Inventory and monitoring of Western Burrowing Owls for the

Inventory and
monitoring of
Western Burrowing
Owls for the
Coachella Valley
MSHCP
Final Report
July – December 2009
J. T. Rotenberry, Ph.D.
Project Director
K. D. Fleming
Field Biologist
Q. Latif, Ph.D.
Post Doctorate Researcher
R. Johnson
GIS Specialist
C.W. Barrows, Ph.D.
co Principal Investigator
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. ii
TABLE OF CONTENTS
Table of Contents ................................................................................................................................ ii
Table of Figures .................................................................................................................................. iii
Introduction ......................................................................................................................................... 1
Methods ................................................................................................................................................ 2
Field Surveys .......................................................................................................................................... 2
Database Development ......................................................................................................................... 4
Niche Modeling ...................................................................................................................................... 4
Habitat variables .................................................................................................................................... 5
Occupancy Modeling............................................................................................................................. 6
Results ................................................................................................................................................... 8
Survey Results ........................................................................................................................................ 8
Niche Modeling Results ...................................................................................................................... 10
Occupancy Models Result .................................................................................................................. 12
Niche Model-Occupancy Model Comparisons ............................................................................... 13
Discussion ..........................................................................................................................................14
Literature Cited ..................................................................................................................................18
Appendix A - Burrowing Owl (Athene cunnicularia) Survey Protocol ..........................................20
Recommended survey protocol ......................................................................................................... 20
Research Questions .............................................................................................................................. 24
Data and Analyses ............................................................................................................................... 25
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. iii
TABLE OF FIGURES
Figure 1. Burrowing Owl survey locations implemented by the CCB in 2009. Black squares
indicate wildland plots, and green lines depict survey routes. ........................................................... 3
Figure 2. Distribution of burrowing owls located during the 2009 breeding season. ...................... 9
Figure 3. Comparison of detection rates (proportion of survey attempts during which owls were
detected in sites where owls were known to occur) among three survey methods on two survey
routes known to be occupied by burrowing owls during the breeding season. Vertical lines
indicate one standard error. ................................................................................................................... 10
Figure 4. Comparisons of survey method effectiveness during winter and breeding season
surveys. Graph on the left shows total detections whereas the graph on the right shows only
those detections that were independent of detections made during linear surveys. ..................... 10
Figure 5. Mapped representation of the niche modeling results within the Coachella Valley.
Areas indicated in pale yellow had HSI values ≥ 0.55, in orange had HSI values ≥ 0.75, and in
dark brown had HSI values ≥ 0.95........................................................................................................ 11
Figure 6. Mapped representation of predicted occurrence probabilities from the occupancy
model best-fitting breeding survey data. Probabilities are depicted on a scale from beige (low
probability) to green (high probability). Brown indicates cells that were not sampled and too
dissimilar in environmental space from sampled cells to estimate occupancy rates based on the
fitted models. ............................................................................................................................................ 13
Figure 7. Our best performing Occupancy Model (yellow shaded areas) overlain upon our best
performing Niche Model (orange shaded areas), HSI/Psi_hat ≥ 0.75. Red circles indicate all
documented burrowing owl locations, historic and recent, utilized in this study. ........................ 14
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 1
INTRODUCTION
Western burrowing owl (Athene cunicularia hypugaea) populations that inhabit the southern part
of North America from northern Mexico through Texas and California tend to not migrate in
significant numbers but instead winter as resident birds (Korfanta et. al. 2005). Burrowing owls
are currently classified as a “species of special concern” by the state of California due to losses
in habitat and other anthropogenic impacts which have led to its widespread decline through
the western United States (James and Espie 1997). Because they are a ground dwelling species,
burrowing owls are threatened by extensive anthropogenic habitat modification including
annual disking of fallow fields, as well as vegetation and pest management in agricultural flood
control channels and suburban development. Other threats include vehicle collisions, predation
from domestic pets and coyotes, and off road vehicle use that can collapse burrows.
Prior to this study, of the 74 known locations within the Coachella Valley Multiple Species
Habitat Conservation Plan (CVMSHCP) area, 41 burrowing owls were previously found within
the proposed reserve system (CVMSHCP 2007). Volunteer based surveys conducted by the
Institute for Bird Populations during the 1991-1993 years reported at least 9000 breeding pairs in
California with the majority located in agricultural areas of the Imperial and Central Valleys.
DeSante et al. (USFWS 2003) believed when comparing these numbers to surveys done in the
1980s that burrowing owls had been extirpated from many regions throughout California
including the Coachella Valley. Despite premature predictions of their demise, burrowing owls
are found in a variety of habitats within the Coachella Valley including adjacent to suburban
and urban development, washes, fallow fields, sand dunes, agricultural drains, and creosote
dominated landscapes. This generalist habitat character makes it difficult to use any specific
vegetation types to classify suitable habitat, especially when compared to other features such as
soil types and the presence of occupiable burrows. While broadly distributed in the Coachella
Valley, this species is uncommon and the trend in its population is unknown. As such it is
among the 27 covered, focal species of the CVMSHCP.
The objectives for burrowing owls within the CVMSHCP are to maintain and ensure
conservation of occupied burrows on current conserved lands, as well as identify and conserve
other lands, minimize harmful effects to the species, and finally to identify and implement
monitoring and management to sustain the population within the plan area (CVMSHCP 2007).
In support of these objectives, UC Riverside’s Center for Conservation Biology’s (CCB) tasks
included:
Task 1: Develop a reliable methodology for monitoring burrowing owls throughout the Plan
area. These protocols should be appropriate for the multiple scales of resolution associated with
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 2
the distribution of owls throughout the Plan area, ranging from historical or known nesting
sites to designated Conservation Areas of widely differing sizes to the regional network of
dikes, berms, and levees. Moreover, the methods must be statistically robust, with sufficient
statistical power to detect meaningful changes in owl distributions through time or in response
to management actions. These methods should be portable to other areas of similar scales, and
to other species.
Task 2: Use these protocols to document the distribution and abundance of owls throughout the
Plan area during the breeding season, initially focusing on the berms, levees, and dikes for
which we expect to obtain winter distributional data.
Task 3: Develop two complementary types of statistical models describing species habitat
relationships: site occupancy models (MacKinzie et al. 2002, 2006) and ecological niche models
(Rotenberry et al. 2006). Robust models will allow us to make inferences about potential habitat
for burrowing owls in other regions, as well as provide templates for constructing similar
models for other species.
Based upon these objectives, specific deliverables included:
•
A burrowing owl survey protocol tailored to the Coachella Valley landscape.
•
A geodatabase and template for data analyses.
•
A spatially explicit niche model describing the distribution of potential suitable
habitat for burrowing owls in the Coachella Valley.
•
A spatially explicit occupancy model based on the distribution and detectability of
burrowing owls in the Coachella Valley in 2009.
METHODS
Field Surveys
We concentrated our efforts on sampling along 42 routes within the 170,295 ha (420,629 ac.) area
of the Coachella Valley of Riverside County, California (Figure 1). To adequately sample across
a diverse landscape, routes were identified with lengths between 0.5 km and 40 km along
roadways that included rural, suburban, and urban landscapes as well as irrigation/flood
control areas. Our surveys were split into periods of winter 08-09 (January – March 2009),
breeding season (April – August 2009) and winter 09-10 (September- December 2009). Three
survey methods were evaluated on each of the survey routes: linear, point counts and audio-
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 3
playback point counts. Each route was surveyed at least three times to ensure the 95%
probability of detection if owls were present, using at least one of the survey methods during
the winter 08-09 period (Conway and Simon 2003). During both the breeding season and winter
09-10 surveys, each route was surveyed at least three times, with multiple visits per survey
method to each route in each season.
Figure 1. Burrowing Owl survey locations implemented by the CCB in 2009. Black squares indicate
wildland plots, and green lines depict survey routes.
Linear surveys were accomplished by driving the routes at a speed below 20 km/hour (15 mi/hr)
while scanning roadside habitat continuously, and stopping only when an owl was detected.
Visual detections for these surveys were limited to the line of sight from the roadway, and
binoculars were used at locations where owls were observed so as to count the number of
individuals at that location. Along certain storm-water drainage routes and riverbeds, where
driving the route was impossible, surveys were completed on foot. These routes include
Mission Creek, Coachella Valley (Whitewater) Storm-water Channel, Buchanan Street Drain,
San Gorgonio River, and the Little Morongo Wash.
Point counts and audio point counts were conducted along these same routes at 800 m (0.5 mi)
intervals. Data was collected at these established points, where we conducted a four-minute
visual search for owls throughout the season. At each stop, the observer would scan the
landscape for owls using binoculars for four minutes from the roadside. Audio point counts
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 4
were conducted in the same manner as regular point counts at these same points, only with a
field speaker and iPod that looped burrowing owl calls over a four minute playback period. For
these, two burrowing owl calls originating from observations at the Salton Sea were
downloaded onto an iPOD and broadcast with an amplified field speaker at a volume adjusted
to adequately cover a 400 meter radius in all directions from the call. The overall broadcast
lasted four minutes and consisted of 12 coo-coo calls (common male call) lasting 30 seconds
followed by 30 seconds of silence repeatedly over three minutes. One full minute was then
broadcast of the defensive (chuck and chatter) call followed by a full minute of silence for
observation. Wildland plot counts were taken by walking along previous established plots
within the Coachella Valley Preserve for 10 x 100 meters and recording all tracks, visual, and
audio observations that occur along the transect and within visual and audible distance.
Database Development
Database development for the project provides a record of fieldwork activity and findings and
provides support for analysis and reporting. It offers a medium that can incorporate the mix of
spatial, temporal, and tabular descriptive information collected during the year. All survey
fieldwork locations and owl observation are based on point locations that were manually
transcribed from GPS-measured coordinate values. All fieldwork activity and observations
include a location, date, and other descriptive attributes that were initially entered in Excel
worksheets. These in turn were imported into an Access relational database management
system as tables where domain restrictions could be applied to numeric, temporal, and
descriptive attribute values. Queries were used to build tabular data subsets based on field
methods, observation results, location, and time to meet reporting, modeling, and cartographic
requirements. All point locations and their attributes were initially converted to shape files and
used as reference for creation of linear transect alignments and the plot area locations (still
represented as a point). The tabular data and representative geometry were subsequently
imported into an ESRI-format file geodatabase.
Niche Modeling
We used the Mahalanobis distance statistic (D2) (Clark et al. 1993; Rotenberry et al. 2002, 2006;
Browning et al. 2005) to model the distribution of available suitable habitat for the burrowing
owl in the Coachella Valley. The Mahalanobis statistic yields for any location an index of its
habitat similarity (HSI) to the multivariate mean of the habitat characteristics at the target
species’ locations (the calibration data set). This statistic has several advantages over other GIS
modeling approaches, the foremost being that only species-presence data are required for the
dependent variable. Because only positive occurrence data are required, historic location
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 5
records from museums and field notes can be used, regardless of survey methodology, as long
as there is sufficient precision in the site location. This also avoids the uncertain assumption of
correct identification of unoccupied habitats (Knick and Rotenberry 1998, Rotenberry et al. 2002,
Browning et al. 2005), an assumption that becomes even more difficult to defend when trying to
model historic distributions.
The Mahalanobis statistic may be further refined by partitioning it into separate, additive
components (Dunn and Duncan 2000; Rotenberry et al. 2002, 2006). This partitioning is based on
an eigen analysis (principal components) of the variables and observations comprising the
calibration data set. The partition or component with the smallest eigenvalue is associated with
the combination of habitat variables that has the least variation among locations, often
indicating minimum habitat requirements. This approach is based on the assumption that the
full range of habitat variation has been captured in the location data, and that variables that
have low variance are more likely to represent limits to a species’ distribution than those that
take on a wide range of values where a species is present. Identifying the variables that
demonstrate the least variability may be more appropriate for modeling potential or historic
distributions in changing environments (Dunn and Duncan 2000, Rotenberry et al. 2002, 2006).
We calculated Mahalanobis distances and their partitions with SAS code provided in
Rotenberry et al. (2006).
The calibration data set consisted of 67 burrowing owl locations recorded incidentally by CCB
researchers while conducting other species surveys during 2003 - 2008, and from historic
burrowing owl locations provided by the Coachella Valley Association of Governments. The
relatively low number of historic owl locations was insufficient to randomly extract a subset for
validation purposes, so only a calibration data set was used.
Habitat variables
Independent variables were derived from existing GIS layers, or were calculated from
algorithms that convert Digital Elevation Models (DEMs) to topographic metrics and average
annual precipitation (normal precipitation 1971 – 2000, http://prism.oregonstate.edu/docs/meta/
ppt_30s_meta.htm, PRISM Group, Oregon State University). To avoid model over-fitting, we
maintained a variables-to-observations ratio of approximately 1:10 (one variable per 9 to 10
observations). Forty independent observations were used as a minimum threshold for
modeling, allowing the inclusion of at least four independent variables for the creation of each
niche model. Only those observations that were spatially independent were included; no more
than one observation per 180 x 180 m cell was used to construct the niche model. This spatial
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 6
independence helped avoid biases or over-weighting observations from easy access or
especially high visitation areas.
Variables used to model burrowing owl suitable habitat included the median terrain slope
value for 18 x 18 neighborhood of 10m cells (Slope), mean maximum temperatures in July from
1971-2000 (Max Temperature July), mean annual precipitation from 1971-2000 (Mean
Precipitation), median value for 18 x 18 of Sappington analysis results from 3x3 10m
neighborhood (Ruggedness 3x3), Sappington analysis for 18 x 18 10m neighborhood
(Ruggedness 20x20), slope curvature (median terrain curvature value for 18 x 18 neighborhood
of 10m cells), and average soil water content as fraction of volume (Sappington et al. 2007).
Variables considered but rejected when their addition failed to improve the model’s median
HSI value included percent sand-soil composition, percent clay-soil composition, and
vegetation type.
Once a model was created, it was used to calculate HSIs for each Mahalanobis distance partition
for every cell on the map. Following Rotenberry et al., (2006), HSI was rescaled to range from 01, with 0 being the most dissimilar and 1 being the most similar to the mean habitat
characteristics of the target species based on the calibration data set. ArcGIS 9.1 (ESRI 2005) was
used to provide a spatial model (niche map) of the similarity to the species mean for each cell.
These modeled areas were then screened for anthropogenic changes to the soil surface with GIS
layers for agricultural development and urban-suburban to derive an estimate of current
suitable habitat availability.
Occupancy Modeling
We developed single-season, single-species occupancy models (MacKenzie et al. 2002, 2006)
with the intention of calculating spatially explicit probabilities of the occurrence of burrowing
owls across the Coachella Valley. Generally, models of this class are designed to analyze
presence-absence data generated from surveying sites of which some substantial portion are
visited more than once, and the probability of detecting a species at a site given its presence
(detection probability) is < 1. Underlying assumptions include constant occupancy status of
individual sites across surveys and no un-modeled heterogeneity in either occupancy or
detection probabilities across sites and surveys. An occupancy model contains both a logistic
model describing occupancy probabilities and a second logistic model describing detection
probabilities, where detection is dependent on a site's occupancy status. As such, models can
include covariates that are explicitly related to environmental characteristics, detectability, or
both. Multiple models with various combinations of covariates can be fitted to the data and
compared under an information-theoretic framework (Burnham and Anderson 2002). From the
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 7
best-fitted models, one can calculate occupancy rates for both surveyed and non-surveyed sites
corrected for imperfect detection and from these infer a species’ range within the study area.
We fitted occupancy models to the data collected from surveys for burrowing owls described
previously. Within a GIS framework (ArcGIS 9.3), we assumed 540×540km cells (approximately
the size of an individual owl’s home range; [Rosenberg and Haley 2004]) to be the “sites”
surveyed. The data consisted of either a 0 (non-detect) or 1 (owl detection) for each day for each
cell that was sampled by a survey. Occupancy models included the following covariates for
detection probabilities: (1) the survey method used, (2) the area of the cell within the detection
range of the surveyed location (linear survey=130m, point=400m [based on post-hoc
examination of the distributions of distances between survey locations and owl locations], or
the plot area intersected with the cell area), and (3) interactions between method and area
(Method × Area) Date (day-of-year of surveys) and Date + Date2 (the quadratic transformation
of date). We considered 12 environmental covariates of occupancy probabilities: elevation,
slope, terrain curvature, maximum July temperature, minimum January temperature, average
precipitation, Ruggedness 3x3, average soil water content, road density, % aeolian sand, %
development, and % agriculture. Additional variables (described previously) were highly
correlated with one or more of these variables and therefore not considered. In recognition of
two distinct seasons in the burrowing owl life cycle, we divided the data by season (breeding
season = April 1 – August 31; non-breeding season = Winter 08-09 – Winter 09-10). In this report,
we only describe occupancy models fitted to the breeding season data (future analyses will
consider the wintering range).
Initially, we fitted models that included all combinations of detection covariates assuming
constant occupancy probability (using Presence 2.0). Subsequent models fitted to the data
included all covariates in the best-fit detection model and various combinations of occupancy
covariates. These included a global model, which included all 12 environmental covariates, and
a model with only the six variables identified as important in niche models (see Niche modeling
results; slope, curvature, maximum temperature, Ruggedness 3x3, and soil water content).
Additional models either added or subtracted variables from these. To assess the value of
occupancy models for predicting burrowing owl occurrence, we conducted a goodness-fit-test
on the global model (Presence 2.0), which provides an index of overdispersion of residual count
data (c = χ2data/ χ2BS; BS = boot-strapped mean; MacKenzie et al. 2006).
Using the best-fit occupancy model, we extrapolated the range of the burrowing owl during the
breeding season within the Coachella Valley during 2009. Occupancy probabilities were
estimated using model parameters applied to cell-specific scores for environmental variables
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 8
(MacKenzie et al. 2002, 2006). We only estimated occupancy probabilities for non-surveyed cells
that were similar in multivariate, environmental space to the surveyed cells.
RESULTS
Survey Results
In 2009 we found 53 locations occupied by burrowing owls across the Coachella Valley, with a
majority (65%) occurring within the northwestern valley within the Desert Hot Springs area,
especially along the Mission Creek and Little Morongo washes (Figure 2). Based on data from
locations where there were multiple surveys and where owls were detected at least once by at
least one of the methods, we found no differences in probabilities of detection among the
methods. Winter survey 08-09 detectability varied from 0.50-0.66 (proportion of survey attempts
during which owls were detected in sites where owls were known to occur) across all survey
methods, whereas breeding season survey detectability varied from 0.55-0.56 across methods.
Although our detection rates were generally similar across survey methods, some difference in
detection rates was measured when comparing methods within individual routes (Figure 3).
Also within the breeding season detection rates for the wildland plot counts (0.30) differed from
rates for other survey methods, even though winter counts resulted in a similar detection rate
(0.60) as other survey methods. Additional criteria used to select a preferred survey method
included overall numbers of owls detected and coverage of the habitats included in the
Coachella Valley.
Our preliminary results indicate that audio point counts and linear surveys detected roughly
the same number of owls in winter 08-09 (eight and nine respectively) (Figure 4A), although
during the breeding season audio counts appeared much less effective. Linear surveys detected
39 active burrows whereas the audio point counts yielded just nine, only one of which had not
been detected with the linear surveys (Figure 4B). In winter the two methods detected different
owls with only three of the detections being at the same active burrows. Winter point counts
(without audio) yielded just three detections (two of which were not detected by either the
linear or audio point counts), as did plot counts (all of which were undetected by any previous
method), for a total winter 08-09 count of 19 independent occupied locations. Breeding season
point counts also detected nine active burrows (three of which had not been detected by other
methods) and wildland plot counts detected five active burrows, all of which were undetected
by any previous method. Total detections of independent occupied locations during the
breeding season were 48.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 9
These results indicate that linear surveys conducted during the breeding season provide the
greatest number of burrowing owl detections (89% of the total detections). Furthermore, since
they are continuous rather than point based, linear surveys cover the most area of the four
survey methods. Although some of the linear surveys were conducted on foot along flood
control levees, the vast majority were conducted using vehicles along passable roadways,
allowing coverage of large areas of the Coachella Valley floor. Because wildland reserves are
largely roadless, on-foot surveys are necessary in these areas. A previous analysis reported that
foot surveys were of minimal value as the owls avoided detection at the approach of people
(Conway and Simon 2003). However, our survey protocols allow owls to be detected from their
distinctive tracks in loose aeolian sand, so owl avoidance of surveyors was not an issue.
Figure 2. Distribution of burrowing owls located during the 2009 breeding season.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 10
Mean Detection Rates at Known Occupied Sites
1
Coachella Valley Storm Channel
0.9
Suburban Desert Hot Springs
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Linear Surveys
Audio Point Counts
Point Counts
Figure 3. Comparison of detection rates (proportion of survey attempts during which owls were detected
in sites where owls were known to occur) among three survey methods on two survey routes known to
45
8
40
35
Winter
Breeding
30
25
20
15
10
5
0
Number of Active Burrowing Owl Locations Detected
Number of Occupied Burrowing Owl Locations Detected
be occupied by burrowing owls during the breeding season. Vertical lines indicate one standard error.
7
Winter
Breeding
6
5
4
3
2
1
0
Linear Surveys
Audio Point Counts
Point Counts
Wildland Plots
Surveys
Audio Point Counts
Point Counts
Wildland Plots
Surveys
Figure 4. Comparisons of survey method effectiveness during winter and breeding season surveys.
Graph on the left shows total detections whereas the graph on the right shows only those detections that
were independent of detections made during linear surveys.
Niche Modeling Results
A niche model describing the potential suitability of the Coachella Valley for burrowing owl
occupancy was developed from historic burrowing owl locations (Figure 5). The best
performing niche model, based on the highest median HSI value (0.908), was the model
partition that accounted for all the variance in the calibration data set and was comprised of
variables that described slope, slope curvature, slope ruggedness, mean maximum July
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 11
temperature, and mean precipitation. Those variables described a multivariate mean condition
for historic locations, and included a mean precipitation of 126 mm, a mean maximum July
temperature of 41°C (both moderate for the Coachella Valley), two ruggedness variables, slope,
soil water and curve, all consistent with a preference for flat terrain on gentle well drained
south-facing slopes. All these variable values were consistent with what is known of burrowing
owl habitat preferences.
The mapped niche model can be depicted as the distribution of cells that have HSI values at or
exceeding any a priori value, with the understanding that HSI values approaching 0 have little
or no similarity to conditions of owl locations used to create the calibration model, and those
approaching 1.0 would be almost identical to the calibration locations. In Figure 5, we show
niche models with HSI values ≥ to 0.55, 0.75, and 0.95. The area mapped as suitable with HSI
values ≥ 0.55 was 88,800 ha; for HSI values ≥ to 0.75 there were 73,800 ha and for HSI values to
0.95 there were 32,565 ha. Seventy-four percent of the burrowing owls located during the 2009
surveys occurred on lands mapped as having HSI values ≥ 0.55. The lack of a closer fit between
owl occurrence and modeled habitat is likely a product of a lack of a strong dependence of owls
on the environmental variables available to us for use in modeling.
Figure 5. Mapped representation of the niche modeling results within the Coachella Valley. Areas
indicated in pale yellow had HSI values ≥ 0.55, in orange had HSI values ≥ 0.75, and in dark brown had
HSI values ≥ 0.95.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 12
Occupancy Models Result
Selected occupancy models were similar to selected niche models, both in the environmental
parameters indicated as important to owls as well as in their reduced precision in predicting
owl occurrence. Of models with detection-only covariates, the best-fit model included all
detection covariates (∆AIC = -83.9 over a constant detection and constant occupancy model). Of
models that included all the detection covariates and occupancy covariates, the best-fit model
included most of the environmental parameters contained in best-performing niche models (all
except precipitation). These included negative effects of slope, curvature, maximum
temperature, and soil water, as well as a positive effect of ruggedness ( ∆AIC =-6.1 relative to
the best-fit detection-only model). These results should be considered preliminary since we did
not try all possible combinations of occupancy covariates since we limited ourselves to use the
variables used in the niche modeling process.
Future modeling exercises could produce better-fitting models by examining all possible
combinations of covariates. However, also similar to niche modeling results, occupancy models
did not provide very precise predictions of owl occurrence. The over-dispersion score for the
global model was high (c = 6.5; c > 1 indicates over-dispersion and c > 4 indicates extremely poor
fit; Burnham and Anderson 2002), indicating the suite of covariates considered here to be
inadequate for developing a reasonably predictive model. Consequently, predictions derived
from better-fitting models with unexamined covariate combinations are unlikely to be much
more than those of the best-fit model found here.
A map of the estimated occupancy probabilities based on our best-fit model is depicted in
Figure 6. Of note are the isolated pockets of high probability of occupancy cells (green)
surrounded by mid-to-low-occupancy cells (orange/beige). Cells in which owls were observed
were always assigned a high-occupancy status, but cells surrounding occupied cells were often
dissimilar from the average occupied cell, resulting in the isolated high-occupancy cells. Also, of
note is the “ring” of high-occupancy cells surrounding much of the region surveyed for
burrowing owls. We are uncertain of whether this phenomenon represents an artifact of model
structure or an actual biological process (e.g., owl selection for specific slope, elevation, or
ruggedness). Finally, a large portion of the Coachella Valley was dissimilar in environmental
space from surveyed areas (depicted in brown). Future attempts to model owl occupancy rates
may benefit from surveys of some of these areas.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 13
Figure 6. Mapped representation of predicted occurrence probabilities from the occupancy model bestfitting breeding survey data. Probabilities are depicted on a scale from beige (low probability) to green
(high probability). Brown indicates cells that were not sampled and too dissimilar in environmental space
from sampled cells to estimate occupancy rates based on the fitted models.
Niche Model-Occupancy Model Comparisons
The niche and occupancy models utilize different statistical approaches. The distribution of
relatively high habitat suitability (HSI/Psi_hat ≥ 0.75) revealed little overlap between the two
modeling approaches (Figure 7). Nevertheless the two models combined captured all of the
historic and recent locations occupied by burrowing owls and so in that sense they may capture
the broad range of habitats in the Coachella Valley that are suitable for burrowing owls.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 14
Figure 7. Our best performing Occupancy Model (yellow shaded areas) overlain upon our best
performing Niche Model (orange shaded areas), HSI/Psi_hat ≥ 0.75. Red circles indicate all documented
burrowing owl locations, historic and recent, utilized in this study.
DISCUSSION
Our linear driving surveys, point counts, and audio point counts were limited to accessible (non
4 wheel drive) roads and so with the exception of the flood control channels, nearly all were
detected next to roads. Along the wildland flood channels, many nesting sites were at or
clumped very near to a road or bridge. This may be due to the high disturbance of soils in these
areas, which encourages usage by burrowing animals (a prerequisite for owl nest sites), or may
be related to amount of soil water content attributed to more available runoff next to the road.
We found very few owls within the vast agricultural areas, and all were clustered near
“guzzlers” or private aqueducts. Based our results, we recommend the use of linear surveys,
coupled with plot surveys, to best characterize the population of breeding burrowing owls in
the Coachella Valley. A proposed survey protocol is described in detail in Appendix A.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 15
Our field surveys revealed 53 independent locations of owls throughout the Coachella Valley.
Thirty-seven of those locations (70 %) were found to be within planned conservation lands,
leaving 16 locations (30 %) outside the planned conservation area. Although our surveys were
distributed across the Coachella Valley, our observations of owls were largely clumped in and
around the Desert Hot Springs area. Many of the nest sites found were along Mission Creek,
Little Morongo Wash, Hidden Springs (now in development), and other washes, as well as a
half dozen nest sites directly adjacent to suburban development. In these areas, many nest sites
were observed on the edge of empty development pads, left vacant by the slow in construction
during this economic recession. Ground squirrels move in to create burrows in the disturbed
soils, and the owls use these burrows opportunistically. The ability of these owls to take
advantage of burrows in high proximity to development makes it imperative that future studies
and monitoring efforts catalog potential nests in these areas along roadside surveys or
environmental analyses for impact reports. The majority of owls nesting within Mission Creek
drainage and Little Morongo Wash are within the areas to be incorporated into the conservation
of those corridors. Unless managed, off-highway vehicle use, dumping, and domestic animal
predations could be threats in these areas given their proximity to suburban and urban areas.
The largest accumulation of nest sites found in the southern Coachella Valley was along the
Whitewater (Coachella Valley) Storm-water Channel, which ultimately drains into the Salton
Sea. These observations were scattered when compared to those found within the Desert Hot
Springs area. As in the Imperial Valley, the availability of burrows within the Coachella Valley
flood control canals, including the Coachella Valley Storm-water Channel, depend greatly on
the intensive management practices of the Coachella Valley Water District and surrounding
farms. Most of the burrows within this route were created by water erosion resulting in large
crags in the bank, which are usually initially occupied and enlarged by burrowing rodents. The
roadside agricultural drains of the Coachella Valley tend to be much shallower than those that
provide habitat for the owls in the Imperial Valley. Many local roadside drains are < 1 m deep,
while Imperial’s can be up to 9 m deep, providing more protection for the nest sites. In
Coachella Valley drains vegetation maintenance and use of tractors to flatten crags challenges
the ability of owls to use this habitat. Rosenburg and Haley (2004) found in their studies within
the Imperial Valley that the availability of burrows is positively correlated with the abundance
of ground dwelling rodent species, and negatively correlated with the intensity of the dredging
of canals and drains. In the future, management activities along these sites may wish to include
collaboration with monitoring efforts to minimize destruction of nests.
Our results differ somewhat from those reported by Conway and Simon (2003) who used
driving linear surveys to detect new nest sites and then used those and other known nest sites
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 16
to determine detection rates for point counts and audio point counts. Using previously known
nest sites, Conway et al. (2003, 2007) found higher detection rates associated with point counts,
especially audio point counts during breeding season. In contrast, our surveys were distributed
more generally across the landscape without using previous known sites to guide the creation
of our routes. Also, Conway and Simon (2003) surveyed throughout the day during breeding
season when on the linear driving transects, and focused their point counts in the mornings,
and audio point counts during morning and evening hours during breeding season. Our
surveys were conducted throughout the day (0700 – 1600) during each seasons to test the
efficacy of each method. Haug and Didiuk (1993) found that using audio surveys among known
nest sites greatly increased detectability, but noted a drastic decline in responses, especially by
female burrowing owls, after mid-April. Finally, Conway et al. (2003, 2007) recommended using
audio surveys late in the wintering period and early in the breeding season to increase
effectiveness. The results of these studies correspond to our findings that audio surveys were
more effective in late winter 2008-2009 and the winter 2009-2010 seasons.
Our efforts to model the distribution of burrowing owls should be considered a useful,
preliminary step towards understanding this species’ Coachella Valley range and identifying
suitable habitat. The aspect of burrowing owl biology most reflected by our modeling results is
that owls are generalists. Because they are generalists, a wide range of areas provide potentially
suitable habitat, with one result being the occurrence of owls within this range is difficult to
predict based on available knowledge. Unlike other generalist species (e.g., American crow, redtailed hawk), burrowing owls do not occur throughout the range apparently suitable to them,
suggesting a lack of habitat saturation. A lack of habitat saturation could occur because a
population is recovering from circumstances of the past and is therefore in the process of filling
the available habitat, or because unaccounted-for-factors are limiting the population. Given the
widespread and continued conservation concern for burrowing owls, the latter is more likely in
this case.
Additional research aimed at identifying factors that limit burrowing owls is necessary for
understanding the observed occurrence patterns. The inclusion of variables reflecting these
currently undefined factors in models would likely improve the predictive value of
distributional models. Potential un-modeled factors that might improve predictive power of
models include local-scale vegetation structure and the presence of burrowing mammals (i.e.,
California ground squirrels). Additionally, future analyses involving occupancy modeling may
result in better fitting models by considering a greater number of variable combinations, as well
as transformations of variables designed to improve their normality. Until research is conducted
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 17
that elucidates a more restrictive set of variables defining suitable burrowing owl habitat, the
combined niche and occupancy models can serve to identify the broad extent of suitable habitat.
In order to better understand the status and persistence of burrowing owls in the Coachella
Valley, and to identify key habitat variables, future research on burrowing owls here should
concentrate on:
•
Understanding the movement of burrowing owls: what is their site fidelity, does
wintering habitat differ from breeding habitat, where do juveniles go? Color banding
should be considered as a means to answer these questions.
•
Attempt to assess breeding success and a minimum number of young produced. The
collection and analysis of regurgitated pellets from around the burrow entrance and
surrounding perches would inform our management efforts on which prey species
such as Palm Springs pocket mouse and various orthoptera are essential to their
diets and increase breeding success. York et al. (2002) suggested that burrowing
owls’ breeding success in agricultural areas of the Imperial Valley was reduced due
to the lack of small mammal prey available. If the same is true in the Coachella
Valley conservation of wildland owls, where small mammal prey is generally
abundant, should be the focus.
•
A habitat analysis should also be made of vegetative cover density and exotic plant
species occurrence within occupied areas to understand if there are effects on the
owls from occupying burrows surrounded by invasive species.
•
Creating a niche model for the distribution of California ground squirrels within the
Coachella Valley could facilitate the creation of a GIS layer for the squirrels’
distribution. That layer could then be incorporated into future iterations of spatially
explicit models for burrowing owls
•
Finally, research into the effects of land use and management patterns along flood
control channels and in suburban areas might provide land managers with better
information on annual timing for vegetation control and dredging within these
areas, as well as the effects of rodent management and land use disturbances on this
population of burrowing owls.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 18
LITERATURE CITED
Browning, D. M., S. J. Beaupré and L. Duncan. 2005. Using partitioned Mahalanobis D2 (K) to
formulate a GIS-based model of timber rattlesnake hibernacula. Journal of Wildlife
Management 69:33-44.
Burnham, K. P. and D. R. Anderson. 1998. Model selection and inference - a practical informationtheoretic approach. Springer- Verlag, New York.
Clark, J.D., J. E. Dunn, K. G. Smith. 1993. A multivariate model of female black bear habitat use
for a geographical information system. Journal of Wildlife Management 57:519526.Coachella Valley Multiple Species Habitat Conservation Plan (CVMSHCP). 2007. http://
www.cvmshcp.org/plan_documents.htm Accessed December 15th, 2009
Conway, C.J. and J.C. Simon. 2003. Comparison of detection probability associated with
Burrowing Owl survey methods. The Journal of Wildlife Management, 67:501-511.
Conway, C.J., V. Garcia, M.D. Smith, K. Hughes. 2007. Factor affecting detection of Burrowing
Owl nests during standardized surveys. The Journal of Wildlife Management, 72: 688-696.
Dunn, J. E. and L. Duncan. 2000. Partitioning Mahalanobis D2 to sharpen GIS classification.
Pages 195-204 in C. A. Bebbia and P. Pascolo, eds. Management Information Systems 2000: GIS
and remote sensing. WIT Press, Southhampton, U.K.
ESRI. 2005. ARCGIS 9.1. ESRI, Redlands, California.
Haug, E. A. and A. B. Didiuk. 1993. Use of recorded calls to detect burrowing owls. Journal of
Field Ornithology 64:188-194.
James, P.C. and R.H.M. Espie. 1997. Current status of Burrowing Owl in North America: an
agency survey. Pp. 3-5 in J.L. Lincer and K Steenhoff (editors). The Burrowing Owl, its
biology and management. Raptor Research Report No. 9. Allen Press, Lawrence, KS.
Knick, S.T., and J.T. Rotenberry. 1998. Limitations to mapping habitat use areas in changing
landscapes using the Mahalanobis distance statistic. Journal of Agricultural, Biological, and
Environmental Statistics 3:311-322.
Korfanta, N.M., D.B. McDonald, and T.C. Glenn. 2005. Burrowing Owl (Athene cunicularia)
population genetics: A comparison of North American forms and migratory habits. Auk
122:464-468
MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. A. Royle, and C. A. Langtimm. 2002.
Estimating site occupancy rates when detection probabilities are less than one. Ecology
83:2248-2255.
MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Baily, and J. E. Hines. 2006.
Occupancy Estimation and Modeling. Elsevier Inc.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 19
Rosenburg D.K. and K. L. Haley. 2004. The Ecology of Burrowing Owls in the Agroecoystem of
the Imperial Valley, California. Studies in Avian Biology 24:120-160.
Rotenberry, J. T., S. T. Knick, and J. E. Dunn. 2002. A minimalist’s approach to mapping species’
habitat: Pearson’s Planes of closest fit. Pages 281-290. in J. M. Scott, P. J. Heglund, M. L.
Morrison, J. B. Haufler, M. G. Raphael, W. A. Wall, and F. B. Samson, editors. Predicting
Species Occurrences; Issues of Accuracy and Scale. Island Press, Covelo, California, USA.
Rotenberry, J. T., K. L. Preston and S. T. Knick. 2006. GIS-based niche modeling for mapping species
habitat. Ecology 87:1458-1464.
Sappington, J.M., K.M. Longshore, and D.B. Thomson. 2007. Quantifiying Landscape
Ruggedness for Animal Habitat Anaysis: A case Study Using Bighorn Sheep in the Mojave
Desert. Journal of Wildlife Management. 71(5): 1419 -1426
US Dept of the Interior, Fish and Wildlife Service (USFWS). 2003. California pp. 40-44 in D.S.
Klute, L.W. Ayers, J.A. Shaffer, M.T. Green, W.H. Howe, S. L. Jones, S.R. Sheffield, and T.S.
Zimmerman (Editors), In Prep. Status Assessment and Conservation Plan for the Western
Burrowing Owl in the United States. Biological Technical Publication FTW/BTP-R 6001-2003.
Washington, D.C.
York, M.M., D.K. Rosenberg, and K.K. Sturm. 2002. Diet and food-niche breadth of burrowing
owls (Athene cunnicularia) in the Imperial Valley, California. Western North American
Naturalist 62:280-287.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 20
APPENDIX A - BURROWING OWL (ATHENE CUNNICULARIA) SURVEY PROTOCOL
Burrowing Owls are widely distributed in western North America, occurring in deserts,
grasslands, and shrub-steppe, along with agricultural and suburban-edge modified landscapes.
While they are sometimes clumped in their distribution, especially in anthropogenically
modified areas, more often they are widely dispersed, even within large, seemingly suitable
habitat patches. Site occupancy is likely a complex interaction among soils, slope, vegetation
density and height, prey density, predator density, and the abundance of burrowing mammals
sufficient in size to provide at least the beginning structure of the owls’ underground burrows.
This broad array of habitats, and cryptic factors as to why certain areas are occupied while
others are not, along with potentially large distances between occupied locations, challenges the
use of any single survey method for determining the owls’ occurrence patterns.
The survey protocol described here is designed to meet three objectives:
1. To sample a sufficient portion of the Coachella Valley to accurately characterize the
distribution and relative abundance of owls within the Coachella Valley Multiple
Species Conservation Plan (CVMSHCP) area.
2. To enable detection of changes in both the distribution and relative abundances of owls
through time.
3. To identify environmental correlates of the occurrence, population size, and changes in
demographic parameters. This aspect of the burrowing owl protocol is addressed under
“Research Questions”.
In 2009 UC Riverside’s Center for Conservation Biology conducted an evaluation of four
burrowing owl survey techniques within the Coachella Valley. Those techniques included
linear surveys, audio point counts (using recorded calls to elicit detections), point counts
(without broadcast of recorded calls), and plot counts using an existing network of plots
designed for detections of multiple species occurring within aeolian sand communities. The
results of this pilot study informed the protocol proposed for future monitoring of burrowing
owls, as well as further research questions that need to be addressed in the future
Recommended survey protocol
Burrowing owls winter in the Coachella Valley, but the status of these birds is not fully
understood; are they the same birds that breed here or are they migrants from further north?
Given comparatively low detection rates in winter using any of the methods we tested, we
recommend only formal breeding period surveys for burrowing owls. For the breeding period,
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 21
linear surveys along routes within the Coachella Valley provided the greatest number of owls
detected. To adequately sample across a diverse landscape, we created 42 routes between 1 and
40 km along roadways that included rural, suburban, and urban landscapes as well as
irrigation/flood control areas, mostly along existing roadways within wildland, suburban, and
agricultural locations (see map below). Linear surveys are conducted by visually scanning
roadside habitat continuously, without stopping the vehicle unless a positive sighting was
made, at a speed of no more than 20 km per hour (Conway and Simon 2003, Conway et al.
2008). Visual detections for these surveys are limited to the line of sight from the roadway, and
binoculars are only used at locations where owls were observed so as to count the number of
individuals at that location. Location specifics are recorded using a handheld GPS.
Linear survey routes along with wildland plot counts accounted for 90% of the occupied
burrowing owl locations detected in 2009. Linear routes do survey wildland habitats away from
roads along Mission Creek, Morongo Wash and the San Gorgonio River wash in the Desert Hot
Springs – Snow Creek regions. In addition to these linear surveys, the 100+ established aeolian
sand community plots that occur within the conserved wildlands should be surveyed for
burrowing owls as well. These involve walking 100 x 10 m plots and recording all tracks, visual,
and audio observations, and do not require any additional effort beyond that extended to
conduct the aeolian sand species surveys.
The current plots survey only a small fraction of the wildland habitat available to the owls and
so this approach will likely underestimate presence of burrowing owls in wildland areas.
However, to conduct surveys that would cover even 10% of the available habitat would require
more than an order of magnitude increase in survey effort. Given the low density of owls
occurring in wildland habitats (especially east of Desert Hot Springs), the gain in terms of
additional burrowing owl sightings would be very small. We do suggest that all persons
conducting property evaluations (i.e. hazardous materials searches), or any natural resource
surveys/inventories within the CVMSHCP area be instructed to collect GPS locations for any
burrowing owls they encounter. Those locations would then be confirmed by those responsible
for burrowing owl surveys and added to a sampling frame of historic burrowing owl locations
that would be resurveyed each year.
The inclusion of additional survey methods such as point counts and audio point counts could
facilitate the detection of additional burrowing owls. During the 2009 pilot study, of sites
occupied by burrowing owls detected along roads and levees, 90% were detected using linear
surveys alone, another 5% were added using point counts, and 5% were added by using audio
point counts. If funding priorities dictate more complete survey of owls during the breeding
period, then point counts and/or audio point counts should be added to the recommended
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 22
survey protocol. Those point counts should also be conducted along those same linear survey
routes, with data only collected at established points consisting of a four-minute visual search
for owls, with each point separated by 0.8 km (0.5 miles). Audio point counts are conducted in
the same manner as regular point counts, only with a field speaker and iPod that looped
Burrowing Owl calls for 30 sec, and then 30 seconds of silence, then repeated the cycle over a
four minute playback period. A loop recording of both the male coo-coo call, and the defensive
(chuck and chatter) call are recommended. Binoculars should be used at all audio point count
locations to determine any movement of Burrowing Owls. As with the linear surveys each point
count survey should be repeated at least three times, and include surveys that correspond to the
early, middle and late phases of the owls’ breeding period (April, May-June, July-August).
Adding point counts and audio point triples the overall survey effort. We suggest that the
addition of point counts to an established network of linear surveys be given a low priority.
Based on the results of our pilot study, we recommend linear surveys during the breeding
season (April to August) in conjunction with plot surveys in wildland-aeolian sand areas
(which will be surveyed for other species as well) as the preferred burrowing owl survey
method. Based on detection rates from our results as well as from previous research (Conway
and Simon 2003, Conway et al. 2008), each linear survey route would need to be repeated ≥
three times during breeding season for an accurate assessment of occupancy. Due to lower
detection rates, wildland plot surveys would need to be repeated ≥ three times (current
protocols require six repetitions on these plots).
To maximize survey efficiency we propose an adaptive monitoring approach starting with the
base set of routes and plots surveyed in 2009 (Figure A). The adaptive component of this
approach would be to add new routes or wildland plots whenever there have been new owl
detections, perhaps through cooperator reports. Within this adaptive monitoring framework
there could also be a stratification of monitoring frequency using a dual-frame sampling
approach; for routes where no owls were previously located for three years there would be two
repetitions during the breeding season, whereas for routes where owls have been more recently
located surveys would be repeated at least three times. Once an owl is located the only further
survey required that year would be near the end of the breeding period to confirm whether
breeding occurred and if possible get a count of the number of young owls produced. As part of
adaptive monitoring and dual frame sampling, we could add additional linear routes in areas
previously identified as "marginal" based on niche or occupancy modeling criteria.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 23
##
Y
Y
Y
#
Y#
#
Y
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
##
Y
Y
Y
#
Y
#
##
Y
Y
Y
#
Y
#
Y
#
Y
#
##
Y
Y
Y
#
Y#
#
Y
#
Y
#
Y
#
Y
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y#
#
Y
#
Y
Y#
#
Y
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y#
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
#
Y
Y
#
Y
#
Y#
#
Y
Y
#
Y
#
Y
#
Y#
#
Y#
#
Y
Y
#
Y#
#
Y
#
#
Y
Y
Y#
#
Y
Y
#
Y
Figure A. The distribution of linear survey routes (red lines) and wildland plots (red circles) surveyed in
2009.
This overall burrowing owl survey effort will require the dedication of one full time biologist
between April and September (to accommodate data collection, analyses and report
preparation). This would be in addition to the team conducting surveys on the aeolian sand
plots where burrowing owls could also be detected. While not lessening the need for a full time
biologist for six months, the linear surveys lend themselves to a “citizen science” involvement,
using interested public in ride-along support for surveys as well as searches for additional owls
not yet on a survey route.
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 24
Research Questions
Figure B. Envirogram depicting hypothetical relationships between likely driver and stressor variables
that dictate burrowing owl occupancy and reproductive success. The four ultimate causes are in slightly
larger font at the top and are connected via the proximate causes in smaller font. In bold are the things
that we believe we measure directly or can obtain information on through other sources.
There are information needs regarding the persistence and long-term conservation of
burrowing owls in the Coachella Valley that will require additional data collection and analysis
approaches. Those information needs can be met through addressing the following research
questions:
•
Is the burrowing owls’ breeding success impacted by land use patterns and/or land
management practices?
•
What are the primary drivers of burrowing owl population fluctuations and
distributional limits within the Coachella Valley?
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 25
•
To what extent are burrowing owls distributed within lands currently conserved,
proposed for conservation, or in a “stable” use pattern through zoning or economic
constraints?
The first research question addressing the owls’ breeding success will require that
•
Surveys attempt to assess breeding success and a minimum number of young produced
– accomplished by conducting late breeding period surveys at all known occupied
burrows.
•
Late breeding season surveys include the collection of regurgitated pellets from around
the burrow entrance and surrounding perches, and the pellet contents analyzed.
•
Land management practices such as vegetation management along flood control levees
are documented.
•
We measure vegetation cover as a co-variate of owl occupancy, especially invasive,
exotic species such as Sahara mustard
The second research question addressing the drivers of the owls’ population fluctuations will
require that
•
Multivariate models are constructed that incorporate potential co-variates to the owls’
occupancy such as rainfall, vegetation cover, potential prey abundance and composition,
land uses including recreation, agriculture (types of crops, fallow fields) and
development densities. Many of these variables will be quantified incidental to other
monitoring activities, although fields will need to be added on survey data sheets to
ensure site-specific data are included. As these variables are not currently available as
GIS layers, modeling will be limited to regression or logistic approaches.
The third research question builds on the first two, but addresses a fundamental question of
whether the burrowing owls occupying lands not destined for conservation such as agriculture
and suburban-rural development will persist. Beyond addressing the first two research
questions, this question will be answered through the long-term monitoring and tracking of
spatial correlates surrounding (500 m radius) burrowing owl locations.
Data and Analyses
At the end of each field season a geodatabase will be submitted to the CVCC summarizing the
findings for that year. Data fields will include:
UTM Coordinates for all survey locations
UTM Coordinates for all occupied locations
Inventory and monitoring Western Burrowing Owls for the CV MSHCP
Final Report p. 26
Survey dates
Detection Dates
Survey Method
Person(s) conducting the Survey
Land use type(s) – 500 m radius
Research specific fields:
Breeding success
Maximum number of owls observed
Proportion of Palm Springs pocket mice in the owls’ diet
Proportion of arthropods in the owls’ diet
Vegetation cover
Sahara mustard cover
Subsequent analyses will include occupancy modeling, modeling drivers of
population/breeding dynamics