A NOAA Fisheries science perspective on the conditions during and post COVID-19: Challenges, observations, and some possible solutions, or why the future is upon us

, Abstract Many fisheries and marine science organizations are working to determine how to meet their missions in the midst of the COVID-19 outbreak. As such, it seems prudent to exchange ideas, share knowledge, and initiate a discussion amongst us. As the scientific leadership team for NOAA Fisheries, we wanted to offer some perspectives. Others are also evaluating the impacts of COVID-19 but from the perspective of addressing tactical, day-to-day concerns of restarting operations for various marine and fisheries-oriented organizations. Thus, it seemed appropriate to us to explore the potential challenges posed by COVID-19, and to purposefully ascertain if there are strategic opportunities for improving how we conduct our operations. We need to find ways to mitigate the effects of COVID-19 on our mission and also to glean information from our responses while in the midst of the crisis. We offer some recommendations to that end, and offer these thoughts not as having solved every problem, but to learn from each other, compare across organizations, and engage in dialogue within our discipline to advance much-needed changes.

. As such, it seems prudent to exchange ideas, share knowledge, and foster a discussion amongst us. Others are also evaluating the impacts of COVID-19 in a marine and fisheries context (Kemp et al. 2020;Bennett et al. 2020) but from the perspective of addressing tactical, day-to-day concerns of restarting operations for various marine and fisheriesoriented organizations (Doremus 2020, FAO 2020b. Thus, it seemed appropriate from our vantage point within the NOAA Fisheries' science enterprise to explore the potential challenges posed by , and to ascertain if there are strategic opportunities for improving how we conduct our operationsin the spirit of Winston Churchill, "never let a good crisis go to waste." Ultimately, we need to find ways to mitigate the effects of COVID-19 on our mission. We also need to glean information from our responses while in the midst of the crisis to take advantage of lessons learned that could improve how we conduct our mission moving forward (NOAA 2009, NMFS 2019, Doremus 2020).
The COVID-19 situation is unprecedented, at least in the context of the past 100 years of fisheries science and management. Certainly, there have also been temporary shocks to fisheries systems due to acute, pulse events such as hurricanes, oil spills, etc. (McLaughlin 2008). But mostly those have been short-term and highly regional in nature, not impacting the entire national fisheries science and management system. The closest lessons one can learn would likely be from the Influenza pandemic from circa 1918 (Reid et al. 2001, Niall et al. 2002, deValpine 2015, but the machinery to manage fisheries was not nearly as established then as it is today. Lessons one can learn from the 1918 situation, acute events and the current COVID-19 situation include the need to uphold all the human health and epidemiological guidelines while (often creatively) maintaining our ability to monitor, measure and manage fishes to provide seafood for the nation. The salient point from our current and historical situation is that although what follows focuses on our mission, the health and safety of the many fisheries professionals working at NOAA Fisheries, of our partners, of our stakeholders and of the communities in which we work remains a priority.
We acknowledge that there are observed drastic changes in global and regional seafood and fishing-forhire markets (e.g., Doremus 2020, FAO 2020c, Kemp et al. 2020, NOAA Fisheries 2020a), which have already been recognized as vulnerable to major shocks (Gephart et al. 2016, Cottrell et al. 2019. For the foreseeable future, at least in the medium-term (i.e., next several years), COVID-19 will impact the market dynamics and associated economies. How that influences the status of stocks that support such markets and the science needed to support these shifts remains a challenge (FAO 2020a,b, NOAA Fisheries 2020a.
D r a f t 3 composite, integrated system (at least for a given region; Link et al. 2015, NMFS 2016a, 2016b, Lynch et al. 2018, Peters et al. 2018. To provide mandated management advice for these LMRs and their associated human communities, we need to have meetings, venues, and protocols that use the best scientific advice available, and debate the best options within those scientifically determined conditions. To provide scientific advice, context and conditions, we need to forecast and assess ecosystems, protected resources, habitats, and fished stocks. To execute those forecasts and assessments, we need suitable data. To obtain the requisite data, we need to not only maintain suitable databases, but conduct surveys, sampling, and other forms of information collection. To accomplish all of the preceding, we need professional and well-trained personnel. The situation in 2020, and perhaps well into 2021, is that it will be the year(s) with severely limited data collection. Though some operations are underway, including limited small-scale regional NOAA Fisheries field work and NOAA vessel operations, these have been limited and are operating under significant health and safety restrictions. In fact, we have already had to cancel most of our major cruises and surveys in 2020 (Doremus 2020, The Maritime Executive 2020a,b, NOAA Fisheries 2020b,c; as of this writing we have cancelled over 50 fisheries surveys resulting in a loss of over 1500 on-the-water days-atsea). So that begs the question, how do we obtain our necessary data under these conditions? More so, how do we conduct our analytic efforts to assess the LMRs we are mandated to manage? Finally, how do we evaluate and communicate the increased uncertainty that will result from any unrecoverable data gaps?
The situation is also one without in-person meetings and with highly constrained travel (Viglione 2020). This limits the exchange of scientific information and also limits the ability for scientists, managers, and stakeholders to meet and discuss issues related to marine resource management. Most of these meetings have moved to virtual formats. For example, the Annual American Fisheries Society meeting was held virtually in 2020, whereas the Annual Science Conference of the International Council for the Exploration of the Seas has been postponed to 2021. The release of the FAO State of the World's Fisheries and Aquaculture report was similarly broadcast world-wide in over four languages. A number of smaller scientific meetings are successfully moving to virtual formats. Management-related meetings are also being held virtually. In the U.S., regional fishery management councils and status review groups (SRG) meet 3-6 times per year to manage marine fishery resources, let alone any sub-committee work.
Since March 2020, all of these council and SRG meetings have been held virtually, along with the attendant committee meetings and public comment events.
The impacts from COVID-19 will have different magnitudes. For example, we all can probably recover some functionality via virtual meetings and as travel restrictions begin to lift, but loss of data, disrupted time-series, etc., are significant with the inability to ever collect the data missed in 2020. The cost of that data-gap will diminish over time, and although modeling or imputation methods may be successfully employed to fill in and provide usable time series of data, the actual data gaps can never be filled. The impacts from COVID-19 also vary in time. There are relatively short-term (~next six months) and long term (3-10 years) consequences of the impacts from COVID-19. At NOAA Fisheries we have long D r a f t recognized the need to move beyond the normal, "bread and butter" assessment approaches into next generation assessments and forecasts and also begin to adopt more efficient and advanced sampling technologies (e.g. Lynch et al. 2018), but we have not always had as clear and urgent of a driver to force such a change. We have also recognized that the ever-increasing number of in-person meetings comes at a cost in terms of time, money, and the environment. It could be that the COVID-19 situation is a driver of such changes that we have considered and that are needed.
As the scientific leadership team for NOAA Fisheries, we wanted to offer some perspectives to the fisheries and marine science community on these topics in the midst of COVID-19. What follows are some thoughts to help mitigate the current circumstance that could also have bearing on future ways we conduct our mission. Although this work focuses on responses to COVID-19, it could be that any lessons learned also have application for other, future "shocks" to the fishery system in the U.S. or elsewhere. We aim to identify issues related to data, analysis, and meeting as they are affected by COVID-19, and propose some suggestions of how we might address those challenges. We offer these thoughts not as having solved every problem, but to learn from each other, compare across fishery and marine science organizations, and to engage in an active dialogue within our discipline to advance much-needed changes.

Data collection in the context of COVID-19
The basic data needed to assess our LMRs are well documented (Ballance et al. 2017, Lynch et al. 2018, and are not repeated here. To assess habitats and ecosystems may require additional data (NMFS 2016b, Peters et al. 2018), but often those analytical efforts involve using the same basal data as in stock or protected resource assessments, with rearrangement of how the data are treated (Samhouri et al. 2010, Shin et al. 2010, Fay et al. 2013, Large et al. 2015, Slater et al. 2017, Libralato et al. 2019).
Here we address the case of increasing loss of data collection capability due to COVID-19.
Maritime operations and associated logistics are challenging enough under routine circumstances, let alone in a pandemic with added layers of health and safety considerations. Not only are there at-sea operations, but there is also a significant amount of shore-based preparatory work before a ship's departure (e.g., staging, instrument calibration, net repairs, loading of supplies) by personnel, whose access to facilities and laboratories is restricted given enhanced health and safety protocols. Hopefully operations can resume safely, with sufficient COVID-19 mitigation protocols, sooner than later. Some University-National Oceanographic Laboratory System vessels have recently begun new research cruises (Graber-Stiehl 2020), and other oceanographic efforts have begun to be reinstituted (Berwyn 2020, Heslop et al. 2020)m as has a limited number of our cruises. We also note that fishing activities have continued. There is such an array of risk factors for which there remain no simple solutions for marine operation considerations (Addetia et al. 2020). Realistically 2020 (and likely beyond) will be a year with significant data gaps and we will need to adjust our analytical efforts accordingly. Furthermore, although there is significant effort to restart data collection programs now and into 2021, it is unclear D r a f t 5 when the full suite of data collection programs will be back on-line. We recognize the dedication of our survey groups around the country and their commitment to obtaining fisheries independent data. We also want to deftly manage the expectations for realistic data collection in the context of COVID-19 related risks.
Most of our fisheries/LMR data are presently collected via fishery-independent surveys that are mainly conducted on vessels that are either part of NOAA's fleet or are commercially chartered. Halting data collection during the COVID-19 situation resulted in large part from our dependence on these vessels as our principal data collection platforms, which in turn depend on at-sea deployment of personnel for their operation. Similar data collection stoppages in the future can be lessened through diversification of sampling approaches, which in addition to offsetting the risk of data loss, can increase the quantity and breadth of our data and information collection. Recognizing that diversified and increased data collection will also require parallel efforts in calibration, data storage, and analysis, below we suggest possibilities for replacing or augmenting our traditional data collection platforms during COVID-19 and beyond.
Assuming that challenges to conducting fishery-independent surveys and collection of associated data, marine mammal abundance surveys, and similar sampling efforts will persist, or that those surveys will be severely compromised in the rest of 2020 and potentially into the future, it seems prudent to emphasize platforms of opportunity and sampling of catches -fisheries dependent data -when and where we can do so safely. We acknowledge that fisheries dependent data are also affected by COVID-19 related restrictions on data collection, i.e., sea sampling, at-sea observation, port sampling, and recreational fishery angler intercept sampling have all been curtailed by disruptions. We also acknowledge that commercial and recreational fishery monitoring are differentially affected. However, fisheries are operating, and people are returning to work and providing food for the nation. Thus, the next steps in the effort to obtain such information might wisely be focused on dealers, processers and captains/leaders of fishing consortia at key ports. Admittedly deploying fisheries observers (NMFS 2020a,d, Kearns 2020) has presented significant challenges in the context of balancing data collection with health and safety concerns (NMFS 2020d, NOAA Fisheries 2020e). Certainly, there are concerns of bycatch, mis-reporting, under-reporting, and how one ultimately considers observers, electronic monitoring, or other forms of independent verification of these fisheries dependent data (Ames et al. 2007, van Helmond et al. 2019. Intercept sampling of recreational anglers will also continue to be limited for the foreseeable future; however off-site mail, telephone, and internet surveys are still able to continue to collect these recreational trip/effort data. What we are proposing is that this sampling emphasis would be primarily executed by port agents to double check and work with the reporters of landings (e.g., dealers, processors, association leaders), using extant methods to ascertain the degree of catches that we already have reported to us, while following COVID-19 guidelines (e.g., social distancing, masks). Providing these total catch or even total landings data would afford us the opportunity to at least employ "catch-only" or catch and primary production methods (noted below) and would allow us to project some assessments forward with these data inputs.
D r a f t 6 Other important fisheries dependent data include social and economic data. We acknowledge drastic changes in global and regional seafood and fishing-for-hire markets (e.g., Doremus 2020, FAO 2020c, Kemp et al. 2020, NOAA Fisheries 2020a, and the need to track these facets of fisheries remains important to ascertain the impacts of fisheries advice, fisheries management decisions, and COVID-19 effects. Much of the same efforts noted above -focused on dealers, processers and captains/leaders of fishing consortia at key ports, as well as recreational fishing surveys-can also be used to obtain necessary social and economic data. NOAA Fisheries' Cooperative Research Program (NMFS 2020f) is another approach to how we collect either primary or auxiliary data to help fill information gaps that are currently missing or limited. This program already works with many partners to collect a range of environmental and fisheries-relevant data, for example temperature measurements on lobster traps. There may be a possibility to expand cooperative research with commercial fishers in an attempt to do survey sampling for us in lieu of their normal fishing operations. This can provide information from any missing components of the survey and offset economic loss from the fishery due to COVID-19. Similar to the loss of large amounts of fishery independent data in 2020, our collection of ecosystem-level data is lacking in a year where many ecosystem changes are occurring. These programs may also be of added value in advancing the research, development, and operationalization of new sampling technologies. There are numerous cooperative data collection efforts across the country and these could be augmented where possible to support filling these knowledge gaps. Similar to the analytical considerations noted below, cooperative efforts take time; perhaps the COVID-19 situation provides further impetus for diversifying data collection among these efforts (NRC 1998).
Crowd-sourcing of suitably validated data (e.g., "citizen science"; Fairclough et al. 2014, McKinley et al. 2017 remains an option to collect necessary data (Fairclough et al. 2014, NOAA Science Advisory Board 2018). The current situation provides an excellent opportunity to further consider other forms of citizen science, such as further developing the use self-reporting (Mangi et al. 2013, Jiorle et al. 2016, NOAA Fisheries Office of Science and Technology 2020), electronic reporting/monitoring methods (Jiorle et al. 2016, NOAA Fisheries Office of Science and Technology 2018a, 2020) and internet search volume (Carter et al 2015). These approaches may be particularly appropriate for recreational fisheries (Jiorle et al. 2016, Venturelli et al. 2016, Crandall et al. 2018. We acknowledge that there are challenges associated with citizen science data collection, especially self-reporting-e.g., levels of opting-in, validation, statistical design, data standards, etc. (Venturelli et al. 2016, Crandall et al. 2018. Certainly compliance in reporting and accuracy of information would remain concerns, but ensuring adequacy of information could be tied directly to the decreasing of buffers for setting annual catch limits noted below (i.e., leading to increases in catch), and doing so would incentivize the industry and NGOs to self-report, selfpolice, and provide an impetus for us to update our database capabilities (Ames et al. 2007, Kindt-Larsen et al. 2011, Hold et al. 2015. Escalating some form of electronic or related monitoring seems D r a f t appropriate in the COVID-19 situation (Kindt-Larsen et al. 2011, CCFA 2020, Gibson and Wozniak 2020, SFP 2020. To efficiently and reliably collect necessary data, developing and beginning to more widely use enhanced forms of passive monitoring seems appropriate. The present situation (and moving forward with budget constraints, an aging fleet, and associated challenges to reliability) reinforces such a need. For instance, some satellite-based examples are noted in the assessment/analytical section below.
Other satellite applications could be the expanded use of Vessel Monitoring Systems (VMS; Witt and Godley 2007). From VMS, one can determine if a vessel is exhibiting fishing behavior (e.g., changes in velocity). From fishing behavior, one can estimate fishing effort (Watson 2017, Watson et al. 2018.
From fishing effort, one can estimate fisheries catch (Bastardie et al. 2010, Watson et al. 2013. From catch, one can estimate landings and use associated analytical methods (Bastardie et al. 2010, Gerritsen and Lordan 2011, Watson 2017. Such estimates might be relatively imprecise estimates, but could be an improvement over no estimates at all. Certainly not every vessel nor every fleet is equipped or required to have VMS, but this would provide a way to get at least a minimum estimate of catches. Plus, most commercial and many recreational vessels have some form of satellite positioning system that could be readily adjusted to VMS needs. Many of these are trackable via web-based applications. Working in partnership with the NMFS Office of Law Enforcement, we have the theoretical capability to at least count fishing vessels, and those vessel counts could be used as a proxy for fishing effort as noted above. Passive acoustics (e.g., set or moored arrays, towed transducers) are in place for many, but certainly not all, regions or species (van Parijs et al. 2009, Heupel andWebber 2012). These can be better utilized to target key taxa, both fished and protected resource species, to obtain indices of relative abundance.
There are nuances to this approach, including whether acoustic signatures to identify taxa of interest exist, if the extent of an array covers a reasonable fraction of the taxa's distribution, the timing of major migration events of certain taxa, and even whether some taxa have been acoustically tagged to groundtruth the array. Yet given those limitations, it may be that these arrays provide a short-term solution, but re-calibrating and using extant arrays might allow us to get such estimates of relative abundance relatively quickly, particularly if staff are not able to be widely deployed on manned platforms.
Similarly, there are many novel, promising, and innovative technologies to sample and survey LMRs (Bradley et al. 2019, Moustahfid et al. 2020, NOAA 2020. Unfortunately, some are not yet at the level of technological development (e.g., eDNA, 'omics) ready for use in surveying LMRs to the scope and extent that is needed and some of them are quite labor and field-work intensive. Yet despite the intensity of labor, some of those options may provide helpful options relative to the alternative. This is particularly true for technologies (e.g., hexacopters, drones) that focus their sampling on land or surface-based organisms, such as many marine mammals (e.g., Torres et al. 2018). Saildrones, autonomous underwater vehicles (AUV), and related, self-contained samplers have acoustic, visual and oceanographic sensor capabilities that are under-utilized as data streams for our data needs (Harris et D r a f t al. 2019, Moustahfid et al. 2020). The technology exists to make these "force-multipliers" for our normal, manned surveys. In some instances (Antarctic, sub-Arctic, and Arctic) these technologies are already used to partially survey given regions. We could redeploy drones and AUVs in the present situation (albeit in a limited sense) and plan to scale up such efforts for future years, making this a clear area for escalation (NOAA 2020); we are already deploying gliders in the Antarctic peninsula and saildrones in the Bering Sea for this purpose (NMFS 2020e). The degree of sampling coverage and breadth of information will be less than we would have in a normal year, but would be more than if not deployed. Certainly, cross-validation and cross-calibration studies would be necessary (NOAA 2020), but those could be done post-hoc after data from these technologies were captured.
While we recommend focusing on catch data to mitigate the immediate impacts of COVID-19, we also need to acknowledge the need for data collection that supports ecosystem indicators that can provide environmental context to expectations for recruitment, vital rates and distribution (Slater et al. 2017, IEA 2019, Dorn andZador, in press). These data can be collected via many of the crowd-sourcing, cooperative research, and similar partner-based efforts noted above. The important point is the acknowledgement that contextual information is needed for any fishery dependent data collected, and the ability to do so is likely more feasible than is typically expected.
Moving forward, we need to continue to evaluate all of our sampling and survey options, but likely with more urgency and a sharpened sense of how exceptional circumstances can impact our mission. Focusing on catch data and the new electronic means to obtain that information seems quite feasible and even more necessary in this COVID-19 situation ( Table 1). Some of these recommendations urgently need to be addressed (e.g., managing expectations, redeploying advanced technologies, port-based sampling, better utilizing crowd-sourced data). Some of these recommendations need to be started now and continued into the future (e.g., better use of vessel monitoring system data, better use of passive acoustics), and some of these recommendations need to be started now, with a sense of urgency, and perhaps used to improve our long-term operations ( Table 1). The present situation has also highlighted our vulnerabilities with respect to being too dependent on any one type of platform and its related data streams. The role of additional study fleets, charters, and other platforms for hire would be wise to more fully explore, as would how to escalate the rate of adopting advanced sampling technologies in an operational sense (NOAA 2016).
We recognize that adding new data series and collection methods to assessments is not "turn-key". There is substantial effort needed for calibration, development and management of data systems, and incorporation into assessments (Ballance et al. 2017, Lynch et al. 2018). LMR management is largely supported by time series, so that status and trends of the resources can be evaluated. New data types without calibration have little long-term perspective and thus their immediate value is limited, but can be built over time. Also, the addition of new data to assessments generally involves peer-review; new data collection approaches need to be synced with assessment review processes. That said, we knew we would benefit from additional fishery independent options even prior to the present situation (NMFS D r a f t 2020b), but the COVID-19 situation is effectively forcing the re-evaluation and adaptation of our data collection methods, as well as the associated urgencies to do so.

Analysis, Forecasts and Assessments in a limited information context
NOAA Fisheries assesses a wide range of fish stocks, protected resources, habitats, and ecosystems, but impacts to data collection will degrade the precision of their results and pose challenges for their execution. Capabilities and methods are already in place for situations that are or become data-poor (e.g. Cope and Punt 2009, Carruthers et al. 2014, Chrysafi and Kuparinen 2015, Geromont and Butterworth 2015, even for full systems assessments (Smythe and Thompson 2015, Francis et al. 2018, Reum et al. 2020). While we conduct our assessments regularly, we do not do all assessments every year. To ensure that the most important needs are met first, we have established a system for prioritizing our assessment work, which not only identifies where assessments are needed most, but also supports decisions about when more streamlined "operational assessments" can be used versus more resource intensive "full assessments" (Methot 2015;Lynch et al. 2018). Our ability to be more efficient is supported by our robust data-bases and increased understanding of system and species' dynamics, and our need to be more efficient is largely driven by rising costs and shrinking daysat-sea (Punt et al. 2020; ICES 2020, NOAA Fisheries 2020d). Thus, we do have protocols in place to scale our assessment efforts as needed.
In the COVID-19 situation, we acknowledge that missing data will make stock, protected resource, habitat, and ecosystem assessments more imprecise and potentially even cause assessment methods to fail. With that being the case, we should strive to identify and account for any uncertainty that such missing data causes, and address that as consistently as possible among stocks, protected resources, and ecosystems. However, the big question is that if there are no data available to update assessments, even where data-poor methods are used, do we still have options to meet our assessment and adviceproviding needs? The answer is that we do, much of our current enterprise has some built-in robustness to these circumstances, at least for the near term and, in other situations, we have an opportunity for creativity, advanced technical and interdisciplinary solutions. What is proposed here is presented for the case when the ability to collect the requisite information is compromised (c.f., ICES 2020), with notations on options at different points in time.
The first and most obvious option is, for stocks assessed with dynamic methods that allow forecasting (Lynch et al. 2018), to simply project the stock forward an additional year with a range of assumptions about stock conditions and catches during 2020 and 2021. In fact, the impetus for this work is largely driven by the reality that there will be instances where some or all of the requisite data are missing for that time period. Using forward projection methods without updated data would increase uncertainty in management advice but would be little different than the current advice framework (NMFS 2020c, Punt et al. 2020) that is routinely based on projections one to several years beyond the terminal year of data in assessment models. As predictions advance further in time, the uncertainty increases and, in some cases, this is reflected in more precautionary advice, though substantial underages in catch could lead to D r a f t increases in future catch advice when a stock is assessed again in the future with new data. Thus, projections of stock status can be done for a future year with suitable caveats about the uncertainty of the projections absent updated data for this current year. Particularly in these situations where standard fisheries data collection frameworks may be impaired (c.f., ICES 2020), socioeconomic data such as fuel receipts, license sales or social media could be useful to characterize key assumptions, such as the magnitude of recreational catch in instances with limited recreational catch surveys during this COVID-19 time-period.
The second option involves "operational" or update assessments that re-estimate model parameters with additional but sparse data. Thus, there are instances where updating the assessments with only some missing data is a possibility. Most modern, simulation-based assessment modeling approaches (Methot 2009, Geromont and Butterworth 2015, Punt et al. 2016) are designed to handle a certain degree of missing data. Even in cases with very little new data, but for stocks whose assessments have been "skipped" within a tolerable range of frequency (e.g., scheduled for this year, not assessed in 1-3 years), this approach would still work and could improve upon strictly updating projections. Though, as one is essentially trading data for assumptions, a full accounting of the increased uncertainty may require a broader range of sensitivity analyses to determine the robustness of the modeling framework to missing data. This all collectively implies that there remains some source of reliable data (e.g., fishery-dependent data collected by industry partners; i.e., catch) to use in the update and to mitigate the data loss, or else recognizing and clearly communicating that the uncertainty will notably increase with each successive year of our forecasts.
The next assessment options are data-poor methods, which have been used when the dynamic stock assessment (i.e. data-rich) methods are unable to be employed (Pitcher 1999, Pitcher and Preikshot 2001, Cope and Punt 2009, Carruthers et al. 2014, Chrysafi and Kuparinen 2015, Geromont and Butterworth 2015. Often such data-poor methods may be less impacted by loss of data and may continue, provided that some form of reliable indicator exists (such as a fisheries independent survey; Shertzer et al. 2008a, Ault et al. 2018, Huynh et al. 2020). The modularity of many modern assessment models (Punt et al 2020) is particularly valuable in this context, where the same platform can serve as a modeling continuum from data-rich to data-poor (Cope 2013), thereby allowing for at least a minimal assessment if data streams erode into the future. To be clear, we are not suggesting that dynamic stock assessment methods should revert to data-poor methods, but the latter does give options for at least a minimal assessment if data streams continue to erode into the future. Any such changes would be multiple years from now, and where possible we recommend retaining the projection methods noted above. We note that many data-poor methods are catch-only approaches and typically rely upon the assumption that a stock was in quasi-equilibrium when the method was applied. This gives a snapshot of advice, but is not intended to be updated with the same equilibrium assumptions. At this point in time, catch data are most certainly influenced by market conditions (Eliasen et al. 2014, Scheld and Anderson 2014) and we understand they are not ideal methods, so would similarly not recommend them unless there are absolutely no other options. That said, this should be an impetus to obtain and D r a f t 11 use fishery-dependent data beyond just catch (e.g., Catch per Unit Effort, CPUE; albeit with the caveat that fishing behavior has almost certainly also been affected, thereby impacting the proportionality of CPUE).
Thus, in many instances, methods exist that can be employed this year for fisheries stock and protected resource assessments having limited data; but those would be for instances that have mostly up-to-date data or for stocks that were scheduled for analysis after having not been assessed within a recent time period (e.g., a year or two, or some reasonable fraction of the taxa's lifespan). For these circumstances, one would need to manage expectations on the limitations of the data and identify ahead of time the stocks where catch data are not viable. The age-structure of longer-lived species gives inertia to those populations such that a data gap is bridged more easily. Where single or multiple year gaps pose a much more acute challenge is for annual and shorter-lived species like shrimp, salmon and small pelagic species, where catch advice comes almost entirely from very recent information. Here is where creativity and scientific innovations outlined below may provide opportunities. These shorter-lived taxa represent situations where fishery-dependent CPUE could particularly play a role.
One strategic tool in widespread application is Management Strategy Evaluation (MSEs; Smith 1994, Punt et al. 2014) developed to find management approaches that are robust to a range of dynamics and uncertainty and that meet multiple, often competing, management objectives. Various MSE approaches have simulated a range of stock, protected resources, and ecosystem statuses based on a range of harvest control rule options, possible sampling, ecosystem conditions, and other, testable assumptions-all relative to some specified management objectives, with the key point that there are feedbacks in MSEs to elucidate the response of the system to various interventions. While they may have utility here for some provision of tactical advice (e.g. next year's TAC) and certainly are useful in evaluating the value of information such as allocation of survey resources, their primary role is in strategic decision making. We see them playing a greater role for decision making under what is likely to be greater environmental uncertainty and even more explicit consideration of multiple objectives for ecosystembased fisheries management, as multiple, conflicting objectives and changing ocean conditions have not gone away during COVID-19. However, their most effective utility now is in considering the insights to be gleaned from information gaps, especially via using existing MSEs. For example, how often did the loss of a single year of data immediately trigger an "exceptional circumstances provision" which would result in notable different management advice; or are management procedures robust to missing data; or, for ecosystem MSEs, how do broad ecosystem indicators perform and inform as tactical advice; all situations commonly explored in MSE evaluations. In most cases, MSEs explicitly consider these situations and most management protocols build in robustness by averaging several data points so that they are inherently designed to avoid common failure points. Perhaps the COVID-19 situation provides an opportunity to expedite ongoing or new MSEs in more of a "crisis management" mode where the rudiments of MSEs already exist. At the least, the COVID-19 situation highlights the need to escalate the development of MSEs for broader use, and provides critical context to evaluate the robustness of our status quo approaches to exceptional circumstances.

D r a f t
In certain cases, management procedures (MPs; de la Mare 1998, Geromont andButterworth 2015, Huynh et al. 2020;sensu Punt et al. 2016) essentially recommend no change if forecasts of fisheries stocks, protected resources, or projected catches, do not fall outside of pre-determined levels (often as tested and established via prior MSEs). Some of these "non-benchmark" approaches exist now, but these would not be that difficult to establish for many stocks occurring in stable situations. This would probably work for most stocks where presumably overfishing is thought to have stopped. These are related to some of the dynamic stock assessment projection methods noted above, and certainly they can use MSEs as the quantitative engine, but what distinguishes MPs here in the more generic sense is that they provide a reasonable course of action for acceptable levels of management options. Therefore, short of formal MSEs or related quantitative analyses, for most stocks and taxa that had previous evidence for a stable or increasing abundance, and are known to not be overfished or threatened, keeping catch or removals constant at the last annual catch limit level should typically work. Thus, in instances where there are data gaps of at least a year, but with stocks or protected resource taxa in previously known positive conditions, and with no known major changes in parameters that impact the ecosystem and taxa of interest, we could make the recommendation of using the prior year's annual catch limits or potential biological removals; we amplify on this further below.
The challenge is for stocks, protected resources, habitats, or ecosystems where there are limited to no data to update even data-poor methods. There exist other options that could be used in such situations. Scenario planning is a structured, expert-driven process that embraces uncertainty and explores plausible alternative future conditions under different assumptions to help manage risk and prioritize actions (Schwartz 1996, Peterson et al. 2003. It can be thought of as a qualitative Management Strategy Evaluation. NOAA Fisheries is beginning to use the method to evaluate science and management actions under climate change (e.g., Atlantic salmon, Borggaard et al. 2019). The approach could be used to evaluate any number of actions under shorter term scenarios, including operating conditions improving or worsening in 2021 or more or less data available in 2021, all in the COVID-19 context. There are comparable, extant, qualitative tools (e.g., conceptual modeling, qualitative network analysis, mental models; Smythe and Thompson 2015, Francis et al. 2018, Reum et al. 2020) that can similarly be used for full ecosystem assessments, with particular consideration of the potential ramifications of various tradeoffs during COVID-19 conditions.
There are similarly some indicator-based approaches that can be projected for overall system status, status of stocks, protected resource status, and even habitat status (Link et al. 2002, Link 2005, Potts 2006, Samhouri et al. 2010, Shin et al. 2010, Kaplan and Leonard 2012, Fay et al. 2013, Large et al. 2015, Lederhouse and Link 2016, Fulton et al. 2019, Libralato et al. 2019, Link and Watson 2019. The value of these approaches is that they are often simple, intuitive, and though they benefit from data updates like any other method, they can be extended to 2-3 year projections using relatively simple and straightforward statistical methods (Link et al. 2002, Link 2005, Shin et al. 2010, Fay et al. 2015, Probst et al. 2013, McDonald et al. 2017. Some of these indicator approaches are based on catch and primary D r a f t 13 production for an entire system (Cury et al. 2015, Libralato et al. 2019, Link and Watson 2019. These indicators rely on satellite imagery to obtain primary production (the base of the food web that can support fishes and protected resources; Smyth et al. 2005, Link andWatson 2019) and can be reverseengineered to estimate how much total catch can be prosecuted in any given ecosystem based on empirical, tested, and theoretically determined limits (Link et al. 2002, Fay et al. 2013, Link and Watson 2019. Allocation projections based on recent precedents are then doable. Certainly, there are nuances of estimating primary production from satellite imagery, as in any of the modeling and assessments we conduct, but this approach provides a means of evaluating our LMR with a series of continuous data. This approach presumes that data-streams from satellites will still be available and will not be directly impacted by COVID-19 for the foreseeable future. Satellite information can also help track major oceanographic features, which once processed into indicators can be used as early warning signals that environmental conditions are changing, thereby adjusting annual catch limits, potential biological removals, or harvest control rules based upon shortterm projections of what would likely be substantially altered environmental conditions. We do this in some regions already using our Ecosystem Status Reports as briefed to our Fisheries Management Councils and other partners (e.g., Slater 2017, IEA 2019, Dorn and Zador, in press), and the approach can be readily expanded elsewhere with widely available and remotely sensed, updated information.
Risk assessments and evaluations are perhaps under-utilized methods to frame the status of LMRs. NOAA Fisheries has pushed to have many of these done with respect to climate (NCSS; Link et al. 2015), and these climate vulnerability analyses are extant for every region (e.g., Crozier et al. 2019, Gaichas et al. 2014, Hare et al. 2016, Spencer et al. 2019; NOAA Fisheries Office of Science and Technology 2019). They are based on repeatable and intuitive methods, can be based on limited information, and can provide important context of what should be prioritized analytically with respect to this risk. There are also Productivity-Susceptibility Analyses (Patrick et al. 2009) extant for most of the managed fish stocks we report on and that can be used to prioritize analytical efforts. These climate vulnerability analyses and related information can be modified to include other factors, and the Productivity-Susceptibility Analyses can be updated for the ones completed approximately 10 years ago (Patrick et al. 2009). Some of our Fisheries Management Councils have conducted related exercises, often using the climate vulnerability analyses results or methods (e.g., Gaichas et al. 2018). The value of these analyses is to not only triage assessment and analytical efforts, but to communicate to all interested parties our rationale for why we are not conducting assessments for some stocks, protected resources, habitats, etc., for the immediate future, largely due to the (lower and differential) level of risk. Although indicators, satellites, and risk assessments offer methods to provide management advice, in many instances these approaches are not currently used, so similar to the broader use of MSEs, the COVID-19 situation provides an opportunity to expedite ongoing or new alternative approaches in a "crisis management" mode.
D r a f t

14
The advice we provide with respect to potential biological removals, annual catch limits, habitat restoration, ecosystem overfishing, etc., can be considered as defaulting to two guiding principles. First is to retain continuity with preceding years. This approach was reinforced when discussing hurricane forecasts with our counterpart physical modelers (NEMoW 3; Townsend et al. 2014); the hurricane modelers have business rules that do not allow them to alter projected hurricane tracks beyond a certain range between forecasts, even if the models, data and information suggest otherwise (NRC 2006, Meuel, et al. 2012J. Beven, pers. comm.). Here we would aim to maintain continuity due to the lack of new information. We need to consider such an approach to mitigate the present situation and in moving forward. This continuity approach is in some respects a type of MP as noted above, and is also a precautionary tactic in a general sense. Yet the effective loss of fisheries-independent data, and lesscertain fisheries dependent data, may result in the need for even more precautionary management. Which leads to the need for more formal precautionary approaches. Thus, the second option is the sensu strictu application of the precautionary approach to fisheries management (Darcy and Matlock 1999, Essington 2001, Hilborn et al. 2001 which is adopted by most Fishery Management Councils, and is a global best practice for LMR management (Darcy and Matlock 1999, Rosenberg 2002, Punt 2008, Rice 2009). The precautionary approach adds buffers to catch advice (hence catch tends to decrease) with increasing scientific and management uncertainty, or known adverse conditions, all of which increase in the absence of data on the stock or the fishery.
We propose a default for many U.S. fisheries that would maintain catches at or around the previously set level, unless otherwise indicated. If conditions arise that warrant it, or are even suspected, managers could, at their discretion, adopt increased uncertainty buffers (e.g., the level of risk or probability of overfishing, e.g., P*; Prager et al. 2003, Shertzer et al. 2008b, Prager and Shertzer 2010, Methot et al. 2014sensu Methot 2009) in setting annual catch limits, potential biological removals, etc. For other situations, ecosystem, social or other indicators may also provide plausible motivations for adding greater precaution. We note that the continuity approach could -and the precautionary approach almost invariably would -lead to reductions in catch under uncertainty and hence tends to only focus on the natural resources, and not on the communities that are dependent on them. COVID-19 may prove to be a more uniquely human perturbation than a stress on natural resources and adopting catch reductions commensurate with the strict precautionary approach coupled with severe economic disruptions could lead to further impacts on fishing communities. Though some short-term economic relief has been provided through the CARES Act (https://www.fisheries.noaa.gov/national/noaafisheries-coronavirus-covid-19-update), the impacts of COVID-19 will be multifaceted and long-lasting beyond the analytical considerations noted here. Socio-economic analysis to explore these COVID-19 impacts will most likely be done in retrospect, but nascent efforts to evaluate them are underway. Hence, we recommend the continuity of advice approach, which unless indicated otherwise, could serve as a default approach that balances resource conservation, human impacts and the substantial analytical and rule-making burden necessary for formally applying a precautionary approach. The major risk with a continuity default is that one assumes status quo, but if something drastic happens in the ecosystem (e.g., occurrences of marine heat waves, harmful algal blooms, etc.) then assumptions of stationary and D r a f t basal conditions are incorrect; this is why it will be important to maintain citizen science efforts (see above) to assist with such observations and why we need to be ready to adopt the more formal precautionary approach.
It may be that we simply support Fishery Management Councils, State Marine Fisheries Commissions, Scientific Review Groups, and other management partners in the normal rule-setting process, and especially those that invoke emergency rules for unassessed stocks and/or maintain current catch and removal levels, aiming to simply postpone our analytical efforts beyond what we can reasonably forecast. However, the items listed above ( Table 1) are some of the alternatives we can consider as we scope and scale those emergency decisions, while the assessments to directly support them may be delayed as we await novel data. Some items noted urgently need implementation now (retaining continuity, employing data-poor methods, continuing short-term projections, using management procedures, conducting risk assessments, using indicators) while others can be viewed as being needed as soon as is feasible but also to spark escalation of their more widespread (and needed) adoption (escalate MSE development, early warning signals; Table 1).

Development of Advice and Debate of Scientifically-based Management Options in a limited travel context
There remains the need to gather and analyze data, implement models that use the data, and review the model outputs. There remains the need to evaluate those outputs and translate them into advice. There remains the need to establish the amount of catch or removal, and the need to debate the allocations thereof. We are always going to need to have our staff, our partners, and our stakeholders interact. The present COVID-19 situation has likely provided a fundamental and possibly lasting shift in how our activities will be conducted moving forward (e.g., Kramer and Kramer 2020).
For at least the next year or so, we expect the criteria for travel approval will likely remain stricter than before COVID-19, but not as strict as the essential/mission-critical travel criteria we currently have in place for COVID-19. That is, within NOAA and similar organizations, this essential travel would be travel limited to that which is necessary to accomplish an immediate, mission-critical, mandated task, and when risks relative to human health are mitigated. Besides federal rules, different states, academic institutions, companies, and NGOs all have separate rules for travel that affect the ability to hold meetings. The diversity of such travel rules makes holding meetings a challenge. Collectively these challenges will reduce the number of in-person meetings, and hence travel, to those activities that can truly only be done in person, causing us to rethink our approach for prioritizing work-related travel. As we all learn more about COVID-19, and as additional mitigation measures are put into place, developing revised criteria for travel will be a useful outcome of the present situation. An intentional evaluation of what has been learned when working and meeting remotely is an effort that could usefully inform D r a f t future meetings. There still needs to be time and opportunity to meet in person, but perhaps in a more strategic and deliberate manner.
The need to shift to online meetings is now feasible, and such online meetings will likely continue more than they did before COVID-19. The efficacy of these online meetings needs to continue to be evaluated. The ability to shift to this more online approach, and the technological capacity to do so, is rapidly evolving. Most of us have adapted to the many video conferences and meetings now common in the COVID-19 situation, and with a few exceptions these are mostly quite functional (Price 2020). One positive outcome of this more "virtualized" working environment is the opportunity for increased collaborative efforts, as barriers to such efforts have been notably lowered by the present situation. Perhaps any savings in more constrained travel could be partially applied to updating the necessary technology and capacity to expand/improve associated online endeavors. If an online presence continues or if it is expanded, the posture of our IT and related staff needs to continue to be, but even more so, one of access and connection over security, not vice-versa (ITU 2020). Issues of database access and information sharing while maintaining reasonable cyber-security warrant further attention in this context.
The future of telework, and its possible expanded use, needs to be assessed. NOAA Fisheries entered mandatory telework in March 2020 and much of our work continued. Certainly, the need to stage field or cruise work, to maintain living organisms, and to conduct laboratory work are not conducive to telework. But much of what we do in our mission can be accomplished via telework or work-from-home scenarios. The implications of this proposed expansion are both potentially positive-less commuting, morale boost, more time to focus on work, potentially less facilities/utilities direct costs, fewer distractions, increased productivity, -and also potentially negative-loss of serendipitous hallway interactions, lack of accountability, loss of camaraderie, increased compute and IT costs, more distractions. The COVID-19 situation has afforded us essentially with a test. The test has effectively shown that the assumptions and concerns of widespread telework at-scale were largely unfounded and warrant being revisited (Orgysko 2020, Thornton 2020); that is, our personnel are finding ways to adjust, and remain remarkably productive (Smith 2020, Thornton 2020. Although productive and still meeting many facets of our mission, we also do not want to overstate the ability of our personnel to maintain productivity given some of the unique challenges arising from work-from-home scenarios, especially over extended periods of time. Nor do we want to understate the differential impacts resulting from work-from-home scenarios, particularly for personnel with dependents at home, impacted from school closures, etc. A clear review of lessons learned during the ongoing COVID-19 situation would be helpful to better clarify when telework works well, and what conditions need to be established for it to be (even more) successful. Any such conditions particularly need to include discussions on computing power, bandwidth, etc. and how those can be maintained and more widely sourced, as well as "empathetic" policies for those with dependents at home. From that, we expect evaluation criteria will evolve and be applied to ascertain the key lessons learned from this de facto teleworking experiment.
D r a f t There are several considerations regarding travel, telework, and how we all conduct meetings during COVID-19 (Table 1). Some of these recommendations needing to be resolved in the short-term (e.g. evaluate effectiveness of online meetings, invest in information technology software, evaluate telework effectiveness etc.), some need to be started now and continued into the immediate future, and some need to be started now and used to improve or revise our long-term operations ( Table 1). The larger, macroeconomic sense is that how we all execute our normal operating protocols of doing business have been disrupted -some maybe irreversibly -due to COVID-19 (Kramer and Kramer 2020; i.e., "will you ever go back to the office again?"). Using the COVID-19 situation as a rationale to evaluate and revise how we conduct parts of our mission seems not only wise, but appropriate as we are in the midst of it, and a real-time examination of this information should minimize the global disruptions thereof (Cho 2020, Kramer andKramer 2020).

A note on organizational excellence
Equal to the LMR-focused aspects of our mission, one of the main priorities for NOAA Fisheries is organizational excellence. What that means is we want NOAA Fisheries to continue to increasingly be an organization that is a workplace of choice, both for our personnel and for interested candidates not yet with us; an organization that takes care of its people; and an organization that is constantly trying to improve how it does business. We suspect that most fisheries and marine science organizations have similar aspirations. That involves a range of discussions on topics that are more science-support than directly science, but which are mission-critical. The COVID-19 situation may provide an opportunity to revisit some of these issues by necessity.
We note that these organizational considerations are being pondered more deeply in other contexts , but here we briefly highlight some important considerations. For instance, what opportunities does COVID-19 afford in terms of professional development, training, continuing education, mentoring, temporary assignments to expand networks and experience, etc.? Related to work-from-home considerations noted, COVID-19 is forcing us to consider other facets of our organization: what facilities do we need, should we consider desk sharing, do we need to revisit organizational structure to perhaps capture "virtual" centers of excellence, are there efficiencies that have arisen in how we conduct our business that we need to more widely adopt, do we need to change standard operating procedures, etc.? The salient point is that we need to continue to evaluate all of our organizational dynamics, and to build on outcomes from a series of program reviews (NMFS 2013(NMFS , 2014a(NMFS ,b, 2015(NMFS , 2016c(NMFS , 2017; NOAA Fisheries Office of Science and Technology 2018b) to continue to improve how we do business in light of the particular exigencies of the COVID-19 situation.

Ad extremum verbum
D r a f t We recognize that many of us in fisheries and marine science organizations are simply trying to deal with the day-to-day challenges of the COVID-19 situation -both professionally and personally -and that ultimately, we all need to find ways to mitigate the effects of COVID-19 on our missions. But we also recognize the need to glean information from our responses while in the midst of the current situation to take advantage of any lessons learned that could improve how we conduct our mission moving forward. Ideally, we would like to implement any of the above recommendations (Table 1) after full mitigation of COVID-19, but short of that we wanted to explore possible solutions that minimize the COVID-19 risk while still affording us the ability to conduct the work needed for our mission. What we present in this work represents some "in-the-moment," real-time attempts to deal with the tactical while also thinking strategically. We present these perspectives in part to signal our intentions, in part to learn what others may be doing, and ultimately to foster a discussion on the topic for the discipline.
We offer these thoughts not as having solved any and every problem, but as a means to foster ongoing discussions within the wider fisheries and marine science communities so that we do not squander this historic opportunity. If this thought-piece results in even one, important next step to advance some much-needed changes in our discipline, we will have succeeded.  Table 1. Summary of key recommendations for core science aspects of the NOAA Fisheries mission.

Analysis, Forecasts and Assessments
With available, mostly up-to-date data Continue with short-term projections for stocks assessed with a dynamic method Employ data-poor methods for stocks with appropriate data If limited or dated data Escalate establishment of management strategy evaluations, ideally for groups of species Establish or adopt pre-determined management procedures Explore scenario planning Use indicator methods, and statistically project forward Adopt catch-satellite based indicators to project total catches Adopt early warning signals (satellite based) using environmental condition indicators Use extant risk assessment and vulnerability analyses to triage what can reasonably be left alone and what needs urgency In absence of any new information, Defaults should: Retain continuity (last year's annual catch limit, potential biological removal) If information suggests otherwise, then Adopt a precautionary approach

Data Collection General Considerations and Focus
Manage expectations of our surveying capacity Maintain port-based and use reporting individuals for catch data Revisit cooperative research emphases Establish crowd-sourced data sources, infrastructure to support it Escalate adoption of electronic monitoring Passive monitoring Expand and better use vessel monitoring system data to estimate effort and catch Expand and better use passive acoustic arrays Advanced sampling technologies Redeploy and use extant drones, hexacopters, autonomous underwater vehicles Escalate broader adoption of advanced sampling technologies

Advice and Debate with No Travel
Continually evaluate travel criteria approval Evaluate effectiveness of online and in-person meetings, with standards for appropriateness of each Assess online meeting strengths and weaknesses, with suggested protocols/guidelines Invest in information technology hardware and software for online meeting Evaluate effectiveness of telework and work from home protocols