| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Rocky Mountain Research Station, Forest Service, U.S. Department of Agriculture, 1221 South Main, Moscow, Idaho 83843
Received for publication June 25, 1998. Accepted for publication October 27, 1998.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: ecological genetics • genecology • genetic variation • microevolution • Ponderosae.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Mitton, 1995
), the genetic structure of isolated, disjunct, or geographically peripheral populations repeatedly draws the attention of those studying the relative effectiveness of evolutionary forces (e.g., Karron, 1991
; Hamrick, Schnabel, and Wells, 1994
; Linhart and Premoli, 1994
; Lewis and Crawford, 1995
; Rehfeldt, 1997
). Because of reduced interpopulation gene flow, finite population sizes, and ecologically extreme environments commonly associated with geographic isolates, either or both strong selection and chance events are given opportunity to control genetic structure (Carson, 1959
; Mayr, 1970
; Yeh and Layton, 1979
; Betancourt et al., 1991
).
For the pines of the Ponderosae subsection of genus Pinus (see Little and Critchfield, 1969
) that inhabit the mountain islands punctuating the Sonoran and Chihuahuan Deserts, an intriguing complex of factors may be influencing genetic structure. First, systematic relationships are so confused that hybridization and introgression among taxa may appear as rampant (Peloquin, 1984
) or incidental (Rehfeldt et al., 1996
), depending on the classification system one accepts. Second, the Ponderosae tend to be confined to mountain islands, and, therefore, populations are isolated and disjunct. Third, the flora of the region as a whole is an admixture of Rocky Mountain species at the southern margin of their distribution and Sierra Madrean species at their northern limits (McLaughlin, 1992
); all populations, therefore, are peripheral. Fourth, moderately steep clines presumably developing from environmental selection (Rehfeldt, 1990
, 1991
, 1993
) have already been documented for broadly distributed continuous populations of the most prominent member of the Ponderosae, Pinus ponderosa Laws., as have the importance of both stochastic (Hamrick, Blanton, and Hamrick, 1989
) and directional forces (Mitton et al., 1977
) in geographically peripheral populations. These isolated, disjunct, and peripheral populations thus exist under conditions that make possible the development and maintenance of unusual genetic structures.
The natural history of the Southwest enhances these possibilities. Although the late Wisconsin vegetation was displaced to elevations as much as 1000 m lower than that of today (Martin and Mehringer, 1965
), at no time were the mountain islands connected by corridors of montane conifers (Van Devender, 1990a
, b
). In addition, the Ponderosae were a decidedly minor component of the Wisconsin flora, being absent from the paleoecologic record in all but the Santa Catalina Mountains (Van Devender, 1990a
, b
). This means that contemporary populations are relicts of those developing from either (1) the expansion of late Wisconsin populations too small to be represented in the paleoecologic record, or (2) long-distant transport sometime during the Holocene. Either case provides a stage ideal for bottlenecks, founder events, or drift to have influenced genetic variation.
This paper deals with the genetic structure of the Ponderosae of the Southwest as revealed by their wind-pollinated seedling progenies. The topic not only presents an evolutionarily intriguing interaction of factors dealing with closely related taxa, but also is relevant to modern issues in population biology. On the one hand, systematics is at the core of conservation biology (Avise, 1996
), and, on the other, the genetic structure of isolated and disjunct populations reflects a natural experiment addressing the effects of fragmentation of natural populations from human encroachment (Karron, 1991
; Lynch, 1996
).
As a prerequisite to implementing this study, cones were collected from natural populations. During these collections, data on cone, needle, and branch morphology were recorded in an attempt to increase the objectivity of field classifications. A multivariate examination of these phenotypic data (Rehfeldt et al., 1996
) demonstrated that none of the three contemporary taxonomic systems used for this region satisfactorily accounted for the variation that occurs across the landscape. The analyses segregated trees into four groups, labeled temporarily as "engelmannii," "scopulorum," "quinquefoliata," and "taxon X." The "engelmannii" group was composed of trees with the characteristics of P. engelmannii Carr.; "scopulorum" with those of P. ponderosa var. scopulorum Engelm.; "quinquefoliata" with those of P. durangensis forma quinquefoliata Mart., P. durangensis Mart., P. ponderosa var. arizonica (Engelm.) Shaw, and P. arizonica Engelm. (sensu Wheeler, 1878
); and "taxon X" with those of P. arizonica Engelm. (sensu Perry, 1991
; Farjon and Styles, 1997
), P. arizonica var. stormiae Mart., and hybrid swarms (Peloquin, 1984
; Rehfeldt, 1993
). In this paper, the validity of these four groups is tested by considering genetic variation among and within groups. At the outset, therefore, each group is considered to be a separate taxon.
Although the taxonomic confusion surrounding the Ponderosae is not to be reviewed herein (for a discussion, see Rehfeldt et al., 1996
), it is pertinent to note that Pinus arizonica references two taxa with different characteristics, both of which are attributed to Engelmann.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
|
), the collections included 78 "engelmannii," 129 "quinquefoliata," 10 "scopulorum," and 73 "taxon X." Trees of each group were dispersed across 
800 m of elevation, with the "engelmannii" trees occurring at a median elevation of 1900 m, "quinquefoliata" at 2200 m, "scopulorum" at 2200 m, and "X" at 2500 m. In making the cone collections, an attempt was first made to obtain a random sample of seeds from the array of genotypes inhabiting each mountain range. A secondary objective was to seek out and include samples from trees exhibiting phenotypes intermediate between taxa. Pinus ponderosa, however, was not sampled proportionally to its abundance particularly in the Black Range and Pinos Altos Mountains.
The controls needed for addressing the relationships among taxa were obtained from locations peripheral to the mountain islands and, therefore, presumably are less subject to possible effects of hybridization and introgression. Thus, cones from ten trees were bulked as population samples from five locations within the botanical distribution of P. p. var. scopulorum, and seeds from 12 populations were obtained from Mexico. Although Mexican populations were labeled as six populations of P. engelmannii (mixtures of seeds from 15 trees), five of P. arizonica (mixtures of seeds from five trees), and two of P. durangensis (mixtures of seeds from five trees), the descriptions of Perry (1991)
suggest that these taxa roughly correspond to "engelmannii," "taxon X," and "quinquefoliata," respectively. Nevertheless, throughout this paper, samples from Mexico are referenced according to the scientific name that accompanied the seeds.
In total, therefore, the experimental materials included seedlings derived from 290 single trees and 18 populations. Without coancestry, seedlings derived from cone collections made from single trees are half-sibs and, therefore, are customarily referred to as a family. Populations are composed of many families proximal to a single location defined by latitude, longitude, and elevation. The experimental materials thus consisted of populations and families, each of which is referenced subsequently as a seedlot.
To assess genetic variation, seedlings from the 308 seedlots were grown in plastic containers (740 cm3) in an experimental design that consisted of nine seedlings growing in row plots in each of five blocks. The containers were arranged in trays that held three plots of nine trees. Seedlings were grown for six mo in a shadehouse (50% shade) at Moscow, Idaho (latitude = 46.7°, longitude = 117°), were transferred to a greenhouse for the winter months where temperatures were maintained above -2°C, and, beginning in March of the second growing season, were exposed to a daytime temperature of 25°C, which was allowed to cool to a minimum of 13°C at night. In mid-May, after shoot elongation and needle growth were completed, the seedlings were again transferred to the shadehouse for the duration of the summer. Because of finite greenhouse space, approximately one-third of the seedlots were sown in three consecutive years. The sowings, hereafter referenced as test 1, test 2, and test 3, contained five common seedlots, two "quinquefoliata" families and three "scopulorum" populations.
Differences in fecundity of parental trees, a variable yield of sound seeds per cone, and differing germination rates meant that all families were not represented by 45 seedlings. While the average number of observations for each family was 40, the minimum was 5. In interpreting the material that follows, it is important to note that cones were collected in the Davis and Chisos Mountains (Fig. 1) in a year of very low fecundity. These locations, therefore, were represented by few families (Table 1), none of which had >15 seedlings.
The relationships among taxa and the genetic structure of taxa were assessed from ten variables, the utility of which had been demonstrated previously for the Ponderosae (Rehfeldt, 1993
). All variables were obtained from measurements on individual seedlings during their second growing season.
Five of the variables described the pattern of shoot elongation. These variables were obtained from predictions made by regression models that describe shoot elongation from a modified logistic function (Rehfeldt and Wykoff, 1981
):
 |
3-d intervals during the second growing season; 13 measurements were available for seedlings in test 1, and 12 were available for tests 2 and 3. Regression statistics produced by this function were used to estimate the (1) initiation of shoot elongation, the day on which 2 mm of elongation had occurred; (2) cessation of elongation, the day on which all but 2 mm of elongation had occurred; (3) duration of elongation, the number of days between initiation and cessation; (4) rate of elongation, the amount of elongation per day over the period during which 20–80% of the annual elongation had occurred; and (5) total elongation.
Additional variables included 2-yr height; number of needles per fascicle, calculated from a sample of ten fascicles distributed along the 2-yr shoot; needle length, measured in July on a single fascicle near the center of the 2-yr shoot; and stem diameter, measured in August at a point intermediate between the cotyledons and the soil surface. The tenth variable was the ratio of the 2-yr height to diameter.
Because a different set and number of seedlots were grown in each of three years, the data needed to be transformed and scaled in order to obtain a single unbiased data set for analysis (Rehfeldt, 1989
). As a first step, the main effects of the tests and blocks within tests were removed by transforming the data to standardized normal deviates within blocks:
 |
and 
are the mean and standard deviation, respectively, for all individuals in block k of test l.
Next, data from test 2 and test 3 were scaled to those of test 1 according to scaling factors determined by the mean performance of the five seedlots common to all tests. Thus, scaled standard deviates (Z'ijk) were defined such that
These transformations produced a single data set within which the main effects of tests and blocks within tests had been eliminated. Using standardized normal deviates also homogenized variances that may have been differentially affected by environmental effects during the disparate testing periods. Scaling assured that the common seedlots had the same mean performance in all tests. Even though the transformations eliminated the main effects of blocks, the error structure of the data, consisting of sampling and experimental errors, was not affected. However, the possibility exists that the scaled data set was confounded by interactions of seedlots and test environments. This possibility needed to be explored before the scaled data set was analyzed.
Statistical analyses
Statistical analyses were performed to (1) assess the effectiveness of standardization, (2) assess the relationships among taxa, and (3) describe the genetic structure of taxa. All analyses used software of SAS (1996)
.
The effectiveness of transforming and scaling of data was assessed for each of the ten variables with an analysis of variance using a model of random effects for the five common seedlots:
 |
Relationships among taxa were examined with principal components and canonical discriminant analyses, both of which used the correlation matrix of ten variables (values of Z') measured on each seedling. The purpose of the principal component analysis was to assess the relationship among taxa along multivariate vectors independently of the taxonomic classifications. The discriminant analysis was made to separate taxa according to the prior classifications. For these analyses, Mexican seedlots were treated as taxa separate from "engelmannii," "scopulorum," "quinquefoliata," and "taxon X." From these analyses, therefore, an attempt was made to separate seven groups. Because the classification of taxa from the Davis and Chisos Mountains (Fig. 1) is uncertain, families from these ranges were identified in the ordinations produced from the multivariate analyses.
The genetic structure of taxa was assessed from analyses of variance performed according to a model of random effects:
 |
2) of an effect to the total of all components (T2). The proportion attributable to the effects of families was expressed by F2/T2, that attributable to the effects of populations (P2) by (M2 + L2)/T2, and that attributable to the total genetic effects (G2) by (P2 + F2)/T2. Simple correlation and regression analyses of performance on geographic variables (latitude, longitude, and elevation) were used to assess the nature of the clines controlling interpopulation variance. These analyses used population means for each location within a mountain range.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
|
Even though principal component analyses summarized variance within the transformed data set independently of taxon classifications, mean values for each seedlot could be calculated for each component. Ordination of these means for the first and fourth principal components provided the best separation of taxa (Fig. 2). This ordination shows: (1) a lack of separation of "scopulorum" from "taxon X"; (2) general separation of "engelmannii" and "quinquefoliata" from each other as well as from "scopulorum" and "taxon X"; (3) alignment of Mexican populations labeled P. engelmannii with the "engelmannii" group; (4) alignment of Mexican populations labeled P. durangensis on the periphery of the "quinquefoliata" group; (5) alignment of Mexican populations labeled P. arizonica on the periphery of "taxon X"; and (6) confusion in the identity of as many as three "engelmannii" families and four "quinquefoliata" families. Principal components other than the first and fourth provided for no additional separation of taxa.
|
Ordination of seedlot means along the first and second canonical vectors (Fig. 3) illustrates many of the same relationships among taxa as described by principal components (Fig. 2), including a lack of separation of "scopulorum" from "taxon X." However, because canonical vectors discriminate among groups, ordinations using canonical variates tended to separate taxa more discretely. As a result, the ordination of Fig. 3 does not substantiate confusion in the identity of the four "quinquefoliata" families that was suggested in Fig. 2, even though those four were among the 20 within the group that had the lowest means for the first canonical vector. Quite similarly, even though "engelmannii" families are discretely separated from "scopulorum" and "taxon X" in Fig. 3, the three "engelmannii" families with the lowest means for the second vector were the same three that were located within the dispersion of "scopulorum" and "taxon X" families in Fig. 2.
|
|
|
|
The status of the seven families whose identity appears confused in Fig. 2 are of particular interest because these families could represent either first-generation hybrids or introgressants. Of these families, the four "quinquefoliata" quite likely represent outliers in the array of genotypes existing within the taxon. This conclusion is suggested by (1) the discrete and distant separation of these families from "scopulorum" and "taxon X" in Fig. 2, (2) variances within families that were typical of other "quinquefoliata" families rather than either the uniformity that should accompany first-generation hybrids or the large variances that would be indicative of transgressive segregation in introgressants, and (3) a lack of ambiguity in the field classifications of Rehfeldt et al. (1996)
. However, the three "engelmannii" families whose identity appears confused in Fig. 2 may contain progenies that are either first-generation hybrids or hybrid derivatives. For these, separation from "scopulorum" and "taxon X" in Fig. 3 was problematic, variances within families were approximately one-half those of the other families, and the phenotypic classifications in the field were confused for two of the parental trees.
Genetic structure of taxa
Because the analyses of parental phenotypes (Rehfeldt et al., 1996
) as well as multivariate analyses of their progenies (Figs. 2, 3) were incapable of separating "scopulorum" from "taxon X," the analyses of genetic variation that follow consider these taxa as members of the same taxon. The analyses thus consider three primary taxonomic groups, "engelmannii," "quinquefoliata," and the combination of "scopulorum" and "taxon X." Analyses of the "engelmannii" group omitted the three families within which intertaxa hybrids may be segregating.
As suggested by the multivariate analyses (Figs. 2–6
), genetic variances within the three primary groups were pronounced. Not surprisingly, therefore, univariate analyses detected differences among families within locations for all variables and for all taxa (Table 4). Whether expressed as proportions (Table 4) or absolute values (Table 5), these family effects tended to be large, accounting for an average of 14% of the total variance in "engelmannii" and 11% in the other two groups. Without coancestry, these intraclass correlations equal one-fourth of the narrow-sense heritability. As a result, an average heritability for the ten variables would be 0.58 or 0.45, and 0.46 for "engelmannii," "scopulorum" and ‘X,’ and "quinquefoliata," respectively. These values correspond to an average individual heritability of 0.55 for the five variables describing 2-yr shoot elongation in families of the broadly distributed P. ponderosa var. ponderosa (Rehfeldt, 1992
). The results thus demonstrate abundant genetic variability within populations of the Ponderosae of the Southwest.
|
|
Structure of "engelmannii"
On average, genetic effects were evenly distributed among and within populations (Table 4). However, of the portion allocated to populations, most of the variance occurred among locations within ranges. Thus, for none of the variables could significant effects of mountain ranges be detected, but for six of the variables, effects of locations within ranges were significant (Table 4). As a result, an average of 4% of the total variance was attributed to the effects of mountain ranges, 10% to the effects of locations within ranges, and 14% to families within locations. In terms of the total genetic effects, these percentages equate to 16, 34, and 50%, respectively.
Despite high variance among populations, statistically significant (P < 0.05) simple correlations related elevation of a population to only needle length (r = -0.64) and number of needles per fascicle (r = 0.45). By using the techniques of Rehfeldt (1993)
, one can calculate that populations of "engelmannii" need to be separated by 370 m in elevation before genetic differentiation becomes a reasonable probability. An interval of 370 m encompasses most of the elevation distribution of "engelmannii" in all but the Chiricahua Range and thus shows that altitudinal clines are so weak that most of the genetic variation among populations remains unexplained.
Assessing the nature of interpopulation clines in "engelmannii," however, is hindered by a distribution limited to four mountain ranges in the Southwest. The populations, therefore, represent only the northern tip of the taxon's distribution. To assess further the steepness of geographic clines, P. engelmannii populations from Mexico were added to the database so that rates of differentiation could be considered across 10Å of latitude. The resulting correlations of performance with latitude were statistically significant (P < 0.05) for only two of the ten variables: needle length (r = -0.41) and 2-yr diameter (r = -0.74). The clines, however, were so weak that populations would have to be separated by 3.75° (420 km) of latitude before there would be a reasonable probability of genetic differentiation.
Together, these results demonstrate that much of the variation among populations occurs nonsystematically across the landscape. Variance unaccounted for by geographic variables could reflect either (1) chance events such as sampling errors, founder events, or genetic drift, or (2) adaptations to microenvironments not reflected by geographic gradients. Regardless, the results mirror those produced by allozyme analyses (Bermejo-Velazquez, 1993
) and suggest that an understanding of this taxon's genetic structure would be aided by knowledge of the mating system.
Structure of "quinquefoliata"
Despite statistical significance for five variables, effects of mountain ranges accounted for an average of only 3% of the total variance, effects of locations within ranges only 1%, and effects of families within locations only 11% (Table 4). As percentages of the total genetic effects, these values equate to 18, 6, and 76%, respectively. Much of the genetic variance, therefore, occurs within populations.
A combination of low interpopulation variances and high intrapopulation variances also typifies Pinus monticola Dougl. (Rehfeldt, 1979
) and, therefore, is not necessarily unusual for forest trees. However, the surprising outcome of the present analyses was that a single population was responsible for most of the population effects. Seedlings from the Black Range had by far the fewest needles per fascicle, averaging three needles in 4.5% of their fascicles and five needles in 66%, while the corresponding percentages for "quinquefoliata" in other populations were 1.7 and 80%, respectively. Seedlings from the Black Range also had the smallest diameters and largest ratios of height to diameter. These seedlings, therefore, provided extreme mean values for the three variables contributing most strongly to the main effects of mountain ranges (Table 4). It seems probable, therefore, that eliminating this population from the analyses would have obliterated interpopulation variances.
In association with low variance among populations, correlation analyses were unable to relate variance among populations to geographic gradients. In total, therefore, the results show that variance among populations is incidental and inconsequential in "quinquefoliata" of the Southwest.
Low interpopulation variance and lack of clines in the Southwest, however, do not necessarily preclude the existence of geographic clines that could not be detected in a sample from the northern margin of the taxon's distribution. If "quinquefoliata" and P. durangensis are considered as belonging to the same taxon, latitudinal clines become detectable (P < 0.05) for 2-yr elongation (r = -0.44), 2-yr height (r = -0.62), and the duration and cessation of shoot elongation (r = -0.65 and r = -0.64, respectively). Because of the absence of samples in the region intermediate between those of "quinquefoliata" families and P. durangensis populations (Fig. 1), these correlations are difficult to interpret. Nevertheless, the available data would suggest that there is a reasonable probability of genetic differentiation (see Rehfeldt, 1993
) between populations separated by 2.65° (300 km) of latitude.
Structure of "scopulorum" and "X"
Like "engelmannii," the taxon combining "scopulorum" and "X" exhibited large genetic variances within and among populations. However, unlike "engelmannii," effects of mountain ranges tended to dominate interpopulation variances (Table 4). On average, the effects of mountain ranges accounted for 12% of the total variance, locations within ranges for 3%, and families within locations for 11%. In terms of the total genetic effects, these percentages equate to 47, 11, and 42%.
In comparison to the other taxa, variances among populations in the combined taxon quite likely have been inflated by the presence of families from the Davis and Chisos Mountains of Texas. Including these families in the database greatly increased the geographic extent of the taxon relative to that of "engelmannii" and "quinquefoliata" (Fig. 1) and may have biased comparisons of interpopulation variances. However, recalculating the statistics of Table 5 without the Texas families still produced average intraclass correlations of 0.07, 0.03, and 0.12 for the effects of mountain ranges, locations within ranges, and families within locations, respectively. Without the Texas families, these same effects accounted for 34, 14, and 52% of the genetic effects, respectively.
Both analyses, therefore, lead toward the same general conclusions: in the taxon combining "scopulorum" and "X," genetic variability is abundant and tends to be prominent both among and within populations. Most of the genetic variation among populations is attributable to the main effects of mountain ranges, although secondary effects can be attributed to locations within ranges.
In "scopulorum" and "X," moreover, clines of genetic variation are prominently associated with geographic gradients. Correlations of performance with elevation of the population were significant (P < 0.05) for four of the variables: needle length (r = -0.64), needles per fascicle (r = 0.37), 2-yr height (r = -0.35), and the height : diameter ratio (r = -0.50). Although these coefficients were calculated from a database containing 24 families and five populations, similar coefficients were obtained for analyses using only family data. These associations suggest that populations must be separated by at least 320 m of elevation before there is a reasonably good probability of genetic differentiation (see Rehfeldt, 1993
). This value corresponds to an interval of 300 m that was estimated independently for P. ponderosa var. scopulorum populations to the north (Rehfeldt, 1993
).
Latitudinal clines assessed after including the five Mexican populations of P. arizonica in the database produced statistically significant (P < 0.05) coefficients for seven variables: 2-yr elongation (r = -0.65), initiation of shoot elongation (r = -0.72), rate of shoot elongation (r = -0.36), duration and cessation of elongation (r = -0.78 and r = -0.73, respectively), and needle length (r = 0.62). The clines, moreover, were moderately steep; on average, populations would have to be separated by 1.5° of latitude (170 km) before genetic differentiation becomes probable. This distance corresponds to an independent estimate of 100 km for populations of P. p. var. scopulorum (Rehfeldt, 1993
). These latitudinal clines, incidentally, readily account for the differences that seemed apparent in Fig. 6 between Mexican populations of P. arizonica and "taxon X" on the one hand and Chisos Mountain families and "taxon X" on the other.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
) reached from analyses of phenotypic variation among trees growing in natural populations. The Ponderosae of the Southwest are readily aligned into three primary taxonomic groups, one of which contains two subgroups. One of the primary groups consists of the "engelmannii" taxon, the characteristics of which are consistent with the descriptions of P. engelmannii Carr. of both the United States and Mexico.
The "quinquefoliata" taxon belongs to a second group to which P. durangensis Mart. also seems to be a member. Historical precedence would demand that the group be labeled P. arizonica, which, according to Wheeler (1878)
, was named by Engelmann as having the predominant five (but also 4, 6, or 7) fine needles per fascicle that typifies "quinquefoliata." This group, therefore, should not include the Mexican taxon with largely three but also four or five coarse needles per fascicle that also is referenced as P. arizonica Engelm. (see Perry, 1991
). Incidentally, according to Farjon and Styles (1997)
, in Mexico, southern populations of P. durangensis commonly exhibit six or seven needles per fascicle, but the number decreases toward 5 with increasing latitude. This observation along with the present results thus supports a conclusion suggesting that "quinquefoliata" of the southwest United States and P. durangensis Mart. are conspecific.
The two subgroups of a third taxonomic group appear to be varieties of P. ponderosa. While "scopulorum" obviously references P. ponderosa var. scopulorum Engelm., a second variety, currently unnamed, would encompass "taxon X," a group that could not be separated discretely from "scopulorum." This unnamed variety also should include the Mexican populations currently labeled as P. arizonica, which primarily have three (but also four) coarse needles per fascicle (see Perry, 1991
). The present analyses also suggest that the nomenclaturally disputed populations inhabiting the Chisos Mountains belong to this unnamed variety.
The results do not support hybridization as a force influencing contemporary systematic relationships. The ordinations of Figs. 2 and 3 along with the results of Rehfeldt et al. (1996)
tend to reject "taxon X" as a hybrid swarm (see Peloquin, 1984
) involving "scopulorum" and "quinquefoliata." In addition, even though an attempt was made at the outset of this study to seek out and include wind-pollinated progenies of trees that were phenotypically intermediate between taxa, at most hybrids and hybrid derivatives may be segregating in only three of the families.
To be sure, results purporting a lack of rampant hybridization and introgression seem incongruous with the reproductive compatibility that exists among taxa (Conkle and Critchfield, 1988
) that co-occur (Table 1). These taxa, however, tend to be isolated both ecologically and phenologically such that intertaxa hybridization may be inhibited. Despite occurring in the same mountain ranges, the taxa are largely parapatric across their altitudinal distributions. In those areas where distributions overlap, some phenologic isolation occurs as a result of intertaxa differences in the patterns of shoot elongation on which strobili maturation is dependent (see Fig. 5 of Rehfeldt, 1993
). Finally, the fact that the taxa have differing ecological requirements further suggests that selection may be eliminating hybrid seedlings before becoming established. Regardless, the present results support the conclusion that hybridization among the Ponderosae is infrequent and inconsequential to the genetic structures that are in evidence today. Molecular markers would be ideal for verifying these results.
Although the present interpretations account for the many and confusing taxonomic descriptions that exist for the Ponderosae of the Southwest, gaps exist in our understanding. Particularly obscure are the relationships between phenotypically similar taxa in the United States and Mexico. It seems opportune, therefore, for a taxonomic re-examination of the Ponderosae. Of particular interest would be a comparison of interrelationships derived from the present studies of experimental plant systematics with those of molecular phylogenetics.
Comparative genetic structures
When arranged into the three primary taxonomic groups defined by the multivariate analyses ("engelmannii," "quinquefoliata," and the combination of "scopulorum" and "X"), Ponderosae taxa inhabiting the mountain islands of the Southwest display very different genetic structures. All taxa exhibit high levels of genetic variability within populations that seem to be similar in magnitude to those of the widespread P. ponderosa var. ponderosa to the north. Differences become apparent, however, in the distribution of genetic variability among populations. In "engelmannii," genetic variation among populations is abundant, clines in association with geographic gradients are weak, and much of the interpopulation variance appears to be random. In "quinquefoliata," genetic variation among populations is depauperate, clines in association with elevation cannot be detected, and latitudinal clines may exist but are weak. In "scopulorum" and "X," genetic variation is abundant among populations, and clines in association with geographic gradients are readily detected and are quite similar to those of P. p. var. scopulorum.
Differences in genetic structure of these taxa have developed despite similarities in their distributions. The taxa currently occur and may have occurred for several thousand years as disjunct populations isolated on mountain islands. Nevertheless, all exhibit levels of genetic variability within populations similar to the widespread P. ponderosa. Quite obviously, the isolated, disjunct, and peripheral conditions under which these taxa exist are neither determinants of, nor, therefore, indicators of genetic structure. For the Ponderosae, either founder events were not unique enough, habitation not long enough, or effective sizes not small enough (Nei, Maruyama, and Chakraborty, 1975
) for anomalous genetic structures to have developed.
Rather than reflecting historical events, the genetic structure of these taxa seems more related to contemporary ecological distributions. Members of the "engelmannii" group tend to occur in pockets of favorable microclimates in valley locations; "quinquefoliata" tends to be broadly distributed across mid-slopes on the north and east aspects; and "scopulorum" and "taxon X" occupy the highest elevations. Interpopulation gene flow in the broadly dispersed "quinquefoliata," for instance, may be sufficient to prevent genetic differentiation, an effect accentuated perhaps by asymmetric gene flow from the middle elevations toward the extremes (Garcia-Ramos and Kirkpatrick, 1997
; Kirkpatrick and Barton, 1997
). In "engelmannii," gene flow among the scattered populations may be limited to the point that stochastic factors such as localized founder effects or genetic drift produce nonsystematic patterns of variation among populations inhabiting the same mountain range. And finally, the structure exhibited jointly by "scopulorum" and "X" merely seems to be a continuation of the clinal variation (Rehfeldt, 1993
) that typifies P. ponderosa var. scopulorum.
Evolutionary implications
Reduced gene flow, directional selection, and small population sizes are factors that may be operating in isolated, disjunct, and peripherally distributed populations to produce anomalous gene frequencies. As a result, means and variances in geographical isolates may be much different than for broadly dispersed populations.
Within the plant kingdom, tests sometimes support (Linhart and Premoli, 1994
) but other times reject (Wilson et al., 1991
; Lewis and Crawford, 1995
) the low marginal variance theory. For forest trees, however, studies dealing with either allozymes or quantitative traits approach unanimity: besides the Ponderosae of the Southwest, in Pinus contorta Dougl. (Cwynar and MacDonald, 1987
), Pinus edulis Engelm. (Betancourt et al., 1991
), Cryptomeria japonica D. Don (Tsumura and Ohba, 1993
), Picea abies Karst. (Tigerstedt, 1973
), Juniperus ashei Buch. (Adams, 1975
), Pinus banksiana Lamb. (Gauthier, Simon, and Bergeron, 1992
), Quercus ilex L. (Michaud et al., 1995
), Pinus nigra Arnold (Kaya and Temerit, 1994
), Eucalyptus urophylla S. T. Blake (House and Bell, 1994
), Pinus washoensis H. Mason & Stockwell (Niebling and Conkle, 1990
), Cupressus spp. (Rehfeldt, 1997
), Pinus strobus L. (Rajora et al., 1998
), and Great Basin conifers in general (Hamrick, Schnabel, and Wells, 1994
), genetic variability remains high in disjunct populations, although in Pinus contorta variability at the margins may be reduced slightly (Yeh and Layton, 1979
). To be sure, if unusual circumstances allow effective sizes to become too small, inbreeding will restrict intrapopulation variances, as apparently happened in Picea chihuahuana Mart. (Ledig et al., 1997
), Cupressus stephensonii Wolf (Rehfeldt, 1997
), and in northern outliers of Picea glauca (Moench) Voss (Tremblay and Simon, 1989
).
The largest documented adverse effects of isolation on genetic variances in a forest tree species, Quercus ilex, involved (1) variation in allele frequencies among populations, and (2) irregularities in the presence or absence of rare alleles (Michaud et al., 1995
). As shown for P. contorta (Cwynar and MacDonald, 1987
), moreover, even though founder events may temporarily result in low allelic diversity in geographic isolates, such effects are ameliorated in time. Little evidence, therefore, can be summoned to show that disjunct, peripheral, or isolated occurrences lead directly toward unusual genetic structures.
Although the effects of isolation on population means has received less attention, the present results, along with those for Larix laricina (Du Roi) K. Koch (Rehfeldt, 1970
), demonstrate that peripherally disjunct populations do not necessarily represent ecotypes specifically adapted to an environment commonly thought to be extraordinarily rigorous at the margin. Although peripheral, such populations frequently occupy environments similar to those occupied by the species within the general distribution (Boyko, 1947
), and, as a result, peripheral populations tend to contain genotypes exhibiting levels of performance typical for populations near the terminus of a cline. This seems true even for populations of Larix laricina (Rehfeldt, 1970
), Juniperus ashei (Adams, 1975
), and the Ponderosae in the Southwest, which have existed in isolated, peripherally disjunct populations for several thousand years.
For forest trees, empirical evidence suggests that unusual means or variances do not necessarily develop in populations that are isolated, disjunct, or peripheral. For trees, therefore, generations are evidently too long and gene flow insufficiently disrupted for anomalous genetic structures to develop and be maintained.
Practical implications
The present results are pertinent to conservation biology, a field dealing with the perpetuation of rare, threatened, or endangered species. From the viewpoint of their genetic structure, populations of the Ponderosae inhabiting the mountain islands of the Southwest certainly are not threatened. Despite distributions that are peripheral, isolated, and disjunct, these taxa display abundant genetic variability; no evidence suggests that isolation in itself has had an adverse effect on their genetic structure. Gene conservation programs, therefore, are not an issue on which perpetuation of these populations is dependent.
Perpetuation, therefore, will depend first on the maintenance of suitable habitats and second on the availability of genotypes appropriate for those habitats. Because clines in "quinquefoliata" and "engelmannii" are weak at best, maintaining appropriate genotypes is applicable primarily to "scopulorum" and "X." The clines described for this taxon imply that populations are physiologically attuned to different portions of the environmental gradient. This means that growth and survival either after disturbance or during periods of environmental change are dependent on the appropriate genotypes being available for colonization. Habitation then will depend on the migration of genotypes to the appropriate environments, a process that can be facilitated greatly with reforestation programs.
Conservation biology also deals with ameliorating adverse effects of human encroachment on the fragmentation of native ecosystems. Because isolated and disjunct populations can be viewed as fragments of a general distribution, the present results thus suggest that fragmentation in itself does not necessarily lead to adverse genetic consequences. As with Acer saccharum Marsh., fragmentation may alter gene flow (Young and Merriam, 1994
) but does not necessarily affect genetic diversity (Fore et al., 1992
). The most dire effects of fragmentation, therefore, may also pertain to migration during periods when environmental gradients are shifting. At such times, maintaining populations in those environments optimal for their growth and survival quite likely will require human assistance.
|
FOOTNOTES |
|---|
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Avise, J. C. 1996 Introduction: the scope of conservation genetics. In J. C. Avise and J. L. Hamrick [eds.], Conservation genetics: case histories from nature, 1–9. Chapman and Hall, New York, NY.
Bermejo-Velazquez, B. 1993 Genetic diversity and the mating system in Pinus engelmannii Carr. Ph.D. dissertation, University of Wisconsin, Madison, WI.
Betancourt, J., W. S. Schuster, J. B. Mitton, and R. S. Anderson. 1991 Fossil and genetic history of a pinyon pine (Pinus edulis) isolate. Ecology 72: 1685–1697.[CrossRef][ISI]
Boyko, H. 1947 On the role of plants as quantitative climate indicators and the geo-ecological law of distribution. Journal of Ecology 35: 138–157.
Carson, H. L. 1959 Genetic conditions which promote or retard the formation of species. Cold Spring Harbor Symposium Quantative Biology 24: 87–105.
Conkle, M. T., and W. B. Critchfield. 1988 Genetic variation and hybridization in ponderosa pine. In D. M. Baumgartner and J. E. Lotan [eds.], Ponderosa pine: the species and its management, 27–43. Washington State University, Pullman, WA.
Cwynar, L. C., and G. M. MacDonald. 1987 Geographical variation of lodgepole pine in relation to population history. American Naturalist 129: 463–469.[CrossRef][ISI]
Farjon, A., and B. T. Styles. 1997 Flora neotropical. Monograph 75 (Pinus). New York Botanical Garden, Bronx, NY.
Fore, S. A., R. J. Hickey, J. L. Vankat, S. I. Guttman, and R. L. Schaefer. 1992 Genetic structure after forest fragmentation: a landscape ecology perspective on Acer saccharum. Canadian Journal of Botany 70: 1659–1668.
Garcia-Ramos, G., and M. Kirkpatrick. 1997 Genetic models of adaptation and gene flow in peripheral populations. Evolution 51: 21–28.[CrossRef][ISI]
Gauthier, S., J. P. Simon, and Y. Bergeron. 1992 Genetic structure and variability in jack pine populations: effects of insularity. Canadian Journal of Forest Research 22: 1958–1965.
Hamrick, J. L., H. M. Blanton, and K. J. Hamrick. 1989 Genetic structure of geographically marginal populations of ponderosa pine. American Journal of Botany 76: 1559–1568.[CrossRef][ISI]
———, A. F. Schnabel, and P. V. Wells. 1994 Distribution of genetic diversity among populations of Great Basin conifers. In K. T. Harper, L. L. St. Clair, K. H. Thorne, and W. M. Hess [eds.], Natural history of the Colorado Plateau and Great Basin, 147–161. University of Colorado Press, Boulder, CO.
House, A. P. N., and J. C. Bell. 1994 Isozyme variation and mating system in Eucalyptus urophylla S. T. Blake. Silvae Genetica 43: 167–176.[ISI]
Karron, J. D. 1991 Patterns of genetic variation and breeding systems in rare plant species. In D. A. Falk and K. E. Holsinger [eds.], Genetics and conservation of rare plants, 87–98. Oxford University Press, New York, NY.
Kaya, Z., and A. Temerit. 1994 Genetic structure of marginally located Pinus nigra var. pallasiana populations in central Turkey. Silvae Genetica 43: 272–277.[ISI]
Kirkpatrick, M., and N. H. Barton. 1997 Evolution of a species' range. American Naturalist 150: 1–23.[CrossRef][ISI]
Ledig, F. T., V. Jacob-Cervantes, P. D. Hodgskiss, and T. Eguiluz-Piedra. 1997 >Recent evolution and divergence among populations of a rare Mexican endemic, Chihuahua spruce, following Holocene climatic warming. Evolution 51: 1815–1827.[CrossRef][ISI]
Lewis, P. O., and D. J. Crawford. 1995 Pleistocene refugium endemics exhibit greater allozyme diversity than widespread congeners in the genus Polygonella (Polygonaceae). American Journal of Botany 82: 141–149.
Linhart, Y. B., and A. C. Premoli. 1994 Genetic variation in central and disjunct populations of Lilium parryi. Canadian Journal of Botany 72: 79–85.
Little, E. L., Jr., and W. B. Critchfield. 1969 Subdivisions of the genus Pinus (pines). United States Department of Agriculture Miscellaneous Publication 1144.
Lynch, M. 1996 A quantitative genetic perspective on conservation issues. In J. C. Avise and J. L. Hamrick [eds.], Conservation genetics: case histories from nature, 471–501. Chapman and Hall, New York, NY.
Martin, P. S., and P. J. Mehringer. 1965 Pleistocene pollen analysis and biogeography of the Southwest. In H. E. Wright and D. G. Frey [ eds], The Quaternary of the United States, 433–451. Princeton University Press, Princeton, NJ.
Mayr, E. 1970 Population, species, and evolution. Belknap Press, Cambridge, MA.
McLaughlin, S. P. 1992 Are floristic areas hierarchically arranged? Journal of Biogeography 19: 21–32.[CrossRef][ISI]
Michaud, H., L. Toumi, R. Lumaret, T. X. Li, F. Romane, and F. Di gusto. 1995 Effect of geographical discontinuity on genetic variation in Quercus ilex L. (holm oak): evidence from enzyme polymorphism. Heredity 74: 590–606.[ISI]
Mitton, J. B. 1995 Genetics and the physiological ecology of conifers. In W. K. Smith and T. M. Hinckley [eds.], Ecophysiology of coniferous forests, 1–36. Academic Press, New York, NY.
———, Y. B. Linhart, J. L. Hamrick, and J. S. Beckman. 1977 Observations on the genetic structure and mating system of ponderosa pine in the Colorado Front Range. Theoretical and Applied Genetics 51: 5–13.
Nei, M., T. Maruyama, and R. Chakraborty. 1975 The bottleneck effect and genetic variability in populations. Evolution 29: 196–218.
Niebling, C. R., and M. T. Conkle. 1990 Diversity of Washoe pine and comparisons with allozymes of ponderosa pine races. Canadian Journal of Forest Research 20: 298–308.
Peloquin, R. L. 1984 The identification of three-species hybrids in the ponderosa pine complex. Southwestern Naturalist 29: 115–122.[CrossRef][ISI]
Perry, J.P., Jr. 1991 The pines of Mexico and central America. Timber Press, Portland, OR.
Rajora, P., L. DeVerno, A. Mosseler, and D. J. Innes. 1998 Genetic diversity and population structure of disjunct Newfoundland and central Ontario populations of eastern white pine (Pinus strobus). Canadian Journal of Botany 76: 500–508.[CrossRef]
Rehfeldt, G. E. 1970 Genecology of Larix laricina (Du Roi) K. Koch in Wisconsin. Silvae Genetica 19: 9–15.
———. 1979 Ecotypic differentiation in populations of Pinus monticola in north Idaho—myth or reality? American Naturalist 114: 627–636.[CrossRef][ISI]
———. 1989 Ecological adaptations in Douglas-fir (Pseudotsuga menziesii var. glauca): a synthesis. Silvae Genetica 37: 131–135.[ISI]
———. 1990 Genetic differentiation among populations of Pinus ponderosa from the upper Colorado River Basin. Botanical Gazette 151: 125–137.[CrossRef]
———. 1991 Models of genetic variation for Pinus ponderosa in the Inland Northwest (U.S.A.). Canadian Journal of Forest Research 21: 1491–1500.<
