Noxubee NWR moist soil unit - 2015 samples
Adam Smith
10 May, 2016
This document is an initial attempt to characterize the seed-producing plant community composition of Noxubee NWR’s moist soil unit, based on 180 1-m2 plots spread across 14 moist soil sub-units in the study area. The plot locations were determined by a GRTS draw stratified based on the area of each unit.
The specific objectives of this document are:
Characterize the community of plants relevant to waterfowl on the moist soil units.
Convert plant community information into duck energy day estimates at the unit level, and aggregate to the entire study area.
Study area
Figure 1 delineates the moist soil unit and identifies individual sub-units and final plot locations. Hover the mouse over a polygon for the unit name or a point for the plot number.
Figure 1.Community composition
There are any of a number of ways to visualize/characterize plant community composition, but it’s something that does not summarize easily, at least when it comes to making comparisons among units.
Here we present a couple of approaches. The first summarizes the proportion of cover of the 19 plant species recorded on the moist soil unit, plus a category for no available food. They are ordered in decreasing order of total cover on all plots.
Figure 2 makes apparent that the most common cover type varies considerably among sub-units.
Figure 2.
A second way to visualize the community composition of units is to use ordination. Ordination tries to reduce the multidimensional community (20 cover types) into a small number of dimension to assist interpretation. Figure 3 illustrates a nonmetric multidimensional scaling view of the moist soil units. The details are beyond the scope of this document, but the figure is informative. The locations of 20 cover types are presented in two dimensions and the 14 units are plotted on this cover type space. Sub-units nearer to each other in their plant community composition are more similar to each other and are associated with the the nearest cover types. Sub-units farther apart have less similar communities associated with the nearest cover types. For example, on average, sub-unit 10 is quite dissimilar from other sub-units and is associated with a high proportion of Smartweed cover; this can be checked using Figure 2. A similar conclusion can be drawn for sub-unit 3 and Panic Grass/Bidens. Note that there is a large cluster of relatively similar plots surrounding the “No Food” cover type, with the composition of less dominant providing only minor separation among the sub-units in this two-dimensional space.
Figure 3.
If we drop the “No Food” category and compare the communities based only on their actual plant composition, we get some better separation of sub-units in ordination space (Figure 4). For example, sub-unit 1 is associated more with Carex than most plots, sub-unit 2 with Juncus, sub-unit 10 with Foxtail and Smartweed, sub-unit 7 strongly with Smartweed, and sub-unit 9S with essentially nothing because it has little food.
Figure 4.
Duck energy days (DEDs) per acre
We can use the estimated energy density and per area yield for each cover species to estimate the duck energy days per acre on each sub-unit (Figure 5). In short, we do this by estimating the DEDs on each individual sample plot and then summarizing by sub-unit. These calculations assume a daily energetic requirement of 292 kCal per duck and the yields and metabolizable energies in Table 1. NOTE: I’ve changed the TME values for Bidens and Smartweed since the original values appeared to be total metabolizable energy rather than true metabolizable energy (new sources: Table 1.2 in Sherfy 1999, Table 1 in Sherfy et al. 2005, Table 1 in Checkett et al. 2002).
| Cover species | Total metabolizable energy (kCal/g) | Yield (kg/ha) |
|---|---|---|
| NoFood | 0 | 0 |
| Sprangletop | 2.62 | 1100 |
| Millet | 2.61 | 1100 |
| Smartweed | 2.61 | 1100 |
| Beakrush | 1.86 | 496 |
| Bidens | 0.5 | 1100 |
| Carex | 0.92 | 1100 |
| Cyperus | 0.92 | 1100 |
| Foxtail | 2.62 | 1100 |
| Paspalum | 1.57 | 496 |
| PanicGrass | 2.62 | 496 |
| Juncus | 1.86 | 496 |
| Rumex | 2.68 | 496 |
| JohnsonGrass | 2.62 | 496 |
| Ludwigia | 1.83 | 496 |
| TealGrass | 2.62 | 1100 |
| CrabGrass | 3.1 | 1100 |
| ButtonBush | 1.83 | 496 |
| ToothCup | 1.83 | 496 |
| Ragweed | 1.82 | 496 |
Figure 5.
Now we can spatially visualize the estimated (median) DEDs per acre by sub-unit (Figure 6).
Figure 6.From this data, we can derive estimates (and some measure of uncertainty) for the total DEDs available on each sub-unit and for the moist soil unit as a whole (Figure 7).
First, we estimate the median total DEDs on each sub-unit (and bootstrapped 95% confidence interval of this median). We use the median as DED estimates from plots in some sub-units are quite skewed (see Figure 5). To generate these estimates, we draw a large number (1000) of bootstrapped samples, with replacement, of the same size (i.e., number of plots) as measured in the field. For each of these samples, we calculate the median DEDs/ac and multiply times the acreage of the given sub-unit. This produces a distribution of total DEDs for a sub-unit from which Figure 7 is derived; the numbers indicate the median total DEDs (in thousands):
Figure 7.
From this bootstrapping exercise we can also generate an estimate, rounded to the nearest thousand, for the total duck energy days present on the entire moist soil unit. This is calculated by summing the total DEDs for each sub-unit in each of the 1000 iterations. The median (and 95% CI) of total DEDs on the moist soil units was 219000 (174000 - 268000) DEDs.