| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Biology, University of New Mexico,Albuquerque, New Mexico 87131
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: Brassicaceae • matechoice • plant matingsystems • pollencompetition • Raphanussativus • self-incompatibility • sexualselection • wild radish
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This statement, however, points to one of the most important reasonsfor the controversy about sexual selection plants. Plants have a varietyof mechanisms for sorting among mates (Marshalland Folsom, 1991). Mating may be nonrandom due to effects ofphysiological incompatibility (deNettencourt, 1977; Richards,1986), lack of success of too closely related pollen donors(Holsinger, 1992; Waller, 1993), differences in quality amongcompatible donors (Marshall, 1991),discrimination against too distantly related pollen donors (Waser and Price, 1994), or failure of pollen fromheterospecific pollinations. Some of these mechanisms of nonrandommating, such as self-incompatibility systems and effects of inbreedingand outbreeding depression, produce very specific interactions betweenpollen donors and maternal plants (e.g., Bertinand Sullivan, 1988). That is, in these cases the success of aparticular pollen donor depends on its relationship with a particularpollen recipient and not on its other characteristics, such as pollentube growth rate. When these interactions are strong, overall selectionon mating performance, or sexual selection, is unlikely because adifferent pollen donor might sire the most seeds on each maternal plant(Waser et al., 1987; Waser and Price, 1993; Rigney, 1995). There is considerable concernabout this problem because incompatibility systems and effects ofinbreeding, in particular, are very common and much studied in plants.Therefore, in order to show that sexual selection can act in plants itis necessary to show that sorting among mates occurs that is not solelydue to the incompatibility system or to other mechanisms that producestrong and specific maternal plant by pollen donor interactions.However, specific tests of this necessary condition for sexual selectionin plants are lacking. Snow and Spira(1991b) made an important step by examining pollen tube growthacross several recipients and by scoring seed paternity across severalplants (Snow and Spira 1996). This studygoes a step further by explicitly ruling out effects of theself-incompatibility system while also testing for the paternity ofseeds.
I tested whether nonrandom mating, not due to overt or subtle effectsof incompatibility, occurred in wild radish, Raphanus sativusL. The mating system of this species is well studied. It is known tohave sporophytic self-incompatibility (Sampson, 1957; Hinata and Nishio, 1980; Pundir, Abbas, and Al-Attar, 1983) with30–50 S-alleles per population (Karron,Marshall and Oliveras, 1990). Effects of several levels ofinbreeding are also known (Nason and Ellstrand,1995). In addition, a variety of previous experiments usingexperimental pollinations suggest that sorting among compatible matesoccurs (Marshall and Ellstrand, 1986,1988; Marshall, 1991; Marshall and Fuller, 1994). In each of theseexperiments, two to six pollen donors, known to be compatible with thematernal plants, were used to pollinate three to ten maternal plants.Pollen donors had overall unequal mating success in all of theseexperiments, which supports the possibility of nonrandom mating amongcompatible donors.
However, because of the small number of maternal plants used, none ofthe previous experiments provides a strong test of consistency of pollendonor performance. In addition, the compatibility relationships withinthe groups of maternal plants or paternal plants were not tested.Therefore, it is possible that several pollen recipients within anexperiment shared S-alleles so that subtle effects of theincompatibility system may have affected the results. It is becomingincreasingly recognized that effects of the incompatibility system arenot always immediate or strong (Seavey and Bawa,1986; Bertin and Sullivan,1988). Thus, if maternal plants shared S-alleles, a subtleeffect of this recognition system might have allowed one pollen donorto have the greatest success in any particular experiment. In order toconvincingly test for the necessary conditions for sexual selection inplants as well as to put the previous data on wild radish in a properperspective, it is essential to examine pollen donor performance acrosslarger numbers of maternal plants and to explicitly address thepotential for effects of the incompatibility system.
I performed this test in two ways. First, I factored out overalleffects of the S-allele system by the design of my experiment. I used aset of maternal plants known to have different S-alleles and askedwhether postpollination performance of pollen donors was consistentacross these plants. If subtle effects of the incompatibility system areimportant, I expected to see strong maternal plant by pollen donorinteraction effects on seed siring ability. However, if the S-allelesystem is not the only sorting mechanism and there are differences inseed siring ability among the pollen donors, I expected to seeconsistent pollen donor performance across the maternal plants. That is,the overall differences in number of seeds sired by each pollen donorwill be larger than any interactions between pollen donors and maternalplants. In a second experiment, I circumvented the incompatibilitysystem by pollination of immature flowers. In this case, I asked whethersorting among compatible mates still occurred after this manipulation. Asimilar manipulation was performed by Cruzan(1993) on petunia and suggested by Lyonset al. (1989). If all sorting involves incompatibility, thenall sorting among mates should be eliminated by circumventing theincompatibility system. However, if sorting occurs among compatiblemates, it should be little affected by the experimentalmanipulations.
In the first experiment, I also considered, more peripherally,whether maternal condition affected the outcome of mating. That is, Iconsidered whether the frequencies of seeds sired by the pollen donorsvaried according to maternal plant treatment. Previous studies of wildradish have shown that altering the condition of maternal plants canaffect seed paternity (Marshall and Ellstrand,1988). If these effects are large, then they may affect thesuccess of pollen donors sufficiently to mask any role of the particularmating characteristics of donors. It is particularly important toconsider those effects in this experiment as it was done in thegreenhouse. If the results are robust across levels of maternal stress,then it is more likely that similar patterns would occur in fieldpopulations.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Experiment1
This experiment tested for differences in pollen donor success acrossmaternal plants known to have different S-alleles. To generate lineageswith distinct S-alleles, a large group of plants was grown from seedscollected from two field populations. These populations were from twofield sites near Riverside, California. They are a few kilometres apartand have similar climate and soils. These plants were screened forcompatibility until 16 fully intercompatible plants were located. Foreach pair of plants, five test crosses were performed in each direction(that is, both plants were used as males and females). Since overallfruit set was 74%, I considered crosses to be compatible if atleast three of the crosses set fruit on each member of the pair. Thatis, a cross was considered compatible only if it was reciprocallycompatible. This procedure was continued until a set of 16 fullyintercompatible plants was identified. Additional crosses were doneamong those plants in order to generate seeds for the experiment.
Since all plants could be reciprocally crossed with all other plants,these plants contained 32 S-alleles (Karron,Marshall, and Oliveras, 1990). (Wild radish has sporophyticincompatibility, so successful mating requires that the maternal plantand pollen donor each have two different S-alleles.) The 16 fullyintercompatible plants were also screened for their isozyme phenotypesfor two loci, phosphoglucose isomerase (Pgi) and leucine aminopeptidase (Lap). (See Ellstrand[1984] for methods.) Next, pairs of these plantswere crossed to create four maternal plant lineages and four pollendonor lineages: a total of eight lineages, each with differentS-alleles. The pairs were chosen so that at least some of theiroffspring would be homozygous for Pgi and Lap. Fourplants were grown from each maternal lineage (plants 1–16) and onefrom each pollen donor lineage (A-D). Since the pollen donor andmaternal plant lineages all had different S-alleles, no overall patternof pollen donor effects can be due to interactions among particularplants with specific S-alleles. While this design was essential toaddressing the questions posed, it constrained the number of plants thatcould be be used. In finding 16 fully intercompatible plants I likelyhad found almost all of the available S-alleles (Karron, Marshall, and Oliveras, 1990). Addingmore plants would have required more S-alleles.
As a consequence of selecting pollen donor lineages that could bedistinguished from each other by isozymes, all four pollen donorlineages happened to have one parent from each of the two originalpopulations. Three of the maternal lineages had two parents from thefirst field population and one had two parents from the second. Possibleconsequences will be considered in the discussion.
The 16 maternal plants for this experiment were divided into twowater-treatment groups so that I could also consider whether maternalcondition affected pollen donor performance. At the time plants beganflowering, two of the maternal plants from each lineage were randomlyselected as control plants and given sufficient water to saturate thepot three times daily. The other maternal plants were given half thecontrol amount of water three times daily. These plants wilted everyday.
In order to measure pollen donor success in siring seeds, allpossible pairwise mixed pollinations (AB, AC, AD, BC, BD, and CD) wereperformed. Pollen was collected by tapping flowers on the bottom ofsmall petri dishes, and all flowers were handled identically. Pollen wascollected from equal numbers of flowers from each donor. Pollen was thenmixed, and applied to stigmas with tissue wrapped forceps so as to coatthe stigmas with pollen. All pollinations were performed in the morningwhen greenhouse temperatures were relatively cool. Only freshly openedflowers were selected. Each cross was repeated 15 times on each maternalplant. In order to stratify the crosses over time, a set of one of eachkind of cross was performed on a maternal plant before going on to thenext set of replicate crosses.
For these mixed pollinations, mature fruits were collected and, foreach fruit, the number of seeds and location of each seed within thefruit (from top to bottom or stylar to basal end) were recorded.Finally, the paternity of mature seeds was scored by starch gelelectrophoresis at two loci, Pgi and Lap (Ellstrand, 1984). The total number of seedsscored for paternity was 5891.
Pollen donor performance was compared by examining the numbers ofseeds sired by each pollen donor. Although only pairwise crosses wereperformed, all crosses were lumped for each plant for analysis becauseequal numbers of each kind of cross were performed. This makes analysisand presentation simpler, but might lead to misinterpretation becauselumping the data may confound the effects of differential siring ofseeds within fruits and differential abortion of fruits. However, therewere no differences in amount of fruit abortion among the pairwisecrosses (
= 4.45, df = 5, P <0.49), so that was not a problem.
I compared numbers of seeds sired by each pollen donor in severalways. First, I asked whether numbers of seeds sired by each pollen donordiffered across the entire experiment by performing a chi-squareanalysis on the lumped data. I then repeated that analysis on eachmaternal family and each maternal plant to test whether the numbers ofseeds sired by each pollen donor were equal across these groups. Thesetests probably overestimated the statistical significance of the resultsas each seed was treated as independent and the paternity of seedswithin fruits is probably not completely independent. Therefore, to besomewhat more conservative, I did an ANOVA in which the number of seedssired within a fruit was the dependent variable and family, stresstreatment, and their interactions were independent variables. Because ofthe breeding design used to generate these plants, all independentvariables were treated as fixed. In this case, since there were twopollen donors per cross, there were two values of number of seeds siredfor each fruit. I used the count of seeds sired rather than theproportion because the proportions of seeds sired will add up to one foreach fruit. In this analysis, effects of maternal family and stresstreatment are effects on seed number per fruit. While this analysis ismore conservative, it still uses several fruits for each plant. Inconsequence, to test for overall differences in the performance ofpollen donors, I also performed two more ANOVAs in which I used only onevalue per pollen donor per maternal plant, reducing the sample size to64 (4 pollen donors x 16 maternal plants). First, I used thearcsine square root transformation of the proportion of seeds sired byeach pollen donor on each maternal plant as the dependent variable in ananalysis of variance where maternal plant and pollen donor were theindependent variables. Both independent variables were treated asrandom. Then, I repeated the analysis using the rank of pollen donorperformance as the dependent variable. In both cases, if pollen donorperformance was consistent, the effect of pollen donor on frequency ofseeds sired will be significant.
To compare performance of pollen donors across maternal stresstreatments, I used a chi-square contingency analysis. I asked whetherfrequencies of seeds sired by the four pollen donors were independentacross stress treatments.
Finally, I used a log-linear analysis to examine the frequencies ofseeds sired by pollen donors across both maternal families and stresstreatments. This analysis used a hierarchical series of models rangingfrom one containing all possible independent variables to those that donot include some variables or their interactions to test the effect ofeach independent variable and their interactions (see Horvitz and Schemske [1995] orCaswell [1989] for acomplete explanation of this kind of analysis). Significance tests areperformed by comparing models that do not contain the variable ofinterest with those that do. Where possible, I performed both marginaland conditional tests of significance, although the results of the twokinds of tests agreed in every case.
To test whether pollen donors were consistently able to set seed andto address whether differences in ability of pollen donors could bemeasured in the absence of pollen competition, single-donor crosses werealso performed. Pollen from each donor (A-D) was used to pollinate eachmaternal plant (1–16). These pollinations were performed in thesame manner as mixed pollinations, but were only replicated ten timeseach. Fruits from these crosses were collected when mature, and theseeds were counted and weighed. Differences in performance among pollendonors were tested in ANOVAs where average seed mass per fruit or numberof seeds per fruit were the dependent variables and maternal family,maternal stress treatment, pollen donor, and their interactions were theindependent variables. Due to the breeding program used to generatethese plants, all factors were treated as fixed.
Experiment 2
Next, I tested whether sorting among compatible mates could occurwhen the incompatibility system was compromised by performing crosses onimmature stigmas that had ineffective incompatibility systems. For thistest, I used five maternal plants (1a-5a), known to be from lineageswith different S-alleles. This experiment was smaller than the previousexperiment because of the difficulty in performing the required budpollinations. Three pollen donors were selected for each maternal plant:that plant (for self pollen), an unrelated compatible donor (donors A- Efor plants 1a-5a, respectively), and a test donor, Z. Donor Z was thesame for all maternal plants, so pollen from a total of 11 differentplants was used. Donor Z was the standard against which the performanceof all other pollen donors were compared. This is important since selfpollen, for example, must be from a different donor (the maternalparent) for every maternal plant examined. Two kinds of mixedpollination were made on each plant in order to test the performance ofself pollen and pollen from an unrelated donor against pollen from donorZ. One mixture consisted of pollen from one of five donors (A-E) mixedwith pollen from the test donor Z. The second mixture was of self pollenmixed with pollen from donor Z.
These crosses were made on flowers of two ages, newly opened flowers,like those used in the first experiment, and immature flowers ~2 dpreanthesis. Test crosses indicated that the immature flowers were youngenough that self pollination could result in seed production in thesenormally incompatible plants. Pollination of newly opened flowers wasperformed as in experiment 1. For pollination of immature stigmas,sepals and petals of preanthesis flowers were removed to expose thestigma. Application of pollen was as in experiment 1. Each cross withimmature flowers was repeated 25 times. Crosses with mature flowers wererepeated ten times on each maternal plant. The difference in sample sizewas meant to account for the possibility of damage to flowers during budpollinations that can lead to fruit abortion. Paternity of mature seedsfrom these mixed pollinations was scored by isozyme analysis as in theprevious experiment.
The data were analyzed in three steps. First, to test whether theself incompatibility system had been compromised, I compared the numbersof seeds sired by donor Z and self pollen on mature and immature stigmasin a chi-square contingency table. The null hypothesis was that numbersof seeds sired by the donors were independent of flower age. Then, totest whether there was nonrandom mating, numbers of seeds sired by eachdonor in a pair were compared against an expectation of equal seedpaternity in chi-square tests. Each cross was tested separately asdifferent donors were used on the five maternal plants. Crosses onmature and immature flowers were analyzed separately because I wasinterested in whether frequencies of seeds sired by the pollen donorswere equal in each case. Due to high fruit and seed abortion, samplesizes were modest (14–80 seeds/maternal plant from mature flowersand 45–122 seeds/maternal plant from immature flowers). Therefore,my ability to detect nonrandom mating was less than in previousexperiments with this species. Finally, I compared performance of thepollen donors in pollinations of immature and mature flowers in twoways. To compare the overall amount of sorting among donors on flowersof the two ages, I used the absolute value of the difference inproportion seeds sired by donor Z and the other compatible donor on eachmaternal plant as the dependent variable in a paired comparisonst test. Then, I performed separate contingency table analyseson each maternal plant to compare the frequencies of seeds sired by thepollen donors across the two flower ages.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
= 8.99, df = 9, P< 0.40). However, when pollen donor performance was compared acrossindividual maternal plants, there was a significant difference infrequency of seeds sired across plants ( = 94.4,df = 45, P < 0.0001). This is due in large part tothe differences across maternal plants that were grown under low-waterand control conditions. The ANOVA of number of seeds sired per fruit confirms that effects ofpollen donor identity are more important than those of maternal familyby pollen donor interactions (Table2). The two strongest main effects were maternal family andpollen donor. Because the analysis was done on number of seeds sired perfruit, the effect of maternal family reflects differences in seed numberper fruit among families. While there was a significant effect ofmaternal family by pollen donor, the variance explained by this effectwas very small (R = 0.014) as compared tothat explained by pollen donor (R = 0.118).Other effects of individual maternal plants are included in the maternalfamily x stress treatment x pollen donor interaction. Whilethis term is statistically significant (P < 0.0436), itexplains very little of the variance in number of seeds sired per fruit(R = 0.0046).
Numbers of seeds sired by the four pollen donors differedsignificantly among stress treatments (Table 3). However, this differencewas not due to a change in rank order among pollen donors. Rank order ofperformance was identical across the treatments. Rather, it was due to aslight evening out of pollen donor performance on the low-water plants.This interpretation is confirmed by the ANOVA (Table 2) in which stress treatmenthas a small effect on number of seeds sired per fruit, but the stresstreatment by pollen donor interaction effect is not statisticallysignificant.
|
|
|
|
= 9.94, df = 1, P < 0.002).The two seeds scored as sired by self pollen on mature stigmas are mostlikely errors in recording. It was then important to ask whether there was nonrandom seedpaternity among this group of compatible donors when used to pollinatemature stigmas. There was statistically significant nonrandom mating incrosses using mature stigmas on two of the five maternal plants(Table 7). This is asomewhat lower amount of nonrandom mating than I have seen in otherstudies (e.g., Marshall, 1991) and isprobably due to the relatively small sample sizes of seeds in thisexperiment.
|
Even so, the patterns of mating were not identical after pollinationof mature and immature stigmas. The amount of sorting among compatibledonors was less in crosses on immature stigmas than in crosses on maturestigmas. The mean of the absolute value of the difference in proportionof seeds sired between the pairs of donors was 0.12 on immature stigmasand 0.34 on mature stigmas (t = 2.78, P= 0.051 in a paired comparisons t test). However, if thefrequencies of seeds sired are compared across mature and immatureflowers on each maternal plant the difference is significant in only oneof five cases (Table7).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In experiment 1, four pollen donors were shown to have highlyconsistent differences in performance across 16 maternal plants fromfour different maternal lineages. The rank order of percentages of seedssired by each donor was absolutely consistent across the four maternallineages even though those lineages were known to have differentS-alleles. This pattern of pollen donor performance cannot be explainedby direct or subtle effects of the incompatibility system. If theincompatibility system had been involved, there would have beendifferences in pollen donor success across the maternal lineages.Another sorting mechanism must be involved.
There were some small differences in pollen donor performance acrossmaternal plants within lineages. Some of those differences were due tothe stress treatments applied to maternal plants. Thus, frequencies ofseeds sired by pollen donors varied somewhat with maternal condition.However, the rank order of number of seeds sired did not vary withmaternal condition. That is, the same pollen donors sired the most andleast seeds overall under both conditions. And, the ANOVA of numbers ofseeds sired per fruit showed that pollen donor identity explained farmore of the variance than either the maternal family by pollen donor orthe maternal family by pollen donor by stress interactions.
These results are consistent with previous work done on wild radish(Marshall and Ellstrand, 1986,1988; Marshall, 1991; Marshall and Fuller, 1994). In some earlierexperiments, there was evidence of maternal plant by pollen donorinteraction effects on seed paternity (Marshalland Ellstrand, 1986). However, in spite of those interactions,the overall performance of pollen donors was sufficiently similar acrossplants to produce strong overall winners and losers. These previousexperiments did not, however, explicitly rule out the effects of theS-allele system.
These results also agree with work on Hibiscus moschutus(Snow and Spira, 1991a). Differences inpollen tube growth among pairs of pollen donors were consistent acrossseveral pollen recipients. In addition, ability to sire seeds wasconsistent among three pollen donors tested across several maternalplants and with several competitors (Snow andSpira, 1996).
Although it is clear that the four pollen donors differed inperformance, the data presented here cannot explain the mechanisms ofthose differences. Pollen was applied directly to stigmas, sopostpollination differences in performance are the most likelyexplanations. But, one prepollination effect is also possible. Sincepollen was collected from equal numbers of flowers from each donor,differences in pollen production per flower could, in part, explain thedifferences in pollen donor performance. However, a preliminary analysisof pollen production suggests that this was not the case. Pollen wascounted from three flowers of each of these pollen donors, collected ona single day. Pollen was suspended in 2% saline and counted usingan Elzone 180 particle counter. Mean pollen grain number was 68 000, 124000, 124 000 and 193 000, for donors A-D, respectively (T. Bennett andD. L. Marshall, unpublished data). Rank for pollen production, fromhighest to lowest, was D, C = B, and A, while overall rank forseeds sired was C, D, B, and A. More importantly, seed paternity wasstill nonrandom when I adjusted for differences in pollen production perflower (Table 8). Sincepollen production can only change among donors (the identity of theintended seed parent cannot change pollen production on the pollenparent), this variable cannot have altered the significance of male byfemale interactions. In addition, differences in pollen production couldnot have explained the patterns of pollen donor performance in at leastone previous experiment (Marshall,1991). In that experiment, rank order of pollen donorperformance was not the same across all two-donor to six-donor mixedpollinations.
|
The results of the second experiment are somewhat less clear. Thegoal was to disrupt the self-incompatibility system and ask whethernonrandom mating among compatible mates could still occur. Whileself-incompatiblity was circumvented to some degree, I was not able tocompletely disrupt the incompatibility system by pollinating immaturestigmas. Self pollen was still at a disadvantage. The results of thisexperiment were consistent with the first in that, where mating wassignificantly nonrandom on mature stigmas, it was also significantlynonrandom on immature stigmas. So, disrupting the incompatibility systemdid not completely eliminate nonrandom mating. However, the amount ofsorting among mates was less on immature than on mature stigmas. Thismay indicate that self-incompatibility and sorting among compatiblemates share mechanisms, both of which were disrupted by pollination ofimmature flowers. Alternatively, the mechanisms of sorting self fromnon-self pollen and sorting among compatible donors might both besomewhat undeveloped in immature flowers. And, although testing thiswas not my goal here, Lyons et al.(1989) point out that this kind of reduction in sorting amongmates on "disabled" plants suggests a role of the maternaltissue in sorting among mates.
The results of experiment 2 are consistent with work on Petuniahybrida (Cruzan, 1993).Pollination of immature flowers of P. hybrida resulted in lesspollen tube attrition in the style than when mature flowers werepollinated. Although, only single pollinations were studied, the resultssuggest that the mechanisms that might result in sorting among pollendonors were less active in the immature flowers.
Using other means to disrupt incompatibility would allow betterexploration of the similarities in mechanisms of the various levels ofnonrandom mating. It is possible, for example, to achieveself-pollination by increasing carbon dioxide concentration (Nakanishi and Hinata, 1975) duringpollination and pollen tube growth. And, it may be possible to washsurface recognition molecules from the pollen coat (Roggen, 1974) or heat pollen (Matsubara, 1984) to achieve self-pollination.By circumventing different parts of the incompatibility system, it maybe possible to detect where sorting among compatible mates has uniquemechanisms or developmental timing.
These experiments demonstrate that sorting among compatible donors ofwild radish cannot simply be a subtle effect of the incompatibilitysystem. There is still sorting among compatible mates when effects ofthe S-allele system are factored out. However, there are other levels ofsorting that are also based on specific maternal plant by pollen donorinteractions that are not explicitly ruled out by this experiment. Ifthese interactions fully explain mating patterns, then sexual selectionis unlikely.
For example, more inbred plants might produce poorer pollen. If donorA was the most inbred and donor C the least, that might explain thepattern of seed paternity. That seems an unlikely explanation however.In order to find sufficient S-alleles, the original screening populationcontained plants from two populations. By chance, all of the lineagesselected to produce the four pollen donors had one parent from eachpopulation. Thus, the pollen donors could not be suffering frominbreeding depression. They might suffer from outbreeding depression,but they should all do so to the same degree. In previous experiments,whether donors were from the same or different population as maternalplants had no predictive value for paternal success (Marshall, 1991).
Discrimination against very closely or distantly related mates canalso occur during mating (Waser and Price,1994). However, it is unlikely that this form of sorting couldhave produced our results. For example, sorting against too distantlyrelated pollen donors would require that, by chance, all the maternalfamilies were most closely related to donor C and least closely relatedto donor A. The probability that all families would have C as the mostclosely related donor is (1/4) or 0.004. The probabilitythat donor A is also the most distantly related to all four families is0.000016. The probabilities would be the same for the sorting to bebased on A being too closely related to all of the maternal families.Therefore, the nonrandom mating in wild radish that I observed cannot beexplained simply by maternal plant–pollen donor interaction. Thekinds of mating events that can result in sexual selection can occur inwild radish. However, exactly which traits are under selection is notyet clear.
Demonstrating that pollen donor performance can be consistent acrossmaternal plants is essential to ascertaining whether sexual selectioncan occur. However, it is not the only criterion for sexual selection tobe important in determining plant characters. While selection acts onthe phenotype, for character change to occur, there must be a geneticbasis for those phenotypes. In this experiment, I do not know whetherthe differences in pollen donor performance had a genetic basis.Environmental effects on the performance of pollen from wild radish(Young and Stanton, 1990) and otherspecies (Lau and Stephenson, 1993,1994; Quesada,Bollman, and Stephenson, 1995) are known. All the pollendonors were grown in the same greenhouse, which should minimizeenvironmental effects, but they are still a possibility. In addition,there might be genetically based, but nonheritable variation among thepollen donors. That is, due to the wide crosses used to produce theplants necessary for this experiment, the pollen donors might differ ingenetic load. This could produce differences in mating ability in thepresent generation that would not necessarily affect mating performanceof progeny. Thus, other kinds of experiments are required to test forthe genetic basis of mating ability.
Likewise, demonstrating a process in the greenhouse does not provethat it occurs in the field. In fact, since the first experiment used 16fully intercompatible plants, these results might seem particularlyinapplicable in the field. However, the events that occurred during thisexperiment do not require such precise conditions. I used plants knownto have completely different S-alleles to factor out any subtle effectsof incompatibility. However, for sorting among compatible mates to occurin the field, only the application of pollen from two compatible donorsis required. This must occur since there is a high frequency of multiplepaternity in the field (Ellstrand andMarshall, 1986), this multiple paternity is likely due to thedeposition of mixed pollen loads (Marshall andEllstrand, 1985), and, with very rare exceptions, onlycompatible donors sire seeds.
Since the pollen donors used in this experiment were all the productsof mating between two populations, one might argue that the kind ofvariability among pollen donors seen in this experiment might not occurin natural populations. While I cannot rule out that possibility, itseems unlikely to me since the differences among pollen donors inseed-siring ability seen here were no greater than those found amongsets of pollen donors taken from the same populations (Marshall and Ellstrand, 1986, 1988).
The environment of the field is different and might also obscuresorting. However, we have shown that, given mixed pollen loads, matingis nonrandom on maternal plants grown in the field (Marshall and Fuller, 1994), and stress did notalter the rank order of pollen donor performance in this experiment.However, these studies do not account for the actions of pollinators(Stanton et al., 1992). Thus, thisstudy documents one, but not all of the necessary conditions, for sexualselection to be important inplants.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bertin, R. I., and M.Sullivan. 1988 Pollen interference and crypticself-fertility in Campsis radicans. American Journal of Botany75: 1140–1147. [CrossRef][ISI]
Campbell, B. 1972 .Sexual selection and the descent of man 1871–1971. Aldine,Chicago, IL.
Caswell, H. 1989 Matrixpopulation models. Sinauer, Sunderland, MA.
Charlesworth, D., and B.charlesworth. 1992 The effects of genetic load onselection in the gametophyte stage. In E. Ottaviano, D. L.Mulcahy, M. Sari-Gorla, and G. Bergammini-Mulcahy [eds.], Angiosperm pollen and ovules, 401–407. Springer Verlag, New York,NY.
———, D. W. Schemske,and V. L. Sork. 1987 The evolution of plantreproductive characters: sexual vs. natural selection. In S. C.Stearns [ed.], Evolution of sex, 317–335. Birkhauser,Basel.
Charnov, E.L. 1979 Simultaneous hermaphroditism and sexualselection. Proceedings of the National Academy of Sciences, USA76: 2480–2484.
Cruzan, M.B. 1993 Analysisof polle-style interactions in Petunia hybrida; the determination of variance in male reproductive success. Sexual PlantReproduction 6: 275–281.
Darwin, C. 1871 The descentof man and selection in relation to sex. Murray, London.
de Nettencourt,D. 1977 Incompatibility in angiosperms.Springer-Verlag, Berlin.
Ellstrand, N.C. 1984 Multiple paternity within the fruits of thewild radish, Raphanus sativus. American Naturalist 123:819–828.
———, B. Devlin, and D.L. Marshall. 1989 Gene flow by pollen into smallpopulations: data from experimental and natural stands of wild radish. Proceedings of the National Academy of Sciences, USA 86:9044–9047.
———, and D. L.Marshall. 1986 Patterns of multiple paternity inpopulations of Raphanus sativus. Evolution 40:837–842.
Grant, V. 1995 Sexualselection in plants: pros and cons. Proceedings of the National Academy of Sciences, USA 92: 1247–1250.
Haig, D., and M.Westoby. 1988 Inclusive fitness, seed resources, andmaternal care. In J. and L. Lovett Doust [eds.],Plant reproductive strategies, 60–79. Oxford University Press,New York, NY.
———, and———. 1989 Parent-specific gene expressionand the triploid endosperm. American Naturalist 134:147–155.[CrossRef][ISI]
Hinata, K., and T.Nishio. 1980 Self-incompatibility in crucifers. In S. Tsunoda, K. Hinata, and C. Gomez-Campo [eds.],Brassica crops and wild allies: biology and breeding, 223–234.Japanese Scientific Societies, Tokyo.
Holsinger, K.E. 1992 Ecological models of plant mating systems andthe evolutionary stability of mixed mating systems. In R.Wyatt [ed.], Ecology and evolution of plant reproduction,169–191. Chapman and Hall, New York, NY.
Horvitz, C. C., and D. W.Schemske. 1995 Spatiotemporal variation in demographic transitions of a tropical understorey herb: projection matrix analysis.Ecological Monographs 65: 155–192. [CrossRef]
Karron, J. D., D. L. Marshall, and D. M.Oliveras. 1990 Numbers of sporophyticself-incompatibility alleles in populations of wild radish. Theoretical and Applied Genetics 79: 457–460. [ISI]
Lau, T. C., and A. G.Stephenson. 1993 Effects of soil nitrogen on pollenproduction, pollen grain size, and pollen performance in Curcurbitapepo (Curcurbitaceae). American Journal of Botany 80:763–768.[CrossRef][ISI]
———, and———. 1994 Effects of soil phosphorus onpollen production, pollen size, pollen phosphorus content and theability to sire seeds in Cucurbita pepo (Cucurbitaceae). Sexual Plant Reproduction 7: 215–220. [ISI]
Lyons, E. E., N. M. Waser, M. V. Price, J.Antonovics, and A. F. Motten. 1989 Sources of variationin plant reproductive success and implications for concepts of sexualselection. American Naturalist 134: 409–433. [CrossRef][ISI]
Marshall, D.L. 1991 Nonrandom mating in wild radish: variation inpollen donor success and effects of multiple paternity among one- tosix-donor pollinations. American Journal of Botany 78:1404–1418.[CrossRef][ISI]
———, and N. C.Ellstrand. 1985 Proximal causes of multiple paternityin wild radish, Raphanus sativus. American Naturalist 126:596–605.
———, and———. 1986 Sexual selection in Raphanussativus: experimental data on non-random fertilization, maternalchoice, and consequences of multiple paternity. AmericanNaturalist 127: 446–461. [CrossRef][ISI]
———, and———. 1988 Effective mate choice in wildradish: evidence for selective seed abortion and its mechanism.American Naturalist 131: 736–759.
———, and M. W.Folsom. 1991 Mate choice in plants: an anatomical topopulation perspective. Annual Review of Ecology andSystematics 22: 37–63. [CrossRef][ISI]
———, and———. 1992 Mechanisms of non-randommating in wild radish. In R. Wyatt [ed.], Ecologyand evolution of plant reproduction: new approaches, 91–118.Chapman and Hall, New York, NY.
———, and O. S.Fuller. 1994 Does nonrandom mating among wild radishplants occur in the field as well as in the greenhouse? AmericanJournal of Botany 81: 439–445. [CrossRef][ISI]
———,———, C. Hatfield, and T.Bennett. 1996 Does interference competition among pollen grains occur in wild radish? Evolution 50:1842–1848.[CrossRef][ISI]
Matsubara,S. 1984 Overcoming the self-incompatibility ofRaphanus sativus by application of plant hormones, aminoacids, and vitamins, and by temperature treatment of pollen.Euphytica 33: 113–121.
Mulcahy, D.L. 1974 Correlation between speed of pollen tube growthand seedling height in Zea mays L. Nature 249:491–493.[CrossRef][ISI]
———. 1979 The riseof the angiosperms: a genecological factor. Science 206:20–23.
Nakanishi, T., and K.Hinata. 1975 Self-seed production by CO2 gastreatment in self-incompatible cabbage. Euphytica 24:117–120.[CrossRef][ISI]
Nason, J. D., and N. C.Ellstrand. 1995 Lifetime estimates of biparentalinbreeding depression in the self-incompatible annual plant Raphanussativus. Evolution 49: 307–316.
Pundir, N. S., R. F. Abbas, and A. A. A.Al-Attar. 1983 Pollen germination and pollen tubegrowth in Raphanus sativus L. following self- andcross-pollinations. Phyton 43: 127–130[ISI]
Queller, D. C. 1983 Sexualselection in a hermaphroditic plant. Nature 305:706–707.[CrossRef]
———. 1987 Sexualselection in flowering plants. In J. W. Bradbury and M. B.Andersson [eds.], Sexual selection: testing the alternatives,165–181. Wiley, New York, NY.
Quesada, M., K. Bollman, and A. G.Stephenson. 1995 Leaf damage decreases pollen production and hinders pollen performance in Cucurbita texana.Ecology 76: 437–443.
Richards, A.J. 1986 Plant breeding systems. George Allen and Unwin,London.
Rigney, L.P. 1995 Postfertilization causes of differentialsuccess of pollen donors in Erythronium grandiflorum(Liliaceae): nonrandom ovule abortion. American Journal ofBotany 82: 578–584. [CrossRef][ISI]
Roggen, H. 1974 Pollenwashing influences (in)compatibility in Brassica oleraceavarieties. In H. F. Linskens [ed.], Fertilization inhigher plants, 273–278. North-Holland, Amsterdam.
Sampson, D. R. 1957 Thegenetics of self-incompatibility in the radish. Journal ofHeredity 48: 26–29.
Searcy, W. A. 1982 Theevolutionary effects of mate selection. Annual Review of Ecology andSystematics 13: 57–85. [CrossRef][ISI]
Seavey, S. R., and K. S.Bawa. 1986 Late acting self-incompatibility inangiosperms. Botanical Review 52: 195–219.
Snow, A.A. 1994 Postpollination selection and male fitness inplants. American Naturalist 144: S69–S83. [CrossRef][ISI]
———, and T. P.Spira. 1991a Differential pollen-tube growth rates andnonrandom fertilization in Hibiscus moscheutos (Malvaceae).American Journal of Botany 78: 1419–1426. [CrossRef][ISI]
———, and———. 1991b Pollen vigor and the potentialfor sexual selection in plants. Nature 352:796–797.[CrossRef]
———, and———. 1996 Pollen-tube competition and malefitness in Hibiscus moscheutos. Evolution. 50:1866–1870.
Stanton, M. L., T. Ashman, L. F. Galloway,and H. J. Young. 1992 Estimating male fitness of plantsin wild populations. In R. Wyatt [ed.], Ecology andevolution of plant reproduction: new approaches, 62–90. Chapmanand Hall, New York, NY.
Stephenson, A. G. and R. I.Bertin. 1983 Male competition, female choice, andsexual selection in plants. In L. Real [ed.],Pollination biology, 109–149. Academic Press, Orlando,FL.
Waller, D. M. 1993 Thestatics and dynamics of mating system evolution. In N. Thornhill [ed.], The natural history of inbreeding andoutbreeding, 97–117. The University of Chicago Press, Chicago,IL.
Waser, N. M., and M. V.Price. 1993 Crossing distance effects on prezygotic performance in plants: and argument for female choice. Oikos68: 303–308. [CrossRef][ISI]
———, and———. 1994 Crossing distance effect inDelphinium nelsonii: outbreeding and inbreeding depression inprogeny fitness. Evolution 48: 842–852. [CrossRef][ISI]
———, ———,A. M. Montalvo, and R. N. Gray. 1987 Female matechoice in a perennial herbaceous wildflower, Delphinium nelsonii.Evolutionary Trends in Plants 1: 29–33.
Westoby, M., and B.Rice. 1982 Evolution and the fitness of plant tissues.Evolution 36: 713–724. [CrossRef][ISI]
Willson, M. F. 1979 Sexualselection in plants. American Naturalist 113:777–790.[CrossRef][ISI]
———. 1982 Sexualselection and dicliny in angiosperms. American Naturalist 119:579–583.[CrossRef][ISI]
———. 1983 Matingpatterns in plants. In R. Dirzo and J. Sarukhan[eds.], Perspectives on plant population ecology,261–276. Sinauer, Sunderland, MA.
———. 1990 Sexualselection in plants and animals. Trends in Ecology andEvolution 5: 210–214.
———, and N.Burley. 1983 Mate choice in plants. PrincetonUniversity Press, Princeton, NJ.
Young, H. J., and M. L.Stanton. 1990 Influence of environmental quality onpollen competitive ability in wild radish. Science 248:1631–1633.
This article has been cited by other articles:
|
|
 |
|
  G. Gleiser, J. G. Segarra-Moragues, J. R. Pannell, and M. Verdu Siring Success and Paternal Effects in Heterodichogamous Acer opalus Ann. Bot., May 1, 2008; 101(7): 1017 - 1026. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. A. Koelling and K. Karoly Self-pollen interference is absent in wild radish (Raphanus raphanistrum, Brassicaceae), a species with sporophytic self-incompatibility Am. J. Botany, May 1, 2007; 94(5): 896 - 900. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. L. Marshall, J. Reynolds, N. J. Abrahamson, H. L. Simpson, M. G. Barnes, J. S. Medeiros, S. Walsh, D. M. Oliveras, and J. J. Avritt Do differences in plant and flower age change mating patterns and alter offspring fitness in Raphanus sativus (Brassicaceae)? Am. J. Botany, March 1, 2007; 94(3): 409 - 418. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. G. M. Shaner and D. L. Marshall How robust is nonrandom mating in wild radish: do small pollen loads coupled with more competing pollen donors lead to random mating? Am. J. Botany, February 1, 2007; 94(2): 266 - 273. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. E. Weis and T. M. Kossler Genetic variation in flowering time induces phenological assortative mating: quantitative genetic methods applied to Brassica rapa Am. J. Botany, June 1, 2004; 91(6): 825 - 836. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Bernasconi, T.-L. Ashman, T. R. Birkhead, J. D. D. Bishop, U. Grossniklaus, E. Kubli, D. L. Marshall, B. Schmid, I. Skogsmyr, R. R. Snook, et al. Evolutionary Ecology of the Prezygotic Stage Science, February 13, 2004; 303(5660): 971 - 975. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. G. M. Shaner and D. L. Marshall Under how wide a set of conditions will nonrandom mating occur in Raphanus sativus (Brassicaceae)? Am. J. Botany, November 1, 2003; 90(11): 1604 - 1611. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. M. Herrera Censusing natural microgametophyte populations: variable spatial mosaics and extreme fine-graininess in winter-flowering Helleborus foetidus (Ranunculaceae) Am. J. Botany, October 1, 2002; 89(10): 1570 - 1578. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. L. Marshall and D. M. Oliveras Does differential seed siring success change over time or with pollination history in wild radish, Raphanus |
