Elsevier

Biological Conservation

Volume 107, Issue 1, September 2002, Pages 99-109
Biological Conservation

A general method for measuring relative change in range size from biological atlas data

https://doi.org/10.1016/S0006-3207(02)00050-2Get rights and content

Abstract

Conservationists need information on trends in the ranges of species in order to set priorities and to assign threat statuses using IUCN criteria. For many taxonomic groups in many countries, the best data available for this purpose come from atlases that map the distributions of species in grid cells. This paper presents a method of calculating an index of relative change in range size using atlas data for two periods. The method allows for the effect of variations in the geographical coverage and intensity of recorder effort. A weighted linear regression model is fitted to the relationship between counts of grid cells (as logit-transformed proportions) in the earlier and later periods. The standardised residual for each species provides an index of its relative change in range size. The method is illustrated by analyses of three British data sets. Previous methods of assessing change from atlas data are briefly reviewed and the advantages and limitations of the new method are discussed.

Introduction

Species threat statuses have played a crucial role in conservation. The publication of threat statuses in Red Data Books allows conservationists to set priorities, and target limited resources with the greatest effect in the battle against extinction.

Rarity was once the sole criterion used to allocate species status, but increasingly there is a focus on decline. This trend can be seen in successive editions of the British Red Data Book for vascular plants. In the first edition (Perring and Farrell, 1977), species were selected for inclusion if they had been recorded from 15 or fewer ‘10-km squares’ (10×10 km cells of the British national grid). The latest edition (Wigginton, 1999) uses the revised threat categories of the IUCN (1994), which have since been further revised (IUCN, 2001). The IUCN threat categories are based on five quantitative criteria, three of which include an assessment of decline in either range or population size.

For most of the biota of even the best-recorded countries, monitoring data do not exist with which to assess declines. However, biological atlas data sets do exist for many groups, and in Britain these provide the main source of data on the conservation status of the majority of the biota (Harding and Sheail, 1992). These data sets generally include the results of at least one national survey, plus collated historical records, and are used to produce distribution maps (e.g. Fig. 1). A general method for quantifying changes in range size from such data sets is needed.

Various ‘change statistics’ have been calculated from biological atlas data. The biases in biological atlas data are widely recognised (Rich, 1998, Dennis et al., 1999), typically including an increase in survey effort over time. However, existing methods to quantify changes in range size usually fail to compensate either for change in the geographical coverage of recording, or for change in recording intensity. Here we present a new method that we have developed to minimise such biases. Although the method proposed here can be used to compare the results of a single survey against accumulated historical data, it is particularly suited to the analysis of distributional data for groups which have been surveyed by comparable methods in two periods, for example the successive surveys of butterflies (Heath et al., 1984, Asher et al., 2001) and vascular plants (Perring & Walters, 1962, Preston et al., in press). After describing the new method and outlining the assumptions it makes, we then present results for three representative distributional data sets.

Section snippets

The general method for calculating relative change in range size

In outline, the method is first to define the set of grid cells that have been surveyed in both survey periods; subsequent calculations are based solely on these cells. For each species, the number of recorded grid cells is counted for each period. These counts of grid cells are expressed as proportions of the total survey area, and then logit-transformed. A linear regression model is then fitted to the relationship between counts of grid cells in earlier and later survey periods. A weighted

Application of the method to three representative datasets

Note that the botanical nomenclature used here follows Stace (1997), and that the nomenclature of other taxonomic groups follows the publications from which examples are drawn.

Calculating the index

The index of relative change in range size was calculated using data sets detailed in Table 1. All calculations were performed using the statistical package Minitab (release 13). Range sizes (numbers of occupied 10-km squares) were counted for each species in each time period. Scatter plots of range sizes in the early and later periods (Fig. 2 a–c) reveal approximately linear relationships between range sizes for all three taxonomic groups. This linear relationship implies that range sizes are

Correlates of change

To demonstrate the potential uses of the change index in detecting correlates of change, three examples are taken from the data sets analysed earlier.

Comparison with other methods of assessing change from atlas data

Previous numerical assessments of the change in the distribution of British species have usually been derived from atlas data by a direct comparison of the numbers of 10-km squares occupied in different recording periods. Gibbons et al. (1993) tabulated the number of 10-km squares in which breeding birds had been found in two recording periods (1968–1972 and 1988–1991), expressing the more recent value as a percentage of the earlier value to obtain a ‘% change’. Of species which were recorded

Acknowledgements

We thank Henry Arnold for preparation of Fig. 1; David Roy and Mark Hill for commenting on drafts of this paper; David Pearman, Martin Luff, Mark Williamson, Petr Pyšek, Vojtı̀ch Jarošik and Phil Shaw for useful discussions. The data sets used are the result of many years of work by volunteer experts to whom we are indebted. We have been encouraged by the interest shown by the Department for Environment, Food and Rural Affairs and the BSBI in the application of this method to the analysis of

References (41)

  • P.D. Carey

    Changes in abundance of Himantoglossum hircinum (L.) Sprengel (Orchidaceae) over the last 100 years

    Watsonia

    (1999)
  • D.R. Cox

    The Analysis of Binary Data

    (1970)
  • R.H.L. Dennis et al.

    Bias in butterfly distribution mapsthe effects of sampling effort

    Journal of Insect Conservation

    (1999)
  • K.J. Gaston

    Rarity

    (1994)
  • D.W. Gibbons et al.

    The New Atlas of Breeding Birds in Britain and Ireland: 1988–1991

    (1993)
  • P.T. Harding et al.

    The biological records centre: a pioneer in data gathering and retrieval

  • J. Heath et al.

    Atlas of Butterflies in Britain and Ireland

    (1984)
  • IUCN Red List Categories, version 2.3. Prepared by the IUCN Species Survival Commission

    (1994)
  • IUCN Red List Categories, Version 3.1. Prepared by the IUCN Species Survival Commission

    (2001)
  • K.J. Kirby et al.

    Seasonal and observer differences in vascular plant records from British woodlands

    Journal of Ecology

    (1986)
  • Cited by (97)

    • The temporal dynamics of a regional flora-The effects of global and local impacts

      2015, Flora: Morphology, Distribution, Functional Ecology of Plants
    View all citing articles on Scopus
    View full text