| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Reproductive Biology |
Department of Biology, University of New Mexico, Albuquerque, New Mexico 87131 USA; Department of Organizational Learning and Instructional Technology, University of New Mexico, Albuquerque, New Mexico 87131 USA; Baker Hall Environmental Residential Academic Program, University of Colorado, Boulder, Colorado 80309 USA
Received for publication May 11, 2006. Accepted for publication December 21, 2006.
ABSTRACT
When more pollen is present on stigmas than needed to fertilize all ovules, selection among pollen grains may occur due to effects of both pollen donors and maternal plants. We asked whether increasing plant age and flower age, two changes in maternal condition, altered the pattern of seed paternity after mixed pollination. We also asked whether changes in seed paternity affected offspring success in an experimental garden. While flower age did not affect seed paternity, there was a dramatic shift in pollen donor performance as plants aged. These differences were seen in the offspring as well, where the offspring of one pollen donor, which sired more seeds on young plants, flowered earlier in the season, and the offspring of another pollen donor, which sired more seeds on old plants, flowered later in the season. Thus, change in maternal condition resulted in altered seed paternity, perhaps because the environment for pollen tube growth was different. The pattern of seed paternity and offspring performance suggests that pollen donors may show temporal specialization.
Key Words: Brassicaceae • flower age • mate choice • offspring fitness • plant age • pollen competition • sexual selection • wild radish
When more pollen grains arrive on stigmas than are needed to fertilize all of the ovules, there is opportunity for unequal mating success among pollen donors. This possibility is intriguing because it could result in sexual selection by means of competition among the pollen grains or choice by the maternal tissue (e.g., Mulcahy, 1974
, 1979
; Charnov, 1979
; Willson, 1979
, 1982
, 1990
; Stephenson and Bertin, 1983
; Queller, 1983
, 1987
; Willson and Burley, 1983
; Snow, 1994
; Delph and Havens, 1998
; Skogsmyr and Lankinen, 2002
). However, the suggestion that sexual selection occurs in plants is controversial for a number of reasons (e.g., Charlesworth et al., 1987
; Lyons et al., 1989
; Grant, 1995
), including the possibility that mating may be nonrandom due to the degree of relatedness of the pollen donors with maternal plants, the difficulty in distinguishing pollen competition from mate choice, and the problems in applying greenhouse results to field conditions (Winsor et al., 2000
).
Despite these difficulties, the possibility of sexual selection in plants continues to receive attention. Sexual selection has been powerful in animals, producing elaborate behaviors and morphologies (e.g., Andersson, 1994
). While more controversial in plants, it is logical to suggest that pollen donors that produce more successful pollen and maternal plants that can mate with superior pollen donors will be favored by selection. Thus, rather than dismiss this intriguing possibility, our goal is to address one of the problems in applying sexual selection to plants.
Distinguishing choice by the female tissue from pollen competition is particularly difficult because competing pollen grains and pollen tubes must be present for the maternal plant to exhibit choice (Marshall and Folsom, 1991
). However, experiments that alter the condition of maternal tissue while holding pollen donor condition constant allow inferences about the role of the maternal plant. That is, if mating patterns change when maternal condition changes, maternal tissue must be involved. In fact, altering the availability of water or nutrients to maternal plants can change mating patterns (e.g., Marshall and Ellstrand, 1988
; Marshall and Fuller, 1994
; Marshall and Diggle, 2001
). While effects of changes in maternal condition confirm a maternal role in mating, they do not exclude pollen donor effects. Pollen grains may respond to differences in stylar conditions created by changes in the maternal environment (Marshall and Ellstrand, 1988
).
Effects of maternal condition are interesting because maternal condition likely changes during the normal course of development in both the greenhouse and the field. For example, as the season progresses and plants begin to set fruit, availability of resources for additional fruits may be reduced (e.g., Marshall and Oliveras, 2001
). On a shorter time scale, condition of individual flowers may also vary. That is, if stigmas are receptive for several days, flowers pollinated on the first day may be in a physiologically different condition than those pollinated a day or two later. Temporal changes in maternal condition might also alter the environment for pollen tube growth, thus providing an opportunity to investigate the effects of maternal condition on mating patterns, responses of pollen grains to changes in the stylar environment, and the range of conditions under which nonrandom mating is possible.
The possibility that plant age affects mating patterns can be studied by examining changes in fruit set, seed set, and seed paternity. Many studies have investigated fruit set and/or number of seeds per fruit in plant species with acropetal inflorescence development (early work reviewed in Diggle, 1995
; also Guitián et al., 2001
, 2004
; Ehlers et al., 2002
; Gomez and Zamora, 2003
; Buide, 2004
; Bañuelos and Obeso, 2005
). Generally, there is a decrease in both fruit set and seed set in those flowers that open more distally (later) (Wesselingh and Arnold, 2003
). Several mechanisms may explain these patterns, including resource limitation, non-uniform pollination, and architectural effects (see Medrano et al., 2000
for a summary). However, few studies have asked whether temporal differences in the maternal environment may affect the outcome of mating, especially seed paternity (but see Marshall and Oliveras, 2001
).
Patterns of mate choice may change as flowers age due to differences in flower physiology. Known changes in flower physiology over time include variation in the degree of self-incompatibility (Ascher and Peloquin, 1966
; Levri, 1998
; Petanidou et al., 2001
; Vogler and Stephenson, 2001
; Goodwillie et al., 2004
; Goodwillie and Ness, 2005
) and declines in female reproductive capacity (Jakobsen and Martens, 1994
; Arathi et al., 2002
). Existing studies have not tested whether these temporal changes are accompanied by variation in the success of compatible pollen donors.
If mating patterns change as plants and flowers age, pollen donor fitness is necessarily altered because pollen donors sire different numbers of seeds. However, to determine whether maternal plant fitness is affected, offspring fitness must be taken into account. If the success of offspring varies with pollen donor identity, then maternal plant fitness must be affected by changes in mating patterns. While most studies involving paternal identity and offspring fitness have focused on inbreeding (e.g., Oostermeijer et al., 1995
; Fischer et al., 2003
; Lienert and Fischer, 2004
; Mitchell et al., 2005
), there is evidence that pollen donor identity affects offspring success. For example, seed paternity affects offspring characteristics in Cassia fasciculata (Lee and Hartgerink, 1986
), Raphanus sativus (Marshall and Whittaker, 1989
), Campsis radicans (Bertin and Peters, 1992
), Eupatorium (Byers, 1998
), Thymus vulgaris (Thompson et al., 2004
), and Betula pendula (Pietarinen and Pasonen, 2004
). However, strong effects of pollen donor identity were not found in Morus moraceae (Burgess and Husband, 2004
), and in some cases, the impact of pollen donor identity varies among maternal plants (Schmid and Dolt, 1994
; Moeller, 1997
; Skogsmyr and Lankinen, 2000
; Galloway, 2001
; Paschke et al., 2002
). Examples where pollen donor mating success and pollen donor effects on offspring fitness are simultaneously examined are particularly rare (but see Bertin and Peters, 1992
; Pasonen et al., 2001
).
We investigated the effects of age of plants and age of flowers on mating success in wild radish, Raphanus sativus (Brassicaceae L.), and then asked whether the pollen donors that were most successful in siring seeds produced the most successful offspring. We chose to study wild radish because of the wealth of previous information about this system. We know that compatible pollen donors sire different numbers of seeds (e.g., Marshall and Ellstrand, 1986
; Marshall and Diggle, 2001
; Marshall and Oliveras, 2001
). In previous studies changing maternal condition altered mating patterns, suggesting that maternal choice can be important (Marshall and Ellstrand, 1988
; Marshall and Fuller, 1994
; Marshall and Diggle, 2001
). However, we have not investigated the effect of floral age either by itself or in combination with changes in plant age. Both of these are effects that likely occur in the field. And, we have not used measures of offspring success to ask whether changes in mating pattern due to plant and flower age alter maternal fitness.
Therefore, we performed a greenhouse study of mating patterns and an experimental garden study of offspring success to address the following questions: As plants or branches age, does pollen donor mating success change? As flowers age, does pollen donor mating success change? Do plant-age- and flower-age-based differences in pollen donor mating success correlate with differences in offspring fitness?
We have two alternative predictions for our results. First, as plants and flowers age, pollen competition and mate choice may diminish (Marshall and Diggle, 2001
). If this is the case, when plants or flowers are older, mating should be closer to random, i.e., more equal among pollen donors. Pollen donors that sire seeds under these conditions may produce less fit offspring. Alternatively, as plants and flowers age, the environment for pollen tube growth may become different. Inasmuch as pollen tubes respond to the local environment, the relative seed siring abilities of the pollen donors would change as the stylar environment changes. If this is true, then there could be selection for offspring that do best in the kinds of conditions represented by the altered stylar environment. In fact, this has been observed in studies (reviewed in Hormaza and Herrero, 1992
) in which an increase in salt content or temperature in styles can select for progeny that are more tolerant of higher salt concentrations and temperatures.
MATERIALS AND METHODS
The plant
Wild radish, Raphanus sativus, is a self-incompatible (Hinata and Nishio, 1980
), weedy annual that grows along roadsides, in ditches, and in agricultural fields in California. Extensive background information on the mating system (Ellstrand, 1984
; Ellstrand and Marshall, 1986
; Ellstrand et al., 1989
) and pollination biology (e.g., Kay, 1976
; Stanton, 1987a
, b
; Stanton and Preston, 1988
; Ashman et al., 1993
) is available. And wild radish is a convenient study organism: Physiological self-incompatibility (Samson, 1957
; Hinata and Nishio, 1980
; Pundir et al., 1983
) makes emasculation of flowers prior to experimental crosses unnecessary, the numerous flowers on each plant allow replication of crosses, the size of the seeds (about 10 mg) facilitates weighing and electrophoresis of individual seeds, and well-developed electrophoretic systems make paternity analysis straightforward (Ellstrand, 1984
).
Experimental crosses
One-hundred twenty-five wild radish plants were grown from field-collected seeds to screen plants for use in this experiment. Plants were grown in 4-L pots in a 4 : 1 : 1 mixture of sand, peat, and perlite. They were watered 1–3 times daily, depending on their size. Beginning at 4 wk, plants were fertilized once weekly with Peter's 20 : 20 : 20 water soluble fertilizer (Scott's Co., Marysville, Ohio, USA) and with Peter's water soluble micronutrient fertilizer (Scott's Co.). Addition of NPK fertilizer was increased to twice weekly as the plants grew larger.
After genotypes were screened at a phosphoglucose isomerase (PGI) locus and tests for compatibility, 10 unrelated maternal plants (1–10) and three pollen donors (A, B, and C) were selected for the experiment. All plants were homozygous for the PGI locus, and the pollen donors had different PGI phenotypes. We screened for compatibility of the pollen donors with the maternal plants by performing test crosses. Crosses were considered compatible if they resulted in the production of multiseeded fruits.
The three pollen donors were used to make three kinds of single pollinations (A-only, B-only, and C-only) and one type of mixed pollination (A+B+C) on each maternal plant. Both single and mixed pollinations were made by collecting pollen in a small petri dish, mixing with tissue-wrapped forceps, and applying the pollen. For mixed pollinations, pollen was collected from equal numbers of flowers for each of the three pollen donors. This method results in application of hundreds of pollen grains per stigma (M. Folsom, University of New Mexico, personal communication). If pollen donors differed in the number of pollen grains per flower, the amount of pollen for each donor would not be equal. However, because of the duration of the experiment, any differences would not confound treatment effects. In addition, previous studies have shown that while pollen production of R. sativus can affect seed paternity, it does not explain all differences in pollen donor success (Marshall, 1998
). All pollinated flowers were marked with small, paper tags.
Crosses were performed on plants and flowers of two ages, and each cross was replicated 15 times for a total of 2400 crosses (10 maternal plants x 4 types of crosses x 2 plant ages x 2 flower ages x 15 replicates). Plant age was determined by time in the experiment. The initial series of crosses was performed on plants that had no maturing fruits (fruits from compatibility crosses were removed as soon as fruit set was assured). Plants at this time had 20–25 inflorescences, each of which produced 1–2 new flowers per d. The first set of crosses, when plants were considered young, was completed in 3–4 d for each plant. This resulted in 5–7 pollinated flowers per branch and a total of 120 pollinated flowers per plant. Half of the crosses (60) were performed on newly opened (young) flowers, and half were performed on flowers that had opened the previous day (old). Of the 30–40 crosses performed on each plant each day, we pollinated an approximately equal number of each type of cross (A, B, C, and mixed). We then waited 9 d to begin the second set of pollinations. At this time, plants were called old because they were already setting many fruits. The second set of crosses was also completed in 3–4 d. These crosses were performed on the same inflorescences that were used when the plants were young, so each branch used had some maturing fruits. Again, each branch had 5–7 additional pollinations, and each plant had an additional 120 pollinated flowers. Half of the pollinations were performed on newly opened flowers and half were performed on 1-d-old flowers.
Allocation to reproduction
We measured resources allocated to reproduction in several ways. Patterns of fruit set were monitored during fruit development. Failures in fruit set could be due to lack of fertilization or failure of fertilized fruits to develop. For consistency with other work, failures in fruit set are called fruit abortions. At maturity, all fruits were collected and seeds were counted and weighed to the nearest 0.01 mg. We also scored seed position within fruits where the position nearest the stem (basal) was numbered one and the position nearest the tip (stylar) had the highest number.
We measured differential success of pollen donors in mixed pollinations by examining the paternity of seeds. Seed paternity for those crosses was assessed through starch gel electrophoresis of the seed tissue followed by scoring for a locus of PGI (Ellstrand, 1984
).
Common garden study
To assess progeny fitness, a subset of the seeds from single pollinations (donor A, B, or C alone) was grown in a common garden. Seeds for the study were selected based on their parentage and their mass. We wished to control for seed mass because in R. raphanistrum, a close relative, seed mass affected plant growth (Stanton, 1984
). Mean seed mass was sufficiently variable across maternal plants that we could not match seed masses among maternal plants. However, it was possible to ameliorate variation in seed mass among pollen donors within maternal plants. Therefore, within maternal plants, we selected the seeds sired by each pollen donor that were closest to the mean seed mass (within 0.05 mg) of that maternal plant. Thus, while initial seed mass might have contributed to maternal effects on progeny growth, it should not contribute to paternal effects. In fact, an analysis of variance in the initial seed mass of the seeds planted showed no significant pollen donor effect (P > 0.05).
We planted 450 seeds (10 maternal plants x 3 pollen donors x 15 replicates) in mid-June. Seeds were sown to a uniform depth in 2 cm x 2 cm peat pots filled with 4 : 1 : 1 sand : peat : perlite. Pots were maintained in a protected, outdoor courtyard and watered twice daily until transplanted to the field site. Five weeks after planting, we transplanted 300 of the seedlings (10 maternal plants x 3 pollen donors x 10 replicates) to an experimental garden. The garden was a 9 x 4.5 m unshaded area adjacent to the University of New Mexico North Golf Course, Albuquerque, New Mexico, USA. While this is outside the normal range of the species, the temperature and soil type were within the limits of normal growth. We provided sufficient supplemental water to produce suitable growing conditions. Prior to planting, we rototilled the garden and added 113 g of Scotts 16 : 16 : 16 slow release fertilizer (Scott's Co.). Half of the seedlings were transplanted on each of two consecutive d. When we measured and harvested these plants (discussed later), we always performed the measures in a 2-d period so that each plant had an equal number of days to grow before measurement.
We placed the seedlings in a randomized block design comprised of 10 blocks with 30 plants each. One replicate of each maternal plant by pollen donor combination was included in each block. Blocks contained two rows of 15 plants. Rows were 40 cm apart and plants within rows were 25 cm apart. This was close enough to allow some competition later in the study. While in the field, plants were naturally pollinated. We watered the plants three times per d until the conclusion of the experiment.
We assessed progeny fitness in several ways. First, after the seeds were planted, we censused the pots daily. We recorded the number of days to emergence for each seedling. Second, we assessed size and growth by measuring the length of the longest leaf, the number of leaves, and plant height four times: 2, 3, 7, and 10 wk after sowing. At 2 and 3 wk, seedlings were still in the protected courtyard, and at 7 and 10 wk, plants were in the experimental garden. Plant height and leaf length were measured to the nearest 0.1 cm. Third, we assessed final plant size after 10 wk. All aboveground biomass was harvested, the tissue was dried at 90°C for 72 h and then weighed to the nearest 0.1 g. While an indirect measure of fitness, adult plant dry mass is highly correlated with fecundity in R. raphanistrum (Stanton, 1984
).
To assess fecundity, we scored reproductive status at 7 and 10 wk and counted total number of flowers at 10 wk. To score reproductive status, we classified plants as vegetative, budding, flowering, early fruiting, or having mature fruits. The number of fruits per plant was also counted at 10 wk as an estimate of realized fecundity. While this is not the total number of fruits a plant could produce in a longer season, measuring fruit number at a particular time is a reasonable estimate of fecundity for this plant. In the field, reproduction of wild radish is typically truncated by drought and not by senescence. The time of drought varies from year to year so that the time for reproduction is quite variable. Although this was an experimental garden, there was ample pollinator service as evidenced by the presence of fruits along entire racemes. Availability of pollinators is not likely to have limited reproduction.
Analysis
Our first concern in examining the data was to decide whether to treat maternal plant and pollen donor as random or fixed effects. Because the plants were not selected at random (they were selected based on compatibility relationships and PGI phenotype), we decided to treat these variables as fixed even though this limits our conclusions to the plants used in this study.
Frequencies of fruit abortion were compared in a categorical analysis (PROC CATMOD, SAS, version 9.1; SAS Institute, Cary, North Carolina, USA). Type of cross, plant age, flower age, and their interactions were used as independent variables in a categorical model where frequency of aborted and non-aborted fruits was the dependent variable.
Other reproductive variables were compared among treatments using ANOVAs (PROC GLM, SAS, version 9.1). Individual seed mass, number of good seeds per fruit, and number of aborted seeds per fruit were used as dependent variables in analyses of variance where pollen donor, maternal plant, plant age, flower age, and their interactions were used as independent variables. Following the ANOVAs, pairwise comparisons among means were made using Tukey's studentized range tests.
Patterns of seed paternity were analyzed in three ways. First, overall differences in seed paternity were examined by comparing numbers of seeds sired by each pollen donor in all mixed pollinations using a chi-square test. We asked whether the numbers of seeds sired were equal across the three pollen donors. Next, numbers of seeds sired by the three pollen donors were compared across maternal plant, flower age, plant age, and their interactions, using a categorical analysis (PROC CATMOD, SAS, version 9.1). This analysis treats each seed as an independent data point. This may be a good reflection of fitness effects for pollen donors, and the experiment was designed for this analysis. However, this technique may overestimate the true sample size because the paternity of each seed in a fruit might not be independent. Therefore, we also calculated the proportion of seeds sired by each donor in each fruit that was the result of a mixed pollination. This created a variable, proportion of seeds sired per fruit by each donor, that was measured at the level of the fruit, the level at which crosses were performed. Because the proportion of seeds sired by each of the three donors will always add up to one for a fruit, we could not use data for all three pollen donors in the same analysis. We used only the data for donors A and C. Then the arcsine square-root transformation of proportion of seeds sired per fruit was used as the dependent variable in an analysis of variance where pollen donor, maternal plant, plant age, flower age, and their interactions were the independent variables.
Offspring characteristics were compared in a number of ways. First, the number of seedlings emerging over time was compared in a survival analysis (PROC LIFETEST, SAS, version 9.1). We tested for equality in seedling emergence curves for pollen donor identity and maternal plant. This procedure does not facilitate testing interaction affects. Therefore, to test if pollen donor identity significantly affected seedling emergence among maternal plants, we compared survival curves for each pollen donor across each maternal plant individually. Model fit was evaluated by visually comparing a plot of log survival function estimates vs. survival time. We found that the Weibull model best fit our data. Thus only Wilcoxon test statistics are reported. Next, we used a repeated measures ANOVA to compare the growth of plants with different parents. Plant height, length of longest leaf, and number of leaves per plant (measured at multiple times) were the dependent variables, and pollen donor, maternal plant, block within the field site, and the maternal plant by pollen donor interaction were the independent variables.
Final measures of plant performance were compared using ANOVAs. Final plant height, length of longest leaf, number of leaves, biomass, and numbers of flowers and fruits were used as dependent variables in ANOVAs where pollen donor, maternal plant, block, and the pollen donor by maternal plant interaction were the independent variables. Number of flowers was log-transformed for analysis. Pairwise comparisons among pollen donors were made using Tukey's studentized range tests. The combined effects of the dependent variables measured at time of harvest were examined with a MANOVA (PROC GLM, SAS, version 9.1).
Reproductive status was compared among the offspring of maternal plants and pollen donors using categorical models (PROC LOGISTIC, SAS, version 9.1). At 7 wk, plants were categorized as being in the vegetative, bud and flower, or fruit stage. At 10 wk, plants were placed in the bud/flower, fruit, or mature fruit stage.
RESULTS
Pollination study: allocation to reproduction
Flowers were much less likely to set fruit on old than on young plants (Fig. 1, 
2 = 44.51, df = 1, P < 0.0001, in a categorical analysis). However, the proportion of fruits that aborted was unaffected by flower age. The difference in fruit abortion between young and old plants changed with type of cross (2 = 7.84, df = 3, P < 0.05).
|
= 6.14) with higher seed mass ( = 10.82 mg) than older plants ( = 5.87 seeds/fruit and = 10.79 mg). However, late seed abortion was unaffected by plant age. Young flowers produced significantly more good seeds per fruit ( = 6.45) than did old flowers ( = 5.52); there were no other effects of flower age. There were no plant age by flower age interaction effects on these reproductive variables. However, subtle interactions may be concealed within the significant three- and four-way interaction terms involving plant age and flower age (Table 1).
|
2 = 191.4, df = 2, P < 0.0001). The overall differences in seed paternity concealed intriguing differences in performance on young and old plants. Pollen donor C sired the most seeds on young plants, while pollen donor A sired the most seeds on old plants (Fig. 2a). The change in paternity for pollen donor A is particularly striking. This plant sired 26% of the seeds on young plants and 62% of the seeds on old plants (Fig. 2a). The effect of plant age on seed paternity was significant in both the categorical analysis of frequencies of seeds sired (Table 2) and in the ANOVA of proportion of seeds sired per fruit (Table 3). The ANOVA also reveals that the effects of plant age vary among pollen donors and among pollen donors within maternal plants (Table 3).
|
|
|
The abilities of pollen donors to sire seeds also differed somewhat among maternal plants. The effects of plant age and flower age varied significantly across maternal plants in the categorical analysis (Table 2), but not in the ANOVA (Table 3).
Effects of pollen donors on offspring performance
Almost all seeds (437/450) germinated within 12 d. The pattern of germination did not differ among pollen donors when all data were lumped (2 from Wilcoxon test = 1.05, df = 2, P = 0.59). However, the pattern of germination did differ among maternal plants (2 from Wilcoxon test = 39.3, df = 9, P < 0.0001).
All of our analyses indicated that patterns of growth and final size were affected by seed paternity. First, there was a significant effect of time on offspring height, leaf length, and number of leaves (Wilks' 
for the effect of time was significant in all repeated measures ANOVAs). Significant interaction effects of time with pollen donor and maternal plant in the repeated measures analysis indicate that growth of offspring depended on parentage. Examination of means (Table 4) shows the following differences in growth: (1) number of leaves did not differ among offspring of the three pollen donors at 3 wk, but by 7 wk, offspring of donor C tended to have the most leaves; (2) progeny of donor C tended to have the fewest leaves at all time intervals, but the difference was largest at 7 wk; (3) progeny of donor C tended to be tallest at 7 wk, but progeny of donor A were taller at 10 wk.
|
|
= 0.769, P < 0.0001), maternal plant (Wilks' = 0.369, P < 0.0001), and block (Wilks' = 0.519, P < 0.0001) on plant height, leaf length, leaf number, biomass, number of flowers, and number of fruits when all variables were considered together. There was no significant pollen donor by maternal plant interaction effect (Wilks' = 0.667, P = 0.544). Examination of the individual variables revealed that, at the end of the experiment, offspring sired by pollen donor A were the tallest and had the most and longest leaves (Table 4), although means for offspring of pollen donor A were not always significantly greater than for offspring of pollen donor B. There were significant effects of pollen donor, maternal plant, and block on all variables except biomass, for which there was only a significant effect of block (Table 6). Additionally, block was not significant for number of flowers. However, there were no significant pollen donor by maternal plant interaction effects.
|
|
|
When pollen grains on stigmas outnumber the ovules that can be fertilized, selection may occur among those pollen grains. While this appears to be a logical possibility, the mechanisms by which that selection might occur remain controversial. Either competition among the pollen grains for access to ovules, choice by the female tissue, or both are possible. Which of these occur and whether such selection can occur under field conditions remains uncertain. We approached this issue by investigating mating patterns on wild radish plants and flowers of different ages. As plants and flowers age, maternal condition changes. If the maternal tissue affects mating, seed paternity should differ among plants and flowers of different ages both because of direct maternal effects and because of pollen grain responses to different selective environments. These are particularly interesting cases to investigate because age-related changes in maternal plants likely occur in the field as well as the greenhouse. Additionally, to consider the effects of changes in mating on fitness, we measured the success of offspring sired by the pollen donors used in the mating study.
We first asked whether allocation to reproduction and mating patterns changed as plants aged. Fruit abortion was higher in older branches, and these branches produced fewer and smaller seeds per fruit (Fig. 1, Table 1), as has been observed in many species (reviewed in Diggle, 1995
). Resource limitation may partially explain these patterns if earlier-developing fruits sequester more resources than later-developing fruits. The reduced number of good seeds per fruit in older plants may be due to early seed abortion or lack of fertilization (Nakamura and Stanton, 1987
).
Mating patterns also varied as plants aged. The rank order of pollen donors changed as plants became older; the performance of pollen donor A increased dramatically (Fig. 2a). The increased performance of pollen donor A on older plants may be due to changes in maternal choice, lessened ability of maternal plants to distinguish among pollen donors, or improved growth of pollen tubes of donor A in the environment of older styles. Our results are somewhat similar to previous work on R. sativus in which the variance in number of seeds sired per fruit was lowest for the earliest pollinations, highest for the second group of pollinations, and then began to decline again as maternal plants aged (Marshall and Oliveras, 2001
). The results of the current study are much more striking because we saw a dramatic reversal in pollen donor performance. This may be due to differences in experimental design. In the 2001 study, crosses were performed continuously such that changes in maternal condition may have created a continuum of environments for pollen tube growth. This experiment was done in two stages such that there may have been a more abrupt shift in stylar environment. Alternatively, in this experiment we may simply have sampled a different range of pollen donor phenotypes.
The change in pollen donor performance (i.e., proportion of seeds sired per fruit) may occur due to characteristics of both the maternal plants and pollen donors. The conditions in the style may change as maternal plants age and fill more fruit. Certainly resources are being diverted to developing fruits and a decrease in resources may be reflected in the stylar environment of newly opened flowers. It is striking that the changes we saw occurred over a relatively short time. We waited only 9 d between the old and young plant treatments. However, since 120 pollinations were performed on young plants, each maternal plant was setting many fruit by the time the second set of pollinations began. Changes over time in the field might be even greater since plants produce flowers over several weeks and even several months, and of the hundreds and possibly thousands of flowers produced by each plant, many more than 120 are likely visited and pollinated in the field. Any subsequent changes in the stylar environment would likely alter the conditions for pollen germination and pollen tube growth. Pollen donors A and C may perform differently in these environments.
We next asked whether flower age was important to patterns of reproduction and mating. The effects of flower age on reproduction were not large. There was a slight reduction in the number of good seeds per fruit in older flowers (Table 1) that may have been due to fewer fertilizations, reduced stigma receptivity (Morse, 1987
; Jakobsen and Martens, 1994
), or a decline in ovule viability (Jakobsen and Martens, 1994
).
Changes in seed paternity as flowers aged were much more subtle than those accompanying changes in plant age. Donor C tended to sire more seeds on older than on younger flowers, but the difference was very small (Fig. 2b). This is in contrast with studies that compared self and outcross pollen. Older flowers often have an increase in selfing (Ascher and Peloquin, 1966
; Levri, 1998
; Vogler and Stephenson, 2001
; Goodwillie et al., 2004
; Davis and Delph, 2005
). Thus, in these wild radish plants, effects of flower age were much less important than changes in plant age. However, we used only fresh and 1-d-old flowers. Wild radish flowers can be receptive for at least another day. Perhaps we would have seen stronger effects on reproduction if we had tested even older flowers.
While changes in number of seeds sired intrinsically alter pollen donor fitness (Marshall, 1991
, 1998
; Marshall and Diggle, 2001
; Marshall and Olivaras, 2001; Morand-Prieur et al., 2003
), maternal plant fitness will only be changed if pollen donor identity affects offspring quality. We found substantial differences in growth and reproduction among the progeny of the three pollen donors. For example, plants sired by pollen donor C bolted first (as indicated by increases in height at 7 wk) and reproduced sooner (Fig. 3). Plants sired by donor A remained in a vegetative phase for a longer time and began producing and maturing fruits later than the progeny of donor C (Fig. 3). At the end of the experiment, plants sired by donor A tended to have greater mass and produce more fruits than donor C (Table 6). Plants sired by donor B were intermediate in timing of bolting and commitment to reproduction; however, donor B had the most fruits at the end of the experiment (Fig. 3). Whether reproducing early like donor C or waiting and reproducing more like donor A confers greater fitness depends on the length of the season. This is relevant to the field because in wild populations, the growing season of R. sativus is water-limited, but the timing of water availability may change substantially from one year to the next.
Thus, the patterns in offspring growth suggest that when maternal plants mate with different pollen donors, maternal fitness may be changed. Other studies support the conclusion that the identity of the pollen donor affects offspring success. For instance, there were similar effects of pollen donors on offspring for a different set of R. sativus pollen donors (Marshall and Whittaker, 1989
), but effects of maternal age on pollen donors' success were not included in that study and the effects of pollen donor identity on offspring success were measured in a greenhouse rather than in the field.
Finally, we considered whether the differences in mating pattern on young and old plants produced differences in offspring success that might be adaptive. The pollen donor that sired the most seed on young maternal plants (C) produced offspring that matured earlier and devoted more resources to early reproduction than early vegetative growth. In contrast, the pollen donor that sired more seeds on older maternal plants (A) produced offspring that matured later and had the potential for much greater reproduction later in the season (Fig. 3).
In the field, a pollen donor such as donor C that allocates resources to flowering early in the season will mostly encounter maternal plants that are young and producing few fruits. Thus, the behavior of its pollen tubes may reflect selection under conditions present in styles early in the season. If the season is truncated due to early drought, pollen donors like donor C would be the most fit.
Alternatively, a pollen donor like A that produces flowers later will usually encounter older maternal plants that are already producing many fruits. Its pollen tube responses may have been selected in the context of the stylar conditions present in older plants. If the season is long, due to an extended rainy season, donors like A will get quite large, produce many flowers, perform well in competition with other donors and be the most fit. The results from our study indicate that the extension of the season need only be a few weeks to see profound differences in pollen donor and offspring success.
The idea that selection among pollen tubes in a particular stylar environment can favor phenotypes that correlate with seedling phenotypes that suit the same environment is not new. Artificial selection has been used to select among pollen tubes in particular stylar environments, such as saline or high temperature conditions. This selection has produced seedlings that are more tolerant of those kinds of conditions than control seedlings (reviewed in Hormaza and Herrero, 1992
). We hypothesize that the same kind of process that produces results under artificial selection can produce adaptive variation in the field.
Our results suggest that changes in maternal conditions over time and pollen tube responses to that variation may act to produce temporal specialization among pollen donors. Pollen donor C appears to specialize on young maternal plants, while pollen donor A may specialize on older maternal plants. While we only studied three pollen donors here, similar results for a larger study would imply that this variation, coupled with variation in the length of the season in the field, would work together to maintain variation in pollen donor characteristics. Environmental change that led to increasing numbers of short or long seasons would favor one type of pollen donor over the other and reduce the current variation in pollen donor abilities.
FOOTNOTES
1 The authors thank T. Bennett and S. Wolterstorff for greenhouse and laboratory assistance, L. Rupert and K. McGee for laboratory assistance, and B. Cabin and B. Milne for field assistance. Funding was provided by the UNM Student Research Allocations Committee and the Biology Department Graduate Student Research Allocations Committee of the University of New Mexico. Additional support was provided by a National Science Foundation REU supplement to NSF grant BSR-8614967. The authors also thank the University of New Mexico North Golf Course for access to the field site. 
2 Author for correspondence (e-mail: marshall{at}unm.edu
)
LITERATURE CITED
Andersson M. B.. 1994. Sexual selection Princeton University Press, Princeton, New Jersey, USA.
Arathi H. S. Rasch A. Cox C. Kelly J. K.. 2002. Autogamy and floral longevity in Mimulus guttatus. International Journal of Plant Sciences 163: 567-573.
Ascher P. D. Pelonquin S. J.. 1966. Effect of floral aging on the growth of compatible and incompatible pollen tubes in Lilium longiflorum. American Journal of Botany 53: 99-102.
Ashman T. L. Galloway L. F. Stanton M. L.. 1993. Apparent vs. effective mating in an experimental population of Raphanus sativus. Oecologia 96: 102-107.
Bañuelos M. J. Obeso J.. 2005. How is fruit production regulated in the dioecious fleshy-fruited shrub Rhamnus alpinus?. Basic and Applied Ecology 6: 249-259.
Bertin R. I. Peters P. J.. 1992. Paternal effects on offspring quality in Campsis radicans. American Naturalist 140: 166-178.
Buide M. L.. 2004. Intra-inflorescence variation on floral traits and reproductive success of the hermaphrodite Silene acutifolia. Annals of Botany 94: 441-448.
Burgess K. S. Husband B. C.. 2004. Maternal and paternal components of fitness of hybrids between red and white mulberry (Morus moraceae). American Journal of Botany 91: 1802-1808.
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]
Charlesworth D. Schemske D. W. Sork V. L.. 1987. The evolution of plant reproductive characters: sexual vs. natural selection. In S. C. Stearns [ed.], Evolution of sex 317-335 Birkhauser, Basel, Switzerland.
Charnov E. L.. 1979. Simultaneous hermaphroditism and sexual selection. Proceedings of the National Academy of Sciences, USA 76: 2480-2484.
Davis S. L. Delph L. F.. 2005. Prior selfing and gynomonoecy in Silene noctiflora (Caryophyllaceae): opportunities for enhanced outcrossing and reproductive assurance. International Journal of Plant Sciences 166: 475-480.
Delph L. F. Havens K.. 1998. Pollen competition in flowering plants. In T. R. Birkhead, A. P. Moller [eds.], Sperm competition and sexual selection 149-173 Academic Press, New York, New York, USA.
Diggle P. K.. 1995. Architectural effects and the interpretation of patterns of fruits and seed development. Annual Review of Ecology and Systematics 26: 531-552.[CrossRef][ISI]
Ehlers B. K. Oleson J. M. Agren L.. 2002. Floral morphology and reproductive success in the orchid Epicactis helleborine: regional and local across-habitat variation. Plant Systematics and Evolution 236: 19-32.
Ellstrand N. C.. 1984. Multiple paternity within the fruits of the wild radish, Raphanus sativus. American Naturalist 123: 819-828.
Ellstrand N. C. Devlin B. Marshall D. L.. 1989. Gene flow by pollen into small populations: data from experimental and natural stands of wild radish. Proceedings of the National Academy of Sciences, USA 86: 9044-9047.
Ellstrand N. C. Marshall D. L.. 1986. Patterns of multiple paternity in populations of Raphanus sativus. Evolution 40: 837-842.
Fischer M. Hock M. Paschke M.. 2003. Low genetic variation reduces cross-compatibility and offspring fitness in populations of a narrow endemic plant with a self-incompatibility system. Conservation Genetics 4: 325-336.
Galloway L. F.. 2001. Parental effects on life history in the herbaceous plant Campanula americana. Ecology 82: 2781-2789.
Gomez J. M. Zamora R.. 2003. Factors affecting intra-fruit pattern of ovule abortion and seed production in Honnathophylla spinosa (Cruciferae). Plant Systematics and Evolution 239: 215-229.
Goodwillie C. Ness J. M.. 2005. Correlated evolution in floral morphology and the timing of self-compatibility in Leptosiphon jepsonii (Polemoniaceae). International Journal of Plant Sciences 166: 741-751.
Goodwillie C. Partis K. L. West J. W.. 2004. Transient self-incompatibility confers delayed selfing in Leptosiphon jepsonii (Polemoniaceae). International Journal of Plant Sciences 165: 387-394.
Grant V.. 1995. Sexual selection in plants: pros and cons. Proceedings of the National Academy of Sciences, USA 92: 1247-1250.
Guitián J. Guitián P. Medrano M.. 2001. Causes of fruit set variation in Polygonatum odoratum (Liliaceae). Plant Biology 3: 637-641.
Guitián J. Medrano M. Oti J. E.. 2004. Variation in floral sex allocation in Polygonatum odoratum (Liliaceae). Annals of Botany 94: 433-440.
Hinata K. Nishio T.. 1980. Self-incompatibility in crucifers. In S. Tsunoda, K. Hinata, C. Gomez-Campo [eds.], Brassica crops and wild alleles: biology and breeding 223-234 Japanese Scientific Society, Tokyo, Japan.
Hormaza J. I. Herrero M.. 1992. Pollen selection. Theoretical and Applied Genetics 83: 663-672.
Jakobsen H. B. Martens H.. 1994. Influence of temperature and ageing of ovules and pollen on reproductive success in Trifolium repens L. Annals of Botany 74: 493-501.
Kay Q. O. N.. 1976. Preferential pollination of yellow-flowered morphs of Raphanus raphanistrum by Pieris and Eristalis spp. Nature 261: 230-232.
Lee T. D. Hartgerink A. P.. 1986. Pollination intensity, fruit maturation pattern, and offspring quality in Cassia fasciculata. In D. L. Mulcahy, G. B. Mulcahy, E. Ottaviano [eds.], Biotechnology and ecology of pollen 417-422 Springer-Verlag, New York, New York, USA.
Levri M. A.. 1998. The effect of timing of pollination on the mating system and fitness of Kalmia latifolia (Ericaceae). American Journal of Botany 85: 1626-1630.
Lienert J. Fischer M.. 2004. Experimental inbreeding reduces seed production and germination independent of fragmentation of populations of Swertia perennis. Basic and Applied Ecology 5: 43-52.
Lyons E. E. Waser N. M. Price M. V. Antovics J. Motten A. F.. 1989. Sources of variation in plant reproductive success and implications for concepts of sexual selection. American Naturalist 134: 409-433.
Marshall D. L.. 1991. Non-random mating in wild radish: variation in pollen donor success and effects of multiple paternity among one- to six-donor pollinations. American Journal of Botany 78: 1404-1418.
Marshall D. L.. 1998. Pollen donor performance can be consistent across maternal plants in wild radish (Raphanus sativus, Brassicaceae): a necessary condition for the action of sexual selection. American Journal of Botany 85: 1389-1397.
Marshall D. L. Diggle P. K.. 2001. Mechanisms of differential pollen donor performance in wild radish, Raphanus sativus (Brassicaceae). American Journal of Botany 88: 242-257.
Marshall D. L. Ellstrand N. C.. 1986. Sexual selection in Raphanus sativus: experimental data on non-random fertilization, maternal choice, and consequences of multiple paternity. American Naturalist 127: 446-461.
Marshall D. L. Ellstrand N. C.. 1988. Effective mate choice in wild radish: evidence for seed abortion and its mechanism. American Naturalist 131: 739-756.
Marshall D. L. Folsom M. W.. 1991. Mate choice in plants—an anatomical to population perspective. Annual Review of Ecology and Systematics 22: 37-63.[Medline]
Marshall D. L. Fuller O. S.. 1994. Does nonrandom mating among wild radish plants occur in the field as well as in the greenhouse?. American Journal of Botany 81: 439-445.
Marshall D. L. Oliveras D. M.. 2001. Does differential seed siring success change over time or with pollination history in wild radish, Raphanus sativus (Brassicaceae)?. American Journal of Botany 88: 2232-2242.
Marshall D. L. Whittaker K. L.. 1989. Effects of pollen donor identity on offspring quality in wild radish, Raphanus sativus. American Journal of Botany 76: 1081-1088.
Medrano M. Guitián P. Guitián J.. 2000. Patterns of fruit and seed set within inflorescences of Pancratium maritimum (Amaryllidaceae): non-uniform pollination, resource limitation, or architectural effects?. American Journal of Botany 87: 493-501.
Mitchell R. J. Karron J. D. Holmquist K. G. Bell J. M.. 2005. Patterns of multiple paternity in fruits of Mimulus ringens (Phrymaceae). American Journal of Botany 92: 883-890.
Moeller A. P.. 1997. Developmental selection against developmentally unstable offspring and sexual selection. Journal of Theoretical Biology 185: 415-422.
Morand-Prieur M. Raquin C. Shykoff J. A. Frascaria-Lacoste N.. 2003. Males outcompete hermaphrodites for seed siring success in controlled crosses in polygamous Fraxinus excelsios (Oleaceae). American Journal of Botany 90: 949-953.
Morse D. H.. 1987. Roles of pollen and ovary age in follicle production of the common milkweed Asclepias syriaca. American Journal of Botany 74: 851-856.
Mulcahy D. L.. 1974. Correlation between speed of pollen tube growth and seedling height in Zea mays L. Nature 249: 491-493.
Mulcahy D. L.. 1979. The rise of angiosperms: a genecological factor. Science 206: 20-23.
Nakamura R. R. Stanton M. L.. 1987. Cryptic seed abortion and the estimation of ovule fertilization. Canadian Journal of Botany 65: 2463-2465.
Oostermeijer J. G. B. Altenburg I. G. M. Den Nijs H. C. M.. 1995. Effects of outcrossing distance and selfing on fitness components in the rare Gentiana pneumonanthe (Gentianaceae). Acta Botanica Neerlandica 44: 257-268.
Paschke M. Abs C. Schmid B.. 2002. Effects of population size and pollen diversity on reproductive success and offspring size in the narrow endemic Cochlearia bavarica (Brassicaceae). American Journal of Botany 89: 1250-1259.
Pasonen H.-L. Pulkkinen P. Kapyla M.. 2001. Do pollen donors with fastest-growing pollen tubes sire the best offspring in an anemophilous tree, Betula pendula, (Betulaceae)?. American Journal of Botany 88: 854-860.
Petanidou T. Ellis-Adam A. C. Den Nijs H. C. M. Oostermeijer J. G. B.. 2001. Differential pollination success in the course of individual flower development and flowering time in Gentiana pneumonanthe (Gentianaceae). Botanical Journal of the Linnean Society 135: 25-33.
Pietarinen P. Pasonen H.. 2004. Pollen performance and male fitness in an anemophilous monoecious tree, Betula pendula. Canadian Journal of Botany 82: 1284-1291.
Pundir N. S. Abbas R. F. Al-Attar A. A. A.. 1983. Pollen germination and pollen tube growth in Raphanus sativus L. following self- and cross-pollinations. Phyton—International Journal of Experimental Botany 43: 127-130.
Queller D. C.. 1983. Sexual selection in a hermaphroditic plant. Nature 305: 705-707.
Queller D. C.. 1987. Sexual selection in flowering plants. In J. W. Bradbury, M. B. Andersson [eds.], Sexual selection: testing the alternatives 165-181 Wiley, New York, New York, USA.
Samson D. R.. 1957. The genetics of self-compatibility in the radish. Journal of Heredity 48: 26-29.
Schmid B. Dolt C.. 1994. Effects of maternal and paternal environment and genotype on offspring phenotype in Solidago altissima L. Evolution 48: 1525-1549.
Skogsmyr L. Lankinen A.. 2000. Potential selection for female choice in Viola tricolor. Evolutionary Ecology Research 2: 965-979.
Skogsmyr L. Lankinen A.. 2002. Sexual selection: an evolutionary force in plants?. Biological Review 77: 537-562.
Snow A. A.. 1994. Postpollination selection and male fitness in plants. American Naturalist 144: S69-S83.
Stanton M. L.. 1984. Seed variation in wild radish: effect of seed size on components of seedling and adult fitness. Ecology 64: 1105-1112.
Stanton M. L.. 1987a. Reproductive biology of petal color variants in wild populations of Raphanus sativus. I. Pollinator response to color morphs. American Journal of Botany 74: 178-187.
Stanton M. L.. 1987b. Reproductive biology of petal color variants in wild populations of Raphanus sativus. II. Factors limiting seed production. American Journal of Botany 74: 188-196.
Stanton M. L. Preston R. E.. 1988. Ecological correlates of petal size variation in wild radish, Raphanus sativus (Brassicaceae). Oecologia 74: 68-76.
Stephenson A. G. Bertin R. I.. 1983. Male competition, female choice, and sexual selection in plants. In L. Real [ed.], Pollination biology 109-149 Academic Press, Orlando, Florida, USA.
Thompson J. D. Tatayre M. Gauthier P. Litricio I. Linhart Y. B.. 2004. Multiple genetic contributions to plant performance in Thymus vulgaris. Journal of Ecology 92: 45-56.
Vogler D. W. Stephenson A. G.. 2001. The potential for mixed mating in a self-incompatible plant. International Journal of Plant Sciences 162: 801-805.
Wesselingh R. A. Arnold M. L.. 2003. A top-down hierarchy in fruit set on inflorescences in Iris fulva (Iridaceae). Plant Biology 5: 651-660.
Willson M. F.. 1979. Sexual selection in plants. American Naturalist 113: 777-790.
Willson M. F.. 1982. Sexual selection and dicliny in angiosperms. American Naturalist 119: 579-583.
Willson M. F.. 1990. Sexual selection in plants and animals. Trends in Ecology and Evolution 5: 210-214.
Willson M. F. Burley N.. 1983. Mate choice in plants Princeton University Press, Princeton, New Jersey, USA.
Winsor J. A. Peretz S. Stephenson A. G.. 2000. Pollen competition in a natural population of Curcurbita foetidissima (Curcurbitaceae). American Journal of Botany 87: 527-532.
| ||||||||||||||||||||||||||||||||||||||||||||||||||
