1 INTRODUCTION

The Patagonian Shelf and Slope is among the most productive areas in the Southwest Atlantic. There are two marine ecosystems, the southern temperate ecosystem and the sub-Antarctic ecosystem which run from the south-western part to the north-eastern part of the Patagonian Shelf through the Falkland (Malvinas) Islands archipelago (Boltovskoy, 2000). A complex hydrographic structure, with a permanent front (the Eastern Falkland Current) over the shelfbreak and three coastal frontal regions (the temperate estuarine zone, the Patagonian tidal zone and the Patagonian cold estuarine zone) and a wide nonfrontal area in between, occurs on the Patagonian shelf (Acha et al., 2004).

In the past several decades, the Patagonian shelf has been a region with a major bottom-trawl fishery for the longfin squid, Loligo gahi (Patterson, 1988), Argentine shortfin squid, Illex argentinus (Laptikhovsky et al., 2013), finfish fishery (Nakamura et al., 1986; Haimovici, 1998) and squid jigging fishery. This has become an important squid fishing ground for the Chinese squid fishery (Chen et al., 2008). Fishery discards include damaged squid, the remains of processed commercial fish and by-catch species, sea urchins and scallops, and other benthos (Laptikhovsky et al., 2013). Those non-natural food sources result in a change in the feeding ecology of fishes and squids in the spatial-temporal scales (Laptikhovsky, 2004). The climate change exacerbates the complex of trophic relationships of fishes and squid in this region. The trophic level of marine fauna on the upper Patagonian shelf, particularly the Río de la Plata estuary (Botto et al., 2011) and lower Patagonian shelf (Ciancio et al., 2008), especially the waters around the Falkland Islands (Quillfeldt et al., 2015) were examined using stable isotope analysis (SIA). This analysis has proved to a powerful and reliable technique to characterize trophic relationships within the Southern Patagonian shelf region (Ciancio et al., 2008; Saporiti et al., 2015). Saporiti et al. (2015) explore the latitudinal variations in the 3 different coastal ecosystems along the Patagonian shelf using SIA. However, the food web of the Patagonian shelf break, particularly the area around the important fishing ground of the squid fishery has not been examined and is poorly understood. Therefore, the aim of this study was to determine the trophic structure of fishes and squid in the productive Patagonian shelf break.

2 MATERIAL AND METHOD 2.1 Sample collection

Samples were collected from the large-scale midwater trawler Longfa between December 2014 to April 2015 in the fishing grounds around the central Patagonian shelf break (Fig. 1). All samples were collected randomly from the fishing locations and stored in a freezer at -20℃ for later analysis in a landbased laboratory.

2.2 Stable isotope analysis

Once in the laboratory, samples were unfrozen. The length (total length for skates, standard length for fish and mantle length for squid, in centimeters) and wet weight (in grams) of the samples were measured. The mantle or white dorsal muscle was extracted and analyzed for the cephalopods, skates and fishes, respectively. The muscle samples were thawed, dried in an oven at 60℃ for 36-48 h, and ground to a fine powder with a mortar and pestle. Lipids were extracted from fish and invertebrate samples with a chloroform/methanol (2:1) solution (Bligh and Dyer, 1959). This is because lipids are depleted in 13C compared with other molecules (DeNiro and Epstein, 1977) and lipid concentration in tissues may vary between and within species. This variation results in an artifactual variation of the overall tissue δ¹³C value. This powder was used directly for carbon and nitrogen analysis.

Stable isotope abundance is expressed in standard δ notation relative to carbonate Pee Dee Belemnite and atmospheric nitrogen. International secondary isotope standards of known C¹³/C¹² and N¹⁵/N¹⁴ ratios, as given by the International Atomic Energy Agency (IAEA)—namely: polyethylene (IAEA CH7, δ¹³C=- 31.8‰), sucrose (IAEA CH6, δ¹³C=-10.4‰), ammonium sulphate (IAEA N1; δ¹⁵N=+0.4% and IAEA N2; δ¹⁵N=+20.3%), potassium nitrate (USGS 34; δ¹⁵N=21.7%), L-glutamic acid (USGS 40; δ¹⁵N=24.6%; δ¹³C=22.6, 2%), and caffeine (IAEA 600; δ¹⁵N=1.0%; δ¹³C=27.7%)—were used for calibration of δ¹³C and δ¹⁵N. The precision for nitrogen was 0.2‰ and for carbon was 0.3‰. Mean C- and N- isotope values were presented for each species. Analyses was performed at the College of Marine Sciences of Shanghai Ocean University (SHOU).

2.3 Food web analysis

Limited availability of baseline data impeded the estimation of the trophic level (TL) of these species directly. However, the samples from Ciancio et al. (2008) were almost all collected from the shelf break (Ciancio et al., 2008; Saporiti et al., 2015), so in order to assign a trophic level to all species, euphausids, a herbivores, were introduced as the lowest trophic level of all species and assigned a value of 2.0 to this species (Ciancio et al., 2008). Using the mean trophic fractionation (TF) factor for ¹⁵N for marine herbivores (2.52‰; Van Der Zanden and Rasmussen 2001), trophic level was calculated by:

2.4 Data analysis

Twelve species were analyzed for carbon and nitrogen isotope composition. Species were grouped into functional guilds, i.e., demersal (Cottoperca gobio, Congiopodus peruvianus, Conger orbignyanus, Raja cyclophora), benthopelagic (Stromateus brasiliensis, Patagonotothen ramsayi, Macruronus magellanicus), bathydemersal (Genypterus blacodes, Merluccius albidus, Breviraja spinosa, Raja bullisi) and pelagic (Illex argentines) guilds according to their biological and ecological features (Sielfeld and Vargas, 1999; Froese and Pauly, 2011; WoRMS Editorial Board, 2014). Kruskal-Wallis test was used to examine the difference in stable C- and N-isotope signatures of guilds.

nMDS Variation in fish species and squid was visualized using non-metric multidimensional scaling (nMDS) on the Bray-Curtis distances with mean Cand N-isotope values. According to Clarke (1993) a stress value below 0.2 indicates that an nMDS plot is useful for interpreting differences between communities.

3 RESULT

The general picture of C- and N-stable-isotope data showed a food web with a narrow range of C-isotope values (between -18.08‰ and -17.04‰), except for S. brasiliensis (-20.75‰) (Table 1, Fig. 2). δ¹³C values for fishes, skates and squid analyzed ranged from -22.01‰ to -16.56‰, -17.53‰ to -16.56‰ and -19.06‰ to -16.52‰, respectively. The δ¹⁵N values for fishes, skates and squid analyzed ranged from 12.86‰ to 17.63‰, 12.64‰ to 14.44‰ and 8.58‰ to 12.61‰, respectively. The lowest and highest δ¹³C values were S. brasiliensis (-22.01‰) and G. blacodes (-16.56‰) for fishes, R. cyclophora (-17.53‰ and -16.57‰) for skates and I. argenitna (-19.06‰ and -16.52‰) for squid. The lowest and highest δ¹⁵N values were P. ramsayi (12.86‰) and G. blacodes (17.63‰) for fishes, R. bullisi (12.64‰ and 14.44‰) for skates and I. argenitna (8.58‰ and 12.61‰).

Except for S. brasiliensis (ANOVA, F=8.416, P=0.020 < 0.05), there were no significant linear relationships between δ¹³C value and the length of fish and squid species (Table 2). However, more linear relationships can be found on δ¹⁵N and the length of fish species, including C. peruvianus and two skates (R. cyclophora and R. bullisi) (Table 3).

The stress value for this nMDS plot is 0.01, meaning that the plot provides a useful representation of the variation in fish species and squid (Fig. 3). P. ramsayi, M. magellanicus and skates (R. cyclophora, B. spinosa and R. bullisi) can be divided into a group. C. gobio, C. peruvianus, C. orbignyanus and M. albidus was categorized as another group. S. brasiliensis, G. blacodes and I. argentinus were significantly different from other species and separated from each other.

The mean C-isotope signatures varied (Kruskal-Wallis H_{3, 78}=17.63, P=0.001 < 0.05) for the four trophic guilds. The guild-average δ¹³C value was less enriched for the benthopelagic guild than for other three guilds, which presented very similar average δ¹³C values (Fig. 4). For the benthopelagic guild, P. ramsayi and M. magellanicus shared similar δ¹³C values, but S. brasiliensis had a significant lower δ¹³C value and lowered the average δ¹³C value of this guild.

A similar significant difference was also found for the stable N-isotope signatures of the four guilds (Kruskal-Wallis H_{3, 78}=28.95, P < 0.001). The average δ¹⁵N value of the pelagic guild (I. argentinus) was remarkable lower than those of the other guilds, which showed very similar and more enriched δ¹⁵N values (Kruskal-Wallis H_{2, 68}=4.10, P=0.13>0.05). The bathydemersal guild (G. blacodes, M. albidus, B. spinosa, and R. bullisi) had more enriched δ¹³C and δ¹⁵N values than those of the other guilds. Generally, δ¹⁵N values increased with the depth of the fish habitat from the surface (pelagic) to the bottom (bathydemersal).

4 DISCUSSION

Stable isotopic analysis was widely used to indicate the feeding habitat of aquatic animals (Fisk et al., 2002; Jackson et al., 2007; Park et al., 2011; Agersted et al., 2014) and has become a common approach to understand the food web structure and the trophic dynamics of different ecosystems (Post, 2002). However, information on diet composition or feeding habits of some fish species in the Atlantic Patagonian shelf region were still derived from gut content analysis (Santos and Haimovici, 1997; Mouat et al., 2001; Nyegaard et al., 2004). Diet composition of I. argentinus and G. blacodes presented an ontogenetic variation (Mouat et al., 2001). Rockcod Patagonotothen spp. and M. magellanicus were key prey species of G. blacodes around the Falkland Islands, but spatial variability in the diet was found in G. blacodes between temperate and sub-Antarctic waters (Nyegaard et al., 2004). However, the stable isotope analysis did not support the ontogenetic diet shift in the above fish species, based on the relation between both isotopes and size. The spatial-temporal heterogeneity of diet composition and limited sample size may contribute to the difference.

The food web structure on the Patagonian shelf were examined by several studies. Because of the introduction of salmonids on the tip of South America in the past decades, Ciancio et al. (2008) examined the trophic relationships of exotic anadromous salmonids on the southern Patagonian shelf inferred from stable isotopes. Drago et al. (2009) revealed the stable isotope values of the food items from South American sea lion in Patagonia. Based on stable isotope analysis, Saporiti et al. (2015) compared the latitudinal changes (the temperature estuarine zone associated with the Río de la Plata plume, the tidal zone off northern and central Patagonia and the cold estuarine zone off southern Patagonia) in the marine food web structures in this region and found the food chain length (FCL) decreases and the trophic redundancy increases as latitude increases in all of the above three regions. The tidal zone off northern and central Patagonia (northern Patagonian region in Fig. 5) has similar latitude range to the present study, but the region of the present study is located offshore and in the shelf break. Significant differences can be found for mean C- and N-isotope values of the same species among the present study and the previous studies (Table 4). The δ¹⁵N value of P. ramsayi in the Patagonian shelf break was lowest and the δ¹⁵N value of P. ramsayi in the northern and central Patagonian shelf was significantly higher than those in the others regions (Fig. 5). Similar lower δ¹⁵N values can be found for other fish and squid species in the shelf break, although the latitudinal changes on the δ¹⁵N values of those species has not occurred. Significant lower δ¹⁵N values of species in present study area may be attributed to complex habitats, oceanographic dynamics and overlapping anthropogenic activities. Squid and finfish fisheries were concentrated in this region in the past decades (Haimovici, 1998; Chen et al., 2008; Laptikhovsky et al., 2013). Humans were fishing down the food web to lower trophic level by selectively removing large fish from the oceans (Pauly et al., 2000). This process has potentially contributed to the lower δ¹⁵N value in this ecosystem. However, data on the feeding ecology of commercial species, particularly finfish, in this region was scarce (Mouat et al., 2001). Therefore more fishery data needed to provide a possible explanation. The δ¹³C values for most of species were concentrated from -18‰ to -17‰. However, the δ¹³C values of some species (for example, G. blacodes and P. ramsayi) in the northern and central Patagonian region were significantly higher than other species. The δ¹³C value of species in the southern Patagonian shelf were higher than those in other regions (Fig. 5). Particularly, the δ¹³C values of S. brasiliensis was significantly lower than other species in the present study, the possible reason could be its residence in shallower and less salty waters which flowed from Argentina or Falkland waters (Fig. 6). Using the ratio of δ¹³C/δ¹⁵N of species (four species, P. ramsayi, G. blacodes, M. Magellanicus and I. argentinus were involved) as an indicator of food web structure. A significant difference can be found between the Patagonian shelf break (the present study) and the northern and central Patagonian region (Saporiti et al., 2015) (paired t-test, t=4.30, P=0.04 < 0.05) (Fig. 5). Lowest mean N-isotope values of species occurred in the Patagonian shelf break. Spatial-temporal heterogeneity of those studies can provide potential explanations for these differences. One of reasons is the possible annual or seasonally variability in stable isotope ratios, as the samples used by Ciancio et al. (2008) and Saporiti et al. (2015) were collected from 2001 to 2005 and from October 2009 to December 2011, respectively. However, the samples collected for the present study were collected from December 2014 to May 2015. The second reason is that the samples collected by Ciancio et al. (2008) were from the southern Patagonian shelf regions between the coastal frontal area and the shelf break and by Saporiti et al. (2015) from the coastal frontal area of northern or southern Patagonian shelf. However the samples in the present study were from shelf break between the regions B and C of Saporiti et al. (2015). Significant differences in the mean N-isotope values of I. argentines (pelagic species), P. ramsayi (benthopelagic species) and G. blacodes (bathydemersal species) between Ciancio et al. (2008) and Saporiti et al. (2015) in a similar region also supported the temporal heterogeneity of the ecosystem.

Complex frontal regime in this region can provide another explanation. The Patagonian shelf has a complicated hydrographic structure, with a permanent front over the shelf break and coastal frontal regions (Acha et al., 2004). This results in the different ecosystem structures in the above habitats. Ciancio et al. (2008) collected most of the secondary consumer samples from the shelf region between the coastal frontal area and the shelf break. The samples of Saporiti et al. (2015) were collected within the coastal frontal regions where are characterized by intense vertical mixing of the water column (Acha et al., 2004) and hence a region with higher δ¹⁵N values (Calvert et al., 1992; Wu et al., 1997; Waser et al., 2000). This can explain why the δ¹⁵N values of samples from Saporiti et al. (2015) were higher than those from Ciancio et al. (2008) and the present study. Furthermore, the FCL in the Patagonian shelf break (FCL=2.16 between G. blacodes and I. argentinus, the present study) was longer than those in the southern Patagonian shelf (FCL=1.4, Ciancio et al., 2008), meaning poleward food web shortening was noted (Saporiti et al., 2015). Compared with the δ¹⁵N values in the northern and central Patagonia, the lower δ¹⁵N values presented in this study area also support this argument. Fishing activities, which has modified the abundance of many species and their diets (Drago et al., 2009; Ramírez et al., 2014; Zenteno et al., 2015) have also modified fatherly food web structures in the exploited regions, particularly in the Patagonian shelf break where heavy commercial squid fishery has occurred in the past decades (Laptikhovsky et al., 2013).

Oceanographic structures, such as fronts, are discontinuities in the marine environment which influence the ecology of marine organisms (Leichter and Witman, 2009) and the distributional pattern of marine organisms at all trophic levels (Alemany et al., 2014). At upper trophic levels, particularly fish and squids, behavioral sensory cues (i.e., temperature, salinity, optical conditions, and trace substances) are involved in actively seek out frontal structures for reproduction, feeding and migration (Olson, 2002). Frontal systems are characterized by high primary and secondary production (Mann and Lazier, 2006) that is transferred to higher trophic levels within the regional food web. The Shelf-break front (SBF, Alemany et al., 2014) and the Eastern Falkland current (EFC, Arkhipkin et al., 2013) cross the present study area (Fig. 6). Argentine drift also reaches the present study area. Similarly, the Southern Patagonia front (SPF) and Argentine drift have an impact on the study areas of Ciancio et al. (2008) and Saporiti et al. (2015). However, the impact of those fronts had different effects on those study areas and further affected trophic levels of the above food web structures.

With the unavailability of stable isotope data of primary producers, the trophic level of the Patagonian shelf break cannot be structured directly. However, there was a clear separation of resources between demersal-benthic and pelagic species of some fish species (Fig. 4 and Fig. 5). Moreover, the present study indicates the Patagonian shelf break can be considered as a separated ecosystem structure with lower δ¹⁵N values.

5 ACKNOWLEDGMENT

We would like to thank the fisheries observers, crew and the officers of the trawler Longfa, LIU Zijun, CHEN Lvfen, WANG Rui, SONG Qi and REN Zeqian at the College of Marine Sciences, Shanghai Ocean University for their helps in processing the samples. We acknowledge Mr Alan Coughtrey of the Shanghai Ocean University for his help in polishing the language. Finally, we would also like to thank two anonymous reviewers for their contributions to improve this manuscript.