Clarke et al.: Elasmobranch bycatch from the shrimp trawl fishery along the Pacific coast of Costa Rica 
9 
Table 3 
Results of the 3 delta-lognormal generalized linear models (delta-GLMs) applied to abundance 
(CPUE) of elasmobranchs from deepwater, monitoring, and commercial surveys conducted along 
the Pacific coast of Costa Rica during 2008-2012: degrees of freedom ( df), deviance change 
(Deviance), residual degrees of freedom (Residual df), residual deviance (Res. dev.), Akaike 
information criterion (AIC), and the probability (P) from the F-test for lognormal submodels or 
chi-square test for binomial models. 
Model 
df 
Deviance 
Residual df 
Res. dev. 
AIC 
P 
Deepwater delta-GLM 
Lognormal submodel 
Intercept 
39 
52.43 
128.34 
Depth 
Binomial submodel 
1 
5.92 
38 
46.51 
125.54 
0.03 
Intercept 
1 
107 
141.45 
143.45 
Depth 
1 
13.12 
106 
128.32 
132.32 
<0.01 
Latitude 
1 
6.10 
105 
122.22 
128.22 
0.01 
Monitoring delta-GLM 
Lognormal submodel 
Intercept 
1 
37 
62.41 
130.69 
Depth 
Binomial submodel 
1 
4.76 
36 
57.65 
129.68 
0.09 
Intercept 
110 
142.65 
144.65 
Depth 
1 
20.33 
109 
122.32 
126.32 
<0.01 
Latitude 
1 
6.08 
108 
116.24 
122.24 
0.01 
Year 
2 
11.00 
106 
105.24 
115.24 
<0.01 
Commercial delta-GLM 
Lognormal submodel 
Intercept 
103 
189.15 
361.35 
Depth 
Binomial submodel 
1 
12.54 
102 
176.61 
356.21 
<0.01 
Intercept 
1 
126 
120.16 
122.16 
Depth 
1 
18.90 
125 
101.25 
105.25 
<0.001 
ing vessels. Although sampling depths from monitor- 
ing surveys were predefined, the location of sampling 
stations was chosen by the captain. Consequently, both 
commercial and monitoring surveys were concentrated 
in the central Pacific region. 
The nonrandom sampling design of both monitoring 
and commercial surveys may have introduced biases in 
the estimates of distribution and abundance that must 
be considered when interpreting the results of our 
study (e.g., elasmobranch abundance and composition 
covaries with shrimp abundance). Moreover, it is like- 
ly that interactions between environmental variables 
drive patterns in both species distribution and commu- 
nity structure; however, interactions were not explored 
because of the small data set. The small sample size 
may also have limited our ability to detect patterns in 
elasmobranch diversity across the examined explana- 
tory variables (e.g., depth, latitude, geographic region, 
year, season, and diel period). This limited ability is 
the most probable cause of the low percentage of the 
variance in species abundance data that was explained 
by the RDAs. 
Our findings indicate that depth is a major factor in- 
fluencing elasmobranch assemblages along the Pacific 
coast of Costa Rica. Both species richness and abun- 
dance peaked in shallow waters and decreased with 
the increasing depth. This feature is common and has 
been reported previously for both demersal (MacPher- 
son, 2003; Massuti and Moranta, 2003; Gouraguine et 
ah, 2011) and pelagic (Smith and Brown, 2002) elasmo- 
branch species. Nearshore environments are very het- 
erogeneous and tend to concentrate a large number of 
species with small depth ranges, whereas a small num- 
ber of species with large depth ranges inhabit homo- 
geneous deepwater environments (Smith and Brown, 
2002; Knip et ah, 2010; Mejia-Falla and Navia 1 ). 
Depth-related changes in environmental factors, such 
as temperature and productivity, may partially explain 
observed trends in species richness (Levinton, 1995). 
Temperature is known to be an important factor 
influencing species richness, given that it may affect 
speciation rates (Allen et ah, 2002). Productivity can 
also influence species richness; for example, areas with 
higher primary productivity tend to have species with 
high trophic levels, large body sizes, and high energetic 
requirements (Smith and Brown, 2002; Leathwick et 
ah, 2006; Knip et al., 2010), including sharks and rays 
(Priede et al., 2006). 
