Northern Prairie Wildlife Research Center
We isolated white blood cells from whole blood in the laboratory and then froze them until needed. DNA was extracted from white cells by standard methods (Sambrook et al., 1989). The Denali and SNF populations had been previously analyzed for variability in 10 microsatellite loci and found to be similar in levels of heterozygosity, allelic diversity, and in the equability of allele frequencies (Roy et al., 1994). Consequently, estimates for various categories of relatedness should be similar in both populations.
In each population, we defined three social groupings based on behavioral criteria: mother-offspring, siblings, and mated pairs. A mated pair was defined as a radio-tagged male and female older than 2 years that travelled together for at least a few weeks. Most mated wolves were also breeding pairs as they remained alone together through the breeding season and produced pups. In larger packs, even when other adults were present in the pack, we identified mated pairs by behavioral attributes such as jointly leading the pack when travelling, close association with one another, and joint attendance at dens.
We defined individuals as mother and offspring if young were observed with the female of the mated pair defined above and if no other adult females were present in the pack. We defined siblings as young born together in a pack with only a single known pair of mated adults. However, the apparent breeding female could conceivably have been a recent replacement of the actual mother of pack offspring, and the putative father could have been incorrectly assigned due to the possibility of extrapair copulations. Extrapair copulations have been documented with molecular genetic techniques in a wide variety of vertebrates, even in species thought to be monogamous based on behavioral observations (e.g. Burke and Bruford, 1987; Creel and Waser, 1994; Gottelli et al., 1994). Consequently, we determined if either of the mated pair could be excluded as a parent by documenting the presence of unique alleles in their putative offspring (Bruford et al., 1992). We calculated the exclusion probability per locus (PEi) following Chakraborty et al. (1988):
with δ and β being the allele frequencies found in an offspring. Combining the probabilities for all loci (Chakraborty et al., 1988) as follows:
yielded the proportion of randomly chosen adults in the population that could be expected to be genetically excluded as the father or mother of a given offspring.
To determine the correspondence of molecular genetic estimates of kinship with known relatedness, we obtained blood samples as above from two captive wolf populations with documented genealogies, the Julian Pack and the Forest Lake colony. The Julian pack is located in Julian, California, USA, and was founded with two wild-caught individuals thought to be from different locations in central Alaska. We obtained samples from the single mated pair and their nine offspring of different years. The Forest Lake colony is located near Forest Lake, Minnesota, USA, and includes individuals from a large pedigree of wolves (Packard et al., 1983) with relationships ranging from inbred siblings to unrelated individuals. The 20 individuals we chose for analysis are a limited subset of the Forest Lake colony wolves, having relatedness (r) values ranging from 0 to 0.5 calculated from the pedigree (Falconer, 1983).
We used twenty GT(n) polymorphic microsatellite loci identified from a domestic-dog genomic library (Ostrander et al., 1993). Detection of microsatellite alleles from genomic DNA was achieved by end-labeling one primer by a standard 32γ-ATP (Amersham) and T4 polynucleotide kinase reaction (Sambrook et al., 1989) and performing 28 cycles of polymerase chain reaction amplification in a 25-ml reaction volume using 50 ng of target DNA, 2 mM MgCl2, and 0.8 U of Taq DNA polymerase (Promega). Reaction condition were denaturation at 94°C for 45 s, annealing at 50°C or 55°C for 45 s, and extension at 72°C for 60 s. We then mixed 3µl of each product were then mixed with 2 Ál of formamide loading dye and heated to 94°C for 5 min before being loaded onto a 6% sequencing gel containing 50% (w/v) urea. A M13 control region was run adjacent to the samples to provide an absolute-size marker for the microsatellite alleles. Gels were then autoradiographed overnight.
Because pedigree data were not known for wild-caught wolves, we used the Queller and Goodnight (1989) index of relatedness (R) to estimate kinship. This index weights each allele inversely by its frequency in the population, so that rare alleles are given a relatively higher weight. If a sample adequately represents a population in a Hardy-Weinberg equilibrium, the index values obtained for parent and offspring or for full siblings should approach 0.5. Overall, the index values vary between -1 and 1. The Queller and Goodnight index of relatedness was calculated for any two individuals (dyads) as follows:
The equation is summed over all loci and alleles. P* is the population frequency of each allele excluding the compared individuals. Px and Py are the frequencies of each allele in the compared individuals, respectively (i.e., 0.5 or 1 depending on whether the individual is a heterozygote or homozygote). This index is not symmetrical, so reciprocal comparisons are not expected to equal each other (Py/Px). To accommodate for this discrepancy, we calculated the denominator values and numerator values for each combination (Py/Px, and Px/Py), and summed them prior to the division. This procedure yields an average estimate of relatedness between the two individuals compared. Standard deviations for the relatedness values were estimated by jack-knifing over all loci (Queller and Goodnight, 1989).
Because of technical limitations, not all individuals could be typed for all 20 microsatellite loci. Consequently, we estimated the number of loci needed to estimate relatedness adequately by rarefaction analysis (e.g., Lehman and Wayne, 1991). We selected a locus at random, calculated the Queller and Goodnight relatedness value, and then selected another locus without replacement and recalculated the relatedness based on these two loci. The sampling was repeated without replacement until all 20 loci were selected. We then expressed the difference between consecutive samplings as a function of the total number of loci drawn. We repeated this procedure 100 times and calculated mean difference values (Figure 1). Descriptive statistics are given as mean values ±1 SD.
|Figure 1. The decrease in the mean difference between consecutive relatedness estimates as a function of the number of microsatellite loci analyzed. The curve is described by the following equation: mean difference = 0.831 (number of loci) -1.41, r = .998. Error bars indicate 1 SD above or below the mean value.|