Home  Newsletter Index    WMC   < Previous  TOC  Next >

Spring 1998

Aquatic/Riparian Effectiveness Monitoring for the Northwest Forest Plan:

A New Approach Based on Integrative Assessment of Watershed Condition

Gordon Reeves, David Hohler, Phil Larsen, Karl Stein and Michael Furniss


Editor's Note: This outline describes the strategy for a large-scale monitoring plan that is currently being formulated to test the effectiveness of the Aquatic Conservation Strategy for the Northwest Forest Plan, covering 24 million acres of public lands in Washington, Oregon, and Northern California. Additional team members have been added to this effort, including WMC's Polly Hays, and Mike Tehan. A final proposal is expected to be completed by July.

BACKGROUND AND DESCRIPTION

Goals

 

Primary Hypotheses

If implementation of the Forest Plan and therefore the ACS is effective, then:

 

Summary of Approach

The core of the proposed monitoring plan consists of the regular evaluation of an array of watersheds selected to represent the distribution of watersheds across the area of the Northwest Forest Plan (NWFP). Each of the watersheds selected for evaluation will be classified into one of three condition classes: properly functioning, compromised, or degraded. Because the set of watersheds to be evaluated will be selected to represent the full array of watersheds across the NWFP region, a frequency distribution of condition classes among the evaluated watersheds represents the frequency distribution of condition classes of watersheds across the NWFP region. This frequency distribution of condition classes will constitute the primary description of status (at regional or provincial scales). Regular evaluation of watersheds (e.g., annually) forms the basis for determining whether improvement, degradation, or no change is occurring across the region or province. The primary types of questions to be answered at this scale are questions like: what fraction of the watersheds across the NWFP region are degraded (regional status)? How is that fraction changing over time (regional trend)?

The evaluations of watershed condition will be based on a comparison of a set of indicators organized by indicator categories, measured or evaluated in a consistent manner on each selected watershed, using a expert panels organized by watershed and basin. Each indicator will be evaluated against what is believed to be the natural potential of the watershed, and the aggregate of the departures of the array of indicators from their natural conditions will be used to assign watershed condition class. Indicators will be of several types: Upslope indicators (such as road density, land use/land cover, harvest history generally attributes that can be mapped for the entire watershed); riparian indicators (e.g., proportion of channel length with mature coniferous riparian forest); channel indicators (physical properties of the channel such as extent and quality of pools, amount of large woody debris); water quality indicators (temperature, chemical constituents); and biological indicators (such as salmonid populations, fish, amphibian, or macroinvertebrate assemblage structure or biotic integrity).

 

 

In addition to condition indicators outlined above, for each selected watershed the array of management activities will be evaluated for their local effectiveness: are the management actions having the desired effects? Summarized across the selected watersheds, the distribution of management actions and their respective effectiveness will indicate the types of management actions most commonly implemented, and the relative effectiveness of the different management actions. Inclusion of indicators of management effectiveness will also provide immediate feedback to local and regional managers about whether the overall strategy encompassed by the ACS objectives and the standards and guidelines are effective, and would allow for ongoing adjustments without having to wait decades for detecting responses of more slowly changing indicators.

We expect that for each selected watershed, each of the indicators will be either completely described (censussed; for example, a map of roads within the watersheds is a complete description; a map of riparian vegetation along all fish-bearing channels within the watershed is a complete description) or described through a representative sample (for example, it is not feasible to sample all macroinvertebrates in stream channels within the selected watersheds, therefore a representative sample will be required to infer watershed condition based on this indicator). Although a core set of indicators is proposed for measurement across all watersheds, circumstances may dictate that certain indicators be dropped for particular watersheds because they are irrelevant, or certain indicators be added to supplement the core set because they are particularly relevant. The use of a core set of indicators, selected in the manner described (i.e., watersheds selected to represent watersheds in the region; indicators selected within watersheds to represent watersheds) also allows tracking of key indicators across the region (or subregions such as provinces) to evaluate indicator- specific trends, regardless of the classification of watershed condition. The following types of indicator-specific questions can be answered: Is the density of high quality pools increasing across the region or across selected provinces? Is the amount of large woody debris in unconstrained channels increasing? Is the amount of mature riparian forest increasing?

A key feature of the proposed monitoring plan is that a central authority would select a statistically representative set of watersheds that would be evaluated on a defined schedule (e.g., annually, or every five years). This feature is important because it is estimated that several thousand watersheds comprise the NWFP region and it will not be feasible to evaluate every watershed. Consequently, a representative sample will be drawn, monitored, and evaluated. Regional (or provincial) inferences will be drawn from the results of the representative sample. Data gathering (mapping and monitoring of core and watershed specific indicators) would occur on each of the selected watersheds during some interval prior to the year of evaluation, such that a consistent database is available across the set of watersheds at the time of evaluation. Simultaneously, standards would be developed and refined over time. Initially, it is expected that standards might be relatively crude, and judgements about watershed condition would be made by expert evaluation of the indicator scores. Over time, it is expected that the evaluation process will improve and that more objective standards will be developed. If consistent protocols are used, it should be possible to retrospectively reevaluate the original condition decisions so that trend evaluation is based on actual changes in indicator scores, and not based on an individual's idiosyncratic interpretation. In addition, tracking of key indicators, separate from watershed classification, can serve as an independent measure of trend: are we getting more/less high quality pools? More/less large wood? More/less fish?

If such a plan is implemented and runs smoothly, our vision is that regular reports (e.g., annual or every five years) could be created that describe current regional condition, and over time the accumulation of the annual descriptions of regional condition would yield a concise summary of trend, whether for individual key indicators, or for watersheds as a whole. By incorporating an evaluation of the effectiveness of management actions, immediate feedback could be available at both local (watershed) and regional scales. By increasing sample sizes (the number of watersheds evaluated each year), the evaluation could be refined to provincial or basin scales.

 

Evaluation Process Principles

The regular, integrated evaluation of watershed condition will be conducted by a hierarchical series of teams at the region, province, and watershed scales and overseen by an independent Science Team. Watershed assessments will be aggregated and analyzed to arrive at a provincial assessment, and provincial assessments will be aggregated into a regional assessment.

The Integrative Evaluation of Watershed Condition approach is:

  1. Modeled on the previous exercises of ecosystem assessment such as Sustaining Ecological Systems (SES), Regional Ecosystem Assessment Project (REAP), the Gang of Four effort prior to FEMAT, etc.,

  2. Coordinated with similar efforts being designed in California (CalOwl) and the Regions 1 and 4 of the Forest Service (add BLM and EPA areas), and the State of Oregon's Coastal Salmon Restoration Initiative (CSRI), and

  3. Grounded in latest discussions of monitoring strategies for aquatic riparian ecosystems from the western U.S. (Naiman et al. 1992, Conquest et al. 1994, Spence et al. 1996).

  4. Integrative of several indicator categories, and their respective indicators, so that the final assessment at each level looks at the whole of the unit. It is based on the ability of interagency, interdisciplinary teams to integrate the understanding of individual indicators into the understanding of the health of larger scale aquatic and riparian ecosystems. We believe that this tends to smooth out differences between assessors and leads to a consensus position on what condition class (functioning, compromised, or degraded) to assign the unit. It is intended to use the best available information, augmented by the professional judgement of the team.

Watershed Evaluation Iterations. The assessment process is intended to be iterative. It is expected that federal agencies will cycle through analysis of the same watersheds over time at some predetermined time interval (annually; every five years). The first iteration will primarily be a qualitative effort based on the compilation of existing information and expert panels evaluation of that information. As the coherent monitoring plan is implemented, subsequent analyses will be based on more rigorous, quantitative information. There should be a general increase in rigor or statistical validity over the course of monitoring cycles, and the information and understanding developed in previous years will result in refinement of subsequent monitoring efforts.

Interagency, Interdisciplinary Integration. The process is based on the ability of teams to integrate the understanding of individual indicators into the understanding of the health of aquatic and riparian ecosystems. It is expected that this process will be conducted by interagency, interdisciplinary teams experienced in assessing the indicators at the locale/scale to be assessed, interpretations and explanation of conditions, causes, and trends. Interagency discussions can lead to a much clearer understanding of the variety of processes and linkages between indicators.

Aggregation of Assessments: Watersheds, Province, and Region. The basic unit for monitoring and assessment is a watershed of 10 - 200 mi2 (sixth field USGS hydrologic accounting unit). The watershed is chosen as the unit of evaluation for several reasons: 1) FEMAT's focus is on integrated condition (watershed/aquatic ecosystems); 2) the set of ACS objectives address all components of aquatic ecosystems; 3) the condition of an aquatic ecosystem is the integrated product of the individual ACS objectives. This size is chosen because the selection: minimizes the variety of processes and management actions relative to larger sized watersheds; is a manageable size for data collection and evaluation; is a commonly delineated unit used by many agencies; is a biological unit in which fish, particularly anadromous salmonids, can complete much freshwater life history. All lands within the selected watershed should be analyzed, regardless of ownership. It is important to understand the watershed as an ecologic system, and to do so requires that all of the resources, allocations, activities, influences and impacts to the aquatic and riparian resources be accounted for. Depending upon the choice of aggregation scale, the information from the individual watersheds that make up a province can be aggregated into a provincial assessment by a Province Team. After the province team has completed and presented its results to the Regional team, all of the province assessment can be aggregated into a regional assessment of aquatic and riparian resources. Clear communication and documentation between assessment teams at each of the different levels will be required for this to be a smooth, efficient process.

Documentation of Rationale. Watershed assessments will be aggregated and analyzed to arrive at a provincial assessment, and provincial assessments will be aggregated into a regional assessment. Documentation of the rationale for conclusions is essential for demonstrating credibility of the watershed condition evaluation. Evaluation teams will be required to clearly and completely document their conclusions as they progress through the process. The power of the process is the ability to produce a useful product at each of the analytic scales and provide information useful for understanding this iteration of the evaluation. Issues which cross watershed boundaries need to be flagged for resolution within provincial and regional level assessments.

Quality Control. Quality control will be provided by 1) a peer review system of each watershed condition determination, 2) an assessment by the team of the quality of information used in the determination, 3) verification by sample survey exercises, and 4) oversight by an independent Science Team. In the presentation of results to the province team, watershed team members will be asked about the data that they used to make their determinations, the limitations of the data, and the ongoing efforts to improve data acquisition. Similar processes will occur at the regional level when the provincial teams present their results to the regional team..

Review. The regional assessment and their results of the lower level assessments will be provided to outside reviewers for comment and review. The objectives of the review process are to:

 

Development Standards and Quantitative Predictive Models

One of the challenges we face is the development of the protocols by which a watershed's condition is classified into functioning, compromised, or degraded. Conceptually, this evaluation will be based on an estimate of watershed's natural condition and range of natural variation. A watershed's condition depends upon the normal functioning of a variety of key processes such as the delivery of water, sediment (fine as well as coarse), energy, wood, the formation and continual function of riparian zones, the development and maintenance of in channel habitat to support natural and desired biotic resources. One choice is to identify watersheds that are functioning naturally whose indicator scores would serve as a standard against which to evaluate the selected watersheds. It is likely that the prevalence of such watersheds will be low because of the extensive history of landscape modification by silviculture practices. Nonetheless, one aspect of the monitoring plan will be to identify and monitor such watersheds.

The proposed monitoring plan relies on the regular and extensive monitoring of a variety of indicators. This can be relatively expensive. A more efficient process might be the development and use of predictive models. These predictive models could rely on measures of key upslope indicators that have been shown to be empirically (or theoretically) related to key riparian or channel habitat or biotic attributes. It may be relatively inexpensive to measure these key upslope indicators (e.g. through remote sensing), making the overall process of monitoring much more efficient than reliance on extensive ground level monitoring.

We propose the development of predictive models and their incorporation into the routine monitoring process. Although this is primarily a research task, we expect it to be closely linked with the ongoing monitoring program to improve the objectivity with which watersheds will be evaluated. Initially the models will use upslope factors such as vegetation conditions, amount and character of roads, riparian vegetation conditions, and relate these characteristics to in-channel conditions, such as amount of wood, number of pools, water temperature, etc. and to biotic conditions such as fish populations. The data for model development will be obtained from existing stream inventories and other available sources, particularly the data on vegetative conditions generated by the Late Successional/Old Growth Monitoring Group. We expect to use multivariate techniques and decision tree analysis to build the predictive models. The latter approach has been used to predict the status of fish populations from physical features and conditions of watersheds considered in the Upper Columbia River Basin Assessment (Lee et al., in press). The models developed in this effort found strong relations between the status of the fish populations and physical features and conditions of the watersheds. Botkin et al. (1995) developed relations between condition of fish populations and the vegetative conditions of watersheds in the Oregon coast. A prototype model of relations between in-channel and upslope features and the ecological conditions of watersheds in the Oregon coast is being developed by the Aquatic Land Interaction Team in Corvallis. Preliminary results are promising. Because of the wide diversity of ecological conditions in the area of the NWFP, it is likely that a model will have to be developed for each physiographic province.

At least initially, we will need to ground truth the model predictions. We will predict the condition of the aquatic ecosystem in a watershed and then verify the predictions by direct observation and examination of available supporting data through the monitoring program. We will also consult with people who are most familiar with the watersheds to determine whether the model predictions match their assessments. Over time, we hope to shift in emphasis from reliance on assessments done strictly on the basis of monitoring to increased reliance on the predictions, relying on less intensive monitoring to provide data with which to verify the continued validity of the models. The degree to which this is done over the long term will depend on the accuracy of the model predictions.


 

Top