SEARCH

SEARCH BY CITATION

Keywords:

  • bottleneck;
  • effective population size;
  • microsatellite;
  • mutation-drift equilibrium;
  • null alleles;
  • Pinus radiata;
  • population structure

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

Most conifer species occur in large continuous populations, but radiata pine, Pinus radiata, occurs only in five disjunctive natural populations in California and Mexico. The Mexican island populations were presumably colonized from the mainland millions of years ago. According to Axelrod (1981), the mainland populations are relicts of an earlier much wider distribution, reduced some 8000 years ago, whereas according to Millar (1997, 2000), the patchy metapopulation-like structure is typical of the long-term population demography of the species. We used 19 highly polymorphic microsatellite loci to describe population structure and to search for signs of the dynamics of population demography over space and time. Frequencies of null alleles at microsatellite loci were estimated using an approach based on the probability of identity by descent. Microsatellite genetic diversities were high in all populations [expected heterozygosity (He) = 0.68–0.77], but the island populations had significantly lower estimates. Variation between loci in genetic differentiation (FST) was high, but no locus deviated statistically significantly from the rest at an experiment wide level of 0.05. Thus, all loci were included in subsequent analysis. The average differentiation was measured as FST = 0.14 (SD 0.012), comparable with earlier allozyme results. The island populations were more diverged from the other populations and from an inferred common ancestral gene pool than the mainland ones. All populations showed a deficiency of expected heterozygosity given the number of alleles, the mainland populations more so than the island ones. The results thus do not support a recent important contraction in the mainland range of radiata pine.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

The distributions of most animal and plant populations have fluctuated over their history and species ranges have contracted and expanded. The variation of local census sizes will influence the genetically effective population sizes over generations. Isolated populations will start diverging due to mutation and drift. If the populations remain in contact, short-term differentiation at neutral loci will be governed by a balance between migration and drift. These demographic events can have long-lasting effects on the pattern of genetic variation in populations (Nei et al., 1975; Wakeley & Aliacar, 2001). Conversely, patterns of genetic variation can provide information about demographic processes.

Many pine species have large continuous distributions with low-genetic differentiation across the range at marker genes (Ledig, 1998). The range of Pinus radiata D. Don is among the smallest of pine species. It is restricted to five disjunctive populations in Western North America, three on the mainland in Northern California, two on Mexican oceanic islands (Fig. 1). The mainland populations are currently large with hundreds of thousands of trees (Moran et al., 1988), whereas the island populations are considerably smaller. In particular, the Guadalupe population had fewer than 400 trees in the late 1900s, with regeneration absent due to foraging by feral goats (Lavery & Mead, 1998). According to Axelrod (1981), a reduction in the range of the P. radiata coastal populations has occurred within the last 8000 years. According to Millar (1997, 2000), on the other hand, a shifting mosaic structure for P. radiata populations has existed over the last 2 million years. The two Mexican island populations, Cedros and Guadalupe, are geographically isolated and morphologically distinct from the mainland populations. The Cedros population (Pinus radiata var. cedrosensis) is supposed to have separated first from the mainland populations up to 10 million years ago (MYA), which translates into about 200 000 generations (assuming 50 years per generation). The evolutionary history of the Guadalupe population (Pinus radiata var. binata) is not as well known. According to Axelrod (1981), P. radiata has colonized the island some time in the Pliocene (1–4 MYA), corresponding to 20 000–80 000 generations.

image

Figure 1. Map of the localities of natural populations of P. radiata in California and on the Mexican islands.

Download figure to PowerPoint

The biogeographical history suggests that P. radiata has independently colonized two oceanic islands with associated isolation and reduced long-term population size and has recently been reduced in population size in one of these islands. On the mainland, population structure has either been a long-lasting metapopulation with semi-isolated populations or a continuous population has been reduced seriously in range and effective population size about 8000 years ago. The demographic history should result in lower population variation and higher differentiation on the islands. Furthermore, a sharp reduction of population size on the mainland should give rise to traces of bottlenecks (excess heterozygosity).

Early genetic studies of natural populations of P. radiata using protein electrophoresis (Moran et al., 1988) have shown that overall genetic diversity in this species is lower than in many other conifers (HT = 0.117) and genetic differentiation between populations is higher (FST =16.2%). Both values differ from average out crossing wind-pollinated species (Hamrick et al., 1992). The whole species had very little variation with respect to cpDNA (Hong et al., 1993). The allozyme study also suggested that the adult genotypic frequencies were in Hardy–Weinberg equilibrium in all populations, suggesting no inbreeds among adult trees (Moran et al., 1988; Hong et al., 1993). In the early allozyme studies (Plessas & Strauss, 1986; Moran et al., 1988) and later ones with RAPD markers (Wu et al., 1999), genetic diversities were quite similar in all populations. A later study with highly polymorphic microsatellites and using more efficient recent statistical methods, showed that in fact selfing or close inbreeding was very rare or absent on the mainland and moderately low on the islands (Vogl et al., 2002). On the islands, inbreeding reduced the mutation load to a point where selfing or close inbreeding became possible, in contrast to most other conifers studied to date. These results suggest that there may have been a bottleneck and a lowered long-term effective population size on the islands.

In the present study, we will use the powerful highly polymorphic microsatellite markers along with recently developed statistical methods to examine how the suggested demographic history is reflected in the genetic structure and compare the results with those from isozyme data. Can we distinguish between alternative scenarios for the recent history of the mainland populations, i.e. significant contraction (Axelrod, 1981) or continuing fluctuations (Millar, 2000)? Are we able to detect the effects of changing population size and bottlenecks in the island populations, where census sizes are smaller than on the mainland?

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

Plant materials

Large seed collections were made earlier in the five natural populations of P. radiata: Año Nuevo, Cambria, Monterey, Cedros and Guadalupe (Eldridge, 1978) (Fig. 1). Material from these populations was planted 1980 at Uriarra near Canberra (latitude 35°17′S, longitude 148 °53′E). In this study, we collected needles from 30 to 34 20-year-old trees from stands representing each population. All trees were from different open-pollinated families. The trees were chosen so as to give equal representation from stands in each population and hence to give good coverage of all populations.

Conifer seeds have haploid megagametophyte storage tissue, which has the same genotype as the corresponding ovule. To estimate the frequency of null alleles, DNA was also isolated from 30 Año Nuevo seed megagametophytes. Each megagametophyte was from the same open pollinated family of the natural population as the diploid sample that we used in our population analysis.

DNA isolation

DNA was isolated from needles and seeds (megagametophytes) using the Fast Prep machine according to the manufacturer's instructions (Savant Instrument Inc., Holbrook, NY, USA) with slight modifications. DNA concentrations were measured using a fluorometer (Hoefer Inc., San Francisco, CA, USA.). Concentrations were between 100 and 600 ng μL−1 for needle samples and between 20 and 55 ng μL−1 for megagametophyte samples in 100 μL total volume.

PCR and microsatellite analysis

Microsatellite loci designed for P. radiata (http://www.ffp.csiroau/tigr/molecular/pinessr.html) (Smith & Devey, 1994; Devey et al., 2002) were amplified. From 30 tested microsatellite loci, 19 were selected for further studies (Table 1), because their bands could be scored in all populations. PCR was performed as described earlier (Devey et al., 2002). Dye-labelled primers for use on the Applied Biosystems 310 Genetic Analyser were synthesized by Life Technologies. The PCR products (1 μL) were mixed with 15 μL of formamide and 0.3 μL Tamra 500 size standard (Applied Biosystems, Foster City, CA, USA) and denatured for 5 min at 95 °C. Each run contained samples from at least three populations, to detect possible migration differences between separate runs. The actual sizes of fragments were determined using the GeneScan® and Genotyper® software packages (Applied Biosystems).

Table 1.  Names, repeat types, allele size ranges, FST and expected heterozygostiy (He) of 19 P. radiata microsatellite loci.
LocusRepeat typeAllele range (bp)FSTHe
Pr9.3(CT)4(CA)1474–1050.1710.717
Pr011(CT)21(CA)8154–1720.0800.873
Pr043-2(CT)22184–1950.0970.685
Pr111(CA)1595–1190.1440.851
Pr28(CT)6(CA)1285–980.0720.692
Pr161(CT)18210–2460.0990.865
Pr4.6(CA)21(TA)6199–2280.1410.808
Pr44(CT)4(CA)1689–1130.2260.830
Pr001(CA)14151–1790.1680.804
Pr114(GA)13114–1300.1680.672
Pr048(CT)17253–2730.1390.744
Pr025(GA)18308–3410.1890.827
Pr102(CT)13118–1390.0640.785
Pr042(CT)14(CA)15170–1860.1010.672
Pr054(GA)19149–1720.1190.874
Pr168(GA)14268–3080.1640.897
Pr174(CT)16(CA)1693–1300.1220.903
Pr203(GA)14117–1360.2220.731
Pr060-2(CT)19272–2840.2140.552

Establishing the molecular pattern of variation at microsatellite loci

Some of the microsatellite alleles from loci Pr9.3, Pr111 and Pr161 were characterized by direct DNA sequencing. PCR was performed in 50 μL volume as described above, but no fluorescent primers were used. PCR products were run on agarose gels and alleles were purified using the QIAquick Gel Extraction Kit (Qiagen Pty Ltd, Doncaster, VIC, Australia). Sequencing was performed by the dideoxynucleotide chain termination method using the 377 sequencer (Applied Biosystems). Both strands were sequenced for each product. Sequences were aligned using the GeneDoc program (Nicholas & Nicholas, 1997).

Data analysis

Generation numbers since separation of the island populations may have been high enough for mutations to influence results (see Nichols & Freeman, 2004). Microsatellite mutation rates times the number of generations may be about one for the Cedros population and maybe an order of magnitude lower for Guadalupe. We therefore reanalysed older isozyme data from the same populations to compare with the microsatellite data analysis. Furthermore, in addition to models based on pure drift we also applied models that incorporate mutation, e.g. RST (Slatkin, 1995). Since it was evident that heterogenous mutation processes contributed to the variation at the microsatellite loci (see Results), we used methods based on the infinite allele model (IAM) and partly on the single-step mutation model (SMM), or its modification, the two-phase model (TPM).

At several microsatellite loci, the observed heterozygosity was lower than expected under Hardy–Weinberg equilibrium. As the level of inbreeding in adult P. radiata is known to be low even on the islands (Moran et al., 1980; Vogl et al., 2002), deviations will likely be caused by undetected null-allele heterozygotes. The analysis of the haploid megagametophytes of seeds, from the same trees that were included in the diploid analysis, allowed us to first estimate the frequency of null alleles and PCR failures resulting in nonamplification of one or the other allele. The seeds containing the haploid megagametophyte were collected from the same trees from which the needles were collected. Null-allele heterozygotes are detected because the haploid megagametophytes of a tree segregate for the presence and absence of the corresponding microsatellite band. We then estimated the frequency of null alleles using Mendelian transmission probabilities. In this separate analysis, we included an F-coefficient and an indicator variable for the maternal inbreeding coefficient (Vogl et al., 2002). The inferred overall frequencies of null alleles and PCR failures were used as parameters in our main analysis. In this main analysis, we used a probabilistic method that estimates allelic frequencies, including those of null alleles and inbreeding coefficients simultaneously. Based on these estimates of allelic frequencies, the mean number of alleles per locus (A), expected heterozygosity (He) and genetic differentiation (FST) among populations were obtained using the GDA program (Genetic Data Analysis) (Lewis & Zaykin, 2001). F-statistics are estimated in this program based on the method described in Weir & Cockerham (1984). Since new mutations after population splits may be important, we also estimated RST (Slatkin, 1995).

Additionally, we estimated Θp, the differentiation from an inferred ancestral gene pool for each population p, concurrently with allele frequencies and inbreeding levels (Vogl et al., 2003). This Bayesian approach allows calculation of posterior intervals incorporating random variation due to the presence of an unknown frequency of null alleles, drift among and unknown levels of inbreeding within the populations and finite sample sizes.

Before using the individual loci for subsequent analysis, we first examined the distribution of the Θp and FST values for all loci separately, using methods similar to Beaumont & Nichols (1996), Vitalis et al. (2001), Porter, 2003 and Beaumont & Balding (2004). Since inbreeding and null alleles are not considered in these approaches, we used the following method: for each locus separately, we calculated the distribution of Θp and compared with the distribution of Θp, when all loci were considered jointly. If the 95% intervals did not overlap, we considered the distributions different. With this procedure, 5% positive results are expected by chance.

Population structure was further characterized by estimating genetic distances among populations using Nei's genetic distance (Nei, 1972). The results are depicted as an UPGMA tree; the confidence of each node was inferred by bootstrapping data a thousand times using Phylip (Felsenstein, 1993).

To infer possible effects of recent bottlenecks, several tests were used to detect departures of populations from mutation-drift equilibrium. The difference between observed gene diversity and equilibrium gene diversity given the observed number of alleles was compared (Watterson, 1978,1986) using the Bottleneck 1.2.02 software (Cornuet & Luikart, 1996; Piry et al., 1999). The same program was also used to test for a deviation from the L-shaped distribution of allelic frequencies typical of equilibrium populations (Luikart et al., 1998). These tests were done on all populations separately and on the inferred ancestral gene pool assuming the IAM, the SMM and TPM. For the SMM and TPM analysis, the allele sizes were binned into classes to fit to the step-wise model.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

Examining the allelism of amplification products and the fit to the step-wise mutation model

Based on the similarity of flanking sequences of different alleles (of the same locus, from different trees), the amplification products from each set of primers were indeed alleles from the same locus. All sequenced microsatellite loci had perfect or compound dinucleotide repeats (Table 1). The length differences not corresponding to the expected change in repeat length were due to single nucleotide insertion–deletion polymorphism in the flanking region or, once, in the middle of the repeat sequence. Thus, with sequencing both mutations concordant and discordant with the SMM were detected in this small sample that we studied in detail.

Estimates of allelic frequency and null alleles

The raw allelic frequencies for the five populations are listed in Appendix 1 (Supplementary Table 1). The observed genotypic frequencies in all populations had a deficiency of heterozygotes compared to random mating expectation. Fixation indices, FI, were as follows: Año Nuevo: 0.145, Cambria: 0.163, Monterey: 0.136, Guadalupe: 0.279, Cedros: 0.258. These high-fixation indices are probably mainly due to undetected null-allele heterozygotes. In later analyses, we used the adjusted allelic frequency estimates where null alleles and inbreeding are accounted for. Most loci had low-null-allele frequencies (<0.1) in all populations. At loci Pr161 and Pr060-2, some populations had null-allele frequencies of about 0.20 and at several loci; null-allele frequencies differed significantly between populations (Pr011, Pr28, Pr054 and Pr114).

Expected heterozygosity within populations and genetic differentiation among populations

All populations had high levels of polymorphism, with average expected heterozygosity of 0.77 in the mainland populations (Table 2). Monterey, the largest mainland population, had the highest mean number of alleles per locus (10), whereas Guadalupe Island, with currently the lowest number of trees, had the lowest (6.73) (Table 2). The mean values for He were similar among mainland populations, but genetic polymorphism was lower on the islands. The small Guadalupe Island population (currently <400 trees) and the considerably larger Cedros Island population (close to 100 000 trees) had similar expected heterozygosity estimates (0.68 and 0.69, respectively) despite their different extant population sizes.

Table 2.  Population sizes and genetic polymorphism in natural populations of P. radiata.
PopulationCensus NnAHe
  1. N, Census size; n, mean number of individuals sampled per locus; A, mean number of alleles per locus; He, expected heterozygosity. Standard errors in parenthesis.

Año Nuevo1 000 00033.98.42 (0.55)0.76 (0.02)
Monterey1 700 0003010 (0.66)0.77 (0.02)
Cambria1 300 000308.47 (0.41)0.77 (0.03)
Cedros80 000317.32 (0.54)0.69 (0.04)
Guadalupe<400316.73 (0.53)0.68 (0.04)

The range of FST values among loci were quite high (from about 0.06 to 0.23, Table 1). We examined whether some loci were statistically significantly different from the remaining loci, but no more than the expected 5% of deviations from a common distribution were found. Thus, all loci are kept in the demographic analyses.

Table 4 shows that substantial genetic differentiation was detected among populations. The average estimate of FST was 14.1% (SD 0.012); RST was similar (data not shown). Estimates of FST from the isozyme data, using the same methods, are similar to those from the microsatellites. This level of differentiation among populations reflects considerable differences in allelic frequencies. None of the loci shared the same most common allele among all five populations. Some alleles were specific for a population, e.g. at locus Pr161 allele 210 bp was only found in Guadalupe (see Appendix 1). Pair-wise FST values between populations also demonstrate that mainland populations had a low level of differentiation, with values ranging between 0.02 and 0.07 (Table 3). Monterey and Cambria were the closest pair of the mainland populations. The two island populations were both quite diverged from the mainland and from each other with FST estimates of 0.17–0.20. Nei's genetic distances (Nei, 1972) also showed that mainland populations clustered in their own group with a high-bootstrap value (Fig. 2). The divergence of populations from a common ancestral population is also described with the Θp estimates (Vogl et al., 2003), which measure the effects of genetic drift (Table 4). The small Guadalupe island population was most diverged, followed closely by Cedros, while the mainland populations form a closely related group, with Θp estimates less than 0.07. Again, estimates from the isozyme data are similar to those from the microsatellites. Only in the case of Monterey, the 0.95 limits do not overlap.

Table 4.  Differentiation of individual populations from the common gene pool in P. radiata, as measured by Θp, means and percentiles.
 Mean2.5%5.0%50.0%95.0%97.5%
Nineteen microsatellite loci
 Año Nuevo0.06250.04630.04700.06080.08230.0866
 Cambria0.04430.02870.03000.04430.05550.0618
 Monterey0.01560.00850.00890.01550.02330.0261
 Cedros0.15160.12040.12390.14960.18350.1895
 Guadalupe0.18030.14330.l4910.18020.21330.222
Five isozyme loci
 Año Nuevo0.05590.02140.02520.05060.10530.1128
 Cambria0.07940.04780.04820.07240.1360.1524
 Monterey0.07060.02970.0340.07310.10780.1102
 Cedros0.22280.15410.15860.21550.3020.3267
 Guadalupe0.15110.07370.08440.15240.2150.2299
Table 3.  Pair-wise estimates of genetic differentiation (FST) between populations of P. radiata based on 19 microsatellite loci below the diagonal and on five isozyme loci above the diagonal.
PopulationAño NuevoCambriaMontereyCedrosGuadalupe
Año Nuevo0.0520.0230.2230.089
Cambria0.0700.0300.1630.071
Monterey0.0480.0270.1690.087
Cedros0.1930.2040.1780.193
Guadalupe0.1720.1920.1750.175
image

Figure 2. UPGMA bootstrap consensus trees of P. radiata populations based on Nei's genetic distance (Nei, 1972). Percentages appearing alongside nodes indicate the proportion of bootstrap replications in which the figured grouping appears.

Download figure to PowerPoint

Is there evidence for departure from mutation-drift equilibrium due to bottlenecks?

The reductions in population sizes associated with the colonization of islands, or with the possible recent reduction in population size on the mainland could result in a change in the level of gene diversity relative to the number of alleles. The expected gene diversities, conditional on the number of alleles observed, were lower than the observed gene diversities (after accounting for null alleles and inbreeding) under all three different mutation models in all populations and in the inferred gene pool. The results for the IAM and SMM are in Table 5. The TPM was intermediate between these two models (results are not shown). This suggests that Pinus radiata as a whole has been expanding rather than contracting recently.

Table 5.  Number of loci with heterozygosity deficiency of the total number of loci (def/total), given the number of alleles in the five populations of P. radiata and the inferred ancestral allele pool assuming the infinite allele model (IAM) and the single-step mutation model (SMM).
PopulationIAM (def/total)SMM (def/total)
Año Nuevo19/1919/19
Cambria18/1918/19
Monterey19/1919/19
Cedros16/1816/18
Guadalupe15/1916/19
All populations17/1919/19

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

The population demography of radiata pine, P. radiata, is quite different from many other pines, which are generally distributed over large continuous ranges with low differentiation in subpopulations. The two island populations of P. radiata differentiated from the mainland populations about 10 MYA (Cedros) and 1–4 MYA (Guadalupe) according to independent evidence. Census sizes on the islands must have been much lower than on the mainland and should have led to reduction in genetic diversity. The population demography on the mainland is less clearly documented; however, the mainland populations are comparatively less differentiated and have larger current and presumably historic census sizes.

If a model of population subdivision by splitting at time t from a large panmictic population is assumed, population differentiation by drift will be approximately equal to t/2Ne. Hence, the time of split weighted by the effective population size of the two island populations is much higher than that of the mainland populations and, consequently, these two island populations are expected to be genetically impoverished compared to the mainland ones.

Furthermore, the two island populations show some selfing (Vogl et al., 2002) as opposed to most other conifers. In P. radiata inbreeding depression must have been purged at least in the island populations. This is only likely in small populations, where mating between relatives are frequent. Hence, we have independent evidence from biogeography and the mating system that effective population sizes in P. radiata have been relatively small on the islands.

Population demography is reflected in the pattern of molecular variation within and between populations. We therefore typed 19 variable microsatellite loci for all five extant populations of P. radiata. All the analyses were based on allelic frequencies corrected for inbreeding effects and null alleles. For the present purposes, i.e. the estimation of levels of variability and differentiation, errors due to ignoring the null alleles would have been less than when estimating inbreeding effects (Vogl et al., 2002). In the special case of conifers, when necessary, it is possible to obtain accurate estimates of null-allele frequencies by genotyping megagametophytes.

Our genetic data show that the three mainland populations had higher levels of genetic polymorphism than the two island populations, measured as expected heterozygosity (He = 0.77 vs. He = 0.68, respectively), or as number of alleles (9 vs. 7, respectively) (Table 2). The two island populations had similar levels of polymorphism, despite very different current population sizes and despite their different demographic histories. The Cedros population was presumably founded 10 MYA, whereas the Guadalupe island was colonized 1–4 MYA, according to other evidence. As drift depletes variability more rapidly in small populations, this is consistent with a lower long-term effective size of Guadalupe than of Cedros. If we assume a generation time of 50 years, these events occurred about 200 000 or 20 000–80 000 generations ago, in Cedros and Guadalupe, respectively. Mutation drift equilibrium is restored slowly by accumulation of new mutations, thus the time may not have been sufficient to fully restore a mutation-drift equilibrium after a severe bottleneck (Nei et al., 1975). The census sizes of the populations, from Moran et al. (1988) are listed in Table 2. The correspondence between the current census size and level of polymorphism is poor. Most evidently, the recent drastic reduction in the size of the Guadalupe population (to 400 trees) has had only little if any effect on the allelic distribution yet. The Guadalupe population may in fact be on its way to extinction before the genetic consequences of drift become observable. On the other hand, the large mainland stands with much higher numbers of trees harbour only little but significantly more variation than the island populations.

Earlier methods had not detected any difference in the level of variability at all (Moran et al., 1988; Wu et al., 1999). Studies on cpDNA variation had detected no polymorphism in any of the populations (Hong et al., 1993). Thus, microsatellites were more efficient in detecting differences in presumably neutral genetic diversity.

Some population genetic studies have detected population size effects in shorter-lived plants (Van Treuren et al., 1991). Several studies in conifer populations have shown that the present conservation status (population size) and polymorphism at marker loci are not correlated. For instance, the Mexican subalpine relict Picea mexicana has about the same level of allozyme diversity as P. radiata (Ledig et al., 2002). The endangered Mexican pine Pinus rzedowskii has even higher allozyme polymorphism (Delgado et al., 1999). Current population sizes are poor predictors of levels of polymorphism, even within a group of closely related species.

Earlier allozyme studies had shown that P. radiata has less isozyme variation than many other pine species (Moran et al., 1988; Muona, 1990; Ledig, 1998). The levels of microsatellite variation are similarly high as in other conifers, e.g. in P. sylvestris (Karhu et al., 1996), P. taeda (Al-Rabab'ah & Williams, 2002) P. contorta (Thomas et al., 1999; Liewlaksaneeyanawin et al., 2004), P. pinaster (Mariette et al., 2001) or Picea abies (Pfeiffer et al., 1997), but detailed across species comparisons are not meaningful because the loci are not a random sample.

Genetic drift leads to loss of diversity within populations, as documented in the previous subsection for the island populations, but also to differentiation between populations. FST and related estimates of population differentiation were deemed suitable for measuring differentiation between the populations, as at least for the mainland populations microsatellite mutation rates are low compared to divergence times of population. Estimates of FST varied widely between loci. Deviating estimates could indicate that the genomic region containing the microsatellite locus has undergone selection (Beaumont & Nichols, 1996; Vitalis et al., 2001). However, we kept all loci in the later analyses, as none of the loci deviated statistically significantly from the rest. The average estimate of FST is 0.14. With isozymes, Moran et al. (1988) estimated that 0.16 of the total variation was between populations. RAPD markers resulted in differentiation levels for three populations (Año Nuevo, Cambria and Guadalupe) between 0.17 and 0.26 depending on the loci scored (Wu et al., 1999). Our reanalysis of the isozyme data gave similar results to those using microsatellites. Hence, the differences in the mutation rate and mutation distribution between isozymes, RAPDs and microsatellites seem not to affect the analysis. This is only expected if the mutation rate is slow compared to the effects of population demography and drift. An earlier study on cpDNA detected hardly any variation within P. radiata and thus did not provide information for describing population differences (Hong et al., 1993). We also analysed the data using statistics designed for the SMM model, such as RST and the results were similar to those above (data not shown).

The analysis using the method of Vogl et al. (2003) showed that the Monterey population's gene pool seems to reflect the general species’ gene pool most closely. The other two mainland populations may still share the same gene pool and exchange migrants or this exchange has stopped recently. According to Vogl et al. (2003) and according to additional simulations, the expected patterns of variation resulting from migration and drift is so similar to that of temporal isolation drift that distinguishing between these two alternatives (Nielsen & Wakeley, 2001) is difficult. The same method has earlier been used for analysis of Drosophila ananassae populations (Vogl et al., 2003), where the marginal populations far from the species centre were also more diverged from the common gene pool.

Among the mainland populations, Monterey and Cambria were closest to each other according to the FST analysis, with just 2% of variation between them, in accordance with the study of Plessas & Strauss (1986). However, Moran et al. (1988) found that Monterey and Año Nuevo are the most similar mainland populations. The traditional view is that Cambria is most distinct, based on many morphological characters (Guinon et al., 1982). Such traits may, however, be influenced by selection and thus evolve at a different rate from neutral markers.

Other available data on morphology and palaeontology suggested demographic fluctuations (at least a drastically recently reduced population size in Guadalupe island) and possibly a reduction in the mainland population some 4–8000 years ago (Axelrod, 1981), perhaps only 200 generations ago. Furthermore, the observed moderate level of inbreeding on the islands suggest purging of the genetic load through (slow) inbreeding (Vogl et al., 2002). Hence, we expect to see signals of population shrinkage in the data as deviations from mutation-drift equilibrium.

An excess of heterozygosity, given the allele number is expected after a bottleneck. If the population is instead expanding, there will be an excess of low-frequency alleles compared to heterozygosity (Watterson, 1986). The test of Cornuet & Luikart (1996) is based on this idea. None of the populations showed heterozygosity excess, as would be expected after a bottleneck, but rather heterozygosity deficiency, perhaps resulting from a population expansion. This however contrasts with other evidence and the previous analysis that showed significant effects of drift especially in the island populations.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

We set out to address two questions: can we distinguish between alternative scenarios for the recent history of the mainland populations, i.e. significant contraction (Axelrod, 1981) or continuing fluctuations (Millar, 2000)? Are we able to detect the effects of changing population sizes and bottlenecks in the island populations? The second question is easily answered with yes: we clearly observe reduced allelic variation on the islands compared to the mainland. This is an improvement over earlier methods (Moran et al., 1988; Hong et al., 1993; Wu et al., 1999) that were not sensitive enough to detect this effect. In addressing the first question, we observe little evidence for the effects of a long-term contraction in our data set. Rather we see the signature of population expansion. But this contradicts other independent evidence and also our findings of significant drift especially in the island populations. The apparent contradiction may be resolved by observing that the signature of expansion may be strong but relatively old, such that more recent contractions and drift had only minor influences and then mainly in the island populations so far. Confirmation of these hypotheses may be obtained through sequence information, where testing for deviations from mutation-drift equilibrium is more powerful.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References

We wish to thank all the members of the plant genetics group for discussion and Pekka Pamilo and Patrik Waldmann for help with analyses. CV wishes to thank Graham Muir for practical help and him and Christian Schlötterer for discussions. Financial support was provided by the Graduate School in Forest Sciences, the Academy of Finland and the Environmental and Biosciences Council to Outi Savolainen.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgments
  9. References
  • Al-Rabab'ah, M.A. & Williams, C.G. 2002. Population dynamics of Pinus taeda L. based on nuclear microsatellites. For. Ecol. Manage. 163: 263271.
  • Axelrod, D.I. 1981. Holocene climatic changes in relation to vegetation disjunction and speciation. Am. Nat. 117: 847870.
  • Beaumont, M.A. & Balding, D.J. 2004. Identifying adaptive genetic divergence among populations from genome scans. Mol. Ecol. 13: 969980.
  • Beaumont, M.A. & Nichols, R.A. 1996. Evaluating loci for use in the genetic analysis of population structure. Proc. R. Soc. Lond. 263: 16191626.
  • Cornuet, J.-M. & Luikart, G. 1996. Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 144: 20012014.
  • Delgado, P., Pinero, D., Chaos, A., Perez-Nasser, N. & Alvarez-Buylla, E.R. 1999. High population differentiation and genetic variation in the endangered Mexican pine Pinus rzedowskii (Pinaceae). Am. J. Bot. 86: 669676.
  • Devey, M.E., Bell, J.C., Uren, T.L. & Moran, G.F. 2002. A set of microsatellite markers for fingerprinting and breeding applications in Pinus radiata. Genome 45: 984989.
  • Eldridge K. 1978. Seed collections in California in 1978. Aust. CSIRO Div. for. Res. Annu. Rep. 19771978.
  • Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Package) Version 3.5c. Distributed by the author. Department of Genetics, University of Washington, Seattle.
  • Guinon, M., Hood, J.V. & Libby, W.J. 1982. A clonal study of intraspecific variability in radiata pine. Aust. For. Res. 12: 191201.
  • Hamrick, J.L., Godt, M.J. & Sherman-Broyles, S. 1992. Factors influencing levels of genetic diversity in woody plant species. N. For. 6: 95124.
  • Hong, Y.P., Hipkins, V.D. & Strauss, S.H. 1993. Chloroplast DNA diversity among trees, populations, and species in the California closed pines (Pinus radiata, P. muricata, and P. attenuata). Genetics 135: 11871196.
  • Karhu, A., Hurme, P., Karjalainen, M., Karvonen, P., Kärkkäinen, K., Neale, D.B. & Savolainen, O. 1996. Do molecular markers reflect patterns of differentiation in adaptive traits of conifers? Theor. Appl. Genet. 93: 215221.
  • Lavery, P.B., Mead, D.J. 1998. Pinus radiata: a narrow endemic from North America takes on the world. In: Ecology and biogeography of Pinus (D.M.Richardson, ed.). 432449. Cambridge University Press, Cambridge.
  • Ledig, F.T. 1998. Genetic variation in Pinus. In: Ecology and biogeography of Pinus (D.M.Richardson, ed.). 251280. Cambridge University Press, Cambridge.
  • Ledig, T.F., Hodgkiss, P.D. & Jacob-Cervantes, V. 2002. Genetic diversity, mating system, and conservation of a Mexican subalpine relict, Picea mexicana Martínez. Conserv. Genet. 3: 113122.
  • Lewis, P.O. & Zaykin, D. 2001. Genetic Data Analysis: Computer program for the analysis of allelic data. Version 1.0 (d16c). Free program distributed by the authors over the internet from http://lewis.eeb.uconn.edu/lewishome/software.html
  • Liewlaksaneeyanawin, C., Ritland, C.E., El-Kassaby, Y.A. & Ritland, K. 2004. Since-copy, species-transferable microsatellite markers developed from loblolly pine ESTs. Theor. Appl. Genet. 109: 361369.
  • Luikart, G., Allendorf, F.W., Cornuet, J.-M. & Sherwin, W.B. 1998. Distortion of allele frequency distributions provides a test for recent population bottlenecks. J. Hered. 89: 238247.
  • Mariette, S., Chagné, D., Lezier, C., PAstuszka, P., Baffin, A., Plomion, C. & Kremer, A. 2001. Genetic diversity within and among Pinus pinaster populations: comparison betwen AFLP and microsatellite markers. Heredity 86: 469479.
  • Millar, C. 1997. Quaternary evolution of Pinus radiata. FRI Bulletin 203: 2225.
  • Millar, C.I. 2000. Evolution and biogeography of Pinus radiata with a proposed revision of its Quaternary history. N. Z. J. For. Sci. 29: 335365.
  • Moran, G.F., Bell, J.C. & Matheson, A.C. 1980. The genetic structure and levels of inbreeding in a Pinus radiata seed orchard. Silvae Genet. 29: 190193.
  • Moran, G.F., Bell, J.C. & Eldridge, K.G. 1988. The genetic structure and the conservation of the five natural populations of Pinus radiata. Can. J. For. Res. 18: 506514.
  • Muona, O. 1990. Population genetics in forest tree improvement. In: Plant population genetics, Breeding, and Genetic Resources (A.H.D.Brown, M.T.Clegg, A.L.Kahler & B.S.Weir, eds). 282298. Sunderland, Mass.
  • Nei, M. 1972. Genetic distance between populations. Am. Nat. 106: 283292.
  • Nei, M., Maruyama, T. & Chakraborty, R. 1975. The bottleneck effect and genetic variability in populations. Evolution 29: 110.
  • Nicholas, K.B., Nicholas H.B. Jr. & Deerfield D.W. II. 1997. GeneDoc: Analysis and Visualization of Genetic Variation. EMBNEW.NEWS 4: 14.
  • Nichols, R.A. & Freeman K.L.M. 2004. Using molecular markers with high mutation rates to obtain estimates of relative population size and to distinguish the effects of gene flow and mutation: a demonstration using data from endemic Mauritian skinks. Mol. Ecol. 13: 775787.
  • Nielsen, R. & Wakeley, J. 2001. Distinguishing migration from isolation: a Markov chain Monte Carlo approach. Genetics 158: 885896.
  • Pfeiffer, A., Olivieri, A.M. & Morgante, M. 1997. Identification and characterization of microsatellites in Norway spruce (Picea abies K.). Genome 40: 411419.
  • Piry, S., Luikart, G. & Cornuet, J.-M. 1999. BOTTLENECK: a computer program for detecting recent reductions in effective population size using allele frequency data. J. Hered. 90: 502503.
  • Plessas, M.E. & Strauss, S.H. 1986. Allozyme differentiation among populations, stands, and cohorts in Monterrey pine. Can. J. For. Res. 16: 11551164.
  • Porter, A.H. 2003. A test for deviation from island-model population structure. Mol. Ecol. 12: 903915.
  • Slatkin, M. 1995. A measure of population subdivision based on microsatellite allele frequencies. Genetics 139: 457462.
  • Smith, D.N. & Devey, M.E. 1994. Occurrence and inheritance of microsatellites in Pinus radiata. Genome 37: 977983.
  • Thomas, B.R., Macdonald, S.E., Hicks, M., Adams, D.L. & Hodgetts, R.B. 1999. Effects of reforestation methods on genetic diversity of lodgepole pine: an assessment using microsatellites and randomly amplified polymorphic DNA markers. Theor. Appl. Genet. 98: 793801.
  • Van Treuren, R., Bijlsma, R. & Van Delden, W. 1991. The significance of genetic erosin in the process of extinction. 1. Genetic differentiation in Salvia pratensis and Scabiosa columbaria in relation to population size. Heredity 66: 181189.
  • Vitalis, R., Dawson, K. & Boursot, P. 2001. Interpretation of variation across marker loci as evidence of selection. Genetics 158: 18111823.
  • Vogl, C., Karhu, A., Moran, G.F. & Savolainen, O. 2002. High resolution analysis of mating systems: inbreeding in natural populations of Pinus radiata. J. Evol. Biol. 15: 433439.
  • Vogl, C., Das, A., Beaumont, M., Mohanty, S. & Stephan, W. 2003. Population subdivision and molecular sequence variation: theory and analysis of Drosophila ananassae data. Genetics 165: 13851395.
  • Wakeley, J. & Aliacar, N. 2001. Gene genealogies in a metapopulation. Genetics 159: 893905.
  • Watterson, G.A. 1978. The homozygosity test of neutrality. Genetics 88: 405417.
  • Watterson, G.A. 1986. The homozygosity test after a change in population size. Genetics 112: 899907.
  • Weir, B.S. & Cockerham, C.C. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 13581370.
  • Wu, J., Krutovskii, K.V. & Strauss, S.H. 1999. Nuclear DNA diversity, population differentiation, and phylogenetic relationships in the California closed-cone pines based on RAPD and allozymes markers. Genome 42: 893908.