Gonzalez-Pestana et al.: Trophic ecology of Sphyrno zygaena off northern Peru 
453 
IRI as a percentage (%IRI; Cortes, 1997). Items rarely 
found in stomachs (e.g., rocks, snails) and parasites (e.g., 
isopoda) were not included in the analysis. 
Statistical analysis 
Cumulative prey curves were constructed to deter¬ 
mine whether an adequate number of stomachs had 
been collected to accurately describe the diet of smooth 
hammerhead (Jimenez and Hortal, 2003). The order in 
which stomachs were analyzed was randomized 1000 
times to eliminate bias. The number of stomachs an¬ 
alyzed is considered sufficient in describing the diet 
when a cumulative prey curve reaches an asymptote. 
Therefore, a slope value less than 0.1 indicates a good 
representation of diet (Soberon and Llorente, 1993). 
We used 2 indices to evaluate trophic niche width 
of prey taxa: Levin index and Berger-Parker index. 
The Levin index was based on %N values. The index 
values range from 0 to 1; low values (<0.6) indicate 
a diet dominated by few prey items, (specialist preda¬ 
tor) and higher values (>0.6) indicate a generalist diet 
(Labropoulou and Eleftheriou, 1997). The Berger-Park¬ 
er index uses the formula of Magurran (1988), d=(n , 
max)/N, where N represents the number of all record¬ 
ed food components (taxa) and max represents the 
number of specimens from taxon i (the most numerous 
taxon in the diet). This index ranges between 1/N and 
1: values closer to 1 represents a specialist feeder and 
a value closer to 1/N indicates a generalist feeder. 
We calculated trophic position on the basis of %IRI 
values of the prey species presented in the stomach 
content. We used the following equation: 
7/,-l!- (I BCij)x(TLj) 
(Christensen and Pauly, 1992), (1) 
where DCjj = the composition of the diet in which j is 
the proportion of preys in the diet of the 
predator I; and 
TLj = the trophic level of the preys. 
The trophic level of the fishes were taken from Froese 
and Pauly 3 and Espinoza (2014) and the trophic level 
of the cephalopods were taken from Cortes (1999) and 
Espinoza (2014). 
We analyzed differences in diet according to 6 fac¬ 
tors: body size and sex of sharks, location of capture, 
season, year, and environmental conditions (El Nino- 
Southern Oscillation event: November 2014 to Decem¬ 
ber 2015). For body size we divided the sharks into 
size classes. This division was based on analyses of 
similarities (ANOSIM) where we chose the size class¬ 
es that showed the highest R- statistic and the lowest 
P-value (Clarke, 1993). The division of the locations 
(north: Zorritos, Acapulco, Cancas, Mancora, and Yaci- 
la; south: San Jose; and Salaverry) was justified be- 
3 Froese R., and B. Pauly. 2012. FishBase, vers. 02/2012. 
[World Wide Web electronic publication; available from 
http://www.fishbase.org.] 
cause of biogeographic characteristics of the Tropical 
East Pacific and Warm Temperate Southeastern Pacific 
marine provinces where the collection sites were locat¬ 
ed (Spalding et al., 2007) (Fig. 1). The division of the 
seasons was based upon the seasonality of chlorophyll- 
a concentration and primary production; for which the 
highest levels occurred during the austral summer and 
fall (Pennington et al., 2006). Therefore, we divided the 
data into 2 seasons: season 1 (austral summer and fall) 
and season 2 (austral winter and spring). 
Nonmetric dimensional scaling (nMDS) ordinations 
generated from a Bray-Curtis similarity matrix on nu¬ 
meric abundance of prey (%N) was used to determine 
whether sex, body size, capture location, season, year, 
or environmental conditions exerted the greatest over¬ 
all influence on the dietary composition of smooth ham¬ 
merhead. ANOSIM was used to test whether dietary 
compositions differed significantly, by generating a 22- 
statistic, stress value, and a P-value. 22-statistic values 
describe the extent of similarity (Clarke, 1993), with 
values near 1 indicating that the 2 groups are entirely 
separate and values close to 0 indicating that there 
are no differences between the 2 groups. Stress value 
measures the goodness-of- fit of the nMDS model to 
the data, where values closer to zero indicate excellent 
representation (without risk of misinterpretation) and 
values larger than 0.2 indicate that the interpretation 
is unreliable (Clarke, 1993). Similarity percentages 
(SIMPER) were employed to determine the dietary cat¬ 
egories that typified particular groups or contributed 
most (or typified a combination of both categories) to 
the similarities between groups (Clarke, 1993). If sig¬ 
nificant differences existed in the diets by factors (e.g., 
sex), then trophic niche width, degree of overlap (based 
on the Bray-Curtis index) and trophic position were 
calculated for each factor. 
Statistical analyses were performed by using RStu- 
dio, vers. 0.96.122 (RStudio, 2012) with R, vers. 3.2.2 
(R Core Team, 2016). 
Results 
A total of 485 samples of gut contents were collected. 
Individual smooth hammerheads measured between 53 
and 294 cm TL, The slope value of 0.002 (less than 0.1) 
for the cumulative prey curve showed that sufficient 
stomach contents were examined to adequately and re¬ 
liably describe the diet of smooth hammerhead. With 
the cumulative prey curve, we calculated that the con¬ 
tents from 39 stomachs would be needed to accurately 
analyze the diet of smooth hammerhead. 
Food items were found in 78% of the stomachs. Of 
these, 92% were in an advanced state of digestion (stag¬ 
es 3 and 4). Prey composition comprised 25 prey items: 
14 teleosts and 11 cephalopods (Table 1). According to 
%IRI values, the most important prey species were Pa¬ 
tagonian squid ( Doryteuthis ( Amerigo ) gahi) (37%) and 
jumbo squid (27%). These 2 species comprised more 
than 60% of the diet. The trophic position was high 
