CHANGES IN THE DEMOGRAPHIC TRENDS OF PINK RIVER DOLPHINS (INIA) AT THE MICROGEOGRAPHICAL LEVEL IN PERUVIAN AND BOLIVIAN RIVERS AND WITHIN THE UPPER AMAZON:MICROSATELLITES AND MTDNA ANALYSES AND INSIGHTS INTO INIA’S ORIGIN - Biology, Evolution and Conservation of River Dolphins Within South America and Asia - page 170

significance level of the binomial distribution was measured using the program Statistics
Sample written by Michael H. Kelly.
The g test focuses on the following facts. This interlocus test shows that for stable
populations the allele size variance is highly variable among loci, whereas in expansion
populations this variance is usually lower. Therefore, allele sizes with sufficiently low
variances are taken as evidence for population expansion. The test used for statistical

Var [



Vj ]

g 

differences is that proposed by Reich & Goldstein (1998):

where Var

4 1

V 2 V

3 6

(V j ) is the observed variance of the allele length variances across the markers employed, and
V is the average variance across loci. A low value of g is taken as a sign of population
expansion. Significance levels for the interlocus test are found on Table 1 of Reich et al.,
(1999), which shows the fifth-percentile cutoffs for the interlocus test. For the present case, g
values lower than 0.17-0.23 will indicate a population expansion. Luikart et al. (1998) noted
that this last test is probably the most powerful for this task.

mtDNA

The mitochondrial gene diversity in each one of the populations studied was measured by
means of the nucleotide diversity () and the  per gene (which is equivalent to k, the average
number of nucleotide differences).
I used two demographic methods for the mtDNA analysis. The first one corresponds to
the mismatch distribution (pairwise sequence differences) calculated following the method of
Rogers & Harpending (1992) and Rogers et al. (1996). The empirical observed distribution
was compared to the two theoretical curves, the first one assuming a constant population size
and the second one assuming a population expansion. The raggedness rg statistic (Harpending
et al., 1993; Harpending, 1994), the Mean Absolute Error (MAE) between the observed and
the theoretical mismatch distribution (Rogers et al., 1996) and the R 2 statistic (Ramos-Onsins
& Rozas, 2002) were used to determine the similarities between the observed and the
theoretical curves. Finally, the second method uses the Fu and Li D and F tests (Fu & Li,
1993), the Fu F S statistic (Fu, 1997) and the Tajima D test (Tajima, 1989a). All these tests
allow for the determination of possible population size changes in the pink river dolphin
ensembles aforementioned (Simonsen et al., 1995; Ramos-Onsins & Rozas, 2002).
Finally, to estimate the divergence times among the mitochondrial haplotypes found in
the diverse pink river dolphins studied, the median joining network (MJ; Bandelt et al., 1999)
was applied by means of the software Network 4.2.0.1 (Fluxus Technology Ltd). Once the
haplotype network was constructed, the  statistic (Morral et al., 1994) was estimated. This
statistic measures the age of an ancestral node in mutational units. This value is transformed
into years by multiplication with the mutation rate. In this case, I took a mutational rate of one
mutation each 20,180 years, because this was the amount found for human mtDNA for the
stretch from 16090 to 16365 np Additionally, the standard deviation () was calculated
(Saillard et al., 2000). The  statistic is unbiased and highly independent of past demography
events. These events could have influenced the shape of a given evolutionary tree, but this
only influences the error of the time estimated and does not increase or decrease this time.