Expected seasonal δ18Oostracode range from water isotope composition
The maximum range of the variation of the expected equilibrium calcites is determined by Tmin-δ18Omax (maximum) and Tmax- δ18Omin (minimum). Mean upper (positive) values are described by Tmax- δ18Omax. At almost all localities the seasonal variation of estimated equilibrium calcites reflects the variation of δ18Ometeoric (see Fig. 3). For instance, the expected equilibrium δ18O values of Florida display largest variations during May/June and October similar to δ18O values of precipitation while during the beginning of the year both calcites and precipitation reveal distinctly smaller variation ranges (Fig. 10).
The monthly variation ranges of Mexican expected equilibrium calcites are largest during winter (November to April) with the most positive δ18Ocalcites. June and October provide not only the smallest variation ranges compared to the other months, but also relatively negative values. August is the summer month with the largest range in corrected δ18Ocalcite (Fig. 11). For two localities18O values of water derived from the literature enabled the calculation of different corrected δ18Ocalcites. The resulting ranges show partly strong deviations. Especially, for MX-BC the δ18Ocalcites corrected show much more negative values with our own water value than those corrected with literature data. The other locality MX-PuL shows a larger range of coincidence for all three estimated δ18Ocalcites.
The estimated equilibrium calcites of Panama provide generally large variation ranges. Largest variation ranges occur between April and October. Smallest ranges prevail in February and March. Summer months (April to November) generally provide negative values (entire variation ranges ≤ 0‰) compared to winter months during which the variation ranges become more positive (Fig. 12). Variation ranges of equilibrium calcites of Colombia vary throughout the year and provide generally negative values (≤ 0‰). Periods with relatively small variation ranges are January to March, and August and September. Large variation ranges occur from April to July, and October to December (Fig. 13). Estimated equilibrium calcites of Brazil present differing variation ranges throughout the year. The variation is solely determined by the range of negative values. Generally large ranges occur from December to May. Largest ranges and most negative values emerge in December. From June to November are the ranges relatively small with minima in August and September (Fig. 14).
The deviation between uncorrected and corrected equilibrium calcites differs between the regions. Corrected δ18O values of equilibrium calcites of Florida display generally much wider and more positive variation ranges than the uncorrected ones. Corrected calcites in Mexico are generally more positive than the uncorrected values. In Panama the corrected values are almost in all cases within the range of the uncorrected ones and exceed them (negatively) only in January. In Colombia the corrected calcites exhibit similar but generally less negative values than the uncorrected calcites. The corrected equilibrium calcites of Brazil differ slightly but are generally relatively similar to the uncorrected calcites. Only the two localities BR-MN-5 and BR-PTO-4 show a more negative range than the other (corrected and uncorrected) values.
Apparent vs. expected oxygen isotope variation of ostracode valves
Florida
Generally, ostracode values vary within the ranges of the expected equilibrium calcites during winter months (November/December to March/April) and become generally more positive than equilibrium calcites during summer months (May to October). The majority of samples shows an offset of ≥ 0.5–1.5‰ between mean δ18Oostracodes and mean δ18Ocalcites during January to April and November to December (Fig. 15). Two samples (FL-CAL-4, FL-CAL-3) provide positive offsets of ~ 1‰ only in January. For all other months the offsets are far beyond the required 1‰ value. The offsets of the last group (FL-EG-3, FL-LX-2, Fl-LX-5, FL-CAL-2) match the required range of ≥ + 1‰ only during July and August. During winter (December to April) they show negative offsets to mean δ18Ocalcites. In November offsets are positive, but too low (0 to + 0.5‰).
Mexico
Ostracode δ18O fall below or are at the lower (negative) margin of the equilibrium calcites during winter months (December to April). A higher similarity between δ18Oostracodes and δ18Ocalcites occurs during summer months from May to November. Considering the differences between mean δ18Oostracodes and mean δ18Ocalcites (Fig. 15) it emerges that except of two samples (MX-LG, MX-Pul-3) offsets ranging between + 0.5 and + 1.5‰ are achieved in May, July, August, and October. Offsets of MX-LG are too positive and lie in the range of + 0.5- +1.5‰ only during February, March, and December. Contrary, MX-Pul-3 provide only negative offsets to equilibrium calcites.
Panama
Ostracode δ18O values are within the ranges of the estimated calcites except for February and March where the δ18Oostracode is below or at the upper margin of the calcites. Offsets between mean δ18Oostracodes and mean δ18Ocalcite correspond only during November to the required range of + 0.5- +1.5‰ (Fig. 15).
Colombia
Ostracode δ18O values are within the ranges throughout the year and exhibit generally a trend to the more positive values relative to the equilibrium calcites. Offsets between mean δ18Oostracodes and mean δ18Ocalcite are far above the ~ 1‰ criterion from April to December with the strongest deviations in May and November. During January to March ostracode values fall within or are very close to the estimated calcites (Fig. 15).
Brazil
The ostracode values mostly coincide with the ranges of the equilibrium calcites from June to December. During the first half of the year, the ostracode values tend to be more positive relative to the estimated equilibrium calcites (Fig. 14). Differences between mean δ18Oostracodes and mean δ18Ocalcite are + 0.5 - +1.5‰ for most localities in May, July, August, and October. Only one locality (BR-MN-3) displays offsets exceeding realistic values of up to + 1.5‰ by far (Fig. 15).
Site specific hydrochemistry and ostracode formation environment
The study area provides quite heterogenous hydrochemical facies (Fig. 4; Supplementary Fig. 1) according to different background geology, and climatic and hydrological conditions. Important contributing process represents mixing of seawater and freshwater which occurs in Florida and Brazil and to a lesser degree also in Mexico (cf. Petrini et al., 2014; Long et al., 2018).
Although saturation of calcite is affected by short-term variations (e.g., seasonal and even diurnal changes; Liu et al., 2007) the SIcalcite values probably roughly reflect the regional background geology differentiating the carbonate platforms of Florida and Mexico from the remaining regions which are mainly composed by silicate rocks and few or no carbonates. However, it is known that the state of the calcite saturation of a solution depends on a variety of factors including not only the calcium and bicarbonate concentrations, temperature, alkalinity, and pH but also indirectly by carbon dioxide pressure, residence time within soil and groundwater areas, as well as photosynthesis rates (Neal, 2002). The reason why Florida and Mexico provide differences in their SIcalcite values despite their similarities in lithological and climatic characteristics is attributed to striking hydrological and geochemical differences. In Florida, the chemical character of water changes systematically downgradient, owing to solution of minerals of the aquifer and corresponding increases in total dissolved solids In Yucatán, no downgradient change exists, and dominant processes controlling the chemical character of the water are dissolution of minerals and simple mixing of the fresh water and the body of salt water that underlies the peninsula at shallow depth (Back and Hanshaw, 1970). The higher amount of salt water in Yucatán provides a higher portion of dissolved solids which reduces the activity of each ion in solution and thus can dissolve more limestone than can freshwater (as in Florida) of much lower dissolved solids concentration.
High correlation coefficient (r2: 0.70) of SIcalcite and pH indicates that SIcalcite is largely controlled by pH (Fig. 4C). With the low SIcalcite values of some solutions (e.g., Fig. 4C) the question arises how ostracodes could have precipitated their valves in solutions undersaturated in calcite. It is known that calcite saturation can be highly variable with diurnal fluctuations between saturation and oversaturation in rivers (Neal et al., 2002). Also, changes from undersaturated to saturated conditions are reported on long-term (decades) and seasonal time scales (Ulloa-Cedamanos et al., 2020). It cannot be ruled out, however, that the water bodies may have changed from undersaturation to saturation within weeks – the time frame assumed between the precipitation of the valves and sampling. Actually, biomineralization by ostracodes is not very well understood. It is known that prior to molting the ostracode accumulates huge amounts of carbonatic and phosphate granules in the outer epidermal cells which form calcitic crystals via an intermediate step. Therefore, it is speculated that ostracodes may store calcium internally what contrasts to the common assumption that the calcium secreted by the animal originates directly from the ambient water (Keatings et al., 2002; Keyser and Walter, 2004). This was, however, rejected since carbon isotope fractionation is not similarly affected by vital effects as oxygen isotopes (Keatings et al., 2002).
Lakes exhibit a broad range of δ13CDIC values, and the variation between lakes is generally larger than seasonal variations within a lake (Bade et al., 2004). It is assumed that biological productivity within a lake is a major control on δ13CDIC signatures (e.g., Leng and Marshall, 2004). However, additional geochemical factors such as the dissolution of limestone in karst areas or different pathways of plants for fixing CO2 (i.e., C3 and C4 plants), or more specifically, their interaction with lake metabolism contribute largely to the δ13CDIC signature (Bade et al., 2004). Although our dataset comprises samples from a variety of water bodies of different sizes, hydrological and productivity characteristics, etc. our analyses displayed a relatively strong (positive) relationship between pH and δ13CDIC. This indicates that fractionation between CO2 and carbonate species at different pH values contributes a certain amount to the variation of δ13CDIC in the solutions. Through carbonate chemistry, the correlation between pH and δ13CDIC is probably not independent from the pattern observed with SIcalcite. This implies that the regional geology (i.e., calcareous vs. siliceous dominance) is responsible for the major picture. However, DIC in aquatic ecosystems is maintained by several mechanisms that include, but are not limited to, atmospheric invasion, terrestrial runoff, stream, and underground inputs, and oxidation of organic matter from the water column and sediments that all can have locally very different influence. Lacustrine systems often receive DIC inputs from more than one source, and the resulting isotope ratios typically reflect the `weighted average´ of these sources (Gu et al., 2004). Therefore, local conditions probably represent a stronger control on δ13CDIC in the solutions than the overall variation in pH and SIcalcite.
It has been observed that in carbonate springs calcite saturation leads to significant δ13CDIC increase (Abongwa and Atekwana, 2015) due to CO2 degassing which preferentially releases 12CO2 (Liu et al., 2003). Concluding from SIcalcite ≥0 this effect can be assumed to a part of the solutions. Lake δ13CDIC variations differ widely in carbonate and carbonate-deficient (or hard and soft water) systems. In carbonate systems the seasonal fluctuations of δ13CDIC display more positive values and distinctly lower amplitudes with max. ~6‰ (Myrbo and Shapely, 2006) compared to soft water lakes in which seasonal δ13CDIC values can span a wide range of ~ 10‰ (Herczeg, 1987). Nonetheless, our data probably show no consistent picture or, more specifically, are hard to interpret due to large differences in geological and hydrological conditions, etc.
In hydrologically closed water bodies, the correlation between [Mg/Ca] and salinity (i.e., conductivity) arises as a result of CaCO3 mineral precipitation due to the differences in mineral saturation, which removes Ca2+ from the lake water and leads to an increase of the [Mg/Ca] ratio (Fukushi and Matsumiyam, 2018). The independence of Mg/Ca and conductivity of the solutions highlights thus the different and rather open hydrologies of the investigated water bodies. Also, many localities are situated close to the sea and provide strong differences in the geological background. Thus, the varying combinations of freshwater mixing with sea water and/or the influence of groundwater draining, e.g., dolomitic rocks probably controls the differences of [Mg/Ca] in the study area (e.g., Surge and Lohmann, 2002).
Parameters controlling the isotopic variability of calcite valves for Cytheridella
The oxygen isotope composition of ostracod valves is a function of the temperature and the isotopic composition of the lake water in which the biominerals were secreted, modified by a vital offset, a systematic, species-specific deviation from the 18O value of a theoretical calcite precipitated under equilibrium conditions (von Grafenstein et al., 1999, Decrouy, 2012). The oxygen isotopic composition of the lake water is determined by the atmospheric component of the global hydrological cycle (e.g., Rozanski et al., 1993). Lake waters reflect the mean oxygen isotopic composition of catchment precipitation, which is primarily a function of latitude, modified by orography and continentality (Schwalb, 2003). Variations in δ18O of precipitation are mainly determined by temperature in middle and high latitudes (Jouzel et al., 2013), while precipitation amount is the main determining factor in the tropics (Lachniet and Patterson, 2006; 2009; Landais et al., 2010). The combination of amount and temperature effects is common in subtropical latitudes (Bowen, 2008). The isotopic composition of lake water depends on the isotopic composition of the precipitation in the catchment (as described above), and the processes that affect the isotopic composition in the lake such as evaporation that changes with relative humidity, temperature, wind stress, relation of lake area vs. volume, and residence time (Schwalb, 2003).
With a few exceptions, ostracode δ18O/δ13C correspond very well to water δ18O/ δ13CDIC (Fig. 8). This highlights the potential of ostracodes as reliable proxies for water chemistry, as shown by previous studies (e.g., von Grafenstein, 1999; Schwalb, 2003; Marco-Barba et al., 2012). Marco-Barba et al. (2012) observed that ostracode oxygen fractionation decreases with pH explained by the shrinking pool of HCO3− which is used by ostracodes for calcification. Although our dataset also comprises a pH range (5.3 to 8.6) in which HCO3− is the dominant inorganic carbon species, our data do not indicate that values of equilibrium calcites have changed in response to changes of pH. The differences between lake water δ13CDIC and δ13Costracode (i.e., Δ13C) exceed known fractionation factors at 25°C that lie in the range of 0.35 to 2.3‰ (Mucci and Morse, 1990) and between 0.9 and 1 (Romanek et al., 1992) by far. While Marco-Barba et al. (2012) found generally negative offsets in the range of 0 to -4‰ our data show no systematics. Especially, removing the Brazilian samples with the unusual δ13C values leads to uncorrelated relationships with temperature, pH, and SIcalcite (Fig. 6E-H). One reason for this might be attributed, again, to the temporal lag between calcification and sampling that leads to comparison with a `wrong´ δ13CDIC value. Observed negative offsets are suggested to be caused by infaunal molting (Decrouy et al., 2011; Marco-Barba et al., 2012). Since most of the values are positive this cannot be attributed to Cytheridella.
The majority of oxygen values of Cytheridella valves display relatively small ranges (~ 2‰) which is typical for small to small-medium open lakes according to the conceptual model by Leng and Marshall (2004). Larger ranges in the δ18O values indicate either seasonal changes or closed basin lakes. Sites with flowing conditions such as the Floridian river sites may therefore reflect seasonal changes. This is supported by measurements on Cyprideis valves sampled in December and July which show a similar range in oxygen isotopes (Meyer et al., 2017a).
The largest range of > 5‰ is displayed by a Mexican cenote (MX-SiNo). Cenotes can be lotic and lentic and are mainly fed by groundwater (Schmitter-Soto et al., 2002). In closed-basin tropical to subtropical lakes with a seasonally dry climate, the δ18O of lake water is controlled mainly by the ratio of evaporation to precipitation (Curtis et al., 1996 and cited references therein). The relatively great difference in oxygen isotope values of groundwater and rainwater (-3.91‰, Curtis et al., 1996; -4 to -5‰, Wassenaar et al., 2009) to the lake water (0.10‰, Table 2) indicates that the cenote MX-SiNo loses a significant fraction of its hydrological budget to evaporation.
Ground and river waters, in general, have typically low δ13CDIC values between − 10‰ and − 15‰ (VPDB; Leng and Marshall, 2004). Higher δ13CDIC (around − 8‰ to -12‰ VPDB) in groundwaters may occur in karstic regions where dissolution of catchment limestones is more pronounced than solid CO2 uptake (e.g., Emblanch et al., 2003; Marfia et al., 2004; Han et al., 2010). In hydrologically closed lakes, carbonates are often high δ13C covariant with δ18O likely reflecting different degrees of equilibration with atmospheric CO2 and preferential evaporative loss of the 16O (Leng and Marshall, 2004). In our dataset there are two localities in Yucatán (MX-LG, MX-Sil) showing both high δ13C and δ18O probably indicating a long exposure of the lake water to surface and exchange with the atmosphere. As discussed in Meyer et al. (2017a), δ13C clearly differs between marsh and river sites which is interpreted as result from high biological activity and differences in the residence time of the water. Photosynthetic activity will decrease 12C from the water, due to the preferential uptake by aquatic plants, while respiration has the opposite effect (e.g., Leng and Marshall, 2004). Marshes are characterized by low water levels, stagnant water, and dense aquatic vegetation. Residence time of the water in those systems is long (Childers, 2006) enabling the accumulation and consumption of organic matter which will contribute to increases of δ13C. In rivers and canals, large-scale processes such as the input and mixing of inorganic carbon from different sources (groundwater, tributaries, etc.) in the catchment is more important than local small-scale processes. Differences in δ13CDIC between different rivers within a region reflect different residence times of water in the tributaries (Atekwana and Krishnamurthy, 1998).
Lowest δ13C values from Colombian sites (CO-ET, CO-AM) correspond well with the Amazon River in the transition between Andean upland and lowland (-14.5‰ ±1.7; Quay et al., 1992). The geology of Colombia provides virtually no carbonates (cf. Parra et al., 2009). Streams draining silicate rocks have low δ13CDIC values possibly due to their relatively low pH values resulting in a relatively high content in dissolved CO2 vs. bicarbonate ions (Hélie et al., 2002). Comparable conditions prevail in Panama contributing to relatively low δ13C values. The low values may be due to enhanced supplies of 13C-depleted DIC from soils and groundwaters from watersheds and/or high oxidation rates of dissolved or particulate 13C-depleted organic carbon (cf. Hélie et al., 2002; Leng and Marshall, 2004).
The very low δ13C values of some Brazilian sites correspond to groundwater sites in which no dissolution of carbonates takes place (Sracek and Hirata, 2002). Although variation with values around − 16 to -21‰ is considerably, this does not provide hints of the much more positive ostracode values from one site in this water body. The most logical explanation for this phenomenon is the local mixing with 13C enriched sources that may vary through time explaining the relatively large range at the site. Pedrozo and Rocha (2007) report a gradient in nutrient and ion concentration, conductivity and other parameters, even within lakes, reflecting different sewage inputs that might be also the case for Lagoa Itapeva. The process leading to the distinctly higher δ13C values could include bacterial methanogenesis that leads to a strong 13C fractionation between CH4 and CO2, because the metabolic pathway of methanogenic bacteria favors the light isotopes. Carbonates formed under this condition have markedly positive δ13C values (Talbot and Kelts, 1990; Schwalb, 2003). The question is if this may occur spatially and temporally restricted in a water body as indicated here.
The within sample variability of ostracodes varies between all sites. Sites with a relatively high within-sample variability (standard deviation ≥ 1‰) occur mostly in Florida. For living populations, some variability can undoubtedly be attributed to temporal fluctuations in water temperature or composition during the ostracode´s life, or small-scale spatial differences, which will mean that the individual valves did not calcify under truly uniform conditions (Holmes, 2008). Contrary to the other regions, most of the sample localities in Florida are rivers or channels whose isotopic composition is usually primarily controlled by precipitation (e.g., Henderson and Shuman, 2010; Price and Swart, 2008). Temporal and spatial fluctuations in river water can result from tributary/lake water mixing, damming regulation and temperature (Wu et al., 2017). In lakes, precipitation water is mixed with a great volume of older evaporated water, buffering the δ18O variations of the water (Leng and Marshall, 2004), and can explain the lower variation of δ18Oostracodes in lakes. An additional contribution to the within-samples variability derives from the life cycle of the ostracodes. Although we used only specimens which were alive at time of sampling which should ensure that the calcification time was closed to the time of sampling, it cannot be excluded that the adult life span of Cytheridella is be long enough to assemble (living) specimens from more than one molting periods in a sample.
As expected, the apparent oxygen isotope fractionation of ostracode―H2O shows a clear correlation with temperature with lesser fractionation at higher temperatures (Fig. 6A). This is in agreement with previous studies which observed that ostracode oxygen isotopes fractionation increases with temperature (e.g., Li and Liu, 2009). Interestingly, at temperatures between 30°C and 37°C fractionation exaggerates equilibrium much stronger than the anticipated vital effect of ~ + 1‰. This could be due to a kinetic effect explained by an amorphous precursor pathway of the precipitated ostracode calcite (cf. Dietzel et al., 2020). However, although ostracodes were not as intensively studied for their biomineralization as other organism groups it is known that the valves of Cytheroidea, to which Cytheridella belongs, are almost completely built of calcite crystals (Keyser and Walter, 2004).
As stated previously it is more probable that the sample solution and measured temperature do not correspond to those at the time of valve calcification. Rapid valve calcification and strong variability of some environmental parameters preclude to correlate values representing exactly the same specific time period. The `unusual´ high fractionations within the temperature range of 30°C to 37°C affects almost exclusively samples from Florida and Mexico. Sampling took place in summer where daily temperature gradients in South Florida account ~ 2°C and increase up to 4°C – 5°C between months (e.g., Price and Overton, 2005).
Accuracy of expected equilibrium calcites
The accuracy of the expected equilibrium calcites depends on a variety of parameters.
Water temperature data of the study area are only available for Florida and for some localities in Yucatán. Due to the strong relationship between mean air temperature and mean water temperature (r²: 0.80; Rollinson and Rowe, 2018) we used air temperature data for the calculation of expected equilibrium calcites. However, this relationship weakens from high to low latitude lakes probably due to the lower inter-annual air temperature variability in the tropics. Additional influences represent geomorphic factors such as lake surface area and lake depth (Kraemer et al., 2015).
In cenotes, water temperature is fairly constant with a gradient of ~ 2–3°C, and seasonal water temperature differences are, thus, negligible (Alocer et al., 1998). If this is true, our field measurements of 31–32°C indicate that the minimum (air) temperatures used for the calculation of the equilibrium calcites are much too low (Table 1; Fig. 2). Indeed, minimum water temperatures of 24.7 or 24.8°C, respectively, (Alocer et al., 1998; Peréz et al., 2010) imply that inclusion of minimum air temperatures is not necessary.
Temperature seasonality is much higher in marginal regions of the study area (Florida, S-Brazil). Temperature time series of Caloosahatchee River demonstrate that lowest mean water temperature occurs in January with 18.7°C and highest mean water in July with 30.4°C (Baldwin and Hunt, 2014). Mean air temperatures of nearby Fort Myers range from 17.4°C in January to 28.1°C in August (climate-data.org). Thus, the water temperatures are slightly higher but correspond very well to the air temperatures. Similar conditions can be assumed for Southern Brazil. A large lake ~ 2–3°C south to the sample area in S-Brazil displays a seasonal temperature range of 8–25°C (Tavares et al., 2019). The range is slightly higher reported for several rivers and lakes in the state of Rio Grande do Sul with 9–28°C, respectively (Garcia et al., 2008). The deviation to the seasonal air temperature range of Porto Alegre with 10.3–29.6°C is relatively small. This implies that the temperature range used for the calculation of the equilibrium calcites represent realistic estimates of lake water temperatures.
Another fundamental assumption for the estimation of the equilibrium calcites is that δ18Oprecipitation represents the major control on lake water δ18O. Spatial variability of δ18O (and δD) composition of precipitation are negatively correlated with temperature, latitude, altitude, distance from the coast, and the amount of precipitation. Further overriding factors that influence the isotopic signature of local precipitation are the continual loss of moisture from an air mass as it moves away from its evaporation source and mixing of different air masses from local vapor sources as well as storm trajectory (Price and Swart, 2008 and references therein). Therefore, the position or more specifically the distance of the GNIP stations to the sample localities is an important point. GNIP stations are unevenly distributed throughout the study area and are relatively far away from the sample localities in cases of Yucatán and Colombia which questions the representativeness of the δ18O precipitation data for the equilibrium calcites.
The air-line distance from Veracruz to Central Yucatán is over 700 km. However, Veracruz and at least the southwestern parts of the Yucatán Peninsula provide similar climatic conditions and annual precipitation amounts of 1200–1500 mm. Isoscape maps of shallow groundwater, that is assumed as proxy for integrating long-term (ca. 5–10 yrs.) precipitation infiltration inputs, show that Veracruz provides values generally ~ 1‰ lower compared to Yucatán (Wassenaar et al., 2009). Thus, the equilibrium calcites for Yucatán could have been estimated too low.
Although air-line distance between Bogota and the sample localities is only between 100 and 230 km there is, however, an altitude difference of ~ 2170 m (difference refers to Villavicencio). In order to test for this effect, we compared the correction with lake water sample and the altitude (Fig. 12). The deviation between the altitude-corrected and the lake water-corrected calcite is negligible small. This indicates that the correction with local water samples is sufficient to compensate the difference in altitude.
The seasonal fluctuations of precipitation amount and its isotopic composition over the major part of South and Central America are controlled by seasonal displacement of the Intertropical Convergence Zone (ITCZ) and the associated changes in the circulation patterns and moisture transport across the continent (Rozanski and Araguás Araguás, 1995). In the tropical Americas δ18Oprecipitation is not only related to one climatic variable but is rather assumed to result from the interplay of different factors which include precipitation amount, temperature, source region contribution, and also the atmospheric circulation (Vuille et al., 2003). For these reasons the inclusion of more than one water sample per site that illustrate seasonal changes of the lake water would be beneficial for the correction of the equilibrium calcites.
Meteoric water lines of surface water in high-humidity regions such as Panama imply that they are not affected by substantial evaporation (Lachniet and Patterson, 2006). Our data which were taken during the rainy season show the same subordinate influence of evaporation on δ18O and δD values of the sample sites (Fig. 5A). Corrections of the calcites that are primarily intended to provide an approximation of the evaporation effect might be therefore not as similarly necessary as in less humid regions and closed basin lakes. However, since all regions except Southern Brazil provide a pronounced precipitation seasonality the necessity of corrections probably differs throughout the year.
The inclusion of literature data to receive further equilibrium calcites shows that the ostracode values coincide stronger with the lake water-corrected calcites than with the literature data-corrected calcites (e.g., Fig. 10). This implies that the lake water has not changed its isotopic composition strongly during calcification of the ostracode valves and (lake water) sampling. The usage of lake water compositions from different dates (i.e., years) might be useful for estimations of different lake conditions for application to e.g., fossil ostracode valves.
Inferences on Cytheridella calcification periods
To identify possible calcification periods for Cytheridella we compared estimated monthly ranges of equilibrium calcites with the ostracode δ18O values (Figs. 10–14). Meyer et al. (2017) excluded months with a large δ18Oeq_ex range as calcification period and assumed that for a plausible calcification time, the ostracode value lies within the range of the theoretical calcite. Further, δ18Oostracode should tend to be more positive due to positive vital effect. These assumptions are, however, imprecise in the way that it is not clear which δ18Oeq value should be exceeded by the δ18Oostracode; for instance, the value calculated on the basis of Tmean and mean δ18Ometeoric, the mean of the calcite range or the upper range margin of the calcite (representing Tmin and δ18Omax). Alternatively, the requirement that the values should tend to be more positive could also indicate that δ18Oostracode is just over the calcite range. This precludes unequivocal identification of possible calcification periods. The approach using the differences between mean δ18Oostracode and the equilibrium calcite value based on Tmean- δ18Omean facilitates the recognition of possible calcification periods.
The most conspicuous finding of this approach is the conform pattern of offsets between mean equilibrium calcites and mean ostracode isotope values during the year (Fig. 15). Although it is possible that through different factors (see discussion above) the equilibrium calcites may have under- or overestimated (displayed by large differences in the offsets) it is conspicuous that there is a general pattern displayed by all regions in form of an `M´. This pattern indicates that winter months can be excluded since offsets are distinctly negative. An exception is provided by Floridian offsets that are relatively positive compared to other regions. This might be attributed to the relatively small ranges of the equilibrium calcites during winter (Fig. 10). Some summer months (in most regions June and September) can be also excluded due to offsets far too positive.
Although each region displays different offsets throughout the year there are periods or months in which almost all offsets coincide with the ~ 1‰ (± 0.5‰) boundary representing the assumed vital effect of Cytheridella (see Fig. 15). Accepting the 1‰-boundary as valid specific offset of Cytheridella, two or three possible calcification periods in spring (April/May), mid-year (July and August) and autumn (October/November) are indicated. This coincides with the observation by Peréz et al. (2010) who reported that Cytheridella molts in spring. It is important to consider that the diagram displays only in which months the offsets are in the suitable range of about + 1‰. Meyer et al. (2017a) assumed also that Cytheridella has a possible calcification period in spring (April) with a second possible calcification time in autumn (October).
The within-sample isotopic variation can provide indirect hints on the life cycle of the ostracode species. The relatively low variation ranges of most sites indicate short and probably seasonally restricted calcification periods of the respective population. Unfortunately, few is known about total life spans in general, and adult life times in particular, of most ostracodes. Within the Podocopida, to which all non-marine ostracodes belong, are the Cytheroidea assumed to exhibit total life spans of ~ 2 months to > 3 years, and adult life times of > 7 days to 3 months (Cohen and Morin, 1990). From Metacypris cordata, a member of the Subfamiliy Timiriaseviinae to which Cytheridella is assigned, is reported to have a total life span of 6–10 months with one generation (Colin and Danielopol, 1979).
So far, our approach ignores variations in δ13C. However, since processes determining carbon isotope values differ almost completely from that controlling oxygen (e.g., Schwalb, 2003; Leng and Marshall, 2004) it might possible to improve our understanding on the relationships between seasonal variations on carbon δ13CDIC and its imprint in authigenic carbonates such as ostracodes’ valves
In littoral zones the δ13CDIC experiences large seasonal variations (Decrouy et al., 2011). The knowledge of these variations could eventually help to determine if Cytheridella develops a second generation in cases when the oxygen isotopes provide inconvincible information. So far, investigations of the morphological variability (Wrozyna et al., 2018a, b; 2019) did not give any hints of e.g., seasonal morphotypes. Ultimately, the population age structure of living Cytheridella samples from different months or seasons would provide the necessary proof of a second generation.
Nonetheless, the similarity of the pattern implies that the calcification time (within a year) of Cytheridella is similar in all regions. The synchronous life cycle of Cytheridella from the different regions is somewhat surprising, since life histories or even development phases of crustaceans in general, are often adapted and coupled to marked shifts in their environment (Olesen, 2018). For ostracodes it is reported that speed of development has been found to be related to environmental factors, particularly temperature, salinity, and habitat (Cohen and Morin, 1990 and references therein). If this would be the case for Cytheridella, another pattern would emerge since our data set covers a wide latitudinal range, different habitats, hydrochemical facies and salinity ranges (within and between the regions). In contrast to temperate regions where seasonality is probably controlled by temperature (Horne, 1983) it has been hypothesized that seasonality of tropical ostracodes might be related to hydrological conditions (i.e., precipitation). As a consequence, the calcification periods of Cytheridella are therefore to be related to the shift between the dry and rainy season (Meyer et al., 2017a). However, according to our data this hypothesis has to be rejected since there are pronounced differences in annual precipitation distribution and amounts between the studied regions (Fig. 2). In particular, Southern Brazil is characterized by high precipitation amounts throughout the year compared to the other regions where precipitation is restricted to summer months.
As an alternative view we can consider that a given life cycle is also the result of its ancestry, meaning that it is best interpreted by implementing not only ecological but also evolutionary interpretations (Oleson, 2018). Thus, phylogenetical relationships are highly relevant to understand shared similarities of life cycles while ecological studies could identify short term and local explanations for differences in certain developmental phases due to environmental factors such as food availability, hydrological changes, etc. Morphometric investigations of appendages and valves have shown that Cytheridella develops distinct regional morphotypes (Wrozyna et., 2016; 2018). Correspondingly, it has been shown that freshwater ostracodes in the Neotropical realm are characterized by a high proportion of endemism (Cohuo et al., 2017). More recent speciation(s) which are usually not detected in qualitative studies (Wrozyna et al., 2019) could explain why morphological divergence is not recognizable. Therefore, the life cycle of Cytheridella seems to be phylogenetically inherited and linked to a supraspecific level. It might have originally been adapted to environmental conditions but has been conserved during the migration and radiation of the group over the Neotropical realm. To support this hypothesis, it must be tested if other taxonomical groups (e.g., species, genera, families) exhibit the same or a similar pattern in order to identify the role of phylogenetic relationships and environmental influences.
This new approach enables a better insight into calcification conditions and resulting geochemical (i.e., δ18O, δ13C) signatures of ostracode valves on a seasonal basis in (sub-)tropical regions where data on lake water hydrochemistry are mostly missing. It also contributes to biological inferences as reconstructed calcification periods of the ostracode species provide hints on its life cycle. Future studies should include monitoring of important hydrochemical parameters (pH, water temperature, δ18O and δ13CDIC, etc.) in order to improve the understanding of temporal fluctuations and interrelationships of involved components on the calcification processes, and repeated sampling of ostracode specimens which would help to verify findings of the `calculated´ calcification periods.