and methodological links see Stoyan and Penttinen 2000 Dale et al 2002 Using

And methodological links see stoyan and penttinen

This preview shows page 19 - 21 out of 24 pages.

and methodological links (see Stoyan and Penttinen2000; Dale et al.2002). Using these different ap-proaches alongside each other is likely to provefruitful, and allow data to be explored from mul-tiple perspectives.Local or global?Localpointpatternanalysismethodshavebeenrelatively little used by ecologists, with most pub-lished research tending to present the results of asingle global statistic. However, the use of localpointpatternanalysesseemstohavemuchtocommend it. It may be incorrect to assume that theresults of a global description of spatial patternapply equally to all parts of the study area, anduseful insights can be gained by exploring spatialvariations in the results. Local variations may besubsumed in global descriptions of spatial pattern;insomecases(e.g.epidemiologicalstudies)theglobal characterisation of pattern would seem to beat odds with the aim of identifying localised clus-tering(Fotheringham1997).AsFotheringham(1997pp. 88–89) comments ‘‘simply reporting one‘average’ set of results and ignoring any possiblevariations in those results is equivalent to reportingameanvalueofaspatialdistributionwithoutseeing a map of the data’’. As demonstrated above,patternsthatapproximateCSRonaveragemayshowsegregationandaggregationindifferentlocationsatdifferentscales(Figs.6,7).Recentcompetition theory emphasises spatial interactionsand the ‘plant’s-eye view’ (Purves and Law2002),andconsiderabletheoreticalresearchhashigh-lighted that mean-field approaches are not alwaysappropriate for describing ecological processes (e.g.see Amarasekare2003; Bolker et al.2003). Thesameseemslikelytobetruewhendescribingecological patterns.Another advantage with both the localised form ofRipley’sKand the SADIE analyses is that theyprovide spatially explicit visual information about thenature of the pattern of interest. Such visualisation isespecially important in the EDA context in whichmany spatial analyses occur in ecology. Althoughthere are criticisms of ‘pure’ visualisation in that (i)the way a map is presented can influence a viewer’sconclusions (e.g. use of class intervals in chloroplethmaps) and (ii) the brain tends to see patterns evenwhere they may not exist (e.g. CSR patterns arefrequently described as clustered by viewers), whenusedobjectivelyalongsidequantitativemeasures,visualisation is a useful way of exploring trends inspatial data (see Unwin1996; Fotheringham1999).Plant Ecol (2006) 187:59–8277123
Using local analyses, in conjunction with appropriatevisualisation tools, variations within plots in aggre-gation can be seen and described in a way not pos-sibleusingsimpleglobalstatisticalmeasures(Wilhelm and Steck1998).

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture