Search USGS⌕ Search

SEARCH · Search USGS

Results for “Ecological Modeling”

Search indexed USGS publications on groundwater, aquifers, geologic maps, mineral resources and earthquakes. Explore source records by subject and place.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 811 records · Page 45Linked to original sources
USGS2015

Potential for real‐time understanding of coupled hydrologic and biogeochemical processes in stream ecosystems: Future integration of telemetered data with process models for glacial meltwater streams

While continuous monitoring of streamflow and temperature has been common for some time, there is great potential to expand continuous monitoring to include water quality parameters such as nutrients, turbidity, oxygen, and dissolved organic material. In many systems, distinguishing between watershed and stream ecosystem controls can be challenging. The usefulness of such monitoring can be enhanced by the application of quantitative models to interpret observed patterns in real time. Examples are discussed primarily from the glacial meltwater streams of the McMurdo Dry Valleys, Antarctica. Although the Dry Valley landscape is barren of plants, many streams harbor thriving cyanobacterial mats. Whereas a daily cycle of streamflow is controlled by the surface energy balance on the glaciers and the temporal pattern of solar exposure, the daily signal for biogeochemical processes controlling water quality is generated along the stream. These features result in an excellent outdoor laboratory for investigating fundamental ecosystem process and the development and validation of process‐based models. As part of the McMurdo Dry Valleys Long‐Term Ecological Research project, we have conducted field experiments and developed coupled biogeochemical transport models for the role of hyporheic exchange in controlling weathering reactions, microbial nitrogen cycling, and stream temperature regulation. We have adapted modeling approaches from sediment transport to understand mobilization of stream biomass with increasing flows. These models help to elucidate the role of in‐stream processes in systems where watershed processes also contribute to observed patterns, and may serve as a test case for applying real‐time stream ecosystem models.

Water Resources Research↗
USGS2017

Experts correctly describe demography associated with historical decline of the endangered Indiana bat, but not recent period of stationarity

Demographic characteristics of bats are often insufficiently described for modeling populations. In data poor situations, experts are often relied upon for characterizing ecological systems. In concert with the development of a matrix model describing Indiana bat ( Myotis sodalis ) demography, we elicited estimates for parameterizing this model from 12 experts. We conducted this elicitation in two stages, requesting expert values for 12 demographic rates. These rates were adult and juvenile seasonal (winter, summer, fall) survival rates, pup survival in fall, and propensity and success at breeding. Experts were most in agreement about adult fall survival (3% Coefficient of Variation) and least in agreement about propensity of juveniles to breed (37% CV). The experts showed greater concordance for adult ( mean CV, adult = 6.2%) than for juvenile parameters ( mean CV, juvenile = 16.4%), and slightly more agreement for survival (mean CV, survival = 9.8%) compared to reproductive rates ( mean CV, reproduction = 15.1%). However, survival and reproduction were negatively and positively biased, respectively, relative to a stationary dynamic. Despite the species exhibiting near stationary dynamics for two decades prior to the onset of a potential extinction-causing agent, white-nose syndrome, expert estimates indicated a population decline of -11% per year (95% CI = -2%, -20%); quasi-extinction was predicted within a century ( mean = 61 years to QE, range = 32, 97) by 10 of the 12 experts. Were we to use these expert estimates in our modeling efforts, we would have errantly trained our models to a rapidly declining demography asymptomatic of recent demographic behavior. While experts are sometimes the only source of information, a clear understanding of the temporal and spatial context of the information being elicited is necessary to guard against wayward predictions.

PeerJ↗
USGS2018

Adaptive management of animal populations with significant unknowns and uncertainties: A case study

Conservation and management decision making in natural resources is challenging due to numerous uncertainties and unknowns, especially relating to understanding system dynamics. Adaptive resource management (ARM) is a formal process to making logical and transparent recurrent decisions when there are uncertainties about system dynamics. Despite wide recognition and calls for implementing adaptive natural resource management, applications remain limited. More common is a reactive approach to decision making, which ignores future system dynamics. This contrasts with ARM, which anticipates future dynamics of ecological process and management actions using a model-based framework. Practitioners may be reluctant to adopt ARM because of the dearth of comparative evaluations between ARM and more common approaches to making decisions. We compared the probability of meeting management objectives when managing a population under both types of decision frameworks, specifically in relation to typical uncertainties and unknowns. We use a population of Sandhill Cranes as our case study. We evaluate each decision process under varying levels of monitoring and ecological uncertainty, where the true underlying population dynamics followed a stochastic age-structured population model with environmentally driven vital rate density dependence. We found that the ARM framework outperformed the currently employed reactive decision framework to manage Sandhill Cranes in meeting the population objective across an array of scenarios. This was even the case when the candidate set of population models contained only naïve representations of the true population process. Under the reactive decision framework, we found little improvement in meeting the population objective even if monitoring uncertainty was eliminated. In contrast, if the population was monitored without error within the ARM framework, the population objective was always maintained, regardless of the population models considered. Contrary to expectation, we found that age-specific optimal harvest decisions are not always necessary to meet a population objective when population dynamics are age structured. Population managers can decrease risks and gain transparency and flexibility in management by adopting an ARM framework. If population monitoring data has high sampling variation and/or limited empirical knowledge is available for constructing mechanistic population models, ARM model sets should consider a range of mechanistic, descriptive, and predictive model types.

Ecological Applications↗
USGS2007

Year-class formation of upper St. Lawrence River northern pike

Variables associated with year-class formation in upper St. Lawrence River northern pike Esox lucius were examined to explore population trends. A partial least-squares (PLS) regression model (PLS 1) was used to relate a year-class strength index (YCSI; 1974-1997) to explanatory variables associated with spawning and nursery areas (seasonal water level and temperature and their variability, number of ice days, and last day of ice presence). A second model (PLS 2) incorporated four additional ecological variables: potential predators (abundance of double-crested cormorants Phalacrocorax auritus and yellow perch Perca flavescens), female northern pike biomass (as a measure of stock-recruitment effects), and total phosphorus (productivity). Trends in adult northern pike catch revealed a decline (1981-2005), and year-class strength was positively related to catch per unit effort (CPUE; R2 = 0.58). The YCSI exceeded the 23-year mean in only 2 of the last 10 years. Cyclic patterns in the YCSI time series (along with strong year-classes every 4-6 years) were apparent, as was a dampening effect of amplitude beginning around 1990. The PLS 1 model explained over 50% of variation in both explanatory variables and the dependent variable, YCSI first-order moving-average residuals. Variables retained (N = 10; Wold's statistic ??? 0.8) included negative YCSI associations with high summer water levels, high variability in spring and fall water levels, and variability in fall water temperature. The YCSI exhibited positive associations with high spring, summer, and fall water temperature, variability in spring temperature, and high winter and spring water level. The PLS 2 model led to positive YCSI associations with phosphorus and yellow perch CPUE and a negative correlation with double-crested cormorant abundance. Environmental variables (water level and temperature) are hypothesized to regulate northern pike YCSI cycles, and dampening in YCSI magnitude may be related to a combination of factors, including wetland habitat changes, reduced nutrient loading, and increased predation by double-crested cormorants. ?? Copyright by the American Fisheries Society 2007.

North American Journal of Fisheries Management↗
USGS2015

Estimating demographic parameters using a combination of known-fate and open N -mixture models

Accurate estimates of demographic parameters are required to infer appropriate ecological relationships and inform management actions. Known-fate data from marked individuals are commonly used to estimate survival rates, whereas N-mixture models use count data from unmarked individuals to estimate multiple demographic parameters. However, a joint approach combining the strengths of both analytical tools has not been developed. Here we develop an integrated model combining known-fate and open N-mixture models, allowing the estimation of detection probability, recruitment, and the joint estimation of survival. We demonstrate our approach through both simulations and an applied example using four years of known-fate and pack count data for wolves (Canis lupus). Simulation results indicated that the integrated model reliably recovered parameters with no evidence of bias, and survival estimates were more precise under the joint model. Results from the applied example indicated that the marked sample of wolves was biased toward individuals with higher apparent survival rates than the unmarked pack mates, suggesting that joint estimates may be more representative of the overall population. Our integrated model is a practical approach for reducing bias while increasing precision and the amount of information gained from mark–resight data sets. We provide implementations in both the BUGS language and an R package.

Alaska↗
USGS2004

Two-dimensional hydrodynamic simulation of surface-water flow and transport to Florida Bay through the Southern Inland and Coastal Systems (SICS)

Successful restoration of the southern Florida ecosystem requires extensive knowledge of the physical characteristics and hydrologic processes controlling water flow and transport of constituents through extremely low-gradient freshwater marshes, shallow mangrove-fringed coastal creeks and tidal embayments, and near-shore marine waters. A sound, physically based numerical model can provide simulations of the differing hydrologic conditions that might result from various ecosystem restoration scenarios. Because hydrology and ecology are closely linked in southern Florida, hydrologic model results also can be used by ecologists to evaluate the degree of ecosystem restoration that could be achieved for various hydrologic conditions. A robust proven model, SWIFT2D, (Surface-Water Integrated Flow and Transport in Two Dimensions), was modified to simulate Southern Inland and Coastal Systems (SICS) hydrodynamics and transport conditions. Modifications include improvements to evapotranspiration and rainfall calculation and to the algorithms that describe flow through coastal creeks. Techniques used in this model should be applicable to other similar low-gradient marsh settings in southern Florida and elsewhere. Numerous investigations were conducted within the SICS area of southeastern Everglades National Park and northeastern Florida Bay to provide data and parameter values for model development and testing. The U.S. Geological Survey and the National Park Service supported investigations for quantification of evapotranspiration, vegetative resistance to flow, wind-induced flow, land elevations, vegetation classifications, salinity conditions, exchange of ground and surface waters, and flow and transport in coastal creeks and embayments. The good agreement that was achieved between measured and simulated water levels, flows, and salinities through minimal adjustment of empirical coefficients indicates that hydrologic processes within the SICS area are represented properly in the SWIFT2D model, and that the spatial and temporal resolution of these processes in the model is adequate. Sensitivity analyses were conducted to determine the effect of changes in boundary conditions and parameter values on simulation results, which aided in identifying areas of greatest uncertainty in the model. The parameter having the most uncertainty (most in need of further field study) was the flow coefficient for coastal creeks. Smaller uncertainties existed for wetlands frictional resistance and wind. Evapotranspiration and boundary inflows indicated the least uncertainty as determined by varying parameters used in their formulation and definition. Model results indicated that wind was important in reversing coastal creek flows. At Trout Creek (the major tributary connecting Taylor Slough wetlands with Florida Bay), flow in the landward direction was not simulated properly unless wind forcing was included in the simulation. Simulations also provided insight into the major influence that wind has on salinity mixing along the coast, the varying distribution of wetland flows at differing water levels, and the importance of topography in controlling flows to the coast. Slight topographic variations were shown to highly influence the routing of water. A multiple regression analysis was performed to relate inflows at the northern boundary of Taylor Slough bridge to a major pump station (S-332) north of the SICS model area. This analysis allows Taylor Slough bridge boundary conditions to be defined for the model from operating scenarios at S-332, which should facilitate use of the SICS model as an operational tool.

Florida↗
USGS2023

Structured decision making remains underused in ecological restoration despite opportunities

Ecological restoration is considered an essential activity as we attempt to repair anthropogenic degradation. Yet, resources are limited and it is important that efforts focus on activities that are effective and yield successful restoration. Structured decision making (SDM) is an organized framework that is designed to incorporate differing values across stakeholders and evaluate alternatives. The SDM framework typically consists of six steps: define the decision problem, define objectives and evaluation criteria, develop alternatives, estimate consequences, evaluate trade-offs, and decide, implement, and monitor. Here, we posit that SDM is well suited for ecological restoration, yet remains underused. Specifically, tools such as stakeholder surveys, conceptual modeling, and multi-criteria decision analysis are notably useful in ecological restoration and can be applied under the SDM framework to ensure robust and transparent decision making. We illustrate the application of SDM to ecological restoration with case studies that used SDM alongside ecosystem service assessments, for species-as-risk management, and to assess action desirability across large and diverse stakeholder groups. Finally, we demonstrate how SDM is equipped to handle many of the challenges associated with ecological restoration by identifying commonalities. We contend that increased use of SDM for ecological restoration by environmental managers has the potential to yield wise use of limited resources and more effective restoration outcomes.

Environment Systems and Decisions↗
USGS2021

Integrating socioecological suitability with human-wildlife conflict risk: Case study for translocation of a large ungulate

Translocations are essential for re-establishing wildlife populations. As they sometimes fail, it is critical to assess factors that influence their success pre-translocation. Socioecological suitability models (SESMs) integrate social acceptance and ecological suitability to enable identification of areas where wildlife populations will expand, which makes it likely that SESMs will also be useful for predicting translocation success. To inform site selection for potential elk Cervus canadensis reintroduction to north-eastern Minnesota, United States, we developed broadscale maps of social acceptance from surveys of local residents and landowners, animal use equivalence (AUE) from forage measured in the field and empirical conflict risk from geospatial data. Resulting SESMs integrated social acceptance favourability scores, AUE and conflict risk, and weighted SESMs showed the relative influences of acceptance and conflict. Social acceptance was positive for local residents and landowners (mean ≥ 5.4; scale of 1–7). AUE (scaled to an elk home range) ranged between 1 and 9 elk/16 km 2 during winter, and from 14 to 83 elk/16 km 2 during summer. Human–elk conflict risk was low (mean ≤ 0.10; scaled 0–1), increasing from north to south. Geographical distributions differed for social acceptance, AUE and conflict risk, and weighted SESMs revealed unsuitable areas that were otherwise obscured. Synthesis and applications . Integrating human–wildlife conflict risk into SESMs shows where social acceptance of translocated species is likely to erode, even where viewed favourably pre-translocation, to inform translocation planning by highlighting interactions between key factors. Such integrated models supplement existing reintroduction biology frameworks by supporting decision-making and knowledge development. In north-eastern Minnesota, natural resource managers who are considering elk reintroductions are using SESMs reported here to identify where human–elk conflict is unlikely to result in an isolated elk population and where addressing concerns for area residents about conflict risk is essential.

Minnesota↗
USGS2010

Empirical models of wind conditions on Upper Klamath Lake, Oregon

Upper Klamath Lake is a large (230 square kilometers), shallow (mean depth 2.8 meters at full pool) lake in southern Oregon. Lake circulation patterns are driven largely by wind, and the resulting currents affect the water quality and ecology of the lake. To support hydrodynamic modeling of the lake and statistical investigations of the relation between wind and lake water-quality measurements, the U.S. Geological Survey has monitored wind conditions along the lakeshore and at floating raft sites in the middle of the lake since 2005. In order to make the existing wind archive more useful, this report summarizes the development of empirical wind models that serve two purposes: (1) to fill short (on the order of hours or days) wind data gaps at raft sites in the middle of the lake, and (2) to reconstruct, on a daily basis, over periods of months to years, historical wind conditions at U.S. Geological Survey sites prior to 2005. Empirical wind models based on Artificial Neural Network (ANN) and Multivariate-Adaptive Regressive Splines (MARS) algorithms were compared. ANNs were better suited to simulating the 10-minute wind data that are the dependent variables of the gap-filling models, but the simpler MARS algorithm may be adequate to accurately simulate the daily wind data that are the dependent variables of the historical wind models. To further test the accuracy of the gap-filling models, the resulting simulated winds were used to force the hydrodynamic model of the lake, and the resulting simulated currents were compared to measurements from an acoustic Doppler current profiler. The error statistics indicated that the simulation of currents was degraded as compared to when the model was forced with observed winds, but probably is adequate for short gaps in the data of a few days or less. Transport seems to be less affected by the use of the simulated winds in place of observed winds. The simulated tracer concentration was similar between model results when simulated winds were used to force the model, and when observed winds were used to force the model, and differences between the two results did not accumulate over time.

Scientific Investigations Report↗
USGS2011

Evaluation of a present-day climate simulation with a new coupled atmosphere-ocean model GENMOM

We present a new, non-flux corrected AOGCM, GENMOM, that combines the GENESIS version 3 atmospheric GCM (Global Environmental and Ecological Simulation of Interactive Systems) and MOM2 (Modular Ocean Model version 2) nominally at T31 resolution. We evaluate GENMOM by comparison with reanalysis products (e.g., NCEP2) and three models used in the IPCC AR4 assessment. GENMOM produces a global temperature bias of 0.6 ◦C. Atmospheric features such as the jet stream structure and major semi-permanent sea level pressure centers are well simulated as is the mean planetary-scale wind structure that is needed to produce the correct position of stormtracks. Most ocean surface currents are reproduced except where they are not resolvable at T31 resolution. Overall, GENMOM captures reasonably well the observed gradients and spatial distributions of annual surface temperature and precipitation and the simulations are on par with other AOGCMs. Deficiencies in the GENMOM simulations include a warm bias in the surface temperature over the southern oceans, a split in the ITCZ and weaker-than-observed overturning circulation.

Geoscientific Model Development↗
USGS2022

Integrated animal movement and spatial capture–recapture models: Simulation, implementation, and inference

Over the last decade, spatial capture–recapture (SCR) models have become widespread for estimating demographic parameters in ecological studies. However, the underlying assumptions about animal movement and space use are often not realistic. This is a missed opportunity because interesting ecological questions related to animal space use, habitat selection, and behavior cannot be addressed with most SCR models, despite the fact that the data collected in SCR studies — individual animals observed at specific locations and times — can provide a rich source of information about these processes and how they relate to demographic rates. We developed SCR models that integrated more complex movement processes that are typically inferred from telemetry data, including a simple random walk, correlated random walk (i.e., short-term directional persistence), and habitat-driven Langevin diffusion. We demonstrated how to formulate, simulate from, and fit these models with standard SCR data using data-augmented Bayesian analysis methods. We evaluated their performance through a simulation study, in which we varied the detection, movement, and resource selection parameters. We also examined different numbers of sampling occasions and assessed performance gains when including auxiliary location data collected from telemetered individuals. Across all scenarios, the integrated SCR movement models performed well in terms of abundance, detection, and movement parameter estimation. We found little difference in bias for the simple random walk model when reducing the number of sampling occasions from T = 25 to T = 15. We found some bias in movement parameter estimates under several of the correlated random walk scenarios, but incorporating auxiliary location data improved parameter estimates and significantly improved mixing during model fitting. The Langevin movement model was able to recover resource selection parameters from standard SCR data, which is particularly appealing because it explicitly links the individual-level movement process with habitat selection and population density. We focused on closed population models, but the movement models developed here can be extended to open SCR models. The movement process models could also be easily extended to accommodate additional “building blocks” of random walks, such as central tendency (e.g., territoriality) or multiple movement behavior states, thereby providing a flexible and coherent framework for linking animal movement behavior to population dynamics, density, and distribution.

Ecology↗
USGS2024

Disentangling drivers of annual grass invasion: Abiotic susceptibility vs. fire-induced conversion to cheatgrass dominance in the sagebrush biome

Invasive annual grasses are often facilitated by fire, yet they can become ecologically dominant in susceptible locations even in the absence of fire. We used an extensive vegetation plot database to model susceptibility to the invasive annual grass cheatgrass ( Bromus tectorum L.) in the sagebrush biome as a function of climate and soil water availability variables. We built random forest models predicting cheatgrass presence or dominance (greater than 15 % relative cover) under unburned (37,219 plots) and burned conditions (6340 plots). We mapped predicted probability of cheatgrass presence and dominance, conditional on burning. We combined predicted susceptibility with burn probability to quantify the 10-year total risk of cheatgrass dominance. Finally, we identified portions of the landscape (1) at risk of fire-induced conversion to cheatgrass dominance, (2) consistently susceptible to cheatgrass dominance, or (3) consistently resistant to cheatgrass dominance. At the scale of the sagebrush biome, we found that abiotic susceptibility to cheatgrass dominance drives total risk, regardless of fire. At local scales (i.e., individual 30 m pixels), burning increased the probability of cheatgrass dominance by a median of 14 %. Threshold-based analyses indicate that 10–31 % of the sagebrush biome was at risk of fire-induced dominance, with 55 % exhibiting abiotic resistance and 5 % exhibiting abiotic susceptibility to dominance regardless of fire. Burn probability was higher in areas predicted to be susceptible to dominance, illustrating how cheatgrass invasion can cause ecosystem conversions that are then sustained by grass-fire cycles. Disentangling the influence of abiotic conditions and fire contributes to our understanding of the mechanisms driving invasion dynamics, and modeling the probability of dominance can help anticipate where ecological transformations are at risk of occurring. Our approach can facilitate the prioritization of management actions in the sagebrush biome and be used as a framework for modeling invasion risk in other disturbance-prone ecosystems.

Biological Conservation↗
USGS2023

Scenario planning and multispecies occupancy models reveal positive avian responses to restoration of afforested woodlands

Scenario planning is a powerful approach for assessing restoration outcomes under alternative futures. However, developing plausible scenarios remains daunting in complex systems like ecological communities. Here, we used Bayesian multispecies occupancy modeling to develop scenarios to assess woodland restoration outcomes in afforested communities in seven wildlife management areas in Arkansas, U.S.A. Our objectives were (1) to define plausible woodland restoration and afforestation scenarios by quantifying historic ranges of variation in mean tree cover and tree cover heterogeneity from 1986 to 2021 and (2) to predict changes in bird species richness and occupancy patterns for six species of greatest conservation need under two future scenarios: complete afforestation (100% tree cover) and woodland restoration (based on remotely sensed historic tree cover). Using 35 years of remotely sensed tree cover data and 6 years of bird monitoring data, we developed multispecies occupancy models to predict future bird species richness and occupancy under the complete afforestation and woodland restoration scenarios. Between 1986 and 2021, tree cover increased in all study areas—with one increasing 70%. Under the woodland restoration scenario, avian species richness increased up to 20%, and four of six species of greatest conservation need exhibited gains in occupancy probability. The complete afforestation scenario had negligible effects on richness and occupancy. Overall, we found decreasing tree cover to historic levels prior to widespread afforestation would provide community-level benefits and would do little harm even to forest-dependent species of conservation concern. Applying multispecies occupancy modeling within a scenario planning framework allows for comparing multiscale trade-offs between plausible futures.

Restoration Ecology↗
USGS2026

Combining branching processes and Allee effects into an integral projection model to assess invasion risk

Invasive species threaten both ecological systems and economies. Successful establishment is governed partly by demographic stochasticity, where random individual-level variation can significantly influence invasion dynamics. A critical factor is the Allee effect, which describes reduced population growth at low densities, potentially serving as a barrier to persistence. Size-dependent growth and survival further shape outcomes based on the size distribution of individuals within the population. To address these complexities, we developed a population-level model integrating an Integral Projection Model (IPM) for the expected size distribution with an Allee effect for recruitment dynamics and a branching process to account for demographic stochasticity. We applied this framework to Silver Carp ( Hypophthalmichthys molitrix ), a species established in the Upper Mississippi River system that poses significant risk of expanding into the Great Lakes. Our analysis compared temporal dynamics and population-size distributions under hypothetical introduction scenarios for large adult and smaller subadult fish. In both cases, initial population size relative to the Allee threshold was a critical determinant of establishment probability, with introductions below the threshold requiring stochastic growth across a critical biomass boundary before self-sustaining dynamics could take hold. Sensitivity analysis further indicated that per-spawner recruitment played a compounding role in invasion risk: higher recruitment rates reduced the critical introduction size for establishment and compressed the time to self-sustaining population size, jointly broadening the window of invasion risk. These findings highlight the framework’s potential for invasion assessments and offer insights into how Allee effects, size structure, and recruitment rates interact to govern establishment outcomes, while underscoring the need for species-specific parameter estimates before quantitative outputs can guide management decisions.

PLOS Ecosystems↗
USGS2017

Minimizing effects of methodological decisions on interpretation and prediction in species distribution studies: An example with background selection

Evaluating the conditions where a species can persist is an important question in ecology both to understand tolerances of organisms and to predict distributions across landscapes. Presence data combined with background or pseudo-absence locations are commonly used with species distribution modeling to develop these relationships. However, there is not a standard method to generate background or pseudo-absence locations, and method choice affects model outcomes. We evaluated combinations of both model algorithms (simple and complex generalized linear models, multivariate adaptive regression splines, Maxent, boosted regression trees, and random forest) and background methods (random, minimum convex polygon, and continuous and binary kernel density estimator (KDE)) to assess the sensitivity of model outcomes to choices made. We evaluated six questions related to model results, including five beyond the common comparison of model accuracy assessment metrics (biological interpretability of response curves, cross-validation robustness, independent data accuracy and robustness, and prediction consistency). For our case study with cheatgrass in the western US, random forest was least sensitive to background choice and the binary KDE method was least sensitive to model algorithm choice. While this outcome may not hold for other locations or species, the methods we used can be implemented to help determine appropriate methodologies for particular research questions.

Ecological Modelling↗
USGS2025

What is a specialist? Quantifying host breadth enables impact prediction for invasive herbivores

Herbivores are commonly classified as host specialists or generalists for various purposes, yet the definitions of these terms, and their intermediates, are often imprecise and ambiguous. We quantified host breadth for 240 non-native, tree-feeding insects in North America using phylogenetic diversity. We demonstrated that a partitioning of host breadth: (1) causes 67% of non-native insects to shift from a generalist to specialist category, (2) displays a reduction in host breadth from the native to introduced range, (3) identifies an inflection point in a model predicting the likelihood of non-native insect ecological impact, with a corresponding change in behaviour associated with specialists versus generalists, and (4) enables three models for strong prediction of whether a non-native forest insect will cause high impacts. Together, these results highlight the primacy of how herbivore host recognition and plant defences mediate whether novel host interactions will result in high impact after invasion.

Ecology Letters↗
USGS2017

A guide to multi-objective optimization for ecological problems with an application to cackling goose management

Choices in ecological research and management are the result of balancing multiple, often competing, objectives. Multi-objective optimization (MOO) is a formal decision-theoretic framework for solving multiple objective problems. MOO is used extensively in other fields including engineering, economics, and operations research. However, its application for solving ecological problems has been sparse, perhaps due to a lack of widespread understanding. Thus, our objective was to provide an accessible primer on MOO, including a review of methods common in other fields, a review of their application in ecology, and a demonstration to an applied resource management problem. A large class of methods for solving MOO problems can be separated into two strategies: modelling preferences pre-optimization (the a priori strategy), or modelling preferences post-optimization (the a posteriori strategy). The a priori strategy requires describing preferences among objectives without knowledge of how preferences affect the resulting decision. In the a posteriori strategy, the decision maker simultaneously considers a set of solutions (the Pareto optimal set) and makes a choice based on the trade-offs observed in the set. We describe several methods for modelling preferences pre-optimization, including: the bounded objective function method, the lexicographic method, and the weighted-sum method. We discuss modelling preferences post-optimization through examination of the Pareto optimal set. We applied each MOO strategy to the natural resource management problem of selecting a population target for cackling goose ( Branta hutchinsii minima ) abundance. Cackling geese provide food security to Native Alaskan subsistence hunters in the goose's nesting area, but depredate crops on private agricultural fields in wintering areas. We developed objective functions to represent the competing objectives related to the cackling goose population target and identified an optimal solution first using the a priori strategy, and then by examining trade-offs in the Pareto set using the a posteriori strategy. We used four approaches for selecting a final solution within the a posteriori strategy; the most common optimal solution, the most robust optimal solution, and two solutions based on maximizing a restricted portion of the Pareto set. We discuss MOO with respect to natural resource management, but MOO is sufficiently general to cover any ecological problem that contains multiple competing objectives that can be quantified using objective functions.

Ecological Modelling↗
USGS2021

Advancing estuarine ecological forecasts: Seasonal hypoxia in Chesapeake Bay

Ecological forecasts are quantitative tools that can guide ecosystem management. The coemergence of extensive environmental monitoring and quantitative frameworks allows for widespread development and continued improvement of ecological forecasting systems. We use a relatively simple estuarine hypoxia model to demonstrate advances in addressing some of the most critical challenges and opportunities of contemporary ecological forecasting, including predictive accuracy, uncertainty characterization, and management relevance. We explore the impacts of different combinations of forecast metrics, drivers, and driver time windows on predictive performance. We also incorporate multiple sets of state-variable observations from different sources and separately quantify model prediction error and measurement uncertainty through a flexible Bayesian hierarchical framework. Results illustrate the benefits of (1) adopting forecast metrics and drivers that strike an optimal balance between predictability and relevance to management, (2) incorporating multiple data sources in the calibration data set to separate and propagate different sources of uncertainty, and (3) using the model in scenario mode to probabilistically evaluate the effects of alternative management decisions on future ecosystem state. In the Chesapeake Bay, the subject of this case study, we find that average summer or total annual hypoxia metrics are more predictable than monthly metrics and that measurement error represents an important source of uncertainty. Application of the model in scenario mode suggests that absent watershed management actions over the past decades, long-term average hypoxia would have increased by 7% compared to 1985. Conversely, the model projects that if management goals currently in place to restore the Bay are met, long-term average hypoxia would eventually decrease by 32% with respect to the mid-1980s.

Chesapeake Bay↗