Molecular Ecology (2004) 13, 2655–2666 doi: 10.1111/j.1365-294X.2004.02281.x © 2004 Blackwell Publishing Ltd Blackwell Publishing, Ltd. Partitioning nuclear and chloroplast variation at multiple spatial scales in the neotropical epiphytic orchid, Laelia rubescens DORSET W. TRAPNELL * and J. L. HAMRICK † *Department of Plant Biology, University of Georgia, Athens, Georgia 30602, †Departments of Plant Biology and Genetics, University of Georgia, Athens, Georgia 30602 Abstract Insights into processes that lead to the distribution of genetic variation within plant species require recognition of the importance of both pollen and seed movement. Here we investigate the contributions of pollen and seed movement to overall gene flow in the Central American epiphytic orchid, Laelia rubescens. Genetic diversity and structure were examined at multiple spatial scales in the tropical dry forest of Costa Rica using nuclear (allozymes) and chloroplast restriction fragment length polymorphism ( RFLP) markers, which were found to be diverse (allozymes, P = 73.3%; H E = 0.174; cpDNA, H E = 0.741). Nuclear genetic structure (F STn ) was low at every spatial scale (0.005 – 0.091). Chloroplast markers displayed more structure (0.073 – 0.254) but relatively similar patterns. Neither genome displayed significant isolation-by-distance. Pollen and seed dispersal rates did not differ significantly from one another (m p /m s = 1.40) at the broadest geographical scale, among sites throughout Costa Rica. However, relative contributions of pollen and seeds to gene flow were scale-dependent, with different mechanisms determining the dominant mode of gene flow at different spatial scales. Much seed dispersal is highly localized within the maternal population, while some seeds enter the air column and are dispersed over considerable distances. At the intermediate scale (10s to 100s of metres) pollinators are responsible for substantial pollen flow. This species appears capable of distributing its genes across the anthropogenically altered landscape that now characterizes its Costa Rican dry forest habitat. Keywords: allozymes, cpDNA, gene dispersal, genetic structure Received 26 January 2004; revision received 19 May 2004; accepted 16 June 2004 Introduction Plants accomplish gene dispersal during two independent life cycle stages, via seeds and pollen. Successful pollen and seed immigration and establishment into extant popu- lations constitute gene flow events while seed dispersal and establishment into vacant sites constitute colonization ( Levin 1981; Hamrick & Nason 1996). Because all natural plant populations originate in this manner, colonization is an important process shaping the genetic structure of species ( Harper 1977; Slatkin 1977; Wade & McCauley 1988). Recognition of the two gene dispersal stages in plants is critical to understanding the distribution of genetic variation within and among populations (Levin 1981; Whitlock & McCauley 1990). Both pollen and seeds disperse biparentally inherited nuclear genes while maternally inherited gene movement depends exclusively on seed dispersal. In most angiosperms, the nonrecombinant chloroplast genome (cpDNA) is inherited maternally (Sears 1980; Corriveau & Coleman 1988; Mogensen 1996; Ennos et al . 1999). By comparing the distribution of genetic markers derived from nuclear (nDNA) and chloroplast genomes, the relative influence of pollen and seed dispersal on gene flow and genetic structure can be determined (Ennos 1994; McCauley 1995). If pollen and seed dispersal rates differ, cpDNA and nDNA should reveal different levels of spatial genetic structure ( McCauley 1994; McCauley et al . 1996). Thus, the distribution of genetic diversity for these two distinct genomes can be used to infer the processes that produce genetic structure. While an increasing number of studies examine the relative contributions of pollen and Correspondence: Dorset W. Trapnell. Fax: 706 542 1805; E-mail: dorset@plantbio.uga.edu