88 
Fishery Bulletin 116(1) 
Table 3 
Summary of results from a multivariate analysis of 
variance (MANOVA) (rc=300) on normalized elliptic 
Fourier descriptors for otoliths of blue jack mackerel 
(Trachurus picturatus) sampled in 2015 from the Ma¬ 
deira archipelago, off mainland Portugal and off the Ca¬ 
nary Islands. Results include the test statistics Pillai’s 
trace and Wilks’ X. df=degrees of freedom (numerator, 
nominator). 
Pillai’s 
trace 
Wilks’ X 
df 
P-value 
Area 
1.09 
0.15 
114.43 
<0.0001 
Sex 
0.23 
0.77 
57.22 
0.27 
Area * sex 
0.43 
0.62 
114.43 
0.41 
nary Islands showed the lowest classification success 
with 62.3%. 
The two discriminant functions were significant 
and discriminated the three areas studied (fl: A.=0.26, 
P<0.0001; f2: ^.=0.85, P<0.0001). The score plots for the 
first 2 discriminant functions (Fig. 8) showed a separa¬ 
tion between the 3 areas studied, but some overlap¬ 
ping can be observed. The first discriminant function 
explained 92.6% of between-group variance. 
Discussion 
The blue jack mackerel population structure in the 
southern part of the northeast Atlantic (Peniche-Ma- 
deira-Canary Islands) was unknown before this study. 
The present results reveal the usefulness of anatomi¬ 
cal geometric morphometric and otolith shape analy¬ 
sis in supporting the existence of three stock units of 
blue jack mackerel in the southern northeast Atlantic. 
The same holistic approach that we applied was also 
used to discriminate the stocks of another species of 
the genus Trachurus in the northeast Atlantic Ocean 
and Mediterranean Sea, the horse mackerel (Trachurus 
trachurus). The body shape (Murta et al., 2008a) and 
otolith shape (Stransky et al., 2008) analysis provided 
evidence of a consistent separation between Atlantic 
and Mediterranean locations, with 90% correct alloca¬ 
tion of individuals of each stock in the otolith shape 
analysis (Murta et al., 2008a; Stransky et al., 2008). 
Ideally, a sampling strategy should focus on a spe¬ 
cific time scale (e.g. spawning season) and on a specific 
length or age range of fish. Adherence to this strategy 
would have enhanced our results, but different fishing 
methods did not allow the implementation of an ideal 
sampling strategy. To overcome this potential weak¬ 
ness, the corrections used in both analyses removed 
the effect of size on shape. High values of classification 
success (78.0% and 73.3%) were achieved in the ca¬ 
nonical discriminant analysis used in both anatomical 
morphometric and otolith shape analysis, respectively, 
Table 4 
Classification matrix of the discriminant analysis per¬ 
formed by using a jackknife procedure for cross-valida¬ 
tion of the otolith shape of individual blue jack mack¬ 
erel (Trachurus picturatus) sampled from 3 areas in 
the northeast Atlantic Ocean in 2015: off the Madeira 
archipelago, off mainland Portugal, and off the Canary 
Islands. Values are percentages of individuals sampled 
in the areas given in rows that were then classified into 
the areas given in columns (values for correct classifica¬ 
tion are presented in bold). Overall classification suc¬ 
cess: 73.3%, Wilks’ X=0.26. In the discriminant analysis, 
13 variables were used. 
Region 
Madeira 
Mainland 
Portugal 
Canary 
Islands 
Madeira 
66.0 
10.0 
24.0 
Mainland Portugal 
9.0 
89.0 
2.0 
Canaries 
31.2 
6.5 
62.3 
Variables 
d7, b9, c8, b8, cl3, cl5, dl3, dl2, 
dl5, c5, a5, cl4, c9 
indicating three phenotypically distinct local popula¬ 
tions. No differences between sexes were observed by 
using either technique. 
Using both analytical techniques, we found the 
highest percentages of misclassification for specimens 
from Madeira that were classified in the Canary Is¬ 
lands group and vice-versa. This overlap between areas 
may be indicative of some degree of migration between 
these two populations. In general, the migratory move¬ 
ments of the blue jack mackerel are driven by feeding 
and spawning requirements (Menezes et al., 2006) and 
where seamounts are used as feeding areas in their 
preadult and adult phase, but it is likely that there 
are immigrants to the seamount from the island shelf 
areas (Menezes et a!. 7 ). Differences in the biological 
and physical environments of hatchery and nursery 
areas can also result in morphometric variations be¬ 
tween these populations (Robinson et al., 1993; Chipps 
et al., 2004; Vila-Gispert et al., 2007). Other, nonen- 
vironmental factors may also contribute to differences 
among populations. For example, a greater growth ca¬ 
pacity in high-latitude populations reflects an adapta¬ 
tion that would counteract a slowing of growth rate 
over the growing season; otherwise a reduction in an¬ 
nual growth rates of individuals with increasing lati¬ 
tude would be expected because of the shorter growing 
seasons at higher latitudes that result in a reduction 
in body size (Conover and Present, 1990; Yamahira et 
al., 2007). Yamahira et al. (2007) observed that individ- 
7 Menezes, G., A. Rogers, H. Krug, A. Mendon^a, B. M. Stock- 
ley, E. Isidro, M. R. Pinho, and A. Fernandes. 2001. Season¬ 
al changes in biological and ecological traits of demersal and 
deep-water fish species in the Azores, 164 p. Univ. Azores, 
Dep. Oceanogr. (DOP), Arquivos DOP Ser. Estud. 1/2001. 
