1. Introduction
Considerable environmental variation in upwelling ecosystems regularly exposes coastal fishery species to varying levels of pH, and [O
2] through space and time [
1,
2,
3,
4,
5,
6,
7]. Environmental fluctuations in seawater properties, including low levels of pH and [O
2], can cause sublethal and lethal effects in loliginids (nearshore squid; [
8]). Within squid habitats, low pH and [O
2] seawater can be tightly associated with upwelling events [
2]. Average pH and [O
2] conditions can be further decreased in southern California during La Niña years as the thermocline shoals [
1]. Early life stages of non-calcifying metazoans exposed to high levels of
pCO
2 and associated low pH (e.g., acidification; [
9,
10,
11]) are affected in several ways, including altered developmental, physiological and behavioral processes [
12,
13,
14,
15]. For cuttlefish and squid embryos, acidified environmental conditions generate additive effects, increasing an already acidified perivitelline fluid that baths embryos within the egg (cuttlefish) [
16,
17] and chorion (squid) [
18,
19]. Thus, environmental hypercapnia could even be more pronounced in early life stages. Further, some mollusks [
20,
21] and fish [
22,
23] are negatively affected when exposed to low levels of oxygen in their early life stages.
Environmental [O
2] is decreasing more quickly along the coast of the Southern California Bight than in the offshore pelagic realm [
24,
25]. Ecological theory suggests that organisms will respond with species-specific shifts in size frequency and biogeographic range [
8,
26]. Many of the knowledge gaps regarding population level effects of [O
2] and pH/
pCO
2 exist because of inadequate tools for assessment, and geochemical proxies have not yet been utilized for squid.
The near-shore squid,
Doryteuthis opalescens, is particularly sensitive to environmental change associated with the El Niño Southern Oscillation (ENSO; [
27,
28,
29]), yet basic knowledge of the market squid population dynamics is lacking. This includes assessment of their population connectivity [
30], as well as knowledge of critical ecological mechanisms controlling their population size [
31]. Fishery boom and bust cycles often correlate with environmental change, such as that associated with ENSO [
27,
28,
29]. Numerous hypotheses for annual catch fluctuation have been presented, but are difficult to test in the field because of sampling method biases [
28,
29,
32]. The developmental, physiological, behavioral and ecological mechanisms that lead to such drastic changes in annual catch have yet to be described, and the magnitude of risk for the population associated with environmental change is unknown. The embryonic life stage may be especially susceptible to low [O
2] and pH/
pCO
2 exposure during La Niña events, because unlike other stages, embryos are site attached to the seafloor and are completely reliant upon a fixed energy reservoir (
i.e., yolk). Without the ability to move, embryos must tolerate exposure to low [O
2] and pH/
pCO
2 levels using their fixed energy supply. One of the most promising tools available to investigators to understand environmental effects at the population level is through the use of statoliths [
30,
33,
34] that are developed during this time.
Statoliths, made of ~95% CaCO
3 in the aragonite crystal form [
35], are used by squid as part of their equilibrium and motion sensory organs (statocysts; [
30,
32,
33,
36]). Paired statoliths develop in each market squid embryo during the last two-thirds of the embryogenesis period (
Figure 1; [
37]), then remain embedded within the statocyst as it grows during each following life stage. After death, the statoliths can sometimes even be preserved in the fossil record [
33,
38,
39]. Statolith aragonite crystal grows with a daily banding pattern, and growth is heavily influenced by the environment [
33,
34,
38,
40]. However, the squid statolith is not in direct contact with the environment, but rather with endolymph fluid within the statocyst [
41].
Environmental chemical effects (e.g., [O
2] and pH) on embryos can be integrated with capsular effects. The structures of the capsule can connect the external chemical environment to the embryo via the capsular membranes, interstitial jelly and the chorion membrane (hereafter, these effects are collectively referred to as “capsular effects”). The pH/
pCO
2 and [O
2] of the perivitelline fluid that surrounds the embryo are impacted additively by the environment and by the physiological processes of the embryo itself [
17,
18,
19,
42,
43,
44,
45,
46,
47]. Elemental incorporation within statoliths can be influenced by the environment [
30,
33,
34,
48], but physiological process impacts on statolith geochemistry, including processes within the statocyst [
41], embryo [
18,
19,
45,
46], as well as within outer-embryo structures [
49,
50], are not well understood. Further, the squid embryonic metabolic rate affects statolith formation [
51,
52]. The cephalopod-embryo metabolic rate greatly increases at the end of development [
42,
45,
46] and is variable among embryos within the capsule [
53]. The molluscan metabolic rate is highly influenced by temperature [
45,
46,
54] and environmental oxygen [
20,
55,
56]; cephalopods can be influenced by environmental pH/
pCO
2 levels at the embryonic stages [
45,
46,
47,
52], but are tolerant at older stages [
57,
58]. As the statolith grows, the volume of the statolith increases exponentially. Thus, as a potential environmental recorder, the geochemistry of embryonic statoliths is weighted towards the end of benthic development. Glycoproteins, Sr
2+, Ca
2+, Mg
2+ and HCO
3− influence the biomineralization process of the squid [
38]. Sr
2+ is required for the initiation of statolith development [
59], and Ca
2+, Mg
2+ and glycoproteins are important for continued statolith growth [
38].
Clear geochemical proxies for exposure to low pH have been established for foraminifera shells [
60,
61,
62], coral skeletons [
63] and mollusk shells [
64] using either δ
11boron or uranium:calcium ratios. In addition, δ
18O has been explored as a proxy for O
2 in statoliths of
Illex illecebrosus [
35]. Here, we explore the element:calcium composition of market squid (
Doryteuthis opalescens) statoliths as a potential proxy of pH/
pCO
2 and [O
2] exposure using levels of pH/
pCO
2 and [O
2] that reflect the highs and lows observed within embryo beds in southern California. This investigation is the first that we are aware of to test for a proxy of pH exposure using squid statoliths.
Any information about these squid could help to fill large knowledge gaps concerning scenarios of rapid climate change. Low pH/high
pCO
2 and low [O
2] (hereafter referred to as “low pHOx”) can cause species-specific negative effects in isolation or in tandem, and the magnitude of each of these effects is likely to be habitat specific [
65,
66]. We conducted experiments to investigate whether statolith geochemistry can reveal squid exposure to low pHOx, low [O
2] only and/or low pH/high
pCO
2 only (hereafter, referred to as “low pH”) during benthic encapsulated stages. We hypothesize that: (1) conditions associated with upwelled seawater observed in the
D. opalescens spawning habitat influence the geochemical composition within embryonic carbonate structures in a manner useful as a proxy and (2) environmental exposure effects can be separated from capsular effects. More specifically, we assessed whether: (A) encapsulated embryos exposed to low pHOx yield distinct individual elemental levels in squid statoliths; (B) multi-elemental signatures can classify statolith exposures independent of individual elemental ratios; and (C) exposure to low pHOx levels yields elemental signatures different from individual effects of low [O
2] or low pH. A goal is to form the ability to assess the exposure of early developmental recruits collected from the field in the absence of seawater pH and [O
2] measurements.
2. Materials and Methods
Treatments were held constant over the entire embryogenesis period to allow for chronic exposure. Encapsulated squid embryos were reared under controlled conditions for the majority of embryogenesis. Experiment 1 (November, 2011) compared the statolith geochemical response between embryos exposed to high pH and [O
2] (high pHOx) and low pH and [O
2] (low pHOx) treatments, and Experiment 2 (March, 2012) compared the statolith geochemical response between embryos exposed to low [O
2] and high pH (low [O
2]) and low pH and high [O
2] (low pH) treatments. Levels were based on field measurements of water conditions made from water depths of 35–88 m water depth, ~6 km off of Del Mar, USA [
1] (see System Overview and Experimental Treatments Section). Unlike near-surface waters, seawater in the lower to mid-shelf depths that are regularly utilized by squid [
32] experiences reduced environmental variability (
i.e., environmental conditions are more stable). In the Southern California Bight, the magnitude of environmental variability (e.g., range of pH and [O
2]) decreases with depth. For example, the range of [O
2] decreased by 37% and the range of pH by 39% from a 7 to 17-m depth [
3]. The variability [O
2] and pH continues to decrease with depth [
1] near areas where squid-embryo capsules were collected (< 10 km). For embryo beds at 80–90 m on the shelf, pH and [O
2] conditions can be near constant over month-long periods. A novel laboratory approach using the Multiple Stressor Experimental Aquarium at Scripps (MSEAS; [
67]) was used to control pH and [O
2] levels. Two experiments were conducted. Each included four tanks: two treatments with two replicate tanks each. For each experiment, capsules were randomized among treatments, aquaria and position within each aquarium.
2.1. Collection of Squid Embryos and Seawater Data
For Experiment 1, newly laid squid capsules (encapsulated embryos) were collected by hand from La Jolla Bay, San Diego, USA (32.86° N, 117.27° W), and from capsules laid at Scripps Institution of Oceanography (squid caught from Del Mar, USA; 32.96° N, 117.28° W). For Experiment 2, newly laid capsules were taken from La Jolla Bay, USA (32.87° N, 117.25° W). All capsules were exposed to treatments for 24 days or more, allowing at least one week of exposure to treatment conditions prior to the initiation of statolith development (statoliths are completely grown under controlled conditions). Environmental data were collected continuously from a 30-m depth, 0.5 m above the seafloor, using a site-attached instrument (SeapHOx: described by Frieder
et al. [
3]) that measured conductivity, temperature, pressure, pH
T and [O
2] from 23 June 2012 to 4 July 4 2013 (location: 32.86° N, 117.27° W).
2.2. System Overview and Experimental Treatments
Embryo capsules of
D. opalescens in Experiment 1 were cultured under treatments of constant low (7.55, 90 µmol·kg
−1) and high levels (7.9, 240 µmol·kg
−1) of pH
T and dissolved oxygen (
Table 1,
Figure 1). Several step-wise changes in the system level setting were conducted during the course of each experiment to maintain stable environmental conditions [
67], and thus, data were not normally distributed. Treatment conditions were distinct for each experiment with respect to pH, Ω
aragonite and [O
2], but not for temperature and alkalinity (Wilcoxon test;
Table 1). Capsules were cultured in 55-L aquaria that had been acid-rinsed in 1 N HCl and then rinsed five times with ultrapure H
2O (resistivity >18.0 MΩ·cm). Treatments were implemented using MSEAS, a manipulated flow-through aquarium design [
67]. Seawater was supplied with seawater pumped from a 5 ± 1.5 m depth off of the Scripps Pier.
Table 1.
Average ±1 standard deviation for temperature (°C), alkalinity (µmol·kg−1), pHT, and [O2] (µmol·kg−1) in tanks during Experiment 1 and Experiment 2. Treatments were distinct for pHT, Ωaragonite and [O2], but not temperature and alkalinity (Wilcoxon test).
Table 1.
Average ±1 standard deviation for temperature (°C), alkalinity (µmol·kg−1), pHT, and [O2] (µmol·kg−1) in tanks during Experiment 1 and Experiment 2. Treatments were distinct for pHT, Ωaragonite and [O2], but not temperature and alkalinity (Wilcoxon test).
Treatment (Tank) | Temp (°C) | Alkalinity (µmol·kg−1) | pHT (in-situ) Total Scale | ΩAragonite | [O2] (µmol·kg−1) |
---|
Experiment 1 | | | | |
High pHOx (1) | 11.3 ± 0.4 | 2215.5 ± 4.8 | 7.938 ± 0.053 | 1.62 ± 0.17 | 241.3 ± 12.3 |
High pHOx (2) | 11.1 ± 0.4 | 2214.2 ± 6.6 | 7.916 ± 0.062 | 1.54 ± 0.21 | 242.6 ± 13.1 |
Low pHOx (1) | 11.4 ± 0.8 | 2214.8 ± 6.3 | 7.578 ± 0.067 | 0.76 ± 0.12 | 82.1 ± 15.8 |
Low pHOx (2) | 11.2 ± 0.9 | 2215.4 ± 5.8 | 7.567 ± 0.065 | 0.74 ± 0.12 | 78.6 ± 21.5 |
Treatment Effect (df = 1, N = 36) | χ2 = 0.02, p = 0.876 | χ2 = 0.01, p = 0.921 | χ2 = 109.35 p < 0.0001 | χ2 = 109.35 p < 0.0001 | χ2 = 90.76, p < 0.0001 |
Experiment 2 | | | | |
Low [O2] (1) | 11.2 ± 0.5 | 2239.1 ± 5.5 | 7.923 ± 0.035 | 1.58 ± 0.10 | 86.4 ± 8.3 |
Low [O2] (2) | 11.6 ± 0.5 | 2241.8 ± 4.5 | 7.908 ± 0.072 | 1.57 ± 0.21 | 83.0 ± 12.9 |
Low pH (1) | 11.3 ± 0.5 | 2241.1 ± 5.8 | 7.559 ± 0.029 | 0.73 ± 0.04 | 241.1 ± 9.1 |
Low pH (2) | 11.6 ± 0.6 | 2244.2 ± 7.1 | 7.552 ± 0.026 | 0.73 ± 0.04 | 241.7 ± 7.6 |
Treatment Effect (df = 1, N = 32) | χ2 = 0.05, p = 0.819 | χ2 = 3.14, p = 0.077 | χ2 = 93.74, p < 0.0001 | χ2 = 93.74, p < 0.0001 | χ2 = 72.74, p < 0.0001 |
Embryos used for both experiments were collected from the field pre-organogenesis, Stages 11–12 [
37,
68,
69]. For Experiment 1, 24 capsules were collected from the field just after being laid and were allowed to acclimate for 3 days at 11 °C. An additional 16 capsules laid in captivity at Scripps Institution of Oceanography were added to the experiment 7 days later. Ten capsules (6 field, 4 aquaria) were randomly assigned and placed into a position in each of four aquaria and were evenly distributed among sources. Twenty capsules were placed into each aquarium. Densities in Experiments 1 (~100 capsules·m
−2) and 2 (density ~200 capsules·m
−2) are within the viable range found in the field, where they are reported to range from 1 to 47,720 capsules·m
-2 [
32,
70].
Figure 1.
Multiple Stressor Experimental Aquarium at Scripps (MSEAS) environmental data. From top to bottom, the graphs depict the (
a) pH
T, (
b) saturation state (Ω
aragonite), (
c) [O
2] (µmol·kg
−1) and (
d) temperature (°C) of the seawater within each tank of (
A) Experiment 1 and (
B) Experiment 2. Purple = Low pHOx; Black = High pHOx; Blue = Low [O
2]; Red = Low pH. The graphic is modified from Bockmon
et al. (2013) [
67]. Blue shading= the estimated period prior to statocyst development (prior to the formation of the statolith).
Figure 1.
Multiple Stressor Experimental Aquarium at Scripps (MSEAS) environmental data. From top to bottom, the graphs depict the (
a) pH
T, (
b) saturation state (Ω
aragonite), (
c) [O
2] (µmol·kg
−1) and (
d) temperature (°C) of the seawater within each tank of (
A) Experiment 1 and (
B) Experiment 2. Purple = Low pHOx; Black = High pHOx; Blue = Low [O
2]; Red = Low pH. The graphic is modified from Bockmon
et al. (2013) [
67]. Blue shading= the estimated period prior to statocyst development (prior to the formation of the statolith).
Embryos were cultured to embryonic-developmental Stages 28 or 29 (near-hatch paralarvae) [
68] to reduce ontogenetic effects [
71] and were collected only within the central portion of the capsule in order to reduce capsule-position effects [
72] (
Figure 2). These stages were indicated by the pigmentation of the ink sac (this occurs earlier in
D. opalescens than in
D. pealeii), the complete covering of the eyes by the cornea, but not a prominent Hoyle’s Organ [
37,
68,
69]. Cultures were maintained at constant temperature (11.3 °C ± 0.3 °C, SD), salinity (33.4 ± 0.2) and light levels using 15 W LED lights on a 12:12 h light:dark cycle to reduce these types of environmental effects on statolith development [
21,
40,
72]. Salinity, temperature and seawater flow rate were constant among treatments. Statolith development takes place at the start of organogenesis [
37], and our experiments exposed embryos to treatments one week or more prior to the statolith formation. The low [O
2] and pH treatments developed more slowly (5–7 d) compared to the other treatments; thus, samples from the 3 treatments (Experiment 1: low pHOx and Experiment 2: low pH, low [O
2]) were gathered at two times, once to match the high pHOx treatment exposure duration and once again 5–7 d later to allow embryos exposed to low [O
2] and/or pH to develop to near-hatch Stages 28–29. The levels of [O
2], pH, Ω
Aragonite and temperature were held constant for each treatment (
Figure 1).
Further, to verify whether or not seawater trace-metal concentration varied among treatments, 100-mL seawater samples were taken weekly using clean-lab protocols to minimize any possibility of contamination [
73]. Seawater samples were filtered, acidified with 50 µL of 12 N optima HCl and stored in darkness at room temperature (21 °C ± 3 °C) until they were analyzed at Arizona State University for magnesium
24(Mg), calcium
48(Ca), strontium
88(Sr), barium
138(Ba) and uranium
28(U). Boron (B) was estimated using discrete salinity values taken daily [
74].
Figure 2.
(a) Two squid-embryo capsules. Capsules are directly exposed to the environment, and each contains between 100 and 300 embryos. Ruler units are in cm; (b) Subsection of the capsule with the capsular membranes removed. Gelatinous material fills the interstitial space between the chorions and the capsular membranes; tick marks at the bottom of the image demarcate mm; (c) Chorion filled with perivitelline fluid and containing a squid embryo; (d) Squid embryo: statocysts are circled in red; (e) Statocysts are filled with endolymph fluid, and each contains a single statolith; (f) Embryonic statolith, made of aragonite.
Figure 2.
(a) Two squid-embryo capsules. Capsules are directly exposed to the environment, and each contains between 100 and 300 embryos. Ruler units are in cm; (b) Subsection of the capsule with the capsular membranes removed. Gelatinous material fills the interstitial space between the chorions and the capsular membranes; tick marks at the bottom of the image demarcate mm; (c) Chorion filled with perivitelline fluid and containing a squid embryo; (d) Squid embryo: statocysts are circled in red; (e) Statocysts are filled with endolymph fluid, and each contains a single statolith; (f) Embryonic statolith, made of aragonite.
2.3. Extraction and Mounting of Statoliths for Elemental Analyses
Statolith extraction and mounting procedures followed clean lab protocols [
73]. Statoliths were removed for analysis when embryos were developed to Stages 28–29 [
68]. Statoliths were removed by dissection, and then chemical digestion of soft parts (encapsulation and embryo proteins) was carried out by placing embryos on a slide with digesting solution (10 µL of ultrapure water and 5 mL of ultrapure 15% H
2O
2 buffered with 0.05 N NaOH in ultrapure H
2O) for 10–20 min, depending on the amount of soft tissue. Digestion solution was removed by pipetting with a clean tip, and statoliths were rinsed three times in ultrapure H
2O. All remaining H
2O was removed by evaporation overnight underneath a hood within a Class 100 clean room. One statolith from each embryo was extracted, and 10–20 embryos were dissected from the center position of each capsule (total = 10–20 statoliths from each capsule). Statoliths were mounted on double-sided tape (Scotch
TM), attached to a slide and prepared for laser ablation. The chemical composition of mounting tape was determined by laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) [
75]; all elemental counts from the tape were at least three orders of magnitude lower than counts found in statoliths. Statolith lengths ranged from 65 to 130 µm.
2.4. LA-ICP-MS Instrument Settings and Methods for Elemental Analyses of Statoliths
Statoliths were analyzed by LA-ICP-MS at low resolution for an elemental menu consisting of boron
11 (B), magnesium
24 (Mg), calcium
48 (Ca), manganese
55 (Mn), copper
63 (Cu), zinc
66 (Zn), strontium
88 (Sr), barium
138 (Ba), lead
208 (Pb) and uranium
238 (U). Ablated material from haphazardly selected statoliths from each slide was introduced via a New Wave UP-213 UV laser, frequency-quadrupled to a 213-nm wavelength with a nominal beam width of 40 µm in the spot beam setting, into a Finnigan Element 2 sector field ICP-MS using a micro-flow nebulizer at 20 µL·min
−1. The laser was set to the spot beam mode at 40% power and a frequency of 20 Hz. Consistent plasma conditions were maintained using 1% HNO
3 during analysis of standards, on instrument blanks and on laser-ablated samples [
75]. Instrument sensitivity was monitored measuring indium (In) and was approximately 1 × 10
6 counts·s
−1 for 1 ppb. All slides were assigned a number and then analyzed in a random order. The laser-ablation process completely vaporized each statolith.
To standardize the mass variation of statolith samples, the concentrations of elements in the statoliths are reported as ratios with respect to Ca, the dominant elemental constituent of the statolith [
76]. Aragonite is a solid, acellular, metabolically inert, crystalline structure [
76] that is ~95% CaCO
3 [
35] and differs from other calcified structures found in cephalopods that are porous, cellular and metabolically active, such as cuttlebone, which have been shown to become less porous (hypercalcification or increased CaCO
3 density) in response to hypercapnia [
47,
77]. Further, if hypercalcification or hypocalcification is induced by a treatment, we would expect all element:calcium ratios to be significantly increased (hypocalcification) or decreased (hypercalcification). Therefore, we would predict that this method is sensitive to changes in calcium concentration. Element:Ca of the sample was determined by using matrix-matched solution standards of known element:Ca and a mass-bias correction [
78].
Several steps were taken to ensure accurate element:Ca of the samples (i.e., the error in measurements using the ICP-MS and laser ablation unit). Throughout the analyses, the ICP-MS tested solution-based measurements, which remained within 5% of each element’s known values. Using a solution standard containing Ca, Mg, Sr, Ba, Ce, Pb, U, Mn and Zn (Spex Certified primary standard solutions), values of the ICP-MS measurements compared to known values were as follows: Mg:Ca (mmol:mol) = 1.24%, Sr:Ca (mmol:mol) = 0.37%, Ba:Ca (µmol:mol) = 2.77%, Pb:Ca (µmol:mol) = 3.09%, U:Ca (µmol:mol) = 0.40%, (N = 20)). The accuracy of the laser ablation method was estimated by using glass standard number 612 from the National Institute of Standards and Technology, Gaithersburg, USA (NIST612). The repeatability (relative standard deviation, % rsd) of the method was determined using the results of the laser-ablated NIST612 standard reference material, except for Sr:Ca, which was determined using the otolith material: B:Ca (mmol:mol) = 0.02%, Sr:Ca (mmol:mol) = 0.61%, Ba:Ca (µmol:mol) = 0.18%, Pb:Ca (µmol:mol) = 11.13%, U:Ca (µmol·mol−1mol) = 4.60% (N = 12). Accuracy for B:Ca was calculated using NIST612, because carbonate and solution-based standards were unavailable, leaving the potential for matrix effects.
Detection limits were defined as the intensities (counts per second) of elements present in the instrument blank plus three times the standard deviation [
79]. The average intensity of each element above the detection limit was as follows (measured as multiples of detection limit): Mg > 1770, Sr > 1,280, Ba > 440, U > 220, Pb and Cu > 50, Zn > 40, Mn > 1.2 and B > 0.8. The percentage of samples where each element was above the detection limit was as follows: Mg 97%, Sr 99%, Ba 100%, U 97%, Pb 98% and B 40%. For the B data, we only included those samples that were above detection limits and eliminated all others from further analysis.
2.5. Statistical Analyses of Seawater and Statolith Elemental Composition
Seawater samples were taken weekly from each tank within each treatment. For statolith samples, tanks were considered replicates for statistical testing of the hypotheses. The average number of embryos (
i.e., statoliths) analyzed per capsule was 5.4 ± 0.5 (standard error). All data were first examined for variance homogeneity and tested for normality by means of residual analysis prior to using ANOVA models. All seawater elemental data were normal, and statolith elemental data were normal for B:Ca, Sr:Ca and Ba:Ca (no transformations needed). Statolith Mg:Ca, Pb:Ca and U:Ca data were right-skewed, with U:Ca being the most intensely skewed of the data sets. Mg:Ca and Pb:Ca data were square-root transformed, whereas U:Ca data were cube-root transformed; data were normally distributed after transformation. Experiments 1 and 2 were tested separately. Treatment effects were tested using a one-way mixed model ANOVAs nesting the tank (fixed) and capsular (random) factors [
80]. In Experiment 1, the effects of treatment on the elemental concentration (element:Ca) of statoliths were tested between groups with high pHOx to those of low pHOx. In Experiment 2, the effects for treatment onto the elemental concentration (element:Ca) of statoliths were tested between groups with low [O
2] and low pH levels.
Multivariate analysis of similarity (ANOSIM, Euclidean Distance, N = 9999 permutations) was used to test the hypothesis that the treatment influences the multi-elemental composition of the statoliths. Principle component analysis (PCA) was used to investigate elemental signatures among groups using a correlation matrix (i.e., data were not transformed). Statistical analyses were conducted using JMP (Version Pro 11) statistical software (SAS Institute, Cary, NC, USA).