|Primary Contact(s)||Created||Required Software||Example Datafile|
|Justen Whittall||12 March 2009||Mesquite, pdap||Aquilegia.nex|
The goal here is to use independent contrasts to test for correlated changes in two continuous traits in a phylogenetic framework. Pollination syndromes are suites of floral traits that attract and reward a single pollinator or a group of similarly functioning pollinators. In addition to the essential pollinator observations conducted in the field, the existence of pollination syndromes would be supported by the correlated evolution of floral traits known to be involved in pollinator attraction, reward and efficiency. In the New World columbines (Aquilegia, Ranunculaceae), most species fall into three non-overlapping regions in floral morphospace. Herein, you'll use independent contrasts partial evidence for the existence of pollination syndromes (a ).
Part One: The Data
B. This file opens with three tabs – the Project, the Character Matrix, and a Tree window. The tree is completely resolved and the Character Matrix has 13 quantitative floral traits and one discrete trait with 3 states (0, 1, and 2). To see multiple tabs simultaneously, use the pop-out arrow on the appropriate tab.
Part Two: The Assumptions
Start an independent contrasts analysis from the tree window by choosing Analysis, New Chart for Tree, PDAP Diagnostic Chart, and choose “floral morphology”.
1. The first analysis in the PDAP Diagnostic Chart window is a standard assumptions check for independent contrasts. You’re looking for a non-significant slope of the regression line in the plot of the absolute value of the contrasts vs. square root of the sum of the corrected branch length (a.k.a. standard deviation). You can get statistics for the regression line (black) by switching to the Text tab near the top of the window and looking for the 2-tailed p-value near the bottom. Are branchlengths properly standardized for this first character (Flower Orientation)?
2. Clicking on any of the datapoints highlights the branch subtending the contrast of interest in the tree window (you can see this best if you have the tree and pdap windows side-by-side and have the ancestral state reconstructions in the tree window by choosing Analysis, Trace Character History, Parsimony Ancestral States, and Floral Morphology).
3. Toggle through the floral characters using the arrow at the bottom left of the pdap window to see if any violate this assumption. If necessary, we could transform character data or branchlengths to remove this correlation.
4. To see the effects of branchlengths on the analysis, start with the pdap graph on Spur Length (#2). Then, in the tree window choose Drawing and select Branches Proportional to Lengths -you’ll see this is an ultrametric tree. Toggle to Tree #2 using the blue triangle in the top left corner of the tree window (called floral traits tree – aflp branchlengths) - this is the phylogram with raw AFLP branchlengths. Notice how the pdap analysis changes with the different branchlength assignments.
Part Three: The Correlations
To test for correlated evolution of floral traits (and therefore evidence of pollination syndromes), compare contrasts in two traits by going to the PDAP.Chart menu and choose Y-contrasts vs. X-contrasts. You can control which two characters are being compared using the blue arrows at the bottom left corner of the chart. Compare contrasts for spur length (#2) and spur hue (#7). The regression line (black) is forced through the origin. Remember, you can get statistics for the regression and a sign test by clicking the Text tabs near the top and scrolling to the bottom. Examine the 2-tailed p-value from the Least Squares Regression and the Sign Test. Are spur length and color significantly correlated? Toggle some more character combinations to determine if they are evolutionarily correlated.
Testing for Correlated Changes in one Continuous and One Discrete Character
Spur Length & Pollination Syndromes: Who's Running Darwin's Race