| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Population Biology |
2Department of Plant Biology, University of Georgia, Athens, Georgia 30602 USA; 3Departments of Plant Biology and Genetics, University of Georgia, Athens, Georgia 30602 USA
Received for publication July 12, 2005. Accepted for publication April 6, 2006.
ABSTRACT
Pollinator behavior plays a central role in determining patterns of pollen-mediated gene movement in zoophilous angiosperms. A species' floral display can strongly influence the behavior of its pollinators and thereby affect its evolutionary pathway. We used paternity analysis to directly measure and describe mating patterns within 15 populations of the epiphytic orchid, Laelia rubescens, in Costa Rican dry forest. Strict correlated mating by orchids allows inference of the precise multilocus diploid genotype of the pollen parents. Our data show that mean effective population sizes were small (11.2 in 1999 and 11.8 in 2000) relative to the number of flowering genets (63 and 56, respectively). Fewer genets were reproductively successful as females than males. The relationship between reproductive success (RS) and floral display within three cluster size classes was consistent between years, with large (>30 inflorescences) and small (
10 inflorescences) clusters often having significantly lower RS than expected, while the RS of medium-sized clusters (11–30 inflorescences) often significantly exceeded expectations. Paternity analysis allowed us to take advantage of the pollination biology of L. rubescens to provide unusually detailed insights into mating patterns, pollen-mediated gene movement and RS for populations of this epiphytic orchid, an herbaceous perennial, distributed in three-dimensional space.
Key Words: Costa Rica • dry forest • effective population size • Orchidaceae • paternity analysis • paternity pool • reproductive success
Angiosperms are uniquely differentiated from all other plant groups by their possession of flowers. The rapid radiation and speciation of angiosperms is often attributed to floral evolution, and it is believed that pollinator traits and behavior have been driving forces behind that evolution. Individuals whose floral display attracts more pollinators are more successful in transmitting genes to the next generation. Zoophilous angiosperms typically offer their pollinator rewards in the form of pollen, nectar, or fragrance to insure multiple visits, and floral displays serve as the visual cue that advertises reward availability. The exception is angiosperms that cheat, using deception to attract pollinators by suggesting the promise of a reward without expending resources to provide one. This can occur through visual mimicry of a sympatric rewarding species or mimicry of a specific reward (e.g., pseudopollen; Dafni, 1984
). Taxa that use deception typically experience lower fruit set than rewarding species.
It is well known that patterns of pollen dispersal are affected by pollinator behavior (e.g., Bateman, 1947
; Wright, 1951
; Linhart et al., 1987
) and that pollen dispersal is central in shaping patterns of gene flow and the partitioning of genetic variation among species' populations. Gene movement and the transmission of genes to the next generation are central to species survival and evolution. Thus, the manner in which a plant's floral display influences the behavior of its pollinators affects the evolutionary trajectory of that species. The reproductive success (RS) of animal-pollinated plants is strongly related to the floral display, which is determined by the number, size, and density of inflorescences (Schmitt et al., 1987
; Kearns and Inouye, 1993
; Widen, 1993
). The RS and hence fitness of a plant is expected to increase with a larger floral display as a result of increased attractiveness to pollinators. Previous studies have shown that large inflorescences have higher pollinator visitation (e.g., Broyles and Wyatt, 1990
; Harder and Barrett, 1996
) and that larger floral displays have a greater likelihood of male and/or female RS (Schemske, 1980
; Firmage and Cole, 1988
; Ackerman, 1989
).
A common observation in orchids is that a large proportion of individuals experience no RS. The few that are reproductively successful produce capsules containing up to eight million, tiny, wind-dispersed seeds that are potentially capable of long-distance dispersal. The clonal growth pattern of some orchids enables individuals to persist over a number of years and thereby improve their chances of lifetime RS. Clonal growth also enables genets to increase their floral displays through the multiplication of flowering ramets and thereby increase their attractiveness to pollinators. The greater synchrony of flowering within larger clusters of self-compatible species, however, also promotes increased pollinations between flowers of the same genet (i.e., geitonogamy; Charpentier, 2002
).
Here we use paternity exclusion analysis to directly measure and describe mating patterns within 15 populations of the animal-pollinated, epiphytic orchid, Laelia rubescens Lindley (Orchidaceae), in the dry forest of Costa Rica. Specifically, we ask what was the effective number of males and females for each population (all individuals on a single host tree) over two years? How large was the pool of potential pollen donors for individual females in these two years? How did male and female RS relate to floral display size? Were reproductively successful clusters in the first year also successful in the second year? Over what distances was pollen dispersed within these populations? Our expectation was that increased RS would be associated with larger floral displays and that, due to the density of flowering clusters, pollen would move over short distances within populations.
Epiphytic orchids have a suite of unusual and unique characteristics that make this work of particular interest. Orchid epiphytes are herbaceous plants that have adopted a growth habit that shares many similarities with trees. The orchid's reproductive potential is enhanced by increased visibility to pollinators and the ability to release seeds higher in the air column for wind dispersal. However unlike the host tree, the orchid population is comprised of multiple genotypes. Epiphytic plants are also unusual in that a population grows in three-dimensional space. As a result, an epiphyte can be surrounded by more conspecific individuals within a distance class than is possible for a terrestrial species. This three-dimensional configuration could increase the floral display density of the population and thus increase population level attractiveness to pollinators. These factors can potentially influence the RS of the species and thus its evolutionary potential.
MATERIALS AND METHODS
Study organism
Laelia rubescens Lindley (Orchidaceae) is a neotropical, long-lived perennial epiphyte that ranges from Mexico to Panama (Williams, 1946
) in dry habitats below 800 m a.s.l. (Mora de Retana and Atwood, 1992
). Its bisexual flowers are exclusively animal-pollinated, with hummingbirds believed to be the primary pollen vectors (D. W. Trapnell, personal observation). Intraflower pollination is not possible, but geitonogamous pollinations do occur (Trapnell and Hamrick, 2005
). Orchids are well suited for paternity analysis because their pollen grains are aggregated in cohesive, sticky pollinia. Each L. rubescens flower has eight pollinia. Four pollinia attach as a unit, via column-derived viscid material, onto the animal vector and are thus transported to a receptive flower. One pollinium possesses enough pollen grains to fertilize every ovule in a recipient flower. The package of four pollinia is also quite large in relation to the stigmatic surface making it mechanically difficult and therefore unlikely that pollinia from multiple sources could be deposited on the stigma. As a result, individual fruits contain seeds pollinated by a single pollen donor and represent full-sib progeny arrays (Trapnell et al., 2004
; Trapnell and Hamrick, 2005
). This attribute allows inference of the diploid genotype of the pollen parent, facilitating direct measurement of pollen flow and detailed descriptions of mating patterns within populations. Such strict correlated mating is only known in orchids (Gentry and Dodson, 1987
; Nilsson, 1992
), milkweeds, mimosoid legumes, and some tropical figs. Analysis of 459 progeny arrays (~6800 seedlings) at the four loci with >2 alleles (i.e., 1836 samples) failed to show a third allele donated by a pollen parent. Thus, our data confirm that if multiple paternity occurs, it is too infrequent to affect our results.
Each fertilization event results in hundreds of thousands of tiny wind-dispersed seeds. Once established on suitable substrate, L. rubescens grows clonally with each fleshy pseudobulb (inflated stem tissue) producing one or two new pseudobulbs/year. Each pseudobulb produces one inflorescence with as many as 20 showy, pink flowers (Halbinger and Soto, 1997
; D. W. Trapnell, personal observation) that mature acropetally. A single inflorescence has been observed to mature up to 11 capsules (D. W. Trapnell, personal observation). Anthesis is over an extended period (January to March) during the dry season. Clusters can possess 100 or more pseudobulbs and produce multiple inflorescences simultaneously. As many as nine distinct genotypes have been observed within an individual cluster (Trapnell et al., 2004
).
Study site
The research site is located in the Pacific lowlands of northwest Costa Rica, in the Tempisque River basin of Guanacaste Province. Laelia rubescens grows on a variety of host trees (Trapnell and Hamrick, in press
) in habitats ranging from primary forests to highly human-modified landscapes. In less disturbed forests, L. rubescens is widely dispersed with relatively few clusters per tree. When the tropical dry forest was cleared for pastures, often one or more shade trees were left. These isolated trees typically have large spreading canopies, and it is on these trees that L. rubescens is most abundant with populations of 350 or more clusters. A population is defined as all L. rubescens occupying a single "host" tree. These trees support several other epiphytic species that occur in densities of a few individuals per tree. These include two orchids (Brassavola nodosa and Encyclia fragans), one bromeliad (Tillandsia schiedeana), and a cactus (Hylocereus costaricensis). The area is classified as seasonally dry tropical forest characterized by semi-deciduous trees and a 6-month dry season (December–May). Study populations are located at Hacienda Solimar (10°17' N and 85°08' W), a privately owned 2000 hectare cattle ranch characterized by isolated trees and small groups of trees in multiple pastures. Hacienda Solimar was established during the mid-1950s and represents a pattern of disturbance typical of many dry forest regions of Costa Rica (Sader and Joyce, 1988
).
Sampling and seed culture
Four large pastures at Hacienda Solimar (AG, CB, GT, and SP), separated by 508 to 1034 m, were selected. Typically, second growth forest ranging in width from 10 to 50 m, occupied stream courses and fencerows between the pastures. Study pastures had one, two, four, and eight trees, respectively, supporting populations of L. rubescens. Host trees included Crescentia alata, Dalbergia retusa, Samanea saman, and Tabebuia rosea. There were 15 populations (i.e., trees) each containing 4–102 clusters of L. rubescens. Every cluster within a population was mapped in three-dimensional space by recording the distance and angle from the center of the tree trunk to the beginning of each cluster. Height was measured with a marked pole or a clinometer where necessary.
At the end of the 1998 flowering season, all old inflorescences within the 15 host trees were removed so that individuals flowering during 1999 could be identified. In April and May of 1999, all mature capsules (211) and subtending maternal leaf material were collected. Subtending leaf tissue from every pseudobulb that flowered but failed to produce capsules was also collected because these individuals represented possible pollen parents. Upon completion of sampling in 1999, all inflorescences were removed as before. During April of 2000 a second reproductive period was similarly sampled with 376 mature capsules collected. Leaf samples from maternal plants and all possible pollen donors were analyzed for their multilocus allozyme genotypes.
Seeds were germinated under sterile conditions. Because nearly all capsules had partially dehisced, it was necessary to sterilize the seeds. Seeds were extracted, sterilized in 10% bleach solution for 10 min, rinsed with autoclaved deionized water for 10 min, and cultured on sterile autoclaved germination medium (Mother Flask Medium IV, G & B Orchid Laboratory, Vista, California, USA). Once the seeds germinated and roots emerged, seedlings were transferred to sterile growth medium (Replate Flask Medium IV, G & B Orchid Laboratory). Seedlings were kept in a growth chamber maintained at approximately 30°C under constant light. Seeds from viable capsules had high germination rates. However, some cultures were lost to fungal contamination by spores that were resistant to multiple sterilization attempts. Upon reaching approximately 1.5 to 2 cm, after 8–14 months, 12–18 seedlings per capsule were assayed for their allozyme genotypes.
Enzyme extraction and electrophoresis
Leaf tissue from maternal plants and possible pollen donors were snap frozen in liquid nitrogen within a few hours of collection and stored in an ultra-cold dry shipper. Samples were sent to the University of Georgia and were crushed in chilled mortars with a pestle, liquid nitrogen, and a pinch of sea sand to disrupt cellular compartmentalization. Seedlings were crushed, without liquid nitrogen or sea sand. Enzymes were extracted with a polyvinylpyrrolidone-phosphate extraction buffer (Mitton et al., 1979
). The resulting slurry containing crude protein extract was absorbed onto 4 x 6 mm wicks punched from Whatman 3 mm chromatography paper. Wicks were stored in microtest plates at –70°C until used for electrophoresis. Wicks were placed in horizontal 10% potato starch gels, and electrophoresis was performed. Seven enzyme stains in three buffer systems resolved 10 putative polymorphic loci. Enzymes stained and the 10 polymorphic loci identified (in parentheses) for each of the three buffer systems were as follows: system 8-, diaphorase (DIA1), fluorescent esterase (FE2), and triosephosphate isomerase (TPI1); system 10, fluorescent esterase (FE1), UTP-glucose-1-phosphate (UGPP1); system 11, malate dehydrogenase (MDH1, MDH3), 6-phosphogluconate dehydrogenase (6-PGD1, 6-PGD2) and phosphoglucomutase (PGM2). All buffer and stain recipes were adapted from Soltis et al. (1983)
except diaphorase and UTP-glucose-1-phosphate, which were taken from Cheliak and Pitel (1984)
and Manchenko (1994)
, respectively. Buffer system 8- is a modification of buffer system 8 as described by Soltis et al. (1983)
. Banding patterns were consistent with those expected for each enzyme system (Weeden and Wendell, 1983
). Seedling allozyme expression was comparable or superior to that of adult leaf tissue. Between two and four alleles were observed at each polymorphic locus making it possible to distinguish 1 749 600 distinct genotypes. The exclusion probability is 0.986 (Paetkau and Strobeck, 1994
). Species-level genetic variation has been reported by Trapnell and Hamrick (2004)
.
Data analyses
Effective number of males and females
Reproductive success was assessed for each population by examining the effective number of males and females as well as the overall effective population size. These measures were determined by assigning paternal and maternal parentage for every capsule within the 15 populations for 1999 and 2000. Each reproductively successful genet was assigned a percentage of the total number of capsules produced within a population according to the number of male and female gametes contributed. Effective number of males within a population was calculated by Ne
= 1/
xi2, where xi is the proportion of male gametes contributed by individual i to total fruit production within the population. Male gametes from pollen donors outside the population were included. Effective number of females within a population was calculated by Ne
= 1/yi2, where yi represents the proportion of female gametes contributed by individual i to total fruit production within the population. Correlation coefficients between the number of inflorescences per population and Ne and Ne were calculated using SigmaPlot (v. 5.0). Total effective population size was determined for each population by Ne = 1/[(1/2)(xi + yi)]2. Because L. rubescens is a long-lived perennial, effective population size for the 2 years pooled together was also determined. The number of migrants per generation (Nem) was found directly for each population by Nem = Ne(1/2)(a/b), where a = number of fruits with pollen donors outside the population and b = total number of capsules produced within the population. This was divided by two to account properly for the haploid nature of pollen. Direct Nem values were averaged across all populations.
These measures were standardized to allow comparisons. Male and female RS relative to the total number of capsules produced within populations was calculated. Specifically, we calculated ratios of (1) effective number of males (Ne) to total number of fruits, (2) effective number of females (Ne) to the total number of fruits, (3) total number of reproductively successful individuals (i.e., individuals that contributed male and/or female gametes) to total fruit, and (4) the ratio of the effective population size (Ne) to total number of genetically distinct individuals (i.e., genets).
Paternity pool
The paternity pool of potential pollen donors for individual females was assessed according to Levin (1988)
. The axial variance, 
2 = pi2/2N, was calculated for each pasture where pi is the distance between a maternal plant (i) and an identified pollen donor. N is the total number of capsules produced within the pasture by pollen donors whose location is known. The paternal pool lies in a circle of radius 3 and area, A = 9
2. The number of potential pollen donors is Ad, where d is the density of pollen producing plants. Because L. rubescens is distributed patchily, d was calculated as the total number of genets in the four pastures divided by the total area, defined as the space occurring within four corners represented by the pastures. Pollen from unidentified fathers was not included in the determination of paternity pools.
Reproductive success relative to floral display
Because floral display is expected to influence RS in animal-pollinated species and the floral display of a cluster (rather than that of a genet) is perceived by the pollinator, RS was also evaluated relative to floral display of clusters regardless of the number of genets contained therein. Male and female RS was assessed for every cluster within the 15 populations by comparing observed and expected RS. Observed male RS (mi) was determined as the number of fruit within the population sired by cluster i. Expected male success of each cluster was calculated as ni(M/N), where ni is the number of inflorescences produced by cluster i, M represents the total number of capsules within the overall study area sired by pollen donors located in the same population as the maternal plant, and N is the total number of inflorescences produced in the site. Significance of the deviation of observed from expected RS was calculated by 
[(mi – niM/N)2/ni(M/N)], df = 1, for three cluster size classes; 10 inflorescences, 11–30 inflorescences and >30 inflorescences. Female RS was similarly determined, with fi representing the number of viable capsules produced by cluster i. Expected female success of each cluster was calculated by ni(F/N), where F represents the total number of capsules within the study site. The significance of the deviation of observed from expected female RS was calculated by [(fi – niF/N)2/ni(F/N)], df = 1, for the three size classes of clusters. Total RS of clusters falling in these size classes was obtained by pooling male and female RS data. Analyses were performed at three spatial scales: (1) pooling all 15 populations, (2) pooling populations within pastures and analyzing the three pastures separately (AG, GT, and SP; CB was excluded because no fruit were produced in either year), and (3) analyzing three of the largest populations separately (AG-468, GT-473, and SP-462).
Reproductive cluster similarity between years
Reproductively successful clusters of orchids (i.e., successfully donated pollen to fruit within the surveyed populations and/or produced viable fruit) were compared for the 2 years to determine the proportion of clusters that transmitted gametes to the next generation in 1999 that were also reproductively successful in 2000. Female RS of genets was also compared for 1999 and 2000.
Pollen movement distances within populations
Because every flowering orchid cluster in the study site was mapped in three-dimensional space, we could generate x, y, and z coordinates for each and calculate the distance separating each cluster from every other cluster within a population. A paternity analysis allowed us to identify the pollen parent within the population and thereby calculate the distance separating maternal and paternal plants. With this information, we calculated the mean distance separating all possible pairs of clusters within populations as well as the mean distance that pollen was transported within populations in effective pollinations. The means were compared and tested for significance using a t test for independent samples (Microsoft Excel, Redmond, Washington, USA).
RESULTS
In 1999, the 15 study populations produced 1884 inflorescences, all representing possible pollen donors. Of these, 149 (7.9%) produced 211 capsules. Individual inflorescences produced up to four capsules (Trapnell and Hamrick, 2005
). An estimated 1.4% of all flowers produced mature fruit. In 2000, these populations produced 2575 inflorescences of which 211 (8.2%) produced 376 fruit. Fruit bearing inflorescences produced up to 11 capsules each. An estimated 1.8% of all flowers set fruit in 2000 (Trapnell and Hamrick, 2005
). Seedlings from 130 capsules in 1999 and 329 capsules in 2000 survived contamination and were used in paternity analyses.
Comprehensive sampling allowed us to determine the multilocus genotype of every flowering individual (i.e., every possible pollen donor) within these 15 populations, while strict correlated mating allowed us to identify the diploid genotype of the pollen parent for every fruit. Multilocus genotypes of the 459 progeny arrays (~6800 seedlings) analyzed provided no evidence of multiple paternity, thus confirming that either multiple paternity does not occur or it occurs too infrequently to affect our results. When allele frequencies of all maternal plants within the study site were compared with the allele frequencies of their pollen donors (from all sources), there was a significant difference (P = 0.05) in one of eight polymorphic loci in 1999 and one of nine polymorphic loci in 2000. However when loci were pooled within years, the difference was not significant. For each of the 2 years, there were four trees with sufficient fruit for comparison. One tree in each year (SP-462 in 1999 and GT-480 in 2000) differed significantly in male and female allele frequencies at one locus. However when loci were pooled, trees AG-468, GT-473, GT-480, and SP-462 had no significant overall difference in male and female allele frequencies in 1999 nor did trees AG-468, GT-473, and SP-462 in 2000. However, tree GT-477 did have a significant overall difference in 2000 due to allele frequency heterogeneity at one of its five polymorphic loci. When allele frequencies of maternal plants were compared between the 2 years, the difference was not significant. However, allele frequencies of 1999 pollen donors differed significantly (P = 0.01) from 2000 pollen donors.
Four mating types were identified: (1) geitonogamous selfing, (2) within tree outcrossing, (3) gene flow events (i.e., pollen originating outside of the target population) either between populations (i.e., trees) within the same pasture or between pastures, and (4) gene flow originating from unidentified sources. This last group resulted either from individuals located in unsampled trees within the study site (relatively few individuals) or from individuals located outside the study area (the majority). When more than one potential pollen parent had the same genotype, paternity was assigned conservatively to the orchid in the most local of three spatial categories relative to the maternal plant (within trees, between trees within the same pasture, or between pastures). If two or more genetically possible pollen donors occurred within the same population as the maternal plant, paternity was randomly assigned. There was no ambiguity in the reported geitonogamous selfing events. Selfing accounted for 10% and 11% of fruit produced in 1999 and 2000, respectively, and within-tree outcrossing was responsible for 56% and 55% of pollinations during the 2 years. Gene flow from identified pollen donors accounted for 17% and 16% of the fruit in the 2 years, while pollen originating from unidentified sources explained 17% and 18% of fruit production (Trapnell and Hamrick, 2005
).
Effective number of males and females
The effective number of pollen donors in populations with 10 or more fruit averaged 12.6 and 26.7 in the 2 respective years, while the mean effective number of females was 9.5 and 12.4 (Table 1). Mean effective population size for populations with 
10 fruit was 15.8 in 1999 and 24.3 in 2000 (Table 1). When the 2 years were pooled, populations with 10 fruit had a mean Ne = 23.5 (2.8–46.6). If all populations were included regardless of size, the mean effective population size when the two years were pooled was 17.0 (1.9–46.6). The number of inflorescences per population was marginally more highly correlated with the effective number of males (r2 = 0.891) than with the effective number of females (r2 = 0.841). In both cases, the relationship was significant (P = 0.01). A direct measure of migrants per generation (Nem) suggested that there were 2.03 and 2.34 migrants in 1999 and 2000, respectively (Table 1).
|
10 fruit, the effective number of males relative to fruit set was 0.6 and 0.4 in the 2 years (Table 2). Where there were <10 capsules, these values increased to 1.0 and 0.8. For the effective number of females, the values were 0.4 in 1999 and 0.2 in 2000 for populations with 10 fruit, while in trees with lower fruit set the values were 0.9 and 0.6. The ratios of reproductive individuals (i.e., genets that contributed male and/or female gametes) to fruit set for populations that had 10 fruit were 1.2 and 0.8 in 1999 and 2000, respectively (Table 2). In populations with less fruit, these values increased to 2.0 and 1.2 (Table 2). In populations with 10 capsules, the ratio of the effective number of individuals to the total number of genets was 0.2 in both years. However in populations with less fruit, the values were 0.3 and 0.2 (Table 2). The effective number of individuals pooled across all 15 populations represented 20% and 24% of all flowering genets in 1999 and 2000 (Table 2).
|
Reproductive success relative to floral display
When the 15 populations were pooled for analysis, observed total RS for small clusters (10 inflorescences) was significantly less than expected in 1999 and 2000 (Table 3). However, medium-sized clusters (11–30 inflorescences) had significantly higher than expected RS in both years (Table 3). Large clusters (>30 inflorescences) displayed no significant difference in either year. Female RS had a similar pattern of significant differences in both years, with the exception of large clusters that revealed significantly less than expected success in 2000. There were fewer significant differences reported for male RS. In 2000 small clusters had significantly less success, and large clusters had significantly greater success (Table 3).
|
Analyses of the three largest populations also had a consistent pattern of reduced RS in small and large clusters and increased RS in medium clusters. The one exception was population GT-473 where small clusters had higher than expected overall success in 2000. Again, male RS deviated significantly from expected in fewer instances.
Reproductive cluster similarity between years
In 1999, 50 clusters across the entire study site sired capsules within the same population, while in 2000, there were 75 clusters. Although there was a 50% increase in the number of clusters that contributed pollen to fruit production within their natal populations, only 40% (20) of the clusters that sired fruit in 1999 also did so in 2000. On the other hand, viable capsules were produced by 54 and 52 clusters in the two years, respectively. Sixty-seven percent (36) of the fruit bearing clusters in 1999 produced fruit the following year. Twenty of the clusters that produced fruit in both years represented the same genetic individuals.
Pollen movement distances within populations
Within populations, pollinators transported pollen a mean distance of 4.36 m (1.54–13.08) in 1999 (Table 4). This measure includes geitonogamous selfing. If only cross-pollination events are considered, then the mean distance of pollen transport was 5.85 m (2.37–13.08). The mean distance separating all possible pairs of flowering clusters within populations was 6.40 m (2.37–8.00). In 2000, pollen moved a mean distance of 4.01 m (0–4.77) or 5.31 m (1.99–6.37) if selfing is excluded (Table 4; Fig. 1), while the mean distance between all possible pairs of flowering clusters within populations was 5.88 m (3.05–7.93). Differences between mean pollen movement distances excluding selfing, and mean distances between all possible pairs of clusters within populations were not significant in either year.
|
|
Despite the impressive reproductive effort evidenced by floral display in the 15 populations (1884 inflorescences and 711 genets in 1999; 2575 inflorescences and 736 genets in 2000), effective sizes of these populations were typically small, never exceeding 36 and 46 individuals per population in the 2 respective years (Table 1). There was a mean of 63 and 56 flowering genets per population in the 2 years, whereas mean Ne values were 11 and 12 (Table 1). When both years were pooled, the mean Ne across all populations was 17.0 and 25.5 for populations with 10 fruit, suggesting that different genets were reproductively successful in the 2 years. This is consistent with the reproductive cluster similarity data showing that, although the number of clusters that sired fruit increased from 50 in 1999 to 75 in 2000, only 20 clusters were reproductively successful as males in both years. Further support is provided by the finding that 1999 pollen donor allele frequencies differed significantly from the 2000 pollen donor allele frequencies. Considering the high rates of pollen flow (34% in both years) and the increased interpopulation distances that pollen was transported in 2000 (Trapnell and Hamrick, 2005
), the difference in pollen allele frequencies is not surprising. Maternal success of individual clusters was more consistent between years. Of the 54 clusters that bore fruit in 1999, 36 also produced fruit in 2000 (of a total of 52 clusters). Twenty (56%) of the clusters that produced fruit in both years represented the same genetic individuals. There was no significant difference between 1999 and 2000 maternal allele frequencies. Thus only a small proportion (18% and 21%) of all flowering genets was effectively contributing gametes during the 2 years of this study. In all but one of the populations, the effective number of females was less than the effective number of males (Table 1). This is consistent with the observation that the effective number of males is marginally more correlated with floral display. These data indicate that, on average, fewer individuals are reproductively successful as females than males but when they are successful, they transmit more gametes to the next generation. In other words, fewer genets bear fruit than sire fruit. This is consistent with the observation of a marginally lower coefficient of variation in female RS (2.90) than male RS (3.01) over the 2 years. The most successful maternal individuals produced 11 and 20 capsules in the two respective years, while the most successful pollen donors sired nine and 11 capsules. In 1999, a single genet contributed a total of 18 male and female gametes to mature capsules, while in 2000 one individual contributed 25 male and female gametes.
The pool of potential pollen donors increased considerably in the second year for two of the three pastures (49-fold increase for AG and three-fold increase for GT), while the paternity pool actually decreased by 14% for SP. If the distances for pollen donors from unidentified sources had been included, the paternity pools would probably be larger, particularly in 2000 when pollen immigration distances increased (Trapnell and Hamrick, 2005
). That genet density was virtually identical in 1999 and 2000 suggests the increase in overall floral display was responsible for the increase in paternity pools. Pasture SP had a particularly large number of pollen donors from unidentified locations in 2000 and would probably have also shown an increase in its potential paternity pool if these locations had been identifiable.
The relationship between RS and floral display within each of the three cluster size classes was surprisingly consistent between years and spatial scales. Small clusters with 10 inflorescences consistently had significantly lower than expected female and overall RS, while medium-sized clusters consistently displayed significantly greater female and overall success. In every instance that medium-sized clusters had a significant difference, RS exceeded expectation. Large clusters with >30 inflorescences were less successful than expected, with few exceptions. Male RS rarely differed significantly from expected regardless of spatial scale and cluster size class. This is consistent with Asclepias exaltata where female RS was more closely correlated with flower number/plant than male RS and where functional males outnumbered functional females (Broyles and Wyatt, 1990
). Our data suggest that pollinators preferentially visit larger clusters with >10 inflorescences.
Laelia rubescens flowers over an extended period, thus not all inflorescences produced in a growing season flower synchronously. Therefore all clusters will have smaller floral displays at any one time than the total number of inflorescences would suggest. This will have the greatest impact on small clusters that produce 10 inflorescences in a growing season; with a sequential floral display, only one or a few inflorescences may be mature at any time. With an optimal forager, such as hummingbirds, it is understandable that small clusters are visited less. The reduced RS of the largest clusters has two possible explanations. Several studies have shown that visitation per flower declines in very large floral displays (e.g., Robertson and Macnair, 1995
). Alternatively, these results may suggest that inbreeding depression is occurring. Large clusters are the most likely to have within cluster (i.e., within genet) pollinations and therefore selfing via geitonogamy (de Jong et al., 1992
; Harder and Barrett, 1995
; Vrieling et al., 1999
; Galloway et al., 2002
). Even though some selfed fruit mature, as evidenced by the observation of 10% and 11% geitonogamous selfing in 1999 and 2000, respectively (Trapnell and Hamrick, 2005
), it is possible that many fruits resulting from selfing suffer from reduced fitness and abort. Intermediate-sized clusters will also have some within-cluster gene movement. However, our data show that multiple genotypes are common within a single cluster (as many as nine genotypes; Trapnell et al., 2004
). Therefore clusters with 11–30 inflorescences (i.e., fewer inflorescences per genotype) are less likely to have matings between inflorescences belonging to the same genotype. Large clusters will also often have multiple genotypes; however, due to their size, it is more likely that multiple inflorescences with the same genotype will be receptive simultaneously, increasing selfing and subsequent abortion. Thus if selfed flowers suffer inbreeding depression, clusters with larger floral displays would experience reduced male and female fitness relative to the medium-sized clusters.
The occurrence of multiple genotypes within a cluster (Trapnell et al., 2004
) has some interesting ramifications. An established cluster of L. rubescens sometimes serves as a nurse plant to conspecifics, providing multiple advantages to recruiting seedlings. In the harsh environment of the seasonally dry tropical forest, an established cluster provides more nutrients, moisture, shade, and lower temperatures than nearby bark substrate. The nurse plant also has the specific mycorrhizal association that is essential to orchid germination. An additional advantage is that a seedling establishing in a larger cluster can benefit from the larger floral display and perhaps become reproductively successful sooner than if it had established on unoccupied substrate. However, recruits will incur a cost to the nurse plant by competing for and reducing available resources. Ultimately though, both individuals will benefit from an increased floral display that will enhance the cluster's attractiveness to pollinators. Even with pollinator movement within the cluster, because there are multiple genotypes present and consequently the possibility of outcrossing, the likelihood of RS is increased for each genotype.
Pollen movement distances within populations also showed a consistent but non-significant trend. With increased floral display in the second year of the study, pollinators generally transported pollen over shorter distances within populations. This is contrary to interpopulation pollen movement patterns observed in 1999 and 2000 whereby pollen was transported across greater distances in 2000 when there was a substantially larger floral display (Trapnell and Hamrick, 2005
). However, the pattern of intrapopulation movement is understandable in light of optimal forager behavior. When more inflorescences are available simultaneously, an optimal forager can and will move shorter distances within the population. However, when a floral display is smaller and fewer inflorescences are available at any one time, the pollinator has to move further to forage effectively.
The availability of the full-sib paternity analysis for this species has allowed an unusually detailed examination of mating patterns within its populations. This study provides insights into pollinator movement patterns, as well as patterns of flowering and their effects on RS. Our data have demonstrated that it is the cluster, not the individual genotype, that is the important functional unit in regards to RS. This is the first examination of intrapopulation mating patterns of an orchid using direct genetic measures. Furthermore, because the canopy of flowering trees is also three-dimensional, our data may provide insights into pollen movement patterns within canopies of zoophilous trees, which, because of the genetic uniformity of the canopy, cannot be studied directly with genetic markers.
FOOTNOTES
1 The authors wish to thank G. Farmer, P. Salazar, K. Hamrick, and numerous workers at Hacienda Solimar for field assistance; R. Pappert, C. Deen, G. Farmer, and other Hamrick lab personnel for valuable lab assistance; R. Clay for orchid seed germination guidance; O. Pacheco for permission to work at Hacienda Solimar and three anonymous reviewers for their valuable comments. This research was supported by NSF Grants (DBI-9602223 and DEB-0211526 to J. L. H.), and NSF Doctoral Dissertation Improvement Grant (022602-01), Botanical Society of America/Karling Student Research Award, Andrew W. Mellon Foundation Organization for Tropical Studies Research Fellowship, and University of Georgia Botany Department Palfrey Grant to D. W. T. 
4 Author for correspondence (dorset{at}plantbio.uga.edu
)
LITERATURE CITED
Ackerman J. D.. 1989. Limitations to sexual reproduction in Encyclia krugii (Orchidaceae). Systematic Botany 14: 101-109.
Bateman A. J.. 1947. Contamination of seed crops. 1. Insect pollination. Journal of Genetics 48: 257-275.[ISI]
Broyles S. B. Wyatt R.. 1990. Paternity analysis in a natural population of Asclepias exaltata: multiple paternity, functional gender, and the "pollen-donation hypothesis.". Evolution 44: 1454-1468.[CrossRef][ISI]
Charpentier A.. 2002. Consequences of clonal growth for plant mating. Evolutionary Ecology 15: 521-530.[CrossRef][ISI]
Cheliak W. M. Pitel J. A.. 1984. Techniques for starch gel electrophoresis of enzymes from forest tree species. Petawawa National Forestry Institute, Information Report P1-X-42, 1–49 Canadian Forestry Service, Agriculture, Chalk River, Ontario, Canada.
Dafni A.. 1984. Mimicry and deception in pollination. Annual Review of Ecology and Systematics 15: 259-278.
de Jong T. J. Waser N. W. Price M. V. Ring R. M.. 1992. Plant size, geitonogamy and seed set in Ipomopsis aggregata. Oecologia 89: 310-315.[ISI]
Firmage D. H. Cole F. R.. 1988. Reproductive success and inflorescence size of Calopogon tuberosus (Orchidaceae). American Journal of Botany 75: 1371-1377.[CrossRef][ISI]
Galloway L. F. Cirigliano T. Gremski K.. 2002. The contribution of display size and dichogamy to potential geitonogamy in Campanula americana. International Journal of Plant Sciences 163: 133-139.[CrossRef][ISI]
Gentry A. H. Dodson C. H.. 1987. Diversity and biogeography of neotropical vascular epiphytes. Annals of the Missouri Botanical Garden 74: 205-233.[CrossRef][ISI]
Halbinger F. Soto M.. 1997. Laelias of Mexico Herbario AMO, Mexico City, Mexico.
Harder L. D. Barrett S. C. H.. 1995. Mating cost of large floral displays in hermaphrodite plants. Nature 373: 512-515.[CrossRef]
Harder L. D. Barrett S. C. H.. 1996. Pollen dispersal and mating patterns in animal-pollinated plants. In D. G. Lloyd, S. C. H. Barrett [eds.], Floral biology: studies on floral evolution in animal-pollinated plants 140-190 Chapman & Hall, New York, New York, USA.
Kearns C. A. Inouye D. W.. 1993. Techniques for pollination biologists University Press of Colorado, Niwot, Colorado, USA.
Levin D. A.. 1988. The paternity pools of plants. American Naturalist 132: 309-317.[CrossRef][ISI]
Linhart Y. B. Busby W. H. Beach J. H. Feinsinger P.. 1987. Forager behavior, pollen dispersal, and inbreeding in two species of hummingbird-pollinated plants. Evolution 41: 679-682.[CrossRef][ISI]
Manchenko G. P.. 1994. Handbook of detection of enzymes on electrophoretic gels CRC Press, Boca Raton, Florida, USA.
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.
Mora de Retana D. Atwood J. T.. 1992. Icones plantarum tropicarum. Fascicle 15. Orchids of Costa Rica, part 2 Marie Selby Gardens, Sarasosta, Florida, USA.
Nilsson L. A.. 1992. Orchid pollination biology. Trends in Ecology and Evolution 7: 255-259.[CrossRef]
Paetkau D. Strobeck C.. 1994. Microsatellite analysis of genetic variation in black bear populations. Molecular Ecology 3: 489-495.[Medline]
Robertson A. W. Macnair M. R.. 1995. The effects of floral display size on pollinator service to individual flowers of Myosotis and Mimulus. Oikos 72: 106-114.[CrossRef][ISI]
Sader S. A. Joyce A. T.. 1988. Deforestation rates and trends in Costa Rica, 1940 to 1983. Biotropica 20: 11-19.
Schemske D. W.. 1980. Evolution of floral display in the orchid Brassavola nodosa. Evolution 34: 489-493.[CrossRef][ISI]
Schmitt J. Eccleston J. Ehrhardt D. W.. 1987. Density dependent flowering phenology, outcrossing, and reproduction in Impatiens capensis. Oecologia 72: 341-347.[CrossRef][ISI]
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][ISI]
Trapnell D. W. Hamrick J. L.. 2004. Partitioning nuclear and chloroplast variation at multiple spatial scales in the neotropical epiphytic orchid, Laelia rubescens. Molecular Ecology 13: 2655-2666.[CrossRef][Medline]
Trapnell D. W. Hamrick J. L.. 2005. Mating patterns and gene flow in the neotropical epiphytic orchid, Laelia rubescens. Molecular Ecology 14: 75-84.[CrossRef][Medline]
Trapnell D. W. Hamrick J. L.. In press. Diversity of phorophytes colonized by the neotropical epiphytic orchid, Laelia rubescens. Selbyana.
Trapnell D. W. Hamrick J. L. Nason J. D.. 2004. Three-dimensional fine-scale genetic structure of the neotropical epiphytic orchid, Laelia rubescens. Molecular Ecology 13: 1111-1118.[CrossRef][Medline]
Vrieling K. Saumitou-Laprade P. Cuguen J. van Dijk H. de Jong T. J. Klinkhamer P. G. L.. 1999. Direct and indirect estimates of the selfing rate in small and large individuals of the bumblebee pollinated Cynoglossum officinale L. (Boraginaceae). Ecology Letters 2: 331-337.[CrossRef][ISI]
Weeden N. F. Wendell J. F.. 1983. Genetics of plant isozymes. In D. E. Soltis, P. S. Soltis [eds.], Isozymes in plant biology 46-72 Dioscorides Press, Portland, Oregon, USA.
Widen B.. 1993. Demographic and genetic effects on reproduction as related to population size in a rare, perennial herb, Senecio integrifolius (Asteraceae). Biological Journal of the Linnean Society 50: 179-195.[CrossRef]
Williams L. O.. 1946. Flora of Panama Part III. Annals of the Missouri Botanical Garden 33: 315-404.[CrossRef]
Wright S.. 1951. The genetical structure of populations. Annals of Eugenics 15: 323-354.[ISI]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
