Response to selection on autogamy in Phlox.
Plant breeders have successfully selected for increased self-fertility in cultivars of several predominantly outcrossing species (Villegas et al. 1971; Richards and Thurling 1973; Busbice et al. 1975; Henny and Ascher 1976; Robacker and Ascher 1978; 1982; Dana and Ascher 1985; Yamada et al. 1989). Selection lines in most cultivars were developed by inbreeding, which would accelerate the rate of response by increasing the genetic variance among individuals. Thus the responses achieved in cultivar selection programs are not likely to be representative of the responses in wild species.
The shift from predominant self-fertilization to predominant cross-fertilization rarely has been documented in flowering plant lineages (e.g., Whalen and Anderson 1981; Barrett and Shore 1987). Selfing species may, in general, lack the capacity to return to outbreeding, "based upon the nature of genetic factors (recessive gene control, integrated gene complexes), rapid fixation of inbreeding genes, morphological character complexes, or ecological factors" (Jain 1976). The few attempts at artificial selection to decrease self-fertility in cultivars have produced varied results (Ockendon 1973; Richards and Thurling 1973; Busbice et al. 1975). The gains that have been achieved have been small and sometimes at the expense of pollen fertility.
The evolvability of breeding systems, either from outcrossing to selfing or from selfing to outcrossing, has never been analyzed through selection experiments on wild plants. The Phlox drummondii-Phlox cuspidata complex is a particularly favorable vehicle for such a study. Phlox cuspidata is largely self-fertilizing (Levin 1978) and is thought to be derived from the almost exclusively outcrossing P. drummondii (Wherry 1955; Erbe and Turner 1962). Phlox cuspidata is the only selfing wild Phlox. Ostensibly the breeding system of the prototype of P. cuspidata passed through a stage where plants had a moderate to high degree of self-fertility but where cross pollen was prepotent to self pollen. These features are found in many cultivars of P. drummondii (Levin 1975; 1989).
The purposes of this investigation were as follows: (1) to determine the facility with which the breeding system of P. drummondii could be moved toward that of P. cuspidata; (2) to determine the extent to which the breeding system trajectory followed in the evolution of P. cuspidata could be reversed; and (3) to determine the response of a partially inbreeding cultivar of P. drummondii, Salmon Beauty, to selection for increased and decreased autogamy. We specifically address the following questions: (1) To what extent can the self-compatible and self-incompatible breeding systems of the two wild species be altered by two cycles of artificial selection? (2) How does the responsiveness of Salmon Beauty to selection compare with that of wild Phlox? (3) What reproductive characters change as correlated responses to selection for autogamy?
MATERIALS AND METHODS
Phlox drummondii Hook. (Polemoniaceae) is an annual endemic to central Texas. It is a predominantly outcrossing species. Pseudo-self-fertility accounts for 1-5% autogamous seed set in unmanipulated flowers (Levin 1985). Pseudo-self-fertility is achieved in the presence of a functional incompatibility system and is characterized by higher seed production with cross pollen than with self pollen, earlier germination of cross pollen and a continuous distribution of self-fertility levels in progeny (Levin 1975, 1989, 1993). Two wild populations of P. drummondii, one from Elgin, Texas, and the other from Nixon, Texas, were chosen for this research as representing distinctive populations. They represent subspecies macallisteri and drummondii, respectively.
Phlox cuspidata Scheele is also an annual endemic in central Texas. It is the only self-compatible, wild Phlox species out of more than 60 species and has a flower morphology that facilitates selfing (Erbe and Turner 1962). The flowers are relatively small compared to the other annual phloxes, and the lowest anther in the corolla tube is generally in contact with the stigma. Considerable fruit and seed set occurs in the absence of pollinators, and levels of selfing in natural populations of P. cuspidata are as high as 80% (Levin 1978, 1989). In addition, P. cuspidata has a relatively low pollen to ovule ratio (Plitmann and Levin 1990), typical of inbreeding species. Two distant populations were considered in this study. One population (Bastrop, Texas) is on the western edge of the species range; the other (Giddings, Texas) is located near the center of the species range.
Seeds for the P. drummondii phase of this study were collected haphazardly from more than 100 plants per population in the spring of 1987 and germinated in the greenhouse. The experiment was initiated in the summer of 1988. Seeds were collected haphazardly from more than 100 individuals in each of the two P. cuspidata populations in the spring of 1989.
The P. drummondii cultivar Salmon Beauty has a 160 year history of inbreeding, selection and genetic drift (Kelly 1915). This cultivar has half the proportion of polymorphic loci (0.11 versus 0.21), one-fourth the level of heterozygosity (0.010 versus 0.042), and one-third the level of genetic diversity (1.24 versus 3.49) of wild P. drummondii populations (Levin 1976). It has an intermediate level of autogamy (ca. 50%, achieved through pseudo-self-fertility) and is quite hardy (Levin 1989). Seeds of Salmon Beauty were obtained from Sluis and Groot Company, Enkhuizen, The Netherlands.
The Breeding Program
Due to limitations on space and manpower, P. cuspidata, wild P. drummondii, and the cultivar were alternated in the greenhouse through three generations with control and selected lines over the course of about four years. The two populations per species, however, were always grown simultaneously and randomly interspersed. Since control and selected lines were not replicated, the strength of this breeding program is in the number of selection experiments performed on autogamy (six) rather than in statistical power.
Two to three weeks after germination, seedlings from each population were transplanted into six inch pots containing a standard greenhouse soil mix. These pots were then randomly assigned to positions on greenhouse benches where they were maintained throughout the duration of the selection cycle. Following minor juvenile mortality, the number of individuals remaining in these base populations were as follows: Elgin, 177; Nixon, 209; Bastrop, 112; Giddings, 114; cultivar, 245. When the plants were about eight to 10 weeks old and had produced about 30 flowers each, they were ready for selection to be imposed. The greenhouse was free of pollinators.
The level of autogamous fruit set was recorded for each plant. This variable does not significantly change with age (Bixby, unpubl. data). The 10% most autogamous plants in each wild P. drummondii population served as the selected plants. They were cross-pollinated with each other such that each one was approximately equally represented as pollen and egg parents. Moreover, every possible combination of pollen and egg parents was made to the extent that a pair of plants produced flowers synchronously. This method maximizes the total number of genotypes in the offspring and thus minimizes the chances of inbreeding in subsequent generations (Mesken 1987). The 10% least autogamous plants in the P. cuspidata populations were identified, and cross-pollinated with each other as described above. In Salmon Beauty, the 10% most autogamous and the 10% least autogamous plants were identified as the "high" and "low" selected parents, respectively. For each population, randomly chosen individuals (numbering approximately 10% of the population) were cross-pollinated with each other to generate seeds for a control group.
In cycle two of the selection process, seedlings representing the selected crosses and control crosses from each of the two populations per species (two selection lines in the case of the cultivar) were planted in a randomized design in the greenhouse. This amounted to at least 125 seedlings for each P. cuspidata group and at least 150 seedlings for each wild and cultivated P. drummondii group. Selection was imposed on the selected-1 generation and seeds produced as above. In total, we had five groups of seeds for each wild population: initial population (base), control-1, control-2, selected-1, and selected-2. For the cultivar, we had seeds from the base population, control-1, control-2, high selected-1, high selected-2, low selected-1, and low selected-2. Hereafter, these groups are referred to as selection treatments.
Following the two cycles of selection, samples of approximately 30 plants from each generation (including base, selected, and control lines) from each population were grown to maturity at randomly assigned positions in the greenhouse. Such collateral cultivation allowed us to observe the changes from one generation to the next while minimizing environmental variation. The level of autogamous fruiting was measured on all plants. Then, on 12 plants randomly chosen from the 30 representatives of each generation, we measured a suite of reproductive characters including self-compatibility, cross-compatibility, anther-stigma proximity, corolla tube length, level of automatic self-pollination, seed-set from automatic self-pollination (autogamy), level of anther indehiscence (P. cuspidata only), and pollen viability (P. cuspidata and cultivar only). In addition, we recorded the total flower number per plant at the time of the autogamous fruit-set survey as an indicator of plant vigor. Total flower number is highly correlated with plant dry weight in P. drummondii across a variety of environments (0.75 [less than or equal to] r [less than or equal to] 1.00; Waitt, pers. comm.). Flower number in P. cuspidata is also significantly, positively correlated with plant dry weight (N = 133, r = 0.8047, P = 0.0001; Bixby, unpubl. data).
Self-compatibility was measured as the proportion of self-pollen grains germinating and sending tubes into the style. With both the wild and cultivated P. drummondii, five flowers from each of the 12 plants per generation per population were haphazardly chosen and labeled prior to flower senescence to prevent experimenter bias toward fruiting or nonfruiting flowers. Then, after senescence, the pistils were harvested and stored in separate vials of 70% ethanol for later processing. With P. cuspidata, 10 flowers per plant were haphazardly chosen from among those that were beginning to senesce - flowers that were estimated to be four days old or more. The 10 pistils from a plant were harvested and stored together in a vial of 70% ethanol. Self-pollen grains from automatic self-pollination were dislodged from the stigmas into solution by sonication, stained with toluidine blue, filtered by vacuum filtration onto paper, and counted under a light microscope. We used a procedure adapted from Linskens and Esser (1957), Martin (1959), and Ramming et al. (1973) to stain pollen tubes. Fixed pistils were washed in tap water and then cleared in 8N NaOH for 12-18 h. After washing again three times in tap water, the pistils were placed in Trisglycine buffer for 1-2 h and counterstained in 0.01% toluidine blue for 15 min. Staining in decolorized Aniline blue (DAB) for 24 h was followed by mounting in a drop of DAB: glycerin (1:1) solution. The pistils were then observed for pollen tubes using fluorescence microscopy.
Cross-compatibility was measured using five flowers from each of the 12 plants per generation per population. Flowers were emasculated before anthesis to prevent contamination from self-pollen. The pollen from several plants of the same species was pooled and used as a cross-pollen source. Cross-pollinations were made with a dissecting needle, applying an average of 43 [+ or -] 4 grains per stigma in P. drummondii, 54 [+ or -] 2 grains in P. cuspidata, and 35 [+ or -] 2 grains in the cultivar. A cellophane hood was then placed over the calyx to simulate the corolla tube and prevent the exposed stigma from drying. Pistils were harvested after 24 h and refrigerated immediately (it was known to take at least 8 h for Phlox pollen to germinate in vivo; Levin 1975), pollen grains were counted under a dissecting scope, and then the pistils were stored in separate vials of 70% ethanol for later pollen tube staining as described above.
Anther-stigma proximity was recorded as the distance (to the nearest 0.1 mm) between the lowest anther and the stigma on the same flowers for which corolla tube length measurements were made. Corolla tube length was measured to the nearest 0.5 min. In wild and cultivated P. drummondii five flowers were haphazardly chosen from among all open flowers on a plant. In P. cuspidata these measurements were made on the same 10 flowers per plant from which values of self-compatibility and self-pollination were obtained. The proportion of these 10 flowers having indehiscent anthers was also recorded.
Up to 15 self-fruits per plant were harvested on most of the 12 plants per generation per population to obtain an estimate of seeds per fruit. Typically, three seeds per ovary is the maximum for these species, but occasionally four were observed.
Protandry was measured as the rate of stigma maturation based on elongation rate. In P. drummondii, anthers dehisce upon flower opening, but the three stigmatic lobes are not fully mature and receptive to pollen germination for several hours to several days later. Five one-day old, five two-day old, and five three-day old flowers were harvested per plant, and the pistils therein were stored in 70% ethanol. Then the style + stigma lengths were recorded to the nearest 0.1 mm, and the rate of elongation was obtained. Because some of the 12 plants per selection treatment were either dead or dying by the time this trait was scored, measurements were made on six wild or seven cultivated plants per selection treatment to maintain a balanced design.
Pollen viability was scored in P. cuspidata and the cultivar to determine if a reduction in autogamous fruiting was due to pollen sterility. The pollen from two flowers per plant and six cultivar or seven P. cuspidata plants per selection treatment was analyzed in a drop of aniline blue in lactophenol under a light microscope. Viable grains stain dark blue. The proportion of grains out of more than 100 per flower which stained dark blue was recorded. This method has been shown to be significantly correlated with pollen germinability in vitro (N = 72, r = 0.7227, P = 0.0001; Bixby, unpubl. data).
The realized heritability and additive genetic variance for autogamous fruiting were calculated from the response to selection as follows:
[h.sup.2] = [R.sub.x]/[S.sub.x] (1)
[V.sub.A] = [h.sup.2][V.sub.p] (2)
where the selection response ([R.sub.x]) is the difference in the mean of the character between the unselected control line and the offspring of the selected parents, the selection differential ([S.sub.x]) is the difference in the mean of the character between the selected parents and the population from which they were selected, and [V.sub.p] is the phenotypic variance in the base population (Allard 1960; Bulmer 1980; Falconer 1989). Because selection was imposed for more than one generation, the cumulative selection response and selection differential were used for [R.sub.x] and [S.sub.x], respectively. Standard errors for realized heritabilities were estimated according to the formula given by Falconer (1989, p. 211).
The realized heritability for autogamous fruiting in the cultivar was also calculated as the divergence between the two selection lines according to the following formula:
[h.sup.2] = ([R.sub.H] + [R.sub.L])/([S.sub.H] + [S.sub.L]) (3)
where [R.sub.H] and [R.sub.L] are the responses to selection in the high and low lines, respectively, and [S.sub.H] and [S.sub.L] are the selection differentials (Mesken 1987).
The additive genetic coefficient of variation (C[V.sub.A]) is a useful measure for making comparisons of the ability of characters to respond to selection (the "evolvability"; Houle 1992). We calculate these values as follows:
[Mathematical Expression Omitted]
where [Mathematical Expression Omitted] represents the population mean before selection.
Characters that are associated with autogamous fruiting via linkage or pleiotropy can affect the rate and/or the direction of evolution if (1) the change in these characters affects fitness; or (2) natural selection is simultaneously shifting the two characters in the same or opposite directions (Hazel 1943; Crow and Nagylaki 1976; Lande 1979). Such associations are reflected by a correlated response to selection in a trait not under direct selection and can be distinguished from genetic drift if the response is in the same direction in both populations of the unidirectional selection programs or if the response is in opposite directions in the two selection lines of the bidirectional selection program and if these responses are relatively consistent in magnitude. Real differences between populations in response to selection, [TABULAR DATA FOR TABLE 1 OMITTED] however, cannot be determined without replication within populations.
To determine the extent of correlated responses to selection, we first tested the residuals of all traits for normality and homogeneity of variances. If these criteria were met, either for the raw data or after transformation procedures, one-way ANOVAs (PROC GLM, SAS 1985) were carried out to determine whether any differences in trait means among selection treatments were significant. When the assumptions for parametric analyses were not met, a nonparametric one-way ANOVA (Kruskal-Wallis test; PROC NPAR1WAY, SAS 1985) was performed. If more than one measurement was made per plant, the mean value per plant was used in the analysis. When a significant selection treatment effect appeared from either parametric or nonparametric tests, unplanned multiple comparison tests were performed using the TUKEY option (PROC GLM, SAS 1985) or paired comparisons using the Wilcoxon two-sample test (PROC NPAR1WAY, SAS 1985), respectively.
Direct Response to Selection
A significant breeding system shift occurred in all six selection experiments. The shift from xenogamy to autogamy was more pronounced than the shift from autogamy to xenogamy. The Elgin and Nixon populations of P. drummondii showed a significant increase in autogamy in the selected line, starting at less than 5% and reaching 41% in Elgin and 56% in Nixon (Table 1). Given that the change in autogamy was due to selection (for reasons which will be discussed later), then autogamous fruiting has a realized heritability of 0.65 (0.02) in the Elgin population and of 1.17 (0.05) in the Nixon population. This is illustrated with the cumulative response to selection plotted as a function of the cumulative selection differential [ILLUSTRATION FOR FIGURE 1 OMITTED]. Additive genetic coefficients of variation were 139.6 (0.4) and 390.0 (1.0) for the Elgin and Nixon populations, respectively. During the collateral cultivation, no control-1 from the Elgin population survived to maturity, and only three base representatives survived. Nixon base and control-1 were not significantly different from control-2 in mean percentage autogamous fruit set or in any other trait.
There was a significant differential between selected and control lines after two cycles of selection in both the Bastrop and Giddings populations of wild P. cuspidata (Table 1). In Giddings the decrease in autogamous fruiting was nearly double that in Bastrop. Heritabilities for autogamous fruiting were estimated to be 0.35 (0.12) and 0.56 (0.17) in the Bastrop and Giddings populations, respectively. The functional relationship between cumulative response to selection and cumulative selection differential is plotted in Figure 1. Additive genetic coefficients of variation for autogamous fruiting were calculated to be 19.1 (1.8) for Bastrop and 28.8 (2.5) for Giddings.
The distribution of autogamous fruiting in the base population of Salmon Beauty was bimodal [ILLUSTRATION FOR FIGURE 2 OMITTED]. Mean autogamous fruit set was 53.9%. The base population was not represented during the collateral cultivation, because none of the seeds germinated. The seeds were five years old and probably inviable. No significant change occurred in the control line over the course of two generations (Table 2). Autogamous fruiting declined dramatically in the low line to 12.1% after just two cycles of selection (Table 2). In the high line autogamous fruiting increased manifestly to nearly 100% after two generations (Table 2). These responses to selection produced realized heritability estimates of 0.87 (0.23) in the high line, 0.34 (0.12) in the low line, and 0.57 (0.22) based on the divergence between the two lines. The additive genetic coefficient of variation is 49.3 (3.4) in the high line, 30.8 (2.1) in the low line, and 39.9 (2.8) based on the divergence between the two lines. The cumulative response is plotted against the cumulative selection differential in Figure 1.
Correlated Responses to Selection
As the level of autogamous fruit set increased in P. drummondii, only self-compatibility showed a significant correlated change in both populations (Table 3). The percentage of self-pollen sending tubes into the style increased from 0.55% to 2.15% in Elgin and from 0.01% to 7.49% in Nixon. Corolla tube length significantly decreased in the Elgin population but did not change significantly in the Nixon population. The distance between stigma and lowest anther significantly declined in the Nixon population but not in the Elgin population (Table 3). The level of automatic self-pollination, cross-compatibility, rate of pistil maturation, and total number of flowers showed no change associated with the change in autogamy (Table 3).
TABLE 2. Mean percentage autogamous fruiting in Salmon Beauty for all selection treatments for all plants represented in the collateral cultivation.
Material N Mean
Low-2 34 12.1 (3.9)a(*) Low-1 29 30.7 (6.7)ab Control-2 28 40.1 (7.0)b Control- 1 23 46.3 (7.7)bc High- 1 28 76.5 (4.4)cd High-2 34 94.6 (1.3)d
* Means followed by the same letter cannot be considered significantly different at the alpha = 0.017 level of probability.
Measurements of reproductive traits in P. cuspidata were made on representatives from control-2, selected-2, and the base groups only, because the base, control-1, control-2, and selected-1 groups did not significantly differ from one another in percentage autogamous fruit set (Table 1). Ultimately, the difference between control-2 and selected-2 is most important. It was also desirable to measure traits on representatives of the base simultaneously with control-2, since control individuals were produced by hand cross-pollination in the greenhouse, whereas most base individuals were likely produced by autogamous pollination in the field.
There was no significant change in self-compatibility in Bastrop P. cuspidata. However, in the Giddings population, the percentage self-compatibility in the selected-2 group was far less than in either the base or control-2 (Table 4). This reduction ostensibly was responsible for a significant decline in self-seeds per fruit (Table 4). Corolla tube length significantly decreased in both P. cuspidata populations; antherstigma proximity was unchanged.
Anther indehiscence, which was very low in both control populations, increased significantly in Giddings (Table 4). Anther indehiscence significantly reduced the level of self-pollination in the Giddings population. Indeed the level in the selected-2 was only one-sixth of that in control-2 (Table 4). The decline in anther dehiscence was not associated with a decline in pollen viability.
[TABULAR DATA FOR TABLE 3 OMITTED]
The increase in autogamy in the high selected line of Salmon Beauty was accompanied by a significant increase in self-compatibility, which more than doubled. Self-seeds per fruit also significantly increased, probably as a result of the increase in self-compatibility (Table 5). Both corolla tube length and anther-stigma proximity significantly decreased (Table 5). The latter has not contributed to the increase in autogamy, because the level of self-pollination was unchanged. There were no significant changes associated with the increase in autogamy in pistil maturation rate, pollen viability, and total flower number (Table 5).
The decline in autogamy in the low line was not accompanied by alterations in self-compatibility or self-seeds per fruit (Table 5). Thus the correlated responses to selection in opposite directions are not mirror images of one another. The rate of pistil maturation significantly decreased, thus enhancing protandry (Table 5). This may have contributed to a decline in autogamy, because pollen grains would be unable to germinate for several hours to a few days. Corolla tube length and the distance between anther and stigma significantly decreased (Table 5). There were no significant changes in self-pollination, cross-compatibility, pollen viability, and total flower number (Table 5).
Whether selection was for increasing or for decreasing autogamy, all selected Phlox groups shifted in the direction of the selected type. Despite the lack of replication, the consistency of these results across populations and across taxa suggests that Phlox breeding systems are quite responsive to selection and that each taxon carries a substantial amount of additive genetic variation for autogamy.
The two- to 10-fold increase in autogamy in wild P. drummondii after just two cycles of artificial selection is comparable to the responses achieved by selection and inbreeding in other species. For example, in Petunia integrifolia two cycles of selection increased the level of self-fertility from ca. 0.0% to 25.3% (Dana and Ascher 1985). In Nemesia strumosa, four generations of inbreeding and selection increased the level of self-fertility from a few percent to 100% (Robacker [TABULAR DATA FOR TABLE 4 OMITTED] and Ascher 1978). One cycle of selection for self-fertility in inbred families of turnip resulted in an increase in self-seed set from about two seeds per fruit to more than five (Richards and Thurling 1973). In our study, selected parents were cross-pollinated rather than inbred to produce the subsequent generations. Inbreeding accelerates a response to selection by increasing the genetic variance among individuals (Falconer 1989).
There have been only two studies in which selection for both an increase and decrease in self-fertility was attempted on the same population. One involved the aforementioned turnip study. By selecting weakly self-compatible individuals from a highly self-compatible family of turnip, Richards and Thurling (1973) achieved a statistically significant reduction in self-fertility after one generation of selection (from 2.32 seeds per fruit to 0.74). Busbice et al. (1975) selected for self-fertility and self-sterility in alfalfa, but no significant differences were obtained after two generations.
The increase in autogamy in the wild and cultivated P. drummondii and the decrease in autogamy in the cultivar ostensibly was due to concordant change in the level of pseudo-self-compatibility. There were no consistent changes in levels of self-pollination and protandry or in floral architecture which could account for the alterations in breeding systems. In P. cuspidata a significant reduction in autogamy was accompanied by a significant reduction in self-compatibility in one of our populations but not in the other. Again, there were no consistent changes in self-pollination, protandry, or floral architecture which could have a bearing on the reduction of autogamy which was achieved in both populations. This reduction did not have a negative impact on cross-fertility, in contrast to what has been found in other species (Wilsie 1951; Whitehead and Davis 1954; Busbice et al. 1975).
Although we achieved a selective reduction in autogamy, this shift was more difficult to achieve than a selective increase. We see this within species and at the species level. One and two cycles of selection on Salmon Beauty yielded a greater change in the high autogamy line than in the low autogamy line. The increase in autogamy in wild P. drummondii was much greater than the decrease in P. cuspidata. The additive genetic coefficient of variation for autogamy was 10 times less in P. cuspidata than in wild P. drummondii. The evolution of reduced autogamy in P. cuspidata would be hampered by a paucity of genetic variation and by anther indehiscence as seen in one of our populations.
The evolution of autogamy in wild P. drummondii ostensibly is opposed by inbreeding depression, which has been demonstrated in the greenhouse and in the field (Levin 1984; Levin and Bulinska-Radomska 1988). The force of inbreeding depression in evolutionary time is evident in the shift from partial autogamy to almost complete xenogamy in escaped cultivars of P. drummondii. The average level of autogamy in these population systems is now no different from that in other wild populations (Levin 1985, 1989; D. Waitt, pers. comm.). Cultivars suffer from inbreeding depression but not quite to the same extent as their wild progenitors (Levin 1989).
The precursor of P. cuspidata must have passed through a stage of partial autogamy, because P. cuspidata is the only autogamous Phlox (Wherry 1955; Levin 1978). Cultivars of P. drummondii are now at this point of partial autogamy, becoming so through inadvertent selection (Levin 1989). We demonstrated with Salmon Beauty that a shift from partial autogamy toward xenogamy could be readily achieved. The shift toward xenogamy in Salmon Beauty was twice as great as that achieved in the Giddings population of P. cuspidata and four times as great as that achieved in the Bastrop population. The pattern we see in Phlox may be representative of that in other phylads. Species with intermediate levels of autogamy may move more readily toward xenogamy than species which are predominantly autogamous.
The authors wish to thank R. Starr for use of his sonicator. Thanks also to J. Fritz, M. Kelly, K. Olsen, and D. Waitt for valuable comments on a previous draft of this manuscript and to E. Bract and D. Sutton for their service in the greenhouse. This research was supported by National Science Foundation grant BSR-8614142 to D.A.L.
ALLARD, R. W. 1960. Principles of plant breeding. Wiley, New York.
BARRETT, S.C. H., AND J. S. SHORE. 1987. Variation and evolution of breeding systems in the Turnera ulmifolia complex (Turneraceae). Evolution 41:340-354.
BULMER, M. G. 1980. The mathematical theory of quantitative genetics. Clarendon Press, Oxford.
BUSBICE, T H., R. Y. GURGIS, AND H. B. COLLINS. 1975. Effect of selection for self-fertility and self-sterility in alfalfa and related characters. Crop Sci. 15:471-475.
CROW, J. F., AND T NAGYLAKI. 1976. The rate of change of a character correlated with fitness. Am. Nat. 110:207-213.
DANA, M. N., AND ED. ASCHER. 1985. Pseudo-self-compatibility (PSC) in Petunia integrifolia. J. Hered. 76:468-470.
ERBE, L., AND B. L. TURNER. 1962. A biosystematic study of the Phlox cuspidata-P. drummondii complex. Am. Mid. Nat. 67:257-281.
FALCONER, D. S. 1989. Introduction to quantitative genetics. 3d ed. Longman Group Ltd., New York.
HAZEL, L. N. 1943. The genetic basis for constructing selection indexes. Genetics 28:476-490.
HENNY, R. J., AND P. D. ASCHER. 1976. The inheritance of pseudo-self-compatibility (PSC) in Nemesia strumosa Benth. Theor. Appl. Gen. 48:185-195.
HOULE, D. 1992. Comparing evolvability and variability of quantitative traits. Genetics 130:195-204.
JAIN, S. K. 1976. The evolution of inbreeding in plants. Annu. Rev. Ecol. Syst. 7:469-495.
KELLY, J.P. 1915. Cultivated varieties of Phlox drummondii. J. NY Bot. Gar. 16:179-191.
LANDE, R. 1979. Quantitative genetic analysis of multivariate evolution applied to brain: Body size allometry. Evolution 33:402-416.
LEVIN, D. A. 1975. Gametophytic selection in Phlox. Pp. 207-217 in D. L. Mulcahy, ed. Gamete competition in plants and animals. North-Holland, Amsterdam.
-----. 1976. Consequences of long-term artificial selection, inbreeding and isolation in Phlox. II. The organization of allozymic variability. Evolution 30:463-472.
-----. 1978. Genetic variation in annual Phlox: Self-compatible versus self-incompatible species. Evolution 32:245-263.
-----. 1984. Inbreeding depression and proximity-dependent crossing success in PHLOX DRUMMONDII. Evolution 38:116-127.
-----. 1985. Reproductive character displacement in Phlox. Evolution 39:1275-1281.
-----. 1989. Inbreeding depression in partially self-fertilizing Phlox. Evolution 43:1417-1423.
-----. 1993. S-gene polymorphism in Phlox drummondii. Heredity 71:193-198.
LEVIN, D. A., AND Z. BULINSKA-RADOMSKA. 1988. Effects of hybridization and inbreeding on fitness in Phlox. Am. J. Bot. 75: 1632-1639.
LINSKENS, H. E, AND K. L. ESSER. 1957. Uber ein specifische Anfurbung der Pollenschlauche im Griffel und die Zahl der Kallosepfroplen nach Selbstung and Fremdung. Naturwissenchaften 44:16.
MARTIN, F. W. 1959. Staining and observing pollen tubes in the style by means of fluorescence. Stain Tech. 34:125-128.
MESKEN, M. 1987. Mass selection for crown height in sugar beets (Beta vulgaris L.). I. Divergent selection in diploids. Euphytica 36:129-145.
OCKENDON, D. J. 1973. Selection for high self-incompatibility in inbred lines of Brussels sprouts. Euphytica 22:503-509.
PLITMANN, U., AND D. A. LEVIN. 1990. Breeding systems in the Polemoniaceae. Plant Syst. Evol. 170:205-214.
RAMMING, D. W., H. A. HINRICHS, AND P. E. RICHARDS. 1973. Sequential staining of callose by aniline blue and lacmoid for fluorescence and regular microscopy on a durable preparation of same specimen. Stain Tech. 48:133-134.
RICHARDS, R. A., AND N. THURLING. 1973. The genetics of self-incompatibility in BRASSICA CAMPESTRIS L. ssp. Oleifera Metzg. II. Genotypic and environmental modification of S locus control. Genetica 44:439-453.
RICK, C. M. 1988. Evolution of mating systems in cultivated plants. Pp. 133-147 in L. D. Gottlieb and S. K. Jain, eds. Plant evolutionary biology. Chapman and Hall, New York.
ROBACKER, C. D., AND P. D. ASCHER. 1978. Restoration of pseudo-self-compatibility (PSC) in derivatives of a high-PSC cross in Nemesia strumosa Benth. Theor. Appl. Gen. 53:135-141.
-----. 1982. Effect of selection for pseudo-self-compatibility in advanced inbred generations of Nemesia strumosa Benth. Euphytica 31:591-601.
SAS INSTITUTE, INC. 1985. SAS user's guide: Statistics. Ver. 5 ed. SAS Institute, Inc., Cary, NC.
STEBBINS, G. L. 1970. Adaptive radiation in angiosperms. I. Pollination mechanisms. Annu. Rev. Ecol. Syst. 1:307-326.
VILLEGAS, C. T., C. P. WILSIE, AND K. J. FREY. 1971. Recurrent selection for high self-fertility in vernal alfalfa (Medicago satira L.). Crop Sci. 11:881-883.
WHALEN, M.D., AND G. J. ANDERSON. 1981. Distribution of gametophytic self-incompatibility and infrageneric classification in Solanum. Taxon 30:761-767.
WHERRY, E. T. 1955. The genus Phlox. Morris Arboretum Monographs III. Philadelphia, PA.
WHITEHEAD, W. L., AND R. L. DAVIS. 1954. Self and cross-compatibility in alfalfa, MEDICAGO SATIVA. Agron. J. 46:452-456.
WILSIE, C. P. 1951. Self-fertility and forage yield of alfalfa selections and their progenies. Agron. J. 43:550-560.
YAMADA, T., H. FUKUOKA, AND T. WAKAMATSU. 1989. Recurrent selection programs for white clover (TRIFOLIUM REPENS L.) using self-compatible plants. I. Selection of self-compatible plants and inheritance of a self-compatibility factor. Euphytica 44:167-172.
|Printer friendly Cite/link Email Feedback|
|Author:||Bixby, Paul J.; Levin, Donald A.|
|Date:||Apr 1, 1996|
|Previous Article:||Inbreeding depression in four populations of Collinsia heterophylla Nutt (Scrophulariaceae).|
|Next Article:||Mating system and asymmetric hybridization in a mixed stand of European oaks.|