Group size estimates were updated throughout the encounter and the largest estimate was used as the provisional group size. Photo-identification was used after the encounter to confirm identified individuals or reveal individuals not identified during the encounters. The final group size for an encounter was a product of in-water

identification and photo-identification afterwards. The end of an encounter occurred when the dolphins moved away or were unable to be observed reliably (e.g., if they were traveling or swimming against a strong current). The researchers moved on to search for another group. Sometimes dolphins from a previous encounter would be sighted again shortly afterwards with other individuals. If the majority of the animals were the same, the researchers resumed the previous encounter. Only if the composition of the group changed by 50% or more, were they ABT 199 considered a different group and a new encounter began. Only groups where more than 50% of individuals were identified were included in analyses. If an individual was resighted twice or more in the same day, they were included in analysis only if there was at least a 50% difference in group composition. Calves were not included in analyses as their associations are dependent on their mothers’ associations. Coefficients of association see more (CoAs) were calculated using the half-weight

index (Cairns and Schwager

1987) with the software program SOCPROG 2.3 (Whitehead 2009). CoAs were calculated for pooled years 1991–1993, 1994–1996, 1997–1999, and 2000–2002. These pooled years permit O-methylated flavonoid enough individuals to be included, while giving representative results. The last year, 2002, was chosen as a cut-off point in the long-term data set because the area was impacted by hurricanes in 2004, after which about 30% of the population was lost (Elliser and Herzing, in press). In the same study area, significant changes in community/social structure were documented in the sympatric bottlenose dolphin population following similar losses of individuals and influx of new immigrants (Elliser and Herzing 2011). CoAs were calculated for pairs of noncalf individuals of known sex using two sighting criteria: (1) those sighted at least six times per pooled period or (2) at least 10 times per pooled period. Similar results were found for both sighting criteria (Elliser and Herzing 2012). The results did not differ using the higher sighting criterion, so we used the six sightings criterion because it allowed for the inclusion of more individuals. In a concurrent study (Elliser and Herzing 2012), SOCPROG was used to conduct permutations to test the null hypothesis of random associations and no preferred/avoided companions (Christal and Whitehead 2001, Whitehead 2009).