Impact of a warm anomaly in the Pacific Arctic region derived from time-series export fluxes

Unusually warm conditions recently observed in the Pacific Arctic region included a dramatic loss of sea ice cover and an enhanced inflow of warmer Pacific-derived waters. Moored sediment traps deployed at three biological hotspots of the Distributed Biological Observatory (DBO) during this anomalously warm period collected sinking particles nearly continuously from June 2017 to July 2019 in the northern Bering Sea (DBO2) and in the southern Chukchi Sea (DBO3), and from August 2018 to July 2019 in the northern Chukchi Sea (DBO4). Fluxes of living algal cells, chlorophyll a (chl a), total particulate matter (TPM), particulate organic carbon (POC), and zooplankton fecal pellets, along with zooplankton and meroplankton collected in the traps, were used to evaluate spatial and temporal variations in the development and composition of the phytoplankton and zooplankton communities in relation to sea ice cover and water temperature. The unprecedented sea ice loss of 2018 in the northern Bering Sea led to the export of a large bloom dominated by the exclusively pelagic diatoms Chaetoceros spp. at DBO2. Despite this intense bloom, early sea ice breakup resulted in shorter periods of enhanced chl a and diatom fluxes at all DBO sites, suggesting a weaker biological pump under reduced ice cover in the Pacific Arctic region, while the coincident increase or decrease in TPM and POC fluxes likely reflected variations in resuspension events. Meanwhile, the highest transport of warm Pacific waters during 2017–2018 led to a dominance of the small copepods Pseudocalanus at all sites. Whereas the export of ice-associated diatoms during 2019 suggested a return to more typical conditions in the northern Bering Sea, the impact on copepods persisted under the continuously enhanced transport of warm Pacific waters. Regardless, the biological pump remained strong on the shallow Pacific Arctic shelves.


Introduction
The Pacific Arctic marine ecosystem is extremely productive due to the persistent flow of nutrient-rich waters fueling high primary production on the shallow northern Bering Sea and Chukchi Sea shelves [1]. Recently, the region showed signs of a warming trend, including international permitting was required as part of the sample collection efforts, and the field studies did not involve endangered or protected species. At CEO, year-round measurements capture physical and biogeochemical parameters in addition to measures of fish, zooplankton, marine mammals, and other ecosystem components [e.g. [22][23][24]. Moorings were deployed and recovered from the R/V Sikuliaq in June 2017 and June 2018 and recovered from the R/V Ocean Starr in July 2019. Sediment trap sample cups were filled with filtered seawater adjusted to a salinity of 38 with NaCl and fixed with formalin (4% final solution) to preserve samples during deployment and after recovery. The carousel holding the sample cups rotated at preprogrammed intervals ranging from one week to one month. As the mooring was recovered before the last rotation of the carousel at DBO2 in June 2018, the sample cup that was open upon recovery was excluded from analysis. Seabird SBE 16plus units were deployed to measure water temperature at three depths at DBO2 and DBO3 and at one depth at DBO4. The unit deployed at the upper depth on each mooring was coupled with a WETStar fluorometer or WET Labs ECO triplet fluorometer to measure fluorescence (Table 1).

Laboratory
In the laboratory, zooplankton and meroplankton were removed from subsamples with forceps and identified to the lowest taxonomic level possible using a dissecting microscope. Sample cups were gently mixed before subsamples (0.1-3 ml) were taken with a modified micropipette to enable the collection of large particles for measurements of chl a, algal cells, fecal pellets, TPM, and POC. Subsamples for chl a measurements were filtered onto GF/F filters (0.7 μm), extracted in acetone for 24 h at -20˚C and measured on a Turner Design fluorometer following the methods outlined in Welschmeyer [25]. Samples were kept cool and in the dark prior to chl a measurements, but may have experienced some degradation. For the enumeration of algal cells, subsample volumes were adjusted to 3 ml with filtered seawater before being placed in an Utermöhl chamber. A minimum of 300 algal cells were counted and identified by inverted microscopy at 100X, 200X or 400X depending on cell size according to the Utermöhl method [26]. Using a dissecting scope, the length and width of fecal pellets (broken or intact) were measured with an ocular micrometer and fecal pellet volumes were calculated according to their shape. Cylindrical pellets were attributed to calanoid copepods while ellipsoidal pellets were attributed to appendicularians [27]. Fecal pellet volumes were converted to fecal pellet carbon (FPC) using a volumetric carbon conversion factor of 0.057 mg C mm -3 for copepod pellets and 0.042 mg C mm -3 for appendicularian pellets [27]. Subsamples for TPM measurements were filtered in triplicate onto pre-combusted (500˚C overnight) and pre-weighed GF/F filters (0.7 μm), rinsed with distilled water to remove salt, dried at 60˚C overnight, and weighed on a microbalance. The filters were then exposed to 1N HCl overnight for removal of inorganic carbon and dried once again at 60˚C overnight before encapsulation for POC measurements. POC was measured using a PerkinElmer CHNS 2400 Series II

Sea ice concentrations
While sea ice was first detected in December 2017 at DBO2 and DBO3, a lasting sea ice cover only formed in January 2018 and sea ice concentrations temporarily decreased in February 2018 at both sites (Fig 2).

Water temperatures
Mooring-derived water temperatures were above 2˚C from June to November at all depths sampled at DBO2 and DBO3 (Fig 3).

Chlorophyll a concentrations and fluxes
While enhanced chl a fluxes often coincided with increased suspended chl a concentrations recorded by the fluorometer 6 to 10 m above the traps, there were also notable differences between the two measurements (Fig 4). Peak chl a fluxes were observed at the end of May and/ or early June for every deployment cycle, reaching values >10 mg m -2 d -1 at DBO2 and DBO3.

Diatom fluxes
Temporal variations in fluxes of diatom cells with chloroplasts were remarkably similar to chl a fluxes, with peak diatom fluxes observed at the end of May and/or early June for every deployment cycle (Fig 5). Whereas diatom fluxes were higher at DBO3 than DBO2 during summer 2017, the composition of the fluxes was similar at both sites, with the centric diatoms

Zooplankton and meroplankton
Copepods, copepod nauplii, and meroplankton were the most numerous animals collected in the sediment traps, with the highest abundances observed during June or July at DBO2 and DBO3 and during late September at DBO4 (Figs 6-8). As morphological separation of all life stages of several dominant copepod species in the Pacific Arctic is difficult, most species were aggregated and reported at the generic level (e.g. Calanus glacialis, C. marshallae; Neocalanus plumchrus, N. flemingeri; Pseudocalanus minutus, P. acuspes, P. mimus, and P. newmani), with the exception of N. cristatus that is recognizable due to its large size compared to its congeners [28]. A distinct dominance of the Pacific copepods Neocalanus was observed at DBO2 and DBO3 from June to August 2017 ( Fig 6). The composition of the copepod community shifted to a dominance of Pseudocalanus at all sites during spring and/or summer 2018, with varying abundances of Metridia, Calanus, and Oncaea among sites. Whereas few nauplii of the genera Calanus, Metridia, and Pseudocalanus were collected during summer 2017, large numbers of them were collected during summer 2018 and 2019, especially at DBO3 (Fig 7). At DBO2 and DBO3, nauplii of all stages were often collected simultaneously from May to October, with a larger proportion of nauplii stages N1 and N2 Meroplanktonic stages were dominated by polychaete larvae during summer 2017 at DBO2 and DBO3, while high abundances of bivalve veliger, polychaete larvae, and barnacle larvae blended from spring to autumn 2018 and during spring and early summer 2019 (Fig 8). At DBO4, barnacle larvae contributed to the vast majority of meroplankton collected from August to November 2018, but were absent during June and July 2019 when few bivalve veliger and polychaete larvae were collected.

Particulate organic carbon and total particulate matter fluxes
POC and TPM fluxes displayed similar temporal variations at DBO2 and DBO3 (Fig 9). At DBO2, POC and TPM fluxes increased simultaneously except during spring when TPM fluxes remained relatively low. At DBO3, the highest POC fluxes of the Pacific Arctic region (>2 g m -2 d -1 ) were observed in early June 2018 and from May to July 2019, except for a week at the end of May 2019. At DBO4, peak POC (~1 g m -2 d -1 ) and TPM fluxes (>50 g m -2 d -1 ) were recorded during the second half of May 2019. High POC and TPM fluxes were also observed during autumn and winter at all sites, often in November and/or December. FPC fluxes contributed most to POC fluxes between June and November at all sites, and sporadically contributed to the complete POC flux between August and October 2017 at DBO3 (Fig 9).

Annual fluxes
Annual fluxes of TPM, POC, FPC, chl a, and diatoms all increased during the 2018-2019 cycle at DBO2 (Fig 10). Annual fluxes were also higher during the 2018-2019 cycle than during the 2017-2018 cycle at DBO3, except for a lower annual FPC flux. At DBO4, annual TPM, POC, and FPC fluxes obtained from August 2018 to July 2019 were higher than previously measured under extended sea ice cover at the same site from August 2015 to July 2016 [21], while annual chl a and diatom fluxes were lower during 2018-2019 than during 2015-2016 [21].

Discussion
A combination of downward export, lateral advection, and resuspension of particles contributed to the fluxes recorded at the three mooring sites of the shallow Pacific Arctic region. As typically observed on Arctic shelves [29,30], TPM and POC fluxes were strongly correlated (R 2 = 0.82, p<0.01) at the DBO sites, with POC fluxes consistently contributing to <7% of the TPM fluxes. Whereas enhanced fluxes in April or May most likely resulted from the release of particulate matter from the melting ice [21], algal export contributed to the elevated POC fluxes during May and June. Diatoms usually dominate phytoplankton abundance and biomass in the northern Bering Sea and Chukchi Sea [31][32][33][34] and typically represent the most important component of algal fluxes across the Arctic Ocean due to their rapid sinking velocities [21,30,35,36]. Diatoms also dominated algal export during spring at the three DBO sites, contributing to daily POC fluxes that were among the highest ever documented in the global oceans [19]. Zooplankton continuously collected at fixed depths indicated a strikingly similar composition of the dominant copepods at the three DBO sites, consistent with the large influence of advection on seasonal zooplankton dynamics in the Pacific Arctic [37,38]. Copepod grazing certainly reduced diatom fluxes during summer, as supported by enhanced FPC fluxes. Similar to copepod and copepod nauplii abundances, benthic larvae abundance generally increased after the onset of the algal bloom, with large numbers of bivalve veliger, polychaete larvae, and barnacle larvae contributing to the grazing pressure on diatoms during summer and autumn. The lasting presence of planktonic larvae of benthic animals indicated productive conditions conducive to reproduction during several weeks at all sites. While copepod fecal pellets enhanced POC fluxes during summer and autumn, elevated TPM and POC fluxes during autumn and winter likely resulted from wind-induced resuspension and dispersion of sediments, as current speeds reached velocities >40 cm s -1 during autumn 2017 at DBO2 and DBO3 [19]. Short-lived increases in chl a concentrations, chl a fluxes, and diatom fluxes sporadically observed during late summer and autumn further support wind-induced resuspension and dispersion of sediments containing diatoms and chl a at the DBO sites. These observations are in agreement with enhanced near-bottom chl a concentrations observed during wind-induced high turbidity events during autumn near Bering Strait [39,40], and with low fluxes of the sympagic diatom biomarker IP 25 and of diatoms containing chloroplasts throughout the polar night on the northeast Chukchi shelf [21,41]. In addition to reflecting the seasonal processes influencing the export of particulate matter in the region, the nearly continuous measurements obtained at a high temporal resolution from June 2017 to July 2019 were used to assess the impact of the enhanced transport of warmer Pacific waters and shorter ice-covered duration on the Pacific Arctic marine ecosystem.

June 2017-June 2018
An earlier onset of warming and later onset of cooling in recent years lengthened the warm (�0˚C) water duration to more than 6 months in the Pacific Arctic region [7]. In 2017, the warm period lasted over 7 months and mean June temperatures were 2˚C warmer than climatology when the sediment traps were deployed [7]. A spring bloom probably occurred prior to the onset of sampling at both sites, as indicated by elevated near-bottom chl a concentrations revealing the occurrence of a spring bloom during May 2017 at a nearby mooring site in the southern Chukchi Sea [17,39]. However, relatively high diatom fluxes mostly composed of the exclusively pelagic diatoms Chaetoceros spp. and Thalassiosira spp. indicated that a pelagic bloom occurred in the DBO3 region during late June and July 2017. The subsequent frequent collection of the boreal diatoms Proboscia spp., cosmopolitan diatom T. nitzschioides, and endemic North Pacific diatom Neodenticula seminae [31,33,42,43] during summer and autumn reflected the enhanced inflow of warm Pacific waters across Bering Strait during 2017.
Although zooplankton community composition and distribution have been shown to vary with water properties and the volume of Pacific water transported to the northern Bering Sea and Chukchi Sea [38,[44][45][46][47], the most abundant groups are generally found throughout the region and differences between communities often depend on the presence or absence of indicator species [4]. Accordingly, peaks in copepod abundance dominated by the Pacific copepods Neocalanus during late June and early July 2017 highlighted an enhanced transport of warm Pacific waters, in agreement with concurrent observations of high abundance of Pacific copepods associated to the inflow of Anadyr Water during June and July 2017 in the northern Bering Sea [44]. The very low numbers of nauplii collected during 2017 reflected the low abundance of the three copepod groups for which nauplii were identified in the trap samples. In particular, the absence of Calanus copepodites and nauplii indicated their decline under warm conditions, as supported by the report of a remarkably low abundance of Calanus in the Pacific Arctic during 2017 [1]. The nearly exclusive presence of polychaete larvae during 2017 also suggested their transport with the enhanced inflow of Pacific waters.
Residual heat from the elevated summer water temperatures delayed sea ice formation to December 2017 at both sites [6]. At DBO2, this delay reduced the rejection of salty, cold, and dense brine that typically sinks to the bottom to form a cold water layer, thereby limiting the cooling of the water column during winter and contributing to significant heat input into the Arctic Ocean [13]. At DBO3, oceanic heat loss was sufficient to reduce water column temperature at the freezing point during winter. As oceanic heat flux through Bering Strait triggers the onset of ice melt in the shallow Pacific Arctic region [7], the enhanced heat input during winter presumably contributed to the early ice breakup observed at DBO2 during spring 2018. High chl a and diatom fluxes dominated by the exclusively pelagic diatoms Chaetoceros spp. and Thalassiosira spp. in late May clearly indicated the occurrence of a large open water bloom following the early sea ice breakup in the northern Bering Sea. While biofouling may explain the concurrent low chl a concentrations recorded at DBO2 at the end of May, high chl a and diatom fluxes at that time were in agreement with the peak near-bottom chl a concentrations recorded at a nearby mooring on May 20 2018 [17]. The early collection of low numbers of the ice-associated pennate diatoms Fragilariopsis spp., F. arctica, and Navicula spp. during the sea ice breakup period (sea ice concentration <50%) hinted at a limited ice algae production while ice still drifted in the region in April, inhibiting the seeding of the spring bloom [48] and likely contributing to the shift in the composition of the diatom fluxes. Peak diatom fluxes at the end of May 2018 occurred approximately seven weeks after sea ice breakup at DBO2, likely due to weaker stratification following the extremely low winter sea ice extent and concentration, as observed during July 2018 in the northern Bering Sea [18]. Hence, the early sea ice breakup inhibited the ice-edge bloom and resulted in a large open water bloom in the northern Bering Sea, similar to previous observations following early sea ice breakup in the southeastern Bering Sea [17,49,50]. At DBO3, the coincident sea ice breakup and export of ice-associated pennate diatoms in early May was followed approximately three weeks later by a peak in chl a and diatom fluxes composed of ice-associated and pelagic diatoms during early June, indicating that colder water temperatures, later sea ice melt, and stronger stratification led to an ice-edge bloom in the southern Chukchi Sea in contrast to the northern Bering Sea. The highest nearbottom chl a concentrations recorded in late May and early June 2018 at a nearby mooring in the DBO3 region [17] confirmed that most of the bloom occurred before mooring turnaround in 2018. In response to the enhanced diatom fluxes, the abundances of copepods, copepod nauplii, and polychaete larvae increased at both sites during May 2018.

June 2018-June 2019
Low chl a and diatom fluxes mostly composed of Chaetoceros spp. and Thalassiosira spp. when sampling was resumed during June 2018 at DBO2 and DBO3 indicated a declining pelagic bloom at both sites. From July to September, the centric diatoms Thalassiosira spp. dominated the low diatom fluxes, a sharp contrast with fluxes composed of pennate diatoms Pseudonitzschia/Nitzschia spp., C. closterium, Fragilariopsis spp., F. arctica, and centric diatoms Chaetoceros spp. collected during the previous summer. This shift in flux composition concurs with a lower abundance of Chaetoceros spp. observed during July 2018 than during July 2017 in the Bering Strait region [49]. The large proportion of Thalassiosira spp. in the low diatom fluxes during July 2018 possibly reflected a rapid depletion of nutrients during the large open water bloom taking place in the northern Bering Sea weeks earlier.
Peaks in copepod abundance during late June at DBO2 and early July at DBO3 probably contributed to the low diatom fluxes observed in early summer through grazing pressure. Most of the copepods collected during June and July 2018 consisted of Pseudocalanus while the abundance of Neocalanus and Calanus remained relatively low at both sites. As Pseudocalanus copepods are approximately three times smaller than Neocalanus and Calanus, their dominance supports observations of a zooplankton community dominated by small copepods during 2018 [15]. While small copepods Pseudocalanus, Acartia, and Oithona similis often dominate copepod abundance, large Calanus copepods often dominate biomass in late summer in the southern Chukchi Sea [38]. As C. glacialis females exploit ice algae to fuel their maturation and egg production [51,52], the abundance and biomass of C. glacialis usually increase under colder conditions in the Bering and Chukchi Seas [53,54]. The dominance of small copepods in 2018 therefore resulted from a low abundance of C. glacialis under warmer conditions. The presence of the Pacific copepods Neocalanus, Eucalanus bungii and Metridia pacifica [32,[44][45][46] from June to October 2018 further illustrated the influence of warm Pacific waters.
At DBO4, low diatom fluxes composed of Fragilariopsis spp., C. closterium, Pseudonitzschia/Nitzschia spp., Navicula spp., Thalassiosira spp., and Chaetoceros spp. at the onset of sampling in August 2018 corresponded to the ice-associated and pelagic diatom groups reported to dominate phytoplankton biomass during summer in the northern Chukchi Sea [33,[55][56][57][58][59], albeit with variations in the relative contribution of each group. Despite the absence of measurements prior to August 2018, the distinct composition of diatoms commonly observed during summer near Hanna Shoal suggests that local algal production was not affected by the anomalous 2018 conditions observed south of Bering Strait. Previous diatom flux measurements at the DBO4 site showed very large fluxes of C. closterium from August to October 2015 during a period of strong winds with frequent direction reversals [21]. Similarly, a strong wind event in September 2013 at a nearby site in the Chukchi Sea induced sufficient vertical mixing to enhance phytoplankton productivity and led to high abundances of C. closterium and Leptocylindrus danicus [60,61]. However, although high water temperatures recorded at depth during November 2018 indicated storm conditions mixing the water column at DBO4, C. closterium fluxes remained low during autumn 2018 when sunlight was sufficient to induce primary production in the region. Peaks in copepod abundance recorded during late August and late September 2018 at DBO4 were dominated by small Pseudocalanus, similar to DBO2 and DBO3. The dominance of small copepods in the northern Chukchi Sea also resulted from a low abundance of C. glacialis under warmer conditions, as confirmed by Calanus that were~5 times more abundant at DBO4 during August 2015 than during August 2018 while the abundance of Pseudocalanus was similar during both years [21]. The collection of Pacific copepods E. bungii and M. pacifica during August and September further illustrated the transport of warm Pacific waters into the Chukchi Sea. Most meroplankton collected during summer and autumn 2018 consisted of barnacle larvae, in agreement with observations of barnacle larvae contributing to the largest meroplankton biomass from 2008 to 2010 in the northeastern Chukchi Sea [62].
Despite higher water temperatures during summer and autumn 2018, coinciding with record high surface seawater temperatures in the Pacific Arctic [20], water temperatures rapidly decreased and sea ice quickly formed in November at DBO4 and in early December at DBO2 and DBO3, leading to the prolonged presence of cold water at all sites during winter 2018-2019 and to a later sea ice breakup at DBO2 in 2019. Cells of the ice-associated diatoms Fragilariopsis spp. were first collected simultaneously to sea ice breakup at the end of April at DBO2 and DBO3 and during mid-May at DBO4, indicative of the latitudinal delay in sea ice breakup. Peak diatom fluxes were consistently observed approximately three weeks after sea ice breakup, resulting in a 2-week delay between the diatom bloom in the northern Bering Sea (DBO2) and in the northern Chukchi Sea (DBO4). Regardless of the delay in the onset of the bloom, the composition of the diatom fluxes was strikingly similar at the three sites during 2019, with Fragilariopsis spp., F. arctica, Navicula spp., Pseudo-nitzschia/Nitzschia spp., Thalassiosira spp., and Chaetoceros spp. composing most of the fluxes. These observations indicate that algal production was likely back to normal in the northern Bering Sea.
Whereas Pacific copepods were not collected in 2019 despite the enhanced transport of Pacific waters observed until September 2019 across Bering Strait [7], the sustained inflow of warmer waters resulted in the near absence of Calanus at DBO2 and DBO3 during June and July 2019. While the majority of Calanus copepods present in the Chukchi Sea are believed to originate from the Bering Sea [38], their increasing abundance at DBO4 in July 2019 suggested an alternate source of Calanus on the northern Chukchi shelf during summer 2019 [63]. No clear peaks in copepod abundance were observed during June or July 2019, consistent with high seasonal and interannual variability in the Pacific Arctic region [62]. By contrast, peaks in nauplii abundance recorded during June 2018 and 2019 at DBO2 and DBO3 revealed a steady timing in copepod reproduction despite variations in the timing of sea ice breakup. Finally, the concurring absence of diatoms and copepods of Pacific origin at the DBO sites during 2019 suggested that less productive conditions further south may have limited their advection into the region that year.

Less ice led to generally lower annual export fluxes
A comparison of annual fluxes obtained for each deployment cycle highlights the large-scale impact of an early sea ice breakup on export in the Pacific Arctic region. In the northern Bering Sea, lower annual TPM, POC, chl a, and diatom fluxes during the June 2017-June 2018 cycle than during the June 2018-July 2019 cycle, despite the export of a large pelagic bloom at the end of May and early June 2018, indicated that elevated fluxes were sustained over a longer period when ice breakup occurred a few weeks later during 2019. It is reasonable to assume that annual fluxes would have been higher for the 2017-2018 cycle without the interruption in sampling during June 2018. However, it is unlikely that these annual fluxes would have been higher than during 2018-2019 as decreasing near-bottom chl a concentrations recorded at a close-by mooring indicated a declining bloom during June 2018 [17]. Lower annual chl a fluxes during 2017-2018 may have also partly arisen from the dominance of the generally small-sized Chaetoceros cells with lower chl a content during the 2018 spring bloom. The large abundance of copepodites and nauplii during summer 2018, together with the larger annual FPC flux during 2018-2019, suggest that the massive pelagic bloom enhanced secondary production and grazing pressure in the northern Bering Sea, supporting the hypothesis of a transition to a pelagic-dominated ecosystem under warmer conditions [64]. The dominance of small-sized diatoms and copepods under higher water temperatures and reduced ice cover in the northern Bering Sea concurs with observations in the Atlantic Arctic sector where a shift in the composition of the phytoplankton and zooplankton communities was observed during a period of anomalously warm Atlantic Water inflow and absence of sea ice cover in the eastern Fram Strait [65]. These observations therefore suggest a similar impact of warmer conditions on the Arctic marine ecosystems influenced by the inflow of Pacific and Atlantic waters. Lastly, lower annual TPM and POC fluxes during 2017-2018 along with similar autumn and winter peaks in daily TPM and POC fluxes during both cycles suggest that the intensity and frequency of resuspension events did not significantly increase during the period of unprecedented sea ice loss during 2017-2018 in the northern Bering Sea.
In the southern Chukchi Sea, annual fluxes were also lower during the 2017-2018 cycle than during the 2018-2019 cycle, except for a slightly higher annual FPC flux. The lower chl a and diatom fluxes collected at DBO3 during 2017-2018, despite sea ice breakup actually occurring later during 2018 than during 2019, highlighted the influence of advected waters from the northern Bering Sea. Indeed, a shorter period of lower diatom fluxes at DBO3 during May 2018 possibly resulted from a reduced inflow of nutrients into the southern Chukchi Sea following the massive pelagic bloom that occurred during the prior weeks in the northern Bering Sea. Whereas higher annual FPC flux during 2017-2018 may be the result of larger fecal pellets produced by larger copepods during summer 2017, the highest copepod abundance during summer 2018 also contributed to relatively high FPC fluxes at DBO3 during the 2018-2019 cycle, further supporting a transition to a pelagic-dominated ecosystem under warmer conditions [64]. Although annual TPM and POC fluxes were approximately half as high during the 2017-2018 cycle as during 2018-2019, mostly due to the enhanced daily TPM and POC fluxes sustained during the bloom period from May to July 2019, the high springtime POC fluxes recorded during both years when compared with POC fluxes from around the world [19] indicate a persistently strong biological pump in the region due to the highly productive and shallow nature of the shelves.
By contrast to the DBO2 site where sea ice conditions were heavier during 2018-2019 than during 2017-2018, sea ice breakup at the DBO4 site took place more than two months earlier during 2019 than when the same site was previously sampled during 2016 [21]. Considering that the day of sea ice breakup on the Chukchi shelf ranged from mid-May to mid-August between 2010 and 2017 [66], sea ice breakup occurred relatively early during mid-May 2019 at DBO4. This early sea ice breakup led to lower annual chl a and intact diatom fluxes during 2018-2019 than under longer lasting sea ice cover during 2015-2016 [21]. While Waga and Hirawake [67] reported an increasing occurrence of fall blooms using satellite remote-sensing data from 2003 to 2017 for the Chukchi Sea, the absence of large diatom fluxes during autumn 2018 in contrast to autumn 2015 partly explains the lower annual chl a and diatom fluxes recorded during 2018-2019. However, the lower annual chl a and diatom fluxes observed near Hanna Shoal also resulted from the short-lived and smaller peak in diatom fluxes recorded after sea ice melt during spring 2019, while elevated diatom fluxes were sustained during several weeks in the presence of ice cover during spring and summer 2016 [21]. Therefore, a loss of sea ice resulted in a shorter period of enhanced springtime diatom fluxes during 2019 at DBO4, similar to observations during spring 2018 at DBO2 and DBO3. By contrast, annual TPM and POC fluxes were higher during 2018-2019 than during 2015-2016 at DBO4 [21], principally due to high daily TPM and POC fluxes following sea ice melt during spring and summer 2019. Elevated TPM and POC fluxes at a time of lower chl a and diatom fluxes during summer 2019 suggest that enhanced resuspension under reduced ice cover likely contributed to these fluxes.

Conclusions
Overall, earlier sea ice breakup resulted in shorter periods of elevated chl a and diatom fluxes following sea ice melt at the three DBO sites on the shallow Pacific Arctic shelves. While POC fluxes represent diverse sources of particles, such as material released from the ice, resuspension events, and lateral advection, chl a and intact diatom fluxes more reliably reflect the impact of reduced sea ice conditions on the biological pump. Chl a and diatom fluxes obtained during the two-year sediment trap time-series therefore suggest that the recent loss of sea ice and ensuing weaker stratification reduced the strength of the biological pump in the Pacific Arctic region. In addition, the sustained transport of warm Pacific waters [7] led to the low abundance or absence of Calanus copepods at the three DBO sites during the study period. The abundance of smaller Pseudocalanus copepods instead either increased or decreased FPC fluxes and the efficiency of the biological pump at the DBO sites, highlighting the need for additional export flux measurements to determine long-term trends. As warming of the Arctic and sub-Arctic continues, major ecosystem changes are expected to occur more frequently in the Pacific Arctic region [14]. In this context, sediment trap time-series provided valuable, nearly uninterrupted measurements of biological parameters at a high temporal resolution to complement the other repeated observations made as part of the DBO. Routine sediment trap deployments would help determine if the changes observed under enhanced transport of warmer Pacific waters and shorter ice-covered duration represent an anomalous period or a new normal for the Pacific Arctic marine ecosystem.