ArticlePDF Available

Removal of cattle grazing correlates with increases in vegetation productivity and in abundance of imperiled breeding birds

Authors:

Abstract and Figures

Livestock grazing is the most prevalent land use practice in the western United States and a widespread cause of degradation of riparian vegetation. Riparian areas provide high-quality habitat for many species of declining migratory breeding birds. We analyzed changes in vegetation and bird abundance at a wildlife refuge in southeastern Oregon over 24 years, following cessation of 120 years of livestock grazing. We quantified long-term changes in overall avian abundance and species richness and, specifically, in the abundances of 20 focal species. We then compared the local responses of the focal species to population-scale trends of the same species at three different large spatial scales. Overall avian abundance increased 23% during the 12 years after removal and remained consistent from then through year 24. Three times as many species colonized the survey sites as dropped out. Of the focal species, most riparian woodland-tree or shrub dependent, sagebrush obligate, and grassland or meadow taxa increased in abundance or remained stable locally. As these species were generally of conservation concern, the population increases contradicted regionally declining or stable trends. In contrast, most riparian woodland-cavity nester species decreased in abundance locally, reflecting disruption of aspen stand dynamics by decades of grazing. Avian nest parasites and competitors of native species declined in abundance locally, matching regional trends. Restoring riparian ecosystems by removing livestock appeared to be beneficial to the conservation of many of these declining populations of migratory birds.
Content may be subject to copyright.
1
Supplementary Information
Removal of cattle grazing correlates with increases in vegetation productivity and
in abundance of imperiled breeding birds
Historical Grazing
Hart Mountain National Antelope Refuge (HMNAR) was established as a wildlife refuge in 1936.
Historically, native grazers at HMNAR included pronghorn (Antilocapra americana), California
bighorn sheep (Ovis canadensis), and mule deer (Odocoileus hemionus; USFWS, 1994). During
the 1920s, approximately 1,000 pronghorn were on or around the lands that became HMNAR.
Bighorn sheep were extirpated from those lands in 1912, but were successfully reintroduced in
1954. Approximately 2,000-3,000 mule deer inhabited HMNAR during the late 1930s and early
1940s.
Cattle (Bos taurus) grazing on and around HMNAR began in the early 1870s on newly-formed
ranches (USFWS, 1994). Cattle were grazed throughout the year without management of animal
numbers or distribution. Domestic sheep (Ovis aries) were introduced to the area around 1900.
HMNAR was heavily grazed by both sheep and cattle during the early 1900s, resulting in
excessive vegetation degradation by the 1920s. In the 1930s after the establishment of HMNAR
as a refuge, management efforts began to reduce livestock numbers. Domestic sheep were
eliminated by 1960. During 1971-1990, grazing management included the delineation of 43
grazing units and 10 exclusion zones, as well as seasonal grazing (April-October), with an annual
average of 12,834 animal unit months (range 10,406-17,228). Cattle were grazed in lower
elevation units during the growing season (April-June) and higher elevation units after the
growing season. Cattle were removed from HMNAR in 1990.
Feral horses (Equus caballus) had been distributed in the southeastern quarter of HMNAR since
the 1960s (USFWS, 1994). Horses were removed by 1999, then populations increased to 200-
300 animals, which were then removed by 2010.
Model Selection
Because the number of bird surveys conducted at a plot differed among phases, we included the
number of visits to a plot as a second random effect (in addition to plot), but this variable did not
improve model fit for 19 of the 20 species or change the responses by phase for 18 of the 20
species. Hence, we report model results without this additional random effect.
For each of the 20 focal species for which we analyzed abundance, the best-performing models
had a variety of structures with and without zero-inflation (Table S3). None of the 20 focal
species occurred at every plot. From 13 to 75 plots had zero occurrences of a given species (out
of 47 plots surveyed in phase 1 and 106 plots surveyed in phases 2 and 3), even when occurrences
were summed across all three phases of the study.
Abundance and Species Richness by Phase
Observers identified 123 bird species at >1 plot over the three phases of our study, 93 during
phase 1, 109 during phase 2, and 111 during phase 3. Observers collected 36,856 individual bird
detections during the study, 10,359 during phase 1, 12,072 during phase 2, and 14,425 during
phase 3. These included 337 individuals for which the species could not be identified.
2
Alternative Explanations for Vegetation Change
We considered the extent to which changes in vegetation we observed could be explained by
factors other than removal of grazing. For example, the increase in NDVI could have been due to
an increase in precipitation, causing greater vegetation growth. To test this, we obtained
precipitation records for May of each year of the study (PRISM Climate Group, Oregon State
University, http://prism.oregonstate.edu). We ran a linear model with square-root transformed
precipitation as the response variable and phase as a predictor variable. During the study, May
precipitation varied over time (F2, 951 = 199.5, P < 0.001) and was higher during phase 1 than
during both phases 2 and 3 (P < 0.001), a pattern inconsistent with an increase in vegetation.
Likewise, 12 plots were burned after the first bird surveys (i.e., after the first phase of surveys for
each plot was completed) as part of HMNAR’s routine management practices. Of these, six plots
were burned more than once. Such burning would have altered patterns in vegetation regrowth.
However, the vegetation changes resulting from these burns were accounted for in our NDVI
measurements. We evaluated NDVI values after burns. They were all within the range of NDVI
values for unburned plots during the same year (i.e., burned plots were not outliers). We then
used a t-test to compare mean percent change in NDVI values before and after burns for those
plots burned only once or more than once. In both cases, mean NDVI increased after the burns,
and the percent change among plots burned once or more than once was not different (t10 = 1.05,
P = 0.317).
3
Fig. S1. Two photographs of the same riparian site at Hart Mountain National Antelope Refuge at
two different time periods: (a) July 1989 and (b) August 2013. The photos show recovery of the
landscape more than two decades after cattle were removed. Photos are two of those used in
Batchelor et al. (2015), and we include them here with permission from W. Ripple.
4
Fig. S2. Average (±SE) relative abundances of six bird species in the riparian woodland-tree or
shrub dependent functional group and their responses during three phases of study after cattle
removal at Hart Mountain National Antelope Refuge, 1991-2014. These represent: (a) late
increase (dusky flycatcher [DUFL], green-tailed towhee [GTTO], orange-crowned warbler
[OCWA], and warbling vireo [WAVI]); and (b) steady increase (MacGillivray’s warbler
[MGWA]) and no change (western wood-pewee [WEWP]). Phase 1 occurred during 1991-1993,
phase 2 occurred during 2000-2002, and phase 3 occurred during 2012-2014. Sample size was 47
plots in phase 1 and 106 plots in both phases 2 and 3.
5
Fig. S3. Average (±SE) relative abundances of three bird species in the riparian woodland-cavity
nester functional group and their responses during three phases of study after cattle removal at
Hart Mountain National Antelope Refuge, 1991-2014. These represent: early decrease (northern
flicker [NOFL] and sapsuckers [SAPS; red-naped sapsucker and red-breasted sapsucker]) and no
change (swallows [SWAL; tree swallow and violet-green swallow]). Phase 1 occurred during
1991-1993, phase 2 occurred during 2000-2002, and phase 3 occurred during 2012-2014. Sample
size was 47 plots in phase 1 and 106 plots in both phases 2 and 3.
6
Fig. S4. Average (±SE) relative abundances of three bird species in the grassland or meadow
functional group and their responses during three phases of study after cattle removal at Hart
Mountain National Antelope Refuge, 1991-2014. These represent: late increase (savannah
sparrow [SAVS]), early decrease (western meadowlark [WEME]), and no change (vesper
sparrow [VESP]). Phase 1 occurred during 1991-1993, phase 2 occurred during 2000-2002, and
phase 3 occurred during 2012-2014. Sample size was 47 plots in phase 1 and 106 plots in both
phases 2 and 3.
7
Fig. S5. Average (±SE) relative abundance of sage thrashers (SATH), a species in the sagebrush
obligate functional group, showing no change in response during three phases of study after cattle
removal at Hart Mountain National Antelope Refuge, 1991-2014. Phase 1 occurred during 1991-
1993, phase 2 occurred during 2000-2002, and phase 3 occurred during 2012-2014. Sample size
was 47 plots in phase 1 and 106 plots in both phases 2 and 3.
8
Fig. S6. Average (±SE) relative abundance of brown-headed cowbirds (BHCO), a species in the
avian nest parasite or native species competitor functional group, showing a steady decrease in
response during three phases of study after cattle removal at Hart Mountain National Antelope
Refuge, 1991-2014. Phase 1 occurred during 1991-1993, phase 2 occurred during 2000-2002,
and phase 3 occurred during 2012-2014. Sample size was 47 plots in phase 1 and 106 plots in
both phases 2 and 3.
9
Table S1. Distributions used in models fit for each of 20 focal bird species in Hart Mountain
National Antelope Refuge, 1991-2014.
Distribution
Description
Poisson
Poisson distribution
Generalized Poisson
Poisson distribution with an additional parameter to model
over-dispersion or under-dispersion
Conway-Maxwell Poisson
Poisson distribution with an additional parameter to model
over-dispersion or under-dispersion
Negative Binomial 1
Negative binomial distribution that has a variance that
increases linearly with the mean
Negative Binomial 2
Negative binomial distribution that has a variance that
increases quadratically with the mean
10
Table S2. Zero-inflation parameters used in models fit for each of 20 focal bird species in Hart
Mountain National Antelope Refuge, 1991-2014.
Zero-inflation parameter
Description
No zero-inflation
Zero-inflation is not modeled
Single parameter
Zero-inflation modeled as equal for all observations (intercept only)
Fixed effects
Rescaled NDVI nested within phase and rescaled elevation
Fixed and random effects
Rescaled NDVI nested within phase and rescaled elevation (fixed
effects) and plot (random effect)
11
Table S3. Best-performing model (i.e., model with the lowest Akaike’s Information Criterion
corrected for small sample size value) for each of 20 focal bird species, by functional group,
explaining the response of bird abundance to Normalized Difference Vegetation Index (NDVI)
over time and elevation in Hart Mountain National Antelope Refuge, 1991-2014. Each model
included 259 observations from 106 study plots.
Species
Distributiona
ZI modelb
Kc
Model weightd
Riparian Woodland-Tree or Shrub Dependent:
American robin
Generalized Poisson
Fixed effects
16
0.92
Dusky flycatcher
Conway-Maxwell Poisson
No zero-inflation
9
0.48
Green-tailed towhee
Generalized Poisson
Single parameter
10
0.48
MacGillivray’s warbler
Conway-Maxwell Poisson
No zero-inflation
9
0.75
Orange-crowned warbler
Conway-Maxwell Poisson
Fixed effects
16
0.99
Spotted towhee
Poisson
No zero-inflation
8
0.36
Warbling vireo
Conway-Maxwell Poisson
No zero-inflation
9
0.42
Western wood-pewee
Conway-Maxwell Poisson
No zero-inflation
9
0.55
Yellow warbler
Conway-Maxwell Poisson
Fixed effects
16
0.62
Riparian Woodland-Cavity Nester:
Mountain bluebird
Poisson
Fixed effects
15
0.26
Northern flicker
Generalized Poisson
Single parameter
10
1.00
Red-breasted/red-naped sapsuckers
Poisson
No zero-inflation
8
0.32
Tree/violet-green swallows
Negative Binomial 1
No zero-inflation
9
0.29
Grassland or Meadow:
Savannah sparrow
Poisson
No zero-inflation
8
0.38
Vesper sparrow
Generalized Poisson
No zero-inflation
9
0.94
Western meadowlark
Generalized Poisson
Fixed effects
16
0.67
Sagebrush Obligate:
Brewer’s sparrow
Generalized Poisson
Fixed effects
16
1.00
Sage thrasher
Poisson
No zero-inflation
8
0.41
Avian Nest Parasite or Native Species Competitor:
Brown-headed cowbird
Poisson
Fixed effects
15
0.49
European starling
Poisson
No zero-inflation
8
0.36
a Distribution is the distribution used for the best-performing model (see Table S1 for descriptions
of each model distribution).
b ZI model indicates which zero-inflation parameters were included (see Table S2 for descriptions
of parameters).
c K is the number of parameters (including intercepts) in a model.
d Model weight is the weight of evidence in favor of each model, based on the model set used for
that species.
12
Table S4. Results of statistical tests of change in avian abundance, species richness, species
diversity, and Normalized Difference Vegetation Index (NDVI) during three phases of study at
Hart Mountain National Antelope Refuge, 1991-2014. Phase 1 occurred during 1991-1993,
phase 2 occurred during 2000-2002, and phase 3 occurred during 2012-2014.
Parameter
Time period
Test statistic
P-value
Abundance
Overall
F2,154 = 32.08
<0.001
Abundance
Phase 2 - Phase 1
t155 = 7.14
<0.001
Abundance
Phase 3 - Phase 1
t155 = 7.72
<0.001
Abundance
Phase 3 - Phase 2
t151 = 0.80
0.701
Species richness
Overall
χ22 = 12.69
0.002
Species richness
Phase 2 - Phase 1
z = -3.04
0.007
Species richness
Phase 3 - Phase 1
z = -0.83
0.687
Species richness
Phase 3 - Phase 2
z = 2.92
0.010
Species diversity
Overall
Kruskal-Wallis χ22 = 1.46
0.482
NDVI
Overall
F2,210 = 45.27
<0.001
NDVI
Phase 2 - Phase 1
t210 = 5.20
<0.001
NDVI
Phase 3 - Phase 1
t210 = 9.50
<0.001
NDVI
Phase 3 - Phase 2
t210 = 4.30
<0.001
13
Table S5a. Bird species that colonized Hart Mountain National Antelope Refuge during three
phases of study, 1991-2014. The same 47 original plots were sampled through all three phases,
but 59 new plots were added to the study in phase 2. Colonized species were those that occurred
at either (i) 0 plots in phase 1 and >1 original plot in phase 3 (“phase 3a”) or (ii) 0 plots in phases
1 and 2 and >1 plot in phase 3 (“phase 3b”). Species categorized as colonized in the 47 original
plots (“phase 3a”) sometimes were detected in phase 2 but only in the 59 new plots; thus, they
were detected in 0% of the original plots in phase 2. Percentages are the number of plots in
which a species was detected divided by the number of plots sampled (indicated under column
headings).
Species
Percentage of plots
Common name
Scientific name
Phase 1
Phase 2
Phase 3a
(original
plots only)
Phase 3b
(all plots)
(n=47)
(n=47)
(n=47)
(n=106)
Ash-throated flycatcher
Myiarchus cinerascens
0.00
0.00
1.89
Black-throated gray warbler
Setophaga nigrescens
0.00
0.00
6.38
California quail
Callipepla californica
0.00
0.00
4.26
California scrub-jay
Aphelocoma californica
0.00
0.00
4.26
Common yellowthroat
Geothlypis trichas
0.00
0.00
3.77
Golden-crowned kinglet
Regulus satrapa
0.00
2.13
4.26
Lark sparrow
Chondestes grammacus
0.00
0.00
1.89
Loggerhead shrike
Lanius ludovicianus
0.00
0.00
2.83
Pacific wren
Troglodytes pacificus
0.00
0.00
3.77
Western bluebird
Sialia mexicana
0.00
0.00
1.89
Wilson’s snipe
Gallinago delicata
0.00
0.00
5.66
Yellow-breasted chat
Icteria virens
0.00
0.00
6.60
Table S5b. Bird species that dropped out at Hart Mountain National Antelope Refuge during
three phases of study, 1991-2014. Dropped out species were those that were detected at >1 plot
in phase 1 and 0 plots in phase 3. Percentages are the number of plots in which a species was
detected divided by the number of plots sampled (47 in phase 1 and 106 in phases 2 and 3).
Species
Percentage of plots
Common name
Scientific name
Phase 1
Phase 2
Phase 3
Canada goose
Branta canadensis
8.51
0.00
0.00
Long-eared owl
Asio otus
4.26
0.94
0.00
Ruby-crowned kinglet
Regulus calendula
17.02
0.00
0.00
Sharp-shinned hawk
Accipiter striatus
10.64
0.00
0.00
14
Table S6. Coefficients, SEs, z-values, and 95% confidence intervals (CI) from the best-
performing models explaining the factors associated with abundances of nine bird species in the
riparian woodland-tree or shrub dependent functional group during three phases of study after
cattle removal at Hart Mountain National Antelope Refuge, 1991-2014. Reference category for
Phase was phase 1; NDVI was the Normalized Difference Vegetation Index.
Species/Model/Parameter
Coefficient
SE
z-value
Lower CI
Upper CI
American Robin
Conditional Model
Intercept
1.000
0.100
10.024
0.804
1.195
Phase 2
-0.305
0.093
-3.289
-0.486
-0.123
Phase 3
-0.558
0.105
-5.295
-0.764
-0.351
Phase 1: NDVI
0.652
0.243
2.682
0.175
1.128
Phase 2: NDVI
0.333
0.144
2.316
0.051
0.615
Phase 3: NDVI
0.706
0.150
4.717
0.413
0.999
Elevation
0.066
0.150
0.437
-0.228
0.359
Zero-inflation Model
Intercept
-1.724
0.812
-2.123
-3.316
-0.132
Phase 2
-4.651
1.792
-2.596
-8.162
-1.140
Phase 3
-4.558
1.682
-2.710
-7.854
-1.261
Phase 1: NDVI
-0.269
1.850
-0.145
-3.894
3.357
Phase 2: NDVI
5.858
2.155
2.718
1.634
10.082
Phase 3: NDVI
6.148
2.467
2.492
1.312
10.984
Elevation
-10.531
3.389
-3.107
-17.174
-3.889
Dusky Flycatcher
Conditional Model
Intercept
-0.180
0.142
-1.274
-0.458
0.097
Phase 2
0.121
0.093
1.310
-0.060
0.303
Phase 3
0.437
0.107
4.099
0.228
0.645
Phase 1: NDVI
0.251
0.292
0.860
-0.321
0.823
Phase 2: NDVI
0.363
0.159
2.281
0.051
0.675
Phase 3: NDVI
0.172
0.176
0.979
-0.172
0.517
Elevation
1.752
0.264
6.645
1.235
2.269
No Zero-inflation
Green-tailed Towhee
Conditional Model
Intercept
-1.344
0.246
-5.470
-1.826
-0.863
Phase 2
0.132
0.077
1.725
-0.018
0.282
Phase 3
0.326
0.097
3.365
0.136
0.515
Phase 1: NDVI
0.598
0.324
1.847
-0.036
1.232
Phase 2: NDVI
0.008
0.207
0.038
-0.397
0.413
Phase 3: NDVI
-0.019
0.235
-0.080
-0.480
0.442
Elevation
1.778
0.427
4.160
0.940
2.616
Zero-inflation Model
Intercept
-2.496
0.370
-6.756
-3.220
-1.772
15
Species/Model/Parameter
Coefficient
SE
z-value
Lower CI
Upper CI
MacGillivray’s Warbler
Conditional Model
Intercept
-1.843
0.326
-5.656
-2.481
-1.204
Phase 2
0.927
0.329
2.814
0.281
1.572
Phase 3
1.535
0.327
4.691
0.894
2.176
Phase 1: NDVI
-1.087
0.902
-1.205
-2.855
0.681
Phase 2: NDVI
0.627
0.279
2.243
0.079
1.174
Phase 3: NDVI
0.479
0.250
1.919
-0.010
0.968
Elevation
0.208
0.265
0.783
-0.312
0.727
No Zero-inflation
Orange-crowned Warbler
Conditional Model
Intercept
-0.800
0.261
-3.071
-1.311
-0.289
Phase 2
-0.121
0.319
-0.380
-0.745
0.504
Phase 3
0.693
0.284
2.444
0.137
1.249
Phase 1: NDVI
0.519
0.612
0.848
-0.681
1.718
Phase 2: NDVI
0.236
0.321
0.736
-0.393
0.866
Phase 3: NDVI
0.122
0.271
0.452
-0.408
0.653
Elevation
1.274
0.225
5.671
0.833
1.714
Zero-inflation Model
Intercept
-5.706
4.323
-1.320
-14.178
2.766
Phase 2
3.487
4.093
0.852
-4.534
11.509
Phase 3
4.393
4.109
1.069
-3.660
12.446
Phase 1: NDVI
-23.038
17.204
-1.339
-56.757
10.681
Phase 2: NDVI
-8.918
4.105
-2.172
-16.963
-0.872
Phase 3: NDVI
-7.464
2.533
-2.947
-12.427
-2.500
Elevation
2.951
1.618
1.824
-0.220
6.121
Spotted Towhee
Conditional Model
Intercept
-3.106
0.730
-4.254
-4.536
-1.675
Phase 2
1.868
0.741
2.520
0.415
3.320
Phase 3
1.918
0.745
2.573
0.457
3.379
Phase 1: NDVI
0.804
2.014
0.399
-3.143
4.750
Phase 2: NDVI
0.834
0.453
1.840
-0.054
1.723
Phase 3: NDVI
0.666
0.424
1.573
-0.164
1.497
Elevation
-1.261
0.420
-3.004
-2.084
-0.438
No Zero-inflation
16
Species/Model/Parameter
Coefficient
SE
z-value
Lower CI
Upper CI
Warbling Vireo
Conditional Model
Intercept
-0.798
0.214
-3.727
-1.218
-0.379
Phase 2
0.095
0.176
0.538
-0.250
0.439
Phase 3
0.398
0.187
2.126
0.031
0.764
Phase 1: NDVI
1.233
0.480
2.570
0.292
2.173
Phase 2: NDVI
0.891
0.244
3.649
0.412
1.369
Phase 3: NDVI
1.048
0.266
3.947
0.528
1.569
Elevation
1.071
0.333
3.219
0.419
1.722
No Zero-inflation
Western Wood-Pewee
Conditional Model
Intercept
-1.041
0.264
-3.938
-1.559
-0.523
Phase 2
-0.491
0.260
-1.891
-1.000
0.018
Phase 3
-0.508
0.279
-1.822
-1.054
0.038
Phase 1: NDVI
1.470
0.573
2.564
0.346
2.593
Phase 2: NDVI
0.883
0.343
2.573
0.210
1.556
Phase 3: NDVI
1.233
0.353
3.495
0.541
1.924
Elevation
0.392
0.361
1.087
-0.315
1.100
No Zero-inflation
Yellow Warbler
Conditional Model
Intercept
0.375
0.149
2.511
0.082
0.668
Phase 2
0.315
0.127
2.473
0.065
0.564
Phase 3
0.528
0.137
3.864
0.260
0.795
Phase 1: NDVI
0.096
0.406
0.237
-0.699
0.892
Phase 2: NDVI
0.384
0.194
1.980
0.004
0.765
Phase 3: NDVI
0.260
0.202
1.284
-0.137
0.657
Elevation
0.322
0.229
1.407
-0.126
0.770
Zero-inflation Model
Intercept
-6.381
3.856
-1.655
-13.938
1.176
Phase 2
-0.899
4.624
-0.195
-9.962
8.163
Phase 3
3.038
3.910
0.777
-4.625
10.700
Phase 1: NDVI
-14.111
8.387
-1.683
-30.549
2.327
Phase 2: NDVI
-14.440
5.253
-2.749
-24.736
-4.143
Phase 3: NDVI
-8.070
2.252
-3.584
-12.484
-3.656
Elevation
2.988
1.292
2.314
0.457
5.520
17
Table S7. Coefficients, SEs, z-values, and 95% confidence intervals (CI) from the best-
performing models explaining the factors associated with abundances of four bird species in the
riparian woodland-cavity nester functional group during three phases of study after cattle removal
at Hart Mountain National Antelope Refuge, 1991-2014. Reference category for Phase was
phase 1; NDVI was the Normalized Difference Vegetation Index.
Species/Model/Parameter
Coefficient
SE
z-value
Lower CI
Upper CI
Mountain Bluebird
Conditional Model
Intercept
0.388
0.435
0.893
-0.464
1.240
Phase 2
-0.620
0.445
-1.393
-1.492
0.252
Phase 3
-2.215
0.533
-4.155
-3.260
-1.170
Phase 1: NDVI
-0.191
0.822
-0.232
-1.803
1.421
Phase 2: NDVI
0.702
0.369
1.903
-0.021
1.426
Phase 3: NDVI
2.517
0.496
5.078
1.545
3.488
Elevation
0.081
0.408
0.199
-0.719
0.881
Zero-inflation Model
Intercept
2.472
0.804
3.074
0.896
4.048
Phase 2
-1.558
1.227
-1.270
-3.963
0.846
Phase 3
-5.165
2.328
-2.219
-9.728
-0.602
Phase 1: NDVI
-3.812
1.867
-2.042
-7.470
-0.153
Phase 2: NDVI
-1.399
1.800
-0.777
-4.927
2.129
Phase 3: NDVI
3.828
2.191
1.747
-0.466
8.122
Elevation
-5.983
1.946
-3.075
-9.797
-2.170
Northern Flicker
Conditional Model
Intercept
-1.111
0.250
-4.440
-1.601
-0.620
Phase 2
-0.346
0.058
-5.942
-0.460
-0.232
Phase 3
-0.343
0.079
-4.363
-0.497
-0.189
Phase 1: NDVI
-1.151
0.207
-5.558
-1.556
-0.745
Phase 2: NDVI
-0.266
0.109
-2.440
-0.480
-0.052
Phase 3: NDVI
-0.132
0.144
-0.919
-0.415
0.150
Elevation
2.896
0.454
6.375
2.006
3.787
Zero-inflation Model
Intercept
-2.635
0.369
-7.137
-3.358
-1.911
Sapsuckers
Conditional Model
Intercept
-0.907
0.281
-3.233
-1.457
-0.357
Phase 2
-0.851
0.311
-2.737
-1.461
-0.242
Phase 3
-0.855
0.339
-2.521
-1.520
-0.190
Phase 1: NDVI
1.240
0.576
2.154
0.112
2.368
Phase 2: NDVI
0.898
0.367
2.449
0.179
1.616
Phase 3: NDVI
1.486
0.390
3.813
0.722
2.250
Elevation
0.910
0.379
2.402
0.167
1.652
No Zero-inflation
18
Species/Model/Parameter
Coefficient
SE
z-value
Lower CI
Upper CI
Swallows
Conditional Model
Intercept
-0.099
0.219
-0.451
-0.528
0.330
Phase 2
-0.001
0.206
-0.007
-0.405
0.402
Phase 3
-0.196
0.230
-0.852
-0.647
0.255
Phase 1: NDVI
1.660
0.482
3.444
0.715
2.604
Phase 2: NDVI
0.981
0.250
3.915
0.490
1.471
Phase 3: NDVI
1.009
0.281
3.590
0.458
1.559
Elevation
0.473
0.284
1.663
-0.085
1.030
No Zero-inflation
19
Table S8. Coefficients, SEs, z-values, and 95% confidence intervals (CI) from the best-
performing models explaining the factors associated with abundances of three bird species in the
grassland or meadow functional group during three phases of study after cattle removal at Hart
Mountain National Antelope Refuge, 1991-2014. Reference category for Phase was phase 1;
NDVI was the Normalized Difference Vegetation Index.
Species/Model/Parameter
Coefficient
SE
z-value
Lower CI
Upper CI
Savannah Sparrow
Conditional Model
Intercept
-2.360
0.448
-5.263
-3.239
-1.481
Phase 2
0.279
0.176
1.583
-0.066
0.625
Phase 3
0.356
0.172
2.074
0.020
0.693
Phase 1: NDVI
0.272
0.451
0.602
-0.613
1.156
Phase 2: NDVI
0.032
0.357
0.090
-0.667
0.731
Phase 3: NDVI
0.453
0.399
1.136
-0.329
1.235
Elevation
-2.060
0.680
-3.028
-3.393
-0.727
No Zero-inflation
Vesper Sparrow
Conditional Model
Intercept
-0.839
0.245
-3.424
-1.319
-0.359
Phase 2
-0.272
0.156
-1.743
-0.578
0.034
Phase 3
-0.094
0.165
-0.573
-0.417
0.229
Phase 1: NDVI
-1.111
0.477
-2.329
-2.046
-0.176
Phase 2: NDVI
-0.544
0.346
-1.573
-1.221
0.134
Phase 3: NDVI
-0.604
0.341
-1.771
-1.272
0.064
Elevation
0.274
0.334
0.822
-0.380
0.928
No Zero-inflation
Western Meadowlark
Conditional Model
Intercept
-2.516
0.454
-5.545
-3.405
-1.626
Phase 2
-0.264
0.079
-3.351
-0.418
-0.110
Phase 3
-0.058
0.084
-0.689
-0.223
0.107
Phase 1: NDVI
-1.248
0.257
-4.859
-1.751
-0.744
Phase 2: NDVI
-0.118
0.207
-0.571
-0.524
0.287
Phase 3: NDVI
0.375
0.233
1.606
-0.083
0.832
Elevation
-2.699
0.644
-4.189
-3.962
-1.436
Zero-inflation Model
Intercept
-3.367
1.794
-1.876
-6.884
0.150
Phase 2
-19.436
17.558
-1.107
-53.849
14.977
Phase 3
1.541
1.852
0.832
-2.089
5.170
Phase 1: NDVI
-2.156
3.084
-0.699
-8.200
3.888
Phase 2: NDVI
-32.542
26.266
-1.239
-84.022
18.939
Phase 3: NDVI
3.225
1.614
1.998
0.061
6.390
Elevation
-0.424
1.616
-0.262
-3.591
2.743
20
Table S9. Coefficients, SEs, z-values, and 95% confidence intervals (CI) from the best-
performing models explaining the factors associated with abundances of two bird species in the
sagebrush obligate functional group during three phases of study after cattle removal at Hart
Mountain National Antelope Refuge, 1991-2014. Reference category for Phase was phase 1;
NDVI was the Normalized Difference Vegetation Index.
Species/Model/Parameter
Coefficient
SE
z-value
Lower CI
Upper CI
Brewer’s Sparrow
Conditional Model
Intercept
-0.133
0.141
-0.939
-0.409
0.144
Phase 2
0.144
0.110
1.299
-0.073
0.360
Phase 3
0.590
0.113
5.206
0.368
0.812
Phase 1: NDVI
-1.157
0.309
-3.747
-1.762
-0.552
Phase 2: NDVI
-0.345
0.189
-1.820
-0.716
0.026
Phase 3: NDVI
-0.303
0.205
-1.482
-0.705
0.098
Elevation
0.051
0.246
0.206
-0.431
0.532
Zero-inflation Model
Intercept
-3.855
1.466
-2.629
-6.730
-0.981
Phase 2
-4.027
2.198
-1.832
-8.336
0.281
Phase 3
-2.485
1.747
-1.423
-5.910
0.939
Phase 1: NDVI
5.704
3.564
1.601
-1.281
12.689
Phase 2: NDVI
3.191
2.427
1.315
-1.566
7.947
Phase 3: NDVI
8.281
2.703
3.064
2.983
13.578
Elevation
-13.141
4.029
-3.262
-21.038
-5.245
Sage Thrasher
Conditional Model
Intercept
-2.027
0.437
-4.639
-2.884
-1.171
Phase 2
-0.368
0.464
-0.792
-1.278
0.542
Phase 3
-0.093
0.471
-0.197
-1.016
0.830
Phase 1: NDVI
-3.308
0.863
-3.835
-4.998
-1.617
Phase 2: NDVI
-1.418
0.704
-2.014
-2.797
-0.038
Phase 3: NDVI
-2.417
0.655
-3.690
-3.700
-1.133
Elevation
0.908
0.445
2.040
0.036
1.781
No Zero-inflation
21
Table S10. Coefficients, SEs, z-values, and 95% confidence intervals (CI) from the best-
performing models explaining the factors associated with abundances of two bird species in the
avian nest parasite or native species competitor functional group during three phases of study
after cattle removal at Hart Mountain National Antelope Refuge, 1991-2014. Reference category
for Phase was phase 1; NDVI was the Normalized Difference Vegetation Index.
Species/Model/Parameter
Coefficient
SE
z-value
Lower CI
Upper CI
Brown-headed Cowbird
Conditional Model
Intercept
1.001
0.126
7.929
0.753
1.248
Phase 2
-0.414
0.159
-2.598
-0.727
-0.102
Phase 3
-1.207
0.188
-6.437
-1.575
-0.840
Phase 1: NDVI
0.005
0.322
0.015
-0.627
0.636
Phase 2: NDVI
0.273
0.218
1.253
-0.154
0.699
Phase 3: NDVI
0.498
0.245
2.035
0.018
0.978
Elevation
-0.524
0.184
-2.846
-0.886
-0.163
Zero-inflation Model
Intercept
-1.173
0.534
-2.197
-2.219
-0.127
Phase 2
-0.608
0.741
-0.821
-2.059
0.843
Phase 3
-2.613
1.292
-2.022
-5.146
-0.080
Phase 1: NDVI
-0.763
1.401
-0.545
-3.509
1.983
Phase 2: NDVI
0.564
1.093
0.516
-1.577
2.706
Phase 3: NDVI
0.623
1.739
0.358
-2.787
4.032
Elevation
-4.953
1.411
-3.511
-7.718
-2.188
European Starling
Conditional Model
Intercept
-1.010
0.343
-2.945
-1.683
-0.338
Phase 2
-0.901
0.214
-4.206
-1.320
-0.481
Phase 3
-1.658
0.314
-5.287
-2.273
-1.043
Phase 1: NDVI
1.613
0.655
2.463
0.330
2.897
Phase 2: NDVI
1.008
0.445
2.264
0.135
1.880
Phase 3: NDVI
1.547
0.509
3.039
0.549
2.544
Elevation
1.132
0.577
1.964
0.002
2.262
No Zero-inflation
22
Table S11. Number of North American Breeding Bird Survey (BBS) trends at three spatial scales
(Oregon, Great Basin, and Western BBS) and responses in abundances of 20 focal bird species
during three phases of study after cattle removal at Hart Mountain National Antelope Refuge,
1991-2014. See Fig. 3 in main text and Figs. S2-S6 for common names associated with species
abbreviations. Number in parentheses is the conservation score for each species (scores >8
indicate species of high or moderate conservation concern). Light gray shaded cells indicate
positive contradiction of trends, dark gray shaded cells indicate negative contradiction of trends,
and cells with no fill (white) indicate matching response (no contradiction).
Oregon
Hart Mountain National Antelope Refuge
BBS
Decreasing
Stable
Increasing
Declining trend
AMRO (5)
BHCO (7)
EUSTb
NOFL (9)
WEME (10)
SATH (11)
SWALa
VESP (11)
WEWP (11)
BRSP (11)
GTTO (11)
MGWA (12)
OCWA (9)
SAVS (8)
SPTO (8)
YEWA (8)
Stable trend
MOBL (12)
None
DUFL (11)
WAVI (9)
Increasing trend
None
None
None
Great Basin
Hart Mountain National Antelope Refuge
BBS
Decreasing
Stable
Increasing
Declining trend
BHCO (7)
EUSTb
WEME (10)
SATH (11)
VESP (11)
DUFL (11)
GTTO (11)
SAVS (8)
Stable trend
AMRO (5)
NOFL (9)
SWALa
BRSP (11)
MGWA (12)
YEWA (8)
Increasing trend
MOBL (12)
WEWP (11)
SPTO (8)
WAVI (9)
Western BBS
Hart Mountain National Antelope Refuge
BBS
Decreasing
Stable
Increasing
Declining trend
AMRO (5)
BHCO (7)
EUSTb
NOFL (9)
WEME (10)
SATH (11)
SWALa
VESP (11)
WEWP (11)
MGWA (12)
OCWA (9)
SAVS (8)
YEWA (8)
Stable trend
MOBL (12)
SAPSc,d
None
BRSP (11)
DUFL (11)
GTTO (11)
SPTO (8)
Increasing trend
None
None
WAVI (9)
23
a Conservation scores for the two swallow species were: tree swallow = 10, violet-green swallow
= 9.
b European starlings were not scored.
c Conservation scores for the two sapsucker species were: red-naped sapsucker = 9, red-breasted
sapsucker = 11.
d In the Western BBS region, each of the two sapsucker species had a different trend, one stable
and one increasing. We assigned a stable trend (the more conservative of the two trends) to
sapsuckers as a whole.
24
Table S12. Number of combined North American Breeding Bird Survey (BBS) trends at three
spatial scales (Oregon, Great Basin, and Western BBS) and responses in abundances of 20 focal
bird species by functional group during three phases of study after cattle removal at Hart
Mountain National Antelope Refuge, 1991-2014. Light gray shaded cells indicate positive
contradiction of trends, dark gray shaded cells indicate negative contradiction of trends, and cells
with no fill (white) indicate matching response (no contradiction). Because trends are combined
across spatial scales, each avian species is represented up to 3 times in each functional group. In
the Western BBS region, each of the two sapsucker species (in the riparian woodland-cavity
nester functional group) had a different trend, one stable and one increasing. We assigned a
stable trend (the more conservative of the two trends) to sapsuckers as a whole.
Riparian Woodland-Tree or Shrub Dependent
Hart Mountain National Antelope Refuge
BBS Combined
Decreasing
Stable
Increasing
Declining trend
2
2
10
Stable trend
1
0
7
Increasing trend
0
1
3
Riparian Woodland-Cavity Nester
Hart Mountain National Antelope Refuge
BBS Combined
Decreasing
Stable
Increasing
Declining trend
2
2
0
Stable trend
4
1
0
Increasing trend
1
0
0
Grassland or Meadow
Hart Mountain National Antelope Refuge
BBS Combined
Decreasing
Stable
Increasing
Declining trend
3
3
3
Stable trend
0
0
0
Increasing trend
0
0
0
Sagebrush Obligate
Hart Mountain National Antelope Refuge
BBS
Decreasing
Stable
Increasing
Declining trend
0
3
1
Stable trend
0
0
2
Increasing trend
0
0
0
Avian Nest Parasite or Native Species Competitor
Hart Mountain National Antelope Refuge
BBS
Decreasing
Stable
Increasing
Declining trend
6
0
0
Stable trend
0
0
0
Increasing trend
0
0
0
25
References
Batchelor, J.L., Ripple, W.J., Wilson, T.M., Painter, L.E., 2015. Restoration of riparian areas
following the removal of cattle in the northwestern Great Basin. Environ. Manage. 55, 930
942.
U.S. Fish and Wildlife Service (USFWS), 1994. Final Environmental Impact Statement for the
Hart Mountain National Antelope Refuge Comprehensive Management Plan. U.S. Fish and
Wildlife Service, Region 1, Portland, OR.
... Some authors have shown that grazing exclusion prevents ecosystem degradation and restores degraded areas (Al-Rowaily et al. 2015;Listopad et al. 2018). Cattle grazing removal also correlates with increases in vegetation productivity and may benefit avian conservation in riparian areas (Poessel et al. 2020). In grassland ecosystems, livestock may be beneficial for biodiversity in suitable density and frequency of cattle grazing (Overbeck et al. 2007;Correa et al. 2019) and has been used as a strategy to improve biodiversity conservation (Verd u et al. 2007;T€ or€ ok et al. 2016). ...
Article
Livestock is a globally widespread farming practice, with benefits and harms for biodiversity. Biodiversity responses to vegetation and soil conditions associated with cattle grazing removal are poorly understood in tropical grassy ecosystems, especially in a long-term chronosequence. In this study, we aimed to identify the main drivers of local conditions on dung beetle taxonomic and functional diversity across a chronosequence of natural grasslands with different cattle grazing removal ages. We expect that vegetation and soil variables associated with environmental changes caused by longer time after cattle grazing removal will drive the taxonomic and functional diversity of dung beetles. We sampled dung beetles and recorded local environmental filters (vegeta-tion and soil variables) in 14 natural grasslands with distinct cattle grazing removal ages (from 3 months to 22 years) and also in 10 reference sites (with cattle grazing). We used structural equation models to evaluate the relationships between the most informative explanations (vegetation and soil variables) of dung beetle taxonomic and functional metrics. Our results provide evidences that local conditions related to vegetation and soil variables in natural grasslands affect the taxonomic and functional diversity of dung beetles differently. Soil conditions (compaction and silt content) had more influence, mostly negative, on taxonomic metrics, while vegetation complexity had more influence on functional metrics, with positive or negative effects depending on the functional metric evaluated. Understanding the temporal change in vegetation and soil conditions can shed light on the process of recovering biodiversity and ecosystem functions in areas with different times of abandonment of livestock farming. Finally, using taxonomic and functional approaches together can also help to better understand how organisms are responding to the process of abandoning livestock farming over time.
... At a public workshop on the refuge, Steve and his students shared their Robinson Draw results to demonstrate grazing impacts. Hart Mountain is now a model for understanding the benefits of livestock withdrawal for sagebrush-steppe ecosystems (Poessel et al. 2020). ...
Article
Full-text available
Globally, croplands and rangelands are major land uses and they have altered lands and waters for millennia. This continues to be the case throughout the USA, despite substantial improvements in treating wastewaters from point sources—versus non-point (diffuse) sources. Poor macroinvertebrate assemblage condition occurs in 30% of conterminous USA streams and rivers; poor fish assemblage condition occurs in 26%. The risk of poor fish assemblage condition was most strongly associated with excess nutrients, salinity and sedimentation and impaired riparian woody vegetation. Although the Clean Water Act was passed to restore and maintain the integrity of USA waters, that will be impossible without controlling agricultural pollution. Likewise, the Federal Land Policy and Management Act was enacted to protect the natural condition of public lands and waters, including fish habitat, but it has failed to curtail the sacred cows of livestock grazing. Although progress has been slow and spotty, promising results have been obtained from basin and watershed planning and riparian zone protections. ## See also https://today.oregonstate.edu/news/study-buffer-zones-better-regulation-needed-prevent-agricultural-pollution-rivers-streams ## This article isn't affiliated w/ any organization for RLV, although my past BC & CA postdocs contributed to my coauthorship. ## Further info: https://today.oregonstate.edu/news/study-buffer-zones-better-regulation-needed-prevent-agricultural-pollution-rivers-streams
Article
Full-text available
Globally, many bird species that rely on native woodland or forest environments are declining due to vegetation clearing for livestock pastures and cereal cropping. In many landscapes, woodland remnants are restricted to waterways and roadsides in narrow, sometimes degraded patches, and not all patches can necessarily provide the resources required to support bird populations. This study investigated the influence of livestock grazing and vegetation characteristics on bird breeding activity in riparian zones in northern Victoria, Australia, where much of the landscape is used for production and has experienced significant loss of woodland. Birds were broadly categorised as ‘woodland’ or ‘non-woodland’ species, based on dependency on woodlands for breeding. The majority of woodland species detected were relatively common, and where riparian zones were heavily grazed, there was significantly lower woodland bird breeding activity compared to non-woodland bird breeding activity (the latter increasing with grazing intensity). Woodland and non-woodland birds had consistently opposite responses to grazing intensity, vegetation and landscape characteristics, suggesting that the factors influencing breeding differ markedly between these two groups. Thus, where riparian zones are intensively grazed, the bird community shifts from predominantly woodland to largely non-woodland species. This has implications for the conservation of both rare and common woodland bird species in southern Australia. Simple changes in land management, for example, livestock exclusion from important breeding habitat, may confer large gains for population persistence of woodland bird species.
Article
We used data from statewide surveys of riparian birds in Utah, 1992–1998, to compare relative-abundance and distance-sampling methods. By generating relative-abundance indices with point-count methods and density with point-transect sampling methods, we examined whether the assumptions underlying each method were met during field surveys for four bird species (Brown-headed Cowbird [Molothrus ater], Bullock's Oriole [Icterus bullockii], Warbling Vireo [Vireo gilvus], and Yellow Warbler [Dendroica petechia]). Point-count methods failed to reasonably meet the fundamental assumption of constant proportionality, with estimated detectability varying 3- to 5-fold despite the use of widely accepted and well-standardized methods. Population trends based on relative abundance were subsequently unstable, often varying in both magnitude and direction with the survey plot radius used (25 m, 50 m, or unlimited distance). Distance-sampling methods appeared to meet critical assumptions, were robust to assumption violations, allowed methodological self-assessment, and were demonstrably efficient in a large-scale, multispecies survey setting. Our data show surveys of birds without estimations of detectability are likely biased because the assumption of constant proportionality is violated to a degree that precludes strict inference and may confound trend analyses.
Article
Riparian zones in the semi-arid West are disproportionately important habitats for both breeding birds and agricultural operations. Despite growing interest in studying avian-habitat relationships to inform land management decisions, few studies have described temporal changes in riparian bird communities. We compared indices of avian abundance and diversity from 3 streamside vegetation associations in east-central Oregon during May-June 2014 with indices from 1993-1994. Our objectives were to identify patterns of change in the avian community with a focus on riparian shrub-dependent species, re-examine previously reported relationships between avian abundance and vegetation volume, and identify possible causes of declines in bird abundance and diversity. We combined historical field protocols to survey birds and measure riparian vegetation with modern analytical techniques. We found few differences in species richness between study periods but documented approximately 30% species turnover. Increases in diversity were driven by increased detections of grassland and wetland bird species and increased evenness due to precipitous declines in detections for 2 of 3 riparian shrub-dependent focal species (i.e., Yellow Warbler [Setophaga petechia] and Willow Flycatcher [Empidonax traillii]). Detections of Song Sparrows (Melospiza melodia), our third focal species, declined by a smaller margin. Large changes in detections of riparian shrub-dependent species did not reflect trends in mesic shrub cover or volume, which had been identified as likely drivers of obligate species abundance. Focal species' declines reflected regional Breeding Bird Survey trends, corroborating our finding that their declines were not a result of changes in local site conditions. Our findings suggest that managing lands to increase wetness and extent of riparian zones can be beneficial for grassland and wetland bird species. However, managing for riparian shrub cover or volume, important metrics of grazing intensity and riparian system health, may be insufficient to conserve riparian shrub-dependent birds because other unidentified local and/or regional factors are likely contributing to habitat suitability of riparian shrub-dependent birds.
Article
Restoring native vegetation in agricultural landscapes can reverse biodiversity declines via species gains. Depending on whether the traits of colonizers are complementary or redundant to the assemblage, species gains can increase the efficiency or stability of ecological functions, yet detecting these processes is not straightforward. We propose a new conceptual model to identify potential changes to complementarity and redundancy in response to landscape change via relative changes in taxonomic and functional richness. We applied our model to a 14-year study of birds across an extensive agricultural region. We found compelling evidence that high levels of landscape-scale tree cover and patch-scale restoration were significant determinants of functional change in the overall bird assemblage. This was true for every one of the six traits investigated individually, indicating increased trait-specific functional complementarity and redundancy in the assemblage. Applying our conceptual model to species diversity data provided new insights into how the return of vertebrates to restored landscapes may affect ecological function.
Article
Grazing by cattle is ubiquitous across the sagebrush steppe; however, little is known about its effects on sagebrush and native bunchgrass structure. Understanding the effects of long-term grazing on sagebrush and bunchgrass structure is important because sagebrush is a keystone species and bunchgrasses are the dominant herbaceous functional group in these communities. To investigate the effects of long-term grazing on sagebrush and bunchgrass structure, we compared nine grazing exclosures with nine adjacent rangelands that were grazed by cattle in southeast Oregon. Grazing was moderate utilization (30 − 45%) with altering season of use and infrequent rest. Long-term grazing by cattle altered some structural aspects of bunchgrasses and sagebrush. Ungrazed bunchgrasses had larger dead centers in their crowns, as well as greater dead fuel depths below and above the crown level compared with grazed bunchgrasses. This accumulation of dry fuel near the meristematic tissue may increase the probability of fire-induced mortality during a wildfire. Bunchgrasses in the ungrazed treatment had more reproductive stems than those in the long-term grazed treatment. This suggests that seed production of bunchgrasses may be greater in ungrazed areas. Sagebrush height and longest canopy diameter were 15% and 20% greater in the ungrazed compared with the grazed treatment, respectively. However, the bottom of the sagebrush canopy was closer to the ground in the grazed compared with the ungrazed treatment, which may provide better hiding cover for ground-nesting avian species. Sagebrush basal stem diameter, number of stems, amount of dead material in the canopy, canopy gap size, and number of canopy gaps did not differ between ungrazed and grazed treatments. Moderate grazing does not appear to alter the competitive relationship between a generally unpalatable shrub and palatable bunchgrasses. Long-term, moderate grazing appears to have minimal effects to the structure of bunchgrasses and sagebrush, other than reducing the risk of bunchgrass mortality during a fire event.
Article
Satellite derived vegetation indices (VIs) are broadly used in ecological research, ecosystem modeling, and land surface monitoring. The Normalized Difference Vegetation Index (NDVI), perhaps the most utilized VI, has countless applications across ecology, forestry, agriculture, wildlife, biodiversity, and other disciplines. Calculating satellite derived NDVI is not always straightforward , however, as satellite remote sensing datasets are inherently noisy due to cloud and atmospheric contamination, data processing failures, and instrument malfunction. Readily available NDVI products that account for these complexities are generally at coarse resolution; high resolution NDVI datasets are not conveniently accessible and developing them often presents numerous technical and methodological challenges. We address this deficiency by producing a Landsat derived, high resolution (30 m), long-term (30+ years) NDVI dataset for the conterminous United States. We use Google Earth Engine, a planetary-scale cloud-based geospatial analysis platform, for processing the Landsat data and distributing the final dataset. We use a climatology driven approach to fill missing data and validate the dataset with established remote sensing products at multiple scales. We provide access to the composites through a simple web application, allowing users to customize key parameters appropriate for their application, question, and region of interest.