Take-Home Exam #2 (McCarthy)

BIOS-870 Multivariate Stats, Spring 2007


 

1.  Using ISI Web of Knowledge, I uncovered over 10,000 references (from the last 5 years) in the primary biological literature using several keywords related to “ordination”.  Pick one paper that is closest to your area of biological interest that utilizes ordination as the primary analytical tool. Provide a detailed critique.  More specifically, explain what was done biologically, experimentally, and analytically. Detail what type of ordination was done, resemblance coefficients used, interpretation made, etc.  Were there better alternatives?  Please provide me with a xerographic copy of the paper when you submit your exam (I will return it). [20 points]

 

 

2.  While collecting the overstory vegetation data set previously provided in question #2 of Exam-1, we also collected a full set of environmental data. Thirty 500 m2 permanent plots were placed throughout the stand. Within each plot, we collected three soil samples that were composited and returned to the lab for analysis. Soils were air-dried, sieved, and assessed for physical and chemical attributes. Sand, silt, clay, (using hydrometer method) and organic matter (loss on ignition) were all assessed as a percentage of the total. pH was assessed using a pH probe in a water:soil slurry and recorded on the standard pH scale of 0-14. Phosphorous, potassium, magnesium, and calcium were all measured on an atomic absorption spectrophotometer and recorded as g kg-1. Light was also measured (at the center of each plot) using hemispherical photographs of the canopy (digital photographs looking up with a 8mm fisheye lens) and evaluated using a software application that documents percent open sky, percent direct beam radiation, and percent diffuse radiation. Do a complete multivariate workup on these data to determine if there are any underlying environmental gradients that are present in this forest. [40 points]

 

 

3. Due to time constraints, we will be unable to cover one relatively new and useful technique in multivariate data analysis: Canonical Correspondence Analysis (CCA). This method is related to CA which we will cover in detail. Basically, CCA allows for a direct assessment of two matrices simultaneously (usually species x samples and environment x samples matrices). Using your textbook (McCune & Grace) first provide a detailed description of what CCA is, how it works, assumptions, strengths and weaknesses, etc. (i.e., provide a written review of Chpt-21). There are only three software applications that I am aware of that do CCA: PC-ORD, CANOCO, and R. Your textbook provides a detailed discussion of the workings of CCA in PC-ORD. CANOCO is a software application specifically designed (by ter Braak) to do CCA and is available on some faculty lab computers. A nice worked example of CCA (by Dave Roberts) using the open source R code can be found here.

Retrieve your Y data matrix (species x samples) for the trees of the Wisconsin vegetation analysis that you worked on in the first problem set. Bring it into the same order and format as the environmental matrix obtained in #2 above. Perform a CCA on the data. Provide input, output, a final publication-grade plot, and a detailed description of the analysis (i.e., interpretation of output and biological conclusions). [40 points]