Introduction
The blackfin notothen, Trematomus scotti (Boulenger), is a relatively small species with a circum-Antarctic distribution, including the South Shetland and the South Orkney islands as the northernmost areas. It is distributed in nearshore and continental waters at depths of 20–800 m (DeWitt et al. Reference DeWitt, Heemstra and Gon1990). Adult populations of this species represent a key component of the demersal fish communities inhabiting the shallow waters around the Antarctic continent (DeWitt et al. Reference DeWitt, Heemstra and Gon1990). Along the western Antarctic Peninsula, its abundance progressively declines from Marguerite Bay to the Bransfield Strait (Daniels & Lipps Reference Daniels and Lipps1982), becoming rare in the southern Scotia Arc (Kock & Jones Reference Kock and Jones2005). On the other hand, the early life stages of T. scotti dominate the larval fish assemblages inhabiting the Bransfield Strait and the Palmer Archipelago with adjacent waters, especially in summer and autumn (Kellermann Reference Kellermann1989, Loeb Reference Loeb1991, Catalán et al. Reference Catalán, Morales-Nin, Company, Rotllant, Palomera and Emelianov2008). Early larvae of 8–16 mm standard length (SL) are caught in late January and February, the smaller ones (8–11 mm SL) with the yolk sac, attaining 12–19 mm SL between March and early April. Larval development extends until June, when dorsal and anal fin formation is complete and metamorphosis takes place (Kellermann Reference Kellermann1990). Pelagic juveniles of 27–41 mm SL are found in the southern Bransfield Strait from November to early February, indicating a relatively long pelagic phase ending with the juveniles’ demersal settlement at the end of the second summer after hatching (Kellermann Reference Kellermann1989).
As observed in many other notothenioids, the Bransfield Strait and adjacent waters represent an important area for larval retention of T. scotti, despite the fact that their adult counterparts probably spawn in neighbouring areas where they are widespread. In the Bransfield Strait, cold and saline coastal water from the Weddell Sea flows westward along the southern shelf and turns to the north and north-east, where it mixes with water from the Bellingshausen Sea flowing in from the west, forming a clockwise gyre (Capella et al. Reference Capella, Ross, Quetin and Hofmann1992, Hofmann et al. Reference Hofmann, Klinck, Lascara and Smith1996, Thompson et al. Reference Thompson, Heywood, Thorpe, Renner and Trasvina2009). The cyclonic water circulation has a great influence on the larval fish assemblage inhabiting the Bransfield Strait, which consists of low-Antarctic species at their southern limit of distribution and high-Antarctic species which are passively transported there from southernmost areas by the incoming water masses (Loeb et al. Reference Loeb, Kellermann, Koubbi, North and White1993). Larval dispersal in oceanic waters further away from the Bransfield Strait is largely reduced, limiting the spatial distribution in suitable habitats for those species which require the permanently cold waters of high-Antarctic shelves (e.g. Trematomus spp.).
The larval period is commonly recognized as the most critical ontogenetic stage, with numerous factors affecting growth and survival. High food seasonality, constant low temperature and strong ocean currents among other factors play fundamental roles in shaping distribution and growth patterns of early life stages of Antarctic fishes (Radtke & Kellermann Reference Radtke and Kellermann1991). Food availability and temperature have a pronounced effect on larval growth, which in turn influences timing and duration of larval dispersal and predation vulnerability, and ultimately, survival rates of early life stages and recruitment strength. It has been reported that sometimes large recruitment fluctuations can be determined by small variations of growth rates and the related durations of larval stages (Houde Reference Houde1987). Driven by environmental constraints like stable low water temperature and pelagic food seasonality, Antarctic fishes evolved different early life history strategies in terms of timing and size at hatching, early growth rates and duration of pelagic phase (North & White Reference North and White1987, Radtke & Kellermann Reference Radtke and Kellermann1991, North Reference North2001). For example, at South Georgia large pelagic larvae (>10 mm SL) hatch during late winter, spring or summer, grow and develop during summer, and then overwinter as pelagic early juveniles or develop into a demersal early juvenile stage before the winter. Alternatively, small pelagic larvae (<10 mm SL) hatch during spring and summer, grow and develop during summer, and then overwinter as pelagic early juveniles or develop into a demersal early juvenile stage before their first winter (North Reference North2001).
Growth curves of the Antarctic fish larvae have been established for a few species, generally using the progression of mean length throughout the summer (Kellermann Reference Kellermann1986, Radtke et al. Reference Radtke, Targett, Kellermann, Bell and Hill1989, Loeb Reference Loeb1991). However, otolith microincrement counts provide more precise estimates of hatching periods than length frequency analysis or the occurrence of the smallest larvae, despite that they represent the most common methods used thus far (Radtke & Kellermann Reference Radtke and Kellermann1991). In addition, the deposition pattern of microincrements may reveal the timing of significant events during the early ontogeny of Antarctic fish, such as larval hatching, yolk absorption, first feeding or changes in habitat (e.g. Ruzicka Reference Ruzicka1996, Kellermann et al. Reference Kellermann, Gauldie and Ruzicka2002, Morley et al. Reference Morley, Belchier, Dickson and Mulvey2005, La Mesa et al. Reference La Mesa, Catalano, Koubbi and Jones2013). The timing and duration of these events have a great influence on feeding strategies of early life stages, as prey size and food availability depend on larval size and season, respectively. In winter, for example, day length and light intensity reduce feeding opportunities for fish larvae, as generally they feed mainly during daylight (North & Ward Reference North and Ward1990).
During the 2010/11 US Antarctic Marine Living Resources (AMLR) field season (Jones et al. Reference Jones, Koubbi, Catalano, Dietrich and Ferm2014), a large number of post-larvae of T. scotti were collected in the Bransfield Strait. Based on otolith microincrement patterns and stomach content analyses, the study aimed to i) extend the knowledge on the spatial distribution of this species in the surveyed area, ii) estimate age and growth of individual fish, providing insight on some key early life history events, such as hatching distribution, and iii) determine food composition and feeding intensity.
Materials and methods
Sample collection
Post-larvae of T. scotti were collected during a mesopelagic and larval fish survey conducted in the Bransfield Strait and adjacent areas aboard the RV Moana Wave from 17 February to 8 March 2011 (Jones et al. Reference Jones, Koubbi, Catalano, Dietrich and Ferm2014). A total of 70 stations covered a wide area off the western Antarctic Peninsula, including the Bransfield Strait and the Gerlache Strait, and off the tip of the Antarctic Peninsula, i.e. north of Joinville Island and around Elephant Island (Fig. 1a). The main fishing gear employed was a four square metre effective mouth multiple opening/closing Tucker Trawl net system, with two panels of 505 μm mesh size and a third panel of 5 mm mesh size (specifically targeting mesopelagic finfish), towed at depth strata between 0–170 m, 170–300 m and 300–600 m. Additional samples were collected using an Isaac Kidd Midwater Trawl (IKMT) single net with a mesh size of 505 μm deployed exclusively between 0–170 m.
Surveyed area across the Bransfield Strait and adjacent areas. a. Complete grid of sampling stations. b. Isaac Kidd Midwater Trawl stations and c. Tucker stations with catches of Trematomus scotti post-larvae. Dots indicate the relative abundance as individuals per 1000 m3 of filtered sea water.
Sample sorting and measurements
After a successful haul, the early life stages of fish were sorted and identified to the species level according to North & Kellermann (Reference North and Kellermann1990). The larval stage of development was assigned to each specimen according to the four-point scale of Koubbi et al. (Reference Koubbi, Duhamel and Camus1990). Fish larvae were measured from the tip of the snout to the end of caudal peduncle (SL) to the nearest mm below. In the laboratory, sagittal otolith pairs were removed under a stereomicroscope using fine needles and one of them randomly selected for further analysis. Otoliths were mounted on slides (medial side down) within a drop of epoxy resin (Petropoxy 154, Burnham Petrographics LLC) cured at 80°C for 12 hours. After complete resin polymerization, otoliths were gently ground with a metallographic paper disc lubricated with 1–3 µm alumina paste and polished with 0.05 µm alumina paste using a lapping wheel (Remet SAS). Otolith sections were then observed on the distal side.
Ageing methodology
The otolith microstructure was examined at high magnification (630x) using a light microscope equipped with a CCD video camera (Leica DFC 420) connected to a digitized computer video system (Leica Application Suite 4.3.0.). Microincrement counts and measurements were made from the primordium to the otolith margin, following as much as possible the same counting path in all otoliths. Care was taken to locate growth discontinuities or checks along the counting path, generally associated with key ontogenetic steps of larval development (e.g. hatching, first exogenous feeding, etc.). The daily periodicity of microincrements in T. scotti were assumed based on similar microincrement patterns of deposition validated in other related species of nototheniids (e.g. T. newnesi Boulenger; Radtke et al. Reference Radtke, Targett, Kellermann, Bell and Hill1989). Therefore, individual age was assigned by counting all microincrements from the primordium to the otolith margin, assuming the first one laid down at larval hatch. For each otolith, two counts were made by the same reader to test the reproducibility of age estimates. Age precision was assessed by calculating the index of average percent error (APE) (Beamish & Fournier Reference Beamish and Fournier1981) and the mean coefficient of variation (CVmean) (Chang Reference Chang1982) compared to the common threshold precision levels (Campana Reference Campana2001). A bias plot was computed to determine the presence of systematic differences in age estimates. The hatch dates distribution was back-calculated by subtracting the age estimates from the date of capture.
Growth modelling
An exponential model was used to describe the growth pattern of T. scotti larval stages, as this generally provides the best fit for modelling early growth rate of Antarctic fishes (e.g. Kellermann Reference Kellermann1986, North Reference North1998, La Mesa et al. Reference La Mesa, Catalano, Koubbi and Jones2013). The model was fitted to the age-length dataset in the following form:
$${\rm log}_{{\rm e}} \,{\rm SL}={\rm log}_{{\rm e}} {\rm a{\plus}gt},$$
where SL is in mm, logea is the initial SL (i.e. the hatching size), g is the instantaneous growth rate and t is the age (days). The instantaneous growth rate (g) was converted to the daily percentage increment in size (G, %SL day-1), which represents a more useful index to compare growth rates estimated for different species or size ranges. Finally, the daily increase in size at the mean SL of the fish sample was calculated as well.
Diet analysis
The stomach contents of each specimen were removed and sorted under a stereomicroscope at low magnification. Prey items were separated into main food categories and counted. The weight of prey was not recorded, as they were generally of small size and at different degrees of digestion. Diet analysis was carried out by assessing the percentage by number (%N) and the frequency of occurrence (%O) of each prey item. To evaluate the rate of feeding activity, the vacuity index was calculated:
$${\rm V}={\rm NeN}^{{{\hbox{-}}{\rm 1}}} ,$$
where Ne is the number of empty stomachs and N is the total number of stomachs examined. A G-test of independence was applied to determine differences in the frequency of empty stomachs in relation to fish size.
Results
Abundance and spatial distribution
Overall, 74 post-larvae (stage 3 of development) of T. scotti ranging from 12 to 20 mm SL were caught at five stations by the IKMT (Fig. 1b) and ten stations by the Tucker Trawl (Fig. 1c). All specimens were collected in the uppermost layer of the water column sampled (0–170 m), except for a single large specimen (20 mm SL) caught in a deeper layer (300–600 m). Catches were unevenly distributed along the Bransfield Strait from north-east of Joinville Island to the Gerlache Strait (Fig. 1b & c). The relative abundance was 0.3–2.6 specimens (mean±standard error, 0.9±0.4) and 0.4–3.6 specimens (1.3±0.2) per 1000 m3 of filtered sea water with the IKMT and the Tucker Trawl, respectively. Peaks of abundance were located in the Gerlache Strait and in the easternmost and westernmost investigated areas of the Bransfield Strait.
Age and growth
Sagittal otoliths of post-larval T. scotti generally had a discoid shape, until the onset of rostrum formation occurring in larger specimens from c. 19 mm SL onwards. Maximum otolith diameter was 103–193 µm. Multiple primordia were delimited by a well-defined check assumed to be laid down at larval hatch, forming a 21–27 µm wide core (Fig. 2). Daily increments encircling the hatch check were relatively thin (1–1.3 μm) and evenly spaced until the otolith margin. They appeared as common bipartite concentric rings, consisting of a narrow and opaque discontinuous zone (D-zone), and a broad and translucent increment zone (L-zone) when viewed with the light microscope (Fig. 3).
Sagittal otolith of post-larval Trematomus scotti showing the core area with multiple primordia (arrow) and hatch check (arrow head).
Sagittal otolith of post-larval Trematomus scotti showing the regular microincrements pattern.
A total of 66 specimens ranging from 12 to 20 mm SL were available for ageing purposes. Most specimens (97%) were successfully aged, providing an age range of 34–67 days with no apparent systematic difference or bias between age estimates (Fig. 4). Both indices of precision were relatively low (APE=1.7%, CVmean=2.5%), revealing good consistency between age estimates. The exponential model fitted to the age-length dataset provided the following relationship (Fig. 5):
$${\rm log}_{{\rm e}} \,{\rm SL}={\rm log}_{{\rm e}} \,{\rm 2}.{\rm 19{\plus}}0.0{\rm 11t}.$$
Applying the exponential function, the larval size at hatch was estimated to be c. 9.0 mm. The daily percentage increment in size (G) was 1.07% SL day-1, and the daily increase at the mean size of 15.8 mm was 0.17 mm. Based on a weekly hatch date distribution (Fig. 6), T. scotti post-larvae collected in the Bransfield Strait hatch over a relatively narrow period between the end of December and the end of January.
Bias plot between age estimates of post-larval Trematomus scotti. Error bars represent the 95% confidence interval.
Age-length relationship estimated for post-larval Trematomus scotti showing the experimental dataset fitted by the exponential model.
Weekly hatch date distribution for post-larval Trematomus scotti back-calculated from the individual age estimate and date of capture.
Diet
Of the 72 post-larvae analysed, 22 had empty stomachs (V=30.5%), with no significant relationship with fish size (G-test, χ2=3.12, df=7, P>0.5). Food composition of post-larval T. scotti consisted exclusively of copepods, most of them at the stage of copepodites (Table I). Feeding intensity ranged between 1 and 14 prey items per stomach, increasing significantly in relation to fish size (Fig. 7a). The number of prey items per stomach was rather different among sampling areas, peaking in the northernmost stations of the Gerlache Strait (GS01, GS05, GS07) and south of King George Island (D1010, D1110) (Fig. 7b).
Feeding intensity of post-larval Trematomus scotti. a. Number of prey items in relation to fish size. b. Number of prey items in relation to sampling station.
Food composition of post-larval Trematomus scotti from the Bransfield Strait and adjacent waters.
Discussion
Consistent with previous surveys carried out in summer across the Bransfield Strait and adjacent waters (Kellermann Reference Kellermann1989, Loeb Reference Loeb1991, Catalán et al. Reference Catalán, Morales-Nin, Company, Rotllant, Palomera and Emelianov2008), the early life stages of T. scotti were among the most abundant fish collected in 2011, accounting for 36% of high-Antarctic species (Jones et al. Reference Jones, Koubbi, Catalano, Dietrich and Ferm2014). Although with slight interannual changes in overall pattern of distribution, generally this species preferred coastal areas within the upper 75–100 m depth, regardless of the influencing water masses (Loeb Reference Loeb1991, Catalán et al. Reference Catalán, Morales-Nin, Company, Rotllant, Palomera and Emelianov2008, present study). Indeed, the bulk of specimens were encountered in the western Bransfield Strait north of Joinville Island, which is typically associated with the Weddell Sea water, as well as in the eastern Bransfield Strait and within the Gerlache Strait, which are characterized by water masses of Bellingshausen Sea origin. According to Loeb (Reference Loeb1991), the larvae of T. scotti collected in the western Bransfield Strait are probably derived from stocks of the Bellingshausen Sea, as supported by the abundance of larval stages of this species found along the western Antarctic Peninsula from the Palmer Archipelago to the Adelaide Island (Kellermann Reference Kellermann1989, Morales-Nin et al. Reference Morales-Nin, Palomera and Schadwinkel1995). Likewise, the potential transport pathways for the larvae of T. scotti caught in the eastern Bransfield Strait might be through predominant water flowing northward from spawning grounds located in the western Weddell Sea, where this species is particularly abundant (Kock et al. Reference Kock, Busch, Holst, Klimpel, Pietschok, Pshenichnov, Riehl and Schöling2008).
The use of sagittal otoliths was confirmed to be a reliable method for estimating age and growth of post-larval T. scotti, as already reported for this stage in other nototheniids (e.g. Radtke et al. Reference Radtke, Targett, Kellermann, Bell and Hill1989, Kellermann et al. Reference Kellermann, Gauldie and Ruzicka2002). The growth rate showed marked individual variability, possibly related to a wide range of environmental conditions experienced by the larvae during the period of advection from the spawning areas. Consistently, the RNA/DNA ratios and the index of recent otolith growth estimated in post-larvae of T. scotti provided evidence of huge differences in both larval condition and short-term growth (Morales-Nin et al. Reference Morales-Nin, Palomera and Busquets2002). The exponential model fitted to the age-length dataset provided an estimated mean daily growth rate of 0.17 mm, which was significantly higher than the growth rate of 0.05 mm day-1 reported from the same area using the progression of length frequency distributions over the entire summer (Loeb Reference Loeb1991). The longer sampling period investigated by Loeb (Reference Loeb1991) and different methodologies used to estimate larval growth rate might partially explain such a large difference. In any case, the growth rate of other nototheniids living off the Antarctic Peninsula showed great variability, both at interspecific and at intraspecific level in relation to sampling season. For example, the growth rate was c. 0.37 mm day-1 for T. newnesi (Radtke et al. Reference Radtke, Targett, Kellermann, Bell and Hill1989), whereas it ranged between 0.04 mm and 0.12 mm day-1 in spring and summer, respectively, for Nototheniops larseni (Lönnberg) (Kellermann Reference Kellermann1986).
According to the classification of the early life history strategies of notothenioids (North Reference North2001), T. scotti has small pelagic larvae (<10 mm SL) which hatch in summer (mainly in January) and then overwinter off the Antarctic Peninsula as pelagic early juveniles until the following summer, when they shift to the demersal habitat. Indeed, overwintering larvae of T. scotti were found in May–June in coastal waters off Biscoe Island (Kellermann & Schadwinkel Reference Kellermann and Schadwinkel1991) and pelagic juveniles of 31–58 mm SL were caught in the Bransfield Strait in January–February (Kellermann Reference Kellermann1989, Catalán et al. Reference Catalán, Morales-Nin, Company, Rotllant, Palomera and Emelianov2008). Off the western Antarctic Peninsula, the estimated size and timing of larval hatching of T. scotti were fairly constant among different studies, either based on the smaller specimens caught (e.g. Kellermann Reference Kellermann1989, Loeb Reference Loeb1991, Catalán et al. Reference Catalán, Morales-Nin, Company, Rotllant, Palomera and Emelianov2008) or based on otolith microincrement counts (present data). Therefore, assuming a fixed spawning season, it can be argued that in this species the egg incubation time is constant and that larval hatch occurs regularly at a species-specific season of the year independent of the interannual variability of the timing of pack ice retreat and onset of the production cycle. Based on adult specimens at an early stage of gonad development and low gonadosomatic index (5–6%) collected in spring and summer in the Weddell Sea (Ekau Reference Ekau1991, Duhamel et al. Reference Duhamel, Kock, Balguerias and Hureau1993), the spawning of T. scotti possibly takes place at the end of winter and the related duration of egg incubation would last c. 120 days. In T. scotti, sexual maturity is attained at 12–13 cm total length (i.e. c. 70% of maximum size), with a relatively low absolute fecundity (4,505–7,840 eggs per female) but a high relative fecundity (100–135 eggs g-1) compared to other species of Trematomus (Duhamel et al. Reference Duhamel, Kock, Balguerias and Hureau1993).
The stomach content analyses of the post-larvae of T. scotti provided useful insights on their feeding strategy. Early stages of copepods (copepodites) represented, overwhelmingly, the most important items in the diet of T. scotti, whereas eggs and adult copepods were rare. The diet composition found here closely resembles a previous report from the same area (Balbontin et al. Reference Balbontin, Garreton and Neuling1986). However, the abundance and frequency of occurrence of eggs and copepodites were similar to each other in the previous study, and the vacuity index was significantly lower (11%) (Balbontin et al. Reference Balbontin, Garreton and Neuling1986), possibly linked to high interannual variability of prey availability. Feeding intensity, in terms of the number of prey items per stomach, was positively correlated with larval size, supporting evidence that larger size and faster growth are likely to confer advantages for survival and potential to recruit (Houde Reference Houde1987). The occurrence of copepods in the stomach contents of T. scotti could be tentatively linked to the scarce presence of krill recorded during the survey within the Bransfield Strait (Dietrich et al. Reference Dietrich, Brooks, Bystrom, Driscoll, Ferm, Hinke, Janssen, Lombard, Pesce, Romain, Thoresen and van Cise2014). When krill is absent, copepods may constitute 40–90% by number of the total zooplankton population of the Scotia Sea and Bransfield Strait (Mujica & Asencio Reference Mujica and Asencio1985). Eggs, copepodites and adult copepods represented the main prey for several other nototheniids composing the larval fish assemblage of the Bransfield Strait, such as Notothenia kempi Norman, Nototheniops larseni and Pleuragramma antarcticum Boulenger (Balbontin et al. Reference Balbontin, Garreton and Neuling1986, Kellermann Reference Kellermann1986). In this context, copepods might not represent a limiting food resource for larval fish especially in summer, when they exhibit many different reproduction peaks and strategies (e.g. Atkinson Reference Atkinson1998). Off the western Antarctic Peninsula, larval hatching of nototheniids was spread over a wide period, with a temporal sequence of different fish species (Kellermann Reference Kellermann1989), which largely contribute to fine resource partitioning among spatially co-occurring larvae.
Acknowledgements
We thank the US AMLR and foreign scientific staff, and the crew members and personnel aboard the RV Moana Wave for their support in sampling activities. We would like to thank Esteban Barrera-Oro and an anonymous reviewer for their helpful advice on the earlier version of the manuscript. This study was carried out within the project 2009/A1.07 funded by the PNRA (Italian National Program for Antarctic Research).
Author contribution
M.L.M conceived the study and wrote the paper, B.C. and C.D.J. provided samples and contributed significantly to the interpretation of data and edited the manuscript before submission.
