Blueprint for Monitoring Plans
Monitoring is an essential tool for successful management of hardwood rangelands. Over time, it allows landowners and managers to measure the impacts of their management activities on the condition of natural resources under their stewardship and then use that information to better manage those resources. As such, monitoring is key to managing livestock and maintaining healthy oak populations.
In order for monitoring to provide useful information to the manager, a monitoring plan should be thoughtfully crafted and complete. A complete plan contains at least seven elements:
- sampling design
- measurement methods
- personnel assignments
- data-management protocols
- methods of data evaluation
An effective monitoring plan takes time to design—often 6 to 18 months, depending on the scope of the plan’s objectives—and external review can be particularly beneficial. Landowners and managers can obtain help designing a monitoring plan from their County UC Cooperative Extension office, local Resource Conservation Districts, the Natural Resource Conservation Service, the Bureau of Land Management, and the US Forest Service. Useful textbooks on sampling are also available through California’s college and university libraries.
An effective monitoring program begins with explicit goals, which are broad statements of the monitoring project’s purpose that indicate clearly how data to be collected will facilitate informed decision-making by the landowner or manager. Ideally, there should be consensus among all participating partners on the specific language of each goal.
From the stated goals, specific questions to be addressed by the monitoring project can be derived. For clarity, each question should focus on one, and only one, parameter to be monitored. A parameter is a measure that describes some characteristic or indicator of interest for purposes of monitoring (e.g. average cover of bare soil). Monitoring is designed to estimate changes in the values of parameters over time. For example, one might examine whether a decrease in the amount of bare soil results in a corresponding decrease in soil erosion. Questions stated in the monitoring plan should be as specific as possible and should include the spatial location and time period of each parameter to be monitored (e.g. What is the total acreage of yellow starthistle in East Pasture in late spring of 2002?).
Because it is usually too costly or otherwise undesirable to take exhaustive measurements, monitoring is often based on sampling. With sampling, measurements are made on a subset rather than a complete census of all units that could possibly be measured. For instance, one would measure height on a sampling of oak trees rather than on each and every oak tree in a large watershed. If a sample is to provide an accurate characterization of the parameter of interest, the sample must be a representative subset of all units that could possibly be measured. This is why statisticians typically recommend selection of a random sample.
In a sample, each unit that contributes to the subset is termed a “sampling unit.” Depending upon what parameter is being characterized, a sampling unit may be an individual oak tree, the position of a vegetation quadrat, a cow-calf pair, etc. For example, if one wanted to estimate average girth of valley oak saplings in a watershed, a natural choice of sampling unit would be the individual valley oak sapling.
The monitoring plan’s sampling design should provide the following: 1) description of the basic sampling unit (e.g. size and shape of a quadrat); 2) number of sampling units to be included in the sample (i.e., sample size); 3) location in space where sampling will occur (e.g. every 10m along a 100m transect); and 4) time schedule for measurements on each sampling unit. Decisions on these four design features are shaped by the parameter to be estimated and, whenever possible, should be refined through preliminary sampling.
The measurement design describes how measurements are to be made for each sampling unit. In a project designed to monitor maximum water depth in vernal pools through the wet season on a ranch, the sample may consist of a random sampling of vernal pools. While the sampling design describes which pools to include in the sample and when they are to be measured, the measurement design describes how to measure maximum water depth within each selected vernal pool (e.g., with a meter stick). Failure to distinguish the sampling design from the measurement design is commonplace and often renders each muddled and incomplete.
Assignment of personnel is an often overlooked but crucial element in most monitoring plans. Though some small monitoring projects may be managed and implemented by a single person, most monitoring projects are group efforts. For a group effort to be effective, each participant should have clearly defined roles and responsibilities that are specified in the monitoring plan.
Data-management protocols encompass 1) data recording, 2) data-quality control, 3) data storage, and 4) data flows. Clear rules should be established for recording data. These rules can cover issues such as conventions for recording dates, missing values, questionable measurements, miscellaneous observational notes, etc. Data-quality controls are procedures implemented throughout the process of data handling for the purpose of minimizing data loss and error. The monitoring plan should detail quality-control procedures for data collection, data entry, data analysis, and data storage. Quality control is particularly important during data collection because errors made during this stage may not be detectable or correctable later in the process. Finally, a data-flow design shows how data are moved, and by which personnel, from data collection, to data entry and processing, and finally to storage.
The last essential element in monitoring plans is a description of methods of data evaluation. Because most monitoring is based on sampling, most analyses focus on parameter estimation. The plan’s questions dictate which parameters are to be estimated. Each estimated parameter should be reported with an estimate of its error. A measure of error provides information about the reliability of the estimates made. Reporting a parameter’s estimate without reporting a measure of its error can give the false impression that the estimate is without error. Standard errors are often reported but confidence bounds are more informative because they account for the sampling distribution of the parameter’s estimator. The organizations mentioned above may be contacted for more details on monitoring natural resources.
prepared and edited by Adina Merenlender and Emily Heaton