Impact of trace metal concentrations on coccolithophore growth and morphology: laboratory simulations of Cretaceous stress

Abstract. The Cretaceous ocean witnessed intervals of profound perturbations such as volcanic input of large amounts of CO2, anoxia, eutrophication and introduction of biologically relevant metals. Some of these extreme events were characterized by size reduction and/or morphological changes of a few calcareous nannofossil species. The correspondence between intervals of high trace metal concentrations and coccolith dwarfism suggests a negative effect of these elements on nannoplankton biocalcification processes in past oceans. In order to test this hypothesis, we explored the potential effect of a mixture of trace metals on growth and morphology of four living coccolithophore species, namely Emiliania huxleyi, Gephyrocapsa oceanica, Pleurochrysis carterae and Coccolithus pelagicus. The phylogenetic history of coccolithophores shows that the selected living species are linked to Mesozoic species showing dwarfism under excess metal concentrations. The trace metals tested were chosen to simulate the environmental stress identified in the geological record and upon known trace metal interactions with living coccolithophore algae. Our laboratory experiments demonstrated that elevated trace metal concentrations, similarly to the fossil record, affect coccolithophore algae size and/or weight. Smaller coccoliths were detected in E. huxleyi and C. pelagicus, while coccoliths of G. oceanica showed a decrease in size only at the highest trace metal concentrations. P. carterae coccolith size was unresponsive to changing trace metal concentrations. These differences among species allow discriminating the most- (P. carterae), intermediate- (E. huxleyi and G. oceanica) and least-tolerant (C. pelagicus) taxa. The fossil record and the experimental results converge on a selective response of coccolithophores to metal availability. These species-specific differences must be considered before morphological features of coccoliths are used to reconstruct paleo-chemical conditions.


Introduction
Trace metal concentrations influence the productivity and species composition of marine algae communities (Bruland et al., 1991;Sunda and Huntsman, 1998).A number of trace metals are important micronutrients (e.g., zinc, iron, copper, nickel), but some of them can become toxic and inhibit marine algal productivity at elevated concentrations (Brand et al., 1986;Sunda and Huntsman, 1992).Others like lead and mercury have no known metabolic functions and can hamper marine phytoplankton growth already at low concentrations (Sunda, 1989;Sunda et al., 2005).
The geological record offers the opportunity to investigate past case histories marked by profound changes in the ocean, such as volcanic injection of large amounts of CO 2 , ocean anoxia, eutrophication and introduction of biologically relevant metals (e.g., Larson and Erba, 1999;Erba, 2004;Jenkyns, 2010;Erba et al., 2015).These events can be seen as "natural experiments" useful to decrypt the ecosystem response to major perturbations at timescales longer than current modifications.Finding out how the changes in seawater composition affect marine biota requires the integration of a long-term and large-scale geological perspective that has been recognized as an essential ingredient for more coherent Published by Copernicus Publications on behalf of the European Geosciences Union.
predictions of how marine organisms might react to future environmental changes.Insights on ocean-atmosphere dynamics under warmer-than-present-day conditions predicted for the end of this century can be obtained by including geological data of past ecosystems, especially those derived from cases of extreme conditions.Well-known perturbations were the oceanic anoxic events (OAEs) which took place during the Mesozoic.These events were caused by intense volcanism that produced large igneous provinces (LIPs; Snow et al., 2005;Neal et al., 2008;Pearce et al., 2009;Erba et al., 2015) that released magmatic fluids delivering metals, mixed with warmed ambient seawater that had enough buoyancy to rise to the surface (Snow et al., 2005;Erba et al., 2015).During the latest Cenomanian OAE 2, for example, less volatile elements, such as nickel and iron (released during the formation of the Caribbean LIP), increased by 8-20 times above seawater background levels while more volatile elements like lead and cadmium (derived from water-rock exchange reactions) increased by about 4-8 times above background level (Orth et al., 1993;Snow et al., 2005).Entering the ocean environment, more and less volatile elements became biologically relevant as evidenced by changes and turnover in marine plankton communities (Leckie, 1985;Leckie et al., 1998;Erba, 2004;Erba et al., 2015).
Studies on calcareous nannofossils documented a size reduction of some coccolithophore species (Biscutum constans, Zeugrhabdotus erectus and Discorhabdus rotatorius) that are coeval with trace metal concentration peaks across both the early Aptian OAE 1a and latest Cenomanian OAE 2 (Erba et al., 2015;Faucher et al., 2017).The fossil record shows that, although most of the Mesozoic nannoplankton taxa did not survive the mass extinction event at the end of the Cretaceous, reconstructed phylogenetic trees (Bown et al., 2004), based on morphological observations of coccolith (shape and ultrastructure), and molecular trees, mostly based on rDNA (de Vargas and Probert, 2004), indicate a link between selected Mesozoic groups and some living coccolithophores.The group to which the four species tested here belong to, namely Emiliania huxleyi, Gephyrocapsa oceanica, Coccolithus pelagicus and Pleurochrysis carterae, evolutionarily diverged from one another since the Late Cretaceous, with the exception of E. huxleyi and G. oceanica that are separated since 250 000 years ago (De Vargas et al., 2007;Liu et al., 2010).Specifically, genera Emiliania and Gephyrocapsa belong to the Cenozoic family Noelaerhabdaceae derived from the extinct Prinsiaceae that, in turn, branched off the Mesozoic family Biscutaceae.Indeed, coccolith dwarfism was observed in genus Biscutum during times of high CO 2 and metal concentrations in both OAE 1a and OAE 2 (Erba et al., 2010;Faucher et al., 2017).It is challenging to unambiguously disentangle the cause(s) of such changes in the fossil record, but evidence of a correspondence between intervals of high trace metal concentrations and coccolith dwarfism suggests a negative effect of these elements on nannoplankton biocalcification processes.Previous work on the response of living coccolithophores to trace metal concentrations focused on Emiliania huxleyi, one of the most abundant species in the world ocean with a nearly global distribution (Westbroek et al., 1989;Winter et al., 2014).Experiments documented a decreasing growth rate under high trace metal concentrations (Vasconcelos et al., 2001;Hoffmann et al., 2012;Santomauro et al., 2016).So far, no such studies have been performed on other coccolithophore species.Furthermore, to our knowledge, this is the first study investigating the effect of high trace metal concentrations on coccolithophore and coccolith morphology and size.The trace metals tested (Zn, V, Ni, Pb) were chosen based on peaks identified in the geological record (Snow et al., 2005) and known trace metal interactions with living coccolithophores to simulate the environmental conditions during OAEs.The main goal of this study is to understand if, similarly to the fossil record, anomalously high quantities of essential and/or toxic metal induce changes in coccolith shape and size and cause coccolith dwarfism in coccolithophore species.
More specifically, we address the following questions: (i) does coccolithophore growth change in response to increasing trace metal concentrations?(ii) Does coccolith size and morphology, as well as coccolithophore size, change in response to high and anomalous trace metal concentrations?(iii) Do trace metal combinations, which mimic OAE conditions, lead to a uniform response among species or to speciesspecific responses on morphological features?(iv) Do coccolith morphometrical features have a potential to serve as a proxy to reconstruct paleo-ocean trace metal concentrations? 2 Materials and methods

Culture conditions
Monospecific cultures of the coccolithophores Emiliania huxleyi (strain RCC 1216), Gephyrocapsa oceanica (strain RCC 1303), Coccolithus pelagicus (strain PLY182G) and Pleurochrysis carterae (no known strain number) were grown as batch cultures in artificial seawater produced as described by Kester et al. (1967).The artificial seawater medium was enriched with 64 µmol kg −1 nitrate, 4 µmol kg −1 phosphate to avoid nutrient limitations, f/8 concentrations for vitamins (Guillard and Ryther, 1962), 10 nmol kg −1 of SeO 2 and 2 mL kg −1 of natural North Sea water (Bach et al., 2011).The carbonate chemistry was adjusted by bubbling with CO 2 -enriched air overnight to reintroduce inorganic carbon, thereby reaching atmospheric CO 2 partial pressure (∼ 400 µatm).All culture bottles were manually and carefully rotated three times a day, each time with 20 rotations in order to avoid cell settling.In the control treatment, the medium was enriched with f/8 concentrations for trace metals (Guillard and Ryther, 1962).
Pb, Zn, Ni and V concentrations were added in low (L), medium (M), high (H ) and extreme treatments because of their high concentrations identified in the Aptian OAE 1a (Erba et al., 2015) and Cenomanian-Turonian OAE 2 (Snow et al., 2005; Table 1).The trace metal chelator EDTA (ethylenediaminetetraacetic acid) was added to the trace metal stock solutions in order to guarantee a constant level of bioavailable trace metals for phytoplankton and prevent metal precipitation.The cultures were incubated in a thermo-constant climate chamber (Rubarth Apparate GmbH) at a constant temperature of 15 • C, a 16 : 8 (hour : hour) light-dark cycle and at a photon flux density of 150 µmol photons m −2 s −1 .
The cultures were pre-exposed to the different treatment conditions (acclimation period) for 7-10 generations, which varied between 6 and 10 days depending on the speciesspecific cell division rates.All cultures were incubated in autoclaved 500 mL square glass bottles (Schott Duran).The initial cell density was relatively low with ∼ 50 cells mL −1 .Final samples were taken when cells were still in their exponential growth phase and cell numbers were low enough to avoid a strong change in the chemical conditions of the growth medium.Therefore, the experimental duration differed among species and among treatments (between 6 and 10 days) due to the different growth rates.Each treatment was replicated three times.Final cell densities in the experiment did not exceed 50 000 cells mL −1 in E. huxleyi, 20000 cells mL −1 in G. oceanica, and 3000 cells mL −1 in C. pelagicus and P. carterae.

Cell abundance and growth rate
Samples for cell abundance were taken every 2nd day with the exception of the control treatment where samples were only taken at the end of the experiment.Incubation bottles were gently turned 10 times in order to obtain a homogenous suspension of the cells before sampling.Cell numbers were immediately measured three times without addition of preservatives using a Beckman Coulter Multisizer.Specific daily growth rates (µ) were calculated from the leastsquares regression of cell counts versus time during expo- nential growth (Eq.1): where c0 and c1 are the cell concentrations at the beginning (t0) and at the end of the incubation period (t1), respectively.

Coccosphere and cell sizes
Cell abundance samples were acidified with 0.1 mmol L −1 HCl to dissolve all free and attached coccoliths and subsequently measured three times in order to obtain cell diameters and volumes (Müller et al., 2012).In this study, we define the coccosphere volume as the coccolith-bearing cell volume and the cell volume as the coccolith-free cell volume.Therefore, the volume of the calcitic portion of the coccosphere (VCP) was estimated as follows: Volume of the calcitic portion of the coccosphere (VCP) = coccosphere volume -cell volume.
Coccolith volume and free coccolith concentrations were also determined for C. pelagicus (Fig. 1).

Scanning electron microscopy (SEM)
Samples were taken from each of the 48 incubation bottles.5-10 mL of sample were filtered by gravity on polycarbonate filters (0.2 µm pore size) and dried directly after filtration at 60 • C. Samples were sputtered with gold-palladium.SEM analysis was performed at the Earth Sciences department of the University of Milan with SEM Cambridge Stereoscan 360.For each sample, 50 specimens were digitally captured and subsequently analyzed with ImageJ software.All pictures were taken with the same magnification (5000x), and www.biogeosciences.net/14/3603/2017/Biogeosciences, 14, 3603-3613, 2017 the scale bar given on SEM pictures was used for calibration (Plate 1).

Coccolith dimensions and E. huxleyi malformation
We measured the length of the distal shield (DSL) and the width of the distal shield (DSW) manually using the public domain program Fiji distributed by ImageJ software (Schindelin et al., 2012).The distal shield area (DSA) was calculated, assuming an elliptical shape of the coccolith, as (Eq.2) Assuming an elliptical shape has been shown to yield nearidentical results to direct measurements of DSA in E. huxleyi (Bach et al., 2012).Malformations were determined for E. huxleyi since in the filters analyzed with SEM coccoliths were very abundant and allow a visual comparison of 100 individual coccoliths for every sample.We sorted the degree of malformation in several categories: our categories were used to describe the morphology of E. huxleyi as "normal", "malformed", "incomplete", and "incomplete and malformed" coccoliths (Langer et al., 2010;Langer and Bode, 2011; for reference images for the categories, see Fig. 2).We considered normal E. huxleyi coccolith with regular shape and well-formed distal shield elements forming a symmetric rim; malformed, malformed coccolith shape or malformed shape of individual elements; incomplete, coccolith with variations in its degree of completion; and malformed and incomplete, coccolith with malformed shape and variations in its degree of completion.

Statistics
Prior to statistical analyses, data were tested for normality and homogeneity of variances.To test the null hypothesis the average values of parameters from triplicate cultures were compared between treatments.Mean µ values, coccosphere and cell diameters, VCP and coccolith sizes of each treatment were compared to the control and among each other.
A one-way analysis of variance was used to determine statistical significance of the main effect of trace metals on the variables.A Tukey post-hoc test was used to identify the source of the main effect determined by ANOVA to assess whether differences in µ and sizes between trace metal treatments were statistically significant.Statistical treatments of data were performed using R software.Statistical significance was accepted for p < 0.05.

Growth rate
In the treatment with extreme trace metal concentrations up to 8 µmol L −1 , none of the four species tested survived the acclimation phase.In L, M and H treatments, however, E. huxleyi, G. oceanica, C. pelagicus and P. carterae all survived.However, the addition of trace metals decreased the growth rate of E. huxleyi, G. oceanica and C. pelagicus com- pared to the control treatment.E. huxleyi growth rate was 1.22 d −1 in the controls and 1.12, 1.16 and 1.10 d −1 in L, M and H trace metal concentration treatments respectively (Table 2; Fig. 3a).G. oceanica growth rate was 0.66 d −1 in the controls.In L, M and H the growth rate was significantly lower compared to the control at values of 0.58 d −1 in L, 0.60 d −1 in M and 0.58 d −1 in H (Table 2; Fig. 3b).C. pelagicus had an average growth rate in the control experiment of 0.56 d −1 .The growth rate was significantly lower in L, M and H compared to the control with values of 0.42 d −1 and 0.43 d −1 in M and H on average (Table 2; Fig. 3c).In contrast, P. carterae showed an increase in growth rate with trace metal addition compared to the control.In the control treatment, the growth rate was 0.52 d −1 and is significantly lower compared to L, M and H treatments with growth rates of 0.57, 0.56 and 0.57 d −1 respectively (Table 2; Fig. 3d).

Coccosphere and cell sizes and volume of the calcitic part of the coccosphere
In E. huxleyi, the mean coccosphere diameters were significantly lower in the L, M, and H treatments compared to the control (mean diameter 4.88 µm; Table 2, Fig. 4a).The VCP was reduced under all increased trace metal treatments compared to the control conditions (Table 2; Fig. 5a), with the lowest VCP recorded in the M treatment.In G. oceanica, the coccosphere diameters were largest in the control treatment (mean diameter 7.25 µm; Table 2; Fig. 4b).L, M and H coccosphere diameters were significantly smaller compared to the control.Specifically, L and M show similar values of 6.58 and 6.60 µm, respectively, while H shows the lowest coccosphere diameter of 6.14 µm (Table 2).Similarly, cell diameters were significantly larger in the control treatment (5.45 µm), intermediate in L and M (L = 5.18 µm; M = 5.19 µm), and smaller in H (4.74 µm).Furthermore, the coccosphere and cell diameters were significantly smaller in H compared to L and M. The VCP was significantly reduced under increased trace metal concentrations compared to control conditions (Fig. 5b), with similar coccosphere VCPs recorded in both L, M and H .
C. pelagicus coccosphere and cell diameters were significantly larger in the control (19.82 and 15.65 µm, respectively) compared to L (17.12 and 10.10 µm, respectively), M (17.05 and 10.46 µm, respectively) and H (16.85 and 10.38 µm, respectively; Table 2; Fig. 4c).However, a significant increase in coccosphere VCP was observed from 1760 in the control to 2102, 2036 and 1954 in the L, M and H treatments, respectively (Fig. 5c, Table 2).P. carterae showed a smaller coccosphere diameter in the control compared to the L, M and H treatments.The coccosphere diameter in L and H (L = 12.11 µm; H = 11.99 µm) is significantly bigger compared to the control (coccosphere diameter 11.70 µm).In M, the coccosphere has a mean diameter of M = 11.88 µm (Fig. 4d).On the other hand, the cell diameters were very similar among all treatments.The VCP was slightly lower in the control compared to the other three treatments (L, M and H ; Table 2.; Fig. 5d).

Coccolith size and C. pelagicus coccolith concentrations
E. huxleyi coccoliths were longer and wider in the control treatment compared to the other three treatments (Table 3).
Increasing trace metal content reduced coccolith length and width significantly (p < 0.05), and the high trace metal treatment showed the lowest distal shield length and distal shield width coccolith size.In G. oceanica coccoliths were longer and wider in the controls compared to the other three treatments (Table 3).However, only in H , coccolith were significantly smaller (p < 0.05) compared to the control treatment.C. pelagicus coccoliths in the L, M and H trace metal treatments were significantly smaller compared to the control replicates (Table 3).Furthermore, a higher number of free coccoliths was present in the trace metal treatments compared to the control replicates (Table 4), and free coccoliths progressively increased with increasing trace metal content.P. carterae coccoliths showed very similar sizes in all the four treatments (Table 3).

Emiliania huxleyi coccolith malformation
Scanning electron microscope analyses of E. huxleyi coccoliths showed changes in the proportion of malformed and incomplete coccoliths.Specifically, malformations and incomplete coccoliths of E. huxleyi increased in all trace metal treatments (L, M and H concentrations) by about 20-35 % compared to the control treatment (Fig. 6).

Growth rate
Whereas in the extreme trace metal treatment none of the species survived the acclimation phase, revealing that the conditions were apparently poisonous, in the L, M and H treatments the four test species responded in different ways to trace metal additions.Previous studies on E. huxleyi responses to trace metal enrichment resulting from volcanic ash showed no significant effects on growth rate for most ashes tested (Hoffmann et al., 2012).Only the addition of pumice, which released low concentration of trace metals, had a beneficial effect and increased E. huxleyi growth.
In one case, however, progressively increased ash content strongly suppressed the growth rate of E. huxleyi in the volcanic ash which contained the highest trace metal concentrations (e.g., Pb from 0.5 up to 2.6 nM L −1 ; Ni from 12 up to 60 nM L −1 ; Hoffmann et al., 2012).Vasconcelos et al. (2001) report a 10-20 % growth rate reduction of E. huxleyi with increasing Pb up to 0.25 µmol L −1 without additions of EDTA.
In our study, E. huxleyi growth decreased with increasing Pb, Zn, Ni and V concentrations, whereby the highest concentration of trace metals up to 0.8 µmol L −1 slowed down E. huxleyi growth by 10 % (Fig. 3a).While E. huxleyi is a species which has been studied intensively, much less work has been done on other coccolithophore species.For example, trace metal effects on G. oceanica have not been tested before, and, in our experiments, we observed a decrease of G. oceanica growth of 12 % under the highest trace metal concentrations (Fig. 3b).A similar negative response to elevated trace metal concentration was observed for C. pelagicus where the growth rate decreased by 31 % in each of the trace metal treatments compared to the control (Fig. 3c).The bigger growth rate reduction observed for C. pelagicus suggests a comparatively high sensitivity of this species to trace metals enrichment.A stepwise increase in trace metal concentration did not induce any progressive growth rate reduction attesting a strong sensitivity of both G. oceanica and C. pelagicus already at low trace metal concentrations.
P. carterae growth rate generally increased with trace metal concentration (Fig. 3d).This beneficial effect of high trace metal quantities (L, M and H ) on P. carterae growth rate might be due to the preferred habitat of this species in eutrophic lagoons and estuaries (Heimdal, 1993), where trace metal concentrations are generally much higher (Sunda and Hunstman, 2005) than in open waters.

Morphometrical analyses
The coccolithophore species tested evidenced a detrimental effect of trace metals on coccosphere, cell and coccolith sizes.Indeed, three species, E. huxleyi, G. oceanica and C. pelagicus, displayed significant size reductions when grown under anomalously high trace metal concentrations.However, the morphometrical responses are highly variable among species (Fig. 7): (1) E. huxleyi reduced its coccolith sizes under high trace metal concentration.This reduction can explain the coccosphere diameter decrease and the concomitant stable cell sizes (Fig. 7a).Additionally, an increase in the percentage of malformed and/or incomplete coccoliths was observed under high trace metal concentrations.
(2) Trace metal concentration also influenced G. oceanica coccosphere and cell sizes (Fig. 7b).Furthermore, an extra size reduction of both parameters, that goes along with coccolith size decrease, occurred at the highest trace metal concentration tested (Fig. 7c).This implies a particularly noxious effect of very high trace metal concentration on G. oceanica growth.(3) C. pelagicus coccosphere, cell diameter and coccolith sizes were negatively influenced by higher trace metal quantities.Increased trace metal induced a reduction of coccosphere, cell and coccolith sizes in C. pelagicus at all the concentrations tested.However, the volume of the calcitic portion of the coccosphere significantly increased under high trace metal concentrations.A plausible explanation is that the size decline of the cell goes hand in hand with an increase in the coccolith numbers that cover the cell (Fig. 7d).Indeed, progressively increased numbers of free/detached C. pelagicus coccoliths go together with a gradual increase in trace metal concentrations.(Table 4).This hints at a beneficial effect of trace metal on the number of C. pelagicus coccoliths produced per cell (Paasche, 1998;Müller et al., 2012).(4) On the contrary, P. carterae does not show any sensitivity to trace metal concentration since coccosphere, cell and coccolith sizes remain stable in all treatments tested.
Coccolithophore algae, therefore, respond differently to changes in trace metal concentrations.This species-specific sensitivity suggests a different degree of adaptation of the species tested.

Analogy and contrast with the fossil record
The trace metals tested here were chosen based on metal peaks identified in the Aptian OAE 1a (Erba et al., 2015) and latest Cenomanian OAE 2 (Snow et al., 2005).Zn and Pb are more volatile elements that are concentrated in magmatically degassed fluids, while Ni and V are found in higher concentrations in water-rock exchange reactions of typical steadystate hydrothermal vents (Rubin, 1997).Therefore, the composure of trace metal tested was intended to simulate OAE conditions.We emphasize that the coccolithophore species chosen for this experiment are linked to the Mesozoic family Biscutaceae based on the fossil record tracing their biocalcification back to ∼ 200 million years ago (Bown et al., 2004).When genomic data sets are considered for reconstruction of coccolithophore evolution, it appears that the selected Coccolithales order (C.pelagicus and P. carterae) diverged from the Isochrysidales order (E.huxleyi and G. oceanica) in the earliest Triassic (De Vargas et al., 2007) or even in the latest Permian (Liu et al., 2010), some 300 million years ago.
Considering that the species tested in this study have a long evolutionary history, it may be justified to conduct a comparison among fossil and living coccolithophore responses.Morphometric analyses of selected nannofossil taxa across Cretaceous OAEs in various geological settings revealed differential species-specific patterns: for instance, Biscutum constans, a cosmopolitan coccolithophore species of the Cretaceous ocean, evidenced size variations in times of environmental change.Specifically, coccolith dwarfism (sensu Erba et al., 2010) occurred at intervals characterized by high trace metal concentrations (Erba et al., 2010(Erba et al., , 2015;;Faucher et al., 2017).Conversely, Watznaueria barnesiae, a cosmopolitan species described as a r-selected opportunistic species (Hardas and Mutterlose, 2007) did not change in size across OAEs (Erba et al., 2010;Bornemann and Mutterlose, 2006;Lübke and Mutterlose, 2016;Faucher et al., 2017).Indeed, a more pronounced ellipticity, interpreted as evidence of malformation, was documented by Erba et al. (2010) during OAE 1a, but not during other times of global anoxia (Bornemann and Mutterlose, 2006;Faucher et al., 2017).The very low variability in W. barnesiae size indicates that this taxon was most adaptable and only marginally affected by the paleoenvironmental stress characterizing Cretaceous OAEs.Finally, D. rotatorius and Z. erectus, species with a meso-eutrophic preference, evidenced inconsistent size trends without a distinct relationship between size and metal peaks (Faucher et al., 2017).
Our laboratory experiments on living coccolithophore algae demonstrate that elevated trace metal concentrations affect coccolith size and/or weight similar to the fossil record.Moreover, as quantified in nannofossil assemblages, our results reveal species-specific responses.In fact, large differences were observed between species, and we could identify the most-tolerant (P.carterae), intermediate-tolerant (E.huxleyi, G. oceanica) and least-tolerant (C.pelagicus) taxa to increasing trace metal concentrations.Parallel changes among fossil and living coccolithophores suggest that trace metal concentrations have the potential to influence coccolith production and sizes.
We stress the fact that both the fossil record and the experimental results converge on a species-specific response of coccolithophores to metal availability.Consequently, the indiscriminate use of coccolith sizes as a proxy of trace metal concentration in seawater should be avoided.Instead, it is crucial to identify the species index/indices that better trace paleoenvironmental stress induced by high (selected) metal concentrations.

Conclusions
With this study, we demonstrated for the first time that a mixture of trace metals affected growth and morphology of coccolithophore species.A size reduction of the coccosphere and cell diameters has been observed in three of the analyzed species.Furthermore, we observed the production of dwarf coccoliths (sensu Erba et al., 2010) under high trace metal concentrations.Our data show a species-specific sensitivity of coccolithophores to trace metal concentration, allowing the recognition of the most-(P.carterae), intermediate-(E.huxleyi and G. oceanica) and least-tolerant (C.pelagicus) taxa.
The comparison of data on living coccolithophores and Mesozoic calcareous nannofossils shows strong similarities, suggesting that laboratory simulations of past extreme condiwww.biogeosciences.net/14/3603/2017/Biogeosciences, 14, 3603-3613, 2017 tions are viable when extant taxa are phylogenetically linked to extinct fossil species.Our study supports the hypothesis that anomalous trace metal conditions in the past oceans significantly contributed to the morphological coccolith changes during Cretaceous OAEs.Laboratory experiments on modern coccolithophore species remain the only means to quantitatively assess the individual or combined role of environmental parameters (e.g., trace metal availability) on coccolith secretion.Our results emphasize the need to consider species-specific differences where coccolith morphological features are used to reconstruct paleo-chemical conditions.

Figure 1 .
Figure 1.Example of Multisizer volume spectra.Black line: spectra of C. pelagicus population (coccolith spectrum and coccosphere spectrum); gray line: C. pelagicus coccolith-free cell after treatment with HCl.

Table 1 .
Trace metal concentrations in the growth medium of the different treatments.

Table 2 .
µ is the growth rate, coccosphere D (µm) is the coccosphere diameter, cell D is the cell diameter (µm) and VCP is the volume of the calcitic portion (d −1 ) of the coccosphere (µm 3 ).Significance was tested using an ANOVA and a Tukey post-hoc test (p < 0.05).Asterisks indicate significant difference from the control treatment.