10 Heterogeneity Lab:
10.1 Introduction & background
Apparent uniformity of grassland and woodland phases of a natural landscape as viewed from high altitudes (e.g., as seen on Google Earth images) typically disappears as we travel at ground level. Close observation reveals variation in plant form and stature as well as in species association that can create a mosaic of definable alternatives within basic woodland / shrub land / grassland themes.
For example, within in a forest there is heterogeneity vertically in the density of foliage. As a student, ecologist Robert MacArthur predicted that habitats that are highly variable in the structure of foliage should would provide more distinctive niches, which would support more bird species like warblers that forage for insects moving about in forest understory bushes, small trees and canopy trees. He proceeded to show that bird species diversity is correlated with an arbitrary index termed “foliage height diversity” (FHD). MacArthur and MacArthur (1961) calculated FHD using estimates of the proportion of total foliage in the ground, shrub and canopy in eastern deciduous forests. This is an example of being creative in developing ways to quantify habitat variation in to test ecological hypotheses.
The structural heterogeneity present in an undisturbed forest can be greatly increased by disturbance. A common form of natural disturbance in forested habitat is the death or removal of canopy trees creating “light gaps.” These gaps drive changes (and increase heterogeneity) in previously shaded understory plant populations. An example of how and why ecologists study forest gaps is seen in Lertzman et al. (1996). Note: This author and colleagues later developed the technology we apply in this exercise. Different plants and animals specialize on or rely on particular microhabitats and/or relationships generated by habitat disturbances like treefalls, landslides, fire, and herbivory. Thus some degree of disruption is promotes the diversity of species. Check out the controversy surrounding the “intermediate disturbance hypothesis”.
Since the first 373L classes took these kinds of data in the early 2000’s, there have been major changes in shrub cover in several areas of BFL. In 2002, deer density had peaked at 68 animals and browsing in the forest understory for just 10 years had reduced shrub cover (anecdotal observations). Meanwhile light availability in the understory has increased due to tree disease and drought-caused mortality. Thus in the last 17 years many large live oaks have died from oak wilt and in the past 9 years severe drought has killed large hackberry, elm and juniper trees across BFL. Between 2007 and 2015 coyote predation reduced the BFL deer population from 25 to 4 individuals. As of early 2016 deer are extinct within BFL for the first time since 1991, by which time the impact of this large herbivore on vegetative structure was in steep decline=. The cumulative effect of losing deer browsing appears to be a dramatic increase in the density of tree saplings (e.g., laurel cherry, Texas Ash) and invasive shrubs (e.g., Chinese privet) in the forest understory.
In this week’s project, we will attempt to classify and characterize the heterogeneity of habitat at BFL based upon the density of vegetation at the ground, shrub and canopy levels. In addition to learning some methods to describe degree of canopy disturbance, we will use digital cameras and gap light analysis software to test our ability to subjectively categorize canopy cover in an ecologically meaningful way. For this project, we define “canopy” as that part of the vegetation that shades understory vegetation. This exercise will teach methods for rapid assessment of vegetative structure.
10.2 Questions and Hypotheses
- What are the relationships between canopy, shrub, and ground cover?
- How do the relative abundances of Canopy, Shrub, and Ground cover compare with historical observations?
You’ll also be doing to other things that aren’t specifically hypothesis or question driven, but should be presented in the results.
- Examining the spatial mosaic of canopy heterogeneity.
- Calibrating your subjective estimates of canopy estimates with the Gap Light Analyzer calculations. This in particular isn’t a biological hypothesis, but comparing different methods for addressing the same question is a common feature of biological research.
10.3 Field Methods
After a review of the variety of ground covers, shrub covers and canopy covers encountered at BFL, we will utilize simple, easily distinguished categories for degree of coverage at each layer ranging from 0 (minimum cover) to 3 (maximum cover).
At each site, we will
- score ground cover (0-3) and enter an estimate of percent dicots
- score shrub cover (0-3) and enter an estimate of percent evergreen shrubs
- score canopy cover (0-3) and enter an estimate of percent evergreens and note the tree species above.
The shrub layer will be scored only below eye level. This level roughly corresponds with the top of the browse line when deer were present. We will compare present shrub layer scores with those in a similar drought year when deer were still a forces at BFL.
Each team will walk along 3-4 of the ten permanent transects at BFL, taking a reading of canopy cover (0-3) at each of the numbered transect markers that occur at 20 m intervals. The cumulative data will include readings at approximately 140 sites across BFL. We will calibrate our subjective estimates of canopy cover using digital cameras and fish-eye lenses to record actual canopy cover and shrub layer density using Gap Light Analyzer (FRAZER, 1999), a software package that can calculate the percent canopy openness in the hemisphere above each sample point. Make sure that you take the fish eye photo exactly where the estimate of canopy cover was taken. If a team member takes the photo after others made the visual estimates, a flag should be left as a reference for the photographer.
To take the picture, the long axis of the camera should be oriented north south with the photographer’s right hand on the south side, i.e. you should face due east, then turn the camera toward the sky. This will ensure that the bottom of the picture points east. Make sure that the camera is on a full wide angle (no zoom) and with the flash suppressed. Those taking photos need to carry a compass.
10.4 Analyses
10.4.1 Calibrating canopy estimates
Since most of our analyses rely on subjective measurements, it would be good to calibrate how well the subjective categories predict light measurements estimated by Gap Light Analyzer. Use linear regression on this class’s dataset, with subjective score as the predictor and percent openness as the response. Note that the structure of the data will almost certainly violate some of the assumptions of a linear regression test. We can still do this because this isn’t a hypothesis-driven analysis, but a prediction-driven one; provide the \(R^2\) and fit equation, but not the p-values.
10.4.2 Relationship between canopy and ground cover
Create three contingency tables examining the pairwise relationships between ground, shrub, and canopy cover cover relationship.
Examine interactions between the three levels by creating four more ground x canopy contingency tables, each containing different levels of shrub cover.
Run chi-squared (or Fisher’s exact) tests on each of these.
Don’t include the raw contingency tables in your report; instead, create enhanced versions where the cells are colored by the relative frequency of co-occurrence. Note that adding color information to a table would turn it into a figure, and it should be referred to as such.
10.4.3 Quantifying the spatial mosaic
Use a map of BFL and number/color the canopy level at each point. Connect adjacent areas of the same number. Note whether any spatial patterns in cover vales can be associated with main habitat types based on past land use, substrate types, and history of natural disturbances.
If you do this by hand, scan or photograph it and include it in the Canvas submission.
10.4.4 Historical comparison
How are the current levels of shrub, canopy, and ground cover different from previous years? We will be comparing current data with fall 2004, when deer populations were high. For each cover type (canopy, shrub, and ground), make a contingency table comparing present cover levels vs. those in 2004 and run a chi-squared test. Visualize the contingency tables in the same way you did with the previous set.