Effects of foliar herbivory on male and female reproductive traits of wild radish, Raphanus raphanistrum.
Foliar herbivory can affect plant fitness directly through changes in plant traits, and indirectly by altering pollinator visitation patterns to damaged plants (Strauss et al. 1996, Strauss 1997, Lehtila and Strauss 1997). Floral traits, as well as leaf characters, can be affected by herbivory (Frazee and Marquis 1994, Strauss et al. 1996), including traits specifically associated with male function (pollen characters) and with female function (ovule characters). Traits such as petal size may influence both male and female plant fitness by affecting rates of pollinator visitation. Although herbivory is known to reduce flower, fruit, and seed production in many cases (for review, see Hendrix 1988, Marquis 1992), the effects of leaf damage on male reproductive traits have generally received less attention (but see Quesada et al. 1995, Mutikainen and Delph 1996, Strauss et al. 1996). To date, fitness effects of herbivory have almost exclusively been estimated by measuring solely female reproductive output, because a good estimate of seeds sired through pollen can only be made by tedious paternity analysis with allozymes or DNA markers, and only small populations can be dealt with in this manner. Gronemeyer et al. (1997) present the only consideration of effects of herbivory on male plant fitness. Both male and female functions contribute equally to alleles represented in the next generation; hence, both should be taken into account when estimating the effects of herbivory on plant fitness.
Most papers that explore how herbivory affects male and female reproductive effort consider species with imperfect flowers or those that are dioecious (e.g., Hendrix and Trapp 1981, 1989, Allison 1990, Delph 1990, Snyder 1993). Less is known about how herbivory affects allocation to male and female function in hermaphrodites (but see McKone 1989, Frazee and Marquis 1994, Strauss et al. 1996). The value of studying allocation patterns to the different sexes in hermaphrodites is that "decisions" about how male and female function receive resources are being made simultaneously for both sexes. The degree to which male and female floral traits can recover from, or compensate for, herbivory, and the rate at which each trait is able to do so, may be important in determining the net effects of herbivory on plant fitness.
The aim of this study is to determine how foliar herbivory affects floral reproductive traits (petal size, male and female gamete production, nectar production), as well as seed and fruit production, in the hermaphrodite wild radish, Raphanus raphanistrum. Specifically, we address the following questions: (1) Does foliar herbivory affect male and female functions differentially? (2) Do all traits recover to the same extent, and at similar rates, from the effects of herbivory?
MATERIAL AND METHODS
Wild radish, Raphanus raphanistrum L. (Brassicaceae), is an introduced, annual, self-incompatible plant of roadsides and fields in North America. In addition to the Pieris rapae larvae used as folivores in this study, its herbivores include other lepidopteran larvae, flea beetles, aphids, thrips, and mammals (K. Lehtila and S. Y. Strauss, personal observation); many of these (including P. rapae) are introduced, and are natural herbivores of wild radish in both its novel and native habitats.
To examine the effects of herbivory on male and female floral traits, we conducted parallel experiments in both a greenhouse and a growth chamber. The separate growth chamber experiment was done because of our concern that greenhouse pests might differentially affect some floral characters (e.g., western flower thrips consume pollen, but not ovules). Our fears of outbreaks were unfounded, however, and these two settings thus served as two separate environments in which to test effects of herbivory on floral traits. The two experiments had a similar general design, but differed in the number of treatment levels and families used and in sampling methods used to estimate total seed production. Plants used in these experiments were full-sib progeny produced by controlled crosses of field-collected plants from upstate New York.
In the greenhouse experiment, one member of each of 28 full-sib families was randomly assigned to one of the following herbivory treatments: undamaged control (U); removal of one-quarter of the leaf area from every leaf over the lifetime of the plant (Q); removal of half of the first four (rosette) leaves (D); or removal of one-half of every leaf over the lifetime of the plant (H). The (D) treatment corresponded to [approximately]20% of total leaf area removal, and was the type of damage imposed in several other previously published experiments (Strauss et al. 1996, Lehtila and Strauss 1997). We used families as a sample of the plant population as a whole, without replication of families within treatments. Use of full-sib families allowed us to remove variation in floral characters arising from genetic (Conner and Via 1993) and possible environmental maternal effects. Plants were kept in a 16-h day/8-h night light cycle.
For both experiments, plants were grown in 10-cm pots using Metro Mix 360 soil (Hummert International, St. Louis, Missouri, USA) and were fertilized once with 0.3-g 17-9-13 Osmocote microfertilizer (Grace-Sierra, Milpitas, California, USA). The position of plants within the greenhouse or growth chamber was randomized at least four times weekly throughout the experiments.
In the growth chamber, seeds from 20 families (16 of the 28 families that were represented in the greenhouse experiment) were sown on 12 December 1995. One plant from each family was randomly assigned to one of three herbivory treatments: control (U), D, and H, for a total of 60 plants. The Q treatment was not included in this experiment because of space limitations. Plants were kept in a 12-h day/12-h night light cycle. Light intensity in the growth chamber was 700-800 mol [multiplied by] [m.sup.-2] [multiplied by] [s.sup.-1], which is saturating irradiance for Raphanus sativus (Combe et al. 1988). Temperature was 27 [degrees] C during the day and 17 [degrees] C at night.
We used Pieris rapae caterpillars to damage leaves at the appropriate levels. Caterpillars were kept inside clip cages, which consisted of two 35-mm plastic petri dish bottoms attached to either side of a bent hairclip mounted on a wooden stake. A single larva was enclosed within each cage, which was subsequently placed on one side of the leaf and moved (alongside the midrib) as the larva consumed foliage. Caterpillars and cages were removed after the amount of leaf area designated for each treatment was consumed. The extent and pattern of damage are within the natural range, based on field observations from gardens in three locations in central Illinois. In these gardens, comprised of full siblings of our experimental plants, [approximately]10% of the plants sustained levels of damage as high or higher than that of our most extreme treatment, and [approximately]35% of the population sustained levels as high or higher than that of our intermediate treatment (S. Y. Strauss and K. P. Lehtila, unpublished data).
When plants began to flower, we collected a series of target flowers throughout the lifetime of each plant. For each flower, we measured the following traits: flower size, pollen grain size and number, and ovule size and number. Flower size was estimated by measuring the length and width of two adjacent petals. We then multiplied mean petal length x width to create a single measure of petal size, which we call "petal size index." In a subset of these flowers, we also measured nectar production with a 1-[[micro]liter] micropipette and its sugar content, using a portable refractometer. For pollen collection, we removed half of the anthers from each flower: two from adjacent long stamens and one from a short stamen. Anther collections were made before anthers dehisced. Number and modal size of pollen grains were measured using an Elzone particle counter (Particle Data Inc., Elmhurst, Illinois, USA). For methods, see Young and Stanton (1990a). Carpels were collected and stored in 70% alcohol, and were later dissected under a microscope to count the total number of ovules and to measure the width and length of the two middle ovules. An ovule size index was created by multiplying the mean length and width of the two ovules. All other open flowers were hand-pollinated every other day to allow seed set.
Target flowers for petal measurements and ovule and anther collections were: numbers 3, 10, 20, 30, 40, 50, 75, 100, 150, and 200 in the series from each plant. Nectar was measured from flowers 3, 10, 50, and 200. Because some of our measurements were destructive, we used two adjacent flowers per target sample so that we could measure all traits simultaneously. For example, when the target flower was the third flower, we used flower 3 for petal and nectar measurements and flower 4 for pollen and ovule collections. The first flower number of the pair is used as a data label (e.g., flower 3 indicates data from the third and fourth flower, etc.).
After plants finished flowering, watering was ceased and fruits were collected as they turned yellow. All fruits were counted and seed numbers were estimated from the number of expanded fruit segments. In both experiments, we also measured date of first and last flower for each plant, total flower production, total numbers of fruits and seeds, and mean seed and fruit mass per plant. In the greenhouse experiment only, we labeled fruits harvested from flowers of different age. Seed counts were done by randomly selecting 20 fruits from each plant, counting their seed number, and multiplying the mean seed number by the number of fruits produced. Seeds from these fruits were removed and weighed to get mean seed mass per plant. Vegetative size at the time of harvest was estimated by measuring the height of plants and the diameter at the stem base.
Previous studies had indicated that effects of leaf damage on corolla characters persisted to the 40th-50th flower in plants receiving the D treatment (Lehtila and Strauss 1997). To reduce the total number of samples processed, we subsampled the families and quantified all the traits for every target flower collected, because pollen and ovule measurements were very laborious. Exploratory data analysis and plots of differences between treatments from these samples allowed us to determine that traits of the 20th, 30th, and 40th flowers were well bracketed by the 10th and 50th flower measurements. We therefore sampled a subset of the target flowers in the midlife of the plant: complete measurements spanning the lifetime of the plants in all 32 families were made on flowers 3, 10, 50, 75, 100, and 200.
All analyses were done using SAS procedure GLM with Type III sums of squares (SAS 1989). Assumptions of homogeneity of variance and normality were checked by visual inspection of standard deviations of different treatments, by normal probability plots of residuals, and by the Shapiro-Wilk test (Shapiro and Wilk 1965). Log-transformation was used for nectar volume to satisfy normality and homoscedasticity assumptions.
The independent variables consisted of damage treatment, full-sib family, and, when both experiments were included in the same analysis, experiment. In cases in which the experiments differed in their results, separate analyses are presented. Otherwise, analyses of both experiments together are presented for brevity.
Because floral characters were sampled repeatedly from the same plants, we used multivariate repeated-measures ANOVAs, with cumulative flower number as the time factor. Dependent variables were: petal size index, pollen number and size, and ovule number and size index. Family and experiment were used as blocking factors. Linear trends between dependent variables and log-transformed flower number were tested with polynomial contrasts. Significant differences between treatments in these contrasts indicate that the rate at which characters change through time differs by treatment.
Traits measured only a single time (date of the first flower, average number of flowers produced/day, total flower production, plant size, and fruit and seed set) were analyzed with two- or three-way MANOVAs with damage as a treatment factor and family, or both family and experiment, as blocking factors.
A priori multiple contrasts were used to test whether the most severe damage treatment (H) and the D treatment differed from the undamaged control (U). In addition, for the greenhouse experiment, we did a separate comparison of the Q vs. D treatments to determine whether plants responded differently to continuous 25% leaf area removal (Q) than to early, concentrated damage that removed a similar amount of total leaf area (D). Because contrasts were non-orthogonal, we used a Dunn-Sidak adjustment of probabilities ([Alpha] = 0.05) to control for multiple comparisons (Day and Quinn 1989).
Effects of herbivory on floral traits: overall effects of herbivory and time
Multivariate repeated-measures ANOVA for petal size, pollen number, pollen size, number of ovules per flower, and ovule size indicated significant differences in all factors of the design (P [less than] 0.001) for all factors except for the between-subjects contrast D vs. U (adjusted P = 0.21) and a within-subjects interaction of time x family (P = 0.07).
Effects of herbivory on male fitness traits: petal size, pollen number, and pollen size
Subsequent univariate tests showed that leaf damage significantly decreased petal size, pollen number, and pollen size (Table 1, [ILLUSTRATION FOR FIGURE 1a-c OMITTED]). There was a significant effect of time, or plant age, on all characters (Table 1). Polynomial contrasts indicated that there was generally a decreasing trend over time, such that most traits diminished with plant age, regardless of damage level [TABULAR DATA FOR TABLE 1 OMITTED] (Table 1, [ILLUSTRATION FOR FIGURE 1 OMITTED]). An exception to this pattern was that there was no decrease in pollen size in the greenhouse experiment ([F.sub.1,70] = 0.01, NS; [ILLUSTRATION FOR FIGURE 1 OMITTED]).
Contrasts allowed us to determine which treatments were responsible for significant damage effects. Mean petal size averaged over plant lifetime was significantly smaller in the most severe damage treatment H (removal of half of every leaf) than in controls (Table 1, [ILLUSTRATION FOR FIGURE 1 OMITTED]). In the greenhouse experiment, overall petal sizes in the D and Q (25% of leaf area removed from each leaf) treatments did not differ significantly ([F.sub.1,76] = 0.41, P = 0.52; [ILLUSTRATION FOR FIGURE 1 OMITTED]); thus, more intensive early damage had the same effect on petal size as a similar amount of total damage spread continuously, at lower levels, over the lifetime of the plant.
The effect of damage on petal size decreased linearly over time in both H vs. U and D vs. U contrasts, and there was also a significant interaction (H x U x time and D x U x time contrasts; Table 1, [ILLUSTRATION FOR FIGURE 1a OMITTED]), meaning that petal size decreased at different rates for the H and D treatments.
The number of pollen grains produced per flower was significantly greater in the undamaged control treatment (U) than in D and H treatments (Table 1, [ILLUSTRATION FOR FIGURE 1b OMITTED]). In addition, heavier early damage (D) significantly ([F.sub.1,71] = 6.4, P = 0.013) reduced pollen production relative to continuous lower-level damage (Q). The damage x time interaction was not significant in any of these contrasts (Table 1).
Pollen grains produced by plants in the H treatment were also significantly smaller than those produced by undamaged (U) plants (Table 1, [ILLUSTRATION FOR FIGURE 1c OMITTED]). Linear contrasts of time x D vs. U, and time x H vs. U were significant (Table 1). Again, this result indicates that pollen grain size decreased at different rates with treatment (i.e., was steeper in the U treatment than in the damage treatments). In the comparison of D and Q treatments, although plants in the Q treatments produced more pollen than those in the D treatment, we found that D plants produced larger pollen grains than Q plants ([F.sub.1,70] = 4.7, P = 0.03).
Effects of herbivory on female traits: ovule size and number, fruit and seed production
Leaf damage did not have a significant effect on any female floral traits, including both ovule number and ovule size (Table 1, [ILLUSTRATION FOR FIGURE 1d, e OMITTED]). These traits generally showed a significant decrease with time, except for ovule size, which did not decrease over time in the growth chamber experiment ([F.sub.1,32] = 0.92, P = 0.34).
Results for fruit and seed production differed slightly between the growth chamber and the greenhouse. In the growth chamber, herbivory significantly reduced seed production in the H treatment compared to undamaged controls. (Table 2). In the greenhouse, there were no significant differences in fruit number, seed number, or fruit and seed mass between the treatments (Table 2).
[TABULAR DATA FOR TABLE 2 OMITTED]
There were no significant effects of damage treatment or damage x time interactions on nectar volume or sugar concentration (repeated-measures ANOVA, damage P [greater than] 0.30 for nectar volume and sugar concentration in both experiments). Both characters showed significant linear trends that differed between the experiments. There was a negative trend over time in the volume of nectar produced by plants in the greenhouse experiment (repeated-measures ANOVA, linear contrast, [F.sub.1,79] = 26, P [less than] 0.001), and a positive trend through time in sugar concentration ([F.sub.1,78] - 17, P [less than] 0.001). In contrast, there was a positive trend in nectar volume in the growth chamber experiment ([F.sub.1,35] = 52, P [less than] 0.001), and a negative trend in sugar concentration ([F.sub.1,28] = 5.8, P [less than] 0.02). These results may represent different humidity conditions and/or day length differences in the greenhouse vs. the growth chamber.
Effects of herbivory on other life history attributes: flowering phenology, total number of flowers, and final plant size
Damage did not significantly affect time to production of the first flower, rate of flower production, or the total number of flowers produced (Wilks' [Lambda] = 0.941, [F.sub.6,210] = 1.1, P = 0.37). Because the MANOVA was not significant, we do not present any of the univariate analyses for these traits. Neither plant height nor base diameter differed between treatments (base diameter, F = 0.14, P = 0.93; height, F = 0.91, P = 0.44).
In this study, damage to leaves by caterpillars affected floral traits associated with male fitness more than those with female fitness. Foliar herbivory significantly reduced pollen number, pollen size, and corolla size (see also Strauss et al. 1996, Lehtila and Strauss 1997). Corolla size is a trait we treat as more closely aligned with male fitness for the following reasons: (1) it affects pollinator visitation in many species (Stanton and Preston 1988, Campbell et al. 1991, Eckhart 1991, Stanton et al. 1991), including Raphanus raphanistrum (Conner and Rush 1996); and (2) attractiveness to pollinators is thought to have a larger effect on male than on female fitness in R. raphanistrum, because seed set is usually not limited by pollen and many more pollinator visits are needed to remove all pollen than to fertilize all the ovules (Stanton et al. 1986, Young and Stanton 1990b, Delph and Lively 1992, Rush et al. 1995, but see Wilson et al. 1994). Thus, all traits that we considered to be associated with male plant fitness were negatively affected by herbivory. The magnitude of the herbivory effect was roughly equivalent to differences in floral traits that spanned the lifetime of the plant [ILLUSTRATION FOR FIGURE 1 OMITTED]; for example, early flowers of damaged plants had traits similar in size to those produced much later (100-200+) by undamaged plants. Thus, effects of damage were comparable to the largest effects of aging on traits.
In contrast, female fitness components of ovule number and size were not affected by leaf damage. This result is striking in that, because flowers are hermaphroditic, allocation to male and female function is occurring essentially simultaneously and, therefore, plants may have experienced selection to invest in female traits over male traits. Our finding that herbivory affects male function in floral traits more than female function is consistent with those of several other studies. In Chamaecrista fasciculata (Fabaceae), another hermaphrodite, both anther length and ovule diameter decreased after leaf clipping treatments, but ovule number was not affected (Frazee and Marquis 1994). In addition, ovule size was only 3.7% smaller, relative to controls, in both 25% and 50% leaf area removal treatments, whereas anther length differences translated into 7.5% and 10.3% fewer pollen grains, respectively, in the same treatments. These percentages for ovule size and pollen number are not necessarily comparable in terms of resources invested, but percentages could reflect changes in relative allocation. In plants with imperfect flowers, Allison (1990) reported a significant decrease in production of male strobili of Taxus canadensis after simulated deer browsing, whereas female strobilus production was much less affected. On the other hand, two other studies of nonhermaphrodites have documented the reverse trend, with female flowers exhibiting larger effects of damage than male flowers (Snyder 1993, Quesada et al. 1995).
Why were petals and pollen affected by damage, whereas ovules were not? This response may have a non-adaptive or an adaptive explanation. Responses of petals and pollen may be linked by the relative timing of their development within the bud; however, this is unlikely to be the case. First, leaf damage started prior to initiation of any floral primordia or reproductive characters. Second, although petal size and pollen number are positively correlated in R. sativus (Stanton and Preston 1988) and R. raphanistrum (Strauss and Lehtila, unpublished data), this linkage can be removed by relatively few selective episodes (Stanton and Young 1994). Third, histological examination of the timing of petal, pollen, and ovule development in another crucifer, Arabidopsis, showed that pollen grains mature first, then ovules, and petal elongation lags behind both of these traits (Bowman 1994). Finally, the rate of convergence in size between traits on damaged and undamaged plants differed for petal size and pollen number (Table 2), suggesting that petal and pollen responses are not closely developmentally tied.
Irrespective of possible developmental linkages, ovules may simply be less plastic than other floral traits, due to other constraints; however, ovule number and size can change considerably when plants grow older (Young and Stanton 1990a, Frazee and Marquis 1994) and, thus, are not completely invariant. In addition, ovule number and size varied between our two experimental sites (greenhouse vs. growth chamber), so other environmental conditions did significantly influence these traits, whereas herbivory treatments did not. In general, allocation of resources to male and female reproductive success appears to vary with different environmental stresses. Mazer (1992) has shown that Raphanus sativus plants grown under high densities produced fewer ovules and smaller seeds and showed a tendency for less pollen production per flower, although the latter was not significant. In recent work, costs of induction of secondary compounds caused by leaf damage to R. raphanistrum were expressed only in pollen characters, whereas costs of leaf area removal per se, in the absence of induction, were expressed in seed number (A. A. Agrawal, S. Y. Strauss, and M. J. Stout, unpublished manuscript).
A possible adaptive explanation for the lack of response of ovules to damage is that plants maximize their total fitness by allocating energy to female reproduction when resources are decreased by herbivory. The link from pollen traits to offspring produced by pollen may be weaker than the link between ovule traits and seed production (Wilson et al. 1994). Successful male reproduction of an insect-pollinated plant depends on the ability of a plant to attract pollinators, the proper deposition of pollen grains onto pollinators, the placement of pollen by pollinators onto receptive stigmas of a compatible plant genotype, successful pollen tube competition, and subsequent avoidance of seed abortion. These steps increase variation in the causal link between pollen number and size and numbers of offspring produced by pollen. If ovule size and number have a more direct link to the number of offspring produced than do the same traits in pollen, then female floral traits should be conserved more than male traits (Wilson et al. 1994).
Female function was generally conserved in seed production and in floral traits across all our experimental treatments, with one exception: plants in the growth chamber with the heaviest herbivory levels produced fewer seeds than did controls. They also had smaller flowers and produced fewer flowers and fruits than did plants in the greenhouse experiment. We interpret these patterns to indicate that the growth chamber was a harsher environment than the greenhouse, possibly because of the reduced amount of light and/ or different light quality. Plants could completely compensate for damage in seed production in a better environment, the greenhouse (see also Maschinski and Whitham 1989). Plants in both of our experiments, however, had lifetime flower and fruit production as large or larger than that of conspecific plants in field experiments (e.g., Stanton 1984, Snow 1990), so resource levels in our experiments are probably not stressful relative to field conditions.
One possible explanation for why we found full compensation for herbivory in fruit and seed production (i.e., why there were no differences between damaged and undamaged plants) is that seeds produced by early flowers of undamaged plants may have competed for resources for later flowering and fruiting. There were, however, no significant differences in the number of seeds produced per fruit or number of fruits per flower of early flowers (flower number [less than] 50) among damage treatments (Table 2).
Damage may have a greater impact on floral characters than on fruiting characters because fruit traits are expressed later in the plant lifetime and, therefore, have a longer period over which plants may compensate for damage (Maschinski and Whitham 1989). Tolerance of herbivory expressed through seed set may be further increased through photosynthesis by fruits themselves (Bazzaz et al. 1979, Galen et al. 1989). It should be noted, however, that despite compensation later in the season in seed production, there may still be losses in male fitness due to reduced pollen production early in the season. In R. sativus, early flowers produced more fruits and seeds (Ashman et al. 1993); thus, losses in opportunities to sire seeds caused by reduced pollen production and/or size could affect overall male fitness even if damaged plants were to compensate completely for damage in seed set.
In conclusion, fitness effects of herbivory have almost exclusively been estimated by measuring female reproductive output. It is not clear whether indirect measures such as pollen number and size or pollinator visitation rates are reliable indicators of male plant fitness (Stanton et al. 1992). One implicit assumption has been that when herbivory decreases the resource base for reproduction, it affects investments in male and female functions in a similar way. Our findings, however, show that male investment may be more affected by damage than female investment. It remains to be seen whether male fitness is more affected, and we are currently undertaking experiments to determine whether changes in investment in male function translate into reduced success at siring seeds.
We thank J. Conner, P. Mutikainen, M. L. Stanton, D. DeSteven, and an anonymous reviewer, as well as members of our plant-insect discussion group: L. S. Adler, J. Thaler, A. Agrawal, M. D. Bowers, and R. Karban for their perceptive comments. The work was supported by NSF DEB 94-07362 to S. Y. Strauss, and part of it was undertaken in the Controlled Environment Facility at University of California, Davis.
Allison, T. D. 1990. The influence of deer browsing on the reproductive biology of Canada yew (Taxus canadensis marsh.). Oecologia (Berlin) 83:523-529.
Ashman, T-L., L. F. Galloway, and M. L. Stanton. 1993. Apparent vs. effective mating in an experimental population of Raphanus sativus. Oecologia 96:102-107.
Bazzaz, F. A., R. W. Carlson, and J. L. Harper. 1979. Contribution of reproductive effort by photosynthesis of flowers and fruits. Nature 279:554-555.
Bowman, J. 1994. Flowers. Pages 133-145 in J. Bowman, editor. Arabidopsis: An atlas of morphology and development. Springer-Verlag, New York, New York, USA.
Campbell, D. R., N. M. Waser, M. V. Price, E. A. Lynch, and R. J. Mitchell. 1991. Components of phenotypic selection: pollen export and flower corolla width in Ipomopsis aggregata. Evolution 45:1458-1467.
Combe, L., P. Quetin, and G. Decoux. 1988. Growth and tuberization of the radish (Raphanus sativus) as a function of light duration. Canadian Journal of Botany 66:18341840.
Conner, J. K., and S. Rush. 1996. Effects of flower size and number on pollinator visitation to wild radish, Raphanus raphanistrum. Oecologia (Berlin) 105:509-516.
Conner, J., and S. Via. 1993. Patterns of phenotypic and genetic correlations among morphological and life-history traits in wild radish, Raphanus raphanistrum. Evolution 47: 704-711.
Day, R. W., and G. P. Quinn. 1989. Comparisons of treatments after an analysis of variance in ecology. Ecological Monographs 59:433-463.
Delph, L. F. 1990. Sex-differential resource allocation patterns in the subdioecious shrub Hebe subalpina. Ecology 71:1342-1351.
Delph, L. F., and C. M. Lively. 1992. Pollinator visitation, floral display, and nectar production of the sexual morphs of a gynodioecious shrub. Oikos 63:161-170.
Eckhart, V. M. 1991. The effects of floral display on pollinator visitation vary among populations of Phacelia linearis (Hydrophyllaceae). Evolutionary Ecology 5:370-384.
Frazee, J. E., and R. J. Marquis. 1994. Environmental contribution to floral trait variation in Chamaecrista fasciculata (Fabaceae: Caesalpinoideae). American Journal of Botany 81:206-215.
Galen, C., T. E. Dawson, and M. L. Stanton. 1989. Carpels as leaves meeting the carbon cost of reproduction in an alpine buttercup. Oecologia (Berlin) 95:187-193.
Gronemeyer, P. A., B. J. Dilger, J. L. Bouzat, and K. N. Paige. 1997. The effects of herbivory on paternal fitness in scarlet gilia: better moms also make better pops. American Naturalist 150(5):592-602.
Hendrix, S. D. 1988. Herbivory and its impact on plant reproduction. Pages 246-263 in J. L. Lovett Doust and L. L. Lovett Doust, editors. Plant reproductive ecology: patterns and strategies. Oxford University Press, New York, New York, USA.
Hendrix, S. D., and E. J. Trapp. 1981. Plant-herbivore interactions: insect-induced changes in host plant sex expression and fecundity. Oecologia (Berlin) 49:119-122.
Hendrix, S. D., and E. J. Trapp. 1989. Floral herbivory in Pastinaca sativa: do compensatory responses offset reductions in fitness? Evolution 43:891-895.
Lehtila, K., and S. Y. Strauss. 1997. Leaf damage by herbivores affects attractiveness to pollinators in wild radish, Raphanus raphanistrum. Oecologia 111:396-403.
Marquis, R. J. 1992. The selective impact of herbivores. Pages 301-325 in R. S. Fritz, and E. L. Simms, editors. Plant resistance to herbivores and pathogens: ecology, evolution and genetics. University of Chicago Press, Chicago, Illinois, USA.
Maschinski, J., and T. G. Whitham. 1989. The continuum of plant responses to herbivory: the influence of plant association, nutrient availability, and timing. American Naturalist 134:1-19.
Mazer, S. J. 1992. Environmental and genetic sources of variation in floral traits and phenotypic gender in wild radish: consequences for natural selection. Pages 281-325 in R. Wyatt, editor. Ecology and evolution of plant reproduction. Chapman and Hall, New York, New York, USA.
McKone, M. J. 1989. Intraspecific variation in pollen yield in brome grass (Poaceae: Bromus). American Journal of Botany 76:231-237.
Mutikainen, P., and L. F. Delph. 1996. Effects of herbivory on male reproductive success in plants. Oikos 75:353-358.
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.
Rush, S., J. K. Conner, and P. Jennetten. 1995. The effects of natural variation in pollinator visitation on rates of pollen removal in wild radish, Raphanus raphanistrum (Brassicaceae). American Journal of Botany 82:1522-1526.
SAS Institute. 1989. SAS/STAT user's guide. Version 6, fourth edition. SAS Institute, Cary, North Carolina, USA.
Shapiro, S. S., and M. B. Wilk. 1965. An analysis of variance test for normality (complete samples). Biometrika 52:591-611.
Snow, A. A. 1990. Effects of pollen-load size and number of donors on sporophyte fitness in wild radish (Raphanus raphanistrum). American Naturalist 136:742-758.
Snyder, M. A. 1993. Interactions between Abert's squirrel and ponderosa pine: the relationship between selective herbivory and host plant fitness. American Naturalist 141:866-879.
Stanton, M. L. 1984. Seed variation in wild radish: effect of seed size on components of seedling and adult fitness. Ecology 65:1105-1112.
Stanton, M. L., T.-L. Ashman, and L. F. Galloway. 1992. Estimating male fitness of plants in natural populations. Pages 62-90 in R. Wyatt, editor. Ecology and evolution of plant reproduction. Chapman and Hall, New York, New York, USA.
Stanton, M. L., and R. E. Preston. 1988. Ecological consequences and phenotypic correlates of petal size variation in wild radish, Raphanus sativus. American Journal of Botany 75:528-539.
Stanton, M. L., A. A. Snow, and S. N. Handel. 1986. Floral evolution: attractiveness to pollinators increases male fitness. Science 232:1625-1627.
Stanton, M. L., and H. J. Young. 1994. Selecting for floral character associations in wild radish, Raphanus sativus L. Journal of Evolutionary Biology 7:271-285.
Stanton, M. L., H. J. Young, N. C. Ellstrand, and J. M. Clegg. 1991. Consequences of floral variation for male and female reproduction in experimental populations of wild radish, Raphanus sativus L. Evolution 45:268-280.
Strauss, S. Y. 1997. Floral characters link herbivores, pollinators, and plant fitness. Ecology 78:1640-1645.
Strauss, S. Y., J. K. Conner, and S. L. Rush. 1996. Foliar herbivory affects floral characters and plant attractiveness to pollinators: implications for male and female fitness. American Naturalist 147:1098-1107.
Wilson, P., J. D. Thomson, M. L. Stanton, and L. P. Rigney. 1994. Beyond floral batemania: gender biases in selection for pollination success. American Naturalist 143:283-296.
Young, H. J., and M. L. Stanton. 1990a. Temporal patterns of gamete production within individuals of Raphanus sativus (Brassicaceae). Canadian Journal of Botany 68:480-486.
Young, H. J., and M. L. Stanton. 1990b. Influences of floral variation on pollen removal and seed production in wild radish. Ecology 71:536-547.
|Printer friendly Cite/link Email Feedback|
|Author:||Lehtila, Kari; Strauss, Sharon Y.|
|Date:||Jan 1, 1999|
|Previous Article:||Episodic reproduction in two fire-prone palms, Serenoa repens and Sabal etonia (Palmae).|
|Next Article:||The consequences of floral herbivory for pollinator service to Isomeris arborea.|