enhancement or degradation (e.g. climate change, gradual changes 

 in predation pressure, introduction of a pathogen, etc.). It is 

 only the last source of variation that is of interest in a long- 

 term monitoring study; (1) and (2) are "noise" that must be 

 removed to reveal the long-term trends. The long-term changes 

 can be separated by using a temporally stratified sampling 

 design. Transects are read for 2-4 consecutive years and then 

 reread for 2-4 consecutive years at 5 to 10-year intervals 

 following the baseline period. For example transects might be 

 read yearly in 1991-1993 and then again in 1998-1999. A mixed- 

 model ANOVA can be used to analyze these data. Period (1991-1993 

 and 1998-1999) is the fixed factor with years within period as 

 replications, and plant is the random factor. A significant 

 effect due to period would indicate a trend. Variation due to 

 experimental error and short-term fluctuations comprise the error 

 term in the ANOVA. Data may have to be transformed before 

 analysis. 



A less rigorous analysis for two years of data may be 

 performed by scoring each plant as either bigger or smaller 

 (disregarding those that stayed the same) and examining the data 

 array visually to determine if equal numbers of plants are 

 increasing and decreasing in size. 



12 



