Avian Solar Work Group
Avian Solar Working Group Research Questions Framework
Defining Research Questions1 and Methodological Approaches for Addressing Potential Impacts of PV Solar Plants on Bird Populations
Thomas B. Smith1 (Chair and Co-Scientific Advisor to ASWG), Steven Beissinger2, Wally
Erickson3, Vasilis Fthenakis4, Trevon Fuller5 (Panel Coordinator), Luke George6, Kristen Ruegg5
(Co-Scientific Advisor to ASWG), and Rodney Siegel7.
1Institute of the Environment and Sustainability and Department of Ecology and Evolutionary
Biology, University of California at Los Angeles, Los Angeles, CA; 2Department of
Environmental Science, Policy & Management, University of California at Berkeley, Berkeley,
CA; 3Western EcoSystems Technology, Inc., Cheyenne, WY; 4Earth and Environmental
Engineering, Columbia University, New York, NY; 5Institute of the Environment and
Sustainability, University of California at Los Angeles. Los Angeles, CA; 6Department of Fish,
Wildlife, and Conservation Biology, Colorado State University, Fort Collins, CO; 7 The Institute
for Bird Populations, Point Reyes Station, CA
1 The research questions reflect the range of concerns of the ASWG; they are not a reflection of priorities of the
ASWG or any ASWG members.
ASWG Research Questions Framework Introduction
The Avian Solar Work Group was established in summer 2015 with the mission of advancing
coordinated scientific research to better understand how birds interact with solar facilities. The
group spent the first year collaborating to identify research topics and formulate specific research
questions. In January 2016, the ASWG convened a Research Panel – consisting of experts in
ecology, ornithology, demographics, biostatistical analysis, and environmental engineering – to
reformulate the questions into scientifically testable hypotheses. The result is the Research
Questions Framework, below, which defines target questions and a range of potential
methodological approaches to research the potential impacts of PV solar facilities on bird
populations.
The questions and methodological considerations included constitute a broad range of scientific
inquiries of interest to the ASWG and/or its individual members. Based on discussions between
the Research Panel, Agency Observers, and ASWG members, the ASWG has decided to focus
initial efforts on research that informs whether water birds or other birds are attracted to solar
panels because they perceive them as water bodies, sometimes referred to as “Lake Effect”.
ASWG will also coordinate related research on siting, background mortality, population-level
effects, and feather spots. ASWG envisions its role as facilitating the development of
independent third party research, partnering with others to seek research funding,
and coordinating across NGOs, government, industry sectors, and research institutions.
Concurrent research efforts will be pursued in concert to maximize results while minimizing
time, effort, and expense.
A. Siting
1) Do avian mortality rates at PV solar power plants differ from background rates at
control sites?
An important first step should be reviewing existing published and unpublished literature that
assesses mortality rates at PV solar plants and/or compares them to background rates (e.g.,
McCrary et al. 1986, Kagan et al. 2014). Collecting new information to answer the question
will require selecting an existing survey protocol or developing a new survey protocol for
assessing mortality rates at multiple PV solar power plants. Most likely this will involve
walking pre-established transects and counting feather spots or other signs of apparent
mortalities. Where possible, attempts should be made to identify bird remains by species. A
pilot study and/or literature review may be necessary to assess mean and variance in
mortality encounter rates during survey transects; this information could then be used to
determine minimum necessary sampling effort (i.e., number and length of transects) for
subsequent surveys.
Identify appropriate reference sites on nearby undeveloped lands with vegetation and other
habitat characteristics relatively similar to what was present at the solar facilities before they
were constructed. Any sampling design should encompass the broadest range of habitats
while taking into account heterogeneity and should have sufficient statistical power to inform
decision-making. Whenever possible the industry should work collaboratively to collect and
pool data across sites in order to maximize sample size. Such an approach is much preferable
to single facility assessments as is typically done currently. Reference sites should be close
enough to PV sites to yield inference on the critical siting questions, but not so close as to
jeopardize the study design. For example, if the reference sites are too close to the PV sites,
scavengers could move carcasses from the PV sites to the reference sites. This would
jeopardize the study design by resulting in an underestimate of mortality rate at the PV sites
and an overestimate of mortality rates at the reference sites. The size (or transect length) of
reference sites need not be the same as that of the solar plants. Conduct surveys at PV solar
power plants and reference sites, and compare observed mortality rates, which could be
expressed in relative terms, such as observed mortalities per linear distance of transect
walked. Interpreting survey results will likely require experimental trials to assess possible
differences in carcass detection probability at PV versus reference sites.
Although comparing mortality rates at PV and reference sites is an appropriate way to assess
the effect of PV sites on avian morality fairly rapidly, where possible it would also be
beneficial and inferentially powerful to assess avian mortality at the location of intended PV
facilities before they are built, allowing for a before/after control impact (BACI) study
design, in which mortality rates could be assessed at PV and control sites both before and
after construction of PV facilities (Smith and Dwyer 2016).
2) What is the relationship of mortality rates to site characteristics (e.g., panels, fence
lines, overhead transmission lines, scale/configuration of installations, proximity to
other solar facilities or other natural or human landscape features such as levels of
fragmentation and loss of habitat, migratory flyways and stop over sites, etc.)?
This question operates on at least two distinct spatial scales: within individual solar facilities,
and across multiple solar facilities.
a) Within a facility, are mortalities associated with particular elements of that facility
(e.g., panels, fence lines, overhead transmission lines)?
This could be answered by collating and interpreting existing information on mortalities,
and/or by establishing new mortality monitoring efforts that classify observed mortalities at
one or (preferably) more facilities according to the particular elements with which they are
associated. Statistical modeling, using a frequentist or Bayesian framework, could then be
used to test for associations between mortality and particular facility elements. Comparing
mortality rates associated with particular elements of facilities may require explicitly
assessing carcass detection probability in proximity to those different elements.
b) Among facilities, do location and landscape features affect mortality rates?
Define a priori a small set of site variables (e.g., scale/configuration of installations,
proximity to other solar facilities or other natural or human landscape features such as levels
of fragmentation and loss of habitat, migratory flyways and stop over sites) and statistically
test associations between those variables and mortality rates at different facilities, using data
from past, ongoing, and perhaps new monitoring programs. Comparing mortality rates
between sites may require explicitly assessing carcass detection probability at those different
sites. Site variables could include time series of land use/land cover change and
fragmentation, microclimate, vegetation phenology, and proximity to water bodies measured
at a high spatial resolution using satellites such as ASTER, Landsat, MODIS, and PALSAR;
or distance to known migratory flyway features such as the Colorado River drainage or the
Pacific Coast.
This analysis will likely be heavily constrained by sample sizes (i.e., the number, locations,
and characteristics of facilities providing data) and the difficulty in clearly assigning values
to the features listed, and therefore may not be fully answerable in a statistically rigorous
way.
3) How might siting be optimized to reduce potential impacts on vulnerable bird
populations in a cost-effective manner?
A cost/benefit analysis should be performed to assess the financial costs of siting new
facilities with respect to factors determined to affect bird mortality rates (based on findings
from question 2b). Also important would be to conduct a meta-analysis of vulnerable bird
populations within a GIS framework. This analysis could be used to identify migratory
hotspots for many key species and the most heavily used routes for vulnerable populations
(following the methods in 2b).
B. Population level effects
Are solar sites causing avian mortality that is significant at the scale of the population
for individual species?
Prior to beginning any population-level analysis, it is important to generate a list of the
species of greatest concern. This can be accomplished through discussions with industry,
NGO’s and governmental organizations (such as the Bureau of Land Management and Fish
and Wildlife Service). The list of candidate species can then be forwarded to each agency and
scored according to priority. Finally, Breeding Bird Survey (BBS) and MAPS (Monitoring
Avian Productivity and Survivorship) data can be used to assess populations that may be at
risk. This information can then be cross-referenced with the results of the species
prioritization list to develop a final list.
1) How should populations be defined in this context?
The most robust method currently available for identifying populations of migratory birds is
genome-wide genetic data. Recent studies show that genome-wide genetic data can resolve
migratory bird populations at spatial scales that are similar to existing Bird Conservation
Regions (see Ruegg et al 2014). Other methods include the use of demographic attributes
from long-term monitoring data to delineate natural population structure (Rushing et al 2015)
or the use of hydrogen isotope ratios to define birds from distinct ecoregions (Wassenaar &
Hobson 2001). Given the general lack of resolution using isotope data this method alone
would not be recommended. In some cases it may be possible or desirable to use a
combination of the above approaches for better resolution of populations. There may also be
situations where obtaining the genetic information on populations is not possible. In these
situations it might be possible to use population projection models (e.g., Franklin et al., 2000;
Franklin et al., 2004; Blakesley, 2010; Dugger et al., 2015) or more arbitrary definitions of
populations based upon ecologically defined geographic regions such as Bird Conservation
Region boundaries or political borders (Millard et al. 2012). Regardless of the method used,
all population maps should be made accessible to interested parties and the public so that
multi-species comparative analyses can be conducted across multiple sites.
2) What research and data would be required to determine if mortality associated with
solar sites is additive or compensatory?
Additive and compensatory mortality represent the opposite ends of a spectrum of the effects
of a new form of mortality on annual survival in a population.
If a specific type of mortality is additive, it will increase baseline mortality by an amount
equal to the sum of baseline and the new source of mortality. If a mortality factor is
compensatory, annual mortality will not increase when the new source is “added” to the
population, deaths from the new source of mortality will be compensated for by a decline in
other sources of mortality.
Compensatory mortality implies density-dependent factors are acting on the population, such
that associated losses of individuals are counterbalanced by enhanced survival or
reproduction of other individuals, as a result of having fewer conspecific competitors. In
order to distinguish whether a new source of mortality is additive or compensatory, annual
mortality must be measured in the presence and absence of the new source of mortality. If
there is no change in annual mortality when the new source of mortality is present,
compensatory mortality is occurring. If annual mortality increases by an amount equal to the
new source of mortality, the new source of mortality is completely additive. Of course, it is
also possible that annual mortality will increase less than the sum of baseline and the new
source of mortality in which case the mortality is partly additive. The potential for
compensatory mortality to occur is influenced by the demographic characteristics of the
population and whether the population is above or below its carrying capacity (Péron 2013).
Species with low annual survival and high recruitment (r-selected or “fast” species) are more
likely to exhibit compensatory mortality than species with high annual survival and low
recruitment. In addition, species that are close to or above their carrying capacity are more
likely to exhibit density dependence in both survival and recruitment than species below their
carrying capacity. Although these predictions are supported by analyses of the impacts of
hunting on waterfowl populations (Péron 2013), this question has not been thoroughly
addressed in other groups of birds.
In practice, it is very difficult to directly measure whether a source of mortality is additive or
compensatory because it requires precise measurements of annual mortality in the presence
and absence of the new source of mortality. For example, despite numerous band-recovery
studies involving hundreds of thousands of banded birds, there is still disagreement about
whether hunting mortality acts in an additive or compensatory manner on North American
waterfowl populations (Pӧysӓ et al. 2004, Pӧysӓ et al. 2013, Sedinger and Herzog 2012). In
case of solar arrays, the task would be even more difficult because the new source of
mortality would likely be very small relative to natural sources of mortality requiring even
larger sample sizes than those used in waterfowl studies. Directly addressing the question of
whether mortality associated with the operation of solar arrays is additive or compensatory
would take many years, cost millions of dollars per species, and may not provide a definitive
answer.
In the absence of direct measures of changes in mortality, the question of whether a new
source of mortality is additive or compensatory can be examined using indirect approaches.
For instance, if the new source of mortality generally results in the death of sick or weak
individuals that were likely to die anyway, or of primarily young individuals that faced an
inherently low rate of recruitment into the adult population, the new source of mortality is
likely to be compensatory. On the other hand, if the new source of mortality results in the
death of healthy individuals that were likely to survive the annual cycle, the new source of
mortality is likely to be additive. Thus, the additive and compensatory hypotheses make
different predictions about the characteristics of the individuals that die from the new source
of mortality. If the age, health, body condition, parasite load, etc. of individuals that die from
the new source of mortality is similar to individuals randomly selected from the population,
the new source is likely to be additive. If the individuals that die from the new source of
mortality are generally less healthy (e.g., poorer body condition, higher parasite loads) or are
less likely to survive the annual cycle (e.g., young birds) then the new source of mortality is
likely to be compensatory. Additionally, for many bird species, existing information (e.g.,
MAPS data) may indicate whether density dependence is likely to be an important factor in
population dynamics.
3) How do population impacts differ by species, guild, migratory pathway, taxonomic
unit and classification (threatened versus non-threatened), etc.?
One could investigate how population impacts differ by species, guild, migratory pathway,
taxonomic unit and classification (threatened versus non-threatened) first by defining a set of
genetic or morphological markers that can be used to define each of these units. If
morphological assessment is still possible (depending upon the state of the carcasses), this
method could be used to identify most individuals to species-level. If carcasses can be further
classified into age and sex categories, then this information could also be useful for
determining which segments of a population are most likely impacted. Species could be
placed into guild categories based upon known ecological characteristics. All other finer
scale classification levels such as populations and migratory pathway would likely require
the development of high-resolution genetic markers (genome-wide genetic markers) that
could differentiate between threatened versus non-threatened populations and identify
populations along their migratory pathways. In cases where species level ID is not possible
with morphological traits, lower-resolution genetic markers (mtDNA) could also be used to
identify species. In cases where the appropriate genetic markers have been developed,
population specific flyway use can be augmented with on the ground surveys such as
capturing birds in mist-nets at a site adjacent to a facility and using the resulting genetic
samples to assess population flyway use over time. Such information can be used to help
inform operational mitigation if sensitive species and populations are concentrated during a
limited time of year.
If high-resolution genetic markers identify particular population segments represented among
the mortalities then demographic (e.g., from the MAPS program) and count (e.g., from the
Breeding Bird Survey) data could be analyzed to compare survival rates and population
growth rates of the affected population segments with other population segments.
Furthermore, once populations have been defined it would be possible to identify species and
populations that are at a very low risk for impacts from solar facilities so that resources can
be allocated towards groups with the greatest need (Beston et al 2016). Similarly, even in the
absence of results from high-resolution genetic markers or other means to isolate particular
population segments that may be adversely affected, count and demographic data could be
harnessed to assess overall population growth rates of any species that is frequently
represented among the mortalities.
C. Lake Effect
1) Are water or other birds attracted to solar panels because they perceive them as
water bodies (i.e., a “Lake Effect”)?
a) Is a “Lake Effect” possible according to the geographic and environmental
infrastructure characteristics of sites?
Summarize the current information on bird mortality at PV and trough facilities including
species/taxa impacted, size of projects, type of technology (tracker vs. fixed), distance
between rows, habitat, proximity to waterbodies/marshes of different sizes, bird count data if
available.
Use standardized methods of measuring bird mortality at a number (10 or more) of solar
arrays. Examine the relationship between mortality of all birds and select groups of birds
(e.g., waterbirds) and environmental/physical variables using graphical methods as well as
more sophisticated linear models. We suggest developing an a priori set of models and using
model selection to identify the most parsimonious model. Environmental variables may
include: size of the solar array, east-west extent of the array, a measure of bird migration
density over the site using NEXRAD or some other radar system, the type of solar panels, the
length of power lines adjacent to the site, the presence, number, or spatial extent of water
bodies near (possibly use a number of different distances) the site, and habitat type
surrounding the site. Whenever possible, the industry should work collaboratively to collect
and pool data across sites in order to maximize sample size.
b) Do birds show evidence of attraction to large solar arrays (e.g. show changes in flight
direction or behavior as they approach arrays)?
Quantify flight direction and behavior at solar sites and at nearby “reference” sites. The
reference sites should be in similar habitat and landscape context as the solar site and
sufficiently far away (> 1 km) to ensure the solar facility is not having an impact on flight
behavior.
Evidence for a change in direction towards the solar array or a decrease in altitude as birds
approach a solar array compared to birds at reference sites would constitute support for the
“Lake Effect” hypothesis. In addition, evidence for an increase in “approach behavior” such
as circling or gliding as birds approach solar arrays would also support the “Lake Effect”
hypothesis. Mobile avian radar systems that may allow assignment of individual radar traces
to a particular species or group (e.g., ducks, geese) could be used to quantify flight direction
as birds approach but they could be cost prohibitive. Marine radar systems provide a less
costly alternative for quantifying flight direction but radar traces cannot be assigned to a
species or group so observers would be needed to identify approaching birds. During daylight
hours observers can quantify approach behaviors, at night, infrared cameras may allow
quantification of behaviors near the facilities. However, these approaches will need to be
tested for their effectiveness.
Radio/satellite telemetry of some select species might also provide information on behavior
during migration and could possibly provide information on whether birds are attracted to PV
facilities. As mentioned above, these approaches will need to be tested for their effectiveness
before implementation on larger scales.
c) What types of birds are affected?
Conduct analyses of groups of birds (e.g., ducks, geese, loons, grebes, all waterbirds, all land
birds) using the approaches outlined for questions 1a and 1b. Analyses should focus on
species/groups that suffer high mortality at the solar arrays because they are likely to be the
most susceptible to the Lake Effect. The methods used for each of the analyses will influence
the groups that can be identified. In most cases, birds can be identified to species with visual
observations. Contrasting information on use by species with mortality can help define risk
as well.
d) Is possible mortality due to stranding, strikes or some other process?
Fresh carcasses from carcass surveys within solar arrays should be necropsied. Cause of
death may be able to be identified in fresh specimens. This can also be determined in the
field if evidence of collision is observed in the field (broken necks, wings etc.) or on the
facilities (bird imprints on panels, broken panels). Frequency of live “uninjured birds” should
also be evaluated. For example, water birds may land safely but unable to take off again
without a body of water. Camera systems or visual observations that can document bird
behavior and incidence of collision or stranding might be applied as well. Background
mortality levels and species at reference sites compared to the solar sites may also help to
identify the likely cause of mortality.
e) If the Lake Effect is demonstrated, what cues are causing the birds to mistake the
solar array as a water body (e.g., what wavelength of reflected light are they responding
to)?
Birds may use a variety of cues to identify water bodies including reflected light, changes in
temperature or humidity, sound, and smell. Previous studies may shed light on the cues that
are most likely used by birds to detect water and additional studies may be able to identify
specific cues. For instance, if previous studies suggest that birds respond to reflected light,
studies of the wavelengths of light reflected from water bodies, in conjunction with
laboratory studies of the optical properties of bird vision, may provide insight into specific
wavelengths that birds use to detect water bodies. Satellite images such as Landsat could be
analyzed to compare the reflectance of light by lakes and PV arrays. In the portion of the
electromagnetic spectrum that is visible to humans, the reflectance of clear water is 5-15%
whereas that of metals like aluminum and steel is 50-90% (Richards, 2012). To assess
whether the reflectance of PV arrays and lakes are perceived as similar by birds, it will be
necessary to account for climatic conditions such as rain during the migratory period and the
characteristics of avian vision such as birds’ ability to see polarized light (Muheim et al.
2006).
Meta-analysis could be performed to better understand the sensory information birds use and
design studies to test hypothesized sensory cues.
f) If a Lake Effect can be demonstrated, how might the threat be mitigated or
eliminated?
If the cues that birds use to detect water bodies can be identified, research can focus on
reducing those cues at solar arrays. For instance, if birds are responding to specific
wavelengths of reflected light, coatings on solar panels that reduce reflectance of those
wavelengths may reduce attraction but may be costly and may decrease transmittance of
irradiation to the active layers. Hazing methods (harassing birds with sounds, light, or other
measures) can reduce bird use of areas but their effectiveness may be temporary and often
diminishes with time. Changing the deterrent response over time may diminish habituation.
Layout designs and panels that match the color of existing landscape should be investigated
and tested for effectiveness. If tracking technology is used, and mortality is at levels of
concern and are determined to be occurring at night, stowing the panels in positions that
might decrease attraction could be considered. Measuring behavior and mortality at sites in
an experimental approach where different stow positions are considered might help
determine optimal stow position.
D. What are the avian risk-reduction options that might lower avian mortality?
Research should focus on the decision-making processes to identify risk reduction options.
This may best be accomplished by first developing a risk assessment framework (Sutter
2007). In other words, before embarking on specific risk mitigation options, detailed efforts
should be made toward identifying and characterizing risks and impacts.
If it is determined that micro and macro scale factors such as location of certain arrays or the
entire project influences the level of mortality, then risk-reduction would entail siting the
facility or sections of the facilities into areas that are less risky for birds. Fatality rates at
facilities in different locations can be compared and contrasted to identify factors that may be
correlated with higher mortality. The size of the effect of these factors will affect how much
monitoring data will need to be collected. Focused monitoring during a peak season at
multiple sites in a region may be useful for isolating some factors, for others monitoring at
multiple facilities over time may be necessary.
Other factors such as the size and configuration of the facility, type of technology (e.g.
tracker vs. fixed-tilt), may be found to impact species differently. The level of the potential
impact of such factors on species would also need to be evaluated to determine the necessity
and degree of their implementation as risk reduction measures.
Many factors have been associated with increased collision risk to birds, especially migrating
songbirds, these include: inclement weather, intense facility lighting, and height of structures.
Downlighting and minimizing lights on the facility (e.g. building lighting, substation
lighting) could minimize songbirds being attracted to the facility and colliding with
buildings, panels, overhead lines, and other elements of the facility. For example, as a first
step, lighting’s possible effects on mortality could be explored through a review of existing
literature (e.g. Kerlinger 2000 Kerlinger et al. 2010, Gehring et al. 2009, Gehring et al. 2011,
Smith and Dwyer 2016).
Burying collector lines when feasible, and ensuring overhead lines are built to Avian
Powerline Interaction Committee (APLIC 2006 and 2012) standards for minimizing potential
for bird electrocution and collision are standard practices. Marking sections of lines that are
considered high risk/high mortality may reduce collision risk or siting overhead lines away
from high avian use areas (e.g., near wetlands) may prove to be beneficial.
A recent review of the potential effectiveness and limitations of auditory, visual and other
types of avian-risk deterrents can be found in Erickson et al (2014a). However, there are few
published studies on the effectiveness of deterrents. Therefore, rigorous experimental
approaches to field testing of possible deterrents under different conditions should be
considered. These might best be achieved through BACI studies. An important component of
these studies should be to evaluate the sensitivity of specific deterrents to habituation, a
factor that has limited the successful use of deterrents in the past.
E. Feather spots
1) What do feather spots represent? Can feather spots be better defined and
quantified?
Surveyors conducting carcass searches often discover evidence of a carcass that only
includes feathers of a bird, also called feather spots. At a PV solar energy facility it is
unlikely that collision with a facility component would result in a bird’s body breaking into
pieces to create multiple feather spots. However, birds killed by collision with facility
components could be scavenged creating feather spots prior to being discovered by
personnel. In addition, evidence of bird mortality from other project related causes, such as
water birds that land safely between rows of panels and die of exposure or exhaustion, could
be scavenged creating feather spots. Bird mortality from other natural causes could also be
scavenged creating feather spots. Natural predation by raptors or mammals can also create
feather spots. Thus, when only a feather spot is detected determining an exact cause of death
is problematic.
In avian fatality monitoring studies at solar energy facilities, feather spot is often defined in
the survey protocol.
Finds will be classified as a fatality according to standards commonly applied in California
(Altamont Pass Avian Monitoring Team 2007, CEC and CDFG 2007), which dictate that
when only feathers are found, to be classified as a fatality, each find must include a feather
spot of at least five tail feathers or two primaries within 5 m or less of each other, or a total of
10 feathers. Searchers will make their best attempt to classify feather spots by size according
to the sizes of the species of birds or identifying features of the feathers.
Similar definitions have been used in avian fatality monitoring studies at wind energy
facilities since the mid-90’s. However, when looking at the origin of what constitutes a
feather spot, a justification is not provided for the number of feathers used in the definition.
Most wind energy avian fatality studies include feather spots as a project-caused fatality.
This conservative approach has been used to avoid added monitoring costs for determining
actual cause of death, and determining species of all feather spots. Further, a bird that is hit
by a moving turbine blade could be broken into multiple parts and create feather spots.
The frequency at which feather spots are discovered is a function of many factors, including
size of the area searched, predation rates, and carcass search interval. Higher predation rates
and larger areas surveyed will likely increase the number of feather spots found. Given the
potentially large number of variables involved, perhaps the best ways to determine if feather
spots are due to solar installations would be to statistically compare the number of feather
spots at solar facilities to a similarly-sized reference site (see below).
a) What methods can be used to identify the species and number of individuals that
comprise feather spots?
Utilizing experienced personnel, including those with bird banding (or museum experience),
and repeated practice, is invaluable in identifying feather spots to a species level. Bird
banding/specimen experience can be difficult to find. Failing this, the skills for fatality ID
can be learned with practice.
Time varies widely depending on specimen and circumstance, but based on experience from
one consulting company, they are able to arrive at ID within 5-15 minutes in a lab equipped
with a computer for access to online feather photographs, books, and a microscope. Experts
in feather identification at universities can also be utilized. Contracting with local or national
museums may also be a possibility. For many decades the Smithsonian National Museum of
Natural History in Washington D.C. was considered to have one of the best trained staffs and
have a long history of doing contract work for various agencies. However, whether they
would have the capacity to address all the needs of the solar industry is unlikely, without
more staff.
When identifying spots the following resources should be considered:
1. Pyle, P. 1997. Identification Guide to North American Birds. Part I. Slate Creek Press,
Bolinas, CA.
2. Pyle, P. 2008. Identification Guide to North American Birds, Part II. Slate Creek Press,
Bolinas, CA.
3. Feather Atlas (printed text and online feather guide http://www.fws.gov/lab/featheratlas/index.php)
4. Slater Wing & Tail Imaging Guide (online resource)
While visual inspection has typically been used, DNA testing can reduce uncertainty,
especially for feather spots that contain parts of multiple species.
In summary, the most effective and cost efficient approach for identifying feather spots to
species and determining the number of individuals they represent will be through the use of
DNA technologies. Costs associated with species identification are modest while determining
the number of individuals is also modest, but only if a SNP assay has already been
developed.
b) Are feather spots a reliable indicator of avian strikes and/or fatalities?
Determining cause of death of a bird found at a facility is difficult, especially when the only
evidence is a feather spot. As previously mentioned, the conservative approach that has been
used has been to include all carcasses, including feather spots, as fatalities in the analysis.
However, some studies suggest that the density of feather spots in a control area may be high
enough to question the assumption that all feather spots found at a solar energy facility
represent mortality caused by the facility (Erickson 2014b, Erickson et al. 2014). Additional
and more robust studies of background mortality and solar facility mortality in different
regions are needed to better understand impacts. It is important to ensure the reference sites
are a large enough distance away from the facility to ensure there is no effect of the facility at
the reference site. Placement of cameras on site to record possible collisions and interactions
with panels is one approach to consider. However, because impact events may be rare,
sophisticated systems for streamlining data processing (e.g. screening the videos for possible
mortality events) will be necessary to reduce the resources needed and may not be cost
effective.
Other methods are described below to provide context as to whether it is more likely than
less likely that the feather spots were caused by the facility or other means.
Method 1: Comparison of density of carcasses, including density of feather spots between
reference/control areas (without the solar facilities) and the solar facility.
These comparisons are useful in understanding background mortality for the control areas
and how they compare to the solar facility mortality. However, the solar facility after being
constructed could result in changes in bird densities within the facility, the predation rates,
and other factors that could affect the background non-facility related mortality rate. These
issues aside, a statistical comparison between a solar facility and an adjacent control site will
generally allow one to estimate impacts. Comparing the species composition of the feather
spots between the sites is also likely informative.
Method 2: Comparison of density of carcasses, and density of feather spots, before and after
the facility is built at both the facility area and the control areas.
Using a BACI design by sampling prior to construction for live birds and carcasses at both
reference areas and control areas will provide additional information as to whether the
feather spots are more or less likely to be facility related.
c) Do feather spots from larger carcasses persist in the environment longer than spots
from smaller ones?
Carcass persistence studies are typically a standard component of fatality studies at both
wind and solar facilities. Several studies and meta-analyses have been conducted that show a
very strong pattern that larger carcasses tend to persist longer in the environment than smaller
carcasses. A meta-analysis of existing information from wind and solar studies will likely
provide the strongest demonstration, confirming this hypothesis.
To test this particular hypothesis on a site-specific basis, implementation of standard
protocols for carcass persistence trials can be conducted. The typical approach is the placing
of fresh carcasses of different sizes at random locations throughout the facility and
monitoring their condition over time through periodic site visits or through placement of
cameras, that can also capture the type of predators. The time to transition from a carcass to a
feather spot would be quantified as well as how long the feather spot persists.
F. Climate change and other broader impacts
1) What demographic effects may result from climate change in the absence of large-
scale solar development, and how do these compare with the impacts of solar facilities
for specific bird populations?
Climate change is likely to affect birds in complex ways, including: (a) changes to the onset
of breeding and migration through changes in the timing of peak food resources that could
reduce adult and juvenile survival and nesting success; (b) shifts in the geographic ranges of
species in response to changes in precipitation and temperature that make portions of their
ranges unsuitable to inhabit; (c) episodes of direct mortality from extreme heat waves; (d)
loss of coastal or marsh habitat and flooding of nests from sea-level rise; and (e) increased
mortality from the arrival of new pathogens and diseases that find a favorable habitat in
California with a new climate. As a result of these processes, bird species are expected to
shift their geographic ranges and avian communities are predicted to reshuffle, with new
combinations of species arising.
It is difficult to determine how to compare these projected effects of climate change on birds
with the impacts of solar facilities on specific bird species. Tackling this question could be
done in two main ways. One approach would require developing simulation models to
project the effects of climate change on demography and comparing model outcomes from
various scenarios to simulations that included: (a) the direct effects of solar facilities on the
demography of particular species; and (b) the potential for indirect effects on demography
from the reductions of greenhouse emissions contributed by the solar plant. A second
approach would be oriented toward modeling geographic ranges of birds. This approach
would examine the habitat loss from current and projected solar facilities, and compare this
impact to the impact on bird distributions projected from climate change scenarios in the
absence of CO2 reduction from solar facilities. Both approaches would be greatly limited by
data availability, so should be considered heuristic exercises that were meant to provide
general insights.
2) Using historical and contemporary data on the abundance and distribution of avian
species with future climate projections, what are the predictions for the future avian
distribution and population trends in California? How can this be used to mitigate the
impacts of PV facilities?
In California, we are fortunate to have an unusually well documented historical record that
can be used to examine changes in patterns of diversity over the past 100 years. Between
1904 and 1940, Joseph Grinnell and colleagues at the UC Berkeley Museum of Vertebrate
Zoology documented and collected mammals, birds, amphibians, and reptiles from >700
locations on multiple transects spanning the environmental diversity of California. This effort
resulted in a remarkable snapshot of early 20th century vertebrate diversity which includes
>100,000 specimens, 74,000 pages of field notes including standardized bird count data and
habitat (including vegetation) observations, and 10,000 images.
The Grinnell Resurvey Project http://mvz.berkeley.edu/Grinnell/ couples historical data with
contemporary resurveys to measure avian responses to climate and land-use change and
project responses to future change. Sites throughout California were originally surveyed for
avian diversity from ~1908-1940 by Grinnell and colleagues. Resurveys have been
completed at 240 sites of relatively low land-use change in the Sierra Nevada, Coast Ranges,
and Mojave Desert. Ongoing resurveys will add 50 sites with histories of agricultural
development in the Central Valley, and 30 sites in urbanized areas of the South Coast.
To make predictions on the future of birds in California, changes in species occupancy at
each site can be related to changes in temperature and precipitation using existing historical
climate data and to changes in land use from future efforts to map historic land cover. Bird
responses to future scenarios of climate and land use could then be projected using direct
measures of change over the past 75 years, in contrast to the typical space-for-time models
that is used by species distribution models to predict future species occurrence based only on
the climate that a species currently inhabits.
Projections of future avian occurrence provided by the Grinnell Resurvey Project could then
be used to mitigate the impacts of PV facilities by identifying areas of high importance for
future species range expansions, range persistence, and maintenance of bird community
integrity. The project’s focus on the effects of both climate and land use change enables a
realistic projection of bird distributions in a state undergoing rapid human development. The
project includes sites located in areas of high interest for solar development such as the
Carrizo Plains, southern Central Valley, and Mojave Desert. Projected bird distributions in
these areas of high relevance to PV facilities can be compared to expected distributional
changes in more protected areas such as the Sierra Nevada and refuges of the Central Valley
that provide a baseline of how birds are expected to respond to climate change alone, as well
as to areas expected to undergo alternate forms of human development such as agricultural
areas of the Central Valley and urban areas of the South Coast.
Literature Cited
Altamont Pass Avian Monitoring Team 2007. Altamont Pass Wind Resource Area Bird and Bat
Monitoring Protocols. Available online at: http://www.altamontsrc.org/
alt_doc/m1_apwra_monitoring_protocol_6_5_07.pdf.
Avian Power Line Interaction Committee (APLIC) 2006. Suggested Practices for Avian
Protection on Power Lines: The State of the Art in 2006. Public Interest Energy Research
Program (PIER) Final Project Report CEC-500-2006-022. Edison Electric Institute,
APLIC, and the California Energy Commission, Washington, DC & Sacramento,
California.
Avian Power Line Interaction Committee (APLIC) 2012. Reducing Avian Collisions with Power
Lines: The State of the Art in 2012. Edison Electric Institute and APLIC, Washington,
DC.
Beston, J. A., J. E. Diffendorfer, S. R. Loss, and D. H. Johnson. 2016. Prioritizing avian species
for their risk of population-level consequences from wind energy development. Plos One
11:e0150813.
Blakesley, J. A., M. E. Seamans, M. M. Conner, A. B. Franklin, G. C. White, R. J. Gutierrez, J.
E. Hines, J. D. Nichols, T. E. Munton, D. W. H. Shaw, J. J. Keane, G. N. Steger, and T.
L. McDonald. 2010. Population dynamics of spotted owls in the Sierra Nevada,
California. Wildlife Monographs 174:1-36.
California Energy Commission and California Department of Fish and Game 2007. California
Guidelines for Reducing Impacts to Birds and Bats from Wind Energy Development.
Commission Final Report. California Energy Commission, Renewables Committee, and
Energy Facilities Siting Division, and California Department of Fish and Game,
Resources Management and Policy Division. CEC-700-2007-008-CMF.
Dugger, K. M., E. D. Forsman, A. B. Franklin, R. J. Davis, G. C. White, C. J. Schwarz, K. P.
Burnham, J. D. Nichols, J. E. Hines, C. B. Yackulic, P. F. J. Doherty, L. Bailey, D. A.
Clark, S. H. Ackers, L. S. Andrews, B. Augustine, B. L. Biswell, J. Blakesley, P. C.
Carlson, M. J. Clement, L. V. Diller, E. M. Glenn, A. Green, S. A. Gremel, D. R. Herter,
J. M. Higley, J. Hobson, R. B. Horn, K. P. Huyvaert, C. McCafferty, T. McDonald, K.
McDonnell, G. S. Olson, J. A. Reid, J. Rockweit, V. Ruiz, J. Saenz, and S. G. Sovern.
2016. The effects of habitat, climate, and Barred Owls on long-term demography of
Northern Spotted Owls. The Condor 118:57-116.
Erickson, W. P. 2014a. Review of Potential Bird Deterrent Strategies for Large Scale Solar
Facilities. Western EcoSystems Technology, Inc., Cheyenne, Wyoming.
Erickson, W. P. 2014b. Sources of Avian Mortality and Risk Factors Based on Empirical Data
from Three Photovoltaic Solar Facilities. Western EcoSystems Technology, Inc.,
Cheyenne, Wyoming.
Erickson, W. P., M. M. Wolfe, K. J. Bay, D. H. Johnson, and J. L. Gehring. 2014. A
comprehensive analysis of small-passerine fatalities from collision with turbines at wind
energy facilities. Plos One 9:e107491.
Franklin, A. B., D. R. Anderson, R. J. Gutierrez, and K. P. Burnham. 2000. Climate, habitat
quality, and fitness in Northern Spotted Owl populations in northwestern California.
Ecological Monographs 70:539-590.
Franklin, A. B., R. J. Gutierrez, J. D. Nichols, M. E. Seamans, G. C. White, G. S. Zimmerman, J.
E. Hines, T. E. Munton, W. S. LaHaye, J. A. Blakesley, G. N. Steger, B. R. Noon, D. W.
H. Shaw, J. J. Keane, T. L. McDonald, and S. Britting. 2004. Population dynamics of the
California Spotted Owl (Strix occidentalis occidentalis): a meta-analysis. Ornithological
Monographs 54:1-54.
Gehring, J., P. Kerlinger, and A. M. Manville II. 2009. Communication towers, lights, and birds:
successful methods of reducing the frequency of avian collisions. Ecological
Applications 19:505-514.
Gehring, J., P. Kerlinger, and A. M. Manville II. 2011. The role of tower height and guy wires on
avian collisions with communication towers. Journal of Wildlife Management 75:848-
855.
Kagan, R. A., T. C. Viner, P. W. Trail, and E. O. Espinoza 2014. Avian Mortality at Solar
Energy Facilities in Southern California: A Preliminary Analysis. National Fish &
Wildlife Forensics Laboratory, Ashland, Oregon.
Kerlinger, P. 2000. Avian Mortality at Communication Towers: A Review of the Recent
Literature, Research, and Methodology. US Fish and Wildlife Service, Office of
Migratory Bird Management, Washington, DC.
Kerlinger, P., J. L. Gehring, W. P. Erickson, R. Curry, A. Jain, and J. Guarnaccia. 2010. Night
migrant fatalities and obstruction lighting at wind turbines in North America. Wilson
Journal of Ornithology 122:744-754.
McCrary, M. D., R. L. McKernan, R. W. Schreiber, W. D. Wagner, and T. C. Sciarotta. 1986.
Avian mortality at a solar energy power plant. Journal of Field Ornithology 57:135-141.
Millard, M. J., C. A. Czarnecki, J. M. Morton, L. A. Brandt, J. S. Briggs, F. S. Shipley, R. Sayre,
P. J. Sponholtz, D. Perkins, D. G. Simpkins, and J. Taylor. 2012. A National Geographic
framework for guiding conservation on a landscape scale. Journal of Fish and Wildlife
Management 3:175-183.
Muheim R., Phillips J. B., and Akesson, S. 2006. Compass calibration in migratory songbirds.
Science 313:837-9.
Péron, G. 2013. Compensation and additivity of anthropogenic mortality: life-history effects and
review of methods. Journal of Animal Ecology 82:408-417.
Pӧysӓ, H., J. Elmberg, G. Gunnarsson, P. Nummi, and K. Sjӧberg. 2004. Ecological basis of
sustainable harvesting: is the prevailing paradigm of compensatory mortality still valid?
Oikos 104:612–615.
Pöysä, H., L. Dessborn, J. Elmberg, G. Gunnarsson, P. Nummi, K. Sjöberg, S. Suhonen, and P.
Söderquist. 2013. Harvest mortality in North American mallards: a reply to Sedinger and
Herzog. Journal of Wildlife Management 77:653–654.
Pyle, P. 1997. Identification Guide to North American Birds. Part I. Slate Creek Press, Bolinas,
California.
Pyle, P. 2008. Identification Guide to North American Birds. Part II. Slate Creek Press, Bolinas,
California.
Richards, J. A. 2012. Remote Sensing and Digital Image Analysis. Fifth Edition. Berlin:
Springer.
Ruegg, K. C., E. C. Anderson, K. L. Paxton, V. Apkenas, S. Lao, R. B. Siegel, D. F. Desante, F.
Moore, and T. B. Smith. 2014. Mapping migration in a songbird using high-resolution
genetic markers. Molecular Ecology 23:5726-5739.
Rushing, C. S., M. R. Dudash, and P. P. Marra. 2015. Habitat features and long-distance
dispersal modify the use of social information by a long-distance migratory bird. Journal
of Animal Ecology 84:1469-1479.
Sedinger J. S., and M. P. Herzog. 2012. Harvest and dynamics of duck populations. Journal of
Wildlife Management 76:1108–1116.
Smith, J. A., and J. F. Dwyer. 2016. Avian interactions with renewable energy infrastructure: an
update. The Condor: Ornithological Applications 118:411-423.
Wassenaar, L. I., and K. A. Hobson. 2001. A stable-isotope approach to delineate geographical
catchment areas of avian migration monitoring stations in North America. Environmental
Science & Technology 35:1845-1850.