Using Spatial Interpolation to Quantify Plant Occupancy

Abstract

ð The grid square occupancies of vascular plants are estimated for the periods 1978–1994 and 1995–2011 across Great Britain. ð This was achieved by selecting well-surveyed 4 km grid squares from the pool of all records and deriving detection/non-detection data from them. From these it was possible to extrapolate occupancy probabilities to the whole area. ð Grid squares were selected by setting a minimum threshold for the recording effort for each grid square based upon the number of species recorded and the number of surveys conducted. The spatial distribution of plants was modelled for these selected sites using variography and kriging was used to interpolate across the whole area. This was conducted for the two time periods and differences in the occupancy probability were compared. ð This approach has several advantages over other methods. • Recording effort is assessed and can be compared between time periods and regions. • Absolute occupancy change can be estimated, rather than just relative change. • The differences in the spatial distribution pattern of each species are accounted for . • The occupancy probabilities and their variance can be mapped. ð An important result was that in England each 4 km gained on average one additional neophyte every three years. Figure 1. Mapping the occupancy probabilities and its change of Campanula rotundifolia in Great Britain. a) The combined kriging variances from both time periods. b) The occupancy probability distribution from data between 1995–2011, red indicates high probability of presence. The arrow indicates a discontinuity at the boundary between the data partitions for southern and northern England. c) The change in occupancy probability of between 1978–1994 and 1995–2011. Red indicates an increase in occupancy probability and yellow indicates a decline in occupancy probability. Clean taxonomy, dates and grid references of records Remove duplicate records Select well surveyed grid squares from the pool of all surveyed grid squares Convert records into detection/non-detection data for each species Model the spatial distribution of occupancy using variography Use the data and variograms in kriging to estimate the occupancy probabilities of each species across the whole region. Divide the country into regions with a similar landscape Method Figure 2. For comparison, a presence-only map of Campanula rotundifolia from beween 1995–2011 mapped with 100 km grid squares. Variograms for Modelling spatial autocorrelation. This is an example of a variogram of the distribution of Campanula rotundifolia in southern England using records between 1995 and 2011. By summing the change in occupancy probabilities you can examine the change associated with particular groups of plants. These are the summed changes in occupancy probabilities for neophytes, archaeophytes and natives comparing 1978–1994 to 1995–2011. Conclusions ð Interpolation can be used to quantify occupancy of imperfectly surveyed areas. ð It can be used to quantify occupancy across whole counties, but can also identify local areas of change. Literature Groom, Q.J. (2013) Estimation of vascular plant occupancy and its change using kriging. New Journal of Botany 3(1) in press.