Sinking of Gelatinous Zooplankton Biomass Increases Deep Carbon Transfer Efficiency Globally

Gelatinous zooplankton (Cnidaria, Ctenophora, and Urochordata, namely, Thaliacea) are ubiquitous members of plankton communities linking primary production to higher trophic levels and the deep ocean by serving as food and transferring “jelly‐carbon” (jelly‐C) upon bloom collapse. Global biomass within the upper 200 m reaches 0.038 Pg C, which, with a 2–12 months life span, serves as the lower limit for annual jelly‐C production. Using over 90,000 data points from 1934 to 2011 from the Jellyfish Database Initiative as an indication of global biomass (JeDI: http://jedi.nceas.ucsb.edu, http://www.bco‐dmo.org/dataset/526852), upper ocean jelly‐C biomass and production estimates, organism vertical migration, jelly‐C sinking rates, and water column temperature profiles from GLODAPv2, we quantitatively estimate jelly‐C transfer efficiency based on Longhurst Provinces. From the upper 200 m production estimate of 0.038 Pg C year−1, 59–72% reaches 500 m, 46–54% reaches 1,000 m, 43–48% reaches 2,000 m, 32–40% reaches 3,000 m, and 25–33% reaches 4,500 m. This translates into ~0.03, 0.02, 0.01, and 0.01 Pg C year−1, transferred down to 500, 1,000, 2,000, and 4,500 m, respectively. Jelly‐C fluxes and transfer efficiencies can occasionally exceed phytodetrital‐based sediment trap estimates in localized open ocean and continental shelves areas under large gelatinous blooms or jelly‐C mass deposition events, but this remains ephemeral and transient in nature. This transfer of fast and permanently exported carbon reaching the ocean interior via jelly‐C constitutes an important component of the global biological soft‐tissue pump, and should be addressed in ocean biogeochemical models, in particular, at the local and regional scale.


Introduction
Biological oceanic processes, primarily carbon production in the euphotic zone, sinking and remineralization, govern the global biological carbon soft-tissue pump (Buesseler et al., 2007). Sinking and laterally transported carbon-laden particles fuel benthic ecosystems at continental margins and in the deep sea (Koppelmann & Frost, 2008;Robinson et al., 2010). Marine zooplankton play a major role as ecosystem engineers in coastal and open ocean ecosystems because they serve as links between primary production, higher trophic levels, and deep-sea communities (Lebrato et al., 2012;Robinson et al., 2010;Robison et al., 2005). In particular, gelatinous zooplankton (Cnidaria, Ctenophora, and Chordata, namely, Thaliacea) are universal members of plankton communities that graze on phytoplankton and prey on other zooplankton and ichthyoplankton. They also can rapidly reproduce on a time scale of days and, under favorable environmental conditions, some species form dense blooms that extend for many square kilometers (Condon et al., 2013). These blooms have negative ecological and socioeconomic impacts by reducing commercially harvested fish species (Pauly et al., 2009), limiting carbon transfer to other trophic levels , enhancing microbial remineralization, and thereby driving oxygen concentrations down close to anoxic levels (Frost et al., 2012).
The global biomass of gelatinous zooplankton, herein collectively referred to as "jelly-carbon" (jelly-C), within the upper 200 m of the ocean amounts to 0.038 Pg C (Lucas et al., 2014). Calculations for mesozooplankton (200 μm to 2 cm) suggest about 0.20 Pg C (Moriarty & O'Brien, 2013). The short life span of most gelatinous zooplankton, from weeks up to 2 to 12 months (Ceh et al., 2015;Raskoff et al., 2003), suggests biomass-production rates above 0.038 Pg C year −1 , depending on the assumed mortality rates, which in many cases are species-specific. This is much smaller than global primary production (50 Pg C year −1 ) (Field et al., 1998), which translates into export estimates close to 6 Pg C year −1 below 100 m (Moore et al., 2004;Siegel et al., 2014), depending on the method used. Globally, gelatinous zooplankton abundance and distribution patterns largely follow those of temperature and dissolved oxygen as well as primary production as the carbon source (Lucas et al., 2014). However, gelatinous zooplankton cope with a wide spectrum of environmental conditions, indicating the ability to adapt and occupy most available ecological niches in a water mass. In terms of Longhurst regions (biogeographical provinces that partition the pelagic environment; Longhurst, 1995Longhurst, , 1998, the highest densities of gelatinous zooplankton occur in coastal waters of the Humboldt Current, NE U.S. Shelf, Scotian and Newfoundland shelves, Benguela Current, East China and Yellow Seas, followed by polar regions of the East Bering and Okhotsk Seas, the Southern Ocean, enclosed bodies of water such as the Black Sea and the Mediterranean, and the west Pacific waters of the Japan seas and the Kuroshio Current (Brotz et al., 2012;Condon et al., 2012;Lucas et al., 2014). Large amounts of jelly-C biomass that are reported from coastal areas of open shelves and semienclosed seas of North America, Europe, and East Asia come from coastal stranding data (e.g., http://www.jellywatch.org/). Large amounts of jelly-C are quickly transferred to and remineralized on the seabed in coastal areas, including estuaries, lagoons and subtidal/intertidal zones (reviewed in Lebrato et al., 2012), shelves and slopes (Billett et al., 2006;Lebrato & Jones, 2009;Sweetman & Chapman, 2011), the deep sea , and even entire continental margins such as in the Mediterranean Sea (Lebrato, Molinero, et al., 2013). Jelly-C transfer begins when gelatinous zooplankton die at a given "death depth" (exit depth), continues as biomass sinks through the water column, and terminates once biomass is remineralized during sinking or reaches the seabed, and then decays. Jelly-C per se represents a transfer of "already exported" particles (below the mixed later, euphotic or mesopelagic zone), originated in primary production since gelatinous zooplankton "repackage" and integrate this carbon in their bodies, and after death transfer it to the ocean's interior. While sinking through the water column, jelly-C is partially or totally remineralized as dissolved organic/inorganic carbon and nutrients (DOC, DIC, DON, DOP, DIN, and DIP;Chelsky et al., 2015;Sweetman et al., 2016;West et al., 2009), and any left overs further experience microbial decomposition or are scavenged by macrofauna and megafauna once on the seabed (Sweetman et al., 2014;Tinta et al., 2016). Despite the high lability of jelly-C (Ates, 2017;Sweetman et al., 2016), a remarkably large amount of biomass arrives at the seabed below 1,000 m. During sinking, jelly-C biochemical composition changes via shifts in C:N:P ratios as observed in experimental studies (Frost et al., 2012;Sempere et al., 2000;Titelman et al., 2006). Yet realistic jelly-C transfer estimates at the global scale remain in their infancy, preventing a quantitative assessment of the contribution to the biological carbon soft-tissue pump.
Ocean carbon export is typically estimated from the flux of sinking particles that are either caught in sediment traps (Asper, 1987) or quantified from videography (Jackson et al., 1997), and subsequently modeled using sinking rates (Martin et al., 1987). Biogeochemical models (e.g., Gehlen et al., 2006;Ilyina et al., 2013;Laufkötter et al., 2015) are normally parameterized using particulate organic matter data (e.g., 0.5-1,000 μm marine snow and fecal pellets) that were derived from laboratory experiments (Ploug et al., 2008) or from sediment trap data (Gehlen et al., 2006). These models do not include jelly-C (except larvaceans; Lombard & Kiørboe, 2010; not only because this carbon transport mechanism is considered transient/episodic and not usually observed, and mass fluxes are too big to be collected by sediment traps (e.g., Siegel et al., 2014), but also because models aim to simplify the biotic compartments to facilitate calculations. Furthermore, jelly-C deposits tend not to build up at the seafloor over a long time, such as phytodetritus (Beaulieu, 2002), being consumed rapidly by demersal and benthic organisms (Sweetman et al., 2014) or decomposed by microbes (Tinta et al., 2016). The jelly-C sinking rate is governed by organism size, diameter, biovolume, geometry (Walsby & Xypolyta, 1977), density (Yamamoto et al., 2008), and drag coefficients (McDonnell & Buesseler, 2010). Recently, Lebrato, de Jesus Mendes, et al., 2013 determined the average sinking speed of jelly-C using Cnidaria, Ctenophora, and Thaliacea samples, which ranged from 800 to 1,500 m day −1 (salps: 800-1,200 m day −1 ; scyphozoans: 1,000-1,100 m d −1 ; ctenophores: 1,200-1,500 m day −1 ; pyrosomes: 1,300 m day −1 ). Jelly-C model simulations suggest that, regardless of taxa, higher latitudes are more efficient corridors to transfer jelly-C to the seabed owing to lower remineralization rates . In subtropical and temperate regions, significant decomposition takes place in the water column above 1,500 m depth, except in cases where jelly-C starts sinking below the thermocline. In shallow-water coastal regions, time is a limiting factor, which prevents remineralization while sinking 10.1029/2019GB006265 Global Biogeochemical Cycles and results in the accumulation of decomposing jelly-C from a variety of taxa on the seabed. This suggests that gelatinous zooplankton transfer most biomass and carbon to the deep ocean, enhancing coastal carbon fluxes via DOC and DIC, fueling microbial and megafaunal/macrofaunal scavenging communities. However, the absence of satellite-derived jelly-C measurements (such as primary production; Behrenfeld & Falkowski, 1997) and the limited number of global zooplankton biomass data sets make it challenging to quantify global jelly-C production and transfer efficiency to the ocean's interior.
In this study, we use data from the recently published Jellyfish Database Initiative (JeDI: http://jedi.nceas. ucsb.edu; Condon et al., 2015), public data sets (http://www.bco-dmo.org/dataset/526852), and global biomass estimates of the upper ocean (Lucas et al., 2014) to model jelly-C transfer efficiency based on Longhurst biogeographical provinces. We estimate exit depths for individual gelatinous zooplankton phyla based on field vertical migration data. Combining these data with published sinking rates (Lebrato, de Jesus Mendes, et al., 2013) and vertical temperature profiles (compiled from GLODAPv2; Key et al., 2015) allow us to calculate remineralization curves and jelly-C transfer efficiency globally. Calculations use over 90,000 data points from 1934 to 2011 as a global representation, across the different phyla, and include separate values for each Longhurst Province. Using two different model functions, the aim of this study is to quantitatively address global biomass and jelly-C transfer efficiency, as well as the gelatinous zooplankton contribution to the sinking compartment of the biological soft-tissue pump. Our results suggest a high importance of including jelly-C transfer efficiency in the next generation of local and regional biogeochemical models, to consistently account for its potential role in global calculations.

Data Sources
In this study, gelatinous zooplankton and jelly-C refer to mesozooplankton (0.2 to 20 mm), macrozooplankton (20 mm to 20 cm), and megazooplankton (20 to 200 cm), following the Sieburth et al. (1978) scale. The majority of the observations of sinking jelly-C concern the macrozooplankton and megazooplankton. Gelatinous zooplankton biomass data originated from the Jellyfish Database Initiative (JeDI http://jedi. nceas.ucsb.edu; Condon et al., 2012), which is a global synthesis project that provides a public data set of nearly 500,000 data points, including quantitative, categorical, and presence-absence data for the last 400 years (see Lucas et al., 2014, for details; Figure 1). This project has a number of public databases in their open directories, including vertical migration data, which were accessed through http://www.bco-dmo.org/ dataset/526852, and then arranged for modeling purposes in a CSV file (Figure 2; Appendix A). Readers can access data in the format used in this publication at the NOAA National Center for Environmental Information (NCEI) under Accession Number 0207402 in https://data.nodc.noaa.gov/cgi-bin/iso?id=gov. noaa.nodc:0207402. A limitation of the JeDI data set is that it does not cover the entire ocean (<50%), and it is also biased toward the Northern Hemisphere (43% vs. 23%) ( Figure 1). The Southern Ocean, the West Pacific, the South East Pacific, and the South Atlantic are not well represented. This means that our jelly-C transfer efficiency total estimations from the upper-ocean gelatinous biomass (starting total production of 0.038 Pg C year −1 ) are strongly conservative (see section 4). Although jelly-C exists below 200 m in the mesopelagic zone, for the purpose of this study we only used data from the upper 200 m because data from deeper depths are of poor quality and unreliable (Lucas et al., 2014). We do not use coastal stranding data in any of our calculations, all data are from the open ocean. Biomass data (in mg C m −3 ) are calculated from abundance data following biometric equations (Lucas et al., 2011). These conversion factors are used because they were the most complete database at the time this study was developed and they were ready to be applied to our model. We acknowledge that there are more recent compilations of gelatinous zooplankton carbon estimations (Andersen et al., 2016) that could slightly reshape our modeled estimates. Biomass data are then classified as maximum and mean biomass by Longhurst Province, and finally further classified by phylum (Chordata, Cnidaria, and Ctenophora; Lucas et al., 2014; Figure 1; Appendix B).

Data Mining and Use in the Model
In order to estimate transfer efficiency, we calculate jelly-C microbial decay ratios between 1 and 0 to describe jelly-C remineralization profiles, which are based on vertical temperature profiles, species-specific sinking speeds, and empirical decay rates, for each Longhurst Province. For biomass data, mean and maximum values have been extracted from field data reported by Lucas et al. (2014) for the 43 Longhurst Provinces. Next, carbon transfer efficiency was calculated using combinations of exit depth (b0) and remineralization profiles (see below for details). The original data in Appendix B and Table S1, S2, S3, and S4 were extracted from Lucas et al. (2014) under John Wiley and Sons License Number 4575870755047.
Temperature data were extracted from the GLODAPv2 database (Key et al., 2015) (Figure 2; Appendix C) and classified according to latitude and longitude ranges of the 43 Longhurst Provinces and averaged into one representative vertical temperature profile per province (Appendices A and B). The representative temperature profiles are based on 50 to 100 temperature profiles per province, covering as much area as possible and spread out as evenly as possible using the same relative distance between profiles to avoid uneven weighting of areas within provinces with different densities of available temperature profiles. We acknowledge that a single Longhurst Province can have many different temperature profiles owing to the geographic range it occupies ( Figure 2). Average seabed depth was also calculated from the GLODAPv2 database to feed the model with an end depth (b1) (Appendices A and B).
Based on the range of jelly-C sinking speeds (Lebrato, de Jesus Mendes, et al., 2013), we calculate mean global biomass export for nominal sinking speeds of 500, 1,000, and 1,500 m day −1 . Subsequently, assumed taxonspecific sinking speeds are used to obtain export estimates per phylum: Chordata (800 and 1,200 m day −1 ), Cnidaria (900 and 1,100 m day −1 ), and Ctenophora (500 and 1,300 m day −1 ). Finally, a uniform sinking speed of 1,000 m day −1 is applied for all taxa to provide global biomass export estimates (Appendix B; Tables S1 and S2). We recognize that species-specific jelly-C sinking speeds could improve the calculations, but to the best of our knowledge, these estimates are the only ones available to date. We assume that carcasses sink in one piece, but we acknowledge potential fragmenting during sinking, which can change the sinking rate and also affect remineralization rate since smaller particles have larger surface area to volume ratio.
For exit depth ("death depth" = b0) ( Figure 2) calculations, we assume that gelatinous zooplankton stay within their vertical migration depth range and do not undergo major predation. Since different species, classes, and phyla have different vertical migration ranges, and it is practically impossible to know at which depth they are dying, we assume that their deepest point of migration matches b0. The selection of a deep b0 Figure 2. Latitudinal jelly-C exit depths and temperature profiles. (a) Detailed jelly-C exit depths (b0) vs. latitude for Chordata (Salpidae/Thaliacea), Cnidaria (Hydrozoa/Scyphozoa), and Ctenophora (Nuda/Tentaculata), including all available observations and the averaged b0 data used per Longhurst Province (details in Appendix C). (b) Representative vertical temperature profiles using GLODAPv2 data (Key et al., 2015) for each of the individual 43 Longhurst Provinces, used to model jelly-C export (details in Appendix B). Line colors correspond to latitude bands: Subpolar and Polar (blue), Temperate (green), and Subtropical and Tropical (red). Deepest temperature data coincide with the seabed depth, which allow calculating remineralization curves.
was made on the assumption that jelly-C does not stay floating in the surface, although in some cases it might, e.g., arriving to the shelves/neritic/beach areas. We think jelly-C, in general, starts sinking at depth because when organisms die, they are in a moribund state for a long time, slowly/passively sinking, until they eventually die off and sink faster. This has been observed in cruises/fieldwork around the world, and is often commented in cruise reports that find intact jelly-C at the seabed (e.g., Billett et al., 2006;Cacchione et al., 1978;Lebrato & Jones, 2009). Data are arranged geographically to match Longhurst Provinces and because variability exists in the number of b0 estimates per province (column "N°depthscounts" in Appendix B), depths (typically >50 depths per province) were averaged to provide a single "Longhurst death depth" per phylum and province (Figure 2, Appendices A and B). These estimates should be reliable for biomass data from b0 = 200-600 m, but we are apprehensive when it comes to biomass data deeper than b0 = 600 m due to poor or insufficient data (Lucas et al., 2014).

Model Development
The original model equation used to describe jelly-C transfer efficiency employs a decay rate (k, equation (1)) with exponential dependence on temperature . The model does not account for consumption influencing the decay rate, although it is known that jelly-C are further consumed by other organisms. The exponential relation is founded thermodynamically because many metabolic rates correlate exponentially with temperature, e.g., in aerobic heterotrophs the metabolic rate equals the respiration rate (Brown et al., 2004). Yet, when applying an exponential temperature equation (equation (2)), very quick biomass decay occurs at high temperatures based on a high bacterial remineralization coefficient (decay rate k) that may or may not be realistic. Experimental observations at the upper end of the temperature range used here (about 30°C) are lacking. To account for this uncertainty, we approach this issue by introducing an alternative equation with k depending linearly on temperature (equation (3)) to reduce the effect of high temperatures on the biomass decay rates , and references therein, Figure 3, Appendices D and E). Results are presented for both equations to provide a range of jelly-C transfer efficiencies. Lastly, we compare both equations under different field (latitudinal) temperature conditions (Figure 2, Appendix D).
Jelly-C decay is calculated following first order kinetics with a decay constant (k) depending exponentially (r 2 = 0.98) or linearly on temperature (T), where T is represented by a gradient (exponential equation (2))  2000), and for phytoplankton from Sudo et al. (1978) and Fujii et al. (2002). or a mean (linear equation (3)). While k may be expected to vary with organism size (Titelman et al., 2006), temperature accounts for 98% of the variability in k (equation (2); Figure 3a), even though the data encompass a wide range of sizes (0.1-1 kg wet weight). Solving equation (1) where z R and T (z R−1 ) are the end depth (seabed) and temperature in the water column, respectively. The depth z D was obtained from the lower range of taxa vertical migration data ( Figure 2). The data were independently collected and calculated per Longhurst Province and then applied in equations (2) and (3) below.
Jelly-C microbial decay ratios were estimated by solving (equation (1)) piecewise, dividing depth ranges into discrete depth intervals (Longhurst Province-specific; see Appendix B) using the two parameterizations for k below (Figure 3). In equations (2) and (3), C is Sinking Speed and K T (i) is the local temperature gradient between depths z i−1 and z i . Underlying assumptions, calculations, and derivations of the solutions are available in Appendix E: 1. Exponential equation: In both cases transfer efficiency is calculated as the product of the individual decay ratios of all depth intervals: where n is the number of depth intervals in each Longhurst Province and the local transfer efficiency M (z i ): M (z i-1 ) is obtained with k(T) from either equation (2), for exponential temperature dependence, or equation (3), for linear temperature dependence. An EXCEL spreadsheet is provided in the supplementary material (Appendix D) with the solution of equations (2)-(4), where users can type in their own data (Table 1; Appendix D).

Global Jelly-C Biomass Transfer Efficiency Estimations
In order to work out a global estimate of jelly-C arriving at the seabed, based on transfer efficiency, we grouped all Longhurst provinces data into three latitudinal bands, polar (66°to 90°, N and S), temperate (66°to 23°, N and S), and tropical (23°to 0°, N and S), using Appendix B to classify jelly-C biomass per Longhurst province and per latitudinal band. Then we averaged the representative vertical temperature profiles of all provinces in each latitude band to obtain one representative profile for each latitude band. Using the initial global standing stock of jelly biomass, M (z R ) = 0.038 Pg C from Lucas et al. (2014), we conservatively assume that organisms die after 1 year. Thus, our estimates remain an approximation given the short life span of many species, usually less than 1 year (Ceh et al., 2015;Raskoff et al., 2003). Regional biomass percentages are attributed to all Longhurst Provinces per latitudinal band, to determine how much biomass each latitudinal band contributes to the global standing stock of the database. The overall jelly-C biomass percentages are polar~20%, temperate~35%, and tropical~45%. Using these percentages as initial biomass per latitudinal band plus the vertical temperature profiles, a constant sinking speed of C = 1,000 m day −1 and an exit depth of 100 m is used to determine the amount of jelly-C biomass transferred to each depth interval down to 6,000 m. Total jelly-C biomass transferred is reported below as the sum of the biomass per latitude band at each depth level. A single sinking speed and a single exit depth are chosen for the global estimation as an average of all the different scenarios encountered for all taxa, provinces, and sinking speeds

Results
Most b0 data (over 95%) in the database are within the upper 1,000 m, over 80% within 600 m (Figure 2; Appendices A and B). Only at a few latitudes we find b0 deeper than 600 m, which we exclude from further analysis. In most Longhurst Provinces and phyla, b0 lies within the upper 200 m, with some exceptions near 500 m. The averaged temperature profiles reflect our division of the Longhurst Provinces into tropical, temperate, and polar latitude bands, with permanent, seasonal, and absent thermoclines, respectively ( Figure 2; Appendix C). We consider all export to below the permanent thermoclines as permanently exported to the ocean interior. Except for polar regions, temperature profiles converge between 3°C and 8°C below 1,000 m, with a semiconstant temperature below. The use of exponential and linear equations of decay rate (k) vs. temperature results in different microbial decay ratios for polar and tropical but not for temperate latitudes.
Jelly-C transfer efficiency depends on initial biomass, initial depth, end depth, sinking speed, and the temperature dependence (exponential is preferred for mechanistic reasons) (Figure 4). The microbial decay ratio is very sensitive to the exit depth (b0 = 100, 500, or 1,000 m) and end depth, which largely determine how much jelly-C arrives at the seabed (transfer efficiency >20% in most cases) (Figure 4). Varying sinking speed at b0 between 500 and 1,500 m day −1 changes how much jelly-C is transferred from b0 to the seabed by 5% to 15%. The equation for the temperature dependency (exponential or linear) also affects the estimated jelly-C transfer efficiency, with changes between 5% and 15% for the same b0 and sinking speed (Figure 4). The largest remineralization within the water column occurs in tropical regions, followed by temperate and polar regions. In most polar regions, remineralization is minimal, with over 70% of jelly-C arriving at the seabed. The same patterns are observed when using phylum-based ratios, but differences exist among the phyla ( Figure 5). Each phylum has its own b0 per Longhurst province, which is reflected in the remineralization curves ( Figure 5). Most remineralization occurs above 1,000 m, irrespective of b0 and sinking speed, with low remineralization rates below.
Jelly-C (mg C m −3 ) maxima per Longhurst Province are used to determine biomass transfer efficiency under a mass deposition event scenario, as when a blooming population of gelatinous zooplankton completely collapses ( Figure 6). Simulations for each phylum with multiple sinking speeds show that large quantities of jelly-C can reach the seabed at anytime and anywhere, irrespective of latitude, depth, sinking speed, b0, Note. Depth, temperature, and sinking speed parameters to change depending on working needs (EXCEL file in Appendix S4). z 0 = Exit depth ("death depth"); T (z) = Temperature at depth z; K T = Temperature gradient between individual depths; M/M 0 = Microbial decay ratio between two adjacent depths; C = Sinking rate (m day− 1 ); z D -z R = Accumulated depth interval; M/M 0 final = Multiply individual microbial decay ratios to integrate the previous remineralization intervals; Biomass = Use initial biomass as a function of M/M 0 and then multiply by the previous number to integrate the remineralization intervals.

10.1029/2019GB006265
Global Biogeochemical Cycles or temperature dependence, assuming no major consumption from pelagic organisms (the percentage consumed is not known). During large bloom events in specific areas, with peaks from 1-5 g C m −3 to greater than 10 g C m −3 (averages per Longhurst province range from 5 to 15 mg C m −3 ), most jelly-C biomass reaches the seabed (50% to 90%, Appendix C). Surprisingly, water depth (b1) is not a limiting factor and model predictions show that large quantities of jelly-C (>50% of original biomass) can arrive at 5,000 m and even deeper. In general, Cnidaria and Urochordata have the highest transfer efficiency, followed by Ctenophores. In some specific areas, Cnidaria have biomass maxima of~20 g C m −3 , of which over 9 g C m −3 arrives at 5,000 m, but the average per Longhurst Province remains below 20 mg C m −3 . Given that in many oceanic areas, the seafloor is much shallower than 5,000 m, especially at continental margins, a large fraction of the jelly-C is transferred to the seafloor under a mass deposition event sinking at high speed ( Figure 6). The only exceptions are shallow shelves where jelly-C could be remineralized or resuspended and mixed upward later during winter mixing.
Estimates of mean and maximum jelly-C biomass reaching the seabed differ between Longhurst Provinces, in some cases by several orders of magnitude (Figure 7; Tables S1 and S3). Variations between phyla range from near zero in some provinces to one or two orders of magnitude in others, reflecting the extreme complexity of jelly-C spatial distribution and transfer efficiencies (Figure 8; Tables S2 and S4). The total transfer of global jelly-C biomass across tropical, temperate, and polar regions also varies with the water depth being used as a target depth and temperature dependence used. Assuming a (conservative) global ocean total estimate of 0.038 Pg C year −1 (Lucas et al., 2014), jelly-C biomass transfer summed over all areas is 0.022-0.027 Pg C year −1 at 500 m, 0.016-0.018 Pg C year −1 at 1,000 m, and 0.009-0.012 Pg C year −1 at 4,500 m . Global jelly-C transfer efficiencies using exponential and linear temperature dependencies. The microbial decay ratio (z D ) is calculated for individual averaged temperature profiles obtained from GLODAPv2 (Key et al., 2015) using data from 43 individual Longhurst Provinces. The plots use a matrix of three sinking speeds (500, 1,000, and 1,500 m s −1 ) and three exit depths (b0) (100, 500, 1,000 m) to represent most field scenarios. The colored lines indicate the averaged polar (blue), temperate (green), and tropical (red) profiles.

10.1029/2019GB006265
Global Biogeochemical Cycles ( Figure 9). Jelly-C transfer to deeper than 4,500 m and export of biomass to the seafloor occur regularly (<0.010 Pg C).

Transfer Efficiency of Sinking Jelly-C
Organic carbon originating in primary production follows three major pathways when sinking in the water column: (i) transfer to higher trophic levels during zooplankton grazing/filtering/consumption, (ii) export to depth via zooplankton fecal pellets/housings/biomass, and (iii) export through sinking of intact phytoplankton cells (phytodetritus) and marine snow/aggregates (Legendre, 1990;Peinert et al., 1989; Pilskaln & Honjo, Figure 5. Phylum-province specific jelly-C transfer efficiencies using exponential and linear temperature dependencies. The microbial decay ratio (z D ) is d for individual averaged temperature profiles obtained from GLODAPv2 (Key et al., 2015) using data from 43 Longhurst Provinces. The plots use a matrix of two sinking speeds (phylum-specific using published data) and the 43 Longhurst Province specific exit depths (b0) (see Figure 2) to represent as accurately the field scenarios. The colored lines indicate the averaged polar (red), temperate (green), and tropical (blue) profiles. Figure 6. Phylum-province specific jelly-C biomass remineralization curves to show potential for seabed mass deposition. A summary of mean and maximum export values can be checked in Tables S1, S2, S3, and S4 and in Appendix B. Curves are calculated using exponential and linear temperature dependencies for individual average temperature profiles obtained from GLODAPv2 (Key et al., 2015) using data from the 43 Longhurst Provinces. The plots use a matrix of two sinking speeds (phylum-specific using published data) and the 43 Longhurst Province specific exit depths (b0) (see Figure 2) to represent as accurately the field scenarios. For ID of the Longhurst Provinces vertical profiles see Figure 2b.

10.1029/2019GB006265
Global Biogeochemical Cycles Figure 8. Solutions for exponential and linear temperature dependencies. The solutions are shown for start (phyla-province specific) and end (seabed) maximum and mean biomass for the 43 Longhurst Provinces using one sinking speed (C = 1,000 m day −1 ) and specific exit depth (b0 = specific; see Figure 2 and Appendix C).
Biomass in several provinces exceeds the axis maximum, which values can be checked in Tables S2 (exponential) and S4 (linear).

10.1029/2019GB006265
Global Biogeochemical Cycles 1987). The jelly-C export contributes to pathway (i) by grazing, and to pathway (ii) as one of the major routes for particles reaching the deep ocean and the seabed.
Most knowledge about pathways (ii) and (iii) comes from sediment trap measurements at various depths that reflect particles originating in primary and secondary production, detritus, fecal material, or even material already processed by gelatinous zooplankton. For technical reasons, most sinking jelly-C (corpses, live bodies, decaying material) does not reach the traps and this flux is thus not quantified. Trap data are then extrapolated via parameterizations in global models (Honjo et al., 2008;Martin et al., 1987). About 5% to 25% of global net primary production (~50 Pg C year −1 ; Field et al., 1998) is transferred from the euphotic zone to below 200 m (Buesseler, 1998;Schlitzer, 2000), and about 1% to 3% reaches the deep-sea below 3,000 m. Other estimates suggest that less than 15% of organic carbon leaving the euphotic zone reaches depths greater than 1,000 m because the remainder is remineralized or consumed at shallower depths (Buesseler et al., 2007;Buesseler & Boyd, 2009;Marsay et al., 2015). Open ocean sediment trap data (e.g., Buesseler et al., 2007) reveal total export ratios out of the euphotic zone between 0.20 and 0.50 at 500 m, indicating high remineralization rates, while the Martin curve predicts an export ratio of~0.27 at 500 m. Detailed open ocean sediment trap investigations (summarized in Buesseler & Boyd, 2009) reach the same conclusion: between 80% and 90% of the carbon leaving the euphotic zone is remineralized above 1,000 m, and not more than 20% of the particulate carbon (POC) reaches 500 m. There is also a large contribution from fecal pellets, e.g., of copepod-and euphausiid-dominated zooplankton communities adding up in some cases to >90% of the total particulate organic carbon export (Belcher et al., 2017;Clarke et al., 1998).
Considering the different transfer efficiencies associated with pathways (ii), in particular the contribution of jelly-C, and (iii), the question arises: What are the major controls of transfer efficiency, and what are the implications, e.g., in zooplankton communities dominated by jelly-C, copepods, or euphausiids? Microbially mediated decay rates for copepod fecal pellets and marine snow range from 0.08 to 0.21 day −1 (Ploug et al., 2008;Ploug & Grossart, 2000), and for phytodetritus from 0.05 to 0.30 day −1 (Fujii et al., 2002;Sudo et al., 1978;Sweetman et al., 2014). While these decay rates are much lower than those associated with jelly-C (Figure 3), the phytodetritus and fecal particles and aggregates also remain much longer in the euphotic zone, owing to lower sinking velocities (from <1-100 m day −1 for phytodetritus, Mei et al., 2003, to between 100 and 500 m day −1 for fecal particles, reviewed by Turner, 2002). Hence, more time is available for remineralization and grazing, facilitating recycling (Huston & Deming, 2002), and preventing efficient

10.1029/2019GB006265
Global Biogeochemical Cycles transfer to deeper water masses. The loss of carbon from the jelly-C via respiration and finite growth efficiency when acquiring carbon via grazing is considered only implicitly. For scyphozoans, at maximal growth rate, respiration reaches roughly 66% of assimilation, while production reaches 34%, with a net growth efficiency ranging from 35% to 37% (Fraser, 1969;Olesen et al., 1994). For salps, net growth efficiency is above 40% (Madin & Purcell, 1992;Zeldis et al., 1995). The associated carbon losses are all implicitly contained in our estimated flux profiles.
The factors controlling the transfer of phytodetritus, marine snow/aggregates, and small zooplankton fecal pellets from the euphotic zone to depth are interrelated processes, such as aggregation, disaggregation, microbial activity, scavenging, predation, detritus production, and mineral ballasting (De La Rocha & Passow, 2007). The transfer efficiency of jelly-C in pathway (ii) as Cnidaria, Ctenophora, and Thaliacea is estimated at 80% (Polar), 40% (Temperate), and 60% (Tropical) at 1,000 m and 50% (Polar), 20% (Temperate), and 30% (Tropical) at 4,000 m ( Figure 9). In certain areas, jelly-C transfer efficiencies below 3,000 m are about one order of magnitude higher than estimates for other pathways, but the jelly-C fluxes remain transient in nature.
Our findings do not argue against the importance of phytoplankton-and smaller zooplankton-derived carbon, which is much larger than jelly-C, but demonstrate a higher transfer efficiency of the jelly-C export flux than some carbon vectors in certain areas, and situations, in particular around large gelatinous blooms and mass deposition events (e.g., Figure 6). Remineralization and downward-transfer of primary and secondary production by phytoplankton and small zooplankton behaves differently and should be considered separately from jelly-C, even though all ultimately originate from the same carbon source. While phytoplankton, fecal, and marine snow particles are heavily remineralized in the upper water column in relatively close proximity to the atmosphere, jelly-C is quickly transferred to the deep ocean (except in shallow shelf seas), contributing greatly to the deep-ocean pools of DOC and DIC.
Although jelly-C fecal pellets and biomass transfer contribute significantly to the vertical flux of organic carbon, both pathways are highly intermittent, restricted to jelly bloom events and subsequent collapses. Including the pathways associated with jelly-C in local and regional models, and potentially in global carbon export models, should be considered a high priority.

Ecological Consequences of Fast Jelly-C Transfer
Jelly-C and other particulate carbon fluxes contribute greatly to the food demands of both pelagic and benthic organisms. Jelly-C in situ benthic measurements reveal in some cases C and N fluxes of 96% and 160% of the phytodetrital C and N, respectively, arriving at the seabed in the same area (Sweetman & Chapman, 2015). This exemplifies our estimates of maximal jelly-C biomass transfer efficiency for mass deposition events, transferring sudden large pulses of jelly-C to the seabed (Figure 6). During a massdeposition event, starting from 400-500 mg C m −3 , considerable amounts of jelly-C biomass (>100 mg C m −3 ) reach the seabed anywhere due to the relatively low decay rates at the low temperatures below the thermocline. In some temperate and in all polar regions, between 50% and 80% of the upper ocean jelly-C biomass reaches the seabed, thus transferring most of the carbon leaving the exit depth ( Figure 6). Deep-sea long-term monitoring of jelly-C deposition supports some of our model estimates, e.g., Smith et al. (2014) report nearly 40 mg of salp-derived jelly-C m −2 d −1 arriving at 3,400 m in the northeastern Pacific during their 6-month study period. For 2 months, the salp jelly-C covered 90% of the seafloor, and benthic megafaunal organisms increased sevenfold in density after the peak of jelly-C deposition.
An increased jelly-C transfer efficiency has ecological consequences derived from the sinking of large carcass quantities, and the accumulation of the biomass at various depths. We could not find published numerical estimates of jelly-C biomass being consumed by other organisms, which would slightly decrease the transfer efficiency. Pelagic consumption rates would need to be estimated prior to implementing in future models. Jelly-C is first consumed in the water column either as living organisms or upon dying and during sinking. The list of pelagic organisms known to consume jelly-C is growing (20% to 40% of the diet in some species, and up to 80% in others), especially after the increasing use of Stable Isotope Analyses, and includes tuna, billfish, mackerel and penguin species, sun fish, turtles, fish larvae, and juveniles, just to name a few (Cardona et al., 2012;Hays et al., 2018;Llopiz et al., 2010;Sampey et al., 2007;Thiebot et al., 2017). Jelly-C is also consumed by microbes during the initial sinking stages via preferential N-rich gelatinous biomass consumption, leaving behind carbon-enriched jelly-C (Tinta et al., 2016). The presence of jelly-C can also trigger changes in bacterial community structure (Titelman et al., 2006). The influence of sinking jelly-C on the microbial loop is likely rather ephemeral owing to the high sinking speed of jelly-C. On its way to the seafloor, jelly-C is often partly scavenged (Lebrato & Jones, 2009;Sweetman et al., 2014), but it can also arrive intact (Billett et al., 2006). The response of benthic macrofaunal/megafaunal communities to deposited jelly-C is largely unknown, except for baited experiments (Dunlop et al., 2017;Sweetman et al., 2014) and fixed stations at single depths . Part of the jelly-C is eventually consumed and remineralized by bacteria after arrival at the seabed (Sweetman et al., 2016;Titelman et al., 2006), leading to shifts in dissolved nutrients and carbon (Chelsky et al., 2015) and decreased oxygen levels West et al., 2009). Before microbes respire jelly-C at the seabed, much of it is often rapidly consumed by megafaunal organisms including fish, crustaceans, echinoderms, and arthropods (Lebrato & Jones, 2009;Smith et al., 2016;Sweetman et al., 2014). To what extent episodic jelly-C falls can affect macrofaunal and megafaunal communities remains unknown, but it is expected to exert a temporarily strong ecological effect in areas where jelly-C falls with a high transfer efficiency create sudden patches of food at the seabed (Ruhl, 2007).

Toward Including Jelly-C Transfer Efficiency in Biogeochemical Models
The jelly-C biomass transfer efficiency estimates rely on a model which depends strongly on the assumed gelatinous biomass decay rate (k), and its temperature dependence according to the exponential or linear solution ( Figure 3). The exponential temperature dependency has a theoretical foundation in thermodynamics, but there is no theoretical support for the linear dependency. However, the exponential solution predicts very high decay rates at high temperatures, which suggest very fast biomass dissolution in the upper water column, similar to particle flux studies (Martin et al., 1987). This phenomenon may apply to certain jelly-C biomass types (ctenophores, hydrozoans, certain salps), but such results are not commonly observed in the field (Sweetman et al., 2016;Titelman et al., 2006). In fact, benthic surveys show that carcasses from different phyla/species can arrive almost intact in shape at the seabed (Billett et al., 2006;Lebrato & Jones, 2009;Smith et al., 2014). Therefore, until a larger data set on jelly-C biomass decay rates is produced from a range of robust laboratory or field decay experiments at various temperatures across several gelatinous zooplankton taxa, it remains unclear whether exponential or linear formulations should be used. High jelly-C decay rates at high temperatures imply increased microbially mediated decomposition during the initial sinking stages from 100 to 500 m, whereas decay rates slow down below due to lower temperatures. Jelly-C biomass from deeper water masses, e.g., from the mesopelagic zone, starts from a deeper b0, which translates into reduced decay rates with the potential to arrive in a more intact condition at the seabed.
Jelly-C biomass transfer efficiency is highly dependent on sinking rate. Owing to the large size of the carcasses, most sink faster than 800 m day −1 , and we assume a global mean of 1,000 m day −1 (Lebrato, Molinero, et al., 2013). Yet the sinking rate could also be affected by the remineralization itself and carcass fragmentation (affecting remineralization rate too; Frost et al., 2012), which will make sinking speed variable. Another factor not considered in this study is that jelly-C "particles" have a wide range of sizes, from milligrams to many kilograms, thus decay rates could be higher in smaller particles, which has not been considered owing to the absence of empirical data. To date, only the study of Titelman et al. (2006) on jellyfish has considered decay rate as a function of size, suggesting that the temperature effect is dominant over size, at least for very large particles such as jelly-C. For smaller particles, size and temperature both are critical during remineralization (Brown et al., 2004;Cavan et al., 2017). Yet this has only been tested at low temperatures. Experiments need to address decay rates including the size component in a temperature range comparable to the ocean at various latitudes to accurately target a wide variety of gelatinous zooplankton groups. These aspects cannot be accounted for in this study because there are no experimental studies linking sinking rate, remineralization, fragmentation, and size. Fast sinking guarantees quick jelly-C transfer from 1,000 m down to 4,000-5,000 m. Transfer profiles differ among Longhurst Provinces, as they are based on the corresponding temperature profiles. The most efficient transfer from the euphotic zone occurs in polar regions (Figure 4) due to low temperatures in the whole water column, preventing major decay. When b0 is below the euphotic zone and the relatively warm thermocline, e.g., 500 or 1,000 m, the jelly-C transfer efficiency is similar across all latitudes. This is because the strong thermoclines in the upper ocean at temperate and tropical Longhurst provinces govern the jelly-C remineralization profiles (Figures 2 and 4). When grouping organisms by taxa and species, jelly-C export per Longhurst Province has a taxon-or speciesspecific component, which is presently difficult to determine. We acknowledge a variable jelly-C sinking speed and a variable b0 in the upper 500 m ( Figure 5) depending on life history, which we have tried to account for as much as possible using wide ranges for sinking rate and b0. Deep jelly-C export originating at greater depths is difficult to assess because no reliable biomass and b0 data exist below 200-500 m, which could be used to estimate deep carbon export and transfer efficiency. This is somewhat unfortunate because large gelatinous zooplankton populations do live deeper than 500 m in the mesopelagic and bathypelagic zones (Robinson et al., 2010).
Other aspects to consider are that if a jelly-C sinking event is from a population living below 500 m, the carbon flux derives from already exported carbon from the upper ocean (potentially reprocessed already by other gelatinous zooplankton), since export flux is commonly defined at carbon being exported below 150 m. Since all jelly-C ultimately derives from primary production, the actual effect of fast sinking jelly-C is an acceleration of the transfer between the upper 500 m and the deep ocean, thus an increased transfer efficiency. The amounts of jelly-C and microbial decay rates defined here as well as the high sinking speeds guarantee an accelerated carbon transfer to the deep ocean. It also needs to be considered that scyphozoans net growth efficiency is above 30% (Fraser, 1969;Olesen et al., 1994), and that for salps, it is above 40% (Madin & Purcell, 1992;Zeldis et al., 1995), which constitutes a loss of fixed carbon from upper levels, that is then not exported to depth. Corrected for the Longhurst Province area, 59-72% of the jelly-C leaving the upper 200 m arrives at 500 m, 46-54% at 1,000 m, 43-48% at 2,000 m, 32-40% at 3,000 m, and 25-33% at 4,500 m. The~0.038 Pg C year −1 upper ocean gelatinous zooplankton production translates into about 0.02-0.03, 0.02, and 0.01 Pg C year −1 being transferred down to 500, 2,000, and 4,500 m, respectively ( Figure 9). The jelly-C biomass production rate of 0.038 Pg C year −1 can be considered conservative considering that the JeDI database covers less than 50% of the global ocean. This previously ignored carbon export flux is significant compared to total ocean carbon export as estimated from sediment-trap data, permanently transferred to the ocean interior, i.e., below the permanent thermocline, and thus forms an important component of the global biological soft-tissue pump to be urgently considered in local, regional, and potentially global biogeochemical models.