5
Late-acting inbreeding depression in both male and female function
of Echium vulgare (Boraginaceae).
Summary
We present data on late-acting inbreeding depression in pollen performance, siring success and seed production in Echium vulgare. Pollen viability and rate of pollen tube growth were both lower for pollen from plants derived from selfing than for pollen from plants derived from outcrossing. Pollen tube numbers within the styles did not differ for pollen from plants derived from selfing or outcrossing. A pollination experiment with two mixtures of pollen from plants derived from selfing or outcrossing, revealed a significant decline of 55% in siring success for pollen from plants derived from selfing. A second experiment with a complete diallel design revealed inbreeding depression for both siring success of the offspring (32.8%) and a decline in seed production of the offspring (34.8% - 40.6%). In addition, results indicated a heritable component for seed number per flower.
Offspring fitness, measured as seed production and siring ability, can be severely affected by late-acting inbreeding depression. Inbreeding depression values for male and female functions were not correlated. Both functions must therefore be considered when calculating inbreeding depression.
C. Melser, A. Bijleveld and P.G.L. Klinkhamer. 1999. Heredity 83: 162-170.
Introduction
Inbreeding depression is considered a major force in the evolution of plant reproductive systems (Darwin 1876, 1877, Maynard Smith 1978, Lloyd 1980, Charlesworth & Charles- worth 1987). Inbreeding depression is the reduction in fitness of offspring produced by polli- nation with closely related pollen donors, compared to the fitness of offspring after pollination with unrelated pollen donors. Ideally, the entire life cycle of the progeny, from seed to seed production, is measured to determine offspring fitness, and thus the magnitude of the inbreeding depression. Yet most published estimates of inbreeding depression are predominantly based on early life stages, i.e. seed production of the maternal plant, germination of the seeds, survival of the seedlings and size of the offspring (e.g. Waser &
Price 1993, Trame et al. 1995, survey in Husband & Schemske 1996, Fischer & Matthies 1997, Byers 1998, del Castillo 1998, Hardner 1998). Estimates of early inbreeding depression may underestimate the cumulative lifetime inbreeding depression. Although offspring have been screened for quality at reproductive stages several times, data are mostly restricted to measurements of e.g. days to flowering, number of flowers or number of seeds of the offspring (Waser & Price 1993, survey of 25 species in Husband & Schemske 1996). The reduction in the female fertility as a result of inbreeding depression ranges from -0.09 in Eichhornia paniculata to 0.74 in Clarkia tembloriensis with a selfing rate of 0.49 (reviewed in Husband & Schemske 1996). The number of seeds measures only the female function of a hermaphrodite plant and on average thus only half of the reproductive success. The effect of inbreeding depression on the male function of plants needs to be studied to get a more reliable estimation of the magnitude of inbreeding depression in plant populations.
Late-acting inbreeding depression at the stage of pollen production has been described for Mimulus guttatus (Willis 1993, Carr & Dudash 1997). Offspring derived from selfing pro- duced approximately 60% fewer pollen grains than offspring derived from outcross polli- nations after one generation of selfing (Fig. 1 in Carr & Dudash 1997). In addition, stainability of the pollen after one generation of selfing was 15% to 40% lower (Fig. 1d in Willis 1993), declining to 60% lower after four generations of selfing (Fig. 1 in Carr &
Dudash 1997). Collinsia heterophylla also showed a decline in the stainability of pollen from plants derived from selfing (Mayer et al. 1996). However, this effect was less than 7% and occurred only in two populations out of four (Mayer et al. 1996). In Phacelia dubia the progeny derived from outcrossing tended on average to have a significantly higher frequency of normal pollen grains than those derived from plants which were produced by selfing, with a difference of 9% (del Castillo 1998). Selecting for high and moreso for low ovule number in Malva moschata by means of selfing also produced severe inbreeding depression in via- bility of the pollen (T.J.Crawford, pers. comm.). These studies of differences in pollen quality were not continued up to the siring of seeds. As far as we know, only the study by Jóhannsson et al. (1998) has measured effects of inbreeding on siring success directly as functional siring ability. Pollen from plants of Cucurbita texana derived from selfing had significantly slower growing pollen tubes in vitro. This effect was on average approximately 8% (Fig. 1 in Jóhannsson et al. 1998). The pollen from plants derived from selfing also sired fewer seeds under conditions of pollen competition with a tester line on Cucurbita pepo. Their experimental design with pollination against a tester line under pollen competition does not
identify separate effects of slower pollen tube growth or post-fertilization causes of differential siring success.
Here we present a study on late-acting inbreeding depression in pollen performance, siring success and seed production in Echium vulgare. In an earlier study of E. vulgare on average no differences in seed production were found between self- and outcross pollinations (Chapter 4). Thus, there are no indications of early inbreeding depression during seed production (Chapter 4) of this mainly outcrossing species (Rademaker 1998). However, late- acting inbreeding depression could influence its reproductive dynamics. By comparing the total reproductive success of selfed and outcrossed progeny we quantify the magnitude of late-acting inbreeding depression during reproductive stages of the progeny for both male and female function.
Materials and Methods
Species
Echium vulgare (L.) is a rosette-forming monocarpic perennial. From the main flowering stem, cymes diverge on which flowers develop sequentially (Nicholls 1987). Each day new flowers open on each cyme. Flowers are hermaphrodite with five anthers and four ovules. The four ovules are arranged in a square. Flowers are protandrous: first the anthers present the ripe pollen in the male phase, then after about one day the style elongates and the two stigmatic lobes diverge and become receptive to pollination. Although protandry and herkogamy reduce self-pollination within one flower, selfing by geitonogamy can still occur because flowers in the male and female phase are present on the same plant simultaneously. The selfing rate of male-fertile plants in the field in 1996 was estimated with molecular paternity analyses and ranged between 0 and 30% in six individuals (Rademaker 1998). Echium vulgare is a gynodioecious species. In Meyendel, near The Hague, where plants were collected, about 12% of all individuals are male-sterile (Klinkhamer et al. 1991). This male sterility is heritable (Klinkhamer & de Jong, unpubl. data). A clear distinction can usually be made between (self-compatible) hermaphrodite individuals with perfect flowers that produce fertile blue pollen, and male-steriles with female flowers that produce infertile yellow pollen. The yellow pollen appears to be collapsed and unstainable with methylene blue when viewed under the microscope. Apparent male-steriles with yellow pollen did not occur in our experiments.
Selection and cultivation of the plants
In earlier experiments about a third of the individuals produced more seeds after selfing than after outcrossing (Chapter 4 and 7), but over all parental combinations used, there was no difference on average between the number of seeds produced after selfing compared to out- crossing (Chapter 4 and 7). For our experiments we selected the parental individuals (P gene- ration) and their number of descendants (F1 plants) to conform to these proportions of seed production: in earlier hand-pollinations, two parental individuals (i and j) produced relatively many seeds after selfing compared to outcrossing, whereas two parental individuals (a and b)
produced relatively few seeds after selfing. The different parental combinations of i, j, a and b were included in the experiment with both selfing and outcrossing. The comparison between selfed and outcrossed plants is thus a comparison within relatives and partially eliminates the contribution of genotypic maternal effects (Lynch 1988). The labelling of the plants is as in Chapter 4.
Seeds were germinated on filter paper; seedlings were potted in 3-L pots filled with 50% sand and 50% potting soil and randomly placed in a growth chamber under controlled conditions. Day and night temperatures were, respectively, 20˚C and 15˚C, and relative humi- dity ranged between 60% and 85%. To induce flowering, after 7 weeks the plants received a cold treatment at 5˚C. After six weeks of cold, the temperature was raised to 22˚C and 18˚C during, respectively, day and night. Additionally the plants received four droplets of gibberellin (21.75% in H2O) in the middle of the rosette two times a week to ensure the production of a flowering stalk (Wesselingh et al. 1994). The final height of the plants ranged between 70 and 90 cm and did not differ between selfed and outcrossed plants. The plants used in the experiments were all at the same developmental stage.
We studied pollen viability, number of pollen grains on the stigma, number of pollen tubes and pollen tube growth and performed two experiments on siring success. Unless other- wise stated, two-sided levels of significance are given.
Pollen viability
To obtain a rough estimate (Heslop-Harrison et al. 1984) of the level of late-acting inbreeding depression in the viability of the pollen in the anthers, we stained the pollen and determined the percentages viable and non viable pollen.
Two individuals of each parental combination derived from selfing were used, based on the availability of individuals of different genotypes: progeny of the crosses (j x j) (=maternal individual x paternal individual), (a x a) and (b x b). One individual of the parental combinations derived from outcrossing was used from the crosses (j x a), (a x j), (j x b) and (b x j). Pollen was collected from several ripe anthers of each plant, diluted in methylene blue (Dafni 1992) and analysed under a light microscope. Percentages of stainable pollen were scored for each plant for 330 to 760 pollen grains. Differences between plants in percentage stainable pollen were tested with Wilcoxon tests.
Pollination method
Flowers were emasculated with forceps before the style elongated and before anthers dehis- ced. Flowers were pollinated by rubbing the pollen firmly on the lobes of the stigma with the end of a toothpick, which was covered with parafilm. With this method of hand-pollination, 90% of the flowers received at least five visible pollen grains that adhere permanently to the stigma (unpubl. data, cited in Chapter 4). The treated flowers were marked with a small drop of paint to identify the pollen donor. All plants were pollinated within one period of 50 days.
Approximately three weeks after the last pollinations, the number of developed seeds per flower was counted.
Number of pollen grains on the stigma, number of pollen tubes and pollen tube growth Based on availability of individual plants, the parental combinations derived from selfing, (i x i) and (j x j), and the parental combinations derived from outcrossing, (j x a) and (j x b), were used, with two individuals of each parental combination. For each parental combination of individual plants, in the F1 generation five flowers were pollinated. After 5 h. the styles were collected, fixed in ethanol with acetic acid (4:1) and stored in 70% ethanol. The tissue was softened overnight in 8N KOH, stained in 0.006% aniline blue in 0.15 M K2HPO4 and viewed with a fluorescence microscope (Martin 1959). The number of pollen grains still adhering to the visible side of the stigma was counted and analysed with Kruskal-Wallis tests.
Stigmas without any visible pollen grains were excluded for the analysis of pollen tubes. The number and the length of the pollen tubes was recorded, using an ocular micrometer. The differences between the parental combinations in number and length of the pollen tubes were analysed with Kruskal-Wallis tests.
Siring success and number of seeds per flower Experiment 1
To examine late inbreeding depression in siring success of pollen of F1 plants, a mix of pollen from plants derived from either selfing or outcrossing was applied on five randomly chosen recipient plants. These recipient plants were collected at least 100 m apart in the nature re- serve of Meyendel, near The Hague.
One pollen mixture was made from six plants derived from selfing (from two sibships from each of the three parental combinations (j x j), (a x a) and (b x b)), and another pollen mixture was made from four plants derived from outcrossing (parental combinations (j x a), (a x j), (j x b) and (b x j)). The mixtures were made with an equal number of anthers of each parental combination. Each mixture was applied to 20 to 30 flowers of the recipient plants, a total of 40 to 60 flowers per maternal individual. All pollinations were carried out within one day. The resulting numbers of seeds per flower were counted. The binomially distributed data of the numbers of seed per flower were analysed with a GLM procedure with a logit link function (McCullagh & Nelder 1989, SAS Institute 1993). The effects of factors entered later in the analysis are adjusted for the effects of the earlier entered factors (SAS PROC GENMOD, type 1). The numbers of seeds per flower were adjusted for the effect of maternal individual as a main factor.
Experiment 2
To examine late inbreeding depression on seed-set and siring success of F1 plants derived from selfing and outcrossing, we pollinated flowers with pure pollen of different pollen donors in a greenhouse and counted the resulting number of seeds per flower. The ten plants used were the same individuals as were examined for pollen viability. All plants were polli- nated in a complete diallel design. Each plant received pollinations with pure pollen from each other individual; 25 to 30 flowers were pollinated for each parental combination, which yields a total of 250 to 300 flowers per maternal individual. The resulting numbers of seeds per flower were counted.
Because number of seeds per flower and siring success may have a heritable com- ponent, we first tested for this by regression of the F1 generation on midparent values. For the seed number per flower, this regression was significant (see results section) and we therefore had to separate the effect of heritability and the effect of late-acting inbreeding depression. To test for late-acting inbreeding depression, we regressed the F1 recipient plants derived from outcrossing against their midparent values and subsequently tested whether the residuals of this regression line for the F1 recipient plants derived from selfing were significantly smaller than zero. To quantify the reduction in number of seeds caused by inbreeding depression, the difference between the slopes of the two regression lines for either plants derived from out- crossing or plants derived from selfing was calculated.
For the male function (siring success) there is no significant heritability (Chapter 4 and this experiment). The binomially distributed data of number of seeds per flower were analysed for the effect of late-acting inbreeding depression on the male function (siring success) with a GLM procedure (SAS PROC GENMOD, type 1). The number of seeds per flower was adjusted for the effect of week of pollination, and the effect of the different pollination types was analysed within individual plants. We distinguished three different pollination types: i) self-pollination within one plant (denoted as self); ii) outcross-pollination with pollen from F1 plants derived from selfing (denoted as S); and iii) outcross-pollination with pollen from F1 plants derived from outcrossing (denoted as C). All three pollination types were on F1 recipient plants derived from selfing (denoted as S) and on F1 recipient plants derived from outcrossing (denoted as C). So, for example, C x S denotes a maternal plant derived from outcrossing, pollinated with pollen from a father derived from selfing, and S self denotes a recipient plant derived from selfing that is self-pollinated. An example of the combination of the pollination types on different types of recipient plants is shown in Fig. 1.
Multiple comparisons between the means of the different pollination types on the recipient plants either derived from selfing or from outcrossing were analysed with t-tests. Significance levels were corrected for multiple comparisons with an improved Bonferroni test (Haccou &
Meelis 1992).
P generation
F1 generation
a x j x b
S x S
a x a(S) a x j(C)j x a(C) b x j(C)j x b(C) b x b(S)
S x C C x C C x S
C self S self
Figure 1:
Example of the different pollination types on maternal plants. a, j and b denote different individuals of the P generation. They produce the F1 generation by selfing and outcrossing, labeling the F1 gene- ration with S and C respectively.
Relation between male and female reproductive success
Pollen viability, siring success and number of seeds per flower were measured on the same individuals. We could therefore calculate the correlation between these two components of male fitness and female fitness.
Results
Pollen viability
The percentages of viable pollen differed between the group of F1 plants derived from selfing and the group of F1 plants derived from outcrossing (normal approximation Z=3.24;
p=0.0012). Pollen from plants derived from selfing contained on average 89.9% (se=0.011) stainable pollen grains (ranging from 84.0% to 95.1% for parental combinations (j x j) and (b x b), respectively), and pollen from plants derived from outcrossing contained on average 93.2% (se=0.016) stainable pollen grains (ranging from 87.3% to 96.9% for parental combinations (j x a) and (b x j), respectively).
Pollen number on stigma
On average, 8.89 (se=0.331) pollen grains were adhering permanently to the visible part of the stigma after hand-pollination. The back side of the stigma on the slide is not visible under the microscope, so the number of pollen present on the stigma is underestimated. The number of pollen grains adhering to the visible part of the stigmas did not differ significantly between pollen donors (normal approximation Z=−0.418; p=0.675)
Pollen tube number
The number of pollen tubes in the style after approximately 5 h. was on average 1.13 (se=0.109). There was no difference between F1 maternal plants derived from selfing and out- crossing (Fig. 2A; p=0.9589) nor between F1 pollen donors in the six combinations of pollina- tions (see Fig. 2A; p=0.1095).
Pollen tube growth
The average length of the pollen tubes in the style after 5 h. ranged between 0.92 and 2.28 mm. Within the styles of maternal plants derived from selfing, the pollen tubes grew more slowly than in the styles of plants derived from outcrossing (Fig. 2B; p=0.0071). The length of the pollen tubes depended on the origin of the pollen donor (Fig. 2B; p=0.0087).
Excluding the self-pollinations, pollen from plants derived from selfing grew on average 1.12 mm (se=0.146), whereas pollen from plants derived from outcrossing grew on average 2.05 mm (se=0.162). Self-pollinations within the different maternal plants (S self and C self) did
Table 1:
GLM analysis of the number of seeds per flower in Echium vulgare after pollinations on five randomly collected individuals with a pollen mixture derived from selfing and a pollen mixture derived from outcrossing.
Factor d.f. F-value P-value
Individual plant 4 8.24 <0.0001 Pollen mixture self or outcross 1 19.72 <0.0001
not differ significantly in pollen tube lengths compared to outcross pollinations (compare S self with SxS and C self with CxC). There was thus no immediate effect of selfing on pollen tube growth, but pollen from plants derived from selfing grew slower in the next generation.
Number of seeds per flower: Experiment 1
Averaged over all randomly chosen maternal plants, the mean number of seeds per flower was 0.55 (se=0.058). Maternal individual influenced the mean number of seeds (Table 1).
The mean number of seeds per flower ranged from 0.36 to 1.20. However, the significance level of the difference between siring success of pollen from plants derived from selfing or outcrossing was even greater (Table 1). The mixture of pollen from plants derived from outcrossing sired on average 0.77 (se=0.096) seeds per flower, whereas that from plants derived from selfing sired on average only 0.34 (se=0.064) seeds per flower, a decrease of 55.8%.
2.00
1.60
1.20
0.80
0.40
Average number of pollen tubes (mm) 0.00
3.00
2.40
1.80
1.20
0.60
0.00
S self S x S S x C C self C x S C x C
Parental combination
Average length of pollen tubes (mm)
s c
s c c s c
A
B
Figure 2:
A Average number and B length of pollen tubes (se) in the style for Echium vulgare in a complete diallel pollination experiment, classified by the origin of the recipient plant (either derived from selfing S or outcrossing C), pollination- type of selfing within one plant (S self and C self), outcross- pollination with pollen from plants derived from selfing (SxS and CxS) and outcross pollination with pollen from plants derived from outcrossing (SxC and CxC).
Number of seeds per flower: Experiment 2 Female reproduction
Averaged over all parental combinations in the diallel pollination design, the mean number of seeds per flower was 0.42 (se=0.016).
The number of seeds produced by the F1 generation plants was correlated at the margin of significance (R2=0.383; n=10; p=0.056) with the mid-parent value of the number of seeds produced by the P generation (Fig. 3), which indicates a slight heritable component for the number of seeds produced. The regression line for the outcrossed F1 recipient plants with an intercept fixed at 0 has a slope of 0.996 (s.e.=0.179). For this regression line, the residuals of the maternal plants derived from selfing (X) were significantly smaller than zero
(X(avg) = -0.203; t5=5.027; p=0.002). This represents 34.8% of the mean number of seeds per
flower, compared with the plants derived from outcrossing (X = 0.584)). Thus plants derived from selfing produce fewer seeds than expected on the basis of their midparent values, indicating late-acting inbreeding depression. The difference between the slopes of the regres- sion lines for the number of seeds produced derived from selfing vs. outcrossing gives an alternative estimate (40.6%) of the decrease in number of seeds produced in the F1 generation as a result of late-acting inbreeding depression.
Figure 3:
Regression of the number of seeds per flower of the F1 generation in experiment 2 on the mid-parent value of the P generation for each parental combination for Echium vulgare. Closed dots are plants derived from selfing; open dots are plants derived from outcrossing. Regression lines are based on all data (continuous line; y=0.764x), on plants derived from selfing (dashed line; y=0.592x) and on plants derived from outcrossing (dotted line; y=0.996x).
Male reproduction
There was no significant correlation between the siring success of the F1 generation plants and their mid-parent values, indicating that the siring success is not significantly heritable (p=0.8057). The effect of paternal inbreeding depression was significant (Table 2; Fig. 4).
Over all parental combinations, pollen from plants derived from selfing (S self, SxS and CxS) sired on average 0.34 (se=0.020) seeds and pollen from plants derived from outcrossing (C self, SxC and CxC) sired on average 0.52 (se=0.027) seeds. For the outcross pollinations, pollen from plants derived from selfing (SxS and CxS) sired 32.8% (0.39 vs. 0.58) fewer seeds than pollen from plants derived from outcrossing (SxC and CxC).
For the four plants derived from outcrossing, early-acting inbreeding depression is detectable in the zygotes: fewer seeds per flower were produced after self-pollination (C self) compared to pollinations with pollen from other individuals that were derived from outcrossing (CxC; Fig. 4). For the six plants derived from selfing, no effect of early-acting inbreeding depression was found. Over all ten individual plants in the experiment, on three individuals more seeds per flower were produced after selfing compared to outcrossing (Table 3). The interaction between pollination type and maternal plant was not significant and was excluded from the final analysis.
Relation between pollen viability and seed number per flower
With a one-sided test, there is a significant correlation between pollen viability and siring success (r=0.462; n=10; p=0.045).
Relation between maternal and paternal seed production
There is no significant correlation between the average seed number per flower produced as a maternal plant and the average seed number per flower sired as a pollen donor (r=0.178;
n=10; p=0.311).
Table 2:
GLM analysis of the number of seeds per flower in Echium vulgare after single-donor pollinations in experiment 2 with three pollination types: i) self-pollination within one plant; ii) outcross-pollination with pollen from F1 plants derived from selfing; and iii) outcross-pollination with pollen from F1 plants derived from outcrossing.
Factor d.f. F-value p-value
Week 6 8.00 <0.0001
Maternal plant derived from selfing or outcrossing 1 78.53 <0.0001
Individual plant 8 20.36 <0.0001
Pollination type 20 3.39 <0.0001
Discussion
The results of our study unequivocally show that late-acting inbreeding depression can seve- rely reduce the fertility of offspring derived from selfing. Not only was the female function of the hermaphroditic flowers of the offspring derived from selfing reduced by 34.8 to 40.6%, but also a decline in siring success of 32.8% to 55.8% was detected. Different stages of pollen performance are affected. Pollen stainability was lower in offspring derived from selfing, but the difference was small. However, pollen viability is only one component of male fitness.
Moreover, low pollen stainability was not reflected in a lower number of pollen on the stigma, nor in fewer pollen tubes. The length of the pollen tubes shows a greater effect of late-acting inbreeding depression. Pollen tubes from plants derived from selfing grew slower than tubes from plants derived from outcrossing, in accordance with the findings of Jóhannsson et al. (1998). Pollen competition in the style can cause a shift from the percen- tages of paternal genotypes present on the stigma to the resulting percentages of the paternity of the seeds (Mulcahy 1979, Snow and Mazer 1988, Snow & Spira 1991, 1996). In each mixed pollination the pollen from plants derived from outcrossing would then outcompete the pollen from plants derived from selfing. Our results probably underestimate the effects of selfing on siring ability, because we used pure, single-donor pollinations, giving all pollen donors equal access to ovules. In Chapter 4, however, no differences in the success of pollen donors at siring seeds were found between single- and mixed-donor pollinations.
Early inbreeding depression in seed-set and survival of the offspring also has to be included in calculations of the life-time inbreeding depression. Among the parental generation that was tested for selective abortion in Chapter 4, inbreeding effects at the stage of seed production were on average zero. In contrast, the data in the present experiment suggest
1.00
0.80
0.60
0.40
0.20
0.00
S self S x S S x C C self C x S C x C Parental combination
Average number of seeds per flower
ab a b A A B
s c
s c c s c
Figure 4:
Average number of seeds per flower (se) for Echium vulgare in experiment 2, classified by the origin of the recipient plant and pollen donor (either derived from selfing S or outcrossing C), pollination- types of selfing within one plant (S self and C self), outcross pollination with pollen from plants derived from selfing (SxS and CxS), and outcross pollination with pollen from plants derived from outcrossing (SxC and CxC). Capital and small letters denote different Tukey tests.
Late-acting inbreeding depression 85
Average number of seeds per flower in Echium vulgare for the different parental combinations in experiment 2. Italic numbers denote selfing.
Maternal individual
j x j(1) j x j(2) b x b(1) b x b(2) a x a(1) a x a(2) j x b b x j j x a a x j Mean in
outcross
pollinations
Paternal individual j x j(1) 0.174 0.417 0.100 0.161 0.167 0.233 0.081 0.200 0.600 0.680 0.293 j x j(2) 0.267 0.679 0.029 0.167 0.300 0.100 0.240 0.250 0.790 0.417 0.284 B x b(1) 0.750 0.211 0.000 0.000 0.360 0.433 0.240 0.333 0.367 0.840 0.393 B x b(2) 0.484 0.500 0.067 0.267 0.320 0.154 0.286 0.667 0.654 0.500 0.403 A x a(1) 0.400 0.267 0.040 0.100 0.200 0.000 0.350 0.581 0.550 0.882 0.352 A x a(2) 0.222 0.261 0.036 0.323 0.500 0.048 0.483 0.375 0.677 0.960 0.426 j x b 0.655 0.889 0.133 0.200 0.733 0.300 0.097 0.655 1.000 1.250 0.646 B x j 0.700 0.407 0.133 0.167 0.267 0.467 0.567 0.633 1.035 0.885 0.514 j x a 0.655 0.407 0.133 0.167 0.167 0.400 0.226 0.800 0.667 0.840 0.422 A x j 0.433 0.786 0.133 0.233 0.516 0.419 0.645 0.552 0.966 0.517 0.520
Mean in
outcross pollinations. 0.507 0.460 0.089 0.169 0.370 0.279 0.346 0.490 0.737 0.806 0.425
that selfing results in fewer seeds than outcrossing (at least in plants that were derived from outcross pollination). An extensive study of early inbreeding depression in seed germination, plant growth and survival under field conditions is described in Chapter 7. Results indicate a decline of circa 10% in germination and early seedling survival. In the literature, a life-time inbreeding depression of 50% has been mentioned as the threshold limit for selfing to be selectively favoured (Lloyd 1980, Lande & Schemske 1985). The threshold of 50% will be even lower if pollen discounting affects the availability of pollen for outcross pollinations (de Jong et al. 1993, Holsinger 1994). With an overall level of late-acting inbreeding depression of 34 - 48%, the life-time inbreeding depression of E. vulgare is close to or even exceeds 50%, and this preliminary estimate suggests that outcrossing should be favoured.
As the selfing rate of E. vulgare under natural conditions is between 0 and 30%
(Rademaker 1998), the species would be considered by Husband & Schemske (1996) as an outcrossing species (selfing rate < 45%). The considerable inbreeding depression of over 30%
in the reproductive stage in E. vulgare supports the hypothesis that, with a low selfing rate, considerable inbreeding depression remains. Nearly half of the 40 outcrossing species ana- lysed by Husband & Schemske (1996) showed inbreeding depression during growth and fe- male reproduction (i.e. number of flowers and number of seeds) of the offspring.
Although the number of different individual parents in this experiment is limited to five, the seed material was collected from those which showed the most extreme variation in number of seeds from selfing vs. outcrossing in earlier pollination experiments (Chapter 4), reducing parent sampling error (Lynch 1988). Our experiment conforms to the fraction of one-third of the plants producing more seeds after selfing compared to outcrossing in the parental generation. Also, in our experiment the average number of seed per flower after self- pollination in the parental generation is equal to that after cross-pollination. Strikingly, in this experiment, the parental combination (j x j) produced relatively more seeds per flower after self-pollinations than after cross-pollinations and parental combinations (a x a) and (b x b) produced relatively fewer seeds per flower after self-pollinations. The parental generation of j, a and b in Melser et al. (1997) has similar differences in seed production between self- and cross-pollination, revealing a heritable component for the production of seeds per flower.
The inbreeding depression in both male and female fitness observed in this study sug- gests that this phenomenon occurs in both genders, and is not correlated between genders.
Estimates of late-acting inbreeding depression, accounting only for the female function, can not therefore legitimately be extrapolated to overall late-acting inbreeding depression.
Furthermore, estimates of overall inbreeding depression, solely based on early inbreeding depression, may greatly underestimate the lifetime inbreeding depression in E. vulgare.
Acknowledgements
We thank H. Nell and K. van Veen-van Wijk for assistance in the field and cultivating the plants and M. Brittijn for drawing the figures. T.J. Crawford kindly allowed us to mention his unpublished data. D. Charlesworth, T.J. de Jong, E. van der Meijden and two anonymous referees gave useful comments on earlier drafts.
