Otolith elemental composition reveals separate spawning areas of anchoveta, Engraulis ringens, off central Chile and northern Patagonia

1 Programa de Magíster en Ciencias con mención en Pesquerías, Universidad de Concepción, Chile. (MJC) E-mail: mariajosecuevasc@gmail.com. ORCID iD: https://orcid.org/0000-0001-7487-4905 2 Centro COPAS Sur-Austral y Laboratorio de Oceanografía Pesquera y Ecología Larval (LOPEL), Departamento de Oceanografía, Universidad de Concepción, Chile. (LC) (Corresponding author) E-mail: lecastro@oceanografia.udec.cl. ORCID iD: https://orcid.org/0000-0001-7665-7883 3 Departamento de Ecología, Facultad de Ciencias y Centro de Investigación en Biodiversidad y Ambientes Sustentables (CIBAS), Universidad Católica de la Santísima Concepción, Concepción, Chile. (KG) E-mail: konrad.gorski@uach.cl. ORCID iD: https://orcid.org/0000-0003-1154-7717 4 Instituto de Ciencias Marinas y Limnológicas, Facultad de Ciencias, Universidad Austral de Chile, Valdivia, Chile. 5 Departamento de Sistemas Acuáticos, Facultad de Ciencias Ambientales, Universidad de Concepción, Concepción, Chile. (AV) E-mail: aurelien.vivancos@gmail.com. ORCID iD: https://orcid.org/0000-0002-3229-1047 6 Department of Chemistry, University of Otago, New Zealand. (MR) E-mail: malcolm.reid@otago.ac.nz. ORCID iD: https://orcid.org/0000-0002-4129-0214


INTRODUCTION
The anchoveta, Engraulis ringens (Jenyns, 1842), is a coastal pelagic anchovy that is widely distributed throughout the eastern Pacific Ocean, from northern Peru (4°30′S) to Chiloé Island in southern Chile (44°14′S) (Serra 1983, Pauly andTsukayama 1987), and is the target of one of the largest pelagic fisheries in the world. This species forms very dense shoals and is strongly influenced by biotic and abiotic environmental factors at all stages of its life cycle (Alheit and Ñiquen 2004). A number of early life history characteristics and reproductive traits, such as egg size, larval size and yolk sac size of recently hatched larvae of the anchoveta, have been described to vary along most of its range of distribution (Llanos-Rivera and ). At its southern limit of distribution (41°-44°S), however, few studies of biology and ecology of adult anchoveta are available. The southern zone of the species distribution, named as the central-southern Chilean fishery zone for administrative purposes, covers a wide latitudinal range (33°S and 44°S; Fig. 1) in which several historic spawning locations have been identified along the coast . In recent years, early developmental stages of anchoveta were found in the interior sea of Chiloé Island (northern Chilean Patagonia; 41°-43°S) (Landaeta and Castro 2006, Landaeta et al. 2009. Asynchrony and significant differences in spawning season have been documented within the central-southern Chilean fishery zone of anchoveta, where spawning occurs from July to December, with higher gonadosomatic indices from August to November below latitude 38°S and from September to October above latitude 38°S , Claramunt et al. 2014. Furthermore, significantly larger females have been found above latitude 38°S ).
Evidence suggesting some spawning stock and population connectivity in this central-southern fishery has been proposed. Soto-Mendoza et al. (2012) used individually based models and concluded that a significant northward transport of eggs and larvae occurs from Corral (39°52'S) towards central Chile (36°-37°S). Other methodological approaches, including analyses of morphometrics, electrophoretic protein genetics, DNA and parasites (Ferrada et al. 2002, Valdivia et al. 2007, George-Nascimento et al. 2015, have been used to assess spawning stock structure of anchoveta from northern to central Chile but have not been conclusive (and have sometimes been contradictory) in terms of the number of stock units reported. The stock structure of this fishery remains poorly defined. Furthermore, it is unknown whether the individuals found in the interior/exterior sea of Chiloé Island constitute a separate unit (Zúñiga and Canales 2014) or are part of the central-southern fishery unit that migrates during the annual cycle.
Understanding of fish population structures and determination of discrete populations or spawning stock units is essential for sustainable management of fisheries resources (Sinclair 1988, Sinclair and Smith 2002, Cadrin et al. 2005. Spawning stock identifica-tion is a complex task in the pelagic environment. Most common methods for distinguishing between marine populations include analyses of morphometric variation, parasitic fauna, biological markers (gonadosomatic index and sexual maturity curve) and genetic markers (Cadrin et al. 2005). In recent years, trace element composition of otoliths has become a common and effective tool for identifying the spatial heterogeneity of populations and has helped identify migratory patterns and differentiate fish stocks (Hicks et al. 2010, Morais et al. 2010, Avigliano and Volpedo 2016. This technique is based on the fact that otoliths originate during the embryonic stage and grow continuously throughout the life of the fish. As they grow, otoliths show characteristic morphologic and chemical composition patterns. Otolith growth that occurred during the pelagic larval stage can be distinguished from growth during the juvenile or later adult period. The ability to correlate a particular part of an otolith with a discrete point in time is one of the most valuable aspects of this structure (Campana 1999, Chang and Geffen 2013, Górski et al. 2015. Furthermore, otoliths do not seem to experience chemical reabsorption for the majority of elements and therefore reflect the physico-chemical characteristics of water, forming a permanent record of the past environmental conditions experienced by fish (Bath et al. 2000, Rooker et al. 2003, Thorrold et al. 2007.
The objective of this study was to evaluate whether otolith elemental composition could indicate changes or persistence in habitats occupied by anchovetas and thereafter help to determine whether the central-south fishery of Chile can be divided into sub-units corresponding to different areas of spawning and growth or, alternatively, the level of mixing of specimens throughout the area shows that such sub-units did not exist. To address this question, the elemental composition of otoliths from individuals caught off central-southern Chile in 2014 and 2015 was analysed.
At both locations, specimens of Engraulis ringens were obtained from commercial catches in autumn and spring 2014 and summer 2015; only the general catch area was reported, as the exact catch locations are commercially sensitive (Fig. 1). Fish samples were selected at random. Subsequently, sex, weight and total length (TL) were recorded for each fish that was selected for otolith analysis. The fish were sorted into two age groups following the age-length key developed by Araya et al. (2008) and otolith ring counts. Group 1 included adult fish (11.5 and 13.4 cm TL; one year old) and Group 2 included adults larger than 13.4 cm TL (>two years old) ( Table 1). Selected specimens were stored in a freezer (-20°C) until analysis at the Fisheries Oceanography and Larval Ecology Laboratory (LOPEL) of the University of Concepción, Chile. Individuals obtained from the central zone and northern Patagonia were similar in size, ranging from 13.1 to 16.1 cm in the central zone and 13.1 cm to 15.6 cm in northern Patagonia. However, during April and January only individuals two years old or older were present, with the largest specimens being observed in April in the central zone (Table 1).

Otolith preparation
After the sagittal otoliths had been extracted from each fish and any adhering tissue had been cleaned away by rinsing with Milli-Q water, the dry samples were stored in 0.5 mL polypropylene micro tubes. The left otolith of each individual was then mounted on a glass slide with epoxy resin and sonicated. The slides were stored in polyethylene bags for later microchemical analysis.

Microchemical analyses
The elemental composition of each otolith was analysed using laser ablation inductively coupled mass spectrometry (LA-ICP-MS) at the Centre for Trace Element Analysis at the University of Otago in Dunedin, New Zealand (the centre uses an ultraclean laboratory for analyses of this type). Depth profiles were obtained along ablation transects from the distal otolith surface (convex side), through the core, to the proximal surface (concave side) (Macdonald et al. 2008, Warburton et al. 2016). This was done using an ICP-MS Agilent 7500cs coupled to an ASI Resolution M-50 laser ablation system powered by a Coherent 193 nm ArF excimer laser.
The otoliths were mounted on a glass slide and the core was located visually through a video imaging system at 400× magnification. Each sample was ablated continuously in a cross section of the unpolished otolith using a 75 micron laser beam. The laser was operated at 5 Hz with an on-sample fluence of 2.2 J cm -2 . Ablation occurred in an atmosphere of pure helium to minimize the re-condensation of the materials and potential elemental fractioning (Eggins et al. 1998). For each otolith ablation, data were collected for up to 450 seconds to ensure that the transect completely traversed the core. To reduce the delay time associated with cleaning, no signal smoothing collector was used (Woodhead et al. 2008). Count rate data were collected for the ions of the following isotopes: 24 Mg, 31 P, 27 Al, 43 Ca, 55 Mn, 60 Ni, 66 Zn, 88 Sr, 63 Cu, 85 Rb, 138 Ba and 208 Pb. Concentrations of these elements are usually sufficiently high in the environment (Campana 1999). NIST 612 calibration glass and MACS-3 otolith reference material were run, bracketing groups of four otoliths.

Data analyses
Raw count rate data were reduced to molar ratios to calcium using the IOLITE software package (Iolitesoftware.com version 2.5). The means of the MACS-3 analyses were typically within 5% of the recommended values, with a precision generally better than 3% (relative standard deviation) for most elements. The relationship between Ca normalized concentrations of the elements Mg, Al, P, Mn, Ni, Zn, Cu, Rb, Sr, Ba and Pb was recorded. Mg:Ca and P:Ca concentration are known to decrease throughout the ontogeny of sardine, because of physiological factors influencing the deposit rate of these elements on the otolith (Javor and Dorval 2016). This happened to also be the case in our data from anchoveta otoliths, as Mg:Ca and P:Ca were always strongly correlated over time (Pearson productmoment correlation p<00.1 for each transect). Hence, Mg:Ca and P:Ca often showed a "butterfly pattern"  on the otolith transects, with a high concentration at emergence, slowly decreasing over time and mirroring on each side of the nucleus. Therefore, we used the concentration of these elements to locate the nucleus on the transects (Fig. 2). The element:calcium ratios were averaged for core and edge regions of each of the studied otoliths. For both edges and cores, 20 s of data were averaged, corresponding to 15 micrometres. Fifteen micrometre averages were used to ensure that only the early larval period was covered in core samples (Plaza and Cerna 2015). Mg:Ca and P:Ca were always strongly correlated and therefore only Mg:Ca was used in the statistical analyses. A permutational multivariate analysis of variance (PER-MANOVA, Anderson 2001) was used to determine differences in the element concentrations (Mg, Al, Mn, Ni, Zn, Cu, Rb, Sr, Ba, Pb) of the otoliths between catch zones and period of capture. This provided an analysis of variance for a set of explanatory factors based on dissimilarity measures, thereby allowing differences to be tested at a multivariate level and across a wide range of distributions of empirical data. The analyses were based on Euclidean distance matrices. The tests were performed separately for microchemical concentrations of the otolith edge and core.
Principal components analysis was used to illustrate differences in the multi-elemental composition of cores and edges of otoliths among samples (Mg, Al, Mn, Ni, Zn, Cu, Rb, Sr, Ba and Pb). Subsequently, data of each element were tested for normality and homogeneity of variance using the Shapiro-Wilk and Levene tests (original and log(x+1)-transformed data). Only ratios of Sr:Ca, Mg:Ca, Ba:Ca, Zn:Ca, Rb:Ca and Mn:Ca were normally distributed (Shapiro-Wilk test, p>0.05) for both original and log-transformed data. However, only the Sr:Ca ratio met the assumption of homogeneity of variance (Levene test, p>0.05). Therefore, differences between zones were tested using a non-parametric test (Kruskal-Wallis) followed by a Wilcoxon test. The differences were tested separately for cores and edges.

RESULTS
Differences between locations and among seasons were observed in the mean seasonal sea surface temperature along the central-southern coast of Chile. In the Talcahuano area, temperatures were about 1° to 2°C higher than in the Inner Sea of Chiloé area in all seasons (Fig. 3A). The 11-year time series of mean sea surface temperatures in the peak spawning months of anchoveta at each location also showed that mean temperature values fluctuated inter-annually, but mean temperatures in those months were consistently higher in the Talcahuano area than in the Inner Sea of Chiloé (Thno, 11.2°C-12.5°C vs. ISCh, 10.4°C-11.1°C; χ 2 0.05,1 , 22.80, p<0.05) (Fig. 3B). This study used a total of 102 adult Engraulis ringens specimens. Otoliths were extracted for laser ablation from 29 specimens that were one-year-old and from 73 that were two years old or older. In total, 56 specimens came from central Chile and 46 from northern Patagonia (Table 1). The concentrations of different trace elements at the otolith edges showed no significant differences in elemental concentrations for any of the factors (zone, month, group and sex; Table  2, Fig. 4). Furthermore, univariate tests for each of the elements separately revealed significant differences only for Mg (p<0.05), which showed higher concentrations in the central zone than in northern Patagonia (Table 3, Fig. 5).
The PERMANOVA revealed significant differences in core elemental concentrations among zones and months (PERMANOVA; pseudo-F=36.7, p<0.001, pseudo-F=6.7, p<0.001) but not among age groups and sexes (Table 2, Fig. 4). April was the only month that did not differ between zones in pairwise comparisons (p>0.05; Table 4). Univariate analyses revealed differences between zones for some of the elements. Specifically, posthoc comparisons indicated significant differences between zones for Ba, Mg and Pb (p<0.05) and strong trends for Rb and Zn (p<0.09) (Table 3, Fig. 5). Cores of otoliths of specimens from the central zone had

DISCUSSION
The objective of this study was to evaluate whether the elemental composition of otoliths could serve as indicators of changes or of persistence in habitats occupied by the anchoveta Engraulis ringens and, thereafter, to determine whether the central-south fishery of Chile could be divided into sub-units corresponding to different areas of spawning and growth. The results showed significant differences in the elemental composition of otolith cores between the two fishing zones that form the central-south fishery, suggesting that there are two distinct spawning areas and that most individuals spent their early life stages in the area where they hatched. However, low differences in the elemental composition of otolith edges in some short periods indicated the possibility that adults from different origins mixed during the year.
In the otolith cores, the most important differences in terms of elemental composition were higher Ba, Mg, Pb and Zn concentrations in the central Chile zone than in northern Patagonia. In some fish species, magnesium has been positively associated with seawater temperature (Schuchert et al. 2010, Javor andDorval 2017), which is consistent with the sea surface latitudinal temperature gradient along the Chilean coast, where warmer waters were found off central Chile (>12°C) and colder waters occurred farther south (<11.5°C). This pattern is consistent with our results of differences in sea surface temperatures among seasons and during the 11-year time series during the months of peak spawning at both locations, and has also been reported previously in other studies (Cubillos et al. 2015, Lillo et al. 2015. Similar to patterns observed in juvenile sardines Sardinops sagax (Javor and Dorval 2017), here we report a strong correlation of Mg:Ca and P:Ca ratios for anchoveta Engraulis ringens. These concentrations were strong discriminants between sites with significant water temperature differences, and decreased throughout the ontogeny. Therefore, ontogenetic effects should be always considered when spatial patterns and stock structure of anchoveta are evaluated on the basis of microchemical signatures of otoliths.
We found higher concentrations of Ba in otolith cores from fish collected in central Chile (upwelling zone) than in those collected from northern Patagonia (a non-upwelling zone). Ba has a nutrient-like vertical distribution with much higher concentrations in deeper waters (Lea et al. 1989, Nozaki 2001, and its concentrations in surface waters significantly increase during upwelling events (Hsieh and Henderson 2017). Therefore, the higher Ba:Ca ratios we found in fish in the central zone are most probably an effect of upwelling. These findings are concordant with previous observations of Ba:Ca in seawater and fishes in other upwelling areas where increased Ba:Ca has been detected under upwelling conditions (Wheeler et al. 2016, Hampton et al. 2018. A similar nutrient-like distribution might be expected for Sr (De Villiers 1999), though we found no significant differences in Sr:Ca rations in otoliths between central Chile and northern Patagonia. Wheeler et al. (2016) concluded that individual ratios of Sr:Ca in fish otoliths showed little variation in the California upwelling system and did not serve as valuable proxies of oceanic conditions.
Regarding other elements, off central Chile, Ahumada (1995) found higher metal concentrations (Pb, Zn, Cu, Cr, Ni and Cd) from anthropogenic origins in San Vicente Bay (central Chile, 36°44′S, 73°09′W). Reports from other bays in the same area were consistent with these results. For example, Pb was the predominant trace metal found in the seawater of Concepción Bay (central Chile; 36°40′S, 73°01′W) (Carrera et al. 1993). The greater concentrations of Pb and Zn recorded in the anchoveta otolith cores from central Chile in this study might be a result of high heavy metals recorded for this area traced to nearby iron, steel and petrochemical industries. Zinc concentrations in otoliths have previously been used to indicate particular habitats because they are influenced by fish diet (Ranaldi and Gagnon 2008) and tends to be higher in urban coastal zones due to anthropogenic sources (Sturrock et al. 2012).
Unlike the elemental concentrations of the otolith cores of anchoveta, elemental concentrations of the otolith edges did not differ between zones. There are at least three potential explanations for these core differences and edge similarities in elemental composition. Differences among cores and similar edge concentrations might occur if the adult habitat is environmentally heterogeneous, wide in extension and covering inshore and offshore zones, but the spawning habitat where small larvae occur is restricted to the coastal zone given their limited swimming capacities. These across-shelf ontogenetic differences in elemental composition have been observed in other anchovy species in other systems (e.g. E. encrasicolus in the Gulf of Cadiz; Catalán et al. 2014). Under this scenario, if environmental conditions inshore in central Chile and northern Patagonia differ significantly but are similar offshore, the otolith edge elemental concentrations might not differ between individuals collected in the two zones, while their otolith core concentrations might. Heavy metal concentrations are usually reported to be several times higher in coastal than in oceanic waters due to greater continental contributions from urban areas or sites of high industrial concentration (Luoma 1990, Giordano et al. 1992, French 1993. In our study, besides Ba, these higher concentrations allowed us to infer the presence of two discrete spawning grounds. Prior research cruises (September-October 2009Cubillos et al. 2010Cubillos et al. , 2011Cubillos et al. , 2013 verified that, spatially, spawning grounds occur in coastal areas not exceeding 15 nautical miles offshore. However, larger individuals could be found as far as 30 to 40 nautical miles from the coastal spawning grounds, as depicted from the current industrial fishing areas. Consequently, across-shelf movements would homogenize elemental composition differences in the otolith edges of older fishes.
A second explanation might be that the greater swimming capacity of the larger individuals might allow them to migrate north and south, could they could sometime access other spawning grounds where they would mix with specimens from local spawning and nursery areas. If this movement were frequent and involved large fractions of the populations, differences in the core elemental composition would tend to decrease significantly, and any difference in edge elemental composition (such as Mg, Table 4) indicative of different environmental temperature regimes probably associated with latitude should vanish. In this study during autumn (April), only adult fish older than one year of age were analysed (Table 1), and no differences were observed in core elemental composition between individuals from the two analysed zones. This suggests that adult individuals from both zones might mix before returning to the spawning habitats. This pattern of segregation of population units and a short period of mixing in some seasons is common in a number of fish species displaying homing behaviour associated with estuarine conditions, e.g. Engraulis encrasicolus L. in the Mediterranean Sea (Aldanondo et al. 2010, Guidetti et al. 2013. A third potential explanation is that these two spawning units represent pocket populations of a broader stock that migrates and that might shift from year to year depending on annual variations in temperature, prey abundance or other undetermined factors. Unfortunately, because the length of our data series is relatively short (one year), we cannot make strong inferences on the restricted or periodical movements across several years. In other systems, elemental differentiation has been observed in cohorts of small pelagic fishes in some years but not in others (Guidetti et al. 2013). While changes in the main spawning locations along the central-southern fishery zone have been reported at 5-to 10-year periods or longer (Bernal et al. 1983, Serra 1983, Castro et al. 1997, shorter-term (inter-annual) changes in migration patterns have not been reported for anchoveta units along their southern distribution range.
In summary, this study of the elemental composition of otoliths provides evidence of two discrete spawning areas of Engraulis ringens along the centralsouth fishery zone: one in central Chile and one in northern Patagonia, and mixing of adults during some non-spawning periods of the year. This finding agrees with the spatial and structural asynchrony in the reproductive processes in the two areas. Future studies should assess whether this pattern is consistent year to year or varies over longer time periods.

AKNOWLEDGEMENTS
This study was financed by the CONICYT Basal Programme COPAS Sur-Austral PIA PFB31. During the study, the first author was supported by graduate fellowships from CONICYT and COPAS Sur-Austral CONICYT PIA PFR31. LC was also supported by COPAS Sur-Austral CONICYT PIA APOYO CCTE AFB170006 and FONDECYT 1161131. The authors thank J. Betzhold, G. Vásquez, and R. Riffo for collections at the artisanal fishery port and vessels and colleagues at the Laboratorio de Oceanografía Pesquera y Ecología Larval (LOPEL) at the Universidad de Concepción for technical support.