Standardized North American Marsh Bird Monitoring Protocol
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Standardized North American Marsh Bird Monitoring Protocol
COURTNEY J. CONWAY1
U.S. Geological Survey, Arizona Cooperative Fish and Wildlife Research Unit, 325 Biological Sciences East,
School of Natural Resources and the Environment, University of Arizona, Tucson, Arizona, 85721,USA
1
Present address: U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit, P.O. Box 441141,
CNR Room 103E 6th & Line Streets, University of Idaho, Moscow, ID 838444, USA
E-mail: cconway@usgs.gov
Abstract.—Little is known about the population status of many marsh-dependent birds in North America but recent
efforts have focused on collecting more reliable information and estimates of population trends. As part of that effort, a
standardized survey protocol was developed in 1999 that provided guidance for conducting marsh bird surveys through-
out North America such that data would be consistent among locations. The original survey protocol has been revised to
provide greater clarification on many issues as the number of individuals using the protocol has grown. The Standardized
North American Marsh Bird Monitoring Protocol instructs surveyors to conduct an initial 5-minute passive point-count
survey followed by a series of 1-minute segments during which marsh bird calls are broadcast into the marsh following a
standardized approach. Surveyors are instructed to record each individual bird from the suite of 26 focal species that are
present in their local area on separate lines of a datasheet and estimate the distance to each bird. Also, surveyors are re-
quired to record whether each individual bird was detected within each 1-minute subsegment of the survey. These data
allow analysts to use several different approaches for estimating detection probability. The Standardized North American
Marsh Bird Monitoring Protocol provides detailed instructions that explain the field methods used to monitor marsh
birds in North America. Received 26 January 2011, accepted 2 April 2011.
Key words.—bitterns, call-broadcast surveys, detection probability, marsh birds, rails, tape playback.
Waterbirds 34(3): 319-346, 2011
The amount of emergent wetland habitat ican Bittern (Botaurus lentiginosus), Least Bit-
in North America has declined sharply during tern (Ixobrychus exilis), Pied-billed Grebe
the past century (Tiner 1984; Dahl 2006; Sted- (Podilymbus podiceps), Limpkin (Aramus
man and Dahl 2008). Populations of many guarauna), American Coot (Fulica americana),
marsh birds that are dependent on emergent Purple Gallinule (Porphyrula martinica) and
wetlands appear to be declining (Tate 1986; Common Moorhen (Gallinula chloropus). The
Eddleman et al. 1988; Conway et al. 1994; Con- U.S. Fish and Wildlife Service (USFWS) has
way and Sulzman 2007). Despite evidence of identified Black Rails, Yellow Rails, Limpkins
population declines and the need to set re- and American Bitterns as Birds of Conservation
sponsible harvest limits, a monitoring pro- Concern because they are relatively rare and
gram specifically designed to determine sta- basic information on status and trends is lack-
tus and estimate population trends of marsh ing in most areas (U.S. Fish and Wildlife Ser-
birds is lacking. The North American Breed- vice 2008). Moreover, Yellow Rails, Black
ing Bird Survey includes survey data on some Rails, Clapper Rails and King Rails are four of
secretive marsh birds, but does not adequate- the 139 “Focal” species that USFWS has given
ly sample emergent wetlands (Bystrak 1981; management priority because they pose spe-
Robbins et al. 1986; Gibbs and Melvin 1993; cial management challenges (U.S. Fish and
Lawler and O’Connor 2004). Marsh birds in- Wildlife Service 2005). Black Rails, Yellow
clude all species that primarily inhabit marsh- Rails and Saltmarsh Sharp-tailed Sparrows
es (i.e. marsh-dependent species), and many (Ammodramus caudacutus) are three of the 20
of these species are considered “inconspicu- species on the National Audubon Society’s
ous” or “secretive.” Primary species of con- national ‘Watchlist’ because they are the
cern in North America include King Rail (Ral- ‘most imperiled’ species (National Audubon
lus elegans), Clapper Rail (Rallus longirostris), Society 2007). Many U.S. states consider these
Virginia Rail (Rallus limicola), Sora (Porzana species threatened or of special concern for
carolina), Black Rail (Laterallus jamaicensis), similar reasons. King Rails are federally en-
Yellow Rail (Coturnicops noveboracensis), Amer- dangered in Canada, Least Bitterns are feder-
319320 WATERBIRDS
ally threatened in Canada (COSEWIC 2002) bird survey data at locations throughout
and Black Rails are federally endangered in North America beginning in 1999. The pro-
Mexico (Diario Oficial de la Federacion tocol is outlined below and is also available
2002). on the internet (see website at http://
Populations of marsh birds may be affect- www.cals.arizona.edu/research/azfwru/Na-
ed by accumulation of environmental con- tionalMarshBird/).
taminants in wetland substrates because they
consume a wide variety of aquatic inverte- Objectives
brates (Odom 1975; Klaas et al. 1980; Eddle-
man et al. 1988; Gibbs et al. 1992; Conway The Standardized North American
1995). Marsh birds are also vulnerable to in- Marsh Bird Monitoring Protocol is intended
vasion of wetlands by many invasive plant to provide guidance to individuals planning
species (e.g. Lythrum salicaria, hybrid Typha, to survey marsh birds to address different ob-
Phalaris arundinacea, Phragmites, etc.) (Gibbs jectives. The most commonly-stated objec-
et al. 1992; Meanley 1992). Hence, marsh tives include: 1) document presence or dis-
birds may represent “indicator species” for tribution of marsh birds within a defined ar-
assessing wetland ecosystem quality, and ea, 2) estimate or compare density of secre-
their presence can be used as one measure tive marsh birds among management units,
of the success of wetland restoration efforts wetlands or regions, 3) estimate population
(Lewis and Casagrande 1997). Marsh birds trend for marsh birds at local or regional
also have high recreational value; many of scale, 4) evaluate effects of management ac-
these species are highly sought-after by rec- tions (often actions that target other spe-
reational birders. Finally, several rails are cies) on secretive marsh birds, and 5) docu-
hunted in many U.S. states and Canadian ment habitat types or wetland conditions
provinces yet we lack the necessary informa- that influence abundance or occupancy of
tion on population trends and status upon marsh birds. The Standardized North Amer-
which to set or adjust sustainable harvest lev- ican Marsh Bird Monitoring Protocol allows
els. For these reasons, numerous federal data sharing and comparisons among sites.
agencies have been cooperating to monitor U.S. Fish and Wildlife Service has a vested in-
marsh bird populations in North America terest in marsh bird populations and their
with the hope of gaining better knowledge habitats because marsh birds are a trust spe-
on status and distribution of these birds and cies, under the protection of the USFWS.
improving estimates of population trends. The National Wildlife Refuge System of the
Continued monitoring will also allow re- USFWS has participated in conducting stan-
source managers to evaluate whether man- dardized marsh bird surveys based on this
agement actions or any other activities ad- protocol since its inception because the ref-
versely impact wetland ecosystems. Any ac- uge system has a disproportionate amount of
tion that alters water levels, alters salinity, re- wetland within their boundaries, and the
duces mudflat/open-water areas, alters management actions employed by refuges
invertebrate communities or reduces the have the potential to dramatically affect
amount of emergent plant cover within marsh bird populations. However, the proto-
marsh habitats could potentially affect habi- col has also been used (and continues to be
tat quality for marsh birds (Conway 1995). used) by biologists in a wide variety of gov-
To help achieve these goals and to help en- ernmental and nongovernmental agencies
sure that marsh bird survey data collected and academic institutions.
throughout North America would be collect- Density, Abundance and Detection Probabili-
ed in a consistent manner, I developed a ty. Abundance is the total number of birds
standardized marsh bird survey protocol. within a defined area of interest. Density is
The Standardized North American Marsh abundance divided by area; for example, the
Bird Monitoring Protocol (or previous ver- number of birds per hectare of wetland (or
sions of it) has been used to collect marsh birds/ha of emergent vegetation within aMARSH BIRD MONITORING PROTOCOL 321
wetland). Surveys rarely count all individuals vides estimates of population trends for
present in the sampling area because detec- some species of marsh birds, but has insuffi-
tion probability is typically less than 100%. cient data to estimate trends of the more se-
Estimates of abundance or density rely upon cretive or rare species.
estimates of detection probability and either
1) a consistent positive correlation between Survey Routes
number of individuals detected during a sur-
vey and number of individuals actually The number of survey points to include
present in the area sampled (i.e. low spatial within a state, local refuge or management
and temporal variation in detection proba- area (or the size of the survey area selected)
bility), or 2) incorporating environmental is often dictated by personnel time available
covariates into the estimation process that ef- and other logistical constraints. A survey
fectively adjust for most of the variation in route is a permanent grouping of points that
detection probability. Few reliable estimates are surveyed together by the same surveyor
of detection probability during marsh bird on the same date during a morning (or
surveys are currently available (but see Con- evening) survey. Each survey point should
way et al. 1993; Legare et al. 1999; Conway belong to one (and only one) permanent
and Gibbs 2001; Bogner and Baldassarre survey route. The number of points per sur-
2002; Nadeau et al. 2008). However, these vey route can vary among routes based on
survey protocols incorporate methods for es- the number of points that one surveyor can
timating several components of detection survey in a morning (or evening) survey win-
probability (see Conway et al. 2010 for an ex- dow (see section below titled Time of day for
ample of how estimates of detection proba- surveys for a definition of ‘survey window’). A
bility derived from these methods can be surveyor may only be able to survey a small
useful). Some authors have expressed skepti- number of points (e.g. six or eight) in a
cism about the value of incorporating meth- morning or evening if points are far apart.
ods intended to estimate detection probabil- These 6-8 points would constitute a “survey
ity into surveys (Johnson 2008), but others route”. If travel between adjacent points is
have advocated for such methods (Burham relatively easy, a surveyor may be able to com-
1981; Thompson et al. 1998; Thompson plete 15 or more points in one morning or
2002; Rosenstock et al. 2002). Focal marsh evening survey session and hence have 15
bird species were those identified by a group points on that survey route. All the survey
of marsh bird biologists as species for which points that make up one survey route do not
we lack quality information on status or pop- necessarily have to be associated with the
ulation trends (Ribic et al. 1999). same patch of marsh. Including fewer points
Population Trend. Population trend is of- per survey route and surveying an additional
ten not well-defined, but a common defini- morning or evening (rather than fewer
tion is the percent annual change in popula- routes with lots of points) will typically result
tion size for a particular species at some de- in more detections (but will require addi-
fined spatial scale over some defined time tional survey days) because marsh birds are
period. Estimates of population trend allow typically most vocal in the two hours sur-
managers to determine whether local or re- rounding sunrise and the two hours sur-
gional marsh bird populations are declining rounding sunset (Conway et al. 2004).
and the rapidity at which they are declining. Once you choose the direction with
Managers can establish a priori population which you conduct a particular survey route,
trend thresholds or trigger points below be consistent (e.g. always survey the points
which immediate management action along route #1 in descending order: point 12
should be taken. Such actions can prevent is surveyed first and point 1 is surveyed last).
local extinctions by identifying population Being consistent in this respect will assure
problems before they become severe. The that each survey point is completed at ap-
North American Breeding Bird Survey pro- proximately the same time of day during322 WATERBIRDS
each replicate survey. Consistency in the 3) along a public road with emergent vege-
chronological order in which points on a tation on both sides,
route are surveyed will help to reduce the 4) along a public road with emergent vege-
sampling variation created by diurnal de- tation on only one side,
creases in vocalization probability of marsh 5) along a grassland/emergent edge,
birds as the morning (or evening) progress- 6) along a scrub-shrub/emergent edge,
es (Conway et al. 2004). Even though metrics 7) along a forest/emergent edge,
to estimate variation in detection probability 8) along an open water/emergent edge,
are incorporated into the survey protocol, 9) within a narrow water channel or tidal
any effort to minimize sampling variation is creek with emergent vegetation on both
still advantageous. sides,
10) within a contiguous patch of emergent
vegetation (also record distance from
Location of Survey Points edge), or
Fixed, permanent survey points are cho- 11) other (and provide description of point
sen and marked with inconspicuous markers placement).
in the field. Each survey point receives a
A point is considered “along a public
unique identification number. Record the
road” if the surveyor is within 25m of the
latitude and longitude of each survey point
roadside during the survey. Surveyors should
using a GPS receiver. If possible, locations of
also record the type of road (gravel, dirt,
all survey points should also be plotted on
paved, etc.). Select the choice that best de-
maps of each wetland. Maps should include
scribes the placement of the point. These da-
the direction in which the speakers are
ta are meant to provide analysts with a meth-
pointed during the survey at each point.
od by which they can evaluate whether pop-
Which direction to orient the speakers is not
ulation trend or density estimates vary de-
always obvious to someone who has not sur-
pending on where survey points are located.
veyed the route before, and may create un-
If estimates vary in this regard, analysts have
wanted variation in numbers detected if
the option of stratifying their estimates to
speaker direction is not consistent. Survey
take variation in the placement of points in-
points are located on either the upland-
to account.
emergent vegetation interface or the open
water-emergent vegetation interface. Con-
ducting surveys at points where observers Point Spacing
stand within contiguous patches of emer-
gent marsh vegetation may not be practical Point spacing in previous studies has var-
in many inland wetlands because of the dis- ied from 40m to 800m (Conway and Gibbs
turbance to emergent plants (and to calling 2001). For setting up new routes associated
rates of marsh birds) caused by walking with the Standardized North American
through the dense vegetation. However, con- Marsh Bird Monitoring Protocol, we recom-
ducting surveys from upland edges, roadside mend 400m between adjacent survey points
edges, and open water edges may create to increase the total area covered by moni-
some bias in estimation of population toring efforts. If points are too close together
trends. In order to determine the extent to (i.e.MARSH BIRD MONITORING PROTOCOL 323
would then at least have the option of using need not conduct surveys at points that no
data only from a subset of points (those that longer have any emergent vegetation. If sur-
are 400m apart) at that particular site for the veyors do not conduct a survey at one or
shared (pooled) data set if they choose to do more existing points, they must record in the
so. However, birds are much more likely to database the reason why a survey was not
have heard the calls broadcast at prior points conducted at those points:
(and hence alter their vocal behavior) when
points are spaced closer than 400m apart. In 1) lack of suitable habitat (due to temporary
areas where survey routes have already been change such as flooding, drought, mow-
established and surveyed in past years, retain ing, etc.),
the original point spacing; do not delete, ig- 2) lack of suitable habitat (due to perma-
nore or move existing survey points even if nent change),
spacing between adjacent points is very dif- 3) survey not attempted due to logistical
ferent than 400m. In marshlands that have reasons.
access throughout the marsh, points should
be in a 400m grid system (hence, one point Time of Day for Surveys
per 16 ha of marsh). If not all possible points
in the grid system can be surveyed, a random Surveys can either be conducted in the
or systematic selection of points that can be morning or evening. However, once one of
surveyed should be selected from the poten- these two time periods is chosen, that time
tial survey points. Placement of survey points period for those points along the survey
within the wetland is a sampling design issue route cannot be changed. The choice of
and observers or survey coordinators should morning or evening survey period (and the
consult a statistician. In many locations, length of each period) should correspond
emergent marsh occurs in small patches less with when marsh birds are most vocal in your
than 16 ha in size. In cases like these, include area. Vocalization probability is typically
at least one survey point at all marshes >0.5 highest in the two hours surrounding sun-
ha within the management area. Additional rise and the two hours surrounding sunset
survey points should be added at small (Conway et al. 2004). Choose the optimal
marsh patches as long as they are 400m away daily survey time(s) for your region and use
from all other survey points. them each year. Including both morning
and evening surveys into a standardized
What if Area around an Existing Point is no monitoring protocol provides added flexibil-
longer Suitable Marsh Bird Habitat? ity and more potential survey hours for field
personnel.
Original survey points are never dropped Morning surveys begin 30 minutes be-
from the survey and are always visited in sub- fore sunrise and should be completed prior
sequent years. If no suitable habitat is to the time when marsh birds cease calling
present at an existing survey point during a (this varies regionally, but is often two hours
particular year (e.g. due to drought or after sunrise in southern latitudes and three
change in water flow), then the surveyors hours after sunrise in northern latitudes
should still make an entry for that point on (Conway et al. 2004; Gibbs and Melvin 1993).
the datasheet and in the database but write The time in the morning when marsh birds
in the Comments column “no survey conduct- cease calling also varies with temperature
ed because suitable emergent vegetation is and time of year (Conway and Gibbs 2001,
not present”. Although some of the focal 2011).
species will very occasionally use wetlands Evening surveys should begin two hours
with open water that lack emergent vegeta- before sunset and must be completed by 30
tion, these species are all much more abun- minutes after sunset. When conducting
dant in wetlands with emergent vegetation evening surveys, surveyors should start their
during the breeding season and so surveyors survey route such that they finish the last324 WATERBIRDS
point at the time when darkness precludes often do not know the local timing of the
them from seeing their datasheet. The half breeding cycle of the various marsh bird spe-
hour between sunset and complete darkness cies at the outset of their survey effort (Rehm
is often the time when detection probability and Baldassarre 2007). Finally, including
is highest (C. Conway, unpubl. data). three or more surveys per year will allow for
estimation of the proportion of survey routes
Number of Surveys per Year and Seasonal occupied by each species (MacKenzie et al.
Timing of Surveys 2002). However, if for some reason it is not
possible to conduct a minimum of three sur-
Optimal seasonal timing for surveys will veys on the area, data collected can still be
vary regionally depending on breeding chro- used for some purposes.
nology of the focal marsh birds (Appendix Contact the program coordinator (see
1) in your area. Conduct at least three sur- contact information below) if the three an-
veys annually during the presumed peak of nual survey windows do not adequately cap-
the marsh bird breeding season. The peak ture the peak breeding seasons of the marsh
breeding season in each location will vary bird species in your area.
among the coexisting marsh birds in that ar-
ea. For example, American Bitterns often Surveys in Tidal Marshes
breed earlier than both Least Bitterns and
rails in some regions, and Clapper Rails and When possible, surveys in tidal marshes
King Rails breed earlier than Virginia Rails should always be conducted at a similar tidal
and Soras in some regions (also see Rehm stage for each replicate survey both within
and Baldassarre 2007). To account for varia- and across years. The tidal stage within
tion in breeding phenology (and hence vari- which to conduct local marsh bird surveys
ation in optimal survey timing) among coex- should be based on when highest numbers
isting species, at least one survey should be of marsh birds are likely to be detected in
conducted within each of the three 15-day your area; optimal tidal stage for surveys may
survey windows. The three survey windows vary among regions. Many salt marsh passe-
vary regionally and are based on the average rines are forced to renest during the peak
minimum temperatures in May in each loca- spring high tide, and detection probability is
tion (Appendix 2). The three survey win- highest during the week after a high spring
dows increase probability of conducting at tide. Clapper Rail surveys conducted since
least one survey during the seasonal peak in 1972 have been timed to coincide with a
vocalization probability for all focal marsh high tide at San Francisco Bay National Wild-
bird species in the area. In many areas, mi- life Refuge, but high tide was a period of re-
grants are still moving through when the duced vocalization probability for Clapper
breeding season is well underway for local Rails in southern California (Zembal and
breeders. Hence, some surveys will occur pri- Massey 1987) and for Black Rails in northern
or to when migration is completed for many California (Spear et al. 1999). In Mississippi,
marsh birds. detection probability and tidal height were
A common goal of marsh bird surveys is positively correlated for Clapper Rails and
to estimate trends over time in the number Seaside Sparrows (Ammodramus maritimus),
of breeding adults of each species, so it is op- but negatively correlated for Marsh Wrens
timal to complete all three annual surveys (Cistothorus palustris) (Rush et al. 2009).
prior to the initiation of juvenile vocaliza- If no local data are available on optimal
tions. At least three surveys are needed to tidal stage for conducting marsh bird sur-
confirm seasonal presence or absence of veys, surveyors should try to conduct surveys
some marsh bird species in a wetland with on days when high or low tide does not fall
90% certainty (Gibbs and Melvin 1993). within the morning (or evening) survey win-
Three replicate surveys per year are also war- dow (i.e. conduct surveys when tides are
ranted because personnel organizing surveys coming in or out). Record the following: 1)MARSH BIRD MONITORING PROTOCOL 325
time of the closest high tide (either the high quence coincides with this protocol. The
tide before or after the survey - whichever is broadcast sequence should include exactly
closer in time) for each survey point, and 2) 30 seconds of calls of each of the focal marsh
tidal amplitude (difference in water level in bird species that are expected breeders in
meters between the highest and lowest tide your area interspersed with 30 seconds of si-
on that day) on the day of the survey. These lence prior to the next focal species’ calls.
tidal features have been shown to influence The 30 seconds of calls consist of a series of
numbers of birds detected during marsh the most common calls for that species inter-
bird surveys (Nadeau et al. 2010). spersed with approximately 5 seconds of si-
lence. For example, an entire survey se-
Survey Methods quence might look like this:
These standardized survey methods for 5 minutes of silence (include a verbal state-
marsh birds originated from suggestions ment at the end of each minute to alert
during two multi-agency workshops at Patux- surveyors)
ent Wildlife Research Center designed to aid
30 seconds of calls of first focal species con-
agencies developing marsh bird monitoring
figured thus:
programs (Ribic et al. 1999; U.S. Fish and
Wildlife Service 2006), and incorporate sug- Three Least Bittern coo-coo-coo calls
gestions from Conway and Gibbs (2001, Six seconds of silence
2011) and recent methodological advances Three Least Bittern coo-coo-coo calls
in estimating detection probability and ob-
Six seconds of silence
server bias (Nichols et al. 2000; Farnsworth et
al. 2002; MacKenzie et al. 2002). Because Four series of Least Bittern kak calls
many marsh birds are secretive, seldom ob- 30 seconds of silence
served and vocalize infrequently, the Stan- 30 seconds of calls of second focal species
dardized North American Marsh Bird Moni- configured like this:
toring Protocol instructs surveyors to broad-
Two Sora whinny calls
cast calls to elicit vocalizations during surveys
(Gibbs and Melvin 1993; Conway et al. 2004; Five seconds of silence
Conway and Gibbs 2005; Conway and Three Sora per-weep calls
Nadeau 2010). However, because we want to Five seconds of silence
estimate detection probability, estimate den-
Four Sora keep calls
sity using distance estimators, analyze data
without the biases associated with call-broad- 30 seconds of silence
cast (Conway and Gibbs 2001, 2011) and sur- 30 seconds of calls of third focal species, etc.
vey (for some participants) non-focal spe-
cies, surveyors will also record birds during a Include a verbal “stop” at end of the final
5-minute passive period prior to broadcast- 30 seconds of silence so that surveyors know
ing marsh bird calls. Hence, surveyors will when to stop the broadcast (and stop the sur-
record all focal species (Appendix 1) detect- vey at that point).
ed during both a 5-minute passive period pri-
or to broadcasting recorded calls and during Broadcast Equipment and Placement
a period in which pre-recorded vocalizations
of focal marsh birds are broadcast into the The broadcast player should be placed
marsh. upright on the ground (or on the bow of the
The recorded calls should be obtained boat), and sound pressure should be 80-90
from the Marsh Bird Survey Coordinator dB at 1 m in front of the speaker. Use a
(contact info below); request digital record- sound-level meter to adjust volume of the
ings of the focal species that breed in your ar- broadcast player at the beginning of each
ea, and we will ensure that the broadcast se- day. If sound quality distorts when volume on326 WATERBIRDS
your broadcast equipment reaches 80-90 dB, clude in the broadcast sequence. For general
you should obtain higher-quality broadcast inventory to document status and distribu-
equipment. If the ground is wet, place the tion, one would include all possible focal
speaker on an object as close to the ground species in their area.
as possible. Surveyors should stand 2 m to
one side of the speaker while listening for vo- Species to Include in the Call-broadcast Se-
cal responses (standing too close to the quence
speaker can reduce the surveyors’ ability to
hear calling birds). Surveyors should point In general, surveyors should include in
the speaker toward the center of the marsh their survey all of the following species that
and should not rotate the speaker during the are thought to breed in the marshes in their
call-broadcast survey. Speakers should be area: King Rail, Clapper Rail, Virginia Rail,
pointed in the same direction for all repli- Sora, Black Rail, Yellow Rail, American Bit-
cate surveys. At points where the direction in tern, Least Bittern, Pied-billed Grebe,
which to point the speakers is not obvious Limpkin, American Coot, Purple Gallinule
(i.e. on a road or in a canal bisecting two and Common Moorhen. The number of
marshes), surveyors should record the direc- species included on the call-broadcast por-
tion of the speakers at each point on a map tion of the survey increases the duration of
and on their datasheets and refer to this in- the survey by 1 min per species at each
formation on all replicate surveys. Visit the point. So, with eight species, you will spend
program website to see a list of suitable 13 minutes (including the initial 5 min pas-
equipment for broadcasting calls: http:// sive listening period) at each point. If a sur-
www.cals.arizona.edu/research/azfwru/Na- veyor is within the breeding range of the
tional MarshBird/). American Coot, Common Moorhen or
Pied-billed Grebe, broadcasting calls of any
Species to Include in the Survey Effort of these species is considered optional but
strongly recommended (Appendix 3).
Surveyors must make three decisions re- However, all surveyors should still record all
garding the species to include in their survey detections of these species (see Appendix
effort: 1) which species will be recorded on 3), even if they do not include one (or all
their datasheet, 2) of those species recorded, three) of these “focal” species in their
which species will be recorded during the broadcast sequence. Participants conduct-
one-minute segments (i.e. each individual ing surveys on National Wildlife Refuges
bird of these species will be recorded on a should see the guidance on the program
separate row on the datasheet), and 3) of website on which focal species to include in
those species recorded, which species’ calls their broadcast sequence for their refuge.
will be included in the call-broadcast se- Use that list only as guidance and contact
quence (Appendix 3). Staff from the U.S. the Program Coordinator (see end of docu-
Fish and Wildlife Service’s National Wildlife ment) if you can recommend corrections to
Refuge System have provided guidance on the list of suggested species for a particular
making these decisions for the >500 refuges refuge. The guidance on the website is
in the U.S. The program website provides based on maps of the breeding ranges of the
this guidance along with a map overlaying focal species, but these maps are not always
the breeding range of each focal species by accurate in identifying the species that
USFWS Region (http://www.cals.arizo- breed locally at a particular refuge. If some-
na.edu/research/azfwru/NationalMarsh- one has good local knowledge of the species
Bird/, then click on Focal Species and then of secretive marsh birds that breed on a par-
breeding distribution). Surveyors should exam- ticular refuge, they should contact the Pro-
ine this map to help determine which focal gram Coordinator to have the species list
species likely breed in their area and use this updated or verified (even if no changes are
information to determine the species to in- necessary). As that occurs, we will updateMARSH BIRD MONITORING PROTOCOL 327
the website to indicate which refuges have Estimating Distance to Each Focal Bird
had their list of species verified. Keeping
the list of recommended species on the Surveyors should estimate the distance
website up-to-date is important, so please from the survey point to each individual
verify the list of focal species for your sam- bird. Recording distance to each individual
pling location by asking people who are fa- bird will allow analysts to use distance sam-
miliar with the marsh birds in your area and pling techniques to estimate density for each
send any suggested modifications to the species in each habitat type and for each sur-
Program Coordinator. veyor. Surveyors need to estimate the dis-
The broadcast sequence includes calls tance to each bird when the bird was first de-
of the focal marsh bird species that are ex- tected during the survey. Several authors
pected breeders in that area. The calls are have suggested that secretive marsh birds of-
broadcast on a portable CD or MP3 player ten move toward the broadcast source prior
with amplified speakers attached. The to vocalizing (Legare et al. 1999; Erwin et al.
marsh birds included in the call-broadcast 2002), and systematic movement toward the
sequence will vary among survey areas, but surveyor violates an important assumption of
will always be consistent within a particular distance sampling. More research is needed
survey route across years. Recommended to address the magnitude of this potential
species to include in the call-broadcast se- problem for each focal species, but analysts
quence for a particular survey route (i.e. at will likely use distance estimates only from
a particular refuge or management area) birds detected during the initial passive por-
are listed in Appendix 3. The goal is to in- tion of the survey (i.e. those that were detect-
clude all of the focal species believed to be ed prior to being exposed to call-broadcast).
potential local breeders (species for which Estimating density from only a subset of
you might reasonably expect to get respons- birds detected (those initially detected dur-
es during the breeding season). Order of ing the 5-min passive period in this case)
calls start with the least intrusive species does not introduce bias in distance sampling
first and follow this order: Black Rail, Least as long as the other assumptions of distance
Bittern, Yellow Rail, Sora, Virginia Rail, sampling are met (Buckland et al. 2001).
King Rail, Clapper Rail, American Bittern, Density indices by habitat type are useful be-
Common Moorhen, Purple Gallinule, cause they allow managers to extrapolate
American Coot, Pied-billed Grebe, survey data to estimate a minimum number
Limpkin. The order of species on the of each marsh bird species on their entire
broadcast sequence was based on recom- management area. The distance at which
mendations by Ribic et al. (1999). The calls most individuals are detected varies among
used for broadcast include the primary ad- the focal species (Conway and Nadeau
vertising call of each species (e.g. ‘whinny’ 2006). Surveyors are encouraged to use a
for Sora, ‘grunt’ for Virginia Rail, ‘clatter’ for range finder to help them determine the dis-
Clapper Rail, ‘click-click-click-click-click’ for tance to specific landmarks surrounding
Yellow Rail, ‘coo-coo-coo’ for Least Bittern, each survey point, which will help estimate
‘pump-er-lunk’ for American Bittern). Other the distance to calling marsh birds. Other
calls associated with reproduction are also methods for improving one’s ability to esti-
included for many of the species. Including mate distance include: 1) tying surveyors’
all the common calls associated with repro- flagging at 50m and 100m away from each
duction of each species on the broadcast se- survey point in each cardinal direction, and
quence will likely increase detection proba- 2) carrying aerial photos of the marsh with
bility during different times of the breeding 50m-, 100m- and 200m-radius circles drawn
season and can help surveyors learn the less around each survey point. Estimating the
common calls of each of the focal species. A distance to some individual birds will involve
list of common calls for each focal species is a lot of uncertainty (i.e. estimating distance
provided in Appendix 4. to birds 5m from the surveyor is much easier328 WATERBIRDS
than estimating distance to birds that are or to the beginning of the survey, write down
>100m away). Indeed, distance sampling of- the day, month and year at the top of the da-
ten led to overestimates of density of song- ta sheet. Write out the month or use a three-
birds in eastern deciduous forest (Alldredge letter abbreviation to avoid confusion be-
et al. 2007b, 2008). However, some surveyors tween day and month (i.e. so that 6 May is
(those that had received prior training in dis- not confused with 5 June). Also write the full
tance estimation to calling marsh birds) were name of all persons present during the sur-
able to estimate distance to calling secretive vey. If more than one person was present,
marsh birds that they could not see with min- write down who recorded the data and all
imal bias (average difference between esti- persons that helped identify calling birds.
mated and actual distance varied among the Because detection probability can differ sub-
focal species from 0 to 24 m in Arizona and stantially among surveyors (Kendall et al.
California based on 29-115 trials per species; 1996; Link and Sauer 1998; Conway et al.
Nadeau and Conway, unpublished data). 2004; Sauer et al. 2004), analysts may wish to
More studies are needed to estimate the bias control for observer bias when estimating
and accuracy of surveyors’ distance estimates trend (similar to the approach used by ana-
to calling marsh birds and the factors that af- lysts of BBS data; Sauer et al. 2008), so partic-
fect bias and accuracy. Surveyors should en- ipants should record any and all surveyors
ter on the datasheet and in the database who contributed to marsh bird detections
which of the following distance estimation (see the multiple-observer surveys section).
aids they used: Unaided, Distance Markers, Write down the name of the survey route and
Range Finder, Range Finder and Maps, Maps the name of the survey area or management
or Aerial Photos or Distance Not Recorded. unit.
The Data Sheet Recording Detections of Focal, Broadcast
Species
An electronic copy of a datasheet
should be obtained from the Survey Coor- When you arrive at the first survey point,
dinator or the program website (http:// write down the unique identification num-
www.cals.arizona.edu/research/azfwru/Na- ber of the survey point and the time. Start
tionalMarshBird/) to ensure that all perti- the survey. When an individual of a focal spe-
nent data are recorded properly. These cies is detected, write the species name in the
datasheets can then be tailored by each sur- “Species” column. You can use the four-letter
veyor to meet local needs as long as none of acronym for the species or write the full spe-
the standards in this protocol are compro- cies name. A list of standard four-letter spe-
mised. The number of species columns on cies acronyms is provided in Appendix 1. Put
the datasheet will differ across survey areas. a “1” in each detection column in which that
For example, if calls of only three species individual is detected aurally and put an “s”
will be broadcast, then the survey duration in each column in which the individual is de-
will be eight minutes at each point (five tected visually (including flying overhead).
minutes of passive listening and one minute For example, if an individual Virginia Rail
of call-broadcast for each of three species) vocalizes during the first one minute of pas-
and will need a datasheet with eight re- sive listening, put a “1” in the first column.
sponse columns. If calls of five species will Regardless of whether that individual calls
be broadcast, then the survey duration will once or many times during the first minute,
be ten minutes at each point (five minutes only put one “1” in the first column. If that
of passive listening and one minute of call- same individual bird is still calling during the
broadcast for each of five species) and will second minute of passive listening, then also
need a datasheet with ten response col- put a “1” in the second column. If the same
umns. A hypothetical example of a complet- individual calls during the 30 seconds when
ed datasheet is provided in Appendix 5. Pri- Sora calls are being broadcast or the 30 sec-MARSH BIRD MONITORING PROTOCOL 329 onds of silence immediately following the are for your own use (not entered into data- Sora sequence, put a “1” in the column for base) and will aid your future identification “SORA”. If that same individual bird calls of unknown calls if that call is heard repeat- again during the Virginia Rail sequence, also edly. If time permits, the surveyor can return put a “1” in the “VIRA” column, and so on. to the point with another expert birder who Hence, if an individual bird is calling con- may be able to help identify that “unknown” stantly throughout the survey period, you bird or with sound recording equipment so will have a “1” in every column for that indi- that they can send the recording of the call vidual. If the individual is heard and seen, to the Program Coordinator for identifica- put both a “1” and an “s” in the appropriate tion. Some species of marsh birds give paired column(s). If you hear a call of the same spe- duets and surveyors can often distinguish cies but from a different individual (or from pairs of birds during surveys. Always record an individual of another species), start a new each member of a pair on its own individual row on the data sheet and follow the same row of the datasheet. Record “pair” in the protocol just described. Recording whether Comments columns for each of the two birds each individual bird is or is not detected dur- that are thought to be members of a mated ing each 1-min segment allows analysts to use pair. removal models or time-of-detection meth- ods (Farnsworth et al. 2002; Alldredge et al. Recording Detections of Focal, Non-broad- 2007a) to estimate detection probability (see cast Species Conway et al. 2010 for an example). Survey- ors may have difficulty determining whether Whenever possible, these species (see a call is coming from a new individual or an Appendix 3) are recorded the same way as individual detected earlier at that survey ‘focal, broadcast species’ above, but their point. Surveyors must often decide whether calls are not broadcast during the call-broad- a call is that of a new individual or one al- cast portion of the survey. If surveyors are ready detected without seeing the bird by us- overwhelmed by the number of focal birds ing their best judgment (this is a challenge detected, then they record these species dif- on all bird surveys regardless of the protocol ferently than the focal, broadcast species used). In general, be conservative and as- (see the What to do if the surveyor becomes over- sume that a call is from the same bird if the whelmed with too many detections section). call came from the same general location (i.e. a similar direction and not too far from Recording Detections of Non-focal Species the location of the original call). The num- (OPTIONAL) ber of rows filled out on the datasheet will differ among survey points and will corre- We recommend that surveyors do not spond to the total number of individual focal record non-focal species during surveys (also marsh birds detected at each point. If no see Johnson et al. 2009). However, some sur- marsh birds are detected at a survey point, veyors will want to record all species detected record the point number and starting time, (including passerines, waterfowl, raptors, and write “no birds” in the Species column. A etc.) or perhaps a subset of all species detect- sample datasheet is included as an example ed (i.e. include marsh-dwelling passerines, of what survey data might look like (Appen- wading birds but not other species) during dix 5). If the surveyor hears a marsh bird but their marsh bird surveys. Surveyors need to is unsure of its identity, the surveyor should be certain that they focus their attention on write “unknown” in the Species column and the focal marsh birds (especially in areas record all data for this individual as de- where densities of secretive marsh birds are scribed above. Make a verbal description of relatively high). If a surveyor feels strongly the unknown call in the Comments column that they must record species in addition to (e.g. ‘soft “kak-kak-grr” - sounds like BLRA the focal species listed in Appendix 3, the but harsher’). Descriptions of unknown calls shared database can accommodate these data
330 WATERBIRDS
but data on non-focal species should be re- Different call types have different functions
corded differently. At each point, record the (see Appendix 4) and can indicate pairing
total number of each non-focal species de- status and stages of the nesting cycle in a local
tected within each of three distance catego- area (allowing refinement of local survey win-
ries (≤50m, 51-100m, and >100m). Individu- dows). Moreover, detection probability and
al birds of non-focal species do not receive observer bias differ with different call types
their own line on the data sheet and survey- (e.g. Least Bittern ‘kak’ and the first part of a
ors do not need to record detections of non- Virginia Rail ‘tick-it’ can be confused with
focal birds using the 1-min segments (Con- Clapper Rail ‘kek’ calls) and accuracy of dis-
way and Droege 2006). tance estimation may vary with call type (Con-
The non-focal species included by a sur- way and Nadeau 2006; Conway, unpubl. da-
veyor will depend on the marsh birds of in- ta). Hence, estimates of population trends
terest at that refuge, management area or based on data from only those call types that
physiographic region. For example, survey- have low observer bias might increase power
ors may want to include non-focal species to detect true population trends due to reduc-
which are thought to be declining or which tions in temporal variation in counts. Further
are not sampled well by other survey efforts. research is needed to determine the extent
However, analysts will need to know which that accuracy of trend estimates can be im-
additional species were being recorded so proved (if at all) by including only those birds
that these data are meaningful (i.e. if no YH- that gave certain types of calls, but there are
BLs are recorded at a point, analysts need to additional benefits of having observers record
know whether a surveyor detected zero YH- call types on surveys (Conway and Gibbs 2001,
BLs or merely did not record YHBLs on their 2011). For these reasons, surveyors should
survey). Therefore, each surveyor must en- record all types of calls given for each focal
ter in the database their list of “non-focal” marsh bird detected in the Calls column on
species that they were recording during their the datasheet (see sample datasheet; Appen-
survey. Please take into consideration that dix 5). Refer to the program website to listen
the number of “non-focal” species included to examples of each common call type: http:/
in your survey effort may reduce your ability /www.cals.arizona.edu/research/azfwru/Na-
to record all the relevant data for the 26 focal tionalMarshBird/.
species (Appendix 1) that are the focus of
the Standardized North American Marsh Birds Detected at a Prior Survey Point or be-
Bird Monitoring Protocol. Moreover, many tween Points
of the non-focal species may be adequately
sampled already by the North American If surveyors detect a new bird immediate-
Breeding Bird Survey. Indeed, Johnson et al. ly after the survey period at a particular point
(2009) cautioned against surveyors record- (or while walking between points) they can
ing non-focal species for fear that the survey- record these birds on a separate row and
or would miss focal species while paying at- record “yes” in the Outside Survey column.
tention to non-focal species. Recording birds detected outside of the ac-
tual survey period may be useful information
Record Types of Calls Given for some of the focal species that are partic-
ularly rare at a local site (species for which
Knowing seasonal patterns of different detections during the actual surveys are
call types in a local area provides useful infor- rare). If a surveyor detects a focal bird dur-
mation. For example, the frequency of differ- ing a survey and the surveyor believes that
ent calls given (e.g. single clatter, paired clatter, the call was given by the same individual bird
kek or kek-burr for a Clapper Rail) varies which was detected and recorded at a previ-
throughout the season (Conway et al. 2004; ous survey point, the surveyor should record
Conway, unpubl. data). Frequency of differ- all the relevant data for that bird and then
ent calls given may also vary across regions. enter a “Yes” in the Detected at a Previous PointMARSH BIRD MONITORING PROTOCOL 331
column on the datasheet. When in doubt, be have received different management ac-
conservative as to whether an individual bird tions. Two adjacent areas with different man-
detected at the current point was the same agement histories presents a problem if
individual recorded at a previous point (i.e. some birds detected at a survey point are
record “Yes” when in doubt). within one area but others are within anoth-
Recording birds that were detected out- er area (with a different management histo-
side of the standardized survey times (i.e. ry). For clarity, some participants who use
outside of the 10-min survey at a point) can this protocol will have certain survey points
be useful because these birds are secretive that were located specifically to count birds
and rarely vocalize. For inventory purposes, within a certain “target area”. For example,
surveyors may not want to ignore these de- four points along a survey route were located
tections, especially if, for example, they rep- with the intent to count marsh birds within
resent the only Black Rail detected all day or “impoundment A” and five points along that
all year. However, a problem arises if one of same survey route were located with the in-
these birds detected outside of the standard- tent to count marsh birds within “impound-
ized survey period is then detected at a sub- ment B”. However, surveyors at these nine
sequent point during the standardized sur- points detect birds both within these im-
vey period. For example, if: 1) the surveyor poundments and also in adjacent areas out-
detects a Black Rail after the 10-min survey side these impoundments. Participants often
period at point #3 and records that bird on would like to count all birds detected at each
its own row on the datasheet (and writes point, but also delineate which ones were
“No” in the Detected at a Previous Point column within these impoundments. Hence, survey-
and “yes” in the Outside Survey column), and ors should record, to the best of their ability,
2) the surveyor then detects the same Black whether each bird detected was or was not in
Rail during the 10-min survey period at their “target area”. In the example above,
point #4. Recording “Yes” in the Detected at a the “target area” is impoundment A for
Previous Point column for the entry at point points 1-4 and the “target area” is impound-
#4 creates a problem because: For many ment B for points 5-9. If some (or all) of your
analyses (including estimates of population points are associated with a “target area”, the
trend) analysts may want to ignore all entries name of that target area should be identified
that have a “Yes” in the Detected at a Previous in the database for each point. Some partici-
Point column and all entries that have a “Yes” pants may not have any “target area” associ-
in the Outside Survey column so that all indi- ated with any of their survey points, and can
vidual birds are counted only once. In the therefore leave the “Target Area” column
scenario above, the Black Rail would have blank on the datasheet (and in the data-
been ignored altogether from trend analy- base). For example, you may not have any ex-
ses. Hence, for the situation described plicit “target areas” associated with any of
above, the surveyor should write “No” in the your survey points if you are conducting sur-
Detected at a Previous Point column for the en- veys primarily to estimate population trend,
try at point #4, and then go back and change determine status and distribution, or to
the “No” to “Yes” in the Detected at a Previous identify habitat relationships. If the marshes
Point column for the initial entry for this bird you survey undergo different management
at point #3 (when the bird was detected after actions, then identifying a distinct “target ar-
the 10-min survey period). ea” associated with each survey point may be
useful, whereby surveyors can record wheth-
Recording Whether Focal Birds are Within er each bird they detect at each point was or
the “Target Area” was not within that “target area”. The “target
area” may be different at each survey point
A common goal of marsh bird surveys is along a survey route. If a participant has dif-
to document the effects of management ac- ferent management units or specific marsh-
tions on marsh birds, but adjacent areas may es that some of their survey points are in-332 WATERBIRDS
tended to monitor (or if the participant sees 1) Include a circle on each row of the
value in differentiating among units or datasheet and make a ‘tick’ on each circle
marshes), they should add a column to their identifying the general direction of that
datasheet titled “In target area” and record a individual (the ticks on a circle, along
“Y” or an “N” for each focal bird detected with the distance estimate, will help you
(see Appendix 5). differentiate one individual from other
individuals of that species as more are de-
If the Surveyor Becomes Overwhelmed with tected at that point - see the column titled
too Many Detections “Direction” in Appendix 5).
2) If the problem is common on a particu-
Because many of the focal species occur lar survey route, reduce the number of
at relatively low densities through much of species in your call-broadcast sequence.
their range, many surveyors will detect few For example, eliminate American Coots,
or no individual birds at any given survey Common Moorhens and Pied-billed
point. However, some survey points within a Grebes from your call-broadcast se-
survey area will have so many marsh birds quence so that you have fewer species on
calling that surveyors will have difficulty re- your call-broadcast sequence. In other
cording data during each 1-min segment in words, still record data for all individuals
which each individual focal bird is detected. of all focal marsh bird species in the
For example, a surveyor may see or hear >20 same way, but just reduce the number of
American Coots at one survey point. When columns on the datasheet (and length of
many birds are calling simultaneously, sur- the call-broadcast sequence).
veyors can have difficulty 1) deciding wheth-
er they are hearing new individuals or previ- 3) For those focal species that are of lower
ously-detected ones, 2) writing new individ- management or conservation interest in
uals on a new line of the datasheet, and 3) your survey area (e.g. American Coots,
finding the correct line on which they had Common Moorhens, Pied-billed
noted previously-detected birds. In these sit- Grebes), simply write down an estimate
uations, here are a few comments, observa- of the total number of individuals de-
tions and suggested remedies. First, individ- tected within each of three distance cat-
ual surveyors do improve, with practice, at egories (100m) for
recording the required information and that particular species at that point (e.g.
making the necessary decisions even with write “AMCO: 0; 12; 23” on one line of
relatively high numbers of calling birds at a the data sheet - see example on sample
point. However, everyone has a threshold at data sheet attached; Appendix 5). Use
which the numbers of calling marsh birds the 1-min segments only for the focal
get too high at a particular point. Exceeding species of higher management concern
one’s ‘tolerance threshold’ occurs more fre- (e.g. Black Rails, Yellow Rails, King Rails,
quently when a surveyor has many species in Clapper Rails, bitterns). Surveyors must
their call-broadcast sequence (and hence record on the datasheet (and in the da-
many detection columns on their tabase) the points at which they were
datasheet). Below is a list of solutions for overwhelmed and could not record data
when a surveyor becomes overwhelmed, in for individual birds on separate rows of
decreasing order of preference (try those the datasheet (for focal species).
nearer to the top of the list before resorting 4) At points at which the surveyor becomes
to those nearer to the bottom). A surveyor overwhelmed, the surveyor should en-
often does not know until after the survey sure that all individuals detected are re-
has started at a particular point that (s)he is corded on the datasheet even if that
becoming overwhelmed and is not effective- means failing to record distance esti-
ly assigning the correct calls to the correct mates to each focal bird and failing to in-
columns (individuals). clude each bird on a separate line (i.e. atYou can also read
