Data collection (continued)

Lecture graphics

Typical sampling methods

Some measure of abundance of individual taxa (not necessarily species, though this is usu. attempted)

Quadrat sampling

Determination of quadrat and sample size:

1. Species-area method

   nested plots, each 2 x previous --> species-area curve

   
   

   various cutoff rules devised to determine where curve is "flat"

   
   

2. Statistical methods

   compute standard errors of the mean for dominant (or common) species for various quadrat sizes and number

   
   

3. Convention

Attributes commonly used to signify importance:

frequency

density (w/ easily-distinguishable individuals)

biomass (aboveground)

cover (usu. foliar)

usu. visual estimation of precise cover or assignment to cover classes (or both, w/ former preceding latter)

various scales are used for cover classes; results depend on scale used

Data collection limitations:

1. Plant distributions and environmental factors involve many spatial scales, so any sample (quadrat) size may be appropriate for some species, too large for others, too small for others

2. Accuracy usu. can be improved by increasing size and number of quadrats (to a point), w/ cost as a trade-off

Typical data management methods:

1. Transformations to reduce heterogeneity of variance

   W/ ecological data, variance usu. incr. w/ incr. in mean cover, density, or biomass --> log transformation

   y_new = log_e(y_old)

   or, w/ values of 0, y_new = log_e(y_old + 1)

   van der Maarel's cover/abundance scale is comparable to log-transformed data for cover values (5-9); abundance values (1-4) all unrealistically represent identical cover values

2. Remove rare species (5% frequency)

   more on the effects of this later

3. Eliminate outliers