Evolution of the pygmy-forest edaphic subspecies of Pinus contorta across an ecological staircase.
The terraces were cut during the Pleistocene epoch by wave action of a rising sea during interglacial periods, covered with beach materials as the sea level dropped during glacial periods, and sequentially raised above sea level by tectonic forces. The five terraces were formed approximately 115,000, 285,000, 610,000, 725,000 and 1.2 mya. The parent rock and process of terrace formation are the same for all five terraces; however, time has produced progressively larger differences among them in soil (Jenny et al. 1969), resulting in different associated vegetative communities.
The first (youngest) terrace has a grassland soil, common elsewhere along the northern California coast, that is well drained, with a moderate pH and many available nutrients. It supports scattered stands of Pinus contorta subspecies contorta (shore pine) and Pinus muricata (bishop pine). Soils of the third, fourth, and fifth terraces have a pH decreasing to as low as 2.9 (Jenny et al. 1969), podzolization with an iron or clay hardpan, and low amounts of available nutrients and organic matter. Anaerobic conditions on the perched water table persist for several months a year, and severe drought occurs in summer for plants with roots limited by the hardpan. These soils support a stunted pygmy forest consisting of an assemblage of plants that include four endemic taxa: Carex californica, Arctostaphylos numularia, Cupressus pygmaea (pygmy cypress), and Pinus contorta ssp. bolanderi (Bolander pine). Extremely stunted forests on the fourth and fifth terraces are dominated by Bolander pine and pygmy cypress, more moderate pygmy-forest sites by bishop pine and Bolander pine. Second-terrace soils, intermediate to those of coastal and pygmy-forest types, are highly variable among locales (Jenny et al. 1969; Sholars 1979, 1982) and, as a result, support highly heterogeneous forests containing Pinus contorta but dominated by Pseudotsuga menziesii (Douglas-fir), bishop pine, and Sequoia sempervirens (coast redwood).
The upper (fourth and fifth) terraces commonly have frontal, seaward dunes. The dune soils and the young soils on the eroded cliff slopes between terraces support mixed forests, mostly bishop pine, Douglas-fir, coast redwood, and tanoak (Lithocarpus densiflorus). These dune and cliff forests are considerably taller and denser than pygmy-forest trees or trees in coastal grasslands, and they separate the populations of shorter trees on the first, third, fourth, and fifth terraces. Populations of Bolander pine on each of the upper terraces are also discontinuous, separated by tall forests, mostly redwood and Douglas-fir, in steep canyons cut from east to west by watercourses emerging at the coast.
Westman (1975), in a survey of the vegetative composition and soil properties of forest types in the pygmy-forest regions, concluded that the primary determinant of species distribution and growth rate was soil pH, followed by water-holding capacity of the soil profile. Soil properties differ most between terraces 1 and 3, and relatively little between moderate and extreme pygmy-forest sites.
Pinus contorta differs sufficiently in morphology from one end of the soil chronosequence to the other to have been considered two species, P. contorta and P. bolanderi, by the taxonomist Parlatore (1868). On the basis of cone and leaf morphology and crossability, Critchfield (1957) reclassified P. contorta on the first terrace and elsewhere near the Pacific Coast as ssp. contorta, and P. contorta in the pygmy forest on the third, fourth, and fifth terraces as ssp. bolanderi.
McMillan (1964) planted Mendocino-area seedlings of both coastal and pygmy-forest origin in the pygmy forest, and found ssp. bolanderi to have significantly higher survival and growth rates after 10 yr. In a later common-garden planting on a fertile sierran soil at Placerville, California, shore and Bolander pines survived and grew similarly, but the morphological traits of form, needles, and cones used by Critchfield (1957) remained true to type.
Subsequent studies using biochemical markers showed ssp. bolanderi to be distinct from other P. contorta populations in allozyme composition (Wheeler and Guries 1982) and terpene composition (Forrest 1980). However, the allozyme study compared only one fifth-terrace Bolander-pine population with shore pine from geographically distant populations of coastal Washington state, confounding short-distance edaphic and long-distance geographic differentiation. The terpene samples were collected from only one pygmy-forest population (fourth terrace), and a nearby coastal population was apparently misclassified as a second pygmy-forest population. These studies have been insufficient for drawing conclusions about subspecific differences among local populations of shore and Bolander pine.
Jenny et al. (1969) suggested an attractive hypothesis for P. contorta differentiation that fits well with the geological history and soil development in this area. In it, ancestors of modern shore pine colonized the first, newly emerged marine terrace (current terrace 5) between 800,000 and 1 mya, when it had soils similar to those on the current lowest terrace (terrace 1). Over several hundred thousand years, shore pines colonized each successive terrace as it arose and continuously occupied the terraces as they were being uplifted. As the upper terraces weathered and podzolized, the pines adapted to slowly changing conditions - becoming disjunct because of high forests on the deeper soils that separated the terrace populations, and evolving into Bolander pine.
However, while a pollen profile from a nearby bog including only the Late Postglacial period shows P. contorta to be dominant for the last few thousand years, a profile from Cape Mendocino approximately 120 km north shows P. contorta pollen to be scarce or absent throughout the Hypsithermal period (approximately 8500 to 3000 yr ago) and common in the Late Postglacial period (last 3000 yr) (Heusser 1960). Pinus contorta pollen is distinct from that of other pines in the region; however, ssp. contorta and ssp. bolanderi cannot be distinguished. Currently, no populations of P. contorta occur between Cape Mendocino and the pygmy-forest region (Griffin and Critchfield 1972). Pollen evidence from these and other sites support a relatively recent southward migration of P. contorta into California. Another problem with a hypothesis of continuous occupation and adaptation is the more than 100,000 yr of terrace-2 soil development during which P. contorta would have been mostly excluded by larger trees that can live on soil of intermediate development. But patches of dwarfed forest do occur on the second terrace, and similar populations in the past may have provided an adaptation route from shore pine to Bolander pine.
Our primary objectives were to examine current and evolutionary relationships between these two subspecies, and to test the hypothesis of Jenny et al. (1969). In the process, we studied P. contorta populations on and off pygmy-forest sites and analyzed patterns of genetic variation within and among populations, using allozymes, generally considered to be neutral or near-neutral genetic markers.
[TABULAR DATA OMITTED]
MATERIALS AND METHODS
Cone and Seed Collection and Handling. - Unopened Pinus contorta cones were collected from 23 to 25 parent trees in each of 11 pygmy-forest (ssp. bolanderi) and 6 coastal (ssp. contorta) populations. Seventeen to 21 wind-pollinated families (at least 10 germinants for inclusion) were studied for each population. Four of the six ssp. contorta coastal populations are near or within the latitudinal range of ssp. bolanderi, one is 29 km south, one 150 km north. Of the eleven ssp. bolanderi populations, four are on terrace 3, three on terrace 4, and four on terrace 5. All fifth-terrace populations sampled are within a 6-km latitudinal spread; fourth-terrace populations are within 15 km; and third-terrace populations are within 17 km. The latitudinal range of the third- and fourth-terrace populations was similar, and fifth-terrace populations were central to the range of the lower terraces. In each population, cones were collected from the tree nearest the intersection point on a 25- x 25-m grid to ensure similar spacing regardless of population size and density, and to reduce the probability of collecting cones from closely related individuals.
Seeds were extracted by one of two methods, depending on the status of the cone scales at collection. Resin-sealed cones (100% of pygmy-forest samples and a small proportion of the coastal samples) were boiled in water for 40 s, then oven-dried for 24 h at 35 [degrees] C. Cones with scales not sealed were oven-dried only.
Seeds were soaked overnight in water, stratified at 4 [degrees] C for at least 2 wk, then germinated at room temperature. Six haploid megagametophytes were dissected from each wind-pollinated family. Each megagametophyte was crushed over ice in one drop of extraction buffer (10 ml 0.2 M phosphate buffer with 250 mg ascorbic acid, 50 mg D-glucose-6-phosphate, 50 mg bovine albumin, and 10 mg dithiothreitol). Samples were frozen at -80 [degrees] C overnight, thawed, absorbed into 2- x 12-mm filter-paper wicks, and loaded into starch gels.
Allozyme Analysis. - Starch gels were prepared and run according to Conkle et al. (1982). Variations in substrate and the cofactor amounts, and details on the four gel-buffer systems and 25 loci investigated are provided elsewhere (Aitken 1989). Poorly resolved individual stains resulted in missing data in some runs. Where possible, these data were collected from reruns, but not all could be replaced.
Statistical Analyses. - Genetic analyses of data were made with BIOSYS-1 (Swofford and Selander 1981); regression analyses, analyses of variance, and principal component analyses were made with SAS (SAS Institute 1987). Allele frequencies were calculated separately for all populations. A principal-component analysis of allele frequencies for all polymorphic loci was performed to reduce the number of variables and to identify general multilocus patterns of differentiation. Frequencies of all alleles common to most populations, except the allele with the lowest frequency, were included in the analysis. The four measures of variation studied were: number of alleles per locus (A); percentage of loci polymorphic at the 5% level (% P); and observed and expected heterozygosities ([H.sub.o], [H.sub.e]). The observed level of heterozygosity in maternal parent trees was adjusted for the number of gametophytes from which genotype was inferred by means of the formula (Morris and Spieth 1978):
[H.sub.o](adj) = [H.sub.o]/(1 - [0.5.sup.k - 1]), (1)
where k is the number of gametophytes sampled (six, except where unresolved data could not be replaced). Relationships of each of the four measures of allozyme variation with latitude, longitude, and elevation, and between the frequencies of individual alleles and the geographic variables were determined by simple linear-regression analyses. One-way ANOVAs were performed for frequencies of alleles by terrace and by subspecies. F-statistics for estimating the de-free of population subdivision (Wright 1965) were calculated with all loci for each of the 17 populations and for combinations of terraces on the assumption that subdivision is wholly caused by genetic drift of neutral alleles, although in reality it may also be affected by direct and indirect (hitchhiking) selection with adaptive alleles. Chord distances (Cavalli-Sforza and Edwards 1967) were calculated for estimates of the allozyme-based genetic differences among all populations, then averaged for genetic distances among population subsets, for example, pygmy-forest populations only, coastal populations only, fifth-terrace populations only. The unbiased genetic distances of Nei (1978), while not necessary for our analyses, were calculated and are provided here to facilitate comparison with other studies that have used that measure.
Genetic Variation. - Pygmy-forest populations (terraces 3, 4, and 5 combined) had significantly less allelic variation (P [less than] 0.01) than did terrace-1 coastal populations, which averaged 1.8 alleles per locus, 38% polymorphic loci, 11.9% observed heterozygosity, and 12.9% expected heterozygosity. Pygmy-forest populations averaged 1.65 alleles per locus, 28% polymorphic loci, 9.6% observed heterozygosity, and 10.5% expected heterozygosity.
Few differences in allozyme allele frequencies among Pinus contorts populations and among subspecies were significant. For example, for all loci except Adh2 and Acp1, the most common allele in all populations was the same. Locus Adh2 had two common alleles with approximately equal frequencies in all populations (observed Adh2-1 frequencies from 0.443 to 0.773). Locus Acp1 had three common alleles, each varying in frequency in all populations, either Acpl-1 or Acp1-2 being most common. None of the allele frequencies at the 25 analyzed loci were significantly correlated with the elevation, latitude, or longitude of population origin, indicating that allozyme clines up or along the chronosequence of terraces were weak or absent. An ANOVA revealed that only Alap1-1 varied significantly between ssp. contorta and ssp. bolanderi (P = 0.038) largely because of the presence of Alapl-3 in coastal, but not in pygmy-forest, populations. No allele frequencies varied significantly among terraces 1, 3, 4, and 5. When only terraces 3, 4, and 5 were analyzed, Acp1-1 (P = 0.065), Acp1-2 (P = 0.054), and Gdhl-2 (P = 0.066) varied across pygmy-forest populations, but given the number of analyses, this number of "statistically significant" differences could be due to chance alone (for details, see Aitken 1989).
Private Alleles. - Eleven apparently private alleles (i.e., those found in samples of only one population) were observed: three in ssp. bolanderi pygmy-forest (P) populations (average 0.27 private alleles per population), eight in coastal ssp. contorta (C) populations (average 1.33 per population). Average frequency of private alleles in C populations was 0.082, in P populations 0.042. Populations P9 on terrace 3, P1 and P7 on terrace 4, and C3 and C5 each had one private allele; C4 and C6 had three. No private alleles were observed on terrace 5. Ten "semiprivate" alleles (i.e., those occurring in one subspecies only) were found in ssp. contorta and seven in ssp. bolanderi. Of these, all but one in ssp. bolanderi populations, but only 4 of 10 in ssp. contorta populations, were at a frequency of less than 1%.
Genetic Differentiation. - A scatter plot of the first and second principal-component scores for populations, identified by terrace of origin, shows overlap of ssp. contorta (terrace 1) and ssp. bolanderi (terraces 3, 4, and 5). However, the scatter of populations is greatest for terraces 1 and 4, less for terrace 3. Terrace-5 populations are tightly clustered, reflecting their relative uniformity.
With the exception of P1 (terrace 4), the first and second principal-component points for pygmy-forest sites cluster fairly tightly. It is noteworthy that outlier P1 is from an atypical pygmy-forest site considerably south of the other populations. While pygmy-forest sites and plant associations occur south of P1, none with P. contorta have been identified. Site P1 is sloped rather than flat, with more moderate soil characteristics than other terrace-4 sites, as shown by greater tree growth; thus, the P1 population may be an outlier because of a recent founder event, or because of less severe selection pressures.
A cluster diagram of the genetic relationships of all populations sampled, based on megagametophyte allozymes used in calculating chord distance (Cavalli-Sforza and Edwards 1967), shows that populations of coastal and pygmy-forest origin do not form the phylogenetic dichotomy expected with different subspecies or species. Pygmy-forest populations are generally more closely related (average chord distance 0.105) than are coastal populations (average chord distance 0.143). The average chord distance between the two subspecies is only 0.128, less than that among coastal populations. Other estimates of genetic distance (Nei 1978) reveal the same pattern: unbiased genetic distance averaging 0.0045 between pygmy-forest populations, 0.0091 between coastal populations, and [TABULAR DATA OMITTED] 0.0078 between pygmy-forest and coastal populations.
When we focus only on the pygmy forest and C2-C5 populations in adjacent coastal sites (eliminating the distant C1 and C6), a distinct edaphic pattern appears - the more severe the site, the lower the genetic chord distance: 0.092 among those on fifth-terrace sites; 0.141 among those on coastal sites. Coastal populations are, on the average, isozymically more similar to some pygmy-forest populations than to other coastal populations; allele frequencies vary more among them than among pygmy-forest populations, and, on more severe sites, pygmy-forest populations have more central and more uniform allele frequencies than do their coastal counterparts. Such relationships also occur in third-terrace and higher pygmy-forest populations.
Population Subdivision. - Estimated [F.sub.ST] values (Wright 1965) given in table 3, indicate that the largest differences are among those populations on terraces 1 and 3. However, these differences between subspecies are not much greater than population differentiation among populations within subspecies. The [F.sub.ST] value among contorta populations on terrace 1 is nearly as large (0.057) as that among populations on terraces 1 and 3 (0.065). The differences among bolanderi populations within and between terraces 4 and 5 are indicated as being substantially smaller (0.029 to 0.039).
Allozymic Differentiation. - The genetic patterns that should be evident in a derivative and progenitor species pair are that the progenitor should contain all, or nearly all, the allelic variation of the derivative, and that the derivative should contain only, or mostly, a subset of progenitor variation and few or no "unique" alleles (Gottlieb 1977, 1981). The distribution of private alleles and the pattern and quantity of genetic variation in allozymes of Pinus contorta ssp. contorta and ssp. bolanderi fit these criteria.
It appears that P. contorta ssp. bolanderi is derived from nearby populations of ssp. contorta. The lack of strong allozyme differentiation between nearby populations of the subspecies contrasts sharply with their substantial morphological and physiological differences (Critchfield 1957; McMillan 1964). This suggests that, while selection in pygmy-forest and coastal environments has produced distinctly different adaptive traits in ssp. contorta and ssp. bolanderi, the duration of parapatric separation has not been sufficient to allow large differentiation at allozyme loci. A similar argument is that repeated gene flow over a longer time has prevented differentiation of neutral or near-neutral allozymes, while selection for more adaptive morphological and physiological traits has sharply differentiated the populations. Allozymes of populations on a terrace with severe site conditions have differentiated less than those on terraces with less severe conditions, possibly because more recent divergence or strong edaphic selection on adaptive loci has not yet allowed divergence at linked, neutral loci.
Spatial trends support this hypothesized progenitor-derivative relationship. A significant decrease in the percentage of polymorphic loci with increasing elevation is largely a function of the difference between coastal and pygmy-forest populations. No trend in genetic diversity across pygmy-forest populations alone (terraces 3, 4, and 5) is statistically significant.
Principal-component analysis shows populations C2, C4, and C6 to be relative outliers to other coastal populations and to pygmy-forest populations. This may be due to sampling error alone, or may indicate that the pygmy-forest populations originated from coastal stands C1, C3, and C5. The latter thesis is supported for C4 and C6 by a higher frequency of private alleles (private alleles were not included in the principal-component analysis), and by their greater separation by chord distance.
The average [F.sub.ST] values between populations on terraces 1 and 3 are larger than those among higher terraces, providing some support for contorta and bolanderi being subspecies. But this is weak evidence compared to the subspecific differentiation observed for morphological, survival, and growth traits. The allozyme-based estimates of [F.sub.ST] between the two subspecies are relatively low compared to those of many other studies (Hamrick 1987), indicating little population subdivision in spite of the apparently valid subspecific differences between contorta and bolanderi.
TABLE 3. Average [F.sub.ST] values among populations within terraces and among populations on adjacent terraces. Within terraces Adjacent terraces Terrace Terrace no. [F.sub.ST] no. [F.sub.ST] 1 0.057 3 0.051 1, 3 0.065 4 0.029 3, 4 0.049 5 0.035 4, 5 0.039
The cline in allozyme diversity across terraces is such that populations on terraces 4 and 5, with more severe conditions, have greater affinity among populations and within terraces than do populations on terraces 1 and 3. The greater affinity among terrace-5 populations might be attributed to smaller geographical spread rather than to environmental similarity among sites. However, in comparisons of nearby populations on different terraces, genetic distances do not reflect geographic proximity: populations P6, P7, and P8 form an east-west transect across terraces 5, 4, and 3 but do not cluster on the basis of chord distance. Likewise, adjacent populations P10 and P11 on terraces 4 and 3 do not cluster. This might also be attributed to greater gene flow among upper-terrace sites, but prevailing winds from the west and the population structure make that unlikely. There is seldom more than a slight breeze at crown level in the pygmy forests, which are isolated islands of short, stunted vegetation in tall forests, even when the wind is blowing strongly in the area. We find it difficult to hypothesize a mechanism for greater movement of pollen or seeds among pygmy-forest populations than among coastal populations. However, gene flow among populations is typically high in out-crossing, wind-pollinated conifer species (Hamrick 1987; Schuster et al. 1989), and differential gene flow must be considered an alternative hypothesis for the observed patterns.
Gene Flow. - The effective rate of gene flow, quantified by the statistic Nm, which is the product of population size, N, and the proportion of individuals migrating in a given generation, m, can be estimated from [F.sub.ST] values or from frequencies of private alleles (Slatkin 1985b). The data contained insufficient numbers of private alleles for quantifying Nm; however, it can be estimated from [F.sub.ST] as
Nm = (1/[F.sub.ST] - 1)/4 (2)
(Slatkin 1985a). Values of Nm greater than 1, indicating levels of gene flow sufficient to counter genetic drift, are to be expected for wind-pollinated conifers like a pine. The average estimated Nm value for all populations in this study is 4.1. Estimated values for terraces 5, 4, 3, and 1 are 6.9, 8.4, 4.6, and 2.9, respectively.
Although estimates of this parameter may have large sampling errors, it seems unlikely that all four estimates would be greater than I because of chance alone, and the apparent gradient suggests greater gene flow on the upper terraces. However, [F.sub.ST], and therefore these estimates of Nm, may also be affected by differences in time of divergence, degree of disequilibrium, and selection among the terraces; thus there is ambiguity. The lower frequencies of private alleles in the pygmy-forest populations support the pattern of estimated effective gene flow but may also indicate more recent colonization of the upper terraces, in contrast to the hypothesis of Jenny et al. (1969). Therefore, while we may consider that the differences in estimated levels of Nm reflect real differences in effective gene flow among terrace populations, it is possible - even likely - that other causes may better explain the differences.
Colonization Sequence. - If all five terraces were simultaneously and similarly colonized, the observed variation in allozymes, morphology, and edaphic adaptation could be explained by some combination of the following: stronger or more focused selection on higher terraces; different effective population sizes in subdivided populations on each terrace (allowing drift to create greater neutral differentiation on terraces with populations having smaller [N.sub.e] or more frequent bottlenecks); or differential migration in which genes move from upper to lower terraces more commonly than in the reverse direction. Current data and observations support the first component but refute the second and third components of the explanation. Prehistoric subpopulation sizes, bottlenecking, and wind direction may have been very different; if not, the first component selection would have to overwhelm the others to produce the present genetic patterns.
It seems to us more likely that the terraces were not simultaneously and similarly colonized. In estimating the colonization sequence and time of divergence, the following are helpful to consider: sequence of terrace emergence; estimated ages of terraces; occurrence of pine pollen in nearby cores; pattern of genetic diversity; pattern of private-allele occurrence; and time since divergence estimated from genetic distance.
Likely colonization sequences of the terraces are:
(1) 1-(2)-3-4-5 or possibly 2-1 and 3-4-5;
(2) 5-4-3-2-1 as proposed by Jenny et al. (1969); or,
(3) 5-4-3-2-1-3 (from 1 or 2) to 4-5.
In the second sequence, P. contorta continually occupies and evolves on each terrace as it becomes available. The third sequence would have shore pine arriving on the coastal terrace 5, migrating to terraces 4, 3, 2, and 1 as they emerged and developed soils favorable to shore pine, being extirpated as increasingly older terraces evolved unfavorable soils, then recolonizing terrace 3 or leaving a few adapted survivors to evolve and recolonize terraces 4 and 5. As such, the third sequence is conceptually little different from the first, the main difference being the previous history of the shore pine colonizing the lower terraces. The important contrast is therefore between the second sequence and the others (i.e., between long-term sequential and short-term nearly simultaneous colonization).
The genetic data support the first and third sequences, colonization of recent shore-pine sites by shore pine, followed by adaptation and evolution of Bolander pine on terrace 3, and by further adaptation and colonization of the upper terraces, probably in the order 3-4-5. If so, populations of shore pine providing the pre-Bolander ancestors that colonized terrace 3 from terraces 1 or 2 (sequence 1), or that migrated from terrace 3 to terrace 2 leaving survivors (sequence 3), are the focus of interest. These shore pines must have arrived on the first or second terrace after soils favorable to pine developed - on terrace I within the past 115,000 yr, or on terrace 2 within the past 285,000 yr.
Pollen records, although sketchy, indicate recent arrival of P. contorta in the area, as mentioned earlier, and provide evidence against the long-term colonization suggested by sequences 2 and 3. Pollen from a postglacial bog near Fort Bragg, in the immediate vicinity of the study area, indicates that P. contorta was abundant over the past 3000 yr; older cores from about 120 km north of the study area indicate that P. contorta probably arrived from the north less than 8000 yr ago (Heusser 1960).
Thus, under the first and third sequences, estimates of terrace emergence place an upper bound on the time since shore pine colonized terrace 1 or 2 at 115,000 to 285,000 yr ago, and pollen records suggest a lower bound of 3000 yr - less if colonization occurred after pines were abundant, and surely less than 8000 yr if the Cape Mendocino core indicates the absence of P. contorta in the region.
Time Since Divergence. - Given the known effective population sizes and an absence of gene flow, generations since divergence ([G.sub.d]) can be estimated from allele frequencies expressed as chord distance, [D.sub.d] (Felsenstein 1985), where [G.sub.d] = [D.sub.c](2[N.sub.e]). Of course, prehistoric population sizes are unknown, but current census counts are 50 to 100 adult trees per hectare for shore pine and 700 per hectare for Bolander pine. With a conservative 1000 [N.sub.e], the estimate is 300 generations since divergence. With 60 yr as the average mean age of Bolander pine (Westman and Whittaker 1975), equivalent to a generation length, the estimate is 18,000 yr. If bottlenecks occurred, or if [N.sub.e] values are smaller, this is an overestimate; however, to the degree that effective gene flow occurred between diverging populations, it is an underestimate. The estimate fits comfortably below the upper bound of 285,000 yr, and thus does not support the hypothesis of nearly a million years since divergence (Jenny et al. 1969). If the effect of smaller [N.sub.e] or bottlenecks exceeded the effect of gene flow, the time could easily fall below or within the 3000- to 8000-yr lower bound.
With the assumption of equal mutation rates ([Mu]), neutrality of alleles, and linkage equilibria, effective population sizes can be calculated from expected heterozygosity, [H.sub.e] (Kimura and Crow 1964), where [H.sub.e] = 4[N.sub.e][Mu]/(4[N.sub.e][Mu] + 1). This model yields 2820 for shore pine and 2355 for Bolander pine, suggesting a greater [N.sub.e] for shore pine than for Bolander pine, which is substantially at odds with current census numbers and population densities for the two subspecies. The discrepancy might be explained by linkage of the allozyme loci to strongly selected genes, by founder effects during colonization and perhaps subsequent bottlenecking after such events as fires in the droughty pygmy sites, or - less likely - by selection on some or all of the allozyme loci.
However, the proposal of Jenny et al. (1969) with respect to the development of the ecosystem in concert with the dynamics of podzol formation (in which carboxyl groups from conifer needles and ericaceous leaves create acids leading to podzolization) is not contradicted by our data. Several ericaceous species, plus bishop pine, pygmy cypress (Cupressus pygmaea (Lemm.) Sarg.), and redwood were probably present and could have contributed to podzolization. They may have participated in the development of soil and vegetation, even if P. contorta was a latecomer to the system.
The lack of divergence of allozymes relative to growth and morphology has been observed in other conifers. Merkle et al. (1988) studied allozyme variation in Douglas-fir in southwest Oregon, an environmentally heterogeneous region. They found significant geographic variation in allozyme frequencies; the geographic patterns did not correspond to patterns of growth and morphology in seedling common-garden studies. Linhart et al. (1989) also found no concordance of morphological and allozyme traits in the offspring of crosses between varieties of Pinus ponderosa. However, Grant and Mitton (1977) found allozyme differences concordant with differing growth forms of Engelmann spruce (Picea engelmannii) at treeline, and Millar (1983) found strong differences in allele frequencies associated with a steep morphological cline of two races of Pinus muricata in the pygmy-forest region.
Millar's results may be profitably compared with those of this study of Pinus contorta in the same region. She first investigated a steep cline between a southern "green-needle" and a northern "blue-needle" race of bishop pine. Across this cline, populations showed marked changes in frequencies of both specific allozymes and foliage traits. In a second study, Millar (1989) investigated 18 populations on both coastal and higher elevation sites, the latter including pygmy-forest soils, well-drained high-forest soils and shallow, sandy soils. The allozymes clearly separated the 8 predominantly green-needle populations from the 10 predominantly blue-needle populations. However, within those sampled sets, allozyme analysis did not separate coastal from ridge populations, nor did it cleanly cluster populations on pygmy-forest soils separately from those on deep, well-drained high-forest soils. In this respect, her results are similar to those for P. contorta, but in contrast, Millar found less allozyme variation in coastal than in ridge populations. She concluded that the green-needle race of bishop pine is migrating north and replacing the blue-needle bishop pine, except on pygmy-forest sites.
The historical dynamics of P. muricata seem different from those of P. contorta, where the Bolander subspecies seems to be recently derived from shore pine and in invading pygmy-forest sites where shore pine did not previously exist. Millar (1989, p. 878) also concluded that although allozyme divergence is minor, "differences in other traits . . . are as great as those that distinguish species in related pines." Thus, although historical dimensions and sequences appear to be different, rapid divergence of adaptive traits and lack of strong divergence of allozyme markers appear to be consistent for these pine species occupying similarly contrasting edaphic sites on California's north coast.
Our data suggest that shore pine colonized a recently developed shore-pine site, perhaps much less than 100,000 yr ago. This contradicts the hypothesis of Jenny et al. (1969) that Bolander pine and the severe soil conditions of the pygmy forest developed simultaneously over a much longer period of time. The data suggest that individuals from coastal populations of shore pine-colonized terrace 3 and their offspring adapted to it after its pygmy-forest soils developed; that individuals from the newly adapted populations on terrace 3 then colonized terrace 4, and their offspring further adapted to its more severe soil; and that individuals from terrace 4 then colonized, and their offspring further adapted to, terrace-5 severe soils. The decrease in the average number of alleles per locus, the differences in genetic distances from lower to higher terraces, and the frequencies and distributions of private and semiprivate alleles support this time frame and sequence.
Pinus contorta in this region exemplifies taxa in which different classes of genetic data provide different kinds of information (Libby and Critchfield 1987). Traits such as cone morphology, tree form, and survival and growth under different edaphic conditions have sharply diverged between shore pine and Bolander pine with little corresponding divergence in allozymes. The selection pressures in the pygmy-forest ecosystem seem likely to have been strongly directional, and Bolander pine appears to have differentiated rapidly from shore pine in traits adapting it to stringent conditions. The allozyme data indicate either that the time since separation has been relatively short, or that repeated migration has been countered by selection for divergent adaptive traits while swamping differences at the allozyme loci.
This research was inspired by the late W. B. Critchfield. M. T. Conkle, A. Schultz, M. Slatkin, and R. Westfall provided helpful guidance and comments throughout the course of this research. C. Millar thoroughly and thoughtfully reviewed the manuscript. The Institute of Forest Genetics, Pacific Southwest Forest and Range Experimental Station, USDA Forest Service, Berkeley, and particularly M. T. Conkle, generously provided access to laboratory facilities, supplies, and computing time. S.A. was supported by fellowships from the Natural Science and Engineering Research Council of Canada, the Graduate Division of the University of California, Berkeley, and the Department of Forestry and Resource Management, University of California, Berkeley. This research was also supported in part by grants to S.A. from the University of California Natural Reserve System, and the California Native Plant Society. This is Paper 2716 of the Forest Research Laboratory, Oregon State University, Corvallis.
Aitken, S. N. 1989. Population genetics of Pinus contorta on coastal and pygmy-forest sites in Mendocino County, California. Ph.D. diss. University of California, Berkeley.
Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. Phylogenetic analysis: models and estimation procedures. Evolution 21:550-570.
Conkle, M. T., P. D. Hodgkiss, L. B. Nunnally, and S. C. Hunter. 1982. Starch-gel electrophoresis of conifers: a laboratory manual. USDA Forest Service General Technical Report PSW-64. Pacific Southwest Forest and Range Experiment Station, Berkeley, Calif.
Critchfield, W. B. 1957. Geographic variation in Pinus contorta. Maria Moors Cabot Foundation Publication 3. Harvard University, Cambridge, Mass.
Felsenstein, J. 1985. Phylogenies from gene frequencies: a statistical problem. Systematic Zoology 34: 300-311.
Forrest, G. I. 1980. Geographical variation in the monoterpenes of Pinus contorta oleoresin. Biochemical Systematics and Ecology 8:343-359.
Gottlieb, L. D. 1977. Electrophoretic evidence and plant systematics. Annals of the Missouri Botanical Garden 64:161-180.
-----. 1981. Electrophoretic evidence and plant populations. Pp. 1-46 in L. Reinhold, J. B. Harborne, and T. Swain, eds. Progress in phytochemistry, vol. 7. Pergamon, Oxford.
Grant, M. C., and J. B. Mitton. 1977. Genetic differentiation among growth forms of Engelmann spruce and subalpine fir at tree line. Arctic and Alpine Research 9:259-263.
Gregory, R. P. G., and A. D. Bradshaw. 1965. Heavy metal tolerance in populations of Agrostis tenuis Sibth. and other grasses. New Phytologist 64:131-143.
Griffin, J. R., and W. B. Critchfield. 1972. The distribution of forest trees in California. USDA Forest Service Research Paper PSW-82. Pacific Southwest Forest and Range Experiment Station, Berkeley, Calif.
Hamrick, J. L. 1987. Gene flow and distribution of genetic variation in plant populations. Pp. 53-67 in K. M. Urbanska, ed. Differentiation patterns in higher plants. Academic Press, New York.
Heusser, C. J. 1960. Late Pleistocene environments of north Pacific North America. American Geographical Society Special Publication 35.
Jain, S. K., and A. D. Bradshaw. 1966. Evolutionary divergence among adjacent plant populations. I. The evidence and its theoretical analysis. Heredity 21: 407-441.
Jenny, H., R. J. Arkley, and A. M. Schultz. 1969. The pygmy forest-podzol ecosystem and its dune associates of the Mendocino coast. Madrono 20:60-74.
Kimura, M., and J. F. Crow. 1964. The number of alleles that can be maintained in a finite population. Genetics 49:725-738.
Libby, W. J., and W. B. Critchfield. 1987. Patterns of genetic architecture. Annales Forestales 13:77-92.
Linhart, Y. B., M. C. Grant, and P. Montazer. 1989. Experimental studies in ponderosa pine. I. Relationship between variation in proteins and morphology. American Journal of Botany 76:1024-1032.
Loveless, M. D., and J. L. Hamrick. 1984. Ecological determinants of genetic structure in plant populations. Annual Review of Ecology and Systematics 15:65-95.
McMillan, C. 1964. Survival of transplanted Cupressus and Pinus after thirteen years in Mendocino County, California. Madrono 17:250-253.
McNeilly, T. 1967. Evolution in closely adjacent plant populations. III. Agrostis tenuis on a small copper mine. Heredity 23:99-108.
Merkle, S. A., W. T. Adams, and R. K. Campbell. 1988. Multivariate analysis of allozyme variation patterns in coastal Douglas-fir from southwest Oregon. Canadian Journal of Forest Research 18:181-187.
Millar, C. I. 1983. A steep cline in Pinus muricata. Evolution 37:311-319.
-----. 1989. Allozyme variation of bishop pine associated with pygmy-forest soils in northern California. Canadian Journal of Forest Research 19:870-879.
Morris, R. S., and P. T. Spieth. 1978. Sampling strategies for using female gametophytes to estimate heterozygosity in conifers. Theoretical and Applied Genetics 5:217-222.
Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89:583-590.
Parlatore, P. 1868. Coniferae. Pp. 361-521 in A. P. de Candolle, ed. Prodromus systematis universalis regni, vegetabilis, vol. 16, sect. 2. Masson, Paris.
SAS Institute. 1987. SAS/STAT guide for personal computers, Version 6. SAS Institute, Cary, N.C.
Schuster, W. S., D. L. Alles, and J. B. Mitton. 1989. Gene flow in limber pine: evidence from pollination phenology and genetic differentiation along an elevational transect. American Journal of Botany 76:1395-1403.
Scholars, R. E. 1979. Water relations in the pygmy forest of Mendocino County. Ph.D. diss. University of California, Davis.
-----. 1982. The pygmy forest and associated plant communities of coastal Mendocino County, California: genesis, soils, vegetation. Black Bear Press, Mendocino, Calif.
Slatkin, M. 1985a. Gene flow in natural populations. Annual Review of Ecology and Systematics 16:393-430.
-----. 1985b. Rare alleles as indicators of gene flow. Evolution 39:53-65.
Swofford, D. L., and R. B. Selander. 1981. BIOSYS-1. A computer program for the analysis of allelic variation in genetics. Release 1. University of Illinois, Urbana.
Westman, W. E. 1975. Edaphic climax pattern of the pygmy forest region of California. Ecological Monographs 30:279-338.
Westman, W. E., and R. H. Whittaker. 1975. The pygmy forest region of northern California: studies on biomass and primary productivity. Journal of Ecology 63:493-520.
Wheeler, N. C., and R. P. Guries. 1982. Population structure, genic diversity, and morphological variation in Pinus contorta Dougl. Canadian Journal of Forest Research 12:595-606.
Wright, S. 1965. The interpretation of population structure by F-statistics with special regard to systems of mating. Evolution 19:395-420.
|Printer friendly Cite/link Email Feedback|
|Author:||Aitken, Sally N.; Libby, William J.|
|Date:||Aug 1, 1994|
|Previous Article:||Inbreeding load and heterosis in relation to population size in the mint Salvia pratensis.|
|Next Article:||Hybridization and chloroplast DNA variation in a Pinus species complex from Asia.|