Skeletal mineralogy of marine calcifying organisms shaped by seawater temperature and evolutionary history—A case study of cheilostome bryozoans

calcifying organisms shaped by seawater temperature


| INTRODUC TI ON
Biomineralizing marine invertebrates, also known as calcifiers, exhibit remarkable ecological and evolutionary diversity.Calcifiers are found worldwide across a wide range of depths and salinities and possess diverse anatomies, physiologies and ecological functions.
As early as the Cambrian, calcifiers are already very diverse, spanning several different animal phyla including sponges, cnidarians, molluscs, echinoderms and bryozoans (Stanley, 2008), contributing significantly to global biodiversity, habitat construction and biogeochemical processes in marine ecosystems.Such biogeochemical processes include carbon cycling and the sequestration of CO 2 , making calcifiers vital carbon reservoirs.Additionally, almost any metal present in the environment can potentially be incorporated into the calcareous structure, hence calcifiers also play a crucial role in the cycling of elements, including magnesium, strontium, manganese, copper and lead, between the oceans, sediments and the atmosphere (Gilbert et al., 2022).
The production of hard skeletal structures, such as plates, tests or shells, characteristic and vital to calcifiers, begins with the organisms' capture of calcium and carbonate ions from the environment and their subsequent transformation into crystalline structures.The ambient availability of such ions is hence a key in the biomineralization process.The precipitation or dissolution of minerals is thought to be influenced by factors such as temperature, salinity and pressure/depth (Falini & Fermani, 2013;Gilbert et al., 2022;Grenier et al., 2020;Thomsen et al., 2018).Variability in the concentration of Ca 2+ , a major cation in seawater, is almost exclusively controlled by calcium carbonate (CaCO 3 ) precipitation and dissolution, with the concentration and availability of Ca 2+ in surface open waters being relatively stable and dependent on salinity (He et al., 2020).In contrast, CO 3 2− availability is strongly dependent on dissolved inorganic carbon (DIC, mainly composed of CO 2 , H 2 CO 3 , HCO 3 − , CO 3 2− ) and total alkalinity (the capacity of seawater to neutralize acids).Both DIC and alkalinity, in turn, are affected by factors such as temperature, as well as the cycling of organic and inorganic carbon through processes like photosynthesis and respiration (He et al., 2020;Jiang et al., 2015).
Common forms of organismal CaCO 3 are calcite and aragonite, each possessing different densities and crystal structures (Gilbert et al., 2022).The physico-chemical and mechanical properties of these mineral forms are different: calcite is thermodynamically more stable than aragonite, consequently being more resistant to dissolution in cold water, and requiring less energy for its formation and maintenance.Furthermore, calcite has a lower packing density of calcium and carbonate ions, lower organic content and is more brittle than aragonite (Weiner & Addadi, 1997).These properties lead us to speculate that calcite and aragonite may be preferentially produced under different environmental and ecological conditions.At a broad scale, the mineralogical variability of calcifiers tends to reflect that of the oceans.Like temperature, the aragonite saturation state of seawater (Ω arag ) tends to decrease with latitude, peaking in the shallow waters of tropical seas.Vertically, Ω cal and Ω arag are highest in the mixed surface layer, while Ω CaCO 3 values decline with increasing depth (Carter et al., 2014;Jiang et al., 2015).This pattern primarily arises from the influence of environmental factors on seawater DIC and total alkalinity, combined with the dissolution characteristics of different CaCO 3 polymorphs (He et al., 2020;Jiang et al., 2015).Despite these generalities, the relationship between biomineralogy and environmental condition is not straightforward.
Calcifiers of different organismal groups produce either calciteonly skeletons (with minor Mg substitution), entirely aragonitic structures or a mixture of both (bimineralic), sometimes with an apparent disregard to the environmental predictions suggested earlier.
For example, reef-building scleractinian corals, the major carbonate producers in shallow tropical oceans, produce skeletons almost exclusively of aragonite (Stolarski et al., 2021;Wild et al., 2011).Yet, when phylogenetic relatedness is unaccounted for.Bryozoans in lower latitudes, characterized by higher seawater temperatures, have higher aragonite concentrations.By accounting for phylogenetic structure using a subset of 87 species for which we have topological information, 40% of the observed mineralogical variability could be attributed to present-day temperature.In contrast, depth and salinity played minor roles, explaining less than 1% of the mineralogical variation each.
Main conclusions: This study emphasizes the influence of evolutionary history on the mineralogical variability of calcifying organisms, even when it can be shown that a single environmental factor (temperature) explains a substantial amount of this variability.When confronted with changing temperature, calcifiers such as bryozoans are likely to respond in diverse ways, depending on the species, given their phylogenetic relatedness and the external conditions they meet.
To offer insights into the relative importance of environment versus phylogenetic history in determining mineralogy, this study explores the modern-day global distribution of mineralogies in cheilostome bryozoans.Within the phylum Bryozoa, the order Cheilostomatida stands out as the most evolutionarily recent.In contemporary marine ecosystems, cheilostomes dominate, comprising over 80% of living bryozoan species (Bock & Gordon, 2013).
Their evolutionary success is attributed to their ability to undergo morphological changes at an accelerated rate compared to other bryozoan orders.Key evolutionary innovations among cheilostomes include adaptations such as larval brooding and protective features for the colony, such as calcified frontal shields and polymorphic zooids like avicularia (Grant et al., 2023;Schack et al., 2019Schack et al., , 2020;;Taylor & Waeschenbach, 2015).The 5000+ described living species of cheilostome bryozoans (Bock, 2023) predominantly inhabit marine environments.These mostly sessile and almost entirely colonial invertebrates are known to attach themselves to various marine substrates such as stones, shells and plants forming large diversity of encrusting and erect colonies.A handful of species form mobile colonies (Håkansson et al., 2023;Taylor, 2020).Cheilostomes are known from the intertidal to abyssal zones (Figuerola et al., 2012) and span from the poles to the equator (Kuklinski & Taylor, 2009) in marine and brackish ecosystems (Piwoni-Piórewicz et al., 2022;Taylor et al., 2010Taylor et al., , 2015)).Aragonite-precipitating cheilostome species are thought to be most numerous in warm, shallow tropical regions, while calcite skeleton dominance increases with greater depths and towards higher latitudes (Figuerola et al., 2023;Taylor et al., 2016).Existing studies on cheilostome bryozoan mineralogy have mainly focused on polar and temperate regions (Figuerola et al., 2023), with limited mineralogical data from tropical areas, making global investigations challenging.This contribution aims to fill the gaps by procuring new data on bryozoan mineralogy, encompassing 437 species (872 colonies), most of whose mineralogy has never been studied before, including 226 species from the tropics (c.23° N to 23° S).Along with previously published data, the data we present comprise the mineralogy of 981 species and 4595 colonies with linked environmental variables.Our new cheilostome mineralogy database (BryoMinBase) serves as the foundation for testing various research questions on skeletal mineralogy, including those we present here.Specifically in this article, we (i) quantify the relative contributions of temperature, depth, salinity and their interactions to mineralogy prediction; (ii) quantify the importance of evolutionary history in explaining the variation in mineralogical composition using taxonomy as a proxy for phylogeny in order to utilize the majority of mineralogical data we have and (iii) investigate evolutionary constraints on mineralogical composition using recent molecular phylogenies with a subset of the mineralogical data.

| Data sources
BryoMinBase (Appendix 1) consists of new mineralogical data generated solely for this investigation, combined with published data from the literature.The database contains data for 4595 colonies from 96 families and 299 genera with 981 assigned species collected from temperatures ranging from −1.7 to 29.9°C, salinities in practical salinity unit (psu) from 6.6 to 40.6 and depths from 1 to 4790 m.
Sub-samples of the colonies were ground into fine powder with an agate mortar and pestle after removing visible epibiotic contaminants.Mineralogical analysis was carried out using an Enraf-Nonius PDS 120 X-ray diffractometer (Natural History Museum, London).The powdered samples were brought into suspension with a few drops of acetone and placed on a quartz crystal plate (zero background holder).The XRD system was equipped with a primary monochromator (germanium 111) and an INEL 120° curved position sensitive detector.The X-ray tube was operated at 40 kV and 35 mA and pure Co Kα 1 radiation was selected using slits settings of 0.14 × 5 mm after the monochromator.The 2-Theta linearity of the detector was calibrated with Y 2 O 3 as an external standard.
Diffractograms were collected in asymmetric flat-plate reflection geometry without angular movement of tube, sample or detector position.The tilting angle between the incoming monochromatic beam and the sample holder was kept constant at ~6°.During the analysis, the sample was rotated to increase the number of crystallites in the X-ray beam.
The weight percentages (wt%) of calcite and aragonite in the skeleton were determined by fitting sample pattern with standard patterns, generated from 100% aragonite (BM 53533) and 100% calcite (ground Iceland Spar).The error associated with this method was estimated to be within ±1% (Piwoni-Piórewicz et al., 2017).Peak assignment and fitting were performed using the Highscore software (Malvern Panalytical).

| Environmental data
The geographic coordinates of the colonies collected in this study, as well as those described in the literature, were used to obtain the ambient environmental conditions.Temperature and salinity were obtained from GIS rasters (5 arc-min spatial resolution) sourced from Bio-ORACLE (v2.2) (Assis et al., 2017;Tyberghein et al., 2012), using the coordinates of the samples.This approach was chosen because, in many cases, variables were not recorded at the time of collection, and we decided to standardize the database.Moreover, by providing mean data, we aimed to minimize the seasonal variability of the environment to some extent, as the colonies form (and mineralize) over an extended period.
From the available Bio-ORACLE raster layers representing present data, benthic layers were selected, indicating long-term mean values for an average depth in each raster cell.These data were calculated based on global ocean re-analyses of satellite and in situ observations spanning from 2000 to 2014.
In cases where depth was not available from collection metadata, bathymetric data were extracted from MARSPEC data (30 arcsecond resolution, Sbrocco & Barber, 2013).The 'extract()' function from the R package raster (Hijmans, 2022) facilitated the retrieval of both bathymetric and environmental data.All data were extracted from a radius of 1000 m at each location or, if not possible, from 2000 to 5000 m.Data extracted from raster cells within the radius were mean averaged.

| Environmental factors associated with mineralogy
To explore the extent to which each of the considered environmental variables explains the observed mineralogical composition of cheilostome species, we first compared several simple models.Environmental data for the response variable (mineralogy) were associated with specific colonies.Colonies with missing environmental data were excluded from the analysis.Where species were represented by multiple colonies (replicates), median observed values (calcite wt%, aragonite wt%, temperature, salinity and depth) across these colonies were used in the analyses.Where colonies were identified only to genus level, we assumed that congeneric colonies from different locations were separate species, determined by geographical coordinates: latitude and longitude (location_IDs).We combined these data with samples that were identified to species level (species_IDs_only), resulting in a list of 981 species representing the analysed database (species_IDs_only_location_Ids; Appendix 1).As a test of whether our results were sensitive to species-level variability in mineralogy, we re-ran the models outlined 100 times by randomly selecting individual colonies representing a particular species.
We first modelled the response variable as binary (where 0 indicates no aragonite and only calcite, while 1 indicates only aragonite and no calcite) using a binomial generalized linear model (GLM) with the explanatory variables: temperature, salinity and depth additive and multiplicative combinations.Only the genera/species with entirely aragonite or calcite mineralogy were used in the analyses.
Although most of our mineralogy data had values of 0 or 1, there are bimineralic cases that cannot be ignored.To include these cases, first, we use ordinal logistic regression (OLR), where the response variables are: 0 = only calcite, 1 = only aragonite and 2 = bimineralic; again, using the same environmental factors as independent variables in different combinations.As a second approach, a linear model was used, where small values (0.001, 0.01 or 0.1) were added to both 0 and 1 values to allow linearization of the response variable using the logit transformation, defined as log(p/1−p) where p is mineralogy.This last model is presented in the main text, while other results (consistent with the main text results) are given in the Supplementary Material.

| Evolutionary history and mineralogy
While we expected that environment likely explained some of the variation in mineralogy, we wanted to test whether phylogenetic information would improve model fit.To include information on evolutionary history into the analysis, we first used 'clade membership' as an additional explanatory variable in the models described earlier.
We were able to assign 768 species (3497 colonies) in BryoMinBase (Appendix 1) to the seven major cheilostome phylogenetic clades (A through G) following Orr et al. (2022).
In a more nuanced approach, we used a smaller subset of our data, representing 87 species (122 colonies) that were successfully sequenced in Orr et al. (2022).We used phylogenetic generalized least squares models in the R package caper (Orme, 2023) analogous to those of our linear model in Table 1.Pagel's lambda, λ (Pagel, 1999) ranges from zero to one, where zero indicates a complete absence of phylogenetic signal of the trait in question (mineralogy).
For all statistical calculations and visualization, we used R ver.
4.2.2GUI 1.79 (R Core Team, 2022).The R script associated with this study is provided in the Supplementary Materials.

| Mineralogy of bryozoans
The skeleton of cheilostome bryozoans exhibit three primary mineralogical forms: calcite, aragonite and bimineral.Of the 981 species studied, fully calcitic skeletons dominated (n = 635), 197 species were bimineralic and 149 species produced only aragonite (Figure 1).Of the environmental factors studied, temperature best explained the mineralogical variability of cheilostome bryozoans (Table 1).Aragonite wt% content in skeletons tended to increase with increasing seawater temperature (Figure 2).Among the studied species, 433 had two or more replicates.Mineralogical variability was observed in 58 species (out of 433, i.e. about 13%) and occurred either between aragonitic and bimineralic or calcitic and bimineralic states.Considering the intra-specific mineralogy of cheilostome bryozoans, it became evident that the variability existed both at the same location (Figure 2) and across different locations (Appendix 1).

| Modelling only the environmental factors
The best model according to the adjusted Akaike information criterion (AICc) was a model that included only temperature as a predictor variable for mineralogy in analyses using all the data that we had available (Table 1).Figure 3 shows the prediction of this model, where the logit transformation is used to linearize the response variable.Temperature alone explained 20% of the mineralogical variation (R 2 = 0.20) within cheilostome bryozoan skeletons; large differences in mineralogy between species remained unexplained (at least 80%, Figure 3).Only 1% of the mineralogical variance was explained by depth, while salinity alone accounted for just 0.3% (Table 1).When a binomial GLM was applied, using only purely calcitic or aragonitic data (there were 197 fewer species in this dataset than in the linear model analyses), the best model (Table S1) has a multiplicative effect between temperature and depth, though the next best model included only temperature, where the removal of depth reduced the R 2 by only 0.01.Finally, when an OLR was applied, the best model (Table S1) also had a multiplicative effect of temperature and depth (pseudo R 2 = 0.11, Figure 4).

| Modelling with categorical phylogenetic information (clades)
Cheilostome bryozoans (768 species and 3497 colonies) belonging to seven major cheilostome phylogenetic clades (A through G, see

TA B L E 1
The linear models of mineralogical variability of cheilostome bryozoans.By incorporating species' affiliation with a specific phylogenetic clade as an additional predictor to the simple linear models, the best model included an interaction between temperature and species' clade membership (Table 2).Here, 28% of the observed variation in mineralogy was explained by temperature.The relationships between mineralogy and temperature varied across clades (Figure 5), with the steepest relationships in clades G (slope = 0.44) and C (slope = 0.38).Among the different clades, the slope is positive for four (clades G, E, F, D), negative for one (clade A) and for the remaining clades, there is no significant effect of temperature on mineralogy (Table S3, Orr et al., 2022).

| Modelling with phylogenetic tree
The phylogenetic generalized least squares (PGLS) models were based on the results of linear models that identified temperature as the primary environmental factor explaining the mineralogical variability observed in cheilostome bryozoan skeletons.
Considering both temperature and the phylogenetic relationships among bryozoans for a smaller subset of data (see Methods), the model explained 40% of the variance in transformed aragonite content (Table 3).Notably, organisms with similar mineralogy exhibited a phylogenetic relationship, which is particularly evident in the context of aragonite skeletons (Figure 6).The presence of a strong phylogenetic signal in the mineralogical variability is supported by the Pagel's lambda (λ) value of 0.95 (Table 3).This signal indicated that mineralogical traits of closely related species tended to be more similar.The dataset used for the PGLS model was smaller (87 species) than the dataset input to the linear models have exclusively calcitic skeletons (Figure 1), corroborating previous observations based on smaller datasets (Figuerola et al., 2023;Smith et al., 2006Smith et al., , 2016;;Taylor et al., 2015).Therefore, the pattern of calcite as a major component of bryozoan skeletons seems to  Gilbert et al., 2022;Smith et al., 2012;Smith, Berman, et al., 2013;Smith, Riedi, et al., 2013;Vinn, 2021).Nevertheless, clear evidence suggests that aragonite also contributes significantly to biogenic CaCO 3 production.Aragonite is the predominant biomineral found in the skeletons of organisms such as scleractinian corals and molluscs (Sulpis et al., 2021(Sulpis et al., , 2022)).
The reasons why different taxa evolved to utilize one mineral over another remain largely unknown, however.Although there is a vast literature on the interactions between organisms and their environments during mineral production, variability in mineralogy is influenced by both environmental and biological controls (Cusack & Freer, 2008;Gilbert et al., 2022;Grenier et al., 2020;Stanley, 2008;Taylor et al., 2015;Telesca et al., 2019).This creates a dependence of skeletons on the environment to varying degrees, contingent upon the species.
To add to the complexity of the situation, biogenic calcite may also have varying content of magnesium (Mg) added in calcifers (Lebrato et al., 2016;Taylor et al., 2015).Solubility and hence the energetics of maintaining the skeleton may be modified by inclusion of The effect of temperature and depth on the mineralogy of cheilostomes.The figure shows the probability of the occurrence of fully calcitic, fully aragonitic and bimineralic skeletons under the influence of interaction between temperature and depth, based on the ordinal logistic regression (Table S1).

TA B L E 2
The linear models of mineralogical variability of cheilostome bryozoans including clade membership.S3.For details on species and their clade membership, see Appendix 1. Clade membership was established based on molecular sequences from Orr et al. (2022).
Although it must be acknowledged that our understanding of bryozoan calcification may be modified by incorporating Mg into the unfolding story (see section 'Environmental predictors of minerology'), we chose not to consider the currently (sparsely) available data on Mg for this study.The main reason for this is a much lower data availability compared to calcite-aragonite composition and a resulting lack of information for robust conclusions.
Numerous studies have indicated that mineralogical variability has an evolutionary context (Gilbert et al., 2022;McDougall & Degnan, 2018;Taylor & Waeschenbach, 2015).The majority of phyla with CaCO 3 skeletons made their first appearances in the fossil record during the Cambrian period (Kouchinsky et al., 2012;Zhang et al., 2021), a time when the chemical conditions of seawater, mainly Mg/Ca seawater ratio, favoured the production of The relationships between mineralogy and seawater temperature across various phylogenetic clades of cheilostome bryozoans.The plot shows predictions of the relationship between mineral composition (logit-transformed aragonite content) of cheilostomes and the impact of temperature (derived from the best linear model Table 2).The inset in the top-left corner provides an overview of the relationships among the main bryozoan clades (Orr et al., 2022).For details on families and genera belonging to each studied cheilostome clade, see Table S2.

TA B L E 3
The phylogenetic generalized least squares (PGLS) analysis of the mineralogical variability of cheilostome bryozoans.calcite (Conci et al., 2021;Li et al., 2023).Data of Porter (2007Porter ( , 2010) ) show that aragonitic clades emerged during the subsequent aragonite sea phase in the Earth history, while calcitic clades appeared after the transition to a calcite sea.Consequently, a correlation exists between the mineralogy of newly-evolved skeletons and dominant seawater chemistry.Although an exception was recently found in molluscs (Li et al., 2023), this relationship has been supported by various studies (Balthasar & Cusack, 2014;Stanley, 2006;Taylor, 2008).Yet, it is essential to highlight that once mineralized skeletons had evolved, fluctuations between calcite and aragonite seas appear to exert little influence on skeletal mineralogy.Some groups of organisms retain their original F I G U R E 6 Phylogenetic clustering of skeletal traits and environment.The cheilostome tree topology shows the phylogenetic relationships between bryozoan species with calcite skeletons marked in white, aragonite skeletons marked in black and bimineralic skeletons presented in grey scale from low aragonite content (light dots) to high aragonite content (dark dots).Each species has environmental characteristics attached as median values for the given species.Increasing temperature is marked on a scale from blue to red, and increasing depth is marked on a scale from dark green to light green.
mineralogy to this day, such as calcitic echinoderms, brachiopods or cyclostome bryozoans.In contrast, many taxa switched mineralogies, and their mineralogical variability does not appear to reflect changes in aragonite-calcite seas.For example, skeletons of molluscs and tubes of serpulid polychaete ancestrally were aragonitic, and cheilostome bryozoans were calcitic (Taylor et al., 2009), yet today all these groups of organisms produce calcite, aragonite and bimineralic skeletons (Figure 1; Gilbert et al., 2022).Due to the enormous diversity of marine calcifying invertebrates characterized by different evolutionary pathways, structural complexities, functional differences and degrees of biological complexity, there is no clear answer to the question about the factors responsible for the mineralogical variability observed in today's living organisms.Porter (2010) suggested that while there may be increased costs associated with producing mineralogy not favoured by ambient seawater, the costs of switching mineralogy in response to changing environmental conditions over time may be even greater.
Therefore, organisms have developed a number of adaptations, including polymorph-specific proteins involved in the biomineralization process (Zhang & Zhang, 2006).Additionally, there may be many other reasons for mineralogical variability related to the design and function of mineralized structures, skeletal strength, dissolution or density, resulting from the physico-chemical properties of calcite and aragonite.For example, the skeletons, in many cases, have a protective function that may determine predation rates.Skeletons also have importance for supporting biological structures which themselves may become less or more important at different times in evolutionary history and in different regions of the world ocean (Gilbert et al., 2022;Porter, 2010;Taylor & Reid, 1990;Zhuravlev & Wood, 2009).The results of these studies indicate that the mineralogical variability of cheilostome bryozoans depends on the environment, but this is not true for all clades (Figures 3 and 4).It is difficult to explain such relationships, but biological factors seem to be responsible for inter-specific variability in mineralogical response to environmental conditions.
The Cheilostomatida mineralogy database (BryoMinBase) not only reveals inter-specific mineralogical variability but also highlights intra-specific variations.In general, the mineral composition of individual colonies from the same species remains relatively constant (Smith & Girvan, 2010).However, among the 433 species analysed with at least two replicates, mineralogical variability was evident in 58 species, both within the same location and across different locations (Figure 2 and Appendix 1).Several factors may contribute to or cause this intra-specific variability, including the potential biological response of a given species to environmental changes (Swezey et al., 2017a(Swezey et al., , 2017b)), increased aragonite content with the age of the organism, growth rates, breeding cycles or food availability (Barnes et al., 2007;Loste et al., 2003;Smith, 2007;Smith & Girvan, 2010), as well as potential phenotypic plasticity.However, bias related to taxonomy cannot be ruled out.Even small differences in morphological characters can indicate a new species (Baptista et al., 2022;Caceres-Chamizo et al., 2017;Grischenko et al., 2022;Harmelin et al., 2011).It is plausible that our examination may have overlooked some important minute morphological characters and the observed intra-specific variation in skeletal mineralogy may indicate the presence of two sibling species.In other words, what exactly drives these intra-specific variations in this study requires further investigation.

| Environmental predictors of mineralogy
Among the environmental factors investigated, temperature emerges as the most influential predictor of skeletal mineralogy for 981 cheilostome bryozoan species (20% of known species to date), explaining 20% of the observed variability (Table 1).Although there are gaps in both our species and spatial sampling, there are no glaring biases that could contribute to the numbers we observed.This observation underscores the established relationship between mineralogical variability in cheilostome bryozoans and a latitudinal gradient, corresponding with an increase in aragonite content as seawater temperature rises (Figuerola et al., 2023;Krzeminska et al., 2016;Kuklinski & Taylor, 2009;Piwoni-Piórewicz et al., 2020;Taylor et al., 2009Taylor et al., , 2015Taylor et al., , 2016)).This pattern is also evident in other groups, such as bimineralic bivalve Mytilus californianus (Olson et al., 2012) or the serpulid tubeworm Hydroides elegans (Chan et al., 2013).The Despite the variability of environmental conditions, many marine organisms exhibit consistent skeletal mineralogies.For example, aragonitic pteropods thrive in Arctic waters, while calcite echinoderms flourish in tropical regions (Anglada-Ortiz et al., 2021;Lebrato et al., 2016), suggesting a strong influence of evolutionary and biological factors on skeletal mineralogy.While calcite and aragonite differ significantly in their crystal structures, density, elemental composition and protein content, impacting their physical properties and energetic costs of production (Weiner & Addadi, 1997), biological factors such as sex, age, skeletal organic matter, biomineralization strategy and ecological functions are also key determinants.As already mentioned, magnesium is sometimes incorporated into the calcite structure (MgCO 3 ) in bryozoans, potentially altering skeletal characteristics, as the solubility of calcite is known to rise with its Mg content (Ries, 2011).Figuerola with co-authors (2023) observed a tendency for bryozoans in cold waters to secrete less soluble calcite with lower MgCO 3 content compared to those in warmer waters.Similar findings have been reported in echinoid skeletal calcite (Duquette et al., 2018;Smith et al., 2016) and in benthic calcitic foraminifera (Maeda et al., 2017).However, the mechanisms regulating the Mg content in marine calcitic skeletons remain poorly understood and exhibit varied effects across different organisms and species.The Mg content of skeletal structures has been observed to co-vary with changes in seawater temperature, as well as with factors such as seawater CaCO 3 saturation state, growth rate variation and food availability (Figuerola et al., 2015;Lebrato et al., 2016;Smith et al., 2016).Therefore, it is conceivable that among cheilostome bryozoans, there might be a tendency towards an increased proportion of not only aragonite but also Mg in calcitic skeletons in warm waters, a hypothesis that warrants further investigation.
Salinity and depth, closely tied to the seawater carbonate system and hence the seawater CaCO 3 saturation state (Ω), have the potential to influence the formation of carbonate skeletons (Jiang et al., 2015).
According to the implemented models, the influence of factors such as salinity and depth on the mineralogical variation of studied cheilostome bryozoans is relatively small compared to the pronounced effect of temperature (Tables 1 and S1).However, these factors contribute to a more comprehensive understanding of this variability, particularly when focusing on the probability of occurrence of a given mineralogical form rather than the weight percentages of calcite and aragonite.
This study emphasizes the interaction between temperature and depth as a crucial predictor of mineralogical variability (Figure 4).The It is noteworthy that all analysed organisms come from ocean zones where the water is saturated with CaCO 3 , above the compensation depth for aragonite and calcite (Law et al., 2018;Sulpis et al., 2021).From the ocean's surface, the level of ΩCaCO 3 gradually diminishes with increasing depth until the seawater attains ΩCaCO 3 saturation less than one, resulting in the dissolution of carbonate minerals occurring faster than their production (Sulpis et al., 2021).
This decline is attributed to the increased solubility of CaCO 3 under elevated pressure, decreasing temperature and escalating CO 2 levels (Sulpis et al., 2021;Woosley, 2018).It is important to note that variations in physical factors and carbon cycle processes contribute to the differences in compensation depth across different oceanic regions.Furthermore, the distinct thermodynamic properties of calcite and aragonite result in the calcite compensation depth being deeper than the aragonite compensation depth (Jiang et al., 2015;Luo et al., 2016).The results of this study suggest that depth may explain the mineralogy of skeletons through the ΩCaCO 3 gradient along it (Figure 4), and the influence of environmental factors on the skeletal mineralogy is a system of mutual dependencies.
The influence of salinity on cheilostome bryozoan mineralogy may have been masked in our study because most samples originated from areas with fully marine salinity (Appendix 1).Previous studies along salinity gradients suggest that the majority of cheilostome bryozoans do not tolerate low salinity (Grabowska & Kukliński, 2016;Piwoni-Piórewicz et al., 2022).Only a few species, such as Einhornia crustulenta and Electra pilosa, were collected at salinities below 20, and they exhibited entirely calcitic skeletons.However, although at higher salinities ranging from 32 to 40 a significant mineralogical variability is observed, it appears to be unrelated to salinity.Thus, under typical marine conditions, salinity plays a minor role in influencing the mineralogical variability of cheilostomes.Yet, at lower salinities, it may become more important, though additional samples/species from such environments are necessary to confirm this hypothesis.

| Phylogenetic history and mineralogy
Changes in seawater parameters over geological time scale are believed to have exerted significant influence on the biomineralization of marine organisms (Gilbert et al., 2022;Stanley, 2008;Taylor et al., 2010).Biominerals are thought to have emerged after the divergence of multiple phyla, with calcifying organisms evolving distinctive strategies for biomineralization within skeleton-forming clades (Stanley, 2006;Stanley & Hardie, 1998;Pawlowski et al., 2003).
When incorporating the temperature predictor within a phylogenetic framework, the clade interaction model explains 28% of mineralogical variability, surpassing the explanatory power of a temperatureonly model (Table 2), which suggests that understanding of the clade membership allows for a more comprehensive explanation of mineralogical variability than solely considering temperature.Notably, the relationship between mineralogy and temperature varies across phylogenetic clades (Figure 5, Table S3).et al., 2022).Therefore, the appearance of aragonite occurred during the evolution of cheilostomes could be associated with the differentiation of morphological features within this group of organisms (Orr et al., 2022;Ostrovsky, 2013;Schwaha et al., 2020;Taylor & Waeschenbach, 2015;Waeschenbach et al., 2012).Mineralogical variability among phylogenetic clades has been observed previously; for instance, Loxton with co-authors (2018) revealed a strong phylogenetic signal for calcite content in the skeletons of Scottish bryozoan species.Similar findings have been reported in other organism groups.Phylogenetic analyses have demonstrated, for example, that within the gastropod family Littorinidae, most clades exhibit aragonitic structures, but three distinct clades have independently evolved calcitic layers (Taylor & Reid, 1990).
When considering the cheilostome phylogeny, 40% of mineralogical variability can be attributed to temperature (Table 3) where aragonitic species tend to inhabit shallow, warm waters, similar to bimineralic species with high aragonite content (Figure 6).Increasing evidence supports the pivotal role of evolutionary factors in the biomineralization process (Gilbert et al., 2022).Our study is in agreement with other studies that included phylogeny in their analyses of biomineralogy: e.g.models accounting for phylogenetic relatedness outperformed those neglecting evolutionary relationships among serpulids (Smith, Riedi, et al., 2013).
Although studies linking mineralogical variation with molecular phylogeny are not yet commonplace, various evolutionary mechanisms, such as gene expression, play a leading role in controlling mineralogical variability, for example, among bivalvia (Kobayashi & Samata, 2006;McDougall & Degnan, 2018;Wang et al., 2014) and octocorallia (Conci et al., 2021).In addition, Smith et al. (2012) noted the presence of a phylogenetic signal within clades in the calcite/aragonite ratio among calcifying sponges (Smith, Berman, et al., 2013).The mineralogical diversity among calcifying organisms is extensive and largely influenced by their position within the phylogenetic tree.The more we understand the evolutionary history of a given group of organisms, the better we can explain the mineralogical variability of CaCO 3 skeletons in relation to the environment.Among calcifiers, cheilostome bryozoans exhibit significant potential for environmental studies concerning temperature.

| CON CLUS IONS
This study advances our understanding of bryozoan mineralogy with

F
An overview of mineralogical distribution among 981 species and 4595 colonies of cheilostome bryozoans.(a) Number of analysed species categorized by the source of data into those generated in this study (437 species) and literature data (601 species), of which the mineralogy of 380 species is presented for the first time.(b-d) Geographic distribution of all species with mineralogical type available in the data from (a).The size of the points indicates the number of colonies/specimens within species at given location, while shape indicates the data source.14668238, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13874by Nat Prov Indonesia, Wiley Online Library on [07/06/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License (981 species).In a linear model of those 87 species, temperature explained 28% of the mineralogical variation.4 | DISCUSS ION 4.1 | Mineralogy of CaCO 3 skeletons To our knowledge, the BryoMinBase dataset presented here is the most extensive examination of the skeletal mineralogy of cheilostome bryozoans in terms of geographic scope and species diversity, including relevant environmental factors (Appendix 1).The mineralogy of 380 species is presented for the first time, with 226 species from the least explored tropical region.Calcite is the predominant mineral in the skeletal structures of cheilostome bryozoans.A large majority of species (n = 635, 65%) be consistent and aligns with a broader trend of calcite dominance among calcifying organisms on a global scale.Calcite is widely recognized as the most stable CaCO 3 mineral and is estimated to constitute the majority of the oceanic CaCO 3 reservoir (Sulpis et al., 2021).It forms the skeletons of various modern calcifying F I G U R E 2 Intra-specific variability in skeletal mineralogy of cheilostome bryozoans and seawater temperature.Plot presents the variability of species mineralogy along the temperature gradient.Each point represents one species, the size of the circles indicates the number of colonies/specimens available in this study and whiskers indicate standard error of aragonite content and temperature.F I G U R E 3 The relationship between mineralogy of cheilostomes and temperature.Here on the y-axis, 0 = fully calcitic skeletons, 1 = fully aragonitic skeletons and values in between indicate bimineralic skeletons.The solid red line represents the predicted aragonite content in skeletons derived from the best linear model outlined in Table 1, incorporating coefficients transformed using the inverse logit function.The dotted blue lines correspond to the 95% confidence interval for the prediction.Each black point on the graph denotes the median values of mineralogy in relation to temperature for individual species.organisms,including coccolithophores, foraminiferans, barnacles, brachiopods, echinoderms and cyclostome bryozoans, all of which typically have monomineralic skeletons(Sulpis et al., 2021(Sulpis et al., , 2022;;Ullmann et al., 2018).Molluscs, octocorals, calcareous sponges, serpulid worms and cheilostome bryozoans are known to produce calcite and aragonite, having mineralogically diverse skeletons, both monomineral or bimineral (Figure1;Conci et al., 2021; Aragonite values are the logit-transformed response variable, while environmental variables and cheilostome bryozoan phylogenetic clades are predictors, incorporating main (+) and interaction (*) effects.The model with the lowest corrected Akaike information criterion (AICc) score shown in bold indicates the variable with the best fit.D, depth; S, salinity; T, temperature.Adjusted R-squared (adjR 2 ) values indicate the proportion of the variance in the response variable explained by the predictors.Detailed results of interaction between temperature and clades are provided in Table general trend and individual examples indicate a mechanism promoting an increase in aragonite proportion in warmer waters, at the same time pointing to the complexity of the biomineralization process and the influence of many other factors (as indicated by the deviations of empirical data from the general prediction of the model in Figure 3), both environmental and biological, on the composition and properties of skeletons.
models are constructed based on mineralogical variability observed across a depth gradient ranging from 0 to 4654 m.Of the 120 cheilostome species from deeper than 200 m in this study, four contained aragonitic skeletons and seven had a bimineralic composition.The deepest occurrence of aragonite was noted at approximately 520 m in one bimineralic species, while calcite occurred below this depth up to 4654 m in 35 species, highlighting the dominance of calcite in the deep regions of the oceans (Appendix 1).The relatively modest effect of depth on mineralogical variability, as indicated by the model calculations, may be attributed to the fact that most bryozoan species occur and were collected in shallow zones, resulting in a less abrupt gradient of this environmental factor.

14668238, 0 ,
Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13874by Nat Prov Indonesia, Wiley Online Library on [07/06/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License within clade A, favouring calcite.Clades A and B are represented by bryozoans of a 'malacostegan grade', which is the most ancient cheilostome clade characterized by simple skeletal morphology (Grant et al., 2023).Organisms belonging to this group are typically weakly calcified and lack any biomineralized frontal wall.Contrary, clade G, being phylogenetically the youngest, exhibits variability that reflects temperature most accurately among all clades (Figure 5, Orr

a
substantial amount of new data.It emphasizes the importance of considering both evolutionary and environmental factors in interpreting mineralogical variations among calcifying organisms.We corroborate that temperature is a crucial factor explaining the skeletal mineralogy of cheilostome bryozoans, revealing a distinct latitudinal gradient characterized by increased aragonite content with rising seawater temperature.The adoption of a phylogenetic approach ensures accurate and contextually meaningful interpretations of environmental conditions based on mineralogical variability.This research underscores that the susceptibility of calcifying organisms to environmental changes is influenced by evolutionary factors, and different species may exhibit a wide range of responses to climate change.
14668238, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/geb.13874by Nat Prov Indonesia, Wiley Online Library on [07/06/2024].See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions)on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 2.1.2| New mineralogical data Bryozoans were collected from tropical and subtropical zones (Australia, Mediterranean Sea, Red Sea, South Africa, New Zealand, Aragonite values are logit-transformed response variable while environmental variables and used as predictors, taking into consideration known molecular phylogenetic relationships among cheilostome species (N = 87).The model includes main effects (+) and interaction effects (*).The predictor with the lowest corrected Akaike information criterion (AICc) score shown in bold indicates the variable with the best fit.D, depth; S, salinity; T, temperature.D:T means the interaction between depth and temperature, S:T is the interaction between salinity and temperature.Adjusted R-squared (adjR 2 ) values indicate the proportion of the variance in the response variable explained by the predictor.Pagel's lambda (λ, 0-1) indicates the strength of phylogenetic signal.