By K.M. Olson-Rutz,
C.B. Marlow, K. Hansen, L.C. Gagnon, and RJ. Rossi*
Abstract
We evaluated the impact of packstock
grazing on a dry, upper timberline meadow. Horses were
picketed on 15 m ropes for different durations, months,
and frequencies over 3 summers. Before horse grazing,
we estimated vegetal, bare soil, litter, rock, and moss
cover, measured grass and forb plant heights, counted
grass and forb stems per area, and determined the percent
of plants grazed. These measurements were repeated 1
growing season later. More bare ground and less litter
and vegetal cover were recorded 1 year following single
8- or 18-hour grazing events. Single grazing events
of 4-hour duration had no effect on cover. Decreases
in vegetal cover were associated with reduced stem numbers.
Eighteen hour picket durations reduced subsequent year
production of grass and forb stems. We discuss the difficulties
encountered in this study, including estimates of necessary
sample sizes, to help in the design of future studies.
Introduction
Management of wilderness or natural
areas is no longer a question of whether it is necessary
but how it should be carried out. Although wildland
ecosystems may sustain some human or domestic animal
use without altering the processes that sustain the
native plant and animal communities, the need to manage
such areas (hereafter referred to as wildland areas)
has been recognized for many years (Krumpe and McLaughlin
1987). Policy makers and managers of wildlands are faced
with the paradox of protecting unique ecosystems or
areas with little human development while allowing recreational
use (Kuss and Graefe 1985, Cole 1987, McClaran 1989,
McClaran and Cole 1993).
One impact of wildland recreational
use is packstock grazing. The level of grazing impact
is affected by the intensity, frequency, and season
of defoliation and the status (grazed or ungrazed) of
neighboring plants. In areas with short growing seasons,
grazing
impacts on plants may not be expressed
until the following growing season (Olson and Richards
1988). From livestock grazing studies, we know that
heavy and multiple defoliations will result in changes
that are apparent the following growing season such
as reductions in biomass production (Cook et al. 1958,
Trlica et al. 1977, Miller and Donart 1981), leaf or
culm length (Cook et al. 1958, Trlica et al. 1977, Edwards
1985, Olson and Richards 1988), number of seed heads
or flowers (Cook et a!. 1958, Mueggler 1975, Edwards
1985), crown size, root and crown TNC concentrations
(Trlica et al. 1977, Miller and Donart 1981), root growth
(Cook et a!. 1958, Richards 1984), and number of tillers
per area (Stout et a!. 1981). Declines in some or all
of these measures suggest an adverse affect on plant
growth and reproduction which can create opportunity
for invasion or expansion of other species (Forcella
and Wood 1986, Silvertown and Smith 1989). Invasion
of wildland plant communities by species that are not
native to the area or expansion of species restricted
to certain habitat types will change the character of
the area, possibly reducing its pristine wilderness
value.
While the principles of livestock
grazing may apply to pack-stock grazing, the response
of high elevation plant communities has received little
study. Also, management objectives of wild-land areas
often differ from those of livestock production systems
(McClaran and Cole 1993). Wildland management objectives
may include maintenance of species diversity or to protect
specific organisms. The ideal may be to manage for no
change, yet limits of acceptable change should be defined
(Krumpe and McLaughlin 1987). Grazing management research
in wildland areas needs to address how such areas respond
to use, and which community responses can be used to
indicate acceptable or unacceptable change (McClaran
and Cole 1993). Our objective was to determine the tolerance
of a mountain meadow plant community to packstock grazing.
Such information can help managers develop packstock
grazing guidelines.
Methods
The study site is a dry, southwest sloping, upper timberline
meadow at 2660 m on the Burntfork of the Bacon Rind
drainage in the Lee Metcalf Wilderness Area in southwestern
Montana. The area is classified as a Festuca idahoensis/Elymus
trachycaulus habitat type (Mueggler and Stewart 1980)
with fine textured soils of the Cryoboroll or Cryochrept
group (Montagne et al. 1982). Records of the Hebgen
Ranger District, Gallatin National Forest, indicate
that the area was never part of a livestock grazing
allotment. There are no lakes or streams and the area
is closed to hunting, therefore, the meadow is used
little by recreational packstock, but is used frequently
by elk (Cervus ciaphus elaphus).
By picketing horses (Equus caballus)
on this meadow we were able to impose a known amount
and season of horse use. In 1988, horses were picketed
on a set of circles for 4 durations (0, 4, 8, 18 hours)
in each of 3 months (early July, mid August, mid September).
The 4-hour circles were grazed for 4 consecutive hours,
the 8-hour circles for 4 hours in both morning and evening,
and the 18-hour circles for 9 hours a day for 2 consecutive
days. There were 4 replicate picket circles per grazing
treatment (duration by month) for a total of 48 circles.
Horses were initially assigned randomly to duration
and replicate. When possible, the same horse was assigned
to the same duration in subsequent months.
In 1989 we repeated the treatments
on another ungrazed part of the same meadow. In addition
to the earlier grazing duration by month treatments,
we added another treatment involving repeated use of
the same picket circles. To achieve this treatment,
4 circles were grazed for 4 hours and another 4 were
grazed for 8 hours each month (July, August, and September).
This treatment was identified as JAS. In 1990, the 1989
circles were regrazed with the same grazing treatment
as applied in 1989.
Circles were 15 m in diameter. Four
transect lines (bearing N, 5, E, W) were marked in each
circle. A 2 X 5 cm frame (Morris 1973) was placed perpendicular
to the transect line at 0.30 m intervals. Twenty-five
such frames were read on the N and S transects. Only
the outer 4 m of the E and W transects (14 frames each)
were sampled to avoid over-sampling the circles' centers.
Before grazing the following data
were collected from each frame:
-
estimated percent cover of mineral
soil, rock, moss/lichen, litter, and basal vegetation
in 10% increments,
-
separate stem counts of grasses
and forbs,
-
height class (0 = no plants,
1 = 0-2 cm, 2 = 2-4 cm, 3 = 4-12 cm, 4 = 12-24 cm,
5 = > 24 cm) of the tallest plant material of
the dominant (according to stem count) vegetation
type (grass or forb) in 1988, and of the tallest
grass and forb thereafter,
-
whether grasses or forbs were
grazed to quantify elk grazing (except in 1988),
and
-
penetration resistance of the
top 1 cm soil layer with a pocket ring penetrometer.
We collected four, 20 cm soil cores from each circle
to determine percentage of soil water content. In
September 1988, 6 inches of snow prevented us from
taking pre-grazing measurements.
All parameters were re-measured in
August 1989 on the circles grazed in 1988, and in July
1990 and 1991 on the circles grazed in 1989 and re-grazed
in 1990. All data were summarized to a mean measurement
per circle. We calculated the proportion of frames with
plants in each height class and the percent of frames
with grazed plants. Cover, height class, and grazed
plant frequency were aicsine-squareroot transformed
to achieve a near normal data distribution to accommodate
analysis of variance.
Initially we calculated a relative
index of change (Cole 1987) from pre-grazing values
to measurements taken 1 year later. However, high standard
errors and high index values for the controls (sometimes
greater than 7,000 when they should be 1), made the
validity of this index for these data questionable and
the index was not used. To control for external environmental
variables and differences among circles before grazing,
we calculated the difference between the initial, and
1 or 2 year later values (1989-1988, 1990- 1989, 1991-1989).
These values include an effect of month when pre-grazing
data were collected, therefore we can only compare grazing
durations within a month.
We used analysis of variance with
duration (hours on picket), vegetation type (grass or
forb), and vegetation type by duration interaction in
the model. Replication within duration was used as the
error term. We included soil moisture, soil penetration
resistance, and soil, vegetal, and litter cover as covariates
in the analyses of stem counts. Covariates were excluded
if they were not significant or parameter estimates
(slope or intensity of the influence) were biologically
insignificant. We used a = 0.10 for all tests.
We failed to find statistically significant
differences among treatments. Therefore, we calculated
power curves (Rotenberry and Wiens 1985) to estimate
the number of replicate picket circles (N) necessary
for us to detect differences among grazing durations
(a = 0.10, B = 0.25, effect size selected according
to variable tested) or the difference among durations
(effect size) necessary for us to detect a grazing effect
with 75% probability
(a= 0.10, N =4, B = 0.25).
Results
Four picket circle replicates were
insufficient to detect a statistically significant effect
of grazing duration on bare soil, litter, vegetation,
rock, and moss cover, and stem counts the year following
grazing. Therefore, we present treatment means and standard
errors (Table 1, Table
2, Fig. 1) and discuss patterns
of change which are consistent among years and with
known possible influences of grazing on plant communities.
Ground Cover
To detect a difference among grazing
durations of a change in cover of 10 % (e.g. treatment
A changed from 10 to 20% litter, while treatment B changed
from 10 to 30% litter), we would have needed 4 to 190
replicate circles, depending on cover type, month, and
year grazed. However, the means suggest a pattern of
grazing induced change (Table
1). Eighteen hours grazing in August 1988, and 4,
8, and 18 hours grazing in August and September 1989
reduced the percentage of vegetal cover relative to
the ungrazed circles the following year. There was a
corresponding increase in percentage of bare soil. Grazing
did not affect litter or rock cover differently from
the ungrazed controls. Changes in moss cover varied
from year to year but were not influenced by grazing.
Circles grazed repeatedly through the summer (JAS) were
not impacted more than circles grazed only once.
After 2 years of grazing (Table
2) most grazing treatments had more bare soil with
a corresponding loss of litter rather than vegetal cover.
All treatments (grazed and ungrazed) had less moss cover
in 1991 than in 1989, but slightly more vegetal cover.
Rock cover did not change over this period.
Plant Stem Numbers
Stem counts on the ungrazed circles
averaged 4 per 10 cm2 for grasses, and ranged from 3
to 7 for forbs, depending on month. To detect differences
among grazing durations in the change in stem counts,
1 grazing treatment would have to change by more than
3.5 stems than another treatment during the year. For
example, a grazing treatment would have to increase
from 4 stems per 10 cm2 1 year to 12 stems the next
year to be significantly different from a treatment
which changed from 4 to 8 stems. To reduce the detectable
difference (3.5 stems) to 1 stem would require 10 to
50 replicate circles depending on vegetation type, month,
and year grazed.
There were, however, trends in grazing
impacts on grass and forb stem counts (Fig.
1). Forb stem counts declined from 1988 to 1989
on all July circles, but the decline was less on the
grazed than on the ungrazed circles (Fig.
1a). Eighteen hours of grazing in August 1988 reduced
both grass and forb stem counts the following year (Fig.
1a).
The meadow grazed in 1989 responded
similarly. Eight and 18 hours of grazing in July increased
the number of forb stems relative to the ungrazed circles.
Forb stem counts were not affected by grazing in August
and September (Fig. 1b). Grazing
for 18 hours in July 1989, any grazing in August and
September 1989, and repeated grazing during the summer
1989 (JAS) reduced grass stern counts in 1990 when compared
to ungrazed circles (Fig. 1b).
Changes in stem counts from 1988 to 1989 and 1989 to
1990 were not influenced by soil moisture, soil penetration
resistance, or ground cover.
After 2 consecutive years of grazing,
the July 8- and 18-hour circles had less grass and more
forb sterns (Fig. 1c) than the
controls. August and September grazing had little influence
on forb numbers. However, 8 hours of grazing in August
and 8 and 18 hours of grazing in September reduced grass
stem counts (Fig. 1c). Two summers
of repeated grazing during the summer (JAS) did not
affect grass or forb stem counts (Fig.
1c).
The grazing treatments did not influence
elk grazing the following summer. Elk grazed 14.7 ±
1.1% (mean ± standard error) of the grasses and 9.9
± 1.0% of the forbs by August 1989 across all circles
grazed by horses in 1988. By July 1990, elk had grazed
7.7 ± 0.8 and 8.9 ± 0.7% of the grasses and foibs, respectively,
on circles grazed by horses in 1989. After 2 years of
horse grazing the elk grazing was uniform across all
circles. By July 1991, elk had grazed 6.9 ± 0.6% of
the forbs and 7.3 ± 0.7% of the grasses.
Discussion
While the meadow we worked on appears
to be resilient to 2 summers of moderate to heavy grazing,
our ability to detect significant grazing effects may
have been limited by too few picket circle replicates.
Yet, consistent patterns among grazing treatment means
indicate that picketed horses could cause some changes
on grazed areas depending on season and duration of
grazing.
Picketing for 8 or 18 hours in mid-
to late summer increased bare soil the following year
and decreased vegetal or litter cover. The decrease
in basal vegetal cover was reflected in reduced grass
stem counts. Grasses grazed during flowering produce
fewer tillers per unit area (Stout et al. 1980, Stout
Ct al. 1981) or have lower tiller replacement (Olson
and Richards 1988) than ungrazed plants. Forbs similarly
produce fewer inflorescence if defoliated just before
or during flowering (Blaisdell and Pechanec 1949, Mueggler
1967, Edwards 1985).
Mid-summer 1988 was extremely dry
(NOAA 1988) and the grasses were flowering in July.
Grazing at this time may have given the forbs a competitive
advantage, thus the smaller decrease in forb stem counts
from 1988 to 1989 on the grazed than the ungrazed circles
(Fig. 1a). By August the forbs
were either flowering, and thus sensitive to defoliation,
or dried and brittle, and thus broken by trampling.
The grasses were more resistant. After 18 hours of grazing
both grasses and forbs were defoliated by grazing or
trampling and produced fewer stems the following year
(Fig. 1a).
Even though grazing in July 1989
occurred during the flowering phase of most grasses
the number of grass stems were not reduced until the
grazing duration reached 18 hours (Fig.
1b). Forbs appeared to benefit from heavy grazing
(8 and 18 hours) in July (Fig. 1b),
but were not influenced by grazing in August or September.
In contrast, grazing during August and September or
repeated grazing (JAS circles) reduced the number of
grass stems per 10cm2 when compared to the ungrazed
controls during 1990.
Defoliation may reduce leaf (Mueggler
1972, 1975, Edwards 1985) or stem lengths (Mueggler
1967, 1972, Trlica et al. 1977, Stout et al. 1980, Stout
and Brooke 1987, Olson and Richards 1988) the following
growing season. However, our height class categories
were too broad (<2, 2-4, 4-12, 12-24, and >24
cm) for us to detect changes in plant heights in response
to grazing the previous year. Plant height reductions
following defoliation in other studies (Mueggler 1972,
1975, Trlica et al. 1977, Stout et al. 1980, Stout and
Brooke 1987) would not have resulted in placing the
plants in lower height classes in our study. We also
had high variability among circles, making data interpretation
difficult. Therefore, these data were not presented.
We encountered several difficulties
in the design, analysis, and interpretation of this
study. First, stem counts and plant heights vary through
a summer and among species. By counting stems and measuring
height classes across species within a vegetation type,
the data had high variance. Grazing and sampling the
number of picket circles required to adjust for such
variance is not feasible. This point is critical because
wilderness rangers often do not have the time to measure
more than a few points within a meadow of interest during
any given year. Although managers hope to set grazing
guidelines which minimize community changes, we may
need to select keystone genera or species because the
systematic categorization of all plant community types'
response to grazing may be an unrealistic goal. Use
of a keystone species would allow a more rigorous evaluation
of the effects of intensity, timing, and frequency of
defoliation on selected plant species in high elevation
plant communities. Measuring changes in ground cover
over time may require less intensive sampling but may
miss impacts on critical plant species.
Evaluation of community response
becomes difficult from a statistical perspective. Ideally,
measures on all grazed circles would be taken at the
same time, regardless of month grazed. However, high
elevation meadows change rapidly through the summer.
Some species such as those in the Liliaceae family go
through their life cycle within weeks and may not be
accounted for if the meadow is "measured" at only 1
point during the growing season. Yet, by taking pregrazing
measurements at different times during the summer, we
could not statistically compare effects of grazing in
different months.
The cumulative impact of several
years' grazing depends on plant phenology at the time
of grazing. Our circles were grazed according to calendar
dates rather than plant phenology because of logistical
constraints. Plant phenology on a given date differs
from year to year. Therefore, we did not get a strong
cumulative effect after just 2 years grazing. From an
ecological perspective, grazing by calendar dates makes
interpretation of plant response difficult. However,
wildland area managers have limited resources and may
be unable to manage packstock grazing according to plant
phenology. Therefore, from a manager's perspective,
studying community response based on calendar dates
may be most appropriate. Either way, the plant communities
must be monitored over more than 2 years to assure objectivity
when determining packstock grazing effects.
Finally, some plant communities have
evolved with disturbance, for example by heavy elk grazing
or burrowing animals. These communities may show little
change in response to the added disturbance of 2 or
3 years of packstock grazing. The meadow we worked on
may be such a type and many years of packstock use may
be necessary before demonstrable changes occur.
Conclusion
We had insufficient replication (N
=4) per grazing treatment to detect statistically significant
changes in stems counts, ground cover, and plant heights
after 1 and 2 years of grazing. The required sample
sizes, as well as other difficulties encountered in
a study are probably indicative of the monitoring limitations
wildland managers face. The data do, however, suggest
that a single period of heavy grazing (18 hours per
picket circle) or moderate (8 hours) repeated grazing
through a summer can reduce vegetal and litter cover,
increase bare soil cover, and reduce grass stem counts.
These changes could be the precursors to a shift in
plant community composition.
This project was funded by the Intermountain
Research Station, Forest Service, USDA, and the Montana
Agricultural Experiment Station. We appreciate the advice
from David Cole (USFS, Missoula, Mont.) and the help
of all the field workers. We thank John Lacey, Matt
Lavin, Jim Pflster, and 2 Journal of Range Management
reviewers for their comments.
Manuscript accepted 20 Dec. 1995.
* Authors
are former research associate, Animal and Range Sciences
Department, Montana State University, Bozeman, Mont.,
59717; associate professor, Animal and Range Sciences
Department, Montana State University, Bozeman, Mont.,
59717; associate professor, Earth Sciences Department,
Montana State University, Bozeman, Mont., 59717; associate
professor, Animal and Range Sciences Department, Montana
State University, Bozeman, Mont., 59717; and assistant
professor, Math and Computer Sciences Department, Montana
Tech, Butte, Mont., 59701.
