Page 4 - Bracciali_2016

C. Bracciali et al. / Marine Environmental Research 113 (2016) 116e123 119

Fig. 2. Representation of morphometric measurements on Chromis chromis (Dulcic, 2005).

Table 2 size (small, <60.00 mm SL; medium, 60.00e64.95 mm SL; and
Correlation values of each morphometric at standard length increase (SL). large, 65.00 mm SL; see body condition results). In both cases,
body height-SL relationships were compared using the ANCOVA.
Morphometrics-SL relationship
The relationships between each morphometric variable and the SL
Morphometrics R N P
were also tested by using the ANCOVA where the slope b was
Head length (HL) 0.92 120 <0.0001 considered as the allometric coefficient. When there were no sig-
Jaw length (JL) 0.61 120 <0.0001 nificant differences between slopes, the intercept a was used as
Eye diameter (ED) 0.77 120 <0.0001
Preocular distance (PreOD) 0.50 120 <0.0001 index of morphometric traits length at recruitment time at the
Postocular distance (PostOD) 0.86 120 <0.0001 school. Lastly, a Principal Component Analysis (PCA) was per-
Dorsal fin length (DL) 0.96 120 <0.0001 formed as an ordination tool of morphometrical traits (Flury, 1988)
Predorsal length (PreDL) 0.93 120 <0.0001 with respect to hydrodynamic conditions, after to have removed
Pectoral fin length (PL) 0.89 120 <0.0001 the size effect (Elliott et al.,1995). To test if hydrodynamics can have
Prepectoral length (PrePL) 0.93 120 <0.0001
Ventral fin length (VL) 0.72 120 <0.0001 effect on morphometrics, a Permutational Multivariate Analysis of
Preventral length (PreVL) 0.93 120 <0.0001 Variance (PERMANOVA) was used considering hydrodynamic
Anal fin length (AL) 0.86 120 <0.0001 condition as a fixed factor (HYDRO; 2 levels; high vs. low). The log-
Preanal length (PreAL) 0.96 120 <0.0001 transformed matrix of sixteen morphometrical variables was used
Body height (BH) 0.81 120 <0.0001
Peduncle height (PH)x 0.95 120 <0.0001 to estimate the Euclidean distances, all p-values were calculated
using 9999 permutations of the residuals under a reduced model
(Anderson, 2001). STATISTICA rel. 10.0 software (StatSoft Inc., USA)
was used to perform ANOVAs and PCAs.
Table 3
Student-Newman-Keuls post hoc comparison test results of the BCI analysis. SLc ¼ SL
class; LOW ¼ LOW-HYDRO site of Cammello Bay; HIGH ¼ HIGH-HYDRO site of Punta
Bassana. 3. Results
SLc Site LOW LOW LOW LOW LOW LOW LOW
The total damselfish captured were 1368 and 1569 in LOW- and
<50.00 HIGH *** HIGH-HYDRO sites, respectively. Lengths (SL) and weights (TW)
50.00e54.95 HIGH ***
55.00e59.95 HIGH *** were highly correlated at both sites (Fig. 3), with the length-weight
60.00e64.95 HIGH ns regression slopes significantly lower in the LOW-HYDRO than in
65.00e69.95 HIGH ** the HIGH-HYDRO sites (2.78 vs. 3.36; ANCOVA, F 1,2935 ¼ 153.01,
70.00e74.95 HIGH *** P < 0.05 [***]). The growth coefficient (the slope b of the Age-SL
>75.00 HIGH ns
relationship) was not different between LOW- and HIGH-HYDRO
fish (ANCOVA, F 1,370 ¼ 0.17, P > 0.05 [ns], Fig. 4). Instead, the
intercept a was significantly greater for individuals living in HIGH-