The great taxonomic diversity of the Orchidaceae is often attributed to adaptive radiation for specific pollinators driven by selection for outcrossing. However, when one looks beyond the product to the process, the evidence for selection is less than overwhelming. We explore this problem by discussing relevant aspects of orchid biology and asking which aspects of reproduction explain the intricate pollination mechanisms and diversification of this family. We reaffirm that orchids are primarily pollination limited, the severity of which is affected by resource constraints. Fruit set is higher in temperate than in tropical species, and in species which offer pollinator rewards than those that do not. Reproductive success is skewed towards few individuals in a population and effective population sizes are often small. Population structure, reproductive success and gene flow among populations suggest that in many situations genetic drift may be as important as selection in fostering genetic and morphological variation in this family. Although there is some evidence for a gradualist model of evolutionary change, we believe that the great diversity in this family is largely a consequence of sequential and rapid interplay between drift and natural selection.

Reduction in the number of pollinator species per plant species is a mechanism that may lower the cost of pollen transfer. Using efficient pollinators may have an evolutionary significance. It is hypothesized that an evolutionary trend from many pollinators to few pollinators per plant species should be observable when species from ancestral versus recently derived monophyletic taxon are compared. Three different orchid phylogenetic sequences are used; two of the phylogenies show a reduction in the number of pollinator species per orchid species from the most ancestral to the most recently derived subfamilies. The third classification did not show this trend. It is thus possible to observe macroevolution of pollinator specialization of a monophyletic plant taxon. Key words: evolution, pollination, systematics, Orchidaceae, evolutionary ecology.

No abstract

S U M M A R YLittle is known about non-mycorrhizal endophytic fungi in tropical orchids; still less is known about how endophytes vary within and between individual orchid plants, b'ungal endophytes were isolated from roots and leaves of epiphytic and lithophytic orchids in the genus Lepanthes; seven species, from rainforests in Puerto Rico, were sampled. The endophytes observed most frecjuently were Xylaria species and Rhizoctonia-\'\ke fungi, found in 29 % of roots and 19 % of leaves, and 45 "o of roots and 31 "" of leaves, respectively. Five deuteromycete genera were also isolated, occurring in 19 "" of roots and 43 ",, of leaves (combined). At least nine species oi Xvlaria were found, with several species sometimes occurring in a single plant. Differences between roots and leaves in frequency of Xylaria and Rhizoitoiiia isolates were not significant, although differences among orchid species in number and types of endophytes were. Heterogeneity of endophytes in single plants and plant organs was greater than diflerences between species. Many Lepanthes species are very restricted in distribution, and knowledge of their interactions with endophytes might be useful in species management.

Genetic drift can play an important role in population differentiation, particularly when effective population sizes are small and gene flow is limited. Such conditions are suspected to be common in the species-rich Orchidaceae. We investigated the likelihood of genetic drift in natural populations of three endemic species of Lepanthes (Orchidaceae) from Puerto Rico. We estimated effective population size, Ne, using three ecologically based methods. Two of the three estimates were based on variance in reproductive potential and the third was based on coalescence time. All estimates of Ne were usually <40% of the standing population size, resulting in values of <20 individuals per population. Based on starch gel electrophoresis of isozymes, Nm estimates suggest restricted gene flow among populations in the range of one or less successful migrant per generation. Genetic differentiation among populations is expected under such conditions from random genetic drift. Indeed we observed high genetic differentiation among populations (L. rubripetala, FsT, G , 8; 0.248, 0.266, 0.293; L. rupestris, 0.148, 0.169, 0.138; L. eltomensis, 0.251, 0.219, 0.218, respectively). Genetic drift is likely to be important for population differentiation in Lepanthes as a result of small effective population sizes and restricted gene flow. 0 2001 The Linnean Society of London ADDITIONAL,