The Common Nighthawk in northern Minnesota

A THESIS SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL

OF THE UNIVERSITY OF MINNESOTA BY

Stephen Robert Kolbe

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE

Dr. Gerald Niemi and Dr. Matthew Etterson

May 2018

© Stephen Robert Kolbe 2018

i

Acknowledgments

I wish to thank the members of committee, Dr. Matthew Etterson, Dr. Gerald Niemi, and

Dr. Richard Green for advice and guidance from beginning to end of this project. I

especially appreciate their patience during my “silent period” as I readjusted to academic

life after five years in the field. I would also like to thank those who helped with surveys

either on the breeding grounds or during migration: Karl Bardon, Josh Bednar, Annie

Bracey, Jessica Chatterton, Allison Coffman, Lizzie Condon, Paul Dolan, Laura

Erickson, Tom Erickson, Lance Gauer, Jan Green, John Green, Alexis Grinde, Ted

Keyel, Alex Lamoreaux, Carly Lapin, Julia Luger, Hannah Panci, Cameron Rutt, Sara

Schutte, Ryan Steiner, Nick Walton, and Ed Zlonis. Larry Snyder was instrumental in

providing access to the count location – thank you. I thank the Integrated Biosciences

graduate program at the University of Minnesota Duluth and the Biology Department for

financial support by way of teaching assistantships and the Natural Resources Research

Institute for a research assistantship. Thanks to JG Bennett, Skye Haas, Kate Maley,

Stephen Nelson, John Richardson, Tom Reed, and Ryan Steiner. Finally, thank you to my

family, especially Richard and Lois Kolbe, for their unwavering support and love.

ii

Dedication

To the visible migration true believers.

iii

Abstract

Developing methods for monitoring bird species that do not exhibit typical breeding

behaviors is difficult. Species that do not sing, are sparsely distributed, are not active in

the early morning, or are secretive are often impossible to monitor using traditional

methods such as point counts. The Common Nighthawk (Chordeiles minor), like other

species in the nightjar family, is a low-density breeder in the boreal forest and is not

adequately surveyed by point counts due to its secretive nature and crepuscular activity.

Specific surveys have been developed for nightjars, but never tested for utility with the

Common Nighthawk. I compared nightjar survey routes censused during a crepuscular

time period to those run during a nocturnal time period. Significantly more nighthawks

were detected during the crepuscular window, a time period that is not used during the

official survey. While the effect of time of survey was significant, the surveys were labor-

intensive and relatively few observations of Common Nighthawks were made. However,

large numbers of this species occur each autumn along the north shore of Lake Superior.

With average annual counts of nearly 19,500 individuals, the autumn migration of

Common Nighthawks is the largest known concentration of this species in the world.

Visible migration counts of nighthawks were conducted for three weeks each year from

2008 to 2017 in Duluth, Minnesota, USA. This daily evening count has elucidated the

weather variables that most often lead to large flights: lighter and westerly winds, and

warmer temperatures. Many of these conditions are not often associated with autumn

migration. These weather effects on nighthawk migration intensity may be tied to aerial

insect availability during migration. While the precise geographic origin of these migrant

birds is unknown, many arrive from the Canadian boreal forest, where this species has

undergone a significant decline and is listed as threatened. Trend analysis from autumn

migration counts does not show such a decline. Counts such as the annual autumn survey

of migrating Common Nighthawks along the north shore of Lake Superior are likely the

best and most cost-effective way to census the boreal forest breeding Common

Nighthawks and determine population trends for this declining aerial insectivore.

iv

Table of Contents Acknowledgements...............................................................................................................i

Dedication............................................................................................................................ii

Abstract...............................................................................................................................iii

List of Tables.......................................................................................................................v

List of Figures.....................................................................................................................vi

List of Appendices.............................................................................................................vii

CHAPTER 1: TESTING BREEDING GROUND COMMON NIGHTHAWK

(CHORDEILES MINOR) SURVEY METHODOLOGY

Introduction..............................................................................................................1

Methods....................................................................................................................4

Results......................................................................................................................7

Discussion................................................................................................................9

CHAPTER 2: VISIBLE MIGRATION COUNTS OF COMMON NIGHTHAWKS

(CHORDEILES MINOR) ALONG THE NORTH SHORE OF LAKE SUPERIOR

Introduction............................................................................................................12

Methods..................................................................................................................15

Results....................................................................................................................21

Discussion..............................................................................................................34

Literature Cited..................................................................................................................39

Appendices.........................................................................................................................43

v

List of Tables Chapter 1 Table 1. Common Nighthawks detected along Nightjar Survey Network routes................8 Chapter 2 Table 1. Summary statistics of count data and weather variables................................23-24 Table 2. Logistic regression model selection table............................................................25 Table 3. Negative binomial model selection table.............................................................26 Table 4. Linear regression model selection table...............................................................27 Table 5. Top model coefficients........................................................................................28

vi

List of Figures Chapter 1 Figure 1. Nightjar Survey Network route locations in northern Minnesota, USA..............5 Chapter 2 Figure 1. Location of the count site...................................................................................16 Figure 2. Daily distribution of Common Nighthawk counts.............................................29 Figure 3. Predicted probability plots..................................................................................30 Figure 4. Effect of wind direction on Common Nighthawk flights...................................31 Figure 5. Geometric mean trend 2008 to 2017..................................................................32 Figure 6. Date and weather adjusted geometric mean trend 2008 to 2017........................33

vii

List of Appendices Appendix 1: Instructions for conducting the Nightjar Survey Network............................43 .

1

CHAPTER 1.

TESTING BREEDING GROUND COMMON NIGHTHAWK

(CHORDEILES MINOR) SURVEY METHODOLOGIES

INTRODUCTION

Collecting reliable data that can be used for developing population trends is a

considerable challenge in ornithology (Rosenstock et al. 2002). Survey design and

methodology have a large effect on bird detection, and improper methods can result in

biased or inaccurate results that, in turn, cascade into implementation of poor

management or conservation decisions (Romesburg 1981, Rosenstock et al. 2002). Time

of survey is one of the most important methodological considerations because avian

activity varies daily and seasonally (Ralph et al. 1995). Fortunately, many species are

properly surveyed by “catch-all” methods such as the North American Breeding Bird

Survey (BBS), a survey that is conducted during early morning hours near the peak of the

breeding season (Sauer et al. 2013). The BBS efficiently detects and monitors species

that are visually or aurally conspicuous during this time period, but other methods are

needed to survey for species that cannot be readily detected in this way (Dunn et al.

2005). A central tenet of the BBS is obtaining a large sample size (the number of routes

with a given species present over multiple years) with a relatively small amount of effort

(Bart et al. 2004). This strategy is effective in areas where relative abundances for a given

species are high, but can be challenging in regions where birds are rarely encountered.

Traditional population monitoring methods such as the BBS are often insufficient

for monitoring species that are secretive (i.e. difficult to detect), occur at low densities, do

not follow traditional paradigms of territorial singing in the early morning hours, and/or

2

have inaccessible breeding grounds (Dunn and Hussell 1995). BBS-based assessments

are especially weak for populations that escape detection by both the BBS and Christmas

Bird Count (CBC) by breeding in the boreal forest and wintering in the Neotropics (Dunn

et al. 2005, Sauer et al. 2017). It is essential to optimize survey methodologies for poorly-

monitored species because conservation planning relies heavily on count data

(Rosenstock et al. 2002).

Multiple sources suggest an overall decline in Common Nighthawks (Chordeiles

minor) throughout much of their North American breeding range (Brigham et al. 2011,

Sauer et al. 2013). From 1966 to 2015, BBS data show a 1.93% annual decline in North

American Common Nighthawk populations (Sauer et al. 2017). A 1.82% annual decline

has been documented in the United States during this time period, along with a 3.41%

decline in Canada (Sauer et al. 2017). All of these values were significant declines. Based

on this alarming trend, the Common Nighthawk was listed as threatened in Canada in

2007 (COSEWIC 2007). Common Nighthawks are a member of the aerial insectivore

guild that is exhibiting range-wide declines, with especially sharp declines in northeastern

North America (Nebel et al. 2010).

Confidence in BBS-produced population trends for Common Nighthawk is low

because most surveys of breeding birds do not sample nighthawks and other nightjars

(family Caprimulgidae) at the appropriate times (Dunn et al. 2005). While BBS trends

may show accurate population declines, the reliability of these trends is diminished due to

low sample sizes, low precision, and low relative abundance, especially in the boreal

forest (Dunn et al. 2005). Nevertheless, the BBS is the only source of long-term

monitoring data for Common Nighthawks (Sauer et al. 2013). It is essential to remedy

3

this situation and develop a targeted long-term monitoring method to track population

trends in the Common Nighthawk, especially in the poorly-surveyed northern portion of

the range of this species, an area in which it may be undergoing the steepest declines

(Brigham et al. 2011).

To address some of the shortcomings with BBS and other typical point count data

for nightjar population estimates and trends, a nationwide program called the Nightjar

Survey Network (NSN) was established in 2007 to collect baseline data specifically

targeting nightjars (Wilson 2008). This survey was designed to maximize detections of

Eastern Whip-poor-wills (Caprimulgus vociferus), but it is uncertain whether the method

is useful for Common Nighthawks (Wilson 2008). Surveys begin at least 30 minutes after

sunset and end at least 15 minutes before sunrise. While the NSN protocol may be better

suited to detect Common Nighthawks than the BBS, it may be failing to obtain the

highest possible detection rates due to the natural history of this species.

Common Nighthawks are crepuscular and frequently forage for extended periods

of time by hawking insects in flight, while Eastern Whip-poor-wills are nocturnal and

typically forage from a low stationary perch, briefly sallying out to catch passing insects

(Brigham et al. 2011, Cink et al. 2017). These differences in foraging strategy and

behavior influence an observer’s ability to detect these species during a survey. Common

Nighthawks are easily detected visually during their foraging flights whereas most other

nightjars are rarely seen. Even a survey designed specifically for nightjars such as the

NSN may not be appropriate for sampling all members of the Caprimulgid family. The

NSN method may be inadequate for surveying Common Nighthawks because of this

species’ crepuscular activity and the survey’s lack of opportunities for visual detection.

4

Because the BBS and NSN surveys used to determine the population status of Common

Nighthawk are likely unable to sample this species properly, the utility of these data is

limited.

The objective of this study was to determine whether NSN surveys adequately

survey Common Nighthawks. I developed an alternative crepuscular survey methodology

using the same survey route to specifically target Common Nighthawks. I hypothesized

that nocturnal surveys will inadequately detect nighthawks because of their crepuscular

habits. Furthermore, the NSN survey is under-sampling Common Nighthawks because it

does not take into account the differential foraging strategy and diel activity of this

species. I predicted that surveys conducted during the crepuscular window of activity

would detect more Common Nighthawks than those conducted during the nocturnal

period.

METHODS

Study area

Nightjar surveys were conducted on secondary roads located in the Superior

(approximately 47.62 N, 91.35 W) and Chippewa (approximately 47.35 N, 94.22 W)

National Forests in the boreal hardwood transition bird conservation region of northern

Minnesota (U.S. Fish and Wildlife Service 1999). Routes were determined by the NSN

and were placed on existing BBS routes (Figure 1). Common Nighthawks are not

abundant breeders in this region, but can be observed foraging over bodies of water in the

evening (S.K. personal observation).

5

6

Nightjar surveys

Nightjar Survey Network routes were surveyed twice during each visit: once

during a crepuscular window (hereafter the crepuscular survey) and once during a

nocturnal window (hereafter the nocturnal survey). Both surveys were conducted

throughout June and each lasted approximately 90 minutes. Following NSN protocol

(Appendix 1), nocturnal surveys began at least 30 minutes after sunset and ended no later

than 15 minutes before sunrise. Each survey required the moon to be above the horizon

and not covered by clouds. Crepuscular surveys were initiated 60 minutes before sunset

and concluded 30 minutes after sunset. In this way, the crepuscular survey and nocturnal

survey could be run sequentially by a single observer.

Each route in this study consisted of ten survey points spaced one mile apart. Six

minutes were spent at every survey point during which time all nightjars detected either

visually or aurally from this stationary position were recorded. No playback, spotlighting,

or any other means to attract or detect nightjars was utilized. The survey window at each

point was partitioned into six one-minute segments. Detections of new and previously

detected individuals were both recorded each minute. Every attempt was made to detect

movement of individual birds during the survey to avoid double counting. Additionally, it

was noted whether a bird detected was thought to be the same individual from a previous

point. Standard meteorological and survey condition data were collected at each point

regardless of visit protocol used. Surveys were not conducted in rain or strong wind.

Statistical analyses

I analyzed the influence of survey time (crepuscular versus nocturnal) on

detection of Common Nighthawks. Comparisons between the crepuscular survey and the

7

nocturnal survey allowed examination of the influence of survey timing on detection of

Common Nighthawks. Wilcoxon signed-rank test was used to compare the incidence of

detection between surveys. All statistical tests were performed in R (R Core Team 2013).

RESULTS

In June 2016 and June 2017, 18 nightjar routes were surveyed both during the

crepuscular and nocturnal windows (Figure 1). Observers detected Common Nighthawks

on six of the 18 routes (33%) in 2016 and 2017 (Table 1). However, nighthawks were

only detected on 2.2% of all route stops. Eight total Common Nighthawks were counted

during the surveys. Two Common Nighthawks were detected on two routes; the

remaining four routes had single detections. Most of the routes on which nighthawks

were detected were in the northeastern region of the state (Figure 1). All eight

nighthawks were detected during the crepuscular survey and zero were detected during

the nocturnal survey window. A Wilcoxon signed-rank test indicated a significant effect

of time of survey (Z = -2.44, p = 0.03).

8

9

DISCUSSION

My results indicate that time of survey is critical when designing surveys

specifically targeting Common Nighthawks, especially in regions where this species is

uncommon such as the boreal forest. The crepuscular period is the optimal time to target

Common Nighthawks for the highest survey detection rates. My prediction was supported

because nighthawks were only detected during the crepuscular surveys. Information

about detection is essential to optimize survey methodology given limited resources and

the importance of count data for conservation planning (Rosenstock et al. 2002).

Common Nighthawks rely on acute vision to forage for flying insects in the late evening

and individuals are most active and vocal during the crepuscular period. The activity of

Common Nighthawks is constrained by the amount of ambient light as they do not utilize

echolocation and are unable to see in complete darkness (Brigham et al. 2011). Neither

BBS routes nor NSN routes provide coverage during this time of day.

This study is one of the first to investigate timing of surveys aimed at detecting

nightjars breeding in the boreal forest. More information about the threatened Canadian

populations of Common Nighthawks is needed (Brigham et al. 2011), and this study

presents useful findings about proper monitoring techniques in the region. It is necessary

to increase the sample size and relative abundance measures for this species in the boreal

forest (Dunn et al. 2005); increasing these metrics will provide more power and precision

in analyzing population trends. This is especially essential for Common Nighthawks as

this species is also inadequately surveyed by the CBC because it winters in South

America.

10

I suggest that observers conducting NSN routes run an antecedent route that

specifically targets Common Nighthawks and other crepuscular nightjars, especially

Lesser Nighthawk (Chordeiles acutipennis). While this new method would double the

amount of actual survey time required of volunteers, the number of visits to the survey

location would not change. Especially in remote areas, the cost of conducting surveys lies

in the logistics of reaching the survey location, not the survey itself. The amount of

participation would be analogous to that of a volunteer-run BBS route, of which there are

approximately 3,000 conducted each summer (Sauer et al. 2013).

I also suggest an expansion of the NSN in the boreal forest region. The boreal

hardwood transition and the boreal forest host declining populations of Common

Nighthawks that are poorly surveyed and even more difficult to detect than in other

regions due to the high percentage of survey locations situated in dense forest. It will

require a concerted and substantial effort to survey Common Nighthawks properly in this

region, but survey information is essential to understanding the decline of boreal forest

breeding nighthawks.

In heavily forested areas like my study region, additional survey methods could

be employed to increase Common Nighthawk detections. Conducting surveys in

locations with views of the sky, especially near or over water, would greatly enhance

nighthawk surveys, particularly if this species concentrates in these areas. Stationary

surveys for Common Nighthawks over bodies of water (e.g. from boat launches or other

public water access locations) during the short crepuscular activity window of this

species would likely detect more nighthawks than roadside surveys through heavily

forested areas. Telemetry data suggest that individuals in upland habitats will leave home

11

territories to forage over bodies of water (Brigham et al. 2011). Current methods for

surveying Caprimulgids do not take advantage of the highly aerial and visible foraging

strategy of Common Nighthawks.

Breeding surveys targeting aerial insectivores are limited or nonexistent, but the

sharp decline within this guild points to the need for intensive and focused monitoring.

Aerial insectivores are likely environmental indicators whose declines may be used as an

early warning about ecosystem health (Nebel et al. 2010, Smith et al. 2015). An

integrated approach that combines BBS data, targeted surveys, migration surveys (e.g.

visible migration counts; see Chapter 2), and wintering ground surveys will be the most

useful and powerful tool for monitoring populations of aerial insectivores in the future.

12

CHAPTER 2.

VISIBLE MIGRATION COUNTS OF COMMON NIGHTHAWKS

(CHORDEILES MINOR) ALONG THE NORTH SHORE OF LAKE SUPERIOR

INTRODUCTION

Accurate estimates of long-term trends are important for providing assessments of

population status, especially for species of conservation concern (Sauer et al. 2003).

Developing trends from traditional population monitoring methods such as breeding bird

surveys is frequently challenging for species that are secretive (i.e. difficult to detect),

occur in low densities, do not follow traditional paradigms of territorial singing in the

early morning hours, and/or have inaccessible breeding grounds (Dunn and Hussell

1995). Assessments are especially weak for populations that escape detection during both

the North American Breeding Bird Survey (BBS) and Christmas Bird Count (CBC)

because they breed in remote boreal forest and winter in the Neotropics (Dunn et al.

2005, Sauer et al. 2017). In such cases, alternative methods have been used to detect

population trends; visual counts of birds migrating between breeding and nonbreeding

grounds are one such method.

Use of long-term visible migration counts for population monitoring in North

America has been largely limited to raptors and other taxa that utilize soaring flight (e.g.

Hoffman and Smith 2003, Farmer et al. 2007, Martín et al. 2016). Species or guilds that

migrate diurnally and are poorly surveyed by traditional breeding bird surveys are

candidates for monitoring via visible migration counts. While past population trends

obtained from visible migration counts have generally corresponded with other

population monitoring efforts, the utility and validity of migration count data has been a

13

source of contention (e.g. Kerlinger 1989). Migration counts have been criticized for

reflecting variation caused by factors such as weather patterns that have little or no

relation to underlying population levels and trends (Dunn and Hussell 1995). However, in

recent years the use of migration counts has been more widely accepted as a way to either

provide basic information or, at minimum, additional information for population trend

development (Titus and Fuller 1990, Hoffman and Smith 2003).

Weather has large effects on bird migration, and the study of avian migration and

weather conditions has been fruitful for decades (reviewed in Richardson 1978, 1990;

Liechti 2006, Newton 2007). The number of birds migrating on a given day is influenced

by temperature, wind speed, wind direction, barometric pressure, humidity, precipitation,

and thermal development, among other elements (Richardson 1990, Liechti 2006). It is

difficult to assign causality because weather metrics are often strongly correlated

(Richardson 1990, Newton 2007). However, there is broad agreement that wind speed

and direction, temperature, and precipitation are among the most important predictors of

migratory activity. In autumn, birds are more likely to migrate during conditions with

falling temperatures, light and/or tail winds, and no precipitation (Richardson 1990,

Liechti 2006). Weather conditions that affect the volume of diurnally migrating raptors

and other soaring species have been well-studied, but those that affect other diurnally

migrating groups are less well known, especially in North America. While some authors

suggest it is not as important to account for weather in estimating trends as previously

thought (Farmer et al. 2007), at least for novel taxa it is vital to understand weather

influences on the migratory behavior of every species when using visible migration

14

counts to develop population trends (Hussell 1981, Richardson 1990, Shamoun-Baranes

et al. 2010).

Common Nighthawks (Chordeiles minor) are declining throughout North

America (Brigham et al. 2011, Sauer et al. 2013). Nighthawks are aerial insectivores, a

guild that may serve as biological indicators as they integrate ecosystem processes across

large geographic regions (Nebel et al. 2010, Smith et al. 2015). Aerial insectivores are

declining at a faster rate than other groups of birds in North America (Bohning-Gaese et

al. 1993, Nebel et al. 2010, Smith et al. 2015). However, confidence in these BBS-

derived population trends for Common Nighthawk is low because surveys of breeding

birds do not sample nighthawks and other nightjars (Caprimulgidae) at the appropriate

times (Dunn et al. 2005). Common Nighthawks breed in relatively low densities, are

secretive and crepuscular, and are poorly surveyed by the BBS and CBC. Consequently,

this species is insufficiently monitored, especially in the northern portion of its range

(Brigham et al. 2011). In this region, focused efforts to survey breeding nighthawks will

be logistically and financially difficult due to low road density and the short nighthawk

breeding season.

During early autumn, large numbers of Common Nighthawks habitually

concentrate and migrate along the north shore of Lake Superior near Duluth, Minnesota

(Hendrickson and Eckert 1991). Single-day flights of over 5,000 nighthawks are nearly

annual events, and these counts represent a sizeable portion of the northern Common

Nighthawk population. Common Nighthawks are a good candidate for use of visible

migration counts to assess population trends, and the western Lake Superior region is a

suitable location to study the effects of weather on migrating Common Nighthawks. The

15

migration and migratory behavior of Common Nighthawks has received little attention

(but see Taylor 1996, 2009).

I analyze visible migration counts of Common Nighthawks along the north shore

of Lake Superior using a 10-year data set. My goals were: (1) describe the largest known

migratory concentration of this species, (2) investigate and describe the relationship

between weather and Common Nighthawk migration intensity, and (3) assess the trend of

the population of nighthawks that moves through the region and if possible correct for

weather effects. I hypothesize that weather conditions that typically stimulate autumn

migration – falling temperatures, tailwinds, and light winds – will be correlated with

nighthawk migration. Additionally, I hypothesize that population trends derived from

visible migration counts of Common Nighthawks will mirror the significant decline for

this species found by the BBS.

METHODS

Study Area

Visible migration counts were conducted from the third story rooftop of a condominium

complex in Duluth, MN (46.83 N, 92.00 W) (Figure 1; Bardon 2012). At this site, most

migrant birds appear from the northeast, and the count location provides an unobstructed

view in this direction. Duluth sits at the western tip of Lake Superior, a large migration

corridor, and many birds concentrate along the shore during southbound autumn

migration as they avoid crossing the large body of water (Hofslund 1966).

16

17

Migration Counts

From 2008 to 2017, counts were conducted daily from August 15th to September

1st commencing at 16:00 CST and continuing until sunset; in some years counts were

made outside of this window. Counts were halted during periods of precipitation. The

number of nighthawks migrating per day was recorded by one or two experienced

observers using 7x to 10x binoculars. On most days, nighthawks were counted

individually, but during peak passage (e.g. 2,000+ birds/hour), large flocks were counted

by fives or tens. Every effort was made not to double-count the migrating nighthawks,

and this effort was aided by their straight-line flight during active migration – a behavior

very much unlike their foraging flights. Counters recorded all nighthawks flying in a

southwesterly direction. Historical counts of nighthawk flights in Duluth were compiled

from published accounts (e.g. Hendrickson and Eckert 1991).

Local Weather

Weather conditions were recorded at 16:00 CST and obtained from the National

Centers for Environmental Information (NCEI; https://www.ncdc.noaa.gov/). These data

were measured at the Duluth International Airport which is located 14 km west of the

count location. Weather covariates predicted to influence the migration of Common

Nighthawks were measured: temperature (°C), percent relative humidity, wind direction

(degrees), wind speed (kph), and barometric pressure (in. Hg). Humidity and temperature

values were highly correlated so humidity was eliminated from further consideration, as

these metrics were likely describing similar weather phenomenon. Wind direction in

degrees (polar degrees) was first converted into Cartesian degrees, then into radians, and

finally transformed using sine and cosine transformations following Fisher (1993):

18

!"#$""% = 450 − +,-!!,$"/0,1- ifwinddirection ≥ 90 1

!"#$""% = 450 − +,-!!,$"/0,1- − 360 ifwinddirection < 90 2

$C!,C- = !"#$""%D180

3

FC0,0G!,-CF+,-! HCFG" = sin $C!,C- 4

F1-#,0G!,-CF+,-! HCFG" = cos $C!,C- 5

This constrained wind direction to values between -1 and 1. To examine the

effects of latitudinal (East/West) winds, cosine values range from W = -1 to E = 1.

Likewise, to examine the effects of longitudinal (North/South) winds, sine values range

from S = -1 to N = 1. All meteorological values were normalized by z-score prior to

inclusion in models.

Statistical Analysis

Regression Modeling

Three analytic methods were used to investigate the influence of weather on the

migration of Common Nighthawks. I constructed generalized linear mixed effects models

(GLMMs) with weather and date variables as fixed effects and year as a random effect. I

first modeled by maximum likelihood the odds of a nighthawk “flight day” by fitting

binary logistic regression models on predictor weather variables. A flight day was

defined as a day on which >1000 migrating nighthawks were counted. Logistic regression

was employed to minimize concerns about modeling raw counts due to observer effects.

Additionally, logistic regression allowed inclusion of 21 historical flight days; I assumed

that the binary nature of logistic regression would accommodate protocol and observer

differences because any count method would be able to distinguish between a flight day

and a non-flight day. I also modeled the raw count of systematically counted nighthawks

19

from 2008 to 2017 on the predictor weather variables using negative binomial regression.

I used this method because the dependent variable consists of over-dispersed count data.

Finally, I constructed linear mixed effects models to predict the number of nighthawks

counted from 2008 to 2017. I log +1 transformed the dependent variable because it was

highly skewed with many more small counts than large counts. The addition of one was

to avoid problems associated with taking the log of zeros. I chose to investigate both raw

and log-transformed count data regression because there is disagreement about which

practice is best for count data (O’Hara and Kotze 2010, Ives 2015).

For all models, I included ordinal date (days from 1 January) and ordinal date2 as

fixed effects due to the non-linear effect of date over the three week migration season.

Three interactions with possible biological significance identified a priori were also

considered (latitudinal wind direction and wind speed, longitudinal wind direction and

wind speed, and latitudinal and longitudinal wind directions). I also investigated a set of

seven models developed a priori that were based on observations of migrating Common

Nighthawks, their natural history, and their aerial prey. For all models, I performed

manual backward selection on the global model to identify a best model among the

candidate set. Model selection was performed using AIC adjusted for small sample size

(AICc; Burnham and Anderson 2002). I considered a variable to have an effect on

Common Nighthawk migration if it was included in the top model and the 95%

confidence interval for the model coefficient failed to overlap zero. I also compared the

signs of coefficients from the various modeling techniques.

Trend Analysis

20

I used the daily visible migration counts to develop count indices and used these

to develop population trends for Common Nighthawk from 2008 to 2017. Count effort

varied among years, so a seasonal passage window of 95% was calculated and any counts

outside of this window were excluded. To analyze the ten-year trend of Common

Nighthawk counts in the study area, I used a geometric mean passage rate index

following Farmer et al. (2007):

F- JKL + 1 = CN + CLOL

P

LQR

+ "KL 6

whereJKL is the number of nighthawks counted on day i in year j, CN is the intercept

estimated by the regression, CLis the year coefficient estimated by the regression for each

year OL and "KL is the error of the regression. F- JKL + 1 is the transformed geometric

mean (TGM)j of nighthawks in year j. The transformed geometric mean was then back-

transformed into the geometric mean:

ST L = " UVW XYZ/\) − 1 7

where V is the variance of the raw counts pooled over all years (Farmer et al. 2007).

To account for weather effects on the counts of migrating nighthawks, I also used

a date-adjusted and weather-adjusted geometric mean daily count following Hussell

(1981) and Farmer et al. (2007) with the general model:

F- JKL + 1 = CN + CLOL + _`,` + !abaKL + "KL

c

aQR

d

`QR

P

LQR

8

where JKL is the count of nighthawks on day i in year j; CN is the intercept estimated by

the regression, CL, _`, and !a are parameters of year, day, and weather covariates

estimated by the regression, respectively. The geometric mean was calculated as above

21

for each day and then averaged over each year. I analyzed both of these population trends

using linear regression with standard t-tests to assess trend significance. I performed all

statistical tests in R (R Core Team 2013).

RESULTS

Between 2008 and 2017, 194,721 migrating Common Nighthawks were counted

in 645 hours and 202 observation days (containing 37 flight days), for an average yearly

count of nearly 19,500. At least one nighthawk was detected on 76% of all days. The

mean daily count was 937 ± 2182 sd nighthawks/day (Table 1). Peak flights during the

ten-year study period included 28,054 on Sept 1, 2015; 14,081 on August 21, 2013; and

10,403 on August 19, 2016 (Table 1). Cedar Waxwings (Bombycilla cedrorum), Cliff

Swallows (Petrochelidon pyrrhonota), and dragonflies (order Odonata) were the only

other taxa that strongly participated in concurrent evening flights.

An additional 21 flight days from years between 1985 and 2007 were added for

use in logistic regression and at least one count was used from 22 different years. Peak

historical flights included 43,690 on August 26, 1990 (the largest flight ever recorded;

Hendrickson and Eckert 1991) and 15,173 on August 23, 2000. Days of >1000

nighthawks occurred as early as August 15th and as late as September 2nd (Figure 2).

Table 1 shows annual and overall average nighthawk counts, temperature, humidity, wind

speed, and pressure from 2008 to 2017. Additionally, conditions during all flight days

during this time period are listed.

Logistic regression analysis identified weather factors associated with the

occurrence of nighthawk flights. Results indicate that significant covariates were

latitudinal and longitudinal wind direction and their interaction, wind speed, pressure, and

22

temperature (Table 2). Similarly, both negative binomial and linear mixed models

indicated that latitudinal wind and wind speed and their interaction, as well as pressure,

and temperature were significant (Table 3, Table 4). The best models from all analyses

included latitudinal winds, wind speed, pressure, and temperature (Table 5).

I found effects of four model coefficients from all top models, including

latitudinal winds, wind speed, pressure, and temperature (Table 5). Common Nighthawk

counts were higher on days with westerly winds, lower wind speeds, lower pressure, and

higher temperatures (Table 5, Figure 3, Figure 4). Westerly winds appear to be the most

important weather variable in all top models followed by wind speed and finally

temperature (Table 5). Regardless of assumed error distribution, all coefficients share the

same sign for each covariate: negative for pressure, wind speed, and wind direction

(indicating westerly component), and positive for temperature. The coefficient for the

interaction between longitudinal and latitudinal winds was negative in two models,

perhaps indicating a slight preference for migration on northwesterly winds (Table 5).

The 95% passage window used for trend analysis from 2008 to 2017 fell between

August 15th and September 1st (ordinal days 227-244). Trends for Common Nighthawk

migration in Duluth based on the geometric mean passage rate and the date-adjusted and

weather-adjusted geometric daily count are shown in Figures 5 and 6. Both methods

show stable or non-significantly increasing trends from 2008 to 2017.

23

24

25

26

27

28

29

30

31

32

33

34

DISCUSSION

I modeled the effects of weather on the visible migration counts of Common

Nighthawks along the north shore of Lake Superior during autumn. The overall

agreement of my best models using various regression techniques adds confidence in my

interpretations. Contrary to my predictions, Common Nighthawks preferred to migrate

during days with warmer temperatures (Figure 3). This observation is surprising because

the number of birds migrating in the autumn typically peaks during days of colder

temperatures (Richardson 1990, Newton 2007). This species’ preference for westerly

winds was also contrary to my predictions. Westerly winds act as a headwind for birds

migrating in a southwesterly direction along the shore of Lake Superior, and migrant

birds often avoid these conditions (Richardson 1990). In addition, it is surprising that no

significant difference was detected between northerly and southerly winds because

northerly winds are generally correlated with autumn migratory flights. Regional

geography may partially explain the strong preference for wind with a westerly

component: west winds push southerly migrating birds toward Lake Superior and east

winds push them away from the lake and out of sight, no longer available to be counted.

Light wind speeds are often associated with larger migratory movements of non-raptors,

and this holds true for Common Nighthawks migrating in this region (Figure 3).

Weather conditions associated with major flights of Common Nighthawks in the

study area are also conducive to the presence of aerial insects in the migratory airspace

(e.g. Finlay 1976). A positive correlation between temperature and aerial insect

abundance has been well established (Hardy and Milne 1938, Glick 1939). Abundance of

flying insects decreases with increasing wind speed (Glick 1939, Møller 2013).

35

Richardson (1978) reviews observations of upwind movements of aerial insectivores that

were presumably due to aerial prey availability or accessibility. These observations

combined with results from my fitted models indicate the importance of aerial insects for

migrating nighthawks. Common Nighthawks may time their migratory movements along

the north shore of Lake Superior during the short migratory window each autumn to

maximize foraging efficiency on days with large numbers of aerial insects. Migratory

flight requires deposition of subcutaneous fat and thus a marked increase in food

consumption (Newton 2007). Past reports have noted very large amounts of certain insect

species in the guts of migrating Common Nighthawks, indicating the importance of insect

hatches for autumn migratory movements (Blem 1972). Common Nighthawks may

concentrate on the Lake Superior shore because of late summer insect hatches in this

location, as this species seems to be an opportunistic predator of abundant aerial insects.

It is also conceivable that nighthawks are following migrating insects: on many days,

large numbers of small unidentified insects were detected flying directionally (i.e.

following the same path as migrating birds) along the shore of Lake Superior (S.K.

personal observation). Future studies should focus on the insectivorous habits of this

species during migration.

Peak Common Nighthawk movements along the North Shore of Lake Superior

were highly synchronous. All large flights occurred between August 15th and September

2nd with a steep ramp-up and drop-off on either side of this window (Figure 2). This

points to strong endogenous and/or exogenous control of migratory behavior (Newton

2007). Nighthawks appear to have a tightly regimented passage window in which they

choose to migrate when conditions are optimal. These conditions may be weather and/or

36

food related. The distribution of nighthawk passage was highly clumped on days with

certain weather characteristics. The current study is one of the first to quantify weather

conditions that lead to a migratory flight of Common Nighthawks, and this information is

useful for predicting likely peak passage days. Monitoring must occur on flight days, and

a single count on these days would detect a high percentage of the annual seasonal

nighthawk movement.

Contrary to my prediction, Common Nighthawk migratory counts in this region

do not mirror the declines from BBS surveys with or without weather adjustments

(Figures 5, 6). The limited time period of my counts may be unable to detect a small

decline, but should be able to detect a significant or catastrophic decline. There is some

indication that Common Nighthawk populations are stabilizing, and this may also play a

role in the ability to detect a trend. As with raptor trends (Farmer et al. 2007), adjusting

for weather effects does not appear necessary for producing robust population trends for

Common Nighthawks in the study region.

Major knowledge gaps exist in the understanding of the migratory period in

Common Nighthawks that, if filled, would enhance the use of visible migration counts of

nighthawks for trend development. For example, the catchment area of migrant

nighthawks through this region in autumn is unknown. Additional research using tracking

and/or modeling techniques is needed to address this issue. This region does not

experience a concentrated return flight of Common Nighthawks in the spring. Research

using tracking technology or exploratory spring visible migration counts to identify

spring movement corridors would be beneficial in understanding the full life cycle of this

species, especially during the migratory period. Limited anecdotal reports and the

37

continental trend of a more westerly spring migration point to river corridors west of the

Great Lakes region as the most likely geographic feature that would concentrate

migrating Common Nighthawks in the spring. Investigation of return flights is especially

important in population monitoring (Farmer and Smith 2010).

In this study, I describe the largest known concentration of Common Nighthawks

in the world. The airspace over Duluth and the north shore of Lake Superior is

disproportionally critical habitat to the conservation of Common Nighthawks during the

autumn migratory period each year (Diehl 2013, Peterson et al. 2015), especially on peak

flight days. Migration is the most dangerous season of the annual cycle for migrant birds

(Sillett and Holmes 2002, Klaassen et al. 2014). Specific dangers to Common

Nighthawks in the study region during migratory flight could include towers, wind

turbines, and automobile collisions. On some days, Common Nighthawks have been

struck by traffic in Duluth (S.K. personal observation). Thus, it is essential to understand

where, when, and why threatened and declining species such as Common Nighthawks

spend time and concentrate during migration.

The utility of visible migration counts for population monitoring is most obvious

for diurnal migrants that are difficult to monitor with traditional methods. Migration

counts can often be conducted by skilled volunteers and require little cost to implement

(Bildstein 1998). Migration counts are an efficient and cost-effective way to monitor the

population of Common Nighthawks that migrates along the north shore of Lake Superior

each autumn. Given that large concentrations of this species occur predictably each year

in this region and that other monitoring efforts (such as the BBS or targeted surveys) are

lacking, Common Nighthawks are an ideal species for which to utilize visible migration

38

counts for population monitoring. Common Nighthawk breeding surveys in the study

region in 2016 and 2017 detected only eight individuals in 36 hours of surveying (S. K.

unpublished data; see Chapter 1). To detect the seasonal migration average of 19,500

nighthawks would take a monumental effort on the breeding grounds, but can be

achieved by one observer in less than three weeks in the autumn. Other locations that

detect migrating Common Nighthawks should implement or continue nighthawk

monitoring so that observers can form a network like that developed for hawk migration

throughout North and Central America (Bildstein 1998). In fact, many hawk migration

sites could be used for these counts in the evenings given the proper coverage. Using the

study of migratory raptors as a guide, an integrated approach of breeding, nonbreeding,

and migration surveys will enhance the reliability of population trends for this species

(Dunn and Hussell 1995, e.g. Paprocki et al. 2017).

Visible migration counts aimed at estimating population trends should not be

limited to raptors and other soaring birds. This study is one of the first to use these types

of counts to monitor populations of non-raptors in North America, and this method shows

promise for other species or groups. Visible migration counts will be useful for avian

species or families such as the rapidly declining aerial insectivores (swifts, swallows,

other nighthawks), waxwings, Blue Jays (Cyanocitta cristata), and blackbirds because

these species migrate diurnally and BBS-type surveys fail to monitor them effectively.

Migration counts can be a time-effective and cost-effective way to augment information

about poorly surveyed species and, much like the BBS, can be accomplished through

citizen science.

39

LITERATURE CITED

Bardon, K. J. (2012). Fall diurnal migration of passerines and other non-raptors at Hawk Ridge and Lake Superior, Duluth (2007-2011). The Loon 84:8-20. Bart, J., K. P. Burnham, E. H. Dunn, and C. M. Francis (2004). Goals and strategies for

estimating trends in landbird abundance. The Journal of Wildlife Management 68:611–626.

Bildstein, K. L. (1998). Long-term counts of migrating raptors : A role for volunteers in

wildlife research. The Journal of Wildlife Management 62:435–445. Blem, C. R. (1972). Stomach capacity in the Common Nighthawk. The Wilson Bulletin

84:492–493. Bohning-Gaese, K., M. L. Taper, and J. H. Brown (1993). Are declines in North

American insectivorous songbirds due to causes on the breeding range? Conservation Biology 7:76–86.

Brigham, R. M., J. Ng, R. G. Poulin, and S. D. Grindal (2011). Common Nighthawk

(Chordeiles minor) version 2.0. In The Birds of North America Online (P.G. Rodewald, editor). Cornell Lab of Ornithology, Ithaca, New York, USA. https:doi.org/10.2173/bna.213.

Burnham, K. P., and D. R. Anderson (2002). Model selection and multimodel inference:

A practical and information-theoretic approach. 2nd Ed. Springer, New York. Cink, C. L., P. Pyle, and M. A. Patter (2017). Eastern Whip-poor-will (Antrostomus

vociferus), version 3.0. In The Birds of North America (P.G. Rodewald, editor). Cornell Lab of Ornithology, Ithaca, New York, USA.

COSEWIC (2007). COSEWIC assessment and status report on the Common Nighthawk

Chordeiles minor in Canada. Committee on the Status of Endangered Wildlife in Canada 2007.

Diehl, R. H. (2013). The airspace is habitat. Trends in Ecology and Evolution 28:377–

379. Dunn, E., B. Altman, J. Bart, C. Beardmore, H. Berlanga, P. Blancher, G. Butcher, D.

Demarest, R. Dettmers, W. Hunter, E. Iñigo-Elias, et al. (2005). High priority needs for range-wide monitoring of North American landbirds. Partners in Flight Technical Series No. 2.

Dunn, E. H., and D. J. T. Hussell (1995). Using migration counts to monitor landbird

populations: Review and evaluation of current status. In Current Ornithology (D. M. Power, Editor). Springer Boston MA.

40

Dunn, E. H., D. J. T. Hussell, and R. J. Adams (1997). Monitoring songbird population

change with autumn mist netting. The Journal of Wildlife Management 61:389–396. Farmer, C. J., D. J. T. Hussell, and D. Mizrah (2007). Detecting populations trends in

migratory birds of prey. The Auk 124:1047–1062. Farmer, C. J., and J. P. Smith (2010). Seasonal differences in migration counts of raptors:

Utility of spring counts for population monitoring. Journal of Raptor Research 44:101–112.

Finlay, J. C. (1976). Some effects of weather on Purple Martin activity. The Auk 93:231–

244. Fisher, N.I. (1993). Statistical analysis of circular data. Cambridge University Press. Glick, P. A. (1939). The distribution of insects, spiders, and mites in the air. Technical

Bulletin No. 673. Hardy, A. C., and P. S. Milne (1938). Studies in the distribution of insects by aerial

currents. Journal of Animal Ecology 7:199–229. Hendrickson, M., and K. Eckert (1991). A record movement of Common Nighthawks.

The Loon 63:68–69. Hoffman, S., and J. Smith (2003). Population trends of migratory raptors in western

North America, 1977-2001. The Condor 105:397–419. Hofslund, P.B. (1966). Hawk migration over the western tip of Lake Superior. The

Wilson Bulletin 78:79-87. Hussell, D. J. T. (1981). The use of migration counts for monitoring bird populations

levels. Studies in Avian Biology 6:92–102. Ives, A. R. (2015). For testing the significance of regression coefficients, go ahead and

log-transform count data. Methods in Ecology and Evolution 6:828–835. Kerlinger, P. (1989). Flight strategies of migrating hawks. University of Chicago Press. Klaassen, R. H. G., M. Hake, R. Strandberg, B. J. Koks, C. Trierweiler, K. M. Exo, F.

Bairlein, and T. Alerstam (2014). When and where does mortality occur in migratory birds? Direct evidence from long-term satellite tracking of raptors. Journal of Animal Ecology 83:176–184.

Liechti, F. (2006). Birds: Blowin’ by the wind? Journal of Ornithology 147:202–211. Martín, B., A. Onrubia, A. de la Cruz, and M. Ferrer (2016). Trends of autumn counts at

41

Iberian migration bottlenecks as a tool for monitoring continental populations of soaring birds in Europe. Biodiversity and Conservation 25:295–309.

Møller, A. P. (2013). Long-term trends in wind speed, insect abundance and ecology of

an insectivorous bird. Ecosphere 4:1–11. National Oceanic and Atmospheric Administration. https://www.ncei.noaa.gov/ (2018).

National Centers for Environmental Information. Nebel, S., A. Mills, J. D. Mccracken, P. D. Taylor, S. Nebel, A. Mills, J. D. Mccracken,

and P. D. Taylor (2010). Declines of Aerial Insectivores in North America Follow a Geographic Gradient. Avian Conservation and Ecology 5:1.

Newton, I. (2007). The migration ecology of birds. London: Academic Press. O’Hara, R. B., and D. J. Kotze (2010). Do not log-transform count data. Methods in

Ecology and Evolution 1:118–122. Paprocki, N., D. Oleyar, D. Brandes, L. Goodrich, T. Crewe, and S. W. Hoffman (2017).

Combining migration and wintering counts to enhance understanding of population change in a generalist raptor species, the North American Red-tailed Hawk. The Condor 119:98–107.

Peterson, A. C., G. J. Niemi, and D. H. Johnson (2015). Patterns in diurnal airspace use

by migratory landbirds along an ecological barrier. Ecological Applications 25:673–684.

R Core Team (2013). R: A language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.

Ralph, C. J., J. R. Sauer, and S. Droege (1995). Monitoring bird populations by point

counts. USDA Forest Service General Technical Report:1–181. Richardson, W. J. (1978). Timing and amount of bird migration in relation to weather: a

review. Oikos 30:224–272. Richardson, W. J. (1990). Timing of bird migration in relation to weather: updated

review. In Bird Migration (G. E., Editor). Springer, Berlin, Heidelberg, pp. 78–101. Romesburg, H. C. (1981). Wildlife science: Gaining reliable knowledge. The Journal of

Wildlife Management 45:293–313. Rosenstock, S. S., D. R. Anderson, K. M. Giesen, T. Leukering, and M. F. Carter (2002).

Landbird Counting Techniques: Current Practices and an Alternative. Source: The Auk 119:46–53.

42

Sauer, J. R., J. E. Fallon, and R. Johnson (2003). Use of North American Breeding Bird Survey data to estimate population change for bird conservation regions. Journal of Wildlife Management 67:372–389.

Sauer, J. R., W. A. Link, J. E. Fallon, K. L. Pardieck, and D. J. Ziolkowski (2013). The

North American Breeding Bird Survey 1966–2011: Summary analysis and species accounts. North American Fauna 79:1–32.

Sauer, J. R., D. K. Niven, K. L. Pardieck, D. J. Ziolkowski, and W. A. Link (2017).

Expanding the North American Breeding Bird Survey analysis to include additional species and regions. Journal of Fish and Wildlife Management 8:154–172.

Shamoun-Baranes, J., W. Bouten, and E. E. Van Loon (2010). Integrating meteorology

into research on migration. Integrative and Comparative Biology 50:280–292. Sillett, T. S., and R. T. Holmes (2002). Variation in survivorship of a migratory songbird

throughout its annual cycle. Journal of Animal Ecology 71:296–308. Smith, A. C., M.-A. R. Hudson, C. M. Downes, and C. M. Francis (2015). Change Points

in the population trends of aerial-insectivorous birds in North America: Synchronized in time across species and regions. PloS ONE 10:1–23.

Taylor, P. (1996). Apparent decline of migrant Common Nighthawks near Pinawa,

Manitoba. Blue Jay 54:35–38. Taylor, P. (2009). Late-summer feeding and migration behaviour and numerical trends of

Common Nighthawks, Chordeiles minor, near Pinawa, Manitoba, 1976-2009. Canadian Field-Naturalist 123:338–345.

Titus, K., and M. R. Fuller (1990). Recent trends in counts of migrant hawks from

northeastern North America. The Journal of Wildlife Management 54:463–470. U.S. Fish and Wildlife Service. (1999). A proposed framework for delineating

ecologically-based planning, implementation, and evaluation units for cooperative bird conservation in the U.S. U.S. Fishn and Wildlife Service, Washington, D.C., USA.

Wilson, M. D. (2008). The Nightjar Survey Network: program construction and 2007

Southeastern nightjar survey results. Technical Report Series CCBTR-08-001. Center for Conservation Biology, Williamsburg, Virginia, USA.

43

44

45

46

47