Journal of
Ecology 2003
91, 234 – 239
Temporal heterogeneity of soil moisture in grassland
and forest
Blackwell Publishing Ltd.
SARAH E. JAMES, MEELIS PÄRTEL*, SCOTT D. WILSON and
DUANE A. PELTZER†
Department of Biology, University of Regina, Regina, Saskatchewan, Canada S4S 0A2
Summary
1 Differences between growth forms in the spatial heterogeneity of associated soil
resources, such as water, are well-documented. We tested for differences in the temporal
heterogeneity of soil moisture between natural grassland, shrubland and Populus
tremuloides forest at the northern edge of the Great Plains.
2 Weekly measurements of soil moisture over a year, and daily measurements during a
growing season, both showed significant interactions between habitat and time. Soils
under grassland and shrubs were wettest at the start of the growing season but driest at
the end.
3 The coefficient of variation of soil moisture content over time during the growing
season was significantly higher in grassland than in forest. Similar results were found for
whole-year measurements, and at two depths beneath the soil surface (10 and 30 cm).
4 We found little interception of rainfall in any vegetation type. The net effect of vegetation on soil moisture during a drying period, however, was significantly greater in grassland than in forest, suggesting that differences in the temporal heterogeneity of soil
moisture content are related to resource uptake.
Key-words: evapotranspiration, interception, resource uptake, shrubs, trees
Journal of Ecology (2003) 91, 234–239
Introduction
Environmental heterogeneity is a well-recognized
influence on plant communities because species differ
both in their responses to, and effects on, patchiness
(Hutchings et al. 2000). Plant species also differ in their
responses to temporal heterogeneity in resource supply
(Grime 1994), but the ability of plants to create temporal heterogeneity has received little attention (Wilson
2000). Previous studies of the influence of vegetation
on temporal heterogeneity in soil moisture content
have either found no significant heterogeneity (Farley
& Fitter 1999) or no differences between growth forms
(Mitchell et al. 1993). These studies were carried out in
relatively mesic sites, however, and significant effects of
plants on resources are more likely to emerge in cases
where the resource is in short supply. Here we test for
differences in vegetation effects on temporal hetero-
© 2003 British
Ecological Society
Correspondence: Scott Wilson (tel. 306-585-4287; fax 306585- 4894; e-mail [email protected]).
*Present address: Institute of Botany and Ecology, Tartu
University, Tartu 51005, Estonia.
†Present address: Landcare Research, Lincoln 8152, New
Zealand.
geneity in soil moisture content in a more xeric region, the
North American Great Plains.
In semiarid environments such as the Great Plains,
water becomes available in pulses (Sala & Lauenroth
1982), creating the potential for different species to
exploit different amounts of temporal heterogeneity
(Goldberg & Novoplansky 1997). On a global scale,
between-year variability in rainfall is significantly
greater for grasslands than forests (Knapp & Smith
2001), suggesting that grasses have faced selection for
coping with greater temporal heterogeneity in soil
moisture content, which may also allow them to exploit
within-year heterogeneity. An ability of grasses to
amplify temporal heterogeneity in soil moisture content might contribute to the exclusion of trees from
grasslands (Wilson 2000).
Grasses might increase the temporal heterogeneity
of soil moisture through water uptake. Fine root length
per unit volume of soil is 20 times greater in grassland
than in forest (Jackson et al. 1997) and, in the Great
Plains, the ability of grasses to reduce soil moisture is
several times greater than that of woody plants on a
per-gram basis (Köchy & Wilson 2000). These facts
suggest that potentially higher rates of water uptake by
grasses may increase the temporal variability of soil
235
Temporal
heterogeneity in
grassland and
forest
moisture content in grassland. In support of this, a 2year field study in our region showed that the coefficient of variation (between sample periods) of moisture
under grassland (62%) was twice that under forest
(33%; data from Peltzer 2001).
Contrasting mechanisms, however, could cause temporal heterogeneity to be higher in forest habitats.
Transpiration, for example, is largely driven by leaf
area index, which is much higher in forest than grassland (Kleb & Wilson 1997). Trees could also increase
the temporal heterogeneity of soil moisture content
if their relatively massive canopies intercept all the
moisture in small rainfall events but allow throughfall
during larger rains. The coefficient of variation of soil
moisture content over time was significantly higher
under woody vegetation than under grasses, both in
England and Saskatchewan (Wilson 2000).
In summary, there is uncertainty about the magnitude and direction of differences in the temporal heterogeneity of soil moisture content between vegetation
types. The causes of any differences are also unclear.
Our goal was to test for differences in daily and weekly
variability of soil moisture between grassland, shrubland and forest habitats, and to examine the contributions of canopy interception and plant uptake to any
observed differences.
Methods
© 2003 British
Ecological Society,
Journal of Ecology,
91, 234–239
We worked at White Butte Recreation Area (58°28′ N,
104°22′ W, 20 km east of Regina, Canada) in a mosaic
of three habitats: grassland (mixed-grass prairie dominated by Stipa comata, Carex eleocharis, Bouteloua
gracilis), shrubland (clonal stands of Symphoricarpos
occidentalis) and forest (Populus tremuloides). Nomenclature follows Looman & Best (1987). Symphoricarpos is about 50 cm tall (i.e. closer in height to grasses
than trees). Shrubs and trees have spread clonally into
grassland in this region (Archibold & Wilson 1980).
Vegetation, resources and spatial variation at the site
are described elsewhere (Wilson 1993; Kleb & Wilson
1997; Köchy & Wilson 1997; Peltzer 2001; Pärtel &
Wilson 2002). The climate is continental, with average
temperatures in July and January of 19 °C and −18 °C,
respectively. Annual precipitation is 384 mm, which
mostly falls in May and June (Environment Canada,
unpublished data).
We examined temporal variability in soil moisture
content at 30 sites (10 in each of grassland, shrubs and
forest) scattered along a 500-m stretch on the edge of an
invading aspen forest. Forest sites were 10–20 m inside,
and grassland and shrubs sites 10–30 m outside, the
edge of the forest. Sites were close together in order to
keep precipitation and soils as similar as possible
between habitats. At each site we installed a tube (PVC,
5 cm diameter, 90 cm long, of which 15 cm was left
above the soil surface), which allowed access for a soil
moisture probe (Sentry 200-AP, Troxler Electronic
Laboratories Inc., Research Triangle Park, North
Carolina) without repeated soil disturbance. Tubes
were covered with caps to keep out rain. The probe
measured soil moisture at 10 and 30 cm below the soil
surface, using a form of time domain reflectrometry
(Rundel & Jarrell 1991). Moisture was measured in a
disk-shaped electrical field 10 cm deep (centred 10 and
30 cm beneath the soil surface) that extended 10 cm out
from the side of the access tube. We investigated soils
near the surface because they contain most roots, both
in general (Jackson et al. 1996) and at our study site
(Wilson & Kleb 1996), and are most influenced by
precipitation (Sala et al. 1997).
We measured soil moisture between 12.00 and 15.00,
every day during the growing season (May to August
1999) and thereafter at weekly intervals until May
2000. We included measurements from outside the
growing season because vegetation is likely to influence
soil moisture at any time. Soil moisture was not measured during rain, because water on the probe influences
readings, nor during January to March when the soil
was frozen, nor during a brief period in August due to
instrument failure. We transformed each probe reading
to relative soil moisture content (actual value minus the
annual minimum value, divided by the annual range,
Dyer & Rice 1997), using the annual minimum and
maximum readings (both depths included) for each
trio of sites closest to each other that contained all three
habitats. Relative soil moisture content allowed us to
examine changes over time in the absence of spatial
variation. Repeated measures  compared relative
soil moisture content between habitats at each depth,
both for the whole year (weekly measurements) and for
the growing season alone (daily data).
Temporal variability of soil moisture content was
defined as the coefficient of variation (CV; Kleb &
Wilson 1997; Knapp & Smith 2001) of relative soil
moisture content. Coefficient of variation expresses
variance as a proportion of the mean, allowing comparisons that are independent of scale (Webster &
Oliver 1990). CV was calculated for each depth at each
site, both across the whole year and across the growing
season.  was used to test for differences between
habitats in the temporal variability of soil moisture
content.
We tested whether the interception of rain by vegetation differed between habitats by collecting precipitation and throughfall daily during a period with
frequent rain (June and early July 1999). Precipitation
was measured at each grassland site in three plastic
vials (46 mm diameter, 71 mm deep) separated from
each other by 10 cm and elevated on stakes that were
60 cm tall. Throughfall was measured at each site in all
habitats using three buried vials, with their tops at the
soil surface. Vials collected only throughfall because
the sandy soil of the study site prevents run-off. For
each of the 12 rainfall events during this period, we calculated mean throughfall for each habitat and mean
precipitation.  was used to test whether the relationship between throughfall (dependent variable) and
236
S. E. James et al.
precipitation (independent variable), and thus interception, differed between habitats.
We tested whether water uptake by plants differed
between habitats using daily measurements of soil
moisture content (at both depths) in vegetated and
unvegetated sites during a period with relatively little
rain (late July and August 1999), when day-to-day
changes in soil moisture content reflected plant uptake.
Three unvegetated sites were created in grassland by
removing all vegetation within a circle 2 m in diameter,
centred on an access tube, by spraying with glyphosate
(3%) and clipping 1 week later. To reduce uptake by
neighbouring roots, soil on the circumference of the
circle was cut to a depth of 20 cm. We determined the
mean soil moisture content of the three unvegetated
sites and in vegetated sites in each habitat on each day.
Soil moisture in unvegetated sites reflected the influence only of evaporation, whereas soil moisture in vegetated sites reflected the effects of evapotranspiration.
Thus, any departure from unity in the relationship
between soil moisture in unvegetated and vegetated
sites represented transpiration. Vegetation may have
other effects on soil moisture, such as those produced
by shading or reduced wind speed at the soil surface,
but our main goal was to examine biological effects,
and the largest of these during a dry period is likely to
be transpiration. We used  to test whether the
relationship between soil moisture in vegetated sites
(dependent variable) and unvegetated sites (independent variable) differed between habitats, to determine
whether plant uptake differed between habitats.
Precipitation data were obtained from Regina, the
closest weather station to the study site (Environment
Canada, unpublished data), 20 km west.
Fig. 1 The weekly sum of precipitation (a), and relative soil
moisture content under grassland, shrubs and forest, 10 cm
(b) and 30 cm (c) below the soil surface. Each point represents
the mean of 10 locations for each habitat. For clarity, only one
of the seven soil moisture readings taken each week during the
growing season is shown.
Results
© 2003 British
Ecological Society,
Journal of Ecology,
91, 234–239
Precipitation during the study period was typical for
the region, occurring as pulses and declining through
May and June (Fig. 1a).
Soil moisture varied significantly with time (P <
0.001), but not between habitats (P > 0.05). The interaction between time and habitat was significant
(P < 0.001). Similar results were found at both depths,
and for both the whole year and the growing season
(Fig. 1b,c).
Significant interactions between time and habitat
arose because differences in soil moisture between
habitats changed with time. At 10 cm depth, soils
were driest under forest during May and June 1999,
under grassland in July 1999, and under shrubs for the
rest of the year (Fig. 1b). At 30 cm depth, soils were
driest under forest in May and June 1999 and May
2000, but driest under grassland for the rest of the year
(Fig. 1c). In summary, the significant interactions
between habitat and time (at both depths and for both
whole-year and growing-season data) reflect differences between habitats in temporal patterns of soil
moisture.
Temporal variability (measured as the coefficient
of variation, CV) of soil moisture over the whole year
differed significantly between habitats (10 cm depth,
F 2,27 = 4.7, P < 0.05; 30 cm depth, F 2,27 = 14.6,
P < 0.001). At a depth of 10 cm, the CV of soil moisture was significantly (P < 0.05) greater under shrubs
(mean + SD: 73.9 + 17.4) than under grassland (57.6 +
13.0) or forest (54.7 + 15.5). At a depth of 30 cm, the
CV of soil moisture under grassland (27.0 + 4.2) was
significantly greater than under shrubs (20.2 + 3.5) or
forest (17.7 + 4.2).
Temporal variability in soil moisture during the
growing season also varied significantly between
habitats (10 cm depth, F 2,27 = 8.7, P < 0.001; 30 cm
depth, F 2,27 = 5.8, P < 0.01). At a depth of 10 cm, the
CV of soil moisture was significantly greater under
grassland (24.9 + 4.0) and shrubs (22.0 + 3.6) than
forest (17.2 + 5.0). At a depth of 30 cm, the CV of
soil moisture under grassland (15.3 + 4.7) was significantly greater than that under forest (10.1 + 3.8);
the CV under shrubs was intermediate in value
(11.6 + 3.2).
habitats (F2,87 = 0.7, P = 0.521), suggesting that plant
effects on soil moisture at this depth were similar in the
three vegetation types. At a depth of 30 cm, however,
the slope of the relationship was significantly lower for
forest than for grassland or shrubs (Fig. 3, F2,87 =
14.6, P < 0.001). This occurred because grassland and
shrub soils were wetter than those under forest when
soil moisture was high (Fig. 3, right side) but drier than
soils under forest during dry conditions (see data
points at the left side of Fig. 3).
237
Temporal
heterogeneity in
grassland and
forest
Discussion
Fig. 2 The relationship between throughfall and precipitation
in grassland, shrubs and forest. Each point is the mean of
10 sites measured after one precipitation event. Regression
relationships between throughfall and precipitation did not
differ significantly between habitats. Shading represents the
area above the major axis. Data beneath this area suggest that
interception occurred.
Fig. 3 The relationship between soil moisture content in
vegetated and unvegetated plots in grassland, shrubs and
forest 30 cm below the soil surface. The slope for forest is
significantly lower than that for grassland or shrubs, suggesting that plant uptake of water from dry soils was less for
forest than for grassland or shrubs.
© 2003 British
Ecological Society,
Journal of Ecology,
91, 234–239
In all cases with significant differences, relatively
low-stature vegetation (grassland or shrubs) had
greater temporal heterogeneity in soil moisture than
did forest.
Throughfall increased with precipitation, and all
values in the relationship were close to the major diagonal, indicating that little interception occurred
(Fig. 2). Regression slopes did not differ between habitats (F 2,30 = 2.3, P = 0.119), suggesting that interception did not differ between habitats.
Soil moisture content was much greater in unvegetated sites than in vegetated sites (Fig. 3, note axis
ranges), reflecting vegetation effects on soil moisture
during a drying cycle. At a depth of 10 cm, the slope of
the relationship between soil moisture in unvegetated
and vegetated plots did not differ significantly between
Temporal heterogeneity in soil moisture content was
significantly greater under grassland and shrubs than
under forest, because grassland and shrubs had higher
soil moisture content at the start of the growing season
but lower soil moisture content at the end (Fig. 1. b,c).
This difference between habitats was reflected by the
significant interaction between time and habitat in the
s for soil moisture content.
In contrast, a common-garden experiment in mesic
Alabama showed no difference between the effects of
grass and tree species on temporal heterogeneity of soil
moisture (Mitchell et al. 1993). Further, no evidence
for temporal variability in soil moisture content was
found in an English forest studied over the course of a
year, even though nutrient availability varied over time
(Farley & Fitter 1999). In both cases, the relatively high
availability of water may have decreased the impact of
plants on soil moisture variability. Plant effects on
resource heterogeneity may be most apparent where
resources are in short supply. Plant effects on heterogeneity may be less apparent following disturbance
(e.g. Guo et al. 2002), when resources are relatively
abundant.
There were no differences between vegetation types
in interception of precipitation (Fig. 2). This was surprising in light of the fact that shoot mass is 70 times
greater in forest than grassland (Wilson 1993). Interception, however, may increase more with total leaf
area than with leaf mass, and fine grass leaves have
higher areas per unit mass than do the coarser leaves
and stems of trees. High shoot mass in forest may be
counterbalanced by high leaf area in grassland, making the two vegetation types similar in interception
properties.
Throughfall was almost identical to precipitation
(points in Fig. 2 occur close to the major diagonal),
suggesting that interception was unimportant in our
region. Interception might have been low at our site
because precipitation usually falls in large amounts
during short periods (Fig. 1a), whereas interception is
typically greatest during small rainfall events (Clark 1940).
The greater temporal heterogeneity of soil moisture
content under grassland reflected the relatively high
ability of grasses to reduce soil moisture content during
periods with little rain (Fig. 3). This may reflect the fact
that fine root length is 20 times greater in temperate
238
S. E. James et al.
© 2003 British
Ecological Society,
Journal of Ecology,
91, 234–239
grassland soils than in forest soils (Jackson et al. 1997)
and that the root : shoot ratio is nearly 30 times greater
in grassland than forest at our study site (Wilson 1993).
As a result, the ability of grass to reduce soil moisture
is nearly five times that of woody vegetation, expressed
on a per-gram basis (Köchy & Wilson 2000). Other
factors may lower temporal variability further in forest.
Mean growing season temperature at our study site
is 1.4 °C lower in forest than in adjacent grassland
(Köchy & Wilson 1997), which could decrease evaporation. Abundant litter (1854 g m−2 in forest vs. 159 g m−2
in grassland, Wilson 1993) and reduced windspeed
may also reduce water loss from forest soils.
Differences detected in soils beneath contrasting
growth forms may be related more to root allocation
than to woody stems. Whenever there were significant
differences between shrubs and trees (see Results),
temporal heterogeneity was greater under shrubs, i.e.
the effects on soil properties of the short-stature
shrub Symphoricarpos occidentalis were generally more
similar to those of grassland than forest. Further, no
differences were found between growth forms in
interception, a response driven by shoots.
Soil moisture content is generally lower under grasslands than forest (Belsky 1994; Li & Wilson 1998), as
seen in our study at the end of the growing season
(Fig. 1). A previous study at our site, however, showed
that moisture was lower in grassland than in forest
throughout the growing season (Peltzer 2001). Peltzer
(2001) measured soil moisture using lysimeters (plastic
tubes), which excluded root uptake, whereas we measured soil that included active roots. Thus, the difference
between the two sets of results provides further evidence that uptake by plants is a mechanism contributing to greater variability in soil moisture content in
grassland.
Soil moisture was always greater at 30 cm than at
10 cm, but the decline of soil moisture during the growing season was similar at both depths (Fig. 1. b,c). Both
patterns may be due to the sandy soils of the study area,
which allow water to infiltrate quickly, thus increasing
soil moisture content at 30 cm and causing temporal
patterns to be generally concordant between depths.
Evidence for rapid infiltration is provided by the
response of soil moisture content to the study period’s
largest rainfall event in late August (Fig. 1a): soil moisture content at 10 cm depth showed no response
(Fig. 1b), whereas moisture at 30 cm depth increased
by nearly 50% (Fig. 1c).
Grasses and woody plants might differ in the depth
from which they obtain most of their soil water, with
fibrous-rooted grasses specializing in acquisition close
to the surface and woody-rooted trees specializing in
acquiring water from greater depths (Walter 1971).
Removal experiments involving shrubs and grasses
obtained some support for this hypothesis (Knoop &
Walker 1985; Golluscio et al. 1998). In our study, however, soils were driest under grasses at both depths at
the end of the growing season (Fig. 1. b,c). Deuterium
addition experiments on the Colorado plateau also
failed to obtain evidence of depth partitioning of water
uptake between species (Schwinning et al. 2002). Overlap of water uptake depths is not surprising in light of
the overlap between grasses and woody plants in root
profiles (Jackson et al. 1996; Wilson & Kleb 1996).
In summary, temporal patterns of soil moisture content differed significantly between vegetation types,
and temporal heterogeneity was significantly greater in
grassland and shrub communities than in forest. The
ability of grasses to reduce soil moisture content to low
levels in the absence of rain is consistent with intense
below-ground competition in grassland (Wilson 1998).
Acknowledgements
We thank J. McClaren and S. Yakimowsky for field
assistance, M. Hutchings, L. Haddon, S. Wiser and
anonymous reviewers for improving the paper, Saskatchewan Environment and Resource Management
for access to White Butte, and the Natural Sciences and
Engineering Research Council of Canada for support.
References
Archibold, O.W. & Wilson, M.R. (1980) The natural vegetation of Saskatchewan prior to agricultural settlement.
Canadian Journal of Botany, 58, 2031– 2042.
Belsky, A.J. (1994) Influences of trees on savanna productivity: tests of shade, nutrients, and tree-grass competition.
Ecology, 75, 922 – 932.
Clark, O.R. (1940) Interception of rainfall by prairie grasses,
weeds, and certain crop plants. Ecological Monographs, 10,
243–277.
Dyer, A.R. & Rice, K.J. (1997) Intraspecific and diffuse competition: the response of Nassella pulchra in a California
grassland. Ecological Applications, 7, 484 – 492.
Farley, R.A. & Fitter, A.H. (1999) Temporal and spatial
variation in soil resources in a deciduous woodland. Journal
of Ecology, 87, 688 – 696.
Goldberg, D. & Novoplansky, A. (1997) On the relative
importance of competition in unproductive environments.
Journal of Ecology, 85, 409 – 418.
Golluscio, R.A., Sala, O.E. & Lauenroth, W.K. (1998) Differential use of large summer rainfall events by shrubs and
grasses: a manipulative experiment in the Patagonian
steppe. Oecologia, 115, 17 –25.
Grime, J.P. (1994) The role of plasticity in exploiting environmental heterogeneity. Exploitation of Environmental
Heterogeneity by Plants (eds M.M. Caldwell & R.W. Pearcy),
pp. 1–19. Academic Press, San Diego.
Guo, D., Mou, P., Jones, R.H. & Mitchell, R.J. (2002)
Temporal changes in spatial patterns of soil moisture following disturbance: an experimental approach. Journal of
Ecology, 90, 338 – 347.
Hutchings, M.J., John, E.A. & Stewart, A.J.A. (2000) The
Ecological Consequences of Environmental Heterogeneity.
Blackwell Science, Oxford.
Jackson, R.B., Canadell, J., Ehleringer, J.R., Mooney, H.A.,
Sala, O.E. & Schulze, E.D. (1996) A global analysis of
root distributions for terrestrial biomes. Oecologia, 108,
389 – 411.
Jackson, R.B., Mooney, H.A. & Schulze, E.-D. (1997) A global
budget for fine root biomass, surface area, and nutrient contents. Proceedings of the National Academy of Sciences USA,
94, 7362 –7366.
239
Temporal
heterogeneity in
grassland and
forest
© 2003 British
Ecological Society,
Journal of Ecology,
91, 234–239
Kleb, H.R. & Wilson, S.D. (1997) Vegetation effects on soil
resource heterogeneity in prairie and forest. American
Naturalist, 150, 283 –298.
Knapp, A.K. & Smith, M.D. (2001) Variation between
biomes in temporal dynamics of aboveground primary
productivity. Science, 291, 481– 484.
Knoop, W.T. & Walker, B.H. (1985) Interactions of woody
and herbaceous vegetation in a southern African savanna.
Journal of Ecology, 73, 235 – 253.
Köchy, M. & Wilson, S.D. (1997) Litter decomposition and
nitrogen dynamics in aspen forest and mixed-grass prairie.
Ecology, 78, 732 –739.
Köchy, M. & Wilson, S.D. (2000) Competitive effects of
shrubs and grasses in prairie. Oikos, 91, 385 – 395.
Li, X. & Wilson, S.D. (1998) Facilitation between woody
plants establishing in an old field. Ecology, 79, 2694–
2705.
Looman, J. & Best, K.K. (1987) Budd’s Flora of the Canadian
Prairie Provinces. Publication 1662. Agriculture Canada
Research Branch, Hull.
Mitchell, R.J., Zutter, B.R., Green, T.H., Perry, M.A.,
Gjerstad, D.H. & Glover, G.R. (1993) Spatial and temporal
variation in competitive effects on soil moisture and pine
response. Ecological Applications, 3, 167–174.
Pärtel, M. & Wilson, S.D. (2002) Root dynamics and spatial
pattern in prairie and forest. Ecology, 83, 1199 –1203.
Peltzer, D.A. (2001) Plant responses to competition and soil
origin across a prairie–forest boundary. Journal of Ecology,
89, 176 –185.
Rundel, P.W. & Jarrell, W.M. (1991) Water in the environment.
Plant Physiological Ecology (eds R.W. Pearcy, J.R. Ehleringer,
H.A. Mooney & P.W. Rundel), pp. 29 –56. Chapman &
Hall, London.
Sala, O.E. & Lauenroth, W.K. (1982) Small rainfall events: an
ecological role in semiarid regions. Oecologia, 53, 301–304.
Sala, O.E., Lauenroth, W.K. & Golluscio, R.A. (1997) Plant
functional types in temperate semi-arid regions. Plant
Functional Types (eds T.M. Smith & H.H. Shugart),
pp. 217–233. Cambridge University Press, Cambridge.
Schwinning, S., Davis, K., Richardson, L. & Ehleringer, J.
(2002) Deuterium enriched irrigation indicates different
forms of rain use in shrub/grass species of the Colorado
Plateau. Oecologia, 130, 345 –355.
Walter, H. (1971) Ecology of Tropical and Sub-Tropical Vegetation. Oliver & Boyd, Edinburgh.
Webster, R. & Oliver, M.A. (1990) Statistical Methods in Soil
and Land Resource Survey. Oxford University Press, New York.
Wilson, S.D. (1993) Belowground competition in forest and
prairie. Oikos, 68, 146 –150.
Wilson, S.D. (1998) Competition between grasses and woody
plants. Population Biology of Grasses (ed. G.P. Cheplick),
pp. 231–254. Cambridge University Press, Cambridge.
Wilson, S.D. (2000) Heterogeneity, diversity, and scale in
plant communities. Ecological Consequences of Habitat
Heterogeneity (eds M.J. Hutchings, E.A. John & A.J. Stewart),
pp. 53–69. Blackwell Science, Oxford.
Wilson, S.D. & Kleb, H.R. (1996) The influence of prairie and
forest vegetation on soil moisture and available nitrogen.
American Midland Naturalist, 136, 222 –231.
Received 9 July 2002
revision accepted 5 December 2002