scispace - formally typeset
Search or ask a question
Journal ArticleDOI

Spatiotemporal Dynamics in Identification of Aircraft–Bird Strikes

10 Jun 2015-Transportation Research Record (Transportation Research Board, National Research Council, National Academy of Sciences USA)-Vol. 2471, Iss: 2471, pp 19-25
TL;DR: In this paper, the authors analyzed 105,529 U.S. civil aviation strike records from 1990 to 2012 in the FAA's National Wildlife Strike Database to examine patterns of collisions involving unidentified birds.
Abstract: A primary concern for human-wildlife interactions is the potential impacts resulting from wildlife (primarily birds) collisions with aircraft. The identification of species responsible for collisions with aircraft is necessary so that airport management can develop effective strategies to reduce strikes with those species. Of particular importance in developing such strategies is the identification of regional, seasonal, and temporal patterns in collisions with unidentified bird species that may limit the effectiveness of regional habitat management to reduce bird strikes. The authors analyzed 105,529 U.S. civil aviation strike records from 1990 to 2012 in the FAA's National Wildlife Strike Database to examine patterns of collisions involving unidentified birds. Factors that affected identification were airport certification class, FAA region, mass of struck species, state species richness (if damage was reported), and interactive effects between the last four factors. Identification varied by region and declined with increasing species richness; this identification was greater for general aviation (GA) airports and the mass of struck species, especially when damage was reported. Species identification might be improved by increasing reporting efforts relative to species richness, especially by GA airport managers and operations staff, who may have a higher propensity of reporting bird strikes, and by collecting more field-based data on avian populations. The results can provide guidance for the development of airport management and personnel training.

Summary (2 min read)

Statistical Analyses

  • The authors modeled the relationships between bird identifications (response variable) by using logistic regression within generalized linear mixed models in R 3.03 (R Development Core Team, Vienna, Austria) with the lme4 package (31) and ranked models by using a sequential modeling approach using Akaike’s information criterion (AIC) and weights (32).
  • Models with a ΔAIC <4 were considered competitive, as this value reduces the potential for errors involv- ing interpretation of models with ΔAIC <2 (33).
  • Coefficients and 85% confidence intervals for biologically significant effects were also examined.
  • The fixed effects included main effects and additive combinations of location characteristics (FAA region, airport type), species richness, estimated mass for struck species, and reported damage, if any.
  • State, airport, and year were classified as random effects in the model, and random effects structure was determined by fitting the full model with multiple combinations of the specified random effects.

RESULTS

  • The authors reviewed 105,529 bird strikes associated with 1,363 airports in the database, with 47% of strikes not identified to species or group (e.g., family or genus classification).
  • The proportion of identified bird strikes, damaging strikes, and mass and species richness varied by FAA region (Table 1).
  • In addition, two peaks, corresponding to spring and autumn migration periods, occurred from March to June and August to November.
  • Identification was 15% lower when damage was reported, but damage had a significant interaction with species mass: larger species responsible for damaging strikes were 1.56 times as likely to be identified.
  • For R2, fixed effects in the best-fit model accounted for 9.4% of total variance, and the full model explained 34.9% of variance in the data set.

DISCUSSION OF RESULTS

  • Recent research has focused on the importance of managing strike risk through ecological assessments of hazardous species (6, 9, 35–37).
  • The models in this research considered the influence of species richness on proper identification, but other population metrics, such as abundance, could be a better predictor of struck-species identification.
  • Efforts by the FAA to increase reporting led to an 11% increase in strike reports at GA airports between 2011 and 2012 (29); such an increase suggests that underreporting may be prevalent.
  • Improving identification of species in bird strikes also has benefits beyond aviation flight risk and airport wildlife management, including for aviation engine manufacturers.
  • Understanding the limitations of the current data set and increasing reporting efforts through training, improving reporting standards, and collecting more field-based data on avian populations will help to increase the industry’s knowledge of avian species and to reduce the aviation risk for both humans and wildlife.

ACKNOWLEDGMENTS

  • This work was supported by Mississippi State University, the U.S. Department of Agriculture, the FAA, and the 2013–2014 Graduate Research Award Program on Public-Sector Aviation Issues from the Transportation Research Board of the National Academies.
  • The authors thank L. Goldstein, L. Howard, R. Nicholson, and R. Samis for providing technical support and advice.
  • Data from eBird were obtained in accordance with the terms of use from the Cornell Lab of Ornithology.
  • Additional funding was provided by FAA; USDA National Wildlife Research Station, Sandusky, Ohio; the Transportation Research Board of the National Academies; and the Forest and Wildlife Research Center and the Department of Wildlife, Fisheries and Aquaculture at Mississippi State University.

Did you find this useful? Give us your feedback

Content maybe subject to copyright    Report

University of Nebraska - Lincoln
DigitalCommons@University of Nebraska - Lincoln
!%7-32%0#-0(0-*))6)%5',)27)57%>
8&0-'%7-326
!)4%571)273*+5-'80785)2-1%0%2(0%27
)%07,264)'7-32)59-')

Spatiotemporal Dynamics in Identi!cation of
Aircra"Bird Strikes
Tara J. Conkling
Mississippi State University7.'1667%7))(8
James A. Martin
Mississippi State University
Jerrold L. Belant
Mississippi State University.&)0%27'*51667%7))(8
Travis L. Devault
USDA/APHIS/WS National Wildlife Research Center 5%9-6)"%807%4,-686(%+39
3003:7,-6%2(%((-7-32%0:35/6%7 ,B46(-+-7%0'311326820)(8-':(1$86(%2:5'
%573*7,) -*)'-)2')6311326
?-657-'0)-6&538+,773;38*35*5))%2(34)2%'')66&;7,)!)4%571)273*+5-'80785)2-1%0%2(0%27)%07,264)'7-32)59-')%7
-+-7%0311326!2-9)56-7;3*)&5%6/%-2'3027,%6&))2%'')47)(*35-2'086-32-2!%7-32%0#-0(0-*))6)%5',)27)57%>8&0-'%7-326
&;%2%87,35-<)(%(1-2-675%7353*-+-7%0311326!2-9)56-7;3*)&5%6/%-2'302
32/0-2+ %5%%57-2%1)6)0%27)5530(%2()9%807 5%9-64%7-37)1435%0;2%1-'6-2()27-@'%7-323*
-5'5%A=-5(75-/)6 USDA National Wildlife Research Center - Sta" Publications
,B46(-+-7%0'311326820)(8-':(1$86(%2:5'

19
Transportation Research Record: Journal of the Transportation Research Board,
No. 2471, Transportation Research Board of the National Academies, Washington,
D.C., 2015, pp. 19–25.
DOI: 10.3141/2471-03
A primary concern for human–wildlife interactions is the potential
impacts resulting from wildlife (primarily birds) collisions with aircraft.
The identification of species responsible for collisions with aircraft is
necessary so that airport management can develop effective strate-
gies to reduce strikes with those species. Of particular importance in
developing such strategies is the identification of regional, seasonal, and
temporal patterns in collisions with unidentified bird species that may
limit the effectiveness of regional habitat management to reduce bird
strikes. The authors analyzed 105,529 U.S. civil aviation strike records
from 1990 to 2012 in the FAA’s National Wildlife Strike Database to
examine patterns of collisions involving unidentified birds. Factors that
affected identification were airport certification class, FAA region, mass
of struck species, state species richness (if damage was reported), and
interactive effects between the last four factors. Identification varied by
region and declined with increasing species richness; this identification
was greater for general aviation (GA) airports and the mass of struck
species, especially when damage was reported. Species identification might
be improved by increasing reporting efforts relative to species richness,
especially by GA airport managers and operations staff, who may have a
higher propensity of reporting bird strikes, and by collecting more field-
based data on avian populations. The results can provide guidance for the
development of airport management and personnel training.
The ecological implications of climate change, urbanization, and
other factors influencing bird populations and migration patterns can
affect species interactions with humans (1–5). One primary concern
for these interactions is the potential impacts from bird strikes with
aircraft. Airports and surrounding landscapes are often grasslands
that are perceived by wildlife as habitat (6, 7). In addition, factors
influencing damage sustained to aircraft include aircraft speed and
the mass and number of struck individuals, the latter of which is
dependent on flocking behaviors of each avian species (8–12). Bird
abundances and flocking and flight behaviors in airport vicinities
may be related to seasonal changes in migration patterns, weather,
food availability, and predation risk (13–17). As a result, development
of models that incorporate ecological data is important for increasing
the understanding of spatiotemporal factors that hinder aviation safety.
The richness of avian species demonstrates spatial variation, with
decreasing richness at greater latitudes (18, 19); this richness may
also be affected by elevation, climate dynamics, existing habitat and
geographic features, and potential food resources (2, 14, 20–22).
Furthermore, site-specific richness of species varies seasonally
because of (a) the presence of breeding or wintering avian species
and (b) temporal pulses during spring and autumn migrations. Dis-
tributions of migrating birds may also be influenced by geographic
features (e.g., mountain ranges or coastlines) that may concentrate
migrating populations (23), and all of these factors can adversely
affect aviation safety (24, 25). Many species known to be detri-
mental to aircraft [e.g., Canada geese (Branta canadensis), gulls, etc.]
are also increasing in numbers and easily adapting to urbanized
environments where they may be more likely to occur at airports
(26, 27). By incorporating measurements of the richness of avian
species into training and reporting procedures, researchers could
improve identification proficiency by accounting for spatiotemporal
variation in avian populations and the relative influence on aviation
strike risk.
The objectives of this study were to (a) examine incidents involv-
ing unknown bird species relative to the total incidents (unidentified
bird ratio) and (b) model nonidentification of bird strikes to identify
potential regions and factors adversely influencing species identifi-
cation, including spatial, temporal, and management variables. This
study examined specific hypotheses that could potentially influence
nonidentification rates related to species richness, location, airport
classifications, estimated mass of struck species, and the occurrence
of damage.
Species richness may negatively influence correct classification of
struck species if the ease of identification for observers is dependent
on the number of similar candidate species. Therefore, the authors
predicted species nonidentification to be greater in FAA regions with
higher overall species richness and also expected nonidentification to
increase during peak migration periods in spring and autumn, when
species richness is greater. Airportcentric management practices
may also influence identification. Certificated–classified airports
often have a trained airport biologist, whereas general aviation (GA)
airports generally do not, and therefore, personnel at GA airports
may be less likely to identify species because they lack specialized
training (28). If any of the above factors is important to species
identification, modifications in training relative to influential factors
may improve identification, strike reporting, and the effectiveness
of hazardous species management.
METHODS
Databases
The authors analyzed U.S. civil aviation strike records from 1990
to 2012 in the FAAs National Wildlife Strike Database to deter-
mine patterns of collisions involving unidentified birds and factors
Spatiotemporal Dynamics in Identification
of Aircraft–Bird Strikes
Tara J. Conkling, James A. Martin, Jerrold L. Belant, and Travis L. DeVault
T. J. Conkling, J. A. Martin, and J. L. Belant, Department of Wildlife, Fisheries
and Aquaculture, Mississippi State University, Mississippi State, MS 39762.
T. L. DeVault, U.S. Department of Agriculture, Animal, and Plant Health Inspection
Service, Wildlife Services, National Wildlife Research Center, Sandusky, OH 44870.
Corresponding author: T. J. Conkling, tjc191@msstate.edu.
This document is a U.S. government work and
is not subject to copyright in the United States.

20 Transportation Research Record 2471
influencing nonidentification. Since 1990, the FAA has compiled
voluntary reports of strikes involving civil aircraft and wildlife
(29); this information determines economic costs of strikes and
provides important data regarding the wildlife species involved, the
type of damage, and aircraft and location details. The analyses were
restricted to civil aviation operations involving avian species
within the contiguous United States. Each record was classified in
the database as identified or unidentified, and all bird strikes were
considered identified if they were at least classified to family level
(e.g., Anatidae).
Each incident was categorized in the data set by month, airport,
FAA region, state, Part 139 airport classification type (certificated
or GA), estimated mass of the struck species, and reported damage,
if any. The richness of avian species was determined for each state on
a mean weekly basis by using data collected through citizen science
and submitted to the eBird website (www.ebird.org) from 1990 to
2010 (30). eBird is a citizen science website that allows users to
submit data on numbers of observed avian species. The numbers of
registered observers and submitted lists vary widely by location, so
the observations were pooled at the state level for subsequent analyses.
Observation data queried from eBird presented species observations
as the proportion of checklists containing that species divided by all
checklists submitted for a given week (e.g., first week of January)
for all years combined. For example, Canada geese occurred on
16% (n = 5,737) of California checklists for the first week of January.
The authors determined the species richness for each state and month
by calculating the total species present on more than 1% of reported
checklists during each weekly interval to exclude single records of
vagrant species. In addition, weekly species records were pooled to
estimate the total species observed in a state by month.
Statistical Analyses
The authors modeled the relationships between bird identifications
(response variable) by using logistic regression within generalized
linear mixed models in R 3.03 (R Development Core Team, Vienna,
Austria) with the lme4 package (31) and ranked models by using a
sequential modeling approach using Akaike’s information criterion
(AIC) and weights (32). Models with a ΔAIC <4 were considered
competitive, as this value reduces the potential for errors involv-
ing interpretation of models with ΔAIC <2 (33). Coefficients and
85% confidence intervals for biologically significant effects were
also examined. The fixed effects included main effects and additive
combinations of location characteristics (FAA region, airport type),
species richness, estimated mass for struck species, and reported
damage, if any. Then the best model from the previous step was used
to determine whether interactive effects between species richness
and region or mass and damage reported improved model fit. State,
airport, and year were classified as random effects in the model, and
random effects structure was determined by fitting the full model
with multiple combinations of the specified random effects. Next
pseudo-R
2
was calculated to examine the proportion of variance
explained by the best-fit models and to assess model fit (34).
RESULTS
The authors reviewed 105,529 bird strikes associated with 1,363 air-
ports in the database, with 47% of strikes not identified to species or
group (e.g., family or genus classification). The proportion of identified
bird strikes, damaging strikes, and mass and species richness var-
ied by FAA region (Table 1). Bird strikes associated with certificated
airports accounted for 94.7% of the records. Approximately 8.4% of
strikes reported damage to the aircraft, and the mean mass standard
deviation (±SD) of struck species involved in damaging incidents
(1,317.5 ± 1,271.2 g) was 2.3 times as large as when no damage
was reported (456.5 ± 616.8 g). The number of bird strikes increased
annually; however, the proportion of reports with unidentified birds
relative to identified species declined (Figure 1).
The richness of avian species varied by month and location, with
the highest species richness within the Southern, Northwest, South-
west, and Western Pacific FAA regions (Table 1). In addition,
two peaks, corresponding to spring and autumn migration periods,
occurred from March to June and August to November. Identification
of species in strike incidents was best explained by certification class,
damage reported, species mass, and an interactive effect between
FAA region and species richness and an interaction between dam-
age reported and species mass (Table 2); no other models were
within 531 ΔAIC. Identification was highest in New England; other
FAA regions were 50% (Northwest) to 79% (Southern) less likely
to identify strikes by species group (Table 2). Bird strikes declined
TABLE 1 Summary Statistics for Total Bird Strikes, Percentage of Unidentified Strikes,
Percentage of Strikes with Damage, Estimated Mass of Struck Species, and Annual Species Richness,
by FAA Region for Incidents Reported in FAA National Wildlife Strike Database, 1990–2012
Total Bird
Strikes (n)
Unidentified
Strikes (%)
Damage
Reported (%)
Species Mass (g) Species Richness
FAA Region Mean SD Mean SD
New England 4,559 31.9 7.8 602.0 885.5 331.6 69.4
Central 5,548 44.1 8.2 441.9 715.0 294.5 47.3
Eastern 20,261 39.7 8.7 597.1 832.4 354.8 80.8
Great Lakes 17,131 41.7 8.1 540.3 795.3 333.0 48.5
Northwest 11,031 42.0 7.9 490.9 699.2 377.1 54.5
Southern 20,499 58.5 8.8 544.0 728.6 355.5 88.8
Southwest 13,670 52.7 6.9 392.9 500.1 516.4 123.2
Western Pacific 12,830 51.0 10.1 571.0 666.8 563.4 84.4
Note: SD = standard deviation.

Conkling, Martin, Belant, and DeVault 21
with increasing species richness, and strikes at GA airports were
1.27 times as likely to be identified as those at certificated airports
(Table 2). Identification was 15% lower when damage was reported,
but damage had a significant interaction with species mass: larger
species responsible for damaging strikes were 1.56 times as likely
to be identified. For R
2
, fixed effects in the best-fit model accounted
for 9.4% of total variance, and the full model explained 34.9% of
variance in the data set.
DISCUSSION OF RESULTS
Recent research has focused on the importance of managing strike risk
through ecological assessments of hazardous species (6, 9, 35–37).
The results in the current research provide further support that
inclusion of basic spatial and temporal ecological data improves
researchers’ ability to interpret patterns of strike risk and nonidenti-
fication. Species richness significantly influenced strike identification
efforts, although this effect varied by location and season. The greatest
frequency of strikes involving unknown species occurred during
periods of avian migration (especially autumn), when species rich-
ness was the greatest (Table 3) and large populations of new juve-
niles from the recent breeding season attempted their first migration
(38). Observers who account for the influence of species richness by
placing more emphasis on identification effort, especially during peak
migration periods, will not only increase identification rates but also
provide more data on species posing collision risks to aircraft. Future
research incorporating migration data within spatial models to identify
hot spots for risk may also reduce risk to both aircraft and migrating
birds (39).
0
1,000
2,000
3,000
4,000
5,000
6,000
7,000
8,000
9,000
10,000
1990 1994 1998 2002 2006 2010
Total Bird Strikes
Unknown
Identified
Total Birds
Year
FIGURE 1 Number of bird strikes and number of bird strikes involving unidentified and
identified species reported in FAA National Wildlife Strike Database for contiguous
United States, 1990–2012.
TABLE 2 Model Selection Results for Generalized Linear Mixed Models to Analyze Factors Influencing Bird Identification
for Incidents in FAA National Wildlife Strike Database, 1990–2012
Model Log Likelihood AIC ΔAIC Weight
Certification class + FAA region × species richness + damage × mass 63,189.81 126,425.60 0.00 1.00
Certification class + FAA region + species richness + damage × mass 63,456.78 126,957.60 531.94 0.00
Certification class + FAA region × species richness + damage 63,551.79 127,145.60 719.97 0.00
Certification class + FAA region × species richness 63,595.50 127,231.00 805.39 0.00
Certification class + FAA region + species richness 63,658.86 127,343.70 918.12 0.00
Species richness 63,681.28 127,372.60 946.95 0.00
Damage reported 63,988.39 127,986.80 1,561.16 0.00
Certification class + FAA region 64,004.27 128,032.50 1,606.93 0.00
FAA region 64,016.49 128,055.00 1,629.37 0.00
Certification class 64,029.42 128,068.80 1,643.22 0.00
Null 64,040.70 128,089.40 1,663.78 0.00

22 Transportation Research Record 2471
The models in this research considered the influence of species
richness on proper identification, but other population metrics, such
as abundance, could be a better predictor of struck-species identifi-
cation. An area could have relatively low species richness but still have
high populations of a few common species; if these common species
(e.g., geese) are also hazardous to aircraft, local abundances could
have a substantial impact on strike risk. Although widely available
citizen science data on bird populations such as eBird provide infor-
mation on the number of species detected in a given state, the avail-
ability of site-specific data varies. The North American Breeding
Bird Survey provides annual roadside monitoring data for breed-
ing birds in the United States and Canada; however, survey data are
restricted to May and June (40). Year-round abundance data for avian
assemblages do not exist for most locations within the United States.
Airports could benefit from employees or volunteers conducting
regular surveys that include point counts or line transects to determine
species richness and abundance of avian species to supplement the
existing data set (41–43).
The current results also demonstrated regional variation in iden-
tification, even when species richness and management-specific
factors such as airport classification are taken into account (Figure 2).
Although the number of flight operations in a given region can
affect the number of bird strikes reported, this potentiality does
not explain the difference in the proportion of strikes identified by
species group. The three regions with the lowest species richness
(Central, Great Lakes, and New England) were also most likely to
identify struck species (Figure 2). The Western Pacific region had
the greatest mean species richness, but the model predicted species
identification to be higher there than in other regions located along
major flyways, especially in locations where convergence of northern-
based migrants and funneling land features may concentrate bird
populations (Eastern, Southern, and Southwest). Tailoring training to
emphasize both identification efforts (even at the level of the species
group) and submission of samples of unidentified remains to the
Smithsonian Feather Identification Lab (44, 45), especially during
periods of increased species richness (i.e., migration), will help
improve the overall data set.
Also important is continuing to obtain baseline data on strike
frequencies, locations, and the species involved through strike
reporting to the FAA National Wildlife Strike Database; however,
many of these data are still sparse partly because of the voluntary
reporting standards. The number of reported incidents in the data-
base increased annually over the past two decades (29), and high-
profile incidents, such as US Airways Flight 1549 crash landing
in the Hudson River in 2009 after colliding with a flock of Canada
geese (46), may have encouraged pilots and airports to report bird
strikes. However, because reports are voluntary, the accuracy of
these numbers in reflecting total incidents and the cause of this
increase—whether from more strikes or simply increased reporting
effort—are unknown (35).
TABLE 3 Model Coefficients, 95% Confidence Intervals, and Odds Ratios for Parameters
in Best-Fit Generalized Linear Mixed Models to Analyze Factors Influencing Bird Identification
for Incidents in FAA National Wildlife Strike Database, 1990–2012
Confidence Intervals
Parameter
a
Estimate SE Lower 95% Upper 95% Odds Ratio
(Intercept) 0.77 0.25 0.29 1.27
Central 1.28 0.37 2.02 0.58 0.28
Eastern 0.76 0.32 1.38 0.14 0.47
Great Lakes 0.79 0.31 1.40 0.20 0.46
Northwest 0.69 0.32 1.33 0.07 0.50
Southern 1.55 0.31 2.17 0.96 0.21
Southwest 1.11 0.34 1.79 0.46 0.33
Western Pacific 0.75 0.39 1.53 0.01 0.47
Species richness 0.34 0.10 0.53 0.15 0.71
Certification class
b
(general aviation) 0.23 0.07 0.09 0.37 1.27
Damage reported
c
0.16 0.03 0.22 0.10 0.85
Mass 0.002 0.01 0.02 0.02 1.00
Central: species richness 0.14 0.13 0.40 0.11 0.87
Eastern: species richness 0.55 0.11 0.76 0.33 0.58
Great Lakes: species richness 0.27 0.11 0.49 0.05 0.77
Northwest: species richness 0.24 0.11 0.02 0.46 1.27
Southern: species richness 0.28 0.11 0.50 0.07 0.76
Southwest: species richness 0.15 0.11 0.38 0.07 0.86
Western Pacific: species richness 0.04 0.12 0.28 0.19 0.96
Damage
c
: mass 0.45 0.02 0.41 0.49 1.56
Note: SE = standard error; — = not applicable.
a
Reference FAA region = New England.
b
Reference certification class = certificated.
c
Reference condition = no damage reported.

Citations
More filters
Journal ArticleDOI
TL;DR: Travis L. DevauLT, U.S. Department of Agriculture, Wildlife Services’ National Wildlife Research Center, 6100 Columbus Avenue, Sandusky, OH 44870, USA richarD a. washBurn as discussed by the authors
Abstract: Travis L. DevauLT, U.S. Department of Agriculture, Wildlife Services’ National Wildlife Research Center, 6100 Columbus Avenue, Sandusky, OH 44870, USA MichaeL J. Begier, U.S. Department of Agriculture, Wildlife Services’ Airports Wildlife Hazards Program, 1400 Independence Avenue SW, Washington, DC 20250, USA JerroLD L. BeLanT, Center for Resolving Human–Wildlife Conflicts, Mississippi State University, Mississippi State, MS 39762, USA BraDLey F. BLackweLL, U.S. Department of Agriculture, Wildlife Services’ National Wildlife Research Center, 6100 Columbus Avenue, Sandusky, OH 44870, USA richarD a. DoLBeer, U.S. Department of Agriculture, Wildlife Services’ Airports Wildlife Hazards Program, 6100 Columbus Avenue, Sandusky, OH 44870, USA JaMes a. MarTin, Department of Wildlife, Fisheries, and Aquaculture, Mississippi State University, Mississippi State, MS 39762, USA ThoMas w. seaMans, U.S. Department of Agriculture, Wildlife Services’ National Wildlife Research Center, 6100 Columbus Avenue, Sandusky, OH 44870, USA Brian e. washBurn, U.S. Department of Agriculture, Wildlife Services’ National Wildlife Research Center, 6100 Columbus Avenue, Sandusky, OH 44870, USA

12 citations

References
More filters
Book
19 Jun 2013
TL;DR: The second edition of this book is unique in that it focuses on methods for making formal statistical inference from all the models in an a priori set (Multi-Model Inference).
Abstract: Introduction * Information and Likelihood Theory: A Basis for Model Selection and Inference * Basic Use of the Information-Theoretic Approach * Formal Inference From More Than One Model: Multi-Model Inference (MMI) * Monte Carlo Insights and Extended Examples * Statistical Theory and Numerical Results * Summary

36,993 citations

06 Oct 2015
TL;DR: The core computational algorithms are implemented using the Eigen C++ library for numerical linear algebra and RcppEigen``glue''.

8,543 citations

Journal ArticleDOI
TL;DR: In this article, the authors make a case for the importance of reporting variance explained (R2) as a relevant summarizing statistic of mixed-effects models, which is rare, even though R2 is routinely reported for linear models and also generalized linear models (GLM).
Abstract: Summary The use of both linear and generalized linear mixed-effects models (LMMs and GLMMs) has become popular not only in social and medical sciences, but also in biological sciences, especially in the field of ecology and evolution. Information criteria, such as Akaike Information Criterion (AIC), are usually presented as model comparison tools for mixed-effects models. The presentation of ‘variance explained’ (R2) as a relevant summarizing statistic of mixed-effects models, however, is rare, even though R2 is routinely reported for linear models (LMs) and also generalized linear models (GLMs). R2 has the extremely useful property of providing an absolute value for the goodness-of-fit of a model, which cannot be given by the information criteria. As a summary statistic that describes the amount of variance explained, R2 can also be a quantity of biological interest. One reason for the under-appreciation of R2 for mixed-effects models lies in the fact that R2 can be defined in a number of ways. Furthermore, most definitions of R2 for mixed-effects have theoretical problems (e.g. decreased or negative R2 values in larger models) and/or their use is hindered by practical difficulties (e.g. implementation). Here, we make a case for the importance of reporting R2 for mixed-effects models. We first provide the common definitions of R2 for LMs and GLMs and discuss the key problems associated with calculating R2 for mixed-effects models. We then recommend a general and simple method for calculating two types of R2 (marginal and conditional R2) for both LMMs and GLMMs, which are less susceptible to common problems. This method is illustrated by examples and can be widely employed by researchers in any fields of research, regardless of software packages used for fitting mixed-effects models. The proposed method has the potential to facilitate the presentation of R2 for a wide range of circumstances.

7,749 citations

Journal ArticleDOI
TL;DR: This work has shown that predation is a major selective force in the evolution of several morphological and behavioral characteristics of animals and the importance of predation during evolutionary time has been underestimated.
Abstract: Predation has long been implicated as a major selective force in the evolution of several morphological and behavioral characteristics of animals. The importance of predation during evolutionary ti...

7,461 citations

Journal ArticleDOI
19 Aug 2011-Science
TL;DR: A meta-analysis shows that species are shifting their distributions in response to climate change at an accelerating rate, and that the range shift of each species depends on multiple internal species traits and external drivers of change.
Abstract: The distributions of many terrestrial organisms are currently shifting in latitude or elevation in response to changing climate Using a meta-analysis, we estimated that the distributions of species have recently shifted to higher elevations at a median rate of 110 meters per decade, and to higher latitudes at a median rate of 169 kilometers per decade These rates are approximately two and three times faster than previously reported The distances moved by species are greatest in studies showing the highest levels of warming, with average latitudinal shifts being generally sufficient to track temperature changes However, individual species vary greatly in their rates of change, suggesting that the range shift of each species depends on multiple internal species traits and external drivers of change Rapid average shifts derive from a wide diversity of responses by individual species

3,986 citations

Frequently Asked Questions (13)
Q1. What have the authors stated for future works in "Spatiotemporal dynamics in identification of aircraft–bird strikes" ?

Future research incorporating migration data within spatial models to identify hot spots for risk may also reduce risk to both aircraft and migrating birds ( 39 ). Although the number of flight operations in a given region can affect the number of bird strikes reported, this potentiality does not explain the difference in the proportion of strikes identified by species group. Continued development of the reporting website and social media could further enhance reporting of strike incidents ( 29 ). Efforts by the FAA to increase reporting led to an 11 % increase in strike reports at GA airports between 2011 and 2012 ( 29 ) ; such an increase suggests that underreporting may be prevalent. 

The authors modeled the relationships between bird identifications (response variable) by using logistic regression within generalized linear mixed models in R 3.03 (R Development Core Team, Vienna, Austria) with the lme4 package (31) and ranked models by using a sequential modeling approach using Akaike’s information criterion (AIC) and weights (32). 

Of particular importance in developing such strategies is the identification of regional, seasonal, and temporal patterns in collisions with unidentified bird species that may limit the effectiveness of regional habitat management to reduce bird strikes. 

The authors determined the species richness for each state and month by calculating the total species present on more than 1% of reported checklists during each weekly interval to exclude single records of vagrant species. 

This study examined specific hypotheses that could potentially influence nonidentification rates related to species richness, location, airport classifications, estimated mass of struck species, and the occurrence of damage. 

site-specific richness of species varies seasonally because of (a) the presence of breeding or wintering avian species and (b) temporal pulses during spring and autumn migrations. 

An area could have relatively low species richness but still have high populations of a few common species; if these common species (e.g., geese) are also hazardous to aircraft, local abundances could have a substantial impact on strike risk. 

The three regions with the lowest species richness(Central, Great Lakes, and New England) were also most likely to identify struck species (Figure 2). 

This work was supported by Mississippi State University, the U.S. Department of Agriculture, the FAA, and the 2013–2014 Graduate Research Award Program on Public-Sector Aviation Issues fromthe Transportation Research Board of the National Academies. 

The objectives of this study were to (a) examine incidents involving unknown bird species relative to the total incidents (unidentified bird ratio) and (b) model nonidentification of bird strikes to identify potential regions and factors adversely influencing species identification, including spatial, temporal, and management variables. 

The North American Breeding Bird Survey provides annual roadside monitoring data for breeding birds in the United States and Canada; however, survey data are restricted to May and June (40). 

Future research incorporating migration data within spatial models to identify hot spots for risk may also reduce risk to both aircraft and migrating birds (39). 

Airports could benefit from employees or volunteers conducting regular surveys that include point counts or line transects to determine species richness and abundance of avian species to supplement the existing data set (41–43).