Am. J. Bot. Join the BSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (15)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Stacy, E. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Stacy, E. A.
Agricola
Right arrow Articles by Stacy, E. A.
(American Journal of Botany. 2001;88:1041-1051.)
© 2001 Botanical Society of America, Inc.

Population Biology

Cross-fertility in two tropical tree species: evidence of inbreeding depression within populations and genetic divergence among populations1

Elizabeth A. Stacy

Department of Biology, Boston University, 5 Cummington Street, Boston, Massachusetts 02215 USA

Received for publication March 4, 2000. Accepted for publication September 28, 2000.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Knowing the spatial patterns of cross-fertility in natural plant populations yields key insight into biparental inbreeding depression, isolation by distance, and, ultimately, speciation. Three adults each of two tropical tree species (Syzygium rubicundum and Shorea cordifolia) were each crossed with five conspecific pollen donors ranging from self to trees occurring in separate forest reserves (12 and 35 km distance for S. rubicundum and Sh. cordifolia, respectively). Cross-fertility was estimated as fruit set, seed germination, and seedling survivorship and height at 1 yr. Means of most cross-fertility measures increased steadily with outcrossing distance, peaking at 1–2 km for S. rubicundum and 1–10 km for Sh. cordifolia, and then declining at the between-forest crosses. However, seed germination and seedling height for Sh. cordifolia suggested hybrid vigor in between-forest crosses. The mean fitness cost of nearest-neighbor mating relative to crossing with more distant neighbors was 45% for S. rubicundum and 0% for Sh. cordifolia. The mean fitness cost of between-forest crosses was 52% and 70% for the two species. Crossing effects on fitness diminished between the stages of fruit set and 1-yr-old seedlings. Results indicate a strong potential for inbreeding depression within forest tree populations and partial reproductive isolation among forests in Sri Lanka's wet zone.

Key Words: hand-pollination • inbreeding depression • outbreeding depression • Shorea • Sinharaja • Sri Lanka • Syzygium • tropical trees


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Estimating cross-fertility at multiple spatial scales within a plant species' range allows insight into the significance of genetic incompatibility and inbreeding and outbreeding depression under natural conditions. At smaller spatial scales, the restricted seed dispersal characteristic of many plants should result in relatedness among near-neighbors in natural populations. Where neighboring plants are related, mating between neighbors (i.e., biparental inbreeding; Uyenoyama, 1986 ) may lead to reduced reproductive fitness due to inbreeding depression or genetic incompatibility between mates. While patterns of fine-scale genetic relatedness have been revealed for a number of herbaceous species (e.g., Epperson and Clegg, 1986 ; Waser, 1987 ; Schoen and Latta, 1989 ), for woody species, particularly angiosperms, data on within-population genetic structure are limited. Some studies have revealed spatial clustering of adult genotypes (or alleles) in animal-pollinated forest trees (Acer saccharum—Perry and Knowles, 1991 ; Swartzia simplex—Hamrick, Murawski, and Nason, 1993 ; Quercus rubra—Sork, Huang, and Wiener, 1993 ; Quercus laevis—Berg and Hamrick, 1995 ; Cordia alliodora—Boshier, Chase, and Bawa, 1995 ), while others have failed to support the expectation of genetic relatedness among neighboring adults (Psychotria nervosa—Dewey and Heywood, 1988 ; Alseis blackiana and Platypodium elegans—Hamrick, Murawski, and Nason, 1993 ). Regardless of patterns of relatedness, mating is often restricted to local neighborhoods in herbaceous species (Levin and Kerster, 1974; Endler, 1979 ; Waser and Price, 1983 ) and in woody species, particularly when flowering adults are clustered (Linhart, 1973 ; Stacy et al., 1996 ). The potential for biparental inbreeding depression in natural populations has been demonstrated for a number of herbaceous and coniferous tree species (Coles and Fowler, 1976 ; Park and Fowler, 1982 ; Levin, 1984 ; Schemske and Pautler, 1984 ; McCall, Mitchell-Olds, and Waller, 1991 ; Heywood, 1993 ; Waser and Price, 1994 ; Trame, Coddington, and Paige, 1995 ; Byers, 1998 ). Very little is known, however, of biparental inbreeding depression in natural populations of woody angiosperms.

Inbreeding depression is frequently cited as an unavoidable consequence of anthropogenic disturbance to tropical forests (e.g., forest fragmentation, logging), where theory predicts that normal mating patterns within already low-density tree populations are shifted to favor short-distance crosses. To date, however, the consequences of elevated near-neighbor mating for population fitness in tropical trees have yet to be quantified empirically. Two fundamental questions to be addressed are: Do adults avoid maturing seed derived from near-neighbor crosses and, if not, how fit are near-neighbor-derived progeny relative to others? This study assesses the consequences of near-neighbor mating in two tropical tree species directly through fitness comparisons of crosses between nearest neighbors with crosses involving more distant mates.

At larger spatial scales, geographically separated populations may differ genetically due to random genetic drift or selection (Wright, 1943 ). Crosses between members of widely separated populations, therefore, may yield progeny of suboptimal fitness due to outbreeding depression (Bateson, 1978 ; Price and Waser, 1979 ). Most evidence of outbreeding depression in plants derives from crosses between populations (e.g., Ritland and Ganders, 1987 ; Sobrevila, 1988 ; Fischer and Matthies, 1997 ). However, evidence of outbreeding depression within populations has been reported for a few species (Waser and Price, 1983, 1989, 1991, 1993, 1994 ; Schemske and Pautler, 1984 ; Waser et al., 1987 ; McCall, Mitchell-Olds, and Waller, 1988 ). Ultimate consequences of outbreeding depression should include reinforcement of reproductive isolation by distance and increased genetic differentiation over some spatial scale, possibly leading to speciation. In tropical forests, models for the evolution of high tree species diversity typically invoke genetic divergence of populations over some spatial scale (e.g., Fedorov, 1966 ; Ashton, 1969 ). Very little is known, however, of the spatial scale of genetic divergence in tropical tree species.

While trends toward outbreeding depression have been documented for a number of plant species (citations above), there are no published reports of outbreeding depression in any forest tree (Coles and Fowler, 1976 ; Park and Fowler, 1982, 1984 ; Crome and Irvine, 1986 ; Hardner, Potts, and Gore, 1998 ). This study assesses the significance and approximate spatial scale of outbreeding depression (as a proxy for genetic divergence) in two tropical tree species through fitness comparisons of long-distance crosses with crosses over short and moderate distances.

For each of two tropical tree species, the following questions were addressed: (1) How does cross-fertility vary with outcrossing distance? (2) What is the potential for inbreeding depression in near-neighbor crosses? (3) What is the potential for outbreeding depression over greater distances within and between forest reserves? and (4) How do the strengths of inbreeding and outbreeding effects vary among early life history stages, including seed set, seed germination, and the survivorship and growth of seedlings?


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 LITERATURE CITED
 
Study site and species
This work was done in the area of the aseasonal Sinharaja World Heritage Site, an 8800-ha, UNESCO Man and the Biosphere Reserve, located in the southwest wet zone of Sri Lanka (Fig. 1). The forests of southwest Sri Lanka are severely fragmented as a result of clearing for coffee and rubber plantations that took place during the British colonial period (>100 yr ago). Compared to many tropical rain forests, the flora of Sinharaja is relatively species-poor (~230 species of woody plants), but characterized by high endemism (Gunatilleke and Gunatilleke, 1980 ). The study species (Syzygium rubicundum and Shorea cordifolia) were selected to represent two of the major canopy-dominant, economically important genera of the island's rain forests. Both species are pollinated predominantly by bees (Apis spp.) and produce single-seeded fruits. The two species differ in mode of seed dispersal (bird- or bat-dispersal vs. wind- or gyration-dispersal) and are expected to exhibit different population genetic structures in natural stands (Loveless and Hamrick, 1984 ; Hamrick, Murawski, and Nason, 1993 ). The populations targeted for this study, however, occurred in 25+-yr-old logged forest at Sinharaja Reserve. Adults of each species are highly clumped in the logged forest relative to their distributions in unlogged stands. This contrast suggests that the population structures of these species have been altered in the high-light, post-logging environment, where regeneration likely resulted in the establishment of groups of sibs or otherwise related individuals. The study populations, therefore, offered an ideal opportunity to evaluate the potential for biparental inbreeding depression in forest trees.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 1. Schematic of Sri Lanka showing the general locations of pollen donors and maternal trees used in the cross-fertility experiment. Maternal trees occurred at Sinharaja (S): 400–500 m above sea level. Pollen donors occurred at Pitikale (P): 350–500 m, Morapitiya (M): 300 m, Walankanda (W): 600–900 m, Gilimale (G): 400–500 m, and Sinharaja. Sinharaja Reserve is located between 8°21' and 8°34' E and between 6°21' and 6°26' N.

 
Syzygium rubicundum (Myrtaceae), which typically flowers annually, is a dominant canopy species of mid-slope areas in and around Sinharaja. Adult density is highly variable in both unlogged and logged forest within Sinharaja Reserve (Stacy, Harischandran, and Gunatilleke, in press ). Flowers of this species are small and brush-like, and the drupes are primarily bird-dispersed.

Shorea cordifolia (Dipterocarpaceae) is a locally abundant main canopy species that flowers heavily at irregular supra-annual intervals (I. A. U. N. Gunatilleke et al., unpublished data). Flowers of this species are white and short-lived, and the winged fruits are dispersed by wind or gravity. Because of its highly restricted seed dispersal, genetic relatedness among near neighbors in natural forest is expected to be high. In logged forest at Sinharaja, Sh. cordifolia usually occurs in clumps of ~5–20 adults, intermixed with smaller stems (personal observation).

Cross-fertility experiments
In January 1997 three adults of each study species were selected as maternal trees for the experimental hand-pollinations. Because flowering is roughly simultaneous among maternal trees and 2–3 workers are required to complete the experimental pollinations at each maternal tree each day, working with more than three maternal trees per species was not feasible. To permit daily access to flowers, a platform was constructed in the canopy of each maternal tree, accessible by a series of interlocking aluminum ladders wired to the bole. Safety equipment was used for climbing and while working on platforms.

Throughout the period of flowering (~2 wk per individual), each maternal tree was hand-pollinated daily using pollen from five donors representing a range of crossing distances. Pollen donors included the maternal tree itself, its nearest neighbor, and three progressively more distant trees. The most distant pollen donors used occurred 12 and 35 km from the maternal trees, for S. rubicundum and Sh. cordifolia, respectively (Table 1). Selection of pollen donors was restricted by the spatial distribution of forest fragments and populations of the study species. For both species, all but the most distant pollen donors used (i.e., those from Walankanda for S. rubicundum and Gilimale for Sh. cordifolia) occurred within the Sinharaja Forest Reserve. Both Walankanka and Gilimale refer to large forest reserves that are not contiguous with Sinharaja (Fig. 1). In addition to tests of self- and cross-fertility, tests of apomixis and autogamy were done for S. rubicundum, as the breeding system of this species was unknown. Ultimately, a few to several hundred flowers were treated with pollen from each donor at each maternal tree (Table 1). Additionally, a large number of untreated and unbagged flowers on five of the six maternal trees were monitored for estimation of fruit set rates under natural, open-pollinated conditions (Table 1).


View this table:
[in this window]
[in a new window]
  		Table 1. Number of flowers pollinated and corresponding crossing distances for each combination of maternal tree and pollen donor of (A) Syzygium rubicundum and (B) Shorea cordifolia. Shared uppercase superscripts within a treatment indicate the sharing of a pollen donor by maternal trees. Each maternal tree had its own unique pollen donor for the selfed, nearest-neighbor, and distant-neighbor treatments

 
The crown of each maternal tree was divided into five sections (branches). Experimental inflorescences for each pollination treatment were then selected randomly, such that each branch included all five treatments. This approach avoided within-crown position effects on seed set and facilitated analysis of variance of fruit set rate. Paper pollinator exclusion bags were used to prevent pollen contamination of experimental flowers by visiting insects.

To permit daily access to pollen donors, large, flower-bearing branches were collected from each pollen donor (including each maternal tree, for self-pollination treatments) with the aid of local tree climbers. Branches were placed cut-end in a stream near Sinharaja Research Station and replaced every 5 d. During this time, blooming of inflorescences on these branches proceeded normally and was indistinguishable from that of intact branches. Throughout the flowering period, newly opened flowers were collected every morning from bagged inflorescences on pollen branches and transported directly to the maternal trees. For each pollen donor, hand crosses were conducted by touching the stigmas of receptive flowers into pollen collected on a glass microscope slide.

At each maternal tree, developing fruits on experimental branches were counted every 5 d throughout the period of heaviest fruit loss (first 45 d), and then every 10 d until fruits were fully mature. Mesh bags were placed on experimental branches just prior to fruit maturation to catch falling fruits. All seeds resulting from hand-pollinations were sown in a temporary greenhouse under partial shade (~50% sunlight) for estimates of percentage seed germination, and rates of survival and growth of seedlings over a 1-yr period (washed seeds of S. rubicundum and whole, single-seeded fruits of Sh. cordifolia). Seed trays and seedlings were labelled by maternal tree and crossing treatment. The greenhouse was located at Sinharaja Field Research Station, and seeds and seedlings were sown and potted, respectively, in local soil. Conditions for estimation of progeny fitness were therefore not unlike those in the maternal environment.

Data analysis
A standardized, cumulative index of cross-fitness was calculated for each combination of maternal tree and pollen donor, based on mature fruit set, seed germination, and survivorship and growth of seedlings. For each species, mixed-model analysis of variance was used to assess the effects of crossing treatment (fixed effect; with maternal tree included as a random effect) on the percentage of hand-pollinated flowers setting mature fruit, rates of seed germination and seedling survivorship, seedling size at 1 yr, and cumulative fitness. Several models were tested using ANOVA: (a) including all treatments, (b) excluding unbalanced treatments, to permit evaluation of interaction terms, (c) minus selfing treatment (as maternal trees were largely or completely self-incompatible), and (d) grouping all within-Sinharaja outcrossing treatments to test the effect of within- vs. between-forest crossing. The effect of crossing distance on each parameter was further tested using linear or quadratic regression analysis, depending on the shape of the relationship. To improve normality, all proportion data were transformed prior to analyses. Lastly, for each maternal tree, the consequences of nearest-neighbor and long-distance mating were estimated through indices of biparental inbreeding depression and outbreeding depression, respectively, based on cumulative fitness values.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 LITERATURE CITED
 
Fruit set
Syzygium rubicundum
Fruit abortion was heavy for all trees, resulting in low fruit set (range across treatments: 2.0–9.7%; Fig. 2a). The timing of abortion was not discernable across treatments. Self-compatibility was low, but variable, across maternal trees (Fig. 2a). Flowers used for tests of apomixis (N = 360) and autogamy (N = 582) failed to set fruit. All analyses of variance in fruit set revealed a highly significant treatment effect and significant maternal tree effect, but no significant interaction between treatment and maternal tree (Tables 2A and 3A). For all three trees, the percentage of experimental flowers setting mature fruit showed a consistent increase with crossing distance, followed by a severe decline in fruit set with the distant between-forest treatment (Fig. 2a). The relationship between crossing distance and fruit set was nearly identical for the three maternal trees and significant with or without the self-pollinated treatment included in the model (quadratic regression model: arcsine square-root [fruit set] = crossing distance [km] + crossing distance2; results without self-pollinated treatment: F2,57 = 8.25, P < 0.0007, R2 = 0.47). Peak mean fruit set occurred at a crossing distance of 1–2 km (distant within-forest treatment) and was 1.7–4.7 times greater than mean fruit set rates for other hand-pollination treatments, averaged across maternal trees. Mean fruit set rate for the distant within-forest treatment was significantly greater than those for all treatments except distant-neighbor and open-pollinated, but consistently exceeded fruit set of open-pollinated flowers (Fig. 2a).



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 2. Percentage of treatment flowers yielding mature fruit from six experimental crosses using (a) Syzygium rubicundum and (b) Shorea cordifolia. Mean, high, and low values for three maternal trees are shown for each species. Pooled sample sizes are shown in parentheses. Shared superscripts indicate no significant difference between means. Figure Abbreviations: Treatments: SELF = self, NNHBR = nearest neighbor, DNHBR = distant neighbor, D-WF = distant within-forest (i.e., distant pollen donor occurring within Sinharaja Forest), D-BF = distant between-forest (i.e., distant pollen donor selected from a separate forest reserve), OPEN = open-pollinated flowers.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Summary of analyses of variance for mature fruit set in (A) Syzygium rubicundum and (B) Shorea cordifolia. Data were arcsine square-root transformed prior to analysis

 
Shorea cordifolia
Fruit set was also low for Sh. cordifolia (range across treatments: 0–5.3%; Fig. 2b). Again, the timing of fruit abortion was not discernable among treatments. Selfed and distant between-forest treatments resulted in 0% and <1% fruit set, respectively. Fruit set from the intermediate-distance cross-pollinations varied across maternal trees, but with one exception (nearest-neighbor treatment at Tree number 1) indicated optimal fruit set at an outcrossing range of ≥2 km (distant neighbor treatment; Fig. 2b). All analyses of variance in fruit set revealed a highly significant treatment effect, but no maternal tree effect (Tables 2B and 3B). The relationship between crossing distance and fruit set was significant only when the selfed treatment was excluded (quadratic regression model: arcsin square root [fruit set] = crossing distance [km] + crossing distance2; F2,57 = 5.71, P < 0.006, R2 = 0.41). At each maternal tree, fruit set rate for open-pollinated flowers was greater than that for all hand-cross treatments, suggesting that some aspect of the hand-pollination procedure (e.g., flower handling, bagging) caused reduced fruit set in Sh. cordifolia.

For both species, within-treatment variation among maternal trees in fruit set was substantial for all outcrosses involving pollen donors within Sinharaja Reserve. In contrast, variation in fruit set rate was very low for between-forest crosses (Fig. 2). For Sh. cordifolia, fruit set for the distant between-forest treatment ranged from only 0.5 to 0.6% and was significantly lower than the mean fruit set rate for all within-forest outcrossing treatments combined (mean = 2.71%, F1,58 = 9.94, P < 0.0003). For S. rubicundum, mean fruit set for the distant between-forest treatment (2.67%) was low relative to mean fruit set rate for all within-forest outcrossing treatments combined (mean = 5.97%). The difference was nearly significant (F1,58 = 3.78, P < 0.06).

Seed germination
Syzygium rubicundum
Seed germination was highly staggered in both natural (Stacy, Harischandran, and Gunatilleke, in press ) and greenhouse settings, extending over >4 mo. At 100 d after sowing (when germination was ~95% complete), the percentages of seeds germinated were moderate, ranging across treatments from 21 to 49% of sown seeds (mean = 36%; pooled across maternal trees; Fig. 3a). Excepting the unusually high germination rate observed for the nearest-neighbor treatment at Tree number 3 (83.3%), the relationship between crossing distance and seed germination was nearly identical to that for fruit set for this species (Fig. 3a). Including all data points, mean seed germination was not significantly different among treatments, nor among maternal trees (Table 3A). Excluding the single outlier did not change the ANOVA results. Among hand-pollination treatments, peak mean seed germination coincided with peak fruit set at an outcrossing distance of 1–2 km and was nearly identical to the mean germination of open-pollinated seeds (Fig. 3a). The relationship between crossing distance and seed germination was not significant using either a linear or quadratic regression model.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 3. Percentage germination of seeds derived from six crossing experiments using (a) Syzygium rubicundum and (b) Shorea cordifolia. Mean, high, and low values for three maternal trees are shown for each species. Pooled sample sizes are shown in parentheses. Shared superscripts indicate no significant difference between means. Germination was monitored in a greenhouse for 120 d and 35 d for S. rubicundum and Sh. cordifolia, respectively. For S. rubicundum, percentage germination estimates for open-pollinated seeds were obtained from a separate study (Stacy, Harischandran, and Gunatilleke, in press )

 

View this table:
[in this window]
[in a new window]
  		Table 3. Summary of one-way analyses of variance for five fitness measures in (A) Syzygium rubicundum and (B) Shorea cordifolia. Prior to analysis, mature fruit set and seedling survivorship data were arcsine square-root transformed, seed germination data were square-root transformed, and cumulative fitness scores were log10 transformed

 
Shorea cordifolia
Germination of experimental seeds was rapid and fairly even across treatments (mean = 62%; range: 55–72% at 45 d; Fig. 3b). A steady increase in mean seed germination with outcrossing distance is apparent, but not significant (F1,9 = 1.02, P < 0.34, R2 = 0.40; Fig. 3b). Mean seed germination was not significantly different among treatments, nor among maternal trees (Table 3B). Interestingly, mean seed germination for all hand-pollination treatments exceeded that of open-pollinated seeds (Fig. 3b). While the reason for this observation is not known, the latter estimates were based on much larger sample sizes and therefore may be more robust.

Seedling survivorship and growth
Due to attrition from one life stage to the next, sample sizes of some treatment classes were low for estimates of seedling survivorship and growth, which may have impeded statistical tests. No significant effects of either crossing treatment or maternal tree were found for either measure in either species (Table 3). Excluding selfed treatment seedlings of S. rubicundum (N = 2), the mean survivorship rate of seedlings throughout their first year was high (>80%) for all treatments of both species. For both species, peak mean seedling survivorship coincided with the distant within-forest treatment.

For S. rubicundum, mean seedling height at 1 yr postgermination increased with outcrossing distance, peaking at an intermate distance of 1–2 km, and dropping slightly with the longest distance crosses (Fig. 4a). For Sh. cordifolia, mean seedling height increased steadily with outcrossing distance, peaking at the longest-distance cross (Fig. 4b). For both species, within-treatment variation in seedling height was high (Fig. 4). For S. rubicundum, variation in seedling height can be explained, at least in part, by staggered seed germination.



View larger version (19K):
[in this window]
[in a new window]
  		Fig. 4. Box plots of plant height at ~1 yr post-emergence for seedlings derived from crossing experiments with (a) Syzygium rubicundum and (b) Shorea cordifolia. Solid squares and boxes indicate mean values ± 1 SE; whiskers show high and low data points. Pooled sample sizes are shown in parentheses. Shared superscripts indicate no significant difference between means. For S. rubicundum, the SELF treatment comprised only a single seedling (34.0 cm in height) and is not shown

 
Cumulative fitness
For each species, the cumulative fitness of each treatment class was calculated as the standardized product of the four fitness measures (see Fig. 5). For both species, the distribution of cumulative fitness with respect to crossing distance was similar to that for fruit set (Fig. 5). For S. rubicundum, variation in cumulative fitness among treatments was highly significant (Table 3A); mean cumulative fitness of the distant within-forest treatment was 4–25 times the means for other hand-pollination treatments and twice that of open-pollinated flowers. The relationship between outcrossing distance and cumulative fitness was nearly significant (quadratic regression: F2,10 = 3.84, P = 0.056). Mean cumulative fitness (relative) was not significantly different between the distant between-forest treatment (mean = 0.07) and all within-forest outcrossed classes combined (mean = 0.24, F1,12 = 1.69, P = 0.219), due to the strong near-neighbor effect on cumulative fitness.



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 5. Cumulative indices of crossing fitness for (a) Syzygium rubicundum and (b) Shorea cordifolia. Mean, high, and low values for three maternal trees are shown for each species. Pooled sample sizes are shown in parentheses. Shared superscripts indicate no significant difference between means. Each index equals the standardized product of proportion fruit set, proportion seed germination, proportion seedling survivorship, and relative height at 1 yr (i.e., height relative to the tallest seedling present). Per-treatment mean seedling survivorship rates were substituted for actual treatment-by-maternal tree means, as the latter, due to small samples sizes, were not robust

 
For Sh. cordifolia, cumulative fitness was maximum at both the distant neighbor and distant within-forest treatments (Fig. 5). Peak cumulative fitness observed for these intermediate crossing treatments was 2 and 4.3 times the cumulative fitnesses of the nearest-neighbor and distant between-forest treatments, respectively, although these differences were not significant (Table 3B). The relationship between outcrossing distance and cumulative fitness was not significant (quadratic regression: F2,8 = 3.70, P = 0.073). Mean cumulative fitness of open-pollinated flowers exceeded that of all other hand-pollination treatments, again likely due to the detrimental effect of the hand-pollination procedure on fruit set in this species. As for fruit set, mean cumulative fitness (relative) of the distant between-forest treatment (mean = 0.12) was significantly lower than that of all within-forest outcrossed classes combined (mean = 0.51, F1,12 = 8.77, P = 0.012).

Inbreeding and outbreeding depression
For each maternal tree, the consequences of nearest-neighbor and long-distance mating were estimated through indices of biparental inbreeding depression and outbreeding depression, respectively, based on cumulative fitness values. As the study species were largely or completely self-infertile, inbreeding depression through selfing was ignored. Biparental inbreeding depression was strong but variable for S. rubicundum, for which mating with the nearest neighbor represented an average fitness cost of 45% (range = –46 to +94%), relative to crossing with moderately more distant neighbors (Fig. 6). For Sh. cordifolia, the nearest-neighbor mating effect was more ambiguous and variable across trees; combining all measures of crossing fitness, inbreeding depression averaged close to zero (–6%, range = –109 to +92%; Fig. 6). In contrast, the effect of between-forest crossing was substantial for both species (mean outbreeding depression = 52 and 70% for S. rubicundum and Sh. cordifolia, respectively; Fig. 6).



View larger version (34K):
[in this window]
[in a new window]
  		Fig. 6. Mean (±1 SE) cumulative cost of mating with the nearest neighbor (biparental inbreeding depression; open columns) and a pollen donor from a separate forest reserve (outbreeding depression; shaded columns) for both study species. Indices were calculated for each maternal tree as: 1 – (Wextreme/Wmoderate), such that W = mean cumulative fitness, "extreme" = either nearest neighbor (NNHBR) or distant between-forest (D-BF) cross, and "moderate" = moderate-distance cross. Both indices can be calculated multiple ways, depending on the definition of "moderate-distance cross" (i.e., for biparental inbreeding, moderate = distant neighbor (DNHBR), or DNHBR and distant within-forest (D-WF); for outbreeding depression: moderate = NNHBR and DNHBR, or DNHBR and D-WF, or all three classes). To present an unbiased picture of extreme-distance crossing effects, all possible values of each estimate for each maternal tree are included. The number of values contributing to each mean is shown in parentheses

 
Variation in expression of inbreeding and outbreeding depression
Apparent crossing effects on fitness declined between the stages of fruit set and 1-yr-old seedling size. Of the four fitness measures, only fruit set was influenced significantly by crossing treatment. Due to the attrition in sample sizes from the stage of mature seeds to 1-yr-old seedlings, however, and the consequent large variation among treatment-by-maternal tree classes in both seedling survivorship and height, the power to detect treatment effects also decreased over time. Accordingly, failure to observe significant crossing effects on progeny fitness (seedling survivorship and growth in particular) should not be considered conclusive.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
LITERATURE CITED
 
The exceptionally low fruit set of selfed flowers of both species may be due to strong inbreeding depression or to genetic self-incompatibility (Bawa, 1992 ). If low self-fertility is attributable to genetic incompatibility, the discrimination of self-pollen is not complete for either species (S. rubicundum, this study; Sh. cordifolia, Dayanandan et al., 1990 ). The predominant outcrossing observed for these trees is characteristic of woody species throughout the tropics (e.g., Ashton, 1969 ; Bawa, 1974, 1992 ; Chan, 1981 ; Bawa, Perry, and Beach, 1985 ; Dayanandan et al., 1990 ; Murawski and Hamrick, 1991 ; Oliveira and Gibbs, 1994 ).

In addition to a striking and predictable selfing effect for both species, the relationship between outcrossing distance and fruit set was significant for each. Peak fruit set corresponded to an outcrossing distance of ~1–2 km. Although not statistically significant, peak fruit set for S. rubicundum consistently exceeded those of open-pollinated flowers, suggesting that the bulk of the natural pollen distribution is restricted to distances <1–2 km. More specifically, similarity between fruit set of open-pollinated flowers and distant-neighbor crosses may indicate that these trees naturally cross with mates occurring very roughly within a radius of 80–500 m (the range of outcrossing distances used in the distant-neighbor treatment). Unfortunately, as the hand-pollination procedure had a negative impact on fruit set in Sh. cordifolia, comparisons of fruit set rates or cumulative fitnesses between hand-pollination and open-pollination treatments are not useful.

While outcrossing distance did not significantly affect the remaining measures of crossing fitness, with few exceptions, mean values of fitness measures followed a largely consistent trend across treatments. For both species, mean values for all measures gradually increased with crossing distance, peaking at an intermediate distance (1–2 km for S. rubicundum and 1–10 km for Sh. cordifolia), and then dropping at the between-forest treatment. Departures from this trend are seen for seed germination and seedling size in between-forest crosses of Sh. cordifolia, which exceeded the mean values for all other treatments, including the open-pollinated treatment (see Hybrid vigor in Shorea cordifolia below). When all fitness measures were combined in cumulative fitness scores, however, the trend of peak fitness at intermediate-distance, within-forest crosses persisted for both species.

Near-neighbor crossing effect
The effect of nearest-neighbor mating on cumulative fitness was significant for S. rubicundum and persisted for crosses ≤500 m (distant neighbor treatment). Mean cumulative fitnesses of outcrosses ≤500 m were 18–26% that of crosses 1–2 km in distance. For Sh. cordifolia, mean cumulative fitness of nearest-neighbor crosses was 50% that of within-forest crosses ≥2 km. For both species, the reduced fitness of near-neighbor crosses was due primarily to reduced fruit set. As for selfing, reduced fruit set in near-neighbor crosses may be due to inbreeding depression or to genetic incompatibility. While the influence of genetic incompatibility was not tested here and therefore cannot be disregarded outright, the consistently lower progeny fitnesses observed for nearest-neighbor crosses relative to longer distance crosses strongly suggest the influence of inbreeding depression. Furthermore, of the six maternal trees used, self-infertility was complete in Sh. cordifolia, but considerably less so in S. rubicundum. Given complete self-infertility due to incompatibility in the three adult Sh. cordifolia, compromised fruit set in nearest-neighbor crosses would be expected if cross-incompatibility between neighboring plants is of the same genetic basis as self-incompatibility. This expectation assumes conspecific neighbors in logged forest are related. In contrast to this expectation, fruit set from nearest-neighbor crosses was not significantly compromised for Sh. cordifolia, as it was for S. rubicundum. This finding suggests, for Sh. cordifolia at least, that the contribution, if any, of genetic incompatibility to reduced crossing success between neighboring trees is of secondary importance to inbreeding depression.

The logged forest at Sinharaja afforded an ideal setting for investigating the potential for biparental inbreeding depression in natural stands of forest trees. The high light conditions in the post-logging environment likely promoted regeneration of high-density stands of related individuals, especially of the light-demanding S. rubicundum (P. M. S. Ashton, personal communication, Yale University). While apparent inbreeding depression in near-neighbor crosses was significant for S. rubicundum, its significance for Sh. cordifolia was more ambiguous, as the nearest-neighbor mating effect was highly variable among the three maternal trees. Interspecific differences in observed biparental inbreeding depression may result from interspecific differences in the genetic relatedness of near-neighbors in the study populations. Alternatively, the discrepancy may be due to historical differences in the breeding structures of the two species. While the species share similar stature and pollinators (primarily bees), they possess very different modes of seed dispersal (S. rubicundum, dispersal by birds or bats; Sh. cordifolia, dispersal by wind or gyration). Because of differences in dispersal potential, a history of biparental inbreeding is expected for Sh. cordifolia, but not for S. rubicundum. As biparental inbreeding depression is expected to be most intense for populations with little natural inbreeding in their recent evolutionary history (and thus little opportunity for the purging of deleterious recessive alleles; Heywood, 1993 ), inbreeding depression from nearest-neighbor crosses in logged forest should have been greater for S. rubicundum than for Sh. cordifolia, as was observed. Planned microsatellite-based studies of fine-scale genetic structure in the study populations will permit evaluation of this conclusion through an interspecific comparison of patterns of relatedness among neighboring trees.

Results of this study indicate that the potential for significant biparental inbreeding effects is strong for tree populations, when neighboring trees are related. The degree of inbreeding depression experienced by a population, however, will vary among species, possibly as a function of their recent evolutionary histories. While the significant, and near significant, cumulative fitness effects of nearest-neighbor mating observed for the two study species in logged forest indicate the potential for such effects in natural stands, both the actual degree of relatedness among near-neighbors and the degree of biparental inbreeding depression experienced in undisturbed forest remain to be determined.

Near-neighbor crossing effects have been demonstrated for a number of coniferous species (Coles and Fowler, 1976 ; Park and Fowler, 1982, 1984 ; Latta et al., 1998 ), but only three studies have yielded evidence of near-neighbor crossing effects in woody angiosperms (Syzygium cormiflorum—Crome and Irvine, 1986 ; Schiedea spp.—Sakai, Karoly, and Weller, 1989 ; Eucalyptus globules—Hardner, Potts, and Gore, 1998 ). In another study suggestive of biparental inbreeding depression, mean fruit set rates were significantly lower for intraspecific crosses <0.5 km distance than for crosses >1 km distance for three subcanopy tree species (Inga spp.) in Costa Rica (Koptur, 1984 ). In fact, it may be that biparental inbreeding depression is common in natural populations of forest trees, but that estimation of its potential through experimental cross-pollinations has been limited to only a few species due to the obvious difficulty of working in the canopy. To my knowledge, there are no published reports of failed attempts to find near-neighbor crossing effects in natural populations of forest trees.

Between-forest crossing effect
For both species, cumulative fitness dropped significantly between long-distance crosses within Sinharaja to crosses involving separate forest populations. This between-forest crossing effect was consistent between species in spite of the large interspecific difference in dispersal potential. For both species, variation in fruit set among maternal trees was least for the between-forest crosses, suggesting a universal poor interfertility between trees occurring in separate forest reserves.

Mechanisms underlying outbreeding depression may be of a genetic or an ecological nature (Price and Waser, 1979 ; Shields, 1982 ). Outbreeding depression involving between-population crosses is most often ascribed to the genetic mechanism involving disruption of coadapted gene complexes (Templeton, 1986 ). According to this model, intrinsic coadaptation involving relatively few loci develops through restricted gene flow among populations and genetic drift within populations (Templeton, 1981 ; Schierup and Christiansen, 1996 ). Crossing disparate genomes results in outbreeding depression through the disruption of coadaptation between homologous chromosomes in the F1 generation and between coadapted portions of individual chromosomes in F2 progeny. The outbreeding depression observed in this study, which was restricted to between-forest crosses over 12- and 35-km distances, may be explained in part by disruption of intrinsic coadaptation. The observation of hybrid vigor in seedlings of Sh. cordifolia is also consistent with this model (Templeton, 1986 ; see below).

In contrast, the ecological mechanism for outbreeding depression involves reduced fitness of wide outcrosses due to adaptation to local biotic and abiotic conditions, such that wide outcrossing yields F1 progeny with alleles maladapted to either of the parental environments (Endler, 1977 ). Although selection-driven divergence is typically associated with intrapopulation outbreeding depression (e.g., Waser and Price, 1989 ), selection-driven divergence between populations seems a plausible contributor to the reduced interfertility between populations observed in this study. Through direct selection on fitness traits, habitat heterogeneity will promote genetic differentiation within and among plant populations (Jain and Bradshaw, 1966 ; Linhart and Grant, 1996 ). The considerable environmental heterogeneity of southwest Sri Lanka is likely sufficient to cause genetic differentiation of tree populations over a scale of tens of kilometers. The ridge and valley system of southwest Sri Lanka comprises elevations ranging from 300 m to >1000 m. Across elevations, variation in temperature, cloudiness, and rainfall (<2500–5000 cm) occurs (Gunatilleke et al., 1998 ).

The apparent outbreeding depression observed in fruit set and cumulative fitness for between-forest crosses in both species indicates some degree of genetic isolation among tree populations occupying the separate forest reserves of Sri Lanka's wet zone. This result is somewhat surprising given the large stature of the species and the small geographic area involved, and it suggests that conditions favorable for speciation in tropical trees may arise over a scale of only several to tens of kilometers. The geographical heterogeneity of southwest Sri Lanka, however, may be of a finer scale than that of the majority of tropical forested landscapes (Ashton and Gunatilleke, 1987 ). It would be desirable to see whether poor cross-fertility between forests is universal for tree species in the wet zone. Unfortunately, plans to continue this study in 1998, and to include other species of Syzygium and Shorea, were thwarted due to a general lack of flowering in the region that year. From a conservation perspective, observation of even minor reproductive isolation between forest reserves suggests that even where tree species are shared among reserves, each forest represents a singular genetic resource worthy of preservation.

Outbreeding depression was not detected in crosses over what is presumably the normal range of pollen flow for either species. The lack of evidence of outbreeding depression within continuous-forest populations in this study is consistent with the literature in which examples of between-population outbreeding depression in plants far outnumber those of within-population outbreeding depression. Given the recent nature of deforestation north of Sinharaja, however, delineation of S. rubicundum into separate populations in the Sinharaja and Walankanda Reserves may not accurately reflect the recent demographic history of this species. Walankanda and Sinharaja Reserves were part of one continuous forest until only 30–40 yr ago (P. S. Ashton, personal communication, Harvard University). This is probably less than the generation time for these trees and indicates the potential for recent genetic connectivity between the two populations. As S. rubicundum is generally restricted to mid-slope areas, however, it is likely that this species was not present in abundance in the valley between Sinharaja and Walankanda Reserves prior to the clearing of forest in that area (P. S. Ashton, personal communication, Harvard University). Regardless, the two forests are separated at present by a deforested strip only 4 km wide. Gene flow between tree populations occupying these forests since the separation is therefore at least plausible (e.g., White, Powell, and Boshier, 1998 ). For these reasons, observation of outbreeding depression in crosses between tree populations occupying Sinharaja and Walankanda Reserves is unexpected, and it indicates that genetic divergence of tree populations can occur over very short distances even in continuous habitat.

Hybrid vigor in Shorea cordifolia
For Sh. cordifolia, trends observed among treatments for seed germination rate and seedling height suggest hybrid vigor or luxuriance in progeny derived from between-forest crosses. This finding is tentative, however, due to the small sample sizes involved with the between-forest crosses for the seed germination and later stages. Hybrid vigor in F1 progeny is consistent with the model for outbreeding depression through the disruption of coadapted gene complexes. According to this model, F1 hybrid vigor results from increased heterozygosity, with subsequent hybrid breakdown in the F2 generation from the disruption of parental genomes during F1 gametogenesis (Templeton, 1986 ). It remains to be seen whether hybrid vigor can occur at some level of genetic differentiation between mates without subsequent F2 breakdown (Shields, 1982 ). Hybrid vigor in interpopulation crosses followed by a drop in F2 fitness has been reported for several herbaceous species (e.g., Clausen, 1951 ; Kruckeberg, 1957 ; Vickery, 1959 ; Grant and Grant, 1960 ; Gottlieb, 1971 ; Grant, 1971 ; Hughes and Vickery, 1974 ; Price and Waser, 1979 ). In woody angiosperms, hybrid vigor in interpopulation crosses has been documented for at least one species, Syzygium cormiflorum, a subcanopy species of Australia's rainforests (Crome and Irvine, 1986 ). Unfortunately, due to the long generation times of trees, study of F2 generations in these species is usually not feasible.

Timing of inbreeding and outbreeding depression
Evidence of inbreeding and outbreeding effects decreased between the stages of fruit set and 1-yr-old seedlings. This finding is tentative, however, as the power to detect crossing effects also declined over the same period. In their review of the timing of inbreeding depression in plants, Husband and Schemske (1996) concluded that for outcrossing species, inbreeding depression is greatest for the stages of seed set and growth and reproduction, and much less important for seed germination. Results of this study are consistent with the general conclusion that embryo abortion is an important, if not primary, component of inbreeding depression in outcrossing plants (e.g., Levin, 1984, 1989 ).

Little is known of the timing of outbreeding depression in plants. Within the confines of the fitness measures used in this study, the results are consistent with the findings of McCall, Mitchell-Old, and Waller (1991) of optimal outcrossing in Impatiens capensis in which the effect of intermate distance decreased between the stages of mature seeds and offspring sexual maturity. Both studies agree with Husband and Schemske's (1996) consensus for inbreeding depression that crossing effects are most notable at the stage of seed set and less so at the stages of seed germination and early seedling growth (but see Waser and Price, 1989 ). Low inbreeding depression observed in germination and survival might be due to the short duration of, and few genes involved in, these stages relative to the stages of seed maturation and reproduction (Husband and Schemske, 1996 ).

Viewed together, the theories of inbreeding and outbreeding depression predict that there should exist some intermediate crossing distance between mates at which both inbreeding and outbreeding depression are avoided (i.e., the "optimal outcrossing distance" of Price and Waser, 1979 ; and Waser and Price, 1983 ). Optimal outcrossing refers to those crosses that achieve greatest reproductive fitness relative to other crosses and therefore signal optimal genetic compatibility between mates. Little is known of optimal outcrossing in woody species. Although identifying an optimal outcrossing distance was not an objective of this study, the results indicate that optimal outcrossing for canopy species in tropical forests may occur over a range of roughly one to several kilometers. While tracking pollen flow at this spatial scale in continuous forest is exceedingly difficult, a handful of recent studies have demonstrated at least low levels of natural cross-pollinations over distances of one, or in some cases, several kilometers in tropical trees (Nason, Herre, and Hamrick, 1996, 1998 ; Nason and Hamrick, 1997 ; Apsit, 1998 ; White, Powell, and Boshier, 1998 ). Outcrossing distance alone, however, does not explain variation in crossing success for the two study species. Similar outcrossing distances yielded very different results in the two species (high fitness at 10 km for Sh. cordifolia and very low fitness at 12 km for S. rubicundum). Rather, for these trees, the critical determinant of crossing success over long distances appears to be whether or not mates occur within the same forest reserve.

Summary of conclusions
For two tree species in Sri Lanka's wet zone forests, fruit set increased significantly with outcrossing distance, peaking at intermediate-distance within-forest crosses (1–10 km depending on species). In crosses between trees occupying separate forest reserves, however, fruit set was significantly reduced (or nearly so) for both species. In contrast, seed germination and seedling height at 1 yr for Sh. cordifolia suggested hybrid vigor in between-forest crosses. The effects of nearest-neighbor mating varied among trees and species; the mean fitness cost of nearest-neighbor mating relative to mating with moderately more distant neighbors was 45% for S. rubicundum and 0% for Sh. cordifolia. In contrast, the fitness effects of between-forest crossing were substantial for both species (52 and 70% relative to within-forest crosses for the same two species). Crossing effects diminished between the stages of fruit set and 1-yr-old seedling size; only the former was significant for both species. Results indicate a strong potential for biparental inbreeding depression within forest tree populations and partial reproductive isolation among trees occupying the remaining forest reserves in Sri Lanka's wet zone.


   FOOTNOTES
 
1 The author thanks S. Gunatilleke and N. Gunatilleke at the Department of Botany, University of Peradeniya for guidance and logistic support; the Forest Department of Sri Lanka for permission to conduct this work within the Sinharaja World Heritage Site; the Botany Department, University of Peradeniya for use of its facilities; S. Harischandran, H. Gamage, and M. Gunadasa for assistance with hand-pollinations; and S. Harischandran and A. Anura for tending greenhouse seedlings. P. Ashton, K. Bawa, N. Gunatilleke, J. Hamrick, L. Kaufman, and J. Mitton provided helpful discussions; P. Ashton, S. Dayanandan, and C. Schneider commented on earlier drafts; and B. Wieninger assisted with the figures. This work was supported in part by funds from the Conservation, Food, & Health Foundation, Inc. Back


   LITERATURE CITED
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Apsit V. J. 1998 Fragmentation and pollen movement in a Costa Rican dry forest tree species. Ph.D. dissertation, University of Georgia, Athens, Georgia, USA

Ashton P. S. 1969 Speciation among tropical forest trees: some deductions in the light of recent evidence. Biological Journal of the Linnean Society 1: 155-196

———, and I. A. U. N. Gunatilleke 1987 New light on the plant geography of Ceylon I. Historical plant geography. Journal of Biogeography 14: 249-285

Bateson P. P. G. 1978 Sexual imprinting and optimal outbreeding. Nature 273: 259-260[Medline]

Bawa K. S. 1974 Breeding systems of tree species of a lowland tropical community. Evolution 28: 85-92

———. 1992 Mating systems, genetic differentiation and speciation in tropical rain forest plants. Biotropica 24: 250-255[CrossRef][ISI]

———, D. R. Perry J. H. Beach 1985 Reproductive biology of tropical lowland rain forest trees. I. Sexual systems and incompatibility mechanisms. American Journal of Botany 72: 331-345[CrossRef][ISI]

Berg E. E. J. L. Hamrick 1995 Fine-scale genetic structure of a turkey oak forest. Evolution 49: 110-120[CrossRef][ISI]

Boshier D. H. M. R. Chase K. S. Bawa 1995 Population genetics of (Cordia alliodora) (Boraginaceae), a Neotropical tree. III. Gene flow, neighborhood, and population structure. American Journal of Botany 82: 484-490[CrossRef][ISI]

Byers D. L. 1998 Effect of cross proximity on progeny fitness in a rare and a common species of Eupatorium (Asteraceae). American Journal of Botany 85: 644-653[Abstract]

Chan H. T. 1981 Reproductive biology of some Malaysian Dipterocarps. III. Breeding systems. Malaysian Forester 44: 28-36

Clausen J. 1951 Stages in the evolution of plant species. Hafner, New York, New York, USA

Coles J. F. D. P. Fowler 1976 Inbreeding in neighboring trees in two white spruce populations. Silvae Genetica 25: 29-34

Crome F. H. J. A. K. Irvine 1986 "Two bob each way": the pollination and breeding system of the Australian rain forest tree Syzygium cormiflorum (Myrtaceae). Biotropica 18: 115-125

Dayanandan S. D. N. C. Attygala A. W. W. L. Abeygunasekera I. A. U. N. Gunatilleke C. V. S. Gunatilleke 1990 Phenology and floral morphology in relation to pollination of some Sri Lankan Dipterocarps. In K. S. Bawa and M. Hadley [eds.], Reproductive ecology of tropical forest plants, UNESCO, Man and Biosphere series, vol. 7, 103–133. Paris and Parthenon Publishing Group, Carnforth, UK

Dewey S. E. J. S. Heywood 1988 Spatial genetic structure in a population of Psychotria nervosa. I. Distribution of genotypes. Evolution 42: 834-838[CrossRef][ISI]

Endler J. A. 1977 Geographic variation, speciation, and clines. Princeton University, Princeton, New Jersey, USA

———. 1979 Gene flow and life history patterns. Genetics 93: 263-284[Abstract/Free Full Text]

Epperson B. K. M. T. Clegg 1986 Spatial-autocorrelation analysis of flower color polymorphisms within substructured populations of morning glory (Ipomoea purpurea). American Naturalist 128: 840-858[CrossRef][ISI]

Fedorov A. A. 1966 The structure of the tropical rain forest and speciation in the humid tropics. Journal of Ecology 54: 1-11[CrossRef]

Fischer M. D. Matthies 1997 Mating structure and inbreeding and outbreeding depression in the rare plant Gentianella germanica (Gentianaceae). American Journal of Botany 84: 1685-1692[Abstract]

Gottlieb L. D. 1971 Evolutionary relationships in outcrossing diploid annual species of Stephanomeria (Compositae). Evolution 25: 312-329[CrossRef][ISI]

Grant V. 1971 Plant speciation. Columbia University, New York, New York, USA

———, andA. Grant 1960 Genetic and taxonomic studies in Gilia XI. Fertility relationships of the diploid cobwebby Gilias.. Aliso 4: 435-481

Gunatilleke C. V. S. I. A. U. N. Gunatilleke 1980 The floristic composition of Sinharaja—a rain forest in Sri Lanka with special reference to endemics. Sri Lankan Forester 14: 171-180

Gunatilleke C. V. S. I. A. U. N. Gunatilleke P. M. S. Ashton P. S. Ashton 1998 Seedling growth of Shorea (Dipterocarpaceae) across an elevational range in southwest Sri Lanka. Journal of Tropical Ecology 14: 231-245[CrossRef][ISI]

Hamrick J. L. D. A. Murawski J. D. Nason 1993 The influence of seed dispersal mechanisms on the genetic structure of tropical tree populations. Vegetatio 107/108: 281-297

Hardner C. M. B. M. Potts P. L. Gore 1998 The relationship between cross success and spatial proximity of Eucalyptus globulus ssp. globulus parents. Evolution 52: 614-618[CrossRef][ISI]

Heywood J. S. 1993 Biparental inbreeding depression in the self-incompatible annual Gaillardia pulchella (Asteraceae). American Journal of Botany 80: 545-550[CrossRef][ISI]

Hughes K. W. R. K. Vickery Jr 1974 Patterns of heterosis and crossing barriers resulting from increasing genetic distance between populations of the Mimulus luteus complex. Journal of Genetics 61: 235-245[ISI]

Husband B. C. D. W. Schemske 1996 Evolution of the magnitude and timing of inbreeding depression in plants. Evolution 50: 54-70

Jain S. K. A. D. Bradshaw 1966 Evolutionary divergence among adjacent plant populations. 1. The evidence and its theoretical analysis. Heredity 20: 407-441

Koptur S. 1984 Outcrossing and pollinator limitation of fruit set: breeding systems of neotropical Inga trees (Fabaceae: Mimosoideae). Evolution 38: 1130-1143[CrossRef][ISI]

Kruckeberg A. R. 1957 Variation in fertility of hybrids between isolated populations of the serpentine species, Streptanthus glandulosus Hook. Evolution 11: 185-211[CrossRef][ISI]

Latta R. G. Y. B. Linhart D. Fleck M. Elliot 1998 Direct and indirect estimates of seed versus pollen movement within a population of ponderosa pine. Evolution 52: 61-67[CrossRef][ISI]

Levin D. A. 1984 Inbreeding depression and proximity-dependent crossing success in Phlox drummondii.. Evolution 38: 116-127[CrossRef][ISI]

———. 1989 Proximity-dependent cross-compatibility in Phlox.. Evolution 43: 1114-1116[CrossRef][ISI]

———, and H. W. Kerster 1974 Gene flow in seed plants. Evolutionary Biology 7: 139-220

Linhart Y. B. 1973 Ecological and behavioral determinants of pollen dispersal in hummingbird-pollinated Heliconia.. American Naturalist 107: 511-523[CrossRef][ISI]

———, and M. C. Grant 1996 Evolutionary significance of local genetic differentiation in plants. Annual Review of Ecology and Systematics 27: 237-278[CrossRef][ISI]

Loveless M. D. J. L. Hamrick 1984 Ecological determinants of genetic structure in plant populations. Annual Review of Ecology and Systematics 15: 65-95

McCall C. T. Mitchell-Olds D. M. Waller 1988 Might a plant that often selfs have an optimal out-crossing distance?. Bulletin of the Ecological Society of America 69: 223.

———, ———, and ———. 1991 Distance between mates affects seedling characters in a population of Impatiens capensis (Balsaminaceae). American Journal of Botany 78: 964-970[CrossRef][ISI]

Murawski D. A. J. L. Hamrick 1991 The effect of the density of flowering individuals on the mating systems of nine tropical tree species. Heredity 67: 167-174[ISI]

Nason J. D. J. L. Hamrick 1997 Reproductive and genetic consequences of forest fragmentation: two case studies of Neotropical canopy trees. Journal of Heredity 88: 264-276[Abstract/Free Full Text]

———, E. A. Herre J. L. Hamrick 1996 Paternity analysis of the breeding structure of strangler fig populations: evidence for substantial long-distance wasp dispersal. Journal of Biogeography 23: 501-512[CrossRef][ISI]

———, ———, and ———. 1998 The breeding structure of a tropical keystone plant resource. Nature 391: 685-687[CrossRef]

Oliveira P. P. Gibbs 1994 Pollination biology and breeding systems of six Vochysia species (Vochysiaceae) in Central Brazil. Journal of Tropical Ecology 10: 509-522[ISI]

Park Y. S. D. P. Fowler 1982 Effects of inbreeding and genetic variances in a natural population of Tamarack (Larix laricina (Du Roi) K. Koch) in eastern Canada. Silvae Genetica 31: 21-26

———, and ———. 1984 Inbreeding in black spruce (Picea mariana (Mill.) B. S. P.): self fertility, genetic load, and performance. Canadian Journal of Forest Research 14: 17-21

Perry D. J. P. Knowles 1991 Spatial genetic structure within three sugar maple (Acer saccharum March) stands. Heredity 66: 137-142

Price M. V. N. M. Waser 1979 Pollen dispersal and optimal outcrossing in Delphinium nelsonii.. Nature 177: 294-296

Ritland K. F. R. Ganders 1987 Crossability of Mimulus guttatus in relation to components of gene fixation. Evolution 41: 772-786[CrossRef][ISI]

Sakai A. K. K. Karoly S. G. Weller 1989 Inbreeding depression in Schiedea globosa and S. salicaria (Caryophyllaceae): subdioecious and gynodioecious Hawaiian species. American Journal of Botany 76: 437-444[CrossRef][ISI]

Schemske D. W. L. P. Pautler 1984 The effects of pollen composition on fitness components in a neotropical herb. Oecologia 62: 31-36[CrossRef][ISI]