Otolith chemistry analyses indicate that water Sr:Ca is the primary factor influencing otolith Sr:Ca for freshwater and diadromous fish but not for marine fish

Water chemistry is thought to be the primary factor influencing fish otolith chemistry. Experimental results with freshwater and diadromous fish have been consistent with this paradigm, but with marine fish, they have often been ambiguous or contradictory. A review of water chemistry data indicated that Sr:Ca (mmol:mol) levels were higher in marine water than in most freshwater systems and that Sr:Ca variability was lower in marine water than in most freshwater systems. We therefore hypothesized that lifetime otolith Sr:Ca profiles of freshwater fish would exhibit low levels of Sr:Ca with moderate variability, of diadromous fish would exhibit highly variable Sr:Ca levels, and of marine fish would exhibit high levels of Sr:Ca with low variability. Otolith Sr:Ca profiles from 81 species of freshwater, diadromous, and marine fish revealed that freshwater fish had low levels of Sr:Ca and lower variability than expected relative to marine fish, diadromous fish had Sr:Ca levels and variability that were consistent with expectations, and marine fish had high maximum Sr:Ca levels, as expected, and high Sr:Ca variability, similar in magnitude to diadromous fish, which was not expected. These findings indicate that water Sr:Ca is the primary factor influencing otolith Sr:Ca variation for freshwater and diadromous fish but not for marine fish. Résumé : On croit que la chimie de l’eau est le facteur principal qui influence la chimie des otolithes de poissons. Les études expérimentales utilisant des poissons d’eau douce ou des poissons diadromes s’accordent à ce paradigme, mais celles faites sur les poissons marins sont souvent ambiguës ou contradictoires. Un examen des données de chimie de l’eau indique que les valeurs de Sr:Ca (mmol:mol) sont plus élevées dans les eaux marines que dans la plupart des systèmes d’eau douce et que la variabilité de Sr:Ca est plus faible dans les eaux marines que dans la plupart des systèmes d’eau douce. Nous avons donc émis l’hypothèse selon laquelle les profils Sr:Ca des otolithes des poissons d’eau douce sur l’ensemble de leur vie présentent de faibles valeurs de Sr:Ca avec une variabilité modérée, ceux des poissons diadromes des valeurs de Sr:Ca très variables et ceux des poissons marins de fortes valeurs de Sr:Ca avec une faible variabilité. L’étude des profils de Sr:Ca de 81 espèces de poissons dulcicoles, diadromes et marins montre que les poissons d’eau douce ont de valeurs faibles de Sr:Ca et une variabilité plus faible qu’attendu par comparaison aux poissons marins, que les poissons diadromes ont des valeurs de Sr:Ca et une variabilité qui correspondent aux attentes et que les poissons marins ont des valeurs maximales fortes de Sr:Ca, tel qu’attendu, et une forte variabilité de Sr:Ca, de même importance que celle des poissons diadromes, ce qui n’était pas prévu. Ces observations indiquent que le Sr:Ca de l’eau est le facteur principal qui influence la variation de Sr:Ca chez les poissons dulcicoles et diadromes, mais pas chez les poissons marins. [Traduit par la Rédaction] Introduction Fish populations around the world are being profoundly impacted by exploitation, development activities, and climate change (Hilborn et al. 2003). Understanding fish populations is essential to effective management and long-term preservation of these resources. The science of fish otolith chemistry is being used increasingly to improve our understanding of fish populations by identifying qualities such as essential habitats, migration patterns, and population structure (Campana 1999). Some applications of fish otolith chemistry are becoming almost routine, but there continues to be active research into environmental and physiological factors influencing the incorporation of trace elements into otoliths in the hopes of expanding our capabilities with the science. The underlying paradigm of almost all otolith chemistry studies is that water chemistry is the primary factor influencing otolith chemistry. Strontium (Sr) has been the most important trace element investigated for several reasons: (i) its concentration (or molar ratio to calcium (Ca)) is relaReceived 20 January 2009. Accepted 8 July 2009. Published on the NRC Research Press Web site at cjfas.nrc.ca on 13 October 2009. J21005 Paper handled by Associate Editor Steven Campana. R.J. Brown.1 US Fish and Wildlife Service, 101 12th Avenue, Room 110, Fairbanks, AK 99701, USA. K.P. Severin. Department of Geology and Geophysics, University of Alaska Fairbanks, Box 755780, Fairbanks, AK 99775-5780, USA. 1Corresponding author (e-mail: [email protected]). 1790 Can. J. Fish. Aquat. Sci. 66: 1790–1808 (2009) doi:10.1139/F09-112 Published by NRC Research Press tively stable within but varies widely among aquatic habitats; (ii) it is metabolically inert and thought to move passively across membrane barriers separating water from blood and blood from endolymph; (iii) it has a 2+ valence in solution and substitutes for Ca in precipitating otolith mineral; and (iv) it is found in relatively high concentrations in water and otoliths compared with other trace elements (Campana 1999; Secor and Rooker 2000). But, does this paradigm hold equally true for all teleost fish? Experimental studies with juvenile diadromous fish have established unambiguous, positive correlations between the salinity of ambient water and the resulting otolith Sr concentration or molar ratio of Sr to Ca (Sr:Ca) (e.g., Secor et al. 1995; Farrell and Campana 1996; Zimmerman 2005). Field investigations with known life history, diadromous species, often using salinity as a proxy for Sr concentration or Sr:Ca in the water, have produced results consistent with experimental findings (e.g., Limburg 1995; Zimmerman and Reeves 2000; Arai et al. 2003). Practical applications of otolith Sr analyses with diadromous species have reportedly been successful in describing complex life history details and migration patterns for certain fish populations (e.g., Limburg et al. 2003; Yang et al. 2006; Brown et al. 2007). Experimental studies with juvenile marine fish have not established consistent positive correlations between salinity treatment levels and otolith Sr concentration or Sr:Ca levels, or there have been significant interactive effects between salinity and environmental or physiological factors (e.g., Bath et al. 2000; Elsdon and Gillanders 2002; Martin et al. 2004). Field investigations of marine species with samples from known locations or life history qualities have produced more ambiguous results than for diadromous species. Many authors have suggested that temperature or physiological events such as spawning interact with water chemistry in complex ways to influence otolith chemistry for marine species (e.g., Campana et al. 1994; FitzGerald et al. 2004; Ashford et al. 2005). Practical applications of otolith Sr analyses with marine species are sometimes reported to be successful in describing life history details such as stock origins or geographic distribution for certain fish populations; however, they usually required the consideration of additional information including concentrations of multiple elements, isotopes, temperature data, and demographic information (e.g., Severin et al. 1995; Thorrold et al. 1998; Rooker et al. 2003). Although physiology is frequently identified as a factor influencing otolith chemistry of marine fish, it is rarely considered to be a factor for freshwater or diadromous fish. Campana (1999) and Secor and Rooker (2000), in their reviews of the otolith chemistry literature, showed that average otolith Sr levels reported in the literature were greatest in marine, least in freshwater, and intermediate in diadromous fish. This pattern supports the general understanding that Sr levels and Sr:Ca are greater in marine water than in freshwater. Most marine teleost fish species maintain slightly greater fluid solute concentrations than freshwater fish, but both average close to one-third that of marine water (Black 1957). Fish are therefore hypotonic to marine water, where they must work to reduce solute uptake and retain water, and hypertonic to freshwater, where they must work to retain solutes and reduce water uptake. Diadromous species migrate between marine and fresh water and would experience a wider range of osmotic and Sr:Ca conditions than other life history categories. Though it is well established that Sr precipitation in otoliths is a competitive process among 2+ valence ions in the endolymph (Campana 1999; Kraus and Secor 2004), the different osmotic conditions faced by fish in these different environments may affect the manner in which Sr ions become available to the endolymph. Summaries of average water chemistry parameters indicate that freshwater systems (e.g., Livingston 1963; Martin and Meybeck 1979; Goldstein and Jacobsen 1987) are uniformly lower in Sr concentration and Sr:Ca molar ratios than marine systems (e.g., Culkin and Cox 1966). Early work by Odum (1957) and more recent work by Kraus and Secor (2004), however, indicate that freshwater Sr:Ca levels are variable among drainage systems and that a small percentage of rivers exhibit Sr:Ca levels greater than reported for marine water. We reviewed water chemistry data for freshwater and marine systems from online databases and from the formal literature to clarify the nature of Sr:Ca environments faced by fish living within these two habitats. We then modeled the estuarine Sr:Ca levels from a small number of river systems to illustrate this transitional environment. This review provides a more comprehensive understanding of Sr:Ca level and variability among habitats, as they relate to fish and otolith chemistry, than was previously available. We tested patterns of otolith Sr:Ca in freshwater, diadromous, and marine categories of fish for consistency with the paradigm that water chemistry was the primary factor influencing otolith chemistry. Mature-sized, wild fish from these three life history categories were selected. Lifetime Sr:Ca profiles were created with core-to-margin transects to examine Sr:Ca level and variability for each fish. Unknown temperature and physiological effects were expected to influence Sr incorporation into otoliths of all three categories of fish and were considered to be common background factors. Based on our understanding of Sr:Ca differences between freshwater and marine environments, we hypothesized that freshwater fish would exhibit relatively low levels of otolith Sr:Ca with moderate variability, diadromous fish would exhibit high maximum Sr:Ca levels and high variability, reflecting the large range of Sr:Ca environments that they encounter, and marine fish would exhibit relatively high levels of otolith Sr:Ca with low variability. Materials and methods Water chemistry review We reviewed water chemistry data from the United States Geological Survey (USGS 2007) database (n = 4138) and from the geochemistry literature (citations in Appendix A; n = 1861) to improve our understanding of the withinsystem variability and among-category differences of Sr and Ca in freshwater rivers and lakes, marine systems, and estuaries. Only the dissolved components of Sr and Ca were considered. They were standardized to mg kg–1, converted to molar concentrations, and presented as the molar ratios of Sr to Ca (Sr:Ca) reported in units of mmol Sr to mol Ca. Strontium and Ca concentration data were found for 58 Brown and Severin 1791 Published by NRC Research Press lake systems and 786 mainstem and tributary rivers from all continents except Antarctica. Data from closed-basin salt or alkaline lakes were not included in our analyses or figures because the extreme chemistry in these environments often precludes fish occupation and the outlier Sr:Ca values presented scaling problems in our figure. Marine water chemistry data consisted of 171 individual records from published accounts of voyages into most of the major oceans and seas of the world, low and high latitudes, from surface waters to depths of more than 5000 m, and from coral reefs in coastal areas to mid-ocean regions. Data from mid-ocean vents and black smokers were not included. In all, almost 6000 individual measurements of water Sr and Ca concentrations were compiled. Sr:Ca level and variability within and among freshwater and marine habitats were illustrated with a histogram. The variability of Sr:Ca within individual river systems was examined and graphically compared with the marine system using box plots. Sr:Ca data from 20 rivers were examined. Ten large river drainages from several continents for which Sr:Ca data were available from many mainstem and tributary sites were selected to illustrate the Sr:Ca variability within a drainage. Ten large North American rivers for which Sr:Ca data were available for multiple years and seasons from lower-drainage, mainstem sampling sites were selected to illustrate the Sr:Ca variability within a limited reach of a drainage over a long period of time. All 171 marine Sr:Ca observations were included. Three rivers with median Sr:Ca levels less than marine water and one river with a median Sr:Ca level greater than marine water were selected to illustrate the relationship between Sr:Ca and salinity in estuary waters. The selected rivers included the Indigirka (Sr:Ca = 2.32) and the Indus (Sr:Ca = 3.58), which had relatively low levels of dissolved solutes, and the Mississippi (Sr:Ca = 2.01) and St. Johns (Sr:Ca = 13.83), which had relatively high levels of dissolved solutes. Concentrations of Sr and Ca were modeled across an estuarine salinity gradient from 0 to 35 practical salinity units (psu) with a progressive series of complementary mixing fractions of fresh and salt water using the following equation: cai 1⁄4 1⁄2ðcfiÞ ð1 pÞ þ 1⁄2ðcmiÞ ðpÞ where cai is the concentration (mg:kg) of element i in ambient estuarine water, cfi is the concentration of element i in freshwater, cmi is the concentration of element i in marine water, i represents dissolved Sr or Ca, and ambient salinity is p 35, where p ranges from 0 to 1. Freshwater Sr and Ca concentrations used in this model were median values for the selected rivers. Average marine water Sr and Ca concentrations from our literature sources were 7.64 (standard deviation (SD) = 0.53) and 405.87 (SD = 27.49) mg kg–1, respectively, and the mean Sr:Ca level was 8.61 mmol:mol. The variable ‘‘p’’ ranged from 0 to 1 and reflected ambient salinity as a proportion of marine salinity, which was modeled at 35 psu. End-member concentrations of Sr and Ca were used to model Sr:Ca (mmol:mol) values and were plotted against salinity. Fish otolith selection, preparation, and analysis Otoliths from 28 freshwater, 21 diadromous, and 32 marine fish species were examined in this study (Appendix B). Taxonomy and nomenclature in this paper are consistent with Fishbase (2008). Mature-sized, wild-caught fish were used in all cases. They were collected from locations and in circumstances ensuring they had lived only in freshwater, in both freshwater and marine water, or only in marine water, respectively. One sample per species was included in among-habitat analyses. Within-species variability was explored with additional analyses of 20 individuals each from freshwater drum (Aplodinotus grunniens), a freshwater species, Bering cisco (Coregonus laurettae), a diadromous species, and sablefish (Anoplopoma fimbria), a marine species. One otolith from each fish was thin-sectioned through the core and mounted on a glass slide. Most were sectioned in the transverse plane, but some unusually shaped otoliths were sectioned in other planes. Each section was 200– 300 mm thick, and growth increments from the core region to the margin were visible with transmitted light. They were polished on a lapidary wheel with 1 mm diamond abrasive and coated with a conductive layer of carbon in preparation for microprobe analyses. A wavelength-dispersive electron microprobe (microprobe) was used for chemical analyses of otoliths. The technology functions by directing a focused beam of electrons to points on a sample surface. Atoms within the material are ionized by the electron beam and emit X-rays unique to each element. Spectrometers are tuned to count the X-rays from elements of interest. The electron beam used here was 5 mm in diameter and was operated at an accelerating voltage of 15 kilo-electron volts and a nominal current of 20 nanoamperes. According to Gunn et al. (1992), the electron beam penetrates less than 3 mm into otolith material. Strontium and Ca X-rays were counted for 25 s at each point along a core-to-margin transect (core precipitated early in life, margin precipitated late in life) for each otolith. Centerto-center distance between points was 8 mm. Elementspecific X-ray counts at each point are proportional to the concentration of the element in the sampled material (Reed 1997; Goldstein et al. 2003). Strontianite and calcite standards were used with established quantitative procedures to develop conversion factors between X-ray count data and elemental concentration estimates in units of mg kg–1, as detailed by Brown et al. (2007). Elemental concentrations of Sr and Ca were converted to molar concentrations using the following equivalencies: