Introduction

Loliginid squids have a high commercial value and play an important role in the ecosystem, as they are eaten by fish and marine mammals and are predators themselves and have been recognised as keystone species in marine ecosystems (Jackson 2004; Gasalla et al. 2010; Arkhipkin et al. 2015). Loliginids, like most cephalopods, are fast-growing animals with a short life cycle of one to two years, or even shorter in medium-small sized squids (Villanueva et al. 2016). The hatchlings are planktonic and, since there is no metamorphosis, are known as paralarvae (Young and Hartman 1988). Differences in growth rates between individuals of the same age are likely to be driven at least partly by environmental factors such as temperature and food availability (Vidal et al. 2002). Upwelling areas are characterised by both high values (during the upwelling season) and high variability of nutrient concentrations, which might be another critical factor in the growth and survival of hatchlings, influencing the interannual differences in recruitment (Raya et al. 1999).

The Western Iberian Peninsula (WIP) is situated at the northern boundary of the Canary Current upwelling ecosystem (CanC). During summer the coastal upwelling occurs with high intensity, leading to a high concentration of nutrients and high primary production supporting upper trophic levels (Arístegui et al. 2009). Squid hatching during the upwelling season will tend to have high growth rates due to the high levels of food intake, whereas those hatching during downwelling will tend to have lower growth rates (Rocha and Guerra 1999), although other factors such as temperature might also have important effects on early growth rates (Raya et al. 1999). Moreover, Moreno et al. (2005) identified two cohorts for L. vulgaris, originating from two spawning seasons, referred to as the warm cohort (WC) and cold cohort (CC) with reference to the temperatures experienced by hatchlings in Portuguese waters. The WC cohort matures at an earlier age than the CC (Moreno et al. 2005, 2012).

On the WIP coast, the loliginid paralarvae are represented by Loligo vulgaris, L. forbesii, Alloteuthis subulata and A. media. Earlier studies identified all the loliginid paralarvae found in the WIP as Loligo vulgaris (González et al. 2010) but more recent molecular studies identified two additional species in the same area, A. media and A. subulata, co-occurring with L. vulgaris (Olmos-Pérez et al. 2018; Roura et al. 2019; García-Mayoral et al. 2020). Chromatophore patterns of paralarvae tend to disappear when specimens are preserved in ethanol and morphological identification to species level can thus be complicated. Hence, for correct identification of different species of loliginid paralarvae, it is useful to apply molecular techniques.

In NE Atlantic waters, both Loligo species are typically recorded as Loligo spp. in the fish markets due to their similarity, while A. media and A. subulata may be reported as Alloteuthis spp. or (mistakenly) as juveniles of Loligo spp. (Arkhipkin et al. 2015; Jereb et al. 2015). Since the 1990s, the abundance of L. forbesii has declined in the WIP (Chen et al. 2006), so that nowadays most Loligo spp. landed in the region is L. vulgaris, which is an important commercial resource in both Spanish and Portuguese fisheries (González et al. 2005; Moreno et al. 2009). In contrast, Alloteuthis spp. are arguably an underexploited species, probably in part because they may be mistaken for juvenile Loligo spp. and are often under the minimum landing size for Loligo spp. in Spain and Portugal. However, they are specifically targeted by Galician small-scale fisheries. Since cephalopods are not quota species under the EU Common Fisheries Policy, stocks are usually not formally defined and (as noted above) landings are often not identified to species, so assessment of stock status is problematic (ICES WGCEPH 2019).

In Galician waters (NW Spain), mean reported annual landings of L. vulgaris varied between 482 t in 2012 and 176 t in 2021. Meanwhile, landings of Alloteuthis spp. between 2012 and 2021 ranged from 24 to 15 t (https://www.pescadegalicia.gal/estadisticas/). Other barriers to stock assessment, and hence also a limitation for management of the fisheries, include biological knowledge gaps: it is essential that assessment and management account for life cycle biology, including short life, environmental sensitivity, fast metabolism and rapid growth and ontogenetic migrations, as well as the resulting strong seasonality and high year-to-year variability of abundance.

To permit appropriate management of fishing on squid resources, age and growth studies are very useful to improve the knowledge of population dynamics and population structure, facilitating the understanding of within and between year changes in the population biomass resulting from natural and fishing mortality (Rocha and Guerra 1999), as well as helping to identify and characterise cohorts and microcohorts within the population. To date, age and growth studies in the WIP calculating the number of daily increments in statoliths of loliginid squid were mainly carried out in adults of L. vulgaris (Guerra and Rocha 1994; Rocha and Guerra 1999; Moreno et al. 2005, 2007; Boavida-Portugal et al. 2010) and A. subulata (Moreno 1993). Some growth studies based on statolith increments for L. vulgaris and A. subulata were also carried out in other areas, e.g. the West Saharan Bank (Arkhipkin and Nekludova 1993; Arkhipkin 1995), the Mediterranean (Natsukari and Komine 1992; Villanueva et al. 2003) and in Portuguese waters (Moreno et al. 2012), but using embryos, juveniles or adults. Most studies of the age and growth of loliginid paralarvae were based on laboratory culture experiments (Lipinski 1986; Turk et al. 1986; Yau 1994; Villanueva 2000a, b). The only study of the length at age of wild loliginid paralarvae was carried out in Galicia (NW Spain), but it did not distinguish between the different species (González et al. 2010). However, knowledge of the growth patterns of the different loliginid paralarvae is essential to understand growth and maturation in younger animals, since they potentially affect adult size and life cycle phenology (Pecl and Jackson 2008).

Thus, the objective of this work was to study the growth rates during the paralarval phase and determine the hatching peaks of the three loliginid species found in Galician waters (NW Spain): L. vulgaris, A. subulata and A. media, based on counting daily growth increments in statoliths. The results should allow us to distinguish the different cohorts and identify any differences in the growth curves of the three species. We also used the material to briefly describe the morphology of the statoliths of paralarvae of the three species.

Materials and methods

Sample collection

Zooplankton samples were collected during oceanographic surveys along the Northeast of Spain (Fig. 1A–D). In the north of Galicia (Fig. 1A), zooplankton samples were obtained on board R/V Sarmiento de Gamboa during June 2017 within the project DREAMER, with a bongo net of 500 µm mesh size. In the southwest of Galicia, zooplankton samples were obtained in the Ría de Pontevedra and Ría de Vigo (Fig. 1B) on board R/V Mytilus using a Multinet type Mammoth (Hydro-Bios) sampler with 250 µm mesh size, between June and November in 2017 and from May to October in 2018. The multinet was lowered close to the seafloor and pulled up gradually to the surface, filtering 200 m3 within each of seven fixed depth strata: 105–85, 85–55, 55–35, 35–20, 20–10, 10–5 and 5 to 0 m. In the Ría de Vigo, six surveys were undertaken at night along four transects (T2 to T5, Fig. 1C) and four 24 h surveys were carried out along six transects around the Cíes and Ons islands (TA-TF, Fig. 1D). The samples were fixed in 96% alcohol on board and stored at – 20 ºC. Subsequently, loliginid paralarvae were sorted from the zooplankton samples in the laboratory and stored in 70% alcohol.

Fig. 1
figure 1

Map showing the sampling sites. A Sampling from the north, with depths ranging from 50 to 500 m and B the southwest of Galicia, showing in detail the transects carried out in C Ría de Vigo and D around the Cíes and Ons Islands, where sea depths ranged from 10 to 105 m. T2, T3, T4, T5, TA, TB, TC, TD, TE, and TF are the name of the transects along which sampling was carried out

Full size image

Identification of paralarvae

A total of 312 loliginid paralarvae was sorted from the zooplankton samples. Species identification using morphological characters was not feasible due to the small size of paralarvae, and the loss of colour from the chromatophores resulting from preservation in ethanol. Accordingly, molecular methods were used to distinguish the loliginid paralarvae from others and to identify them to species. Overall, 278 loliginid paralarvae were identified genetically. For mtDNA extraction, mantle tissue from each paralarva was taken using the QIAamp DNA Micro Kit (QIAGEN), applying the manufacturer’s protocol. Universal primers HCO2198 and LCO1490 were used to amplify a fragment of cytochrome c oxidase subunit I (COI) (Folmer et al. 1994) and sequencing of PCR products was carried out in StabVida (Portugal). For species identification, all sequences obtained were examined using the BLAST tool (Basic Local Alignment Search Tool) through web servers of the National Center for Biotechnology Information (NCBI). Each individual was assigned a specific ID number. The paralarvae and the statoliths subsequently extracted from them were labelled accordingly.

Measurements and statolith extraction

All the individuals identified that were not damaged (n = 233) were fixed and measured. The dorsal mantle length (DML) was measured (to 0.05 µm) (Fig. 2a) after fixation following Roper and Voss (1983), using a LEICA M205 stereomicroscope and LEICA Application Suite (LAS) image analysis system. Statoliths were removed by dissection under a stereomicroscope with fine dissecting needles of 20 µm of tip diameter. Statoliths were obtained from a total of 264 damaged and undamaged individuals. Each statolith was placed on a microscope slide covered with Crystalbond. The statolith rings were clear enough to estimate the age with confidence without grinding and were observed under a Nikon ECLIPSE 80i microscope connected to a Nikon DS-Fi camera. The software Nikon NIS-Elements D was used to count the rings manually and measure the statolith length (SL) (Fig. 2b). The reliability of ring counts was checked by having two replicate counts for all the individuals. Hatching dates for each individual were back-calculated from the capture date, assuming daily increment formation in the statoliths. Photographs were taken to permit description of the statolith morphology in the paralarvae of the three species.

Fig. 2
figure 2

Measurements of the paralarval squids (a) and statoliths (b): DML, Dorsal mantle length and SL, statolith length

Full size image

Only paralarvae with both an estimated age and an undamaged mantle (hence also a DML measurement) were used for growth analysis. Out of 222 paralarvae, 96 paralarvae corresponded to Alloteuthis media, 84 to Loligo vulgaris and 42 to A. subulata.

For each species, DML versus age and SL versus age (i.e. with age as the explanatory variable), also SL versus DML (with SL as the response variable) were modelled using generalised additive models (GAMs; Hastie and Tibshirani 2017). GAMs permit visualising the relationships without imposing a specific mathematical form, providing information on which parametric growth model is likely to be most appropriate and also allowed us to test for differences between species and seasons. We also fitted linear, power, exponential, and logarithmic equations, to determine which type of equation gave the best fit, based on residual plots and the coefficient of determination (r2). We also tested the residuals for normality (based on the Shapiro test).

For the GAMs, to achieve a better approximation to a Gaussian distribution and homogeneity of variance, the variable DML was log-transformed. Differences in size at age between species were determined by joining together the data for all three species, then fitting GAMs which included a DML-species interaction (i.e. with separate smoothers representing the DML effect for each species (as well as including species as a factor) and comparing this model with a model without the interaction (i.e. with a single, common, smoother for the effect of DML for the three species) using an F-test. For each species separately, we then explored differences between seasons, spring (May–July) vs. autumn (September to November), using the date of capture for the GAM models of DML at age, SL at age and DML vs SL, by including season as an additional explanatory variable. Interactions between effects of season and the other explanatory variable in the model (i.e. DML or SL) were tested using the methodology described above for the effect of species. We also repeated the above analyses using information on the hatching month (back-calculated from the estimated age) rather than the capture month. All GAMs were fitted using R and RStudio software (http://www.rstudio.com/).

Growth rates

The growth rates of L. vulgaris, A. media and A. subulata were calculated for each 10-day interval (age class) for those combinations of species, age class with at least 8 individuals and for which a significant linear regression line could be fitted to DML versus age, based on the slope of the regression.

Following Forsythe and Van Heukelem (1987), growth rate is estimated as

  1. a.

    Instantaneous relative growth rate (G, % DML d−1)

    $$G= \left(\frac{ln {(\text{DML}}_{2})- ln {(\text{DML}}_{1}) }{{T}_{2}-{T}_{1}}\right)\times 100$$
  2. b.

    Absolute daily growth rate (DGR, mm d−1)

    $$\textit{DGR}= \left(\frac{{\text{DML}}_{2} - {\text{DML}}_{1} }{{T}_{2}-{T}_{1}}\right)$$

    where DML1 and DML2 represent the dorsal mantle length in mm at the beginning and the end of the time interval T2–T1 (in days). In the present analysis, these values were derived using the age and DML values from two points along the fitted regression lines—and DGR corresponds to the regression slope for DML versus age. Regression analysis was carried out using Minitab (Minitab Ltd).

Results

Genetic identification and distribution of paralarvae

In total, 278 out of 310 loliginid paralarvae were identified to species, of which 124 were Alloteuthis media, 55 A. subulata and 99 Loligo vulgaris (Table 1). Most of the A. media and L. vulgaris paralarvae were found along the southeastern coast of Galicia while A. subulata was more frequently collected in the northern waters of the region. As indicated in Table 1, most of the paralarvae were found during spring, at the beginning of the upwelling season. In addition, despite similar sampling effort, the number of paralarvae obtained in Galician waters decreased from 2017 to 2018 (Table 1).

Table 1 Number of paralarvae collected and identified genetically with COI in the North and South-west of Galicia during 2017 and 2018 in spring (S) and autumn (A) and in the different months
Full size table

Statolith morphology

The statoliths of A. media and A. subulata were very similar in shape and quite different from L. vulgaris (Fig. 3). In the youngest paralarvae (i.e. 3–4 days old), both Alloteuthis species had rounded and elongated statoliths with no defined rostrum, whereas L. vulgaris had larger statoliths with a pronounced fingertip shaped rostrum. All statoliths had a well-developed and smooth wing. As the paralarvae grew, the rostrum became more pronounced, and the wing started to form.

Fig. 3
figure 3

Comparison of statolith morphology of the three species of loliginid paralarvae captured off the coast of Galicia (NW Spain) at different ages. Scale bar 50 μm

Full size image

Age and growth of wild paralarvae

The DML of L. vulgaris paralarvae ranged from 1.61 to 6.01 mm, in A. media, the range was from 1.31 to 6.58 mm and in A. subulata, it was 1.64 to 6.95 mm. The paralarvae of L. vulgaris had the lowest maximum number of increments (NI), with NI ranging from 1 to 35, while individuals of A. subulata and A. media had NI of between 1 and 54. The statolith length varied from 0.12 to 0.35 mm in the case of L. vulgaris, from 0.08 to 0.41-mm for A. media and from 0.11 to 0.38 mm for A. subulata.

The number of paralarvae decreased with increasing size of the paralarvae. For A. media and L. vulgaris, the most abundant size class was between 2 and 3-mm DML (Fig. 4), making up 44% and 70% of the samples for the respective species. On the contrary, the most frequent size for A. subulata was between 3 and 4 mm, making up 36% of the total for this species. In all species, individuals from the larger size classes were found less frequently and the smallest size class was less abundant than next largest size class.

Fig. 4
figure 4

Dorsal mantle length (mm) distribution of the three species of loliginid paralarvae captured off the coast of Galicia (NW Spain)

Full size image

Mantle length versus age

GAMs confirmed that the dorsal mantle length (DML) versus age (NI) relationship was strongly positive but non-linear. Even after log-transforming DML, the fitted curves were non-linear. The GAM results also indicate differences between the species in the DML–NI relationship: the final model contained significant effects of both of age and species (P < 0.001 in both cases, Table 2, Fig. SI 1). There was no significant interaction between the two explanatory variables and the interaction term was dropped. Between-species comparisons from this model showed that estimated values of DML at age for A. media and A. subulata was lower than for L. vulgaris, but there was no significant difference between the two Alloteuthis species (Table 2).

Table 2 Results of generalised additive models fitted to DML versus age and species for the three species of loliginid paralarvae
Full size table

For all species analysed separately, when an effect of season was added to the GAM of DML versus age, both the main effect of season and its interaction with the effect of mantle length were non-significant. The same results were obtained if season was defined according to hatching month rather than capture month.

Of the four parametric models, the exponential model yielded the best fit for the DML–NI relationship (Fig. 5) and the estimated growth rate differed depending on the species. The exponential fits also suggested that Loligo vulgaris had a larger DML compared to Alloteuthis sp. at the same age, while A. media had the smallest size at hatching but grew faster than A. subulata, achieving a larger size from the 20th day of life (Fig. 5).

Fig. 5
figure 5

Exponential regression showing the relationships between the number of increments (days) and dorsal mantle length (mm) for Alloteuthis media (r2 = 0.720), A. subulata (r2 = 0.834) and Loligo vulgaris (r2 = 0.702) paralarvae

Full size image

Statolith length versus age and statolith length versus mantle length

Fitting GAMs for SL versus age (NI), combining data for the three species, there were again significant effects of both age and species. The age effect was non-linear. Results for the main effect of species revealed that the SL at age was significantly smaller in both Alloteuthis species than in L. vulgaris, although A. subulata did not show significant differences from A. media (Table 3). Adding an interaction term, there was a weakly significant interaction (P = 0.0284) between the effects of age and species, indicating differently shaped growth curves (Fig. SI 2). While A. media showed linear growth of SL, the growth curves in the other two species were non-linear. A. subulata showed linear growth in SL up to 30 days but the curve then flattened, while L. vulgaris showed an increasing rate of growth in SL between the ages of 10 and 30 days.

Table 3 Results of generalised additive models fitted to statolith length versus age and species for the three species of loliginid paralarvae
Full size table

In the GAM of SL versus DML and species for all three species together, there was a strong positive effect of DML on SL, with the slope of the relationship gradually decreasing in larger animals (Fig. SI 3). The main effect of species was non-significant (P = 0.254). Adding an interaction term revealed no differences in the curves for the three species (Table 4). Thus, the SL versus DML relationships is essentially the same for all three species.

Table 4 Results of generalised additive models fit SL in relation to DML for the three species of loliginid paralarvae
Full size table

Comparing the three parametric models, the relationships of SL with age (NI) and SL with DML were best described by linear equations for each species (Fig. 6a and b). While this sounds contradictory to the GAM results, it should be noted that the linear model performed better than two specific non-linear models, i.e. nature of the nonlinearity detected by the GAMs was not well represented by the non-linear parametric models tested.

Fig. 6
figure 6

A linear regressions fitted to statolith length (mm) versus the number of increments (days) for loliginid paralarvae species: Alloteuthis media (r2 = 0.766), A. subulata (r2 = 0.891) and Loligo vulgaris (r2 = 0.604). b Linear regressions fitted to statolith length (mm) versus dorsal mantle length (mm) for A. media (r2 = 0.927), A. subulata (r2 = 0.897) and L. vulgaris (r2 = 0.903)

Full size image

Growth rate (G) and daily growth rate (DGR)

Estimates of instantaneous relative growth rate (G, as % of lnDML d−1) and the overall average daily growth rate (DGR) for the three species overall were quite similar (in the range 2.3–2.7% and 0.072 to 0.084, respectively), although it should be noted that the L. vulgaris samples include a narrower range of ages (maximum age 35 days) compared to the other species. In L. vulgaris, it is apparent that growth rate increased over the first 35 days of life, with the highest values (4.37% and 6.4 mm d−1) seen in the 21–35 days old group. Although growth rates appear to have been slightly higher in autumn, the autumn sample also contained more older animals. In both Alloteuthis species, again there was evidence that growth rate increased in older paralarvae, being higher in the 21–30 day age group than in the 1–10 day age group. No estimates could be obtained for the 11–20 day age group in these species, due to small sample size in A. subulata and highly variable data in A. media (Table 5).

Table 5 Dorsal mantle length (DML) growth rates, based on linear regression for DML (mm) versus age (NI, days) for each age range (10-day groups: 1–10, 11–20 days, etc.) and season for L. vulgaris, A. media and A. subulata paralarvae
Full size table

Hatching season

The average age of A. media paralarvae increased from May to September, when the oldest individuals were collected (Fig. 7). From September the mean age fell to November. On the contrary, L. vulgaris showed a decreasing trend in average age from May to September–October, when the youngest paralarvae were collected, followed by a slight increase to November. Most of the paralarvae of A. subulata (Table 1) were caught in the north of Galicia during June when a wide range of ages were observed. The mean age of A. subulata in June was similar to November.

Fig. 7
figure 7

Box and whiskers plots for the age of paralarvae of each species in each month: 5, May; 6, June; 7, July; 9, September; 10 October; 11, November. The horizontal line in each box indicates the mean

Full size image

Figure 8 depicts the frequency distribution of hatching months for each species, calculated from the collection dates and ages of each paralarva. The paralarvae of L. vulgaris and A. media show two seasonal hatching peaks, one in late spring–early summer and another peak in late autumn. During the autumn season, hatchlings of A. media hatched 1 month before those of L. vulgaris. Finally, A. subulata had the earliest hatching peak in spring, before the other two loliginid species. The absence of a clear second hatching peak in A. subulata may be because it was found mainly in the north of Galicia and autumn samples were only collected in the southwest Galicia.

Fig. 8
figure 8

Frequency distribution of hatching months for the three species of loliginid paralarvae (values expressed as percentages). Months corresponding to: 4, April; 5, May; 6, June; 7, July; 8, August; 9, September; 10 October; 11, November

Full size image

Discussion

Size, age, statolith morphology and hatching season

All three loliginid species were caught in both spring and autumn months, although few A. subulata individuals were taken in autumn. The size of L. vulgaris paralarvae was slightly bigger than that of Alloteuthis media and A. subulata. The sizes of paralarvae of L. vulgaris from this study were smaller than those reared in laboratory (Von Boletzky 1973; Turk et al. 1986), whereas hatchlings of A. subulata cultured—up to 10 days—were of similar size as those obtained in this study (Lipinski 1986; Yau 1994). Previous culture experiments with paralarvae of L. forbesii reported mean hatching sizes between 3.40 and 4.90-mm mantle length (ML) (Hanlon et al. 1987), which is double the size of L. vulgaris found here. Comparisons between our study and Gonzalez et al. (2010) suggest that the lowest values of DML obtained in their study (1.26-mm) might correspond to A. media and not L. vulgaris. It is also worth noting that the oldest paralarvae in both Alloteuthis species were somewhat older (50 days or more) than the oldest L. vulgaris paralarvae, possibly indicating that pelagic phase of the latter species is shorter or higher net avoidance in larger individuals.

The paralarvae of L. vulgaris presented the most developed statolith with a well-defined lateral and dorsal dome and a long rostrum at different ages compared to Alloteuthis spp. These differences were also mentioned by Arkhipkin and Nekludova (1993) who described statoliths in adults of Alloteuthis sp. with lateral and dorsal domes that were poorly developed compared to those of Loligo sp. statoliths.

Back-calculating hatching dates based on age, it is also evident that eggs of all three species hatched in both spring and autumn. The paralarvae of L. vulgaris and A. media reflected two reasonably well-differentiated hatching peaks, one in late spring-early summer and another peak in late autumn. A hatching peak for A. subulata occurred in spring, apparently earlier than in the other two loliginids, and a second peak that was almost negligible in our data due to the very low number of paralarvae of this species in the autumn samples.

Our data suggest that two hatching peaks exist for these three species: a more prominent peak at the beginning of the upwelling season and another less pronounced at the end of the upwelling season. These peaks coincide with the results obtained in Gonzalez et al. (2010) even though they considered that all the loliginid paralarvae examined were L. vulgaris. Bettencourt et al. (1996) reported similar results for L. vulgaris, indicating a main hatching peak in spring and a less defined peak in autumn. Considering that the embryonic development of L. vulgaris lasts between 25 and 70 days (Villanueva 2000b), there is a probably a main spawning peak in late winter to early spring and a second, smaller, peak in late summer-early autumn. This agrees with Rocha and Guerra (1994), who noted that L. vulgaris spawns in Galician waters all year round but with intensive spawning activity between December and April.

Moreover, A. subulata has an embryonic phase of 2 to 3 weeks (Jereb et al. 2015) which means that the main spawning peak would take place in late winter with a secondary peak in autumn. Alloteuthis media would spawn during spring and summer, if we assume the same duration for the embryonic development as in A. subulata. Spawning activity for A. subulata was reported to be in June in Portuguese waters (Moreno 1993) and slightly earlier in west Saharan waters, between March and May (Arkhipkin and Nekludova 1993), which coincides with our results. The fact that the second hatching peak of A. subulata is almost negligible could be because they are mainly found in the northern waters of Galicia (García-Mayoral et al. 2020) and sampling from that area was carried out only during June. Another possible explanation could be their short life span, of six months, as calculated by Arhkipin and Nekludova (1993).

Even so, our results showed that the two Alloteuthis do not coincide in hatching times and also differ in their spatial distribution. Alloteuthis subulata is more abundant in the north of Galicia while A. media is more abundant in south Galicia (García-Mayoral et al. 2020), which might be a way to diminish interspecific competition. This should be considered for future studies since the oceanography of the two areas is markedly different (Arístegui et al. 2009; Bode et al. 2009), which would be expected to influence the spawning and hatching time as well as the growth rates, even if no significant difference in growth rate between areas was detected in the present analysis.

The presence of two hatching peaks could indicate the occurrence of two distinct cohorts. Population structure of several loliginid species can be defined by the presence of two cohorts with different hatching seasons, which could mean that squids with different length at the same age will be found (Markaida et al. 2004), albeit not simultaneously; if the cohorts coexist at some point during the year they will clearly be at different life stages.

Growth patterns

Statistical analysis of relationships between dorsal mantle length (DML), statolith length (SL) and age (SI) in paralarvae of the three loliginid species suggested that growth in DML and SL is non-linear and, as might be expected, growth is more rapid in the (larger) L. vulgaris than in the two (smaller) Alloteuthis species. Arkhipkin and Nekludova (1993) reported that Alloteuthis spp. are among the slowest growing of the squids in the family Loliginidae. The DML–SL relationship did not differ significantly between species.

No seasonal differences in growth of DML could be detected despite likely differences in environmental conditions between spring and autumn, possibly because other sources of variation in conditions experienced by paralarvae (e.g. between years and between locations) mask any seasonal pattern.

The best representation of paralarval growth in terms of DML at age among the parametric models tested was an exponential regression, for each of the three species of loliginids. Other studies reported exponential growth during the early life stages of squid (first phase) (Villanueva 2000b; González et al. 2010) but, after the first 3 to 4 months post-hatching (second phase), the growth slowed down and could be represented by either power or exponential functions (Rodhouse and Hatfield 1990; Bettencourt et al. 1996; Rocha and Guerra 1999; Hatfield 2000; Hatfield et al. 2001).

It has been suggested that the relationship between SL and age could be useful to reveal information on squid growth without the need for counting the increment number of statoliths (Jackson 1994) and, therefore, also avoiding a possible underestimation of the age. González et al. (2010) had hypothesised that underestimation of age in adults could be possible due to statolith rings post-hatching getting closer to each other as the individual grows, making it difficult to count the increment numbers. Previous studies revealed that temperature influences statolith growth and, in consequence, increment deposition. Low temperatures cause statolith growth to slow down and thus, the distance between daily rings will be narrower and the rings will be more difficult to distinguish than those rings laid down during high temperatures (Villanueva 2000b; Keyl et al. 2011). However, in the present study, the R2 values for the relationship between statolith length and age in days were relatively low (see Fig. 6a) and Moreno et al. (2012) found no significant relationships between the number of increments post-hatching and the statolith size in juveniles and adults of L. vulgaris.

For those subsets of data with sufficient sample size and in which a DML-age regression had a non-zero slope, there was evidence in all three species that growth rate (expressed as G or DGR) was higher in older paralarvae. Comparisons between species and seasons were inconclusive. Overall, loliginid paralarvae displayed notable variability in their growth rate values, both within and between species. This high variability meant that growth rate could not be estimated for some age classes, since no significant regression line could be fitted. This contrasts with the relatively stable growth patterns observed in adults. This variability in paralarval growth is likely to be influenced by the high nutrient input and primary production provided by the Canary Current (CanC), especially in the upwelling season as well as the associated variation in temperature (Rocha 1999; Rocha and Guerra 1999; Villanueva 2000b; Forsythe 2004; Arístegui et al. 2009; Moreno et al. 2012).

The previous study in of L. vulgaris in Portuguese waters by Moreno et al. (2012) indicated that those paralarvae that hatched at the beginning of the upwelling season (from May to August) also known as the “warm cohort”, sensu Moreno et al. 2012) have higher growth rates caused by the warmer average sea surface temperatures (SST), and a planktonic stage lasting only two months. Squid hatched at the end of the upwelling season (from September to November, also known as the “cold cohort”, sensu Moreno et al. 2012) will experience colder average SST and lower growth rates, spending up to three months in the paralarval stage (Moreno et al. 2012). Culture experiments with paralarvae of L. vulgaris showed that both measures of growth (G and DGR) in length under summer conditions were double than those obtained under winter conditions (Villanueva 2000b). Turk et al. (1986) obtained high growth rates in reared L. vulgaris, which they attributed to the high diversity and abundance of food available and to the warm temperatures during the experiments. In wild populations, adults of L. vulgaris had higher growth rate values in summer than in winter (Arkhipkin 1995).

Differences in seasonal growth at age were also related to temperature in other squid species such as Doryteuthis gahi, which had higher growth rates at higher temperatures during the early stages of development (Hatfield 2000), and Photololigo edulis, which showed higher growth rates in the warm season than in the cold season (Natsukari et al. 1988). Dosidicus gigas had lower growth rates during cold seasons (Markaida et al. 2004). Miyahara et al. (2006) noted that, in addition to temperature, food availability was another factor that affected early-stage growth in Thysanoteuthis rhombus.

It should be noted that in the present study, paralarvae were sampled over 2 years and at a number of different sites and that we joined data from several months to represent “spring” and “autumn” seasons due to the rather low density of paralarvae in the plankton samples (possibly related to low catchability) and hence small sample sizes. Ideally, further work with a considerably larger sample size is needed to fully understand variability in growth rates.

Since temperature affects growth rate, it is also likely that it will influence the maximum size reached by mature squid. Thus, information on temperature could be useful to predict the size at the squid will be caught by fisheries (Hatfield 2000) as well as their abundance (Moustahfid et al. 2021). Thus, further studies on temperature influences on the growth of paralarvae would also potentially be useful for fisheries management.