Spatial and temporal patterns in size and maturation of Loligo plei and Loligo sanpaulensis ( Cephalopoda : Loliginidae ) in southeastern Brazilian waters , between 23 ° S and 27 ° S

Patterns of population dynamics of Loligo plei and Loligo sanpaulensis in southeastern Brazil were investigated with samples obtained from commercial catches and research cruises from 1999 to 2000 and from 2002 to 2003. Size and maturity structure of the two species varied according to depth and season. Body size decreased with depth for L. plei, whereas for L. sanpaulensis size increased with depth up to 100 m and decreased again in deeper areas. GSI and incidence of mature animals decreased with depth in both species. L. plei females matured at a larger size in summer, while size at maturity in L. sanpaulensis was greater in autumn. For L. plei, reproductive events occurred in the late winter and spring, in depths up to 40 m, and during summer in inshore waters. L. sanpaulensis increased reproductive activity in summer, winter and spring between depths of 30 and 80 m. The presence of high proportions of immature squid offshore suggests that juveniles might develop in these areas and, upon maturation, migrate back to inshore waters to spawn, particularly in spring and summer. However, differences found in aggregation patterns in different depth strata, as well as the seasonal differences found in the size structure of L. plei and L. sanpaulensis, could ultimately be due to changes in the water column related to food availability. Considering that L. plei and L. sanpaulensis are both caught as by-catch by shrimp trawlers throughout the year, we also discuss hypotheses on the life cycle of the species and its implications for fishery management. We recommend measures to protect squid spawning grounds by creating spring and summer protected areas where trawling would be prohibited until 60 m depth, i.e. in the area and time of year when mature individuals concentrate.


INTRODUCTION
Loliginid squid play a significant economic and ecological role in coastal marine ecosystems worldwide as many species are fishery resources and represent a key link in marine food webs (Pierce and Guerra, 1994;Santos and Haimovici, 2002;Boyle and Rodhouse, 2005;Pierrepont et al., 2005;Rodhouse, 2005;Staudinger, 2006).Along the southern Brazilian coast, Loligo plei (Blainville, 1823) and Loligo sanpaulensis (Brakoniecki, 1984) are the most abundant species of Loliginidae, and are fishery resources of local interest (Perez et al., 2005;Gasalla et al., 2005a,b).
Loligo plei is commercially caught from Cape Frio (22°S) to Cape Santa Marta Grande (28°S), and is the most important Loliginid species in the fisheries of the Brazilian states of São Paulo and Santa Catarina.In these areas, L. plei is mainly fished as a by-catch of the pink-shrimp fishing trawlers (Perez et al., 2005) and seasonally by coastal hand jigging (Gasalla, 2004;Gasalla et al., 2005b;Martins et al., 2006).Catches by all gears peak in summer.However, Loligo sanpaulensis is of secondary importance in commercial fishing in the southeastern Brazilian states of São Paulo and Santa Catarina, but is the main fished loliginid squid species in the states of Rio de Janeiro and Rio Grande do Sul (Gasalla et al., 2005a).
In Brazil, studies of the two species began during the 1970s through sampling undertaken in the FAUNEC survey, conducted by the University of São Paulo (Juanicó, 1979).Later, studies concentrated mainly on the L. sanpaulensis population along the coasts of Rio Grande do Sul and Cape Frio (Haimovici and Andriguetto, 1986;Costa and Fernandes, 1993;Andriguetto and Haimovici, 1996;Santos and Haimovici, 1998).In the mid 1990s, studies on squid population dynamics were initiated in Santa Catarina and São Paulo based on commercial fisheries, and since then studies have focused on Loligo plei (i.e.Perez and Pezzutto, 1998;Perez, 2002;Perez et al., 2002;Gasalla et al., 2005a,b).
In summary, previous knowledge has shown that both Loligo species have short life cycles and are semelparous.Males have a broader size range distribution and larger maximum sizes than females and their length-frequency distribution is approximately log-normal, while female lengths usually show normal distributions.In addition, some hypotheses on spatial patterns in size of maturation in relation to season, depth ranges and inshoreoffshore migration have been proposed (Juanicó, 1979;Andriguetto andHaimovici, 1991, 1996;Perez, 2002).One hypothesis suggests that there might be more than one population of L. sanpaulensis in Brazilian waters (Gasalla et al, 2005a).This hypothesis refers specifically to work carried out by Juanicó (1979) that showed four distinct groups with different morphological characteristics within L. sanpaulensis samples caught between Rio de Janeiro (22ºS) and Rio Grande (32ºS).The northern-most and southern-most groups seem to be much alike, and are separated spatially by the intermediate Santa Catarina (27ºS) group (Juanicó, 1979), which tends to suggest that the differences could be environmentally driven.
Considering that knowledge of the life history and population structure of a species is crucial for fishery management, this paper presents results of a study on spatial and temporal variation in size and maturation of L. plei and L. sanpaulensis along the southeastern coast of Brazil.

MATERIALS AND METHODS
The study area is located in the South Brazil Bight (SBB), spanning southern Rio de Janeiro state (23ºS) to Itajaí, in Santa Catarina state (27ºS).From August 1999 to December 2000, and between August 2002 and July 2003, L. plei and L. sanpaulensis were obtained as by-catch of pink-shrimp (Farfantepenaeus spp) trawl fishing, at depths of 30 to 100 m.Monthly or, when possible, weekly samples were collected at the ports of Santos and Guarujá.Size-stratified random subsamples (Sparre and Venema, 1998) were collected and, when catches were very small, all landed squids were sampled.Skippers were interviewed to obtain information about the fishery, such as location and depth of capture.In some cases, complementary information was obtained from landing records held in the Instituto de Pesca "ProPesq®" fishery database.Complementary data were obtained from demersal fish surveys of the REVIZEE National Research Program.In this case, specimens of L. sanpaulensis were collected with the bottom-trawler "Soloncy Moura" at depths ranging from 100 to 150 m in February, April and June 2002.
Mantle length (ML, mm) and total body weight (BW, g) were recorded in the laboratory.Sex and maturity stages were identified according to a scale proposed by Juanicó (1983) and adapted by Perez et al. (2002): (I) immature, (II) in maturation, (III) mature and (IV) spawners.Individuals that could not be sexed were classified as "juvenile".
Length-weight relationships (LWR) were calculated using the expression: BW = aML b , the parameters a (intercept) and b (slope) were estimated by non-linear regression.Analysis of covariance (ANCOVA) was used to test differences between the slopes of the LWR obtained for each sex.In view of the supposed existence of more than one population of L. sanpaulensis in Brazilian waters, the LWR was also calculated for two separate groups: (a) individuals caught between Santos (24ºS) and Itajaí (27ºS), and (b) those captured in the northern part of the studied area (between 23°50'S and 23°00'S).
Gonad weight (GW) and accessory reproductive organ weight (AOW) (spermatophoric Complex + Needham's sac + penis, for males, and oviduct + nidamental glands + oviducal gland, for females) were obtained.Appropriate standard indices of maturity and reproductive activity were tested before adopting the gonadosomatic index (GSI) (sensu Macy, 1982;Raya et al., 1999;Perez et al., 2002), which is defined here as: Commercial trawl data were used to describe the seasonality of the maturation process, expressed by the bimonthly variation in GSI values for males and females.The sex ratio was calculated monthly for the different depth zones, and significant deviations from 1:1 were tested using Chi-squared (χ²) tests (Zar, 1996).
Maturity ogives were estimated from the proportion of mature squids for each sex per size class, with a logistic curve (King, 1996).
The mean length at which 50% of the individuals are mature was calculated as ML 50% = -a/b, were "a" and "b" are the intercept and the slope of the plot between mean length and proportion of mature individuals.
Data from commercially trawled and survey samples were used to investigate the spatial and temporal distribution of population strata of L. plei and L. sanpaulensis.Mean values of ML and GSI of males and females, and the proportion of each maturity stage, were analyzed (a) throughout the year, and (b) from four different depth zones (A1: 30 to 39 m; A2: 40 to 60 m; A3: 61 to 80 m; and A4: 100 to 150 m).
Since not all length distributions displayed a normal distribution, a maximum likelihood estimate (Haddon, 2001) was used to test for differences in mean lengths throughout the year, and between four different depth zones.The "normal" likelihood function πσ σ μ was adopted for females, and a "log-normal" function was used for males, based on observed length frequency distribution characteristics, where x is observed ML and µ is the mean value (expected ML).Likelihood profiles were constructed to evaluate statistical differences between mean ML values (Hilborn and Mangel, 1997), where overlapping distributions indicate similar average ML values.

Size structure
A total of 1584 L. plei specimens were analyzed (Table 1), of which, complete depth and latitude data were available for 1341 individuals.Males ranged from 40 to 300 mm in mantle length, and reached greater sizes than females, whose length was between 30 and 180 mm.The overall length-frequency distribution for females was approximately normal whereas that for males was approximately log-normal (Fig. 1).
A total of 938 L. sanpaulensis individuals were analyzed, of which, 623 came from the fishery and 315 from fish research surveys (Table 2).Mantle length frequencies showed unimodal distributions for both sexes.Males reached larger sizes than females, and they varied from 25 to 163 mm and from 29 to 111 mm respectively (Fig. 1).

Length-weight relationships
The relationships between mantle length and body weight of male and female Loligo plei were BW = 0.00828ML 1.765 (r 2 = 0.92, n = 862) and BW = 0.00349ML 1.963 (r 2 = 0.84, n = 654) respectively.The slopes ("b") of the LWR differed significantly between the sexes (ANCOVA; F= 6.055, df 1, P=0.014) and the females were heavier at a given length than males.
The LWR calculated using all females of L. sanpaulensis showed a low coefficient of determination (r 2 = 0.28), which indicates that the regression did not fit the data perfectly.This was due to the fact that length and weight ranges varied, mainly in females, between squids caught between Santos and Itajaí, which showed larger sizes, and those captured in the northern portion, which were smaller.Thus, data were separated into two different sets, which resulted in higher r 2 values.Then, the lengthweight relationships for males and females for individuals caught between Santos and Itajaí were BW = 0.00409ML 1.954 (r 2 = 0.877, n= 114) and BW = 0.00117ML 2.289 (r 2 = 0.739, n= 138).For individuals caught in the northern area the relationships were BW= 0.00377ML 1.980 (r 2 = 0.867, n= 236) and BW = 0.00312ML 2.039 (r 2 = 0.524, n= 111) respectively.

Seasonal variation in length and length-atmaturity
Seasonal differences in mean ML of Loligo plei were observed; the smallest immature males and fe-males (stage I) were found during summer months, and the largest individuals were present during winter.Winter mature individuals (stage III and IV) were also smaller than those captured in summer.Mean lengths of immature and mature females showed strong differences between summer and winter, as shown by likelihood profiles, but these differences were not consistent among the males (Fig. 2).Females of L. plei matured at a larger size in summer (ML 50% = 139 mm) while males showed similar size at maturity in summer (ML 50% = 160 mm), autumn (ML 50% = 155 mm) and winter (ML 50% = 154 mm) (Fig. 3).
Mean ML in Loligo sanpaulensis was significantly different at different periods of the year, but only for mature individuals (Fig. 2).The largest immature males and females were found during winter, and in the case of mature individuals both sexes showed larger sizes during autumn.Immature and mature individuals showed lower mean ML during autumn and spring respectively (Fig. 2).Size at maturity in L. sanpaulensis was greater in autumn for both sexes: ML 50% was 82 mm and 119 mm respectively (Fig 3).

Reproductive index, maturation and sex ratio
A clear pattern of size and maturation in relation to depth was found for L. plei, as both the mean ML and mean GSI decreased with increasing depth (Fig. 4).This fact was confirmed by likelihood estimates that demonstrated similarities in body size only between depth classes A3 and A4 (80 to 150 m) for both sexes (Fig. 5).Two groups of mature individuals could be distinguished, one abundant and represented by larger individuals located inshore, and another, less abundant and with smaller speci- mens, located offshore in depths between 61and 150 m (Fig. 6).Significant variation in length in relation to depth was also confirmed by ANOVA (P<0.001).
The frequencies of occurrence of immatures and juveniles were highest in inshore areas, concentrated between 61 and 100 m, and during autumn and summer (Figs. 6 and 7).Mature L. plei individuals were found throughout the year (Fig. 7), with maturation peaks for both sexes between September/October and January/Feb-  ruary (Fig. 8).An ANOVA for monthly mean GSI values (Table 3) showed highly significant differences (P<0.001) in all cases.For males, little variation was observed in GSI, except in July/August 1999, January/February 2000 and 2003, and September/October 2002.Females showed a strong GSI peak between September/October 2000 (late winter and spring) and a secondary peak in January/February 2000 (summer).There were significantly more males than females (P<0.05) in the total sample.The predominance of males was significant in depths between 40 and 60 m (χ 2 = 10; P<0.05) and 61 and 80 m (χ 2 = 23; P<0.05).In months with GSI peaks the sex ratio was 1:1 (Fig. 8 and 9), followed by a trend of mature male predominance just after the peak (Fig. 9).
The ML and GSI of L. sanpaulensis increased with depth up to depths of 100 m, and decreased again with depths greater than 100 m (Fig. 4).Likelihood  estimates did not confirm differences in mean ML between depth strata A1 and A2 in females or between A2 and A3 in males (Fig. 5).The same pattern was observed for the mean length of mature individuals (Fig. 6).The highest percentages of matures were found at depths of 30 to 80 m, and immatures (stage I   and II) at depths greater than 100 m, especially during summer and autumn (Figs. 6 and 7).Mature individuals of L. sanpaulensis were present throughout the year in high frequencies (Fig. 7).There was limited variation in GSI data, mainly for males, but the GSI peaked in summer and winter for both sexes (Fig. 8).Highly significant differences (ANOVA; P<0.001) were recorded between months in all GSI data (Table 3) Males were significantly predominant over females in areas deeper than 40 m (A2, χ² = 5.9; A3, χ² = 35; A4, χ² = 12; P<0.05), and in almost the entire sample (χ² = 32; P<0.05).Seasonally, females showed predominance over males only in November 1999 (P<0.05), while males predominated just after the breeding peak in November/December 1999 and January/February 2003.The 1:1 sex ratio proportion was observed after the breeding peak in July/August 2002 (Fig. 9).

DISCUSSION
Sexual dimorphism in size is well known in loliginids, and males of many species show larger maximum sizes than females (Moreno et al., 1994;Pierce et al., 1994;Raya et al., 1999;Rocha and Guerra, 1999;Perez et al., 2002).This was clearly seen in the present study for both Loligo species.In addition, the slopes of the length-weight relationship and the weight per unit length were greater in females than in males, possibly because females invest more in the reproductive organs than in somatic growth (Forsythe and Van Heukelem, 1987).
Better fits in the L. sanpaulensis length-weight relationship were found when data were sorted into two different areas, which may be related to the occurrence of different stocks in the studied area, as already suggested by Juanicó (1979) and Gasalla et al. (2005a).
Our study showed that mature squids are caught as a by-catch of the pink-shrimp commercial trawling throughout the year, which suggests that spawning occurs in the fishing area all year round, with increases in reproductive activity in some seasons.The variation range of mean GSI values was narrow for L. sanpaulensis and for males of L. plei, but seems to be comparable with other studies (i.e.see ΔIGS in Guerra and Rocha (1994) for L. vulgaris and L. forbesi; Perez et al. (2002) for L. plei, and Olyott et al. (2006) for L. vulgaris).Keeping this in mind, L. plei showed slightly more reproductive activity during late winter and spring (September/October) as well as summer (January/February).Despite a gap in data for L. sanpaulensis, the periods of increased spawning activity for this species appear to be in summer and winter, which is also consistent with previous observations by Andriguetto and Haimovici (1996).
The 1:1 sex ratio observed in and around reproductive seasons suggests aggregation for mating in the studied area, similar to other cases (i.e.Barón and Ré, 2002).There are indications of higher postmating female mortality or migration as mature males predominated just after some reproductive seasons, which was also found by Augustyn (1990) and Perez et al. (2002).There was also a higher incidence of spawner males than females throughout the study period, which could be explained by the hypothesis of greater longevity of males, as suggested for L. sanpaulensis by Aguiar (2006) and for L. plei by Jackson and Forsythe (2002).
Temperature seasonality and food availability are the main factors that typically influence the growth of cephalopods (Boyle and Rodhouse, 2005).The seasonal variation and the segregation patterns of maturity, mean length and size-at-maturity in deeper areas found for L. plei and L. sanpaulensis in this study could be associated with seasonal differences in temperature and food availability observed along the southeastern coast of Brazil, due to a complex and dynamic hydrographic variation (Campos et al., 1995;Borzone et al., 1999).The influence of South Atlantic Central Water (SACW) is frequently detected in the continental shelf at the end of spring and summer, which causes changes in the water column temperature due to an influx of colder (less than 18ºC), less saline waters (Campos et al., 1995;Castro and Miranda, 1998).Due to the coastal upwelling of this nutrient-rich water mass, the primary productivity is higher, and this is reflected in higher trophic levels.During autumn and winter the SACW moves back to the upper slope, and in this period, the water column is homogenized on the shelf and a decline in biological production is observed (Campos et al., 2000).
Thus, seasonal environmental changes and food availability can be important factors that shape Loligo spp aggregation patterns, as previously noted for other loliginids worldwide (Hatfield and Cadrin, 2002 for L. pealeii;Arkhipkin et al., 2004 for L. gahi;Olyott et al., 2006 for L. vulgaris).Moreno et al. (2005) show that immature individuals of L. vulgaris exposed to more favourable environmental conditions (more available food and/or warmer temperatures) grow faster and these individuals attain the minimum size for maturity earlier.In individuals exposed to colder waters and lower food availability in the first life stages, maturity may be delayed and investment in growth prolonged, which would result in maturation at larger sizes.
Considering the hypothesis raised by Perez et al. (2006) that L. plei off southern Brazil might live up to around 9 months of age, it is possible that larger summer (January/February) spawners found in the present study could have been the progenitors of the smaller winter-spring (September/October) matures.Due to the possibly greater growth rates that immature individuals show during spring and summer (Perez, pers. comm.), it is likely that our late winter (September) spawners were the parents of larger mature individuals responsible for the reproductive event in summer.
Figure 10 summarizes the main hypotheses raised on displacement and life cycles of L. plei and L. sanpaulensis.L. plei showed aggregation for reproduction in shallow waters during the summer, and in greater depths during the spring and winter.In L. san- paulensis, reproductive events of summer were concentrated between depths of 30 and 80 m.The results suggest that juveniles of both species arising from spring and summer spawning migrate from inshore nursery grounds to deeper waters (between 100 and 150 m) in the northern zone of the studied area.Thus, the segregation patterns of size and maturity of L. plei show that larger-sized individuals and mature individuals concentrate closer to the coast, and smaller sized and immature squid concentrate in deeper areas.High concentrations of mature L. plei individuals inshore during summer might take advantage of the high temperature and high food availability associated with the SACW intrusion.However, since mature L. plei are caught throughout the year, some portion of the population may not always find these favourable conditions and still might be competent to breed.Such life cycle plasticity due to the combination of variable reproductive characters in individuals is the key to cephalopod survival in variable environments (Boyle and Rodhouse, 2005), and their life cycle is probably timed to take advantage of good conditions.
Compared with L. plei, an opposite size segregation pattern seems to occur for L. sanpaulensis, as the size of individuals and predominance of mature individuals increase with depth and decrease again in depths greater than 100 m.This has also been suggested by other authors in different sub-regions, such as Juanicó (1979) and Andriguetto and Haimovici (1991) based on oceanographic surveys off Rio Grande do Sul (southern Brazil) and Costa and Fernandes (1993) in Cape Frio (northern-most southeast Brazilian shelf).According to Barón and Ré (2002), immature L. sanpaulensis individuals in the north eastern coast of Patagonia (Argentina), migrate offshore during certain periods of the year, and return inshore when mature.
It should be emphasized that age determination is still necessary to confirm interpretations of the relationships between hatching season and the differences in length and size-at-maturity of seasonal groups, as found in temperate ecosystems (e.g.Boyle et al., 1995;Moreno et al., 2005).
Nevertheless, our results seem to suggest important sustainable management implications for squid populations.Firstly, considering that L. plei and L. sanpaulensis are caught as by-catch by pink-shrimp trawlers, broader strategies rather than single targetspecies management would be beneficial (i.e.multispecies approaches).However, in view of the costs of ecosystem management in the short term, simpler, focused measures could be more reliable.One possible suggestion for protecting squid spawning grounds would be to implement spring and summer protected areas where trawling would be prohibited until 60 m depths, i.e. in the area and time of year when mature individuals concentrate.Regulating trawling activities and broadening previously established seasonal closures for the shrimp fleet may allow potential squid spawners to escape, shrimp to recruit (mainly pink-shrimp), and other by-catch fishery stocks to recuperate.

Fig. 2 .
Fig. 2. -Likelihood profiles showing seasonal differences in mean mantle length (mm) for mature and immature males and females of L. plei and L. sanpaulensis.

Fig. 8 .
Fig. 8. -Bimonthly variation of the mean GSI with standard deviation for females and males of L. plei and L. sanpaulensis.

Fig. 10
Fig. 10.-Schematic diagrams of population displacement in different strata during spring-summer and autumn-winter of L. plei and L. sanpaulensis in the studied area.The dark zone indicates the intrusion and retreat of the South Atlantic Central Water (SACW).BC is the core of the Brazil Current; TW, Tropical Water and CW, Coastal Water.

Table 1 .
-Numbers of Loligo plei measured for each class, sex and maturity stage from samples taken by commercial shrimp trawlers in1999-2000 and 2002-2003.ML = mantle length.

Table 2
. -Numbers of Loligo sanpaulensis measured for each class, sex and maturity stage from samples taken by commercial shrimp trawlers in 1999-2000 and 2002-2003 and ground fish cruises of the REVIZEE program.ML = mantle length.

Table 3
. -Results of the ANOVA test for differences in GSI between months for L. plei and L. sanpaulensis data sets.n: number of squid sampled; d.f: degrees of freedom; MS: mean square