To further assess the occurrence of sexual dimorphism using geometric morphometric techniques, pictures were also taken from the same external view of the right second pleopod of all five males and females selected using the same magnifier, which had an integrated Leica DC100 video camera and a DC200 digital imaging system. The first landmark was placed on the dorsal maximum convex curvature of the protopod, the second and fourth were placed at the anterior maximum curvature where the exopod and endopod are inserted in the protopod, the third at the maximum concavity between these two landmarks, the fifth at the maximum perpendicular width of the protopod in relation to the first landmark, the sixth at the ventral maximum curvature in the posterior part of the protopod and the seventh at the maximum concave curvature on the posterior dorsal margin of the protopod (Fig. 1).

Using Tps.Dig2 v2.16 software (Rohlf 2010a), the x,y coordinates of the landmark dataset were obtained. Generalized procrustes superimposition analysis was performed to align and to eliminate scale and orientation distortions before shape analysis using tpsRelw v1.49 software (Rohlf 2010b). Statistical analysis of shape variance was assessed using the MorphoJ software (Klingenberg 2011). We then applied procrustes ANOVA, which is a two factorial ANOVA that performs analysis of both size (univariate) and shape (multivariate). This procedure is therefore a useful method for discovering localized variation (Klingenberg and McIntyre 1998). Finally, a principal component analysis (PCA) was applied to quantify the contribution of the principal factor of landmark displacement that explains the shape variance between genders. This PCA was run using the covariance matrix generated from shape variation components, which were previously derived from the coordinates landmark dataset, pooled within-group with sex as a covariate (Viscosi and Cardini 2011).

We assessed the power of telling apart the gender in adult individuals by just visual observation, since this would facilitate sex identification in field work. A total of 187 individuals of sizes comprised between 8 and 24 mm CL were visually examined for sex identification based on pleopod morphology during one fishery research cruise. Three volunteers were asked to identify the sex visually based on the information provided by the morphological drawings obtained in the section on pleopod morphology. A posteriori one of the authors checked whether each assignation was correct based on the presence of appendix masculina under a stereoscope microscope, as usually performed for caridean shrimps. The proportion of correct assignation by size class was then calculated.

Sex ratio, as the proportion of males, was calculated within each 1 mm CL size class separately for samples taken in the Alboran Sea (the westernmost basin of the Mediterranean Sea), the intermediate Alacant area (between Cape Palos and Cape La Nao) and the Catalan Sea (in the NW Mediterranean). These sectors were established according to previous results (Simão et al. 2015), in which differences in size at maturity were observed between Alboran Sea populations and the rest of the sampled populations along the Iberian Peninsula Mediterranean coasts. Sex ratio was not calculated in size classes with fewer than 25 individuals. The significance of the deviation from an expected 1:1 sex ratio was calculated using a heterogeneity G-test (Sokal and Rohlf 1981, Abelló et al. 1990). The relationship of sex ratio with depth was also calculated; only samples containing more than 25 sexed individuals larger than 15 mm CL were considered for this analysis; this cut-off size was chosen to ensure that sex ratio was calculated over adult specimens to minimize error in gender assessment (see below), and also to encompass the whole adult cohort as visually assessed through size frequency distributions.

In order to verify whether size frequency distributions (SFDs) differed between geographic populations and between depth strata, the Kolmogorov-Smirnov test within VITMAN fisheries software (Lleonart and Salat 1992) was used. Combinations of geographic area and depth strata resulting in less than 100 individuals measured (i.e. at both extremes of the bathymetric range of the species in each area) were not used in the analyses.

The difference in protopod morphology between genders of Pasiphaea sivado was quite readily identified by sight (Fig. 2). Overall, in all five pleopod pairs, adult female protopods were clearly much thinner than those in adult males. Additionally, the anterior edge of the male protopods was clearly much more convex than in females, while the posterior edge was approximately straight in both males and females. As shown in Figure 3, the ratio between pleopod protopod length and width (L/W), which can be considered a protopod thinness index, was much higher in females than in males, thus characterizing the thinner, rectangular, elongated shape in females versus the more robust shape present in males. A significant difference between genders was obtained for all five pleopod pairs (Table 1). Additionally, no setae were present on the anterior edge of the first pleopod protopod in any of the adult males examined, and in the rest of the pleopods anterior setae were much scarcer in males than in females.

The use of geometric morphometrics techniques further evidenced the differences between genders in pleopod protopod shape. Both P. sivado males and females grouped as two well-differentiated clusters in the analysis of the shape of their second pleopod protopods (Fig. 4A). The overall consensus shape (Fig. 4B) of the pleopod protopods, which is based on the geometric mean position of each landmark, clearly differentiated between the thinner, elongated shape of the females versus the broader shape found in males. PCA showed that the first principal component (PC1), which accounted for 56.1% of the total shape variability, clearly differentiated males from females. The shape variance between genders explained by the PC1 was significantly different (ANOVA F_10,80=27.24 p<0.0001) andwas not affected by the centroid size effect (ANOVA F_1,8=1.87 p=0.209). The next five principal components explained 16.5%, 10.4%, 6.2%, 3.5% and 1.26% of the total shape variance, respectively.

Regarding the visual sex assessment on board, a total of 542 gender assignations were made on the examined individuals by three independent observers. Figure 5 shows the percentage of correct sex identification by pleopod morphology based on independent visual assessment by all observers. Total reliability was attained in individuals larger than 18 mm CL, and the percentage of correct assignation decreased with decreasing size.

The overall SFDs of sexed individuals by geographic area (Fig. 6) were biased towards adult individuals in all populations. The Alboran Sea population showed a higher proportion of individuals in the largest sizes than in those obtained in the Alacant area and the Catalan Sea. The comparison between male and female SFDs within each area showed that they were not significantly different in any case (main diagonal in Table 2). The comparison of male or female SFDs among areas also showed non-significant values (Table 2).

The analysis of sex ratio by size class showed that in all sectors females were dominant at sizes smaller than around 18-19 mm CL, while the proportion of males was significantly higher at larger sizes, therefore evidencing the occurrence of a slight sexual size dimorphism in P. sivado (Fig. 6). When special attention was paid to the precise size at which the dominance of males occurred, a decreasing gradient was found in the study area, with the proportion of males being significantly higher at CL sizes larger than 21 mm in the Alboran Sea, 20 mm in the Alacant area and 19 mm in the Catalan Sea (Fig. 6).

When the shape of the SFDs was analysed by depth strata, a high frequency of juveniles in relation to adults was evident in the shallowest depth stratum (200-300 m), especially in the Alboran Sea (Fig. 7), where the population showed a marked SFD displacement towards larger sizes with increasing depth. No noticeable variation in SFD displacement was evident in the Alacant area and Catalan Sea populations (Fig. 7 and Table 3). The comparison of the SFDs with the Kolmogorov-Smirnov test showed the occurrence of significant differences by depth only in the Alboran Sea population, in both males and females (Table 4). No significant differences in SFD variation with depth were detected in the Alacant area and Catalan Sea populations. It is worth noting that despite intensive sampling, the densities of P. sivado at certain depths and areas are very low (Simão et al 2015): the species is hardly found shallower than 200 m, except in the Alboran Sea, and deeper than 600 m throughout the study area.

A total of 84 samples were used to analyse the sex ratio across the bathymetric distribution of the species (Fig. 8). Most of the samples (70.2%) did not show a significant deviation from a 1:1 sex ratio. However, sex segregation was identified in the rest of samples, with 11.9% of them being biased towards females and 17.9% towards males. The highest proportion of significant deviation towards females was found at the shallowest extreme of the bathymetric distribution of the species but also in the deepest strata. On the other hand, samples with a significantly higher proportion of males were located in the central depth strata. Overall, no significant tendency of gender segregation across the bathymetric distribution was found (r=0.0803, p=0.4677).

The occurrence of sexual dimorphism in pleopod shape was detected in Pasiphaea sivado by both visualization and measurement methods: the traditional linear measurement and geometric morphometric analysis. An important, useful application of the present observations and the description of the occurrence of sexual dimorphism in pleopod shape in adult P. sivado is, from now on, the ability to quickly assess the gender of individuals when population biology studies are performed in the field.

A similar sexual dimorphism in pleopod shape has also been reported in Pasiphaea japonica, but it also seems to occur in Alainopasiphaea nudipeda, as evidenced by Figures 1 and 2d in Hayashi (1999), as well as in Pasiphaea semispinosa (authors’ personal observations based on specimens deposited in collections: CBR at the ICM-CSIC in Barcelona and CCDE at the IEO-Cádiz). Other personal observations, not yet quantified, on the congeneric species Pasiphaea multidentata—a species that coexists with P. sivado across the Mediterranean Sea (Company et al. 2001, Abelló et al. 2002)—indicate that this species does not show this dimorphism.

The phylogenetic relationships within the family Pasiphaeidae are far from being properly understood (Hayashi 1999, 2006, Matzen da Silva et al. 2011). Based on present findings, the occurrence of sexual dimorphism in pleopod shape may well have a significant role in determining phylogenetic relationships within the family Pasiphaeidae, in particular within the Pasiphaea sivado group (Hayashi 1999, 2006), since both P. sivado and P. japonica (which also belongs to this group) have been shown to display this morphological characteristic (Doi 1975, Nanjo and Ohtomi 2009). Insufficient information is as yet available concerning other species in the genus or the family.

The present study also provides evidence on the occurrence of significant size dimorphism in P. sivado between genders, a fact previously unreported in this species, with females accumulating in size classes smaller than around 20 mm CL and males accumulating in the largest size classes. Sex ratio as a function of size clearly suggests the occurrence of differential growth rates between genders in P. sivado, as reported in many species of crustaceans (Abelló and Cartes 1992, Kevrekidis and Thessalou-Legaki 2006, Oh et al. 2006). This fact is usually attributed to a higher investment of females in reproduction rather than growth. This pattern appears to follow the so-called anomalous pattern described by Wenner (1972), in which no differences in sex ratio occur in juveniles, so a predominance of females is due to a lower female growth rate, with males finally accumulating their frequency in the largest sizes. Several explanations may account for this pattern, with different growth rates between the genders from the puberty moult or size at maturity being the most likely (Ben Mariem 2004, Fantucci et al. 2009). In the present study, mature females were identified at sizes larger than around 15 mm CL, coinciding with results from Company et al. (2001), and size at 50% maturity was found to vary geographically between 20.6 mm CL in the Catalan Sea and 23.0 mm CL in the Alboran Sea. Some biological characters estimated by Company et al. (2001) for P. sivado in the Catalan Sea support the sex ratio pattern with size reported in the present study. In that work males showed a faster growth rate (k=0.62), a larger asymptotic size (L_∞=27.5 mm) and a higher growth performance index (Φ=0.13) than females (0.55, 26 mm and 0.06, respectively). Reproductive aspects of the species, such as the continuous reproduction throughout the year and the higher volumetric index of females (2.92) compared with males (2.76) (Company et al. 2001), also suggest the smaller amount of energy investment in size growth by P. sivado females in relation to males. Another factor that may also limit growth in females is the moulting delay of ovigerous females in order to avoid loss of attached embryos, a phenomenon characteristic of crustacean species with egg-carrying biological features (Guerao et al. 1994, Oh et al. 1999, Viegas et al. 2007), such as caridean shrimps, lobsters, and brachyuran and anomuran crabs.

The size at which differences in sex ratio become significant has also been suggested to be an indirect indicator of the size at puberty moult and/or maturity size/age of the species (Abelló et al. 1990). In the present study, the precise size at which males outnumbered females progressively decreased from the Alboran Sea to the Alacant area and the Catalan Sea, corroborating the patterns of size structure and size at maturity observed in this area (Simão et al. 2015) and implying the occurrence of environmental and ecological processes affecting growth rates.

Overall, no significant gender segregation in P. sivado populations as a function of depth was found, suggesting that this species does not perform differential migration between sexes over the sea bottom. In the Catalan Sea, the absence of gender segregation by depth was also reported by Company et al. (2001) throughout the year, and could be related to the gregarious behaviour of macroplankton/micronekton organisms (McManus and Woodson 2012). During the daytime, P. sivado inhabits the epi-/suprabenthic environment over the upper and medium slope, with high densities recorded when sampled by demersal trawl nets during fishery research surveys (Cartes et al. 2007, García-Rodríguez et al. 2011, Simão et al. 2015), while during the night the species has been reported to migrate several hundred metres up the water column to waters near the surface to feed on plankton (Froglia and Giannini 1982, Cartes 1993, Simão et al. 2014).

Controversially, a relatively high number of samples showed biased sex ratios, towards both a higher proportion of males and of females. In the absence of direct observations, sex ratio–deviated groups, indicative of aggregative behaviour within genders, are suggestive of differential reproductive behaviour and/or selective feeding in relation to the energetics and quality needs depending on gonadal developmental stages. An increase in stomach fullness and dietary energy intake has been recorded in pre-reproductive females of the deep-sea shrimp Aristeus antennatus, parallel to a significant increase of the gonadosomatic index and fecundity (Cartes et al. 2008). Differences in diet composition between genders and maturity stages have also been found in other shrimps, Plesionika heterocarpus and Plesionika martia (Fanelli and Cartes 2008). Aggregative behaviour within genders has also been related to other causes, such as the need to look for appropriate habitat for moulting (Reid et al. 1994, 1997, Sampedro and González-Gurriarán 2004) or for hatching (i.e. females look for suitable sites to improve dispersal and survival of recently hatched larvae) (Hicks 1985, Zeng and Naylor 1997, Carr et al. 2004). Survival of larvae may also be related to larval aggregation behaviour in addition to dispersal patterns, which are assumed to be low given the large size of the eggs and lower number of embryos that P. sivado females carry, when compared with other caridean shrimps (Company et al. 2001).

In attention to the concept of operational sex ratio (OSR), defined as the proportion of receptive females to sexually active males (Correa and Thiel 2003), the results of the present study suggest that the populations of P. sivado may overall be close to the OSR optima. The histograms of Figure 6 show that the overall shape of the SFDs of males and females is very similar, but males are displaced towards the right (largest sizes) by just one size class. If we quantify the total adult males and females bigger than the size class preceding the one in which the sex ratio starts to be biased towards males, we will find a sex ratio close to equilibrium. Even with the existence of a size sexual dimorphism or some samples with biased sex ratio, no overall sex segregation at population level was found.

The present study has shown the occurrence of morphological differences between genders in the benthopelagic shrimp P. sivado, namely in the pleopod shape of adult individuals. This characteristic has been shown to be useful for determining sex in field studies. Additionally a slight size dimorphism has been identified, as well as geographic population differences in size structure. This knowledge may also be useful for studying population biology patterns in still poorly-known species inhabiting exploited demersal ecosystems.