5.2: Site descriptors and Covariates

Video Presentation

Learning Guide

In this module we’re going to discuss things that you measure and record at a site that are not, in themselves, direct indicators for your study or monitoring. We can collectively call these site descriptors and covariates.

Site Descriptors

Site descriptors are pretty much just what the term implies. They are descriptive information you record about the site you are measuring. This could include unstructured information on where the site is located, ownership and access information, indications of recent disturbance or management, and even photographs of the plot. The purpose of this type of unstructured site information is to give you reference information if you need to revisit a site or understand why you got certain values for indicators at the site. In other words, this kind of general site descriptive information can be very valuable to refer back to when you are analyzing your data.

Examples of unstructured site decriptors include plot drawings, directions, notes, or other general text information.


Structured Information

Another class of site descriptive information is structured information you collect on the physical properties of the site. This could include measures such as elevation, slope, landform, precipitation, and temperature. For these properties, specific measured values are recorded or are made according to established categories. These properties can either be measured at the site (slope and landform are easily measured in the field), from nearby locations (which may be appropriate for temperature and precipitation if there is a weather station close by), or from GIS layers (such as for elevation or climate information).


Structured site descritors include measurements or observations about a sites physical properties.


Guided Observation

Structured site descriptors can also take the form of guided observations. For example, is there evidence of erosion, grazing, fire, or disturbance? If so, a rubric can help categorize these observations into consistent classes in a repeatable way to create data that could be for analysis. Consider the signs of soil erosion that are recorded as part of the plot observation data sheet in the Monitoring Manual for Grasslands, Shrublands, and Savanah Ecosystems (Herrick et al. 2017). In this instance, observers use five classes to evaluate whether the site exhibits any signs of soil erosion for six different aspects. These kinds of observations are generally reliable and quick to implement, and can provide useful information for interpreting monitoring data.

Guided observations such as this rubric for indicators of soil erosion are useful site descriptive information.


Soils Information

The soils at a site are one of the primary determinants of land potential and the type and amount of vegetation that can occur there. Because monitoring data need to be analyzed and interpreted relative to the land’s potential, having good soils data collected from the site is important for studies and monitoring programs. Collection of soil data does not need to be complex, but it does mean that 1) you need to have some basic training and skill in collecting soils data, and 2) yes, you need to dig a small soil pit. Typically, data are collected by soil horizon or depth for soil texture class, percent rock fragment, percent clay, effervescence, color, and structure. While this may seem like a lot of work, the value of the soils data you will get from it will be worth it!

Soils data from monitoring sites are important for determining land potential



A covariate is a variable that is used to help explain the value of or change in an indicator. Covariates are not indicators themselves, but things that we record about a site that are related to, or in some cases control, the value of an indicator. Many of the site descriptors, especially structured ones like elevation, precipitation, or soils, can be used as covariates. We can also use other site measurements as covariates. For example, cover of perennial grasses at a site might be influenced by the cover and height of encroaching shrubs like juniper or mesquite. Here, perennial grass cover is the indicator, and encroaching shrub cover (also measured at the site) is a covariate. A helpful rule of thumb for determining covariates is, “If it can be measured, but it’s not one of your indicators, then it is potentially a covariate.” Many of the properties we measure or record as site descriptors are often used as covariates in analyzing monitoring data. You can find more information on covariates at the URL on the bottom of this slide.


Example: Using covariates to understand monitoring data

Consider this example of the potential for covariates to help explain monitoring data. Monitoring data were collected on 32 sample sites in a sagebrush ecosystem.

One of the primary indicators for this monitoring was cover of perennial grasses because of the strong relationship between perennial grasses and rangeland health. From this graph, we can see that the majority of the sites have more than 20% cover of perennial grasses, which looks pretty good, right? However, there are a handful of sites with lower perennial grass cover. It would be tempting to conclude that these sites are in bad shape because they lack perennial grasses, but let’s take a closer look first.


One of the covariates measured as part of this effort was precipitation from the PRISM climate layers. The precipitation information might give us a window into what is going on with perennial grasses in this area.

This graph is a plot of perennial grass cover by site precipitation. Looking at this graph, we can see there is a fairly strong relationship between precipitation and perennial grass cover. So, what does this mean? Well, one explanation is that the amount of precipitation a site receives helps determine the potential of that site to produce perennial grass. Thus, just because a site has little perennial grass does not mean we can just conclude that it is in poor shape.


Let’s take this analysis another step further, and look at the influence of soils on perennial grass cover.

This graph is the same scatter plot as the previous slide, but the points have been colored according to each plot’s ecological site. Ecological sites will be discussed in more detail in the next module, but briefly, ecological sites are determined primarily by soil composition and landform and are an expression of the site’s potential to produce a type or amount of vegetation. In other words, ecological sites tell us about land potential. If we look at the sites with low perennial grass cover, we can see that almost all of them are Black Sagebrush sites, which are inherently lower productivity sites than the other sagebrush ecological sites in the area.


The take-home message here is that we need to interpret our monitoring data within the context of land potential and other management activities or ecological processes that may be occurring at the site. Covariates help us control for these other factors in our data analyses.


Module Summary

In review, site descriptors are information that is recorded about a sampling location that can aid in analysis or interpretation of monitoring data. These can include unstructured data like general site descriptions, location information, and other notes you record. They also include structured data that you collect at a site that are not directly indicators themselves such as elevation, slope, precipitation, and soils variables. Finally, a covariate is a variable that is used to help explain an indicator value or change in an indicator that you observe at a site.

One Comment

Leave a Reply

Your email address will not be published. Required fields are marked *