We studied the resilience of southeastern Norwegian old-growth Picea abies forest floor vegetation to experimental disturbance. Five treatments, differing in depth of removal of vegetation and soil l...

Sets of presence records used to model species’ distributions typically consist of observations collected opportunistically rather than systematically. As a result, sampling probability is geographically uneven, which may confound the model's characterization of the species’ distribution. Modelers frequently address sampling bias by manipulating training data: either subsampling presence data or creating a similar spatial bias in non‐presence background data. We tested a new method, which we call ‘background thickening’, in the latter category. Background thickening entails concentrating background locations around presence locations in proportion to presence location density. We compared background thickening to two established sampling bias correction methods – target group background selection and presence thinning – using simulated data and data from a case study. In the case study, background thickening and presence thinning performed similarly well, both producing better model discrimination than target group background selection, and better model calibration than models without correction. In the simulation, background thickening performed better than presence thinning when the number of simulated presence locations was low, and vice versa. We discuss drawbacks to target group background selection, why background thickening and presence thinning are conservative but robust sampling bias correction methods, and why background thickening is better than presence thinning for small sample sizes. Particularly, background thickening is advantageous for treating sampling bias when data are scarce because it avoids discarding presence records.

Species composition is a vital attribute of any ecosystem. Accordingly, ecological restoration often has the original, or “natural,” species composition as its target. However, we still lack adequate methods for predicting the expected time to compositional recovery in restoration studies. We describe and explore a new, ordination regression‐based approach (ORBA) for predicting time to recovery that allows both linear and asymptotic (logarithmic) relationships of compositional change with time. The approach uses distances between restored plots and reference plots along the successional gradient, represented by a vector in ordination space, to predict time to recovery. Thus, the approach rests on three requirements: (a) the general form of the relationship between compositional change and time must be known; (b) a sufficiently strong successional gradient must be present and adequately represented in a species compositional dataset; and (c) a restoration target must be specified. We tested the approach using data from a boreal old‐growth forest that was followed for 18 years after experimental disturbance. Data from the first 9 years after disturbance were used to develop models, the subsequent 9 years for validation. Rates of compositional recovery in the example dataset followed the general pattern of decrease with time since disturbance. Accordingly, linear models were too optimistic about the time to recovery, whereas the asymptotic models provided more precise predictions. Synthesis and applications. Our results demonstrate that the new approach opens for reliable prediction of recovery rates and time to recovery using species compositional data. Moreover, it allows us to assess whether recovery proceeds in the desired direction and to quantitatively compare restoration speed, and hence effectiveness, between alternative management options.

Restoration of degraded ecosystems may take decades or even centuries. Accordingly, information about the current direction and speed of recovery provided by methods for predicting time to recovery may give important feedback to restoration schemes. While predictions of time to recovery have so far been based mostly upon change in species richness and other univariate predictors, the novel ordination‐regression based approach (ORBA) affords a multivariate approach based upon species compositional change. We used species composition data from four alpine spoil heaps in western Norway, recorded at three time points, to predict time to recovery using ORBA. This approach uses distances between restored plots and reference plots along a successional gradient, represented by a vector in ordination space, to model linear or asymptotic relationships of compositional change as a function of time. Results from ORBA were compared with results from models of more generic univariate attributes, that is total cover, species richness and properties of the physical environment as functions of time. ORBA predictions of time to species compositional recovery varied from less than 60 years with linear models to 115–212 years with asymptotic models. The long estimated time to recovery suggests that the restoration schemes adopted for these spoil heaps are likely to be suboptimal. Much shorter time to recovery was predicted from some of the more generic univariate attributes, that is species richness and total cover, than from species composition. Given the current rates of recovery, most spoil heaps will reach reference levels for total cover and species richness within 50 years, whereas predictions indicate that 67–111 years are needed to restore levels of soil organic matter and pH. Synthesis and applications. Species composition and soil conditions provide information of generally higher relevance for evaluation of ecosystem recovery processes than the most commonly used metric to assess restoration success, species richness. Species richness is decoupled from species compositional recovery, and likely to be a generally poor measure of restoration success. We therefore encourage further improvement of methods like the ordination‐regression based approach that use species compositional data to predict time to recovery.