| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Population Biology |
2Laboratoire Génétique et Environnement, C.C. 065, ISEM (UMR 5554), Université Montpellier II, 34095 Montpellier Cedex 05, France; 3Station Biologique de la Tour du Valat, Le Sambuc, 13200 Arles, France; 4Centre de Recerca Ecològica i Aplicacions Forestals, Universitat Autònoma de Barcelona, Bellaterra 08193, Barcelona, Spain; 5Laboratoire d'Écologie (UMR 7625), C.C. 237, Université Pierre et Marie Curie (Paris VI), 7 quai St Bernard, 75252 Paris Cedex 05, France
Received for publication September 13, 2001. Accepted for publication February 26, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: conservation biology • Marsilea strigosa • microsatellites • multilocus analyses • population structure
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Young, Boyle, and Brown, 1996
). A meta-analysis of published data sets showed that heterozygosity was also significantly correlated to population size and that genetic variation was significantly reduced in rare species, as compared to widespread ones (Frankham, 1996
). Comparing patterns of neutral genetic variation between rare and widespread plant congeners, Gitzendanner and Soltis (2000)
also found that genetic diversities (except the total diversity HT) were significantly lower in rare species. Whether the lack of genetic variation is the cause or the consequence of rarity constitutes the chicken and egg problem of conservation biology. Therefore, management recommendations based only on low genetic variation assessments may be of limited significance (Holsinger, Mason-Gamer, and Whitton, 1999
). Moreover, other factors may be of primary importance for assessing extinction risks (Lande, 1999
). More interestingly, understanding how genetic variation is distributed within and among natural remnant populations of rare species may help to identify the evolutionary forces that shape this variation. The estimation of demographic parameters from neutral genetic variation may, among other expertise, give clues to the management priorities.
During the 20th century, Mediterranean wetlands were dramatically reduced in size and numbers (Blondel and Aronson, 1999
). These habitats constitute an important source of richness for biodiversity, and they are often associated with seasonal flooding dynamics. The Mediterranean climate creates the conditions for ephemeral flooding and harsh summer drought in ponds and shallow wetlands. Flooding usually lasts from late fall to spring, before water evaporates during summer, leaving the pool beds dry. The plant species that may be found in those habitats generally share some life-history traits especially with regards to life-cycle duration (annuality), seed dormancy, and dispersal (Blondel and Aronson, 1999
). Temporarily flooded pools provide a unique situation for population biologists. In contrast with many natural situations, the habitat is generally readily demarcated and population boundaries are sharply defined (Holland and Jain, 1981
). Yet, the data are still scarce concerning the population biology and genetics of such ephemeral plant populations (Elam, 1998
). Those communities of species are generally characterized by small and/or fluctuating population sizes (Elam, 1998
), and the pattern of species diversity resembles that predicted by insular biogeography theory (Holland and Jain, 1981
). In the Great Valley (California), vernal pools are characterized by high endemism, interpreted as the consequence of low dispersal abilities of component species and high extinction risks (Holland and Jain, 1981
). Human activities are the major cause of the degradation of wetlands. Because temporary ponds and marshes are continuously lost, many taxa face increased rarity and extinction risks (Jain, 1994
).
Marsilea strigosa Willd. (Marsileaceae, Pteridophyta) is a rare water fern, whose leaves resemble those of a four-leaf clover. This species is listed in the French red book of endangered species (Olivier et al., 1995
), in the Bern convention on the conservation of European wildlife and natural habitats (1979), and in the European "Habitats" directive (1992). It is found in the Mediterranean (France, Spain, Algeria, Morocco, Egypt, Italy, and Sardinia), as well as in the northern part of the Volga Delta (Tutin et al., 1964
). Marsilea strigosa is found in seasonally wet habitats, where it grows in shallow water and at the edges of ponds. All the members of section Marsileaceae produce spores within specialized structures called sporocarps. The establishment of new sporophytes is remarkably rapid, which makes Marsileaceae able to grow in intermittent and ephemeral habitats. Sporocarps are drought-resistant, bean-shaped reproductive structures that allow M. strigosa populations to persist over winter. The sporocarps may remain viable even after very long periods of dry storage (Allsopp, 1952
; Johnson, 1985
; Soltis and Soltis, 1986
). Sporocarps open in flooded ponds (on hydration, a distinct gelatinous sorophore breaks through an opening in the sporocarp) to produce chains of sori, each sorus bearing series of mega- and microsporangia. Sporangia (that remain attached to the sporocarp) release their spores. The gelatinous outer layer of the spores (perine) expands, thereby decreasing the spore density. The spores drift toward the surface of the ponds (Schneider and Pryer, in press
). Female and male gametophytes develop within the mega- and microspores (endospory) and the fertilization occurs at the air/water interface. In vitro observations of sporocarp germination show that the gelatinous matrix prevents sporangia from freely moving around in water and that M. strigosa is self-compatible, indicating a possible predominantly autogamous reproductive system. Within 48 h after the release of spores from the sporangia, embryos are settled on the water/soil interface (Schneider and Pryer, in press
). Then, sporophytes grow through two phases: one aquatic phase with floating smooth leaves, followed by one terrestrial phase with smaller hairy leaves. During this later phase, plants propagate clonally. At the end of the growing season, the sporocarps are produced at the stem base, attached to the rhizome. As such, they do not constitute obvious dispersal units. Thus, one may ask whether what appear as metapopulations of ponds actually behave as sets of isolated populations or as sets of populations linked by dispersal through, e.g., recolonization following local extinctions (i.e., "true" metapopulations).
If the main evolutionary forces that shape neutral genetic variation and its distribution correlate to the species' biological characteristics (e.g., self compatibility, possibly associated to selfing, long-term storage of sporocarps in the soil, low dispersal capabilities), one expects the population genetic structure to be highly pronounced. However, the geographical scale at which such processes occur is hardly known: for example, flooding may temporarily connect ponds at a very local scale, promoting gene flow between populations within a metapopulation. We thus investigated the level and distribution of genetic variation at two contrasted scales. In this paper, we first study the multilocus structure at the Mediterranean scale (France, Morocco, and Spain; see Fig. 1). We then compare the structure to that observed at a finer scale in the highly subdivided natural reserve of Roque-Haute (France). Because genetic variation was totally absent at 12 putative enzymatic loci (seven systems) among French and Spanish populations (Vitalis et al., 1998
), we used seven microsatellite loci recently developed for this species (Vitalis, Dubois, and Olivieri, 2001
).
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). All the individuals from French and Spanish populations were maintained in a vegetative state in a greenhouse before leaves were sampled for DNA extraction. Sampled leaves from Moroccan populations were dried in silica gel. The sample sizes ranged from 3 to 30 individuals, depending on the actual density of plants and the size of ponds (Table 1).
|
Microsatellite analyses
DNA extractions were performed on frozen or dried leaves, as described in Vitalis, Dubois, and Olivieri (2001)
. Polymorphism was assayed on each DNA sample at seven microsatellite loci (GenBank accession numbers range from GBAN-AF317636 to GBAN-AF317642; the prefix GBAN- has been added to each GenBank accession to link the online version of American Journal of Botany to GenBank but is not part of the actual accession number). All seven loci are dinucleotide repeats, two of which are imperfect tandem repeats (IE3 and IVH5). The polymerase chain reaction (PCR) conditions have been described in Vitalis, Dubois, and Olivieri (2001)
. Amplification products (10 µL) were mixed with 6.7 µL formamide loading dye and then electrophoresed in 6% acryl-bisacrylamide and 8 mol/L urea sequencing gels, for 3–5 h at about 1800 V. Sequencing reactions of pUC19 vector (Appligene) were also loaded adjacent to the samples, to serve as a single base ladder. Once dried, gels were exposed to X-ray films for 24–48 h. Loci IIIB8, IIIC8, and XIIE3 were multiplexed, as were loci IA10 and IF7. Due to the low variation of locus IE3 at the Mediterranean scale (and to the absence of variation within all populations, including Roque-Haute), Roque-Haute 1999 samples were not scored at this locus. Locus IVH5 was monomorphic within Roque-Haute (for both 1994 and 1999 samples).
Statistical analyses
One-locus analyses of local mating system
For each locus and within each sample, departure from random union of gametes was tested by a score test (U test), with the alternative hypothesis of heterozygote deficiency. This test has proven to be more powerful than the exact Hardy-Weinberg test (probability test) of, e.g., Guo and Thompson (1992)
when the alternative hypothesis is heterozygote excess or deficiency (see Rousset and Raymond, 1995
). The tests were performed using genepop, version 3.3 (Raymond and Rousset, 1995
). Global tests across loci or across samples were performed using the multisample score test of Rousset and Raymond (1995)
. This test assumes the independence of loci. Gene diversities (heterozygosities) were estimated as (1 – Q0) and (1 – Q1), where the Qi's are the estimates of probabilities of identity in state (IIS) of pairs of genes, either within (Q0) or between (Q1) individuals within populations. One-locus Wright's F statistics, FIS (Wright, 1951
), were estimated by the estimator of Weir and Cockerham (1984)
. Multilocus estimates were computed in a slightly different manner than in Weir and Cockerham (1984)
or in Weir (1996)
. Here, the multilocus statistic was computed as the ratio of the sum of one-locus IIS probability estimates (Q0 – Q1), over the sum of one-locus IIS probability estimates (1 – Q1). This gives more weight to the Qi estimate of a more intensively typed locus, in proportion to the intensity of typing (see Rousset, 2001)
.
One-locus analyses of population differentiation
Genotypic differentiation was tested by a log-likelihood based exact test (Goudet et al., 1996
) over all samples and over all sample pairs. Unbiased estimates of the associated P values were calculated using the Markov chain method computed by genepop version 3.3 (Raymond and Rousset, 1995
). One-locus Wright's F statistics FST (Wright, 1951
) were estimated by the estimator of Weir and Cockerham (1984)
. Multilocus estimates were computed as a ratio of the sum of one-locus IIS probability estimates (Q1 – Q2), over the sum of one-locus IIS probability estimates (1 – Q2) (Q2 being an estimate of the IIS probability of pairs of genes among populations). To take into account not only the probability of identity between pairs of genes but also the mutational processes at microsatellite loci, the intraclass correlations for allele size, 
ST (see Rousset, 1996
), have been estimated in an analysis of variance framework (Michalakis and Excoffier, 1996
). Multilocus ST estimates of ST were computed similarly to multilocus estimates (ratios of the sum of one-locus estimates). The 95% confidence intervals were obtained by bootstrapping over loci as advocated by, e.g., Weir (1996)
. We derived the approximate bootstrap confidence intervals (ABC) using the method described by DiCiccio and Efron (1996)
. This procedure is an analytic version of the bias-corrected and accelerated (BCa) algorithm for producing confidence limits from bootstrap distributions that applies to nonparametric problems (DiCiccio and Efron, 1996
). The confidence interval endpoints are analytically approximated, rendering Monte Carlo simulations unnecessary. Isolation by distance was analyzed by computing the regression of pairwise (or ST) estimates between pairs of populations to the natural logarithm of their geographical distance (Rousset, 1997
). Geographic distances were calculated with reference to the World Geodetic System 1984 or WGS84 (Defense Mapping Agency, 1987
). Rank correlations were tested using the Mantel permutation procedure (Mantel, 1967
). For both within- and between-populations analyses, sequential Bonferroni correction was applied to test for significance, whenever multiple tests were performed (Rice, 1989
).
One-locus analyses of genetic distances
Pairwise genetic distances between pairs of populations were computed using Cavalli-Sforza and Edwards' chord distance (Cavalli-Sforza and Edwards, 1967
), with the gendist program (phylip package version 3.5c; Felsenstein, 1993
). We also computed the pairwise genetic distance matrix, which takes into account the average allele size differences between pairs of populations, i.e., the (
µ)2 distance of Goldstein et al. (1995)
. Fitch-Margoliash trees were constructed from those distance matrices with the program fitch (phylip, version 3.5c), which implements the method of Fitch and Margoliash (1967)
. It does not assume an evolutionary clock, so it results in an unrooted tree. The robustness of each node was evaluated by bootstrapping data over loci, for 2000 replications (Hedges, 1992
), using the seqboot program (phylip, version 3.5c). The resulting consensus tree was obtained through the program consense (phylip, version 3.5c) and displayed with the program treeview (Page, 1996
).
Two-locus analyses of linkage disequilibria
Tests for the independence of genotypes across loci (genotypic linkage disequilibrium) were performed (Fisher exact tests) for all pairs of loci within each sample. Unbiased exact P value estimates were obtained by the Markov chain method computed by genepop version 3.3 (Raymond and Rousset, 1995
). Linkage disequilibria were estimated by the composite Burrows' 
ij estimator (Cockerham and Weir, 1977
) using the linkdos program (Garnier-Gere and Dillmann, 1992
) adapted from Black and Krafsur's (1985)
program. This estimator does not require the gamete frequencies to be known, and it is not biased by departure from panmixia. Genotypic linkage disequilibria across populations were tested against the null hypothesis (linkage equilibrium) by a 
2 statistic (see Weir, 1979
). To test for significance, sequential Bonferroni tests were performed whenever multiple tests were carried out (Rice, 1989
). We also calculated Ohta's variance components of linkage disequilibrium (Ohta, 1982b
) with the linkdos program. Ohta (1982b)
defined such components to account for within- and between-subpopulation effects, in analogy with Wright's (1951)
F statistics. She proposed a test to discriminate between epistatic selection and genetic drift as possible explanations for large observed linkage disequilibria by comparison of appropriate variance components (Ohta, 1982a
).
Multilocus analyses
Assuming that the mutation pattern of the microsatellite loci employed may be approximated as a stepwise model (SMM; Kimura and Ohta, 1978
), the difference in the number of repeats between any two alleles at a locus may reflect the time since the divergence of those alleles. Thus, a genetic distance between multilocus genotypes was calculated as
) were computed with the program minspnet, provided with the software arlequin version 2.000 (Schneider, Roessli, and Excoffier, 2000
). |
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
), ranged from 0.000 to 0.061 (Appendix 1). Gene diversities between individuals (1 – Q1), which are equivalent to Nei's HS, ranged from 0.000 to 0.419 (Appendix 1). Yet, the low variability observed was not due to limited sample size (Spearman rank order correlation between gene diversity estimates and sample sizes: rs = 0.223, P = 0.464). We tested the null hypothesis that the alleles at one locus were randomly distributed among individuals within populations (alternative hypothesis = heterozygote deficiency). Among 34 possible tests performed (out of 91 population x locus combinations), 28 were significant at the 
= 0.05 level (for both the exact test and the score test). The sequential Bonferroni test performed over the 34 exact tests (respectively score tests) showed that 16 tests (respectively 15 tests) were significant at the = 0.05 level. One-locus estimates, 's, of the parameter FIS ranged from –0.125 to 1.000, over all samples (Appendix 1). Multilocus estimates ranged from 0.591 to 1.000. In all populations but Roque-Haute, positive estimates were unlikely to be attributable to Wahlund effect (all the samples within a locality were taken within a single pond). From the equation 
= 2/(1 + ) (see, e.g., Pollak, 1987
), the estimated selfing rates within all populations but Roque-Haute ranged from 0.743 to 1.000. All the multiple tests performed per locus for all populations and per population for all loci were significant at the = 0.05 level, indicating departure from Hardy-Weinberg equilibrium.
One-locus analyses of population differentiation
Genotypic differentiation among populations was highly significant at each locus (log-likelihood based test; P < 0.0001). One-locus estimates of parameter FST ranged from 0.742 to 1.000 (Table 2). The global test across loci was also highly significant (Fisher's method; P < 0.0001) and the multilocus estimate of FST was equal to 0.845. Intraclass correlation for allele size estimates (ST) ranged between 0.855 and 1.000. The ST estimates were significantly larger than estimates (one-tailed Wilcoxon's signed-rank test; P = 0.009). Pairwise FST estimates were significantly correlated with geographic distance (Mantel test; P = 0.018), as were ST estimates (P = 0.003).
|
chord distance (Fig. 2). Geographically close populations (Vendres and Roque-Haute; Sinarcas ponds 1 and 2; Ben Slimane dayas 1 and 2; Balearic samples) were grouped together, respectively. This indicates that close populations resemble each other more than they do more distant populations. As might be seen from Fig. 2, the chord distances between Balearic/Moroccan samples and all other samples were larger on average (range 0.314–0.340) than all the distances between any sample and all the others (range 0.281–0.340). The bootstrap values (expressed as the percentage of the number of occurrences of a node, out of 2000 bootstrap samples) ranged from 22 to 100%. All the nodes within the subtree made of Guadalajara, Valdepeñas, Balearic, and Moroccan populations were satisfactory resolved (bootstrap values 
68%). All the other nodes, except those uniting very close populations (Vendres and Roque-Haute, on the one hand, and Sinarcas ponds 1 and 2, on the other hand), were not supported (bootstrap values <50%), indicating that the tree topology was not yet resolved for the populations from central Spain. Withdrawing Balearic and Moroccan samples did not change the topology nor the bootstrap values, of the tree part that linked the remaining samples. This ensured that long-branch attraction phenomena (Felsenstein, 1978
) were not responsible for the lack of consistency among Spanish samples.
|
µ)2 distance of Goldstein et al. (1995)
. A Fitch-Margoliash tree was drawn from this pairwise distance matrix (Fig. 3). Geographically closer populations were grouped together, except populations from Moroccan dayas and the two ponds from Sinarcas, which were not significantly grouped (bootstrap value <50%). Surprisingly, Valdepeñas population was found to be more related than Ben Slimane daya 2 to the group made of Ben Slimane daya 1 and Balearic populations. The branch lengths were far longer with Goldstein et al.'s (1995)
(µ)2 distances than with Cavalli-Sforza and Edwards' (1967)
distances. However, some small branch lengths made difficult the identification of clear population clusters, at least as compared with the chord distance Fitch-Margoliash tree (Fig. 2). As with the chord distance Fitch-Margoliash tree, some nodes uniting Spanish and French populations were not well supported (bootstrap values <50%).
|
= 0.05 level (five, when the sequential Bonferroni procedure was computed). Burrows' ij estimates of genotypic linkage disequilibria between pairs of loci (Cockerham and Weir, 1977
) were highly significantly different from zero, even after Bonferroni correction for multiple tests (P < 0.0001). Table 3 gives Ohta's variance components of linkage disequilibrium (Ohta, 1982b
). By analogy with F statistics, Ohta derived the variance components of linkage disequilibrium, as the intraclass correlations of genes at different loci of gametes within a class, relative to genes in another class. The variance components of the disequilibrium within a population relative to that of the total population (noted D'2IS, the correlation of genes at two loci from one gamete relative to the total and D2ST, the correlation of genes from different gametes of one subpopulation relative to the total) are fairly large, as is the total variance of linkage disequilibrium (D2IT). Moreover, D'2IS (respectively D2ST) estimates were significantly larger than D'2ST, the variance of the total disequilibrium (respectively D2IS, the variance of the disequilibrium within a subpopulation) estimates (one-tailed Wilcoxon's signed-rank test; P < 0.0001). Thus, the relationships D'2IS > D'2ST and D2ST > D2IS were fulfilled, which indicated that genetic drift and limited dispersal among populations may be mainly responsible for the observed linkage disequilibrium (Ohta, 1982a
).
|
|
|
= 0.05 level (for both the exact test and the score test). The sequential Bonferroni test performed over the 15 exact tests showed that ten tests were statistically significant at the = 0.05 level (and so were the score tests). One-locus estimates, , of the parameter FIS ranged from –0.020 to 1.000, over all samples (Appendix 2). Multilocus estimates ranged from 0.923 to 1.000. Estimated selfing rates thus ranged from 0.960 to 1.000. All the multiple tests performed per locus for all populations were highly significant. Among seven possible tests performed per pond for all loci (26 combinations), six were significant at the = 0.05 level. Genotypic differentiation among ponds was highly significant at each locus (log-likelihood based test; P < 0.0001). One-locus estimates of parameter FST ranged from 0.592 to 0.872 (Table 4). The global test across loci was also highly significant (Fisher's method; P < 0.0001) and the multilocus estimate of FST was equal to 0.723. Intraclass correlation for allele size estimates (ST) ranged between 0.470 and 0.885. ST estimates were not significantly different than estimates (two-tailed Wilcoxon's signed-rank test; P = 0.686). Pairwise estimates were significantly correlated with geographic distance (10 000 permutations Mantel test; P < 0.0001), as were ST estimates (P < 0.0001).
|
In 9 ponds out of 13, multilocus genotypes were identical in years 1994 and 1999. Yet, the genotype B was found in pond 64 in 1994 and not in 1999, and the genotype O was found in pond 129 in 1994 and not in 1999. The genotype A was present in both ponds, in 1994 and 1999. The genotype D was found in pond 58 in 1994 and not in 1999 (but the genotype J was found there in 1994 and 1999). In pond 180, the genotypes C and J were found in 1994 and not in 1999, while genotypes M and O, which were found in 1999 were absent from the 1994 sample. Yet, the genotype A was found in pond 180 in 1994 and 1999. Note that the two genotypes M and O are genetically close to genotype A (see Fig. 6).
|
) was constructed from the pairwise allele size difference distance of Eq. 1. Note that the distances separating two nodes on Fig. 6 (each node representing a single sampled genotype) are proportional to the actual distances between two multilocus genotypes (Eq. 1). Whenever a pond sample was composed of more than one genotype, those genotypes were closely related (Fig. 6: see, e.g., ponds 81, 118, 163, and 180). |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Microsatellite variation was low, with an average number of alleles over all Mediterranean samples equal to 1.65 (SD = 0.61; see Appendix 1). The difference between the two classes of markers is probably due to the difference in mutation rates. The mutation rates of microsatellite loci have been reported to fall in the range of 10–6–10–2 (Jarne and Lagoda, 1996
; Goldstein and Schlötterer, 1999
), i.e., up to 104 times larger than that of allozymes (see, e.g., Voelker, Schaffer, and Mukai, 1980
). The effective number of alleles per locus (estimated as the inverse of the IIS probability for pairs of genes within the total pooled Mediterranean sample) was positively correlated with the weighted average number of repeats (Spearman correlation rs = 0.57; P = 0.180). The lack of significance was due to locus IVH5. With this locus removed, the correlation was higher and significant (rs = 0.89; P = 0.019). This could indicate that replication slippage may be an important factor in generating variation at those microsatellite loci (see, e.g., Innan, Terauchi, and Miyashita, 1997
). Because available data and simulation-based tests are not yet appropriate to test whether microsatellite loci mutate according to a stepwise mutation model (SMM; Kimura and Ohta, 1978
), an infinite allele model (IAM; Kimura and Crow, 1964
) or a mixture of both (two-phase mutation models or TPM; Slatkin, 1995
), there is no real consensus about the actual mutational processes at play along microsatellite sequences (see, e.g., Jarne and Lagoda, 1996
). Although the TPM has been considered the most appropriate model (Di Rienzo et al., 1998
), neither the SMM nor even the IAM could be rejected (Valdes, Slatkin, and Freimer, 1993
).
Marsilea strigosa reproductive system
Within-population lack of variation (fixation) at most loci prevented the estimation of FIS parameters and the application of exact tests for deviation from Hardy-Weinberg proportions in most cases (57 out of 91 for the Mediterranean data set; 115 out 130 for the Roque-Haute data set). Exact tests performed per locus and per sample were far from being all significant. In nearly all cases, however, the absence of significance was due to small sample sizes. Such low statistical power for largely inbred samples is not unexpected (see Robertson and Hill, 1984
; Rousset and Raymond, 1995
). Moreover, all the multisample score tests but one (pond 59) were significant. Selfing rate estimates based on multilocus estimates overall ranged from 0.743 to 1.000 and should be considered as underestimating the actual selfing rate. Indeed, in a finite population of size N, where the overall amount of selfing is s, FIS = 
/(2 – ), with = s – 1/N (see, e.g., Rousset, 1996
). Thus, what is really estimated is the actual rate at which selfed progeny are produced, minus the inverse of (unknown) population size. Although the presence of male and female spores (heterospory) prevents intragametophytic selfing (the fusion of male and female gametes derived from the same gametophyte), M. strigosa clearly reproduces mainly through intergametophytic selfing (the fusion of male and female gametes from different gametophytes derived from the same sporophyte, which is analogous to seed plant selfing). Were M. strigosa clonally reproducing only, strong heterozygote excesses would be expected. Therefore, our results imply that M. strigosa reproduces sexually through an annual life cycle and that sporocarps are the only means for M. strigosa to survive to winter harsh conditions.
These results are consistent with a recent study by Schneider and Pryer (in press)
of the structure and function of spores in the Marsileaceae. Although Schneider and Pryer (in press)
argued that the mechanism to float at the water/air interface for an extended period could favor the mixing of spores of different individuals (and thereby increase the potential for outbreeding), we did not find any evidence for an effective outbreeding mating system in M. strigosa. Yet, Schneider and Pryer (in press)
observations that a small chamber is formed at the aperture point between the two outer layers (perine and exine) of the megaspore may give a clue to the functional aspects of M. strigosa reproductive system. Schneider and Pryer (in press)
argued that this free space (referred to as a sperm lake) between the solid and gelatinous layers of the megaspore might function primarily as a trap for male gametes. Because of the large size of megaspores, a vortex-like effect would indeed occur and would force sperm cells in the local neighborhood to be caught up in the vortex and to be propelled into the sperm lake. Similar observations have been made in aquatic flowering plants with hydrophilous pollination (see, e.g., Cox, 1988
).
One-locus population differentiation at contrasted geographic scales
We observed considerable and highly significant differentiation at both Mediterranean and Roque-Haute scales (multilocus estimates = 0.845 and 0.723, respectively; see Tables 2 and 4). This indicates that gene flow is highly restricted, not only between Mediterranean populations (average pairwise distance = 576 km, SD = 357 km) but also between ponds within the Roque-Haute natural reserve (average pairwise distance between sampled ponds = 267 m, SD = 157 m). Among the Mediterranean samples, ST estimates of the intraclass correlations for allele sizes (ST) were significantly larger than estimates of FST while the same was not true among Roque-Haute 1999 samples (see Tables 2 and 4). This result is in agreement with the analytical predictions by Rousset (1996)
, who showed that, with stepwise mutations occurring, FST increases much less than ST in the more distant populations. Although those exact results hold in equilibrium models of population structures (Rousset, 1996
), a similar lack of variation for FST measures was also obtained through simulations of nonequilibrium situations (Slatkin, 1993
). A similar relationship was obtained on Bulinus truncatus, a highly selfing freshwater snail, at smaller but comparable geographic scales (Viard et al., 1996
). Nonsignificant differences between estimates of FST and ST have also been reported (Michalakis and Veuille, 1996
; Estoup et al., 1998
). According to Estoup et al. (1998)
, the absence of significant differences could be the result of large dispersal rates and/or recent populations divergence, two situations under which differentiation does not depend much on the mutation process (Slatkin, 1995
; Rousset, 1996
). Our results are thus consistent with theoretical expectations, with significantly larger ST values than values at the Mediterranean scale and nonsignificant differences at the Roque-Haute natural reserve scale, where the average time of divergence among pools is expected to be more recent than among Mediterranean populations and where (even limited) gene flow is expected to be higher at smaller geographic scales.
There are but few studies to compare our results with. With the exception of few studies using polymorphic DNA markers (see, e.g., Schneller et al., 1998
; Keiper and McConchie, 2000
; Landergott et al., 2001
; Pryor et al., 2001
) most of the population genetic studies in pteridophytes have been carried out with allozymes. Moreover this is, to our knowledge, the first report of the population genetic structure of a heterosporous fern (heterospory being a characteristic trait of Marsileaceae and Salviniaceae: see, e.g., Pryer, 1999
). Keiper and McConchie (2000)
reported high FST estimates (average = 0.783) at 469 amplified fragment length polymorphisms (AFLPs) among eight natural populations of the umbrella fern Sticherus flabellatus (R. Br) St John in Australia. With the exception of Hemionitis palmata, with an overall estimated FST of 0.698 (Ranker, 1992), such a high population differentiation is not common in pteridophytes (Keiper and McConchie, 2000)
. Maki and Asada (1998)
showed that gene flow among populations of Polystichum otomasui (a narrow endemic fern in Japan) was high enough to impede genetic differentiation. Maki and Asada (1998)
argued that, as for most homosporous ferns, aerodynamic properties of spores make them likely to be highly efficient dispersal units in this species. Rare but effective long-distance dispersal events were also invoked to account for the distribution of genetic variation in Dryopteris remota, an apomictic homosporous fern (Schneller et al., 1998
). In Marsileaceae however, spores are produced within the sporocarps, which therefore must be considered as the units of long-distance dispersal (Johnson, 1985
). Our results suggest that sporocarps may not be as efficient dispersing units as homosporous fern spores are likely to be.
Multilocus population structure
At the Mediterranean scale, some pairs of loci showed significant genotypic disequilibria within samples. For each pair of loci, Burrows' ij estimates of genotypic linkage disequilibria across all populations (Cockerham and Weir, 1977
) were highly significantly different from zero. Ohta's variance components of linkage disequilibrium (Ohta, 1982b
) within subpopulations relative to that of the total populations (D'2IS and D2ST) were very large and significantly greater than the variance of the total disequilibrium (D'2ST) and that of disequilibrium within a subpopulation (D2IS), respectively (see Table 3). Because limited migration is very effective in increasing differentiation of gamete types among subpopulations, this result is expected through random genetic drift alone (Ohta, 1982a
). This very strong multilocus genotypic structure across Mediterranean populations, together with several suspicions that the microsatellite loci typed here fitted with a SMM (see above), led us to consider the multilocus genotypic population structure of M. strigosa.
The minimum spanning network (MSN) constructed from the pairwise distance given by Eq. 1 between sampled multilocus genotypes was strikingly representative of the actual population clustering (Fig. 4). In a single case, one genotype was found in two populations (genotype 16 in populations 3 and 4). The relative branching of populations was congruent with the (µ)2-based tree (Fig. 3), with, e.g., Ben Slimane daya 1 branching to Valdepeñas population. Although the similarity between multilocus genotypes MSN and (µ)2 distance-based dendrogram might not be expected in general situations, this can be interpreted here as the consequence of the quasi absence of shared multilocus genotypes between populations, as well as the low frequency of shared alleles between populations, at most loci. The clustering of multilocus genotypes reflects the quasi absence of gene flow among Mediterranean populations. The source of genetic variation within populations seems therefore to be limited to dramatically rare events of mutation and migration, followed by effective recombination (although the latter should also be considered as a rare event, given the high rate of selfing inferred from the data).
Fine-scale population structure
At a smaller scale, within the Roque-Haute natural reserve metapopulation, we observed a very strong clustering of multilocus genotypes (Fig. 5). Except genotype N (ponds 115 and 180), all the genotypes common to two or more populations were strongly geographically clustered. A striking result comes from the relationships between the multilocus genotypes within single (polymorphic) ponds (Figs. 5–6). Within all the polymorphic ponds (see, e.g., ponds 81, 118, 163, and 180) multilocus genotypes are closely linked in the MSN based on allele size differences (Eq. 1) (see Fig. 6 and the box in Fig. 5). This is consistent with the strong one-locus differentiation. There was no clear pattern between long-term presence of M. strigosa in a given pond and the genetic diversity retained in that pond. For example, no M. strigosa sporophyte was observed in pond 81 for 5 yr, and yet the averaged heterozygosity within this pond was the highest of all ponds: (1 – Q1) = 0.367 (see Appendix 2). Long- or mid-term storage of sporocarps in the soil has the same effect on genetic variation as seed banks. Because sporocarp storage promotes the age structuring of populations, it may increase the effective population size and thus dampen the effect of genetic drift (see, e.g., Rousset, 1999
). In the homosporous fern Athyrium filix-femina, spore banks were shown to retain a considerable amount of genetic variation (Schneller, 1998
). Moreover, the newly colonized ponds in 1999 do not undermine the multilocus structure observed. In all cases, the multilocus genotypes present in the newly colonized ponds are compatible with short-scale dispersal events from adjacent ponds. This was particularly striking for polymorphic ponds 62 and 163. The overall pattern of genetic structure is therefore consistent with rare dispersal events occurring at a short scale (isolation by distance). Interestingly, this demonstrates that, reminiscent of the Mediterranean pattern of multilocus population structure (see Fig. 4), genetic variation may arise primarily as the result of mutation and recombination.
Implications for conservation
Our results show that gene flow is extremely reduced between Mediterranean populations, as well as between close ponds within a single site. The low dispersal ability of M. strigosa surely jeopardizes its chance to recolonize empty sites, even if rare events (e.g., the odd climatic conditions such as the late reflooding in spring 1999) may favor the movements of dispersal units or agents from time to time. There are only a few studies that aim at documenting the level and the distribution of genetic variation of such ephemeral plant populations (Elam, 1998
). It is therefore difficult to extrapolate our results to other plant species living in temporarily flooded habitats. Yet, the presence of M. strigosa in far distant populations at the Mediterranean scale (see Fig. 1) and its apparently low ability to disperse is puzzling. We cannot discriminate between a scenario of ongoing habitat fragmentation (in which case the actual populations are remnant of some more common wetlands in a distant past) and a scenario of ongoing loss of dispersal agents (such as, e.g., water birds). For both scenarios, however, the low genetic variation maintained within populations should not be considered as the major threat to M. strigosa. Instead, some agricultural practices (e.g., draining, overpumping of ground water), along with many other threats (invasion of alien species, pollution, etc.), imperil the existing populations. In France, the third known M. strigosa population (Saint-Estève, Eastern Pyrenees) disappeared in 1982, following the management of a temporarily flooded pond into a permanent one (Amigo, 1987
). Yet, a new M. strigosa population was found close to the extinct one in 1996, within an 82-ha wetland in Torremilla (J. Molina, Conservatoire Botanique National Méditerranéen de Porquerolles [France], personal communication). Given the low dispersal capabilities of this species, management strategies should thus first focus on the protection and the conservation of M. strigosa remnant populations and suitable habitats.
|
|
|
|
FOOTNOTES |
|---|
6 Author for reprint requests, current address: School of Biological Sciences, Royal Holloway, University of London, Egham, Surrey TW20 0EX, UK (vitalis{at}isem.univ-montp2.fr
)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Amigo J. J. 1987 Exit la mare temporaire de Saint-Estève (Saint Estève, Pyrénées-Orientales, France). Naturalia Ruscinonensia 1: 71-136
Black W. C. E. S. Krafsur 1985 A Fortran program for the calculation and analysis of two-locus linkage disequilibrium coefficients. Theoretical and Applied Genetics 70: 491-496[CrossRef][Web of Science]
Blondel J. J. Aronson 1999 Biology and wildlife of the Mediterranean region. Oxford University Press, Oxford, UK
Cavalli-Sforza L. L. A. W. F. Edwards 1967 Phylogenetic analyses: models and estimations procedures. American Journal of Human Genetics 19: 233-257
Cockerham C. C. B. S. Weir 1977 Digenic descent measures for finite populations. Genetical Research 30: 121-147[Web of Science]
Cox P. A. 1988 Hydrophilous pollination. Annual Review of Ecology and Systematics 19: 261-279
Defense Mapping Agency. 1987 Department of Defense World Geodetic System 1984, its definition and relationships with local geodetic systems. DMA Technical report 8350.2. Washington, D.C., USA
Di Rienzo A. P. Donnelly C. Toomajian B. Sisk A. Hill M. L. Petzl-Erler G. K. Haines D. H. Barch 1998 Heterogeneity of microsatellite mutations within and between loci, and implications for human demographic histories. Genetics 148: 1269-1284
DiCiccio T. J. B. Efron 1996 Bootstrap confidence intervals. Statistical Science 11: 189-228[CrossRef][Web of Science]
Elam D. R. 1998 Population genetics of vernal pools plants: theory, data and conservation implications. In C. W. Witham, E. T. Bauder, D. Belk, W. R. Ferren Jr., and R. Ornuff [eds.], Ecology, conservation, and management of vernal pool ecosystems—Proceedings from a 1996 Conference, 180–189. California Native Plant Society, Sacramento, California, USA
Estoup A. F. Rousset Y. Michalakis J.-M. Cornuet M. Adriamanga R. Guyomard 1998 Comparative analysis of microsatellite and allozyme markers: a case study investigating microgeographic differentiation in brown trout (Salmo trutta). Molecular Ecology 7: 339-353[CrossRef][Medline]
Felsenstein J. 1978 Cases in which parsimony and compatibility methods will be positively misleading. Systematic Zoology 27: 401-410[CrossRef][Web of Science]
Felsenstein J. 1993 phylip (Phylogeny Inference Package) version 3.5c. Distributed by the author. Department of Genetics, University of Washington, Seattle, Washington, USA
Fitch W. M. E. Margoliash 1967 Construction of phylogenetic trees. Science 155: 279-284
Frankham R. 1996 Relationship of genetic variation to population size in wildlife. Conservation Biology 10: 1500-1508[CrossRef][Web of Science]
Garnier-Gere P. C. Dillmann 1992 A computer program for testing pairwise linkage disequilibria in subdivided populations. Journal of Heredity 83: 239
Gitzendanner M. A. P. S. Soltis 2000 Patterns of genetic variation in rare and widespread plant congeners. American Journal of Botany 87: 783-792
Goldstein D. B. A. Ruiz Linares L. L. Cavalli-Sforza M. W. Feldman 1995 Genetic absolute dating based on microsatellites and the origin of modern humans. Proceedings of the National Society of Sciences, USA 92: 6723-6727[CrossRef]
Goldstein D. B. C. Schlötterer 1999 Microsatellites: evolution and applications. Oxford University Press, Oxford, UK
Goudet J. M. Raymond T. de Meeus F. Rousset 1996 Testing differentiation in diploid populations. Genetics 144: 1933-1940[Abstract]
Guo S. W. E. Thompson 1992 Performing the exact test of Hardy-Weinberg proportion for multiples alleles. Biometrics 48: 361-372[CrossRef][Web of Science][Medline]
Hedges S. B. 1992 The number of replications needed for the accurate estimation of the bootstrap P value in phylogenetic studies. Molecular Biology and Evolution 9: 366-369[Web of Science][Medline]
Holland R. F. S. K. Jain 1981 Insular biogeography of vernal pools in the Central Valley of California. American Naturalist 117: 24-37[CrossRef][Web of Science]
Holsinger K. E. R. J. Mason-Gamer J. Whitton 1999 Genes, demes, and plant conservation. In L. F. Landweber and A. P. Dobson [eds.], Genetics and the extinction of species, 23–46. Princeton University Press, Princeton, New Jersey, USA
Innan H. R. Terauchi N. T. Miyashita 1997 Microsatellite polymorphism in natural populations of the wild plant Arabidopsis thaliana. Genetics 146: 1441-1452[Abstract]
Jain S. K. 1994 Genetics and demography of rare plants and patchily distributed colonizing species. In V. Loeschcke, J. Tomiuk, and S. K. Jain [eds.], Conservation genetics, 291–307. Birkhäuser Verlag, Basel, Switzerland
Jarne P. P. Lagoda 1996 Microsatellites, from molecules to populations and back. Trends in Ecology and Evolution 11: 424-429[CrossRef]
Johnson D. M. 1985 New records for longevity of Marsilea sporocarps. American Fern Journal 75: 30-31[CrossRef]
Keiper F. J. R. McConchie 2000 An analysis of genetic variation in natural populations of Sticherus flabellatus [R. Br. (St John)] using amplified fragment length polymorphism (AFLP) markers. Molecular Ecology 9: 571-581[CrossRef][Medline]
Kimura M. J. F. Crow 1964 The number of alleles that can be maintained in a finite population. Genetics 49: 725-738
Kimura M. T. Ohta 1978 Stepwise mutation model and distribution of allelic frequencies in a finite population. Proceedings of the National Society of Sciences, USA 75: 2868-2872[CrossRef]
Kruskal J. B. J. 1956 On the shortest spanning subtree of a graph and the traveling salesman problem. Proceedings of the American Mathematical Society 7: 48-50
Lande R. 1999 Extinction risks from anthropogenic, ecological, and genetic factors. In L. F. Landweber and A. P. Dobson [eds.], Genetics and the extinction of species, 1–22. Princeton University Press, Princeton, New Jersey, USA
Landergott U. R. Holderegger G. Kozlowski J. J. Schneller 2001 Historical bottlenecks decrease genetic diversity in natural populations of Dryopteris cristata. Heredity 87: 344-355[CrossRef][Web of Science][Medline]
Maki M. Y.-J. Asada 1998 High genetic variability revealed by allozymic loci in the narrow endemic fern Polystichum otomasui (Dryopteridaceae). Heredity 80: 604-610[CrossRef][Web of Science]
Mantel N. 1967 The detection of disease clustering and generalized regression approach. Cancer Research 27: 209-220
Michalakis Y. L. Excoffier 1996 A generic estimation of population subdivision using distances between alleles with special reference for microsatellite loci. Genetics 142: 1061-1064[Abstract]
Michalakis Y. M. Veuille 1996 Length variation of CAG/CAA trinucleotide repeats in natural populations of Drosophila melanogaster and its relation to the recombination rate. Genetics 143: 1713-1725[Abstract]
Nei M. 1973 Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Sciences, USA 70: 3321-3323
Ohta T. 1982a Linkage disequilibrium due to random genetic drift in finite subdivided populations. Proceedings of the National Society of Sciences, USA 79: 1940-1944[CrossRef]
Ohta T. 1982b Linkage disequilibrium with the island model. Genetics 101: 139-155
Olivier L. J. P. Galland H. Maurin J. P. Roux 1995 Livre rouge de la flore menacée de France. Tome 1: espèces prioritaires. Collection Patrimoines naturels, vol. 20. Muséum National d'Histoire Naturelle, Conservatoire Botanique National de Porquerolles & Ministère de l'Environnement, Paris, France
Page R. D. M. 1996 treeview: an application to display phylogenetic trees on personal computers. Computer Applications in the Biosciences 12: 357-358
Pollak E. 1987 On the theory of partially inbreeding finite populations. I. Partial selfing. Genetics 117: 353-360
Pryer K. M. 1999 Phylogeny of marsileaceous ferns and relationships of the fossil Hydropteris pinnata reconsidered. International Journal of Plant Sciences 160: 931-954[CrossRef][Web of Science][Medline]
Pryor K. V. J. E. Young F. J. Rumsey K. J. Edwards M. W. Brudford H. J. Rogers 2001 Diversity, genetic structure and evidence of outcrossing in British populations of the rock fern Adiantum capillus-veneris using microsatellites. Molecular Ecology 10: 1881-1894[CrossRef][Medline]
Ranker T. A. 1992 Genetic diversity, mating systems, and interpopulational gene flow in neotropical Hemionitis palmata. Heredity 69: 175-183[Web of Science]
Raymond M. F. Rousset 1995 genepop (Version 1.2): a population genetics software for exact tests and ecumenicism. Journal of Heredity 86: 248-249
Rice W. R. 1989 Analyzing tables of statistical tests. Evolution 43: 223-225[CrossRef][Web of Science]
Robertson A. W. G. Hill 1984 Deviations from Hardy-Weinberg proportions: sampling variances and use in estimation of inbreeding coefficients. Genetics 107: 703-718
Rousset F. 1996 Equilibrium values of measures of population subdivision for stepwise mutation processes. Genetics 142: 1357-1362[Abstract]
Rousset F. 1997 Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance. Genetics 145: 1219-1228[Abstract]
Rousset F. 1999 Genetic differentiation in populations with different classes of individuals. Theoretical Population Genetics 55: 297-308
Rousset F. 2001 Inferences from spatial population genetics. In D. Balding, M. Bishop, and C. Cannings [eds.], Handbook of statistical genetics, 239–269. John Wiley, Chichester, UK
Rousset F. M. Raymond 1995 Testing heterozygote excess and deficiency. Genetics 140: 1413-1419[Abstract]
Schneider H. K. M. Pryer In press Structure and function of spores in the aquatic heterosporous fern family Marsileaceae. International Journal of Plant Sciences.
Schneider S. D. Roessli L. Excoffier 2000 Arlequin, Version 2000: a software for population genetics data analysis. Genetics and Biometry Laboratory, Department of Anthropology, University of Geneva, Geneva, Switzerland
Schneller J. J. 1998 How much genetic variation in fern populations is stored in the spore banks? A study of Athyrium filix-femina (L.) Roth. Botanical Journal of the Linnean Society 127: 195-206[CrossRef]
Schneller J. R. Holderegger F. Gugerli K. Eichenberger E. Lutz 1998 Patterns of genetic variation detected by RAPDs suggest a single origin with subsequent mutations and long-distance dispersal in the apomictic fern Dryopteris remota (Dryopteridaceae). American Journal of Botany 85: 1038-1042[Abstract]
Slatkin M. 1993 Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47: 264-279[CrossRef][Web of Science]
Slatkin M. 1995 A measure of population subdivision based on microsatellite allele frequencies. Genetics 139: 457-462[Web of Science][Medline]
Soltis D. E. P. S. Soltis 1986 Active enzymes from megaspores of Marsilea and Regnellidium. American Fern Journal 76: 17-20[CrossRef][Web of Science]
Tutin T. G. V. H. Heywood N. A. Burges D. M. Moore D. H. Valentine S. M. Walters D. A. Webb 1964 Flora Europaea, vol. 1. Cambridge University Press, Cambridge, UK
Valdes A. M. M. Slatkin N. B. Freimer 1993 Allele frequencies at microsatellite loci: the stepwise mutation model revisited. Genetics 133: 737-749[Abstract]
van Treuren R. R. Bijlsma W. van Delden N. J. Ouborg 1991 The significance of genetic erosion in the process of extinction. I. Genetic differentiation in Salvia pratensis and Scabiosa columbaria in relation to population size. Heredity 66: 181-189[CrossRef][Web of Science]
Viard F. P. Bremond R. Labbo F. Justy B. Delay P. Jarne 1996 Microsatellites and the genetics of highly selfing populations in the freshwater snail Bulinus truncatus. Genetics 142: 1237-1247[Abstract]
Vitalis R. B. Colas M. Riba I. Olivieri 1998 Marsilea strigosa Willd.: statut génétique et démographique d'une espèce menacée. Ecologia Mediterranea 24: 145-157
Vitalis R. M.-P. Dubois I. Olivieri 2001 Characterization of microsatellite loci in the endangered species of fern Marsilea strigosa Willd. (Marsileaceae, Pteridophyta). Molecular Ecology Notes 1: 64-66[Web of Science]
Voelker R. A. H. E. Schaffer T. Mukai 1980 Spontaneous allozyme mutations in Drosophila melanogaster: rate of occurrence and nature of the mutants. Genetics 94: 961-968
Weir B. S. 1979 Inferences about linkage disequilibrium. Biometrics 35: 235-254[CrossRef][Web of Science][Medline]
Weir B. S. 1996 Genetics data analysis II. Sinauer, Sunderland, Massachusetts, USA
Weir B. S. C. C. Cockerham 1984 Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370[CrossRef][Web of Science]
Wright S. 1951 The genetical structure of populations. Annals of Eugenics 15: 323-354[Web of Science]
Young A. T. Boyle T. Brown 1996 The population genetic consequences of habitat fragmentation for plants. Trends in Ecology and Evolution 11: 413-418[CrossRef]
This article has been cited by other articles:
|
|
 |
|
  B.-H. Song, A. J. Windsor, K. J. Schmid, S. Ramos-Onsins, M. E. Schranz, A. J. Heidel, and T. Mitchell-Olds Multilocus Patterns of Nucleotide Diversity, Population Structure and Linkage Disequilibrium in Boechera stricta, a Wild Relative of Arabidopsis Genetics, March 1, 2009; 181(3): 1021 - 1033. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. KANG, Q. YE, and H. HUANG Genetic Consequence of Restricted Habitat and Population Decline in Endangered Isoetes sinensis (Isoetaceae) Ann. Bot., December 1, 2005; 96(7): 1265 - 1274. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
