Integrating precipitation, grazing, past effects and interactions in
long-term vegetation change
Morris, C., Badik, K. J., Morris, L. R., & Weltz, M. A. (2016). Integrating
precipitation, grazing, past effects and interactions in long-term vegetation
change. Journal of Arid Environments, 124, 111-117.
doi:10.1016/j.jaridenv.2015.08.005
10.1016/j.jaridenv.2015.08.005
Elsevier
Version of Record
http://cdss.library.oregonstate.edu/sa-termsofuse
Journal of Arid Environments 124 (2016) 111e117
Contents lists available at ScienceDirect
Journal of Arid Environments
journal homepage: www.elsevier.com/locate/jaridenv
Integrating precipitation, grazing, past effects and interactions in longterm vegetation change
Christo Morris a, *, Kevin J. Badik b, Lesley R. Morris c, 1, Mark A. Weltz a
a
USDA-ARS, Great Basin Rangelands Research Unit, 920 Valley Rd., Reno, NV, 89512, USA
University of Nevada, Ecology, Evolution, and Conservation Biology Program, 1664 N. Virginia St., Reno, NV, 89557, USA
c
USDA-ARS, Forage and Range Research Lab, 696 North 1100 East, Logan, UT, 84322, USA
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 25 September 2014
Received in revised form
14 July 2015
Accepted 3 August 2015
Available online 15 August 2015
Determining the causes of vegetation change in arid and semi-arid environments can be difficult and
may involve multiple factors, including disturbance, inter-annual climatic variation, soils, effects from
years past and interactions between these factors. Theoretical models describing vegetation change in
these systems have generally focused on a single aspect as the primary driver. The integration of these
factors into a single model may be what is required to fully understand the drivers of vegetation change
in desert systems. To test the contributions of these various factors, we analyzed a long-term (1979
e2011) vegetation dataset using multiple linear regression.
While precipitation and livestock density were important variables for explaining vegetation change,
the consistency with which past effects and interactions significantly improved the models underscores
their importance. Past effects were included in every model except for shrub diversity, and included both
precipitation and livestock density effects. A novel approach to addressing the interaction between
grazing and precipitation was included by dividing precipitation by stocking density. Grass density had a
high positive correlation with this metric, while shrub cover had a small negative correlation. These
results support the integration of multiple factors to explain vegetation change.
Published by Elsevier Ltd.
Keywords:
Artemisia tridentata ssp. vaseyana
Bodie Hills
Disturbance
Succession
Transition
1. Introduction
Vegetation change in arid and semi-arid ecosystems can be can
difficult to understand and can involve complex combinations of,
and interactions between, inter-annual climatic variation, soils,
disturbance, and effects from previous years. Theoretical models of
vegetation change on rangelands have evolved over time from
linear (Clements, 1916; Dyksterhuis, 1949) to multi-equilibrial
state-and-transition models (Westoby et al., 1989; Laycock, 1991;
Briske et al., 2005). Linear succession describes a smooth transition, driven by internal ecosystem processes, such as climate, while
the state-and-transition models illustrate discreet states driven by
external ecosystem process such as disturbance (Briske et al.,
2003). Successional models adapted for rangelands were based
* Corresponding author. Present address: Agricultural Sciences and Natural Resources Program, Oregon State University, One University Blvd., La Grande, OR
97850, USA.
E-mail address: Christo.Morris@Oregonstate.edu (C. Morris).
1
Present address: Agricultural Sciences and Natural Resources Program, Oregon
State University, One University Blvd., La Grande, OR 97850, USA.
http://dx.doi.org/10.1016/j.jaridenv.2015.08.005
0140-1963/Published by Elsevier Ltd.
on grazing as the driving factor (Briske et al., 2005) and drought
was assumed to be additive (Walker, 1993). In other words, the
effects of drought could be offset by a reduction in grazing. Additionally, both factors were assumed to temporarily arrest secondary
succession (Briske et al., 2003). This approach was found to be
inadequate to describe the multitude of states that can occur on a
given site, which lead to the development of state-and-transition
models (Allen-Diaz and Bartolome, 1998; Bestelmeyer et al.,
2003; Briske et al., 2003); however, even as the state-andtransition model was being proposed, it was suggested that a
more accurate description may lie somewhere between the extremes of these two models (Westoby et al., 1989). For instance,
state-and-transition models focus on disturbance as drivers of
change ignoring the contribution of climate and other internal
ecosystem factors (Briske et al., 2003). However, inter-annual climatic variation is often considered a controlling factor in vegetation
change, even more so than management (Fynn and O'Connor,
2000; West, 2003b; Mashiri et al., 2008). Rare or extreme climatic events can drive change in semi-arid systems (West et al.,
1979; Walker, 1993; Holmgren and Scheffer, 2001) or in some
cases the convergence of multiple rare events are responsible
112
C. Morris et al. / Journal of Arid Environments 124 (2016) 111e117
(Wiegand and Milton, 1996).
It has also been suggested that the two competing models may
address different temporal scales of change and that the integration
of the two is what is necessary to formulate an accurate explanation
of rangeland vegetation dynamics (Fuhlendorf et al., 2001; Briske
et al., 2003). Change may occur episodically and at varying rates
depending on interactions between disturbance, vegetation characteristics and inter-annual climatic variation spanning multiple
years (Iglesias and Kothmann, 1997; Fuhlendorf et al., 2001; Curtin,
2002; Perlinski et al., 2014). Multiple studies have shown that the
interaction between inter-annual climatic variation and disturbance, or between multiple disturbances, are required to understand vegetation change (Fuhlendorf and Smeins, 1997; West and
Yorks, 2002). Previous research has suggested that what is
missing to make the conceptual models predictive and applicable
to management is the integration of the highly stochastic and
variable climate regime that exists in semi-arid regions (Walker,
1993; Wiegand and Milton, 1996; Bestelmeyer et al., 2004;
Hardegree and Van Vactor, 2004; Briske et al., 2005; McClaran
and Wei, 2014). Another factor that may be important are lag effects (Walker, 1993; Wiegand and Milton, 1996). Lags at short time
scales, such as the effect of the previous year's precipitation, have
been shown to affect plant diversity (Adler and Levine, 2007), while
cover for shrubs and grasses were shown to be correlated with
precipitation from two to four years previous (Anderson and
Inouye, 2001) and productivity of desert grasses can take 2e4
years to recover to expected levels after drought (Moran et al.,
2014).
Most ecological studies last less than six years (Tilman, 1989);
yet, on arid and semi-arid rangelands it may take as long as 20e25
years for a site to receive a representative range of precipitation
(West, 2003a). Therefore, long-term data are necessary to detect
phenomena that would otherwise be beyond the scope of human
observation, such as slow, sporadic or rare events (Hobbie et al.,
2003). The use of permanent plots is considered a valuable
method for this type of temporal analysis (Stohlgren, 2007), though
they are a rare resource. Long-term studies which include disturbances are considered key to understanding rangeland systems
(Allen-Diaz and Bartolome, 1998). The objective of this study is to
investigate the role of both precipitation and grazing, along with
their interactions and past effects, with changes in vegetation
occurring over a 30-year span of data collected from permanent
plots.
precipitation for the water year (Oct.eSep.) has declined by
approximately 44%, while winter (Dec.eMar.) precipitation has
declined by 33% and spring precipitation (Apr.eJun.) has declined
by 24% (Morris et al., 2014).
The majority of the Bodie Hills area is managed by the USDI
Bureau of Land Management (BLM) and is organized into four
grazing allotments (Aurora Canyon, Bodie Mountain, Mt. Biedeman
and Potato Peak) totaling approximately 38,000 ha. Since 1962
California Department of Parks and Recreation has managed
approximately 400 ha around the historical town site of Bodie as
Bodie State Historic Park. This site was a major mining area from
1875 until the mid-1880's, producing $70 million in gold and silver
over the years and hosting a peak population of between
7000e12,000 people in 1880 (Sprague, 2003). Mining activity
continued intermittently until World War II, at which point the
town was deserted.
Despite the pattern of resource over-exploitation associated
with historical mining operations (Young and Budy, 1979; Sprague,
2003), survey records from 1942 describe range conditions in the
Bodie Hills area as generally good and “underutilized” (Bell, 1943).
BLM records from the late 1950's indicate that the number of sheep
animal unit months (AUM) in the Bodie Hills area ranged from
16,000e18,000 (BLM, 1958, 1959, 1960) plus an additional 500e700
cattle AUMs. In 1964 the number of sheep AUMs on the Aurora
Canyon, Mt. Biedeman and Potato Peak allotments alone was
around 7500, plus an additional 1200 cattle AUMs (BLM, 1964).
During the 1960's and 1970's the total number of AUMs decreased
and many of the allotments were converted from sheep to cattle
grazing. Over the last 30 years, the BLM has continued to reduce
stocking rates for both sheep and cattle on all four allotments in the
Bodie Hills (Fig. 1; BLM, 2011, 2012). During this time period, sheep
have generally been herded within each allotment, while cattle
have been allowed to graze freely, from June through October.
Adjustments in AUMs were periodically made by the BLM to account for dry years, but the correlation between AUMs and precipitation was low (R2 ¼ 0.04) for current water year and total
stocking rate [data not shown]). In addition to domestic livestock,
wild grazers and browsers reside in the Bodie Hills (BLM, 2008),
2. Methods
2.1. Site description
The Bodie Hills are situated on the CaliforniaeNevada border
southeast of Bridgeport, CA and north of Mono Lake. Elevations
range from 2100 m to 3100 m. Parent materials originate from the
Pliocene and are primarily of volcanic origin, including andesite,
dacite, rhyolite and welded tuff (O'Neil et al., 1973). Soils consist of
sandy, ashy, and gravelly loams with clays and cobbles at depth.
Depth to a restrictive layer is generally greater than 200 cm, though
in some cases shallow soils can be discerned based on shifts in
vegetation type and overall productivity.
Mean temperature for the period from 1965 to 2011 was 3.18 C,
measured at an elevation of 2550 m (WRCC, 2010). Over this time
period, mean maximum annual temperature has increased by
about 2.1 C and mean minimum annual temperature has increased
by about 0.39 C (Morris et al., 2014). Average annual precipitation
was 323 mm for the period 1965e2011, also measured at an
elevation of 2550 m. The majority of precipitation falls as snow
during winter months. Over the past 45 years average total
Fig. 1. Changes in livestock numbers and precipitation from 1979 to 2011 in the Bodie
Hills, CA. The dashed line in the lower panel shows the trend line for precipitation over
this time frame.
C. Morris et al. / Journal of Arid Environments 124 (2016) 111e117
including mule deer (Odocoileus hemionus), pronghorn antelope
(Antilocapra americana), feral horses (Equus caballus), greater sagegrouse (Centrocercus urophasianus), sagebrush vole (Lemmiscus
curtatus), pygmy rabbit (Brachylagus idahoensis) and American pika
(Ochotona princeps).
Current vegetation consists primarily of montane sagebrushsteppe (Artemisia tridentata ssp. vaseyana (Ryd.) Beetle) communities, punctuated by aspen (Populus tremuloides Michx.) stands at
higher elevations, and mountain mahogany (Cercocarpus ledifolius
Nutt.), tobacco brush (Ceanothus velutinus Douglas ex. Hook), wet
m,
meadows and pinyon-juniper (Pinus monophylla Torr. & Fre
Juniperus osteosperma (Torr.) Little) woodlands at lower elevations.
Herbaceous species include Thurber's needlegrass (Achnatherum
thurberianum (Piper) Barkworth), needle-and-thread (Hesperostipa
comata (Trin. & Rupr.) Barkworth), Sandberg bluegrass (Poa secunda
J. Presl.), bottlebrush squirreltail (Elymus elymoides (Raf.) Swezey),
Lo
€ ve), prairie
basin wildrye (Leymus cinereus (Scribn. & Merr.) A.
Junegrass (Koehleria macrantha (Ledeb.) Schult.), Astragalus
(Astragalus sp.), granite prickly phlox (Linanthus pungens (Torr.) J.M.
Porter & L.A. Johnson), phlox (Phlox sp.), Anderson's lupine (Lupinus
andersonii S. Watson), Great Basin lupine (L. argenteus Pursh) and
the locally endemic Bodie Hills cusickiella (Cusickiella quadracostata
(Rollins) Rollins) (Provencher et al., 2009; Low et al., 2010). Vegetation at the time data collection began was characterized by low
shrub, grass and forb cover (Morris et al., 2014). Over the course of
the dataset, foliar cover of shrubs across all four allotments
increased from 3% to 20%, while basal cover of grasses increased
from 3% to 8% (Fig. 2). Mean shrub density showed a slight declining
trend for one allotment, but doubled from 2.5 to 5 shrubs m2 for
the other three, though the results were nonsignificant. Grass
density fluctuated, but remained overall unchanged for the Bodie
Mt. allotment, but increased significantly in the other allotments
113
from 15 to 100 plants m2. Shrub diversity across all four allotments
trended upward, while grass diversity remained virtually
unchanged.
2.2. Vegetation sampling and data manipulation
Twenty-one permanent plots of two sizes (either 2.25 m2 or
0.83 m2) were established in either 1969 or 1973 and measurements for cover and density along with plant position were recorded by BLM employees. Cover values were based on ocular
estimates and were aided by dividing each plot into nine subsections. Each subsection was overlaid by a grid with
7.5 cm 7.5 cm cells to aid in accurate cover estimates (Habich,
1992). Foliar cover was estimated for shrubs and forbs, while
basal cover was estimated for grasses. Within each plot density
values were collected by counting all individual shrubs, mature
(reproductive) perennial grasses, perennial grass seedlings and
perennial forbs. Annual species were not included in density calculations. For analysis, density data from the smaller plots were
scaled proportionally to match the area of the larger plots. Measurements were taken by BLM employees once for a given sample
year during the months of July, August, September or October and
were repeated at irregular intervals ranging from 1 to 13 years until
2011. We calculated diversity using Shannon's Diversity Index (SDI)
based on cover values of each species within a plot (Shannon, 1948),
which accounts for both species number and evenness of distribution. The importance of adequate training to reduce observer
error in long-term datasets is known (Stohlgren, 2007). The issue of
observer error was addressed in this dataset by using welldocumented methods, by breaking the sample plots into smaller
units and by using the same datasheets over the course of data
collection. We excluded plots that fell within areas dominated by
Artemisia arbuscula, because of differences in overall productivity
and potential differences in responses to disturbance and climate.
The remaining 18 plots spanned all four allotments and were all on
sites dominated by A. tridentata ssp. vaseyana and categorized as a
montane sagebrush steppe plant community (Provencher et al.,
2009; Low et al., 2010).
2.3. Statistical analysis
Fig. 2. Mean vegetation cover for Bodie Mountain allotment. Error bars represent one
standard error. P-values are displayed for datasets that were analyzed with ANOVA and
were significant at a ¼ 0.05. Means with letters in common are not significantly
different at a ¼ 0.05. Cover values are based on foliar measurements for all groups
except for grasses, where it is based on basal area. For non-normal data, data that had a
non-significant model fit or data that had no significant differences between means, a
trend line is displayed along with the slope of the line (adapted from Morris et al.,
2014).
In order to understand what factors were associated with
changes in vegetation cover, density and diversity, we ran stepwise
multiple regressions for shrub, grass, perennial forb and annual
forb cover, density and diversity using JMP Version 9.0 (SAS Institute Inc., Cary, NC, 1989e2012). Data were transformed to meet the
assumptions of normal data distribution using Box-Cox transformations. Data that could not be successfully transformed were
not analyzed. Model selection was based on the lowest value of
Akaike Information Criterion (AIC) (Akaike, 1987), which balances
the goodness of fit for the model with the number of parameters.
Model terms were selected from a variety of independent variables
as well as modifications to these variables (Table 1). It should be
noted that we utilized the stepwise approach in order to explore a
broader suite of variables than are traditionally examined in
modeling vegetation change in semi-arid environments. While the
magnitude and direction of variables in a stepwise regression is
dependent on the other variables in the models, this approach
provides a method to consider these less-common factors.
Interactions between precipitation and grazing were evaluated
by crossing precipitation and stocking density for current water
year (WY), previous WY, and two-, three- and five-year averages
within the model. An inverted interaction was calculated by
dividing the various precipitation terms by their respective stocking density term. This calculation allowed for a more meaningful
114
C. Morris et al. / Journal of Arid Environments 124 (2016) 111e117
Table 1
Independent variable terms plus modifications selected from for inclusion in stepwise multiple regression models.
Independent variables
Water year precipitation (WY)
Winter precipitation
Spring precipitation
Cattle density
Sheep density
Total livestock density
Sand content
Clay content
Plot
Sampling day
Modifications
Previous year(s)a
Averages
1
1
1
1, 2 & 3
1, 2 & 3
1, 2 & 3
2,
2,
2,
2,
2,
2,
3
3
3
3
3
3
&
&
&
&
&
&
5
5
5
5
5
5
year
year
year
year
year
year
Interactionsb
Standard, Inverted
Standard, Inverted
Standard, Inverted
a
Refers to lag effects.
Interactions were between independent variables, their modifications and the corresponding cattle, sheep and total livestock density (e.g. 3-year average of WY
precipitation and the 3-year average of sheep density).
b
interpretation of extreme events than a multiplicative interaction
since it highlights the periods when precipitation and grazing were
not synchronized (e.g. precipitation was low, but grazing was high
or vice versa). This is not a standard interaction, but it fits the
general definition of an interaction, whereby the effect of one independent variable depends on the value of a second independent
variable (Sokal and Rohlf, 1981). For years when stocking rates were
zero for either class of livestock, a proxy value of 0.0001 was used to
allow for the calculation. This value was at least two orders of
magnitude lower than the lowest stocking density within the
dataset. Stocking density was based on the number of AUMs
calculated for each allotment for each year divided by the area of
the allotment. Grazing data was not available prior to 1979 for
Bodie Mountain allotment and before 1976 for the other allotments,
so those years (including the years required to calculate averages)
were excluded from analysis. This resulted in n ¼ 129 for all models.
Since climate data were only collected at one location in the
Bodie Hills area, modeled precipitation data at the 4 km scale were
used to establish unique values for each plot based on elevation and
location (PRISM, 2012). Year and stocking rate were highly correlated (R2 ¼ 0.55), so whichever one was the most biologically
relevant for each vegetation growth form was selected for inclusion
in the model (Graham, 2003). Shrubs are utilized less by livestock in
the summer and are slower growing, so year was considered the
more relevant factor. Herbaceous species are more impacted by
livestock utilization and grow much more quickly, so stocking
density was considered the more relevant independent variable.
Due to the difference in factors affecting woody and herbaceous
growth forms, data could not be combined to analyze for total
vegetation cover, density or diversity.
3. Results
For cover data, the top model for shrubs had an adjusted R2 of
0.70 and grasses had an adjusted R2 of 0.49 (Table 2). Cover data for
both perennial and annual forbs were unable to be fitted to a statistical model and were excluded from analysis. For shrubs, the
effect of plot had the highest coefficient, followed by year, which
had a high positive relationship with shrub cover. The two-year
average of water year precipitation divided by total livestock density was negatively associated with shrub cover, while there was a
slight positive relationship with the two-year average of spring
precipitation. Grass cover had a positive relationship with current
year winter precipitation and a negative coefficient with current
year sheep stocking densities. The effect of cattle stocking densities
was not as strongly correlated as sheep and was equal to the plot
effect. Grass cover was positively associated with soil sand content
and negatively associated with winter precipitation from two years
previous. Lastly, there were small positive correlations with overall
stocking density from two years previous, current year spring
precipitation and soil clay content.
Mature grass was the only density data that could be fitted to a
model and had an adjusted R2 value of 0.37. The term with the
highest effect was plot, followed by the 2-year average of winter
precipitation divided by sheep density. Percent soil sand content
and the 3-year average of winter precipitation both had similar
small effects, though sand was positively correlated while winter
precip was negative. For shrub diversity the model had an R2 value
of 0.55 with plot having the largest effect, followed by a positive
correlation with year. For grass diversity the model had an R2 value
of 0.40, with plot again having the largest effect. There were small
positive correlations with total livestock density and the 3-year
average of spring precipitation.
4. Discussion
This analysis supports the integration of disturbance, precipitation, past effects and interactions into understanding vegetation
change over time. This finding is consistent with studies from drier
climates (Fuhlendorf and Smeins, 1997; Fuhlendorf et al., 2001; Yao
et al., 2006; Mashiri et al., 2008; Butterfield et al., 2010; Moran
et al., 2014). Plot, grazing and precipitation had the largest coefficients, but the consistency with which past effects appear
highlights their importance. Some form of past effect was significant in every model except for shrub diversity. The interaction
between grazing and precipitation was significant in two models in
the form of the inverted interaction. This suggests that this metric
may have utility in future applications.
Plot was a significant factor in every model and consistently had
a high coefficient. Although the use of permanent plots in multiple
regression violates the assumption of independence between
samples, we used it to account for the heterogeneous nature of
shrublands, the long lifespans of desert plants (both shrubs and
herbaceous plants) and the rareness of recruitment events (West,
1979; West et al., 1979). As such, the significance of plot in the
models indicates that plant cover and density are likely to be
correlated with previous sampling events. For example, plots with
grasses or shrubs already present were very likely to have the same
plants present later in the dataset, while plots with high cover of
bare ground were likely to continue to have high cover of bare
ground.
4.1. Shrubs
The large effect of plot on shrubs is related to the spatial heterogeneity of shrub-dominated systems and the long-lived nature
C. Morris et al. / Journal of Arid Environments 124 (2016) 111e117
115
Table 2
Results from stepwise multiple regression. Coefficients for terms are standardized for comparison within each whole model based on a mean of zero and variance of one.
Dependent variable
Growth form
Term
Coefficient
P-value
Whole model Adj R2
Cover
Shrubs
Plot
Year
2-year avg. of WYa precip./total livestock density
2-year avg. of spring precip.
Winter precip.
Sheep density
Cattle density
Plot
% sand
2 years winter precip.
2 years total livestock density
Spring precip.
% clay
Plot
2-year avg. of winter precip./sheep density
% sand
3-year avg. of WY precip.
Plot
Year
Plot
Total livestock density
3-year avg. of spring precip.
0.68
0.54
0.20
0.11
0.55
0.51
0.34
0.34
0.31
0.30
0.19
0.12
0.11
0.54
0.46
0.22
0.20
0.69
0.23
0.58
0.19
0.13
<0.0001
<0.0001
0.008
0.0345
<0.0001
<0.0001
0.0023
<0.0001
0.132
0.0028
0.0168
0.1618
0.111
<0.0001
<0.0001
0.0056
0.0063
<0.0001
0.0001
<0.0001
0.0084
0.0576
0.70
Grass
Density
Grass
Diversity
Shrub
Grass
a
0.49
0.37
0.55
0.40
WY ¼ water year.
of desert shrubs (West et al., 1979). A plot that was initially located
in an interspace between shrubs is much less likely to have high
shrub cover, even 30 years later; whereas a plot that initially contained shrubs would be much more likely to have high shrub cover
later on. The positive correlation with time is not surprising,
especially since shrub cover was so low at the beginning of the
dataset. The negative effect of the interaction between the two-year
average of water year (WY) precipitation divided by total livestock
density is more difficult to interpret. Since the sign of the coefficient
is negative, that means that the highest shrub cover is associated
with years having low precipitation and high stocking densities,
while low shrub cover is associated with high precipitation and low
stocking densities. It may seem counterintuitive that shrub cover
would be high at times of low precipitation, unless we consider that
the combination of low precipitation and high stocking densities
would also result in the lowest cover of grasses. We see evidence of
this in the positive association of grass cover with winter precipitation and the negative association with both classes of livestock,
and the positive association of grass density with the two-year
average of winter precipitation divided by sheep density. The
negative relationship between shrubs and grasses can best be
explained by competition for soil and water resources (Caldwell
et al., 1987).
The correlation between shrub cover and spring precipitation
was unexpected, considering that spring precipitation is associated
more with shallow soil moisture and shrubs are generally associated with deeper soil moisture extraction (Ogle and Reynolds,
2004). While A. tridentata is known to have both shallow and
deep roots and can extract soil moisture from up to 2.5 m in depth
(Campbell and Harris, 1977), shifting precipitation from winter to
spring, while maintaining overall quantities did not increase
growth for A. tridentata (Bates et al., 2006), illustrating that spring
precipitation doesn't confer any particular advantage. However,
young shrubs are known to experience more water stress during
the summer season than adult shrubs, including A. tridentata
(Campbell and Harris, 1977), due to a less-developed root system.
This would make young shrubs more dependent on and responsive
to spring and summer precipitation for growth and survival. So, the
growth of shrubs from spring precipitation may be the contribution
of young shrubs responding to decreased or prolonged spring
drought. There is less research available on water relations for
Purshia tridentata, though evidence suggests a similar pattern. In a
snow depth manipulation experiment, increased snow with subsequent higher soil moisture levels lasting slightly longer into the
spring did not increase cover of P. tridentata, even after 50 years
(Loik et al., 2013). P. tridentata is thought to have few shallow roots
and responded more slowly to summer precipitation events than
A. tridentata (Loik, 2007). It also relies primarily on deep soil
moisture from winter snowfall, leaving seedlings susceptible to
drought until their roots extend to greater depths.
4.2. Grasses
The stronger effect of winter precipitation on grasses as opposed
to spring precipitation was surprising, since herbaceous plants have
shallower root systems capable of rapidly capitalizing on spring and
summer precipitation before evaporation occurs (Schenk and
Jackson, 2002) and winter precipitation is associated with deeper
soil moisture (Ogle and Reynolds, 2004). However, it is also noted
that seasonality of precipitation also effects rooting depth in
addition to mean annual precipitation, which may be important for
a site like Bodie, where the majority of precipitation falls as snow.
Of herbaceous plants, grasses are one of three families that root
significantly deeper than the global average (Schenk and Jackson,
2002). Whether this is deep enough to tap winter precipitation
reserves in the soil, or whether the grasses in the Bodie Hills are
rooting deeper than average would require additional research.
The negative correlation between grass cover and both sheep
and cattle grazing was expected. Herbivory directly reduces plant
mass and photosynthetic capacity and can indirectly reduce growth
through altered competitive interactions (Caldwell et al., 1987;
Briske and Richards, 1994). The more negative association of
sheep over cattle, may be partially attributed to the negative effects
of trampling by sheep (Laycock et al., 1972). The effect of plot was
lower for grasses than for shrubs, but still substantial and related to
the long-lived nature of mature desert grasses (West, 1979; West
et al., 1979). Plots that had grasses established initially may have
had the same individuals present decades later. Since the conditions for establishment can be rare and episodic (Holmgren and
Scheffer, 2001), plots that already had existing grass cover would
116
C. Morris et al. / Journal of Arid Environments 124 (2016) 111e117
have been the most likely to increase grass cover. The positive
correlation with both sand and clay content of the soil may be a
statistical anomoly since the soils fell mostly into the sandy loam
and sandy clay loam categories and didn't span the full range of
textural classes. The two remaining factors for grass cover both
relate to past effects and are difficult to explain. The negative
relationship with winter precipitation two years previous may be
related to competition with other growth forms, but why the effect
would be delayed is unknown. A study using long-term data in a
Chihuahuan desert grassland found a negative correlation with
grass cover and current-year precipitation and also speculated that
the effect was mediated through competition for soil moisture (Yao
et al., 2006). The positive relationship between grass cover and total
livestock density from two years previous could be characterized as
a compensatory response to grazing (Ferraro and Oesterheld,
2002), but this effect has only been documented over a period of
months (Oesterheld and McNaughton, 1991). It is not known if the
effect would span multiple years.
The effect of plot on mature perennial grass density is the same
as for grass cover; plots that initially had grasses established were
more likely to have those same individuals in subsequent samplings. The positive association with the 2-year average winter of
precipitation divided by the 2-year average of sheep density
means that for years when winter precipitation was low and sheep
density was high, grass density was low; conversely, when winter
precipitation was high and sheep density was low, grass density
was high. This mirrors the positive correlation with winter precipitation and negative correlation with sheep density for grass
cover, except that the effect seems to be amplified with respect to
density to the point that it emerges as an interaction. This effect
may be due to the cover data being based on basal area, rather than
foliar cover, which is much less variable and less affected by
grazing in the short term than foliar cover (Elzinga et al., 1998). It is
unclear why mature perennial grass density, which would usually
be less variable than even basal cover (Elzinga et al., 1998), is
exhibiting the interaction. Regardless, these results, along with the
negative correlation between shrub cover and this interaction,
highlight the importance of adjusting stocking rates based on
received precipitation in order to preserve grass and reduce the
tendency for grazing to increase shrub cover on late-seral phases
due to decreased competition from grasses (Evers et al., 2013). The
positive correlation with soil sand content suggests that the higher
infiltration rates of sandier soils confer an advantage for grasses,
but again, since not very many texture classes were represented,
the effect may be limited within a small range of textures. The
negative correlation of grass density with the three-year average of
WY precipitation again may be related to shrub competition but is
unclear in this analysis.
Two factors that weren't included in this analysis that have been
found to be important in other analyses are slope and topographic
position (Fynn and O'Connor, 2000; Alados et al., 2004). The diffuse
nature of the reductions in grazing over time also made it impossible to test for longer-term lag effects related to demographic
processes, which have been found to last from 15 to 25 years after
the complete cessation of grazing in semiarid systems (Anderson
and Holte, 1981; Somodi et al., 2004). In addition, the use of plots
as the experimental unit instead of grazing management unit
presents a case of pseudo-replication (Hurlbert, 1984). Access to
permanent plots in a greater number of grazing units, as in other
studies (Mashiri et al., 2008), would have been preferred but was
unavailable. Given the large allotment sizes (2000e22,000 ha) and
dispersion of our plots we feel enough heterogeneity in soil type,
elevation (2308e2810 m) and precipitation (325e485 mm year1)
was captured to interpret our results despite probable statistical
dependence in both time and space of the plots.
5. Conclusion
While precipitation and livestock density were important variables for explaining vegetation change, the consistency with which
past effects, such as lag effects and averages, and interactions
significantly improved the models highlights the need for researchers and managers to examine variables beyond those traditionally associated with vegetation change. In some cases the lag
effects resulted in a negative correlation with precipitation, which
we speculated was related to competitive interactions. Dividing
precipitation by stocking density, as opposed to a traditional
interaction, was the only significant interaction. This is a novel
approach to investigating the relationship between these two factors, but provides a more meaningful interpretation than the
traditional approach. It also suggests more work is needed to understand the relationship between stocking rates, precipitation and
vegetation dynamics.
Acknowledgments
We would like to thank all the BLM employees who helped
collect and archive the data used in this analysis as well as the
employees from the National Archives Administration who helped
recover information on past stocking rates. We would also like to
thank two anonymous reviewers who offered insightful comments
on an early version of this manuscript.
References
Adler, P.B., Levine, J.M., 2007. Contrasting relationships between precipitation and
species richness in space and time. Oikos 116, 221e232.
Akaike, H., 1987. Factor analysis and AIC. Psychometrika 52, 317e332.
Alados, C.L., Pueyo, Y., Barrantes, O., Escos, J., Giner, L., Robles, A.B., 2004. Variations
in landscape patterns and vegetation cover between 1957 and 1994 in a
semiarid Mediterranean ecosystem. Landsc. Ecol. 19, 543e559.
Allen-Diaz, B., Bartolome, J.W., 1998. Sagebrush-grass vegetation dynamics:
comparing classical and state-transition models. Ecol. Appl. 8, 795e804.
Anderson, J.E., Holte, K.E., 1981. Vegetation development over 25 years without
grazing on sagebrush-dominated rangeland in southeastern Utah. J. Range
Manag. 34, 25e29.
Anderson, J.E., Inouye, R.S., 2001. Landscape-scale changes in plant species abundance and biodiversity of a sagebrush steppe over 45 years. Ecol. Monogr. 71,
531e556.
Bates, J.D., Svejcar, T.J., Miller, R.F., Angell, R.A., 2006. The effects of precipitation
timing on sagebrush steppe vegetation. J. Arid Environ. 64, 670e697.
Bell, W.R., 1943. Range Survey Report and Range Management Plans of the Bodie
Hill-Bridgeport-Conway-Mono Basin Grazing Units, Virginia City Grazing District, Region 3. United States Grazing Service, Reno, NV.
Bestelmeyer, B.T., Brown, J.R., Havstad, K.M., Alexander, R., Chavez, G., Herrick, J.,
2003. Development and use of state-and-transition models for rangelands.
J. Range Manag. 56, 114e126.
Bestelmeyer, B.T., Herrick, J.E., Brown, J.R., Trujillo, D.A., Havstad, K.M., 2004. Land
management in the American Southwest: a state-and-transition approach to
ecosystem complexity. Environ. Manag. 34, 38e51.
BLM., 1958. Unit License Report. Bureau of Land Management. Bishop Field Office,
CA.
BLM., 1959. Unit License Report. Bureau of Land Management. Bishop Field Office,
CA.
BLM., 1960. Unit License Report. Bureau of Land Management. Bishop Field Office,
CA.
BLM., 1964. License File. Bureau of Land Management. Bishop Field Office, CA.
BLM., 2008. Environmental Assessment: Livestock Grazing Authorization. Bureau of
Land Management, Bishop Field Office, CA.
BLM., 2011. Aurora Canyon, Bodie Mtn., Mt. Biedemann and Potato Peak Allotment
Files. Accessed 9/11. Bureau of Land Management, Bishop Field Office, CA.
BLM., 2012. Record Group 49, Bureau of Land Management; Bakersfield CA, District
Land Office. Grazing Permits Box # 049-98-0001. Accessed 2/12. National Archives and Records Administration, Riverside, CA.
Briske, D.D., Fuhlendorf, S.D., Smeins, F.E., 2003. Vegetation dynamics on rangelands: a critique of the current paradigms. J. Appl. Ecol. 40, 601e614.
Briske, D.D., Fuhlendorf, S.D., Smeins, F.E., 2005. State-and-transition models,
thresholds, and rangeland health: a synthesis of ecological concepts and perspectives. Rangel. Ecol. Manag. 58, 1e10.
Briske, D.D., Richards, J.H., 1994. Physiological response of individual plants to
grazing: current status and ecological significance. In: Vavra, M., Laycock, W.A.,
Pieper, R.D. (Eds.), Ecological Implications of Herbivory in the West. Society for
C. Morris et al. / Journal of Arid Environments 124 (2016) 111e117
Range Management, Denver, CO.
Butterfield, B.J., Betancourt, J.L., Turner, R.M., Briggs, J.M., 2010. Facilitation drives 65
years of vegetation change in the Sonoran Desert. Ecology 91, 1132e1139.
Caldwell, M.M., Richards, J.H., Manwaring, J.H., Eissenstat, D.M., 1987. Rapid shifts in
phosphate acquisition show direct competition between neighbouring plants.
Nature 327, 615e616.
Campbell, G.S., Harris, G.A., 1977. Water relations and water use patterns for Artemisia tridentata Nutt. in wet and dry years. Ecology 58, 652e659.
Clements, F.E., 1916. Plant Succession: an Analysis of the Development of Vegetation. Carnegie Institute of Washington, Washington, DC.
Curtin, C.G., 2002. Livestock grazing, rest, and restoration in arid landscapes. Conserv. Biol. 16, 840e842.
Dyksterhuis, E.J., 1949. Condition and management of rangeland based on quantitative ecology. J. Range Manag. 2, 104e105.
Elzinga, C.L., Salzer, D.W., Willoughby, J.W., 1998. Measuring and Monitoring Plant
Populations. Technical Reference 1730-1. U.S. Bureau of Land Management,
Denver, CO.
Evers, L.B., Miller, R.F., Doescher, P.S., Hemstrom, M., Nielson, R.P., 2013. Simulating
current successional trajectories in sagebrush ecosystems with multiple disturbances using a state-and-transition modeling framework. Rangel. Ecol.
Manag. 66, 313e329.
Ferraro, D.O., Oesterheld, M., 2002. Effect of defoliation on grass growth: a quantitative review. Oikos 98, 125e133.
Fuhlendorf, S.D., Briske, D.D., Smeins, F.E., 2001. Herbaceous vegetation change in
variable rangeland environments: the relative contribution of grazing and climatic variability. Appl. Veg. Sci. 4, 177e188.
Fuhlendorf, S.D., Smeins, F.E., 1997. Long-term vegetation dynamics mediated by
herbivores, weather and fire in a Juniperus-Quercus Savanna. J. Veg. Sci. 8,
819e828.
Fynn, R.W.S., O'Connor, T.G., 2000. Effect of stocking rate and rainfall on rangeland
dynamics and cattle performance in a semi-arid savanna, South Africa. J. Appl.
Ecol. 37, 491e507.
Graham, M.H., 2003. Confronting multicollinearity in ecological multiple regression. Ecology 84, 2809e2815.
Habich, N., 1992. Rangeland Inventory and Monitoring: Supplemental Studies
Addition. Addendum to Technical Reference 4400-5. Bureau of Land Management, Denver, CO.
Hardegree, S.P., Van Vactor, S.S., 2004. Microclimatic constraints and revegetation
planning in a variable environment. Weed Technol. 18, 1213e1215.
Hobbie, J.E., Carpenter, S.R., Grimm, N.B., Gosz, J.R., Seastedt, T.R., 2003. The US long
term ecological research program. Bioscience 53, 21e32.
Holmgren, M., Scheffer, M., 2001. El Nino as a window of opportunity for the
restoration of degraded arid ecosystems. Ecosystems 4, 151e159.
Hurlbert, S.H., 1984. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54, 187e211.
Iglesias, R.M.R., Kothmann, M.M., 1997. Structure and causes of vegetation change in
state and transition model applications. J. Range Manag. 50, 399e408.
Laycock, W.A., 1991. Stable states and thresholds of range condition on North
American rangelands: a viewpoint. J. Range Manag. 44, 427e433.
Laycock, W.A., Buchanan, H., Krueger, W.C., 1972. Three methods of determining
diet, utilization, and trampling damage on sheep ranges. J. Range Manag. 25,
352e356.
Loik, M.E., 2007. Sensitivity of water relations and photosynthesis to summer
precipitation pulses for Artemisia tridentata and Purshia tridentata. Plant Ecol.
191, 95e108.
Loik, M.E., Griffith, A.B., Alpert, H., 2013. Impacts of long-term snow climate change
on a high-elevation cold desert shrubland, California, USA. Plant Ecol. 214,
255e266.
Low, G., Provencher, L., Abele, S., 2010. Enhanced conservation action planning:
assessing landscape condition and predicting benefits of conservation strategies. J. Conserv. Plan. 6, 36e60.
Mashiri, F.E., McClaran, M.P., Fehmi, J.S., 2008. Short- and long-term vegetation
change related to grazing systems, precipitation and mesquite cover. Rangel.
Ecol. Manag. 61, 368e379.
McClaran, M.P., Wei, H., 2014. Recent drought phase in a 73-year record at two
117
spatial scales: implications for livestock production on rangelands in the
Southwestern United States. Agric. For. Meteorol. 197, 40e51.
Moran, M.S., Ponce-Campos, G.E., Huete, A., McClaran, M.P., Zhang, Y.,
Hamerlynck, E.P., Augustine, D.J., Gunter, S.A., Kitchen, S.G., Peters, D.P.C.,
Starks, P.J., Hernandez, M., 2014. Functional response of U.S. grasslands to the
early 21st-century drought. Ecology 95, 2121e2133.
Morris, C., Badik, K.J., Weltz, M.A., 2014. Analysis of Long-term (1969e2011)
Vegetation Data from Bodie Hills, CA: Integrating Weather, Disturbance, Lag
Effects and Interactions in Vegetation Change. USDA-ARS Great Basin Rangeland Research Unit, Reno, NV.
O'Neil, J.R., Silberman, M.L., Fabbi, B.P., Chesterman, C.W., 1973. Stable isotope and
chemical relations during mineralization in the Bodie mining district, Mono
County, California. Econ. Geol. 68, 765e784.
Ogle, K., Reynolds, J.F., 2004. Plant responses to precipitation in desert ecosystems:
integrating functional types, pulses, thresholds, and delays. Oecologia 141,
282e294.
Oesterheld, M., McNaughton, S.J., 1991. Effect of stress and time for recovery on the
amount of compensatory growth after grazing. Oecologia 85, 305e313.
Perlinski, A.T., Paige, G.B., McClaran, M.P., 2014. Evaluating a state-and-transition
model using a long-term dataset. Rangel. Ecol. Manag. 67, 173e182.
PRISM., 2012. PRISM Climate Group. Oregon State University. http://prism.
oregonstate.edu (created 4 Feb 2004).
Provencher, L., Low, G., Abele, S., 2009. Bodie Hills Conservation Action Planning.
Final report to the Bureau of Land Management Bishop field office. The Nature
Conservancy, Reno, NV.
Schenk, H.J., Jackson, R.B., 2002. Rooting depths, lateral spreads and below-ground/
above-ground allometries of plants in water-limited ecosystems. J. Ecol. 90,
480e494.
Shannon, C.E., 1948. A mathematical theory of communication. Bell Syst. Tech. J. 27,
623e656.
Sokal, R.R., Rohlf, F.J., 1981. Biometry: the Principles and Practice of Statistics in
Biological Research. WH Freeman & Co., New York, NY.
Somodi, I., Viragh, K., Aszalos, R., 2004. The effect of the abandonment of grazing on
the mosaic of vegetation patches in a temperate grassland area in Hungary.
Ecol. Complex. 1, 177e189.
Sprague, M., 2003. Bodie's Gold: Tall Tales and True History from a California
Mining Town. University of Nevada Press, Reno, NV.
Stohlgren, T.J., 2007. Measuring Plant Diversity. Oxford University Press, New York,
NY.
Tilman, D., 1989. Ecological experiments: strengths and conceptual problems. In:
Likens, G.E. (Ed.), Long-term Studies in Ecology: Approaches and Alternatives.
Springer-Verlag, New York, NY, pp. 20e71.
Walker, B.L., 1993. Rangeland ecology: understanding and managing change. Ambio
22, 80e87.
West, N.E., 1979. Survival patterns of major perennials in salt desert shrub communities of southwestern Utah. J. Range Manag. 32, 442e445.
West, N.E., 2003a. History of rangeland monitoring in the U.S.A. Arid Land Res.
Manag. 17, 495e545.
West, N.E., 2003b. Theoretical underpinnings of rangeland monitoring. Arid Land
Res. Manag. 17, 333e346.
West, N.E., Rea, K.H., Harniss, R.O., 1979. Plant demographic studies in sagebrushgrass communities of southeastern Idaho. Ecology 60, 376e388.
West, N.E., Yorks, T.P., 2002. Vegetation responses following wildfire on grazed and
ungrazed sagebrush semi-desert. J. Range Manag. 55, 171e181.
Westoby, M., Walker, B., Noy-Meir, I., 1989. Opportunistic management for rangelands not at equilibrium. J. Range Manag. 42, 266e274.
Wiegand, T., Milton, S.J., 1996. Vegetation change in semiarid communities simulating probabilities and time scales. Vegetatio 125, 169e183.
WRCC., 2010. Historical Climate Information. Western Regional Climate Center.
Available online at: www.wrcc.dri.edu (Accessed 1/10).
Yao, J., Peters, D.P.C., Havstad, K.M., Gibbens, R.P., Herrick, J.E., 2006. Multi-scale
factors and long-term responses of Chihuahuan Desert grasses to drought.
Landsc. Ecol. 21, 1217e1231.
Young, J.A., Budy, J.D., 1979. Historical use of Nevada's pinyon-juniper woodlands.
J. For. Hist. 23, 113e121.