Lizhu Wang

Quantitative Identification of Disturbance Thresholds in Support of Aquatic Resource Management

The identification of disturbance thresholds is important for many aspects of aquatic resource management, including the establishment of regulatory criteria and the identification of stream reference conditions. A number of quantitative or model-based approaches can be used to identify disturbance thresholds, including nonparametric deviance reduction (NDR), piecewise regression (PR), Bayesian changepoint (BCP), quantile piecewise constant (QPC), and quantile piecewise linear (QPL) approaches. These methods differ in their assumptions regarding the nature of the disturbance-response variable relationship, which can make selecting among the approaches difficult for those unfamiliar with the methods. We first provide an overview of each of the aforementioned approaches for identifying disturbance thresholds, including the types of data for which the approaches are intended. We then compare threshold estimates from each of these approaches to evaluate their robustness using both simulated and empirical datasets. We found that most of the approaches were accurate in estimating thresholds for datasets with drastic changes in responses variable at the disturbance threshold. Conversely, only the PR and QPL approaches performed well for datasets with conditional mean or upper boundary changes in response variables at the disturbance threshold. The most robust threshold identification approach appeared to be the QPL approach; this method provided relatively accurate threshold estimates for most of the evaluated datasets. Because accuracy of disturbance threshold estimates can be affected by a number of factors, we recommend that several steps be followed when attempting to identify disturbance thresholds. These steps include plotting and visually inspecting the disturbance-response data, hypothesizing what mechanisms likely generate the observed pattern in the disturbance-response data, and plotting the estimated threshold in relation to the disturbance-response data to ensure the appropriateness of the threshold estimate.

Defining and Characterizing Coolwater Streams and Their Fish Assemblages in Michigan and Wisconsin, USA

Abstract Coolwater streams, which are intermediate in character between coldwater “trout” streams and more diverse warmwater streams, occur widely in temperate regions but are poorly understood. We used modeled water temperature data and fish assemblage samples from 371 stream sites in Michigan and Wisconsin to define, describe, and map coolwater streams and their fish assemblages. We defined coolwater streams as ones having summer water temperatures suitable for both coldwater and warmwater species and used the observed distributions of the 99 fish species at our sites to identify coolwater thermal boundaries. Coolwater streams had June-through-August mean water temperatures of 17.0–20.5°C, July mean temperatures of 17.5–21.0°C, and maximum daily mean temperatures of 20.7–24.6°C. We delineated two subclasses of coolwater streams: “cold transition” (having July mean water temperatures of 17.5–19.5°C) and “warm transition” (having July mean temperatures of 19.5–21.0°C). Fish assemblages in coolwater stream...

A Hierarchical Spatial Framework and Database for the National River Fish Habitat Condition Assessment

Abstract Fisheries management programs, such as the National Fish Habitat Action Plan (NFHAP), urgently need a nationwide spatial framework and database for health assessment and policy development to protect and improve riverine systems. To meet this need, we developed a spatial framework and database using National Hydrography Dataset Plus (I-.100,000-scale); http://www.horizon-systems.com/nhdplus). This framework uses interconfluence river reaches and their local and network catchments as fundamental spatial river units and a series of ecological and political spatial descriptors as hierarchy structures to allow users to extract or analyze information at spatial scales that they define. This database consists of variables describing channel characteristics, network position/connectivity, climate, elevation, gradient, and size. It contains a series of catchment-natural and human-induced factors that are known to influence river characteristics. Our framework and database assembles au river reaches and t...

A River Valley Segment Classification of Michigan Streams Based on Fish and Physical Attributes

Abstract Water resource managers are frequently interested in river and stream classification systems to generalize stream conditions and establish management policies over large spatial scales. We used fish assemblage data from 745 river valley segments to develop a two-level, river valley segment-scale classification system of rivers and streams throughout Michigan. Regression tree analyses distinguished 10 segment types based on mean July temperature and network catchment area and 26 segment types when channel gradient was also considered. Nonmetric multidimensional scaling analyses suggested that fish assemblages differed among segment types but were only slightly influenced by channel gradient. Species that were indicative of specific segment types generally had habitat requirements that matched segment attributes. A test of classification strength using fish assemblage data from an additional 77 river valley segments indicated that the classification system performed significantly better than random...

A spatially constrained clustering program for river valley segment delineation from GIS digital river networks

River valley segments are adjacent sections of streams and rivers that are relatively homogeneous in hydrology, limnology, channel morphology, riparian dynamics, and biological communities. River valley segments have been advocated as appropriate spatial units for assessing, monitoring, and managing rivers and streams for several reasons; however, methods for delineating these spatial units have been tedious to implement or have lacked objectivity, which arguably has limited their use as river and stream management units by natural resource agencies. We describe a spatially constrained clustering program that we developed for delineating river valley segments from geographic information system digital river network databases that is flexible, easy-to-use, and improves objectivity in the river valley segment delineation process. This program, which we refer to as the valley segment affinity search technique (VAST), includes a variety of options for determining spatial adjacency in stream reaches, as well as several data transformation methods, types of resemblance coefficients, and cluster linkage methods. The usefulness of VAST is demonstrated by using it to delineate river valley segments for river network databases for Michigan and Wisconsin, USA, and by comparing river valley segments delineated by VAST to an expert-opinion delineation previously completed for a Michigan river network database.

Landscape-Based Assessment of Human Disturbance for Michigan Lakes

Assessment of lake impairment status and identification of threats’ type and source is essential for protection of intact, enhancement of modified, and restoration of impaired lakes. For regions in which large numbers of lakes occur, such assessment has usually been done for only small fractions of lakes due to resource and time limitation. This study describes a process for assessing lake impairment status and identifying which human disturbances have the greatest impact on each lake for all lakes that are 2xa0ha or larger in the state of Michigan using readily available, georeferenced natural and human disturbance databases. In-lake indicators of impairment are available for only a small subset of lakes in Michigan. Using statistical relationships between the in-lake indicators and landscape natural and human-induced measures from the subset lakes, we assessed the likely human impairment condition of lakes for which in-lake indicator data were unavailable using landscape natural and human disturbance measures. Approximately 92% of lakes in Michigan were identified as being least to marginally impacted and about 8% were moderately to heavily impacted by landscape human disturbances. Among lakes that were heavily impacted, more inline lakes (92%) were impacted by human disturbances than disconnected (6%) or headwater lakes (2%). More small lakes were impacted than medium to large lakes. For inline lakes, 90% of the heavily impacted lakes were less than 40xa0ha, 10% were between 40 and 405xa0ha, and 1% was greater than 405xa0ha. For disconnected and headwater lakes, all of the heavily impacted lakes were less than 40xa0ha. Among the anthropogenic disturbances that contributed the most to lake disturbance index scores, nutrient yields and farm animal density affected the highest number of lakes, agricultural land use affected a moderate number of lakes, and point-source pollution and road measures affected least number of lakes. Our process for assessing lake condition represents a significant advantage over other routinely used methods. It permits the evaluation of lake condition across large regions and yields an overall disturbance index that is a physicochemical and biological indicator weighted sum of multiple disturbance factors. The robustness of our approach can be improved with increased availability of high-resolution disturbance datasets.

A Landscape-Based Classification of Fish Assemblages in Sampled and Unsampled Lakes

Abstract We related fish species patterns and landscape-scale environmental data from 216 Michigan lakes to identify repeatable types of fish assemblages, identify environmental factors related to assemblage types, and classify fish assemblages in unsampled lakes. Multivariate regression tree modeling of fish species abundances identified six assemblage types that were explained by degree-days during the ice-free period, lake surface area, and mean lake surface temperature. Warmwater species dominated southern lakes, while coolwater and coldwater species had higher abundances in northern lakes. Coolwater species were present in large southern lakes, whereas warmwater species were excluded from northern lakes that had low mean surface temperatures or low degree-days. These results suggest that patterns of lake fish assemblages are shaped by differences in climate as well as lake-specific differences in surface temperature regimes and in vertical availability of coldwater and coolwater habitats. Because we ...

Comparison between model-predicted and field-measured stream habitat features for evaluating fish assemblage-habitat relationships

The use of model-predicted, local-scale habitat data as inputs in analyses intended to evaluate multiscale fish assemblage-habitat relationships in streams has become increasingly common as the scale at which such studies are conducted has increased. We used fish assemblage and habitat data from 208 wadeable streams in Wisconsin and Michigan to determine whether model-predicted habitat data would yield results similar to those of field-measured data in multiscale analyses of fish assemblage-habitat relationships. Predictions of local habitat features from landscape variables were generated by means of generalized additive modeling with likelihood-based boosting. Relationships between fish assemblage measures and landscape and local habitat features were studied via partial constrained multivariate ordination analyses. The total variation explained in the fish assemblage data sets was similar for model-predicted local habitat features and field-measured data, as was the proportion of variation explained that was due independently to local and regional (i.e., landscape) habitat features. We observed dissimilar results in the magnitude of ordination scores for local habitat features and the directional relationships between local habitat ordination scores and individual species and assemblage metric scores. Our findings indicate that model-predicted, local-scale habitat data can be useful for evaluating the relative strengths of local and regional habitat features in structuring fish assemblages, but caution may be necessary when evaluating species-habitat or assemblage metric-habitat relationships.