Open-Ocean Minima in δ¹³C Values of Particulate Organic Carbon in the Lower Euphotic Zone

Introduction

Particulate organic carbon (POC) is the major carrier of carbon from the surface to the deep ocean (the biological pump); therefore, quantifying POC and understanding its origins and degradation are important priorities. POC is defined as all combustible carbon captured on filters of pore size ∼0.7 μm and remaining after removal of carbonates by acidification. POC therefore encompasses diverse organic structures contained in the biomass of phytoplankton, heterotrophic and chemoautotrophic microbes, and detrital materials such as dead cells, cell fragments, fecal pellets, other aggregated material, and terrigenous or resuspended allochthonous organic matter. The naturally occurring stable carbon isotope ratio of POC (δ¹³C_POC) is one of the only chemical characterizations that captures the entire POC pool, since individual, measurable chemical components constitute only a small proportion of the total (Kharbush et al., 2020; this issue). Accordingly, δ¹³C_POC values preserved in seafloor sediments are important factors in interpreting earth’s past carbon cycle (Kump and Arthur, 1999).

δ¹³C_POC values in the surface ocean have been relatively well-characterized, largely motivated by the possibility of using preserved δ¹³C_POC values as an indicator of photosynthetic responses to past and evolving sea surface CO₂ concentrations (e.g., Rau et al., 1989; Freeman and Hayes, 1992; Francois et al., 1993; Goericke and Fry, 1994; Young et al., 2013). Open-ocean δ¹³C_POC values vary from −35‰ at high latitudes to −16‰ at low-mid latitudes (Goericke and Fry, 1994). In contrast, the δ¹³C value of dissolved CO₂ used as a substrate for photosynthesis only varies by ∼2‰ across surface waters (Rau et al., 1989). To explain the δ¹³C_POC variations, the expressed isotopic fractionation (ε_P) between CO₂ (δ¹³C_CO2) and photosynthetic biomass (δ¹³C_P) (Hayes, 1993, 2001) was found to correlate to some degree with CO₂ concentration, but also with phytoplankton growth rates and cell geometry (Laws et al., 1995; Popp et al., 1998, and others). Additional influence on ε_P can arise from the use of carbon concentrating mechanisms and the active uptake and/or fixation of bicarbonate (see Wilkes and Pearson, 2019).

In contrast, relatively few studies have focused explicitly on vertical patterns of δ¹³C_POC extending below near-surface depths (Jeffrey et al., 1983; Saino, 1992), and to our knowledge no global data compilation exists. Individually, several studies have noted large (2–5‰) decreases in δ¹³C_POC values between the near-surface ocean and the lower euphotic zone or upper pycnocline in open ocean settings, a depth range over which the values of δ¹³C_CO2 usually vary by only 0.5–2.0‰ (Schmittner et al., 2013; Meyer et al., 2016). Hypothesized mechanisms for creating vertical δ¹³C_POC patterns include both photosynthetic and degradative processes. In this short review we compile data from published studies reporting δ¹³C_POC values in the upper 250 m of open-ocean water columns in order to explore whether global patterns or environmental drivers can be detected, briefly review existing hypotheses for the observed vertical patterns, and make suggestions for future work.

Data Compilation of Global Ocean δ¹³C_POC Profiles

Average δ¹³C_POC Minimum in the Lower Euphotic Zone or Local Pycnocline

We compiled suspended δ¹³C_POC data from 11 published studies representative of the global open ocean and containing 51 vertical profiles of δ¹³C_POC extending from near-surface waters to depths of 250 m. Study locations spanned latitudes from −53° to 50° (Figure 1A and Supplementary Tables S1, S2) in the oligotrophic Atlantic (Pedrosa-Pàmies et al., 2018), equatorial Atlantic (Bishop et al., 1977; Jeffrey et al., 1983), Gulf of Mexico, Caribbean Sea, Eastern Tropical Pacific (Jeffrey et al., 1983), South Atlantic (Hurley et al., 2019), Southern Ocean (O’Leary et al., 2001; Trull et al., 2008), and North Pacific (Saino, 1992; Minagawa et al., 2001; Hernes and Benner, 2002; Druffel et al., 2003). Non-tabulated data points (Saino, 1992) and coordinates locations (Trull et al., 2008) were extracted using a built-in MATLAB function (ginput); data are compiled in Supplementary Table S2. We excluded some published studies from our numerical analysis due to a scarcity of depths sampled in the upper water column or due to likelihood of terrigenous inputs, low-oxygen metabolisms, or experimental manipulations (Williams and Gordon, 1970; Eadie and Jeffrey, 1973; Druffel et al., 1996; Benner et al., 1997; Trull and Armand, 2001; Hernes and Benner, 2006; Close et al., 2014; Krishna et al., 2018; Liu et al., 2018). We also did not include data from the Arctic region due to the prevalence of terrigenous or advected POC in these areas (e.g., Griffith et al., 2012; Xiang and Lam, 2020).

Figure 1. (A) Data locations (open circles) shown over mapped annual climatological surface ocean chlorophyll concentrations from Melin (2013). Symbols are color-coded according to the ranges in Δδ¹³C_POC values observed in the upper 250 m (overlying maximum minus underlying minimum; see Figure 2). Insets show examples of four vertical profiles of δ¹³C_POC values (filled black circles): each inset spans from the surface down to 250 m on the vertical axis and a δ¹³C range of −27 to −21‰ on the horizontal axis (see also Supplementary Figure S1); horizontal lines indicate depths of the mixed layer and euphotic zone as indicated in the legend. (B,C) Normalized Δδ¹³C_POC values among all compiled data: medians (center vertical lines), interquartile range (boxes), 99% intervals (brackets), and outliers (individual points); (B) when grouped by standardized euphotic zone depth intervals (100% E_z, depth at which irradiance is 0.1% of surface values); (C) when grouped by local mixed layer depth (MLD) and maximum annual mixed layer depth (MLD_max). Δδ¹³C_POC indicates δ¹³C_POC values normalized to the minimum δ¹³C_POC value found within the upper 250 m at each station, as described in the text. Depth ranges and statistical results are further outlined in Supplementary Table S3.

To compare across sites, we standardized reported sampling depth in two different ways: as a proportion of total euphotic zone depth, defined as 0.1% surface PAR (E_z, Figure 1B), and relative to the monthly and maximum annual mixed layer depths (MLD, Figure 1C). For the E_z standardization we retrieved global 1% surface PAR depths from the mapped MODIS-Aqua satellite data product (NASA Goddard Space Flight Center, 2018) produced using the method of Lee et al. (2007); determined to be within ∼14% of in situ measured values). The 4 × 4 km grid location closest to each published study site was extracted; results were similar (within 0.3 ± 1.7 m) when using an average of the surrounding data points. E_z was calculated based on first-order attenuation from 1% PAR at each location and ranged from 61 to 187 m. We retrieved monthly and maximum annual MLD from Holte et al. (2017), who compiled data from Argo float profiles and used a hybrid density algorithm to find MLD. We found the nearest points on the resulting 1 × 1° grid and used an average of the values at these points.

To account for known latitudinal variations in absolute values of δ¹³C_POC, the δ¹³C_POC data for each individual vertical profile were normalized to the minimum δ¹³C_POC value anywhere within the upper 250 m of the water column, i.e., the local minimum value was subtracted from each data value in an individual profile to calculate Δδ¹³C_POC (Figures 1B,C). After depth and isotopic data were normalized, we compiled all data to assess global vertical trends. We compared the distribution of all Δδ¹³C_POC values in the upper 0–50% and lower 50–100% of the E_z and below (Figure 1B and Supplementary Figure S2A). For the MLD standardization we compared all Δδ¹³C_POC values above the monthly MLD, between the monthly and annual maximum MLD (local pycnocline), and below the annual maximum MLD (maximum pycnocline; Figure 1C and Supplementary Figure S2B). Due to normalization of the isotopic data to a zero value, the distributions were skewed and statistical tests were chosen accordingly for non-normally distributed data. A Wilcoxon rank sum test was used to assess significance pair-wise between the different depth ranges (Supplementary Table S3). δ¹³C_POC values were found to be significantly lower in the lower euphotic zone compared to the upper euphotic zone, with a difference in median Δδ¹³C_POC values of 1.4‰ (p < 10^–8; Figure 1B and Supplementary Table S3). δ¹³C_POC values were also significantly lower in depths between the monthly and maximum MLD when compared to those above the monthly MLD, with a difference in median Δδ¹³C_POC values of 0.9‰ (p < 0.01; Figure 1C and Supplementary Table S3).

Generalizing the statistical results above, within the 51 compiled profiles, δ¹³C_POC values in the lower euphotic zone and/or below the local mixed layer are significantly lower than values in the upper euphotic zone and/or above the local mixed layer. Further, when considering data distribution in relation to the euphotic zone depth, δ¹³C_POC values in the lower euphotic zone (50–100% of E_z) defined a local minimum; they are lower than δ¹³C_POC values both above and below this depth range (p < 0.02; Figure 1B and Supplementary Table S3).

Comparison of Individual δ¹³C_POC Vertical Profiles to Environmental Data

The analysis above considered the compiled global data as a whole and examined differences between median δ¹³C_POC values within coarse divisions of the upper water column. We also examined the overall magnitude of variation in δ¹³C_POC values found within individual water columns, i.e., the upper water column max-min range in δ¹³C_POC values (Δδ¹³C_POC max-min). We calculated this range as the difference between the minimum δ¹³C_POC value found anywhere in the upper 250 m of individual water columns and the maximum δ¹³C_POC value found above this depth in order to focus on the pattern of higher δ¹³C_POC overlying lower δ¹³C_POC values. We found that 63% of individual water columns had a max-min range > 2‰, with the largest ranges >5‰ (Figure 2A). We attempted to assess whether the observed Δδ¹³C_POC max-min range correlated with broad scale environmental features extracted from global datasets. Using linear regressions (MATLAB function polyfit), no correlation was found between the local max-min range and the absolute depth of the euphotic zone, chlorophyll a, nor modeled surface phytoplankton size or fecal pellet export (Melin, 2013; Siegel et al., 2014; linear regression, r² < 0.1). The only linear regressions resulting in r² >0.15 were those comparing the local Δδ¹³C_POC max-min range to net primary production (Figure 2B) and maximum annual mixed layer depth (Figure 2C).

Figure 2. (A) Distribution of upper water column ranges in Δδ¹³C_POC values in the 51 individual water columns in this study, calculated as the overlying maximum δ¹³C_POC value minus subsurface minimum δ¹³C_POC value (i.e., maximum values occurring at depths below the minimum are not considered here). Negative values indicate profiles in which the minimum δ¹³C_POC value was found at the shallowest data point. (B) Comparison of Δδ¹³C_POC max-min ranges in (A) to climatological net primary production (NPP) at each location. (C) Comparison of Δδ¹³C_POC max-min ranges in (A) to maximum annual mixed layer depth (MLD_max) at each location. One extreme outlier was omitted from (B) with an NPP value of 1250 mg m^–2 d^–1 and Δδ¹³C_POC max-min range of 1.8‰.

Where possible, we extracted coinciding in situ vertical profiles of concentrations of POC, dissolved inorganic carbon (DIC), O₂, chlorophyll, and nutrient concentrations from the published studies, as well as C:N ratios. We also obtained annual climatological DIC concentrations, alkalinity, salinity, pH, and temperature data mapped at 1 × 1° resolution from GLODAP and World Ocean Atlas databases (Key et al., 2004), and we calculated CO₂ concentration profiles for each δ¹³C_POC data location using the CO2SYS tool (Lewis and Wallace, 1998). We tested for pair-wise correlations between vertical profiles of these variables and δ¹³C_POC values (both absolute values and Δδ¹³C_POC values) where possible, and we did not find any significant relationships. Measured in situ δ¹³C_DIC values were available from O’Leary et al. (2001) and Hurley et al. (2019). We normalized δ¹³C_DIC and δ¹³C_POC to near-surface values, using the data point closest to the surface to account for the known latitudinal variation in surface ocean δ¹³C_DIC and δ¹³C_POC values. In the nine individual profiles where this comparison was possible, we found that the vertical range in δ¹³C_POC values was on average approximately four times larger than the vertical range in δ¹³C_DIC over the same depths (slope = 4.17, r² = 0.53, linear regression).

Discussion

Data Availability and Quality

While our compilation of published data here is likely not exhaustive, it is to our knowledge the first global examination of vertical profiles of δ¹³C_POC in the upper water column. Global mapped 3-dimensional oceanographic data products for δ¹³C_POC currently do not exist, and most major oceanographic programs do not include vertical δ¹³C_POC as a standard parameter. Anecdotally, vertical δ¹³C_POC values might be measured much more frequently than they are published or interpreted (e.g., Bishop et al., 1999). The scarcity of published data and lack of a database for δ¹³C_POC values were barriers to identifying environmental commonalities underlying the observed patterns. We also note that low vertical resolution in many profiles likely means that the true minimum and maximum δ¹³C_POC values in a given water column were not captured (Supplementary Figure S1).

Of the 11 studies from which we compiled data, 6 collected POC using in situ or submersible pumps to filter large volumes of water (>70 L). Jeffrey et al. (1983); Saino (1992), Wu et al. (1999); Minagawa et al. (2001), and Hernes and Benner (2002) instead collected seawater in large volumes (10–30 L) from Niskin bottles and filtered the water shipboard. Filtering POC from large volumes of water helps in obtaining a statistical sampling of the particle population; small sample volumes are subject to interference from the inconsistent capture of sparse large particles as well as a larger proportional blank contribution from dissolved organic matter (DOM) sorbed to the filter when POC loading is low. Discussion of DOM sorption and differential particle retention based on filtration media or pressures has been extensive in terms of effects on POC and pigment concentrations (e.g., Gardner et al., 2003; Nayar and Chou, 2003; Bishop et al., 2012). However, the effects of sampling methods and blank correction on δ¹³C_POC values are under-discussed (Lorrain et al., 2003) and are of particular importance in the subsurface where POC concentrations are lower and blank contribution is proportionally higher. In addition, the most common, elemental analysis-based methods for bulk carbon isotope analysis are >99% inefficient, necessitating large sample quantities; broader use of “nano-scale” analytical methods could improve our ability to obtain better three-dimensional data coverage for oceanic δ¹³C_POC measurements (see Close, 2019).

The size fraction(s) of particles studied also can be important. Most studies described here collected total POC onto 0.45, 0.7, 0.8, or 1.0 μm glass or quartz fiber filters, but the studies of Bishop et al. (1977) and Hurley et al. (2019) in the mid-latitude Atlantic included a prefilter of 53 μm mesh size. In these cases, the observed patterns of δ¹³C_POC in the lower euphotic zone were a particular feature of small particles (∼0.7–53 μm), which constituted the majority of POC. Trull et al. (2008) included measurements of several particle size classes in a study of the Southern Ocean – where large phytoplankton and particles are more abundant that the above studies – and found heterogeneities in the δ¹³C values across both size and depth, possibly relating to a combination of physiological effects on ε_P, different sinking rates, and temporal offsets between surface production and deeper particles (see below).

Photosynthetic Hypotheses for Origins of Low δ¹³C_POC Values in the Lower Euphotic Zone or Upper Pycnocline

Several physiological, community, and environmental features of in situ photosynthesis have been suggested as drivers of systematic variations in δ¹³C_POC values between the upper and lower euphotic zones. Saino (1992) noted “a subsurface minimum near the base of the euphotic zone” when examining δ¹³C_POC values at study sites adjacent to the Kuroshio Current and suggested photosynthetic mechanisms. Models and culture studies (e.g., Popp et al., 1998; Cassar et al., 2006) have demonstrated that ε_P depends on an interplay between the growth rate of an organism and the supply rate of CO₂; higher CO₂ supply rates and/or slower growth rates will lead to a larger ε_P, and therefore lower values of δ¹³C_P. Saino (1992) suggested that both conditions likely exist in the lower euphotic zone: CO₂ concentrations here are higher than the surface due to the net removal of CO₂ by photosynthesis at the surface and the net respiration of CO₂ in the lower euphotic zone, and growth rates are likely lower due to light limitation. O’Leary et al. (2001) also found low δ¹³C_POC values in the lower euphotic zone in a study of the subantarctic water column, and their model results pointed to slow photosynthetic growth rates as the cause. Decreases in photosynthetic growth rate with increasing depth in the euphotic zone are frequently observed in the open ocean (e.g., Laws, 2013), and relationships between growth rates and photosynthetic fractionation of carbon isotopes have been demonstrated in culture (Laws et al., 1995; Bidigare et al., 1997; Popp et al., 1998).

A smaller average cell size of phytoplankton in the lower euphotic zone/upper pycnocline than in the upper euphotic zone/mixed layer also is a frequent observation in open-ocean settings (e.g., Poulton et al., 2006; Barone et al., 2015). Trull and Armand (2001) demonstrated that a smaller average cell size (higher surface area: volume ratio) of phytoplankton in the lower euphotic zone could result in lower δ¹³C_POC values, an isotopic relationship previously established in phytoplankton cultures and models by Rau et al. (1996); Popp et al. (1998), and others, stemming from faster CO₂ diffusion rates.

Conversely, targeted studies of individual photosynthetic taxa or biomarkers have sometimes suggested a pattern of increasing or unchanging δ¹³C values over increasing depth in the euphotic zone (Prahl et al., 2005; Popp et al., 2006; Tolosa et al., 2008; Radabaugh et al., 2014). The recent model of Wilkes and Pearson (2019) suggests that increasing δ¹³C values over depth could be a phenomenon specific to eukaryotic phytoplankton with intracellular CO₂ partitioning or carbon concentrating strategies, with carbon isotopic fractionation varying according to whether growth is limited by low nutrient concentrations (upper euphotic zone/mixed layer) or other factors such as light [lower euphotic zone/upper pycnocline; also explored by Laws et al. (2002) and Cassar et al. (2006)]. Some phytoplankton also employ active uptake and/or fixation of bicarbonate (HCO₃^–), thereby also complicating relationships between cell size, CO₂ concentrations, and δ¹³C_P (e.g., Bentaleb et al., 1998; Cassar et al., 2004). Due to such variations in photosynthetic physiologies, the isotopic sensitivity of a whole phytoplankton community to its local seawater environment (CO₂ concentrations, light and nutrient availability) will depend also on the taxonomic composition of photosynthesizing organisms present (e.g., Leboulanger et al., 1995; Bentaleb et al., 1998; Trull et al., 2008).

Saino (1992) also discussed how slow isotopic equilibrium relative to chemical equilibrium within the DIC system could lead to relatively high values of δ¹³C_CO2 in the mixed layer compared to the lower euphotic zone. Notably, δ¹³C_CO2 usually is calculated from measured δ¹³C_DIC using constants that assume the system is at both chemical and isotopic equilibrium (Freeman and Hayes, 1992).

Some compound-specific isotope analyses of lipids have supported a photosynthetic origin of low δ¹³C_POC values in the lower euphotic zone. Close et al. (2014) found that δ¹³C values of lipid biomarkers for in situ production paralleled δ¹³C_POC values throughout the euphotic zone in the Eastern Tropical North Pacific, which ruled out the degradative accumulation of lipids (see section “Degradative or Dynamical Hypotheses for Origins of Low δ¹³C_POC Values in the Lower Euphotic Zone or Upper Pycnocline”) as a driver behind low δ¹³C values in the lower euphotic zone. Similarly, O’Leary et al. (2001) demonstrated that δ¹³C values of photosynthetic sterols paralleled vertical profiles of δ¹³C_POC in subantarctic waters.

Degradative or Dynamical Hypotheses for Origins of Low δ¹³C_POC Values in the Lower Euphotic Zone or Upper Pycnocline

Equally common in published interpretations of upper water column δ¹³C_POC variations are hypotheses relating to degradation or alteration processes. Jeffrey et al. (1983) noted that “the upper part of the pycnocline (i.e., below the local mixed layer) is marked by lighter POC-δ¹³C values at almost all the stations occupied” in a broad-ranging study of the Eastern Tropical North Pacific, Gulf of Mexico, Caribbean, and western south Atlantic. These authors suggested a degradative mechanism, following the interpretations of Eadie and Jeffrey (1973) and based on the results of Degens (1969): below the local mixed layer POC concentrations decrease rapidly due to high rates of respiration. Amino acids and sugars comprise quickly degrading, ¹³C-enriched components of POC, potentially leaving remaining POC at these depths with a relatively high abundance of ¹³C-depleted lipids. This explanation was cited by Druffel et al. (2003), who noted relatively low δ¹³C values of POC in the upper 200 m at study sites in the North Central Pacific and Sargasso Sea.

Bishop et al. (1977) noted “a depletion in ¹³C of 4‰ over the depth interval between 32 and 50 m,” but could not attribute this to the degradative explanation of Eadie and Jeffrey (1973) because a preponderance of lipids should result in an increase in C:N ratios. Instead C:N ratios were invariant over these depths, and the authors suggested that secondary production by bacteria could lead to low δ¹³C values. However, the proportional contribution of lipids may increase without being accompanied by an increase in C:N, if proteinaceous (low C:N) material is also abundant, as suggested by Pedrosa-Pàmies et al. (2018). There are few broad organic compositional studies of POC that are accompanied by carbon isotopic data. One such study by Sannigrahi et al. (2005) did not find a strong relationship between organic composition and δ¹³C_POC values in the upper water column at Station ALOHA [isotopic data previously reported by Hernes and Benner (2002)]; in particular the lipid composition was not proportionally higher at the lower euphotic δ¹³C_POC minimum than it was at shallower depths.

Hernes and Benner (2002, 2006) suggested that low subsurface δ¹³C_POC values in the oligotrophic subtropical Pacific and Sargasso Sea were the result of an advected source of terrigenous organic matter, based on a correlation with lignin concentration. However, this correlation existed mainly in POM below the euphotic zone. The local minimum in δ¹³C_POC values observed in the lower euphotic zone did not correlate with the proportional contribution of lignin to total POC at that depth. Advection of POC from adjacent waters often cannot explain the range of δ¹³C_POC values observed in a single water column (e.g., Lourey et al., 2004), although this is more commonly cited in the Arctic (Griffith et al., 2012; Xiang and Lam, 2020).

Overall, organic compositional effects on δ¹³C_POC values seem to be more apparent at advanced stages of organic matter degradation that occur below the euphotic zone, in deep mesopelagic and bathypelagic waters. Here decreases in δ¹³C_POC values are more clearly related to increasing C:N ratios or changing biochemical composition than they are in the upper water column (see Minagawa et al., 2001; Hwang and Druffel, 2003; Sannigrahi et al., 2005). Analogous to the patterns recognized for particulate δ¹⁵N values (Altabet et al., 1991), shallower, earlier degradation of macromolecules may lead first to increases in δ¹³C_POC values as bonds containing ¹²C are more quickly hydrolyzed or respired in upper mesopelagic waters (e.g., Bishop et al., 1977; Jeffrey et al., 1983; Wu et al., 1999). Accordingly, slightly higher δ¹³C_POC values were found below E_z compared to the lower euphotic zone in our data compilation (Figure 1B and Supplementary Table S3), and this pattern is more apparent in individual water column profiles (Supplementary Figure S1; see also Cavagna et al., 2013).

Particle, seasonal, and water column dynamics also may lead to vertical variations in δ¹³C_POC. Studies of size-fractionated POC or plankton frequently find size-based differences in δ¹³C values. Typically large particles are assumed to sink more quickly through the euphotic zone; those originating in the surface can carry their δ¹³C_POC values below the mixed layer, possibly also being disaggregated into small particles therein. Large phytoplankton often have higher δ¹³C_POC values than small phytoplankton and therefore would tend to contribute ¹³C-enriched rather than ¹³C-depleted material into the subsurface through such a mechanism (Bishop et al., 1977; Fry and Wainright, 1991; Wu et al., 1999; Trull and Armand, 2001; Trull et al., 2008). Heterogeneous sinking rates and δ¹³C values have also been used to explain temporal offsets in the origins of upper and lower water column δ¹³C_POC values, especially in locations with strong seasonality. That is, POC in the subsurface may represent export from the surface ocean during growth conditions (growth rates, CO₂ concentrations, phytoplankton community composition) typical of preceding seasons (Lourey et al., 2004; Cavagna et al., 2013). In high latitude areas seasonal changes in MLD can also distribute surface biomass throughout the euphotic zone and create temporal and spatial offsets between the formation and observation of POC (Grenier et al., 2015).

Zooplankton egesta and/or methane production by gut microbiota also have been cited as possibly affecting δ¹³C_POC values (Cavagna et al., 2013). Fecal pellets can be highly ¹³C-depleted in comparison to zooplankton diet (Tamelander et al., 2006). In addition, relatively ¹³C-depleted methane has been observed reaching maximum concentrations in the pycnocline of oxygenated open-ocean water columns (Holmes et al., 2000; Sasakawa et al., 2008), likely linked to bacterial degradation of dissolved organic compounds (Repeta et al., 2016). Bacteria may incorporate this methane into biomass during oxidative metabolism (Sasakawa et al., 2008), but it is unclear if biomass quantities would be sufficient to affect overall δ¹³C_POC values.

Unique Geochemical Signatures of the Lower Euphotic Zone or Upper Pycnocline: Potential Implications

One objective of this review is to motivate wider measurement and/or reporting of vertically resolved δ¹³C_POC data. Regardless of the underlying primary or degradative drivers of this pattern, better data coverage – especially at higher vertical resolution – will allow more precise definition of the depths at which δ¹³C_POC minima occur and the overall magnitude of variation in upper water column δ¹³C_POC values; both could prove to be a useful diagnostic of underlying ecosystem processes or carbon dynamics as described above.

The current data compilation suggests that there is a widespread pool of POC in the ocean with δ¹³C values that are significantly different from those modeled as “source” photosynthetic values in the surface ocean. POC in the lower euphotic zone or upper pycnocline is abundant and serves as a food source for subsurface biota; studies such as Liu et al. (2018) have discussed how variations in δ¹³C_POC values in deep chlorophyll maxima can influence how δ¹³C values are interpreted in food web studies. The depths of the lower euphotic zone or upper pycnocline, which usually include deep chlorophyll maxima when present, can also be a location of net particle formation and export (Kemp et al., 2000; Benitez-Nelson et al., 2001; Umhau et al., 2019). Recognizing subsurface sources for exported material can be important to the interpretation of the sedimentary geochemical record (e.g., Luo et al., 2014; Hurley et al., 2018). Interestingly, an analog exists in terrestrial systems; plant matter growing in the upper canopy of forests consistently has higher δ¹³C values than plant matter growing below (the “canopy effect”; Van der Merwe and Medina, 1991). The differential preservation and dietary usage of these two signatures can affect the interpretation of terrestrial soil records and food webs (e.g., Bonafini et al., 2013).

Finally, global biogeochemical models include the carbon isotope fractionation of photosynthetic organisms according to relationships with growth rate and CO₂ concentrations established in culture (Tagliabue and Bopp, 2008; Schmittner et al., 2013). However, it is unclear whether current models already can adequately reproduce observed vertical δ¹³C_POC patterns; these models have instead focused on validation of surface ocean δ¹³C_POC values or vertical δ¹³C_DIC patterns. A more comprehensive dataset of vertical δ¹³C_POC values could help validate such isotope-enabled carbon cycle models.

Conclusion

Existing published works demonstrate that δ¹³C_POC values can be significantly lower in the lower euphotic zone or upper pycnocline compared to the upper euphotic zone or local surface mixed layer in a wide range of global open-ocean locations. While individual studies have offered hypotheses for the origin of this vertical pattern, a global compilation of δ¹³C_POC profiles with coinciding oceanographic data is necessary to test whether there are consistent environmental drivers. We suggest that high-quality isotopic data coverage of the global open ocean is needed, particularly with better data density throughout the euphotic zone and with coinciding CO₂ system data, cell size characterization, and δ¹³C_DIC data. The large water sample volume needed to obtain sufficient material for δ¹³C_POC measurements is currently a barrier to including this parameter in many global-scale oceanographic programs, but improvements in the efficiency of isotopic measurement could mitigate this issue. Compound-specific isotope approaches, autotrophic and heterotrophic rate measurements (including substrate-specific photosynthetic uptake, bicarbonate vs. CO₂), particle dynamics characterization, and relevant enzyme expression data all would aid in discerning between potential hypothesized mechanisms. Further, we suggest that an isotope-specific database such as the emergent IsoBank (Pauli et al., 2017)¹ would be an appropriate place to aggregate both published and unpublished δ¹³C_POC data that currently may be dispersed across various oceanographic databases.

Author Contributions

HC and LH designed the study. LH conducted the data analysis and contributed to the manuscript. HC wrote the manuscript. Both authors contributed to the article and approved the submitted version.

Funding

Publication was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) project number 422798570. This work was partially inspired by discussions at a workshop on Marine Organic Biogeochemistry that was held in Delmenhorst, Germany in April 2019 and funded by the Hanse-Wissenschaftskolleg and the Geochemical Society. LH was supported by a University of Miami Fellowship.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

We thank Ellen Druffel and Rut Pedrosa-Pámies for providing tabulated data from Druffel et al. (2003) and Pedrosa-Pàmies et al. (2018), Igor Belkin for advice in recovering data from Saino (1992), and David Siegel for providing modeled climatologies from Siegel et al. (2014). We are grateful for comments by MM, Yuan Shen, and TT which resulted in considerable improvements to this mini-review.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmars.2020.540165/full#supplementary-material

Footnotes

1. ^ http://isobank-qa.tacc.utexas.edu/en/

References