effects of population size and pollen diversity on reproductive success ...

American Journal of Botany 89(8): 1250–1259. 2002.

EFFECTS

OF POPULATION SIZE AND POLLEN

DIVERSITY ON REPRODUCTIVE SUCCESS AND OFFSPRING SIZE IN THE NARROW ENDEMIC

COCHLEARIA BAVARICA (BRASSICACEAE)1 MELANIE PASCHKE,2,4 CLEMENS ABS,3

AND

BERNHARD SCHMID2

2 Institut fu¨r Umweltwissenschaften, Universita¨t Zu¨rich, Winterthurerstrasse 190, 8057 Zurich, Switzerland; and Fachgebiet Geobotanik, Technische Universita¨t Mu¨nchen, Am Hochanger 13, D-85350 Freising-Weihenstephan, Germany

3

In small, fragmented populations of self-incompatible plant species, genetic drift and increasingly close relationships between plants may restrict the number of genetically different pollen donors, the availability of compatible mates, and the opportunity for pollen competition and selection. These restrictions may reduce the siring success or increase the probability of inbreeding depression in the offspring. To test if this was the case, we hand-pollinated maternal plants in small and large populations of the rare, endemic plant Cochlearia bavarica (Brassicaceae) with pollen from one, three, or nine donors from the same population or with nine donors from a different population. In one additional population of intermediate size, maternal plants were hand-pollinated with ten donors located at a distance of 1, 10, 100, or 1000 m. We then recorded seed and offspring characters. On average, offspring from small populations were smaller than normal and fewer survived to maturity. Increasing the number of pollen donors had a positive effect on reproductive success in small and large populations, but at the highest pollen diversity this occurred at the expense of slightly reduced offspring fitness. Because the total amount of transferred pollen was held constant, these effects could not be attributed to increasing pollen load. Rather, the increasing pollen diversity may have increased the chances of selecting a particularly ‘‘good’’ donor for fertilization— an example of a sampling effect of diversity. Pollen from outside a population or from 10–100 m away resulted in higher reproductive success and greater offspring size. Effects of population size and pollination treatments on reproductive success and offspring fitness were additive. Apparently, there is no obvious size threshold above which the potential of inbreeding depression can be ignored in C. bavarica. Key words: Brassicaceae; Cochlearia bavarica; endemic plant species; inbreeding depression; Munich, Germany; pollen competition; pollen diversity; pollination distance; population size; sampling effect.

Many plant species are distributed in small, isolated populations owing to habitat fragmentation by humans (Czech, Krausman, and Devers, 2000). Small populations face higher risks of extinction because of environmental, demographic, and genetic stochasticity (Shaffer, 1987; Bond, 1996). Within a small population genetic variation decreases as a result of genetic drift (Ellstrand and Elam, 1993; Montgomery et al., 2000) and the average relatedness between individuals will increase (Falconer, 1989). This increases the probability of inbreeding and its negative effects on reproductive success and offspring fitness (Lacey, 1987; Ouborg and van Treuren, 1994; Fischer and Matthies, 1998a; Groom, 1998). In endemic species, a long history of genetic drift, isolation, and frequently occurring bottlenecks may have reduced genetic variation (Karron, 1987) and allowed high loads of recessive deleterious mutations to accumulate (Lynch, Conery, and Bu¨rger, 1995). Thus, breeding between relatives enforced by a further decline in population size in such endemic species may be particularly problematic because the high genetic load will lead to inbreeding depression (Kirkpatrick and Jarne, 2000). Although this situation is the precondition for purging the genetic load (Charlesworth, Morgan, and Charlesworth, 1992; Husband and Schemske, 1996), it also presents an exManuscript received 17 August 2001; revision accepted 5 March 2002. The authors thank Dagmar Frielingsdorf and Christina Aus der Au for help with the fieldwork, Markus Fischer, two anonymous reviewers, Lilli Strasser, and Giorgina Bernasconi for constructive comments, and the Swiss National Science Foundation for financial support to B. Schmid (Swiss Priority Programme Environment, project number 5001-44628). 4 Author for reprint requests (e-mail: [email protected]). 1

tinction risk, thereby raising considerable concern about conservation. If mating within a population is accompanied by inbreeding and inbreeding depression, crossing with pollen from outside the population may increase heterozygosity and thus fitness of descendants (Hauser and Loeschke, 1994). However, if coadapted gene complexes within the parental genomes break down (Templeton, 1986; Dudash, 1990), local adaptations are disrupted (Fenster and Dudash, 1994), and outbreeding depression may result in reduced fitness of descendants (Fischer and Matthies, 1997). If both inbreeding and outbreeding depression occur together and populations or species have a structure with spatial autocorrelation of genetic similarity, an optimal outcrossing distance should exist, in which the positive effects of unrelatedness of the pollen donor are greater than the negative effects of dissimilarity (Waser and Price, 1989; Schierup and Christiansen, 1996). There is a further possibility to reduce effects of inbreeding depression in small populations, which has recently been suggested in animals (Tregenza and Wedell, 2002). If individuals in small populations engage in multiple matings, they may sample a greater range of genetically different parents, thereby increasing the potential to select ‘‘good’’ gametes during the fertilization process. In plants, this selection could occur by some sort of female choice or pollen competition on the stigma or in the style (Aizen, Searcy, and Mulcahy, 1990; Marshall, 1991). If selection of pollen is possible, crossings by many compared with few or single pollen donors could have a beneficial effect on reproductive success and offspring fitness (Niesenbaum, 1999), particularly if it occurs in small popu-

1250

August 2002]

PASCHKE

ET AL.—POPULATION SIZE AND POLLEN DIVERSITY IN

lations. Self-incompatible species represent a simple example of this effect. If the diversity of pollen that a plant within a small population can receive declines, the probability that pollen and ovules are compatible will decrease (Byers and Meagher, 1992; Gigord, Lavigne, and Shykoff, 1998). In other plant species, selfing or single-donor pollination can lead to reduced offspring fitness (Montalvo, 1992). Population sizes of the endemic Cochlearia bavarica Vogt have been declining since the late 1980s and small populations of the species show a pattern of reduced genetic variation, viability, and individual fitness when compared with larger populations (Paschke, Abs, and Schmid, 2002). In this study we used an experimental approach to analyze if lower pollen diversity in small populations may be a cause of the observed pattern in C. bavarica. We artificially crossed maternal plants from small and large populations with different kinds and numbers of pollen donors. To minimize potential confounding effects of pollen load, we used the same number of pollen grains for all pollinations. We asked the following questions: Do offspring of plants from small populations show lower fitness than offspring of plants from large populations? Is decreasing pollen diversity accompanied by reduced reproductive success and fitness depression in offspring? Do pollinations between populations yield fitter offspring than pollinations within populations? Are there interactive effects between population size and pollination treatments such that pollen diversity and pollination between populations are relatively more beneficial to the fitness of offspring in small than in large populations? Is there an optimal outcrossing distance? MATERIALS AND METHODS Study species—The narrow endemic Cochlearia bavarica (Brassicaceae), first described by Vogt (1985), is an allohexaploid (2n 5 36) plant that presumably originated from hybridization between C. pyrenaica DC. (2n 5 12) and C. officinalis L. (2n 5 24) (Koch, Hurka, and Huthmann, 1996). Cochlearia bavarica has a narrow distribution and grows in only two regions, west and southeast of Munich, Germany (Abs, 1999). It occurs in 21 sites in approximately 30 populations. The species is monocarpic and mostly self-incompatible. However, individual plants differ in self-incompatibility. While some plants are strictly self-incompatible, about 40% within a population are able to self-fertilize (M. Fischer, M. Hock, and M. Paschke, unpublished data). Plants flower from May to June and develop about 300 flowers on a ramified primary inflorescence and several secondary inflorescences. Fruit set is, on average, 40%. Each fruit develops a maximum of six seeds. Insect pollinators are generalists, such as flies, bumble bees, other bees, or small moths (M. Paschke, personal observation). Experiment 1: hand pollination with different pollen diversities within a population and with pollen from a nearby population—To test for inbreeding and outbreeding depression in large and small populations, we selected five small and five large populations for hand-pollination experiments in May 1998. Small populations had less than 100 flowering plants, while large ones had more than 1000. In each population we marked five flowering plants at random and estimated their size by counting the number of inflorescences. On the primary inflorescence of each marked plant, we randomly selected six branches and bagged five of them with nylon fabric (mesh size ,0.25 mm). The unbagged branch served as control for open pollination (treatment 1). We hand-pollinated four of the bagged branches with pollen from one (treatment 2), three (treatment 3), and nine donor plants (treatment 4) of the same population or with pollen from nine donor plants of a neighboring population 200–1000 m away from the experimental population (treatment 5). The last bagged inflorescence branch was not pollinated to control for seed set after autonomous selfing (treatment 6). We marked pollen-donor plants at the beginning of the experiment. Within each population, we selected at random

C. BAVARICA

1251

one neighboring pollen-donor plant about 0.5 m from each of the receiving plants (treatment 2) or a set of three (treatment 3) and nine pollen-donor plants (treatment 4), respectively. The pollen donors for treatments 2–4 were always different plants. In nearby populations, we also chose nine pollen-donor plants at random (treatment 5). We used these pollen-donor sets for all maternal plants within each population. To guarantee that we used the same quantity of pollen for each treatment, we collected pollen from 60–90 (treatment 2), 30 (treatment 3), or ten flowers (treatments 4 and 5), respectively, from each of the marked pollen-donor plants in each population. We dried the anthers for about 4 h so that they had time to open and release the pollen grains. One at a time, we hand-pollinated all open flowers on each of the marked inflorescence branches with the corresponding pollination treatment, moving a soft brush covered with pollen over the stigma. All plants were hand-pollinated four times at weekly intervals (it was unlikely that flowers could open and wilt between pollination events and thus remain undetected). Thirty days after the first pollination, we determined the fruit set (ratio of developed to developed 1 undeveloped fruits) of the flowers that were open during the pollination treatments. We harvested ten randomly chosen developed fruits per plant and pollination treatment and specified seed set as the proportion of the maximum possible number of seeds set (i.e., 60 5 6 seeds per fruit 3 10 fruits). We weighed ten randomly chosen seeds per plant and treatment. For germination, we put all developed seeds of each plant and treatment on wet filter paper in petri dishes and exposed them to a day-night regime of 14 h light at 168C and 10 h dark at 108C. Seeds started to germinate after 3 d. After 15 d, there was hardly any new germination. Thus, we measured the germination rate after 15 d, as well as the size (root and leaf length) of three randomly chosen seedlings per petri dish. Immediately following the measurements, we planted a single randomly chosen seedling per mother plant and treatment in a 6 3 6 cm pot containing a mixture of one part sand and two parts soil (M. De Baat BV, Coevorden, The Netherlands). We put the pots in a completely randomized arrangement in a snail- and slug-protected plot in our experimental garden at the University of Zurich. We measured survival and plant performance 210 and 300 d after the start of the experiment (i.e., after the first pollination). We counted the leaves and measured (to the nearest millimeter) the height of the longest leaf tip. As an overall estimate of plant performance, we used total plant size, defined as the product of the number of leaves and plant height. After 420 d, in spring 1999, we counted the plants that had become reproductive adults. In addition to the vegetative characters mentioned above, we also measured the height of the uppermost inflorescence and counted the number of inflorescences and the number of flowers of the primary inflorescence on these plants. The total number of flowers was defined as the product of the number of flowers and the number of inflorescences. Experiment 2: hand pollination with pollen from different distances—To test for optimal outbreeding distance, in May 1997 we marked nine flowering individuals at random in one population of about 100 flowering plants and hand-pollinated them with pollen from different distances. On the primary inflorescence of these plants, we randomly selected six branches and bagged five of them with nylon fabric (mesh size ,0.25 mm). The unbagged branch was used as control for open pollination (treatment 1). We hand-pollinated four of the bagged side branches with pollen from neighboring plants located 1 m (treatment 2) and 10 m (treatment 3) away, and with pollen from other populations located 100 m (treatment 4) and about 1000 m (treatment 5) away. The fifth bagged inflorescence branch was not pollinated to control for seed set after autonomous selfing (treatment 6). We repeated the hand pollinations every 3 d over 21 d in May 1997. For each pollination treatment, we collected pollen from ten plants selected at random. We hand-pollinated all open flowers, one at a time, on each of the marked inflorescence branches with the corresponding pollination treatment, moving a soft brush laden with pollen over the stigma. In June, at fruit maturity, we determined the fruit set (ratio of developed to developed 1 undeveloped fruits) for each treatment and plant. For ten randomly chosen developed fruits per plant and treatment, we counted the number of seeds per fruit and measured the seed mass of ten randomly chosen seeds.

1252 TABLE 1.

AMERICAN JOURNAL

OF

BOTANY

[Vol. 89

Skeleton analysis of variance for variables of Cochlearia bavarica measured in pollination experiment 1. Source of variation

Mean square (MS)

Population size group Population identity [population size group] Pollen recipient [population identity] Hand vs. open pollination Hand pollination within vs. between populations Log-linear effect of pollen diversity Log-quadratic effect of pollen diversity Population size group 3 C1 Population size group 3 C2 Population size group 3 C3 Population identity [population size group] 3 C1 Population identity [population size group] 3 C2 Population identity [population size group] 3 C3 Residual

MSPS MSP MSF MSC1 MSC2 MSC3 MSC4 MSPS3C1 MSPS3C2 MSPS3C3 MSP3C1 MSP3C2 MSP3C3 MSR

Variance ratio

MSPS/MSP MSP/MSF MSF/MSR MSC1/MSP3C1 MSC2/MSP3C2 MSC3/MSP3C3 MSC4/MSR MSPS3C1/MSP3C1 MSPS3C2/MSP3C2 MSPS3C3/MSP3C3 MSP3C1/MSR MSP3C2/MSR MSP3C3/MSR —

Note: C1 5 hand-pollination vs. open pollination, C2 5 hand pollination within vs. between populations, C3 5 log-linear effect of increasing pollination diversity of hand pollination within populations (1, 3, 9 pollen donors), C4 5 log-quadratic effect of increasing pollination diversity of hand pollination within populations, PS 5 population size group, P 5 population identity, F 5 pollen recipient, R 5 residual. Nesting of terms is indicated with square brackets. The interactions of C4 with population size group and with population identity are pooled with the residual.

Statistical analysis of experiment 1—The measured variables of reproductive success and offspring fitness were transformed when they deviated significantly from a normal distribution. In a sequential analysis of variance (ANOVA), we tested the effects of population size group, population identity, maternal plant, pollination treatments, and interactions of these factors on the measured variables. A skeleton ANOVA with the respective variance ratios is presented in Table 1. Mean maternal plant size was considered as a covariate but not included in the final ANOVA models because it only explained a small and nonsignificant proportion of the variance of the other measured variables. We specified four orthogonal contrasts for the pollination treatments. The first contrast, C1, distinguished between open pollination and hand pollination and was thus used as a control for the efficiency of hand pollination itself. The second contrast, C2, distinguished hand pollination within vs. between populations and was used to test for outbreeding depression. The third and fourth contrasts, C3 and C4, formed a second-order polynomial that accounted for the increasing pollen diversity within the hand-pollination treatments using fathers from the same populations (log3 of one, three, or nine pollen donors). Thus, C3 tested for log-linear effects associated with the increasing number of pollen donors, while C4 tested for possible deviations from the log-linear relationship (general linear test as recommended, e.g., by Neter et al., 1996). Pollination treatment 6 was omitted from the analyses because we found no spontaneous self-pollination in this experiment. In addition to ANOVAs for individual characters, we also calculated overall multivariate ANOVAs (MANOVAs) to check for the consistency of effects across the different characters. For these MANOVAs, we grouped the response variables into two sets: early characters (seed characters and offspring characters, including germination and seedling performance) and late characters (offspring performance after 210–420 d). The flowering characters of offspring after 420 d were not included in the MANOVA because only a few plants reached the flowering stage within this time. We calculated the magnitude of fitness depression, d in each character w for the two low levels of pollen diversity within populations (treatments 2 and 3), wi, and for the pollinations between populations (treatment 5), wo, in comparison with the high level of pollen diversity within populations (treatment 4, population averages), wc, using the coefficient of Johnston and Schoen (1996): d 5 1 2 (wi /wc )

or d 5 1 2 (wo /wc )

Positive coefficients indicate fitness depression, whereas negative coefficients indicate fitness increase. We tested whether the magnitude of fitness depression was stage-specific by using the coefficient defined above as dependent variable. Time, expressed as the number of days that had passed since pollination, and all interactions

with time were integrated in the model as a source of variation after population size group, population identity, plant identity, and pollination treatment. We included the linear and quadratic term of time to avoid the problem of serial correlations in repeated-measures analysis (see, e.g., Elashoff, 1986). Statistical analysis of experiment 2—To study the influence of the pollination treatment on offspring performance in experiment 2, we again used an ANOVA with contrasts for the pollination treatments. The first contrast distinguished between open pollination and hand pollination. The second and third contrasts were linear and quadratic contrasts for the effect of the logarithm of pollination distance. Cubic deviations, given the four pollen distances, were the residual term. The contrast mean squares were tested against their respective interactions with maternal plant identity for fruit set, seeds per fruit, and seed mass.

RESULTS Experiment 1: hand pollination with different pollen diversities within a population and with pollen from a nearby population—To demonstrate the effects of pollination treatments on reproductive success and offspring fitness over time, we first analyzed individual characters with separate univariate ANOVAs, then the fitness coefficient in a repeated measures ANOVA, and finally the group of early and late characters with two MANOVAs. Effects of population size, population identity, and maternal plant identity—In large populations fruit set, germination rate of juveniles, survival of offspring after 420 d, and total number of flowers of flowering offspring were higher than in small populations, although some of the effects were only marginally statistically significant (Table 2), possibly because of substantial variation among populations and pollen recipients (Table 3). This latter variation, indicating specific maternal effects due to genetic or environmental variation, was particularly high for the early characters—i.e., fruit set, seed mass, and germination rate—but also significant for total plant size and inflorescence height at some stages later in offspring life (Table 3). Pollination treatments—We found no spontaneous selfing with the exception of two plants in one population. Therefore, the corresponding pollination treatment was not included in

August 2002]

PASCHKE


TABLE 2. Least square means of maternal and offspring characters in small and large population size groups of Cochlearia bavarica. Population size group Character

Maternal Fruit set (%) Seed set (%) Seed mass (mg) Offspring Germination rate (%) Root length of seedlings (cm) Leaf length of seedlings (cm) Survival rate of juveniles (%) Total plant sizea after 210 d Total plant size after 300 d Total plant size after 420 d Proportion of flowering plants (%) Inflorescence height (cm) Total number of flowers

Small

56.55 (0.14) 41.24 (0.22) 1.21 (0.08) 56.52 2.38 0.34 75.41 25.71 96.14 39.41 28.80 18.40 108.60

Large

63.60 (0.15)* 42.72 (0.21) 1.43 (0.08)

(0.34) 69.09 (0.48) 2.88 (0.02) 0.36 (0.78) 90.30 (1.30) 27.50 (4.02) 95.10 (2.76) 40.90 (0.63) 33.11 (1.08) 20.00 (10.55) 142.52

(0.33) (0.45) (0.02) (0.80) (1.24) (3.73) (2.47) (0.62) (0.99) (9.79)*

Note: Data were pooled over all pollination treatments. Characters with marginally significant (P , 0.10) differences between size groups are shown in boldface type; characters with significant (P , 0.05) differences in addition are marked with an asterisk (significances are taken from Table 3). Standard error is given in parentheses. a Number of leaves 3 leaf height.

the analyses. The fruit set of open-pollinated flowers was significantly higher than that of hand-pollinated flowers (Fig. 1A, Table 3). Offspring derived from open pollination also had a higher survival rate after 420 d than those derived from the average, but not from the three-pollen-donor hand-pollination treatment (Fig. 1G). Fruit set and seed set were very similar but seed mass was higher for hand pollinations between vs. within populations (Table 3). However, this difference was not apparent if only the highest pollen diversity within populations was compared with the between-population pollination (Fig. 1C). Total plant size after 420 d was always higher for the hand pollinations between than within populations. For the hand-pollination treatments within populations, fruit set and seed set continually increased with pollen diversity, whereas root length and leaf length of seedlings and survival after 420 d peaked at intermediate levels (Fig. 1, Table 3). This indicates that pollen diversity increases reproductive success but perhaps at the expense of reduced offspring fitness at the highest diversity level. Interactions—Interactions between population size or identity and pollination treatments were rarely significant and therefore omitted from Table 3. In particular, only in one case did a pollination treatment affect plants from small vs. large populations differently: hand pollination had a positive effect on the total number of flowers in large but not in small populations (interaction ‘‘population size 3 hand vs. open pollination’’ significant at P , 0.05). When the interactions between maternal plants and the three pollination contrasts C1–C3 were included in the statistical analysis and tested against the interaction ‘‘pollen recipient 3 pollination contrast C4’’ as error term, they were highly significant for seed mass (P , 0.005) but significant for only a few other characters, despite the large degrees of freedom and, therefore, considerable statistical power. Thus, specific maternal effects (e.g., Schmid and Dolt, 1994), due to genetic or environmental variation did not alter the effects of pollination

C. BAVARICA

1253

treatment on offspring fitness. Therefore, these interactions were pooled into a combined error term for the ANOVAs presented in Tables 1 and 3. Timing of fitness depression—Population identity, pollination treatment, and time had significant effects on the fitness coefficient d (Fig. 2, Table 4). On average, the coefficient was positive, indicating strong fitness depression, for the singlepollen-donor treatment, around zero for the between-population pollination treatment, and even slightly negative for the three-pollen-donor treatment. This indicates again that the nine-pollen-donor treatment resulted in somewhat lower offspring fitness than the three-pollen-donor treatment. Furthermore, the coefficient declined over time, again in agreement with the observation that increased reproductive success after pollination with nine donors from within a population may be at the expense of decreased offspring fitness. Population size group and its interactions had no effect on the fitness coefficient, but population identity effects and their interactions with time and with pollination treatments were substantial. Multivariate analysis of variance (MANOVA)—The effects of population size group and pollination treatments were consistent if compared within early characters or within late characters (Table 5). Small population size had a significant negative effect on early maternal and seedling characters (fruit set to seedling size) but not on characters of later stages (offspring size after 210–420 d). Therefore, the results of the MANOVA confirm the results described in the preceding section—namely, that fitness depression decreased over time. The variation among pollen recipients (5 maternal plants) was significant for both early and late characters, supporting the finding of specific maternal effects in the previous analyses (see Table 3). As in the univariate ANOVAs, the pollination treatment (all contrasts combined) had a particularly strong effect on the early but a weaker (and partly reversed) effect on the late characters in the MANOVA. The significant interaction between population identity and pollination treatment for both early and late characters suggests that fitness depression lasted longer in some populations than in others. Experiment 2: hand pollination with pollen from different distance—The nine plants used as pollen recipients had significantly different fruit set, number of seeds per fruit, and seed mass (Table 6). As in experiment 1, we found no spontaneous selfing in experiment 2 and, therefore, omitted the corresponding pollination treatment from the analyses. Open pollination resulted in higher fruit set and more but slightly lighter (P , 0.1) seeds per fruit than did hand pollination (Fig. 3, Table 6), indicating a potential seed number vs. seed mass trade-off. The relationship between the logarithm of pollination distance within the hand-pollination treatments and seed mass followed a quadratic curve, with a maximum at intermediate distances of 10–100 m. Interactions between maternal plants and open vs. hand pollination or pollination distance within hand-pollination treatments were tested against the error—i.e., the remainder of the maternal plant-by-pollination treatment interaction (equivalent to the interaction maternal plant-by-deviation from quadratic of the pollen-distance treatments). The first of these interactions was significant for all three characters (Table 6), indicating that the reproductive success of some pollen recipients was more dependent on pollination treatment than that of other

A) Source of variation

1 8 — 1 1 1 1

df

0.5 1.4 1.5 3.4 1.4 0.3 0.5

0.253 0.067 — 0.228 0.018 0.480 0.349

MS

3.8 3.9 — 67.7 0.8 6.4 20.5

F

Survival after 420 d

147.59 314.85 184.04 821.64 204.43 29.42 64.76

F

After 210 d

4.0 1.4 2.3 1.2 0.1 0.1 0.1

F

,0.05 ,0.005

,0.005

,0.05

P

,0.05

P

,0.005

P

0.7 13.7 1.1 2.7 3.2 0.1 4.3

F

Root length

16.15 23.21 1.69 6.71 6.81 0.06 6.68

MS

P

,0.05

,0.005

Offspring characters: seedlings

,0.05

,0.05

P

0.0 2.8 0.7 0.0 0.2 0.1 4.7

F

After 300 d

51.13 2484.72 884.59 84.35 381.99 276.86 5895.68

MS

,0.05

P

1 8 16 1 1 1 1

df

6.24 18.49 8.30 21.48 1.09 5.75 0.20

MS

0.3 2.2 2.8 1.9 0.4 7.0 0.1

F

Inflorescence height P

Offspring characters: survival and reproduction

1 8 40 1 1 1 1

df

F

0.1 2.7 1.1 2.8 1.0 10.1 0.2

Seed set

0.007 0.057 0.021 0.067 0.019 0.393 0.003

MS

Offspring characters: growth (total plant size)

1 8 40 1 1 1 1

df

1 8 40 1 1 1 1

df

1 8 16 1 1 1 1

df

1 8 40 1 1 1 1

df

1 8 40 1 1 1 1

df

1 8 40 1 1 1 1

df

F

15 444.91 1711.35 2212.45 177.89 6916.09 514.02 1764.34

MS

9.0 0.8 0.4 0.5 1.4 0.3 0.3

F

Total no. flowers

0.2 0.7 1.5 0.5 6.5 0.9 0.3

After 420 d

89.29 538.22 796.46 240.65 2728.14 183.59 148.30

MS

F

0.8 5.9 0.8 1.1 1.7 0.1 4.8

Leaf length

0.031 0.041 0.007 0.009 0.035 0.002 0.043

MS

F

3.0 3.3 1.9 0.4 6.3 4.8 0.0

Seed mass

2.380 0.789 0.237 0.065 0.573 0.809 0.001

MS

,0.01

P

,0.05

,0.06

P

,0.05

,0.005

P

,0.05

,0.01 ,0.005

P

BOTANY

Note: See Table 1 for abbreviations. The interaction terms and the residual shown in Table 1 were also fitted but are only mentioned in the text. F-values were calculated according to Table 1. For the overall consistency of patterns of significances, see also the MANOVAs in Table 5. For the analysis of survival, data were grouped for populations, therefore variation among recipient plants could not be estimated.

Population size group Population identity Pollen recipient Hand vs. open pollination Hand pollination within vs. between populations Log-linear effect of pollen diversity Log-quadratic effect of pollen diversity

D) Source of variation

1 8 40 1 1 1 1

MS

1.383 0.347 0.250 0.146 0.008 0.013 0.012

MS

,0.01 ,0.05

,0.005 ,0.005

,0.05

P

OF


df

1 8 40 1 1 1 1

df

7.8 0.8 1.8 103.4 0.1 13.3 4.1

F

Germination rate

0.303 0.039 0.051 3.413 0.003 0.434 0.116

MS

Maternal characters

AMERICAN JOURNAL

C) Source of variation


B) Source of variation

1 8 40 1 1 1 1

df

Fruit set

Sequential analysis of variance for maternal and offspring characters of Cochlearia bavarica in pollination experiment 1.


TABLE 3.

1254 [Vol. 89

August 2002]

PASCHKE


C. BAVARICA

1255

Fig. 1. The effect of five different hand-pollination treatments on measures of reproductive success and offspring fitness in Cochlearia bavarica (data from small and large populations combined). The number of pollen donors is given on the x-axis (n 5 variable number of pollen donors in the case of open pollination). SED 5 standard error of difference between means.

recipients. The effects of pollination distance and optimal distance on seed mass also varied among pollen recipients (remaining two interactions in Table 6), presumably because of genetic or environmental variation among maternal plants. DISCUSSION Effects of population size, population identity, and maternal plant identity on reproductive success and offspring size—Reduced reproductive success or offspring fitness in small plant populations have been reported for a number of

Fig. 2. The effect of time since hand pollination within or between populations on fitness depression (or fitness increase) in Cochlearia bavarica (described in MATERIALS AND METHODS: Statistical analysis of experiment 1). The reference for the fitness comparison was always the handpollination treatment with nine pollen donors within populations. Negative values indicate fitness increases rather than depressions. Figure-wide standard error of difference between means 5 0.04.

species (Menges, 1991; Heschel and Paige, 1995; Fischer and Matthies, 1998a, b; Fischer, van Kleunen, and Schmid, 2000; Ke´ry, Matthies, and Spillmann, 2000; Luijten et al., 2000). Our pollination experiments with the rare, endemic plant Cochlearia bavarica also showed a tendency for reduced reproductive success and offspring fitness in small as compared with large populations. However, this effect was difficult to detect because of the large residual variation between populations (effect of population identity) and among maternal plants within populations (effect of pollen recipient). The negative effect of small population size might have been due to reduced availability of genetically different pollen donors, especially because C. bavarica is a self-incompatible species. As we will discuss below, increasing the chances of pollen recipients to obtain genetically different pollen did have a positive effect especially on reproductive success, albeit both in small and large populations. The large variation among maternal plants with regard to seed and offspring characters declined with offspring age as may be expected for maternal carryover effects (Schmid and Dolt, 1994). However, since we found no significant relationship between maternal plant size, used as a covariate, and seed and offspring characters, the maternal effects probably also reflected genetic differences between individual pollen recipients. Maternal plant size is usually seen as a good indicator for maternal resource availability and thus maternal carryover effects (Gorchov, 1988). Differences between hand pollination and open pollination—Although open pollination and hand pollination were equally successful for most of the scored seed and offspring characters of C. bavarica, fruit set was significantly higher after open pollination than after hand pollination (see Fig. 1A).

1256

AMERICAN JOURNAL

OF

BOTANY

[Vol. 89

TABLE 4. Repeated-measures analysis of variance, using sequential sum of squares, for the fitness coefficient d of Cochlearia bavarica measured in Experiment 1. Skeleton analysis Source of variation

Population size group Population identity Pollen recipient Pollination treatment PS 3 PT P 3 PT Pollen recipient 3 PT Day linear Day squared PS 3 day linear PS 3 day squared P 3 day linear P 3 day squared PT 3 day linear PT 3 day squared P 3 PT 3 day linear P 3 PT 3 day squared Residual

Mean square (MS)

Fitness coefficient

Variance ratio

MSPS MSP MSF MSPT MSPS3PT MSP3PT MSF3PT MSD MSDx3D MSPS3D MSPS3D3D MSP3D MSP3D3D MSPT3D MSPT3D3D MSP3PT3D MSP3PT3D3D MSR

MSPS/MSP MSP/MSF MSF/MSF3PT MSPT/MSP3PT MSPS3PT/MSP3PT MSP3PT/MSF3PT MSF3PT/MSR MSD/MSP3D MSD3D/MSP3D3D MSPS3D/MSP3D MSPS3D3D/MSP3D3D MSP3D/MSR MSP3D3D/MSR MSPT3D/MSP3PT3D MSPT3D3D/MSP3PT3D3D MSP3PT3D/MSR MSP3PT3D3D/MSR —

df

MS

F

1 8 39 2 2 16 75 1 1 1 1 8 8 2 2 16 16 1188

1.37 0.81 0.28 1.77 0.56 0.37 0.26 5.79 3.03 0.04 0.18 0.49 0.82 0.03 0.26 0.23 0.09 0.13

1.7 2.9 2.1 4.8 1.5 1.4 2.0 11.9 3.7 0.1 0.2 3.7 6.2 0.1 2.9 1.8 0.7

P

,0.05 ,0.05 ,0.005 ,0.01

,0.005 ,0.005 ,0.05

Note: The skeleton analysis is on the left, the actual analysis on the right. PS 5 population size group, P 5 population identity, F 5 pollen recipient, PT 5 pollination by one or three pollen donors within a population (wi) or by nine pollen donors from a different population (wo) compared against pollination by nine pollen donors within a population (wc) (see MATERIALS AND METHODS: Statistical analysis of experiment 1), D 5 time (measured as number of days since hand pollination), R 5 residual.

This may have been due to a negative effect of bagging the inflorescences for the hand-pollination treatments. Effects of pollen diversity—We found that increasing the number of pollen donors from one to three to nine had a positive effect on reproductive success (fruit set and seed set), whereas offspring characters peaked at the intermediate level of pollen diversity (root length and leaf length of seedlings, offspring survival after 420 d) (see Fig. 1). Because the total amount of pollen applied was the same for each level of pollination diversity (see MATERIALS AND METHODS section), these effects could not be explained as effects of pollenload size. Rather, it seems likely that both effects (increase in female reproductive output with the number of pollen donors and optimum of offspring fitness for an intermediate level of multiple matings) resulted from variation in genetic diversity among pollen donors. Several, mutually nonexclusive mechanisms are possible. The increase in fruit and seed set could reflect an increased investment by the recipient plant for multiply-sired fruits, as has been found in wild radish (Marshall and Ellstrand, 1986). Preferential investment for multiply-sired fruits may be adaptive for the recipient plant because, analogous to the sampling effect of diversity (Loreau, 2000), an increasing number of pollen donors increases the chances for

compatible pollen grains or among compatible ones for pollen grains with high fertilizing abilities (Bernasconi and Keller, 2001) or less closely related to the recipient plant (Tregenza and Wedell, 2002). Increased offspring fitness among multiply-sired progeny for intermediate levels of multiple pollination may arise through better resource partitioning among half than full sibs (Karron and Marshall, 1990), a mechanism analogous to niche complementarity (Loreau, 2000). Such an effect may follow an optimality curve if competition becomes predominant over complementarity for high levels of pollen diversity. Alternatively, the smaller seedling size and survival of offspring from the highest than the intermediate diversity treatment may indicate a trade-off between fertilizing ability and ‘‘father quality’’ or that optimal levels of polyandry differ for the male and female function in plants, as they do for instance in insects (Arnqvist and Nilsson, 2000). However, such trade-offs to our knowledge have not been described so far in plants. Future studies will have to assess to which degree multiple pollination also results in multiple paternity among the seeds. The low average fruit set resulting from the one-donor treatment was expected because the probability that a single pollen donor is not compatible with the recipient plant is relatively high in the endemic C. bavarica with sporophytic self-incom-

TABLE 5. Multivariate analysis of variance for early characters (fruit set, seed set, seed mass, germination rate, and root length and leaf length of seedlings) and late characters (leaf numbers, leaf heights after 210, 300, and 420 d) of Cochlearia bavarica in pollination experiment 1. Early characters Source of variation

Population size group Population identity Pollen recipient Pollination treatment Population size group 3 PT Population identity 3 PT

df

6, 48, 240, 24, 24, 192,

35 177 734 127 127 730

Late characters

Exact F

P

4.09 3.27 1.22 3.64 0.47 1.27

,0.001 ,0.005 ,0.05 ,0.001 ,0.05

Note: PT 5 pollination treatment. F values are based on Wilks’ lambda (SAS, 2000).

df

6, 48, 240, 24, 24, 186,

35 177 454 123 123 451

Exact F

0.12 1.33 1.60 1.40 0.79 1.23

P

,0.001 ,0.05

August 2002] TABLE 6.

PASCHKE


C. BAVARICA

1257

Analysis of variance, using sequential sum of squares for variables of Cochlearia bavarica measured in pollination experiment 2. Fruit set

Seeds per fruit

Seed mass

Source of variation

df

MS

F

P

df

MS

F

P

df

MS

F

P

Pollen recipient Hand vs. open pollination Distance of pollen donor D3D Pollen recipient plant 3 C1 Pollen recipient plant 3 D Pollen recipient plant 3 D 3 D Residual

8 1 1 1 8 8 8 8

1172.94 5946.64 219.80 816.43 533.32 129.82 191.34 61.41

19.1 11.2 1.7 4.3 8.7 2.1 3.1

,0.005 ,0.01

8 1 1 1 8 8 8 293

26.26 46.99 0.15 0.06 7.63 3.08 4.51 2.67

9.9 6.2 0.1 0.0 2.9 1.2 1.7

,0.005 ,0.05

8 1 1 1 8 8 8 357

0.44 2.15 0.02 1.65 0.28 0.21 0.22 0.04

11.0 7.7 0.1 7.5 7.0 5.3 5.5

,0.005

,0.01

,0.005

,0.05 ,0.005 ,0.005 ,0.005

Note: C1 5 hand pollination vs. open pollination, D 5 log-linear effect of increasing pollination distance of hand pollination (pollen from 1, 10, 100, 1000 m distance), D 3 D 5 log-quadratic effect of increasing pollination distance of hand pollination. Each distance contrast is tested against its corresponding interaction with the pollen recipient.

patibility (M. Fischer, M. Hock, and M. Paschke, unpublished data). Further, if a single pollen donor was compatible (there are degrees of compatibility in C. bavarica; for a further example see Gigord, Lavigne, and Shykoff, 1998), it may still have been closely related to the recipient plant and thus inbreeding depression may have resulted. Indeed, we found that the stages of seed development until seed dispersal (30–60 d) expressed large fitness depressions after hand pollination with only one pollen donor (see Fig. 2). After 60–300 d, fitness depression was not apparent, but it did show up again at the last investigated stage (420 d after pollination). Surprisingly, interactions between pollination treatments and population size group, population identity, or maternal plant identity (5 effect of pollen recipient) were small and statistically almost never significant. Thus, the mentioned positive effects of increasing pollen diversity on reproductive success and of the intermediate pollen diversity on offspring char-

acters were consistent across populations and maternal plants. Hence, we conclude that C. bavarica in both small and large populations may benefit from pollinator services that lead to multiple mating. Effects of pollination distance and of pollen from outside the population—We found evidence for an optimal outcrossing distance between 10 and 100 m. Pollen collected within this distance, and representing a mix from ten donor plants each time, showed higher siring success after hand pollination than did pollen from 1 m away or from more than 100 m away. This optimal outcrossing distance could be explained by the appearance of inbreeding and outbreeding depression within the same species. Plants in the neighborhood of a maternal plant are probably more closely related than plants farther away, and thus the likelihood of inbreeding increases in lowdistance pollinations (see, e.g., Fenster, 1991). On the other

Fig. 3. The effect of pollination from different distances on measures of reproductive success in Cochlearia bavarica. SED 5 standard error of difference between means.

1258

AMERICAN JOURNAL

hand, plants from much farther away may be too different, such that co-adapted gene complexes could break down or local adaptations disrupt. Optimal outcrossing distances have for instance also been observed in Ipomopsis aggregata (Pursh) V. Grant (1–10 m; Waser and Price, 1989) and in a few other plant species (Waser, 1993; Fischer and Matthies, 1997). Besides the negative effects of the largest pollination distances, we found significantly positive effects of pollen from nearby plants in a different population rather than the same population for small and large populations alike on seed mass and on offspring size after 420 d (see Fig. 1C, F). The positive effects of pollen from outside a population (and for pollen from 100 m away, which also came from a neighboring population in the pollination-distance experiment) showed that all populations might suffer some inbreeding depression. Sheridan and Karove (2000) reported similar positive effects of intersite crosses in another rare plant. Implications for conservation—Our pollination experiments suggest that pollination by multiple pollen donors or by pollen donors at some distance from a maternal plant has a beneficial effect on reproductive success and offspring fitness in the narrow endemic species C. bavarica. Pollination by single donors may result in inbreeding depression. Therefore, it is essential that pollinator services can be maintained within and among neighboring populations of C. bavarica. Because pollinator availability may be threatened particularly in small populations (Paschke, Abs, and Schmid, 2002), and because small populations independent of pollination treatment had reduced reproductive success and offspring fitness in the present study, conservation efforts should focus on these small populations. If possible, one should try to increase the size of these populations or connect them to neighboring populations. LITERATURE CITED ABS, C. 1999. Differences in the life history of two Cochlearia species. In K. Marhold, B. Schmid, and F. Krahulec [eds.], Ecology of closely related plant species, 39–53. Opulus Press, Uppsala, Sweden. AIZEN, M. A., K. B. SEARCY, AND D. L. MULCAHY. 1990. Among- and within-flower comparisons of pollen-tube growth following self- and cross-pollinations in Dianthus chinensis (Caryophyllaceae). American Journal of Botany 77: 671–676. ARNQVIST, G., AND T. NILSSON. 2000. The evolution of polyandry: multiple mating and female fitness in insects. Animal Behaviour 60: 145–164. BERNASCONI G., AND L. KELLER. 2001. Multiple mating by females affects their sons’ reproductive success in the red flour beetle Tribolium castaneum. Journal of Evolutionary Biology 14: 186–193. BOND, W. J. 1996. Assessing the risk of plant extinction due to pollinator and disperser failure. In J. H. Lawton and R. M. May [eds.], Extinction rates, 129–146. Oxford University Press, Oxford, UK. BYERS, D. L., AND T. R. MEAGHER. 1992. Mate availability in small populations of plant species with homomorphic sporophytic self-incompatibility. Heredity 68: 353–359. CHARLESWORTH, D., M. T. MORGAN, AND B. CHARLESWORTH. 1992. The effect of linkage and population size on inbreeding depression due to mutational load. Genetical Research 59: 49–61. CZECH, B., P. R. KRAUSMAN, AND P. K. DEVERS. 2000. Economic associations among causes of species endangerment in the United States. BioScience 50: 593–601. DUDASH, M. R. 1990. Relative fitness of selfed and outcrossed progeny in a self-compatible, proteandrous species, Sabatia angularis L. (Gentianaceae): a comparison in three environments. Evolution 44: 1129–1139. ELASHOFF, J. D. 1986. Analysis of repeated measures designs. BMDP technical report 83. BMDP Statistical Software, Los Angeles, California, USA.

OF

BOTANY

[Vol. 89

ELLSTRAND, N. C., AND D. R. ELAM. 1993. Population genetic consequences of small population size: implication for plant conservation. Annual Review of Ecology and Systematics 24: 217–241. FALCONER, D. S. 1989. Introduction to quantitative genetics, 2nd ed. Longman, London, UK. FENSTER, C. B. 1991. Gene flow in Chamaecrista fasciculata (Leguminosae). II. Gene establishment. Evolution 45: 410–422. FENSTER, C. B., AND M. R. DUDASH. 1994. Genetic considerations for plant population restoration and conservation. In M. L. Bowles and C. J. Whelan [eds.], Restoration of endangered species, 34–62. Cambridge University Press, Cambridge, UK. FISCHER, M., AND D. MATTHIES. 1997. Mating structure and inbreeding and outbreeding depression in the rare plant Gentianella germanica (Gentianaceae). American Journal of Botany 84: 1685–1692. FISCHER, M., AND D. MATTHIES. 1998a. RAPD variation in relation to population size and plant fitness in the rare Gentianella germanica. American Journal of Botany 85: 811–819. FISCHER, M., AND D. MATTHIES. 1998b. Effects of population size on performance in the rare plant Gentianella germanica. Journal of Ecology 86: 195–204. FISCHER, M., M. VAN KLEUNEN, AND B. SCHMID. 2000. Genetic allee effects on performance, plasticity, and developmental stability in a clonal plant. Ecology Letters 3: 530–539. GIGORD, L., C. LAVIGNE, AND J. A. SHYKOFF. 1998. Partial self-incompatibility and inbreeding depression in a native tree species of La Re´union (Indian Ocean). Oecologia 117: 342–352. GORCHOV, D. L. 1988. Effects of pollen and resources on seed set and other fitness components in Amelanchier aborea (Rosaceae: Maloideae). American Journal of Botany 75: 1275–1285. GROOM, M. J. 1998. Allee effects limit population viability of an annual plant. American Naturalist 151: 487–495. HAUSER, T. P., AND V. LOESCHKE. 1994. Inbreeding depression and matingdistance dependent offspring fitness in large and small populations of Lychnis flos-cuculi (Caryophyllaceae). Journal of Evolutionary Biology 7: 609–622. HESCHEL, M. S., AND K. N. PAIGE. 1995. Inbreeding depression, environmental stress, and population size variation in scarlet gilia (Ipomopsis aggregata). Conservation Biology 9: 126–133. HUSBAND, B. C., AND D. W. SCHEMSKE. 1996. Evolution of the magnitude and timing of inbreeding depression in plants. Evolution 50: 54–70. JOHNSTON, M. O., AND D. J. SCHOEN. 1996. Correlated evolution of selffertilization and inbreeding depression: an experimental study of nine populations of Amsinckia (Boraginaceae). Evolution 50: 1478–1491. KARRON, J. D. 1987. A comparison of levels of genetic polymorphism and self-compatibility in geographically restricted and widespread plant congeners. Evolutionary Ecology 1: 47–58. KARRON, J. D., AND D. L. MARSHALL. 1990. Fitness consequences of multiple paternity in wild radish, Raphanus sativus. Evolution 44: 260–268. KERY, M., D. MATTHIES, AND H.-H. SPILLMANN. 2000. Reduced fecundity and offspring performance in small populations of the declining grassland plants Primula veris and Gentiana lutea. Journal of Ecology 88: 17–30. KIRKPATRICK, M., AND P. JARNE. 2000. The effect of a bottleneck on inbreeding depression and the genetic load. American Naturalist 155: 154– 167. KOCH, M., H. HURKA, AND M. HUTHMANN. 1996. Chloroplast DNA restriction site variation and RAPD-analyses in Cochlearia (Brassicaceae): biosystematics and speciation. Nordic Journal of Botany 16: 585–603. LACY, R. C. 1987. Loss of genetic diversity from managed populations: interacting effects of drift, mutation, immigration, selection and population subdivision. Conservation Biology 1: 143–158. LOREAU, M. 2000. Biodiversity and ecosystem functioning: theoretical advances. Oikos 91: 3–17. LUIJTEN, S. H., A. DIERICK, J. G. B. OOSTERMEIJER, L. E. L. RAIJMAN, AND H. C. M. DEN NIJS. 2000. Population size, genetic variation and reproductive success in a rapidly declining, self-incompatible perennial (Arnica montana) in the Netherlands. Conservation Biology 14: 1776–1787. LYNCH, M., J. CONERY, AND R. BU¨RGER. 1995. Mutation accumulation and the extinction of small populations. American Naturalist 146: 489–518. MARSHALL, D. L. 1991. Nonrandom 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., AND N. C. ELLSTRAND. 1986. Sexual selection in Ra-

August 2002]

PASCHKE


phanus sativus: experimental data on nonrandom fertilization, maternal choice, and consequences of multiple paternity. American Naturalist 127: 446–461. MENGES, E. S. 1991. Seed germination percentage increases with population size in a fragmented prairie species. Conservation Biology 5: 158–164. MONTALVO, A. M. 1992. Relative success of self and outcross pollen comparing mixed- and single-donor pollinations in Aquilegia caerula. Evolution 46: 1181–1198. MONTGOMERY, M. E., L. M. WOODWORTH, R. K. NURTHEN, D. M. GILLIGAN, D. A. BRISCOE, AND R. FRANKHAM. 2000. Relationship between population size and loss of genetic diversity: comparisons of experimental results with theoretical predictions. Conservation Genetics 1: 33–43. NETER, J., M. H. KUTNER, C. J. NACHTSHEIM, AND W. WASSERMAN. 1996. Applied linear statistical models, 4th ed. Irwin, Chicago, Illinois, USA. NIESENBAUM, R. A. 1999. The effects of pollen load size and donor diversity on pollen performance, selective abortion, and progeny vigor in Mirabilis jalapa (Nyctaginaceae). American Journal of Botany 86: 261–268. OUBORG, N. J., AND R. VAN TREUREN. 1994. The significance of genetic erosion in the process of extinction. IV. Inbreeding load and heterosis in relation to population size in the mint Salvia pratensis. Ecology 48: 996– 1008. PASCHKE, M., C. ABS, AND B. SCHMID. 2002. Relations between population size, allozyme variation, and plant performance in the narrow endemic Cochlearia bavarica (Brassicaceae). Conservation Genetics, in press.

C. BAVARICA

1259

SAS. 2000. JMPt statistics and graphics guide, version 4. SAS Institute, Cary, North Carolina, USA. SCHIERUP, M. H., AND F. B. CHRISTIANSEN. 1996. Inbreeding depression and outbreeding depression in plants. Heredity 77: 461–468. SCHMID, B., AND C. DOLT. 1994. Effects of maternal and paternal environment and genotype on offspring phenotype in Solidago altissima L. Evolution 48: 1525–1549. SHAFFER, M. 1987. Minimum viable populations: coping with uncertainty. In M. E. Soule´ [ed.], Viable populations for conservation, 69–86. Cambridge University Press, Cambridge, UK. SHERIDAN, P. M., AND D. N. KAROVE. 2000. Inbreeding, outbreeding, and heterosis in the yellow pitcher plant, Sarracenia flava (Sarraceniaceae), in Virginia. American Journal of Botany 87: 1628–1633. TEMPLETON, A. R. 1986. Coadaptation and outbreeding depression. In M. E. Soule´ [ed.], Conservation biology: the science of scarcity and diversity, 105–160. Sinauer, Sunderland, Massachusetts, USA. TREGENZA, T., AND N. WEDELL. 2002. Polyandrous females avoid costs of inbreeding. Nature 415: 71–73. VOGT, R. 1985. Die Cochlearia pyrenaica Gruppe in Zentraleuropa. Berichte der Bayerischen Botanischen Gesellschaft 59: 133–135. WASER, N. M. 1993. Population structure, optimal outbreeding, and assortative mating in angiosperms. In N. Thornhill [ed.], The natural history of inbreeding and outbreeding: theoretical and empirical perspectives, 173–199. University of Chicago Press, Chicago, Illinois, USA. WASER, N. M., AND M.V. PRICE. 1989. Optimal outcrossing in Ipomopsis aggregata: seed set and offspring fitness. Evolution 43: 1097–1109.