Valuable habitat and low deforestation can reduce biodiversity gains from development rights markets

Abstract

Illegal private land deforestation threatens global biodiversity, even in areas with native habitat requirements stipulated by law. Compliance can be improved by allowing landholders to meet legal reserve requirements by buying and selling the rights to have deforested land through a Tradeable Development Rights system (TDR). While this policy mechanism may prevent native habitat area loss, the spatial pattern of reserved areas will shift, creating novel landscape patterns. The resulting altered fragmentation and connectivity of habitat will impact biodiversity. TDR may also allow landholders to earn rent on land they never intended on converting, resulting in additional deforestation elsewhere and net habitat loss. We construct a simulation model to explore the potential implications for biodiversity when development rights can be traded, compared with the landscape resulting from enforced individual compliance with deforestation laws. We find that where future deforestation is very likely, a TDR market can provide better outcomes for both biodiversity and agriculture, resulting in more connected habitat networks with larger fragments and fewer edge effects. However, the TDR market can be harmful if future deforestation is unlikely, or if one habitat type is tightly spatially correlated with high economic returns from agriculture. Policy implications. Allowing landholders to buy and sell the rights to keep more cleared land than legally stipulated will result in transformed multiuse landscapes. Losses of native habitat in some areas will be offset in others. We conclude that trading forest development rights has the potential to improve habitat configurations, but that careful consideration should be given to current species distributions and likely future deforestation scenarios.

Interpolating missing land cover data using stochastic spatial random forests for improved change detection

“…3). We performed this on a rectangular section of the image to mimic a row of land holder plots or grids (Helmstedt & Potts, 2018).…”

“…To create missing data we simulated cloud patterns independently of the land cover data. We used a distance grid with a multivariate distribution and spatial autocorrelation structure that decays with distance following the process described in Holloway et al (2019) and Helmstedt & Potts (2018). We overlaid the simulated cloud with our land cover images using the Raster package (Hijmans, 2017) and deleted the land cover information for the regions under the clouds.…”

Stochastic spatial random forest (SS-RF) for interpolating probabilities of missing land cover data

“…In order to produce our own missing data, we simulated missing data in cloud patterns on the images t + 1 and t + 2 . We simulated the missing data patterns independently of the NDVI based forest presence data using the process described in [16] based on [33]. We applied the SS-RF method to the same pixels and neighbourhoods in each image, identified by their geographical location (longitude and latitude), to ensure we were examining the same observations over time and could make fair assessments of model performance when interpolating the values at future time points t + 1 and t + 2.…”

A Decision Tree Approach for Spatially Interpolating Missing Land Cover Data and Classifying Satellite Images

“…A simulated cloud mask was applied to random areas of the image to intentionally create 'missing' spectral data. The cloud-like shapes are simulated independently of the FPC data, while using a distance grid and applying a multivariate normal distribution, which is spatially autocorrelated with a covariance that exponentially declines with distance [32]. The simulated cloud layer was overlaid on the FPC layer and the corresponding pixels under the 'clouds' had their spectral information that was removed in the FPC layer.…”