During the late austral winter and spring of 2010-2012, 13 dusk and nocturnal coastal surveys (1930 to 2300 h) were conducted at El Quisco Bay (33°24′S, 71°43′W) on board an artisanal vessel (Fig. 1). Oblique hauls of a Bongo net (60 cm diameter, 300 μm mesh size) with one TSK flowmeter mounted in the frame of the net were performed for 15-20 min from a depth of 20 m. Seawater filtered by the net ranged from 13.1 to 437.4 m³ (mean±one standard deviation: 141.7±102.5 m³). Subsequently, the nets were washed on board and all zooplankton samples (n=166) were initially fixed with 5% formalin buffered with sodium borate and preserved in 96% ethanol after 12 h. Although there are no formal studies about reproduction seasonality of the two species, the period of study was selected during a season when abundance of larval stages of both species is conspicuous and greater than in other seasons (Hernández-Miranda et al. 2003).

All fish larvae from plankton samples were separated, counted and identified. Larval Pacific sandperch P. jugularis were identified according to features described by Vélez et al. (2003), and Chilean sandperch P. chilensis were identified following Neira et al. (1998). Both larvae have a robust body with a large head bearing small preopercular spines; they differ in the presence (absence) of internal melanophores behind the nape in larval P. jugularis (P. chilensis) (Fig. 2). Abundance was standardized to individuals per 1000 m³, taking into account the number of larvae captured and volume of the seawater filtered by the net. Standard length (SL) and upper jaw length (UJL, from the tip of the snout to the posterior end of the maxilla) of all intact larval P. chilensis and P. jugularis (n=159 and 83, respectively) were measured under an Olympus SZ61 stereomicroscope attached to a camera (Motic moticam 2500, resolution 5.0 MPixel) using Motic Image Plus 2.0 software.

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, except for indigestible prey remains such as setae, under a Motic BA310 microscope. The maximum SL and width of each prey item (maximum prey width, MPW) were measured with the microscope and a Motic moticam 2500 camera (resolution 5.0 megapixels) using the Motic Image Plus 2.0 software. The volume of each prey item was estimated using the three-dimensional shape that most closely resembled the item, following Cass-Calay (2003) and Sun and Liu (2003). The prosome length of copepodite prey was measured because the urosome was often missing.

The relationships between the SL and UJL of the larvae and between the UJL and MPW were determined separately for each species by linear regression analyses, and comparisons of adjusted means were performed with one-way ANCOVA (Zar 1999).

Feeding incidence (FI) was calculated as a percentage of the total number of larvae that had any gut content out of the total number of larvae examined for each species and for larval size ranges. Comparisons of FI among size ranges and between species were carried out with contingency tables.

All larvae with identifiable prey in their guts were used for the analysis. The diet was described using the percentage frequency of occurrence (%F) of a diet item in larvae with food in their guts, the percentage of the total number (%N) of diet items that were examined and the percentage of volume (%V) of each item out of the total volume of prey items. An index of relative importance (IRI) was calculated as IRI=(%N+%V)×%F. To readily allow comparisons among prey items, the IRI was standardized to %IRI for each prey item i (Cortés 1997).

To estimate the feeding success of larvae during their development, three measures were compared: number of prey items per gut (NPPG, number), MPW (mm) and total prey volume per gut (TPVG, mm³) (Reiss et al. 2002, Landaeta et al. 2011). To determine whether these indicators of feeding success were related to the growth of larvae, non-parametric correlation tests were run (Spearman rank tests, rs); if correlation was significant (P<0.05), then the feeding success was related to the larval size, and simple linear regressions were carried out between indicators (NPPG, MPW, and TPVG) and SL for both larval species. In order to compare the feeding success between species, one-way ANCOVAs were run if parameters were significantly correlated with SL. If not, a Mann-Whitney U-test was run.

To compare feeding composition among larval species and size groups, in terms of prey number and volume, two-way PERMANOVAs were run, using the Bray-Curtis similarity index and 9999 permutations, with Past 3.13 software (Hammer et al. 2001).

Intraspecific differences in prey size were examined throughout larval development. Pearre’s trophic niche breadth (Pearre 1986) was adopted to analyse the relationship between prey size and predator size. This model uses the standard deviation (SD) of the log10-transformed prey size as a measure of trophic niche breadth. In this analysis, fish larvae were classified according to body length at 0.1-mm intervals. Only classes with >2 prey item in the gut were used for further analysis. The mean and SD of the log10-transformed prey width was calculated for each available size class of larval fish. The relationship between body length and the corresponding mean and SD of the log10-transformed prey size was examined using linear regression analysis to determine any shifts in niche breadth with growth.

Diet overlap between larval P. jugularis and P. chilensis was measured using Schoener’s overlap index α (Schoener 1970), a robust measure of diet similarity (Wallace 1981): α=1–0.5 Σ|P_xi – P_yi|, where P_xi and P_yi are the frequencies of prey category i in predator species x and y. Index value ranges from 0 (no overlap) to 1 (complete overlap), and overlap, in general, is considered to be biologically significant when the value is >0.6 (Schoener 1970, Keast 1978). Where species differ markedly in their diet, overlap levels will be low; however, where common items make up a significant percentage of the diet of the two species, they will be high, and (theoretically) the species will be in potential competition (Keast 1978). Comparisons were made for similar size ranges of the upper jaw length: <300, 300-500, 500-700, 700-900 and >900 μm UJL.

For the two larval species, size range was similar, varying from 3.83 to 10.80 mm SL for P. jugularis (mean±standard error, median; 5.58±0.07 mm, 5.61 mm), and from 3.49 to 7.71 mm SL for P. chilensis (6.02±0.14 mm, 6.00 mm). Similarly, the UJL of the two larval species was similar in range, 259.4-1485.50 μm for P. jugularis and 256.00-1066.60 μm for P. chilensis. For both species, the relationship between SL and UJL was linear, showing larger variability in P. jugularis (R²=0.667) than in P. chilensis (R²=0.955, Table 1, Fig. 3). The latter may be due to greater variation in the allometric growth of mouth gape and snout in larval P. jugularis (Vélez et al. 2003). For larval P. jugularis, mouth gape corresponded to 5.9% to 19.1% SL (mean±SD, 11.4±3.2% SL), while in larval P. chilensis, mouth opening corresponded to 6.9% to 13.8% SL (9.7±1.5% SL). In the comparison of UJL of the two species (corrected by body length and for the same size range, 3-8 mm), larval P. jugularis had a larger mouth gape than P. chilensis (one-way ANCOVA, F=23.29; P<0.001), and both increased their gape through development at similar rates, 175–183 μm mm^–1 (homogeneity of slopes, F=0.39; P=0.531, Table 1).

Both species showed high FI throughout their larval development (Fig. 4); empty stomach accounted for 19 out of 159 individuals in larval P. chilensis and 6 out of 83 individuals in larval P. jugularis. FI varied from 81.8% to 100% in larval P. jugularis, and from 69.2% to 100% in P. chilensis. There was no significant difference in the FI between species (χ²=1.63, P=0.202), or among size ranges for larval P. jugularis (χ²=5.01, P=0.543); however, the largest sizes of larval P. chilensis showed a significant decrease in FI during development (χ²=19.7, P=0.003).

The diet composition of P. jugularis and P. chilensis was similar (22 and 24 prey items), dominated mainly by copepod nauplii, followed by calanoid copepodites, calyptopes, invertebrate eggs, zoea and early stages of gasteropods and polychaetes (Table 2 and 3).

Nauplii were the most important prey throughout larval development (%IRI~80%-90%, %F>70%), and were numerically important from small to larger larvae (80-70 %N). In terms of %V, copepod nauplii items in P. jugularis remained constant at ~30% in larvae of less than 7 mm, but in P. chilensis they decreased from small larvae to large larvae. In the gut contents of P. jugularis, tintinnids, appendicularians, Calanoides patagoniensis, Clausocalanus arcuicornis, Metridia lucens, euphausiid eggs, myses and fish eggs were also identified (Table 2). In the case of P. chilensis, they fed on balanidae larvae, polychaeta larvae and copepodites of Aetideus armatus, Centropages brachiatus, Corycaeus sp., Oncaea sp. and Pleuromamma gracilis (Table 3). Numerically and volumetrically, copepodites were slightly more important than egg items. In terms of %F, in both diets the most important items were nauplii, invertebrate eggs and Paracalanus indicus. Finally, both diets had similar prey items, such as gasteropod larvae, zoea and some copepodites such as Acartia tonsa, Calanus chilensis and Metridia longa (Table 2 and 3).

Throughout the larval development, the NPPG was low, varying between 1 and 13 prey (mean±SD, 4.2±3.1 prey per gut) in P. jugularis and between 1 and 19 (5.5±3.6) in P. chilensis. In both species, NPPG was independent of larval size (Spearman rs_Pj=–0.14, P=0.080; rs_Pc=–0.029, P=0.732) (Fig. 5A, D).

The total volume per gut (TVPG) ranged from 1.4×10^–4 and 0.027 mm³ (0.008±0.007 mm³) in P. jugularis, while in larval P. chilensis it varied from 3.3×10^–5 to 0.031 mm³ (0.007±0.006 mm³). A significant positive correlation was evident between larval size (SL) and TVPG for both larval sandperches (Spearman rs_PJ=0.259, P<0.01; rs_PC=0.259, P<0.01) (Fig. 5B, E). No significant differences were detected in TVPG between the two species (one-way ANCOVA, F=2.59, P=0.109) (range of comparison, 3-8 mm SL), suggesting that at a given size, total prey volume ingested was equal (P. jugularis 0.006 mm³; P. chilensis 0.004 mm³) (homogeneity of slope test, F=2×10^–4, P=0.987).

MPW ranged between 21.55 and 511.3 μm (105.1±64.5 μm) in P. jugularis, and between 31.4 and 499 μm (82.4±41.7 μm) in P. chilensis. In both species, MPW was positively correlated with larval length (SL) (Spearman rs_PJ=0.417, P<0.01; rs_PC=0.100, P<0.05), suggesting that at larger sizes, larvae select larger prey items (Fig. 5C, F). One-way ANCOVA indicates significant differences in the MPW between P. jugularis and P. chilensis (F=17.72, P<0.01); P. jugularis ingested wider prey (97.3 µm) than P. chilensis (83.0 µm) at the same size (homogeneity of slope test, F=9.85, P<0.001).

Two-way PERMANOVA detected significant differences in prey number and volume composition between the larval species (Table 4), but it found no difference among the size ranges analysed within the species (P>0.05). The latter suggests an absence of differences during the larval development of the two species.

Trophic niche breadth (SD of prey size) was narrow and similar between the two larval sandperches. For P. jugularis it was 0.159±0.07 and for P. chilensis it was 0.156±0.03 (Fig. 6). In both cases, trophic niche breadth was independent of larval size, as suggested by the linear regressions (Table 5).

Schoener’s index of diet overlap varied between 0.291 (no overlap) and 0.909 (high diet overlap) (0.652±0.183) (Table 6). Most of the indices showed a biologically relevant diet overlap, particularly between larvae having an UJL of less than 900 μm, considering intra- and interspecies comparisons. This suggests that resource partitioning is occurring late in the larval development of both sandperches.