Patterns of evolutionary and geographical distribution in neotropical orchids

Tropical material offers special advantages for the study of evolution, since the evolutionary process has progressed in a more continuous form than in the north temperate region, not having been seriously interrupted by glaciations, nor being restricted in space. The members of the orchid family offer special advantages (a) since the family is one of the largest and probably a most recent one phylogenetically, with the taxonomy well advanced and (b) since they have a peculiar type of geographical distribution, the basic pattern, developed through slow expansion of areas, being overlaid by a secondary pattern resulting from long-distance dispersal. The problem, which will be discussed, refers to the relative importance of sympatric-central or sympatric-marginal disruptive evolution as compared to an allopatric one. Two groups of species were taken into consideration, representing two different types of phylogenetic evolution and geographic distribution. The species of bifoliate Cattleya in the Atlantic and central regions of Brazil form several pairs or chains of vicarious species with contiguous, but not overlapping distribution, and a certain overlapping occurs only in the species, C. loddigesii, and then on a subspecific level. The regularity of this behaviour seems to indicate strongly that either the species have evolved by disruptive-central (sympatric) diversification, one original species breaking up into two or more vicarious ones, or by disruptive-marginal diversification, one species changing into another at and beyond its adaptive range. C. loddigesii may thus be considered as an early stage in such a development. In the genus Oncidium only one group of species was mentioned with a very high degree of strictly sympatric diversification. The third group is formed by species of the Maxillariinae, one belonging to Trigonidium and the others to Maxillaria. One set of three species shows an extreme range of distribution:M. rufescens and M. cleistogama in nearly all the American tropics except the Amazon Basin, and T. acuminatum in the Amazon Basin and neighbouring areas. The other set was composed of the 15 species of the Sect. Pictae, with partially sympatric areas in southern, central and eastern Brazil only. The study required the use of statistical methods, and a form of sequential analysis of variance was applied. The whole area of distribution of a species was divided into regions; within regions there were the localities from which material came and the next units were the individual plants collected. The residual error of this sequence was represented by the variance of phenotypic variation between flowers within plants. If diversification should be stimulated by allopatric separation, then the error components would diminish in the sequence given, whereas if sympatric evolution is frequent, the ‘within' errors between regions or localities need not be different. In order to minimize the phenotypic ‘within-plant' variability, the specimens were first maintained under homogenizing conditions. The analysis was carried out for ten species, and in three an additional component was separated referring to indications of infraspecific taxonomic divisions. For the very uniform M. cleistogama the data show that there is no indication of a genetical intrapopulational variability and that geographical separation between localities in Venezuela, Peru and eastern Brazil also caused no significant increase in variability. In the case of the other two species, T. acuminatum and M. rufescens, genetic variability between plants was quite significant, and so was variation between localities. But in both species intraspecific separations were represented by the highest error, and showed a correlation with geographical distribution. This in fact was to be expected, since the vast area of distribution included different floristic-ecological areas and thus some allopatric evolution was bound to occur.