Sediment traps have been deployed for a number of years at two sites, P2 and P3 (Fig. 1), upstream and downstream of South Georgia (at -55.248∘ N, -41.265∘ E and -52.812∘ N, -39.972∘ E respectively) in the Scotia Sea in the Southern Ocean (Manno et al., 2015). The Scotia Sea is mainly located in the eastward flowing Antarctic Circumpolar Current (ACC), which is split by a number of frontal systems including the Southern Antarctic Circumpolar Front (SACCF; Fig. 1). The complex circulation patterns and variability in frontal systems shapes the Scotia Sea ecosystem (Murphy et al., 2007). P3 and P2 are located downstream and upstream of South Georgia respectively, leading to marked differences in community structure with large rapidly sinking diatoms likely to be more prevalent in the iron fertilised downstream region (Korb et al., 2012; Smetacek et al., 2004). Phytoplankton blooms at P3 can be sustained for 3–4 months (Whitehouse et al., 2008), whereas blooms are typically much shorter in the SACCF region where P2 is located (Park et al., 2010), likely influencing the dynamics of the zooplankton community. Variability in regional dispersal or retention by the current systems of the ACC is important for determining the seasonal dynamics of Scotia Sea ecosystems (Murphy et al., 2007; Thorpe et al., 2007).

During cruises in austral spring 2013 (JR291) and 2014 (JR304) aboard the R.R.S. James Clark Ross, samples of sinking particles in the upper mesopelagic were collected using Marine Snow Catchers (MSCs) (Table 1) and zooplankton abundance data using Bongo nets. Sediment trap data were obtained from traps deployed in 2012 and 2013 at P2 and P3, at depths of 1500 and 2000 m respectively. Mean current velocities in December 2012 and 2013 (measured with a Nortek Aquadopp current meter deployed just below the sediment traps, ST) were 7.2 and 4.5 cm s-1, and, 14.2 and 12.5 cm s-1at P3 and P2 respectively. These data agree with mean current velocities at the depth of the ST at both sites of < 10 cm s-1 observed by Whitehouse et al. (2012) in 2008, suggesting that the effects of lateral advection are minimal and as such they are not considered in this study.

All FPs were photographed using an Olympus SZX16 microscope. FPs were classified visually as round, ovoid, or cylindrical using light microscopy. All FPs in each category collected in the MSC were counted, and their length and width measured using ImageJ. For each ST sample, the dimensions of 10–50 FPs of each class were measured and, for MSC samples, all FPs were counted and measured. FP volumes were calculated for round, ovoid, and cylindrical pellets using the formula for a sphere, ellipsoid, and cylinder respectively. Equivalent spherical diameters (ESD) were also calculated. We compare FP volume rather than FP number to avoid bias due to possible fragmentation (Wexels Riser et al., 2010). The carbon contents of FPs were calculated based on conversion factors of 0.035, 0.052, and 0.030 mg C mm-3 for round, ovoid, and cylindrical FPs respectively, based on measurements made on FPs collected from the ST in spring–early autumn (Manno et al., 2015).

Without faecal production experiments of isolated species, it is difficult to ascertain the exact origin of FPs collected in the MSC and ST. Previous studies (González, 1992; González et al., 1994a; González and Smetacek, 1994; Martens, 1978; Wilson et al., 2008; Yoon et al., 2001) suggest that ovoid/ellipsoidal pellets originate from copepods, pteropods and larvaceans, cylindrical pellets from krill and copepods, and spherical pellets from amphipods, small copepods, and crustacean nauplii.

Sinking velocities (w) of a sample of FPs collected in MSC were measured on board both cruises. During JR291, sinking velocities were measured in a graduated glass cylinder in a temperature controlled laboratory (2 ∘C). For each FPs, the sinking velocity was calculated from the average of the time taken to sink past two marked distances (10 cm apart), with the starting point more than 10 cm from the water surface. During JR304, sinking velocities were measured in a temperature controlled (at 4 ∘C) flow chamber system (Ploug and Jorgensen, 1999), suspending FPs in an upward flow and taking the average of three measurements. Only FPs larger than 0.15 mm ESD (i.e. those visible by eye) could be measured. No significant differences were found between sinking velocities measured during JR291 and JR304 by these two different methods (Student's t test, p=0.2).

The median sinking velocity of measured FPs for each MSC was utilised to calculate the sinking FP flux (FPF). FPFnFPm-2d-1=nFPA×wh Here, nFP is the total number of FPs collected at the base of the MSC (excluding krill FPs), A the area of the MSC opening based on inner MSC diameter, and h the height of the snow catcher (1.53 m).

For sediment trap samples, FP fluxes were calculated as follows: FPFnFPm-2d-1=nFP/(A/d), where d is the number of days that the trap was open (15 days) and A is the area of the sediment trap (0.5 m2).

FP collected in the ST and MSC were compared in terms of the number of FPs in each morphological type as well as in terms of carbon. As the absolute number of FPs was vastly different between MSC and ST samples due to attenuation with depth, we compared the percentage abundance and carbon across the size distribution of all FPs from measured FP volumes. As only an average FP size for each morphological type (rather than for all individual FPs) was measured for samples from the ST deployments, we make use of historical sediment trap data (Manno et al., 2015) at the same sites from December 2009 and 2010. The size of all FPs in each sample split were measured in the study of Manno et al. (2015) and hence we use these data to compare size distributions of MSC and ST collected FPs. Manno et al. (2015) also categorised FPs into ovoid, cylindrical, and round, with an additional category of elliptical. We combine cylindrical and elliptical categories due to their similar morphology and to allow for comparison with our MSC data. Although this introduces uncertainty in terms of inter-annual variability between 2009–2010 (full sediment trap data) and 2013–2014 (Marine Snow Catchers data), consistency in the FP types and percentages in each category between years (Fig. S1 in the Supplement) provides confidence in the use of these historical data. Numbers of large cylindrical FPs, probably originating from large euphausiids, were removed from counts given the large potential bias in the quantification of these organisms in the net samples. Again we took into account only the spring data (November and December).

In order to estimate error uncertainty, we take the standard error of our measurements, i.e. multiple Bongo net tows for zooplankton, multiple MSC deployments for mesopelagic FPs, and multiple ST deployments for bathypelagic FP. We compare zooplankton size distributions using a Kolmogorov–Smirnov test. FP size distributions (in terms of % abundance) are also compared using an Anderson–Darling k sample test as this test is more sensitive to differences in the tails and differences in shift, scale, and symmetry when means are similar (Engmann and Cousineau, 2011). All statistics were carried out in RStudio (version 0.98.1091; R Core Team, 2014).

On average, total zooplankton abundances and species compositions were similar at P2 and P3 (Fig. 2), with small microcopepod species Oithona similis, Oncaea sp., and Ctenocalanus sp. outnumbering the main large calanoid copepod species (Rhincalanus gigas, Calanoides acutus, Calanus similimus, C. propinquus, Euchaeta spp., and Metridia spp.) (Table S2, Fig. 2). The number of zooplankton with a PL < 2 mm was similar at P2 and P3 (ratio P3 : P2 of 1.1), but the abundance of larger copepods (4–7 mm PL) at P3 was almost double that of P2 (ratio P3 : P2 of 1.8) (Fig. S2).

The predicted size distribution of egested FP from our mesozooplankton copepod community highlights that most FPs egested in the epipelagic would be in the smallest size category < 0.001 mm3 (97.6 ± 20.3 and 97.0 ± 4.0 % at P2 and P3 respectively) with low contributions (< 2 %) from each of the larger FP size categories (Fig. 3a). The high standard error of FP < 0.001 mm3 at P2 is in part due to very high abundances of Oithona similis during one deployment. Removing this net from the average gives 97.8 ± 13.7 % FP < 0.001 mm3. The predicted size distributions of FPs at P2 and P3 were not significantly different (p > 0.5, Mann–Whitney U test, Kolmogorov–Smirnov test, and Anderson–Darling k sample test).

Sinking faecal pellets collected by the MSC (upper mesopelagic) and the ST (bathypelagic) are described in terms of size and shape to assess changes between these two layers.

Although small FPs were numerically dominant in the MSC, comparison of Figs. 5 and 6 reveals higher contributions of the larger FP size classes to total FPC. This is not unexpected as larger FPs contain a larger amount of carbon. FPC data highlight the importance of the loss of large FPs to the carbon sinking through the water column. Although abundances of small FPs greatly reduced with depth, this does not represent such a large change in terms of carbon.

Sinking velocities of FPs (excluding krill FPs) collected in the MSC ranged from 52 to 382 m d-1 at P2 and 13 to 227 m d-1 at P3, reflecting the range in FP shapes and sizes. Generally, small FPs had lower sinking velocities than larger FPs. We measured FP sinking rates (excluding krill FPs) of 47–120 m d-1 for FP < 0.002 mm3, and 36–270 m d-1 for FP > 0.02 mm3 (Table S3 in the Supplement). Rates measured in this study are consistent with the range of 5–220 m d-1 given by Turner (2002) for copepod FPs.

At P3, the flux of cylindrical and elliptical FPs in the MSC was an order of magnitude higher than fluxes of round or ovoid FPs (190 716 FP m-2 d-1 compared to 32 172 FP m-2 d-1). Similarly at P2, cylindrical and elliptical FPs were the dominant FP type (21 128 FP m-2 d-1), but fluxes of round FPs were also important (14 596 FP m-2 d-1) at this site (Table 2). FP fluxes in the ST were dominated by ovoid FPs at both sites (Table 2).

Our data suggest that small FPs are not transferred efficiently from the epipelagic to the meso- and bathypelagic, and hence make a small contribution to FP fluxes at depth, particularly in terms of carbon. Comparison of estimated copepod FP production with measurements of sinking FPs in the upper mesopelagic (from MSC) gives an indication of the degree of retention in that layer. The community at both P2 and P3 was dominated by microcopepod species which, based on their size, produce small FPs, which are expected to sink more slowly than large FPs (Komar et al., 1981; Small et al., 1979; Stamieszkin et al., 2015). Agreeing with the data presented here, small FPs (< 0.002 mm3) are predicted to have a sinking velocity 3 times slower than larger FPs (> 0.02 mm3) based on the empirical relationship of Small et al. (1979) for copepod FPs.

The longer residence time of small FPs in the upper ocean (due to their slower sinking velocities) means they are exposed to remineralisation processes, such as coprophagous feeding, fragmentation, and microbial remineralisation, for a longer period of time. This type of retention filter and low export efficiency of small FPs has been observed in a number of oceanographic environments (e.g. Dagg et al., 2003; Viitasalo et al., 1999; Wexels-Riser et al., 2001). Wexels Riser et al. (2010) made observations over the upper 200 m of a Norwegian fjord, finding that large FPs produced by Calanus finmarchicus contributed disproportionately to vertical flux despite large numbers of small FPs produced by Oithona similis, agreeing well with the loss of small FPs that we observed in the Scotia Sea.

It is important to acknowledge here that although the 200 µm mesh used in this study is commonly used in zooplankton surveys, this leads to an underestimation of the smaller zooplankton size classes present in the epipelagic. Ward et al. (2012) found that a 53 µm mesh caught 5.87 times more zooplankton than a 200 µm net in the upper mesopelagic of the northern Scotia Sea in spring. However, in this study an underestimation of the small zooplankton size classes serves to reinforce the fact that small FPs dominate the flux of FPs out of the epipelagic and are largely attenuated as they pass through the mesopelagic.

Comparison of freshly egested FP size distributions with the size distributions of FPs sinking through the mesopelagic relies here on the assumption that different species within the copepod community had the same rates of egestion. FP production varies with species, as well as factors such as season and food availability; the range in FP production rates between different copepod species across a number of high-latitude studies is 2–48 FP ind. d-1 (Dagg et al., 2003; Daly, 1997; Roy et al., 2000; Thibault et al., 1999; Urban-Rich et al., 1999). However, as the estimated abundance of egested FPs in the smallest size category (0–0.001 mm3) is between 60 and 250 times greater than the next largest category, the smallest FPs are still likely to dominate the FP community even if egestion rates are varied within reasonable bounds. Therefore, despite our assumptions regarding rates of egestion, our conclusion of rapid attenuation of these small FPs in the upper mesopelagic remains valid.

Our data reveal a change in FP size, shape, and abundance between the upper mesopelagic and bathypelagic of the Scotia Sea suggesting in situ FP production by deeper-dwelling zooplankton. The occurrence of intact and fresh FPs in deep-sediment traps in the Southern Ocean (e.g. Accornero et al., 2003; Manno et al., 2015) may therefore be a result of an indirect, cascade-like transfer through the mesopelagic as they are reprocessed by different zooplankton communities (Miquel et al., 2015; Urrere and Knauer, 1981).

Urrere and Knauer (1981) deployed free-floating traps off the Monterey Peninsula in California. They observed a decrease in numerical FP fluxes in the upper 500 m, but FP fluxes increased by a factor of 2.7 from 500 to 1500 m. This increase was largely due to elliptical FPs, suggesting the presence of deep resident (or overwintering) zooplankton populations (Urrere and Knauer, 1981). The authors concluded that organic material reaches the deep ocean (supporting deep resident zooplankton populations) through in situ repackaging of detritus and via heterotrophy as well as inputs from migrating populations, emulating the “ladder of migrations” first proposed by Vinogradov (1962). More recently, Miquel et al. (2015) deployed drifting sediment traps in the upper 210 m of the Beaufort Sea, observing increases in elliptical FPs with depth and decreases in cylindrical FPs. They explain this by the presence of omnivorous and carnivorous zooplankton in the mesopelagic, whose primary food sources are the vertical flux of organic matter and other organisms. In agreement with our observations, Suzuki et al. (2003) observed large declines in cylindrical FPs between sediment traps deployed at 537 and 796 m in the marginal ice zone of Antarctica, and increases in elliptical FPs over the same depth range. They suggest that coprophagous feeding and new FP production can explain some of the loss of cylindrical FPs, with fragmentation into small sinking particles explaining the rest. As different zooplankton species produce different shapes of FPs, a change in FP shape can suggest a change in zooplankton community structure.

At both P2 and P3, we saw an increase in the contribution of ovoid FPs to the total number of FPs between the upper mesopelagic (MSC samples) and bathypelagic (ST samples), increasing by factors of 4.5 and 8.5 at P2 and P3 respectively. This suggests that there is either an input of ovoid FPs at depth, or that cylindrical–elliptical and round FP are preferentially remineralised in the mesopelagic. We made both size and shape measurements of FPs in the upper mesopelagic and bathypelagic, allowing us to discern if there is indeed production of new ovoid FPs at depth. At both P2 and P3, we observed size classes of ovoid FPs in the ST (0.003–0.008 mm3) that were not present in the MSC, which rules out selective remineralisation. Furthermore, the intact shape of ovoid FPs in the ST argues against fragmentation as a cause of this change in size distribution. In agreement with Manno et al. (2015), we observed that ovoid FPs in the ST showed fewer signs of fragmentation and were more intact than cylindrical or elliptical FPs at both P2 and P3. Estimates of FPC in ST samples indicates that these ovoid FPs also make a large contribution to the flux of POC and, as such, their production at depth represents a mechanism for long-term storage of carbon in the ocean. Hence, we conclude that FP fluxes to depth are augmented by FPs produced in situ at depth.

We can estimate the size class of zooplankton producing the FPs we find at depth based on the FP size class and Eq. (1). We estimate that zooplankton with a PL of 2.6–3.8 and 2.6–3.2 mm could have produced the FPs we observed in the ST, based on dominant size classes of FPs of 0.003–0.008 and 0.003–0.005 mm3 at P3 and P2 respectively. Of the species within these size classes recorded in the Bongo net tows at P2 and P3, Calanoides acutus IV and Metridia gerlachei adults were the most abundant and may be responsible for the flux of these FPs to the ST. C.acutus is a known seasonal migrator in the region, occurring in the upper 200 m in summer but residing deeper (∼ 200–600 m) in spring (Ward et al., 2012). Metridia spp. are also known migrators (Ward et al., 1995, 2006b; Ward and Shreeve, 1999), found to be one of the more abundant species in the 500–1000 m depth range based on Discovery Investigations to the west of the Drake Passage (Ward et al., 2014). Ward et al. (2014) find the most abundant species in this depth range to be Oncaea spp., Oithona frigida and Microcalanus pygmaeus, all of which are too small (≤ 0.5 mm PL) to produce the larger FPs that were dominant in the ST. Similar to the situation in the epipelagic and upper mesopelagic, we suggest that although small species are more abundant, they produce small FPs, which sink slowly and are rapidly remineralised. It is likely that it is the less abundant larger carnivores and recyclers in the lower mesopelagic that are contributing more to the flux of carbon to the deep ocean through the production of large FPs, agreeing with the modelling study of Stamieszkin et al. (2015). Calanoid copepod families Aetideidae, Heterorhabdidae, Metridinidae, and Euchaetidae are also common in the mesopelagic of the Scotia Sea and surrounding area (Laakmann et al., 2009; Ward et al., 1995; Ward and Shreeve, 1999), and are of an appropriate size (as adults or other copepodite stages) to produce the larger FPs that were dominant in the ST. Although we can only speculate as to the possible producers of FPs in the ST, it is clear that appropriately sized zooplankton are sufficiently abundant in the mesopelagic to influence the flux of FPs to the ST.

When comparing data sets collected via different methods (in this case Bongo nets, MSC and ST), it is important to consider the different time and space scales over which they measure. The zooplankton Bongo net samples integrated vertically over the top 200 m and temporally over the period over which replicate samples were taken (a few days at each site for both cruises). MSC samples were an instantaneous snapshot of the particle flux and, at a deployment depth of 110 m below the mixed layer, they integrate over spatial scales of tens of kilometres (based on median sinking rates at P2 and P3 and a current speed of 10 cm s-1). Conversely, ST samples captured the flux over a 15 day period and at a deployment depth of 1500 and 2000 m had a potential sample collection area on spatial scales of hundreds of kilometres (based on the same conditions). If zooplankton communities vary significantly over tens of kilometres then this would reduce the direct comparability of MSC and ST data. Previous studies in the region suggest that much of the Scotia Sea is populated by a single zooplankton “community”, but there are regional differences in the stage of phenological development (Ward et al., 2006a), implying that the species composition may not vary on short spatial scales. Changes in the species stage are likely tied to changes in phytoplankton productivity, as for much of the time, Southern Ocean zooplankton are food limited (Ward et al., 2006a). Cluster analysis of phytoplankton in the Scotia Sea reveals distinct communities (in terms of abundance, community structure, and productivity) on spatial scales of hundreds of kilometres (Korb et al., 2012), and hence we would not expect significant changes in the stage-structure of zooplankton on the spatial resolution of the MSC, making these results more comparable to those of the ST. The high sinking rates of zooplankton FP means that their occurrence in ST is representative of the conditions directly above the ST (Buesseler et al., 2007). Slow-sinking particles spread out more as they sink, which increases our uncertainty in depth comparisons of smaller FPs. However, the spatial scale of zooplankton variability at our study site means that slow-sinking FP particles reaching the ST likely reflect the same zooplankton community structure as occurring directly above the ST. For each of our three methods (nets, MSC, and ST), we take averages over multiple years, which should also reduce the uncertainties associated with the various spatial and temporal resolutions of the three methods. However, we acknowledge that the different spatial and temporal scales of measurement could also contribute to some of the vertical changes in FP shape and size structure that we observed.

Our data suggest that zooplankton residing below the euphotic layer repackage sinking detritus and produce FPs, which are able to pass through the lower mesopelagic and be collected in ST in the bathypelagic. Observations made at P2 and P3 in autumn show that, during the night, the highest zooplankton abundances are in the upper 125 m (C. Liszka, personal communication, 2016). However, corresponding daytime surface abundances are typically lower, which may be partially explained by certain species that migrate vertically in the water column (C. Liszka, personal communication, 2016). We suggest that diel vertical migrators may contribute to the relatively fresh FPs we found at depth. A modelling study by Wallace et al. (2013) suggested that FPs penetrate deeper in the water column when there is zooplankton vertical migration, with the deepest FP production occurring when zooplankton undertake diel vertical migrations rather than foray type feeding (multiple ascents and descents during a day). Resident zooplankton populations were observed below 150 m depth, with a peak at 375–500 m, most notably at P3 (C. Liszka, personal communication, 2016), suggesting that the deeper parts of the community, consisting of non-migrators or seasonal or ontogenetic migrators are also important at our study site and could repackage organic material in the upper mesopelagic, and may have produced some of the intact FPs that we observed in our ST.

The abundance of zooplankton typically declines rapidly over the upper 1000 m of the water column (Ward et al., 1995, 2014; Ward and Shreeve, 1999), suggesting that any new FP production below the depth of our MSC samples is likely to take place in the upper to mid-mesopelagic where zooplankton abundances are higher. Zooplankton are more concentrated in the epipelagic; however, the total abundance of zooplankton in the meso- and bathypelagic can be high due to the large depth extent of these layers. In the Antarctic Zone (to the west of our study site), Ward et al. (2014) found that the total depth integrated zooplankton abundance in the 250–2000 m horizon (extrapolating abundances recorded at 750–1000 m down to 2000 m) is about three-quarters (0.74) of the zooplankton abundance in the top 250 m. Therefore it is likely that there is still substantial production of FPs in the lower mesopelagic, and compared to FPs produced in the epipelagic, FPs produced in the lower mesopelagic are subject to remineralisation processes over a shorter distance, and therefore are more likely to reach the deep ocean intact.

Despite the similarities in copepod abundances at P2 and P3, the numbers of FPs collected at P3 were an order of magnitude higher than at P2. Surface phytoplankton productivity at P3 is typically much higher than at P2, with large blooms occurring in most years (Borrione and Schlitzer, 2013; Korb et al., 2008, 2012). This may in part explain higher FP fluxes at the P3 site, as in good feeding conditions (such as those measured during JR304; Belcher et al., 2016b) FP production rates have been shown to be higher (Besiktepe and Dam, 2002; Butler and Dam, 1994). The zooplankton community structure may also affect the fate of FPs in the mesopelagic. Previous studies have found relationships between POC export and the presence of microcopepod species, suggesting that low POC export may be attributed to coprophagy and/or coprorhexy (Suzuki et al., 2003; Svensen and Nejstgaard, 2003). More recently, several studies have proposed that the main role of small zooplankton species may be to fragment FPs rather than ingest them (Iversen and Poulsen, 2007; Poulsen and Kiørboe, 2005; Reigstad et al., 2005). Regardless of the mechanism, previous studies agree that high microcopepod abundances can lead to increased FP retention. The ratio of small copepods to large calanoids is higher at P2 (Fig. 2), which may result in greater losses of FPs in the epipelagic and mesopelagic, resulting in lower numbers of FPs captured in our MSC and ST at P2. Indeed, we see higher attenuation of FP fluxes at P2 than P3 between our measurement depths (Table 2).

The flux of FPs reaching the deep ocean therefore depends not only on surface production but also on the meso- and bathypelagic zooplankton populations and the balance between FP retention and FP production. Our data implies that in situ FP production in the mesopelagic accounted for additional fluxes of FP to the bathypelagic at both P2 and P3. However, as there is the potential for further working, fragmentation and remineralisation of FPs produced in the mesopelagic, the gross deep FP production cannot be quantified here. We therefore cannot determine whether higher FP fluxes at P3 are due primarily to reduced FP attenuation or to increased FP production at depth; most likely a combination of both mechanisms is taking place. Previous work in the region has however found that in the upper-mesopelagic (mixed layer depth to 200 m) FP attenuation is higher at P2 than P3 (Belcher et al., 2016b).

Our comparison of FP size, shape, and abundance in the upper mesopelagic and lower bathypelagic agrees with previous hypotheses (Accornero et al., 2003; Manno et al., 2015; Suzuki et al., 2003), that in situ FP production augments the flux of FPs to depth in the Southern Ocean. We find that the occurrence of intact FPs in deep ST could be explained by both vertical migrations of zooplankton, and repackaging and in situ FP production by meso- and bathypelagic zooplankton populations (Fig. 7). Taking an integrated surface production of 1 g C m-2 d-1 (based on measurements by Korb et al., 2012, to the northwest of South Georgia), and assuming an assimilation efficiency of 66 % (Anderson and Tang, 2010; Head, 1992) during vertical migration (left panel Fig. 7, scenario 1), we calculate that up to 340 mg C m-2 d-1 could reach the depth of migration (this depth will vary both between species and seasonally). In comparison, if FPs are repackaged multiple times on their transit through the mesopelagic then FPs will be assimilated multiple times, resulting in reduced transfer of carbon when compared to diel vertical migration. For example, FPs that are assimilated twice over the same vertical distance as a typical vertical migration (right panel, Fig. 7, scenario 2), result in up to 115 mg C m-2 d-1 reaching the same depth. The exact difference in carbon transfer between these two routes (scenarios 1 and 2) will depend on the number of repackaging steps over the migration depth, specific assimilation efficiencies of the repackaging copepods as well as loss of FP carbon via remineralisation. However, these calculations highlight that the route by which the FPs are transferred to depth is a key control on the amount of carbon reaching depth. Regardless of the feeding mode of these mesopelagic zooplankton communities (detritivory, omnivory, or carnivory), production of FPs at depth via both the aforementioned scenarios supports the transfer of intact FPs to the deep ocean, supporting the sequestration of carbon on long timescales. There is therefore a need to link meso- and bathypelagic zooplankton communities (particularly the larger size classes) to carbon fluxes within global biogeochemical models by refining the contribution of different zooplankton size classes to carbon fluxes via their differential FP production rates and sinking speed.