DISCUSSION:

The lack of separation in any of the dendrograms represents a lack of genetic structure at the population level relative to the entire data set. The chicozapote is a highly variable, outcrossing plant species (Pennington 1990). Many researchers have found that populations of outcrossing woody plants typically hold more genetic variation within populations than between populations (Hamrick and Godt 1996). Erstad (1996) studied ecotypes of Ribes rubrum in Norway and found more variation within populations than between populations in outcrossing woody species. In a study of four tropical tree species in Costa Rica, Schierenbeck et al. (1997) found that these species exhibited much genetic variation between individuals, but found no population genetic differences along an elevation gradient. They hypothesized that tropical trees in general have large gene pools, possibly as a result of effectively widespread pollen and seed dispersal. M. zapota is believed to be pollinated by bats (Pennington 1990,1991) and the large, sweet fruits are dispersed by birds and other large animals (Klimstra and Dooley 1990). This long distance dispersal could explain why there is no observed genetic structure at the population level, because gene flow tends to homogenize population genetic structure.

However, results of the Fisher's exact tests and the analysis of the distance-data matrix show that the genetic structure in the El Eden swamp and forest populations is significantly different from what would be expected of random variation. A possible explanation of this difference can be hypothesized from the matrix data. The mean distance between forest individuals is very close to the mean of the forest x swamp distances, while the swamp x swamp mean is lower. This could indicate that the swamp population is actually a sub-set of a larger gene pool that includes the forest-dwelling individuals. Certain genes may be required to survive in the inundated conditions of the swamps, which would mean that the genotypes of the individuals in the swamp have been selected from the gene pool of the population as a whole. This would mean that individuals in the swamp would be more similar to each other than to the population as a whole, leading to the observed lower mean distance.

The situation observed in the El Eden populations may represent an intermediate stage in the formation of an ecotype of M. zapota. These trees require from five to eight years to reach reproductive capability. This slow growth, combined with the outcrossing nature of the species (Pennington 1990), would mean that it would take many years for adjacent populations to diverge. The obvious morphological differences between the swamp and forest forms of the tree suggest that a large selective pressure does exist in these habitats, and difference in genetic diversity revealed by this study suggest that the pressure may be having an effect.

An alternative explanation for the observed morphological differences between swamp and forest populations of M. zapota is that the differences are the result of extreme phenotypic plasticity of the species. The chicozapote can grow in various habitats around the world, and has shown enough variation in form to spawn fifty-seven synonyms (Pennington 1990). The wide distribution of chicozapote throughout southern Mexico and Central America indicates that tree has enough plasticity to survive in various habitats, and phenotypic plasticity has been touted as a method of extending the range of a species (Sultan 1987).

Future studies on this plant should include reciprocal transplants of swamp and forest individuals, as well as common gardens of plants grown from seed. These additional experiments would give direct evidence of a presence or lack of a genetic correlate to the observed morphological variation between these populations. Unfortunately, due time constraints, the slow growth of the species, and the remote locations of the sample populations, these experiments were not included in the current study.

In addition, germplasm collections need to be made of the different forms of M. zapota, including the forest and swamp types, as well as the different cultivated varieties. In India, which has the largest amount of cultivation of chicozapote, new varieties only arise by mutation, since all trees are propagated vegetatively (Chadha 1992). A controlled breeding program would be very useful for stabilizing traits of interest and developing new varieties. Finally, agricultural studies should be conducted with the different morphological types in the swamps and other marginal habitats of the Yucatan peninsula, to determine the feasibility of chicozapote as a crop plant in these areas.