Feeding ecology of fish larvae from Chilean Patagonian during austral winter

Feeding habits of the ichthyoplankton from Chilean Patagonia (44-46°30’S) were studied during June-July 2012 (austral winter). Ichthyoplankton assemblage was composed by 10 species, with low abundance (3.8 to 16.73 ind. 1000 m-3). Most abundant larvae were Maurolicus parvipinnis, Leptonotus blainvilleanus and Sprattus fuegensis. These three species fed mainly on calanoid copepodites, Paracalanus indicus and Calanus sp. copepodites, and cyphonautes. Trophic overlap among dominant fish larvae was high (Schoener’s D > 0.69) and no significant differences were detected in prey composition and size. Therefore, no resource partitioning occurred in planktonic fish larvae during winter 2012 in Chilean Patagonia.

and channels (Niklitschek et al. 2014, Osorio-Zúñiga et al. 2018, although a secondary spawning location in the inner seas is used during austral spring (Bustos et al. 2007, Landaeta et al. 2018. In this sense, the oligotrophic nature of the environmental conditions during austral winter may constraint feeding strategy of fish larvae. However, until now, no studies have been done regarding feeding ecology of fish larvae during winter. Therefore, using fish larvae collected during winter 2012, we evaluated the diet, prey size and trophic overlap of the early life stages of marine fishes of south Patagonia. Studying feeding ecology of fish larvae is important to establish potential constraints that the winter season may pose to the trophic interactions, particularly in those species with commercial interest.

Materials and methods
A bio-oceanographic survey was carried out between 16 June 2012 and 08 July 2012 (austral winter) in the inner fjords and channels of Chilean Patagonia between 44º and 46°30'S, on board RV Abate Molina (Fig. 1A). Ichthyoplankton samples were collected at each station (n= 31) with 1 m 2 Tucker trawl (300 μm mesh size), by oblique trawls from surface to 150 m depth performed for 15-20 min. A flowmeter (General Oceanics™, Florida USA) was installed in the frame of the net to estimate the filtered volume. All samples were fixed with 10% formaldehyde buffered with sodium borate, and after 12 h, they were preserved in 96% ethanol. Filtered seawater volume varied from 93.6 to 353.7 m 3 (mean ± SD, 236.5 ± 67.6 m 3 trawl -1 ).

Introduction
The Chilean Patagonia is located in the southeastern border of the Pacific Ocean and covers more than 1,000 km between 41° and 55°S. Fjords therein usually range between 100 and 500 m depth, from 10 to 70 km long, and have a two-layer structure that is vertically limited by a strong halocline between 4 and 20 m depth (González et al. 2011). In the area, the productivity cycle can be divided into productive (late winter/early spring to late autumn) and non-productive (winter) seasons (Montero et al. 2011). In winter, integrated Chlorophyll-a concentration (Chl-a) in the mixed layer varies between 5 and 56 mg m -2 , whereas the spring range is 8.5-165 mg m -2 (González et al. 2011). Picophytoplankton (< 2 μm) make the greatest contribution of all size-fractions to the total Chl-a in winter; on the other hand, large phytoplankton cells (> 20 μm) account for 51% of the total Chl-a during spring (González et al. 2011).
The strong temporal variations in oceanographic conditions and primary and secondary productivity pose restrictions to reproductive tactics of marine fish inhabiting this ecosystem. Tactics of most marine fish species from Chilean Patagonia encompass reproduction during austral spring in order to ensure food availability to their offspring during the first-feeding period, after the yolk-sac absorption (Landaeta & Castro 2006, Bustos et al. 2008a, Landaeta et al. 2015. Other species, mainly Gadiform fishes such as southern hake Merluccius australis and Patagonian grenadier Macruronus magellanicus, spawn during austral winter on the continental shelf-break, out fjords SL and UJL were determined separately for each species by linear regression analysis, and comparisons of adjusted means were performed with one-way ANCOVA (Zar 1999), setting significance at P < α = 0.05.
The gut of each larva was dissected from the body and opened lengthwise with fine needles. Prey items were counted and identified to the lowest possible taxon under a Motic® BA310 microscope. The maximum length and width of each prey item were measured with an accuracy of 1 μm using the same equipment mentioned above. The maximum prey width (MPW) was compared among species with the non-parametric Kruskal-Wallis test. The volume of each prey item was estimated using the three-dimensional shape that most closely resembled the item, according to Cass-Calay (2003) and Sun & Liu (2003). The prosome length of copepodite prey was measured since the urosome was often missing.
Feeding incidence (FI) was calculated as the percentage of the total number of individuals examined with at least one prey item in the gut contents (Vera-Duarte & Landaeta 2017). All larvae with identifiable prey in their guts were used for the analysis. The diets were described with the Sea Surface Temperature (SST) was obtained through MODIS satellite systems (NASA), which were distributed by Ocean Color Web software 1 . For this purpose, the average SST was calculated to show the surface thermal distribution of the Chilean Patagonian Sea during the cruise.
In the laboratory, all fish larvae from plankton samples were separated, counted and identified to the lowest possible taxon, under stereomicroscope following to Fisher (1959), de Ciechomski (1971, Sabatés & Olivar (1990) and Bustos & Landaeta (2005). The fish abundance was standardized to individuals 1000 m -3 .
The standard length (SL, from the tip of the snout to the tip of the notochord or to the base of the hypural plates) and upper jaw length (UJL, from the tip of the premaxilla to the distal end of the maxilla) of undamaged larvae (n= 71) were measured to the nearest 0.01 mm under a stereomicroscope Olympus® SZ-61 with a video camera Moticam® 2500 (5.0MPixel) connected to a PC with Moticam® Image Plus 2.0 software. Larvae measurements were not corrected for shrinkage. The relationships between  (Pinkas et al. 1971), calculated as IRI= (%N + %V) x %F, where %N is the percentage of the total numbers of prey items in their guts, %V is the percentage of volume of each item and %F is the frequency of occurrence of a prey item in larvae with food in their guts. To readily allow comparison among prey items, the IRI was standardized to %IRI for each prey item (Cortés 1997). Additionally, trophic overlap was measured using proportional similarity or Schoener`s D index, D= 1 -0.5 ∑ i (|p i -q i |), where p i is the proportion of prey item i used by species p and q i is the proportion of prey item i used by species q. This index varies between 0 (no trophic overlap) and 1, where values above 0.6 indicate overlap (Schoener 1974, Balbontín et al. 1997.
Size of larvae ranged from 2.91 to 90.95 mm SL (mean ± SD, 27.6 ± 29.29 mm SL), while upper jaw length varied between 0.10 and 5.45 mm (UJL: 0.93 ± 1.12 mm). Linear regressions between SL and UJL were significant in almost all larval species (P < 0.05), except for Sprattus fuegensis larvae (P = 0.425). In addition, M. parvipinnis larvae showed a larger UJL compared with L. blainvilleanus (oneway ANCOVA, F 1, 44 = 16.13, P < 0.001). Larval size range in M. parvipinnis was narrower than in L. blainvilleanus, but the proportion of UJL/SL was smaller for the latter. Differences in mouth gape directly affect feeding incidence, and the ability to capture and handle prey of different size ranges (Sabatés & Saiz 2000, Salas-Berríos et al. 2013. The spatial distribution of the larvae during winter was not homogeneous throughout the area (Fig. 1A). Fish larvae were detected at Guafo mouth, at the northern entrance of the Moraleda channel, along the Puyuhuapi and Aysén fjord, and at the mouth of the Concepción channel. Nonetheless, those abundances were lower compared to those found during austral spring in Patagonian fjords, which usually range between 3 and 370 ind. 1000 m -3 (Bustos et al. 2008b, Landaeta et al. 2009, 2011a. The average conditions of SST indicated the presence of relatively warmer waters in the Guafo mouth (> 10.5 ºC), and slightly cooler waters along the Moraleda channel (< 9.5 ºC). The lowest temperatures were registered inside the Puyuhuapi and Aysén fjords and in the Concepción channel (~8 °C) (Fig. 1B). SST in the area was lower compared to SST in the inner sea of Chiloé during austral spring (> 11 ºC;Landaeta et al. 2015).
Highest feeding incidences (100%) were observed in Sprattus fuegensis, M. australis and Macrouridae larvae ( Table 1). The diet of S. fuegensis included 10 prey items, which was dominated by calanoid copepodites (%IRI, 37.91%), copepod remains (12.60%) and Calanus sp. copepodites (10.08%). Predation of this species upon copepodites has been previously described during austral spring in fjords and channels of southern Chile (Landaeta et al. 2012). M. australis diet was characterized by high IRI values on copepod remains (88.39%). However, low larval abundances were recorded, probably since spawning occurs in austral winter from July to September (Balbontín & Bravo 1993). Unidentified Macrouridae larvae showed presence of copepod remains (88.50%) and chitinous remains (11.49%). Nonetheless, FI results need to be taken with caution due to the small number of individuals collected. Slightly lower feeding incidences were observed in M. parvipinnis and L. blainvilleanus larvae (~81%). Diet composition of these species was similar (17 prey items); while M. parvipinnis showed presence of copepod remains (17.29%), calanoid copepodites (14.52%) and Paracalanus indicus copepodites (11.29%), diet of L. blainvilleanus was based on calanoid copepodites (57.54%), copepod remains (13.09%) and P. indicus copepodites (4.28%) (Table 1). Accordingly, feeding habits of larval M. parvipinnis during austral spring Trophic overlap among M. parvipinnis, L. blainvilleanus and S. fuegensis larvae was high. Low sampling size precluded the comparison amongst other species. In particular, considerable overlaps were found between M. parvipinnis and S. fuegensis (D= 0.803), L. blainvilleanus and S. fuegensis larvae (D= 0.790), and M. parvipinnis and L. blainvilleanus (D= 0.696). These results suggest that resource partitioning was not occurring between larvae collected in Chilean Patagonia (44-46º30`S) during austral winter conditions. Therefore, the main components of the winter larval fish assemblages are feeding on similar prey items and size ranges. It is plausible to consider that the low abundance of larval fishes and the habitat complexity (channels, fjords, inner seas) may reduce the probabilities of competition among species. It is expected that the trophic overlap maintains during spring, because the dominant larval fish fed on copepods (Landaeta et al. 2012(Landaeta et al. , 2014(Landaeta et al. , 2015, but the large abundance of prey items potentially preclude food competition. Additionally, some oceanographic structures, such as salinity and tidal fronts, may help to increase the aggregation of prey items in oligotrophic conditions. However, to test this hypothesis, it is necessary to collect plankton samples and environmental conditions at smaller spatial and temporal scales.