RANGE-WIDE POPULATION SIZE OF THE
LESSER PRAIRIE-CHICKEN:
2012, 2013, 2014, AND 2015
Prepared for
Western Association of Fish and Wildlife Agencies
c/o Bill Van Pelt
WAFWA Grassland Coordinator
Arizona Game and Fish Department
5000 W. Carefree Highway
Phoenix, Arizona 85086
_____________________________________________________________
Prepared by:
Lyman McDonald, Kristen Adachi, Troy Rintz, and Grant Gardner
WEST, Inc.
Western EcoSystems Technology, Inc.
200 South Second Street, Laramie, Wyoming 82070
July 9, 2015
1
TABLE OF CONTENTS
LIST OF FIGURES AND TABLES .......................................................................................................... 3
EXECUTIVE SUMMARY ........................................................................................................................ 4
INTRODUCTION....................................................................................................................................... 5
STUDY AREA ............................................................................................................................................. 6
METHODS .................................................................................................................................................. 6
Probabilistic Samples ................................................................................................................................ 6
Aerial Survey Methods ............................................................................................................................. 9
STATISTICAL METHODS ...................................................................................................................... 9
Probability of Detection ............................................................................................................................ 9
Estimation of Population Parameters in the Short Grass Prairie Region ................................................ 10
Estimation of Precision of Estimated Population Parameters ................................................................. 10
RESULTS .................................................................................................................................................. 10
Number of LPC Detected and Average Cluster Size 2013-2015 ............................................................ 15
Estimated Densities and Abundances of LPC and LPC Leks ................................................................. 15
DISCUSSION ............................................................................................................................................ 23
ACKNOWLEDGEMENTS ..................................................................................................................... 25
LITERATURE CITED ............................................................................................................................ 26
2
LIST OF FIGURES AND TABLES
FIGURE 1 Study area for 2012-2015 LPC surveys...................................................................... 8
FIGURE 2 Histogram of perpendicular distance to clusters of LPC 2012-2015 ....................... 11
FIGURE 3 Histogram of perpendicular distance to clusters of LPC; pooled data ..................... 12
FIGURE 4 Estimated total population size and abundance in each ecoregion........................... 18
FIGURE 5A Estimated total population sizes with 90% confidence intervals .......................... 19
FIGURE 5B Estimated population sizes with 90% confidence intervals ................................... 20
FIGURE 6 Estimated population size of GPC in NW Kansas with 90% confidence intervals . 23
FIGURE 7 Map of drought levels in U.S., June 4, 2013 ........................................................... 25
TABLE 1 Numbers and percentages of detections of clusters of LPC........................................ 12
TABLE 2 Sample sizes for estimation of probability of detection by observers ........................ 13
TABLE 3 Models for probability of detection in field of view of back seat observers .............. 14
TABLE 4 Models for probability of detection as function of distance to cluster........................ 14
TABLE 5 Numbers of LPC detected by ecoregion ..................................................................... 16
TABLE 6 Average cluster sizes of LPC ...................................................................................... 15
TABLE 7 Estimated densities of LPC with 90% confidence intervals ....................................... 16
TABLE 8 Estimated population sizes of LPC with 90% confidence intervals ........................... 16
TABLE 9 Estimated differences in population sizes with 80% confidence intervals ................. 17
TABLE 10 Estimated differences in population sizes with 90% confidence intervals ............... 17
TABLE 11 Estimated densities of LPC leks with 90% confidence intervals .............................. 21
TABLE 12 Estimated abundance of LPC leks with 90% confidence intervals ........................... 22
TABLE 13 Estimated densities of GPC and HPC, NW Kansas, with 90% confidence intervals 22
TABLE 14 Estimated population sizes of GPC and HPC with 90% confidence intervals ......... 22
TABLE 15 Estimated desities and abundance of GPC Leks with 90% confidence intervals ..... 22
3
EXECUTIVE SUMMARY

We estimated lesser prairie-chicken (Tympanuchus pallidicinctus; LPC) population sizes and
lek abundances for four ecoregions of Kansas, Colorado, New Mexico, Oklahoma, and Texas
in 2012, 2013, 2014, and 2015.

Two hundred forty-five grid cells surveyed in 2012 were resurveyed in 2013, 2014, and
2015. Thirty-eight additional cells were surveyed in 2013, 2014, and 2015. The same two
transects were flown in each cell each year. The same field survey and analysis methods
were used in 2012, 2013, 2014, and 2015.

Proportions of LPC, hybrid lesser-greater prairie chickens (HPC), and greater prairie
chickens (Tympanuchus cupido; GPC) in the Kansas portion of the Short Grass-CRP Prairie
Region (SGPR) was estimated using ground survey data collected from 2008 to 2013.

We detected 141 clusters of LPC, GPC, and HPC in 2012, 73 in 2013, 92 in 2014, and 136 in
2015 resulting in a relatively large pooled data set of 442 detections of prairie-chickens in the
aerial surveys. We estimated probability of detection in the survey strips using the pooled
data set which resulted in slight increases in estimates of LPC population sizes in 2012, 2013
and 2014 relative to estimates reported previously in McDonald et al. (2012, 2014a, 2014b).

Of the 442 detections of clusters of prairie-chickens in 2012, 2013, 2014, and 2015, 17%
were in crop-land, 66% in short-grass grassland, 3% in shinnery oak (including other shrub
dominated land), 5% in sand sagebrush prairie, and 10% in tall-grass grassland including
CRP grassland (with little or no shrubs).

We estimated the probability of detection of clusters of prairie-chickens using weighted
averages of distance sampling models scaled by the estimated probability of detection on the
inside edge of the field of view of the rear seat observers. We adjusted counts of LPC, GPC,
and HPC by covariate-specific, scaled, model-averaged probabilities of detection to estimate
population sizes in four ecoregions and the total study area.

We estimated the total population size of LPC to be 37,107 (90 % confidence interval [CI]:
28067, 57000) in 2012, 19,643 (90% CI: 12600, 27846) in 2013, 23,363 (90% CI: 16617,
32538) in 2014, and 29,162 (90% CI: 21661, 41017) in 2015. There was an estimated 25%
increase from 2014 to 2015, however the point estimate was not statistically significant (p >
0.2).

Increases in abundance of LPCH 2014 to 2015 were estimated in 3 of the 4 ecoregions. The
largest was a statistically significant 75% increase in the Sand Sage Prairie Region (SSPR) of
southeast Colorado, southwest Kansas and northern Panhandle of Oklahoma (p < 0.1). The
estimated increases were 30% in the Mixed Grass Prairie Region (MGPR) of northeast
Panhandle of Texas, northwest Oklahoma and south central Kansas, and 27% in the
Shortgrass-CRP Prairie Region (SGPR) of northwest Kansas, however the point estimates
were not statistically significant (p > 0.2).
4

We estimated the number of HPC in the SGPR to be 334 (90% CI: 200, 506) in 2012, 126
(90% CI: 50, 225) in 2013, 91 (90% CI: 60, 144) in 2014, and 248 (90% CI: 161, 371) in
2015.

The population size of GPC in the SGPR was estimated to be 33, 945 (90% CI: 23678,
48114) in 2012, 15,729 (90% CI: 10908, 20766) in 2013, 16,879 (90% CI: 13267, 25299) in
2014, and 23,241(90% CI: 17723, 33668) in 2015.

The 2012 drought was a likely contributing cause for the estimated decreases in population
sizes in spring 2013 in all four regions. The drought continued throughout much of the range
of the lesser prairie-chicken in spring 2013. In the face of the continuing drought of 2013,
estimated populations in the Short Grass-CRP Prairie Region (SGPR) of NW Kansas and
Mixed Grass Prairie Region (MGPR) of NE Texas, NW Oklahoma, and S Central Kansas
showed modest increases in spring 2014. Reproduction and survival in 2014 apparently
continued to improve in the SGPR of northwest Kansas, MGPR of northeast Texas,
northwest Oklahoma, and south central Kansas and SSPR of southeast Colorado, southwest
Kansas, and parts of the Panhandle of Oklahoma.

Ground surveys and counts of LPC on known lekking sites were conducted annually by
personnel of the wildlife departments of Colorado, Kansas, Oklahoma, New Mexico and
Texas. Ground surveys in the portion of the Shinnery Oak Prairie Region (SOPR) in western
Texas indicated a possible decrease in abundance of LPC from 2014 to 2015 (Brad Simpson,
personal communication); however, ground survey data collected in eastern New Mexico
indicated a possible increase in population size from 2014 to 2015 (Grant Beauprez, personal
communication). In this report, we reported an estimated decrease in abundance of LPC in
the Shinnery Oak Prairie Region (SOPR) from 2014 to 2015; however the decrease was not
statistically significant (p > 0.2). Maps of extreme drought and severe drought for 2014
(U.S. Drought Monitor Map Archive;
http://droughtmonitor.unl.edu/MapsAndData/MapArchive.aspx) indicated the intensity of the
drought in this region did not uniformly decrease until late summer or early fall 2014,
potentially contributed to low reproduction and survival of LPC in 2014 and contributed to
high variation in survey methods.
INTRODUCTION
Ascertaining precise estimates of wildlife population size (N) is valuable information for natural
resource agencies in the management of harvested and non-harvested species (Rabe et al. 2002).
Acquiring precise and unbiased estimates of population size required either a complete census or
probabilistic sample of subunits with which to infer population size (Johnson 2002); however,
limited funding and staffing often precluded implementation of these sampling designs. The
result was the development of population indices to monitor population trend or estimate a
minimum population size. The limitation of such data was their unknown relationship to
population size. Further, it must be assumed that indices tracked population dynamics
(McKelvey and Pearson 2001). These assumptions were problematic when knowing the
5
population size was critical to decision makers either in the context of harvest or population
recovery of sensitive species.
Our objectives were to implement consistent, statistically robust survey and analysis methods to
estimate lesser prairie-chicken (Tympanuchus pallidicinctus; LPC) population sizes and lek
abundances from 2012 to 2015. To achieve this, we had to address issues of regional variation as
well as the co-occurrence of greater prairie-chicken (Tympanuchus cupido; GPC) and of hybrid
lesser-greater prairie-chickens (HPC) in northwestern Kansas. We estimated LPC and lek
abundances for four ecoregions:
(1) Shinnery Oak Prairie Region (SOPR) located in eastern New Mexico-southwest
Texas Panhandle,
(2) Sand Sagebrush Prairie Region (SSPR) located in southeastern Coloradosouthwestern Kansas-western Oklahoma Panhandle,
(3) Mixed-Grass Prairie Region (MGPR) located in the northeast Texas Panhandlewestern Oklahoma-south central Kansas, and
(4) Short Grass Conservation Reserve Program (CRP) Prairie Region (SGPR) located in
northwestern Kansas (Figure 1).
We also estimated GPC and GPC lek abundances in the SGPR located in northwestern Kansas.
STUDY AREA
Our study area included the 2011 Estimated Occupied Range of LEPC as defined by the Lesser
Prairie-chicken Interstate Working Group and mapped in the Western Association of Fish and
Wildlife Agencies’ web site http://www.wafwa.org/documents/lpc/april2012/SGPCAT.pdf
(McDonald et al. 2012). In addition, we included habitats with relatively high probability of lek
occurrence in northwest Kansas as measured by the Western Governors Association Southern
Great Plains Crucial Habitat Assessment Tool http://www.ksre.ksu.edu/doc13895.ashx. The
study area for 2012, 2013, 2014, and 2015 was illustrated in Figure 1 where we indicated the grid
cells selected and not selected for survey in 2013, 2014, and 2015. The buffered areas
surrounding the sub-areas delineated an approximate 77.7 km (30 mi) buffer into which the
survey may be expanded in the future.
METHODS
Probabilistic Samples
We ranked 15- × 15-km grid cells in the study area from 1 to 536 by an equal probability
sampling procedure known as Generalized Random Tessellation Stratified (GRTS) sampling
6
(McDonald et al. 2012, 2014a, Stevens and Olsen 2004). Cells selected by the GRTS procedure
maintained a spatially balanced sample for aerial resources such that any contiguous subset, if
taken in order, was an equal probability sample of the target population.
The first 180 cells in the GRTS list were selected for survey in 2012. Funds became available for
survey of additional cells in two regions. Forty additional cells were selected in Kansas and 36
additional cells were selected in New Mexico and western Texas from their respective top ranked
cells in the GRTS list for a total sample size of 256 probabilistically selected cells for aerial
survey in 2012. This resulted in three strata with different sampling intensities. In addition, four
ecoregions were defined and superimposed on the study area during analysis of the results,
effectively re-stratified and ultimately resulted in six strata with different sampling intensities.
The SOPR and SGPR were individual strata with equal probability sampling in 2012: 75 of 123
cells surveyed in SOPR and 80 of 165 cells surveyed in SGPR. The SSPR had two strata in the
conditional design: one with 13 of 37 cells surveyed and one with 16 of 34 cells surveyed.
Ecoregion MGPR also had two strata in the conditional design: one with 35 of 100 cells
surveyed and one with 37 of 75 cells surveyed.
Prior to the 2013 survey, we pre-stratified the study area using the four ecoregions and selected
GRTS equal probability samples from each using the same GRTS list from 2012. This process
resulted in 245 grid cells resurveyed from 2012 and 38 new cells for a total sample size of 283
grid cells in 2013, 2014, and 2015. We flew the same transect lines in cells that were resurveyed
in 2013, 2014, and 2015.
7
FIGURE 1. Study area for 2012, 2013, 2014, and 2015 lesser prairie-chicken surveys illustrated with grid
cells selected for survey in 2013, 2014, and 2015. The colored areas surrounding the study subareas indicated an approximate 77.7 km (30 mi) buffer into which the survey may be expanded
in the future.
8
Aerial Survey Methods
The survey platform used for the surveys was the Raven II (R-44) (Robinson Helicopter
Company, Torrance, CA) helicopter accommodating two observers in the rear left and right
seats, and a third observer in the front left seat. Three helicopters and survey crews operated
simultaneously within the study area each year. Transects were flown north to south or south to
north at nominal values of 60 km per hour and 25 m above ground. Surveys were conducted
from sunrise until approximately 2.5 hours after sunrise during the lekking period: March 15 to
May 15.
Two 15-km north-south parallel transect were selected in each of the survey cells. The starting
point of the first transect was randomly located in the interval (200 m, 7,300 m) on the base of
the cell and the second transect was located 7,500 m to the right of the first transect.
Survey methods were the same in all four years of the survey and were described in detail in
McDonald et al. (2012)
STATISTICAL METHODS
Probability of Detection
We pooled data from 2012, 2013, 2014, and 2015 to estimate the probabilities of detection of
clusters of prairie-chickens because the numbers of detections within each year were limited and
the survey methods remained unchanged. Based on the pooled data set, we estimated population
sizes for 2015 and adjusted estimates of population sizes for 2012, 2013, and 2014. Estimates
for 2012, 2013, and 2014 were expected to differ slightly from results reported in McDonald et
al. (2012, 2014a, 2014b).
We used the R package Mark–Recapture Distance Sampling (mrds) in the R language and
environment (2.14.2; R Development Core Team 2012) and custom R code to fit multiple
covariate distance sampling detection models and conventional distance sampling detection
models for the estimated probability of detection of clusters of prairie-chickens. The estimates
of probability of detection were then scaled by the probability of detection near the transect line
to obtain the covariate-specific, scaled, model averaged probabilities of detection. Analysis and
modeling methods were reported in detail in McDonald et al. (2014a) and remain the same.
9
Estimation of Population Parameters in the Short Grass Prairie Region
The proportions of LPC, GPC, and HPC in the SGPR in NW Kansas were estimated using
ground survey data collected from 2008 through 2013. All ground survey data and initial data
processing were provided by the Kansas Department of Wildlife, Parks and Tourism and Kansas
Biological Survey (Jim Pitman and Mike Houts, personal communication).
Estimation of Precision of Estimated Population Parameters
We used bootstrapping techniques (Manly 2006) to estimate 90% confidence intervals for
density and population totals of LPC, HPC, and GPC individuals as well as leks within each
ecoregion. From each bootstrapped sample we generated new estimates of densities and
population totals. We calculated confidence intervals based on the central 90% of the bootstraped
distribution (the percentile method) for each estimated parameter.
RESULTS
Data collected from surveys from 2012 to 2015 were used to estimate the probability of detection
of clusters of prairie-chickens. Population size and density were estimated for 2015 and
estimates of population size and density were updated for 2012, 2013, and 2014. The data from
2012, 2013, 2014, and 2015 were pooled for estimation of probability of detection within a 600
meter (m) strip as the between year variation was minimal and pooling the data reduced the
variation in the data set (Figures 2 and 3).
10
FIGURE 2. Histograms of the perpendicular distance (m) from the transect line to the center of detected
clusters of prairie-chickens: 2012, 2013, 2014, and 2015.
11
FIGURE 3. Histogram of the perpendicular distance (m) from the
transect line to the center of detected clusters of prairiechickens pooled from 2012, 2013, 2014, and 2015.
We detected 141 clusters of LPC, GPC, and HPC in 2012, 73 in 2013, 92 in 2014, and 136 in
2015 while surveying on transects (i.e., within 300 m of the transect line) for a total of 442
detections of prairie-chickens in the combined data set (Table 1).
TABLE 1. Numbers and percent of detections of leks and non-lekking clusters of LPC, GPC, and
HPC by habitat type in the data sets for 2012, 2013, 2014, and 2015. Habitat types
were: CR = crop land, SGR = short-grass grassland, SO = shinnery oak (including
other shrub dominated land), SP = sand-sage prairie, TGR = tall-grass grassland
including CRP grassland (with little or no shrubs).
Habitat
CR
SGR
SO
SP
TGR
Total
2012
27 (19.1%)
91 (64.5%)
6 (4.3%)
3 (2.1%)
14 (9.9%)
141
2013
14 (19.2%)
49 (67.1%)
2 (2.7%)
7 (9.6%)
1 (1.4%)
73
2014
11 (12.0%)
66 (71.7%)
2 (2.2%)
2 (2.2%)
11 (12%)
92
2015
23 (16.9%)
85 (62.5%)
1 (0.7%)
10 (7.4%)
17 (12.5%)
136
Total
75 (17.0%)
291 (65.8%)
11 (2.5%)
22 (5.0%)
43 (9.7%)
442
12
Of the 442 prairie-chicken clusters detected in 2012, 2013, 2014, and 2015, 66% were in shortgrass grassland, 17% were in crop land, 10% were in tall-grass grassland including CRP
grassland (with little or no shrubs), 5% were in sand-sage prairie, and 3% were in shinnery oak
(including other shrub dominated land; Table 1).
We continued to use the observations of LPC, GPC, and HPC by the front left and back left
observers in “mark-recapture” models. For example, clusters of prairie-chickens seen by the
back left observer were “marked” and some of those clusters were “recaptured” by the front left
observer. These models were used to estimate the probability that at least one of the two
observers will detect a cluster given that it in the field of view of the back left observer (i.e.,
greater than 6.8 m from the transect line). The pooled data collected from 2012 through 2015
increases the sample sizes of the datasets to 162 and 163 for the front left and back left
observers, respectively (Table 2). We gave the detections equal weight for modeling the
components of the covariate specific, scaled, model averaged probability of detection on the
inside edge of the field of view of the back left observer.
TABLE 2. Sample sizes for logistic regression models to
estimate the probability that at least one of the
two observers will detect a cluster.
Year
2012
2013
2014
2015
Front Left
57
24
32
49
Back Left
50
28
39
46
Total
162
163
We dropped ten observations greater than 300 m from the transect line as they were outside the
viewshed defined in the survey protocol. Additionally, Buckland et al. (2001) recommended
dropping up to 5% of observations with the largest distances to the transect line to remove the
influence of outliers prior to modeling probability of detection. Data were grouped into 14
intervals for fitting models for probability of detection with the first interval spanning 0 to 40 m
and all subsequent intervals encompassing 20 m. The first interval was defined at 0 – 40 m in
order to compensate for potential errors in assigning distances near the transect line thus
avoiding artificial “spiking” of the detection probability on and close to the transect line.
Covariates used in the models for probability of detection were perpendicular distance to the
cluster (distance), cluster size (size), and the categorical variable habitat type (habitat; Tables 3
and 4). Due to the similarity of detection probability of prairie-chicken clusters in crop-land and
short-grassland, we combined those habitat types into one habitat category. The four levels
considered for habitat type were: short-grass/cropland, shinnery oak, sand-sage prairie, and tall13
grass grassland. Another categorical variable, flushed or not flushed, was not used in the models
in this report because of the very small number of observed clusters of prairie-chickens that were
not flushed. Weighted average estimates of probability of detection were obtained for
combinations of covariates associated with detections of leks and non-leks using model
averaging with the corrected Akaike Information Criterion (AICc).
TABLE 3.
Logistic Regression models used for estimation of probabilities of detection on the inside edge of
the field of view of the back left observers. Distance = perpendicular distance to detected
clusters, none = no covariates, size = size of cluster, and habitat = habitat type occupied. The
back left observer models estimated the probability that the back left observer detected a cluster
given that the cluster was detected by the front left observer. Similarly, the front left observer
models estimated the probability that the front left observer detected a cluster given that the
cluster was detected by the back left observer.
Back Left Observer Model
Covariates
AICc
distance
distance + size
None
size
habitat
distance + size + habitat
size + habitat
distance + habitat
221.94
221.96
222.06
222.25
226.74
226.76
226.79
226.87
Front Left Observer Model
Model
Weight
0.24
0.24
0.23
0.21
0.02
0.02
0.02
0.02
Covariates
AICc
distance + size + habitat
size + habitat
distance + size + habitat
habitat
distance + habitat
size + habitat
None
distance
214.40
214.65
216.64
216.76
217.18
217.32
220.24
220.39
Model
Weight
0.32
0.28
0.11
0.10
0.08
0.07
0.02
0.02
TABLE 4. Distance sampling models used to estimate probability of detection as a function of
distance from the transect line and other covariates. Distance to detected clusters was
in all models. Size = size of cluster, and habitat = habitat occupied by detected
clusters. Pooled data (n = 442) from 2012, 2013, 2014, and 2015 were used to fit the
distance sampling models. Key Functions were ne = negative exponential model, hr =
hazard rate, and hn = half normal.
Model Covariates
No Adj. Terms
No Adj. Terms
size
habitat
size + habitat
size
size + habitat
No Adj. Terms
habitat
Key Function
ne
hr
hr
hr
hr
hn
hn
hn
hn
AICc
1862.46
1869.19
1869.66
1873.69
1874.11
1884.81
1888.88
1889.26
1892.85
Model Weight
0.936
0.032
0.025
0.003
0.003
< 0.001
< 0.001
< 0.001
< 0.001
14
Number of LPC Detected and Average Cluster size: 2013-2015.
There were 368 LPC detected in 2012, 203 in 2013, 224 in 2014, and 275 in 2015 (Table 5).
Note that fewer cells were in the survey in 2012 (256 cells) while survey effort was increased to
283 cells in 2013, 2014, and 2015. While the number of LPC detected increased in 2015 from
2014, the average cluster size of LPC detected decreased slightly from 3.9 LPC per cluster to 3.3
LPC per cluster from 2014 to 2015 (Table 6).
TABLE 5.
Estimated numbers of lesser prairie-chickens (LPC) detected by ecoregion and overall in 2012,
2013, 2014, and 2015. “On transect” indicated observations were made between start and end
points of transects. “Off transect” indicated observations were made while traveling to and
from selected transect lines or outside of the transect viewshed (300 m from the transect). Two
hundred fifty six (256) cells were surveyed in 2012, 283 cells were flown in 2013, 2014, and 2015.
2012
Region
SOPR
SSPR
MGPR
SGPR
Total
TABLE 6.
Region
SOPR
SSPR
MGPR
SGPR
Overall
2013
2014
2015
on
transect
off
transect
on
transect
off
transect
on
transect
off
transect
on
transect
off
transect
44
22
86
216
368
7
6
0
16
29
24
35
39
105
203
12
5
4
12
33
17
8
70
129
224
10
7
2
9
28
10
14
86
165
275
7
13
19
9
48
Average cluster sizes of LPC detected by ecoregion and overall in 2012, 2013, 2014, and 2015.
2012
3.4
7.3
6.6
4.3
4.6
2013
2.4
5.8
5.6
4.9
4.7
2014
2.4
4.0
4.4
3.9
3.9
2015
1.4
1.8
3.0
3.7
3.3
Estimated Densities and Abundances of LPC and LPC Leks
We adjusted counts of LPC, GPC, and HPC by covariate specific, scaled, model averaged
probabilities of detection to estimate population sizes in four ecoregions and the study area
(Tables 7 and 8, and Figure 4). We estimated the total population size of LPC to be 37,107 (90
% confidence interval [CI]: 28067, 57000) in 2012, 19,643 (90% CI: 12600, 27846) in 2013,
23,363 (90% CI: 16617, 32538) in 2014, and 29,162 (90% CI: 21661, 41017) in 2015.
15
The estimated population size of 29,162 (90% CI: 21661, 41017) in 2015 was a 25% increase in the population size relative to 2014.
The overall increase of 5,799 (80% CI: -3105, 15525) LPC was not statistically significant at the 80% confidence level (p-value >
0.2), however the estimated increase of 377 (90% CI: 132, 1473) LPC in the SSPR of SE Colorado, SW Kansas, and Oklahoma
Panhandle was statistically significant at the 90% confidence level (p-value < 0.1; Tables 9 and 10).
TABLE 7. Estimated densities of LPC per 100 km2 by ecoregion and overall in 2012, 2013, 2014, and 2015 for the study area.
Bootstrapped 90% confidence intervals were reported on the densities of LPC per 100 km2.
Region
SOPR
SSPR
MGPR
SGPR
Overall
14.04
16.71
25.01
55.28
30.77
2012
(8.24, 53.19)
(8.68, 25.49)
(14.67, 38.85)
(34.52, 81.48)
(23.27, 47.26)
7.22
13.34
11.17
29.69
16.29
2013
(4.99, 11.03)
(8.38, 18.14)
(7.47, 14.87)
(12.30, 49.24)
(10.45, 23.09)
5.01
3.16
19.41
36.92
19.37
2014
(2.89, 8.52)
(2.57, 3.54)
(9.92, 31.15)
(21.86, 55.74)
(13.78, 26.98)
2.94
5.51
25.30
46.72
24.18
2015
(1.90, 4.64)
(3.94, 11.99)
(19.63, 35.26)
(27.00, 72.04)
(17.96, 34.01)
TABLE 8. Estimated population sizes of LPC by ecoregion and overall for 2012, 2013, 2014, and 2015 for the study area.
Bootstrapped 90% confidence intervals were reported on the population sizes of LPC.
Region
SOPR
SSPR
MGPR
SGPR
Overall
3,886
2,670
9,903
20,648
37,107
2012
(2280, 14719)
(1387, 4071)
(5809, 15386)
(12895, 30432)
(28067, 57000)
2013
1,999
(1382, 3051)
1,386
2,131
(1339, 2898)
504
4,424
(2957, 5887)
7,685
11,090 (4593, 18391) 13,788
19,643 (12600, 27846) 23,363
2014
(801, 2358)
(410, 565)
(3928, 12335)
(8165, 20821)
(16617, 32538)
814
881
10,019
17,448
29,162
2015
(526, 1283)
(630, 1915)
(7772, 13963)
(10083, 26909)
(21661, 41017)
16
TABLE 9. Estimated differences in population estimates for LPC between years with bootstrapped 80%
confidence intervals on the differences.
Region
SOPR
SSPR
MGPR
SGPR
Total
2013 minus 2012
-1,887
(-5654, -638)
-539
(-1846, 616)
-5,479
(-10055, -1969)
-9,558
(-19048, -1793)
-17,464 (-31564, -8743)
2014 minus 2013
-613
(-1472, 201)
-1,626 (-2151, -1019)
3,261
(163, 6951)
2,698
(-3626, 9743)
3,720
(-3915, 12156)
2015 minus 2014
-572
(-1248, 12)
377
(266, 1249)
2,334
(-1081, 6628)
3,660
(-4672, 11366)
5,799
(-3105, 15525)
TABLE 10. Estimated differences in population estimates for LPC between years with bootstrapped 90%
confidence intervals on the differences.
Region
SOPR
SSPR
MGPR
SGPR
Total
2013 minus 2012
-1,887
(-11952, -297)
-539
(-2262, 970)
-5,479
(-11239, -1017)
-9,558
(-21170, 502)
-17,464 (-38391, -6512)
2014 minus 2013
-613
(-1792, 453)
-1,626 (-2382, -856)
3,261
(-779, 8061)
2,698 (-5609, 11663)
3,720 (-6231, 14951)
2015 minus 2014
-572
(-1612, 192)
377
(132, 1473)
2,334
(-2590, 7883)
3,660
(-6445, 14283)
5,799
(-5177, 19089)
17
FIGURE 4. Estimated total population sizes of LPC in 2012, 2013, 2014, and 2015 for the study area.
SOPR = Shinnery Oak Prairie Region (E New Mexico, W Texas), SSPR = Sandsage Prairie
Region (SE Colorado, SW Kansas, Oklahoma Panhandle), MGPR = Mixed Grass Prairie Region
(NE Texas, NW Oklahoma, S central Kansas), and SGPR = Short Grass-CRP Prairie Region
(NW Kansas).
18
FIGURE 5a. Estimated total population sizes of LPC in 2012, 2013, 2014, and 2015 with 90% confidence
intervals for the study area, MGPR = Mixed Grass Prairie Region (NE Texas, NW Oklahoma,
S central Kansas), and SGPR = Short Grass-CRP Prairie Region (NW Kansas). Note that
scales on the vertical axes were different for Figures 5a and 5b.
19
FIGURE 5b. Estimated total population sizes of LPC in 2012, 2013, 2014, and 2015 with 90% confidence
intervals in the SOPR = Shinnery Oak Prairie Region (E New Mexico, W Texas) and SSPR =
Sandsage Prairie Region (SE Colorado, SW Kansas, Oklahoma Panhandle).
Note that scales on the vertical axes were different for Figures 5a and 5b.
20
We observed an approximate 25% decrease in the estimated abundance of LPC leks in 2015 relative to 2014. The abundance of LPC
leks was estimated to be 3,427 (90% CI: 2251, 5525) in 2012, 1,856 (90% CI: 994, 2753) in 2013, 2,419 (90% CI: 1531, 3621) in
2014, and 1,793 (90% CI: 1231, 2578) in 2015 (Table 12).
TABLE 11. Estimated densities of LPC leks per 100 km2 by ecoregion and overall in 2012, 2013, 2014, and 2015.
Bootstrapped 90% confidence intervals were reported on the densities of LPC leks per 100 km2.
Region
SOPR
SSPR
MGPR
SGPR
Overall
1.59
1.54
2.47
4.72
2.84
2012
(0.85, 5.30)
(0.54, 2.60)
(1.51, 3.77)
(1.98, 8.16)
(1.87, 4.58)
0.30
2.33
0.55
3.16
1.54
2013
(0.00, 0.93)
(1.99, 2.60)
(0.00, 1.04)
(1.07, 5.54)
(0.82, 2.28)
0.62
0.40
1.71
4.04
2.00
2014
(0.00, 1.79)
(0.00, 0.86)
(0.89, 2.62)
(2.03, 6.32)
(1.27, 3.00)
0.28
0.37
2.31
1.99
1.48
2015
(0.00, 0.79)
(0.00, 1.63)
(1.34, 3.33)
(0.77, 3.52)
(1.02, 2.14)
TABLE 12. Estimated abundances of LPC leks by ecoregion and overall for 2012, 2013, 2014, and 2015.
Bootstrapped 90% confidence intervals were reported on the abundances of LPC leks.
Region
SOPR
SSPR
MGPR
SGPR
Overall
439
246
977
1,765
3,427
2012
(234, 1468)
(87, 415)
(597, 1491)
(738, 3048)
(2251, 5525)
84
373
219
1,181
1,856
2013
(0, 257)
(318, 415)
(0, 412)
(398, 2070)
(994, 2753)
171
65
675
1,508
2,419
2014
(0, 495)
(0, 137)
(354, 1039)
(759, 2359)
(1531, 3621)
77
60
914
743
1,793
2015
(0, 219)
(0, 260)
(532, 1318)
(286, 1313)
(1231, 2578)
We estimated the number of HPC in the SGPR to be 334 (90% CI: 200, 506) in 2012, 126 (90% CI: 50, 225) in 2013, 91 (90% CI: 60,
144) in 2014, and 248 (90% CI: 161, 371) in 2015 (Table 14). The population size of the GPC in the SGPR was estimated to be 33,
945 (90% CI: 23678, 48114) in 2012, 15,729 (90% CI: 10908, 20766) in 2013, 16,879 (90% CI: 13267, 25299) in 2014, and
23,241(90% CI: 17723, 33668) in 2015 area (Table 14 & Figure 6).
21
TABLE 13. Estimates of GPC and HPC densities per 100 km2 for 2012, 2013, 2014, and 2015 in the SGPR habitat region of NW
Kansas in the study area. Bootstrapped 90% confidence intervals were reported on the densities of GPC and HPC per
100 km2.
Region
GPC
HPC
90.88
0.89
2012
(63.40, 128.82) 42.11
(0.53, 1.35)
0.34
2013
(29.20, 55.60)
(0.13, 0.60)
45.19
0.24
2014
(35.52, 67.73)
(0.16, 0.39)
62.22
0.66
2015
(47.45, 90.14)
(0.40, 0.99)
TABLE 14. Estimates of GPC and HPC population sizes for 2012, 2013, 2014, and 2015 in the SGPR habitat region of NW Kansas in
the study area. Bootstrapped 90% confidence intervals were reported on the population sizes of GPC and HPC.
Region
GPC
HPC
33,945
334
2012
2013
(23678, 48114) 15,729 (10908, 20766)
(200, 506)
126
(50, 225)
2014
16,879 (13267, 25299)
91
(60, 144)
23,241
248
2015
(17723, 33668)
(151, 371)
TABLE 15. Estimates of GPC lek densities per 100 km2 and abundances of GPC leks for 2012, 2013, 2014, 2015 in the SGPR habitat
region of NW Kansas. Bootstrapped 90% confidence intervals were reported on the population sizes of GPC and
abundances of GPC leks per 100 km2.
Region
Density
Abundance
5.74
2,144
2012
(4.01, 8.50)
(1497, 3176)
4.34
1,619
2013
(2.87, 6.15)
(1071, 2296)
5.04
1,882
2014
(3.68, 7.68)
(1374, 2870)
4.03
1,505
2015
(2.78, 6.64)
(1038, 2479)
22
FIGURE 6. Estimated population sizes of greater prairie-chickens (GPC) with 90% confidence intervals in
2012, 2013, and 2014 in the Short Grass-CRP Prairie Region (NW Kansas).
DISCUSSION
With four years of survey data, we were able to improve the models used to estimate the
probability of detection on the inside edge of the field of view of the back left observers and the
probability of detection as a function of distance from the transect line. These improved models
were used to not only estimate the population sizes in 2015, but also to re-estimate population
sizes in 2012, 2013, and 2014. Use of these models contributed to the slightly larger estimates
for 2012, 2013, and 2014 relative to values reported in McDonald et al. (2014a, 2014b).
23
The overall estimated population size increased approximately 25% from 2014 to 2015. An
increase in estimated total LPC from 2014 to 2015 was observed in the Mixed Grass Prairie
Region (MGPR), Sand Sage Prairie Region, and SGPR (Table 8, Figure 4).
Similar to previous reports, the estimated population size of 19,643 in 2013 was approximately
53% of the estimated population size of 37,107 in 2012, while the estimate of 23,363 in 2014
increased to approximately 63% of the initial value (Table 8, Figure 4). Additionally, the
estimated population size of 29,162 in 2015 was approximately 79% of the initial estimated
population size in 2012.
We estimated the population sizes of greater prairie-chickens in the SGPR (Table 14). The
estimated population size in 2013 decreased to 15,729 (approximately 46% of the 2012 estimate)
and increased to 16,879 in 2014 (approximately 50% of the 2012 estimate) from the initial value
of 33,945 in 2012. Additionally, the estimated population size in 2015 was approximately 68%
of the initial estimated population in 2012 (estimated population size of 23,241 in 2015).
It was well documented that there was an excessive drought throughout the southern great plains
in 2012 (e.g., National Climatic Data Center 2012). The 2012 drought was a likely contributing
cause for the estimated decrease in population sizes in all 4 regions in 2013. The drought
continued throughout much of the range of the lesser prairie-chicken in spring 2013 (e.g.
Figure 7, http://droughtmonitor.unl.edu/mapsanddata/maparchive.aspx; June 4, 2013). In the
face of the continuing drought, estimated populations in the SGPR of NW Kansas and Mixed
Grass Prairie Region of NE Texas, NW Oklahoma, and S Central Kansas showed modest
increases in population sizes in 2014.
Reproduction in 2014 and survival apparently continued to improve in the SGPR of northwest
Kansas and MGPR of northeast Texas, northwest Oklahoma, and south central Kansas giving
rise to 27% and 30% increases respectively in 2015 compared to 2014, however the estimated
increases were not statically significant (p > 0.2). Similarly, the SSPR of southeast Colorado,
southwest Kansas, and parts of the Oklahoma Panhandle had an estimated statistically significant
75% increase in 2015 compared to 2014 (p < 0.1).
Ground surveys and counts of LPC on known lekking sites were conducted annually by
personnel of the wildlife departments of Colorado, Kansas, Oklahoma, New Mexico and Texas.
Ground surveys in the portion of the Shinnery Oak Prairie Region (SOPR) in western Texas
indicated a possible decrease in abundance of LPC from 2014 to 2015 (Brad Simpson, personal
communication); however, ground survey data collected in eastern New Mexico indicated a
possible increase in population size from 2014 to 2015 (Grant Beauprez, personal
communication). In this report, we estimated a decrease in abundance of LPC in the Shinnery
Oak Prairie Region (SOPR) from 2014 to 2015; however the decrease was not statistically
24
significant (p > 0.2). Maps of extreme drought and severe drought for 2014 (U.S. Drought
Monitor Map Archive; http://droughtmonitor.unl.edu/MapsAndData/MapArchive.aspx) indicate
the intensity of the drought in this region did not uniformly decrease until late summer or early
fall 2014, potentially contributed to low reproduction and survival of LPC in 2014 and
contributed to high variation in survey methods.
FIGURE 7. Map produced by U.S. Drought Monitor, June 4, 2013.
ACKNOWLEDGEMENTS
The 2012 survey and reports were accomplished by the financial support of the Great Plains
Landscape Conservation Cooperative, the Bureau of Land Management, and the National Fish
and Wildlife Foundation through grants to the Western Association of Fish and Wildlife
Agencies. Financial support for the 2013, 2014, and 2015 surveys and this report were provided
by grants to the Western Association of Fish and Wildlife Agencies from the wildlife agencies of
25
the states of Colorado, Kansas, New Mexico, Oklahoma, and Texas, the Bureau of Land
Management, the National Fish and Wildlife Foundation, and various oil and gas companies and
associations. The recommended study design and methods were developed with the assistance
of the following members of the Lesser Prairie-chicken Interstate Working Group: Bill Van Pelt,
WAFWA Grassland Coordinator, Arizona Game and Fish Department; Jim Pitman, Kansas
Department of Wildlife, Parks and Tourism; Sean Kyle, Texas Parks and Wildlife Department,
David Klute, Colorado Division of Parks and Wildlife; Grant Beauprez, New Mexico
Department of Game and Fish; and Doug Schoeling, Oklahoma Department of Wildlife
Conservation.
We wish to acknowledge our use of the ground survey data collected by the Kansas Department
of Wildlife, Parks and Tourism and the initial analysis of these data provided by Michael Houts,
GIS/Remote Sensing Specialist, Kansas Biological Survey. We also wish to acknowledge the
assistance of the aerial survey crew members and pilots.
LITERATURE CITED
Buckland, S.T., D.R. Anderson, K.P. Burnham, J.L. Laake, D.L. Borchers, and L. Thomas. 2001.
Introduction to Distance Sampling, Oxford: Oxford University Press.
Johnson, D.H. 2002. The Importance of Replication in Wildlife Research. Journal of Wildlife
Management 66: 919-932.
Manly, B.J.F. 2006. Randomization, Bootstrap and Monte Carlo Methods in Biology. Third
Edition. Chapman and Hall, London.
McDonald, L.L., J. Griswold, T. Rintz, and G. Gardner. 2012. Results of the 2012 Range-Wide
Survey of Lesser Prairie-Chickens (Tympanuchus pallidicinctus). Technical Report
prepared for the Western Association of Fish and Wildlife Agencies. Western
EcoSystems Technology, Inc. (WEST), 200 South 2nd Street, Laramie, Wyoming,
82070.
McDonald, L.L., G. Beauprez, G. Gardner, J. Griswold, C. Hagen, F. Hornsby, D. Klute, S.
Kyle, J. Pitman, T. Rintz, D. Schoeling, B. Van Pelt. 2014a. Range-wide population size
of the lesser prairie-chicken: 2012 and 2013. Wildlife Society Bulletin 38:536–546.
McDonald, L.L., K. Adachi, T. Rintz, G. Gardner, and F. Hornsby. 2014b. Range-wide
population size of the lesser prairie-chicken: 2012, 2013 and 2014. Technical Report
prepared for the Western Association of Fish and Wildlife Agencies. Western
EcoSystems Technology, Inc. (WEST), 200 South 2nd Street, Laramie, Wyoming,
82070.
26
McKelvey, K.S., and D.E. Pearson. 2001. Population Estimation with Sparse Data: the Role of
Estimators Versus Indices Revisited. Canadian Journal of Zoology 79: 1754-1765.
National Climatic Data Center. 2012. NCDC Releases June 2012 U.S. Monthly Climate Report:
U.S. Experiences Warm and Dry June; Drought Expands to 56% of Lower 48.
http://www.ncdc.noaa.gov/news
R Development Core Team. 2013. R: A Language and Environment for Statistical Computing. R
Foundation for Statistical Computing, Vienna, Austria.
Rabe, M.J., S.S. Rosenstock, and J.C. deVos, Jr. 2002. Review of Big-Game Survey Methods
Used by Wildlife Agencies of the Western United States. Wildlife Society Bulletin 30:
46-52.
Stevens, D.L., Jr., and A.R. Olsen. 2004. Spatially Balanced Sampling of Natural Resources.
Journal of the American Statistical Association 99: 262-278.
27