Content area
Evaluar el progreso hacia el alcance de los objetivos de conservación y sustentabilidad del agua dulce requiere la transformación de diversos tipos de datos para hacerlos información útil para los científicos, gestores de recursos y otros tomadores de decisiones. A pesar de que el volumen de datos recolectados alrededor del mundo en cuerpos de agua dulce ha incrementado sustancialmente, muchas regiones y ecosistemas aún no cuentan con suficiente recolección y/o acceso a datos. Ilustramos como los desafíos respecto a la generación y accesibilidad a los datos son el resultado de un conjunto diverso de mecanismos subyacentes y proponemos soluciones que pueden ser aplicadas por individuos y organizaciones. Discutimos diferentes opciones para enfrentar la escasez de datos, incluyendo el uso de ciencia comunitaria, sensores remotos, sensores ambientales y revalorización de conjuntos de datos históricos. Resaltamos la importancia de coordinar los esfuerzos para la recolección de datos entre grupos y programas de monitoreo para mejorar el acceso a los datos. A nivel institucional enfatizamos la necesidad de priorizar el procesamiento de los datos, incentivando la publicación de los datos y promoviendo investigaciones que fortalezca una mayor cobertura y representatividad de los diferentes ecosistemas. Algunas de las estrategias institucionales incluyen aproximaciones tecnológicas y analíticas, pero muchas requieren de un cambio de prioridades e incentivos dentro de las organizaciones, tales como las instituciones de investigación académicas y gubernamentales, grupos de monitoreo, revistas científicas y agencias de financiamiento. Nuestro objetivo es estimular la discusión para reducir las disparidades de datos que dificultan la comprensión de los procesos en cuerpos de agua dulce y sus cambios a través de las diferentes escalas espaciales.
INTRODUCTION
Freshwater ecosystems are vital for the cultures and economies of diverse societies around the world. Freshwater ecosystems also integrate complex biogeochemical processes occurring within their watersheds and airsheds (Schindler, 2009; Williamson et al., 2009). Therefore, understanding the environmental and ecological dynamics of freshwater ecosystems, and making resource-management or conservation decisions, requires collecting a wide range of biotic and abiotic data and transforming it into useful information. Ecology and environmental sciences have entered an era of big data (Hampton et al., 2013; Tosa et al., 2021). This phase is being driven by the automation of physical, chemical, and biological monitoring (Besson et al., 2022; Jankowski et al., 2021; Marcé et al., 2016), the widespread adoption of remote-sensing technologies (Cavender-Bares et al., 2022), the use of the Internet as a vast source of ecologically relevant data (Jarić et al., 2020; Keeler et al., 2015), and more recently the advancements in artificial intelligence (Branco et al., 2023). In some highly studied systems, such as with acid rain in the northeastern United States, data and information are no longer a primary limitation to environmental decision-making (Likens, 2010).
Despite these advances, pronounced disparities in data collection and data access persist across freshwater ecosystems globally, hindering decision-making and management. For example, Lake Tahoe (United States) and Lake Nahuel Huapí (Argentina) are both large, deep freshwater lakes known for their ecological and economic importance; yet, there are 60 times more meteorological stations within 30 km of Lake Tahoe (~1 per 172 residents) than around Nahuel Huapí (~1 per 61,000 residents; Figure 1). In a sparsely monitored system such as Nahuel Huapí, data abundance and spatial coverage are potentially inadequate for science-informed decision-making or management. In a highly monitored system such as Lake Tahoe, conservation and management efforts may instead encounter challenges with “too much” data, such as difficulties in integrating and synthesizing disparate data sources and providing access to increasingly large volumes of data. Data collection and access vary across countries and regions, roughly correlated with gross domestic product (GDP; Hughes et al., 2021), but can also vary at smaller spatial scales due to logistical or socioeconomic factors that are unrelated to the value of a freshwater system in terms of ecosystem services, economic and/or cultural importance, or biodiversity. Whether resulting from “too little” or “too much” data, lack of access to relevant or actionable information constrains efforts to tackle societal challenges in ecosystem management, environmental sustainability, and conservation (Junk & Piedade, 2004; Marcé et al., 2016; Thurstan et al., 2015).
[IMAGE OMITTED. SEE PDF]
Two conditions are necessary to generate useful information for conservation and management: (1) collection of high-quality data (e.g., data must exist at the time and place they are needed and must have adequate metadata to support reuse), and (2) adequate data access (e.g., data must be findable, accessible, interoperable, and reuseable; FAIR; Wilkinson et al., 2016). Challenges to creating useful information can arise from deficiencies in either condition, but it is common for one of these principal mechanisms to be the major obstacle for understanding and sustainably managing a given freshwater ecosystem. Absence or inadequacy of data collection is common in many freshwater ecosystem types (e.g., subterranean ecosystems; Saccò et al., 2024; mountain headwaters; Moser et al., 2019; springs; Fensham et al., 2023) and entire regions (e.g., Global South; Krabbenhoft et al., 2022). Challenging logistics associated with site access (e.g., remote sites or unsafe conditions) may further constrain the type or timing of data collection, especially when combined with financial limitations (Ghosh, 2022). In contrast, regions with sufficient scientific infrastructure and resources to support high-resolution temporal and spatial environmental monitoring and data collection are subject to a separate set of challenges related to providing access to large volumes of data. Large datasets are challenging to organize, share, maintain, and use, though a shift toward open data practices is underway (Fidler et al., 2017; Reichman et al., 2011; Stall et al., 2019). Integrating datasets collected by multiple organizations and agencies is often necessary for regional-scale environmental decision-making, yet a lack of coordination among groups often becomes a bottleneck limiting production and thus availability of useful information.
The result of either insufficient data collection or inadequate data access is the same: scientists, resource managers, and other interest groups lack the inference needed to make informed ecosystem conservation and management decisions. Against this backdrop, we identify a common set of challenges to data collection and access encountered by research and monitoring groups (“data scenarios”) and offer strategies for increasing the generation and accessibility of data and ecological information. A Spanish-language translation of the manuscript is provided as supporting information (Appendix S1).
CONCEPTUAL OVERVIEW
In this section, we identify a set of common challenges with data collection and data access that limit the development of useful information needed for freshwater conservation and management. We define data utility as enabling decision-making at a range of scales, from local to global. High-utility data can be used to address project or system-specific questions and goals but can also be reused for larger-scale syntheses that contribute to regional or global understanding and decision-making. Challenges with data collection include deficiencies in abundance and coverage, representativeness, collection methods, and/or coordination between groups collecting data. Challenges with data access include poor usability or interoperability of datasets (sufficient similarity in data structure that allows for complementary datasets to be used in a single analysis), barriers to sharing or publishing data (e.g., low data availability), and poor data visibility or dissemination (e.g., datasets exist but are difficult to find or use).
Each type of data challenge can have multiple underlying causes, resulting from ecosystem attributes as well as from financial, institutional, and social environments within which research and monitoring are conducted. For example, deficiencies in data abundance and coverage commonly arise from a lack of funding and scientific infrastructure (e.g., in regions with low GDP; Hughes et al., 2021), but also from logistical challenges (e.g., remote or hazardous sites; Ghosh, 2022), or institutional, geographic, and taxonomic biases. Inadequate data access can result from a lack of data quality controls, institutional resistance to data sharing, lack of incentives or resources to publish data, and from language or knowledge barriers (Wilkinson et al., 2016). Though the result of these different data challenges is the same (i.e., less generation of useful information), potential solutions require accurately identifying and addressing the specific underlying causes (Wilkinson et al., 2016). It is also important to note that in some cases, the decision not to share data may be an act (and legal right) of tribal sovereignty to protect tribes from harmful consequences of data misuse (Carroll et al., 2020, 2021).
We conceptualize a “path” to creating useful information as occurring along two axes (Figure 2), one related to data collection (x-axis; e.g., data abundance and coverage, data collection methods, coordination among research and monitoring groups) and one related to data access (y-axis; e.g., dataset useability/interoperability, data availability, and data visibility/dissemination). Ideally, data collection and access increase in tandem, achieving high utility along the shortest path (Figure 2, black arrow). In reality, the trajectories of research or monitoring programs often diverge from this path. In many freshwater ecosystems, data collection, and even more so, data access, remain stalled at low levels, reducing the utility of what is collected (Scenario A). Another common scenario in highly monitored freshwater ecosystems is that data abundance increases much more rapidly than data access, resulting in a “plateau” of low utility data that cannot be transformed into inference (Scenario B). In a third scenario, data may be relatively abundant and accessible, but due to biased sampling, spatial, temporal, or taxonomic coverage is insufficient to achieve high utility (Scenario C). For all three data scenarios (A–C), solutions may be applied to shift the effort toward higher utility (dashed arrows).
[IMAGE OMITTED. SEE PDF]
In the following sections, we describe examples of each data scenario and their typical causes, as well as examples of successful strategies for improving data collection and access. Throughout the paper, we aim to cite existing studies that demonstrate the data challenges described in our conceptual overview and provide potential solutions. However, some of the challenges and strategies we describe have not yet been quantified in peer-reviewed literature, and we identify knowledge gaps where further research may be beneficial. Finally, there are additional considerations for the collection and sharing of data beyond those we discuss in this paper, particularly related to Indigenous data sovereignty (Carroll et al., 2020). Several recent papers have articulated the CARE Principles for Indigenous Data Governance: Collective benefit, Authority to control, Responsibility, and Ethics (e.g., Carroll et al., 2020, 2021), including recommendations on how to apply CARE Principles in ecology and biodiversity research (Jennings et al., 2023). While our work is focused on data challenges in aquatic research and monitoring, we reiterate the importance for users of Indigenous data to apply CARE Principles and FAIR Principles in tandem.
SCENARIO A: INADEQUATE DATA COLLECTION AND LOW DATA ACCESS
The most basic limitation to generating useful information is insufficient data collection, whether due to deficiencies in spatial or temporal coverage, methods, or coordination. Insufficient data collection commonly results from a lack of resources (e.g., funding and scientific infrastructure; Rogers et al., 2023) and logistical challenges (e.g., remoteness, safety, accessibility; Ghosh, 2022), both of which likely contribute to the significant positive correlation between spatial and temporal sampling coverage and per-capita GDP (e.g., Hughes et al., 2021). Insufficient data collection leads to geographic disparities in information needed for freshwater conservation and management, both within and among regions or countries. For example, in the early 2000s, residents in the watershed of Lake Yojoa, the largest lake in Honduras, became concerned about the rapid deterioration of the lake's water quality (Fadum & Hall, 2022). The cause of eutrophication was hotly debated: community residents suspected a large industrial aquaculture operation in the lake as the likely source of excess nutrients, whereas the aquaculture company pointed to a range of activities in the lake's watershed (e.g., agriculture, urbanization) as the source of the nutrients. A principal challenge to addressing the change in the state of Lake Yojoa was a complete absence of continuous monitoring, including nutrient concentrations for the lake or streams in its watershed. This lack of data inhibited management decision-making because not only was the cause of ecosystem change in question, it was also unclear whether the ecosystem had changed at all.
A similar lack of data collection occurs in Patagonia (southern South America), an extensive (1,061,000 km2) region with numerous rivers and lakes that are often the only source of water for human consumption and economic activity. Some of these water bodies show limited human influence, while others have been significantly impacted by human activities, such as hydroelectric and agricultural use, oil and gas extraction, and tourism (Alfonso et al., 2020; Scordo et al., 2020). Patagonia's human population continues to grow, creating higher demand and impact on available freshwater resources. For example, increased water consumption for human use has reduced the surface area and depth of several Patagonian lakes (Scordo et al., 2020, 2023), even causing one lake's outflow river to dry completely, resulting in water scarcity in a region of 17,600 km2 (Scordo et al., 2020). Water scarcity is likely to increase in the future as consumptive demand increases with population growth, concomitant with reductions in water supply due to climate change (e.g., projected precipitation decreases of ~10%–20%; Barros et al., 2015). Analyzing the roles of climatic trends and anthropogenic impacts on lake level and river discharge is crucial for water management planning to ensure current and future water security. However, in Patagonia, there are only a few weather and hydrological stations (Figure 1A), and basic environmental variables such as water temperature, lake level, or river discharge have never been measured in most water bodies. Water quality or biological data that do exist are often collected by local scientists with limited funding or institutional resources (Scordo, Instituto de Investigaciones en Biodiversidad y Medioambiente CONICET – UNCOMA, oral communication, July 19, 2023). The lack of environmental data makes it exceptionally challenging to discern whether the deterioration of water resources in Patagonia is due to climate change, direct human consumptive use, or both. Such a lack of knowledge also makes it difficult to project what the future holds for these resources.
The examples highlighted in Honduras and Patagonia exemplify a common pattern across Latin America (Riveros-Iregui et al., 2018) and the Global South more broadly, in which insufficient spatial and temporal data collection prevents accurate assessments of ecosystem state changes, even when those ecosystems are driving regional economies (Sterner et al., 2020). For example, the global distribution of river gaging stations remains geographically skewed, with entire continents (e.g., Asia) underrepresented both in terms of total number of gages as well as the length of available records (Figure 3A,B). Deficiencies in spatial and temporal coverage are even more extreme for environmental measurements requiring specialized, manual sampling and analytical methods, for example methane fluxes from rivers and streams (Figure 3C,D). These spatial and temporal gaps in basic monitoring data limit not only inference about a particular ecosystem, but also cross-system understanding to inform global-scale analyses.
[IMAGE OMITTED. SEE PDF]
Insufficient data collection tends to co-occur with and contribute to inadequate data access. When fewer data are collected, fewer resources are allocated to developing standardized collection protocols or creating databases or publicly available platforms at the regional or national level (Michener, 2015). In some cases, data are available from governmental agencies (Savage & Hyde, 2014), though navigating the formal request process can be daunting. Data may contain many inconsistencies due to uncalibrated equipment or use of different units for the same parameters, resulting in limited interoperability across datasets (Wilkinson et al., 2016). Incomplete or disorganized data make the development of new ideas or research projects challenging because potential users do not know the data exist or are available (e.g., lack of preliminary data to obtain funding), resulting in a self-reinforcing feedback loop that limits new data collection (Wilkinson et al., 2016).
Challenges related to publishing research and obtaining funding reinforce the condition of low data collection and access. Language barriers can create an obstacle to developing and sharing environmental data. For example, mono-lingual literature searches can unintentionally exclude available data (Nuñez et al., 2021; Zenni et al., 2023). This issue can be particularly problematic in ecological fields where global coverage is important, such as studies of biodiversity (Amano et al., 2023). Though not as well quantified as geographic and demographic disparities in publication (Liu et al., 2023; Martin et al., 2012; Nuñez et al., 2021), it is likely that journal editors and reviewers implicitly favor research conducted in well-known, data-abundant, or iconic study sites. For example, a single iconic lake with long-term data in a data-abundant region (e.g., Lake Tahoe) is more likely to be considered a “generalizable” system of broad significance given the presence of similar data within the region, whereas it is more difficult to argue the generalizability of a single lake in a data-scarce region lacking a long-term record (e.g., Nahuel Huapí; Figure 1) since there are no comparable data within the region. To our knowledge, feedbacks between data abundance, likelihood of publication or funding success, and their role in maintaining data disparities among regions or study systems have not been explicitly quantified for freshwater research. Still, more casual observations of individual scientists have suggested that studies conducted in highly monitored ecosystems may be more likely to be published in high-profile journals, yielding a feedback loop of greater research interest, funding opportunities, and data accumulation, perpetuating existing data and information gaps (Ghosh, 2022). Efforts by reviewers, editorial boards, and funding agencies to consider these dynamics can support more multifaceted perspectives on study novelty (Lieurance et al., 2022; Mahdjoub et al., 2022).
Having identified various underlying mechanisms that result in insufficient data collection and access, in the following sections we provide examples of approaches to enhance information generation in data-limited freshwater ecosystems.
Legacy datasets
In the absence of continuous monitoring that adequately captures the range of intra- and interannual variation within a freshwater ecosystem, identifying legacy data and repeating measurements years later can be used to accurately characterize state changes. In Lake Yojoa, identifying a legacy dataset from the gray literature that had sampled Lake Yojoa every 2 weeks for 2 years (Vaux & Goldman, 1984) and then repeating an analogous sampling effort for 2 years 40 years later determined that (1) the ecosystem had changed from a mesotrophic to a hypereutrophic lake, and (2) that the accumulation of nitrogen in the hypolimnion was the driver of the lake's shifting trophic state (Fadum & Hall, 2022). Similarly, in Patagonia, comparisons of legacy and contemporary data showed that soluble reactive phosphorus and dissolved inorganic nitrogen in seven Patagonian lakes had increased up to 7.4 times (Scordo et al., 2024) relative to measurements obtained 30 years prior (Diaz et al., 2007; Pizzolon, 1998; Quirós, 1988). Though in both of these examples, it was not possible to elucidate when nutrient concentrations began to increase or to definitively discern drivers, it was evident that the lakes had changed over the past 30–40 years, highlighting the value of legacy data in data-scarce ecosystems. Datasets that bookend large temporal or spatial gaps in data coverage can also be complemented by remotely sensed data (see Remote-sensing and model reanalysis), although interpolating data using different data types creates its own challenges. Though continuous long-term datasets are preferable to data with large temporal gaps, it is possible to accurately identify directional change in data-scarce systems through rigorous comparisons of legacy and contemporary data.
Remote-sensing and model reanalysis
Remote-sensing products can be used to interpolate data over time and space to create more comprehensive datasets in freshwater systems lacking long-term or high-resolution data collection. Satellite imagery can be used to observe parameters related to water quantity and quality, such as water surface elevation (Papa et al., 2023), water temperature (Herrick et al., 2023), water color (Gardner et al., 2021), and chlorophyll-a (e.g., algal blooms; Flores-Anderson et al., 2020). Many remote-sensing products are now freely available worldwide due to initiatives such as Google Earth Engine, European Space Agency (ESA) Copernicus, as well as via specific research efforts such as lake Collection 2 Surface Temperature retrieval (lakeCoSTR; Herrick et al., 2023). Freely available remote-sensing products offer a broad spectrum of radiometric, spatial (1 m–1 km), and temporal (1–15 days) resolution with global coverage, providing information on spatial and temporal variation in environmental factors worldwide. For example, Hydrography90m uses a global digital elevation model to derive networks of Earth's stream channels at high spatial resolution, including some headwaters (Amatulli et al., 2022; Schürz et al., 2023). While remote-sensing products cannot provide information prior to the satellite era or from deeper depths of the water bodies, their utility for characterizing the status and trends of surface waters will grow as these images continue to be collected.
Climate model reanalysis data such as the gridded climate data initiatives by the National Aeronautics and Space Administration (NASA), the National Oceanic and Atmospheric Administration (NOAA), and SPEI (standardized precipitation-evapotranspiration index) are also valuable sources of meteorological data in systems lacking in situ weather stations. A climate reanalysis is a comprehensive and long-term analysis of historical weather and climate data that uses advanced data assimilation techniques to combine various sources of observational data from ships, satellites, ground stations, radiosonde observations, and radar. This process generates a spatially and temporally continuous dataset of weather and climate conditions, which can be used to understand changes in freshwater ecosystems. For example, the North American Land Data Assimilation System (NLDAS) has been used in a deep learning model to generate daily estimates of lake water surface temperature (median lake specific error of 1.24°C) across 185,549 North American lakes from 1980 to 2020, only 12,227 of which had in situ temperature observations (Willard et al., 2022). In the absence of in situ meteorological data in many regions of Argentina, the SPEI has been used to show significant correlations between changes in the surface area of lakes and climate variability (Seitz et al., 2020). However, for remote-sensing and reanalysis data to provide useful information in freshwater systems with limited environmental monitoring, local capacity building and access to the technology that allows for analyses of remotely sensed data are necessary. Collaboration with international organizations, governments, and nonprofits has the potential to help leverage the full potential of remote-sensing for environmental studies in data-scarce ecosystems (Sheffield et al., 2018). Alternatively, program and/or agency managers could encourage or require capacity building as a component of remote-sensing and reanalysis data product publications.
Environmental sensors
In situ environmental sensors, many of which have become increasingly affordable relative to conventional analysis techniques, provide high-resolution data at an unprecedented scale (Marcé et al., 2016). Many new sensors are small, portable, robust to physical disturbance and temperature extremes, have a long battery life, and require little need for calibration or maintenance over the course of the entire year. This makes sensors a valuable tool for monitoring freshwater systems where site accessibility (remote, mountainous regions, under ice, etc.), scientific infrastructure and expertise, or research funding to support intensive manual sampling may be limited. Relatively inexpensive sensors can be used to infer ecosystem parameters or processes that may otherwise require expensive equipment or personnel time to quantify. For example, vertical arrays of temperature sensors in the water column can be used to estimate the attenuation of visible light (Kd; Read et al., 2015), which is a critical parameter for predicting primary productivity in aquatic systems. Inexpensive light-intensity sensors can be used to accurately detect ice breakup in lakes, which are often inaccessible during shoulder seasons due to safety issues (Block et al., 2019). Although currently 10 times more expensive than temperature and light sensors, the development of dissolved oxygen sensor technologies that go beyond electrode-based measurements has substantially reduced costs (Rode et al., 2016). These dissolved oxygen sensors can be used to monitor seasonal patterns in oxygen availability for aquatic organisms, as well as to estimate ecosystem metabolic rates (e.g., gross primary production, ecosystem respiration) that are integral to understanding carbon and nutrient cycling or food web productivity (Jankowski et al., 2021).
Strategic sensor placement within freshwater ecosystems can yield useful information for relatively low cost. In Lake Yojoa, Honduras, researchers installed temperature and pressure sensors in each of the lake's six main tributaries to provide daily estimates of discharge (Fadum et al., 2024). Discharge estimates, coupled with weekly sampling for water chemistry, were used to estimate watershed nutrient loading, compare it with nutrient inputs from industrial aquaculture into the lake, and allow determination of drivers of eutrophication. Using remotely sensed data, or by focused work in single systems (Wik et al., 2016), it is possible to determine the necessary coverage and placement of sensors a priori. For example, in Lake Yahuarcocha (Ecuador) researchers were able to reduce instrumented sites from seven to four by using spatial and temporal statistical approaches and pattern recognition techniques (Jácome et al., 2018). Local development of sensors can also increase the usage and implementation of these instruments. Initiatives or networks such as Red Latino Americana de Tecnologías Libre (REGOSH), Sensores Comunitarios (CoSensores), or Estaciones de Monitoreo Ambiental Costero (EMAC) use open access technology for both software and hardware to develop sensors to monitor environmental variables at a fraction (as low as 20%) of the cost of commercial sensors (Vitale et al., 2018).
Model transfers
Mechanistic models, statistical models, or even conceptual models developed in highly studied freshwater systems can be used to predict events or ecological characteristics in sparsely monitored systems. Examples of such model transfers are common, particularly in hydrology and species distribution modeling. Species distribution models are regularly used to project probability maps representing the potential distribution of a species of interest, both in data-abundant and data-scarce ecosystems (e.g., groundwaters; Mammola & Leroy, 2018). Hydrologic models are routinely used to predict streamflow in ungauged or poorly gauged catchments (Hrachowitz et al., 2013). However, models are overwhelmingly developed and calibrated in regions with abundant data, leaving open the question of how their extrapolations are biased and, therefore, useful in data-scarce systems (Yates et al., 2018). Model transfers are best ensconced in a framework of adaptive learning, where priority lies in testing predictions and updating model assessments and forecasts as more data become available (data-model integration; Dietze et al., 2018; Sequeira et al., 2018). Such data-model integration additionally prioritizes new data collection by focusing on testing model predictions that local experts or interest groups prioritize or find most uncertain, then honing sampling regimes. In this way, model transfers can accelerate data collection in data-scarce systems by highlighting data needs or priorities.
Community science
Community science (also referred to as “citizen science”) has a rich legacy of increasing the spatial and temporal frequency of data collection, improving inference capabilities for remote sites (e.g., Kennedy et al., 2016), or for addressing questions where high-resolution data add substantial value (Weyhenmeyer et al., 2017). For example, in the remote Grand Canyon reach of the Colorado River, ecologists outfitted recreational river rafters with scientific equipment to collect data about insect emergence (Kennedy et al., 2016) and bat activity (Metcalfe et al., 2023), greatly increasing spatial and temporal data coverage. Expanded data coverage revealed important linkages between the daily hydropeaking wave below dams and the spatial patterns in insect egg-laying (Kennedy et al., 2016), as well as between insect emergence and bat activity (Metcalfe et al., 2023).
Smartphone apps, such as iNaturalist () and eBird (), are a popular approach to community science that empowers members of the public to collect valuable scientific data (Di Cecco et al., 2021). By facilitating people's documentation of observations of flora, fauna, or environmental quality in their local environments, these apps enable rapid collection of large datasets that would be challenging for research or monitoring programs to amass. For example, the AppEAR mobile app allows users to evaluate the habitat quality of rivers, streams, lakes, and estuaries from their mobile devices by means of questionnaires and photographs (Cochero, 2018). The AppEAR project incorporated about 1000 users who sent information on over 100 sampling sites throughout Argentina. Community science initiatives may be most effective for collecting relatively simple but informative metrics such as Secchi depth (Kirby et al., 2021), water depth (Millar et al., 2023), stream flow (Kampf et al., 2018), and particulate air pollution (Lu et al., 2022). Metrics that can be collected from a mobile phone app are most accessible to the broadest population, but metrics that require a purchased sensor have also been employed successfully (Lu et al., 2022; Malthus et al., 2020; Metcalfe et al., 2023).
Training opportunities
Lack of resources and expertise in data-related fields, including data science, analytics, and technology, contributes to inadequate data access in many parts of the world. Training opportunities can empower individuals and communities to harness the benefits of data and technology for their development and well-being (Carroll et al., 2021). One strategy is to establish remote mentoring programs where experienced data professionals can provide guidance and support to learners about data organization, analysis, and visualization. Identification of specific training needs through collaboration with local universities, nongovernmental organizations (NGOs), and community groups can provide tailored programs that align with local priorities. Global training programs that focus on practical skills and hands-on experience and provide scholarship support can further enhance training opportunities. Examples of existing global training programs include those funded by the Inter-American Institute for Global Change Research (IAI) or the Global Lake Ecological Observatory Network (GLEON) Graduate Student Fellowship Program. A society journal honor program at the Association of the Sciences of Limnology and Oceanography (ASLO) provides mentorship and open access fee waivers to early career researchers that lack funding for publication costs, with an emphasis on scientists from the global south (Hotaling et al., 2023). ASLO has also developed a program called “Amplifying Voices” that gives underrepresented early career scientists the opportunity to present their work in a virtual format to an international society of listeners (Meinikmann et al., 2022).
SCENARIO B: HIGH DATA COLLECTION, LOW DATA ACCESS
Research and monitoring programs in highly monitored freshwater ecosystems face a distinct set of data-related challenges compared with ecosystems where data collection is limited (Scenario A). Data access, rather than data collection, can often pose a greater obstacle to creating useful information for managers and other interest groups in data-abundant freshwater ecosystems (Figure 2, Scenario B). Lack of reusability or interoperability, defined as datasets that can be easily integrated to create broader datasets, is a major challenge to providing data access. Lack of interoperability is especially acute when large volumes of data are collected by automated sensors, by long-term monitoring programs, and/or by multiple research groups or agencies (e.g., Figure 1, weather stations operated by multiple agencies near Lake Tahoe). As data collection protocols adapt to new technologies, integrating new and old operating procedures to provide consistently accurate and reproducible measurements can take significant effort and may be hindered by staffing turnover due to budget fluctuations and inflation pressures (Caughlan & Oakley, 2001). Maintaining data access is particularly challenging for long-term research programs spanning decades of technological innovation and personnel change. For example, high-frequency sensors are now relatively inexpensive and have been integrated into traditional point-based sampling within ecological monitoring programs (e.g., National Science Foundation Long Term Ecological Research sites), vastly increasing data volume and thus the need for updated data quality assurance (QA) quality control (QC), processing, and storage procedures.
Data access is facilitated by coordination among research and monitoring entities to create shared quality control protocols and to achieve sufficient spatial or temporal coverage. Multiple government agencies and academic research groups often simultaneously monitor the same ecosystem, yet often groups use different methods, sampling approaches, or sample at uncoordinated spatial or temporal resolution (Stompe et al., 2020), limiting data utility. The water quality monitoring program at Lake Powell Reservoir in the southwest United States provides a concrete example of Scenario B (Figure 2)—Lake Powell contains one of the longest water quality records from a human-made reservoir (Deemer et al., 2023), yet this dataset has not been integrated into relevant national or global analyses of lake and reservoir water quality (e.g., O'Reilly et al., 2015) due to its lack of organization and low visibility. Data are collected from Lake Powell by multiple federal agencies with different priorities, contributing to the lack of coordination and data sharing. The National Park Service has primarily been concerned with Escherichia coli contamination (Hoffman et al., 2025) and the establishment of invasive Quagga mussels in Lake Powell (Wenzel, 2009), the U.S. Bureau of Reclamation is primarily concerned with reservoir salinity (Deemer et al., 2023) and with the ability to predict water release temperatures and salinities under a variety of management scenarios (Williams, 2007), and the U.S. Geological Survey is primarily concerned with how reservoir water releases impact the downstream ecosystem within the context of an adaptive management program (Gloss et al., 2005). Because data collection was not coordinated with data sharing a priori, high data latency hindered data sharing and synthesis, and the Lake Powell datasets were not used to understand larger-scale trends in water quality.
Lack of institutional resources and priorities devoted to organizing, sharing, or publishing data also prevents researchers from making their datasets accessible to others (Borycz et al., 2023). Researchers typically lack sufficient time or training in best practices to publish datasets (Borycz et al., 2023). There are rarely external incentives for researchers to publish datasets, since performance evaluations, particularly in academia, usually focus on manuscript publication rather than on dataset publication (Borycz et al., 2023). In extreme cases, historical datasets literally “pile-up” as undigitized paper data sheets as monitoring programs move toward electronic data capture and storage. If digitization and QA/QC of older data are not prioritized, historical datasets are essentially lost and cannot contribute to the development of useful information for managers or other interest groups (Specht et al., 2018).
In highly monitored freshwater ecosystems where many datasets are nominally available to the public, low data visibility can also stymie data access (Haddaway & Bayliss, 2015). Many datasets are stored in non-peer-reviewed “gray literature” such as industry reports or technical reports by governments, NGOs, private firms, or through private or public utilities. For example, public utilities in Wisconsin, USA (e.g., NEW Water, the brand of the Green Bay Metropolitan Sewerage District: ), collect and manage a large database of water quality data for Green Bay in Lake Michigan (Harris et al., 2018). Although these “cryptic” datasets are publicly available through direct requests to government databases that store and manage them, in practice they can be difficult to obtain without guidance, and the data may be difficult or impossible to use if not accompanied by sufficient descriptive metadata.
In the following sections, we describe strategies to advance data utility in freshwater ecosystems where data are abundant but access remains limited.
Improve coordination among research entities
A priori planning and coordination among data collecting entities is likely to lead to reusable and interoperable datasets in data-abundant ecosystems, especially when regional-scale or cross-jurisdictional data collection are needed. The Upper Mississippi River (UMR, USA), which spans multiple state boundaries, provides one example of a data-abundant ecosystem with a highly coordinated, regional-scale data collection effort, implemented by the Upper Mississippi River Restoration (UMRR) program. The UMRR program includes both state and federal partners, with the U.S. Army Corps of Engineers implementing restoration and long-term resource monitoring (LTRM) programs, the U.S. Geological Survey coordinating the monitoring component, and five state agencies (Minnesota, Wisconsin, Iowa, Illinois, and Missouri) cooperating to collect, analyze, and publish monitoring data. River monitoring includes water quality, fisheries, and aquatic vegetation data. Data collected through LTRM is made public annually as both downloadable data files as well as viewed through a data viewer (). Sampling design and methodology are highly standardized across the ecosystem, and data management is centralized. These data are used to inform adaptive management of a highly important economic and ecological resource and to aid in habitat restoration and rehabilitation to meet specific objectives. Additionally, the long-term nature and standardization of protocols within the monitoring network allow for assessment of environmental change (Houser, 2022). Funding within the program also supports hypothesis-driven science to inform management of the ecosystem by teams of federal, state, academic, and NGO partners. These projects range in ecological scale from genetic information on fish populations (Shi et al., 2023) to hydrogeomorphic change at regional scales (Van Appledorn et al., 2021).
Successful global-scale efforts to coordinate and integrate existing monitoring efforts into networks of high temporal and spatial resolution data coverage also exist. For example, Global Ecosystem Research Infrastructure (GERI) is composed of ecological research networks covering all seven continents (Loescher et al., 2022)—this network's integrated datasets are being used to understand ecological drought at cross-continental scales. A more grassroots example of global coordination among research entities is the GLEON. GLEON-associated researchers from across the world share data in working groups to develop research papers and data analysis tools (Weathers et al., 2013). In addition to data sharing and synthesis groups, GLEON has promoted coordination among research entities by building data harmonization tools, such as ecocomDP, an R software tool (R Core Team, 2024) for formatting and reformatting data collected by different entities to promote synthetic research (O'Brien et al., 2021).
Prioritize data curation and synthesis
Research and monitoring programs can achieve higher data utility by prioritizing data curation and synthesis. To improve data access and utility by the Lake Powell water quality monitoring program, the U.S. Geological Survey funded a protocol evaluation panel through the Glen Canyon Dam Adaptive Management Program in 2017 and identified data management, metadata development, and QA/QC procedures as major areas in need of improvement (Hamilton et al., 2018). This aligned with the recent prioritization of open public access to data associated with U.S. federally funded science (U.S. Geological Survey, 2023). As a result of these efforts, the full Lake Powell dataset was recently described in a data paper (Deemer et al., 2023) and is available as a data release (Andrews & Deemer, 2022).
NGOs and synthesis centers can promote coordination in data collection and data sharing among research programs lacking existing structures or resources to do so. At the lower “barrier to action” end of the spectrum, NGOs can bring multiple groups together around a shared challenge to discuss ongoing data collection efforts and gaps, while also creating opportunities for coordination, cooperation, and new partnerships. For example the NGO CalTrout organizes regular meetings to bring together individuals from academia, government agencies, conservation groups, and Tribal Nations to determine restoration priorities and coordinate data collection efforts for salmonid fishes in a northern California, USA, watershed (“Eel River Forum”: ). More resource-intensive efforts include engaging synthesis centers to facilitate knowledge production and communication. For example, the recent “State of Alaskan Salmon and People” working group, convened by the National Center for Ecological Analysis and Synthesis (NCEAS), focused on data compilation, analyses, and visualization to identify knowledge gaps and priorities for future research and monitoring in Alaska (e.g., see project overview: and resultant data portal: ). In the San Francisco Estuary and Sacramento-San Joaquin River Delta, California, USA, a coalition of federal and state agencies (Interagency Ecological Program: ) has collected monitoring data spanning physical, chemical, and biological parameters, and has recently begun to facilitate working groups to synthesize and publish datasets collected by its multiple agency and academic partners (e.g., zooplankton: Bashevkin, Hartman, et al., 2022; e.g., water temperature: Bashevkin, Mahardja, et al., 2022). These syntheses have allowed ecosystem-wide assessments of climate-change-driven water temperature trends that were previously unclear (Bashevkin, Mahardja, et al., 2022), as well as better assessments of fish species abundances (Stompe et al., 2023), improving the utility of previously collected data.
Incentivize data publication
Data sharing has slowly become standard practice in ecology and environmental sciences, and many journals now require data sharing as a condition of manuscript publication (Tedersoo et al., 2021). Free public repositories such as Figshare (free up to 20 GB of data), Open Science Framework (OSF), or the Environmental Data Initiative (EDI) provide the infrastructure needed to publish ecological and environmental datasets. However, many journals still do not require data publication, and many researchers opt not to do so given the time and effort involved. Some journals accept statements such as “the authors will make data available upon reasonable request”; however, compliance with requests is not enforced and often has been inadequate (Sholler et al., 2019; Tedersoo et al., 2021; Vines et al., 2014). For example, Vines et al. (2014) found that the availability of research data “available upon request” declines rapidly with article age (falling by 17% per year) and suggested that the main obstacles to data sharing are broken email chains and obsolete storage devices. Because many researchers fear they may lose credit or be “scooped” by others using their published data, data repositories typically require those reusing datasets to cite the dataset and its authors (Silvello, 2018), though the compliance rate is unknown. In addition, dataset citations are often valued less highly than citations of published manuscripts, both by researchers and those evaluating their productivity, disincentivizing the nonstrivial effort required for dataset publication (Borycz et al., 2023).
To encourage researchers to publish data, some journals now encourage submissions of “data papers” which appear in regular issues of journals such as Ecology and Limnology and Oceanography Letters. Other journals, such as Scientific Data, exclusively publish data papers. Data papers range in format from basic (e.g., dataset with an abstract and enriched metadata; Chavan & Penev, 2011), to more integrative; for example, papers that show how multiple data streams can be combined to support ecosystem models, wherein the “data” published are the model output (Appling et al., 2018). Data papers are now included in services that generate publication lists and metrics (e.g., Google Scholar, Scopus) whereas datasets published in traditional repositories are not. However, some of these data journals have high article processing charges that constrain the capability of researchers to publish their data, especially in regions with low GDP (Mekonnen et al., 2022) or for researchers who must also pay the cost of English editing services (Ramírez-Castañeda, 2020). Revising incentive structures to account for overall research output (including manuscripts, data papers, and published datasets) is a systemic issue (Davies et al., 2021); however, individual reviewers or committees can begin to change norms by recognizing and rewarding researchers who publish data (Borycz et al., 2023).
SCENARIO C: LOW DATA COVERAGE AND REPRESENTATIVENESS
Freshwater ecosystems that are relatively data-abundant may still have inadequate spatial, temporal, or taxonomic data coverage. Sampling biases commonly occur when data are logistically difficult to collect, often in combination with a perception that certain sites, seasons, or species are unimportant. For site locations (e.g., subterranean), seasons (e.g., winter), or times of day (e.g., night) that pose physical access challenges, there is greater cost, effort, and safety risk involved to collect relatively fewer data (Ghosh, 2022). In one recent biodiversity study, 80% of reported records were obtained within 2.5 km of a road, resulting in over-representation relative to roadless areas (Hughes et al., 2021). As of 2015, only 2% of peer-reviewed freshwater lake research included data collected during periods of ice cover (Hampton et al., 2015), despite approximately 50% of the world's lakes developing seasonal ice cover (Verpoorter et al., 2014). Similar biased approaches are applied for saline and hypersaline lakes, which are almost always sampled when wetted despite being dry for extended periods of time and harboring crucial dormant resident biota in dried lakebed sediments (Saccò et al., 2021). Sampling bias by system size may also occur, where larger systems are relatively oversampled relative to their smaller counterparts (DeWeber et al., 2014) or vice versa (Hanson et al., 2007). For example, streams with small watersheds (<10 km2) were highly underrepresented when comparing the active U.S. Geological Survey gage sites in October 2010 to the “census population” available in the National Hydrography Dataset (NHDPlusV1, DeWeber et al., 2014; U.S. Environmental Protection Agency and U.S. Geological Survey, 2005).
Spatial and temporal data gaps often result from preexisting perceptions that there are certain sites or times where an ecological or biogeochemical process is more important. For example, many lake ecologists justified their focus on ice-free seasons under the paradigm that ecological processes were negligible in wintertime, but a recent push to better understand limnology “under the ice” has challenged this assumption (Hampton et al., 2017). In other cases, under-sampling may have more to do with the limited capability to map certain ecosystem types. For example, the probabilistic survey design implemented in the United States Environmental Protection Agency National Lakes Assessment uses the National Hydrography Dataset to select lakes and reservoirs, and this dataset can only reliably resolve waterbodies down to 0.01 km2, thus excluding the smallest ponds from the survey (U.S. Environmental Protection Agency, 2024).
Biased spatial, temporal, or taxonomic data coverage can lead to a limited or even incorrect understanding of global processes. For example, spatial and temporal bias in aquatic methane sampling has challenged efforts to upscale fluxes and integrate freshwaters into the global carbon cycle (Deemer & Holgerson, 2021). Despite a growth in the number of ecosystems where greenhouse gas emissions have been measured, many of these systems may have been selected due to their propensity to be high emitters (e.g., urban or nutrient enriched systems; Deemer & Holgerson, 2021; Stanley et al., 2016), measurements from small ponds are underrepresented relative to their surface area (Deemer & Holgerson, 2021), and few ecosystems have measurements across all seasons (Stanley et al., 2023; Zhuang et al., 2023; Figure 3D). In biological or biodiversity studies, there are taxonomic disparities in the type of biological data collected; even in highly monitored regions, most biological data are collected from charismatic organisms (e.g., vertebrates; Mammola, Fukushima, et al., 2023). Researchers tend to study, and hence accumulate more data for organisms that are of large size, colorful, useful, or harmful to humans, and those with large distribution ranges (Mammola, Adamo, et al., 2023).
The sociopolitical environment in which research and monitoring are conducted can influence temporal and spatial patterns in data collection as well. Political instability and regional conflicts create large spatial and temporal gaps in data collection (Suchikova et al., 2023). Conversely, countries undergoing large, rapid, or highly visible environmental change may have more funding and/or political will to monitor certain aspects of freshwater ecosystems. For example, some of the earliest, methodologically complete studies of reservoir methane emissions that incorporated both bubble and diffusive fluxes were done in the Brazilian Amazon. The Brazilian studies aimed to compare the greenhouse gas emissions from hydroelectricity generation with emissions from fossil fuel-based energy sources (Fearnside, 1995; Rosa et al., 1996), presumably to inform future policies on land use change. In contrast, early methane flux studies in temperate and boreal reservoirs generally only captured diffusive fluxes (Deemer et al., 2016). Because of the priority differences across regions, analyses of global datasets found latitude to be a key predictor of reservoir methane emissions, with lower latitude Amazonian reservoirs emitting much more methane than their temperate counterparts (Barros et al., 2011). Follow-up work comparing only methodologically similar (diffusive + bubble) studies found that ecosystem productivity and morphology are better predictors than latitude, and that methane emissions from Amazonian reservoirs are statistically indistinguishable from reservoir methane fluxes in other regions (Deemer et al., 2016; Deemer & Holgerson, 2021).
In the following sections, we describe strategies to improve data coverage and representativeness in freshwater ecosystems.
Redesign monitoring networks
Monitoring networks can be designed to improve data coverage and representativeness of study sites, increasing data utility without imposing additional financial or logistical costs. Strategically selecting monitoring site locations a priori can ensure that data coverage supports research goals and alleviates spatial gaps effectively. Optimal sensor locations for the United States National Ecological Observatory Network (NEON) were chosen by delineating ecoregion boundaries via statistical analysis of existing geographic and meteorological data (Hargrove & Hoffman, 2004). Existing monitoring networks can also be redesigned to improve data coverage and utility; in the Ganges River Basin in India, researchers leveraged an existing, data-rich network of rain gauges, removing sensor locations that provided redundant information and redeploying those resources in data-scarce locations (Tiwari et al., 2020).
Fund and publish studies filling data gaps
As mentioned previously for Scenario A (low data collection, low data access), low data coverage and representativeness are also reinforced by biases in the publication and funding processes. To address this issue, journal editors, reviewers, and funders can give greater consideration to place-based studies that fill knowledge gaps in data-scarce ecosystems, even if (especially if) they replicate studies done elsewhere. This could include requests for proposals that are place-or community-based, motivated by known data gaps. At the same time, editors and funders can lessen emphasis on “global” syntheses reliant on a subset of data-abundant, but potentially nonrepresentative, study systems. Including associate editors with a diversity of geographical expertise on journal editorial boards, in addition to subject matter diversity (e.g., Lieurance et al., 2022; Mahdjoub et al., 2022), may help facilitate positive feedback loops (ex., more publications lead to more funding and data collection) that improve data coverage and representativeness. Journals can explicitly encourage submissions that replicate existing measurements in poorly characterized ecosystems. For example, the journal Estuaries & Coasts publishes a series of brief reports focused on comparative studies conducted in data-scarce estuaries around the world. Similarly, funding bodies could do more to support projects that expand data collection into underrepresented areas, even if the studies are not as methodologically or conceptually novel as research proposed in data-abundant ecosystems.
CONCLUSION
Insufficient data collection and/or data access creates barriers to achieving science-informed freshwater sustainability and conservation goals. Lack of accessible data can stymie understanding and decision-making in both data-scarce and data-abundant freshwater ecosystems, though these data challenges result from a diverse set of underlying mechanisms. Despite these challenges, we highlight many examples of successful strategies for increasing data utility that can be widely adopted by the freshwater research and monitoring community. Some of these strategies include technological and analytical approaches, but many require changing the priorities and incentives of organizations, including academic or government research institutions, monitoring groups, journals, and funding agencies. Evaluating progress toward societal goals related to water quality, water supply, and freshwater biodiversity requires addressing the data challenges presented here and creating more comprehensive, representative data coverage of the world's precious freshwater ecosystems.
AUTHOR CONTRIBUTIONS
Adrianne P. Smits, Ed Hall, Matthew R. Pintar, Bridget R. Deemer, and Facundo Scordo conceived the idea for the manuscript. Adrianne P. Smits, Albert Ruhi, Carolina C. Barbosa, Bridget R. Deemer, and Facundo Scordo made the figures. Adrianne P. Smits led manuscript writing, and all authors wrote and edited the manuscript.
ACKNOWLEDGMENTS
This paper emerged from a working group at the Revisiting the Freshwater Imperative workshop, Fort Collins, CO (NSF DEB 2017732). We thank the workshop leaders, Erin Hotchkiss, Mike Vanni, Catherine O'Reilly, Steve Sadro, and Kathy Cottingham, for facilitating and bringing this author group together. We thank Steven Sadro, Samantha Oliver, and Katie Willi for helpful comments on a previous version of this manuscript. Dario Galindo translated this manuscript into Spanish, and this version is available as a supplemental file (Appendix S1); we very much appreciate his efforts to increase the accessibility of this work. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
DATA AVAILABILITY STATEMENT
All datasets used to generate figures are publicly available and cited in the paper.
Alfonso, M. B., F. Scordo, C. Seitz, G. M. Mavo Manstretta, A. C. Ronda, A. H. Arias, J. P. Tomba, L. I. Silva, G. M. E. Perillo, and M. C. Piccolo. 2020. “First Evidence of Microplastics in Nine Lakes across Patagonia (South America).” Science of the Total Environment 733: [eLocator: 139385].
Allen, G. H., and T. M. Pavelsky. 2018. “Global Extent of Rivers and Streams.” Science 361: 585–588.
Amano, T., V. Berdejo‐Espinola, M. Akasaka, M. A. de Andrade Junior, N. Blaise, J. Checco, F. G. Çilingir, et al. 2023. “The Role of Non‐English‐Language Science in Informing National Biodiversity Assessments.” Nature Sustainability 6: 845–854.
Amatulli, G., J. G. Marquez, T. Sethi, J. Kiesel, A. Grigoropoulou, M. M. Üblacker, L. Q. Shen, and S. Domisch. 2022. “Hydrography90m: A New High‐Resolution Global Hydrographic Dataset.” Earth System Science Data 14: 4525–4550.
Andrews, C. M., and B. R. Deemer. 2022. “Limnology Data from Lake Powell, Desert Southwest USA.” US Geological Survey Data Release. USGS ScienceBase Catalogue.
Van Appledorn, M., N. R. De Jager, and J. J. Rohweder. 2021. “Quantifying and Mapping Inundation Regimes within a Large River‐Floodplain Ecosystem for Ecological and Management Applications.” River Research and Applications 37: 241–255.
Appling, A. P., J. S. Read, L. A. Winslow, M. Arroita, E. S. Bernhardt, N. A. Griffiths, R. O. Hall, et al. 2018. “The Metabolic Regimes of 356 Rivers in the United States.” Scientific Data 5: [eLocator: 180292].
Barros, N., J. J. Cole, L. J. Tranvik, Y. T. Prairie, D. Bastviken, V. L. M. Huszar, P. del Giorgio, and F. Roland. 2011. “Carbon Emissions from Hydroelectric Reservoirs Linked to Reservoir Age and Latitude.” Nature Geoscience 4: 593–596.
Barros, V., C. Vera, E. Agosta, D. Araneo, I. Camilloni, A. Carril, M. Doyle, et al. 2015. “Cambio climático en Argentina; Tendencias y Proyecciones (Climate Change in Argentina: Trends and Projections).” http://3cn.cima.fcen.uba.ar/3cn_informe.php.
Bashevkin, S. M., R. Hartman, M. Thomas, A. Barros, C. E. Burdi, A. Hennessy, T. Tempel, and K. Kayfetz. 2022. “Five Decades (1972–2020) of Zooplankton Monitoring in the Upper San Francisco Estuary.” PLoS One 17: [eLocator: e0265402].
Bashevkin, S. M., B. Mahardja, and L. R. Brown. 2022. “Warming in the Upper San Francisco Estuary: Patterns of Water Temperature Change from Five Decades of Data.” Limnology and Oceanography 67: 1065–1080.
Besson, M., J. Alison, K. Bjerge, T. E. Gorochowski, T. T. Høye, T. Jucker, H. M. R. Mann, and C. F. Clements. 2022. “Towards the Fully Automated Monitoring of Ecological Communities.” Ecology Letters 25: 2753–2775.
Block, B. D., B. A. Denfeld, J. D. Stockwell, G. Flaim, H. F. Grossart, L. B. Knoll, D. B. Maier, et al. 2019. “The Unique Methodological Challenges of Winter Limnology.” Limnology and Oceanography: Methods 17: 42–57.
Borycz, J., R. Olendorf, A. Specht, B. Grant, K. Crowston, C. Tenopir, S. Allard, N. M. Rice, R. Hu, and R. J. Sandusky. 2023. “Perceived Benefits of Open Data Are Improving but Scientists Still Lack Resources, Skills, and Rewards.” Humanities and Social Sciences Communications 10(1): 1–12.
Branco, V. V., L. Correia, and P. Cardoso. 2023. “The Use of Machine Learning in Species Threats and Conservation Analysis.” Biological Conservation 283: 110091.
Carroll, S. R., I. Garba, O. L. Figueroa‐Rodríguez, J. Holbrook, R. Lovett, S. Materechera, M. Parsons, et al. 2020. “The CARE Principles for Indigenous Data Governance.” Data Science Journal 19: [eLocator: 43].
Carroll, S. R., E. Herczog, M. Hudson, K. Russell, and S. Stall. 2021. “Operationalizing the CARE and FAIR Principles for Indigenous Data Futures.” Scientific Data 8: [eLocator: 108].
Caughlan, L., and K. L. Oakley. 2001. “Cost Considerations for Long‐Term Ecological Monitoring.” Ecological Indicators 1(2): 123–134. [DOI: https://dx.doi.org/10.1016/S1470-160X(01)00015-2].
U.S. Geological Survey. 2023. Public Access to Results of Federally Funded Research at the U.S. Geological Survey: Scholarly Publications and Digital Data (ver. 2.0). Reston, VA: U.S. Department of the Interior.
Vaux, P. D., and C. R. Goldman. 1984. El Cajon Project: Ecology, Limnology and Fisheries Program 286. Davis, CA: Ecological Research Associates.
Verpoorter, C., T. Kutser, D. A. Seekell, and L. J. Tranvik. 2014. “A Global Inventory of Lakes Based on High‐Resolution Satellite Imagery.” Geophysical Research Letters 41: 6396–6402.
Vines, T. H., A. Y. Albert, R. L. Andrew, F. Débarre, D. G. Bock, M. T. Franklin, K. J. Gilbert, J. S. Moore, S. Renaut, and D. J. Rennison. 2014. “The Availability of Research Data Declines Rapidly with Article Age.” Current Biology 24(1): 94–97.
Vitale, A. J., G. M. E. Perillo, S. A. Genchi, A. H. Arias, and M. C. Piccolo. 2018. “Low‐Cost Monitoring Buoys Network Tracking Biogeochemical Changes in Lakes and Marine Environments – A Regional Case Study.” Pure and Applied Chemistry 90: 1631–1646.
Weathers, K. C., P. C. Hanson, P. Arzberger, J. Brentrup, J. Brookes, C. C. Carey, E. Gaiser, et al. 2013. “The Global Lake Ecological Observatory Network (GLEON): The Evolution of Grassroots Network Science.” Limnology and Oceanography Bulletin 22: 71–73.
Wenzel, J. M. 2009. “Aquatic Nuisance Species Management: Potential Distribution of the Invasive Zebra/Quagga Mussel in Southwestern United States.” MS thesis, Prescott College.
Weyhenmeyer, G. A., M. Mackay, J. D. Stockwell, W. Thiery, H.‐P. Grossart, P. B. Augusto‐Silva, H. M. Baulch, et al. 2017. “Citizen Science Shows Systematic Changes in the Temperature Difference between Air and Inland Waters with Global Warming.” Scientific Reports 7: [eLocator: 43890].
Wik, M., B. F. Thornton, D. Bastviken, J. Uhlbäck, and P. M. Crill. 2016. “Biased Sampling of Methane Release from Northern Lakes: A Problem for Extrapolation.” Geophysical Research Letters 43: 1256–1262.
Wilkinson, M. D., M. Dumontier, J. Ij, G. Aalbersberg, M. Appleton, A. Axton, N. Baak, et al. 2016. “The FAIR Guiding Principles for Scientific Data Management and Stewardship.” Scientific Data 3: [eLocator: 160018].
Willard, J. D., J. S. Read, S. Topp, G. J. Hansen, and V. Kumar. 2022. “Daily Surface Temperatures for 185,549 Lakes in the Conterminous United States Estimated Using Deep Learning (1980–2020).” Limnology and Oceanography Letters 7(4): 287–301.
Williams, N. T. 2007. “Modeling Dissolved Oxygen in Lake Powell Using CE‐QUAL‐W2.” MS thesis, Brigham Young University.
Williamson, C. E., J. E. Saros, W. F. Vincent, and J. P. Smol. 2009. “Lakes and Reservoirs as Sentinels, Integrators, and Regulators of Climate Change.” Limnology and Oceanography 54: 2273–2282.
Yates, K. L., P. J. Bouchet, M. J. Caley, K. Mengersen, C. F. Randin, S. Parnell, A. H. Fielding, et al. 2018. “Outstanding Challenges in the Transferability of Ecological Models.” Trends in Ecology & Evolution 33: 790–802.
Zenni, R. D., J. Barlow, N. Pettorelli, P. Stephens, R. Rader, T. Siqueira, R. Gordon, T. Pinfield, and M. A. Nuñez. 2023. “Multi‐Lingual Literature Searches Are Needed to Unveil Global Knowledge.” Journal of Applied Ecology 60: 380–383.
Zhuang, Q., M. Guo, J. M. Melack, X. Lan, Z. Tan, Y. Oh, and L. R. Leung. 2023. “Current and Future Global Lake Methane Emissions: A Process‐Based Modeling Analysis.” Journal of Geophysical Research: Biogeosciences 128: [eLocator: e2022JG007137].
© 2025. This work is published under http://creativecommons.org/licenses/by/4.0/ (the "License"). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.