ABSENCE OF CONSPECIFIC POLLEN ADVANTAGE IN THE

Transcription

1 American Journal of Botany 87(6): ABSENCE OF CONSPECIFIC POLLEN ADVANTAGE IN THE DYNAMICS OF AN IPOMOPSIS (POLEMONIACEAE) HYBRID ZONE 1 RUBEN ALARCÓN AND DIANE R. CAMPBELL 3 Department of Ecology and Evolutionary Biology, University of California, Irvine, California 9697 USA; and Rocky Mountain Biological Laboratory, Crested Butte, Colorado 814 USA The frequency of hybrid formation in angiosperms depends on how often heterospecific pollen is transferred to the stigma and on the success of that heterospecific pollen at fertilizing ovules. Even if heterospecific pollen is capable of effecting fertilization it may perform poorly when conspecific pollen is also available on the stigma. We applied pollen mixtures to stigmas to determine how pollen interactions affect siring success and the frequency of hybrid formation between two species of Ipomopsis (Polemoniaceae) in Colorado. Plants of both parental species and natural hybrids were pollinated with I. aggregata and I. tenuituba pollen in ratios of 100:0, 80:0, 50:50, 0:80, and 0:100 by mass. Plants were homozygous for different alleles at an isozyme marker, allowing us to distinguish the type of pollen parent for 166 viable seeds from 73 fruits. In contrast to studies of many other hybridizing taxa, there was no evidence of an advantage to conspecific pollen, nor did composition of the stigmatic pollen load affect seed set. Instead, the frequency of seeds sired by a given species was proportional to its representation in the pollen load. In this hybrid zone, both the frequency of first-generation hybrid formation and the relative male fitness of the two parental species should be predictable from the rates of pollen transfer to stigmas. Key words: gene flow; hybridization; hybrid zone; interspecific pollen competition; Ipomopsis; male fitness; Polemoniaceae; pollen interactions; reproductive isolation. Natural hybridization appears common in many groups of flowering plants (Knobloch, 197; Ellstrand, Whitkus, and Rieseberg, 1996). Where two related species come into contact, they may form a zone of hybridization. The evolutionary dynamics of such a zone depend on gene flow between the species and the fitness of hybrids relative to the parental species (Harrison, 1993; Arnold, 1997). Both interspecific gene flow and relative fitness in turn depend on events that take place during pollination and post-pollination. Interspecific gene flow requires first that heterospecific pollen be transferred to stigmas. However, hybrids will only be formed if that heterospecific pollen goes on to fertilize ovules. Even if heterospecific pollen is capable of fertilization it might not do so if conspecific pollen is also available on the stigma. Similarly, relative male fitness of a genotype in a hybrid zone could depend not only on how much pollen it transfers to stigmas but also on interactions between its pollen and that of other genotypes. Several recent studies have employed genetic markers to study pollen interactions in hybrid zones (Carney, Cruzan, and Arnold, 1994; Wilson and Payne, 1994; Rieseberg, Desrochers, and Youn, 1995; Carney, Hodges, and Arnold, 1996; Hauser, Jorgensen, and Ostergård, 1997). In these cases heterospecific pollen had a lower chance 1 Manuscript received 8 April 1999; revision accepted 10 September The authors thank D. Massart and D. Mitchell for assistance in the lab and the field and S. Carney, J. Reithel, and N. Waser for comments on the manuscript. Funding was provided by National Science Foundation grant DBI to RMBL and grants DEB and DEB to DRC. Current address: Department of Biology, University of California, Riverside, California 951 USA. 3 Author for correspondence ( drcampbe@uci.edu). 819 of siring a seed than did conspecific pollen when applied in a mixture on the stigma even though pure pollination with heterospecific pollen could easily produce hybrids. We refer to this phenomenon as conspecific pollen advantage. In the Louisiana Iris complex, conspecific pollen advantage is partly attributable to differences in the rate of pollen tube growth (Carney, Hodges, and Arnold, 1996; Emms, Hodges, and Arnold, 1996). Other possible mechanisms include: production of compounds that either facilitate or inhibit pollen germination or the growth of pollen tubes (Cruzan, 1990), physical blocking of pollen by grains that arrive earlier on the stigmatic surface (Thomson, 1989), usurption of ovules by prior penetration of pollen tubes (Waser and Price, 1991), and selective abortion of zygotes (Hauser, Jorgensen, and Ostergård, 1997). These studies indicate that a conspecific pollen advantage may act as a barrier to gene flow between species. However, to evaluate the importance of this phenomenon in natural hybrid zones, we need to know not only the impact of pollen mixtures on rates of fertilization but also the frequency with which flowers receive mixed pollen loads of different compositions. We are taking this dual approach in studying hybrid zones between Ipomopsis aggregata and Ipomopsis tenuituba. These two species come into sympatry in several locations throughout the western United States (Grant and Wilken, 1988). At our study site at Poverty Gulch in Gunnison County, Colorado, the only common pollinators in most years are broad-tailed and rufous hummingbirds (Selasphorus platycercus and S. rufus). Hummingbirds fly frequently between the species in artificial mixed populations (Campbell, Waser, and Meléndez-Ackerman, 1997) and are capable of transferring large amounts of pollen between them. These pollinators prove equally proficient at trans-

2 80 AMERICAN JOURNAL OF BOTANY [Vol. 87 ferring I. aggregata pollen to either species when flowers are presented in mixed sequences, and I. tenuituba receives a particularly high percentage of interspecific pollen (Campbell, Waser, and Wolf, 1998). Furthermore, pollen loads deposited on the stigma usually contain pollen from many individuals (Campbell, 1998), which should heighten the level of interspecific pollen flow (Leebens- Mack and Milligan, 1998). Together these observations suggest that many flowers in hybridizing populations will receive loads of pollen containing a mixture from the two species. In the current study, we examined the impact of such pollen mixtures on the post-pollination success of I. aggregata and I. tenuituba pollen. Pollen was applied in realistic mixtures to the stigma, and isozyme markers were used to determine the identity of the sire for successful progeny. Because previous studies predict that I. aggregata should receive more conspecific pollen while I. tenuituba receives more heterospecific pollen (Campbell, Waser, and Wolf, 1998), we varied the ratio of pollen applied over a wide range. Our study is unusual not only in the context provided by detailed information on pollination, but also in our inclusion of natural hybrids as well as the parental species as pollen recipients. We asked three specific questions. First, how does the relative proportion of seeds sired by a particular species depend on the type of recipient (I. aggregata, I. tenuituba, or hybrid) and the pollen ratio applied to stigmas? If there is conspecific pollen advantage, I. aggregata pollen should sire relatively few seeds when applied in mixture to I. tenuituba even if it performs as well as the conspecific I. tenuituba pollen when applied alone. Second, we compared the two species in overall success of pollen at siring seeds. Third, we asked whether seed set depends on the ratio of pollen applied. A species difference in performance of pollen, or a dependence of seed set on composition of the pollen load, might contribute to variation in relative male fitness of the two species and thereby to overall fitness differences that influence the stability of the hybrid zone. MATERIALS AND METHODS Study system At Poverty Gulch, populations of I. aggregata and I. tenuituba are found at lower (900 m) and higher (300 m) elevations, respectively, while populations composed of hybrid individuals, displaying intermediate characteristics, are in between (Grant and Wilken, 1988; Campbell, Waser, and Meléndez-Ackerman, 1997; Meléndez- Ackerman, 1997). The exact genetic identity of plants in hybrid populations (i.e., F1, F, backcross, etc.) is unknown, and we refer to these hereafter simply as hybrids. Ipomopsis aggregata is characterized by red trumpet-shaped corollas, while I. tenuituba displays longer and narrower white, pink, or pale violet corollas. Both species have protandrous and self-incompatible flowers. See Campbell, Waser, and Meléndez- Ackerman (1997) for further description of the study system. Pollen replacement experiment We screened plants at an isozyme locus in order to choose pollen donors and recipients that were homozygous for particular alleles. Flower buds were collected from plants in I. aggregata (L), hybrid (H), and I. tenuituba (A and B) populations during June to July 1997 (letters give the locations of populations described in Campbell, Waser, and Meléndez-Ackerman, 1997), with the exception of one hybrid individual (I10) potted in The buds were placed in a cooler and returned on ice to the Rocky Mountain Biological Laboratory (Gothic, Colorado). Upon arrival the buds were assayed at the slower migrating locus of 6-phospho-glucose dehydrogenase (6- PGD), following the procedures in Campbell and Dooley (199). We identified individuals homozygous for the fast (F), medium (M), and slow (S) alleles to be used in the experimental pollinations. These plants were potted and placed in a greenhouse at the RMBL. Only individuals homozygous for the S or the M allele were used as pollen donors. Pollen recipients were also homozygous for the S or M allele, with the exception of two hybrids homozygous for the F allele. Since pollinator visitation rates are very low in the natural populations (Campbell, Waser, and Meléndez-Ackerman, 1997), when flowers do receive mixtures of pollen from two taxa they probably do so in a single load. We thus subjected flowers to different mixes of pollen deposited simultaneously rather than with any time delay. Experimental pollinations employed five ratios of pollen, 100:0, 80:0, 50:50, 0:80, and 0:100 I. aggregata to I. tenuituba, based on anther mass. Ratios of anther mass are expected to correspond well with ratios of pollen grain number for the following reasons. Anther mass correlates strongly with pollen grain number in I. aggregata (r 0.75 in Campbell, 199), and the two species have pollen grains of similar size. We compared the sizes of pollen grains by using an ocular micrometer to measure the mean diameter for a sample of 50 grains on each of ten plants per species collected from the same populations used in the isozyme study. The diameters for I. aggregata (mean SE m) and I. tenuituba ( m) were indistinguishable (F 1,18 0.4, P 0.53). We used eight each of I. aggregata, I. tenuituba, and hybrid plants as pollen recipients (total 4 recipients), and all 16 of the I. aggregata and I. tenuituba plants, along with four additional plants (two per species), as pollen donors. To avoid self-pollinating and to minimize the chance of crossing incompatible individuals (Carney, Hodges, and Arnold, 1996) we placed the I. aggregata and I. tenuituba plants into four groups. A group included two I. aggregata plants homozygous for the same allele (S or M) and two I. tenuituba plants homozygous for the other allele (groups 1 and 3: I. aggregata MM; groups and 4: I. aggregata SS) that served as both donors and recipients. Each of the four additional donors (one plant homozygous for each allele per species) was added to two groups, making three MM donors of one species and three SS of the other in each group. Each pollen recipient then received pollinations from two groups that did not include itself and that differed in which species was assigned the SS genotype. To prepare a pollen mixture from a particular group, we collected and weighed freshly dehisced anthers from the three plants per species in a small microcentrifuge tube. Using round toothpicks to transfer pollen, we saturated the stigmas with pollen (following Campbell and Halama, 1993) to provide ample opportunity for pollen interactions if they exist. Two female-phase flowers per ratio per genotype assignment were hand-pollinated on each of the 4 recipient plants. Reversing the genotype assignment between the two donor groups ensured that any variation in fitness associated with the 6-Pgd locus could be accounted for in the analysis. In total we attempted to pollinate 480 flowers (4 recipient plants 5 ratios genotype combinations flowers) but were only successful in pollinating 416 due to the availability of male and female phase flowers. An additional 3 pollinations (leaving us with 384) were discarded from genetic analysis because we discovered an initial error in genotyping one pollen donor. After the fruits matured, the number of seeds was recorded. We then proceeded to screen all the seeds, using horizontal starch gel electrophoresis of seeds, to determine the genotype of the progeny and the siring success rate of the donor species. Statistical analysis We first conducted a preliminary analysis to see whether siring success depends on the genotype at the 6-Pgd locus. To do so we compared the performance of I. aggregata pollen assigned to the SS vs. MM genotype for the 50:50 pollen mixture. For each genotype assignment we calculated the proportion of seeds sired by I.

3 June 000] ALARCÓN AND CAMPBELL POLLEN INTERACTIONS IN IPOMOPSIS 81 TABLE 1. One-sample t test comparing the proportion of seeds sired by Ipomopsis aggregata with the expected proportion based on the pollen ratio. Expected proportion Mean observed proportion df t P aggregata for the two replicate flowers combined. A paired t test (on arcsine-transformed data) was then used to compare the two treatments matched by recipient plant. To compare the two species in siring ability, we performed one-sample t tests, comparing the proportion of seeds sired by I. aggregata to the expected outcome based on the proportion of I. aggregata pollen in the mixture. One test was performed separately for each pollen mixture, i.e., with expected ratios equal to 0., 0.5, and 0.8. In each case we used a plant (including up to four flowers that received pollen in a particular ratio) as a single replicate. The two flowers that received pollen from a given set of donors would by themselves provide an independent replicate. However, the number of seeds set by only two flowers was often too low to provide a meaningful estimate of a proportion. Combining the results from both genotype assignments was justified by the lack of an effect of the 6-Pgd genotype per se in our preliminary analysis. To find out whether siring success of a particular species depended on either pollen ratio or recipient species we again used as the dependent variable the proportion of seeds sired by I. aggregata (arcsine transformed) for the flowers belonging to a particular plant and pollen ratio. The ANOVA model contained the following factors: recipient species, plant nested within species, pollen ratio, and the interaction of ratio recipient species. Recipient species was either I. aggregata, I. tenuituba, or hybrid. To allow more general inferences about siring success of the two species we supplemented this analysis with the conservative approach of treating each of the four sets of pollen donors as supplying only a single replicate. In this case we used a one-sample t test for each recipient species to compare the proportion of seeds sired by I. aggregata with the expected ratio based on representation in the pollen pool. For analysis of seeds per flower, each replicate was the mean number of seeds across the two replicate flowers for a given plant, pollen ratio, and genotype assignment. We used the same ANOVA model as described above to test for the effect of pollen ratio or an interaction between ratio recipient species. Some pollinations (111 of 384) led to an aborted fruit with no seeds. To distinguish the contribution of variation in fruit set, as opposed to variation in seeds per successful fruit, we also analyzed the proportion of flowers for a given plant and pollen ratio that set a fruit. In this case each replicate included the flowers from both genotype assignments. For analyses of seeds per flower and fruits per flower we excluded data from hybrid I10, which was collected in 1996, because it alone had received nutrient supplements prior to this experiment. RESULTS Of the 384 pollinations for which we had sufficient information to identify the sire, 73 produced mature fruits, containing a total of 166 viable seeds. All 118 seeds from the mixed pollinations were screened at the 6-Pgd locus, while the seeds from single-species pollinations were assumed to be sired by just one species. The siring success of I. aggregata relative to I. tenuituba pollen did not depend on its 6-Pgd genotype (SS or TABLE. ANOVA for proportion of seeds sired by Ipomopsis aggregata following pollination with mixed pollen loads. Values were arcsine transformed prior to analysis. Plant(species) was considered a random effect; all other variables were fixed. The effect of recipient species was tested over plant(species). Source of variation Species Plant(species) Ratio Species ratio Residual error df Type III SS MS F P a a Because of unequal sample sizes this F test is approximate. Restricted maximum likelihood estimation implemented in PROC MIXED (SAS, 1996) also found no significant effect of species. MM). When pollen was applied in a 50:50 mix, approximately the same proportion of seeds was sired by I. aggregata regardless of whether it had the SS genotype (mean proportion 0.56) or the MM genotype (mean 0.53). A paired t test detected no significant difference between these estimates (t 0.44, df 8, P 0.694), justifying our decision to combine data from both genotype assignments for the subsequent analyses. Flowers pollinated with mixtures containing 0.0, 0.50, and 0.80 I. aggregata pollen produced seeds sired by I. aggregata in frequencies of 0.75, 0.506, and 0.771, respectively. These observed results did not deviate significantly from the expected values (Table 1), suggesting that the two species of pollen did not differ overall in ability to sire seeds. Surprisingly, the recipient species (I. aggregata, hybrid, or I. tenuituba) had no detectable effect on the proportion of seeds sired by I. aggregata (Table ). In fact, regardless of the recipient species, I. aggregata sired seeds in proportion to its frequency in the pollen loads (Fig. 1). This proportion thus varied significantly with the ratio of pollen from the two species (P ), but the effect of the ratio applied did not change with the recipient species (interaction P 0.357). Results were consistent across the four sets of pollen donors. For example, when pollen was applied in a 50:50 mix, the proportion of seeds sired by I. aggregata on I. aggregata recipients ranged from 0.40 to Treating each set of donors as a single replicate also yielded no significant differences between the observed proportion of seeds sired by I. aggregata and the expected value of 0.5 on any of the recipient species (one-sample t tests for 50:50 mixes, df 3, all P 0.0). We found no evidence whatsoever for a conspecific pollen advantage. The ratio of pollen from the two species also had no detectable effect on seeds set per flower (Table 3; Fig. ). Nor, was there an interaction between recipient species and pollen ratio, as would be expected if plants produced more seeds when receiving pollen mixtures composed mostly or entirely of conspecific pollen. Seeds per flower from pure pollinations were statistically indistinguishable regardless of whether conspecific or heterospecific pollen was applied (e.g., 4.5 vs. 5.0 for I. tenuituba recipients in Fig. ). Similar results were obtained for fruits per flower, which did not vary significantly with pollen ratio, recipient species, or the interaction (all P 0.0). Plants of I. tenuituba tended to produce the fewest

4 8 AMERICAN JOURNAL OF BOTANY [Vol. 87 TABLE 3. ANOVA of seeds per flower. All five pollen ratios, representing the three mixed pollinations and the two pure pollinations, were included in the analysis. The effect of recipient species was tested over plant(species). Source of variation Species Plant(species) Ratio Species ratio Residual error df Type III SS MS F P a a This F test is approximate. PROC MIXED (SAS, 1996) also found no significant effect of species. Fig. 1. The proportion of seeds sired by Ipomopsis aggregata as a function of the proportion of I. aggregata pollen in the mixture applied to stigmas. The solid line depicts the expected relationship under the null hypothesis that all pollen performs equivalently. Means and standard errors across plants are shown for each type of recipient plant and each pollen ratio. Data for the three types of recipients are slightly offset to enable visualization of the error bars. seeds per flower and hybrids the most, but the trends were not significant (P for effect of species). The only factor that significantly influenced seeds per flower was the identity of the maternal plant (P ). DISCUSSION The idea that conspecific pollen has an advantage over heterospecific pollen has a long history, tracing back at least to Kölreuter (1763). Darwin (1859) suggested that pollen from a plant s own species is prepotent over foreign pollen from another species. This possibility has been confirmed using genetic markers in several genera, including Iris (Carney, Cruzan, and Arnold, 1994), Helianthus (Rieseberg, Desrochers, and Youn, 1995), and Brassica (Hauser, Jorgensen, and Ostergård, 1997), which, like Ipomopsis, hybridize in the wild. In all these systems, conspecific pollen sired more seeds than did heterospecific pollen when applied in equal amounts, although the effect was often more pronounced for one of the recipient species (Emms, Hodges, and Arnold, 1996). It has been further suggested that a conspecific pollen advantage plays an important role in reducing the frequency of formation of F1 hybrids in nature, and there is much evidence for this scenario in the Louisiana irises. Following introduction of Iris hexagona into natural populations of its congener I. fulva, 1% of seeds were hybrids (Arnold, Hamrick, and Bennett, 1993). Moreover, established F1 hybrids are rare in natural populations despite high survival of hybrids once formed (Burke, Carney, and Arnold, 1998). Our results with Ipomopsis are strikingly different. We created conditions conducive to pollen competition by hand-pollinating with excess pollen, a method that produces seed sets higher than typically found in local natural populations (Campbell and Halama, 1993). For example, I. aggregata averaged 6.1 seeds per flower (Fig. ) compared to 3.1 for open-pollinated field plants (Campbell and Halama, 1993). Despite that situation, we found no evidence that conspecific pollen was superior when applied in a mixture. Instead, the frequency of seeds sired by either I. aggregata or I. tenuituba equaled its frequency in the pollen load. For example, with a 50: 50 mix of pollen applied to plants of the two pure species, 51.5% of seeds on average were hybrid. Pollen simply performed in proportion to its representation, regardless of the kind of recipient plant, I. aggregata, I. tenuituba, or natural hybrid. The two species of pollen did not differ significantly in overall success at siring seeds, and there was no asymmetry in the success of backcrosses with hybrid recipients. Furthermore, composition of the pollen load had no detectable influence on the number of seeds produced per flower. The difference between our results for Ipomopsis and the conspecific advantage detected elsewhere cannot be explained by low statistical power. Standard errors around our estimates were not large (Fig. 1). For the 50: 50 mix of pollen applied to plants of the two pure species, Fig.. Mean seeds per flower for each type of recipient plant and each pollen ratio. Values plotted are means and standard errors across the two sets of pollen donors and up to eight plants of a given type of recipient.

5 June 000] ALARCÓN AND CAMPBELL POLLEN INTERACTIONS IN IPOMOPSIS 83 the 95% CI for proportion of seeds sired by foreign pollen was Based on the level of variation we saw in this experiment, we would have had sufficient statistical power to detect a conspecific advantage if the true rate of success for foreign pollen had been 30% or lower (power 86% based on one-tailed test). And in other genera, lower rates have been found following pollination with an equal pollen mix, i.e., 3 8% in Cucurbita (Wilson and Payne, 1994), 7 9% in Hibiscus (Klips, 1999), and 4% in Helianthus (Rieseberg, Desrochers, and Youn, 1995). The absence of a conspecific pollen advantage in this Ipomopsis hybrid zone means that the frequency of hybrid formation between the parental species should be predictable from the mixture of pollen that lands on the stigma. Any barrier to interspecific gene flow would have to take place prior to or during pollination. For this Ipomopsis system we have sufficient data on pollination to make some realistic predictions about hybrid formation. Because the flowering phenologies are so similar (Meléndez-Ackerman, 1997), the frequency of natural hybridization and interspecific gene flow will be determined largely by pollinator behavior. In experimental populations containing equal numbers of the two species and hybrids, pollinators make transitions between all three types of plants (Campbell, Waser, and Meléndez-Ackerman, 1997). Combining these frequencies of transitions with the amounts of pollen successfully exported and deposited on receptive stigmas, we expect moderately high rates of interspecific pollen flow in settings where the parental species are in close proximity (Campbell, Waser, and Wolf, 1998). Moreover, plants of I. tenuituba are expected to receive more foreign pollen than does I. aggregata. Thus, the formation of F1 hybrids is predicted to be asymmetrical. The evolutionary impact of that asymmetry will ultimately depend on relative survival rates of the two kinds of hybrid progeny. We do not yet know how F1 pollen fares in mixtures with conspecific pollen on a stigma, but we do have information on siring success of pollen from natural hybrids, which presumably include advanced generation recombinants. In mixed-donor pollinations on both I. aggregata and I. tenuituba, hybrid pollen sired fewer seeds than pollen from the parental species (Meléndez-Ackerman and Campbell, 1998). It is thus possible that pollen interactions reduce the formation of backcross offspring, while having no effect on production of F1 hybrids. This possibility is supported by paternity analysis in experimental mixed populations (Meléndez-Ackerman and Campbell, 1998). In that setting, relatively few backcross seeds with hybrid fathers and I. aggregata mothers were formed even though pollen transfer was predicted to be high for this kind of mating (Campbell, Waser, and Wolf, 1998). Because conspecific and heterospecific pollen perform similarly in siring seeds, and seed set varies little with composition of the pollen load, the relative male fitness of the two species in the hybrid zone should also be predictable from amounts of pollen exported to receptive stigmas. In most years pollinators strongly prefer to visit I. aggregata, and that preference leads to the prediction that I. aggregata has the highest success at exporting pollen, with hybrids intermediate between the parental species (Campbell, Waser, and Wolf, 1998). These patterns in pollination success are echoed in the high success of I. aggregata, compared to the other two types of plants, at siring seeds in experimental populations (Meléndez- Ackerman and Campbell, 1998). We are now engaged in estimating male and female fitness over the entire life cycle, including variation in survival, in order to predict the consequences for hybrid zone stability. In contrast to recent studies of other taxa, for the Ipomopsis hybrid zone at Poverty Gulch we saw no evidence that a conspecific pollen advantage reduces gene flow. This unusual result (but see Mather, 1947) raises questions about the circumstances under which a conspecific pollen advantage is important. The hybrid zone studied here is one with extensive hybridization. In some other areas in the western United States where I. aggregata and I. tenuituba come into contact, hybrids appear less common (Grant and Wilken, 1988). Although there are other plausible explanations, it is possible that geographic variation in hybridization relates in part to differences in the level of conspecific pollen advantage. Species with longer styles often have faster growing pollen tubes (Arnold, 1997), and style lengths do vary among contact sites. Repeating our experiment at contact sites with varying levels of hybridization would allow testing the hypothesis that pollen interactions help determine levels of hybridization in nature. LITERATURE CITED ARNOLD, M. L Natural hybridization and evolution. Oxford University Press, Oxford, UK., M. L., J. L. HAMRICK, AND B. D. BENNETT Interspecific pollen competition and reproductive isolation in Iris. Journal of Heredity 84: BURKE, J. M., S. E. CARNEY, AND M. L. ARNOLD Hybrid fitness in the Louisiana Irises: analysis of parental and F1 performance. Evolution 5: CAMPBELL, D. R Variation in sex allocation and floral morphology in Ipomopsis aggregata (Polemoniaceae). American Journal of Botany 79: Multiple paternity in fruits of Ipomopsis aggregata (Polemoniaceae). American Journal of Botany 85: , AND J. L. DOOLEY The spatial scale of genetic differentiation in a hummingbird-pollinated plant: comparison with models of isolation by distance. American Naturalist 139: , AND K. J. HALAMA Resource and pollen limitations to lifetime seed production in a natural plant population. Ecology 74: , N. M. WASER, AND E. J. MELÉNDEZ-ACKERMAN Analyzing pollinator mediated selection in a plant hybrid zone: hummingbird visitation patterns on three spatial scales. American Naturalist 149: ,, AND P. G. WOLF Pollen transfer by natural hybrids and parental species in an Ipomopsis hybrid zone. Evolution 5: CARNEY, S. E., M. B. CRUZAN, AND M. L. ARNOLD Reproductive interactions between hybridizing irises: analysis of pollen tube growth and fertilization success. American Journal of Botany 81: , S. A. HODGES, AND M. L. ARNOLD Effects of differential pollen tube growth on hybridization in the Louisiana Irises. Evolution 50: CRUZAN, M. B Pollen-pollen and pollen-stylar interactions during pollen tube growth in Erythonium grandifolium (Liliaceae). American Journal of Botany 77: DARWIN, C On the origin of species. John Murray, London, UK. ELLSTRAND, N. C., R. WHITKUS, AND L. H. RIESEBERG Distri-

6 84 AMERICAN JOURNAL OF BOTANY [Vol. 87 bution of spontaneous plant hybrids. Proceedings of the National Academy of Sciences, USA 93: EMMS, S. K., S. A. HODGES, AND M. L. ARNOLD Pollen tube competition, siring success, and consistent asymmetric hybridization in Louisiana Irises. Evolution 50: GRANT, V., AND D. H. WILKEN Natural hybridization between Ipomopsis aggregata and I. tenuituba (Polemoniaceae). Botanical Gazette 149: HARRISON, R.G.[ED.] Hybrid zones and the evolutionary process. Oxford University Press, Oxford, UK. HAUSER, T. P., R. B. JORGENSEN, AND H. OSTERGÅRD Preferential exclusion of hybrids in mixed pollinations between oilseed rape (Brassica napus) and weedy B. campestris (Brassicaceae). American Journal of Botany 84: KLIPS, R. A Pollen competition as a reproductive isolating mechanism between two sympatric Hibiscus species (Malvaceae). American Journal of Botany 86: KNOBLOCH, I. W Intergeneric hybridization in flowering plants. Taxon 1: KÖLREUTER, J. G Vortsetzung der Vorläufigen Nachricht von einigen das Geschlecht der Pflanzen betreffenden Versuchen und Beobachtungen. Gleditshischen Handlung, Leipzig, Germany. LEEBENS-MACK, J., AND B. G. MILLIGAN Pollination biology in hybridizing Baptisia (Fabaceae) populations. American Journal of Botany 85: MATHER, K Species crosses in Antirrhinum. I. Genetic isolation of species majus glutinosum and orontium. Heredity 1: MELÉNDEZ-ACKERMAN, E. J Patterns of color and nectar variation across an Ipomopsis (Polemoniaceae) hybrid zone. American Journal of Botany 84: , AND D. R. CAMPBELL Adaptive significance of flower color and inter-trait correlations in an Ipomopsis hybrid zone. Evolution 5: RIESEBERG, L. H., A. M. DESROCHERS, AND S. J. YOUN Interspecific pollen competition as a reproductive barrier between sympatric species of Helianthus (Asteraceae). American Journal of Botany 8: SAS SAS system for windows, release 6.1. SAS Institute, Cary, North Carolina, USA THOMSON, J. D Germination schedules of pollen grains: implications for pollen selection. Evolution 43:0 3 WASER, M. M., AND M. V. PRICE Reproductive costs of selfpollination in Ipomopsis aggregata (Polemoniaceae): are ovules usurped? American Journal of Botany 78: WILSON, H. D., AND J. S. PAYNE Crop / weed microgametophyte competition in Cucurbita pepo (Cucurbitaceae). American Journal of Botany 81: