| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
What's this? |
Population Biology |
2Departamento de Biología Vegetal, Universidad de Valencia, Burjassot, Valencia, Spain; 3Departamento de Biología Vegetal, ETSIA, UPV, Valencia, Spain; 4Departamento de Agricultura y Economía Agraria, EPS de Huesca, Universidad de Zaragoza, Spain; 5Department of Agricultural Botany, University of Reading, UK; 6Instituto de Biología Vegetal y Biotecnología, Universidad de Talca, Talca, Chile
Received for publication June 2, 2004. Accepted for publication January 31, 2005.
ABSTRACT
Cedrus atlantica (Pinaceae) is a large and exceptionally long-lived conifer native to the Rif and Atlas Mountains of North Africa. To assess levels and patterns of genetic diversity of this species, samples were obtained throughout the natural range in Morocco and from a forest plantation in Arbúcies, Girona (Spain) and analyzed using RAPD markers. Within-population genetic diversity was high and comparable to that revealed by isozymes. Managed populations harbored levels of genetic variation similar to those found in their natural counterparts. Genotypic analyses of molecular variance (AMOVA) found that most variation was within populations, but significant differentiation was also found between populations, particularly in Morocco. Bayesian estimates of FST corroborated the AMOVA partitioning and provided evidence for population differentiation in C. atlantica. Both distance- and Bayesian-based clustering methods revealed that Moroccan populations comprise two genetically distinct groups. Within each group, estimates of population differentiation were close to those previously reported in other gymnosperms. These results are interpreted in the context of the postglacial history of the species and human impact. The high degree of among-group differentiation recorded here highlights the need for additional conservation measures for some Moroccan populations of C. atlantica.
Key Words: Cedrus atlantica • genetic diversity • Morocco • Pinaceae • population structure • RAPD • Spain
Successful management and preservation of forest tree species depend on accurate assessment of genetic diversity to address questions regarding genetic relationships among individuals as well as levels and structure of genetic variation. In particular, knowledge of population genetic structure provides a historical perspective of evolutionary changes that characterize a species and allow us to predict how populations will respond to future events of natural and artificial origin (Wallace, 2002
). In recent years, a series of techniques and genetic markers have been developed to analyze and estimate genetic diversity, but no single technique is universally ideal; each available technique has both strengths and weaknesses (Mueller and Wolfenbarger, 1999
).
Random amplified polymorphic DNA (RAPD) markers have found the widest application in analyses of genetic variation below the species level, despite their drawbacks (see Hedrick, 1992
), particularly in investigations of population structure and differentiation, including estimation of FST analogues and genetic variation within populations (Nybom and Bartish, 2000
). Corroborating it, a recent comparison of different nuclear DNA markers for estimating intraspecific genetic diversity in plants demonstrated that STMS (sequence tagged microsatellite sites)-based among-population diversity estimates are similar to those derived by the dominantly inherited markers (Nybom, 2004
). Such analyses are crucial for conservation genetics, and the rapidity with which RAPD markers can be generated is an advantage because they can deliver crucial information within the time constraints frequently demanded by urgent conservation decisions (Mueller and Wolfenbarger, 1999
). Furthermore, recent papers demonstrate that new statistical approaches, particularly Bayesian methods, for estimating population genetic parameters significantly alleviate the bias related to their dominant nature (Zhivotovsky, 1999
; Holsinger et al., 2002
).
Cedrus atlantica Manetti (Pinaceae) is the most important timber resource in Morocco, covering a surface of 132 000 ha, representing 2.3% of the national forest. The species is now distributed in two unequal and highly fragmented areas (Ezzahiri and Belghazi, 2000
): the Rif (16 000 ha) and both the Middle and High Atlas (116000 ha). Cedrus atlantica now contributes 12% to the annual consumption of coniferous wood, compared to the 17% it contributed 10 years ago. In fact, the cedar forest has an important socioeconomic role in the Middle Atlas. It provides jobs for a large number of people and constitutes the main source of income for rural families in this mountainous area where agricultural activities are limited (Benchekroun, 1994
). Because of their exceptional adaptability to a variety of environmental conditions, cedar trees have been successfully introduced in many countries outside their natural distribution as either ornamental or reforestation species.
In spite of their biological and economic importance, only limited information is available on the extent, distribution, and nature of genetic variability in Cedrus. This is unfortunate because such information is crucial for devising strategies to protect and preserve the genetic resources of the cedar forests. Panetsos et al. (1992
, 1994)
and Scaltsoyiannes (1999)
studied allozyme variation in the genus, but they did not report on population genetic structure. Other investigations have focused on the study of allozyme segregation patterns in a French plantation of C. atlantica (Fallour et al., 2001
), karyotypic analysis revealing the interspecific differentiation within the genus (Bou Dagher-Kharrat et al., 2001a
), and marker-assisted fingerprinting of the offspring resulting from controlled pollination experiments among the Mediterranean cedar taxa (Fady et al., 2003
). Finally, Bou Dagher-Kharrat et al. (2001b)
reported on cedar genetic diversity using amplified fragment length polymorphisms (AFLP).
The main aim of this study was to measure levels of genetic variability and population differentiation of C. atlantica using RAPD markers. Specific aims include measuring levels of genetic variability within populations, identifying levels of genetic differentiation among populations, and providing guidelines for the conservation, management, and restoration of the species. The genetic diversity levels were determined from natural Moroccan populations and 40-yr-old Spanish plantations of C. atlantica. These plantations, of unknown origin, provide an interesting model to study the secondary evolution of an exotic gene pool outside its natural range.
MATERIALS AND METHODS
Plant material
Leaf samples were obtained from a total of 77 C. atlantica individuals (Table 1) from (1) eight populations located in the natural distribution of the species in Morocco: Western Rif (population M1), Western Middle Atlas (M4, M5, M7, M8, M10, and M11), and Eastern High Atlas (M17); and (2) five discrete plantations within a managed Mediterranean forest located in Arbúcies, Girona, Spain (B, F, T, V, and X). Trees were chosen at random throughout each population and sampled directly using silica gel to dry and preserve the leaf tissue until DNA extraction. The number of sampled individuals in each population and the geographical distances among the populations within each region (Morocco and Spain) are shown in Table 1. All Moroccan populations except M10 are of natural origin (MAMVA, 1997
).
|
. Sixty decamer primers (Series OPA, OPB, and OPC from Operon Technologies, Alameda, California, USA) were initially screened to identify well-amplified, polymorphic bands among populations. Six primers from the initial screening process (OPA18, OPA20, OPB7, OPB8, OPB12, and OPC2) that had a high level of polymorphism and the best readability were chosen for further study of the 77 individual genotypes. Duplicate reactions were run for all primers and all individuals. Only those bands consistently reproduced in different analyses were considered. Bands of similar molecular mass and migration distances across individuals were assumed to be homologous. Homology assessments were made across gels based on an amplified standard individual and a Gene Ruler DNA ladder mix (Fermentas AB, Vilnius, Lithuania), that were run on each gel. Control samples containing all reaction material except DNA were used to test that no self-amplification or DNA contamination occurred. To test reproducibility and repeatability in the RAPD profiles, this procedure was performed twice.
Statistical analysis
Amplified fragments, named by the primer used and the molecular mass in base pairs (bp), were scored as either present (1) or absent (0) in each DNA sample, and three binary matrices of the different RAPD phenotypes were assembled (hereafter referred to as data sets Cedro-140, Marruecos-140, and Arbúcies-140). We refrained from pruning the loci that fulfilled Lynch and Milligan's (1994)
criterion because that may lead to significant bias in estimates of population genetic parameters (Lee et al., 2002
). Note also that this bias is substantially eliminated if a high number of polymorphic dominant markers are generated (Krauss, 2000
). The binary data were used to compute pairwise similarity matrices based on the Apostol coefficient using RAPDistance software (Armstrong et al., 1996
).
Both neighbor-joining (NJ) cluster analysis and principal coordinates analysis (PCoA) were carried out on the Apostol distance matrices using the PHYLIP 3.57c (Felsenstein, 1993
) and NTSYS-pc (Rohlf, 1997
) packages, respectively. The reliability of the NJ dendrogram was tested by bootstrap analysis with 500 replications.
An alternative Bayesian-based clustering method was also applied to infer how many clusters or subpopulations (K) were most appropriate for interpreting the data without prior information on the number of locations at which the individuals were sampled. The original data matrices were imported into the STRUCTURE, version 2.1 program (Pritchard et al., 2000
). Ten independent runs of K = 1–13 were performed at 100 000 MCMC (Marcov chain Monte Carlo) repetitions and a burn-in period of 30 000 iterations using no prior information and assuming correlated allele frequencies and no admixture. The posterior probability was then calculated for each value of K using the estimated log-likelihood of K to choose the optimal K.
The distance matrices between RAPD patterns were also used to calculate pairwise genetic distances between populations. These distance matrices were used to construct dendrograms using the NJ method. The relationships between matrices of genetic and linear geographic distances were examined with a Mantel test in NTSYS. Resulting r values were interpreted as correlation coefficients. All dendrograms were displayed and printed using TREEVIEW software (Page, 1996
).
Two different analyses of molecular variance (AMOVA) were performed to study the partitioning of genetic variation within and among C. atlantica populations using the WINAMOVA 1.5 program (available from L. Excoffier, Genetics and Biometry Laboratory, University of Geneva, Switzerland). First, the selected bands were analyzed directly as phenotypes using the Euclidean distance matrix of Excoffier et al. (1992)
. Second, we performed a genotypic AMOVA as extended by Stewart and Excoffier (1996)
for RAPD data assuming random mating (S = 0). An alternative Bayesian approach (Holsinger et al., 2002
) was also used to obtain independent estimates of FST (
B and Gst-B, the Bayesian analogs of FST and Nei's GST, respectively). The original data matrix was imported into HICKORY, version 1.0 (Holsinger and Lewis, 2003
) and used for a full model, f (the inbreeding within populations) = 0 model, B = 0 model, and f-free model. We conducted several runs with default sampling parameters (burn-in = 50 000, sample = 250 000, thin = 50) to ensure that the results were consistent. The deviance information criterion (DIC; Spiegelhalter et al., 2002
) was used to estimate the fit between the data and a particular model and to choose among models (Holsinger and Wallace, 2004
). The differentiation among populations was finally analyzed using the exact test for population differentiation of Raymond and Rousset (1995)
as implemented in the TFPGA program (Miller, 1997
), which has algorithms designed for dominant markers.
Within-population genetic diversity was assessed for each population as (1) percentage of polymorphic loci; (2) Nei's (1978)
unbiased expected gene diversity (H), assuming Hardy-Weinberg equilibrium, as implemented in the TFPGA program; (3) Bayesian gene diversity (HB) as implemented in the HICKORY program, which does not assume Hardy-Weinberg equilibrium within populations (see before); and (4) Shannon's information index as implemented in POPGENE software (Yeh et al., 1997
).
RESULTS
RAPD fingerprinting
The six selected primers generated a total of 140 consistently and well-amplified bands, ranging in size from 400 to 2250 bp. Most of these bands (98%) were polymorphic among the 13 populations (Table 2). Reflecting this high level of genetic polymorphism, no individual had the same band pattern over all the primers used.
|
Relationships among RAPD phenotypes
First, the 140 RAPD phenotypes (the Cedro-140 data set) were analyzed using the Apostol coefficient. The unrooted NJ dendrogram based on these data clustered the 77 individuals within distinct groups roughly according to their geographical origin. Two main clusters were formed: one included the populations M4, M5, M10, and M11 from Morocco (the most geographically closely located populations; see Table 1), and the other included the rest of the Moroccan populations (M1, M7, M8, and M17) and those from Arbúcies (B, F, T, V, and X). In this second cluster, individuals from each region formed two independent subclusters. In the Arbúcies subcluster, all five populations formed minor clusters of their own. In Morocco, the individuals from populations M4, M5, and M11 (integrated in the first cluster) and those from population M1 (included in the second) also formed minor clusters of their own; the remaining individuals appeared intermingled in their corresponding subcluster. A majority-rule consensus tree derived from the 500 replications revealed similar relationships (Fig. 1).
|
|
1.000) was the model with K = 7; the other models (K = 1–6 and K = 8–13) were completely insufficient to explain the data (P 0). According to these results, seven is the most probable number of clusters (subpopulations) containing the 77 C. atlantica individuals. The membership frequency of each predefined population (1 to 13) in each of the seven inferred clusters is shown in Table 3. The 52 individuals from Morocco were grouped into three clusters, with mean membership proportions of 1 (populations M4, M5, M10, and M11), 0.87 (M1 and M7), and 0.86 (M8 and M17). Note that while cluster 1 regrouped individuals exclusively from the four most geographically close populations (see Table 1), clusters 2 and 3 had individuals from the three Moroccan regions (Rif, Middle Atlas, and High Atlas) in which they were sampled. In Arbúcies, most of the individuals from populations B, F, and X grouped in their respective original subpopulations (membership proportions of 1, 0.91, and 1), while individuals from populations T and V grouped in a single cluster (mean membership proportion of 0.93). The consistency of these results was corroborated by executing the program STRUCTURE with the Marruecos-140 and Arbúcies-140 data sets. The only models that explained the data sufficiently in Morocco and Arbúcies were K = 3 (P 1.000) and K = 4 (P 1.000), respectively. Thus, the cluster pattern based on Bayesian statistics indicates the existence of a clear structure in C. atlantica populations at both inter- and intraregional levels.
|
showed that populations M4, M10, M11, and M5 were clearly different from the rest of populations. Thus, the matrix of combined probabilities for each pairwise comparison demonstrated that significant differences were only obtained when populations M4, M10, M11, and M5 were compared with the remaining nine populations (data not shown). Note also that the topology of dendrogram obtained from the Cedro-140 data set is in good agreement with the seven subpopulations inferred from the program STRUCTURE (Table 3).
|
Population genetic structure
AMOVA analysis from the phenotypic distance matrix (Cedro-140 data set) permitted a partitioning of the overall variation into three levels (Table 4). Although most of this variation was found within populations (63.8%), there was also evidence for a significant phenotypic structure of the populations (
ST = 0.362, P < 0.0003). The geographical structure among populations was maintained when the two regions (Morocco and Arbúcies) were separately analyzed (30.9% and 20.6% of population differences, respectively).
|
ST = 0.285, P < 0.0002). Note, however, that this analysis showed that the partitioning of genetic differentiation between Morocco and Arbúcies was not significant (CT = 0.054, P = 0.1339). For the within-regions analyses, the highest proportion of genetic variation was between individuals within populations, but again a significant population structure was evident in each region, especially in Morocco (ST = 0.302 vs. 0.105 in Arbúcies).
The genotypic AMOVA procedure was also implemented to study separately the genetic structure of the two main groups of populations in Morocco (group 1: populations M4, M5, M10, and M11; group 2: populations M1, M7, M8, and M17; see Table 5). Among-population genetic differentiation within each group (ST = 0.148, P < 0.001 for group 1 and ST = 0.193, P < 0.001, for group 2) was smaller than that observed when AMOVA was performed considering the eight populations (ST = 0.362). These results show that among-population differentiation in group 2, encompassing populations located 280 km apart, is only slightly higher than that observed for group 1, which had the geographically closest populations (see Table 1).
|
B, the Bayesian estimate of population structure, is based on the f-free model from the Hickory program (Holsinger and Lewis, 2003
). In this model, the posterior mean of B incorporates all of the uncertainty in the prior of f and does not appear to be greatly affected by the prior of f when there are a large number of polymorphic loci (Holsinger and Lewis, 2003
). For the 13 populations (Cedro-140 data set), this model (DIC = 4306.76) generated a mean B of 0.303 ± 0.016 (95% confidence intervals 0.271 and 0.334) and a Gst-B of 0.286 ± 0.014 (95% confidence intervals 0.258 and 0.312). The B = 0 model (DIC = 6821.28) provides evidence of the genetic structure among populations. Among Moroccan populations (Marruecos-140 data set), the f-free model (DIC = 2638.80) yielded a mean B of 0.304 ± 0.019 (95% confidence intervals 0.267 and 0.341) and a Gst-B of 0.276 ± 0.015 (95% confidence intervals 0.247 and 0.304). The B = 0 model (DIC = 4180.71) indicated that there is compelling evidence for the existence of population genetic structure. The Arbúcies-140 data set yielded similar results: DIC values of 1607.13 and 1972.37 for the f-free and B = 0 models, respectively; B = 0.178 ± 0.022 (95% confidence intervals 0.136 and 0.222) and a Gst-B of 0.148 ± 0.017 (95% confidence intervals 0.116 and 0.182). The Bayesian approach was also used to estimate FST in the two groups of Moroccan populations (group 1: populations M4, M5, M10, and M11; group 2: populations M1, M7, M8, and M17). Mean B and Gst-B values for group 1 were 0.165 ± 0.022 (95% confidence intervals 0.124 and 0.209) and 0.129 ± 0.015 (95% confidence intervals 0.100 and 0.159), respectively (DIC values of 1210.70 and 1517.08 for the f-free and B = 0 models, respectively). Mean B and Gst-B values for group 2 were 0.091 ± 0.019 (95% confidence intervals 0.056 and 0.132) and 0.070 ± 0.015 (95% confidence intervals 0.042 and 0.100), respectively (DIC values of 1280.59 and 1390.77 for the f-free and B = 0 models, respectively). All these results show that the f-free model yielded estimates of FST in concordance with those previously described in the extension of the AMOVA analysis (see Tables 4 and 5).
Within population diversity
The percentage of polymorphic loci (Pp) ranged from 38.6% (population M5) to 62.1% (population M1), with an overall mean of 51.0%. Values of Nei's gene diversity (H) assuming Hardy-Weinberg equilibrium ranged from 0.151 (population M5) to 0.233 (population F), with a mean value of 0.194. Both Bayesian (HB) and Shannon's index (I) estimates were slightly higher (0.239 and 0.270, respectively; Table 6). Nevertheless, the correlation between these estimates among populations was high (r = 0.83, P = 0.0005 for H vs. HB; r = 0.993, P = 0.000 for H vs. I; and r = 0.852, P = 0.0002 for HB vs. I). The Mann-Whitney U test between Morocco and Arbúcies populations showed no significant difference for any of the four parameters used to estimate within-population genetic diversity (data not shown). No significant correlation could be found between population sample size (see Table 1) and Pp, H, or I (data not shown). In contrast, Bayesian diversity estimates were slightly higher in those populations with a smaller sample size (N = 13, r = –0.63, P = 0.02).
|
The use of dominant markers to assess genetic variability among individuals and populations is promising because many polymorphic loci can be obtained fairly easily, in a relatively short time and at low cost, without any prior knowledge of the genome of the species under study (Mueller and Wolfenbarger, 1999
; Nybom and Bartish, 2000
; Nybom, 2004
). Therefore, RAPDs should find major population genetics applications, notably in the field of genetic conservation, where molecular markers need to be developed at a reasonable cost (Hardy, 2003
).
In our study, RAPD markers proved to be a powerful method for the detection of spatial genetic variation. With six primers, we obtained 137 polymorphic markers and could differentiate the 77 C. atlantica individuals analyzed, reflecting a rich allelic diversity in the populations.
The genetic relationships observed between some individuals in our study are noteworthy. The distance-based clustering methods (NJ and PCoA analyses) revealed a strong structure among the 13 investigated populations of C. atlantica; three population groups were roughly differentiated. One group, located in Morocco (populations M4, M5, M10, and M11), seems to be genetically isolated, whereas the remaining two groups located in Morocco (populations M1, M7, M8, and M17) and Spain (populations B, F, T, V, and X) are genetically more similar to each other. Furthermore, the exact test for population differentiation revealed significant differences among populations only when populations M4, M5, M10, and M11 were compared with the remaining nine populations.
Distance-based clustering methods are more suited to exploratory data analysis than to precise statistical inference (see Pritchard et al., 2000
). Because of this, we also used a model-based clustering method based on Bayesian statistics to analyze our RAPD data. This approach indicates clearly that C. atlantica subpopulations are not a panmictic population, and at least seven subpopulations exist within the two sampled areas (three in Morocco and four in Spain). Our model-based clustering method also confirms that subpopulations M4, M5, M10, and M11 represent a genetically distinct population.
RAPDs indicate that C. atlantica appears to maintain a high level of genetic diversity (Pp = 51.0%; H = 0.194), as observed in most woody species based on isozymes (Pp = 59.5%, H = 0.183 for angiosperms, and Pp = 71.1%, H = 0.169 for gymnosperms; Hamrick et al., 1992
). In several coniferous species, RAPD and isozyme markers generally show similar levels of polymorphism (Aagaard et al., 1998
; Lee et al., 2002
and references therein), and C. atlantica was not an exception. Thus, our results were quite similar to those based on allozymes (Pp = 50%, H = 0.183) in six C. atlantica populations from France, Morocco, and Algeria (Scaltsoyiannes, 1999
). Our estimate of intrapopulation variability was, however, higher than that based on AFLP markers (0.11) in a Moroccan population (Bou Dagher-Kharrat et al., 2001b
). Furthermore, our H values are within the 99% confidence limits for expected heterozygosity (0.189–0.227) in 50 pine species (Ledig, 1998
; Delgado et al., 2002
). The Bayesian (HB) and Shannon's index estimates of intraspecific genetic diversity within C. atlantica were slightly higher (0.239 and 0.270, respectively). Nevertheless, the three measures were significantly correlated. Shannon's information index has general applications in ecology and is relatively insensitive to the skewing effects caused by the inability to detect heterozygous loci (Dawson et al., 1995
). Unfortunately, comparison with Shannon's indices derived for other species is complicated by the fact that other studies have employed different approaches for its calculation (Gillies et al., 1997
; Wolf et al., 1997
). In this study, we adopted the approach employed by Yeh et al. (1997)
, which gives a range of values between 0 and 1 (maximum phenotypic diversity). In the few cases where Shannon's information indices could be compared, the extent of variation recorded in C. atlantica populations (0.210 to 0.316) was lower than in other gymnosperms such as Fitzroya cupressoides (0.343 to 0.636; Allnut et al., 1999
) or Pilgerodendron uviferum (0.337 to 0.716; Allnut et al., 2003
). To date, very few studies have reported Bayesian gene diversity estimates based on RAPD markers in coniferous species. Note, however, that Bayesian estimates of expected heterozygosity in two populations of Pinus silvestris (see Zhivotovsky, 1999
) are comparable to those obtained in our C. atlantica populations using the Bayesian approach of Holsinger and Lewis (2003)
.
Our RAPD-based AMOVA studies revealed that most of the variation in C. atlantica was recorded within populations, a result consistent with those from most other woody perennial, outbreeding plant species, especially conifers (Hamrick et al., 1992
). In a comprehensive review, Nybom and Bartish (2000)
assembled reported RAPD-based GSt and AMOVA-derived 
St values for about 100 plant species and confirmed the tendency of gymnosperms to display lower values of population differentiation than angiosperms, as indicated by previous studies using isozymes (Hamrick et al., 1992
). However, in contrast to the previously mentioned studies, RAPD-based values for between-population diversity increased with increasing distributional range of the species (Nybom and Bartish, 2000
). Although the reason for this difference between marker systems is not fully understood (see Nybom and Bartish, 2000
), this finding could explain the high degree of genetic differentiation recorded between populations in our among-regions analysis (ST = 0.362; B = 0.303; Gst-B = 0.286). However, our estimates of the fixation indices for the Moroccan populations (ST = 0.309; B = 0.304; Gst-B = 0.276) are significantly higher than those reported for other conifers and cannot be explained with the theoretical predictions regarding isolation by distance. For example, in a large allozyme data compilation, a mean GST value of only 0.073 was recorded for the 121 species of gymnosperms examined (Hamrick et al., 1992
). Similarly, the mean genetic diversity among populations in 50 pine species, as measured by allozymes, was 0.129 (Ledig, 1998
; Delgado et al., 2002
). Comparison with AMOVA-derived ST estimates based on RAPD analyses of other conifer species also indicates that the value recorded for the Moroccan C. atlantica populations is relatively high (0.309 vs. 0.24; Nybom and Bartish, 2000
).
Both distance-based and model-based clustering analyses, as well as the exact test for population differentiation, showed that C. atlantica populations from Morocco comprise two genetically distinct groups (populations M4, M5, M10, and M11 on the one hand, and populations M1, M7, M8, and M17 on the other). Within each group, Bayesian FST or GST estimates for population differentiation (0.165 or 0.129 and 0.091 or 0.070, respectively) are close to those given earlier for the analysis of population structure in other gymnosperms. Furthermore, the FST value found in our study for populations M1, M7, M8, and M17 was similar or somewhat lower than that reported by Scaltsoyiannes et al. (1999)
in Abies pinsapo and A. numidica using isozymes (0.102). Both Abies and Cedrus taxa may have a common evolutionary history in Africa (Scaltsoyiannes et al., 1999
). Thus, we suggest that the high FST value obtained in the analysis of the eight Moroccan populations may be due to genetic isolation of the populations M4, M5, M10, and M11. This result is surprising, given their geographical proximity to populations M7 and M8, which in turn are more genetically related to the most geographically distant populations (M17 and, especially, M1). Such differences between the two groups of populations suggest either that they have been genetically isolated at some time in the past, and/or that gene flow between them is restricted at the present time, providing scope for genetic differentiation as a result of drift or selection. Because of this, it is pertinent to consider these results in the context of the postglacial history of C. atlantica, as indicated by pollen analysis. The continuous presence of Cedrus pollen during several middle and upper Quaternary interglacial–glacial cycles (see Cheddadi et al., 1998
; Magri and Parra, 2002
, and references therein) suggest that cedar, although with possible geographical shifts, was continuously located in the Maghreb region rather than in distant refugia. Thus, the present close relationships among populations more than 250 km distant may only reflect historical gene flow between C. atlantica populations. If so, the differentiation of M4, M5, M10, and M11 populations could result from modern causes. In fact, anthropogenic exploitation of the area since c. 1300 has had a greater effect on the forest ecosystem than any of the Holocene intervals (Lamb and van der Kaars, 1995
). Corroborating this, we note that population M10 has an artificial origin.
In the course of early domestication, substantial unintentional genetic change may occur as a consequence of the sampling process, natural selection, and alterations in reproductive behavior. This may generate significant differences between seed samples harvested from wild stands and those derived from domesticated populations (Stoehr and El-Kassaby, 1997
; Zheng and Ennos, 1999
). Although the Spanish C. atlantica plantation is of unknown origin, our RAPD data suggest that it probably originated from populations M1, M7, M8, or M17, or related populations. Using the data from natural populations as a baseline, we can explore the extent of genetic change detectable with RAPD markers in the five C. atlantica managed populations sampled here. The results of this study demonstrate that managed populations harbor levels of genetic variation that are similar to the variation in their natural population counterparts.
Our results have a number of implications for the development of conservation strategies for C. atlantica. The detection of population differentiation using techniques such as RAPDs may assist in the definition of appropriate units for conservation, thus providing an appropriate focus for conservation management or monitoring (Newton et al., 1999
). In C. atlantica, the two main population groupings detected within Morocco may merit individual conservation attention. The populations M4, M5, M10, and M11 appear to be particularly distinctive and should perhaps be accorded highest priority for conservation. The definition of such management units will also be of value for informing the sampling of germplasm for ex situ conservation activities and for sourcing material for restoring degraded populations of the species.
Despite the limited number of sampled trees in some populations (but see Dodd and Kashani, 2003
), coherent results of the molecular analysis with different approaches were obtained in our RAPD survey of the Moroccan and Spanish populations of C. atlantica. Furthermore, observed variabilities within and between populations coincided with expectations. Thus, RAPDs represent a useful tool for the discrimination of populations and for comparative variability studies in Cedrus. The results of our study also demonstrate that C. atlantica individuals could be uniquely identified based on their RAPD fingerprints. Finally, this study highlights the importance of molecular analysis in understanding the genetic systems governing natural populations and contributes to the knowledge of conservation of genetic resources in the Moroccan flora.
FOOTNOTES
1 The authors thank the European Union (Contract number: ERBIC18-CT97-0177) and the Generalitat Valenciana (Groups 03/102) for financial support, J. Garolera for facilities when sampling populations from Arbúcies, and D. Lindsay for revision of the manuscript. 
7 E-mail: juan.segura{at}uv.es
LITERATURE CITED
Aagaard J. E. K. V. Krutovskii S. H. Strauss 1998 RAPDs and allozymes exhibit similar levels of diversity and differentiation among populations and races of Douglas-fir. Heredity 81: 69-78[CrossRef][Web of Science]
Allnutt T. R. A. C. Newton A. Lara A. Premoli J. J. Armesto R. Vergara M. Gardner 1999 Genetic variation in Fitzroya cupressoides (alerce), a threatened South American conifer. Molecular Ecology 8: 975-987[CrossRef][Medline]
Allnutt T. R. A. C. Newton A. Premoli A. Lara 2003 Genetic variation in the threatened South American conifer Pilgerodendron uviferum (Cupressaceae), detected using RAPD markers. Biological Conservation 114: 245-253[CrossRef][Web of Science]
Armstrong J. A. Gibbs R. Peakall G. Weiler 1996 RAPDistance programs: version 1.04 for the analysis of patterns of RAPD fragments. Australian National University, Canberra, Australia
Benchekroun F. 1994 The economy of the Moroccan cedar forest and its impact on the development of local collectivities. Annales de la Recherche Forestiere au Maroc 27: 713-724
Bou Dagher-Kharrat M. G. Grenier M. Bariteau S. Brown S. Siljak-Yakovlev A. Savouré 2001a Karyotype analysis reveals interspecific differentiation in the genus Cedrus despite genome size and base composition constancy. Theoretical and Applied Genetics 103: 846-854[CrossRef][Web of Science]
Bou Dagher-Kharrat M. S. Mariette B. Fady F. Lefevre G. Grenier C. Plomion S. Arnould 2001b Analysis of Cedrus genetic geographic diversity using AFLPs. In C. Plomion [ed.], Wood, breeding, biotechnology and industrial expectation, Bordeaux, France, June 2001, 93. INRA, Bordeaux, France
Cheddadi R. H. F. Lamb J. Guiot S. van der Kaars 1998 Holocene climatic change in Morocco: a quantitative reconstruction from pollen data. Climate Dynamics 14: 883-890[CrossRef][Web of Science]
Dawson I. K. A. J. Simons R. Waugh W. Powell 1995 Diversity and genetic differentiation among subpopulations of Gliricidia sepium revealed by PCR-based assays. Heredity 74: 10-18
Delgado P. A. Cuenca A. E. Escalante F. Molina-Freaner D. Piñero 2002 Comparative genetic structure in pines: evolutionary and conservation consequences. Revista Chilena de Historia Natural 75: 27-37[Web of Science]
Dodd R. S. N. Kashani 2003 Molecular differentiation and diversity among the California red oaks (Fagaceae; Quercus section Lobatae). Theoretical and Applied Genetics 107: 884-892[CrossRef][Web of Science]
Excoffier L. P. E. Smouse J. M. Quattro 1992 Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131: 479-491[Abstract]
Ezzahiri M. B. Belghazi 2000 Synthèse de quelques résultats sur la régénération naturelle du cèdre de l'Atlas au Moyen Atlas (Maroc). Sécheresse, Science et Changements Planétaires 11: 79-84
Fady B. F. Lefèvre M. Reynaud G. G. Vendramin M. Bou Dagher-Kharrat M. Anzidei R. Pastorelli A. Savouré M. Bariteau 2003 Gene flow among different taxonomic units: evidence from nuclear and cytoplasmic markers in Cedrus plantation forests. Theoretical and Applied Genetics 107: 1132-1138[CrossRef][Web of Science][Medline]
Fallour D. B. Fady F. Lefèbre 2001 Evidence of variation in segregation patterns within a Cedrus population. Journal of Heredity 92: 260-266
Felsenstein J. 1993 PHYLIP: phylogeny inference package, version 3.5p. Department of Genetics, University of Washington, Seattle, Washington, USA
Gillies A. C. M. J. P. Cornelius A. C. Newton C. Navarro M. Hernández J. Wilson 1997 Genetic variation in Costa Rican populations of the tropical timber tree Cedrela odorata L. (Spanish cedar), assessed using RAPDs. Molecular Ecology 6: 1133-1145[CrossRef]
Hamrick J. L. M. J. W. Godt S. L. Sherman-Broyles 1992 Factors influencing levels of genetic diversity in woody plant species. New Forests 6: 95-124[CrossRef]
Hardy O. J. 2003 Estimation of pairwise relatedness between individuals and characterization of isolation-by-distance processes using dominant genetic markers. Molecular Ecology 12: 1577-1588[CrossRef][Medline]
Hedrick P. 1992 Shooting the RAPDs. Nature 355: 679-680
Holsinger K. E. P. Lewis 2003 HICKORY: a package for analysis of population genetic data V1.0. Department of Ecology and Evolutionary Biology, University of Connecticut, Storrs, Connecticut, USA. Available at website http://www.eeb.uconn.edu
Holsinger K. E. P. O. Lewis K. D. Dipak 2002 A Bayesian approach to inferring population structure from dominant markers. Molecular Ecology 11: 1157-1164[CrossRef][Medline]
Holsinger K. E. L. A. Wallace 2004 Bayesian approaches for the analysis of population genetic structure: an example from Platanthera leucophaea. Molecular Ecology 13: 887-894[CrossRef][Medline]
Krauss S. L. 2000 Accurate gene diversity estimates from amplified fragment length polymorphism (AFLP) markers. Molecular Ecology 9: 1241-1245[CrossRef][Medline]
Lamb H. F. S. van der Kaars 1995 Vegetational response to Holocene climatic change: pollen and palaeolimnologic data from the Middle Atlas, Morocco. The Holocene 5: 400-408
Ledig F. T. 1998 Genetic variation in Pinus. In D. M. Richardson [ed.], Ecology and biogeography of Pinus, 251–280. Cambridge University Press, Cambridge, UK
Lee S.-W. F. T. Ledig D. R. Johnson 2002 Genetic variation at allozyme and RAPD markers in Pinus longaeva (Pinaceae) of the White Mountains, California. American Journal of Botany 89: 566-577
Lynch M. B. G. Milligan 1994 Analysis of population genetic structure with RAPD markers. Molecular Ecology 3: 91-99[Medline]
Magri D. I. Parra 2002 Late quaternary western Mediterranean pollen records and African winds. Earth and Planetary Science Letters 200: 401-408[CrossRef][Web of Science]
MAMVA (Ministère de l'agriculture et de la mise en valeur agricole). 1997 Les ressources génétiques forestières au Maroc, vol. 2, Cédre de l'Atlas (Cedrus atlantica Manetti). Service des Semences et Pépinières, Salé, Morocco
Miller M. 1997 Tools for population genetic analyses (TFPGA) 1.3: a windows program for the analysis of allozyme and molecular population genetic data. Department of Fisheries and Wildlife, Utah State University, Logan, UT, USA. Available at website http://bioweb.usu.edu/mpmbio/ index.htm
Mueller U. G. L. L. Wolfenbarger 1999 AFLP genotyping and fingerprinting. Trends in Ecology and Evolution 14: 389-394
Nebauer S. G. L. del Castillo-Agudo J. Segura 2000 An assessment of genetic relationships within the genus Digitalis based on PCR-generated RAPD markers. Theoretical and Applied Genetics 100: 1209-1216[CrossRef][Web of Science]
Nei M. 1978 Estimation of average heterozygosities and genetic distances from a small number of individuals. Genetics 89: 583-590
Newton A. C. T. R. Allnutt A. C. M. Gillies A. J. Lowe R. A. Ennos 1999 Molecular phylogeography, intraspecific variation and the conservation of tree species. Trends in Ecology and Evolution 14: 140-145
Nybom H. 2004 Comparison of different nuclear DNA markers for estimating intraspecific genetic diversity in plants. Molecular Ecology 13: 1143-1155[CrossRef][Medline]
Nybom H. I. V. Bartish 2000 Effects of life history traits and sampling strategies on genetic diversity estimates obtained with RAPD markers in plants. Perspectives in Plant Ecology, Evolution and Systematics 3: /2 93-114
Page R. D. M. 1996 TREEVIEW: an application to display phylogenetic trees on personal computers. Computer Applications in the Biosciences 12: 357-358
Panetsos K. P. A. Christou A. Scaltsoyiannes 1992 First analysis on allozyme variation in Cedar species (Cedrus sp). Silvae Genetica 41: 339-342[Web of Science]
Panetsos K. P. A. Scaltsoyiannes M. Tsaktsira 1994 Genetic variation in allozymes of Cedrus libani A. Rich and Cedrus atlantica Mannetti. Annales de la Recherche Forestiere au Maroc 27: 420-434
Pritchard J. K. M. Stephans P. Donnelly 2000 Inference of population structure using multilocus genotype data. Genetics 155: 945-959
Raymond M. L. F. Rousset 1995 An exact test for population differentiation. Evolution 49: 1280-1283[CrossRef][Web of Science]
Rohlf F. J. 1997 NTSYS: numerical taxonomy and multivariate analysis system, v. 2.0. Exeter Software. Setauket, New York, USA
Scaltsoyiannes A. 1999 Allozyme differentiation and phylogeny of cedar species. Silvae Genetica 48: 61-68[Web of Science]
Scaltsoyiannes A. M. Tsaktsira A. D. Drouzas 1999 Allozyme differentiation in the Mediterranean firs (Abies, Pinaceae). A first comparative study with phylogenetic implications. Plant Systematics and Evolution 216: 289-307[CrossRef][Web of Science]
Spiegelhalter D. J. N. G. Best B. P. Carlin A. van der Linde 2002 Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society, Series B 64: 583-639[CrossRef]
Stewart C. N. L. Excoffier 1996 Assessing population genetic structure and variability with RAPD data: application to Vaccinium macrocarpon (American cranberry). Journal of Evolutionary Biology 9: 153-171[CrossRef][Web of Science]
Stoehr M. U. Y. A. El-Kassaby 1997 Levels of genetic diversity at different stages of the domestication cycle of interior spruce in British Columbia. Theoretical and Applied Genetics 94: 83-90[CrossRef][Web of Science]
Wallace L. E. 2002 Examining the effects of fragmentation on genetic variation in Platanthera leucophaea (Orchidaceae): inferences from allozyme and random amplified polymorphic DNA markers. Plant Species Biology 17: 37-49
Wolf K. S. El-Akkad R. J. Abbot 1997 Population structure in Alkanna orientalis (Boraginaceae) in the Sinai Desert, in relation to its pollinator behavior. Molecular Ecology 6: 365-372
Yeh F. C. R. C. Yang T. Boyle Z. H. Ye J. X. Mao 1997 POPGENE, the user friendly shareware for population genetic analysis. Molecular Biology and Biotechnology Centre, University of Alberta, Canada
Zheng Y. Q. R. A. Ennos 1999 Genetic variability and structure of natural and domesticated populations of Caribbean pine (Pinus caribaea Morelet). Theoretical and Applied Genetics 98: 765-771[CrossRef][Web of Science]
Zhivotovsky L. A. 1999 Estimating population structure in diploids with multilocus dominant DNA markers. Molecular Ecology 8: 907-913[CrossRef][Medline]
CiteULike Complore Connotea Del.icio.us Digg Facebook Reddit Technorati Twitter What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Terrab, A. Hampe, O. Lepais, S. Talavera, E. Vela, and T. F. Stuessy Phylogeography of North African Atlas cedar (Cedrus atlantica, Pinaceae): Combined molecular and fossil data reveal a complex Quaternary history Am. J. Botany, October 1, 2008; 95(10): 1262 - 1269. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Dubreuil, M. Riba, and M. Mayol Genetic structure and diversity in Ramonda myconi (Gesneriaceae): effects of historical climate change on a preglacial relict species Am. J. Botany, May 1, 2008; 95(5): 577 - 587. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Terrab, O. Paun, S. Talavera, K. Tremetsberger, M. Arista, and T. F. Stuessy Genetic diversity and population structure in natural populations of Moroccan Atlas cedar (Cedrus atlantica; Pinaceae) determined with cpSSR markers Am. J. Botany, September 1, 2006; 93(9): 1274 - 1280. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
= individuals from M4, M5, M10, and M11 populations; 
= individuals from M1, M7, M8, and M17 populations; 
= individuals from B, F, T, V, and X populations. Population abbreviations are defined in 