Sex-associated variations in coral skeletal oxygen and carbon isotopic composition

Introduction Conclusions References

rate is known to influence the δ 18 O and δ 13 C isotope record to a lesser extent. Recent published data show differences in growth parameters between female and male coral; thus, skeletal δ 18 O and δ 13 C are hypothesized to be different in each sex. To assess this difference, this study describes changes in the skeletal δ 18 O and δ 13 C record of four female and six male Porites panamensis coral collected in Bahía de 10 La Paz, whose growth bands spanned 12 years. The isotopic data were compared to SST, precipitation, PAR, chlorophyll a, and skeletal growth parameters. Porites panamensis is a known gonochoric brooder whose growth parameters are different in females and males. Splitting the data by sexes explained 81 and 93 % of the differences of δ 18 O, and of δ 13 C, respectively, in the isotope record between colonies. Both iso- 15 tope records were different between sexes. δ 18 O was higher in female colonies than in male colonies, with a 0.31 ‰ difference; δ 13 C was lower in female colonies, with a 0.28 ‰ difference. A difference in the skeletal δ 18 O implies an error in SST estimates of ≈ 1.0 • C to ≈ 2.6 • C. The δ 18 O records showed a seasonal pattern that corresponded to SST, with low correlation coefficients (−0.45, −0.32), and gentle slopes (0.09 ‰ • C −1 , 20 0.10 ‰ • C −1 ) of the δ 18 O-SST relation. Seasonal variation in coral δ 18 O represents only 52.37 and 35.66 % of the SST cycle; 29.72 and 38.53 % can be attributed to δ 18 O variability in seawater. δ 13 C data did not correlate with any of the environmental variables; therefore, variations in skeletal δ 13 C appear to be driven mainly by metabolic effects. Our results support the hypothesis of a sex-associated difference in skeletal 25 δ 18 O and δ 13 C signal, and suggest that environmental conditions and coral growth parameters affect skeletal isotopic signal differently in each sex.
18796 tion, which affect salinity (Epstein et al., 1953). Depletion in carbonate δ 18 O occurs as temperature increases in inorganic and biogenic carbonates (Allison et al., 1996). In tropical and subtropical oceans, variations in salinity caused by evaporation, rainfall, or river run-off affect skeletal δ 18 O and need to be considered when establishing a skeletal δ 18 O-SST relationship (Cole and Fairbanks, 1990;Carriquiry et al., 1994 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | fixation of 12 CO 2 increases, which leads to an increase in 13 CO 2 in the coral carbon pool. Hence, coral skeletons formed during periods of high photosynthesis contain greater amounts of 13 C (Swart, 1983;McConnaughey, 1989;McConnaughey et al., 1997). During seasons with lower photosynthetic activity or when the photosynthesis to respiration ratio falls, coral skeletons would have lesser amounts of 12 C. Changes 5 in the photosynthesis-respiration ratio are influenced by photoperiods, photo-intensity, and temperature; where longer photoperiods and higher temperatures promote higher photosynthesis-respiration ratios (higher 13 C). If maximum solar radiation occurs during summer, skeletal δ 13 C will be inversely related to δ 18 O; if the maximum photoperiod occurs during colder seasons, δ 13 C and δ 18 O will be positively related (Swart et al., 1996b). Since zooplankton have generally low isotope levels, compared to coral skeletons and zooxanthelae, an increase in the heterotrophic activity of coral should reduce the δ 13 C of coral skeletons (Grottoli and Wellington, 1999). Felis et al. (1998), and Bernal and Carriquiry (2001) demonstrated that levels of coral skeletal δ 13 C decrease during upwellings, with high concentrations of zooplankton related to decreasing zoox-15 anthellae photosynthetic activity, and an increase in coral heterotrophic feeding (Cole et al., 1993;Quinn et al., 1993). The δ 18 O and δ 13 C in coral skeletons are depleted in 18 O and 13 C, in comparison to inorganic aragonite precipitated under isotope equilibrium (Weber and Woodhead, 1972;McConnaughey, 1989). This departure from equilibrium is referred to as "the 20 vital effect" and appears to be constant in the coral growth axis (Land et al., 1975;McConnaughey, 1989;Barnes and Lough, 1992;Barnes et al., 1995;Wellington et al., 1996). Isotope disequilibrium of coral skeletons results from coral precipitating their skeletons too quickly to attain isotope equilibrium (McConnaughey, 1989). Hence, all coral skeletons contain appreciable amounts of carbon and oxygen, which have not 25 been allowed to equilibrate with the ambient conditions and are isotopically depleted. Variations in coral skeletal growth parameters (skeletal density, extension, and calcification rate) are possible sources of deviation from oxygen and carbon isotope fractionation, which affect the external controls of the isotopes (Allison et al., 1996;Lough Introduction Barnes et al., 1995;Cohen and Hart, 1997). Skeletal growth parameters in coral have sex-based differences in some gonochoric species (Cabral-Tena et al., 2013;Carricart-Ganivet et al., 2013), so it is possible for the sex of a coral colony to be another cause of deviation in oxygen and carbon isotope fractionation. The influence of metabolic effects, such as reproduction, is another factor affecting the δ 18 O and δ 13 C 5 signal in skeletons (Kramer et al., 1993;Gagan et al., 1994;Barnes et al., 1995;Allison et al., 1996;Cohen and Hart, 1997;Lough et al., 1996;Swart et al., 1996b). The stony coral Porites panamensis has a wide distribution along the eastern tropical Pacific, from Mexico to Ecuador, and tolerates a wide range of environmental conditions, including low temperature and high-turbidity that are often stressful to other coral species (Halfar et al., 2005;Reyes-Bonilla et al., 2007). This coral has extension rates ranging from 0.4 to 1.2 cm yr −1 , along the coast of Mexico and Costa Rica (Guzmán and Cortés, 1989;Halfar et al., 2005;Cabral-Tena et al., 2013), where extension and calcification rates are different in males and females (Cabral-Tena et al., 2013). P. pana- 15 mensis is a gonochoric brooder with reproductive activity throughout the year (Glynn et al., 1994;Carpizo-Ituarte et al., 2011;Rodríguez-Troncoso et al., 2011). This study describes changes in the skeletal isotopic oxygen and carbon record of six male and four female P. panamensis coral, collected in Bahía de La Paz, with growth density banding covering 12 years. Oxygen and carbon isotope recording was used to 20 assess a possible sex-associated variation in the coral skeletal δ 18 O and δ 13 C signal related to differences in the "vital effect" of colonies between sexes. The isotopic record was compared to surface seawater temperature (SST), rainfall, photosynthetically active radiation (PAR), concentration of chlorophyll a, and skeletal growth data. Introduction  (Glynn et al., 1994;Carpizo-Ituarte et al., 2011;Rodriguez-Troncoso et al., 2011). The specimens were collected 5 at depths of 3-4 m. Divers used hammer and chisel to remove the colonies from the substrate. A fragment from each colony was fixed in Davison's solution for a histological examination and identification of sex (Howard and Smith, 1983). Coral fragments were first decalcified for 24 h in a solution containing 10 % HCl, 0.7 g EDTA, 0.008 g sodium potassium tartrate, and 0.14 g sodium tartrate in 1 L of dis-10 tilled water (Glynn et al., 1994). The tissue was then rinsed under running water until free of acid, and placed in 70 % ethanol until processed by conventional histological techniques (Humason, 1979). Transverse 8 µm sections were prepared with a rotator manual microtome, and stained with hematoxylin and eosin. After staining, the samples were studied under a compound microscope. The colonies were labeled female if 15 any planulae or oocytes were observed, regardless of their stage of development; the colonies were labeled male if any spermatocytes were observed in the slide section.

Growth parameters
From each colony, three slices (7-8 mm thick) were cut along the major growth axis. Slices were air-dried and X-rayed with a digital mammograph machine (Senographe 20 600 T, GE Healthcare, Little Chafont, UK). Images were made at 36 kVp for 980 mAs and 30 cm source-to-subject distance. X-ray films were digitized with a Kodak DirectView Classic CR System, at 75 dpi resolution. An aragonite step-wedge was included on each X-radiograph as a reference for calculating skeletal density. The stepwedge was built from eight blocks cut from a shell of Tridacna maxima; each block Introduction sity tracks were placed in the maximum growth axis in the digital X-radiography of each slice; density was measured using the ImageJ 1.44 image processing program (http://imagej.nih.gov/ij). A data series of absolute density vs. distance was generated and dated backwards for each slice, using photodensitometry (Carricart-Ganivet and Barnes, 2007). The coral year starts in the summer, with the highest SST at the sam-5 pling site (Hudson et al., 1976). The maximum and minimum density for each year (1993 through 2009) were identified in each density series.

Isotope analysis
After the skeletal growth analysis, one slice covering the most extensive chronological extension of each of the ten colonies was selected for isotope analysis. Continuous 10 samples of aragonite powder were collected along each coral's maximum growth axis using a drill with a 0.1 mm bit. Each sample was ∼ 1 mm apart. Aragonite powder was analyzed using an isotope ratio mass spectrometer (Delta V Plus, Thermo Scientific, Waltham, MA) with an automated system for carbon analysis in an acid bath (Finnigan Gas Bench II, Thermo Electron, Madison, WI). Each isotope 15 sample had < 0.05 ‰ error. Reference NBS-19 (International Atomic Energy Agency, Vienna, Austria) was used as the isotope standard. The seasonal pattern of δ 18 O was used to establish chronology. This is supported by the consistent pattern of annual density-band pairs described for Porites by Lough and Barnes (2000) ANCOVA test was used to assess the differences between slopes and the y intercept of lineal equations of δ 13 C vs. δ 18 O plots of the results of male and female data.

Environmental data
SST, PAR, and concentration of chlorophyll a data were obtained from the NOAA live access server (http://las.pfeg.noaa.gov/oceanWatch/oceanwatch.php), and in situ ther-5 mograph temperature data (2003)(2004)(2005)(2006)(2007)  relationships between environmental data and isotope data of both sexes. Regime shift index for environmental and isotope data were calculated with the Sequential Regime Shift Detection Software (Rodionov, 2004).

Skeletal growth 20
All specimens were collected in March, a period of low SST in Bahía de La Paz. All Xradiographs had a low-density annual growth band in the apex of the slice. This means that P. panamensis form a low-density band in winter. Annual growth bands in each colony were dated and the sampling resolution for isotope analysis was determined. Introduction The average yearly extension rate was 1.05 ± 0.04 cm yr −1 for female colonies, and 1.27 ± 0.04 cm yr −1 for male colonies. The average skeletal density was 0.94 ± 0.01 g cm −3 for females, and 0.95 ± 0.01 g cm −3 for males. The average calcification rate was 0.97 ± 0.04 g cm −2 yr −1 for females, and 1.24 ± 0.03 g cm −2 yr −1 for males.

Skeletal isotope composition and environmental data 5
The δ 18 O records of female and male coral colonies show a seasonal pattern ( Fig. 1) that was strongly correlated between sexes (r = 0.45, p > 0.000 001). δ 18 O in female colonies, was higher than in male colonies (Fig. 1). The overall average δ 18 O in female colonies was −2.89 ± 0.33, and −3.20 ± 0.37 ‰ in male colonies (Table 1) δ 13 C showed a cyclic pattern in female and male colonies (Fig. 2), that was correlated between both genders (r = 0.19, p = 0.005). The skeletal δ 13 C of female colonies was lower than the skeletal δ 13 C of male colonies (Fig. 2). The overall average of δ 13 C in female colonies was −1.66 ± 0.38, and −1.38 ± 0.37 ‰ in male colonies (Table 1). The overall average of δ 13 C in females is significantly lower than in males (t 498 = −8.01, 20 p > 0.00 001). No regime shift was found in the δ 13 C data of either sex.
The δ 18 O skeletal data series corresponds to the SST (Fig. 1). Table 2 shows correlation coefficients between the δ 18 O isotope data of coral colonies and environmental variables. The correlation coefficient between the isotope average time series data and SST was −0.45 (p = 0.00 003) for female colonies, and −0.32 (p = 0.0005) for 25 male colonies; the r to Z transformation showed that both correlation coefficients are equally strong (Z = −1469; p = 0.07). No significant correlation was found between the BGD 12, 2015 Sex-associated variations in coral skeletal oxygen and carbon isotopic composition The departure from isotope equilibrium of our samples was estimated with the equations by Grossman and Ku (1986), for δ 18 O, and Romanek et al. (1992)

Skeletal isotopic composition and skeletal growth
The analysis showed that high density bands are depleted in 18 O and 13 C, which are deposited during summer; low density bands are enriched in 18 O and 13 C, which are deposited during winter. In female colonies, a strong negative correlation between the mean annual coral δ 18 O and skeletal density was found (Table 4; r = −0.78, p = 0.001) 5 (Table 4). This suggests that denser skeletons are more depleted in δ 18 O, compared to less dense skeletons, and no significant correlation was found between δ 18 O and other skeletal growth parameters in female colonies; no significant correlations between mean annual coral δ 13 C and any growth parameters were found. In male colonies, there was a strong negative correlation between mean annual coral δ 18 O and the lin-10 ear extension and calcification rates (Table 4; r = −0.50 and −0.44, p = 0.045 and 0.0008). This suggests that faster growing and calcifying colonies are more depleted in δ 18 O. No significant correlation was found between δ 18 O and skeletal density in male colonies; no significant correlation between any coral growth parameter and mean annual coral δ 13 C was found.  (Gagan et al., 1994), Costa Rica (Carriquiry, 1994), Panama (Wellington and Dunbar, 1995), and the Galapagos Archipelago (McConnaughey, 1989 (Linsley et al., 1999), Fiji (Le Bec et al., 2000), and Guam (Asami et al., 2004). Asami et al. (2004) Swart et al. (1996b) suggest that this means that the maximum photoperiod in Bahía de La Paz occurs during winter (high δ 18 O = low SST, high δ 13 C = high photosynthesis). When the SST peaks in the summer and surface seawater generally becomes depleted of nutrients, zooxanthellae disperse (Hoegh-Guldberg, 1999; Barton and Casey, 2005). Hence, photosynthesis might be 20 less intense until the nutrient-rich waters of winter promote the growth of zooxanthellae and restore photosynthesis intensity (Jokiel, 2004;Franklin et al., 2006). Skeletal δ 13 C (Fig. 2) was higher in both genders, between November and January (lowest SST and PAR), and lower from June through August (highest SST and PAR), suggesting a positive relationship between δ 13 C and photosynthesis, and a dominant 25 role of light-induced photosynthesis on seasonal changes of δ 13 C in coral. Still, the δ 13 C-PAR regressions and correlations were not significant, meaning that photosynthesis was not stimulated or inhibited by light, and remained near its maximum efficiency during the whole year, according to Sun et al. (2008) China. They suggest that other factors may be affecting photosynthesis in addition to light, such as abundance of dissolved nutrients. High concentrations of chlorophyll a occurred during periods of relative enrichment of 13 C in the coral skeleton (November through January), when fixation by algae of the isotopically lighter carbon enriches δ 13 C in coral skeletons (Allison et al., 1996); however, the correlations of skeletal δ 13 C 5 and chlorophyll a were not significant in any case. Trends in coral skeletal δ 13 C reflect seasonal variations in metabolic effects, that is, modifications of photosynthesis to respiration ratios in the δ 13 C pool of coral. Higher coral respiration reduces coral δ 13 C (McConnaughey, 1989;McConnaughey et al., 1997). Respiration normally increases with temperature and lowers 13 C in coral skele-10 tons, which is reflected in our results, high SST = low δ 13 C. No other environmental variables considered in this work explained this pattern in coral δ 13 C, driven mainly by metabolic effects as described by Sun et al. (2008) in Porites coral of the South China Sea. We found a negative correlation (r = −0.78, p = 0.001) between δ 18 O and the skele- 15 tal density in female colonies, i.e. more dense skeletons are depleted in δ 18 O. This is not consistent with studies that have observed that coral skeletal high-density bands are enriched in 18 O (Klein et al., 1992;Al-Rousand, 2007). This may be due to a difference in timing of skeletal density bands in different Porites coral species, as described by Lough and Barnes (2000). In male coral, we found a negative correlation 20 between the δ 18 O and linear extension and calcification rates (r = −0.50, p = 0.045 and r = −0.44, p = 0.0008), meaning that the faster a colony grows and calcifies, the more it is depleted in δ 18 O. This is consistent with the observations of other authors of Porites spp. coral (McConnaughey, 1989;Felis et al., 2003). In Porites corals, SST is a dominating control of variations in growth parameters and of δ 18 O; the skeletal 25 extension and calcification rate increases with SST, while skeletal density decreases (Lough and Barnes, 2000), so the growth parameters of both sexes and δ 18 O behave as expected; that is, an increase in SST = a decrease in density = δ 18 O enrichment in females, and an increase in SST = an increase in extension and calcification rate = BGD 12,2015 Sex-associated variations in coral skeletal oxygen and carbon isotopic composition and skeletal growth parameters in either males or females, meaning that regardless of the skeletal extension rate, density or calcification rate, P. panamensis deposited a widely varying δ 13 C, as reported by Allison et al. (1996) in Porites coral from South Thailand, and by Swart et al. (1996b) in Montastrea annularis in Florida, USA.

5
General consensus states that all coral skeletons contain appreciable amounts of carbon and oxygen in isotopic disequilibrium, and are depleted in 18 O and 13 C because of kinetic variations due to differences in coral growth. Larger isotopic disequilibrium occurs when coral grows faster (Land et al., 1975;McConnaughey, 1989;Aharon, 1991). McConnaughey (1989)  the isotope record between coral growing in the same environment are attributed to differences in the "Vital effect" of each colony (Linsley et al., 1999;Felis et al., 2003). several Porites species (not detailed by the authors), in three sites in the northern part of the Gulf of Aqaba. None of the mentioned works considered the sex of the colony as a factor explaining differences in the "Vital effect" of coral colonies. If we pool the isotopic data of both sexes together, the differences between our isotopic records are BGD 12,2015 Sex-associated variations in coral skeletal oxygen and carbon isotopic composition  Felis et al., 2003). If we split our data by sex, the differences in the isotopic records drop to 0.07 ‰ in the δ 18 O, and to 0.02 ‰ in the δ 13 C. In our data, the sex of the colony explains 81 % (δ 18 O) and 93 % (δ 13 C) of the differences in the "Vital effect" of coral colonies. Thus, the main source of differences in the isotope record is attributed to differences in the "Vital effect" associated to colony sex, for which we offer two explanations; a simple one, and a complex one: Energy expenditure during the formation of gametes causes differences in the formation of skeletal density bands, and carbon isotopic depletion in coral skeletons (Kramer et al., 1993;Gagan et al., 1994). Cabral-Tena et al. (2013), and Carricart-Ganivet 10 et al. (2013) found sex-dependent effects on the growth parameters and timing of density band formation of coral, related to metabolic effects. We found that P. panamensis female colonies grew slower in comparison to male colonies (1.05 ± 0.04 cm yr −1 vs. 1.27 ± 0.04 cm yr −1 ). Faster growing coral are more depleted in 18 O and more enriched in 13 C, relative to slower-growing coral (McConnaughey, 1989;Felis et al., 2003), this 15 may be the origin of the isotope data difference between sexes (higher δ 18 O and lower δ 13 C in females), so a simplistic approach might be that since the growth rates are different between sexes, the "Vital effect" will also be different between sexes, thus explaining the differences we found in δ 18 O and δ 13 C between sexes. A more complex explanation for this sex-associated difference in coral isotopic data 20 could result from the role Ca-ATPase (enzyme strongly associated with coral calcification) activity has in the mechanism of the "Vital effect". Adkins et al. (2003), and Rollion- Bard et al. (2003) found that the Ca-ATPase activity in deep sea and symbiotic coral establishes a pH gradient between the coral cell wall and the extracellular calcifying fluid (ECF). The pH gradient (more basic in the ECF) promotes a passive CO 2 25 flux into the ECF and controls the mixing of carbon with isotopically heavier signature from the seawater-dissolved inorganic carbon, thus, the intense activity of Ca-ATPase will result in a carbon heavier skeleton. Oxygen isotopes also respond to the pH of the ECF, proportions of the dissolved carbonate species are pH dependent. At low pH  . This complex mechanism of the origin of the "Vital effect" might explain why we found a sex-associated variation in coral skeletal oxygen and carbon 15 isotopic composition of Porites panamensis. Kramer et al. (1993), and Gagan et al. (1994) suggested that energy expenditure during the formation of gametes may cause differences in the isotopic depletion in coral skeletons; Kramer et al. (1993) observed depletions in isotope data during reproductive seasons, regardless of the sex of the coral, and found minimum δ 13 C values 20 in skeletons of Oribicella faveolata during spawning seasons (summer), although this phenomenon was also observed in other coral species which produce gametes the whole year (O. faveolata has only one reproductive event per year). The results obtained by Kramer et al. (1993) were inconclusive, but suggested a lag effect of isotope signal, associated with the initiation and duration of the reproductive cycle. It is possible 25 that the sex-associated variation we found in isotope data is due to the reproductive strategy of P. panamensis. P. panamensis is a gonochoric brooding species with reproductive and larval release events through the whole year in the Pacific coast of Mexico Rodriguez-Troncoso et al., 2011) duction in gonochoric spawners are lower than in gonochoric brooding species where energy is required not only for egg production, but also for larval development (Szmant, 1986). This implies that there should be sex-associated variations in the coral skeletal isotope data of other gonochoric brooding coral, as some massive Porites (which can be spawners or brooders; Glynn et al., 1994;Baird et al., 2009).

5
Considering δ 18 O of coral skeletons is used to estimate SST in different sites and conditions, the next part of the discussion seeks to exemplify what would a difference in δ 18 O between sexes would represent in terms of errors in SST estimation. Using the widely accepted paleotemperature equations for calcite (Epstein et al., 1953) and aragonite (Grossman and Ku, 1986), a ∼ 0.31 ‰ difference between sexes would represent 10 an error in SST estimates of ∼  , 1989), and ∼ 1.47 • C in SST estimates, for the commonly admitted paleotemperature calibration in coral (0.21 ‰ • C −1 ). δ 13 C of coral skeletons has been used as a proxy for the photosynthetic activity of zooxanthellae (mainly driven by light). Until now, no general rule applies to how much δ 13 C means how much radiance (like the dependence of δ 18 O to SST resulting 20 in paleotemperature equations), but a difference of ∼ 0.28 ‰ in coral δ 13 C between sexes should be taken into account for this kind of applications, since it may influence the descriptions of the variability in δ 13 C of coral skeletons. δ 13 C of coral skeletons is also used to correct the δ 18 O data when estimating the SST at which coral grew, by using the regression line equations obtained from the δ 13 C vs. δ 18 O plots (Smith et al., 25 2000). When we compared the regression line equations obtained from the δ 13 C vs. δ 18 O plots of both sexes, the ANCOVA showed that both the slope (F 498 = 9.619, p = 0.002) and the y intercept (F 498 = 222.5, p < 0.00 001) are different between equations (Fig. 4.). Also, Fisher's r to z transformation (z = −2.34, p = 0.01) showed that the