Age determination is fundamental for understanding and properly estimating various parameters that determine the biology of cephalopod species, such as growth rate, population structure, longevity, mortality, productivity and the recruitment process (Boyle and Rodhouse 2005). Furthermore, our knowledge of these processes, as well as conditioning factors, is essential to the proper management of fisheries resources. Due to the high intrinsic growth variability and consequent mixing of individuals of different ages (micro-cohorts) within the same size range, cephalopod studies on age and growth based on length and body weight measures are not considered as accurate as those undertaken using hard structures (Jackson 1994). However, the methodologies based on the external morphology of specimens are still employed because they are low-cost and easy to apply in most fisheries (Semmens et al. 2004).

Direct ageing methods, based on reading growth increments in hard structures, have been developed to determine age in cephalopods. Particularly in octopods, beaks have provided some positive results in reading growth increments of Octopus vulgaris paralarvae (Hernandez-Lopez et al. 2001) and adults (Canali et al. 2011, Perales-Raya et al. 2010). Stylet Increment Analysis has been applied and daily growth increment deposition has been validated in Octopus vulgaris (Hermosilla et al. 2010), Octopus pallidus (Doubleday et al. 2006), Octopus maya (Rodríguez-Domínguez et al. 2013) and Octopus (cf) tetricus (Leporati and Hart 2014).

The horned octopus, Eledone cirrhosa (Lamarck, 1798), is a medium-size benthic octopod that is widely distributed over the continental shelves of the northeastern Atlantic and Mediterranean Sea (Boyle 1983). This species breeds throughout the year, although in the northwest Iberian Peninsula reproduction mostly occurs between late spring and early summer (Regueira et al. 2013). The reproductive peak in the western Mediterranean, at a similar latitude to the above-mentioned study, is similarly timed (Lefkaditou and Papaconstantinou 1995, Mangold-Wirz 1963, Moriyasu 1988). However, studies performed in Scotland indicate that egg-laying occurs later, between July and September (Boyle and Knobloch 1983), suggesting that latitude, through day-night duration, is an important parameter influencing sexual maturation (Regueira et al. 2013).

Hatching season for this species in northern Europe has been assumed to happen mainly between October and December, about three or four months later than the sexual maturation peak (Boyle 1983). Little is known about early life stages of this species. Given the size of the eggs (about 7 mm), considered intermediate compared with those of other octopuses, hatchlings are certainly planktonic, but probably only for a matter of days, during which they remain near the bottom (Boyle 1983). In view of all the above, a 20-month lifecycle has been postulated for this species in north European waters (Boyle and Knobloch 1982), although this is unverified due to the lack of studies focusing on age and population structure of the species.

E. cirrhosa landings in Galician commercial harbours show a marked seasonality, typically associated with cephalopods, which have a short life span and rapid population turnover (Sánchez and Martín 1993). Amount of captures also showed an increasing trend over the last ten years (Regueira et al. 2014). However, the biology of this octopus is still poorly studied in this region.

The main objective of the present study was to estimate age and growth of the horned octopus Eledone cirrhosa during its juvenile and adult stages using growth increments in stylets and to obtain new insights into aspects of the lifecycle of this species, such as life span and hatching season, in northwestern Iberian Peninsula waters.

From March 2009 to March 2010 samples of commercially caught horned octopus (E. cirrhosa) were collected monthly in three Galician (northwestern Iberian Peninsula) fishing harbours: Burela (north) and Ribeira or Bueu (west). This species is caught as a by-catch in the multi-species trawl fishery operating in this whole area. Vessels landing their captures in Burela operate on northern fishing grounds, while those landing in Bueu and Ribeira operate on western fishing grounds (Fig. 1).

In total, 2220 individuals of E. cirrhosa (306 males and 595 females from the northern fishing ground; and 379 males and 940 females from the western fishing ground) were processed. Measurements of each specimen included body weight (BW), eviscerated body weight (EBW) and dorsal mantle length (DML). Lengths and weights were measured to the nearest 0.1 cm and the nearest 0.1 g, respectively. Since specimens were mostly intact, no bias in weight measures was expected. The macroscopic maturity scale of Inejih (2000) was adapted to assign a specific maturity stage to each individual. For males this scale was as follows: I, immature; II, maturing; III, pre-spawning, with some spermatophores in Needham’s sac; and IV, mature, fully developed spermatophore. For females the scale was as follows: I, immature; II, maturing; III, pre-spawning; IV, mature; and V, post-spawning. Stylets of 30 specimens of each sex and fishing ground were taken monthly and stored in 70 % ethanol until preparation.

In order to identify size classes (grouped in 5 mm DMLs) and structure sampling, a modal progression analysis (MPA) was performed separately for males and females from both northern and western fishing grounds using the Bhattacharya (1967) method and FISAT software. Subsequently, three individuals representing each size class, sex and fishing ground were taken for stylet increment analysis.

Permanent stylet preparations were made following four steps according to the Barratt and Allcock (2010) methodology. First, transversal sections of stylet approximately 1-2 mm thick were obtained using a razor blade. Second, sections were dehydrated by successive immersions for one hour in 70%, 90% and absolute ethanol. Third, sections were embedded in low-viscosity acrylic resin (LR WhiteTM, London Resin Company). Fourth, in order to achieve a better light transmission and therefore a clearer view of the growth rings, the samples were ground and polished on both sides.

Samples were observed under a transmitted light microscope connected to an image analysis device equipped with LAS v1.4 software. Stylet sections were photographed at 400× magnification under full light power, with the condenser adjusted to optimize brightness and contrast. A series of pictures were taken of each stylet section, covering its entire surface. Each complete image comprised 4 to 54 partial pictures.

The stylet sections consisted of a striped-concentric surface surrounding a core, where each growth increment appears as a double band, comprising an optically translucent ring and a dark one (Fig. 2). Each pair of bands was considered as one growth increment and, therefore, as a daily time indicator in the life of E. cirrhosa.

In total, 264 stylets were prepared and photographed, and 123 were considered unreadable due to focusing or sample quality deficiencies. Two non-consecutive counts of growth increments were carried out in the remaining 141 stylets, from the outer to inner zone. When the difference between counts exceeded 10% (Doubleday et al. 2006, Hermosilla et al. 2010), an extra count was undertaken. After eliminating samples with differences greater than 10%, 122 samples were accepted. These stylets represented 27 females and 39 males from the western fishing ground, and 33 females and 23 males from the northern fishing ground.

A preliminary observation of the photos taken from these stylet sections showed that the core area of the stylet was useless for reading, as found in other similar studies (e.g. Doubleday et al. 2006). Thus, Core Area (CoA, mm²) and Counted Area (CA, mm²) of each stylet were measured from pictures. The number of increments in the CoA was estimated by extrapolation following the increment deposition rate found in CA. This deposition rate for each stylet was obtained by fitting the potential relationship between CA and the number of counted increments. After logarithmic transformation, a t-test was preformed to compare both the slope and the intercept in growth increment deposition rate between males and females (Zar 1984).

Based on previous studies on age validation by stylet increment analysis performed in three octopus species (Doubleday et al. 2006, Hermosilla et al. 2010, Leporati and Hart 2014) a daily deposition rate of growth increments in the stylet of E. cirrhosa was assumed. Thus, age of the specimens was calculated by adding the estimated growth rings in the CoA and those already counted in the CA, and hatching dates were back-calculated from capture date. Estimated age was then compared between sexes and fishing ground and was also related to BW, EBW and DML of the specimens.

Instantaneous relative growth rate (G, % BW d^–1) was calculated for each 90-day age class in order to obtain growth trends during different ontogenic periods, using the equations given by Forsythe and Van Heukelem (1987):

$$\text{G = }\frac{{\text{LnW}}_{\text{2}}{\text{– LnW}}_{\text{1}}}{{\text{t}}_{\text{2}}{\text{– t}}_{\text{1}}}\text{100}$$

where W₂ is the BW at age t₂ and W₁ is BW at age t₁. To estimate G, maximum and minimum BW for each 90-day age class were used.

Given that assumptions of normality and homoscedasticity were not fulfilled, Spearman-rank correlation (r_s) was used to evaluate covariation between numerical variables. Differences between categorical predictors were tested using non-parametric Kruskal-Wallis test for multiple comparisons and Mann-Whitney for paired comparisons, adjusting alpha by the Bonferroni approach (α_t) with the purpose of controlling Type I error across tests (Zar 1984). All data were treated with the STATISTICA 6.0 statistical software.

Our results show that E. cirrhosa hatches throughout the year in northwestern Iberian waters, with a peak in January. The comparison of our data with the reproductive cycle in this area, where the maximum gonadosomatic index of females peaks in June (Regueira et al. 2013), indicates that the period between maturity peak and egg hatching is about six months. Considering that embryonic development at 16°C takes about 110 days in this species (Mangold et al. 1971) and since sea bottom temperature on the northwestern Iberian continental shelf ranges between 11°C and 14°C (Gago et al. 2011), this longer time period should be expected. Experiments in culture indicate that females of E. cirrhosa may copulate several weeks before egg laying begins and can preserve sperm for up to six weeks (Mangold et al. 1971). Therefore, although the reproductive peak occurs in June, egg laying may be delayed until the summer and hatching until December.

A significant differential longevity between sexes was observed in E. cirrhosa, with females living longer than males. This cannot be established as a general tendency in octopods on the basis of previous studies. Neither Hernandez-Lopez et al. (2001) nor Canali et al. (2011) found significant differences in lifespan between sexes in O. vulgaris. Conversely, Leporati et al. (2008) found significant differences in age by reading O. pallidus stylets, although males were significantly longer-lived than females. This lack of uniformity in the published results could suggest inconsistencies in the sampling method rather than physiological differences between species. These inconsistencies may be due to a different catchability between sexes during different life stages, as might result from females of E. cirrhosa migrating to shallow water to spawn (Regueira et al. 2014); during this reproductive migration, they may be more vulnerable to fishing, while a proportion of males remain in deeper areas, outside the fishing grounds worked by trawlers. This could be the reason why females seem to be much more abundant than males in this fishery. On the other hand, females of E. cirrhosa lay their eggs on rocky substrate in a similar way to O. vulgaris, and certainly not on open smooth muddy or sandy bottoms, and probably brood the eggs (Mangold et al. 1971). If this is so, females might not be captured during this period, while males could be captured throughout their life. The absence of post-spawning females in this study supports this fact. Given the above, lifespan difference between sexes should be up to almost eight months. Combining this with the results on age estimate obtained in this study (maximum age of 517 days), the life span of female E. cirrhosa should be around two years, agreeing with the hypothesis of Boyle (1986) for Scottish populations.

According to minimum estimated ages, recruitment to the fishery of E. cirrhosa in northwestern Iberian waters should occur when animals reach 3-4 months old. This also agrees with the fishery catch pattern of the species in northwestern Galician waters, in which a peak of captures corresponding to new recruits takes place around May (Regueira et al. 2014). Thus, the concluded chronological pattern matches that proposed by Mangold et al. (1971) when analysing embryonic development of the species.

Despite the significant correlation found between size and age, in view of our results, MPA seems not to provide a suitable vision of age structure in E. cirrhosa populations. The application of MPA as a tool for inferring age structure of cephalopod populations has been rejected before, mainly due to the high intrinsic variability in growth rates (Forsythe and Van Heukelem 1987) and also because neither length nor weight are good indicators of the size of a soft-bodied animal (Forsythe 1984). Although some authors have suggested its validity if it can correct the observation bias introduced by the size-selective fishing gear (Keyl et al. 2011), our results support the former hypothesis, given the high size diversity between same-age specimens. This individual diversity is also illustrated by the wide variety of mathematical models used to define the growth of cephalopods, due to the unsuitability of a single model account for the wide variability in size of individuals belonging to a single age class (Semmens et al. 2004). Given this lack of consensus, it seems likely that cephalopods have different types of growth in different stages of development, so the growth rates set out in this work are only valid for the benthic life stage and cannot be extrapolated to embryonic and planktonic stages.

Based on size at first maturity (DML50%) estimated for this species on northwestern Iberian coasts (Regueira et al. 2013) and DML-Age relationships in this paper, females should achieve maturity at around one year old, while males should do so at approximately 8 or 9 months.

Estimated growth rates are similar results to those estimated for other octopods in the wild, such as Octopus mimus (0.09-1.25%) (Cortez et al. 1999). According to experiments in the laboratory, octopuses generally demonstrate two growth phases: an initial rapid exponential phase and a slower power growth phase (Semmens et al. 2004). Therefore, a progressive moderation of growth rate would be expected, or at least in young individuals it should remain stable. Given the limited amount of successfully aged individuals, although the observed trends are plausible, further analysis will be needed to accurately determine such changes in growth rates as well as for embrionary and planktonic phases. Moreover, as in other cephalopods, different micro-cohorts may have different growth rates depending on the time of hatching (Pierce and Guerra 1994). The lack of precise catch location data could explain part of the variability in our results, as well as reported segregated distribution by sex and size (Regueira et al. 2014), since individuals caught in different areas may be subjected to diverse abiotic and biotic conditions affecting growth processes (Pierce et al. 2008).

The method for sample preparation described by Barrat and Allcock (2010) and applied herein provided durable samples, which were kept throughout the process of experimentation and photography. This proves the suitability of the method for age studies in Eledone cirrhosa.

In the future, the results of the present study will need further validation, not only because the deposition rate of growth rings could have a different periodicity to a day, but also because stylet formation and growth processes remain unknown during embryonic development and early life stages and, as observed in other structures such as statoliths (González et al. 2010), the central area of the stylet may undergo compaction during early growth. That would make it harder to estimate the number of rings in this area and could bias the results on estimated age. However, the concordance between this interpretation of the population structure by mean of age estimates, the reproductive cycle described for the species and the pattern of catches supports the validity of what is concluded in this paper.