The role of sand lances ( Ammodytes sp.) in the Northwest Atlantic Ecosystem: A synthesis of current knowledge with implications for conservation and management

The American sand lance ( Ammodytes americanus , Ammodytidae) and the Northern sand lance ( A. dubius , Ammodytidae) are small forage fishes that play an important functional role in the Northwest Atlantic Ocean (NWA). The NWA is a highly dynamic ecosystem currently facing increased risks from climate change, fishing and energy development. We need a better understanding of the biology, population dynamics and ecosystem role of Ammodytes to inform relevant management, climate adaptation and conservation efforts. To meet this need, we synthesized available data on the (a) life history, behaviour and distribution; (b) trophic ecology; (c) threats and vulnerabilities; and (d) ecosystem services role of Ammodytes in the NWA. Overall, 72 regional predators including 45 species of fishes, two squids, 16 seabirds and nine marine mammals were found to consume Ammodytes . Priority research needs identified during this effort include basic information on the patterns and drivers in abundance and distribution of Ammodytes , improved assessments of reproductive biology schedules and investigations of regional sensitivity and resilience to climate change, fishing and habitat disturbance. Food web studies are also needed to evaluate trophic linkages and to assess the consequences of inconsistent zooplankton prey and predator fields on energy flow within the NWA ecosystem. Synthesis results represent the first comprehensive assessment of Ammodytes in the NWA and are intended to inform new research and support regional ecosystem-based management approaches.


| INTRODUC TI ON
Sand lances and sandeels (Ammodytes sp.) are considered a "quintessential forage fish" in the Northern Hemisphere (Robards, Willson, Armstrong, & Piatt, 1999). Despite their ecological importance, most aspects of their ecology, population dynamics and vulnerability to current and future stressors in the Northwest Atlantic Ocean (NWA) are poorly understood. A few key historical studies of the biology, life history, distribution and ecology of Ammodytes exist in the region. However, recent and projected environmental and ecological changes (Alexander et al., 2018;Saba et al., 2016;Thomas et al., 2017) are making some of this information obsolete as most of it was collected in the 1970s and 1980s (Nelson & Ross, 1991;Scott, 1968Scott, , 1973Winters, 1981Winters, , 1983. The NWA is a highly dynamic ecosystem currently facing myriad impacts from climate change, fishing, aquaculture, oil and gas development, as well as emerging and unknown risks from alternative energy sources (e.g., offshore wind energy facilities), nearshore and shoreline alterations from activities such as sand mining and coastline armouring (Fisheries & Oceans Canada, 2009;Pershing et al., 2015;Saba et al., 2016). Ammodytes may be vulnerable to changes brought about by any one or combination of these anthropogenic threats. Currently, we have an incomplete understanding of what drives high spatio-temporal variability in distribution and density of Ammodytes, which severely limits our ability to make predictions and assess risk. Although we can draw on experiences elsewhere (e.g., North Sea and Pacific Northwest) where the species and its habitats are well studied, the set of challenges to Ammodytes in the NWA region is unique. These knowledge gaps further impede the evaluation of cascading indirect impacts and trophic consequences of Ammodytes variability on predators, co-occurring forage species and their sensitivity to disturbances in the greater community. The objective of this synthesis is to summarize the current state of knowledge and identify information gaps, about the two primary species, A. dubius and A. americanus (hereafter collectively referred to as NWA Ammodytes) throughout their range in continental shelf waters of the NWA from North Carolina (USA) to Greenland. A range of basic and applied questions related to the life history, trophic ecology and vulnerability to growing anthropogenic threats were identified by a diverse working group of scientists, natural resource managers and conservation practitioners from state, federal, academic and non-governmental organizations with interests and expertise in NWA Ammodytes ecology or their predators. Although previous literature syntheses have covered Ammodytes on a global scale (Robards, Willson, et al., 1999) and in the Northeast Atlantic Ocean (Green, 2017), to the best of our knowledge this is the first comprehensive assessment in the NWA region. Results are intended to inform new research, and to help guide conservation and management efforts by regional Fishery Management Councils, regulatory agencies, fishing communities, conservation organizations and coastal development groups, all of whom share responsibilities and interests in NWA Ammodytes and their predators.

| LIFE HIS TORY
The family Ammodytidae contains 33 described species globally, with eight in the genus Ammodytes (Fricke, Eschmeyer, & Van Der Laan, 2019;Orr et al., 2015). The genus Ammodytes is composed of zooplanktivorous fishes that span coastal temperate to polar waters in the Northern Hemisphere. Their occurrence depends on the presence of coarse-grained sand from which they emerge on diel and seasonal cycles to feed in the water column (Auster & Stewart, 1986;Holland, Greenstreet, Gibb, Fraser, & Robertson, 2005;Reay, 1970;Scott, 1968;Wright, Jensen, & Tuck, 2000).
There are two phenotypically similar, congeneric NWA species: the American sand lance Ammodytes americanus (DeKay 1842) and the Northern sand lance A. dubius (Reinhardt 1837). Their distinction is primarily based on differences in distribution, maximum size and a few meristic (i.e., countable) traits. A. americanus primarily occurs in shallow nearshore habitats (<20 m, though often <2 m) from Delaware, USA, to the Labrador coast, Canada (Auster & Stewart, 1986;Nizinski, 2002;Nizinski, Collette, & Washington, 1990). In contrast, A. dubius tends to occur in deeper, more offshore waters (2-100 m, though often > 20 m) between Cape Hatteras, USA, and Greenland (Nizinski et al., 1990). While the smaller A. americanus rarely exceeds 20 cm in standard length, A. dubius appears to be deeper-bodied and can grow to over 30 cm (Nizinski et al., 1990). In addition, A. dubius generally shows higher meristic counts of certain morphological features than A. americanus, with a greater number of plicae (folds of the skin around the lateral musculature) and vertebrae (Nizinski et al., 1990;Scott, 1968). The meristic separation between the two species appears to increase with latitude, thereby aiding identification in the more northern locations of their range. The subtle differences between A. dubius and A. americanus have likely been confused in the literature. For example, Meyer, Cooper, and Langton (1979) referred to Ammodytes on Stellwagen Bank as A. americanus, yet the offshore specimens (and probably nearly all those observed in the study) were A. dubius owing to the depth of the bank and its distance from shore. The difficulty in distinguishing the species morphologically has likely also confounded genetic differentiation between the two species. For example, previous studies suggest that the mitochondrial genomes of A. dubius and A.
americanus are nearly identical, and standard DNA barcoding techniques using the cytochrome oxidase 1 (COI) gene cannot be used to differentiate between them (Horne, Mcbride, Lighten, Bradbury, & Bentzen, 2016;McCusker, Denti, Van Guelpen, Kenchington, & Bentzen, 2013;Orr et al., 2015); however, morphological characters used to identify these specimens were not provided. In contrast, COI sequence data for Ammodytes collected from waters along the east coast of the USA grouped into two distinct clusters (D. Richardson, personal observation). One grouping matched sequences currently reported as A. dubius and A. americanus in public databases (GenBank and Boldsystems). The second group of sequences differed by ~4% from the first, was a close match to A. hexapterus (Ammodytidae) from the Pacific and Arctic, and came from specimens with the meristics of A. americanus. These results suggest that it is possible to genetically differentiate these two species.

| Growth and reproduction
Little is known about the biology and life history of A. dubius and A. americanus in the NWA. A few studies were conducted in the 1970s and 1980s to assess age distributions, growth rates and reproduction, but may now be out of date given the rapid environmental changes occurring in the region during recent decades Pershing et al., 2015;Thomas et al., 2017). Indeed, Ammodytes populations that are in close proximity (<60 km) to each other but experience different ocean temperature regimes can exhibit mark-counter-gradient latitudinal growth (Baumann & Conover, 2011), although the exact mechanisms behind this pattern are not understood (Nelson & Ross, 1991;Winters, 1983).
Both species of Ammodytes in the NWA are gonochoristic (i.e., reproductively distinct between sexes) and exhibit 1:1 sex ratios (Nelson & Ross, 1991). The diameter of oocytes is unimodal for A. americanus, suggesting once-a-year spawning for this species (Westin, Abernethy, Meller, & Rogers, 1979), which is consistent with recent observations for A. dubius from Stellwagen Bank (H. Baumann, personal observation). Both species spawn in fall and winter along the Northeast USA: A. dubius and A. americanus develop ripe gonads in the fall, and larvae are prevalent in the water column throughout the winter and spring (Dalley & Winters, 1987;Nelson & Ross, 1991;Potter & Lough, 1987;Walsh, Richardson, Marancik, & Hare, 2015). Near Greenland, the timing of spawning for A. dubius occurs earlier in the year, likely due to the colder temperatures and a truncated foraging season in this northern region (Danielsen, Hedeholm, & Grønkjaer, 2016). The duration of spawning times of A. americanus and A. dubius is unknown. The historical literature suggests a long spawning season ranging from December through May on the Grand Banks (Dalley & Winters, 1987); however, a recent analysis of A. dubius captured on Stellwagen Bank in 2016 and 2017 indicates that this species has a truncated spawning period, lasting ~1-2 weeks in late November (Murray, Wiley, & Baumann, 2019).

| Early life history
Fertilized eggs of NWA Ammodytes are demersal and adhesive, and are thought to develop on sandy substrates over the course of a two month period . Time to hatch in the wild has been observed in a single study in the Gulf of Alaska at 67 days of total incubation . Laboratory studies revealed that development is highly temperature-dependent for A. americanus and the European congener A. marinus, which can result in interannual and regional variability in hatch phenology for these species (Régnier, Gibb, & Wright, 2018;Smigielski et al., 1984). Yolk-sac larvae begin to appear in ichthyoplankton tows in February in Nova Scotia and in December on Georges Bank, Nantucket Shoals, and Stellwagen Bank, though peak hatching is in January for these more southerly regions (Dalley & Winters, 1987;Potter & Lough, 1987;J. Llopiz, unpublished data). Larvae range from 4 to 7 mm at hatch, and A. americanus larvae in Long Island Sound consume phytoplankton at first feeding before shifting to copepod species such as Temora sp. and Acartia sp. throughout their early life history (Auster & Stewart, 1986;Monteleone & Peterson, 1986). In the laboratory, first feeding in A. americanus can occur up to 16 days post-hatch, indicating they may be resilient to short-term delays in accessing food during the first few weeks of life . NWA Ammodytes larvae live in the water column for the first 3-4 months until reaching sizes of 35-50 mm, at which point they begin to settle into demersal habitats (Auster & Stewart, 1986;Scott, 1973).
Historically, settlement has occurred in May along the Northeast USA and June to July in Nova Scotia (Potter & Lough, 1987;Scott, 1973); however, observed shifts of later winter larval phenology during recent decades in the NWA (Walsh et al., 2015) could affect settlement timing.

| Diet and key prey
The adult diet of Ammodytes in the NWA is poorly known, with most existing studies focusing on A. dubius (Danielsen et al., 2016;Meyer et al., 1979) and the last comprehensive foraging study within Northeast U.S. shelf waters conducted in the late 1970s (Bowman, 2000). Where diets have been characterized, large and energy-rich copepods, primarily members of the genus Calanus, are prominent and thought to affect recruitment success and productivity (Bowman, 2000;Danielsen etal., 2016;van Deurs, van Hal, Tomczak, Jónasdóttir, & Dolmer, 2009;van Deurs, Jørgensen, & Fiksen, 2015;Lindegren et al., 2018;Régnier et al., 2018;Scott, 1973). Of ecological significance, in areas where NWA Ammodytes are particularly abundant, such as Georges Bank, they can exhibit top-down effects on zooplankton, consuming significant proportions of total annual production (Gilman, 1994).
Calanus species, notably C. finmarchicus (Calanidae), were historically abundant in the deep waters (>75 m depth) of the Gulf of Maine (Bigelow, 1926;Durbin, Gilman, Campbell, & Durbin, 1995;Runge & Jones, 2012), where densities have been as high or higher than anywhere across the NWA, even though this area represents the southernmost margin of their subarctic range (Melle et al., 2014). C. finmarchicus exhibits high interannual and seasonal variability in the Gulf of Maine . Historically, older stages were transported from the Gulf of St. Lawrence during summer and fall into the eastern Gulf of Maine, with contributions by the subsurface Labrador Subarctic Slope Water (Head, Harris, & Petrie, 1999;MERCINA Working Group et al., 2001;Record et al., 2019) in the Nova Scotia Current (Appendix S1: Supplemental 1) (Kane, 2007;Pershing et al., 2005).
However, shifts in seasonal oceanographic conditions, circulation and the phenology of lower trophic level species are affecting established patterns in regional timing and availability of resources (see Sections 4.1 and 4.2; Staudinger et al., 2019;Thomas et al., 2017). Production from the Maine Coastal Current also supplied predators, including NWA Ammodytes, in the western Gulf of Maine during summer and fall with lipid-rich older stages of C. finmarchicus (Ji et al., 2017;Runge et al., 2015). In Greenland waters, late-stage Calanus have been found to constitute the majority of the summer diet of A. dubius (Danielsen et al., 2016). Smaller copepods, such as Centropages typicus (Centropagidae), Temora longicornis (Temoridae), Oithona sp. (Oithonidae) and Pseudocalanus sp. (Clausocalanidae), are also known prey of NWA Ammodytes, particularly A. americanus, likely due to overlapping occurrence in coastal habitats (Bowman, 2000).
Bottom-up effects were postulated where significant correlations existed between A. marinus recruitment in the North Sea and the production of Calanus eggs Régnier et al., 2018). Intraguild trophic relationships among forage fishes may also contribute to oscillations in relative abundances (Irigoien & de Roos, 2011). Such dynamics have been suggested in the NWA region where Atlantic herring (Clupea harengus, Clupeidae) and Atlantic mackerel (Scomber scombrus, Scombridae) abundances were observed to oscillate out of phase with NWA Ammodytes during 1969-2010 (Fogarty, Sissenwine, & Cohen, 1991;Richardson, Palmer, & Smith, 2014;Sherman et al., 1981) as well as for capelin (Mallotus villosus, Osmeridae) and A. hexapterus in Alaskan waters . This out-of-phase oscillation is noteworthy given its consistency over such a long time span. Because NWA Ammodytes, Atlantic mackerel and Atlantic herring diets differ appreciably throughout much of the region, competition for prey is unlikely to be the primary driver of this oscillation (Bowman, 2000;Suca et al., 2018). However, when diet overlap is high among zooplanktivorous predators, the prey base can be reduced to the point where all forage fish growth and survival are compromised (Purcell & Sturdevant, 2001). Top-down control is more likely as both Atlantic herring and mackerel are known to prey on larval NWA Ammodytes (Fogarty et al., 1991;Suca et al., 2018; also see Section 3). When Atlantic herring and mackerel are at low population levels, NWA Ammodytes can be released from predation, resulting in a competitive advantage, and vice versa (Polis, Myers, & Holt, 1989). Further, intercohort cannibalism (adult Ammodytes consuming larvae) can occur in regions and years with low abundances of alternative prey (Eigaard et al., 2014, North Sea). Because generalist predators typically consume the most abundant prey available, out of phase cycles of NWA Ammodytes with other forage fishes in the region could have important trophic effects on higher-level predators. Indeed, in past decades predator diets (e.g., Atlantic cod) have mirrored trends in oscillating abundances of NWA Ammodytes and Atlantic herring (Fogarty et al., 1991;Nelson & Ross, 1991;Richardson et al., 2014). Environmental drivers and fishing pressure likely influence these complicated species interactions directly and indirectly ( Figure 1).
The availability of NWA Ammodytes appears to be highly patchy across temporal and spatial scales and differs substantially from that of other forage fishes, in part due to habitat requirements for coarse-grained sandy bottom substrates that allow them to bury and hide from predators (Nizinski, 2002). In comparison, Atlantic herring make broad movements throughout the year and are less confined to a single type of substratum, making them more widely distributed across continental shelf habitats, except during the fall spawning season (Jech & Stroman, 2012;Munroe, 2002). Dependency on sandy substrates leads to high densities of NWA Ammodytes in regions such as the northwest and southwest corners of Stellwagen Bank in the Gulf of Maine. High densities of NWA Ammodytes in a predictable location attracts Richardson et al., 2014) and is likely advantageous to predators, both resident and those that move to occupy such habitats during times of peak food abundance. Further, changes in higher-level predator abundance can create strong top-down pressures that control NWA Ammodytes dynamics in areas where they are concentrated. Predatory release due to overfishing of Atlantic cod (Gadus morhua) and other piscivorous fishes was one explanation for observed population increases in NWA Ammodytes in Canadian waters from 1990 to 2010 (Frank et al., 2013;Frank, Petrie, Fisher, & Leggett, 2011); however, this remains an open question as changes in the vertical distribution of pelagic forage fishes provides an alternative explanation, and gear bias may have confounded interpretation of demographic trends (Jech & McQuinn, 2016;McQuinn, 2009).
Gathering reliable data on NWA Ammodytes abundance, distribution and population dynamics in the region has been difficult.
In addition to the absence of fishery-dependent data, Ammodytes are not caught consistently or detected readily in state and federal bottom trawl survey methods due to the mesh sizes used (Miller et al., 2010;Politis, Galbraith, Kostovick, & Brown, 2014). Their narrow, anguilliform morphology and burrowing behaviour also make Ammodytes japonicus (Ammodytidae) in the Northwest Pacific, however, undergoes aestivation in the late summer through fall prior to spawning, indicating that dormancy strategies are variable according to species and climate (Inoue, 1967;Kuzuhara et al., 2019;Sekiguchi, 1977). A winter dormancy period has been suggested for F I G U R E 1 Diagram of top-down and bottom-up controls on Northwest Atlantic (NWA) Ammodytes. Panels from bottom to top: Environment-Average monthly temperature anomaly (°C) for the Gulf of Maine calculated from NOAA's Extended Reconstructed Sea Surface Temperature V5. Prey-Copepod size index data (small copepod-large copepod; average across Northeast U.S. Shelf) are adapted from Perretti et al. (2017). Intraguild Competition/Predation-Mean standardized anomaly in NWA Ammodytes and Atlantic herring indices are adapted from Richardson et al. (2014). Predation-Cod diet data represent per cent herring and NWA Ammodytes by mass in the diet of cod (with 95% confidence interval) collected by the Northeast Fisheries Science Center Food Web Dynamics Program. Policy and Management-historical management decisions impacting NWA Ammodytes and other forage fishes A. dubius, yet no rigorous study of dormancy timing currently exists in the NWA region (Gilman, 1994). In each case, vigorous feeding prior to dormancy appears to contribute to maturation and survival, thus making this genus potentially vulnerable to changes in the spatio-temporal dynamics of their zooplankton prey (Kuzuhara

| ROLE A S PRE Y
Although Ammodytes are recognized as important forage fish, a comprehensive evaluation of the extent and variation of their role as prey in the diets of higher trophic levels has not been completed for the NWA region. To address this need, we synthesized available literature and diet datasets associated with three major predator groups: fishes, seabirds and marine mammals. Using Web of Science and Academic Search Premier, the scientific and common names of known predator species were searched in combination with "diet" and "prey." In addition, we queried "Ammodytes," "Ammodytes americanus" and "Ammodytes dubius," on their own and in combination with "prey." Studies were reviewed for relevance (i.e., geographical scope), and dietary metrics describing consumption of Ammodytes (e.g., % mass) were compiled. In addition, a query of the long-term NMFS/NEFSC Food Web Dynamics Program database (Smith & Link, 2010; https://inport.nmfs.noaa.gov/inpor t/hiera rchy/1368) yielded detailed information on 40 predatory fishes. Data were identified by searching for all records where "Ammodytes" was identified as a prey item in a predator stomach. Predation was summarized as the relative proportion with 95% confidence intervals (CI) In total, 45 species of fishes, 2 squids, 16 seabirds and 9 marine mammals were reported to consume Ammodytes in the NWA region.
The methodology for assessing predator diets varied from direct observations of stomach contents to visual assessments of prey deliveries and observations of surface foraging behaviours ( Figure 4).
Fish predator data yielded the most quantitative assessments of diet followed by seabird predators. In contrast, information on marine mammal diets was often based on opportunistic assessments and largely qualitative.   In contrast, haddock, windowpane flounder and winter skate consumed very high amounts (>31% M) of NWA Ammodytes during summer.

| Importance to fishes and squids
Geographically, NWA Ammodytes was consumed by the greatest diversity of fishes in Southern New England (N predators = 28) waters. A few notable predators were seasonal migrants to the region (striped bass and bluefish (Pomatomus saltatrix, Pomatomidae)). Most fish diets from TA B L E 1 A summary of all fish predators caught by the Northeast Fisheries Science Center on the Northeast Continental Shelf from 1973 to 2015 and the percentage by mass (% Mass) and 95% Confidence Intervals (CI) of their overall diets found to contain Northwest Atlantic Ammodytes  Exact dietary values and other study details can be found for fishes and squids in Appendix S1: Supplemental Table 6, for seabirds in Appendix S1: Supplemental Table 7, and for marine mammals in Appendix S1: Supplemental Table 9 species) compared to more recent decades (N 2000s = 36 predators; Predator-prey body-size data were available for 35 fishes in the NMFS trawl series (Table 1). The mean size ± SD of NWA Ammodytes consumed by all predators across all years was 117.3 ± 46.7 mm.
Based on size-at-age estimates from Monteleone and Peterson (1986) and Nelson and Ross (1991), the majority of NWA Ammodytes  Ammodytes in the region (Figure 4; Appendix S1: Supplemental 6).
American plaice in Newfoundland waters also showed historically high amounts (16% FO) of A. dubius during the 1990s (Zamarro, 1992), as well as pollock (7%-13% FO) in the Bay of Fundy in the 1950s and 1960s (Carruthers et al., 2005). Several studies determined that bluefin tuna, especially smaller and younger fish (Logan et al., 2011(Logan et al., , 2015, relied heavily (up to 69% M of their diet) on NWA Ammodytes during summer and fall in the Mid-Atlantic Bight, Southern New England, Gulf of Maine and Georges Bank regions during the late 1980s to early 2000s (Chase, 2002;Logan et al., 2011Logan et al., , 2015, while low amounts (<5% M) were found in sympatric yellowfin tuna in the early 2000s (Teffer et al., 2015).
In summary, NWA Ammodytes were found in the diets of several fishes of high conservation concern including Atlantic cod, Atlantic

| Importance to seabirds
A total of 16 species of seabirds including terns, alcids, gulls, cormorants, murres, shearwaters, gannets and some ducks were reported to consume NWA Ammodytes in notable amounts, according to published (Appendix S1: Supplemental 7) and unpublished (Appendix S1: Supplemental 8) sources. This was either as adults or as provisioned to chicks along the eastern coasts of the United States and Canada (Figure 8).  (Goyert, 2015).  Ammodytes provide a highly nutritional source of lipids and proteins to seabirds (and other predators). Post-larval NWA Ammodytes showed higher caloric content than capelin (Baillie & Jones, 2003) and Atlantic herring (Hislop, Harris, & Smith, 1991)  where YOY and juveniles are more likely to occur (Breton & Diamond, 2014;Chapdelaine & Brousseau, 1996). Small-sized Ammodytes, although of lesser nutritional value compared to larger individuals, may be selected because of their availability, size or morphology; smaller individuals are also easier for adults to carry and for chicks to swallow (Bradstreet & Brown, 1985;Burger & Piatt, 1990;Burke & Montevecchi, 2008;Gaston & Woo, 2008).
Ammodytes role as prey in the NWA ecosystem has important implications for seabirds of conservation concern. State wildlife manage- Abemayor, unpublished data; Goyert, 2015). As Ammodytes specialists, roseate terns show limited flexibility in their foraging strategies, which makes them particularly vulnerable to changes in availability (Goyert, 2015). During the breeding season, feeding areas have been documented within 10-30 km of breeding colonies (Heinemann, 1992), but recent tagging studies suggest that adults may travel as far as 50 km to find suitable prey (Loring et al., 2019). Productivity and chick survival rates of roseate terns have been attributed primarily to the availability of high-quality prey (Kirkham, 1986), and thus, changes in the abundance and distribution of NWA Ammodytes relative to breeding colonies and staging locations could significantly affect their population dynamics and breeding success if they are not able to exploit an alternative prey source of comparable nutrition. In addition, direct links between dietary importance and productivity have yet to be quantified explicitly in the region.  (Kenney, Payne, Heinemann, & Winn, 1996;Selzer & Payne, 1988 MacLeod, Santos, Reid, Scott, & Pierce, 2007), and starvation events coincided with Ammodytes declines (but see Thompson et al., 2007). Eastern Canada during the early 1990s were linked to forage availability, including NWA Ammodytes (Stevick et al., 2006). Humpback whale calf survival after weaning in the Gulf of Maine has also been linked to the availability of NWA Ammodytes and Atlantic mackerel (Robbins, 2007). Information on other large whales was indirect and derived from relating whale occurrences with known locations of NWA Ammodytes or their habitat (Overholtz & Nicolas, 1979;Payne et al., 1986Payne et al., , 1990Weinrich, Martin, Griffiths, Bove, & Schilling, 1997).

| Importance to marine mammals
One of the only areas in the NWA with direct diet data for minke whales is in Greenland. NWA Ammodytes were found in 92% FO of whale diets from offshore areas of West Greenland in the 1980s; however, they were less important during the early 1990s and replaced by capelin (Neve, 2000). In one study of individually identified minke whales off the coast of MA, sightings were less common in years when local NWA Ammodytes abundance was low, providing indirect evidence of their possible dietary importance (Murphy, 1995). Evidence from other areas of the world (e.g., Japan, Iceland and Norway) shows that minke whales consume adult Ammodytes (Lydersen, Weslawski, & Øritsland, 1991;Murase et al., 2009;Sigurjónsson, Galan, & Víkingsson, 2000;Tamura et al., 2009;Víkingsson et al., 2015). In waters off Scotland, minke whale diets contained high amounts (66% Number, 62% M) of A.
marinus (Pierce, Santos, Reid, Patterson, & Ross, 2004), and their distribution in surrounding waters corresponded to Ammodytes-associated habitats (Macleod et al., 2004;Olsen & Grahl-Nielsen, 2003). A body condition study off Iceland did not successfully link minke whale blubber thickness with trends in Ammodytes abundance, but this was considered likely due to there being multiple prey species of importance (Christiansen, Víkingsson, Rasmussen, & Lusseau, 2013).

| Interspecific interactions
Within the broader NWA food web, the predictability and persistence of forage species such as Ammodytes is highly important to the multiple predator groups that specialize on them (e.g., Atlantic sturgeon, roseate terns and harbour seals; Figure 4, Appendix S1: Supplements 6-9). Interspecific interactions may provide additional opportunities for socially flexible and opportunistic predators. The pelagic realm is a dynamic environment where prey aggregations (e.g., bait balls) can attract multiple predators that engage in facilitative (e.g., commensal) or competitive feeding frenzies (Goyert et al., 2018;Goyert, Manne, & Veit, 2014). During these social interactions, predatory fishes (e.g., tunas) and marine mammals can drive prey upwards towards seabirds feeding at the air-sea interface (Safina, 1990;Veit & Harrison, 2017). In addition, fish, seabird and marine mammal predators may

| Changes in regional climate and oceanographic patterns
The NWA region is experiencing rapid warming due to climate change, with rates as high as 0.4-0.3°C per decade since the 1980s

| Fluctuations in primary prey resources
Since 2010, evidence points to a shift in the external source that supplies C. finmarchicus, NWA Ammodytes' primary prey during the spring season, into the Gulf of Maine as more warm and saline (Calanus-poor) Atlantic slope water and less cold (Calanus-rich) Nova Scotia shelf current enters at depth through the Northeast Channel .  .
In addition to changes in supply and transport, C. finmarchicus abundance is also driven by changes in the timing and magnitude of local primary production. One source of variability is the match or mismatch of food available to C. finmarchicus emerging from diapause in spring, which has already advanced in time, and is predicted to continue advancing, due to earlier spring warming (Maps et al., 2011;Pierson, Batchelder, Saumweber, Leising, & Runge, 2013).
Shifts in the timing of seasonal events during recent decades are linked to changes in primary productivity and growth cycles leading to large earlier cohorts of C. finmarchicus, despite reduced overwintering stock Staudinger et al., 2019). Changes in spring production affect the supply of C. finmarchicus in subsequent seasons, particularly on Stellwagen Bank, Georges Bank and elsewhere in coastal waters of Southern New England (Greene & Pershing, 2007). Whether this favourable match between primary and secondary production in the western Gulf of Maine continues in the future is uncertain. Long-term habitat modelling (that does not take into account advective supply) suggests a long-term decline in regional C. finmarchicus abundance (Grieve, Hare, & Saba, 2017). As C. finmarchicus represents the primary source of lipids to pelagic consumers in the Gulf of Maine, a reduction in its availability may have consequences not only for NWA Ammodytes but also for the broader regional food web .

| Climate change impacts
NWA Ammodytes have been ranked as "moderately vulnerable" to climate change relative to 81 other marine fishes and invertebrates along the Northeast USA (Hare et al., 2016). Under a high emissions scenario (RCP 8.5) for the time period of 2005-2055, NWA Ammodytes are expected to have high climate exposure from increasing sea surface temperatures (high-very high exposure), ocean acidification (very high) and sea level rise (high), among other factors (Hare et al., 2016). Biological and ecological attributes that influence their sensitivity to climate impacts confer moderate-to-high restrictions on mobility, and moderate habitat specificity. This is due to their strong association with sandy sediments, often with patchy and ephemeral distribution, located in relatively shallow water depths of <100 m. It remains unresolved how changes in coastal hydrology could impact habitat suitability, particularly for A.
americanus. Aspects of their spawning cycle, early life history and sensitivity to increasing temperatures also influence their moderateto-high climate vulnerability ranking (Hare et al., 2016). However, projections of future variability and long-term changes in circulation are uncertain (Brickman et al., 2018). Similarly, acidification trends of Gulf of Maine waters and parts of the greater NWA shelf have so far been masked by the unusually strong temperature and salinity increases (i.e., lower CO 2 solubility and higher alkalinity, buffering capacity; Salisbury & Jönsson, 2018). A reversal of these decadal trends, combined with predicted increases in freshwater input, could lead to a more rapid acidification of NWA coastal shelf waters in the coming decades, as compared to current model predictions for this region (~−0.35 pH units by 2099, Bopp et al., 2013).
Emerging research on A. dubius indicates that this species exhibits biological characteristics that may make them particularly vulnerable to ocean warming and acidification, more so than previously thought. As a fall/winter spawner, NWA Ammodytes release embryos and larvae into cold and cooling water; warmer temperatures in fall may affect hatch timing and survival characteristics with uncertain consequences (Laurel, Hurst, Copeman, & Davis, 2008 (Nye, Link, Hare, & Overholtz, 2009;Walsh et al., 2015). These studies provide important evidence and insights on the sensitivity and adaptive capacity of species to climate impacts (Beever et al., 2016). The centre of biomass of 24 out of 36 (67%) fish stocks generally shifted either poleward and/or to greater depth in the NWA Shelf ecosystem (Nye et al., 2009). Catch data from the NMFS bottom trawl survey (Pinsky, Worm, Fogarty, Sarmiento, & Levin, 2013) 1977-1987 and 1999-2008 showed a significant shift in spatial distribution and seasonal phenology of larval Ammodytes in the region (Walsh et al., 2015). Timing of larval NWA Ammodytes occurrence was significantly later during winter and showed a simultaneous spatial shift in distribution northward and to deeper waters along the continental shelf; changes in the distribution of adult NWA Ammodytes were also evaluated but showed no trend (Walsh et al., 2015), possibly due to the high dependency of older life stages on specific bottom substrates.  (Morley et al., 2018).
Collectively, these results suggest that larval stages of Ammodytes in the NWA are responding to changing environmental conditions. The lack of evidence for shifts in distribution, range and phenology of adults could be a function of stochasticity or data scarcity. Alternatively, adult NWA Ammodytes may lack the capacity to adapt their distribution or behaviour to track optimal conditions, or perhaps they are adapting in place (Parmesan, 2007). If NWA Ammodytes are not able to keep pace with changing environmental conditions at any life stage, they will be at increased risk of population declines given projections of declining suitable habitat (Morley et al., 2018).

| Fisheries
There as such, there are concerns by the conservation and management community an Ammodytes fishery could still be pursued in the NWA as a replacement species (e.g., for bait) or new industry (e.g., fish oil), and if this happened there would be detrimental consequences on the broader community of predators known to rely on them as prey.
Stock assessments of species like Ammodytes that undergo boom and bust cycles in a Maximum Sustainable Yield context require understanding the implications of highly stochastic recruitment dynamics (Arnott, Ruxton, & Poloczanska, 2002;Deyle et al., 2013).
This is difficult to estimate and subject to high uncertainty; therefore, precautionary and ecosystem-based management approaches that account for direct and indirect sources of fishing and other mortality will be important (see Section 5 of this synthesis  (Tien et al., 2017). Strong associations with certain bottom substrates increase the risk of Ammodytes to localized depletion, particularly when capture rates remain high during low abundance years in areas where fish exhibit dense aggregations (Csirke, 1988). Bottom fishing gear can also disturb the demersal eggs of Ammodytes during developmental periods and older individuals during winter dormancy when they are buried in the sediment.
Recently, the Mid-Atlantic Fishery Management Council has taken steps to move towards an ecosystem-based approach to fisheries management. For example, they recognized forage species as key components of the regional ecosystem and increased protections by Fishery Conservation and Management Act to strengthen key protections and promote responsible management of forage fishes nationally. All of these efforts could benefit regional populations of NWA Ammodytes, though indirect effects (e.g., to intraguild feedback loops with Atlantic herring) can be complicated and have unintended outcomes that require regular data collection to determine outcomes and effectiveness (Fogarty et al., 1991;Sherman et al., 1981).

| Energy development and resource extraction
During recent decades, energy development and exploration in the U.S. Atlantic has increased, at least in part as a result of U.S. policy. In an effort to identify shared risks and opportunities across seascapes, regional ocean partnerships such as the Northeast Regional Ocean Council ( Offshore infrastructure development and construction projects can transform the coarse-grain sediment habitats that Ammodytes rely on into artificial reef habitat that supports hard-bottom associated communities (Lindeboom et al., 2011). This would exclude Ammodytes from previously occupied areas. Indirect effects of these activities may also impact Ammodytes; for example, it has been hypothesized that artificial reef effects lead to an increase in predators that exert additional topdown pressure on Ammodytes (e.g., by Atlantic cod; Lindeboom et al., 2011). The cumulative effects from multiple closely located development sites would pose additional risk (Leonhard, Stenberg, & Støttrup, 2011). The development has the potential to act at different time scales by altering Ammodytes habitats immediately through direct disturbance, and then incurring lagged ecological effects as the community stabilizes (Gray, 2006;Petersen & Malm, 2006).
Alternatively, some development activities could have neutral or positive effects. In the North Sea, the presence of A. marinus and A.
tobianus was assessed before and after the construction of wind turbines. Short-term effects of the facilities were either not observed (Lindeboom et al., 2011) or showed increases in density at wind development sites after construction (Degraer, Brabant, Rumes, & Vigin, 2016;Leonhard et al., 2011;Stenberg et al., 2015;van Deurs et al., 2012;Vandendriessche, Hostens, Courtens, & Stienen, 2011). Positive shortterm effects were attributed to increased or neutral effects on sediment quality, increases in juvenile abundance, associations of midwater feeding schools with structure and/or reductions in predators during construction (Leonhard et al., 2011;van Deurs et al., 2012). Reductions in fishing activities during construction and operation of wind farms also have the potential to benefit Ammodytes, though this could lead to localized increases in competitors and predators that experience concomitant reductions in fishing mortality. Alternatively, these patterns could be due to, or in synergy with, the aggregation effects of the offshore structures through flow or shade effects. No long-term effects on Ammodytes were seen in such areas, despite an increase in species diversity due to artificial reef effects (Degraer et al., 2016;Leonhard et al., 2011;Stenberg et al., 2015;van Deurs et al., 2012).  (Staudinger, 2011). NWA Ammodytes live within or near sandy bottom habitats, which generally do not retain contaminants as much as silty, muddy habitats. This aspect of their life history plus their relatively low position in the food chain likely limits exposure. However, following oil spills and other chemical disasters, Ammodytes have accumulated toxic chemicals from fuel oils and dispersants (Calbet, Saiz, & Barata, 2007;Hansen, Altin, Olsen, & Nordtug, 2012). Such exposure can have immediate lethal or long-term chronic sublethal effects on Ammodytes and their predators (Hjermann et al., 2007). Negative effects of oil spills on Ammodytes include physiological haemorrhaging when Ammodytes burrow into oil-contaminated sands and reduced time buried in oil-contaminated sand, which increases exposure to predators (A. hexapterus, Pearson, Woodruff, Sugarman, & Olla, 1984;Pinto, Pearson, & Anderson, 1984), as well as mass die-offs in response to oil and/or the subsequent detergents that are used to contain or clean up spills (e.g., Torrey Canyon clean up in the United Kingdom; Reay, 1970;Simpson, 1968). Additional toxic burdens (e.g., paralytic shellfish poisoning associated with red tides) have led to mass mortalities of higher trophic level predators of conservation and management concern (Jessup et al., 2009).
The interactive impacts of a changing climate and other anthropogenic stressors on Ammodytes are largely unknown, but as a key prey species, these cumulative effects warrant careful consideration.

| ECOSYS TEM S ERVI CE S AND ECOSYS TEM -BA S ED MANAG EMENT
In marine systems, increasing attention has been focused on characterizing the direct and indirect ecosystem services provided by forage fishes (Alder et al., 2008;Pikitch et al., 2004). As demonstrated by this synthesis, NWA Ammodytes provide extensive support to higher trophic levels through energetic transfer. Many of the 72 predators that rely on NWA Ammodytes directly contribute to regional economies and food resources through commercial and recreational fisheries, as well as cultural and other recreational benefits such as tourism and viewing activities (Nelson et al., 2013). The bottom-up contributions of NWA Ammodytes to the landings of commercially exploited fish predators in the NWA are likely substantial. Their indirect value may outweigh any commodity services arising from direct harvest and sale of NWA Ammodytes if a large-scale commercial fishery was ever pursued in the region. A formal valuation analysis of existing supporting and projected commodity services (e.g., based on harvest targets), derived from NWA Ammodytes, would be an important preliminary step to assess potential trade-offs, stakeholder conflicts and competing demands within the NWA region (Hunsicker et al., 2010;Koehn et al., 2017).
Ecosystem services are crucial for developing ecosystem-based management (EBM) and ecosystem-based fisheries management (EBFM) plans and are increasingly being integrated into sustainability efforts in the world's oceans (Altman, Boumans, Roman, Gopal, & Kaufman, 2014;Francis, Hixon, Clarke, Murawski, & Ralston, 2007;Patrick & Link, 2015;Ruckelshaus, Klinger, Knowlton, & DeMaster, 2008). EBFM approaches to resource management incorporate interactions among species (e.g., predator-prey, competitive) and with their environment (e.g., climate change), account for direct and indirect effects of human activities and view humans as an integral component of ecosystems (Boumans, Roman, Altman, & Kaufman, 2015;Patrick & Link, 2015;Van Dyne, 1969). Important Minimum realistic models (MRMs) that account for natural predation and fishing mortality rates provide an example of how ecosystem considerations have been approached in the NWA region (Gamble & Link, 2009;Link et al., 2011). This type of model has been used to answer research questions pertaining to some forage fishes including Atlantic herring, Atlantic mackerel and butterfish; however, they have yet to be directly linked to species stock assessments or management actions. To date, the role of NWA Ammodytes as prey has not been explicitly accounted for in stock assessments, despite their apparent importance in years of low Atlantic herring and mackerel abundance (Gamble & Link, 2009;Link & Sosebee, 2008;Moustahfid, Link, Overholtz, & Tyrrell, 2009;Overholtz, Jacobson, & Link, 2008;Overholtz & Link, 2006;Tyrrell, Link, & Moustahfid, 2011 (Boumans et al., 2015). Efforts are underway to incorporate environmental variables and the effects of climate change into projections of food web dynamics over the coming century. However, NWA Ammodytes remain underrepresented in many regional EBM plans, and their explicit consideration would improve overall understanding of ecosystem dynamics in the NWA.
Scaling reference points based on confidence in scientific knowledge and assessments is recommended (Pikitch et al., 2012) and currently used to set regional catch limits. Since NWA Ammodytes are unmanaged forage fishes, information on their biology and population dynamics is woefully inadequate compared to managed species (e.g.,

Atlantic herring). Lastly, the Magnuson-Stevens Fishery Conservation
and Management Act identifies the maintenance of forage fishes for all components of the ecosystem as an important consideration in setting optimal yields for harvested species. In support of these conservation and management goals, we conclude this synthesis by outlining remaining gaps in knowledge and high-priority basic and applied research needs for NWA Ammodytes populations in the NWA region.  (Green, 2017;Wanless, Harris, Newell, Speakman, & Daunt, 2018). Statistical and ecological modelling studies have investigated the trophic implications of long-term changes in Ammodytes population size structure, abundance and nutrition, revealing direct links to seabird breeding success at multiple sites in the North Sea (Frederiksen, Wanless, Harris, Rothery, & Wilson, 2004, Frederiksen, Elston, Edwards, Mann, & Wanless,2011Wanless, Wright, Harris, & Elston, 2004;Wanless et al., 2018).

| A PATH FORWARD: RE S E ARCH NEEDS
Similar to the NWA, the North Sea is experiencing rapid warming due to climate change (Rutterford et al., 2015), which has been linked to several detrimental effects on Ammodytes populations. Warming has been associated with increases in the metabolic rate of YOY Ammodytes and subsequent reductions in growth and energy reserves, which compromise their ability to attain adequate body sizes needed to survive overwintering (van Deurs, Hartvig, & Steffensen, 2011). Increasing temperatures have also been shown to reduce the reproductive potential of adults (Wright, Orpwood, & Scott, 2017). In addition, climate change is altering the availability of key Ammodytes prey in the North Sea where C. finmarchicus is being replaced by a warmer water conspecific, C.
helgolandicus (Calanidae). Calanus helgolandicus differs in phenology, size, and nutrition (Frederiksen et al. 2011, van Deurs et al., 2009van Deurs et al., 2015), which has implications for energy flow to higher-level predators (von Biela et al., 2019). Examples from the Northeast Atlantic presented here and throughout this synthesis show how the competing demands from the commercial fishery relative to fish, marine mammal and seabird predators can be managed (Furness, 2002) when Ammodytes are explicitly considered. Further, regional fisheries management in the North Sea provides guidance on the tools and data needed to advance research and conservation in the NWA.
An immediate and high-priority need in the NWA is to resolve long-term patterns and drivers of Ammodytes abundance and distribution. Ichthyoplankton surveys (e.g., NEFSC Ecosystem Monitoring (EcoMon)) have effectively tracked the distribution and abundance of early life stages of NWA Ammodytes, as well as co-occurring plankton and oceanographic conditions (Sherman et al., 1981;Walsh et al., 2015). Continued support for this programme is imperative to prevent data gaps in one of the few long-term time series of NWA The sensitivity and adaptive capacity of NWA Ammodytes to climate change remains an area of high uncertainty due to a lack of regional and species-specific studies. It is currently unknown if A.
dubius and A. americanus are exposed and respond to environmental stressors equally. Climate change impacts on Ammodytes have been reasonably well investigated in the Pacific Northwest (von Biela et al., 2019;Robards, Anthony, Rose, & Piatt, 1999;Robards et al., 2002) and the North Sea von Biela et al., 2019;Burthe et al., 2012;Burthe, Wanless, Newell, Butler, & Daunt, 2014;Wanless et al., 2004Wanless et al., , 2018. In contrast, very few studies have directly evaluated climate impacts on Ammodytes in the NWA (Danielsen et al., 2016;Dixon, Dempson, Sheehan, Renkawitz, & Power, 2017). Preliminary evidence suggests A. dubius may be a critical indicator of climate change and system thresholds. Early-stage development appears to be highly sensitive to ocean acidification and temperature (Murray et al., 2019) and is likely to affect other aspects of life history that influence vulnerability (Hare et al., 2016).
There is strong concern about climate-induced shifts in C. finmarchicus distribution in northern areas of the NWA .
Ammodytes populations that are heavily reliant on Calanus sp. are expected to be relatively more vulnerable to changes in availability and nutrition, particularly during spring (Friedland et al., 2015;Morse, Friedland, Tommasi, Stock, & Nye, 2017;Thomas et al., 2017), and may be indicators of shifting ecosystem dynamics and energy transmission processes, as has been suggested for the Northeast Pacific congener A. personatus (von Biela et al., 2019). Climate-induced changes in the distribution and availability of C. finmarchicus could also intensify competitive interactions between NWA Ammodytes and planktivorous whales such as the critically endangered North Atlantic right whale (Payne et al., 1990). Although orders of magnitude different in size, the presence of NWA Ammodytes as well as other forage fishes (e.g., herring, mackerel) inhibits their feeding behaviour.
Current management measures, including the possession limit implemented in the Mid-Atlantic region in 2017, and small-mesh and exempted fishery regulations in New England, have likely kept regional fishing mortality rates on NWA Ammodytes low; however, their designation as an unmanaged forage species and absence of a species stock assessment has, until recently, precluded the acquisition of basic biological data, assessments of mortality rates and accurate quantitative population assessments. Based on what is known from systems outside the NWA (e.g., North Sea), Ammodytes may be highly vulnerable to overfishing. The sandy substrates preferred by NWA Ammodytes are relatively resilient to physical disturbance (Auster & Langton, 1999) and may be repopulated after acute fishing or construction events.
However, timelines to recovery and further consequences are not well understood (Green, 2017;Wanless et al., 2004Wanless et al., , 2018. Catches could initially remain high, even after repeated fishing attempts in the same area, while inflicting long-term impacts, such as increased dispersal and exposure to predation as fish relocate in search of undisturbed habitat.
In addition, if fishing, dredging, sand mining or offshore development activities occur during spawning periods, these disturbances could disrupt early life history through damage to eggs laid in/on the substrate.

| CON CLUS IONS
This synthesis provides a comprehensive summary of the current state of knowledge of Ammodytes populations in the NWA. A diverse set of at least 72 species of predators were found to rely on NWA Ammodytes as prey. Collectively, these results show that changes in the availability and distribution of NWA Ammodytes could affect numerous regional species that are highly valued as commercial fisheries (e.g., bluefin tuna, Atlantic cod), as endangered species (roseate terns, Atlantic salmon, Atlantic sturgeon), and iconic wildlife that support cultural and recreational activities throughout the region (e.g., humpback whales, Atlantic puffins). The amassed data are readily available to calculate key metrics, populate initial models and facilitate broad-scale assessments of NWA Ammodytes, their dependent predators, and linked human systems (Smith et al., 2011). Ecosystem (e.g., Ecopath with Ecosim, Atlantis) and empirical dynamic models are potential tools to explore the connections and consequences of previously unresolved community changes (e.g., top-down versus bottom-up controls; intraguild competitive relationships) and disturbance scenarios (e.g., climate and fishing levels) on NWA Ammodytes populations (Glaser et al., 2014;Klein, Glaser, Jordaan, Kaufman, & Rosenberg, 2016;Plagányi & Essington, 2014). However, reliable estimates from any new research initiatives are dependent on filling the remaining data gaps outlined here. Several predator groups require expanded diet data to fully comprehend their dependence on NWA Ammodytes, including all marine mammals, adult seabirds, estuarine and inshore fish predators, and small pelagic/intraguild competitors. Updated evaluations are also needed to understand ecosystem changes occurring during the most recent two decades (2000-present) that capture potentially unprecedented changes in trophic interactions due to rapid warming in the region (Saba et al., 2016;Thomas et al., 2017). NWA Ammodytes have been consistently abundant and were consumed by the greatest diversity of fish predators in the Southern New England region, making this an ideal focal area for targeted sampling and analyses of population size structure and related changes in predatory demand and energy transfer.
Paramount to resolving almost all of the remaining questions outlined in this study is the need for information on the underlying environmental and ecological factors driving NWA Ammodytes' spatial and temporal variability over multiple scales. Retrospective analyses of the conditions surrounding periods of peak abundance (e.g., the early 1980s, and around 2010) and at known sites of locally high abundance (e.g., Stellwagen Bank) could provide important insights. In addition, new data collected from alternative and novel approaches such as hydroacoustic surveys , geospatial analytical techniques , composite indices and predators as biological samplers (Piatt et al., 2018;Richardson et al., 2014) could address vertical and horizontal availability over diel, seasonal and interannual scales.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are publicly available in National Oceanic and Atmospheric Administration (NOAA)