Survival and hunter reporting rate estimates of mourning doves based on 2003-2013 Pennsylvania bird banding data 1 S.M. Amburgey 1 , A. Barenblitt 2 , K.M. Serno 2 , M. Toenies 1 , J. Vernon 3 , 1 Intercollege Graduate Degree Program in Ecology, The Pennsylvania State University, Forest Resources Building, University Park PA 16823 2 Department of Ecosystem Sciences and Management, The Pennsylvania State University, Forest Resources Building, University Park, PA 16823 3 The Graduate Program in Acoustics, The Pennsylvania State University, Applied Science Building, University Park, PA 16823 1 This report was conducted as a class project for WFS 560, Population Estimation and Modeling, under the direction of Duane R. Diefenbach.
16
Embed
Survival and hunter reporting rate estimates of mourning ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Survival and hunter reporting rate estimates of mourning doves based on 2003-2013 Pennsylvania bird banding data1
S.M. Amburgey1, A. Barenblitt2, K.M. Serno2, M. Toenies1, J. Vernon3,
1 Intercollege Graduate Degree Program in Ecology, The Pennsylvania State University, Forest Resources Building, University Park PA 16823
2 Department of Ecosystem Sciences and Management, The Pennsylvania State University, Forest Resources Building, University Park, PA 16823
3 The Graduate Program in Acoustics, The Pennsylvania State University, Applied Science Building, University Park, PA 16823
1 This report was conducted as a class project for WFS 560, Population Estimation and Modeling, under the direction of Duane R. Diefenbach.
INTROD
M
American
annual su
are widel
(Seaman
their rang
southwar
hunting p
traveled b
Figure 1: M
DUCTION
Mourning dov
n birds (KDW
urvival and h
ly distributed
s and Sander
ge. Mournin
rd to Mexico
practices or o
between sum
Mourning dov
ves (Zenaida
WPT, 2014)
hunter report
d throughou
rs, 2014; Fig
g doves may
o and Centra
other threats
mmer and wi
e range map (N
a macroura)
), and therefo
ting rates for
ut North Ame
g. 1); thus na
y reside in th
al America (A
s. As surviva
inter ranges
NatureServe, 2
) are one of t
ore it is impo
r effective p
erica and int
ational-scale
heir breeding
Aldrich, 199
al and harves
and range of
007)
the most freq
ortant to obt
opulation m
to Canada an
e, robust data
g range throu
93), exposing
st rates likely
f origin, regi
quently hunt
tain reliable
management.
nd Central A
a are availab
ughout winte
g these birds
y vary based
ional analys
ted North
estimates of
Mourning d
America
ble for much
er or migrate
s to various
d on distance
es are impor
f
doves
of
e
e
rtant.
Mourning dove abundances have generally increased over the last 48 years in the eastern
portion of the range, but in the last five years, abundance estimates were stable (Seamans and
Sanders, 2014). Pennsylvania mourning dove populations appear stable based on recent
assessments despite a significant increase in calling along survey routes reported in 2012
(Seamans and Sanders, 2012). To better inform managers and assess current practices, regular
analysis of banding data should occur as data become available. The Pennsylvania Game
Commission, via the U.S. Geological Survey’s Bird Banding Lab, tracks mortalities of animals
as reported by hunters (including year and cause of death) and records information about the sex
(Male, Female, Unknown), age (Hatch Year [HY], After Hatch Year [AHY], Unknown), and
region (Piedmont, Non-Piedmont) where animals are banded and harvested (see Methods for
more details).
When modeling survival and reporting rates, better estimation of responses to hunting
can be obtained by using demographic and regional information. Male and female survival and
reporting rates are expected to be relatively similar due to shared parental duties (e.g. incubation
and feeding) (Hitchcock and Mirarchi, 1984) and similar hunter harvest rates are likely because
the sexes are indistinguishable in the field.. The primary difference between Piedmont and non-
Piedmont regions is of primary importance for this report. This roughly represents the Northwest
and Southeast portions of Pennsylvania (Fig. 2). Hunting pressure is hypothesized to be different
between the regions, with higher hunting rates in the Piedmont due to an increased amount of
farmland ideal for hunting (L. Williams, pers. comm.).
Figure 2: Mnon-Piedm
METHO
W
survival a
before th
birds. So
Bands ar
for mour
W
recovery
regions (
Map of Pennsymont for purpos
ODS
We utilized d
and reportin
he hunting se
me birds ban
re routinely a
ning doves t
We used ArcM
regions of a
Fig. 3). Thi
ylvania regionsses of the analy
data collected
ng rates of m
eason and du
nded in 2004
applied to do
typically last
Map 10.2.2
all individua
is allowed us
s. The Piedmonysis (Zimmerm
d by L. Willi
ourning dov
uring the hun
4 received re
oves in the su
ts from Sept
(ESRI 2014)
ls, separatin
s to test regio
nt region is labman, 2011)
iams (Penns
ves from 200
nting season
eward bands
ummer betw
tember to Jan
) to plot prov
g mourning
onal differen
beled, with all o
sylvania Gam
03-2013. Do
hunters repo
s, but this inf
ween July and
nuary (Seam
vided coordi
doves into P
nces in survi
other regions b
me Commiss
oves were ba
orted harves
formation w
d August. H
mans and San
inate data of
Piedmont an
ival.
being summariz
sion) to asse
anded each y
sted or dead
was not provi
Hunting seas
nders, 2014)
f banding an
nd Non-Piedm
zed as
ss
year
ded.
son
.
nd
mont
Figure 3: M
W
rates by r
a Seber D
reporting
interval.
(i.e. (1-S)
cannot be
D
banding l
Map of Pennsy
We used Prog
region, year,
Dead Recove
g rates (r) ba
Recovery ra
)*r), but due
e estimated.
Due to limite
locations oc
ylvania depictin
gram MARK
, sex, or a co
ery model (S
sed on numb
ates are depe
e to a lack of
d data, only p
curred throu
ng all banding
K (White & B
ombination o
Seber, 1970)
ber of doves
endent on th
f reward ban
parameters w
ughout sever
locations betw
Burnham, 19
of parameter
, allowing fo
banded and
he individual
nd informatio
with sufficient
al ecoregion
ween 2003 and
999) to estim
rs (Table 1).
or estimation
d dead doves
l doves being
on across eac
t data availabl
ns within Pen
2013
mate surviva
We analyze
n of survival
s recovered (f
g killed and
ch year of th
le were consi
nnsylvania, w
l and reporti
ed the data us
l (S) and
(f) for each t
being report
he project, f
dered. Whil
we grouped
ing
sing
ime
ted
le
banding locations into the Northwest and Southeast regions based on hypotheses regarding
hunting pressure. Due to limited numbers of reported doves in all categories, only AHY animals
were used for our initial analysis. Only doves with known sex were included in the analysis.
Table 1: Description of original models run using Program Mark
Model* Description S(s)r(t) Survival varies by sex, reporting varies by time (years) S(.)r(t) Survival is constant, reporting varies by time (years) S(.)r(reg) Survival is constant, reporting varies by region S(t+reg)r(.) Survival varies by time and region, reporting is constant S(t)r(.) Survival varies by time, reporting is constant S(t*reg)r(.) Survival varies by time by region, reporting is constant S(s)r(.) Survival varies by sex, reporting is constant S(.)r(s) Survival is constant, reporting varies by sex S(reg)r(.) Survival varies by region, reporting is constant S(t)=r(t) Survival varies by time, reporting varies by time S(t*reg)=r(t*reg) Survival varies by time by region, reporting varies by time by region S(reg)=r(reg) Survival varies by region, reporting varies by region
We observed no support for a difference in survival by sex and were able to further
investigate the potential differences by age groups by including unknown sex animals in our
dataset. A secondary set of models was run focusing on differences in age, region, and time
(Table 2).
We ran bootstrap goodness of fit (GOF) tests using the most parameterized models in
both the original and secondary sets of models. We were unable to run a GOF test on a more
global model with all parameters varying due to a lack of convergence of such a model. Model
selection was performed using QAICc as an adjustment for small sample size and a lack of
model fit (Burnham and Anderson, 2002).
Table 2: Description of secondary models run after adding in unknown individuals
Model Description
{S(age) r(t)} Survival varies by age, reporting varies by time (years) {S(t*reg) r(.)} Survival varies by time (years) by region, reporting is constant {S(age) r(.)} Survival varies by age, reporting is constant {S(t+reg) r(.)} Survival varies by time (years) and region, reporting is constant {S(t) r(.)} Survival varies by time (years), reporting is constant {S(.) r(age)} Survival is constant, reporting varies by age {S(.) r(t)} Survival is constant, reporting varies by time (years) {S(reg) r(.)} Survival varies by region, reporting is constant {S(.) r(reg)} Survival is constant, reporting varies by region {S(age) r(reg)} Survival varies by age, reporting varies by region {S(t) = r(t)} Survival varies by time (years), reporting varies by time (years) {S(t*reg) = r(t*reg)}
Survival varies by time (years) by region, reporting varies by time (years) by region
{S(.) r(age)} Survival is constant, reporting varies by age {S(reg) = r(reg)} Survival varies by region, reporting varies by region {S(t*age) r(t*age)}
Survival varies by time (years) by age, reporting varies by time (years) by age
We also investigated other model structures on reporting rate for both original and
secondary sets of models based on visualizations of reporting rates over the years (See Results,
Fig. 6 for more detail). This model had survival varying by age and reporting rates for the years
2003, 2010-2013 held equal and reporting rates for 2004 – 2009 held equal (S(age)
r(1=8,9,10,11, 2-7)).
RESULTS
The total number of mourning doves banded from 2003-2013 was 15,570 (Fig. 4). Of the
banded birds, 2,437 were banded in the Piedmont. Only 87 of the 532 reported birds were
reported in the Piedmont.
Figure 4: T
E
banded a
Our seco
reported
although
Total animals b
Excluding an
animals (3,69
ondary analy
(approximat
fewer birds
banded per yea
imals with u
90 males, 1,0
sis was cond
tely 4%). On
were origin
ar by region (to
unknown sex
003 females)
ducted on 14
nly 83 of the
nally banded
op) and reporte
x, our origina
), of which o
4,703 banded
e 586 reporte
in the Piedm
ed per year by r
al analyses w
only 197 wer
d animals, of
ed birds were
mont (2,215)
region (bottom
were conduc
re reported (
f which only
e from the P
).
m)
cted on 5,369
(3.7%; Fig. 5
y 586 were
Piedmont reg
9
5).
gion,
Figure 5: T
T
our hypo
bootstrap
model c-
reporting
were held
7)). From
rates wer
(non-Pied
Confiden
Total numbers
To investigate
otheses conce
p GOF test o
hat by 2.471
g rates that v
d equal and r
m this model
re found to b
dmont, years
nce Intervals
of banded and
e differences
erning mour
on the most p
1. Our top ra
ary by regio
reporting rat
S is estimat
be 0.0519 (Pi
s 1, 8-11), an
s are provide
d reported male
s in sex, regi
rning dove su
parameterize
anked model
ons and by ye
tes for 2004
ted as 0.528
iedmont, yea
nd 0.030 (no
ed in Table 4
e and female bi
ion, and tim
urvival (S) a
ed model (S(
(w = 0.999)
ears: reportin
– 2009 were
(SE ± 0.041
ars 1, 8-11),
on-Piedmont
4.
irds per region
e, we ran mo
and reporting
(t*reg)r(.)) p
) allowed for
ng rates for t
e held equal
, 95% CI =
0.0243 (Pie
t, years 2-7).
n from 2003-20
odels (n = 13
g rate (r) (Ta
prompted us
r a constant
the years 20
(S(.) r(reg,
0.446 – 0.60
edmont, year
. Standard er
013
3) to investig
able 3). The
to adjust the
survival rate
003, 2010-20
1=8,9,10,11
09). Reportin
rs 2-7), 0.118
rrors and 95
gate
e
e and
013
, 2-
ng
8
%
Table 3: Survival and reporting rate original models for mourning doves banded in Pennsylvania from 2003 – 2013