[ Login ]

Biogeographical Affinity  

Biogeograpical affinity is sometimes treated separately from similarity in ecological studies. An author may report similarity between (let us say) lists of taxa from a number of study areas; and, also, measure the affinity between the study area lists and a set of "canonical" lists representing well known geographical regions more or less remote from the study areas geographically. The argument of Tulloss (1997) justifying the tripartite similarity metric (index, coefficient) appears to have converted some authors away from using other similarity coefficients analyzed in that paper as well as in Hayek (1994). On the other hand, such an author may use Simpson's Coefficient for measuring biogeographical affinity in a paper in which the tripartite similarity metric is used for computing similarity. Do similarity and biogeographical affinity require two different solutions? It all depends on what "affinity" means, and this meaning is in the hands of a paper's author. One could say that "affinity" is measured by the percentage of taxa in a study area that are also taxa in a given canonical list of taxa. Then, as long as the list of taxa in the study area is smaller than the canonical list being compared, Simpson's Coefficient will give the desired result. However, if the area of study is very diverse and has a list larger than a canonical list, one may well get an undesired result from Simpson's Coefficient -- if a given canonical list is smaller thant the list for which affinity is being computed, Simpson's Coefficient will yield not the percentage of species in the latter list that are also in the canonical list, but the percentage of species in the canonical list that are in the study area list. This is because the denominator of Simpson's Coefficient is the size of the smaller of the two lists being compared. In other words, there is potential for ambiguity in the interpretation of a percentage output from a computation of Simpson's Coefficient. Hence, if "affinity" were intended to mean the percentage of taxa in a sample list that also occur in a canonical list, it would appear that it is more reasonable to simply compute that percentage than to use Simpson's Coefficient with it's possibility for ambiguity. In fact, as observed by Tulloss (1997), the number of taxa in the larger of two compared lists plays no role whatever in the computation of Simpson's Coefficient except that of being eliminated from consideration. To go a bit further in our discussion: Let us assume a list from a given study area contains 120 taxa. We are going to compare this to two canonical lists. One of these lists contains 300 taxa; the other contains 3,000 taxa. In the first case, the study area includes 50 of the taxa in the list of 300. In the second, the study area includes 50 taxa in the list of 3,000. The biogeographical affinity (computed as a percentage of the study area list) to the regions represented by the canonical lists will be the same in both cases: 50/120 or 42%. On the other hand, the tripartite similarity indices of the study area list in comparison to the two canonical lists will be significantly larger in the case of the list of 300 than in the case of the list of 3,000. Hence, assuming that that the meaning of "affinity to biogeographical region X" is "percentage of taxa in the present list that appear in the canonical list for X," biogeographical affinity with region X is in fact something distinct from similarity to canonical list X. Secondly, the Simpson Coefficient is both an unduly complex tool for computing affinity so-defined and is potentially ambiguous as described above. Closing thought: It appears useful to present affinity to a canonical list and similarity to a canonical list together. The second index provides additional interpretative value for the percentage yielded by the first computation.

© 2003, Amanita-Bear Consulting.
- Last modified: 06/15/2004