Biology of Sanzo ’ s goby , Lesueurigobius sanzoi ( de Buen , 1918 ) , off Cabo Raso , Portugal *

The presence of Sanzo’s goby, Lesueurigobius sanzoi (de Buen, 1918), off Cabo Raso, Portugal, is recorded for the first time. The biology of this population was studied using 11 samples collected between April 2000 and September 2001 at depths of 70-100 m on muddy and sandy bottoms. The females reach a standard length of 110 mm and the males 103 mm. It attains an age of 24 months, which equals two breeding seasons. The otoliths form two translucent rings each year, which corresponds to the winter growth and the reproduction period. Sanzo’s gobies can be classified as “premature” and “standard” individuals, according to whether the first breeding occurs before or after the first winter respectively. “Standard” individuals, which have an extended first growth, reach a longer length than “premature” individuals, whose first growth is much shorter. Although discrete, sexual dimorphism was noticeable even outside the spawning season. The breeding season occurs between May and September/October with more than one spawning. The minimum length for which the maturity stage occurs is 64 mm, at an age of 0+. The feeding activity is quite constant throughout the year, and the diet consists predominantly of small crustacea and molluscs, although polychaets also feature commonly in its diet.


INTRODUCTION
Gobies are among the most successful teleost groups in temperate and tropical waters, and form the dominant taxa in the small fish fauna of benthic habitats (Fouda et al., 1993).According to several authors (Fonds, 1973;Gros, 1980;Nash, 1982;Miller, 1984;Magnhagen and Kvarnemo, 1989;Nellbring, 1993;Ota et al., 1996), the reproductive behaviour of gobioid fishes is very complex.Females deposit demersal eggs on the ceiling of the nest constructed by the territorial male in the muddy substrate or under empty shells.Brood care is usually done by the male parent until hatching.The majority possess a long breeding season with several spawnings per individual, after which they usually die (Miller, 1984).
The distribution range of Sanzo's goby extends from the south coast of Portugal to Mauritania in the eastern Atlantic and into the Alboran sea in the western Mediterranean (Miller, 1986).Sanzo's goby, Lesueurigobius sanzoi (de Buen, 1918) was studied in the 1980's off northern Namibia, on the west coast of Africa (Lloris et al., 1984;Olivar, 1989).
In Portugal, there is a considerable lack of knowledge about the biology, ecology and ethology of this species, essentially due to the null commercial value of this species, which is usually discarded by fishermen.
This study, besides being the first record of L. sanzoi off Cabo Raso, presents some aspects of the species biology, such as its reproduction, age and growth, and feeding behaviour.This new knowledge will allow other studies on the ecology and ethology of Sanzo's goby, and even a deeper study on its biology.

MATERIAL AND METHODS
Samples were obtained off Cabo Raso (Fig. 1) from the by-catch of a commercial otter trawler at 70 to 100 m depths.The study area presents bottom sediment composed of mud, sand and sandy mud, with a few rocky spots.This is an oligotrophic area, with a mean air temperature ranging from 9 to 23ºC, and a mean water temperature ranging from 14 to 19ºC.Salinity varies between 35.000 and 36.000, and the pH between 8.0 and 8.2 (Anonymous, 2000).All fish were caught with a 40 m long otter trawl with a 65 mm knot to knot mesh size between April 2000 and September 2001.Due to bad weather conditions and some logistic problems, it was not possible to obtain samples on a monthly basis.
A total of 915 specimens were caught during the 11 sampling sessions.The fishes were put on ice aboard and frozen on arrival at the laboratory.The laboratorial analysis procedure consisted in the observation of the external morphology, measured to the nearest 1 mm (standard length, SL), after which they were eviscerated and weighed to the nearest 0.0001 g (eviscerated weight, EW).After dissection, gonads and gastrointestinal contents were removed and weighed separately to the nearest 0.0001 g.Sagittal otoliths were removed from all fish, cleaned from the adjacent tissues and stored in 96º alcohol for at least two weeks.After polishing, total otolith radius (OR) and each growing ring radius were measured using a dissecting microscope with incident light.
A total of 472 otoliths were successfully read to study the population age structure and the species growth.Based on the method described by Bagenal and Tesh (1978), the condition factor (K) and the theoretical length of each fish was back-calculated using the relationship between SL and OR.Backcalculated mean standard lengths were compared with a t-test (Sokal and Rohlf, 1995).
The fish were sexed and the gonads were classified into six stages of maturation (N=509) based on Miller's criteria (Miller, 1961), and the oocytes were measured using micrometer, which was adapted to the eyepiece lens of the magnifier.The gonadosomatic index (GSI) was determined as the ratio of gonad weight to EW.The sex ratio was established as the ratio of the number of females to the number of males for all mature individuals (i.e.SL ≥ size of first maturity).This comparison was based on a Gtest of independence.
The total gastrointestinal contents were removed from 508 Sanzo's gobies.The proportion of sexes in the monthly samples was compared using a G-test of independence (Sokal and Rohlf, 1995) and food items were counted and identified to the lowest taxonomic level possible.The frequency of occurrence (FO), the numerical frequency (Fi), the vacuity index (VI) and the fullness index (FI) were estimated according to Hyslop (1980) and Assis (1992).The comparison of the monthly fullness index and the proportion of empty stomachs in each of the monthly samples were performed using a Kruskal-Wallis test and a G-test of independence respectively (Sokal and Rohlf, 1995).Due to the small size of most preys and to the fact that some of the food items were already partially digested, it was decided to use only a numerical approach to characterise the species diet.The degree of overlap between diets during reproduction and non-reproduction periods for males and females was calculated using the Shorigin index (Wallace, 1981).The diet diversity was also estimated using the Shannon index (Ludwig and Reynolds, 1988).

Age and growth
The otoliths of L. sanzoi are large, rhomboidal in outline and plano-convex in section, and exhibit translucent and opaque rings.In practice, after polishing the otoliths in this species are normally read with ease.However, some otoliths were illegible, and a few exhibited structural variations, such as excrescences on the otolith edge or an uncommon shape.Nevertheless, excluding two cases, these differences only occurred in one otolith of the pair.
By recording the type of edge observed on the otolith, it was noticed that throughout the sampling period there were always individuals presenting a translucent edge, even in summer (Fig. 2).According to the otolith edge type and the month of capture (Fig. 3), the individuals were classified as premature, i.e. fish that spawn before the first winter, or standard, i.e. fish that spawn after the first winter.Using this classification it was possible to form six groups of individuals based on the type of edge and BIOLOGY OF SANZO'S GOBY 287 the number of translucent rings present in the otoliths: 0+: premature individuals aged 0+ with a complete translucent ring (spawning mark) and translucent edge (winter growth), which corresponds to the individuals captured between January and April; I: standard individuals aged 1+ with a complete translucent ring (winter growth) and a translucent edge (spawning mark), captured between May and October; II: premature individuals aged 1+ with 2 complete translucent rings (1 st spawning mark + winter growth) and a translucent edge (2 nd spawning mark), captured between May and October; III: standard individuals aged 1+ with 2 complete translucent rings (winter growth + 1 st spawning mark) and a translucent edge (2 nd winter growth), captured between January and April; IV: premature individuals aged 1+ with 3 complete translucent rings (1 st spawning mark + 1 st winter growth + 2 nd spawning mark) and a translucent edge (2 nd winter growth), captured between January and April; V: standard individuals aged 2+ with 3 complete translucent rings (1 st winter growth + 1 st spawning mark + 2 nd winter growth) and a translucent edge (2 nd spawning mark), captured between May and October.
The back-calculated mean standard length (SL-) attained by males and females at the first winter check, for both premature and standard individuals, showed no significant differences between sexes (Table 1).However, between premature and standard individuals there was a significant difference in the SL-.
Due to the small number of individuals that reached the second year of life (N = 8), it was not possible to make a thorough analysis of the data.Apparently, only standard individuals showed three summer growing rings in their otoliths.
Condition factor (K) showed a slight decrease in October and attained a minimum in January.The increase in K observed in April reflected a gain in the fish condition prior to the breeding season although in July, when the fishes spent a considerably amount of energy on gonad maturation and reproductive behaviour, K also showed a low decrease (Fig. 4).

Reproduction
According to our observations, all males possess more coloured anal and caudal fins and all females possess soft grey pigments near the genital opening.This shows a smooth sexual dimorphism independent of the reproductive period.
Six stages of maturation were reported (Table 2).According to the monthly variation in maturity stages, males and females start ripening in April and spawn from May until September/October (Fig. 5).Stage IV appeared in April for both sexes, and males and females in stage IV/VI first appeared in May and increased their number in June.From April to September, a great number of individuals were at a ripening stage.All males collected in October were at spent stages, and in January all individuals were at recovering stages.The greatest number of immature individuals appeared in March and April.L. sanzoi showed a breeding season that extended from May to September/October, with a maximal gonad activity during June.The mean GSI (Fig. 6) attained a minimum in January, which was a rest period.During spring and summer, the fish invested in gonad development, which resulted in an increase in the mean GSI values.
The samples collected in September included two premature females (SL = 64 mm) and one premature male (SL = 65 mm) aged 0+ exhibiting mature stage IV/VI.These fish possessed the minimal length for which the maturity stage was ≥ IV, so 64 mm was considered to be indicative of the dimension of the first spawning for premature individuals.In fact, 82% of the females sampled during the sampling season with a SL between 60 and 65 mm were mature specimens, or showed ovaries in an advanced stage of maturation.
BIOLOGY OF SANZO'S GOBY 289  Statistical analysis by G-test of independence showed that the proportion of males and females did not vary between the samples and was close to 1:1 ratio in most of the months (Table 3).

Feeding ecology
Monthly variation of the vacuity index (VI) and fullness index (FI) showed that there were some fluctuations during the sampling period (Fig. 7).According to the G-test of independence, during the first and the last month of the breeding season, May (G W = 2.582; df = 1; P = 0.6170) and October 2000 (G W = 1.131; df = 1; P = 0.7216) respectively, total VI reached the higher values.Only in July 2000 did a significant difference occur between males and females VI (G W = 7.181; df = 1; P < 0.05).In January 2001, only males had empty stomachs, but due to the small sample size the differences were not considered significant (G W = 0.548 df = 1; P = 0.6858).Throughout the year all individuals had food in their guts.According to the Kruskall-Wallis test (KW) used to compare the monthly FI (KW = 10; df = 10; P = 0.44), there were no significant differences considering only the part of the population that was feeding.
A total of 44 different food items were identified in the gastrointestinal contents of the analysed Sanzo's gobies (Table 4).Most items were identified as belonging to the Crustacea class (Fig. 8).This group is doubtless the most important, representing 66.7% of all food items (Fi) and occurring in 83.3% of the analysed gut contents (FO) (Table 4).Polychaeta and Mollusca exhibited relevant values concerning Fi and FO.Shannon diversity index (Fig. 9) showed the lowest diversity among ingested preys in April and May 2000, and between January and April 2001.During the breeding season, males and females showed differences in their diet behaviour (Table 5).Using a G-test of independence (Table 5), it was observed that the three taxa that did not show significant differences between sexes were Bivalvia, Gammaridea and Crustacea n.i.The degree of diet overlap was calculated for males vs. females during reproduction and non-reproduction periods (Table 6).It can be observed that the minimum overlap occurred for males in the reproduction period vs. non-reproduction period.The maximum overlap occured for the non-reproduction period between males and females.

DISCUSSION
The occurrence of Sanzo's goby off Cabo Raso, on the west coast of Portugal at depths of 70 to 100 m in mud and sand bottoms could mean that the distribution of this species has been spreading northward in the last years, following the trend of other southern fish species, probably as a result of a global change in the climate (Cabral et al., 2001).
Contrary to what would be expected for temperate climates (Bagenal and Tesh, 1978), Sanzo's goby forms two translucent rings each year in its otoliths.Comparing the relationship between the breeding season and the deposition of the translucent ring, it was observed that there were two major peaks of individuals with translucent rings, which corresponded to two low values of the gonadosomatic index (GSI).It was also clear that the GSI peak in June 2000 resulted in a high percentage of individuals with translucent edges in July.On the other hand, in January the high percentage of individuals with translucent edges was mostly due to the slow winter growth.The winter ring was formed due to the slow growth rate conditioned by adverse abiotic factors, and the spawning ring was probably due to physio-BIOLOGY OF SANZO'S GOBY 291 logic changes during this period, which conditions otolin mineralisation (Härkönen, 1986).It is possible that although some individuals formed a translucent ring during spawning, they did not actually get to spawn.Sanzo's goby attains a maximum life span of 24 months, corresponding to two breeding periods.Nevertheless, the number of individuals that can reach the second breed was low, since most of them die after the first breeding season.These results are far different from the ones for Fries's goby, which reaches a maximum age of 9 years in the Firth of Clyde (Nash, 1982), and 11 years in the Lynn of Lorne (Gibson and Ezzi, 1978), both in Scotland.The Sanzo's goby population studied was divided into two types of individual, standard and premature.Standard individuals first spawned after the first winter aged 1+, whereas premature individuals first spawned in the last months of the breeding season (September/October), a few months after hatching, aged 0+ and with a minimum length of 64 mm.Theoretical mean growth between premature and standard individuals was different, in spite of an identical growth for both sexes within each group.Premature individuals reached longer lengths than standard ones.As premature individuals hatched at the beginning of the breeding season, they had a first fast growth period longer than that of standard individuals, which hatched late during the breeding sea-son and reached the next spring with shorter lengths than premature ones.In hypothesis, premature individuals' descendents, which will hatch in the last months of the breeding season, will be standard individuals that will spawn during the next breed.
There was a numerical difference between the total number of premature (N = 113) and standard (N = 307) individuals captured.Apparently, a cohort that hatched at the beginning of the breeding season should have greater success because this period has more favourable abiotic conditions, and there is far more food available.Therefore, it is possible that premature individuals have a higher mortality rate or that their progenitresses (standard individuals) have a lower fecundity.The largest fish caught was a oneyear-old premature female with 110 mm standard length (SL), which is larger than the maximum standard length of 70 mm for Namibia and 100 mm for Faro, south Portugal, recorded by Lloris et al. (1984) and the 95 mm recorded by Miller (1986).
Sanzo's goby is an oviparous and gonocoric species with secondary sexual dimorphism and a short life cycle.Ovaries are elongated and rounded, and testes have an elongated and triangle section.Due to the long breeding season, between May and September/October, and the great number of males and females apparently spent but actually "semispent" or with developing oocytes, it can be considered that L. sanzoi has more than one spawning per breeding season, like other Gobiidae species (Miller, 1961;Fonds, 1973;McEvoy and McEvoy, 1992).According to Cole (1954), Sanzo's goby is an abbreviate iteroparous species.The ripe eggs with a diameter of 0.56-0.71mm are similar to Fries's goby eggs with a diameter of 0.5-0.6 mm (Gibson and Ezzi, 1978).
The number of reproduction cycles can be determinant in the life spawn of this species, as both pre-292 C. PAULO-MARTINS et al. mature and standard individuals showed two spawning marks on their otoliths and no fish showed signs of a third spawning.It can therefore be concluded that this Sanzo's goby population only reaches the second year of life and has two breeding seasons during its lifetime, after which they die.This L. sanzoi population had a sex ratio quite close to 1:1, which reveals there was no dominance of either sex and that males and females had an equal vulnerability to capture.A similar result occurs with L. friesii in the Firth of Clyde (Nash, 1982), although Gibson and Ezzi (1978) refer this same species as having a sex ratio in favour of females (female to male sex-ratio 1.2:1).A greater number of females would be expected, mainly during spawning, when males are guarding the nest.It is possible that the samples were collected in an area far from the nest sites, or that the trawl was destroying the nest as it passed.
Feeding activity went through some fluctuation during the year.In May 2000 some individuals were not feeding.This behaviour was possibly due to the males' reproductive behaviour, since a significant number of males showed a high VI in comparison with females.Also, in October 2000 both sexes showed high values of VI, suggesting that the end of the breeding season induces a decrease in the feeding activity of this species, or that the number of available preys was lower at this time of the year.Nevertheless, all the individuals kept feeding during the sampling period, showing a feeding activity that was quite constant throughout the year.
The gastrointestinal content analysis showed that Sanzo's goby fed mainly on Polychaetes, Crustacea Gammaridea and Crustacea Caridea, although it was considered as a generalist species because none of the preys was consumed in quantities exceeding 50%.There was no major fluctuation in prey ingestion during the year, although during the reproductive period diversity increased, suggesting that there was a higher diversity of preys available during this period.In June 2000 and March 2001 maximum diversity showed that Sanzo's goby turned out to be more specialist, ingesting a greater number of Phylocheras sp., Gastropoda and Gammaridea respectively (unpublished data).Once this species has a peculiar reproductive behaviour, it induced some consequences in the diet of both sexes.Considering the need to spend high levels of energy on the primary reproductive effort (Miller, 1984), females would be expected to consume more food than males.During breeding, males preferred molluscs to crustaceans.It is likely that the great stress males suffer due to the territory and nest guard during breeding affects their ability to seek preys.Thus, males eat more molluscs because they are easier to hunt, due to their epibenthonic way of life and weak capacity to make sudden moves in order to escape from a potential predator.Males would be expected to ingest a similar number of Bivalvia and Gastropoda.Therefore, the differences found could result from the fact that Gastropoda show a higher mobility than Bivalvia, and therefore it is possible that their densities in the males' territories are kept more or less constant.
FIG.2.-Variation of translucent edge mean width (TEW) during the sampling period.J "premature" individuals and I "standard" individuals.
FIG. 4. -Variation of the condition factor (K) during the sampling period.Vertical bars report the standard deviation.

TABLE 3 .
-Proportion of males and females in each monthly sample and corresponding mean SL; M SL min, male minimum standard length; M SL max, male maximum standard length; M SL -, male mean standard length; F SL min, female minimum standard length; F SL max, female maximum standard length; F SL -, female mean standard length; G W , G statistics with Williams correction; df, degrees of freedom;G H , heterogeneity G; G T , total G; P, significance level.

TABLE 4 .
-Diet of L. sanzoi based on 1708 preys identified on a sample of 477 full gastrointestinal guts FIG. 9. -Monthly variation of diversity (Shannon index) (full circle and maximum diversity (H' max ) (bars).

TABLE 5 .
-Preys' numerical frequency during breeding season; FN, number of i preys in females' gastrointestinal contents; FNr, total number of preys (except prey i) in female gastrointestinal contents; MN, number of i preys in male gastrointestinal contents; MNr, total number of preys (except prey i) in male gastrointestinal contents; G Y , G statistical with Yates correction; df, degrees of freedom; P, significance level; *, P < 0.05.