# Looking for ants that define prairies

My analysis right now is focused on considerations of alpha and beta diversity. I have been held up by computer issues but here are the initial results of a form of discriminant analysis on ant species occurrence in 21 sites over 3 months in the Fort Worth/Arlington, Texas area.

I used both a biplot scaling (for short gradients — which I think is most appropriate) and a Hill’s scaling (which is meant for long gradients) and got the same results. So, I will continue these analyses with the biplot scaling.  This initial work was done using soil type as the clustering category. (I will continue DA using ecological group, habitat type (disturbance), and the presence of the Comanche harvester ant as clustering groups. So, several more analyses to do.)

This first summary table and plots are for species occurrence data from 21 sites. The analysis called a CCA in CANOCO, the program I am running (CCA is a general form of CVA, and otherwise called a discriminant analysis). I am using this as a clustering analysis: do the ants cluster by soil type basically. Turn it the other way round: Knowing the ants present, can I say what the soil is? Are the ants bioindicators?  This first analysis uses all the species presence data.

Here is the summary table:

 **** Summary **** of CCA for Soil Clustering, biplot scaling, all species Axes 1 2 3 4 Total inertia Eigenvalues 1.000 0.970 0.770 0.706 6.000 Species-environment correlations 1.000 0.985 0.877 0.840 6.000 Cumulative percentage variance of species data 16.7 32.8 45.7 57.4 Cumulative percentage variance of species-environment relation 22.9 45.1 62.7 78.8 Sum of all eigenvalues 6.000 Sum of all canonical eigenvalues 4.370

And the biplot of data: messy, messy, messy. The red are the species, the triangles with the S followed by a number are the soils.

For the second analysis, I only included species that made up more than 1% of the total ants collected from that sample.

Here’s the summary table:

 **** Summary Of CCA with Ecological Descriptors **** Axes 1 2 3 4 Total inertia Eigenvalues 0.864 0.776 0.717 0.608 6.000 Species-environment correlations 0.929 0.881 0.846 0.780 Cumulative percentage variance of species data 14.4 27.3 39.3 49.4 Cumulative percentage variance of species-environment relation 24.7 46.9 67.4 84.8 Sum of all  eigenvalues 6.000 Sum of all canonical     eigenvalues 3.494

And the triplot — here with the addition of circles with numbers — these are the samples.

I haven’t reached interpretations or conclusions yet. This is still pretty raw data…

# Addtional Sampling for Ant Assemblages

One of my committee members suggested I get more samples for my ant assemblage project. I have added three prairie sites and one woodland.

One prairie and one woodland site are in the Southwest Nature Preserve in Arlington, Texas. The Comanche harvester ant is nesting in this prairie. There are about 50 colonies of Pogonomyrmex comanche. Soil here is mixed but very sandy with some bare ground.  Currently working through these samples.

Prairie Site at the Southwest Nature Preserve, Arlington, Texas

The woodland site at the Southwest Nature Preserve, Arlington, Texas

Today I set pitfall traps in the Tandy Hills Prairie and Stratford Prairie in Fort Worth, Texas. Comanche is not found in these sites. The soil here is dark and rocky, more clay and little sand, and no bare ground. I will collect these samples on Tuesday and see what I’ve got.

Tandy Hills Prairie Site, Fort Worth, Texas

Stratford Prairie Site, Fort Worth, Texas