Printer Friendly

Effects of pollen quantity on progeny vigor: evidence from the dessert mustard Lesquerella fendleri.

Intense competition among pollen grains for access to ovules often increases progeny vigor (Mulcahy 1971, 1979; Mulcahy and Mulcahy 1975; Mulcahy et al. 1975; McKenna 1986; Winsor et al. 1987; Bertin 1990; Schlichting et al. 1990; Richardson and Stephenson 1992; Janse and Verhaegh 1993; Quesada et al. 1993, 1996a,b; Palmer and Zimmerman 1994; but see Smith et al. 1990; Snow 1990, 1991). The most common explanation for this is that selection on gametophytes is an efficient means for removing deleterious alleles at loci important for both gametophytic and sporophytic life stages (Charlesworth and Charlesworth 1992; Hormaza and Herrero 1992; Stephenson et al. 1992). Gametophytic selection therefore has been proposed as an important factor in the success of angiosperms (Mulcahy 1979).

Studies of pollen competition and progeny vigor usually involve variation in the number of pollen grains applied to stigmas or of the distance pollen tubes must travel to reach ovules (utilizing natural or artificial variation style length). Studies of the latter type invariably show significant effects on progeny vigor (Mulcahy and Mulcahy 1975; Ter-Avanesian 1978; McKenna and Mulcahy 1983; Lee and Hartgerink 1986; McKenna 1986). However, effects of pollen load size are less consistent (e.g., Winsor et al. 1987; Bertin 1990; Smith et al. 1990; Snow 1990, 1991; Palmer and Zimmerman 1994). Furthermore, most studies to date involve cultivated plant species (but see Snow 1990, 1991; Richardson and Stephenson 1992), which may respond differently than wild plants because of their history of close inbreeding and hybridization.

To better understand the effects of pollen competition on progeny vigor, and why results vary among different studies, more empirical data are needed. Here I report results from a study of the effects of pollen load size on progeny vigor for a native, desert mustard (Lesquerella fendleri; Brassicaceae). I then review published studies of the effects of pollen load size on progeny vigor, considering why results may vary among methods and species.


Lesquerella fendleri is a small perennial common in the deserts of the southwestern United States. The 15-mm diameter, yellow flowers are simultaneously hermaphroditic and are self-incompatible. Flowers in the field receive an average of 319 pollen grains (SE = 43, n = 17) during their one- to two-day existence, and produce 8.5 seeds per mature fruit (SE = 0.9, n = 60). Flowers typically contain about 20 ovules, and are pollinated by small bees and beeflies (unpubl. data).

To assess the effects of pollen load size on progeny vigor, I artificially altered the amount of pollen applied to flowers on 13 potted plants in a greenhouse. All plants were collected from within 100 m of one another at the Sevilleta Long Term Ecological Research area (LTER), 80 km south of Albuquerque, New Mexico, in fall 1992. I applied three pollination treatments to flowers in randomized order within plants, using monofilament fishing line (Kearns and Inouye 1993). To do this, I placed anthers from at least eight plants not otherwise involved in the experiment in a small petri dish, and coated the fishing line with pollen by swirling it in the dish. Then, I either touched the line briefly to the stigma (low pollen treatment), swiped it lengthwise along the stigma once (medium treatment), or swiped it lengthwise along the stigma four or more times (high treatment), reloading the line with pollen after each touch or swipe. I intended these three treatments to provide a wide range of pollen availability, but still allow full or nearly full seed set (Mitchell 1997). I applied six replicates of these three treatments to each plant from June 8-13, 1993 (18 flowers on each of 13 plants, 234 flowers total). More than 72 hours after pollination I collected stigmas, squashed them on a microscope slide with basic fuchsin stain (Kearns and Inouye 1993), and counted pollen grains (Mitchell 1997). I analyzed results from all recovered fruits, including those with no seeds. Along with seed number and individual seed mass (total seed mass of a fruit/number of seeds in fruit), I also analyzed the proportion of pollen that successfully sired seeds (ratio of the number of seeds in a fruit to the number of pollen grains applied; seed:pollen ratio) as an index of the intensity of pollen selection. Unfortunately, I was unable to distinguish self and outcross pollen when counting pollen loads, and attempts to prevent self-pollen deposition were unsuccessful (see Mitchell 1997). Therefore, seed:pollen ratios include both the outcross pollen I applied and passively deposited self pollen (average self pollen loads on unpollinated flowers = 92 grains, n = 74). Trials indicated that prior arrival of self pollen does not decrease seed production (unpubl. data).

In September 1993 I planted seeds into sterilized sand in 6-cm pots in a greenhouse at the University of New Mexico, and sprayed them with 1 g/L gibberellic acid to break seed dormancy (Evans et al. 1996). Two maternal plants did not [TABULAR DATA FOR TABLE 1 OMITTED] produce enough seeds for planting, so I excluded them. I planted seeds from each treatment-plant combination in a randomized block design, with 42 blocks that each contained up to 33 seeds (three treatments for each of 11 plants). I recorded emergence of cotyledons (emergence date) to the nearest half-day, and measured seedling size (length and width of longest leaf, number of leaves) as each seedling reached 21 days postemergence.

To assess effects of pollen load size on progeny performance in the field, I transplanted some of the seedlings described above back to their site of origin. To do so, on November 21, 1993, I constructed 10 new blocks of seedlings. These blocks included as complete a sample of plant-treatment combinations as possible, achieved by randomly choosing plants from those alive in the original planting blocks. After cold-hardening these blocks for one week, I transplanted them to within 100 m of the origin of the parental plants at the Sevilleta LTER.

I used the remaining plants to construct new blocks for greenhouse measurements, in the same manner described for the field seedlings. Because unequal numbers of seedlings remained for each plant-treatment combination (I gave first priority to producing even replication in the field blocks), these blocks were strongly unbalanced. I recorded date of first flowering and date of death for each plant.

For analyses, I used mixed-model ANOVA (Proc GLM; SAS Institute 1989) and Type III sums of squares, considering maternal plant and replicate as random factors, and treatment as fixed. Therefore I tested treatment over treatment x plant (Sokal and Rohlf 1985).


As anticipated, pollen loads differed significantly among treatments, but seed set per fruit and proportion fruit set did not (Table 1). As a consequence, treatments differed significantly in seed:pollen ratios (Table 1). Overall, the seed:pollen ratio differed two-fold among treatments, with the fastest 6% of pollen grains siring seeds in the low treatment, and the fastest 3% in the high treatments, respectively. Thus, all treatments resulted in intense pollen selection (low seed:pollen ratios). Seed size did not vary among treatments. For brevity, Table 1 shows statistical tests only for treatment effects. For most traits, the full ANOVA also revealed significant maternal effects, and significant maternal plant x replicate interactions, which indicate that the magnitude of the differences among plants varied among replicates. In no cases were interactions of treatment with any other factor significant. Sequential Bonferroni adjustments to probability levels (Rice 1989) to account for nonindependence of traits did not change these conclusions.

I found no detectable effect of pollination treatment on any measure of early offspring quality. Pollen load size did not detectably affect date of emergence or proportion of seeds emerging (Table 2). These conclusions persisted when the data were analyzed using the accelerated failure time analysis advocated by Fox (1993; [[[Chi].sup.2].sub.2]P = 0.6). Pollen load size also had no significant effect on any of the postgermination characters measured in the first 21 days (Table 2). There were no significant differences among maternal sibships for any traits except emergence date, proportion emerging, and number of leaves. Significant maternal plant x treatment interactions (for leaf length, leaf width, and leaf number) made it difficult to draw unambiguous conclusions about maternal plant and treatment effects in any case. Earlier emerging seedlings tended to grow larger by 21 days after emergence (the correlation between emergence date and the 21-day measures ranges from -0.16 to -0.28, all P [less than] 0.002), but including emergence date as a covariate still did not reveal significant effects of treatment (unpubl. data). Furthermore, the lack of a detectable effect was not due to low statistical power. Power analysis (Cohen 1969) indicated that the observed magnitude of variation among treatments provided [greater than] 70% probability of detecting significance, even for the small treatment effects that occurred in this experiment. The sole exception to this was leaf number, for which power was 15%.

For plants transplanted to the field, leaf number just before transplantation (54 days after planting) did not vary significantly among treatments (Table 2) although there was significant variation among maternal sibships (P = 0.004; the maternal plant x treatment interaction approached significance, P = 0.06). Unfortunately, all 283 plants unexpectedly died within two weeks of transplanting. The plants grown to maturity in the greenhouse showed no significant differences among treatments in date of flowering or date of death (Table 2).


Despite strong and significant variation among treatments in pollen load size and significant (though small) differences in seed:pollen ratios, I detected no significant effect of pollination treatment on any measure of offspring quality. Indeed, for only two of 10 measures did seeds from high pollen loads tend (nonsignificantly) to perform best. Perhaps any [TABULAR DATA FOR TABLE 2 OMITTED] effects of pollen load size on sporophytic quality would have been more noticeable under field conditions (Lee and Hartgerink 1986; Snow 1990) and after maternal effects dissipated, but the unexpected mortality of all transplants prevented that comparison here. However, other studies have shown effects at the seed and three- to four-week-old seedling stage (e.g., Bertin 1990; Bjorkman 1995), so the abbreviated time scale here is comparable with that in other studies that found positive results.

Why do my results and those of a few others (Smith et al. 1990; Snow 1990, 1991) differ from those for so many other species? Possible explanations include: inadequate statistical power; utilization of "small" pollen loads that were too large (see Schlichting et al. 1990); and study of a wild, self-incompatible plant species that might respond differently than self-compatible or cultivated species (see Snow 1990). The first possibility is readily dismissed as a general explanation, since my study had adequate power to detect reasonably strong effects had they existed (see Results).

To investigate the possibility that studies showing no effect involved "small" pollen loads that nonetheless were large enough to generate such intense pollen competition that no further gains in vigor were possible, I attempted to review all published studies of the effects of pollen load size on progeny vigor. I excluded from consideration any studies using other methods to vary pollen competition, such as cutting the style to different lengths (e.g., Mulcahy and Mulcahy 1975). All such studies showed a significant effect of pollen competition of progeny vigor. I also excluded any that did not report pollen load sizes and seed set per flower (e.g., Smith et al. 1990). For each study, I noted whether pollen load had a significant overall effect on any progeny character, and calculated seed:pollen ratio (as defined in the Methods) as an index of the intensity of pollen selection. When this ratio is small, few pollen grains successfully fertilize ovules, and competition is therefore more intense. I calculated this ratio separately for the lowest and highest pollen load treatments in each study. I used these values to compare studies finding no effect of pollen load size on any measure of progeny vigor with those finding a significant effect on any estimate of progeny vigor. Ideally, information such as pollen germination and attrition could be used to more accurately assess the fraction of potentially successful pollen that actually succeeded in siring seeds, but such information was not available for most studies.

I located 12 studies that met my criteria, five of which involved Cucurbita pepo (Table 3). To avoid oversampling that species, I only report results from one study on C. pepo, and from one on crosses between C. pepo and C. texana. My conclusions were unaltered by this conservative step. Because pollen and seed values for some studies varied or were ambiguous, Table 3 reports both high and low estimates of seed:pollen ratios. For simplicity, I report results based on the lowest values for each study. The trends reported below were unaffected by which estimates were used, although statistical significance of the differences disappeared when the largest estimates were used. Seed:pollen ratios were lower overall in the three studies that found no significant effect of pollen load size on progeny vigor. The mean value for low pollen loads was 0.082 for studies showing no effect, and 0.266 for those showing an effect (Table 3). The difference between means for these groups was biologically large, and marginally significant (one-tailed t-test on arcsine, squareroot transformed values, P = 0.04). Because of the small sample size, only very strong differences were likely to be significant, and I took this to indicate that pollen selection tended to be very intense in the studies, showing no effect, even in the low pollen treatment. The mean seed:pollen ratio for high pollen loads was 0.021 for studies showing no effect, and 0.076 for those showing an effect (Table 3). This difference approaches significance (one-tailed t-test as described above, P = 0.10). Together, these results are consistent with [TABULAR DATA FOR TABLE 3 OMITTED] the proposition (see also Schlichting et al. 1990) that the low treatments in my study and in Snow's (1990, 1991) were not low enough to produce a meaningful difference between treatments in pollen competition. In other words, although the low treatment received less pollen, it may nonetheless have caused such intense selection (in my study only 6% of pollen grains successfully sired seeds) that no additional gains in vigor were possible when the intensity of competition was doubled so that only 3% of L. fendleri pollen tubes were successful.

Figure 1 represents this argument (see also Snow 1986; Schlichting et al. 1990; Richardson and Stephenson 1991; Mitchell 1997). In this hypothetical example, as pollen load increases, pollen competition and progeny vigor both increase nonlinearly, but at different rates. In the region labeled "filling," pollen loads are small enough that only a fraction of ovules are fertilized. This means there is little or no pollen competition, and that more pollen increases seed set, but not seedling vigor. In the region labeled "sorting," all ovules are fertilized (or more ovules are fertilized than can be matured), and seed set per fruit does not increase with pollen load. However, seed:pollen ratio decreases, and an even-smaller fraction of the available pollen tubes is successful. This improves mean offspring quality. In the region labeled "surplus," more pollen does not increase progeny vigor because all available genetic variation for offspring quality has been exploited.

In terms of Figure 1, studies finding no effect may have used pollen loads that were so large that both high and low pollination treatments were in or close to the "surplus" region. Also, there may have been no effect of pollen load on vigor if both treatments were in the "filling" region, but this would be signified by large differences in seed set per fruit among treatments. To detect a significant effect of pollen competition or progeny vigor, one pollination treatment must be in a different region from the others. Studies showing a significant effect probably had the low treatment in the "filling" or "sorting" regions, and the high treatment in the "surplus" region. However, because seeds in few-seeded fruits frequently are larger than those in many-seeded fruits, comparison of vigor for seedlings from the "filling" region with any other region may be suspect (Charlesworth 1988; Stephenson et al. 1988), requiring more complicated experimental designs to remove effects of seed size on vigor (e.g., Schlichting et al. 1990; Quesada et al 1993, 1996a,b).

Is it possible to determine the boundaries between these different regions a priori? I know of no reliable way to predict the upper level of the "sorting" region. However, the upper limit for the "filling" region can be determined from dose-response relationships, since it should occur at the point where seed production no longer increases with pollen load. The dose-response relationship for L. fendleri (Mitchell 1997) indicates that the upper end of the "filling" region should be [approximately] 120 pollen grains. This means that my low treatment was larger than the upper limit of the "filling" region by 20 or so pollen grains, and suggests that all of my three treatments were probably in the "sorting" and "surplus" zones. In retrospect, using a pollen load smaller than the breakpoint of the dose-response curve ([approximately] 50) might have better allowed me to detect any effect of pollen competition on progeny vigor. Note that dose-response curves and this breakpoint may vary among individual plants (Mitchell 1997), further complicating the issue.

Are there other differences between those studies finding a significant effect of pollen load size on progeny vigor and those that do not? Self-compatibility is not associated with a particular outcome (Table 3), but all three studies showing no effect involved wild species. This is not consistent with the hypothesis that cultivated, self-compatible plants (which are likely to have low genetic variation) would show no effect. However, many other hypotheses are consistent with these sparse data. More studies, especially of wild species, are required before these can be tested definitively.

From my results I draw two main conclusions. First, results for L. fendleri show no indication that pollen load size affects early measures of progeny vigor in this species, despite a statistically powerful design. Second, one-third to one-fourth of published studies of the effects of pollen load size on progeny vigor also show no effect. In principle, the reason for this might be that different species respond differently to pollen load size, but my review indicates that it may instead be an artifact of using a low pollen treatment that is still large enough to generate substantial pollen competition. However, more empirical work on wild plants is needed.


J. Avritt, B. Cabin, A. Evans, D. Furcht, P. Gegick, D. Marshall, E. Mitchell, K. Mitchell, and C. O'Dear provided help, support, and advice. Special thanks to N. Waser for suggesting inclusion of Figure 1, and to A. Stephenson, A. Snow, and an anonymous reviewer for many helpful comments on an earlier version of the manuscript. The Sevilleta LTER and the Sevilleta National Wildlife Refuse provided permission to collect and transplant plants. R. Bertin, A. Snow, and A. Stephenson provided unpublished information on pollen and seed production in their studies. This research was supported by National Science Foundation grant DEB 92-03203. This is publication 95 of the Sevilleta LTER.


BERTIN, R. I. 1990. Effects of pollination intensity in Campsis radicans. Am. J. Bot. 77:178-187.

BJORKMAN, T. 1995. The effect of pollen load and pollen grain competition on fertilization success and progeny performance in Fagopyrum esculentum. Euphytica 83:47-52.

CHARLESWORTH, D. 1988. Evidence for pollen competition in plants and its relationship to progeny fitness: a comment. Am. Nat. 132:298-302.

CHARLESWORTH, D., AND B. CHARLESWORTH. 1992. The effects of selection in the gametophyte stage on mutational load. Evolution 46:703-720.

COHEN, J. 1969. Statistical power analysis for the behavioral sciences. Academic Press, New York.

EVANS, A. S., R. J. MITCHELL, AND R. CABIN. 1996. Morphological side effects of using gibberellic acid to induce seed germination: consequences for the study of seed dormancy. Am. J. Bot. 83: 543-549.

FINGERETT, E. R. 1979. Pollen competition in a species of evening primrose, Oenothera organensis, Munz. Master's thesis. Washington State Univ., Pullman.

FOX, G. A. 1993. Failure-time analysis: emergence. flowering, survivorship, and other waiting times. Pp. 253-289 in S. M. Scheiner and J. Gurevitch, eds. Design and analysis of ecological experiments. Chapman Hall, New York.

HORMAZA, J. I., AND M. HERRERO. 1992. Pollen selection. Theor. Appl. Gen. 83:663-672.

JANSE, J., AND J. J. VERHAEGH. 1993. Effects of varying pollen load on fruit set, seed set and seedling performance in apple and pear. Sex. Plant Repro. 6:122-126

KEARNS, C. A., AND D. W. INOUYE. 1993. Techniques for pollination biologists. Univ. of Colorado Press, Niwot.

LEE, T. D., AND A. P. HARTGERINK. 1986. Pollination intensity, fruit maturation pattern, and offspring quality in Cassia fasciculata (Leguminosae). Pp. 417-422 in D. L. Mulcahy, G. B. Mulcahy, and E. Ottaviano, eds. Biotechnology and ecology of pollen. Springer-Verlag, New York.

MCKENNA, M. A. 1986. Heterostyly and microgametophytic selection: the effect of pollen competition on sporophytic vigor in two distylous species. Pp. 443-448 in D. L. Mulcahy, G. B. Mulcahy, and E. Ottaviano, eds. Biotechnology and ecology of pollen. Springer-Verlag, New York.

MCKENNA, M. A., AND D. L. MULCAHY. 1983. Ecological aspects of gametophytic competition in Dianthus chinensis. Pp. 317-322 in D. L. Mulcahy and E. Ottaviano, eds. Pollen: biology and implications for plant breeding. Elsevier Biomedical, New York.

MITCHELL, R. J. 1997. Effects of pollination intensity on Lesquerella fendleri seed set: variation among plants. Oecologia 109:382-388.

MULCAHY, D. L. 1971. A correlation between gametophytic and sporophytic characteristics in Zea mays. Science 171:1155-1156.

-----. 1979. The rise of the angiosperms: a gnecological factor. Science 206:20-23.

MULCAHY, D. L., AND G. B. MULCAHY. 1975. The influence of gametophytic competition on sporophytic quality in Dianthus chinensis. Theor. Appl. Gen. 46:277-280.

MULCAHY, D. L., G. B. MULCAHY, AND E. OTTAVIANO. 1975. Sporophytic expression of gametophytic competition in Petunia hybrida. Pp. 227-232 in D. L. Mulcahy, ed. Gamete competition in plants and animals. North-Holland Publishing, Amsterdam, The Netherlands.

PALMER, T. M., AND M. ZIMMERMAN. 1994. Pollen competition and sporophyte fitness in Brassica campestris - does intense pollen competition result in individuals with better pollen? Oikos 69: 80-86.

QUESADA, M., J. A. WINSOR, AND A. G. STEPHENSON. 1993. Effects of pollen competition on progeny performance in a heterozygous cucurbit. Am. Nat. 142:694-706.

-----. 1996a. Effects of pollen competition on progeny vigor in a Cucurbita pepo x C. texana hybrid. Theor. Appl. Gen. 92:885-890.

-----. 1996b. Effects of pollen competition on the reproductive performance in cucurbit hybrids (Cucurbitaceae): [F.sub.1] and backcross generations. Can. J. Bot. 74:1113-1118.

RAMSTETTER, J. 1988. Pollen competition in Aureolaria. Ph.D. diss. Univ. of Massachusetts, Amherst.

RICE, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.

RICHARDSON, T. E., AND A. G. STEPHENSON. 1991. Effects of parentage, prior fruit set and pollen load on fruit and seed production in Campanula americana L. Oecologia 87:80-85.

-----. 1992. Effects of parentage and size of the pollen load on progeny performance in Campanula americana. Evolution 46: 1731-1739.

SAS Institute. 1989. SAS/STAT user's guide. Vers. 6, 4th ed. Statistical Analysis Systems Institute, Cary, NC.

SCHLICHTING, C. D., A. G. STEPHENSON, L. F. SMALL, AND J. A. WINSOR. 1990. Pollen loads and progeny vigor in Cucurbita pepo: the next generation. Evolution 44:1358-1372.

SMITH, S. E., D. M. CONTA, AND U. BEHCERT. 1990. Pollen load, seed position, and agronomic performance in alfalfa. Crop Sci. 30:561-565.

SNOW, A. A. 1986. Pollination dynamics in Epilobium canum Onagraceae: consequences for gametophytic selection. Am. J. Bot. 73:139-151.

-----. 1990. Effects of pollen-load size and number of donors on sporophyte fitness in wild radish (Raphanus raphanistrum). Am. Nat. 136:742-758.

-----. 1991. Effects of pollen-load size on sporophyte competitive ability in two Epilobium species. Am. Mid. Nat. 125:348-355.

SNOW, A. A., AND S. J. MAZER. 1988. Gametophytic selection in Raphanus raphanistrum: a test for heritable variation in pollen competitive ability. Evolution 42:1065-1075.

SOKAL, R. R., AND F. J. ROHLF. 1985. Biometry. 2d ed. Freeman, New York.

STEPHENSON, A. G., J. A. WINSOR, C. D. SCHLICHTING, AND L. E. DAVIS. 1988. Pollen competition, nonrandom fertilization, and progeny fitness: a reply to Charlesworth. Am. Nat. 132:303-308.

TER-AVANESIAN, D. V. 1978. The effect of varying the number of pollen grains used in fertilization. Theor. Appl. Gen. 52:77-79.

WALSH, N. E., AND D. CHARLESWORTH. 1992. Evolutionary interpretations of differences in pollen tube growth rates. Quart. Rev. Biol. 67:19-37.

WASER, N. M., R. G. SHAW, AND M. V. PRICE. 1995. Seed set and seed mass in Ipomopsis aggregata variance partitioning and inferences about postpollination sexual selection. Evolution 49: 80-89.

WINSOR, J. A., L. E. DAVIS, AND A. G. STEPHENSON. 1987. The relationship between pollen load and fruit maturation and the effect of pollen load on offspring vigor in Cucurbita pepo. Am. Nat. 129:643-656.
COPYRIGHT 1997 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1997 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Mitchell, Randall
Date:Oct 1, 1997
Previous Article:Comparative patterns of craniofacial development in Eutherian and Metatherian mammals.
Next Article:Advantages of multiple matings to females: a test of the infertility hypothesis using lizards.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters