| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ecology |
Department of Plant Biology, 2512 Plant Sciences Building, University of Georgia, Athens, Georgia 30602 USA
Received for publication December 30, 2005. Accepted for publication August 30, 2007.
ABSTRACT
We examined the impact of habitat fragmentation on gene flow in populations of the neotropical tree Bursera simaruba. In particular, we compared the effectiveness of three common techniques to estimate gene flow in the context of a highly disturbed system. Paternity analysis on emerging seedlings from eight small (N = 3 to 9) stands of trees showed that between 45% and 100% of seedlings were sired from outside their stand, indicating pollen moved readily over the isolation distances examined. Based on six populations of 21–24 trees each, estimates of allozyme genetic diversity (Ps = 73.3%; He = 0.244) were higher than those reported for species with similar life history traits. Indirect, FST-based gene flow estimates for these six populations yielded an estimate of 3.57 migrants per generation, although possible violations of model assumptions limit the reliability of the estimate. A twogener analysis showed pollen moved either 320 m or 361 m and that there were only 2.46 effective pollen donors per maternal tree. Despite the potential for long-distance pollen movement, seed abortion was high, especially in stands with fewer than four trees. Population size, rather than isolation distance, appears to limit reproduction in the populations examined.
Key Words: Bursera simaruba • Burseraceae • dry forest • fragmentation • gene flow • paternity analysis • Puerto Rico • twogener
Habitat fragmentation can substantially alter demographic processes of a zoophilous plant species (Gibbs, 2001
), in large part through reductions in pollinator visitations (Fischer and Matthies, 1997
; Groom, 2001
; Knapp et al., 2001
). Reduced population size and lowered pollination visitation can increase genetic drift, augmenting the negative effects of inbreeding (Fischer and Matthies, 1997
; Ledig et al., 1997
; Kuang et al., 1999
), possibly limiting regeneration and increasing the risk of local extinction (Menges and Dolan, 1998
). Gene flow between fragments will reduce the extent of inbreeding and genetic drift and their genetic and demographic effects.
Despite the fundamental importance of gene flow, accurate estimates have proven elusive, and the development of improved techniques for gene-flow estimation is an ongoing effort. Numerous methods use genetic markers to estimate gene flow. One method infers the level of gene flow that would be needed to explain the spatial distribution of genetic diversity for a set of populations (Neigel, 1997
). Although yielding estimates of gene flow with relatively little work, these "indirect" approaches rely on assumptions that may not hold for fragmented environments, most notably that genetic divergence among populations by genetic drift is balanced by the homogenizing effect of gene flow (Wright, 1951
; Slatkin, 1993
). Additionally, if fragmentation lowers pollen movement among remnant populations, a new equilibrium between migration and drift may be established, but it could be several generations before this level is reached (Fig. 1; e.g., Neel and Ellstrand, 2001
). Until that time, gene flow estimates determined indirectly will be too high, and predictions of future genetic change may be in error.
|
Recent studies utilizing paternity analysis have demonstrated that insect-pollinated tropical trees may be capable of dispersing pollen for distances sufficient to reduce the threat of genetic isolation. Frequently, over 25% of a tree's seed crop is sired by pollen originating outside the tree's stand, with pollen often traveling more than 1 km (Hamrick and Nason, 2000
). Chase et al. (1996)
used highly variable simple sequence repeats (SSRs) to conduct a paternity-analysis of the hawkmoth-pollinated tropical tree, Pithecellobium elegans. They showed that pollen moved an average of 142 m, with the greatest measured movement being 350 m. Approximately 29% of the pollen came from outside the study area. In the putatively insect-pollinated tropical tree, Dicorynia guianensis, Latouche-Hallé et al. (2004)
found that 62% of the pollen came from outside of their 40-ha study plot, with pollen dispersal distances between parents often exceeding 1 km. White et al. (2002)
examined gene flow in the tree species Swietenia humilis among several forest fragments in Honduras. Highly variable SSRs allowed them to determine that between 24% and 100% of pollen traveled more than 900 m. In one case, an isolated tree received 71% of its pollen from distances greater than 4.5 km. Similarly, three species of Ficus in Panama received over 90% of their pollen from more than 1 km (Nason and Hamrick, 1997
).
Because of the high numbers of progeny that must be analyzed to obtain dependable estimates of gene flow rates and mating patterns, paternity analyses are often limited to a single population (Sork et al., 1999
). Furthermore, unless many highly variable loci are available, the ability to identify the actual pollen donor in large populations can be greatly compromised. Thus, only rarely are resources available to sample enough populations with a sufficient level of precision to address ecological factors that could impact gene flow, such as plant density or habitat type. As a result, Sork et al. (1999)
called for the development of new analytical protocols for the study of gene movement at the landscape level. In response, Smouse et al. (2001)
published the first of several papers on a novel gene-flow estimation procedure, dubbed twogener, which circumvents many of the assumptions of the indirect approaches, while not being as labor intensive as traditional paternity analyses. The technique performs an AMOVA on the genotypes of the male gametes inferred from offspring arrays of trees spread across a landscape. Similar to traditional indirect approaches, twogener infers levels of pollen movement from patterns of genetic similarity, although it does not require the assumption of migration–drift equilibrium.
Although fewer studies have been conducted with the more recent twogener technique, results suggest shorter pollen movement than has been seen with paternity analysis. Degen et al. (2004)
estimated mean pollen dispersal distances between 27 and 53 m in the bird-pollinated, tropical tree Symphonia globulifera. In the Amazonia tree Dinizia excelsa, Dick et al. (2003)
found mean pollen dispersal distances of 212 m in undisturbed forest and 1.5 km in an open ranch; a previous paternity analysis found that pollen moved up to 3.2 km between pasture trees (Dick, 2001
). These lower estimates of pollen movement are not unexpected, however, because twogener estimates "effective" pollen flow (Smouse and Sork, 2004
).
In this study, we estimated genetic diversity, mating system, and gene flow in fragmented populations of the neotropical tree Bursera simaruba (L.) Sarg. (Burseraceae). Three techniques were used for gene-flow estimation, and we discuss their relative merits and limitations. By evaluating estimates derived from the different techniques, we hoped to determine the role gene flow plays in allowing B. simruba to persist in a highly fragmented landscape. Using direct measures of gene flow, we determined the proportion of pollinations that involve immigrant pollen and attempted to identify populations for which pollen flow is minimal and consequently most at risk for loss of genetic diversity and possible extinction. We also estimated gene flow using an indirect, FST-based approach as well as a twogener analysis. Significantly, this is one of only a few studies to examine gene flow in a Caribbean plant community and thus adds to our understanding of these relatively understudied systems.
MATERIALS AND METHODS
All study populations were in southwestern Puerto Rico, near the Sierra Bermeja mountain range; different populations were used for direct and indirect measures of gene flow, but all populations were nested within the same general study area (Fig. 2). The region is classified as subtropical dry forest (sensu Holdridge, 1967
), receiving less than 1000 mm of rain per year. Much of the land has been cleared for agriculture and cattle pasture; remnant vegetation is found in scattered forest patches, in strips of vegetation along fencerows and ravines, and as isolated trees in fields. Forest clearing in Puerto Rico occurred primarily during the 19th century (Wadsworth, 1950
). The location of the study site, however, was one of the first to be colonized by the Spanish in the 16th century and so may have a longer disturbance history.
|
). It has small (5–7 mm diameter) green flowers that are pollinated by small flies, cerambycid beetles, and other small insects (Stevens, 1983
).
The bird-dispersed fruits are preferred by only a few bird species, most notably vireos (Greenberg et al., 1995
). Seeds ripen for 7–8 months on the tree (Becerra and Venable, 1999
), with the embryo rapidly filling the seed cavity in the final week (Stevens, 1983
). No obvious changes in external appearance accompany this change.
Estimates of genetic diversity
To obtain estimates of genetic diversity, we sampled six populations of 21 to 24 individuals (Fig. 2). The minimum distance separating any two populations was 2.5 km, and the maximum was 25.2 km. All populations had minimal forest cover, being embedded in a matrix of mostly graminoid vegetation, with some shrubs and small trees. Trees averaged 5–8 m in height and 10–20 cm in diameter, with a few as large as 30–40 cm in diameter. Aerial photographs from the Refuge population (site #5, Fig. 2) showed that the area was free of trees as late as 1960, indicating that the trees, which are typical in size for trees at other sites, can be no more than 40 yr old.
At least 20 cm2 of leaf tissue was collected from each tree. Leaves were shipped on ice to the University of Georgia within 48 h of collection. In the laboratory, leaves were crushed in a potassium phosphate extraction buffer (Mitton et al., 1979
), and the resulting extracts were stored on filter paper wicks at –70°C.
Horizontal starch gel electrophoresis was used to assay allozyme diversity. Three buffer and gel–electrode combinations, and 11 enzyme stains resolved 15 loci on 11.0% starch gels. Gel and electrode buffer recipes followed Soltis et al. (1983)
. Enzymes stained (and loci resolved) on System 4 (a Tris–citrate gel and tray buffer) were isocitrate dehydrogenase (Idh-1), shikimate dehydrogenase (Skdh-1), and UTP-glucose-1-phosphate (Ugpp-1). System 11 (a histidine gel and citric acid tray buffer) was stained for malate dehydrogenase (Mdh-1, Mdh-2, and Mdh-3) and 6-phosphogluconate dehydrogenase (6Pgd-1). System 6 resolved diaphorase (Dia-1), glutamate dehydrogenase (Gdh-1), menadione reductase (Mnr-1 and Mnr-2), phosphoglucoisomerase (Pgi-1 and Pgi-2), triose-phosphate isomerase (Tpi-1 and Tpi-2), and peroxidase (Per-1). Stain recipes were modified from Soltis et al. (1983)
except for diaphorase (Cheliak and Pitel, 1984
) and UTP-glucose-1-phosphate (Manchenko, 1994
). For enzymes with more than one locus, allozymes were numbered sequentially, with the lowest number assigned to the most anodal.
Standard measures of genetic variation were calculated at the population (subscript p) and species level (subscript S). Genetic diversity parameters estimate percentage of polymorphic loci (Pp and PS), the mean number of alleles per locus (A) and per polymorphic locus (AP), the effective number of alleles per locus (Ae), observed heterozygosity (HO), and expected heterozygosity (He). The degree to which trees within a population were related to each other was estimated by calculating the number of alleles in common among individuals (Surle et al., 1990
). Genetic structure was analyzed using Weir and Cockerham's (1984)
estimators of F statistics, calculated with the FSTAT software package (Goudet, 1995
). Fixation indices (FIS) for individual populations were calculated with the program Lynsprog, developed by M. D. Loveless and A. Schnabel (College of Wooster, Wooster, Ohio, USA), which provides 
2 values for the indices.
Direct measures of gene flow
Eight small, spatially isolated stands of trees, varying in degree of isolation from 80 m to more than 600 m, were chosen for analysis (Table 1). As in the six populations used to estimate genetic diversity, the degree of relatedness among trees was estimated by calculating the number of alleles in common among individuals (Surle et al., 1990
). Fruit production per tree was estimated by multiplying the average number of fruits on 4–5 branches by the total number of fruiting branches on a tree.
|
Estimates of outcrossing rates were computed with the program MLTR (Ritland, 2002
). Paternity exclusion analysis was used to estimate levels of gene flow into each study population. In short, each seedling's genotype was compared with those of the potential sires in the seedling's population. The percentage of seedlings that could not have been sired by any of the potential fathers yielded an initial estimate of "apparent" gene flow. Allozyme trials yielded an exclusion probability of 0.67 for B. simaruba, indicating that for a given seed, 67% of potential sires will be correctly identified as nonsires (Chakroborty et al., 1988
). The low number of potential pollen donors within each site (3–9) reduced the occurrences of paternal ambiguity with this exclusion probability. The possibility remains, however, that a father from outside the population could produce a pollen gamete indistinguishable from that created by a sire from within the population. In such a case, the resulting seedling would be falsely identified as having originated within the population, leading to an underestimation of gene flow. The sum of this "cryptic" gene flow rate and the "apparent" gene flow rate yields the total gene flow rate.
To estimate cryptic gene flow, we used the program GFLOW (available from B.K.D.), which employs a slight modification of the technique developed by Devlin and Ellstrand (1990)
. This technique uses maximum-likelihood to determine the total (apparent + cryptic) gene flow rate that would be most likely to yield the observed apparent gene flow rate (Devlin and Ellstrand, 1990
). Pollen-pool frequencies, needed in the estimation of cryptic gene flow, were taken from the allele frequencies of the six populations of adult trees used in the estimation of genetic diversity.
Gene flow was estimated for individual maternal trees in a stand. The overall gene flow rate for the stand was calculated as the average, weighted by seed number, of gene flow estimates for all the maternal trees for which seedlings were available.
Twogener analysis
A twogener analysis (Smouse et al., 2001
) yielded an alternative measure of pollen movement. The approach estimates pollen movement at a landscape level, using the genetic intraclass correlation coefficient (
ft) derived from an AMOVA (Excoffier et al., 1992
) on the inferred male gametic genotypes. The estimate of ft can be used to directly estimate the effective number of pollen donors (Nep). Joined with an estimate of population density (d), it can be used to estimate the average pollination distance (Smouse et al., 2001
). Unlike paternity analysis, the twogener approach does not assume complete sampling of all fathers within the area of the study. We used the program Famoz (Gerber et al., 2003
) to conduct the AMOVA and subsequent calculations of Nep and ft, using a density of 0.225 trees/km2, which was estimated previously for B. simaruba in the area of our study (Weaver and Chinea, 2003
).
Indirect measures of gene flow
Indirect estimates of the effective number of immigrants per generation (Nem), were made with the following equation (Wright, 1951
):
|
| (1) |
). A key assumption for using this approach is that the populations are in migration–drift equilibrium, indicated typically by a finding of isolation by distance: isolation by distance is indicated by a statistically significant positive relationship between pairwise estimates of FST/(1 – FST) and the geographic distance separating each population pair (Rousset, 1997
).
For comparing indirect and direct measures of gene flow, an estimate of effective population size (Ne) was combined with a paternity analysis-derived estimate of migration rate (m) to yield Nem, which was compared to the Nm derived through the indirect, FST-based approach. The migration rate (m) was estimated as half of the total gene flow rate estimated via paternity analysis, because the paternal contribution represents half of a seedling's genotype and the maternal contribution (nonmigrating in the current study: seeds were taken from known maternal trees) the other half. Effective population size was calculated as Ne = 1/
((ci + pi)/2)2, where ci and pi are the relative female and male reproductive contribution of the ith individual, respectively (Crow and Denniston, 1988
). Relative fruit production yielded estimates of relative female reproductive success. Relative male reproductive success was calculated by assigning parentage to seedlings using the program CERVUS 2.0 (Marshall et al., 1998
) and determining the relative amount of pollen contributed by each male.
RESULTS
Genetic diversity statistics and population structure
Of the 15 loci surveyed, 11 (73.3%) were polymorphic in at least one of the populations sampled (Table 2). At the species level, there were 2.73 alleles per polymorphic locus (APs), the effective number of alleles (Aes) was 1.50, and the expected heterozygosity (Hes) was 0.244.
|
The percentage of the variation distributed among populations (GST) was 6.6% (Table 3). Patterns of polymorphism (or lack thereof) at four loci suggested geographic patterning. Newac, the westernmost population, was polymorphic for 6Pgd-1, which was monomorphic in the remaining populations. Conversely, Newac was monomorphic for Idh-1, which was polymorphic in all the other populations except its closest neighbor, Refuge. Similarly, Pargue, the easternmost population, was polymorphic for Dia-1, which was monomorphic for all other populations except its closest neighbor, Tinaja. Skdh-1 was polymorphic for all populations except R303B and Tinaja, which were the two populations closest to each other.
|
2 = 14.35; P < 0.10), however, indicating a match with Hardy–Weinberg expectations. This is also demonstrated by the close match of observed heterozygosity (0.220) to expected (0.222) (Table 2). With all six populations considered together, there were fewer heterozygotes than under Hardy–Weinberg expectations (FIT = 0.042), due primarily to a Wahlund effect among populations (GST = 0.066). A 2 analysis for allele frequency heterogeneity among populations revealed significant differences among populations for eight of the 11 polymorphic loci (all except Dia-1, Mnr-1, and 6Pgd-1).
Mating system and direct estimates of gene flow
The multilocus estimate of outcrossing rate (tm) for the species was 0.985 (SE = 0.042). The single-locus estimate was 0.904, leading to an estimate of biparental inbreeding (tm – ts) of 8.1%. Seed germination rates varied from 0.0 to 40.8% (Table 1). Because of low germination rates, only three populations had enough seedlings to allow paternity analyses of seeds from individual maternal trees. In two populations, which had 17 (DC) and 19 seeds (JFH), total gene flow rates were calculated at the population level only.
Stands with more than three trees produced 1279 (SE = 410) fruits/tree, an amount statistically greater than the 309 (SE = 193) fruits/tree produced in stands of only three trees (t test, t = 2.143, df = 28, P = 0.021). Stands that produced the most fruit also yielded seeds with the highest germination rates (Fig. 3). In those stands with lower fruit production, more time was required to find seeds on the ground, and the vast majority of the seeds were hollow. The percentage of seeds that were hollow decreased monotonically with increasing stand size, to a low of 33.8% in the largest stand (N = 9 trees). Larger stands of trees also produced seeds with full-grown embryos that did not germinate. Because embryos fill the seeds only in the last week of an 8–9 mo developmental period, these seeds must have been alive at least that long.
|
|
). Assuming a bivariate distribution, the default distribution used in the absence of other information, the average distance of pollen movement (
) was 320 m. Based on an exponential distribution, which may be a closer fit to reality for most species (Austerlitz et al., 2004
), pollen moved an average of 361 m.
Indirect estimates of gene flow
The five highest pairwise GST values were between the Refuge population and all other populations (range 0.122–0.211). There appeared to be a relationship between genetic and geographic distance, once the Refuge population was removed. The next highest pairwise GST values were between Newac and Pargue (0.063), which also were the farthest apart. Next were R303B and Pargue (0.059), which were the sixth farthest apart, and then Antenna and Pargue (0.038), the third farthest apart. The lowest pairwise GST was between Antenna and R303B (0.009), which had the second lowest geographic separation.
Despite the apparent relationship between genetic and geographic distance, there was only a slightly positive, nonsignificant relationship between pairwise FST/(1 – FST) and interpopulation distances, as would be expected under migration–drift equilibrium (r = 0.006, P = 0.349).
Interestingly, two of the Nem values estimated directly from our gene flow and population data, 2.97 (R307) and 3.00 (JFH), were close to 3.57, the FST-based value of Nm. The values of Nem for JFL (2.40) and JFE (2.05) were slightly lower. The Nem for DC was the lowest (0.71), although having just three trees limited its maximum possible Nem to 1.41.
DISCUSSION
Diversity statistics and population structure
Even with the widespread clearing of forest in the study area, B. simaruba maintained a relatively high amount of genetic variation: diversity statistics were comparable to or exceeded those reported for other long-lived, perennial, outcrossing species by Hamrick and Godt (1996)
. The percentage of polymorphic loci was slightly higher (73.3 vs. 65.5%), while the genetic diversity value (He) was substantially higher (0.244 vs. 0.180). The proportion of variation among populations (GST) was low for B. simaruba (0.066 vs. 0.094), a value even lower than that reported for long-lived, perennial, wind-dispersed species (0.086), which typically have the capacity for long-distance gene movement.
Long-distance pollen movement, combined with almost total outcrossing, is likely responsible for the low levels of population divergence (as measured by GST) and the relatively high genetic diversity values (He) and percentage of polymorphic loci (Pp) found within populations. It is possible, however, that the populations in this study may not have undergone enough generations of drift since fragmentation to appreciably decrease genetic diversity, especially because Puerto Rican forests were fragmented mainly in the 19th century (Wadsworth, 1950
). Low GST values may also be due to the spatial scale of the study. Pollen, and possibly seeds, may readily move over the distances involved (ca. 25 km); notably, many plant allozyme studies sample populations at a scale of several hundred kilometers.
The ability of seed dispersal to introduce variability is illustrated by the Refuge population, which aerial photos show to have been established after 1960. High genetic diversity (He) and relatively constant GST values with the other populations indicate that this site was established by seeds from several individuals and/or populations (immigrant pool model, Slatkin, 1977
) rather than from only a few source individuals (propagule pool model).
Direct measures of gene flow
The high levels of pollen flow observed suggest that even stands with as few as 5–9 trees receive immigrant pollen, even when the nearest B. simaruba tree is further than 600 m (R307). Results from the twogener analysis are consistent with the paternity analysis results. The average distance of pollen movement (320–361 m) exceeded the average distance between trees within a population (57 m), which is consistent with the notion that most successful pollen originates via gene flow.
The twogener analysis demonstrated one important point that the paternity analysis could not detect; namely, the effective number of fathers (2.45) is at the low end of what has been reported for other tree species (see Table 3 in Hardy et al., 2004
). A relatively low density of trees, asynchrony in flowering (B. Dunphy, personal observation), and the polygamodioecious breeding system of the species may be responsible. So, although pollen appears to be capable of moving long distances, individual trees receive most of their pollen from just a few donors.
Germination rates varied considerably with stand size. In the four sites with three trees each, only 17 of 1834 seeds (0.9%) germinated. The three stands with 7–9 trees had germination rates ranging from 20–41%. Similarly, small stands of the tropical tree Spondias mombin on islands in Lake Gatun, Panama had significantly lower germination than did larger fragmented and continuous-forest populations (Nason and Hamrick, 1997
). In B. simaruba and S. mombin, these findings suggest that larger stands receive more foreign pollen, thus losing fewer seed to self-incompatibility and biparental inbreeding. This may be analogous to what has been reported in other insect-pollinated (Platt et al., 1974
; Willson and Rathcke, 1974
), bat-pollinated (Heithaus et al., 1982
), and bird-pollinated (Carpenter, 1976
) plant species, where an increase in flower number led to increased pollination success, presumably due to an increase in pollinator visits.
Curiously, the largest population, R307, had a relatively low percentage (67%) of foreign pollinations. If dead seeds are considered in the count along with living seeds, however, R307 had the second highest percentage of gene-flow seeds (Fig. 3). In addition, the trees of this stand, the largest of all examined, produced more fruit per tree than trees in the other seven stands (B. Dunphy, personal observation). This suggests that R307 may actually receive as much, if not more, foreign pollen as the other stands. If so, a greater tolerance of internally generated pollen, normally filtered out through seed abortion in the smaller stands, may be responsible for depressing the effective gene flow rate.
Conversely, intolerance for within-population pollen likely explains the high gene flow estimate (100%) for the JH site, although this comes at the cost of limiting seed production when foreign pollen is rare, which appears to be the case for the JH site where only 7.5% of seeds germinated. Similarly, a direct analysis of pollen flow in fragmented stands of the mahogany species Swietenia humilis from Honduras showed that the percentage of seeds sired by foreign pollen increased as population size decreased (White et al., 2002
). Local pollen was presumably still being received in smaller stands, but a self-incompatibility system filtered out much of this pollen before fertilization. In B. simaruba, seed abortion likely played a similar role in increasing the representation of foreign sires among seedlings.
Indirect estimates of gene flow
Despite the relative simplicity of the approach, the absence of migration–drift equilibrium among populations of B. simaruba would render unreliable the conclusions of a study based solely on indirect, FST-based estimates of gene flow. If the true migration rate were lower than the estimated rate of Nm = 3.57, for instance, erroneous predictions of further genetic changes could be made, especially if the true rate were less than one migrant per generation, the level that Wright (1931)
demonstrated would allow genetic drift to increase genetic distance among populations. Nevertheless, the relatively low levels of differentiation among populations indicate that historical levels of gene flow must have been relatively high. Low levels of genetic divergence (average GST = 0.055) were also found for 16 common woody species on the 6 x 6 km Barro Colorado Island in Panama (Hamrick and Loveless, 1989
). Direct evidence that pollen moved over 500 m in one species, Tachigali versicolor, and over 200 m for several other species is consistent with high gene flow causing the low GST values (e.g., Hamrick and Murawski, 1990
; Stacy et al., 1996
).
Although a significantly positive relationship between genetic and geographic distances may develop over time, Rousset (1997)
noted that populations separated by small geographic distances generally will not follow expectations of the model used to determine isolation by distance; evidence of a relationship between geographic and genetic distances is typically found only at larger scales (e.g., Kaufman et al., 1998
; Bockelmann et al., 2003
). A study of population pairs in B. simaruba with larger interpopulation distances did indeed find a significant relationship between geographic distance and FST beginning at approximately 25–30 km of separation (B. Dunphy, unpublished data), just beyond the maximum distance separating any two of the six populations used in this study. This suggests that B. simaruba may actually be in migration–drift equilibrium, which may explain the close relationship between the indirect Nm estimates (3.57) and direct Nem (mean = 2.23) estimates of migration.
This similarity between the direct and indirect estimates was nevertheless surprising, given the numerous assumptions that differ between the two approaches, as well as the differences in experimental design, including differences in the number of study subjects and the isolation distances between stands or populations of trees. Both techniques, with the exception of direct estimates in the DC stand, yielded estimates of Nm between 1 and 4, the range within which Wright (1931)
demonstrated that gene flow will tend to predominate over genetic drift. Direct estimates of gene flow do not incorporate seed movement, however, while the indirect estimates do. This should not affect estimates of Nem substantially because gene movement by pollen is generally much greater than that by seeds (Ennos, 1994
). Perhaps more important is the potential impact of population size; specifically, whether direct estimates of Nem for larger stands would be similar to those from smaller stands.
Ecological implications
In tropical forests, conspecific trees tend to occur in small clumps, often with substantial distances separating neighboring clumps (Hubbell and Foster, 1983
). If trees in a clump are related, as might be expected in low-density patches of a bird-dispersed species (Hamrick et al., 1993
) such as B. simaruba, and if they possess a self-incompatibility mechanism, as do a large number of tropical tree species (Bawa, 1974
), then selection should be strong for floral strategies that support the energetic needs of long-distance pollinators.
Inga species in wet montane forests of Costa Rica produce many relatively unspecialized flowers that attract a wide array of pollinators over an extended flowering season (Koptur, 1984
). Although stigmas may receive much pollen from other flowers on the same tree, enough long-distance pollinations presumably occur to justify the vast expenditure of resources on flower production. An obligate outcrosser may therefore be more tolerant of habitat fragmentation than a mixed-mating species because the latter could have an increase in selfed seeds as populations decrease in size and become more geographically isolated (White et al., 2002
). The current study of B. simaruba suggests that in some cases, however, this benefit has to be weighed against the potentially high loss of reproductive output from aborted seeds in smaller, more isolated populations.
Low seed production and germination rates in smaller populations of B. simaruba (i.e., less than five trees) may pose an ecological risk to the survival of those populations. Dry forests present a particularly harsh environment for seedling survival (Ray and Brown, 1995
). If microsites for seedling survival are scarce, then it may be difficult for the relatively small number of viable seeds to find a suitable microsite, and none may survive. This would be a threat not just to remnant populations in a fragmented landscape, but could also greatly slow the spread of this bird-dispersed species, especially given the likelihood that seeds that establish a new site may have come from the same source, thus leading to reproductive compatibility problems later on. The abundance of B. simaruba across its range in a variety of habitats suggests that this problem has not restricted its distribution. One ameliorative mechanism might be facultative apomixis, which has been reported from other Bursera species (Becerra and Venable, 1999
). Under this scenario, trees capable of apomixis would have a selective advantage in small populations with restricted pollen flow and would increase in number. As the population grows, pollinators become more plentiful, leading to a higher frequency of foreign pollen, which in turn yields outbred, viable seeds. These recombinant individuals would themselves become a source of variability as the population grew, and within-population pollinations would eventually become possible.
Summary
Fragmented habitats present a special challenge to the accurate estimation of gene flow, a critical reproductive parameter influencing the capacity for a species to survive and reproduce in the altered environment. Indirect methods of gene flow rely on assumptions that may not be met in such circumstances. The most important of these assumptions is that of migration–drift equilibrium, normally indicated by a correlation between geographic and genetic distances (isolation by distance). Despite the absence of isolation by distance in B. simaruba, there was general agreement between indirect (Nm) estimates of migration and those derived using direct, paternity-based methods (Nem), which suggests that the system may actually be near migration–drift equilibrium. Direct estimates of gene flow further show that small stands of B. simaruba do not appear to be genetically isolated, at least for stands of more than four trees. Gene flow rates were high, and pollen moved long distances, in one case (population R307) over 600 m. High genetic diversity values in larger populations and relatively low genetic divergence among populations support the notion that gene flow occurs over substantial distances. The twogener analysis showed a low number of effective fathers, however, indicating that genetic diversity may still be lost because of limited breeding opportunities in the smaller stands of trees. Very small groups of trees (<4 individuals) may be especially at risk through decreased seed production presumably related to increases in self-pollination and biparental inbreeding.
FOOTNOTES
1 The authors thank the staff of the Caribbean National Wildlife Refuge for access to study sites and logistic support; J. Schwagerl for extensive field assistance; and M. Arnold, J. Avise, R. Malmberg, and C. Peterson for advice and comments on the manuscript. This research was funded by grants from Sigma Xi, the University of Georgia Department of Plant Biology, the University of Georgia Center for Latin American and Caribbean Studies, and the National Geographic Society. 
2 Author for correspondence (e-mail: hamrick{at}plantbio.uga.edu
)
LITERATURE CITED
Austerlitz F. Dick C. W. Dutech C. Klein E. K. Oddou-Muratorio S. Smouse P. E. Sork V. L.. 2004. Using genetic markers to estimate the pollen dispersal curve. Molecular Ecology 13: 937-954..[CrossRef][Medline]
Bawa K. S.. 1974. Breeding systems of tree species of a lowland tropical community. Evolution 28: 85-92..[CrossRef][Web of Science]
Becerra J. X. Venable D. L.. 1999. Nuclear ribosomal DNA phylogeny and its implications for evolutionary trends in Mexican Bursera (Burseraceae). American Journal of Botany 86: 1047-1057..
Bockelmann A. C. Reusch B. H. Bijlsma R. Bakker J. P.. 2003. Habitat differentiation vs. isolation-by-distance: the genetic population structure of Elymus athericus in European salt marshes. Molecular Ecology 12: 505-515..[CrossRef][Medline]
Carpenter F. L.. 1976. Plant–pollinator interactions in Hawaii: pollination energetics of Metrosideros collina (Myrtaceae). Ecology 57: 1125-1144..[CrossRef][Web of Science]
Chakroborty P. Meagher T. R. Smouse P. E.. 1988. Parentage analysis with genetic markers in natural populations. 1. Paternity exclusion and expected proportions of offspring with unambiguous paternity. Genetics 118: 527-536..
Chase M. R. Moller C. Kesseli R. Bawa K. S.. 1996. Distant gene flow in tropical trees. Nature 383: 398-399..[CrossRef][Web of Science]
Cheliak W. M. Pitel J. A.. 1984. Techniques for starch gel electrophoresis of enzymes for forest tree species. Petawawa National Forestry Institute, Information Report P1-X-42, Canadian Forestry Service, Agriculture, Chalk River, Ontario, Canada..
Crow J. F. Denniston C.. 1988. Inbreeding and variance effective population numbers. Evolution 42: 482-495..[CrossRef][Web of Science]
Degen B. Bandou E. Caron H.. 2004. Limited pollen dispersal and biparental inbreeding in Symphonia globulifera in French Guiana. Heredity 93: 585-591..[CrossRef][Web of Science][Medline]
Devlin B. Ellstrand N. C.. 1990. The development and application of a refined method for estimating gene flow from angiosperm paternity analysis. Evolution 44: 248-259..[CrossRef][Web of Science]
Dick C. W.. 2001. Genetic rescue of remnant tropical trees by an alien pollinator. Proceedings of the Royal Society of London, B, Biological Sciences 268: 2391-2396..
Dick C. W. Etchelecu G. Austerlitz F.. 2003. Pollen dispersal of tropical trees (Dinizia excelsa: Fabaceae) by native insects and African honeybees in pristine and fragmented Amazonian rainforest. Molecular Ecology 12: 753-764..[CrossRef][Medline]
Ennos R. A.. 1994. Estimating the relative rates of pollen and seed migration among plant populations. Heredity 72: 250-259..[CrossRef][Web of Science]
Excoffier L. Smouse P. E. Quattro J. M.. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491..[Abstract]
Fischer M. Matthies D.. 1997. Mating structure and inbreeding and outbreeding depression in the rare plant Gentianella germanica (Gentianaceae). American Journal of Botany 84: 1685-1692..[Abstract]
Gerber S. Chabrier P. Kremer A.. 2003. FaMoz: a software for parentage analysis using dominant, codominant and uniparentally inherited markers. Molecular Ecology Notes 3: 479-481..[CrossRef][Web of Science]
Gibbs J. P.. 2001. Demography versus habitat fragmentation as determinants of genetic variation in wild populations. Biological Conservation 100: 15-20..[CrossRef][Web of Science]
Goudet J.. 1995. FSTAT (ver. 1.2): a computer program to calculate F-statistics. Journal of Heredity 86: 485-486..
Greenberg R. Foster M. S. Marquez-Valdelamar L.. 1995. The role of the white-eyed vireo in the dispersal of Bursera fruit on the Yucatan Peninsula. Journal of Tropical Ecology 11: 619-639..[Web of Science]
Groom M.. 2001. Consequences of subpopulation isolation for pollination, herbivory, and population growth in Clarkia concinna (Onagraceae). Biological Conservation 100: 55-63..[CrossRef][Web of Science]
Hamrick J. L. Godt M. J. W.. 1996. Effects of life history traits on genetic diversity in plant species. Philosophical Transactions of the Royal Society of London, B, Biological Sciences 351: 1291-1298..[CrossRef]
Hamrick J. L. Loveless M. D.. 1989. The genetic structure of tropical tree populations: associations with reproductive biology. In J. H. Bock, Y. B. Linhart [eds.], The evolutionary ecology of plants, 129-146. Westview Press, Boulder, Colorado, USA..
Hamrick J. L. Murawski D. A.. 1990. The breeding structure of tropical tree populations. Plant Species Biology 5: 157-165..[CrossRef]
Hamrick J. L. Murawski D. A. Nason J. D.. 1993. The influence of seed dispersal mechanisms on the genetic structure of tropical tree populations. Vegetatio 107–108: 281-297..
Hamrick J. L. Nason J. D.. 2000. Gene flow in forest trees. In T. J. B. Boyle, A. Young, D. Boshier [eds.], Forest conservation genetics: principles and practice, 81-90. Center for International Forestry Research (CIFOR) and Commonwealth Scientific and Industrial Research Organisation (CSIRO), Collingwood, Australia..
Hardy O. J. González-Martínez S. C. Fréville H. Boquien G. Mignot A. Colas B. Olivieri I.. 2004. Fine-scale genetic structure and gene dispersal in Centaurea corymbosa (Asteraceae) I. Pattern of pollen dispersal. Journal of Evolutionary Biology 17: 795-806..[CrossRef][Web of Science][Medline]
Heithaus E. R. Stashko E. Anderson P. K.. 1982. Cumulative effects of plant–animal interactions on seed production by Bauhinia ungulata, a neotropical legume. Ecology 63: 1294-1302..[CrossRef][Web of Science]
Holdridge L. R.. 1967. Life zone ecology. Tropical Science Center, San José, Costa Rica..
Hubbell S. P. Foster R. B.. 1983. Diversity of canopy trees in a neotropical forest and implications for the conservation of tropical trees. In S. L. Sutton, T. C. Whitmore, A. C. Chadwick [eds.], Tropical rain forest: ecology and management, 25-41. Blackwell, Oxford, UK..
Kaufman S. R. Smouse P. E. Alvarez-Buylla E. R.. 1998. Pollen-mediated gene flow and differential male reproductive success in a tropical pioneer tree, Cecropia obtusifolia Bertol. (Moraceae): a paternity analysis. Heredity 81: 164-173..[CrossRef][Web of Science]
Knapp E. E. Goedde M. A. Rice K. J.. 2001. Pollen-limited reproduction in blue oak: implications for wind pollination in fragmented populations. Oecologia 128: 48-55..[CrossRef][Web of Science]
Koptur S.. 1984. Outcrossing and pollinator limitation of fruit set: breeding systems of neotropical Inga trees (Fabaceae: Mimosoideae). Evolution 38: 1130-1143..[CrossRef][Web of Science]
Kuang H. Richardson T. E. Carson S. D. Bongarten B. C.. 1999. Genetic analysis of inbreeding depression in plus tree 850.55 of Pinus radiata D. Don. II. Genetics of viability genes. Theoretical and Applied Genetics 99: 140-146..[CrossRef][Web of Science]
Latouche-Hallé C. Ramboer A. Bandou E. Caron H. Kremer A.. 2004. Long-distance pollen flow and tolerance to selfing in a neotropical tree species. Molecular Ecology 13: 1055-1064..[CrossRef][Medline]
Ledig F. T. Jacob-Cervantes V. Hodgskiss P. D. Eguiluz-Piedra T.. 1997. Recent evolution and divergence among populations of a rare Mexican endemic, Chihuahua spruce, following Holocene climatic warming. Evolution 51: 1815-1827..[CrossRef][Web of Science]
Little E. L. Wadsworth F. H.. 1964. Common trees of Puerto Rico and the Virgin Islands. USDA Forest Service, Agriculture Handbook 249, Washington, D.C., USA..
Manchenko G. P.. 1994. Handbook of detection of enzymes on electrophoretic gels. CRC Press, Ann Arbor, Michigan, USA..
Marshall T. C. Slate J. Kruuk L. Pemberton J. M.. 1998. Statistical confidence for likelihood-based paternity inference in natural populations. Molecular Ecology 7: 639-655..[CrossRef][Medline]
Menges E. S. Dolan R. W.. 1998. Demographic viability of populations of Silene regia in midwestern prairies: relationships with fire management, genetic variation, geographic location, population size and isolation. Journal of Ecology 86: 63-78..[CrossRef][Web of Science]
Mitton J. B. Linhart Y. B. Sturgeon K. B. Hamrick J. L.. 1979. Allozyme polymorphisms detected in mature needle tissue of ponderosa pine. Journal of Heredity 70: 86-89..
Nason J. D. Hamrick J. L.. 1997. Reproductive and genetic consequences of forest fragmentation: two case studies of neotropical canopy trees. Journal of Heredity 88: 264-276..
Neel M. C. Ellstrand N. C.. 2001. Patterns of allozyme diversity in the threatened plant Erigeron parishii (Asteraceae). American Journal of Botany 88: 810-818..
Neigel J. E.. 1997. A comparison of alternative strategies for estimating gene flow from genetic markers. Annual Review of Ecology and Systematics 28: 105-128..[CrossRef][Web of Science]
Platt W. J. Hill G. R. Clark S.. 1974. Seed production in a prairie legume (Astragalus canadensis L.). Oecologia 17: 55-63..[CrossRef][Web of Science]
Ray G. J. Brown B.. 1995. The structure of five successional stands in a subtropical dry forest, St. John, U.S. Virgin Islands. Caribbean Journal of Science 31: 212-222..
Ritland K.. 2002. Extensions of models for the estimation of mating systems using n independent loci. Heredity 88: 221-228..[CrossRef][Web of Science][Medline]
Rousset F.. 1997. Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance. Genetics 145: 1219-1228..[Abstract]
Slatkin M.. 1977. Gene flow and genetic drift in a species subject to frequent local extinctions. Theoretical Population Biology 12: 253-262..[CrossRef][Web of Science][Medline]
Slatkin M.. 1993. Isolation by distance in equilibrium and nonequilibrium populations. Evolution 47: 264-279..[CrossRef][Web of Science]
Smouse P. E. Dyer R. J. Westfall R. D. Sork V. L.. 2001. Two-generation analysis of pollen flow across a landscape. I. Male gamete heterogeneity among females. Evolution 55: 260-271..[CrossRef][Web of Science][Medline]
Smouse P. E. Sork V. L.. 2004. Measuring pollen flow in forest trees: an exposition of alternative approaches. Forest Ecology and Management 197: 21-38..[CrossRef][Web of Science]
Soltis D. E. Haufler C. H. Darrow D. C. Gastony G. J.. 1983. Starch gel electrophoresis of ferns: a compilation of grinding buffers, gel and electrode buffers, and staining schedules. American Fern Journal 73: 9-27..[CrossRef][Web of Science]
Sork V. L. Nason J. Campbell D. R. Fernandez J. F.. 1999. Landscape approaches to historical and contemporary gene flow in plants. Trends in Ecology and Evolution 14: 219-224..[CrossRef]
Stacy E. A. Hamrick J. L. Nason J. D. Hubbell S. P. Foster R. B. Condit R.. 1996. Pollen dispersal in low-density populations of three neotropical tree species. American Naturalist 148: 275-298..[CrossRef][Web of Science]
Stevens G. C.. 1983. Bursera simaruba (Indio Desnudo, Jinocuave, Gumbo Limbo). In D. H. Janzen [ed.], Costa Rican natural history, 201-202. University of Chicago Press, Chicago, Illinois, USA..
Surle S. E. Arnold J. Schnabel A. Hamrick J. L. Bongarten B. C.. 1990. Genetic relatedness in open-pollinated families of two leguminous tree species, Robinia pseudoacacia L. and Gleditsia triacanthos L. Theoretical and Applied Genetics 80: 49-56..[Web of Science]
Wadsworth F. H.. 1950. Notes on the climax forests of Puerto Rico and their destruction and conservation prior to 1900. Caribbean Forester 11: 38-47..
Weaver P. L. Chinea J. D.. 2003. Secondary subtropical dry forest at the La Tinaja tract of the Cartagena Lagoon National Wildlife Refuge, Puerto Rico. Caribbean Journal of Science 39: 273-285..[Web of Science]
Weir B. S. Cockerham C. C.. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370..[CrossRef][Web of Science]
White G. M. Boshier D. H. Powell W.. 2002. Increased pollen flow counteracts fragmentation in a tropical dry forest: an example from Swietenia humilis Zuccarini. Proceedings of the National Academy of Sciences, USA 99: 2038-2042..
Willson M. F. Rathcke B. J.. 1974. Adaptive design of the floral display in Asclepias syriaca L. American Midland Naturalist 92: 47-57..[CrossRef][Web of Science]
Wright S.. 1931. Evolution in Mendelian populations. Genetics 16: 97-159..
Wright S.. 1951. The genetical structure of populations. Annals of Eugenics 15: 323-354..[Web of Science]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
