Riemer et at: Food habits of Eumetopias /ubatus off Oregon and northern California, 1986-2007 
371 
Table 1 
Total number of Steller sea lion (Eumetopias jubatus) scat collected by site, month, and year from haul-outs and rookeries in 
northern California and Oregon, 1986-2007. 
Breeding season month 
Nonbreeding season month 
Location 
Year 
May 
Jun 
Jul 
Aug 
Sep Oct Nov Dec Jan Feb Mar 
Apr 
Total 
Columbia River 1 
2004 
55 
38 
45 
51 
189 
2006 
48 
48 
2007 
20 
61 
3 
84 
Cascade Head 2 
2003 
13 11 
24 
Orford Reef 3 
1990 
41 
41 
2002 
15 
15 
Rogue Reef 7 
1986 
18 
18 
1987 
40 
40 
1988 
20 
20 
1990 
47 
47 
1993 
36 
36 
1994 
33 
33 
1995 
12 
12 
1996 
60 
60 
2001 
70 
48 
46 
164 
2002 
33 
37 
42 
78 
190 
2003 
12 
54 
57 
123 
2004 
33 
33 
2005 
2 
13 
20 
35 
2006 
25 
25 
St. George Reef 5 
1990 
4 
4 
1994 
37 
37 
2002 
35 
35 
2003 
29 
7 
36 
2004 
34 
34 
2006 
33 
33 
Total 
62 
321 
374 
232 
48 67 0 0 11 0 42 
259 
1416 
1 South Jetty (46.233 lat. N, 124.070 long. W). 
2 Sea Lion Cove (45.067 lat. N, 124.013 long. W). 
3 Long Brown Rock (42.791 lat. N, 124.605 long. W ). 
4 Primarily Pyramid Rock (42.444 lat. N, 124.469 long. W), but also surrounding sites including Needle Rock (42.448 lat. N, 124.483 long. W), 
Double Rock (42.449 lat. N, 124.490 long. W), and South Seal Rock (42.436 lat. N, 124.465 long. W). 
5 South Seal Rock (41.813 lat. N, 124.361 long. W). 
data (Legendre and Legendre, 1998). A Jaccard dis- 
tance of zero indicates that two scat shared all of the 
same prey items, whereas a distance of one indicates 
that they had no prey items in common. We paired the 
Jaccard distance matrix with a design matrix consist- 
ing of zeros for between-population distances and 1/ 
(n ; — 1) for within-population distances (where i indicates 
population membership; see Manly, 1997). When used 
with a design matrix the Mantel test is equivalent to a 
nonparametric multivariate analysis of variance (Sokal 
and Rohlf, 1995). 
We compared diets based on prey identified to the 
lowest possible taxon in order to limit the potential for 
spurious differences arising from an arbitrary categori- 
zation of prey types, although this procedure resulted in 
some comparisons where data were not at an equivalent 
taxonomic level. Analysis based on additional categori- 
zation of prey — such as size, ecology, or abundance — al- 
though potentially useful, was beyond the limits of what 
the data could support. We restricted statistical com- 
parisons to selected unpooled collections with at least 
30 samples. For each test, 9999 randomizations were 
used to obtain the distribution for the Mantel test sta- 
tistic (r M ) and to calculate probability (P) values. A sig- 
nificance level of a= 0.002 was used based on a Bonfer- 
roni adjustment of a=0.05 for 26 multiple comparisons. 
Results 
We collected 1416 Steller sea lion scat samples during 
42 collection trips from 1986 through 2007. The number 
