Calculation of collated indices of abundance of butterflies based on monitored sites


Dr D. Moss, Institute of Terrestrial Ecology, Monks Wood, Abbots Ripton, Huntingdon, Cambridgeshire PE17 2LS.



  • 1The Butterfly Monitoring Scheme uses regular transect counts on fixed sites to establish annual indices of abundance of butterfly species in the United Kingdom. The annual change in collated index for each species has hitherto been calculated as a simple ratio between total counts summed over all participating sites.
  • 2A revised method for calculation of collated indices is proposed, which applies a logarithmic transformation to site values so as to downweight the influence on the index of the sites with greatest numbers of a species. Zero counts are handled comparably with non-zero values. An alternative method using geometric mean ratios is also examined.
  • 3Indices calculated using the logarithmic transformation are compared with those calculated using the traditional method. Two internal tests of dependency of collated indices on the sites with highest abundance are made. Another evaluation uses regression analysis for the effects of temperature and rainfall on collated indices of butterfly abundance.
  • 4The first internal test shows that logarithmic transformation reduces the dependence of the collated index on the sites of highest abundance for twenty-five out of twenty-seven species examined; while a second test shows a reduction for twenty-six out of forty-two species, and an increase for one species. The number of significant regressions on temperature variables increases with the use of the logarithmic transformation from 11% to 13% of tests made, and on rainfall from 10% to 11%.
  • 5The geometric mean ratio method introduces considerable biases in its treatment of zero values, for which a remedy is not available.
  • 6Although the revised collated indices calculated using logarithmic transformation are broadly comparable with traditionally calculated values, the reductions in dependency on sites with the most abundant populations render the revised indices less subject to perturbation due to local ‘noise’, and so more suitable for research on factors influencing butterfly abundance.