The evolving genetic history of a population of Lathyrus sylvestris: evidence from temporal and spatial genetic structure.
The majority of studies of population structuring examine factors responsible for genetic differentiation in space. Most populations are structured geographically at different scales, resulting from a variety of factors including mating system and dispersal capabilities. For plants, many reviews focus on correlations between life history traits and spatial genetic differentiation among populations (Brown 1979; Hamrick et al. 1979; Gottlieb 1981; Loveless and Hamrick 1984; Hamrick and Godt 1990), and spatial substructuring within populations (Epperson 1990, 1993; Heywood 1991). The breeding system, of course, strongly affects spatial structuring. Genetic differentiation is far more extensive in selfing than in preferentially outcrossing species. Ecological factors correlated with self-fertilization, such as a short life cycle and clonal growth, are thus also associated with increased spatial genetic structure (Handel 1985; Berg and Hamrick 1994). Successional stage, to a lesser extent, also influences the level of differentiation among populations. Higher levels of differentiation are found among populations of colonizing species. This may be partly due to the correlations between successional stage, breeding system, and life form, as most of the colonizing species are selfing annuals. However, in early stages of succession, populations are likely to be transient and not in equilibrium, and local extinction and colonization events are important factors that may influence spatial structuring. Thus, nonequilibrium dynamics of genetic diversity have to be considered. Theoretical models show that genetic structure of populations is affected by nonequilibrium processes and that founding effects and genetic bottlenecks are critical for the differentiation of populations in a metapopulation (Slatkin 1977; Maruyama and Kimura 1980; Wade and McCauley 1988; Olivieri et al. 1990; Whitlock and McCauley 1990). Experimental data need to be collected by considering temporal variation of population genetic structure.
Studies of genetic differentiation in time are uncommon due to the difficulty of collecting this kind of data. Such studies are usually attempted either with organisms with short generation times, such as bacteria (Caugant et al. 1981), or via comparisons of genetic diversity between juveniles and adults (Zouros and Foltz 1984; Tonsor et al. 1993). Studies involving comparisons between differentiation in time and differentiation in space are also rare. They are more often completed on animal species with short generation time (Drosophila montana: Baker 1975; Musca domestica: Black and Krafsur 1986; D. buzzati: Thomas and Barker 1990) than on plant species (Pinus ponderosa: Linhart et al. 1981; Phacelia dubia: Del Castillo 1994).
However, among plants, perennial species are good candidates for studies of temporal and spatial variation in population structure. Even when it is not possible to follow population demography during consecutive years, age determination of individuals may provide a solution (see Schaal and Levin 1976; Linhart et al. 1981). It allows us, using data on one year's study, to infer historical events in the population, as long as we are aware of the potential bias of not including individual mortality.
In this paper we analyze patterns of microdifferentiation in space and in time within a natural population of Lathyrus sylvestris L., a wild perennial legume in Europe. Information previously collected on reproductive biology, mating system, and population dynamics of this species (Bashar et al. 1987; Hossaert and Valero 1988; Valero and Hossaert-McKey 1991; Hossaert-McKey and Jarry 1992) provided an essential background for the interpretation of data on genetic structuring. By combining data on age structure and spatial genetic structure, we attempt to reconstruct the dynamics of population differentiation. In the studied population, individuals were mapped within a transect and identified for their age and their genotypes at five enzyme loci. Moreover, it was possible to determine if individuals were produced by vegetative propagation or originated from sexual reproduction. Consequently the following questions can be addressed: (1) What are the different levels of spatial demographic structuring within the population? In particular, do the individuals originated from sexual reproduction present a spatial distribution similar to those originated from clonal growth? (2) What is the pattern of recruitment within the population? Are individuals of the same age clumped in space and do levels of spatial structuring differ among age classes? (3) What is the pattern of gene flow within the population? (4) If significant heterogeneity is found for demographic and genetic characteristics, is it related to selection or to nonselective factors such as founder effects and mating among relatives? (5) What conclusions can be reached about the stability of the population structure?
MATERIAL AND METHODS
Biological Characteristics of the Species Studied
Lathyrus sylvestris is a perennial legume widely distributed in Europe. It is a densely branched bushy vine capable of clonal growth by means of an extensive system of underground rhizomes (Valero and Hossaert-McKey 1991; Hossaert-McKey and Jarry 1992). Flowers are pollinated by bumblebees, carpenter bees, and bruchid beetles (Hossaert et al. 1986; Bashar et al. 1987; Valero et al. 1987; Asmussen 1993). It shows a mixed mating system, being essentially outcrossed but self-compatible; self-fertilization can occur after insect tripping (Valero 1987). Most pollinator visitation occurs over short distances but some long-distance pollen carryover can be observed (pers. obs.), as described in other related species: L. japonicus Willd. (Asmussen 1993) and L. latifolius L. (Godt and Hamrick 1993). Consequently pollen is mainly dispersed among flowers of one individual or its proximate neighbors. Seeds are produced in the fall and are dispersed for short distances (about 1.20 m; Hossaert-McKey and Jarry 1992) by dehiscent pods (between 6 to 14 seeds per pod, mean = 12.39 [+ or -] 0.08; Valero and Hossaert-McKey 1991).
Seedlings of L. sylvestris form a taproot and a tuft of ramets from buds on a thickened underground rhizome. As the individual grows, more rhizomes are produced, extending parallel to the surface. On these rhizomes a new taproot can develop, forming a secondary connected tuft of ramets. These individualized tufts of ramets with their own taproots are called clonemates. A genetic individual can be formed of several connected clonemates (from 2 to 17; Magda 1989). In the following study, "individual" refers to the genetic individual or genet or genotype.
Field Sites and Sampling Methods
This study was conducted in 1986, in one L. sylvestris population in southwestern France (Urdos, Atlantic Pyrenees, 42 [degrees] 56 [minutes] N, 0 [degrees] 32 [minutes] W). This population is located in an abandoned (15 yr prior to the study) railway station in the Aspe valley (750 m above sea level). A previous study on patterns of spatial and temporal dynamics of the tillers and seedlings was performed in this same population (Hossaert-McKey and Jarry 1992). After that study, rhizome connections among clonemates of a single genet were revealed by destructive digging. The area studied within the population was a transect of 25 m x 7 m, oriented north-south along the level railway. The positions of all rhizomes, clonemates and individuals (including seedlings) were mapped within this area by an x,y-grid. Seedlings were easily distinguishable by the presence of cotyledons (if very young), the more rounded shape of the leaves and the absence of wings on primary stems.
Determination of the Age of the Individuals
On each clonemate, part of the taproot was sampled to study the annual growth rings for age determination. The method is based on the differential growth of the xylem cells according to seasonal variations in plant development. Cross sections of the collected taproots were immersed in chlorine, then stained with carmine and fast green to reveal differences in the cell sizes (Magda et al. 1988). Ten sections per individual were examined to avoid errors due to irregular growth. Blind tests were previously performed using 10 plants of known age with an error of 10% (an error of 1 yr for 1 individual).
Young leaves (3 to 5) were collected on the 303 clonemates identified in the transect. Leaves were ground in homogenizing buffer (0.1 M Tris-HCL, 5 mM sodium thioglycolic acid, pH 7.6). Samples were centrifuged, and the aqueous supernatant was decanted and stored at -70 [degrees] C. Among the thirteen enzymes tested, four systems were apparently variable and revealed a total of 10 presumptive polymorphic loci. These systems were scored for each individual. Five of the 10 loci were found to be polymorphic and were used in subsequent analyses. Allelic variation was visualized using standard staining schedules (Soltis et al. 1983). Loci and buffer types included: Aspartate-[Alpha]-ketoglutarate aminotransferase-1 (Aat-1), with 1.0 M Tris-HCL buffer, pH 8.0; Acid Phosphatase-1 (Acph-1) with 0.05 M sodium acetate buffer, pH 5.0; Esterase-1 and -4 (Est-1 and Est-4) with [Alpha] and [Beta] naphthyl acetate as substrate and 1.0 M phospate buffer, pH. 6.0; and Leucine amino transaminase-1 (Lap-1) with 1.0 M phosphate buffer, pH 6.0. Enzyme electrophoresis was performed using slab polyacrylamide gel electrophoresis (Ornstein and Davis 1962, modified by Gasquez and Compoint 1976).
Analysis of Spatial Distribution
Except for the analysis of the spatial distribution of clonemates, all the following analyses were concentrated at the individual level. In these analyses the age of a connected individual was determined by the age of its oldest clonemate. For both demographic and genetic characters, spatial autocorrelation analyses were performed (Cliff and Ord 1973). Spatial autocorrelation measures the dependence of a character at each location on the character values at other nearby locations, with no a priori assumptions regarding the scale of the spatial patterns.
Spatial Distribution of Clonemates. - The transect map was transformed into a grid of 0.5 m x 0.5 m quadrats. Choice of quadrat size for analysis was based on previous results on ramet distribution (Hossaert-McKey 1988; Hossaert-McKey and Jarry 1992) and on the necessity of regrouping quadrats in nested blocks of increasing size. According to its position in space, each individual and each clonemate was attributed to a specific quadrat.
The test of the null hypothesis of a random distribution of clonemates among the basic quadrats of the grid was performed by using the index of dispersion Id (variance to mean ratio); Id(n - 1) can be compared to a [[Chi].sup.2] with n - 1 degrees of freedom (Skellman 1952), n being the sample size.
Patterns in spatial distribution of the clonemates for different scales within the grid was examined by using Geary's Index (Geary 1954; Cliff and Ord 1973) applied to spatial autocorrelation matrices (Chessel 1981). The basic quadrats of the grid (0.5 m x 0.5 m, in our case) are noted as [P.sub.i]. The contiguity relationship matrix V is defined by [V.sub.ij] = [V.sub.ji] = 1 if quadrats [P.sub.i] and [P.sub.j] are neighbors (i.e., in the same block), and [V.sub.ij] = [V.sub.ji] = 0 otherwise. The values of a variable x in quadrats [P.sub.i] and [P.sub.j] are noted as [x.sub.i] and [x.sub.j]. For each possible block size of the grid (i.e., resulting from nesting of basic quadrats), we computed the quantity Z = [H.sub.V]/[H.sub.T] with
[Mathematical Expression Omitted]
and the Geary Index IZ is:
IZ = (E(Z) - Z)/[square root of V(Z)].
Under the assumption of random spatial distribution, [H.sub.V] and [H.sub.T] would be approximately equal and the mean E(Z) = 1 (Cliff and Ord 1973). If the values of neighbor quadrats are correlated, the observed values of [H.sub.V] and Z are low and IZ becomes significantly positive (the level being 1.96 for [Alpha] = 0.05). Variance of Z is given by Cliff and Ord (1973).
Spatial Distribution of Age and Genotype of Individuals. - Allozyme genotype and age of each individual located within the transect are also known. With these data we were able to compare the pattern in the spatial distribution of the clonemates with the distribution of the different age classes within the transect and with the spatial distribution of the genotypes.
Spatial autocorrelation analysis (Sokal and Oden 1978a,b; Cliff and Ord 1981) of age and genotype distribution was performed using the SAAP program developed by Wartenberg (1989). This procedure has recently been applied to the study of genetic variation within plant populations (see for review Heywood 1991; Epperson 1993) but, to our knowledge, has never been applied to the study of age structure in plants.
The age of the individuals, as revealed by examination of the annual growth rings, was used directly as a discrete quantitative value. For the genetic data, genotypic results were coded in such a way that single individuals received allele frequency values of 0.0, 0.5, or 1.0 for each allele of every locus. Alleles with a frequency less than 0.01 were excluded. Only one allele was considered at diallelic loci. Moran's Index (Moran 1950) was calculated on these frequencies for each locus and for each age class at the different distance classes. Spatial structuring was tested by comparing Moran's I to its expected value under the null hypothesis of a random distribution of genotypes and age classes. Values of Moran's I higher than E(I) indicate that plant pairs within a distance class are more similar than expected by random distribution.
Patterning in age structure and in allelic frequencies was studied by examining the behavior of the autocorrelation statistics with increasing interindividual distance. This examination was made through the construction of correlogram graphs of Moran's I versus distance class. Euclidian distances between the two individuals of all the possible pairs were calculated, permitting assignment of all pairs to distance classes. Distance classes were defined using an equal number of point pairs within each distance class. The intercept of Moran's I correlogram with the abscissa (E(I) or median) may be used as an estimate of the mean size of homogeneous patches. Overall significance of correlograms was assessed using Bonferroni's corrected P-values (Sakai and Oden 1983). When the mean value of Moran's I for a combination of several loci was used, the experimentwise error technique of Bonferroni was no longer appropriate. Runs tests above and below the median (Sokal and Rohlf 1981, p. 782) were applied. This nonparametric procedure tests whether positive and negative values of I are randomly distributed along the correlograms.
Hierarchical analyses were performed to compare the genetic structure for a subdivided population in space or time. Two kinds of subdivisions were thus examined: (1) different spatial population subdivisions within the transect; and (2) different age classes of the individuals. The population subdivisions were established using the scale revealed for the spatial heterogeneity in the demographic structure and for the spatial distribution of age classes and genotypes.
Allele frequencies for the different loci were calculated for different spatial population subdivisions and for different age classes using the program GENESTATS (Black and Krafsur 1985). Differences in allele frequencies among the classes were compared by a G-test (Sokal and Rohlf 1981, p. 744) using the critical values of the [[Chi].sup.2] distribution based on Sidak's multiplicative inequality to control experimentwise type I error (Sokal and Rohlf 1981, p. 731). When sample size within a class was too small, data for several classes were grouped for comparison tests.
F-statistics were used to identify sources of genetic differentiation in spatial or age structure of the population. Departure from random mating for each locus in subpopulations was tested using F-statistics weighted by allelic frequencies (Weir and Cockerham 1984). The GENESTATS program was used to calculate and compare sample estimates of F-statistics over all loci. When multiple comparisons were performed, Sidak's corrected procedure was applied.
Spatial Demographic Structure and Recruitment
Levels of Spatial Structuring of Clonemates and Seedlings. - Based on existing connections, a maximum of 274 genets of L. sylvestris were present in the study transect; of these, 47 were seedlings. The others were established plants composed of a single tuft of ramets (i.e., a single clonemate: 212 plants) or two or more connected clonemates (44 tufts of ramets, connected in groups of 2 or 3 clonemates comprising a maximum of 15 genets). Using allozymes it was also possible to confirm that each clonemate of a connected individual exhibited the same genotype. Of the 212 single clonemate "individuals," only 15 were in close proximity but not connected to individuals of the same genotype. We cannot exclude the possibility that some separate individuals may have formerly been connected clonemates. Thus the actual number of genets may be somewhat lower than our maximum estimates.
The spatial distribution of seedlings, older clonemates (79 years) and all other clonemates is given in Figure 1. Clonemates are found preferentially in a 3-m-wide median zone oriented north-south (coordinate as 3 m [less than] y [less than] 6 m, [ILLUSTRATION FOR FIGURE 1 OMITTED]).
The analysis of spatial structure is focused on this median zone, and 28 clonemates located outside of this median zone (24/28 are grouped in distinct patches within the transect, [ILLUSTRATION FOR FIGURE 1 OMITTED]), were removed from the analysis.
The value of the dispersion index (I = 2.18, P [less than] 0.001) allows rejection of the hypothesis of a Poisson distribution of clonemates among the basic quadrats (0.5 m x 0.5 m). The spatial distribution thus significantly departs from random and clonemates exhibit a clumped distribution within the transect.
The spatial autocorrelation matrix (6 lines x 50 columns matrix, with 0.5 m x 0.5 m basic quadrats) shows high values only in cases of grouping along the lines, for four quadrat sizes (0.5, 1.0, 1.5, and 3.0 m) of the nested series. Most of the spatial structuring is observed along the x-axis. For each of the four quadrat sizes, values of Geary's I (or [I.sub.Z]) are given as a function of increasing sizes of blocks formed along the transect [ILLUSTRATION FOR FIGURE 2 OMITTED]. Over the entire range of block sizes, four maxima in Geary's I are revealed by the correlograms. Significant values are observed for block sizes of 0.5-3.0 m, 56 m, 12-13 m, and 25 m. The first significant peaks in Geary's I occur between 0.5 and 3 m along the x-axis (north-south axis) of the transect. The multiplicity of peaks at this scale results from small variations in the size of the aggregates. In fact, they should be considered as one large peak corresponding to the average modular unit of the transect. The second significant value at 5-6 m divides the transect in four groups or subpopulations along the same axis. The peak at 12-13 m in the correlogram divides the transect into two parts along the x-axis. A last significant value is observed for 25-m-long blocks. It is the only structuring occurring along the y-axis (east and west parts of the transect). This structuring is due to a higher density of the clonemates at the bottom (east) margin of the transect.
To compare spatial distributions of seedlings and connected clonemates, the transect was divided into two sub-groups along the x-axis (subpopulation 0-12.5 m, and subpopulation 12.5-25 m). This subdivision corresponds to the third peak in the Geary's I observed in the previous analysis. Each subpopulation consisted of more than a hundred clonemates. Seedlings and connected clonemates show significant opposite distribution patterns between the two subpopulations (G-test using Williams' correction, P[14,028; df = 1] [less than] 0.001). In the first 12.5 m (left side of the transect), connected clonemates are more numerous than seedlings (33 versus 17), while in the second part of the transect, seedlings are more frequent (30 versus 11 connected clonemates).
Spatial Distribution of Individuals of Differing Age. - The youngest individuals are the seedlings (age class 0), and the oldest individual is 9 yr old. The two oldest individuals (8 and 9 yr old) are found at the extreme north and south parts of the transect.
Moreover, an age class pyramid [ILLUSTRATION FOR FIGURE 3 OMITTED] is significantly different between the two parts of the transect (Mann-Whitney test, P = 0.0005). The oldest part of the transect occurs within the southern part [ILLUSTRATION FOR FIGURE 1 OMITTED]. Mean ages are respectively 4.07 yr (SE = 0.24) in the first 12.5 m and 3.00 yr (SE = 0.18) in the second part of the transect.
The results of the autocorrelation analysis of ages are summarized in the correlogram in Figure 4a. The overall test of the correlogram gives a highly significant value for spatial autocorrelation (Bonferroni corrected P-value [less than] 0.0001). Thus, the spatial distribution of ages in the population is not random. Eight of the 10 values for Moran's I at different scales [ILLUSTRATION FOR FIGURE 4A OMITTED] were found to depart significantly from E(I). The correlogram shows that individuals of the same age are significantly grouped together at distances lower than 4 m (Moran's I [greater than] 0, [ILLUSTRATION FOR FIGURE 4A OMITTED]). The intercept occurs at a distance between 5 and 7 m [ILLUSTRATION FOR FIGURE 4A OMITTED] indicating the size of homogeneous patches. This distance corresponds to the second scale of spatial structuring previously found for the individuals. The lowest value of Moran's I is found at a distance between 11 and 12 m, which corresponds to the limit of the two subpopulations previously described.
Spatial Genetic Structure
Spatial Autocorrelation of Alleles. - Two of the five loci (Aat-1 and Lap-1) show a moderate level of polymorphism. The three other loci (Est-4, Est-1 and Acph-1) are less polymorphic (Table 1).
All the loci studied, except locus Acph-1, are diallelic, so that only one allele is considered to make the correlogram. For locus Acph-1, allele 1 is found at a frequency lower than 0.01 (Table 1) and was excluded from the analysis, making this locus also diallelic.
Because values of Moran's I obtained with locus Aat-1 are far higher than those obtained with the four other loci, results obtained with this locus were analyzed separately. The correlogram obtained for locus Aat-1 allelic frequencies [ILLUSTRATION FOR FIGURE 4B OMITTED] [TABULAR DATA FOR TABLE 1 OMITTED] shows a remarkable cline of decreasing values of Moran's I with increasing distance.
Values of Moran's I range from 0.77 to -0.89 and all of them except one [ILLUSTRATION FOR FIGURE 4B OMITTED] are found to depart significantly from E(I). Positive spatial autocorrelation is found for short distances and negative ones for distances greater than 9 m. The overall significance of the correlogram is very high according to the Bonferroni criterion (P [much less than] 0.0001). The intercept occurs at distances between 8 and 9 m, indicating the average size of homogeneous patches.
For the other loci, 1 of the 4 correlograms demonstrates overall statistical significance according to the Bonferroni criterion (Est-1: P = 0.006). A total of 3 of the 40 Moran's I statistics were found to differ significantly from E(I). The mean correlogram over the four loci is presented in Figure 4c. In this figure, means over the four Moran's I and standard errors are given for each distance class. The overall significance of the correlogram was tested by a runs test above and below the median.
The runs test demonstrates a significant departure from random distribution of means of Moran's I along the correlogram (P = 0.008). The means of Moran's I over the four loci are positive at shorter distance classes and become negative for distances greater than 7 m. The size of the homogeneous patches is 6-8 m, as shown by the intercept of the correlogram with the median [ILLUSTRATION FOR FIGURE 1C OMITTED]. Thus there is clearly a significant nonrandom spatial distribution of alleles for nearly all the loci studied.
Variation of Allelic Frequencies among Spatial Subpopulations. - According to the results obtained from the spatial demographic and spatial allelic patterns, the population was divided into four subpopulations each 6 m wide (corresponding to the second scale of spatial structuring of individuals [ILLUSTRATION FOR FIGURE 2 OMITTED] and to the distance defined by the intercepts of age [ILLUSTRATION FOR FIGURE 4A OMITTED] and by means for correlograms over four loci [ILLUSTRATION FOR FIGURE 4C OMITTED]). Variation in allelic frequencies among subpopulations is significant for three of the four loci tested (Table 1) when no correction for multiple tests is made. The number of significant tests reduces to two (loci Aat-1 and Lap-1) when the Sidak correction is applied. The lack of significant variation among subpopulations for the other loci is not surprising, given the overall low frequency of one of the two alleles in each case (Table 1). Of the two loci with significant subpopulation differentiation, locus Aat-1 reveals a very strong difference in allele distribution along the transect [ILLUSTRATION FOR FIGURE 5 OMITTED]. The steep cline observed at this locus is explained by the regular decrease of allele-1 frequency along the transect (0.92 in the first 6 m, and 0.17 in the last distance class) and reciprocally for the other allele. These differences in spatial structure over the four distance classes were used to determine the boundaries between subpopulations in the F-statistics.
F-Statistics Analysis of Spatial Genetic Differentiation. - The overall deficit of heterozygotes in the population was found to be very high (Table 2, mean [F.sub.IT] = 0.39, two-tailed test: P [less than] 0.001) and nearly the same for all the loci except for the least polymorphic locus, Acph-1. There is a significant genetic differentiation among the four subpopulations (mean [F.sub.ST] = 0.23, one-tailed test: P [less than] 0.05), as expected from the results presented in the previous paragraph. Indeed, the highest value is obtained for locus Aat-1 ([F.sub.ST] = 0.49). Genotypic structure is not only explained by a difference in spatial distribution of allelic frequencies among subpopulations, but also by a significant heterozygote deficit within subpopulations (mean [F.sub.IS] = 0.21, two-tailed test: P [less than] 0.05). Two of the three more polymorphic loci, Est-4 and Lap-1, show a greater value of [F.sub.IS] than [F.sub.ST], while for locus Aat-1 the differentiation among subpopulations explains by itself the global heterozygote deficit found in the population (Table 2).
Variation in Allele Frequencies among Age Classes. - To test for differentiation in allele frequency distribution, several age classes were pooled (1-2 yr; 3-4 yr; 5-9 yr) and seedlings were excluded from the subsequent analyses because of their low frequency. Differences in allelic frequencies among the three sets of age classes are significant for loci Aat-1 and Lap-1, but only for locus Aat-1 when the Sidak correction is applied (Table 3).
At Aat-1, the frequency of allele 1 is higher in older individuals. The same trend is observed for allele 4 of locus Lap-1, but it is not significant. Genetic differentiation among age classes is thus revealed by this study. However, the patterning in age structure is much less marked than that of spatial differentiation along the transect.
F-Statistics Analysis of Genetic Differentiation among Age Classes. - The F-statistics were performed on the same three sets of age classes. As expected, the overall population heterozygote deficit is similar to the previous values obtained for the spatial structuring (Table 4, mean [F.sub.IT] = 0.39, P [less than] 0.001). The nonequality between both [F.sub.IT] values is due to the removal of the seedlings from the analysis. There is a low but significant differentiation among age classes (mean [F.sub.ST] = 0.03, P [less than] 0.01), as already mentioned in the previous paragraph. The highest values of [F.sub.ST] are obtained for the loci Aat-1 and Lap-1. However, results show that the heterozygote deficit is mainly explained by a high heterozygote deficit within age classes (Table 4, mean [F.sub.IS] = 0.37, P [less than] 0.001).
TABLE 2. Mean F-statistics over all loci (Weir and Cockerham 1984). The subgroups are the four spatial subdivisions in the population (transect divided in 6-m blocks along the x-axis). Locus [F.sub.IS] [F.sub.ST] [F.sub.IT] Aat-1 -0.092 0.495 0.449 Lap-1 0.326 0.044 0.356 Est-4 0.414 0.013 0.422 Est-1 0.404 0.003 0.405 Acph-1 -0.032 0.010 -0.022 Mean 0.211(*) 0.229(*) 0.391(**) SE 0.059 0.087 0.020 SE: standard error of mean. * Significantly different from zero, P [less than] 0.05 (two-tailed test for [F.sub.IS] and one-tailed test for [F.sub.ST]). ** Significantly different from zero, P [less than] 0.01 (two-tailed test).
Heterozygote deficit was calculated within each age class (Table 5) to determine whether age classes contribute equally to the high observed value of [F.sub.IS] (Table 4). The mean [F.sub.IS] values over loci do vary among age classes. [F.sub.IS] is not significantly different from zero for the older individuals (5-9 yr old, P = 0.12) while significant heterozygote deficit is observed in the other age classes using Sidak correction. These results are found for three of the five loci (Lap-1, Est-4, and Est-1; Table 5). At locus Aat-1, [F.sub.IS] values are always high whatever the age class. In contrast, no heterozygote deficit is revealed for the less polymorphic locus Acph-1.
The results illustrate how the "rapprochement" of demographic and genetic approaches leads to a much better understanding of the patterns of structure observed in this population of L. sylvestris. The major events in the history of the population can be reconstructed using the information about the demography and genetic structure of the population.
Occurrence of Different Scales of Spatial Substructuring within the Population: Comparison of Demographic and Genetic Characteristics
A significant clumped spatial structuring exists among either clonemates [ILLUSTRATION FOR FIGURE 2 OMITTED] or individuals [ILLUSTRATION FOR FIGURES 4A, B, C OMITTED] of the transect. In this species which is able to reproduce by clonal growth, several scales of clumping are revealed by our study [ILLUSTRATION FOR FIGURE 2 OMITTED]. As already shown in other plant species (see for reviews Cook 1983; Hamrick and Godt 1990; Heywood 1991), clonal growth induces pronounced patchy structure in plant populations. In this studied population, the smallest aggregate corresponds to the size of the genet (principal clonemate and its connected secondary clonemates). The size of the genet (0.5-3.0 m) is identical to the size found in other [TABULAR DATA FOR TABLE 3 OMITTED] populations and for the closely related specie L. latifolius (Hossaert-McKey 1988; Hossaert-McKey and Jarry 1992).
Two other levels of spatial structuring are revealed by the analysis. They correspond to aggregation among individuals rather than among clonemates. The most striking result is the consistency with both scales detected in the transect for demographic and genetic characteristics. First, spatial clumping of clonemates exists at 5-6 m [ILLUSTRATION FOR FIGURE 2 OMITTED]. At this scale, there is evidence of spatial organization among individuals in autocorrelation analyses of both age classes and allele frequencies. Indeed, the size of a homogeneous patch was found to be 5-7 m, for the age classes, and 6-8 m for the genetic structure [ILLUSTRATION FOR FIGURES 4A, C OMITTED]. Information from these different approaches reflects the occurrence of a spatial structure at a scale of ca. 6 m. This congruence is consistent with our previous knowledge of the pollination biology of the species (Hossaert et al. 1986; Bashar et al. 1987; Valero et al. 1987), and together these data suggest that the spatial structuring at an average of 6 m may correspond to the neighborhood size of the population (cluster of related individuals). The implications of these results will be discussed later.
TABLE 4. Values of F-statistics over all loci (Weir and Cockerham 1984). The subgroups are the different age classes: 1-2 yr, 3-4 yr, and 5-9 yr. Locus [F.sub.IS] [F.sub.ST] [F.sub.IT] Aat-1 0.423 0.052 0.453 Lap-1 0.263 0.029 0.284 Est-4 0.478 0.006 0.481 Est-1 0.532 -0.006 0.529 Acph-1 -0.017 0.001 -0.016 Mean 0.373(***) 0.031(**) 0.393(***) SE 0.026 0.007 0.028 SE = standard error of the mean. ** Significantly different from zero, P [less than] 0.01 (one-tailed test). *** Significantly different from zero, P [less than] 0.001 (two-tailed test).
Second, the clumping of clonemates round at 12-13 m [ILLUSTRATION FOR FIGURE 2 OMITTED] reveals interesting features about the genetic history of the part of the population included in our transect. Again, there is a congruence between the results obtained in individuals for age classes and genotypic data. For age classes, the minimum autocorrelation index is observed at 11-12 m [ILLUSTRATION FOR FIGURE 4A OMITTED], showing that at these distances individuals have the highest probability of being of different ages when random pairs of individuals are compared. For the genotypic analysis, the most interesting result is given by data at locus Aat-1. At this locus, the intercept of Moran's I correlogram around 9 m [ILLUSTRATION FOR FIGURE 4B OMITTED] indicates the size of patches of individuals of rather homogeneous genotypic composition. A closer examination of genotype distribution along the transect reveals the occurrence of two patches at the two extremes (north and south) of the transect, which differ greatly by their allelic frequencies at locus Aat-1 [ILLUSTRATION FOR FIGURE 5 OMITTED].
TABLE 5. Variation of [F.sub.IS] among age classes (Weir and Cockerham 1984). The significance of mean [F.sub.IS] values was tested using critical values of Student's t-distribution based on Sidak multiplicative inequality (Sokal and Rohlf 1981, p. 242). Age classes Locus 1-2 yr 3-4 yr 5-9 yr Aat-1 0.604 0.606 0.556 Lap-1 0.271 0.425 -0.087 Est-4 0.723 0.521 -0.073 Est-1 0.789 0.657 -0.019 Acph-1 not done -0.033 -0.011 Mean 0.543(**) 0.511(***) 0.293(****) SE 0.051 0.040 0.147 SE = standard error of the mean. ** Significantly different from zero, P [less than] 0.01 (two tailed test). *** Significantly different from zero, P [less than] 0.001 (two tailed test). **** Not significantly different from zero, P [greater than] 0.10.
Figure 6 summarizes the combinations of information given by the genetic and demographic results. In this figure, observed frequencies of the three Aat-1 genotypes ([ILLUSTRATION FOR FIGURE 6A OMITTED] for homozygotes 11, [ILLUSTRATION FOR FIGURE 6B OMITTED] for homozygotes 22, and [ILLUSTRATION FOR FIGURE 6C OMITTED] for heterozygotes 12) are plotted as a function of age class for both parts of the transect. In the two halves of the transect the oldest individuals are respectively 9 and 8 yr old and each is homozygous for a different allele. Heterozygous individuals appear only 2 or 3 yr after the establishment of the homozygotes [ILLUSTRATION FOR FIGURE 6C OMITTED]. Even if the historical reconstitution of this sequence could be biased (our data do not include mortality), this scenario strongly suggests founder effects. These founder effects are easily detected because of the arrival of two different homozygotes at each end of the transect; as a result, we have the exceptional opportunity to follow the different steps of the colonization of this part of the population (see discussion below).
Spatial Demographic Structure and Recruitment
In this study it was possible to compare the spatial distributions of individuals originated from sexual reproduction (seedlings) and from clonal growth (connected clonemates). Moreover, additional data on the age and genotypes of these individuals suggest hypotheses regarding patterns of recruitment in this population.
An interesting result is the contrasting distribution between the two parts of the transect in (1) age class pyramids [ILLUSTRATION FOR FIGURE 3 OMITTED]; (2) spatial distribution of the seedlings compared to connected clonemates; and (3) spatial distribution of genotypes at the Aat-1 locus [ILLUSTRATION FOR FIGURES 5 AND 6 OMITTED]. The oldest individuals are found in the left side of the transect (southern part), where the density of connected plants and the density of ramets are also significantly higher (Hossaert-McKey and Jarry 1992), and where most of the genotypes examined at locus Aat-1 are homozygotes for the allele 1 [ILLUSTRATION FOR FIGURE 5 OMITTED]. Two main hypotheses can be advanced to explain the observed negative correlation between vegetative growth and seedling establishment. First, these differences could result from variations in conditions for clonal extension along the transect (Cook 1985; Salzman 1985; Sutherland and Stillman 1988; Navas and Garnier 1990). However, microhabitat conditions are likely to be homogeneous along the studied transect because the population is located on the rock ballast of a levelled railway track and no shading plants existed yet in this young population. Consequently, this hypothesis does not seem to be appropriate to our field situation. Second, this pattern of spatial structuring could be the result of differences in age structure between the two parts of the transect. Lathyrus sylvestris is a partially clonal species characterized by limited lateral growth and module dispersal leading to compact clumps of ramets ("phalanx type," Lovett-Doust 1981), which increase in size with time. This increase in density is reinforced by the fact that in clonal species (Cook 1985) and in particular in L. sylvestris (Magda 1989; Jarry et al. 1990), the probability of producing connections increases with age. The observed spatial correlation between the frequency of connected clonemates and ageing of the population could then be easily explained. Moreover, seedling establishment occurs generally around the mother plant (Hossaert-McKey and Jarry 1992) and seedling survivorship should be more difficult in places where density of connected clonemates is high (Schellner et al. 1982; Eriksson and Bremer 1993). Our data strongly suggest that the left side of the population is older and established by one or few individuals with the Aat-11 genotype [ILLUSTRATION FOR FIGURE 5 OMITTED]. Under these circumstances this side of the population could expand both vegetatively and sexually, producing mainly Aat-11 individuals. When the second colonization (by the Aat-22 individuals) at the other end of the transect occurred later, the pollen pool would have been strongly biased toward the Aat-1 allele; hence the higher frequency of heterozygotes and lower frequency of Aat-22 individuals in the younger age classes at the right side of the population [ILLUSTRATION FOR FIGURE 5 OMITTED].
Potential Sources of Spatial Genetic Variation
Plant populations are often characterized by the occurrence of spatial substructuring of genotypes (see for reviews Levin 1988; Epperson 1990; Heywood 1991). However, assessing which factors (environmental mosaic, mating system, founder effects or genetic drift) are responsible for such a structure is an active area of research. In the current study, two main factors (founder effect and mating system) seem to be responsible for the two scales of spatial genetic variation revealed by the spatial autocorrelation analyses.
The founder effect is mainly revealed by spatial variation in allelic frequency at the Aat-1 locus. At the Lap-1 locus, a less pronounced but significant spatial differentiation does occur, while at the three other loci spatial differentiation is not significant (Table 1). Such a discrepancy among loci has been already observed within populations of other plant species (Liatris cylindracea: Sokal and Oden 1978b; Ipomea purpurea: Epperson and Clegg 1986; Delphinium nelsoni: Waser 1987; Psychotria nervosa: Dewey and Heywood 1988; and Banksia spinula: Carthew 1993). Variation in allelic diversity among loci may, to some degree, account for the observed discrepancy among loci, as the power of statistical tests is strongly reduced in loci exhibiting low levels of polymorphism (e.g., Acph-1). However, Slatkin and Arter (1991) pointed out that "if a population is recently invaded by migrants from a population in which the frequencies of alleles at one locus differ from those of the resident population, transient clines at those loci will be established." Linkage disequilibrium analysis has been used to detect associations between genotypes that have not decayed because of the age of the population. Surprisingly, only a limited trend for linkage disequilibrium in spatial subpopulations between locus Aat-1 and the other loci emerged from the data (Valero unpubl. data).
For the other loci studied, F-statistics analysis among spatial subpopulations (Table 2) suggests that the major cause of heterozygote deficit in the overall transect is probably associated with inbreeding within the subpopulations. In these clusters, crosses between related individuals can be reasonably presumed because of the concordant information given by the analysis of spatial distribution of ages, as mentioned earlier. As L. sylvestris exhibits a mixed mating system, with preferential outcrossing (Valero 1987; Valero et al. 1987), the high values of [F.sub.IS] can be due to both crosses among relatives and selling. The observed weighted mean [F.sub.IS] value over all loci excluding Aat-1 is 0.330 (SE = 0.023), which leads at inbreeding equilibrium to an apparent selfing rate of 0.50 according to Haldane's (1924) calculation. A direct estimation of selfing rate from crossing experiments (Valero et al. 1987) gives a value of 0.23 (SE = 0.18) in this species. This direct measurement is similar to multilocus selfing estimates obtained from allozymic data in a closely related species (Lathyrus latifolius, Valero 1987; Godt and Hamrick 1991, 1993). Thus, although the assumption of inbreeding equilibrium is surely not achieved in this L. sylvestris population, these selfing estimates suggest that biparental inbreeding contributed at least 27% to the apparent rate of selfing. This high rate of biparental inbreeding suggests the occurrence of limited gene flow. First, restricted foraging of the main pollinators (bumblebees, honeybees, and bruchid beetles: Hossaert et al. 1986; Valero et al. 1987) is likely to occur among flowers of the same individual or its closest neighbors because, as revealed by this study, the size of a genet ranges from 0.5 to 3.0 m in diameter. In addition, after pod dehiscence, most seeds are dispersed only short distances from the maternal plants (Hossaert-McKey 1988; Hossaert-McKey and Jarry 1992). Consequently, initial inbreeding caused by the founder effect is likely to be reinforced because neighbors are likely to be related individuals.
Comparison with other perennial species pollinated by insects reveals similarly restricted neighborhood size (Gliddon et al. 1987; Waser 1987; Dewey and Heywood 1988; Schnabel and Hamrick 1990; Perry and Knowles 1991). However, the results presented in this paper are of particular interest for the following reasons: (1) The transect studied was quite homogenous and selective effects are not likely to occur, in contrast to the results obtained on sugar maple stands (Perry and Knowles 1991). (2) All the individuals in the transect were mapped and sampled, so the smallest sample scale was used. This enabled us to reject the hypotheses of an undetected finer scale structuring, such as presumed for Psychotria nervosa (Dewey and Heywood 1988). (3) Pollination biology and life history of the studied species were well known and were very useful in interpretation of the genetic structure. In plant populations, genetic drift and selection interact in a complex process to affect genetic structure, and spatial differentiation may occur in many different ways (Levin 1988). Moreover there is a continuous feedback between genetic structuring and demographic processes (Levin 1988). Consequently, in plant populations, changes occur over time and populations may rarely reach equilibrium. Studies combining spatial and temporal variation are very useful in understanding the time frame of such genetic changes.
Sources of Temporal Genetic Variation
Results of F-statistics analyses of genetic differentiation among age classes vary with the locus considered. At the Aat-1 locus, genetic differentiation exists among age classes, while it is not significant for the other loci (Table 3). This result illustrates once again the occurrence of the founder effects associated with the Aat-1 locus. Nevertheless, at this locus genetic differentiation among age classes ([F.sub.ST] = 0.05, Table 4) is far less pronounced than spatial genetic differentiation ([F.sub.ST] = 0.50, Table 2). A closer look at [F.sub.IS] variation among age classes reveals an increase in heterozygote deficit within the youngest cohorts of individuals for most loci except for Aat-1 (Table 5). This locus shows a large and constant [F.sub.IS] value whatever the age class, which can be easily understood as a strong spatial Wahlund's effect. The observation of variation in heterozygote deficit with aging of the population has been made in mussels (Zouros and Foltz 1984), and in various plant species (Liatris cylindracea: Schaal and Levin 1976; Plantago lanceolata: Tonsor et al. 1993; and Phacelia dubia: Del Castillo 1994).
Two main hypotheses can be advanced to account for such a result. First, a selection hypothesis can be invoked, involving a heterozygote advantage or inbreeding depression. If homozygote individuals are more inbred than heterozygotes, they may suffer from a lower fitness compared to heterozygotes. Consequently, they may die off faster, leading to reduced [F.sub.IS] values with increasing age. This hypothesis is the one generally accepted in the literature (Schaal and Levin 1976; Zouros and Foltz 1984; Tonsor et al. 1993; del Castillo 1994) and may account for the temporal genetic variation revealed in this study. Such selection against homozygotes has already been proposed for a very closely related species, Lathyrus latifolius, from studies of patterns in gene movement (Godt and Hamrick 1993). Nevertheless, critical reviews (Charlesworth and Charlesworth 1987; Mitton 1993; Uyenoyama et al. 1993) illustrate that inbreeding depression, depending on the species, may affect very different stages in plant life cycles, from early seed formation (e.g., Templeton and Read 1983) to later reproduction (e.g., Schemske 1983; Nason and Ellstrand 1995). From our study it is not possible to detect the exact timing of inbreeding effects.
However, a second hypothesis involving neutrality may also be advanced. It posits that, in addition to the initial inbreeding due to the founder effect, the density of related individuals increases with limited gene flow within a cluster over time, leading to an increased frequency of crosses between related neighbors. Many theoretical investigations have described such a process of isolation by distance (Wright 1943; Rohlf and Schnell 1971; Turner et al. 1982; Slatkin 1993). The establishment of such structuring is especially easy in insect pollinated species because pollen is often dispersed only short distances and gene flow is restricted (Levin and Kerster 1974; Hamrick and Godt 1990; Heywood 1991) as previously discussed for L. sylvestris.
As found by Linhart et al. (1981), genetic structuring in time is less pronounced than spatial differentiation. Similarly, no consistent pattern of linkage disequilibrium is observed among age classes, in contrast to the limited trend for linkage disequilibrium in spatial subpopulations between locus Aat-1 and the other loci (Valero unpubl. data).
Why Is Differentiation in Time Smaller than Differentiation in Space?
As discussed above, the studied population is experiencing demographic and genetic variation in time, suggesting that it is not at equilibrium. The following scenario is proposed to explain the genetic dynamics of the population. On the one hand, spatial structuring is becoming less marked due to the recombination of founder genotypes at the Aat-1 locus and thus decay of the founder effect. On the other hand, as establishment of new individuals increases, a new spatial structure emerges due to mating between relatives (6-m-wide patch of clumped individuals of revealed by all loci except Aat-1).
Theoretical work has shown that establishment of spatial differentiation by restricted gene flow is generally a rapid process (Turner et al. 1982; Sokal and Wartenberg 1983; Slatkin 1993). It seems that in our case, we are observing the first steps of the process before equilibrium has been reached. The population studied is located in a railway station abandoned for only 15 yr, supporting our assumption that the establishment of the population is very recent. As population structuring is still in process, spatial differentiation is difficult to reveal, at least for loci other than Aat-1. Moreover, as temporal variation is measured between few overlapping generations, it is more difficult to detect the occurrence of changes in genetic structure in time than in space, the latter being the result of several generations of differentiation (Charlesworth 1980).
This study provides an example of a population of L. sylvestris undergoing a dynamic process of spatial differentiation as a consequence of founder effects and subsequent restricted gene flow. It could be that this population corresponds to a peculiar situation. However, this legume is regularly found in disturbed habitats (railways tracks, road embankments, [see Hossaert and Valero 1988] or in graveyards [Godt and Hamrick 1993]), suggesting that situations of nonequilibrium dynamics should be the rule in this colonizing plant species.
The authors are grateful to the following persons for their assistance in the field studies: G. Tortay, A. Chaib and P. Blanchard, or in conducting the electrophoretic work: B. Caron and A. Youssef. For helpful comments and discussions on the manuscript, we thank D. McKey, Y. Linhart, K. Spitze, F. Bonhomme, C. Williams, and D. Waller. W. C. Black, and E. S. Krafsur kindly provided a copy of GENESTATS for calculation of the F-statistics. Financial support was provided by the CNRS (ATP to MH-M and MJ), and by the Ministere de l'Enseignement superieur (Doctoral fellowship to DM). MH-M and MV want also to thank the University of Miami, Department of Biology for its hospitality during the preparation of last drafts of the manuscript.
ASMUSSEN, C. B. 1993. Pollination biology of the sea pea, Lathyrus japonicus: Floral characters and activity and flight patterns of bumblebees. Flora 188:227-237.
BAKER, W. K. 1975. Linkage disequilibrium over space and time in natural populations of Drosophila montana. Proc. Nat. Acad. Sci. USA 72:4095-4099.
BASHAR, A., G. FABRES, M. HOSSAERT, M. VALERO, AND V. LABEYRIE. 1987. Bruchus affinis and the flowers of Lathyrus latifolius: An example of the complexity of relations between plants and phytophagous insects. Pp. 189-194 in V. Labeyrie et al., eds. Insect-plant relationships. Dr. W. Junk Publishers, Dordrecht, Netherlands.
BERG, E. E., AND J. L. HAMRICK. 1994. Spatial and genetic structure of two sandhills oaks: Quercus laevis and Quercus margaretta (Fagaceae). Am. J. Bot. 81:7-14.
BLACK, W. C., AND E. S. KRAFSUR. 1985. A FORTRAN program for analysis of genotypic frequencies and description of the breeding structure of populations. Theor. Appl. Genet. 70:484-490.
-----. 1986. Temporal and spatial trends in allozyme frequencies in house fly populations, Musca domestica. Theor. Appl. Genet. 71:673-681.
BROWN, A. H. D. 1979. Enzyme polymorphism in plant populations. Theor. Popul. Biol. 15:1-42.
CARTHEW, S. M. 1993. Population genetic structure of Banksia spinulosa. Heredity 70:566-573.
CAUGANT, D. A., B. R. LEVIN, AND R. K. SELANDER. 1981. Genetic diversity and temporal variation in the E. coli population of a human host. Genetics 98:467-490.
CHARLESWORTH, B. 1980. Evolution in age-structured populations. Cambridge Univ. Press, Cambridge.
CHARLESWORTH, D., AND B. CHARLESWORTH. 1987. Inbreeding depression and its evolutionary consequences. Annu. Rev. Ecol. Syst. 18:237-268.
CHESSEL, D. 1981. The spatial autocorrelation matrix. Vegetatio 46:177-180.
CLIFF, A.D., AND J. K. ORD. 1973. Spatial autocorrelation. Pion, London.
-----. 1981. Spatial process: Models and Applications. Pion, London.
COOK, R. E. 1983. Clonal plant populations. Am. Sci. 71:244-253.
-----. 1985. Growth and development in clonal plant populations. Pp. 259-296 in J. Jackson et al., eds. Population biology of clonal organisms. Yale Univ. Press, New Haven, CT.
DEL CASTILLO, R. F. 1994. Factors influencing the genetic structure of Phacelia dubia, a species with a seed-bank and large fluctuations in population size. Heredity 72:446-458.
DEWEY S. E., AND J. S. HEYWOOD. 1988. Spatial genetic structure in a population of Psychotria spinulosa. I. Distribution of genotypes. Evolution 42:834-838.
ENDLER, J. A. 1977. Geographic variation, speciation and clines. Princeton Univ. Press, Princeton., NJ.
EPPERSON, B. K. 1990. Spatial patterns of genetic variation within plant populations. Pp. 229-253 in A. H. D. Brown et al., eds. Plant population genetics, breeding, and genetic resources. Sinauer, Sunderland, MA.
-----. 1993. Recent advances in correlation studies of spatial patterns of genetic variation. Pp. 95-155 in M. K. Hecht, ed. Evolutionary biology. Vol. 27. Plenum Press, NY.
EPPERSON, B. K., AND M. T. CLEGG. 1986. Spatial auto-correlation analysis of flower color polymorphisms within substructured population of morning glory (Ipomea purpurea). Am. Nat. 128: 840-858.
ERIKSSON, O., AND B. BREMER. 1993. Genet dynamics of the clonal plant Rubus saxatilis. J. Ecol. 81:533-542.
GASQUEZ, J., AND J. P. COMPOINT. 1976. Apport de l'electrophorese en courant pulse a la taxonomie d'Echinoria crusgalli L. Ann. Amelior. Plant. 26:345-355.
GEARY, R. C. 1954. The contiguity ratio and statistical mapping. Inc. Stat. 5:115-145.
GLIDDON, C., E. BELHASSEN, AND P. H. GOUYON. 1987. Genetic neighbourhoods in plants with diverse systems of mating and different patterns of growth. Heredity 59:29-32.
GODT, M. J., AND J. L. HAMRICK. 1991. Estimates of outcrossing rates in Lathyrus latifolius populations. Genome 34:988-992.
-----. 1993. Patterns and levels of pollen-mediated gene flow in Lathyrus latifolius. Evolution 7:98-110.
GOTTLIEB, L. D. 1981. Electrophoretic evidence and plant populations. Prog. Phytochem. 7:1-46.
HALDANE, J. B. S. 1924. The mathematical theory of natural and artificial selection. Part II. The influence of partial self-fertilisation, inbreeding, assortative mating and selective fertilisation on the composition of Mendelian populations and on natural selection. Proc. Cambridge Phil. Soc. Biol. Sci. 1:158-163.
HAMRICK, J. L., AND J. W. GODT. 1990. Allozyme diversity in plant species. Pp. 43-63 in A. H. D. Brown et al., eds. Plant population genetics, breeding, and genetic resources. Sinauer, Sunderland, MA.
HAMRICK, J. L., Y. B. LINHART, AND J. B. MITTON. 1979. Relationships between life history characteristics and electrophoretically detectable genetic variation in plants. Annu. Rev. Ecol. Syst. 10:173-200.
HANDEL, S. N. 1985. The intrusion of clonal growth patterns on plant breeding systems. Am. Nat. 125:367-384.
HEYWOOD, J. S. 1991. Spatial analysis of genetic variation in plant populations. Annu. Rev. Ecol. Syst. 22:335-355.
HOSSAERT, M., AND M. VALERO. 1988. Effect of ovule position in the pod on patterns of seed formation in two species of Lathyrus (Leguminosae: Papilionoideae). Am. J. Bot. 75:1714-1731.
HOSSAERT, M., A. BASHAR, AND D. B. MCKEY. 1986. Floral nectar production in relation to insect activity in Lathyrus latifolius L. Pp. 213-224 in A. K. Kaul and D. Combes, eds. Lathyrus and Lathyrism. Third World Medical Research Foundation, New York.
HOSSAERT-MCKEY, M. 1988. Des fleurs comment et a quoi bon! Donnees et reflexions sur la reproduction sexuee de deux especes affines a propagation vegetative: Lathyrus latifolius et Lathyrus sylvestris (Legumineuses: Papilionaceae). Ph.D. diss. Universite Pau, France.
HOSSAERT-MCKEY, M., AND M. JARRRY. 1992. Spatial and temporal patterns of investment in growth and sexual reproduction in two stoloniferous species, Lathyrus latifolius and L. sylvestris. J. Ecol. 80:555-565.
JARRY, M., M. HOSSAERT, M. VALERO, D. MAGDA, AND A. CHAIB. 1990. Construction d'un module et communication dans une equipe plurisdisciplinaire a propos d'une etude sur le mode de reproduction d'une Legumineuse spontanee, Lathyrus sylvestris. Pp. 141-149 in CNRS, ed. La modelisation au confluent des Sciences. CNRS, Paris.
LEVIN, D. A. 1988. Local differentiation and the breeding structure of plant populations. Pp. 305-329 in L. C. Gottlieb and S. K. Jain, eds. Plant evolutionary biology. Chapman and Hall, New York.
LEVIN, D. A., AND H. W. KERSTER. 1974. Gene flow in seed plants. Pp. 139-220 in T. Dobzhansky, ed. Evolutionary biology. Vol. 7. Plenium, New York.
LINHART, Y. B., J. B. MITTON, K. B. STURGEON, AND M. L. DAVIS. 1981. Genetic variation in space and time in a population of ponderosa pine. Heredity 46:407-426.
LOVELESS, M. D., AND J. L. HAMRICK. 1984. Ecological determinants of genetic structure in plant populations. Annu. Rev. Ecol. Syst. 15:65-95.
LOVETT-DOUST, L. 1981. Population dynamics and local specialization in a clonal perennial (Ranunculus repens). 1. The dynamics of ramets in contrasting habitats. J. Ecol. 69:743-755.
MAGDA, D. 1989. A propos de la propagation vegetative de Lathyrus sylvestris (Legumineuse: Papilionaceae). Contribution a l'etude de la dynamique des populations vegetales. Ph.D. diss. Universite de Pau, France.
MAGDA, D., F. R. WAREMBOURG, AND V. LABEYRIE. 1988. Physiological integration among ramets of Lathyrus sylvestris L. Oecologia 77:255-260.
MARUYAMA, T., AND M. KIMURA. 1980. Genetic variability and effective population size when local extinction and colonization of subpopulations are frequent. Proc. Nat. Acad. Sci. USA 77: 6710-6714.
MITTON, J. B. 1993. Theory and data pertinent to the relationship between heterozygosity and fitness. Pp. 17-41 in N. C. Thornhill, ed. The natural history of inbreeding and outbreeding. Univ. of Chicago Press, Chicago.
MORAN, P. A. P. 1950. Notes on continuous stochastic phenomena. Biometrika 37:17-23.
NASON J. D., AND N. C. ELLSTRAND. 1995. Lifetime estimates of biparental inbreeding depression in the self-compatible annual plant Rhaphanus sativus. Evolution 49:307-316.
NAVAS, M. L., AND E. GARNIER. 1990. Demography and growth forms of the clonal perennial Rubia peregrina in mediterranean vineyard and unmanaged habitats. J. Ecol. 78:691-712.
OLIVIERI, I., D. COUVET, AND P. H. GOUYON. 1990. The genetics of transient populations: Research at the metapopulation level. Trends Ecol. Evol. 5:207-209.
ORNSTEIN, L., AND B. J. DAVIS. 1962. Disc electrophoresis, Part I and II. Distillation Products Industries, Rochester, NY.
PERRY, D. J., AND P. KNOWLES. 1991. Spatial genetic structure within three sugar maple (Acer saccharum Marsh.) stands. Heredity 66:137-142.
ROHLF, J. F., AND G. D. SCHNELL. 1971. An investigation of the isolation-by-distance model. Am. Nat. 105:295-324.
SAKAI, A. K., AND N. L. ODEN. 1983. Spatial pattern of sex expression in silver maple (Acer saccharinum L.): Morisita's Index and spatial autocorrrelation. Am. Nat. 122:489-508.
SALZMAN, A. G. 1985. Habitat selection in a clonal plant. Science 228:603-604.
SCHAAL, B. A., AND D. A. LEVIN. 1976. The demographic genetics of Liatris cylindracea Michx. (Compositae). Am. Nat. 110:191-206.
SCHELLNER, R. A., S. A. NEWELL, AND O. T. SOLBRIG. 1982. Studies on the population biology of the genus Viola. IV: Spatial patterns of ramets and seedlings in three stoloniferous species. J. Ecol. 70:273-290.
SCHEMSKE, D. W. 1983. Breeding system and habitat effects on fitness in three neotropical Costus (Zingiberaceae). Evolution 37:523-539.
SCHNABEL, A., AND J. L. HAMRICK. 1990. Organization of genetic diversity within and among populations of Gleditsia triacanthos (Leguminosae). Am. J. Bot. 77:1060-1069.
SKELLMAN, J. G. 1952. Studies in statistical ecology. I. Spatial patterns. Biometrika 39:346-362.
SLATKIN, M. 1977. Gene flow and genetic drift in a species subject to frequent local extinctions. Theor. Popul. Biol. 12:253-262.
----. 1987. Gene flow and the geographic structure of natural populations. Science 236:787-792.
----. 1993. Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47:264-279.
SLATKIN, M., AND H. E. ARTER. 1991. Spatial autocorrelation methods in population genetics. Am. Nat. 138:499-517.
SOKAL, R. R., AND N. L. ODEN. 1978a. Spatial autocorrelation in biology. 1. Methodology. Biol. J. Linn. Soc. 10:199-228.
----. 1978b. Spatial autocorrelation in biology. 2. Some biological implications and four applications of evolutionary and ecological interest. Biol. J. Linn. Soc. 10:229-249.
SOKAL, R. R., AND F. J. ROHLF. 1981. Biometry. Freeman, San Francisco, CA.
SOKAL, R. R., AND D. E. WARTENBERG. 1983. A test of spatial autocorrelation analysis using an isolation-by-distance model. Genetics 105:219-237.
SOLTIS, E. D., C. H. HAUFLER, D.C. DARROW, AND G. J. GASTONY. 1983. Starch gel electrophoreis of ferns: A compilation of grinding buffers; gel and electrode buffers, and staining schedules. Am. Fern J. 73:9-27.
SUTHERLAND, W. J., AND R. A. STILLMAN. 1988. The foraging tactics of plants. Oikos 52:239-244.
TEMPLETON, A. R., AND B. READ. 1983. The elimination in inbreeding depression in a captive herd of Speke's gazelle. Pp. 241-261 in C. M. Schonewald-Cox et al., eds. Genetics and conservation. Benjamin/Cummings, Menlo Park, CA.
THOMAS, R. H., AND J. S. F. BARKER. 1990. Breeding structure of natural populations of Drosophila buzzatii: Effects of the distribution of larval substrates. Heredity 64:355-365.
TONSOR, S. J., S. KALISZ, J. FISHER, AND T. P. HOLTSFORD. 1993. A life-history based study of population genetic structure: Seedbank to adults in Plantago lanceolata. Evolution 47:833-843.
TURNER, M. E., J. C. STEPHENS, AND W. W. ANDERSON. 1982. Homozygosity and patch structure in plant populations as a result of nearest-neighbor pollination. Proc. Nat. Acad. Sci. USA 79: 203-207.
UYENOYAMA, M. K., K. E. HOLSINGER, AND D. M. WALLER. 1993. Ecological and genetic factors directing the evolution of self-fertilization. Pp. 327-381 in D. Futuyma and J. Antonovics, eds. Oxford survey of evolutionary Biology. Vol. 9. Oxford Univ. Press, Oxford.
VALERO, M. 1987. Systeme de reproduction et fonctionnement des populations chez deux especes de Legumineuses du genre Lathyrus. Ph.D. diss. Universite Lille, France.
VALERO, M., AND M. HOSSAERT-MCKEY. 1991. Discriminant alleles and discriminant analysis: Differences between two related species of Leguminosae, Lathyrus latifolius and Lathyrus sylvestris. Bot. J. Linn. Soc. 107:130-161.
VALERO, M., M. HOSSAERT, B. CARON, A. YOUSSEF, AND P. VERNET. 1987. Allocation des ressources chez deux especes de Legumineuses, Lathyrus latifolius and Lathyrus sylvestris. Pp. 141-148 in J. M. Legay, ed. Biologie des populations. CNRS, Paris.
WADE, M. J., AND D. E. MCCAULEY. 1988. Extinction and recolonization: Their effects on the genetic differentiation of local populations. Evolution 42:995-1005.
WARTENBERG, D. E. 1989. SAAP. A spatial autocorrelation analysis program. Vers. 4.3. Exeter Software, Setauket, NY.
WASER, N. 1987. Spatial genetic heterogeneity in a population of the montane perennial plant Delphinium nelsonii. Heredity 58:249-256.
WEIR, B. S., AND C. C. COCKERHAM. 1984. Estimating the F-statistics for the analysis of population structure. Evolution 38:1358-1370.
WHITLOCK, M. C., AND D. E. MCCAULEY. 1990. Some population genetic consequences of colony formation and extinction: Genetic correlations within founding groups. Evolution 44:1717-1724.
WRIGHT, S. 1943. Isolation by distance. Genetics 28:114-138.
----. 1977. Evolution and the genetics of populations. Vol. 3. Experimental results and evolutionary deductions. Univ. Chicago Press, Chicago.
ZOUROS, E., AND D. W. FOLTZ. 1984. Possible explanations of heterozygote deficiency in bivalve molluscs. Malacologia 25: 583-591.
|Printer friendly Cite/link Email Feedback|
|Author:||Hossaert-McKey, Martine; Valero, Myriam; Magda, Daniele; Jarry, Marc; Cuguen, Joel; Vernet, Philippe|
|Date:||Oct 1, 1996|
|Previous Article:||Theory of mast reproduction in plants: storage-size dependent strategy.|
|Next Article:||Are populations islands? Analysis of chloroplast DNA variation in Aquilegia.|