Randomness in Allopatric Speciation

Abstract

Anderson, S., and M. K. Evensen (Department of Mammalogy, The American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024) 1978. Randomness in allopatric speciation. Syst. Zool. 27:421-430.-Data from North American terrestrial vertebrates are used to test hypotheses of randomness in several aspects of allopatric speciation (vicariance). The data do not lead to rejection of the following: the initial parapatry of vicariance tends toward randomness as time passes; barriers are random as to affected species (the congruence of barriers is unusual); ranges of geminate species (a pair formed by vicariance) and subspecies (in species with only two) are random in size. The data lead to modification of the following: barriers that lead to subspecific differentiation arise at random places. [Allopatric speciation; vicariance; randomness; North American mammals.] A species may become two species if a barrier divides the original range and if changes occur in one or both of the segments that are sufficient to make them distinguishable. The degree of distinctness required to make them distinguishable is an interesting subject, but is peripheral to the subject here, as is the criterion of noninterbreeding in nature, another consideration included in various pragmatic and theoretical definitions of species. The process of speciation referred to above is called allopatric speciation or vicariance. The term allopatric speciation has been used in a more restricted sense than this by Endler (1977). The term vicariance has been used with other meanings also; for example, the differentiation that leads to the geminate species or the origin of the barrier that leads to the differentiation (see Croizat, Nelson, and Rosen, 1974:277). e do not use the term in either of these two ways. We raise the question whether any measureable aspects of the process of allopatric speciation occur at random, or to what degree, or in what ways they seem to occur at random? The concept of randomness is not absolute; it can be defined only in reference to a particular model. A model which took into consideration all possible shapes of animal ranges and barriers dividing those ranges would be impossibly complex. In fact, even with the simplifying assumptions that all ranges are circular and all barriers straight lines, there are three plausible models of a random division of those ranges (the interested reader is referred to Bertrand's Paradox [Papoulis, 1965]). We chose the brokenstick model (MacArthur, 1957; Anderson, 1974) because of its simplicity, familiarity to the reader, and the availability of computational techniques. The concept of randomness and the appearance of randomness in sets of data are of interest because they provide a useful background (the of statistics). Unless results depart significantly from the null hypothesis and the null hypothesis can be rejected, the results do not call for further explanation. Most, if not all, speciatioon in sexually reproducing organisms occurs by vicariance. At least, that is the prevailing view and we have no quibble with' it now. Views on modes of animral speciation have been summarized recently by Bush (1975) and Endler (1-977). The questions we address here pertain to possible randomness in various aspects of allopatric