See

discussions, stats, and author profiles for this publication at:

https://www.researchgate.net/publication/226409828

A Modified-Whittaker nested vegetation

sampling method

Article in Plant Ecology · April 1995

DOI: 10.1007/BF00045503

CITATIONS READS

329 667

3 authors, including:

Thomas J. Stohlgren Maurya Falkner

Colorado State University State of California
234 PUBLICATIONS 9,316 CITATIONS 14 PUBLICATIONS 458 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Semi-retirement! View project

All content following this page was uploaded by Thomas J. Stohlgren on 15 April 2014.

The user has requested enhancement of the downloaded file.

Vegetatio 117: 113-121, 1995. 113
(~ 1995KluwerAcademicPublishers. Printedin Belgium.

A Modified-Whittaker nested vegetation sampling method

T.J. S t o h l g r e n 1, M . B . F a l k n e r 2 and L . D . Schell 2
1National Biological Service, Natural Resource Ecology Laboratory, Colorado State University, Fort Collins,
Colorado 80523, USA; 2Natural Resource Ecology Laboratory, Colorado State University, Fort Collins,
Colorado 80523, USA

Accepted6 December1994

Key words: Nested sampling techniques, Plant species richness, Species-area curves, Species diversity

Abstract

A standardized sampling technique for measuring plant diversity is needed to assist in resource inventories and for
monitoring long-term trends in vascular plant species richness. The widely used 'Whittaker plot' (Shmida 1984)
collects species richness data at multiple spatial scales, using 1 m 2, 10 m 2, and 100 m 2 subplots within a 20 m x 50 m
(1000 m 2) plot, but it has three distinct design flaws involving the shape and placement of subplots. We modified
and tested a comparable sampling design (Modified-Whittaker plot) that minimizes the problems encountered in
the original Whittaker design, while maintaining many of its attractive attributes. We overlaid the two sampling
methods in forest and prairie vegetation types in Larimer County, Colorado, USA (n = 13 sites) and Wind Cave
National Park, South Dakota, USA (n = 19 sites) and showed that the modified design often returned significantly
higher (p < 0.05) species richness values in the 1 m 2, 10 m 2, and 100 m 2 subplots. For all plots, except seven
ecotone plots, there was a significant difference (p < 0.001) between the Whittaker plot and the Modified-Whittaker
plot when estimating the total number of species in the 1000 m 2 plots based on linear regressions of the subplot data:
the Whittaker plot method, on average, underestimated plant species richness by 34%. Species-area relationships,
using the Modified-Whittaker design, conformed better to published semilog relationships, explaining, on average,
92% of the variation. Using the original Whittaker design, the semilog species-area relationships were not as
strong, explaining only 83% of the variation, on average. The Modified-Whittaker plot design may allow for better
estimates of mean species cover, analysis of plant diversity patterns at multiple spatial scales, and trend analysis
from monitoring a series of strategically-placed, long-term plots.

Introduction al. 1987) recommend overlaying nested quadrats of

increasing size to quantify species-area curves (Fig. 1a)
Several recent books have focused on the importance - one measure of species richness in an area. Palmer
of biological diversity (e.g. Magurran 1988; Wilson (1990, 1991) compared several methods for estimat-
1988; Soul6 & Kohm 1989; Peters & Lovejoy 1992), ing species richness including: number of observed
but none contain precise methodologies to measure it. species, extrapolation from species-area curves, inte-
There has been a renewed interest to quantify patterns gration of the lognormal distribution, and nonparamet-
of biological diversity at state and national scales (e.g. ric estimators, He cautioned that species-area curves
GAP Analysis; Scott et al. 1993), and local scales may have different forms at different scales. Extrapola-
(Dallmeier 1992; Tilman & Downing 1994), but land- tion of species-area information may also be influenced
scape ecologists have not yet agreed on a standard- by the sampling design used.
ized field methodology for evaluating plant diversity We believe many field ecologists continue to ago-
(Stohlgren 1994). nize about how to select the appropriate quadrat shapes
Textbooks on vegetation sampling methods (e.g. and sizes for particular vegetation types (also see
Mueller-Dombois & Ellenberg 1974; Barbour et Podani et al. 1993). This problem is, of course, magni-
114

da 1984). Shmida (1984) showed a strong semilog

relationship
A. Nested Quadrat
S = b + dlogA (1)
~ ,
between the number of species (S) and quadrat area
(A)(b is a constant, d is the slope). This is similar
8meters to other reported species-area relationships (Miller &
Wiegert 1989; Shafer 1990; but see Pielou 1977).
B° Whittaker Plot (WH) The 20 m × 50 m size of the 'Whittaker plot' (Fig.
lb) was designed initially for more-or-less homoge-
2
neous areas so that replicate sites were easy to find
A
(Shmida 1984). Borrowing perhaps from the nested
50meters quadrat approach (Fig. la), the 1 m x 1 m subplots are
overlaid on the 2 m × 5 m subplots, which, in turn, are
C. ~. Long-Thin Plot (LT) overlaid on the 10 m x 10 m subplot. The most attrac-
tive attributes of the Whittaker plot design are that
---- r it provided plant ecologists with: (1) a standardized
100 meters approach to quantify species richness in different plant
communities; and (2) insights on the effects of quadrat
D. Modified-Whittaker Plot (MW)
0.20 9.20 21.20 33.20 45,20 size when determining species-area relationships.
Stohlgren (1994) pointed out that the Whittaker
o.s- ls.T.5
~. 50.10
! plot has three distinct design flaws. First, if the habitat
r is not strictly homogeneous, species richness is influ-
0,0 15JI 27.1) 39,0 49.5.0
50 meters enced by plot shape. Circular or square plots (with
a reduced perimeter to surface area ratio) will have
Fig. 1. Nested vegetation sampling designs. (A) Nested Quadrat fewer species, in general, than a long-thin rectangle
design (from Mueller-Dombois & Ellenberg 1974), (B) Whittak- covering a more heterogeneous area (Bormann 1953).
er Plot design (from Shmida 1984), (C) Long-Thin Plot design Second, plot size and shape interactions may influence
(from Stohlgren 1994), and (D) Modified-Whittaker Plot design (this
paper; shown with x, y coordinates of subplot and plot comers). species richness (Pielou 1977). Note that the Whittaker
plot design shifts from 1 m x 1 m squares to 2 m ×
5 m rectangles to a 10 m × 10 m square, then back
fled when multiple vegetation types or large, heteroge- to a 20 m x 50 m rectangle, which confounds the
neous landscapes must be studied (Stohlgren 1994). influences of plot shape and size (Table 1; Fig. lb).
Shmida (1984) published a nested vegetation sam- Third, is the problem of spatial autocorrelation. Not
pling method that was developed and used by R. H. only are the ten 1 m × 1 m plots contiguous in one
Whittaker over many years. The primary purpose of small area of the 20 m × 50 m plot (i.e. high spa-
that paper was to present Whittaker's nested vegetation tial autocorrelation), the successively larger plots are
sampling technique for the measurement of species superimposed on the smaller plots (i.e. the plots are
diversity to compare different plant communities from not independent in terms of species richness). Thus,
different regions of the world. Whittaker (1977) had a species-rich area in one of the 1 m × 1 m plots
long realized that patterns of plant diversity can be elu- affects the species richness reported in the larger-sized
cidated only by systematic surveys and by sampling at plots. This last problem is inevitable among nested
multiple spatial scales (i.e. nested quadrat sizes of 1 (overlaid) quadrats (e.g. Mueller-Dombois & Ellen-
m 2, 10 m 2, and 100 m 2 within a 1000 m 2 area). Multi- berg 1974; Pielou 1977; Barbour et al. 1987), and
scale sampling of vegetation allows for: (1) evalua- results may be heavily biased by the starting location
tions of the influence of spatial scale on local species in the field.
richness patterns (Podani et al. 1993); (2) better com- Initiated in May, 1992, the primary objective of our
parisons of community richness than single-scale mea- vascular plant species richness project was to develop
surements provide (Whittaker 1977); and (3) the devel- and test alternate field methods to minimize the design
opment of species-area (or 'collector's curves'; Pielou problems of the Whittaker plot when collecting plant
1977) to estimate larger-scale richness patterns (Shmi- diversity information. In an earlier study, Stohlgren
 115

Table 1. Dimensionsof plots and subplots in various nested vegetation sampling designs.

Design
Dimensions Whittaker Long-Thin Modified-Whittaker
Ten 1 m2 subplots 1 m × 1 m (contiguous, clus- 0.32 m x 3.16 m (systemat- 0.5 m x 2 m (systematic, outer
tered) ic, outer perimeter not con- perimeter not contiguous)
tiguous)
Two 10 m2 subplots 2 m x 5 m (contiguous, over- 1 m x 10 m(systematic, outer 2 m x 5 m (systematic, outer
lapping) perimeter non-overlapping) perimeter non-overlapping)
One 100 m2 subplot 10 m x 10 m (overlapping) 3.16 m x 31.6 m (centered, 5 m x 20 m (centered, non-
non-overlapping) overlapping)
One 1000 m2 plot 20 m × 50 m (overlapping) 10 m x 100 m (overlapping) 20 m x 50 m (overlapping)
Source Shmida 1984 Stohlgren 1994 This paper

(1994) developed and tested an alternate nested vege- We hoped to develop a standardized nested plot
tation sampling design (10 m x 100 m 'long-thin plot'; technique that can be used by plant ecologists for better
Fig. 1c; Table 1) in the Beaver Meadows area of Rocky estimates of local species richness and mean species
Mountain National Park, Colorado, USA. The 400 m cover, analysis of plant diversity spatial patterns, and
x 350 m area (elevation 2800 m) contained portions trend analysis from monitoring a series of strategically-
of a riparian meadow and a lodgepole pine-ponderosa placed, long-term plots.
pine forest. An ecotone area was subjectively defined
as within 100 m of the forest edge to test the differ-
ent plot designs in heterogeneous areas. We randomly Methods and study areas
located the southwest corner (of north-south orient-
ed plots) and superimposed three Whittaker plots and We modified the 20 m x 50 m Whittaker plot design
three long-thin plots in the meadow, ecotone, and forest keeping the attractive features of the long-thin plot and
areas of the study site. original Whittaker plot designs (Fig. 1d; Table 1). Like
In all habitat types, and for all plot sizes, the long- the long-thin plot design, the Modified-Whittaker plot
thin plot design consistently returned higher species design minimizes the problems in the original Whit-
richness values than the Whittaker design. And, the taker design by using consistent rectangle proportions
species richness sampling with the long-thin plot in the subplots to remove subplot size-shape interac-
design more accurately reflected the total species rich- tions (Stohlgren 1994). Like the Whittaker plot design,
ness recorded in a complete vascular plant survey of the the Modified-Whittaker plot is 20 m x 50 m. How-
area (Stohlgren 1994). However, the long-thin design ever, like the long-thin plot (Fig. lc), the 1 m 2 and
was somewhat cumbersome for field crews, and total 10 m 2 subplots are arranged systematically inside the
species richness was not comparable to Whittaker plot perimeter of the 20 m x 50 m plot. Likewise, the
data (Whittaker et al. 1979; Naveh & Whittaker 1979) 100 m 2 subplot is centered in the plot. The three sub-
or data collected by many investigators who used 20 m plot sizes are independent and non-overlapping and
x 50 m plots (e.g. Rice & Westoby 1983; Baker 1990). species richness can be used to construct species-area
So, a further modification of the Whittaker plot design curves. Pielou (1977) used the term 'collector's curve'
was warranted. The objectives of this research were where one is trying to casually compile an exhaus-
to: (1) develop a new nested vegetation plot design tive list of species in an area or where nested subplots
(the Modified-Whittaker plot design) that minimizes are independent (non-overlapping). We prefer the term
the statistical problems of the original Whittaker plot species-area curve to collector's curve because: (!) our
design; (2) compare the two designs in several habitats three subplot sizes are non-overlapping and indepen-
in terms of multi-scale species richness patterns; and dent of each other (and only 13% of the 20 m x 50 m
(3) evaluate how data collected using the two designs plot is not independent); (2) we focus on precise spa-
conforms to established species-area relationship the- tial scale aspects of species richness from subplot data;
ory. and (3) it is more commonly used and understood.
116

We compared the Whittaker and the Modified- Whittaker plot design for estimating total species rich-
Whittaker plot designs by overlaying them in a variety ness in the 0.1 ha plots from subplot data.
of habitat types in Larimer County, Colorado, USA (n Using data from the 1 m 2, 10 m 2, and 100
= 13 sites), and in Wind Cave National Park, South m 2 subplots and 1000 m 2 plots, we found that the
Dakota, USA (n = 19 sites; Appendix). We used the Modified-Whittaker design conformed better to pub-
paired t test (Zar 1974) to compare the number of lished semilog relationships, explaining, on average,
species recorded in the 1 m 2, 10 m 2, and 100 m 2 sub- 92% of the variation (Table 3). Using the original Whit-
plots of both plot designs. taker design, the semilog species-area relationships
We used the semilog expression discussed earlier were not as strong, explaining only 83% of the varia-
(i.e. linear regression) to estimate the total number of tion on average. The mean coefficients of determina-
species in each 1000 m z plot based on the cumula- tion were significantly different for all sites and vegeta-
tive species recorded in the 1 m 2, 10 m z, and 100 m 2 tion types (except the ecotone plots, which followed the
subplots. These 'expected' values for both plot designs same general pattern but were more variable). Thus, the
can be compared to 'observed' values of the total num- Modified-Whittaker plot technique was robust across
ber of species recorded in 20 m × 50 m plots. Since the range of habitats, vegetation types, and land use
the observed values were identical for each pair of characteristics included in this study (Fig. 2, Tables 2
plots, the paired t test was used to compare the mean and 3, and Appendix) and study plots were lumped for
observed-expected values for both plot techniques. The the comparisons that follow.
plot technique with the smallest differences between Based on the semilog linear relationships from the
observed and expected values would be more useful subplots, the Whittaker plot design, on average, under-
in estimating local species richness. We also used the estimated species richness in the 0.1 ha plot by about
semilog linear regressions to evaluate which sampling 34%. The Modified-Whittaker plot design also under-
design conformed better with established species-area estimated total species richness, but only by an average
relationship theory (Shmida 1984; Shafer 1990). The of 9%. The Modified-Whittaker plot design was more
technique that produced higher coefficients of determi- stable over the range of species richness values tested
nation (r2 values) would be more accurate in estimating (Fig. 3). The Whittaker plot design became less stable
the species richness of larger areas (in the same habi- (i.e. greater absolute differences between observed and
tat). The paired t test was used to compare mean coeffi- expected values) as plot species richness increased.
cients of determination of species-area regressions for Fitting average species richness values from all
the different sites and vegetation types. subplots and plots (Fig. 4), we found a stronger
species(S)-area(A; m 2) relationship (S = 25.5 + 4.5
× log A; r 2 = 0.992; p < 0.0001) using the Modified-
Results Whittaker design compared to the original Whittaker
design (S = 13.6 + 7.4 x log A; r 2 = 0.912; p <
The Modified-Whittaker plot design returned signif- 0.0001).
icantly higher (p < 0.05) species richness values in
the 1 m 2, 10 m 2, and 100 m 2 subplots for the com-
bined data (Fig. 2). Comparisons of the Whittaker and Discussion
Modified-Whittaker plot designs also show similar pat-
terns across study sites and in a variety of habitats (Fig. While additional field tests are needed in other vegeta-
2). The forest plots were more variable, but still showed tion types, the Modified-Whittaker nested vegetation
the same general pattern. sampling design looks promising for several reasons.
Species richness in the vegetation communities First, the new design minimizes the problems in the
tested ranged from 10 species/0.1 ha to 69 species/0.1 original Whittaker design. The consistent rectangle
ha. Based on the semilog linear relationships from the proportions remove subplot size-shape interactions,
1 m 2, 10 m 2, and 100 m 2 subplots, we found significant and rectangles generally perform better than squares
differences between observed and expected values of at recovering species richness (Stohlgren 1994). The
the total number of species recorded in 20 m × 50 m subplots in the Modified-Whittaker design have less
plots for all sites and vegetation types (Table 2). With overlap than the subplots in the original design (except
the exception of seven 'ecotone' plots, the Modified- for the largest size plot, of course). Thus, they are
Whittaker plot design was superior to the original influenced less by spatial autocorrelation and non-
 117

COLORADO (n=13) 15°r S.DAKOTA (all9) F......................

COMBINED (n=32) i

24 20
"~ 20

o .... 0 . . . . . .

PRAIRIE (n=lS) FOREST (n-7)

,*50 5o *
~ 40

20~ ~ ~
10

0
1
--
10 100 1000

~/~
20
10

0 fH iJ] Itt t 1
1

WHIT
tO 100

Subplot/Plot Area (m 2)
1
1000

MOD-WHIT
3
20
10

0
1 10 100 1000

• * P < 0.05 * P < 0.01

Fig. 2. Comparison of the mean number of species recorded by subplot/plot area in the Whittaker and Modified-Whittaker plot designs for
different study locations and vegetation types. Vertical bars represent standard errors. Double asterisks show statistical significant differences
at c~= 0.05 (paired t test), a single asterisk denotes a difference at c~= 0.10.

Table 2. Mean observed-expected species richness values (absolute values) for the 20 m × 50 m plots.

Site/Vegetation Type Observed-Expected Species Richness Paired t test p

Whittaker Mean (S.E.) Modified-Whittaker Mean (S.E.)

Colorado Sites (n = 13) 18.8 (2.6) 3.6 (0.8) < 0.001

South Dakota Sites (n = 19) 10.6 (1.2) 3.7 (0.7) < 0.001
Combined Sites (n = 32) 14.0 (1.4) 3.6 (0.5) < 0.001
Prairie Plots (n = 18) 11.0 (1.4) 3.4 (0.6) < 0.001
Forest Plots (n = 7) 17.1 (2.3) 4.3 (1.4) < 0.001
Ecotone Plots (n = 7) 18.3 (4.6) 3.7 (1.3) < 0.1

i n d e p e n d e n c e o f observations (Pielou 1977). B e c a u s e represent small-scale species richness in the 20 m x

v e g e t a t i o n is often clustered spatially (Fortin et aL 50 m plot. This is precisely what our data show.
1989), and m o s t species are rare in c o v e r and abun- T h e greatest difference in the two techniques is in
dance ( B a r b o u r et al. 1987), the ten c o n t i g u o u s 1 m 2 the n u m b e r o f species r e c o r d e d in the 1 m 2 subplots
subplots in the original W h i t t a k e r design are m o r e like- (Fig. 2; Fig. 4). The 1 m 2 subplots using the M o d i f i e d -
ly to miss important patches o f vegetation and under- W h i t t a k e r design a v e r a g e d 26.0 (4-1.8; S.E.) species,
118

Table 3. Mean coefficients of determination (r2) for semilog linear regressions of species (number of species) to
area (m 2) relationships, n.s. = not significant.

Site/Vegetation Type Mean Species-Area r2 Paired t test p

Whittaker Mean (S.E.) Modified-Whittaker Mean (S.E,)

Colorado Sites (n = 13) 0.785 (0.02) 0.912 (0.02) < 0.02

South Dakota Sites (n = 19) 0.855 (0.02) 0.922 (0.01) < 0.05
Combined Sites (n = 32) 0.827 (0.02) 0.917 (0.01) < 0.01
Prairie Plots (n = 18) 0.820 (0.01) 0.910 (0.02) < 0.05
Forest Plots (n = 7) 0.820 (0.01) 0.928 (0.02) < 0.05
Ecotone Plots (n = 7) 0.850 (0.03) 0.927 (0.03) n.s.

while the Whittaker 1 m 2 subplots averaged only 16.7 30 'WIIIT

(3-1.4) species. The original Whittaker design of con- MIII)-WIII'I

tiguous 1 m × 1 m subplots (that combined cover only 25

a 1 m x 10 m in the center of the 20 m x 50 m plot)

2(J
probably provides a very biased view of small-scale
interactions occuring over the plot because the sub-
plots: (1) exclude many species found in the 20 m x
50 m plot (Fig. 2); and (2) they result in artificially- lo
• o •

c~
low variance due to spatial autocorrelation problems 0

(Stohlgren 1994). Particularly for plant species cov- g

er estimates conducted in the 1 m 2 subplots in both d5
rr <~
'9

designs, the systematic subplots are more likely to give o

o 10 20 30 411 50 60 70 80
better estimates of mean species cover for the 1000 m 2 'rO'lTkL SPECIES (0.1 ha)
area. And, the systematic placement of subplots around
the perimeter of the plot allows for the analysis of spa- Fig. 3. Relationship of observed minus expected species richness
tial patterns in plant diversity within a 20 m x 50 m to observed species richness (i.e. total species in each 0.1 ha plot)
area. using the two techniques.
Second, the attractive attributes of the original
Whittaker design are maintained. A nested vegeta-
tion sampling design (i.e. sampling species richness at as long-term study plots. Random placement of sub-
multiple spatial scales) allows for mathematical esti- plots may produce similar species richness data to the
mates of total diversity (Shmida 1984; Palmer 1990, systematic placement used in the Modified-Whittaker
1991). The same 'sample data sheets' presented in design. But, random placement of the non-overlapping
Shmida (1984) can be used. However, species cover- subplots (i.e. x, y coordinates, angle, etc.) would be
abundance data from the 1 m 2 subplots are improved more difficult for field crews, and permanent marking
by recording more species over a broader area of 0.1 of the subplots would be necessary if trend analysis was
ha plot. Strong species-area relationships (Table 3, Fig. a study objective. As stated previously, the long-thin
4) allow for better estimates of local species richness plot design (Fig. lc) was cumbersome for field crews,
from a series of plots (although species-area curves and the increased perimeter to area ratio of the subplots
may not always fit the semilog form; Pielou 1977). made it more difficult for field crews to identify which
And, comparisons of species richness can be made plants should be included or excluded.
among community types or throughout the world by Our botanists outline the 20 m × 50 m with two
Whittaker and others (e.g. Nevah & Whittaker 1979; 75 m tapes, use a 50 m tape for the center 5 m x 20 m
Rice & Westoby 1983; Baker 1990). subplot, then use a snap-together 0.5 m × 2 m PVC
The systematic placement of the subplots makes frame to record species cover in the 1 m 2 subplots adja-
the design easy to use in the field now and in the future cent to the borders (Fig. ld). Decimeter markings on
the PVC frame aid in the cover estimates. As an inter-
 119

and complete surveys of large areas are generally

• WIIIT • MOD-WtllT
cost prohibitive (Stohlgren & Quinn 1992). Cost-
efficient standardized sampling techniques for other
5O components of biodiversity must also be developed
(Stohlgren et al. 1994).
MOD-WIIIT ; ~ ' . - -"~:" "
40
r 1.0.992 - .~ -"""

>
r. 30
,d
Acknowledgements
20
r.; -" WHIT The US Department of Interior National Park Service
z 10 r ~ - e.912
and National Biological Service provided the fund-
0
ing for the research. James Detling and Dennis Oji-
0 1 2 3 4 ma at Colorado State University contributed the South
Dakota plot data, and the staff at Wind Cave National
LOG x (PLOT AREA m 2)
Park added logistical support. Jean Marie Ederer, Rick
Edwards, Emily Galbraith, Kate Healy, Alicia Lizarra-
Fig. 4. Overall species-area relationships for the two sampling ga, Lisa Nelson, Laura Stretch, and Krista Alper pro-
designs based on the mean number of species by subplot/plot area
vided field assistance. Alan Carpenter and two anony-
(all sites combined). Vertical bars represent standard errors.
mous reviewers improved the manuscript considerably.
To all we are grateful.

esting added benefit, we are using the ratio of native

to non-native species at the multiple spatial scales as References
one environmental indicator of ecosystem condition.
This has potentially broad application in comparing Baker, W.L. 1990. Species richness of Colorado riparian vegetation.
J. Veg. Sci. 1: 119-124.
various land use practices and successional patterns, Barbour, M.G., Burk, J.H. & Pitts W.D. 1987. Terrestrial Plant Ecol-
and to investigate species-environment relationships. ogy. Second Edition. Benjamin/Cummings Publishing Company,
This nested design can be shrunk (proportionately) for Menlo Park, California.
smaller-scale studies (e.g. thin riparian zones, tundra Bormann, EH. 1953. The statistical efficiency of sample plot size
and shape in forest ecology. Ecology 34: 474---487.
vegetation, lichen surveys, etc.). It can be expanded Dallmeier, E 1992. (ed.) Long-term Monitoring of Biological Diver-
for larger-scale habitats (such as widely scattered tree sity in Tropical Forest Areas: Methods For Establishment and
habitats), but longer tapes and better field surveying Inventory of Permanent Plots. MAB Digest 11. United Nations
techniques might be needed. Educational, Scientific, and Cultural Organization (UNESCO),
Paris, France.
We are, unfortunately, far from standardizing field Fortin, M., Drapeau, P. & Legendre, P. 1989. Spatial autocorrelation
techniques for assessing plant species diversity at land- and sampling design in plant ecology. Vegetatio 83: 209-222.
scape scales. Decisions must be made about the number Heltshe, J.E, & Forrester, N.E. 1983. Estimating species richness
using the jackknife procedure. Biometrics 39: 1-12.
and the placement of plots needed to describe the plant
Magurran, A.E. 1988. Ecological Diversity and Its Measurement.
species diversity of large study areas and landscapes Princeton University Press, Princeton, New Jersey.
(Stohlgren 1994). The timing of sampling, and species' Miller, R.I. & Wiegert, R.G. 1989. Documenting completeness,
rareness, and the patterns of landscape features must species-area relations, and the species-abundance distribution of
a regional flora. Ecology 70: 16-22.
also be considered. Determining the efficiency of Mueller-Dombois, D. & Ellenberg, H. 1974. Aims and Methods of
sampling in vascular plant species richness projects Vegetation Ecology. John Wiley & Sons. New York.
will depend on species discovery/accumulation rates Nevah, Z. & Whittaker, R.H. 1979. Structural and floristic diver-
and rate of observed changes (Heltshe & Forrester sity of shrublands and woodlands in northern Israel and other
Mediterranean areas. Vegetatio 41: 171-190.
1983; Miller & Wiegert 1989). Trend analysis from Palmer, M.W. 1990. The estimation of species richness by extrapo-
monitoring a series of strategically-placed, long-term lation. Ecology 71:1195-1198.
Modified-Whittaker study plots may be a valuable tool Palmer, M.W. 1991. Estimating species richness: the second-order
for quantifying and detecting trends in vascular plant jackknife reconsidered. Ecology 72:1512-1513.
Peters, R.L. & Lovejoy, T.E. 1992. Global Warming and Biological
species richness. Diversity. Yale University Press. London.
Species-level plant distribution data at landscape Pielou, E.C. 1977. Mathematical Ecology. pp. 285-290. John Wiley
scales are expensive to collect. Most species are rare & Sons, New York, NY.
120

Podani, J., Cz~tr'hn, T. & Bartha, S. 1993. Pattern, area and diversity: Stohlgren, T.J. & Quinn, J.F. 1992. An assessment of biotic inven-
the importance of spatial scale in species assemblages. Abstracta tories in western US national parks. Natural Areas Journal 12:
Botanica 17: 37-51. 145-154.
Rice, B. & Westoby, M. 1983. Plant species richness at the 0.1 Stohlgren, T.J., Quinn, J.E, Ruggiero, M. & Waggoner, G. 1993.
hectare scale in Australian vegetation compared to other conti- Status of biotic inventories in US National Parks. Biological Con-
nents. Vegetatio 52: 129-140. servation 71: 97-106.
Scott, J.M., Davis, E, Csuti, R., Noss, R., Butterfield, B., Groves, Soul6, M.E. & Kohm, K.A. 1989. Research Priorities for Conserva-
C., Anderson, H., Caicco, S., D'Erchia, F., Dewards, Jr., T.C., tion Biology. Island Press. Washington, DC.
Ulliman, J. & Wright, R.G. 1993. GAP Analysis: a geographic Tilman, D. & Downing, J.A. 1994. Biodiversity and stability in
approach to protection of biological diversity. Wildlife Mono- grasslands. Nature 367: 363-365.
graphs 123: 1-41. Whittaker, R.H. 1977. Evolution of species diversity on land com-
Shafer, C.L. 1990. Nature reserves: Island theory and conservation munities. Evolutionary Biology 10" 1-67.
practice. Smithsonian Institute Press, Washington, DC. pp. 189. Whittaker, R.H., Niering, W.A. & Crisp, M.O. 1979. Structure,
Shmida, A. 1984. Whittaker's plant diversity sampling method. pattern, and diversity ofa mallee community in New South Wales.
Israel Journal of Botany 33: 41-46. Vegetatio 39: 65-76.
Stohlgren, T.J. 1994. Planning long-term vegetation studies at land- Wilson, E.O. 1988. Biodiversity. National Academy Press. Wash-
scape scales, pp. 209-241. In: Ecological Time Series. Powell, ington, DC.
T.M. & Steele, J.H. (eds) Chapman & Hall, New York. (In Press). Zar, J.H. 1974. Biostatistical analysis. Prentice-Hall. New Jersey.
 121

Appendix. Study site locations and vegetation types:

Site Name Location, State Elevation Vegetation Comments

1. Button Rock Forest Button Rock Preserve, CO 2134m Pine/C3/C4 Southeast facing, Ponderosa savanna
2. Taft Hill Disturbed Area Fort Collins City Limit, CO 1554m C3 Weedy Species Deep soil, C3 dominated
3. Taft Hill Disturbed Area Fort Collins City Limit, CO 1554m C3/C4 Weedy/Prairie Mix Deep soil, C3 dominated
4. Taft Hill Disturbed Area Fort Collins City Limit, CO 1554m C3/C4 Weedy/Prairie Mix Deep soil, C3 dominated
5. Pineridge Prairie Pineridge Open Space, CO 1631m C3/C4 Weedy Species Prairie dog colony
6. Pineridge Prairie Pineridge Open Space, CO 1631m C3/C4 Weedy/Prairie Mix Prairie dog colony
7. Pineridge Prairie Pineridge Open Space, CO 1631m C31C4Weedy/Prairie Mix Prairie dog colony
8. Pineridge Ecotone Pineridge Open Space, CO 1634m Pine/C3/C4 Prairie Mix Forest/Prairie border, east facing
9. Pineridge Ecotone Pineridge Open Space, CO 1634m Pine/C3/C4 Prairie Mix Forest/Prairie border, east facing
10. Pineridge Ecotone Pineridge Open Space, CO 1634m Pine/C3/C4 Prairie Mix Forest/Prairie border, east facing
11. Pineridge Forest Pineridge Open Space, CO 1695m Pine/C3/C4 Ponderosa, east facing
12. Pineridge Forest Pineridge Open Space, CO 1695m Pine/C3/C4 Ponderosa, east facing
13. Pineridge Forest Pineridge Open Space, CO 1695m Pine/C3/C4 Ponderosa, east facing
14. Red Valley Grass Wind Cave National Park, SD 1316m C3/C4 Prairie Upland, rocky, C4 dominated
15. Fire Tower Prairie Wind Cave National Park, SD 1390m C3/C4 Prairie Deep soil, C3 dominated, west facing
16. Fire Tower Ecotone Wind Cave National Park, SD 1396m Pine/C3/C4 Prairie Mix Forest/Prairie border, west-facing
17. Fire Tower Forest Wind Cave National Park, SD 1402m Pine/C3/C4 Prairie Mix Rocky, pine dominated, west facing
18. Archer Hill Bottom Wind Cave National Park, SD 1122m C3/C4 Prairie Deep soil, C3 dominated
19. Orchard East Midslope Wind Cave National Park, SD 1176m C3/C4 Prairie Midslope east facing
20. Highland Creek E. Prairie Wind Cave National 1322m C3/C4 Prairie Deep soil, C3 dominated, east facing
Park, SD
21. Highland Creek E. Ecotone Wind Cave National Park SD 1335m Pine/C3/C4 Forest Prairie border, east facing
22. Highland Creek E. Forest Wind Cave National Park SD 1341m Ponderosa Pine/C3/C4 Mix Rocky, pine dominated, east facing
23. Boland Ridge N. Prairie Wind Cave National Park SD 1243m C3/C4 Prairie Mix C3 dominated, north facing
24. Boland Ridge N. Ecotone Wind Cave National Park SD 1255m Pine/C3/C4 Prairie Mix Forest Prairie border, north-facing
25. Boland Ridge N. Forest Wind Cave National Park SD 1274m Ponderosa Pine/C3/C4 Mix Rocky, pine dominated, north facing
26. Boland South Midslope Wind Cave National Park SD 1183m C3/C4 Prairie Mix South facing, rocky soil
27. Five Six S. Bottom Wind Cave National Park SD 1158m C3/C4 Deep soils, C3 dominated
28. Five Six S. Midslope Wind Cave National Park SD l170m C3/C4 South-facing
29. Five Six S. Upper Wind Cave National Park SD 1182m C3/C4 C4 dominated, rocky
30. Look Out Upper Wind Cave Nationa/Park SD 1311m C3/C4 C4 dominated, rocky
31. Custer E. Prairie Wind Cave National Park SD 1384m C3/C4 C3 dominated, deep soil
32. Custer E. Prairie Ecotone Wind Cave National Park SD 1396m Pine C3/C4 Forest/Prairie, east facing

View publication stats