Agriculture Reference
In-Depth Information
Box 11.1 Effect of Sampling Design on Diversity Measures
Nested sample designs are the only way to ensure the species-area curve increases
with increasing area. All of the California data presented in
Table 11.2
were
recorded from nested plots with the sample design in
Fig. B11.1.1
.Nestedwithin
this 0.1-ha plot are ten 100-m
2
subplots with two 1-m
2
quadrats nested in the
subplots. One concern about the use of nested designs in species-area studies is with
the statistical analysis, as least squares regression analysis assumes that samples at
different scales are independent. Nested designs potentially can result in depend-
ence between estimates at different scales, although this need not always be the
case. In a study of 90 sites in California chaparral this was tested by randomly
subsampling without replacement so that each scale comprised a different subset of
30 sites. Species-area regression equations, Standard Error and
r
2
were nearly
identical when calculated for these independent unnested plots or when calculated
from the nested plots for all 90 sites (Keeley & Fotheringham
2005
).
Data outside California reported in
Table 11.2
have been collected with
other designs and it has been proposed that plot shape and orientation can
greatly alter community diversity estimates (Harte
et al.
1999
; Stohlgren
et al.
1995
). However, field comparisons fail to show such differences (Keeley &
Fotheringham
2005
). Plot size can have an effect and as a general rule the bulk
of the aboveground and belowground plant should be contained within the
plots where it is recorded, otherwise one is sampling diversity over a larger
niche space than assumed by the plot size. In postfire shrublands, 1-m
2
plots are
large enough to contain both a good sampling of herbaceous plants as well as
seedlings and resprouts of woody species. Where sampling designs may run
into trouble is when the plot size is substantially smaller than the growth forms
being sampled. For example, Condit
et al.
(
1996
) reported that highest diversity
for tropical forests was obtained in very long 2-m bands, very likely because the
roots and canopies of these tropical trees were occupying niche space outside
the sample plot and in effect they were “sampling” from a much larger area.
50m
1m
20m
10m
10m
Fig. B11.1.1
Nested 0.1-ha sample plot used in Keeley & Fotheringham (
2005
)
.
Continued
