INTRODUCTION
The small pelagic fish fisheries produce more than one-third of the annual catches worldwide, but their contribution varies widely among stocks due to strong inter-annual fluctuations in recruitment (van der Lingen & Castro, Reference van der Lingen and Castro2004). Survival varies during early developmental stages (eggs and larvae) due either to the effects of the environment on the reproductive physiology of the female fish through changes in fecundity and spawn quality (maternal effects; Castro et al., Reference Castro, Claramunt, Krautz, Llanos-Rivera and Moreno2009a) or to the effect of the environment directly on the eggs and larvae in the form of mortality through predation, starvation, or transport to areas unsuitable for development (Houde, Reference Houde1987). Despite these multiple causes of mortality in young fish and strong variations in recruitment, some species are able to recover rapidly and reach extremely high biomasses in short periods (i.e. a few years).
The anchoveta, Engraulis ringens, is one such species, having historically shown extreme variations in recruitment and stock biomass along its wide distribution range. The anchoveta, is distributed in the Pacific, ranges from 4° to 42°S (Serra et al., Reference Serra, Rojas, Aguayo, Hinostroza and Cañon1979) and shows three important spawning areas, referred to as northern (north of Peru), central (southern Peru–northern Chile) and southern (central Chile) (Llanos-Rivera & Castro, Reference Llanos-Rivera and Castro2004). In the latter, reproduction is most intense in austral winter (July–September) (Cubillos et al., Reference Cubillos, Canales, Bucarey, Rojas and Alarcón1999) when larval food is less available than in summer, the second most intense reproductive period. In summer, larval food is more abundant due to heightened primary production as well as intensified seaward transport in the Ekman layer due to coastal upwelling (Castro et al., Reference Castro, Claramunt, Krautz, Llanos-Rivera and Moreno2009a). Reports have shown that growth rates for anchoveta larvae may be highly variable among cohorts of individuals spawned within a few days of each other during a same season (0.40 mm/d versus 0.57 mm/d; Castro & Hernández, Reference Castro and Hernández2000) and also that very high growth rates may be achieved by some of these cohorts during the same season (Llanos-Rivera & Castro, Reference Llanos-Rivera and Castro2006). How, then, do larvae achieve such high growth rates in winter, when food production is reduced? Some mechanisms should exist to ensure their chances of obtaining food: perhaps food availability occurs in patches (or other concentrating processes) or turbulence levels remain low enough to facilitate encounters between food and larvae (Sundby, Reference Sundby, Marrase, Saiz and Redondo1996). Furthermore, quality larval preys are necessary throughout larval development in order to ensure adequate nutrition during all larval phases. For the southern spawning area, information on this latter environmental factor (food type availability) is still incomplete (i.e. prey volumes and carbon units) and no estimates have been made to transform the food ingested by larvae into carbon or other units to facilitate their incorporation into classic pelagic trophic webs.
The studies about feeding in anchoveta larvae indicate that they feed principally on phytoplankton species such as dinoflagellates and diatoms, mollusc larvae D and copepods eggs during their first exogenous feeding stage. More developed larvae, on the other hand, switch to feeding on more motile preys such as copepod nauplii or larger preys such as copepodites (Muck et al., Reference Muck, Rojas De Mendiola, Antonietti, Pauly, Muck, Mendo and Tsukayama1989; Valenzuela et al., Reference Valenzuela, Balbontin and Llanos-Rivera1995; Llanos et al., Reference Llanos, Herrera and Bernal1996; Llanos-Rivera et al., Reference Llanos-Rivera, Herrera and Bernal2004). Nevertheless, most studies about larval feeding consider a single factor (i.e. maximum width) to be the only proxy for the size of the preys captured by the larvae (Schmitt, Reference Schmitt1986; Balbontín et al., 1997; Morote et al., Reference Morote, Olivar, Villate and Uriarte2008a, Reference Morote, Olivar, Pankhurst, Villate and Uriarteb). The use of one-dimensional measurement although useful in some cases, may be misleading for energy estimates involving different types of preys (phytoplankton versus zooplankton) or food items of the same width but different lengths. Moreover, a given prey item may vary its aspect ratio during development, all the while retaining a constant width (i.e. some copepod nauplii). For estimates of energy or matter content of ingested larval prey, a few other measurements allow estimations of volume that provide more reliable, relevant information on individual larval requirements for growth (e.g. quantity or quality of food, variations in food during early ontogenetic stages). The present study reports information about feeding incidence, the kind, size, and carbon contents of preys, as well as variations in these from the first feeding stage until reaching 14 mm standard length (SL) for larval Engraulis ringens collected in the southern spawning area (Central Chile). The larvae were collected in early winter (spawning season), when feeding conditions are supposedly less favourable than in summer. The information reported exemplifies the changes in the relative importance of different types of preys when they are considered on different scales of comparison (linear, volumetric and carbon content) as well as the relative importance acquired by the different types of preys as larval development proceeds. The latter aspect is particularly important in young larvae, especially in mid-winter, during the species' main spawning season.
MATERIALS AND METHODS
Field work
Zooplankton samples were collected in the morning (5:30 and 10:30 a.m., sunrise time: 07:56 hours (www.timeanddate.com)) from three stations during an oceanographic survey carried out in mid-July 2007 off Talcahuano (36°32′S 72°56′W). The samples were taken with a Bongo net (60 cm diameter, 300 µm mesh) equipped with a General Oceanics flowmeter; the towing speed was approximately 2 knots. Samples were preserved on-board with 5% buffered formalin. Water samples to quantify potential larval anchoveta food items were collected in the first 15 m (known vertical distribution for larvae and food items) with 10-l Niskin bottles and then mixed in a 20-l container. Subsamples (250 ml) of the seawater from this mixture were preserved with 3 drops of Lugol and stored in the dark for later phytoplankton analyses in the laboratory. Simultaneously and from the same mixture, additional microplankton samples were obtained from prefiltered seawater (2 l) that had passed through a 150-µm sieve and been retained in a 37-µm mesh sieve. The microplankton retained was then preserved with 5% formalin in 125-ml jars.
Laboratory work
At the laboratory, ichthyoplankton was separated from the rest of the zooplankton sample under a stereoscopic microscope fitted with an ocular eyepiece micrometer. Because eye functionality is coupled with the beginning of the first feeding stage (Muck et al., Reference Muck, Rojas De Mendiola, Antonietti, Pauly, Muck, Mendo and Tsukayama1989; Azócar, Reference Azócar2006), this study relied exclusively on larvae with pigmented eyes and functional mouths. SL and mouth width (maximum distance between mouth lateral sides; Llanos-Rivera et al., 1996) were measured and registered for each larva.
Larval gut content dissection was done with entomological needles on a microscope slide; a drop of 50% glycerine–seawater solution was added to increase the viscosity at the moment of dissection. After dissection, the microplankton ingested was observed, identified, and photographed with a digital photographic camera (Canon Power Shot) connected to a Nikon E 200 microscope. Each food item was then measured to determine maximum width and total length using Image J software.
Microplankton groups and species were identified in the laboratory using a light microscope equipped with phase contrast and following specialized literature for microphytoplankton (Cupp, Reference Cupp1943; Taylor, Reference Taylor1976; Rivera, Reference Rivera1981; Sournia, Reference Sournia1986; Sundström, Reference Sundström1986; Ricard, Reference Ricard1987; Rines & Hargraves, Reference Rines and Hargraves1988; Tomas, Reference Tomas1997) and microzooplankton (Trègouboff & Rose, Reference Trègouboff and Rose1957; Bé, Reference Bé1967; Marshall, Reference Marshall1969; Boltovskoy, Reference Boltovskoy1981, Reference Boltvskoy1999; Hu & Song, Reference Hu and Song2001; Zaleski & Clips, Reference Zaleski and Clips2001; Song et al., Reference Song, Al-Rasheid and Hu2002; Fernandes, Reference Fernandes2004). The counting was done with an inverted microscope using the Utermöhl method following the procedures described in UNESCO (Reference Sournia1978), Villafañe & Reid (Reference Villafañe, Reid, Alveal, Ferrario, Oliveira and Sar1995) and Boltovskoy (Reference Boltovskoy1981). The microphytoplankton and microzooplankton abundances are expressed in cell l−1 and ind. l−1, respectively.
Data analysis
Based on the relative larval abundance at different lengths, larval data were classified into three larval size intervals (<6 mm, 6 to 9 mm and >9 mm) from which similar number of larvae were analysed. The first corresponds to the first feeding stage (Ware et al., Reference Ware, Rojas de Mendiola and Newhouse1981). Feeding incidence (FI), defined as the percentage of larvae with at least one prey in the digestive tract (Balbontin et al., Reference Balbontin, Llanos-Rivera and Valenzuela1997), was estimated for each larval interval. Gut contents were identified to the lowest possible taxa.
To characterize the diet, typical larval feeding indices were computed and the prey volume and carbon content were estimated. The relative importance index (RII) corresponds to the product between the numeric occurrence (total number of a certain prey over the total number of all preys in all stomachs, %NO) and the frequency of occurrence (number of stomachs in which one item occurs over the total analysed stomachs, %FO) (Holden & Raitt, 1974 fide Valenzuela et al., Reference Valenzuela, Balbontin and Llanos-Rivera1995). For those larvae with higher RII, the volume of each ingested food item was estimated by assuming the most similar geometric form to the item (Sun & Liu, Reference Sun and Liu2003). The carbon content of each prey present in the larval gut (typically copepods or phytoplankton cells) was estimated by means of the algorithms proposed by Uye (Reference Uye1991) for the copepod Paracalanus sp. (the most abundant copepod genus in this area and period; Castro et al., Reference Castro, Daneri, Escribano, Farías, González, Morales and Pizarro2009b), whereas for phytoplankton preys we used the algorithms proposed by Edler (Reference Edler1979). Linear regressions were carried out between larval lengths and width and between the volume and carbon content of the most important preys (those with higher RII). In these regressions, the preys were identified taxonomically in order to observe the value contributed by each one. Finally, the mean accumulated carbon ingested per larva and the mean number of preys per larva, were plotted for the three larval size intervals, allowing us to differentiate the total carbon consumption and number of preys consumed by each larval size.
To test for differences between FI values among size intervals, a Chi-square (χ-2) analysis was carried out (contingency table). Differences in the width, volume, and carbon content of preys between larval size intervals were evaluated with a Kruskal–Wallis (non-parametric) test. All statistical tests were carried out utilizing the software STATISTICA 6.0.
RESULTS
The larval sizes analysed ranged from 3.98 to 13.93 mm SL and the widths of their preys in the guts ranged 0.16 mm. The prey width range decreased at larger larval lengths. An important proportion of analysed larvae corresponded to yolk sac larvae, that is, with incomplete eye pigmentation and prior to the onset of exogenous feeding. For those larvae in the feeding stage, the %FI changed significantly between size intervals (C-contingency = 0.32; P < 0.01), increasing linearly towards the larger larvae (Table 1).
Table 1. Number of larvae analysed, larvae with contents in their gut, and food incidence (percentage) per larval size interval.
Potential edible plankton collected from the water column during the same cruise (Table 2) showed a high number of diatoms, of which Asterionellopsis, Skeletonema and Chaetoceros were the most abundant. Diplopsalis minor was the only microflagellate collected. Typical microplanktonic items were nauplii, nude ciliates and copepod eggs in groups. Coscinodiscus, Protoperidinium, Gymnodinium, Dinophysis, mollusc larvae and small copepodites were not found in the analysis, although they are usually reported in anchoveta gut contents in other areas (Muck et al., Reference Muck, Rojas De Mendiola, Antonietti, Pauly, Muck, Mendo and Tsukayama1989; Llanos et al., Reference Llanos, Herrera and Bernal1996; Llanos-Rivera et al., Reference Llanos-Rivera, Herrera and Bernal2004).
Table 2. Potential larval food items (phytoplankton and microzooplankton) present in the water column during sampling off Talcahuano. The mean was obtained from three samples taken at 0 and 15 m. Mean and standard deviation (SD) are cells per ml (diatoms and microflagellates) and individual per l (microzooplankton). Items without SD were present in only one sample.
Of the 96 analysed larvae, 60 contained preys in their gut, the largest number of prey items being eight per larva. Both phytoplankton and zooplankton preys were observed in all larval size intervals. The number of different items was lowest in the first of these intervals (<6 mm), with digested remains of phytoplankton (PY), copepods egg (CE), copepod nauplii (NA), pollen grains (PG) and unidentified preys (UN). In the second and third size intervals (6–9 and >9 mm), the number of different items increased with the incorporation of copepodites (CP) and digested unidentified zooplankton items (ZO) (Figure 1).
Fig. 1. Preys consumed at each larval size interval (<6 mm, 6–9 mm and >9 mm). PY, phytoplankton; CE, copepod eggs; NA, nauplii; CP, copepodites; ZO, zooplankton remains; BL, bivalve larvae D; PG, pollen grain; UN, unidentified.
Numerically, the diet of larvae <6 mm SL consisted of 37% PY and 42% NA. A high percentage of nauplii were found in larvae between 5.5 and 6 mm SL. In the middle size interval the copepodite item was incorporated into the diet (17%). Larger larvae (>9 mm) presented an important increase in copepodite preys (reaching 29.3% NO), whereas PY were less abundant and nauplii abundances remained the same as in the other larval size intervals.
The least represented prey items were copepod eggs and bivalve larvae D, which were found in larvae smaller than 6 mm (<13% of total preys). The unidentified digested zooplankton (ZO) started to show up in larvae over 6 mm SL, peaking (9%) in the larger larval size interval, coincident with an increased percentage of zooplankton (>70%; Figure 2).
Fig. 2. Relative importance index (RII) expressed as a percentage of the principal preys consumed.
The analysis of the diet using the RII coincided with the %NO described. PY was an important fraction in the larval diet in the first feeding stage (47.6% RII), decreased in the intermediate larval size interval and reached an RII of only 2% in the largest larvae. For longer larvae, the diet increased in zooplankton items (copepod eggs, copepod nauplii and copepodites); almost doubling the RII of the phytoplanktonic contribution in the last two size intervals (47.6% versus 87.2–88.6% RII). Copepod eggs were a constant component of the diet of these larvae, but they did not exceed a RII of 10% in all size intervals. Copepod nauplii, the most important prey in the diet, increased in % RII along with increasing larval size, although the highest RII value was found in the middle size interval. Copepodites showed a nearly linear increase in RII with larger anchoveta larvae, although these RII values were always far below those of the nauplii (Figure 2). Hence, copepod nauplii were the most important prey quantitatively from the first feeding stage until at least 14 mm SL, as registered for this work.
A positive linear relation existed between larval length and prey width (P = 0.0008) (Figure 3). Maximum prey widths generally corresponded to copepod nauplii and copepodites, whereas the lowest values corresponded mostly to phytoplanktonic preys. Larvae <12 mm SL still fed on small preys (18–100 µm); these were mainly phytoplankton. The consumption of copepodites (73–180 µm mm width) began in larvae of 6.2 mm SL. Larvae between 6 and 10.2 mm SL presented all different types of preys. Likewise, preys with maximum width between 70 and 110 µm were ingested by practically the entire larval size range studied (Figure 3).
Fig. 3. Relationship between larval size and prey width, volume and carbon content.
Prey volume was also significantly related (P = 0.000001) with larval length, showing a higher correlation coefficient than prey width (Figure 3). Copepod nauplii and copepodites presented higher volume values than the other preys, whereas the lowest volumes were presented by PY. Expressing the prey size in terms of its volume accentuates the finding that larger larvae eat the biggest preys.
The carbon content in the diet estimated for each prey was based on their width and/or volume values (Figure 3). This allowed us to determine: (i) a positive relationship (P = 0.0000002) between carbon content in the diet and larval length (longer larvae consume more carbon); (ii) that copepodites were the prey item with the highest carbon content; and (iii) that in the lower larval size interval (<6 mm SL), nauplii contributed the greatest amount of carbon content despite the higher abundance of phytoplankton preys.
Linear regressions between prey width, volume, and carbon content, and the length of the larva that preyed on each item showed negative, non-significant slopes for phytoplankton; positive, non-significant slopes for copepod eggs and nauplii; and positive, significant slopes (P < 0.05) for copepodites (Table 3).
Table 3. Linear regressions between prey widths, prey volume, carbon content and larval length. PY, phytoplankton; CE, copepod eggs; NA, copepod nauplii; CP, copepodites.
**, significant values (P< 0.05).
In terms of individual prey items, the highest carbon content was contributed by copepodites and the lowest by phytoplankton. The nauplii contribution was equivalent to 25% of the carbon contribution of a copepodite (Figure 4).
Fig. 4. Carbon content of principal preys in all anchoveta larvae. PY, phytoplankton; CE, copepod eggs; NA, nauplii; CP, copepodites.
The average number of preys per larva for each size interval (Figure 5) differed among larval size intervals (Kruskal–Wallis; H (2,54) = 7.52; P = 0.0232). The mean total carbon consumption by larval size interval (obtained from total carbon consumed of each larva) also showed significant differences among the three size intervals (Figure 5, Kruskal–Wallis; H (2,54) = 8.87; P = 0.012). In both cases (number of preys and carbon content), the first two larval size intervals were more similar between them than compared with the largest size, which increased markedly, particularly in terms of carbon content.
Fig. 5. Number of preys per larva and carbon content in larval guts estimated for each larval size interval.
DISCUSSION
This study described the diet composition of E. ringens during its early ontogeny using prey characteristics that provide information not reported previously. Our results show that as larval size increases, so does the index of feeding incidence, suggesting greater larval capture ability (mobility, sensory organs and vision). The feeding incidence values reported herein ranked high in comparison with those observed by Balbontín et al. (1997) for E. ringens along the coast of central Chile during a 1990 sampling. For other Clupeiform species (i.e. Sardinella aurita larvae off the southern Catalan coast; SL <9 mm), Morote et al. (Reference Morote, Olivar, Villate and Uriarte2008a) also found high values of incidence (67–77 %), but these decreased as the larval SL increased (9–12 mm). The high values of incidence estimated (45–87 %) in our study are interesting because sampling was done in winter when environmental conditions are usually less favourable for larval feeding than in austral spring or summer, periods of the year with higher productivity due to coastal upwelling (Castro et al., 2009 a, b).
The dietary composition of the anchoveta larvae shows that smaller larvae consumed more phytoplankton and nauplii than copepodites, that other items were incorporated into the diet later and that a change occurred in the diet as the larvae increased in length (phytoplankton consumption decreased, nauplii changed little, copepodite consumption increased). Similar results were reported for the same species along the Peruvian coast in February and March 1982 (Muck et al., Reference Muck, Rojas De Mendiola, Antonietti, Pauly, Muck, Mendo and Tsukayama1989) and for other Clupeiform species such as Sardinella brasiliensis (Kurtz & Matsuura, Reference Kurtz and Matsuura2001). For northern anchovy (E. mordax) off California, this change in diet was observed in laboratory experiments utilizing 10 and 20-mm SL larvae: the latter were able to select the largest available preys (Schmitt, Reference Schmitt1986). Unlike E. ringens, phytoplankton is not highly important in the diet of E. encrasicolus, at least not in the first feeding larvae (Conway et al., Reference Conway, Coombs and Smith1998). Despite phytoplankton was always present in the stomachs of E. encrasicolus larvae, it was not found in large quantities. In other words, it seems most larval anchovy broaden their feeding spectra as they grow but continue to feed on phytoplankton, a situation already described for the Clupeiform Sardinella aurita (Morote et al., Reference Morote, Olivar, Villate and Uriarte2008a). For E. encrasicolus, nauplii are the main prey in the diet, coinciding with our results. Copepods eggs were also primordial prey in the E. encrasicolous diet, but they were not of major relevance for E. ringens.
In terms of species richness, the total number of items found in the digestive tract of the anchoveta was within the range reported for this and other species, although it is strange not to find very frequently items present in most Clupeiforms (i.e. tintinnids, dinoflagellates such as Protoperidinium and the large-sized diatoms Coscinodiscus); these items are also normally present in coastal upwelling areas such as the Talcahuano zone, where this study was carried out. The absence of these items in the diet, nevertheless, coincided with their low numbers in the microplankton samples obtained during the sampling period (Table 2), thereby explaining this difference in the results as compared with other studies carried out on the same fish species in other months (Muck et al., Reference Muck, Rojas De Mendiola, Antonietti, Pauly, Muck, Mendo and Tsukayama1989; Llanos et al., Reference Llanos, Herrera and Bernal1996; Balbontin et al., Reference Balbontin, Llanos-Rivera and Valenzuela1997; Llanos-Rivera et al., Reference Llanos-Rivera, Herrera and Bernal2004).
The larval morphometric relationships determined in the present study are similar to those obtained in other studies on Engraulis ringens (Muck et al., Reference Muck, Rojas De Mendiola, Antonietti, Pauly, Muck, Mendo and Tsukayama1989; Valenzuela et al., Reference Valenzuela, Balbontin and Llanos-Rivera1995; Llanos et al., Reference Llanos, Herrera and Bernal1996). As the mouth gape increases with larval length, so does the width of the preys. These results suggest that larvae might be seeking a particular type of food (i.e. different stages of a same copepod species) but the identification of the intestinal contents evidenced that several items shared a similar corporal width.
The different dietary characteristics for the larval size intervals were more notable when considering the ingested carbon content. Although copepodites were consumed less frequently than nauplii, the former were the most important carbon source in the higher larval size interval. Alternatively, nauplii contributed the greatest amount of carbon in the smallest larval interval. Another interesting aspect was the conspicous (exponential) increase in carbon consumption in the higher larval size interval with respect to the other two intervals. This result is not the rule for larval fish. In larvae of the tuna Auxis rochei, for instance, Morote et al. (Reference Morote, Olivar, Pankhurst, Villate and Uriarte2008b) also observed evident changes in carbon consumption with larval development, although the changes were more notorious at shorter larval lengths (3 mm LS and >5 mm LS).
In summary, larval anchoveta in the Talcahuano area showed a high feeding incidence with a mixed diet initially dominated by phytoplankton (small larvae) and later switching to zooplankton preys (older larvae). While larval anchoveta grow, they feed on preys whose widths do not vary much compared to their body length and volume. These preys might be different taxa or a single species whose widths do not vary much among life stages (i.e. nauplii and copepodites). Differences in carbon content were observed among food items and a conspicuous increase in carbon consumption was observed in the larger larval sizes (>9 mm SL). These ontogenetic changes in feeding are coincident with the increased proportion of larger preys and also with the number of preys consumed by larger larvae. These results contrast with other studies in which the proportion of larger food items increased with larval length but the number of ingested preys decreased (Hillgruber et al., Reference Hillgruber, Kloppman, Wahl and von Westernhagen1997; Morote et al., Reference Morote, Olivar, Villate and Uriarte2008a).
ACKNOWLEDGEMENTS
The authors thank the entire scientific staff at the Laboratory of Reproductive Ecology, Universidad Arturo Prat, Iquique, and the people at the Fisheries Oceanography and Larval Ecology Laboratory (LOPEL), Universidad de Concepción, for their help during cruises and plankton collections. Dr G. Herrera and S. Soto helped with laboratory techniques. Financial support was provided by FONDECYT grant 1070502 to L.R. Castro and G. Claramunt.
