Subalpine fir populations were sampled at 3 elevations on southwest-facing slopes on Klahhane Ridge, Blue Mountain: the upper (treeline), middle, and lower extent of its elevation range. At least 50 trees were sampled from 20 m-wide band transects at each site. In addition, 30 trees were sampled from treeline at Mount Dana, the westernmost extent of subalpine fir in the Olympic Mountains. Only trees with dominant or codominant crown classification (Spurr and Barnes 1980) were sampled to reduce variation in growth trends related to competition. Trees at middle and low elevations on Blue Mountain were sampled from 2 sites (separated by approximately 200 m at each elevation) in order to minimize differences in slope and aspect thereafter Blue Mountain low-1 and low-2, middle-1 and middle-2). Trees with major stem or crown deformities were avoided.
Two cores were extracted with an increment borer at 1.1-1.4 m height on the cross-slope sides of each tree. Cores were mounted in wooden blocks and sanded until individual tracheids were visible. All cores were crossdated (visually -- Swetnam et al. 1985, and verified with COFECHA -- Holmes 1983), and one core from each tree (the core with the fewest instances of reaction wood) was measured to the nearest 0.01 mm with an incremental measuring machine (Robinson and Evans 1980). Basal area growth increments (BAI) were created from DBH and ring width measurements for each tree.
A 15-20 cm section of shade foliage was clipped from the lower branches on the north-facing side of each sample tree. The cuttings were placed on ice and kept cool to suspend biological activity until processing. All foliage samples were processed within four weeks of collection. While processing leaf tissue within a week (the majority of these samples were processed within a week) is recommended to maintain the full compliment of available enzymes (Cheliak and Pitel 1984), we observed little reduction in enzyme quality in samples that were stored for 4 weeks prior to processing. Needles were initially ground with a mortar and pestle in liquid nitrogen and a grinding buffer (after Mitton et al. 1979). Preliminary electrophoretic trials of enzymes ground in this buffer showed poor resolution of banding patterns, and 4 other grinding buffers were also tried in an effort to maximize the number and resolution of enzymes; buffers included a simple seed buffer, simple seed buffer with polyvinyl pyrolidone added, Melody's buffer (Diane Delany USDA Forest Service Albany, CA, personal communication), Soltis's buffer (after Soltis 1983), and a modification of Melody's buffer (with bovine albumin and sucrose added. A comparison of banding patterns resolved from these 5 grinding buffers showed that the modified Melody's buffer provided the largest number of enzymes resolved with relatively clear banding patterns. In addition, the modified Melody's buffer had the advantage of being less noxious than other buffers because it does not contain 2-mercaptoethanol. The solutions obtained from grinding leaf tissue were separated, and solutions from individuals were stored in microtiter trays in an ultralow temperature freezer (-70° C). Samples were subsequently stored on dry ice and flown to the USDA Forest Service Institute of Forest Genetics (IFG) lab in Albany, California, where allozyme analysis was conducted.
Leaf tissue solutions were subjected to electrophoretic separation on starch gels using techniques of Conkle et al. (1982) and Millar (1985), and gel systems of Strauss and Conkle (1986) and Wendel and Weeden (1989). An initial screening was performed to determine enzyme activity for 33 different enzymes on 6 different gel systems. Fifteen loci from 12 enzyme systems showed strong and repeatable resolution: aspartate aminotransferase (AAT-1 and AAT-2, EC 2.6. 1.1), fructose-biphosphate aldolase (ALD, EC 4.1.2.13), fructose-l ,6-diphosphate (FDP, EC 3.1.3.11), â-D-galactoside galactohydrolase (GAL, EC 3.2.1.23), isocitrate dehydrogenase (IDH, EC 1.1.1.42), malate dehydrogenase (MDH-1 and MDH-2, EC 1.1.1.37), peptidase (PEP, EC 3.4.11.1), peroxidase (PER, EC 1.11.1.7), phosphoglucoisomerase (PGI-1 and PGI-2, EC 5.3.1.9) phosphoglucomutase (PGM, EC 5.4.2.2), 6-phosphogluconate dehydrogenase (SIX, EC 1.1.1.44), and UDP-glucose pyrphosphorylase (UGP, EC 2.7.7.9). These loci were scored on 5 different gel systems. Diploid genotypes were interpreted from segregation patterns and comparison of allozyme phenotypes of the same enzymes for related species (Diane Delany USDA Forest Service Albany, CA, personal communication).
Allele frequencies, mean number of alleles per locus, percentage of polymorphic loci (alleles were classified as polymorphic if more than 1 allele was observed), and mean heterozygosity levels (computed as both direct count and Hardy- Weinberg expectations) were calculated for each site. Deviations from Hardy- Weinberg equilibrium were assessed with Chi-square analysis; allele frequencies were pooled when frequencies of some classes were low, and Yates' correction for continuity was applied. Subdivison of genetic structure among sites was assessed with Chi-square tests for differences of allele frequencies, F-statistics (Wright 1965, Nei 1972, 1973, 1977), and Nei's unbiased genetic distance (calculated with exclusion of 6 monomorphic loci, Nei 1978). Chi-square analyses of allele frequencies were compared both by elevation for mountains and between different mountains for all elevations. Unweighted pair-group method (UPGMA) cluster analysis (Sneath and Sokal 1973) was used to visualize differences in genetic distance among groups. All analyses were performed with the program BIOSYS-1 (Swofford and Selander f989). The program GeneStat-PC (Lewis and Whitkus 1993) was used to calculate Hamrick and Godt's (1990) Hes (the average genetic heterozygosity over all loci) in order to compare heterozygosity levels in subalpine fir with other conifer species.
Comparisons
A simple arithmetic mean of basal area increment (BAI) growth for the time period 1950-1990 (1970-1990 for Blue Mountain high elevation) was compared with the number of heterozygous loci for each individual to determine the effect of genetic variability on growth. The time period 1950-1990 was selected for comparison because BAI calculated from ring width measurements has substantially lower growth land variance) during the first 20 years that BAI values are calculated (visual inspection of data). Large errors in BAI calculations occur early in BAI series when there is a large difference between the actual ring width measurements (total of all ring width measurements to the center of the tree) and the distance to the center of the tree estimated from DBH (used in calculated BAI). Because some trees on most sites date back to the early 1900's, the time period 1950-1990 maximizes sample size. The time period 1970-1990 was compared for Blue Mountain high elevation land subsequently Klahhane Ridge high elevation), because many trees dated back to only the 1930's on this site. Mean BAI from the 2 time periods (1950-1990 and 1970-1990) are highly correlated (r² = .95) for all individuals on high elevation sites.
A comparison of BAI growth and different genotypes for polymorphic loci were also compared among individuals for each site. Preliminary analysis revealed similar patterns of both multi-locus and allozyme-specific genetic relationships with growth among elevations. Relationships were therefore summarized by elevation; high elevation sites included only Blue Mountain and Klahhane Ridge (Mount Dana was excluded because of its relatively lower genetic variability). Kruskal-Wallis and Mann-Whitney U-test statistics (Zar 1984) were used to compare growth differences among groups and between pairs of groups of genetically different individuals. These tests were chosen over comparable parametric tests (e.g., Tukey's multiple comparison, one-way ANOVA, or t-statistic for unequal variances) because groups often differ in sample size (most heterozygotes are rare), and the variance in growth response is partially dependent on sample size.
Relationships were explored between tree age, mean basal area growth, and genotypic frequencies for each locus to determine if any observed differences in growth-genetic relationships were related to differences in allele frequencies among age groups (even though growth rate is not associated with age). Trees were first aggregated by elevation, then divided into 2 age groups at each elevation to include equal numbers of individuals in each group. Chi-square analyses of allele frequencies of individuals from different age groups were compared at each elevation.