Eyes and ears: A comparative approach linking the chemical composition of cod otoliths and eye lenses

Abstract Fish eye lenses are a protein‐based chronological recorder of microchemical constituents that are a potentially useful tool for interpretations of environmental, ecological and life‐history experienced by fish. Here, we present the first study with data on the chemical composition of eye lenses from Baltic cod examined using laser ablation inductively coupled plasma mass spectrometry (LA‐ICPMS) and compare these spatially resolved data to otoliths from the same fish also analysed by LA‐ICPMS, measuring the isotopes 27Al, 137Ba, 43Ca, 52Cr, 65Cu, 57Fe, 39K, 7Li, 25Mg, 55Mn, 31P, 208Pb, 85Rb, 45Sc, 29Si, 88Sr, 47Ti, 50V, 149Yb, 66Zn and 90Zr. Comparison of the variation in element concentrations between eye lenses and otoliths from the same individuals showed minor similarities, suggesting a different governance in the uptake processes. A strong overlap between the concentric growth rings in the eye lenses and the otolith Sr periodicity was observed, where each consecutive minima in the chemical profile with high accuracy correspond to the width of each lens ring. No comparable trends were seen between growth rings and all other elements measured from both lenses and otoliths. The characteristic rings observed in cod eye lenses do not seem to represent seasonal fluctuation nor are they found to be directly linked to age. With this research, we provide a baseline study identifying elements in corresponding eye lenses and otoliths that show potential for unravelling the environmental and biological conditions experienced by fish.

composition of otoliths is regulated by a suite of extrinsic (e.g., water concentration, temperature) and intrinsic (e.g., fish size, growth, sex and food) factors (Anon, 2014;Campana et al., 1996;Sturrock et al., 2012Sturrock et al., , 2015. Patterns within and between individuals yield information on life-history events such as migration patterns and stock structure (Campana, 2005;Campana et al., 1996;Campana & Thorrold, 2001;Svedäng et al., 2010), and in recent years even direct age . Other aspects explored with the various chronological tissues have included diets and ecosystem reconstruction (Tzadik et al., 2017).
Among the chronological tissues, eye lenses are a particularly interesting candidate for inferring new knowledge about fishes' life history. Unlike otoliths and many other chronological tissues, eye lenses have a protein-based structure (Pourang et al., 2018). They consist of a wide range of proteins bound in primarily α and γ crystallines that are thought to make up the structural part of the eye lens (Dove, 1999;Wistow & Slingsby, 2016). The exact construction of these crystallines is less well documented, although it is known that a main component is sulphur (Mahler et al., 2013). Studies on the microchemistry of fish eye lenses are limited and tend to focus on a few selected elements. Some of these studies (e.g., Quaeck, 2017;Quaeck-Davies et al., 2018) have examined the structural design of eye lenses from various species and the behaviour of the physical structure of the lens. These studies have established that eye lenses are a good repository for stable isotope-derived information relating to the entire life history of a fish (Young et al., 2022). Other studies (Gillanders, 2001;Kingsford & Gillanders, 2000;Pourang et al., 2018) have examined the use of eye lens microchemistry in a fisheries management context. Some of these studies show that eye lens chemistry has potential in stock discrimination using elemental fingerprinting, similar to the use of otoliths (Campana, 2005). Other studies have linked eye lens trace elements with environmental factors such as depth, spatial scale and heavy metal pollution (Dove, 1999;Dove & Kingsford, 1998). No study exists to link trace elements in the fish blood plasma with eye lens concentrations like for otoliths (Sturrock et al., 2012(Sturrock et al., , 2013(Sturrock et al., , 2014. This is nevertheless an essential feature to fully understand the uptake processes of trace elements to any chronological tissue and to evaluate to what extent trace elements reflect environmental concentrations or physiological processes. In this study, the microchemistry of corresponding eye lenses and otoliths from Atlantic cod (Gadus morhua) was investigated with the aim of (a) establishing a baseline of trace elements occurring in eye lenses using solution inductively coupled plasma mass spectrometry

| Sample collection
Otoliths and eye lenses of 12 Baltic cod were collected randomly from fisheries samples between 2017 and 2019. Following capture, the length, weight, sex, maturity stage and capture location were recorded. Eye lenses and otoliths were dissected out, dried and stored in paper envelopes. An overview of the biological data and the capture positions of the sampled cod are shown in Table 1

| Chemical analysis
The sampled otoliths were cleaned using an ultrasonic bath with deionized water for 10 min, rinsed with deionized water and left to dry in acid-washed trays in a laminar flow hood following the processing protocol by Hüssy et al. (2021). Lenses were cleaned by removing any adhering tissue and dried for at least 12 months. Lenses and otoliths were then embedded in Struers (Detroit Rd. Westlake, cleavland, OH, USA) cold mounting casting epoxy and sectioned through the core with an Accutom-100 multicutter to expose the entire growth axis from core to edge. The surface of each section was polished using 3 μm abrasive paper mounted on Buehler (Waukegan Rd, Lake Bluff, IL, USA) rotating disks, and then cleaned once more as described  Figure 2) and from core to the dorsal edge in the otoliths.
The LA-ICPMS run conditions are listed in Supporting Information Table S1.
The basic analytical approach and data processing techniques used for analysing otoliths at GEUS are described in Serre et al. (2018).
Additional details and LA-ICPMS settings are included in Supporting Information Table S1, including the analytical precision and accuracy of the LA-ICPMS data (Supporting Information Figure S1). The LA-ICPMS analyses are reported in counts per second (cps) for each isotope measured along the transect. To calculate elemental abundances (e.g., in ppm) we use an internal standard element and an external standard reference matrix (e.g., NIST-610 glass) with known element concentrations. For otoliths, Ca was used as the internal standard element due to the otolith's robust calcium carbonate structure, which was assumed to be 38.3 wt.% Ca (Serre et al. 2018 and references therein) for all otoliths analysed. The otolith data used here were previously included in Hüssy et al. (2021). For eye lenses P was selected as the best internal standard element based on its consistency in apparent concentration as measured by the solution-based ICP-MS analyses, and because P in LA-ICPMS pre-run test analyses together with Mg showed the smallest variation of the elements measured across the growth zones of the eye lenses. A P averaged abundance of 0.09 wt% P 2 O 5 (corresponding to ca. 390 ppm P) as determined by solution ICP-MS was assumed to be representative for all the eye lenses analysed. Variations in Ca content across the transects in the otoliths and P in the eye lenses may of course occur, but there was not the scope here to F I G U R E 2 Images of Gadus morhua eye lenses of two selected individuals (fish ID #3 and #5), where the red dots represent the growth rings identified along the laser transect. The left image shows an ideal example with well-defined growth rings and the right image shows an example where the growth rings are difficult to discern examine all otoliths and eye lenses in detail for variations across the growth zones. In this study we thus assume a more-or-less consistent Ca and P content for the otoliths and eye lenses, respectively. Consequently, any significant variation in the abundance of these internal standard elements will affect the accuracy of all other element concentrations, but it does not affect element ratios determined relative to Ca or P for the otoliths and eye lenses, respectively.

| Eye lens ring identification
As for otoliths, growth zones in eye lenses are concentrically and symmetrically formed around the core. Unlike in otoliths, there is no consensus as to what these rings represent or why they form in this manner. For each cod we have used the optical images of the crosssectioned eye lenses to identify these rings. ImageJ (v. 1.53e) was used to measure the width of each ring along the laser transect line from the core to both the left and right lens edges. Not all individuals showed a clear and easily recognizable outline of the rings. Figure 2 shows an example of (a) easy identifiable rings and (b) rings that were difficult to identify. Where the structure was less clear, e.g., where rings appear to have 'merged' or 'dissipated', the complete rings were identified by superimposing a full circle on the image using an elliptical tool in ImageJ. Any 'incomplete' rings were not considered true rings.

| Data analysis
It was assumed that the chemical profile of each individual is statistically independent, thus outlier removal was performed on each eye lens separately, applying the same procedure for all samples (Zuur et al., 2010). Each individual transect represents a timeline and some elements exhibited large shifts in the concentration along the data transect. To account for the shifts that would otherwise offset standard deviation and mean value estimates, the data from each individual were partitioned into four equal parts for each element, respectively, illustrated by the vertical lines in Supporting Information Figure S2.
Within each section of the partitioned data, any measurements greater than section mean ± 4σ were excluded.
The quality of the laser transect data for both the eye lenses and otoliths was examined by calculating the signal-to-noise-ratios (SNRs), defined as mean 2 /S.D. 2 . This determined which elements showed overall potential for meaningful statistical modelling (SNR > 5). measurements for otoliths and 20 for the lenses. We tested for concentration differences between the eye lenses and the otoliths. In addition, we tried to establish if any lens and otolith concentrations showed the same response to biological factors included here. For elements with SNR > 5, mixed linear models were set up using the R function 'lmer' (Kuznetsova et al., 2017). For this analysis, the factor type was used, denoting either lens or otolith concentration. Fish ID number was added as a random effects term to account for differences in data points between lenses and otoliths. The general full model for the 'all' measurements, for each of the elements considered, were constructed as follows: where Y represents each element modelled, i = 1, … 24 (two measure- To correlate periodic variation in the element concentration to the observed lens rings, the distance between each lens ring was used to parse each LA-ICPMS measurement to the corresponding ring. By overlapping the regression lines with the annotated rings, it was then possible to look for coherence in maxima and minima, and the visible rings, as illustrated in Figure 3. To test whether there is a generic correlation in possible patterns, a mixed-effects model between the distance from element core to local minima or maxima, and distance from lens core to visible rings were set up as follows, with fishID included as a random effect: where Y represents local minima or maxima in the given element profiles, i = 1, … n is the number of consecutive minima, d(fishID) $ N (0,σ fishID 2 ) and ε i $ N 0, σ 2 À Á .
3 | RESULTS   In general, element concentration profile plots across the eye lenses along the transect demonstrate coherent trends for almost all individuals. The alkaline earth metals Mg and Sr typically show an increasing abundance from core to edge of the eye lenses and Mg shows a fairly prominent peak domain from the core to edge area.

| Element concentrations and trends in eye lenses
The elements P and Zn show an opposite decreasing trend from the core region towards the edge, but P has similar-looking peak trends prominent in the core area. The elements Cu and K only show decreasing abundances in a fairly narrow field near the core and edge of the eye lenses profiles and otherwise stable fluctuating element profiles across the transect of the eye lenses. Examples of these trends are shown in Figure 5.
Going forward with analysis and comparison of lens and otolith using LA-IPCMS data, nine elements were included for the otolith analyses (Ba, Ca, Cu, K, Mg, Mn, P, Sr, Zn), with an additional seven elements included for the eye lenses (Al, Fe, Li, Pb, Rb, Si, Ti). Element concentrations, together with the standard deviation and SNR value, are shown in Figure 6 and Table 2.
F I G U R E 5 Example of Gadus morhus eye lens element concentration profiles of a typical individual for Mg, Sr, P, Zn, Cu and K. Profiles are shown from core to edge. Also shown is the local second-degree polynomial regression and its confidence band, together with detected minima marked with a star

| Comparison of eye lens and otolith chemical features
Of the elements included in this study, seven (Ba, Cu, K, Mg, P, Sr, Zn) had, when taking the mean over the whole growth axis, SNR > 5 for both the lens and otolith. These elements were modelled individually with model (1). No significance of any factors included was found for P. This implies that a general difference in concentration between lens and otolith could not be demonstrated. For all remaining elements the model could be reduced, excluding the factor sex and the interaction of sex and type. In the reduced model type was found to be significant for all remaining elements (ANOVA, num d.f. = 1, P < 0.05; for further details see Supporting Information Table S2), indicating a general difference between lens and otolith element concentrations, with Ba, K and Sr being significant lower in the lens than otolith and Cu, Mg and Zn being significantly higher.
For data representing the mean 'core' and 'edge', SNR > 5 for both lens and otolith occurs for the elements K, Mg and P in the 'edge' data and for Cu and Zn in the 'core' data. The relevant elements were analysed using model (1) for the 'core' data and model (2) for the 'edge' data. For all 'core' and 'edge' data, the model reduction resulted in exclusion of sex and its interactions. A significant influence of type was found on all elements modelled (ANOVA, d.f. = 1, P < 0.05; for further details see Supporting Information Table S2).
Through these analyses, life-stage-specific trends in P and Mg were discovered, highlighting that element concentrations averaged over the entire tissue mask ontogenetic changes between the two structures, where lens concentrations in P are higher than otolith concentrations in the core, but lower in the edge. Such ontogenetic patterns are also observed in Mg, but with a mirror-like inverse pattern compared to P, both illustrated in Supporting Information
Likewise, several elements have previously also been detected in fish eye lenses, but these were never explored by LA-ICPMS analysis. To our knowledge, our study is the first fish eye lens investigation using LA-ICPMS. It thus serves as a baseline study for future investigations in respect to which elements can be anticipated to occur in fish eye lenses (at least from cod) and which elements may be too scarce or too noisy to be useful when using comparable analytical equipment. By comparing the two analytical methods applied here to the eye lens, it is apparent that for the heavier metals found only in trace amounts of <0.1 ppm (Yb, Sc, Zr, V), solution ICP-MS analysis is preferable to LA-ICPMS. On the other hand, the LA-ICPMS technique not only provides a very high spatial resolution across the measured object, but also demonstrates reliable concentration determination for the elements Al, Ba, Fe, K, Li, Mg, Mn, P, Pb, Rb, Si, Sr, Ti and Zn, whereas results for Cr, Cu and Ca should be further investigated.
Other studies reporting trace element concentrations in fish eye lenses (Dove, 1999;Dove & Kingsford, 1998;Gillanders, 2001;Kingsford & Gillanders, 2000;Pourang et al., 2018;Quaeck, 2017;Quaeck-Davies et al., 2018) used dissolution procedures, thereby missing the spatial resolution. The concentrations reported in these studies largely correspond to the concentrations from the solutionbased analysis reported here, or are at least on the same order of magnitude. The major difference is that our study reports the entire suite of elements that were found in eye lenses, while earlier studies seem restricted to a few elements. When comparing the relative offset of eye lens concentrations to the otolith concentration of the elements Mn, Ba and Pb in Gillanders (2001) to the LA-ICPMS results from this study, they appear to follow the same trends relative to the otolith concentration, whereas Sr clearly differs, having a higher relative concentration to the otolith. Variation in concentrations between T A B L E 2 Overview of the chemical data available for the 12 Baltic cod (Gadus morhua), where both otolith and eye lenses were analysed F I G U R E 7 Linear regression of the distance of otolith Sr minima from the core (see Figure 5) against the corresponding eye lens rings distance from the core of Gadus morhua. The figure shows mean correlation with standard error for each individual separately. Blue, the individuals where the lens rings were considered easy to identify; Black, lens rings were difficult to identify (fish IDs #6 and #11 could not be plotted due to unequal x and y data) studies are to be expected since interspecies differences and differences in environment are bound to have a major impact.
In general, there are large elemental differences between eye lenses and otoliths, and for the most part there is no real correlation in mean concentrations. This is, however, not surprising, as otoliths are composed of calcium carbonate and eye lenses consist of a variety of proteins (Wistow & Slingsby, 2016). While our analyses on a relatively small sample indicated that sex was not a significant driver of the observed trends, it would still be useful to include more physiological features, such as maturity stage, age, somatic condition etc., to identify where the interindividual variation in element concentrations comes from. However, this requires a targeted sampling design with a much larger sample size.
An opposing pattern was observed between eye lens P and Mg concentrations, with high P and low Mg near the core, and a steadily decrease (P) and increase (Mg) towards the edge of the eye lenses.
This trend seems to be proportional, but further investigation of the detailed behaviour of this feature is required. A quantitative analysis of this relationship was not possible here since P and Mg exhibit an inverse relationship that is associated with an instrumental/data reduction artefact when calculating the total concentrations. Another interesting trend regarding P and Mg is demonstrated in the relationship between eye lens and otolith tissues. For a given section of the data transect, when P is significantly higher in the eye lens than in the otolith then Mg is consistently higher in the otolith. This relationship, which seems proportional at all points in the data transect, indicates a complex and interdependent relationship between lenses and otoliths, which suggests that the incorporation of P and Mg is largely regulated by biochemical and physiological processes.
When testing for coherence between the lens growth rings and patterns in otolith element concentration, minima in otolith Sr were found to correspond significantly with the lens rings for all but two of the individuals. In these the number of maxima and rings differed, with one and wo more rings than minima present, although overlap between other rings and minima were still present. It should be considered here that the model which showed strong correlation between lens rings distances and otolith Sr valley distances also did not include the two individuals previously mentioned. The exclusion of theses two individuals thus questions whether this is a general trend or only true for some individuals, and emphasizes the importance of quality control of eye lens ring identification. In otoliths, Sr concentration is known to have a positive correlation with the extrinsic environmental concentration, where environmental Sr concentration typically explains >90% of otolith Sr concentration and other variables like salinity and temperature only have a limited effect (Hüssy et al., 2020 and references therein). Cod in the Baltic Sea experience large seasonal variations in ambient salinity and oxygen concentrations as a result of horizontal and vertical migrations related to spawning and feeding, respectively. During spawning, the eastern Baltic cod moves to deeper and more saline waters, while the feeding areas are predominantly located in shallower areas where the water is less saline (Righton et al., 2010;Schaber et al., 2012). The present results do not provide evidence of the mechanisms driving the observed correspondence in otolith element and eye lens ring patterns. However, the close match suggests a link between seasonal migrations associated with variations in environmental Sr and eye lens ring formation.
Some notable observations can be made between how easy it is to discern the eye lens rings in relation to the element profile plots.
When P and Mg have continuous opposing downward and upward trends, respectively, without major peaks, the lens rings are better defined. We thereby speculate whether these profile trends are connected to changes in the crystalline structure of the eye lens, from alpha to gamma as the fish ages (Dove, 1999). Additionally, the otolith Sr maxima matched the lens growth rings best in the individuals where the rings were less clear. It is not within the scope of this study to try to infer the physiological mechanisms or influence of environmental factors behind these results. Nevertheless, these features certainly require further in-depth investigations to identify the underlying mechanism behind the correlations.
The present study was based on only 12 individuals with unequal distribution of sex, capture date, etc. The sampling strategy has possible implications for the results in that it entrains an unbalanced design where model results are less robust than for larger studies. In addition, the standardization procedures and analytical approach might also be improved, here to use other elements as internal standards, such as K or Si, because these elements are an abundant and omnipresent constituent in eye lenses, further allowing for better analysis on the covariance of P and Mg. An additional challenge, presumably introducing some variability into the interpretation of the eye lens rings, is that no prior knowledge on how to identify the proper rings is currently available. It appears that the lens rings do not follow a consistent pattern, being incrementally closer to one another, as is observed for otolith annuli. The rings near the edge appear closer to one another, but they are very difficult to distinguish from one another and should be subject to further examination.

| CONCLUSIONS AND FUTURE CONSIDERATIONS
This study demonstrates that LA-ICPMS analyses can be conducted in situ on eye lenses with an acceptably high precision and accuracy using the analytical approach and instrumentation employed herein. Robust quantitative analyses have been produced providing reliable results, and thus the technique is shown to be a powerful tool for the interpretation of environmental, ecological and whole life history of fish.
Furthermore, the results of this study suggest that incorporation of trace elements is mostly governed by different processes for otoliths and eye lenses. The rings observed in cod eye lenses do not represent seasonal fluctuations or other age-related measures directly, and the lens element patterns do not seem to correspond to these rings. However, an important result of this study is that patterns in otolith Sr generally match well with eye lens growth rings. Eye lenses may thus provide ecological and biological information on a fish's life that could complement otolith-derived information.
The physiological mechanisms behind our results are still not fully established, but point towards a potential link between eye lens growth ring formation and fish behaviour or other regularly recurring drivers. It is, for instance, unclear from this study whether the observed patterns in Mg and P could be a derived effect from α/β crystalline formation or exchange, where protein structures reach a given length before their growth is terminated and a new growth layer starts to form.
It must be noted that the number of fish individuals used for this study does not allow major conclusions to be drawn on a larger scale and caution should be used in respect to other fish species. Further investigations are required to constrain or confirm our results. However, it does provide some important first results towards a better understanding of the use of fish eye lenses as a tool to unravel ecological and biological questions.
One weakness of our study was the sampling design, where a more balanced sampling design collecting a larger sample size and consideration for equal numbers of sex and maturity stage should be followed in future investigations. A greater contrast in the ages and lengths of individuals is also advisable, including quality assurance of a consistent interpretation of the eye lens growth rings.