Earth and Planetary Science Letters Stylasterid corals: A new paleotemperature archive

Stylasterids are a ubiquitous deep-sea coral taxon that build their skeletons from either calcite, aragonite, or both. Yet, robust geochemical proxy data from these corals are limited. In this study, 95 modern stylasterids, spanning a wide range of depths (63 to 2894 m) and ambient seawater temperatures (0 to 17 ◦ C), were tested for their potential use as paleoceanographic archives. Stable oxygen and carbon isotopic composition ( δ 18 O and δ 13 C) were measured from the main trunk of all specimens and ﬁve specimens were further sub-sampled to assess internal chemical variability. The isotope data show non-equilibrium precipitation from seawater for both δ 18 O and δ 13 C, with the growing tips of colonies yielding the isotopically lowest values. Overall, the calcitic corals showed lower isotope values for δ 18 O and δ 13 C than aragonitic specimens. Within the aragonite corals, we present a δ 18 O:temperature calibration that exhibits a signiﬁcant linear relationship with the equation δ 18 O coral-seawater = − 0.22 ( ± 0 . 01 ) × T ( ◦ C) + 3.33( ± 0.06) across a temperature range of 0 to 30 ◦ C, using samples from this study and published data. This work highlights the potential application of stylasterid coral δ 18 O data to reconstruct paleo seawater temperature. © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Reconstructing the thermal history of the ocean is vital to understanding fundamental aspects of the climate system, including the evolution of ocean heat content (Cheng et al., 2019), past ocean circulation (Marcott et al., 2011), climate sensitivity (Martínez-Botí et al., 2015), and response of marine ecosystems (Hughes et al., 2017). The development of robust paleoceanographic temperature proxies is important to achieve these goals. Pioneering studies in the 1950s first identified the potential use of carbonate fossils as archives for the seawater temperature in which they formed, by finding a temperature dependence of oxygen isotope composition (δ 18 O) in biogenic carbonates (Emiliani, 1955;Epstein et al., 1953;McCrea, 1950;Urey, 1947). For many years, the δ 18 O of foraminifera and shallow-water corals have been important archives of paleotemperatures (Lea, 2014). Although extremely valuable, these archives present inherent limitations to the resolution and geographical extent of the potential reconstructions they provide. In this framework, carbonate-building deep-sea corals are emerging as complementary paleoceanographic archives, as they are unaltered by bioturbation, can be long-lived, and are widely distributed both spatially and bathymetrically (Roberts et al., 2006;Robinson et al., 2014).
Geochemical studies on deep-sea scleractinian and bamboo corals (class Anthozoa) have shown that isotope signatures related to biomineralisation (so-called "vital effects") can mask the temperature dependent fractionation (Emiliani et al., 1978;Heikoop et al., 2000;Smith et al., 2000Smith et al., , 2002Adkins et al., 2003;Hill et al., 2011). This manipulation of the δ 18 O and carbon isotope (δ 13 C) composition of small volumes of seawater taken into the tissue during deep-sea coral biomineralisation makes temperature reconstruction from these taxa challenging. To overcome this limitation, the "lines method" was developed to exploit the linear relationship between stable carbon and oxygen isotope composi-  tion governed by kinetic isotope fractionation (Smith et al., 2000).
In the "lines method," multiple δ 18 O and δ 13 C analyses are used to determine which part of the skeleton precipitated closest to equilibrium with seawater (Smith et al., 2000). A similar approach was also applied to calcitic bamboo corals (Hill et al., 2011). A variety of other geochemical paleotemperature proxies have been tested in coral carbonates, including clumped isotopes of carbon and oxygen (Ghosh et al., 2006;Thiagarajan et al., 2011;Spooner et al., 2016) and trace metal ratios including Sr/Ca (Beck et al., 1992;Smith et al., 1979) and Li/Mg (Case et al., 2010;Montagna et al., 2014). However, all geochemical approaches applied to date have revealed limitations due to "vital effects" and often location and/or speciesspecific calibrations are required (Alpert et al., 2016;Fowell et al., 2016;Spooner et al., 2016). Consequently, high-resolution records of intermediate and deep-water temperatures are challenging to obtain and interpret, although their importance over the timescale of human-induced global warming is unequivocal (Gleckler et al., 2016).
There is a pressing need to unlock the potential of other coral taxa and to broaden the scope of ocean temperature proxy development. Stylasteridae (Fig. 1) is a ubiquitous (Fig. 2) and highly diverse family of corals within the class Hydrozoa (Cairns, 2011). Stylasterids are the only known coral taxon with variable mineralogy, calcifying their skeleton out of aragonite or calcite, and occasionally using both CaCO 3 polymorphs (Cairns, 2007;Lindner et al., 2008;Cairns, 2011). Despite their diversity and abundance, knowledge about their ecology, physiology and evolution is limited. To date few studies have explored the geochemistry of stylasterid corals and their potential use as paleoceanographic archives. Two studies hinted at a relationship between δ 18 O and temperature, with one focusing on the shallow-water genera Distichopora and Stylaster (Weber and Woodhead, 1972a), and the other on a single specimen (genus Errina) from the Azores archipelago (Wisshak et al., 2009). A further study compared taphonomic and diagenetic alteration, as well as δ 18 O and δ 13 C composition of modern and sub-fossil samples of Stylaster erubescens (Black and Andrus, 2012). More recently, 14 C measurements in Antarctic stylasterids (genus Errina) have been used to trace water mass movements on the centennial time scale (King et al., 2018). Their global distribution across a vast depth range, presence in areas where other archives are rarely found (steep walls and low nutrient waters (Cairns, 1992), as well as their tree-ring like growth bands, raise the potential of these deep-sea corals as high-resolution subsurface ocean archives.
Here, we present the first systematic oxygen isotope calibration of stylasterid corals. We measured δ 18 O and δ 13 C in modern samples of seven genera from a wide geographic distribution to test whether this coral taxon can be used as a reliable and useful archive of past subsurface seawater temperature.

Seawater properties
The seawater temperature was estimated for each sample location and depth (Table S2). During cruises NBP0805, NBP1103 and JR15005 CTDs (conductivity-temperature-density) were deployed within 0.5 • latitude and longitude from the coral sampling sites. Similarly, during cruises CE14001, JC094, JC136, DY081 and SJ2007 seawater temperature was measured at the sample collection site via an ROV or manned submersible mounted CTD (±5 m depth from collection site). Seawater temperature for samples collected from the Galápagos archipelago during cruise AL1508 was extracted from the GLODAP v2 bottle data database (Lauvset et al., 2016) (<0.2 • latitude and ∼5 • longitude difference).
All cruise-derived CTD temperatures were compared to the closest available data in GLODAP v2 (Table S2). The degree of agreement between data sources was variable (mean difference = ±0.3 • C), with 75% of the data showing a difference of <±0.4 • C. Hence, we take ±0.4 • C to be the estimated uncertainty on coral growth temperatures. All figures and tables in this manuscript report the temperature measured on site by the co-located CTD, where available.
The seawater δ 18 O and dissolved inorganic carbonate δ 13 C (δ 13 C DIC ) were also obtained for each sample location (Table S1). Seawater samples were collected via Niskin bottles during cruises NBP0805, NBP1103 and DY081 (attached to the rosette), and JC094 (attached to the ROV) and measured for δ 18 O and δ 13 C back in the laboratory (Hendry et al., 2019;Spooner et al., 2016).

Coral mineralogy
The mineralogy of 56 corals, from the 95 selected for this study, was determined by X-ray diffraction (XRD). The subset of 56 corals was selected to include all the specimens of genera known to biomineralise their skeletons of either aragonite or calcite (or both). These included Cheiloporidion sp. (all 8 specimens of this study), Conopora sp. (6 out of 15 specimens), Errina spp. (7 out of The selected specimens were subsampled perpendicular to the vertical growth axis across the main skeletal branch or central trunk of the colony, using a dental drill. Subsamples of ∼0.5 g were powdered using an agate pestle and mortar. Samples were analysed by X-Ray diffraction (XRD) at the University of Manchester using a Bruker D8Advance diffractometer, with a wavelength (λ) of 1.54060 Å and 2θ ranging from 5 to 70 (Davis et al., 1986;Dickinson and McGrath, 2001). This quantitative technique identifies skeletal mineralogy as well as the relative proportions of each polymorph.

Stable isotope composition of the coral skeleton
All 95 coral specimens were analysed for their δ 18 O and δ 13 C by taking subsamples from the surface of the central trunk of the coral colony, thus sampling modern skeletal material, deposited shortly before sample collection (Fig. 1). Surface aliquots were prepared using a scalpel to scrape off ∼0.1 mg from the surface layer of the coral skeleton into a fine powder. For specimens for which only a portion of the colony was collected, the largest available branch was sampled. An additional, twenty-two specimens were analysed for bulk δ 18 O and δ 13 C, thus subsampling skeletal material deposited during the entire lifespan of the specimen. Bulk aliquots were prepared with a dental drill, cutting a section perpendicular to the vertical growth axis across the central trunk of the colony. Samples were further powdered using an agate mortar and pestle.
To evaluate δ 18 O and δ 13 C variation within a single skeleton, five specimens (one calcitic, Cheiloporidion pulvinatum, and four aragonitic, Errina antarctica, Inferiolabiata labiata, Inferiolabiata lowei, and Stylaster densicaulis) were also sampled at the tip of a distal branch and three of these (Cheiloporidion pulvinatum, Errina antarctica and Stylaster densicaulis) also in a secondary branch.
Intra-specimen δ 18 O and δ 13 C variability within the same visible growth band of a single colony was also examined in an aragonitic specimen of Errina antarctica. Eight subsamples were ob-tained from the same growth band at intervals of 5 mm along a branch, parallel to the vertical growth axis. In this specimen δ 18 O and δ 13 C were also measured in a cross-section of a secondary branch, with one sample from each growth band. A total of 13 subsamples were obtained at intervals of ∼0.5 mm (the width of each growth ring) on a cross-section, perpendicular to the vertical growth axis.
Coral skeleton δ 18 O and δ 13 C were analysed at the British Geological Survey using an IsoPrime dual inlet mass spectrometer connected to an attached Multiprep device. The aliquots of coral samples and standards were loaded into glass vials that were evacuated before addition of anhydrous phosphoric acid (H 3 PO 4 ) at 90 • C. The carbon dioxide (CO 2 ) obtained from the digestion of coral CaCO 3 was collected for 15 min, cryogenically purified, and transferred to the mass spectrometer. δ 13 C and δ 18 O values were calculated using an in-house laboratory calcite standard calibrated against NBS-19. To obtain the δ 18 O and δ 13 C of aragonitic samples, an aragonite-acid fractionation factor of 1.00855 was applied to the gas values. The Craig correction was applied to account for δ 17 O (Craig, 1957

Coral skeletal mineralogy
Of the 56 specimens analysed for skeletal mineralogy, 16 were calcitic (>93% calcite) and 40 were aragonitic (>93% aragonite) ( Table S1). The 39 samples that were not analysed for their mineralogy were assumed to be aragonitic based on prior observations that they belong to genera that always present aragonitic skeletons (Cairns and Macintyre, 1992).

Isotope variability
Values of δ 18 O and δ 13 C from the main trunk of aragonitic stylasterid corals ranged from −0.40 to +3.43 and from −3.05 to +2.30 , respectively. Data from calcitic stylasterids exhibited a range of +0.40 to +1.68 for δ 18 O and of −6.63 to +0.19 for δ 13 C, respectively (Fig. 3a). Bulk measurement of δ 18 O and δ 13 C of aragonitic specimens ranged from +0.03 to +3.32 and from −2.68 to +1.73 . Bulk data from calcitic specimens was of +1.73 and +1.78 for δ 18 O, and −4.62 and −4.67 for δ 13 C (Fig. 3a). Linear relationships are observed for the δ 13 C:δ 18 O data of the main trunk samples. Least squares regression analyses yielded a slope of 1.66 and R 2 value of 0.36 (p-value < 0.001). Classifying datasets by genus yield slopes from −0.65 to 4.29, and R 2 values from −0.99 to 0.78 (Table 2).
In the five specimens with repeat analyses, the tips were

∼1
and ∼3 lower in δ 18 O and δ 13 C respectively than main trunk and secondary branches (Fig. 4) The isotopic composition of samples from within a single growth band of Errina antarctica yielded a δ 18 O range from +1.14 to +1.45 and a δ 13 C range from −1.51 to +1.50 . The isotopic composition across the growth bands of the trunk yielded a δ 18 O range from +0.80 to +1.46 , and a δ 13 C range from −1.50 to −0.59 . The sample from the centre of the trunk yielded the lowest δ 18 O and δ 13 C (Fig. 5). Least squares linear regression of   (1969). Equilibrium values for δ 13 C were estimated using fractionation factors from Romanek et al. (1992). The estimated equilibrium values were subtracted from the coral isotopic value (δ 18 O (δ 13 C) coralδ 18 O(δ 13 C) equilibrium ) to obtain the fractionation values for each sample (Fig. 3b). Fractionation values from isotopic equilibrium of aragonitic colonies ranged from −2.41 to −0.16 for δ 18 O and from −6.37 to −1.05 for δ 13 C. Calcitic colonies ranged from −1.52 to   +0.28 for δ 18 O and from −8.09 to −5.13 for δ 13 C (

Isotope fractionation in stylasterid corals
The isotopic fractionation within stylasterid skeletons is not spatially homogeneous, with the growing tips and the central core of the coral being lower than the main trunk samples by up to 1 and 3 for δ 18 O and δ 13 C, respectively (Figs. 4 and 5). The slope in the δ 13 C:δ 18 O relationships is similar to those observed in scleractinians and bamboo corals ( Fig. 4; Weber and Woodhead, 1972b;Smith et al., 2000Smith et al., , 2002Adkins et al., 2003;Hill et al., 2011). Scleractinian and bamboo corals belong to the class Anthozoa, whilst stylasterids are Hydrozoa, so this broad similarity in slope points towards some commonality in the "vital effects" occurring in these three distinct taxa of the Phylum Cnidaria.
Despite the observed similarities in the "vital effects" signal, the process through which corals produce their carbonate skeletons and manipulate the isotopic composition of seawater is yet to be fully understood. Whether this process is common across all the coral taxa, or whether there are different pathways to coral biomineralisation also remains unclear. McConnaughey (1989) proposed a biomineralisation model in which the "vital effects" in azooxanthellate scleractinians, and the observed strong correlation in their skeletal δ 18 O and δ 13 C, was caused by a kinetic fractionation during the skeletal calcification. Subsequently, Adkins et al. (2003) discovered that the densest areas (Centres of Calcification, COC, thought to be associated with rapid mineralisation) in the scleractinian Desmophyllum dianthus do not follow the δ 18 O: δ 13 C trend observed in the rest of the skeletal components; the COCs being more depleted in δ 18 O compared to δ 13 C. This observation led to a second model of biomineralisation, whereby the isotopic composition is affected by the pH of the extracellular calcifying pool from which the coral precipitates. More recently a fully kinetic model of biomineralisation has been proposed by Chen et al. (2018) in which the enzyme carbonic anhydrase modulates the carbonate chemistry of the calcifying fluid controlling the slope of the δ 18 O:δ 13 C relationship. This model suggests that the kinetic mechanism proposed by McConnaughey (1989) dominates the δ 18 O signature, while the Adkins et al. (2003) mechanism explains most of the range in δ 13 C and the COC 'kink' observed as scleractinian deep-sea corals reach a minimum δ 13 C value (∼−8 ) at maximum cell derived CO 2 . The presence of the enzyme carbonic anhydrase has not been studied in stylasterids, hence the interpretation of the proposed biomineralisation processes in stylasterids requires further investigation.
In our data, we also find evidence of non-systematic internal isotopic variability within stylasterid growth bands. Although there are some similarities in the "vital effects" shown by the different coral taxa, stylasterids show a much smaller range in δ 18 O and δ 13 C than scleractinian and bamboo corals ( Fig. 4 and 5; Smith et al., 2000Smith et al., , 2002Adkins et al., 2003;McConnaughey, 2003;Hill et al., 2011).
In summary, the internal variation in the isotopic composition of different parts of the stylasterid coral skeleton reveal both similarities and differences in biomineralisation between stylasterids (Hydrozoa) and scleractinian and bamboo corals (Anthozoa). However, the narrower ranges in the δ 18 O data from stylasterid corals in comparison with these other deep-sea coral families raise the potential of δ 18 O in stylasterid corals as a possible paleoclimate proxy.

Inter coral isotope variability
Given the internal isotope heterogeneity seen in the growth in tips and trunks careful sampling is required to minimise isotopic overprinting by "vital effects". Therefore, samples from only the main branch of all 95 specimens were examined to explore environmental controls (Fig. 3a and 3b).

Calcitic and aragonitic stylasterid corals
Calcitic stylasterid corals all show a lower δ 18 O and δ 13 C signature than the aragonitic colonies collected at the same location (Fig. 3a). Indeed, this result is highlighted by colonies of Cheiloporidion pulvinatum with different mineralogy which show statistically different values. This observation, as expected, shows that the skeletal mineralogy of stylasterid corals has an important control of the isotopic signal of the sample in line with the data from aragonite scleractinians compared to calcitic bamboo corals (e.g. Smith et al., 2000;Hill et al., 2011).

Stylasterid genera
Our data indicate that coral calcification in stylasterids occurs in disequilibrium with seawater, showing a mean fractionation of −1.16 for aragonitic specimens, with Adelopora sp. and Inferiolabiata spp. having statistically the same mean fractionation to Errina spp. and Stylaster spp. Weber and Woodhead (1972a) reported that shallow water specimens of Distichopora (39 specimens) and Stylaster (6 specimens) precipitated in isotopic equilibrium with seawater for δ 18 O across a temperature range of ∼5 • C. However, we recalculated the fractionation of these samples considering δ 18 O of seawater at each location (Schmidt et al., 1999)  Although the existing data does not allow us to rule out genus or species-specific differences in fractionation, we focus the remainder of the discussion on assessing the potential of a "universal" aragonitic stylasterid temperature calibration.

Stylasterid δ 18 O temperature proxy
Our results provide evidence that the δ 18 O of stylasterid corals is temperature dependent and unlocks the possibility of their use as paleotemperature proxies.
To assess the potential of δ 18 O as a paleotemperature proxy in aragonitic stylasterid corals, we present a universal calibration that could provide a predictive relationship for seawater temperature based on the δ 18 O stylasterid values. While the specimens analysed for this calibration are geographically constrained to the Atlantic Ocean and eastern Pacific, the diversity of genera included in the calibration suggests that this calibration is likely to be valid for aragonitic stylasterid corals globally. The least squares regression analysis for all the aragonitic samples in this study (77 main trunk samples and 20 bulk samples) yields the following equation: For the mean δ 18 O value (2.58 ), 95% prediction intervals produced a temperature uncertainty of ±3.5 • C (95% confidence intervals = ±0.3 • C).  Weber and Woodhead (1972a) and Wisshak et al. (2009). Least-squares regression line of the calibration (orange dotted line), 95% prediction intervals (grey dashed line), and 95% confidence intervals (grey continuous line) are included. Symbols and colours represent genera and location, respectively. Colour-filled symbols are main trunk samples of this study. Black-filled symbols outlined in colour are bulk samples of this study. Empty symbols outlined in black are data modified from Weber and Woodhead (1972a) and Wisshak et al. (2009). Each red vertical bar represents δ 18 O variability measured in a single specimen of scleractinian coral (Adkins et al., 2003). The blue dotted line represents the calibration for aragonite precipitating in isotopic equilibrium with seawater from Grossman and Ku (1986).
This stylasterid δ 18 O:temperature relationship has a similar slope to the biogenic aragonite in equilibrium calibration (Grossman and Ku, 1986), with an average offset of ∼1.2 . The only other deep-sea stylasterid δ 18 O data (Errina dabneyi), from the Azores, is consistent with this temperature relationship (Wisshak et al., 2009).
When extended to include data from shallow-water stylasterids, the δ 18 O:temperature relationship is remarkably similar (Weber and Woodhead, 1972a). The data do indicate that there is a smaller fractionation in these shallow water samples, which may be species-specific effects in warm water hydrozoans or analytical artifacts. The δ 18 O:temperature relationship was recalculated for the samples of this study and those available in the literature (Weber and Woodhead, 1972a;Wisshak et al., 2009) In this calibration (Eq. (2)), for the mean δ 18 O value (2.21 ), 95% prediction intervals produced a temperature uncertainty of ±4.4 • C (95% confidence intervals = ±0.5 • C).
The data suggest that there could be differences in isotopic fractionation between genera. To test this, we considered each separate genus using least squares regression analyses ( Fig. S1 and Table  S3). The temperature estimation uncertainty for each of the genera are larger than those given for all genera combined, and in some cases the temperature range is only of ∼2 to 4 • C. Following these limitations, the universal calibration is preferred at this time (Eq. (2)).

Potential for time-series analyses
The temperature dependence of δ 18 O in stylasterid corals, paired with the growth bands observed in skeletal cross-sections, raises the possibility of their use as high-resolution archives of the recent or distant past (e.g. King et al., 2018). A better understanding of biomineralisation processes, growth rates and micro-skeletal structure in stylasterids could aid in the acquisition of a more accurate paleotemperature calibration for this purpose.
Few previous studies have looked at the growth rates and lifespan of stylasterid corals. Imaging and geochemical measurements have produced differing growth rate estimates, but whether this is due to method accuracy or species/site-specific differences is unknown (Miller et al., 2004;Wisshak et al., 2009;King et al., 2018). In-situ photographic observations of Errina dabneyi recorded a linear extension of 4 to 6 mm/yr in the branch length (Wisshak et al., 2009). Calcein staining and imaging methods for the shallowwater Errina novaezelandiae suggested an average linear extension of 7 mm/yr in the colony width but growth was found to be highly variable (Miller et al., 2004). Finally, King et al. (2018) found a maximum growth rate of 0.5 mm/yr in length extension using radiocarbon dating methods on Errina sp. These growth rates give estimated ages of 70 to 100 yr for Errina dabneyi (∼40 cm high), 30 to 40 yr for Errina novaezelandiae (20 to 30 cm in diameter), and ∼ 200 to 400 yr for Errina sp. (9 to 20 cm high).
We find similarities between the internal variability of δ 18 O and δ 13 C from our Errina antarctica sub-sampled across growth bands  (Fig. 7). At face value these ranges in δ 18 O could translate to temperature changes of 2.4 and 2.7 • C in magnitude. This variation in the temperatures recorded by single individuals (at 130 m in the sub-Antarctic and 500 m in the Azores, respectively) residing within the thermocline are not unrealistic. It is not yet clear, however, why the centre of the branches and the growing tips are systematically lower in δ 18 O and δ 13 C compared to the rest of the skeleton and whether this represents a different biomineralisation processes. It might be that the initial calcification and growing tips calcify rapidly, and posterior thickening of the branches occurs at a slower pace, diminishing kinetic fractionation of δ 18 O (Adkins et al., 2003). In any case, additional studies focusing on detailed internal variability within stylasterid corals are needed to understand these mechanisms in order to fully exploit these corals as paleoclimate archives.
Another consideration to be made prior to using stylasterids as paleotemperature archives is to understand potential post-mortem diagenesis. Wisshak et al. (2009) observed the presence of secondary carbonate precipitation infilling the canals of the coral Errina antarctica plotted against the distance from the surface of the colony. Each datapoint corresponds to a visible growth band in the cross-section. (b) δ 13 C and δ 18 O data from a cross-section on Errina dabneyi (Wisshak et al., 2009). skeleton in Errina dabneyi and, although its origins (biotic or abiotic) remained unresolved, the bulk isotope composition of this specimen fits within our overall temperature calibration. Likewise, a study comparing living and sub-fossil samples of Stylaster erubescens found a significant variability in the presence and degree of taphonomic and diagenetic alteration among these samples, but, encouragingly, δ 18 O and δ 13 C data appeared consistent (Black and Andrus, 2012). Therefore, we suggest that aragonitic stylasterid corals represent practical substrates for paleoceanographic temperature reconstruction providing care is exercised in sample selection. Our work shows that, whereas there is risk of inaccurate temperature reconstruction when using highly calcitic specimens, the benefits of developing a temperature proxy that is available from many depths and many locations within the oceans will play an important role in understanding the spatial and temporal pattern of heat absorption into the interior ocean.

Conclusions
We propose stylasterid corals as a new high-resolution archive of intermediate and deep-water temperature as recorded in the oxygen isotope composition of their skeletons. They present a degree of internal variability of up to ∼0.8 and ∼4 for δ 18 O and δ 13 C, respectively. A major finding of this study is that stylasterids present limited internal variability in both δ 13 C and δ 18 O when compared to other deep-sea corals. The tips of the skeleton are always lower than the main trunk or branches for both δ 18 O and δ 13 C and further studies focusing on growth aspects of these corals may elucidate isotope fractionation effects and best sampling practices for time series work. The δ 13 C:δ 18 O linear relationships within individual specimens are similar to those reported in previous studies for deep-sea scleractinian and bamboo corals, and suggest that the isotope composition of stylasterid corals contains "vital effects" similar to other coral groups.
The difference observed in how the δ 18 O and δ 13 C relate within different parts of a single coral indicate that different biomineralisation processes act as the corals grow. We encourage future studies to be directed towards the better understanding of biomin-eralisation mechanisms, the micro-skeletal structure, the δ 18 O and δ 13 C composition, and diagenetic alteration in different genera of stylasterid corals.
We find skeletal mineralogy affects the δ 18 O and δ 13 C composition of stylasterids, with calcitic specimens showing lower δ 18 O and δ 13 C than aragonitic specimens when compared to calculated equilibrium values. For this reason, only aragonitic samples are considered for a temperature calibration.
We present a universal δ 18 O:temperature calibration for aragonitic stylasterids and temperatures ranging from 0 to 30 • C. This demonstrates that temperature is recorded in the skeletal chemistry of stylasterid corals, thus suggesting the use of stylasterid δ 18 O records to produce high spatial and temporal resolution timeseries of past seawater temperature. This similar temperature dependency is robust across the seven aragonitic genera tested here. We recommend future research to further confirm this behaviour across the diverse Stylasteridae family (30 extant genera, Cairns, 2011).
To further explore the potential of stylasterids as temperature proxies, different dating techniques should be tested on these taxa aiming for the obtainment of growth rates, lifespan and age of the specimen. Current methods used for dating corals (U-series, radiocarbon and 210 Pb) each present limitations and advantages and will have to be implemented on stylasterid corals as well to obtain accurate chronologies. Future work directed towards new proxy development in stylasterid corals, specifically trace metals and clumped isotopes, are highly welcomed in order to unravel the multi-proxy utility of this promising taxon.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.