Geology topics

Kelly O. Maloney

Publications and source records attributed to Kelly O. Maloney.

At least 19 recordsLinked to original sources

USGS2026

Long-term predictive modeling of stream condition suggests wide-spread changes within the Chesapeake Bay Watershed, USA

Stream ecosystems worldwide face ongoing degradation, underscoring the urgent need for conservation and restoration. Regional analyses of stream condition have been limited by sparse spatial and temporal data, particularly at long time scales. To address this gap, we used observed data to predict annual biological condition for 360,893 small, nontidal stream reaches in the Chesapeake Bay watershed from 1985 to 2023 (39 years). Predictions were generated using random forest models trained on extensive benthic macroinvertebrate data sets and predictors including natural landscape features, land cover, and climate variables. Four biological metrics were assessed: percent Ephemeroptera, Plecoptera, and Trichoptera excluding Hydropsychidae (EPT-H), percent Ephemeroptera, percent clinger functional group, and the Index of Biological Integrity (IBI). Results revealed degraded biological conditions near Washington, D.C. and Baltimore, Maryland, with declining trends across all metrics in these urbanized areas. Spatial heterogeneity was evident: IBI and clinger percentages increased in many southern streams but declined in northern streams, whereas EPT-H and Ephemeroptera decreased watershed wide except in the Southeastern Plains bioregion. By 2021, watershed wide IBI improvements were predicted for 0.9–1.1% of stream length, falling short of management goals. This study demonstrates the utility of long-term data and machine learning for predicting stream condition, identifying key stressors, and guiding restoration and conservation site selection.

Delaware, Maryland, New York, Pennsylvania, Virgin↗

USGS2026

Stream macroinvertebrate responses vary with region, land use and management practice type

Intensive land use alters hydrology and water quality, threatening freshwater benthic macroinvertebrates. Over 200,000 management practices (MPs) have been implemented across the Chesapeake Bay watershed since the 1980s, yet biological responses remain inconsistent. We synthesized 29 studies from 4 physiographic provinces covering 8 MP categories and evaluated macroinvertebrate responses along MP gradients using structural (richness), functional (biomass), tolerance, and biotic metrics. We hypothesized that MPs enhancing habitat complexity or restoring flow regimes would benefit taxa sensitive to sediment, hydrologic instability and organic pollution, with outcomes shaped by regional context, land use, and chosen metrics. Four themes emerged. (i) Agricultural Riparian Forest Buffers (RFBs) consistently improved sensitive metrics related to abundance, biomass and richness. (ii) Urban streams with Stream Habitat Improvement and Management (SHIM) showed improved richness and diversity, but biomass and tolerance metrics declined or remained neutral, indicating unresolved hydrologic and pollutant stress. (iii) Structural and functional responses diverged: effect sizes for total and feeding-group biomasses (functional metrics) were negative, whereas genus-level Ephemeroptera-Plecoptera-Trichoptera (EPT) richness (structural metric) was positive, indicating that structural shifts may not track underlying production changes. (iv) Physiographic comparisons showed counterintuitive patterns, as RFBs improved EPT richness in Piedmont streams but had negative effects in the Coastal Plain. Evaluating MP effectiveness requires distinguishing a no-MP pathway (stressors → instream conditions → assemblages → responses) from an MP-mediated pathway (practice regime → modified stressors → instream conditions → assemblages → responses), underscoring the need for region-specific, multi-metric monitoring and improved understanding of MP density thresholds and recovery lags.

eastern contiguous United States↗

USGS2025

Assessing streams in the Chesapeake Bay Watershed to guide conservation and restoration activities

Freshwater streams in the Chesapeake Bay watershed are home to numerous aquatic organisms (like fish, amphibians, mussels, and insects) and provide drinking water and recreational opportunities to people living in or visiting the watershed. Land-use changes, such as urban development and increased activities in certain agricultural sectors, have degraded water quality and altered conditions in these streams, thereby affecting their health and function. The U.S. Geological Survey (USGS) is working with Federal, State, and local partners to develop modeled assessments of stream health in freshwater streams and rivers within the Chesapeake Bay watershed. The USGS compiled large datasets for multiple stream health indicators, including instream stressors (salinity, water temperature, physical habitat, and streambank erosion) and living resources (macroinvertebrates and fish communities; fig. 1). These datasets were used by USGS scientists to develop models to predict stream health conditions across the entire region, including areas with little or no monitoring data. Collectively, these stream health assessments provide critical information to natural resource managers who implement restoration and conservation activities in the region.

Chesapeake Bay Watershed↗

USGS2025

Achieving interpretable machine learning by functional decomposition of black-box models into explainable predictor effects

Machine learning (ML) models are often based on complex black-box architectures that are difficult to interpret. This interpretability problem can hinder the use of ML in fields like medicine, ecology, and insurance, and has boosted research in interpretable machine learning (IML). Here, we propose a novel approach for the functional decomposition of black-box predictions, which is a core concept of IML. This approach replaces the prediction function with a surrogate model consisting of simpler subfunctions, providing insights into the direction and strength of the main feature contributions and their interactions. Our method is based on a concept termed “stacked orthogonality”, which ensures that the main effects capture as much functional behavior as possible. To compute the subfunctions, we combine neural additive modeling with an efficient post-hoc orthogonalization procedure. Our method yielded plausible results in an analysis of stream biological condition in the Chesapeake Bay watershed (United States).

Chesapeake Bay watershed↗

USGS2025

Tracking status and trends in seven key indicators of river and stream condition in the Chesapeake Bay watershed

Freshwater streams and rivers are recognized as vital habitats within the Chesapeake Bay watershed, which has been undergoing extensive restoration efforts for more than 30 years. Resource managers need to understand stream and river condition and how these conditions are changing over time to determine whether regional long-term restoration and conservation goals are being met. The objective of this report was to document the spatial and temporal variability of conditions for seven indicators of river and stream health across the nontidal Chesapeake Bay watershed. The framework for the U.S. Geological Survey’s Nontidal Network (NTN), a network of more than 100 nutrient and suspended sediment monitoring locations, was extended to assess conditions for six additional indicators of stream health: temperature, salinity, toxic contaminants, streamflow, hydromorphology, and biological aquatic communities. For each indicator, the latest available data from multiple sources were compiled and harmonized, and key metrics were identified to describe indicator conditions across space and time. A status condition was defined for each indicator to describe overall spatial variability in recent condition, and trend analyses were used to describe changes in each indicator metric over time. The analysis revealed clear differences in spatial and temporal data coverage across the seven indicators, so individual indicator trend analyses were not constrained to a common time interval. However, a status snapshot was conducted across all indicators for the 2015–17 period to simultaneously explore spatial variability across all indicators. The status snapshot highlighted general degraded conditions across multiple indicators in large metropolitan regions, such as the Baltimore–Washington, D.C., metropolitan area. Regression analysis between indicator status metrics and major land cover for the sites suggest urbanization as a potential driver of degraded conditions for many of the indicator metrics, including total phosphorus, salinity, temperature, high-flow frequency, and metrics of habitat and biological assemblage quality. A final analysis exploring the spatial representation of each indicator network showed that some indicator monitoring networks did not cover certain settings, such as small watersheds. These results provided an initial assessment of stream health status and trends and will continue to be leveraged to describe conditions across the Chesapeake Bay watershed to help inform local and regional management decisions. These results also highlighted the need for improved coordination among monitoring organizations to support long-term multi-indicator monitoring and assessment.

Chesapeake Bay watershed↗

USGS2025

Stressor-driven changes in freshwater biological indicators inform spatial management strategies using expert knowledge, observational data, and hierarchical models

Stream ecosystems face continuous pressures from multiple anthropogenic stressors that reshape biological communities and impact ecosystem health and services. Managers can encounter challenges in stewarding ecosystems threatened by multiple stressors, in part because most multiple stressor studies are experimental and, while valuable, offer limited management relevance in targeting these stressors on the landscape. Recent advances in causal inference coupled with large biomonitoring data sets could further understanding of observational stressor-response relationships, aiding management. In this study, we use bioassessment data in the Chesapeake Bay watershed in the mid-Atlantic region of the United States to identify how water quality and physical habitat stressors influence key benthic macroinvertebrate response metrics, considering hierarchical relationships using Bayesian networks. Results suggest water temperature and specific conductivity were prevalent stressors in a mountainous region (northern Appalachians), whereas in an agriculturally dominated region (southern Appalachians) physical habitat alterations were the predominant stressor. In mixed-land use regions (Piedmont & Coastal Plains), specific conductivity was a key stressor, but habitat heterogeneity was important for macroinvertebrate metrics. To illustrate how these stressor-response relationships can be used to guide management decisions, we applied the resist-accept-direct (RAD) framework to develop a portfolio of management options based on predicted changes in macroinvertebrate metrics in response to physical habitat and water quality stressors. For example, accepting changes in areas with co-occurring stressors may be the most feasible option, whereas directing changes through stream restoration or water quality improvements may be effective in areas with single stressor groups. By leveraging observational bioassessment data and causal inference to identify key stressor-response relationships, this research supports decision making by building a simple, strategic management portfolio.

Delaware, Maryland, New York, Pennsylvania, Virgin↗

USGS2025

Physical habitat is more than a sediment issue: A multi-dimensional habitat assessment indicates new approaches for river management

Degraded physical habitat is a common stressor affecting river ecosystems and typically addressed in the United States (US) through a regulatory focus on sediment. However, a narrow regulatory focus on sediment may overlook other aspects of physical habitat and the processes for its creation, maintenance, and degradation. In addition, there exist few “ready-to-use” regional assessments of the multiple dimensions of physical habitat to better understand continuous patterns of condition and prioritize management efforts across a large spatial scale. In this study, we use rapid habitat monitoring data to train a machine-learning ( i.e., random forest) model to predict twelve physical habitat metrics for nearly 120,000 km of nontidal rivers and streams across the Chesapeake Bay watershed, US. We capture a range of habitat conditions driven by both natural variables and anthropogenic pressures. Covariation among habitat metrics indicated two major dimensions of habitat variation: 1) coarse bed substrate and hydromorphic heterogeneity and 2) bank stability and riparian condition. The model predicted localized changes from 2001 to 2019, and the predicted areas of deterioration roughly balanced improvements across the watershed, indicating little progress towards long-term watershed management goals. To evaluate connections to regulatory and management endpoints, we compared our physical habitat predictions to paired estimates of sediment and flow alteration across the region. Sediment concentrations were greater in reaches with less bank stability and lower riparian quality; however, the relation was weak for coarse bed condition metrics, including embeddedness, which is frequently used for establishing regulatory sediment restrictions. For flow alteration, most habitat metrics had lower scores with altered flow metrics, but metrics of instream habitat heterogeneity and coarse substrate condition were most strongly affected. Increased flashy, high flows negatively affected most metrics, but coarse substrate metrics were also negatively affected by greater low flow severity. This study highlights a potential disconnect between a narrow focus on regulatory sediment targets given the multiple dimensions and responses of physical habitat. A more holistic approach to physical habitat in management interventions – one that considers hydromorphic processes, diversity and variability in microhabitats, and explicit consideration of alterations to both low and high flows – may be warranted. By providing direct estimates of multiple aspects of physical habitat, this model can help support managers in the Chesapeake Bay watershed to better understand the range of habitat conditions, identify high-quality reaches for conservation, and target potential management actions tailored to localized conditions.

Delaware, Maryland, New York, Pennsylvania, Virgin↗

USGS2025

Integrating theory and empirical patterns: Fish body size distributions, life history traits, and environmental flows in streams

Individual size distributions (ISDs) are prominent in ecological research and may support resource managers with ecosystem-scale objectives. We use a database of individual size measurements for US stream fishes to test for direct and indirect effects of traits, flow regimes, and land use on the interspecific ISD exponent. Path analysis indicates that traits have strong, direct effects on ISD. Flow and land use effects on the exponent are largely indirectly mediated by their influences on species traits. ISD exponents increase (abundances of larger-bodied individuals increase, relative to smaller-bodied) when environments favor higher trophic levels, warmer thermal tolerances, and periodic life histories. Alternatively, ISD exponents decrease in systems that favor opportunistic life histories. Our flexible modeling framework that includes direct and indirect effects of traits, flow regimes, and land use on ISD could be expanded to incorporate additional variables that interact with flow (e.g., temperature and physical habitat) to assess of effects of multiple stressors on aquatic ecosystem functioning.

Science Advances↗

USGS2025

Leveraging artificial intelligence and machine learning to advance Chesapeake Bay research and management: A review of status, challenges, and opportunities

The Chesapeake Bay and its watershed (hereafter “Chesapeake Bay region”) have been the focus of extensive restoration efforts for several decades. These restoration efforts are guided by the Chesapeake Bay Watershed Agreement (Chesapeake Executive Council 2014) which outlines 10 goals and 31 measurable outcomes. The Chesapeake Bay is globally recognized as a model for coastal restoration due to long-term investments in monitoring, modeling, implementation and research by the Chesapeake Bay Program (CBP) partnership. These monitoring network spans tidal and non-tidal regions and provides data across multiple scales. Artificial intelligence (AI), particularly machine-learning (ML) and deep learning (DL), has emerged as a powerful tool for analyzing large, complex datasets. These techniques have gained widespread adoption across various disciplines, including ecology, hydrology, and environmental science. In the Bay context, AI/ML is increasingly being used to explore drivers of environmental change, analyze system dynamics, and predict conditions in areas with limited monitoring. The CBP partnership, particularly its Scientific and Technical Advisory Committee (STAC), has increasingly recognized the growing role of AI/ML in watershed and estuarine management. Recent Chesapeake Community Research Symposium sessions and initiatives such as the Chesapeake Global Collaboratory highlight increasing regional momentum to apply big data and AI/ML for environmental solutions. Together, these developments underscore the timely need to explore how AI/ML can help advance Chesapeake Bay restoration and management. This STAC workshop, titled “Leveraging Artificial Intelligence and Machine learning to Advance Chesapeake Bay Research and Management: A review of status, challenges, and opportunities,” was held from February 24-25, 2025, in Edgewater, Maryland to bring together over 50 federal, state, and academic scientists and partners to synthesize the current state of AI/ML applications and identify research gaps in Chesapeake Bay research and management. The workshop focused on three main objectives: 1. Summarize recent AI/ML applications and lessons learned in both tidal and nontidal areas of the Chesapeake Bay region. 2. Identify challenges and gaps in applying AI/ML approaches to Chesapeake Bay data. Such challenges and gaps may include data limitations, harmonization issues, ineffective communication of AI/ML insights, and a lack of coordination among research and management institutions. 3. Develop recommendations and identify opportunities for leveraging AI/ML to address issues across the Chesapeake Bay region. Key areas of focus may include generating new information to support watershed management, delivering AI/MLgenerated insights to managers in a clear and actionable way, and fostering greater collaboration among stakeholders within the CBP Partnership. Workshop participants engaged in science presentations and breakout sessions to develop recommendations for advancing the integration of AI/ML techniques into research and management across the Chesapeake Bay region. By synthesizing current applications, identifying challenges, and exploring new opportunities, the workshop has provided valuable insights and recommendations for better leveraging AI/ML approaches to support the success of Bay restoration efforts. Together, these recommendations provide a roadmap for enhancing data-driven, science-based decision making aligned with the goals and outcomes of the Chesapeake Bay Watershed Agreement.

Delaware, Maryland, Virginia↗

USGS2024

Connecting conservation practices to local stream health in the Chesapeake Bay watershed

The Chesapeake Bay Partnership is implementing conservation practices (CPs) throughout the Chesapeake Bay watershed to reduce nutrient and sediment delivery to the Bay. This study intends to provide an integrated and detailed understanding of how local streams respond to these CP-driven management efforts. Key issue: To what extent do CPs positively affect the health of local streams in the nontidal watershed (cobenefits)? Critical unknown: How do CPs change water quality and the stressors that affect stream aquatic life? Which CPs improve stream health more effectively? Critical knowledge to be delivered to stakeholders includes— the effects of CPs on local water-quality conditions, the degree to which these same CPs also provide local stream-ecosystem benefits, and a deeper understanding of local stream-ecosystems, including stressors and CPs, to guide the selection of management efforts that enhance both water quality and overall stream-ecosystem health.

Chesapeake Bay watershed↗

USGS2024

Causal inference approaches reveal both positive and negative unintended effects of agricultural and urban management practices on instream biological condition

Agricultural and urban management practices (MPs) are primarily designed and implemented to reduce nutrient and sediment concentrations in streams. However, there is growing interest in determining if MPs produce any unintended positive effects, or co-benefits, to instream biological and habitat conditions. Identifying co-benefits is challenging though because of confounding variables (i.e., those that affect both where MPs are applied and stream biota), which can be accounted for in novel causal inference approaches. Here, we used two causal inference approaches, propensity score matching (PSM) and Bayesian network learning (BNL), to identify potential MP co-benefits in the Chesapeake Bay watershed portion of Maryland, USA. Specifically, we examined how MPs may modify instream conditions that impact fish and macroinvertebrate indices of biotic integrity (IBI) and functional and taxonomic endpoints. We found evidence of positive unintended effects of MPs for both benthic macroinvertebrates and fish indicated by higher IBI scores and specific endpoints like the number of scraper macroinvertebrate taxa and lithophilic spawning fish taxa in a subset of regions. However, our results also suggest MPs have negative unintended effects, especially on sensitive benthic macroinvertebrate taxa and key instream habitat and water quality metrics like specific conductivity. Overall, our results suggest MPs offer co-benefits in some regions and catchments with largely degraded conditions but can have negative unintended effects in some regions, especially in catchments with good biological conditions. We suggest the number and types of MPs drove these mixed results and highlight carefully designed MP implementation that incorporates instream biological data at the catchment scale could facilitate co-benefits to instream biological conditions. Our study underscores the need for more research on identifying effects of individual MP types on instream biological and habitat conditions.

Maryland↗

USGS2024

Assessing the added value of antecedent streamflow alteration information in modeling stream biological condition

In stream systems, disentangling relationships between biology and flow and subsequent prediction of these relationships to unsampled streams is a common objective of large-scale ecological modeling. Often, streamflow metrics are derived from aggregating continuous streamflow records available at a subset of stream gages into long-term flow regime descriptors. Despite demonstrated value, shortcomings of these long-term approaches include spatial restriction to locations with long-term continuous flow records (commonly, biased toward larger systems) and omission of potentially ecologically important short-term (i.e., ≤1 year) antecedent streamflow information. We used long-term flow regime and short-term antecedent streamflow alteration information to evaluate relative performance in modeling stream fish biological condition. We compared results to understand whether short-term antecedent streamflow information improved models of fish biological condition. Results indicated that models incorporating short-term antecedent data performed better than those relying solely on long-term flow regime data (kappa statistic = 0.29 and 0.23, respectively) and improved prediction accuracy among stream sizes and in six of nine ecoregions. Additionally, models relying solely on short-term streamflow information performed similarly to those with only long-term streamflow information (kappa = 0.23). Incorporating short-term antecedent streamflow metrics may provide added ecological information not fully captured by long-term flow regime summaries in macroscale modeling efforts or perform similarly to long-term streamflow data when long-term data are not available.

Science of the Total Environment↗

USGS2024

Evaluating the effectiveness of joint species distribution modeling for fresh water fish communities within large watersheds

Accurately predicting species’ distributions is critical for the management and conservation of fish and wildlife populations. Joint Species Distribution Models (JSDMs) account for dependencies between species often ignored by traditional species distribution models. We evaluated how a JSDM approach could improve predictive strength for stream fish communities within large watersheds (the Chesapeake Bay Watershed, USA), using a cross-validation study of JSDMs fit to data from over 50 species. Our results suggest that conditional predictions from JSDMs have the potential to make large improvements in predictive accuracy for many species, particularly for more generalist species where single species models may not perform well. For some species there was no added explanatory effect from conditional information, most of which already exhibited strong marginal predictive ability. For several rare species there were significant improvements in occurrence predictions, while the results for two invasive species considered did not show the same improvements. Overall, the optimal number of species to condition upon, as well as the effects of conditioning upon an increasing number of species, varied widely among species.

Chesapeake Bay watershed↗

USGS2023

Observed and projected functional reorganization of riverine fish assemblages from global change

Climate and land-use/land-cover change (‘global change’) are restructuring biodiversity, globally. Broadly, environmental conditions are expected to become warmer, potentially drier (particularly in arid regions), and more anthropogenically developed in the future, with spatiotemporally complex effects on ecological communities. We used functional traits to inform Chesapeake Bay Watershed fish responses to future climate and land-use scenarios (2030, 2060, and 2090). We modelled the future habitat suitability of focal species representative of key trait axes (substrate, flow, temperature, reproduction, and trophic) and used functional and phylogenetic metrics to assess variable assemblage responses across physiographic regions and habitat sizes (headwaters through large rivers). Our focal species analysis projected future habitat suitability gains for carnivorous species with preferences for warm water, pool habitats, and fine or vegetated substrates. At the assemblage level, models projected decreasing habitat suitability for cold-water, rheophilic, and lithophilic individuals but increasing suitability for carnivores in the future across all regions. Projected responses of functional and phylogenetic diversity and redundancy differed among regions. Lowland regions were projected to become less functionally and phylogenetically diverse and more redundant while upland regions (and smaller habitat sizes) were projected to become more diverse and less redundant. Next, we assessed how this model projected assemblage changes 2005-2030 related to observed time-series trends (1999–2016). Halfway through the initial projecting period (2005–2030), we found observed trends broadly followed modelled patterns of increasing proportions of carnivorous and lithophilic individuals in lowland regions but showed opposing patterns for functional and phylogenetic metrics. Leveraging observed and predicted analyses simultaneously helps elucidate the instances and causes of discrepancies between model predictions and ongoing observed changes. Collectively, results highlight the complexity of global change impacts across broad landscapes that likely relate to differences in assemblages’ intrinsic sensitivities and external exposure to stressors.

Global Change Biology↗

USGS2023

Tracking status and trends in seven key indicators of stream health in the Chesapeake Bay watershed

“The Bay Connects us, the Bay reflects us” writes Tom Horton in the book “Turning the Tide—Saving the Chesapeake Bay”. The Chesapeake Bay watershed contains the largest estuary in the United States. The watershed stretches north to Cooperstown, New York, south to Lynchburg and Virginia Beach, Virginia, west to Pendleton County, West Virginia, and east to Seaford, Delaware, and Scranton, Pennsylvania. The watershed is more than 64,000 square miles that contain 150 major rivers and streams, hereafter referred to collectively as streams, that total more than 100,000 miles in length. The watershed contains thousands of smaller creeks and tributaries, large numbers of plants and animals, and, in 2020, more than 18.4 million people. As changes occur in population, land use, and climate within the watershed, so too do the diversity and health of the Bay's ecosystems.

Chesapeake Bay watershed↗

USGS2022

Multispecies approaches to status assessments in support of endangered species classifications

Multispecies risk assessments have developed within many international conservation programs, reflecting a widespread need for efficiency. Under the United States Endangered Species Act (ESA), multispecies assessments ultimately lead to species-level listing decisions. Although this approach provides opportunities for improved efficiency, it also risks overwhelming or biasing the assessment process and would benefit from clear guidance for practitioners. We reviewed multispecies assessments conducted between 1993 and 2019 for ESA listing decisions to identify the ecological basis for combining species, the assessment approach used, and the policy factors influencing their efficacy. We identified 42 cases covering 359 species. Most assessments (81%) included two to five species, although the maximum was 82. A common theme involved grouping narrow endemics or habitat specialists based on taxonomic relatedness, similar distributions, and common threats to persistence. All assessments included a combined threats analysis, but few employed a common species' response model or expert elicitation process. Although ESA risk assessments are distinct from policy decisions, most assessments (50%) supported decisions that all species warranted endangered status. Available guidance has generally emphasized ecological similarity as the key attribute leading to successful multispecies assessments. The challenge with consistently selecting species based on qualitative proxies such as common distributions or threats to persistence is that ecological patterns and processes are scale dependent. Focusing instead on the assessment methods and their potential for bias and increased efficiency may provide a stronger basis for developing consistent and transparent guidance.

Conservation Science and Practice↗

USGS2022

Using fish community and population indicators to assess the biological condition of streams and rivers of the Chesapeake Bay watershed, USA

The development of indicators to assess relative freshwater condition is critical for management and conservation. Predictive modeling can enhance the utility of indicators by providing estimates of condition for unsurveyed locations. Such approaches grant understanding of where “good” and “poor” conditions occur and provide insight into landscape contexts supporting such conditions. However, as assessments are conducted at large extents crossing jurisdictional boundaries, combined datasets are likely not suited for traditional assessment approaches which rely on jurisdictionally-specific reference sites. Here, we used a large dataset compiled from multiple providers to assess the condition of fish habitat for non-tidal streams and rivers in the Chesapeake Bay watershed (CBW), USA. We concurrently used community and species-level analyses to provide a more holistic view of habitat conditions by using random forest models to predict selected metrics and species occurrence with landscape data for inland CBW stream reaches. Community analyses included metrics describing composition, tolerances, habitat preferences, and functional traits of fish communities whereas species-level analyses consisted of distribution models for key sensitive and gamefish species. For community analyses, a final index was calculated as the average of selected metric deciles with higher scores inferring less biologically altered (i.e., better) conditions, providing an alternative to using reference sites. For species analyses, species occurrence was predicted for stream reaches, with presence indicating suitable habitat. Uncertainty was calculated for both approaches using model prediction intervals. Results indicated different numbers of suitable metrics for each region, with most in the Northern Appalachian (15) and least in the Southern Appalachian Piedmont (3). Four species (three sensitive) were suitable for modeling. At the CBW scale, predictions did not vary greatly among deciles for the community or species analyses for 2001, 2006, 2011, and 2016. Most stream reaches did not vary in mean decile rank or in species occurrence between 2001 and 2016; however, the largest community changes occurred in large rivers in the Coastal Plains ecoregion and the largest species occurrence changes occurred in Torrent Suckers in medium-sized rivers. When compared, results from community analyses agreed for one sensitive species (Brook Trout) but not the other three, potentially due to regionally inappropriate tolerance assignment. Comparisons also demonstrated substantial variation among approaches suggesting a lack of redundancy. While each approach traditionally has its targeted audience and respective strengths and weaknesses, concurrent use of these approaches permits direct comparisons and may assuage shortcomings of each approach when considered separately.

Chesapeake Bay watershed↗

USGS2022

Explainable machine learning improves interpretability in the predictive modeling of biological stream conditions in the Chesapeake Bay Watershed, USA

Anthropogenic alterations have resulted in widespread degradation of stream conditions. To aid in stream restoration and management, baseline estimates of conditions and improved explanation of factors driving their degradation are needed. We used random forests to model biological conditions using a benthic macroinvertebrate index of biotic integrity for small, non-tidal streams (upstream area ≤200 km 2 ) in the Chesapeake Bay watershed (CBW) of the mid-Atlantic coast of North America. We utilized several global and local model interpretation tools to improve average and site-specific model inferences, respectively. The model was used to predict condition for 95,867 individual catchments for eight periods (2001, 2004, 2006, 2008, 2011, 2013, 2016, 2019). Predicted conditions were classified as Poor, FairGood, or Uncertain to align with management needs and individual reach lengths and catchment areas were summed by condition class for the CBW for each period. Global permutation and local Shapley importance values indicated percent of forest, development, and agriculture in upstream catchments had strong impacts on predictions. Development and agriculture negatively influenced stream condition for model average (partial dependence [PD] and accumulated local effect [ALE] plots) and local (individual condition expectation and Shapley value plots) levels. Friedman's H-statistic indicated large overall interactions for these three land covers, and bivariate global plots (PD and ALE) supported interactions among agriculture and development. Total stream length and catchment area predicted in FairGood conditions decreased then increased over the 19-years (length/area: 66.6/65.4% in 2001, 66.3/65.2% in 2011, and 66.6/65.4% in 2019). Examination of individual catchment predictions between 2001 and 2019 showed those predicted to have the largest decreases in condition had large increases in development; whereas catchments predicted to exhibit the largest increases in condition showed moderate increases in forest cover. Use of global and local interpretative methods together with watershed-wide and individual catchment predictions support conservation practitioners that need to identify widespread and localized patterns, especially acknowledging that management actions typically take place at individual-reach scales.

Chesapeake Bay Watershed↗