, angle, and leaf structure), i.e., optical traits. A systematic analysis of the contributions and spatio-temporal variability of optical traits for the remote sensing of organismic vegetation patterns has not yet been conducted. We thus use time series of optical trait values retrieved from the reflectance signal using physical models (optical trait indicators, OTIs) to answer the following questions: How are optical traits related among patterns of floristic composition and reflectance? How variable are these relations in space and time? Are OTIs suitable predictors of plant species composition?We conducted a case study of three temperate open study sites with semi-natural vegetation. The canopy reflectance of permanent vegetation plots was measured on multiple dates over the vegetation period using a field spectrometer. We recorded the cover fractions of all plant species found in the vegetation plots and extracted gradients of species composition from these data. The physical PROSAIL leaf and canopy optical properties model was inverted with random forest regression models to retrieve time series of OTIs for each plot from the reflectance spectra. We analyzed these data sets using correlation analyses. This approach allowed us to assess the distribution of optical traits across gradients of species composition. The predictive performance of OTIs was tested in relation to canopy reflectance using random forest models.OTIs showed pronounced relationships with floristic patterns in all three study sites. These relationships were subject to considerable temporal variability. Such variability was driven by short-term vegetation dynamics introduced by local resource stress. In 72% of all cases OTIs out-performed the original canopy reflectance spectra as indicators of plant species composition. OTIs are also easier to interpret in an ecological sense than spectral bands or features. We thus conclude that optical traits retrieved from reflectance data have a high indicative value for ecological research and applications.