Estimation of herbaceous biomass from species composition and cover

Abstract Questions Biomass is an important ecological property, but its measurement is destructive and time‐consuming and therefore generally missing for historical vegetation plots. Here we propose and test indirect estimation of herbaceous biomass using models based on easily obtainable variables, namely plant height and cover. We compare these models with Ellenberg indicator values for nutrients ( EIV s Nutrients ), which are sometimes used as an alternative measure of productivity. Location Czech Republic, western Slovakia. Methods Above‐ground biomass (dry weight; g·m −2 ) was regressed against the following explanatory variables: (1) Cover E 1 , total percentage cover of the herb layer visually estimated in the field; (2) Biomass estimate‐raw, ‐adjusted and ‐median , calculated from plant covers and heights (according to a local flora); and (3) mean EIV s Nutrients calculated per plot. For the analyses, we used four data sets containing a total of 469 plots from different vegetation types: ‘Wet meadows’, ‘Dry grasslands’, ‘Fen–dry grassland transects’ and ‘Forest herb layer’. To test the applicability of different biomass estimates we chose an example of a species richness–productivity relationship in the ‘Wet meadows’ data set and describe differences in resulting patterns. Results Both cover of herb layer and calculated ‘biomass volumes’ were more accurate in predicting biomass dry weight than EIV s Nutrients . The best results were obtained from the Biomass estimate‐median model that combines median stand height and total cover of the herb layer. Cover E 1 showed relatively tight correlations with biomass, particularly in sparse vegetation, but was a rather poor predictor when cover values were high. This was especially noticeable in application of the Cover E 1 model in analysis of the species richness–productivity relationship. Conclusions In contrast to biomass, cover of the herb layer has a fixed upper limit (100%), which may lead to misinterpretations in dense, structurally diverse vegetation. Most promising is the Biomass estimate‐median method, which can be applied both to already sampled plots by calculating median height from average species heights according to local floras and to newly sampled plots using the median of plant heights measured in the field. Therefore, we propose it as a rapid, non‐destructive alternative to biomass harvest.