Electronic Supplementary Material A novel mammalian social structure in Indo-pacific bottlenose dolphins (Tursiops sp.): complex male-male alliances in an open social network METHODS To control for artificial comparative reduction of the range estimation with increasing sample size when the FKDE method of range estimation was used, a random subset of 60 location points was set as the highest number of points included for range estimation for each individual. This cut-off was chosen because when more than 60 location points were included in the analysis, the variance in sub-sampled range sizes of each individual began to show a substantial decrease (although in our sub-sampling trials, where 20 trial range calculations were performed for each individual, with each calculation based on a different number of sighting points randomly selected from that individual's sighting record, FKDE range size decreased indefinitely with increasing sample size). In the case of 2nd order alliances the sub-sampling trials were performed for each alliance using the same procedure as for individuals, and they yielded a cut-off point of 100 sightings, which was therefore the highest number of sighting points for each alliance included in the FKDE range estimation. For those alliances with a higher number of sightings, we randomly sub-sampled 100 points from their sighting records for FKDE analysis. In the case of MCP range estimations, we used all available sighting points (Table S1).
Table S1 Number of alliance sighting points included in the analysis for range calculation using the MCP and FKDE methods, respectively Alliance Sighting points - MCP Sighting points - FKDE BB 84 84 BL 153 100 CB 108 100 XF 150 100 FCB 115 100 GG 73 73 PB 108 100 HH 82 82 HC 149 100 KS 296 100 PHG 26 26 PD 160 100 RHP 118 100 RR 224 100 SK 42 42 SJ 88 88 WC 115 100 The relationship between alliance size and range size was examined using Spearman s rank correlation. This calculation was done for individual males as well as for alliance ranges. The analysis for 2nd-order alliances was then repeated excluding the five lone trios, who were not members of 2nd-order alliances. Trios and pairs without 2nd-order alliance partners were usually older animals whose other 2nd-order alliance partners had previously disappeared [1]. All statistical analyses were performed using SPSS version 11. For testing the normality assumption in the datasets, the Shapiro-Wilk test [2] was used. All tests were two-tailed. RESULTS (a) The size of individual male and alliance ranges
The correlation between MCP range and sample size for individual males was very weak (rho = 0.170, p = 0.063, n = 120). However, the models of social organization tested in this paper are based not on individual but alliance ranges, and here also, the relationship between MCP range size and sample size was weak and not significant (rho = 0.266, p = 0.302, N =17). Average MCP range size for individual males in Shark Bay was 101.7 km 2 (Table S2 ; Table 1 & Figure 2 in the main text). The average 90% volume contour (full) range size calculated using the FKDE method was 76.2 km 2 and the average FKDE 50% volume contour (core) was 22.3 km 2. For the five lone trios and twelve 2nd-order alliances in Shark Bay, the average range sizes were 144.92 km 2 (MCP), 92.3 km 2 (90% FKDE) and 25.6 km 2 (50% FKDE). Table S2 Average range sizes (in km 2 ) for individual males in Shark Bay Mean Standard N Minimum Maximum error MCP 101.71 3.70 120 27.50 216.84 90% FKDE 50% FKDE 76.18 1.84 120 35.97 119.32 22.27 0.58 120 9.68 39.63 Range size for individual males is significantly positively correlated with the size of the respective 2nd-order alliance of which they are members. This holds true for both MCP and FKDE methods of range estimation, and for both full and core ranges (MCP: rho=0.428, P=0.01, N=120; 90% FKDE: rho=0.452, P=0.01, N=120; 50% FKDE: rho=0.438, P=0.01, N=120). Size of alliance ranges (2nd-order and lone trios combined), estimated using MCP and FKDE (at 90% and 50%) methods, is also positively correlated
with the number of dolphins in the alliance (MCP: rho=0.593; P=0.012, N=17; 90% FKDE: rho=0.480, P=0.05, N=17; 50% FKDE: rho=0.565, P=0.018, N=17). When the lone trios are excluded from the analysis, this trend is still visible, but it is only statistically significant for the MCP method of estimation (rho=0.672; P=0.017, N=12); whereas for the FKDE estimation method the significance at the 0.05 level is lost (90% FKDE: rho=0.400, P=0.108, N=12; 50% FKDE: rho=0.400, P=0.102, N=12). The MSD model tested here includes data from six years, which could conceivably obscure mating season territoriality if there had been annual shifts in territories. However, data from a single year with the most data available shows clearly that the pattern of extensive mating season range overlap holds true within as well as across years (Figure S1).
Figure S1 - Alliance core ranges during the peak mating season of a single year (2004), calculated using FKDE method
(b) Male and female ranges Watson-Capps [3] examined the ranges of a much smaller number of adult females (21 versus our sample of 120 males) in a smaller study area (250 km 2 versus our 600km 2 area), restricting the sample size to 50 sightings per individual for both the MCP and FK methods. She did not report individual values, but reported a mean MCP range of 47.8 km 2 and 95% FK range of 52.9km 2 for 21 females (excluding 5 provisioned females from her total of 26). Male ranges vary along the peninsula (unpublished data) and tend to be larger north of the 250km 2 area examined by Watson-Capps [3]. Males (n = 66) ranging predominantly within the smaller 250km 2 area had an average MCP range of 95.8km 2 (range 27.5-180.2) and 90% FK range of 75.4 km2 (range 37.7-199.3). Conservatively, the average MCP range for males with a sample size of 50 or less (n = 15, range 24-50) was 70.1km 2 (range 27.5-98.7) with only 2 values below the female mean of 47.8km 2. The average 90% FK for these same 15 males was 65.9 km 2 (range 37.7-97.0) with only 2 values below the female mean 95% FK value of 52.9 km 2. Ranges for both males and females are highly variable, but within the 250km 2 area examined by Watson-Capps [3], male ranges are, on average, larger than female ranges. DISCUSSION A low cost of locomotion is only part of an explanation for why we find complex alliances in the Shark bay dolphins. Alliance formation is more complex in Shark Bay than in some other Tursiops populations (e.g. Sarasota, Florida) and this may owe, in part, to a higher density of dolphins in Shark Bay [4]. Ranging patterns and population density will combine to determine the rate at which males encounter each other in competitive contexts. Alliance formation is favoured when the rate of such encounters is higher [5].
A population of bottlenose dolphins (Tursiops truncatus) in Tampa Bay, Florida, next to the Sarasota dolphins, has been compared to chimpanzee communities based on a community definition of shared patterns of residency and associations [6]. In our view the key for any chimpanzee-dolphin community comparison is not whether there is a statistically significant degree of association that can be used to define dolphin 'clusters', but whether any females or males from different dolphin 'communities' associate at all and whether 'community' ranges are defended by members of either sex. It would be very surprising to find male or even female chimpanzees wandering over to neighbouring communities to socialize, given the nature of the violent male chimpanzee range defence (see [7]). 1 Connor. R.C., Watson-Capps, J., Sherwin, W.S. & Krützen, M. 2010 New levels of complexity in the male alliance networks of Indian Ocean bottlenose dolphins (Tursiops sp.). Bioletters doi: 10.1098/rsbl.2010.0852. 2 Shapiro, S. S., &Wilk, M. B. 1965. An analysis of variance test for normality (complete samples). Biometrika 52 (3-4), 591 611. 3 Watson-Capps, J.J. 2005. Female mating behavior in the context of sexual coercion and female ranging behavior of bottlenose dolphins (Tursiops sp.) in Shark Bay, Western Australia. Ph. D. Dissertation, Department of Biology, Georgetown University. 4 Connor, R.C., Wells, R., Mann, J. & Read, A. 2000 The bottlenose dolphin: social relationships in a fission-fusion society. In Cetacean societies: field studies of whales and dolphins. (eds. J. Mann, R. Connor, P. Tyack, and H. Whitehead), pp. 91-126. Chicago: University of Chicago Press. 5 Connor, R.C. & Whitehead, H. 2005 Alliances II: Rates of encounter during resource utilization: A general model of intrasexual alliance formation infission-fusion societies. Anim. Behav. 69, 127-132. 6 Urian, K.W., Hofmann, Wells, R.S. & Read. A.J. 2009. Fine-scale population structure of bottlenose dolphins (Tursiops truncatus) in Tampa Bay, Florida. Mar. Mamm. Sci. 25, 619-638. 7 Goodall, J. 1986. The Chimpanzees of Gombe: Patterns of Behavior. Cambridge, MA: Harvard University Press.