Occupancy Modeling of Bird Species in a Subalpine Lake Ecosystem main content.

Occupancy Modeling of Bird Species in a Subalpine Lake Ecosystem

Part of the Young Naturalist Awards Curriculum Collection.

by Ian, Grade 12, New Jersey - 2014 YNA Winner

Introduction and Background

Ryker Lake, Sparta, New Jersey from Survey Site # 1Click to enlarge

Northern New Jersey is home to a number of subalpine lakes that create habitat for both migratory and residential bird species. Ryker Lake is a small, shallow subalpine lake in Sussex County, New Jersey. It is part of the Sparta Mountain Wildlife Management Area ecosystem. The area is used for recreational purposes, including fishing, hiking, and limited hunting (“Sparta Mountain,” “Important Birding Areas”). The area is home to a number of endangered or threatened state species, including the osprey, the red-shouldered hawk, the great blue heron, and especially golden-winged warblers (DeFalco & Dey).

mute swan
Mute SwanClick to enlarge

Previously, as part of New Jersey Audubon’s Citizen Science program, I conducted bird count surveys during 2010–2012 in the Ryker Lake area of the Sparta Mountain Wildlife Management Area. During this research, I noticed the potential impact of human development on both the spatial and temporal diversity of the bird species. Specifically, I noticed that certain survey points closer to developed areas seemed to have fewer birds and fewer different types of birds than other, more secluded sites. I determined that the density and diversity of birds varied seasonally, being highest during the spring and fall and with the fewest species seen during the winter, and that the density and diversity of birds also varied by survey site, with the high noise and frequent human traffic at Site 3 resulting in reduced species diversity and reduced bird numbers.

Although collecting this data was extremely valuable and rewarding, I was concerned that I was not able to identify all the birds all the time, or that I might have missed some important birds since I wasn’t at each site all the time. Working with mentors, my research indicated that there are probability-based methods, called occupancy modeling, that can help address these concerns. As I extended the research into the 2013 season, I sought to answer the following question.

Question 1: Can occupancy modeling better describe the population dynamics of the bird species in the Ryker Lake area?

I was particularly interested in small insectivorous bird species because of their variety, migratory patterns, and endangered status. The golden-winged warbler is a small, insectivorous passerine that breeds in the northeastern United States and winters in South America, especially in Ecuador. It feeds on medium-sized, usually flightless insects, usually by standard gleaning. The warbler nests exclusively in early successional habitat near more mature forest, historically after fires but more recently after clear-cuts. Breeding habitat is currently the primary focus of conservation efforts, given the difficulty of maintaining such edge habitat without fires. As a relatively specialized species with precise ecological needs, the golden-winged warbler needs protection to protect it from human impact on its habitat. During the course of my research, an area nearby Ryker Lake was part of a project near power line paths between PSE&G and the New Jersey Audubon Society to secure habitat for a small population of golden-winged warbler nesting pairs (Pakonis).

Protocol Site Label Garrison Survey Site Name
SM08 Site #2
SM09 Site #3
SM06 Site #4

Table 1: Survey Site Tanslation

In 2013, additional efforts were undertaken by New Jersey Audubon to provide more habitats in the Ryker Lake area for golden-winged warblers (“Important Bird and Birding Areas”). In particular, one such effort was the girdling of several hundred trees in an attempt to open up the woodland canopy and encourage habitat more suitable for the warblers. Unfortunately, no information about the girdling operation was posted, causing some confusion among hikers who frequent the area. Also, I was concerned that the creation of habitat for the golden-winged warblers might impact other resident wildlife species.

Close-up of a tree trunk with a ring of bark cut from around it. Background shows a forest floor covered with dry brown leaves.
One of the hundreds if trees girdled in an attempt to open up the woodland canopy in hopes of attracting more warblers.Click to enlarge

The expanded project now covers approximately three years, from January 2011 to December 2013. It looks at the impact of the conservation efforts that New Jersey Audubon has taken over that time period to protect the threatened golden-winged warbler. These efforts led me to try to answer an additional question.

Question 2: What is the baseline occupancy for certain bird species that might compete with golden-winged warblers in the newly created habitat within the Ryker Lake Wildlife Management Area?

Materials and Methods

Data was collected using a modified New Jersey Audubon Sanctuary Inventory Level 2 Citizen Science bird survey protocol, as described below.  I identified the birds, and I or my assistant (my father) recorded the data on protocol data sheets. We selected four sites from the nine described in the original protocol.

The four survey sites had the following characteristics: one at the border between the forest, the lake, and an open grassy area at the beginning of the lake area’s trail system; one in the middle of the forest, in an area frequently traveled by humans; one in the deep forest and out of the usual traffic area, but within two hundred yards of the main road; and one in the deep forest, out of the way of the usual foot traffic, with an easy view of the lake, and well away from the road. This selection of survey sites allowed us to study the effects of noise and regular human foot traffic on bird numbers, diversity, and activity. Furthermore, girdling of trees was noted at Sites 2 and 3, and on the transect from Site 1 to 2 and from 2 to 4.

While some recreational data was collected at other sites, we have not included any data from those sites in this study. Using the selected sites allowed us to focus directly on the lake ecosystem while also allowing us to compare sites that were more or less affected by human development and interference.

Our protocol was as follows: we would approach the site, wait two minutes (to allow any disturbance from our approach to be resolved) and then would begin the ten-minute count period. We elected not to conduct the optional transect portion of the original protocol, preferring instead to focus upon the site-specific observations. Sightings were classified based on site, species of bird, number of birds, distance from observers, and identification method, as well as whether or not the bird[s] observed were in flight when they were observed. Atmospheric conditions and count start time were recorded at the beginning of each day. Data was transcribed from field data collection sheets to Microsoft Excel for analysis.

Yellow Warbler
Golden-winged warblerClick to enlarge

Identification of birds was conducted using either unaided vision or vision assisted by a set of Nikon Monarch ATB 8 x 36 binoculars. I used David Sibley’s or Ken Kaufmann’s guide for field markings (Sibley; Kaufmann). Images of the lake and wildlife were taken with a Kodak EasyShare Z1012IS digital camera.

For the expanded project, we performed occupancy estimation and modeling on the data. Occupancy modeling is a statistical technique used to analyze the population dynamics of species over area and time. Generally, occupancy is defined as the probability that a randomly selected site in an area of interest is occupied by a species (Mackenzie, et al.; Mattison & Marshall).

ψ =x/sWhere psi is the occupancy, x is the number of occupied sites, and s is the number of total sites.  

Point count surveys used to determine x have a number of limitations, including observer error, that I was worried about. To learn how to conduct occupancy modeling, I utilized an online course offered through the University of Vermont (Donovan & Hines). To conduct the analysis, I used Presence Version 6.2, a free occupancy-modeling software available from the United States Geological Survey website. As a starting point for analysis, I used single-season analytical models to estimate the presence or absence of species at the sites. Models based on probabilities were compared to naïve estimates for goodness of fit and simplicity using the Aikaike Information Criterion (AIC). AIC is a mathematical method used to compare a set of models for appropriateness of fit. Usually, the AIC of the models are compared, and the model with the minimum achieved AIC is considered the most likely. The absolute magnitude of AIC is not very relevant, but the differences in AIC between different models are important. Presence-generated AIC values were used to calculate and analyze model fit.

This analysis should provide good information on local bird populations and their dynamics, which will be helpful in guiding future conservation efforts. My focus in this paper is to compare the occupancy data for the three largest insectivorous passerine groups—flycatchers (Tyrannidae), warblers (Parulidae), and swallows (Hirundidae). These three groups cover the main insectivorous feeding strategies: typical gleaning, hawking, and aerial feeding. I hypothesize that aerial feeding, practiced by swallows over open water, is an advantageous strategy in the increasingly eutrophic lake environment and would not be affected by the creation of successional habitat for golden-winged warblers. Baseline data for the gleaners and hawkers would be important, as these species might be affected by the new habitat creation.

Occupancy models were made and analyzed for each year of data for each of three sites (Site 1, an edge site at the shore of the lake; Site 2, a quiet area in the middle of the forest; and Site 3, a forest site near a fairly busy road and a loud stream) for each group of birds. Furthermore, a detailed occupancy analysis was performed for the yellow-rumped warbler, the predominant warbler species at Ryker Lake, at each of the three sites to provide a baseline of date for when the habitat is cleared out and opened up for golden-winged warblers.

Results and Discussion

Red-winged Blackbird in budding tree
Red-winged blackbirdClick to enlarge

The study period spanned from (late) December 2010 to mid-December 2013. A summary of the survey results is shown in Table 2 and Table 3. We conducted a total of 65 observation days over the three-year study period. We observed 2,110 birds total across the entire observation period. The average number of birds observed per observation day was 32, with a minimum number seen of 5 and a maximum number seen of 202. The biodiversity across all sites included 89 total unique species. A list of all the species observed is provided in Appendix A.

# of observation days 65
Total # of birds observed 2,110
Average 32
Minimum # birds seen in one day 5
Maximum seen in one day 202
Number of different species observed 89

Table 2. Summary of Observations.


Type Number Observed Number of observations
Grand Total 2,110 980
Canada Goose 408 53
Greater Scaup 171 19
Blue Jay 148 98
American Crow 108 78
Northern Mockingbird 90 69
Cackling Goose 81 6
Tufted Titmouse 79 43
Yellow-rumped Warbler 78 35
Unidentified Passerine 73 16
Mallard 68 24
American Robin 59 33
Black-capped Chickadee 55 39
Turkey Vulture 46 34
Gray Catbird 43 34
Tree Swallow 43 22
Song Sparrow 41 25
Wood Duck 41 18
Red-winged Blackbird 37 29
Barn Swallow 33 15
Eastern Phoebe 24 11
Eastern Kingbird 21 16
Gray Catbird 19 15
Carolina Chickadee 18 12
Rusty Blackbird 18 5
American Gold finch 16 5
White-breasted Nuthatch 16 12
Great Blue Heron 15 13
Dark-eyed Junco 14 9
Northern Flicker 13 13
American Redstart 12 4
Fish Crow 12 7
Wild Turkey 12 11
Barrow's Goldeneye 10 1
Downy Woodpecker 10 9
Ring-billed Gull 10 5
Red-bellied Woodpecker 8 8
Red-eyed Vireo 8 5
Baltimore Oriole 7 6
Black-throated Green Warbler 7 6
Common Grackle 7 5
Common Yellowthroat 7 7
Mute Swan 7 4
Louisiana Water Thrush 6 5
Yellow Warbler 6 6
Eastern Wood Pewee 5 5
Ovenbird 5 5
Black Vulture 4 2
Brown Creeper 4 2
Mourning Dove 4 3
Palm Warbler 4 4
Prothonotary Warbler 4 4
Red-tailed Hawk 4 4
Veery 4 4
American Black Duck  3 2
Chipping Sparrow 3 3
Sparrows 3 1
Unidentified Duck 3 2
Yellow-bellied Sapsucker 3 3
American Coot 2 1
Belted Kingfisher 2 2
Black-throated Blue Warbler 2 1
Blue-winged Teal 2 1
Bufflehead 2 2
Cardinal 2 2
Common Raven 2 2
Osprey 2 2
Pileated Woodpecker 2 2
Ring-necked Duck 2 1
Ruby-crowned Kinglet 2 2
Savannah Sparrow 2 1
Unidentified Warbler 2 2
Unidentified Woodpecker 2 1
White-throated Sparrow 2 2
Blackburnian Warbler 1 1
Cedar Waxwing 1 1
Chestnut-sided Warbler 1 1
Cliff Swallow 1 1
Coopers Hawk 1 1
Double-crested Cormorant 1 1
Eastern Towhee 1 1
Green Heron 1 1
Hairy Woodpecker 1 1
House Finch 1 1
Magnolia Warbler 1 1
Northern Water Thrush 1 1
Ovenbird 1 1
Red-shouldered Hawk 1 1
Ruby-throated Hummingbird 1 1
Scarlet Tanager 1 1
Tennessee Warbler 1 1
Unidentified Crow 1 1
Unidentified Hawk 1 1
Unidentified Vireo 1 1
Willow Flycatcher 1 1
Yellow-throated Warbler 1 1

Table 3. Bird Observation and Diversity.

redheaded woodpecker
Redheaded WoodpeckerClick to enlarge

Figure 2 displays the total number of birds observed during a given observation day across all sites as a function of observation date. We observed some interesting “spikes” in the numbers of various bird species, including in the spring of 2011 and the fall of 2012. The spring of 2012 and the fall of 2011 also “spiked” above the mean, but were less than one standard deviation out. This was due to a flock of greater scaup that resided on the lake as soon as it thawed, and a number of yellow-rumped warblers (many likely among the unidentified warbler sightings, which are counted here despite their uncertain identification) that showed up early in the season to get a head start on replenishing their fat stores.

Figure 2

Figure 2

Canadian geese
Canada geese on frozen lakeClick to enlarge

Periods of below-average bird numbers corresponded with the hottest stages of summer and the coldest stages of winter. The smaller rise in numbers in the fall of 2011 is due to several factors, but principally Tropical Storm Irene, which killed some birds and trashed the habitat. The spike in the fall of 2012 is due to a two-week gap in the survey due to Hurricane Sandy; during the period after the storm, many migrants flooded in from up north, due to the timing of the storm (late October, in between the major migration waves). It is highly likely that the extreme weather caused some of the unexpected spikes and drops in bird diversity, especially after the two superstorms. Data from 2013 shows a huge spike in the spring, a result of a mass influx of migrating Canada geese on one particular survey day.

The number of birds and the number of birds per observation both varied by site. Sites 1 and 4, one a border habitat and one with low disruption, showed higher numbers of birds, while Site 3, near a busy road, had both low sightings and a low number of birds per sighting. While I could also often observe the same water birds from Site 4 as I saw at Site 1, I did not double-count them.

Sites 1 and 4 had the highest bird density and diversity (Table 4), due to Site 1 being a border habitat and Site 4 being away from almost all of the regular foot traffic. High concentrations of water birds mean that more birds were seen at Site 1, despite it having a slightly lower number of species seen. Site 3’s proximity to the road appeared to cause extremely low numbers and diversity. Interestingly, mild disruption (such as at Site 2) seems to cause density drops before diversity drops, possibly because some members of disruption-intolerant species (such as some warblers) can tolerate minor disruption (such as regular dog walkers) but not major disruption (such as the close proximity of a major road).

  Observations Total Number of Birds Observed %birds %obs
Site #1 Observations 283 795 38% 29%
Site #2 Observations 239 489 23% 24%
Site #3 Observations 139 187 9% 14%
Site #4 Observations 321 610 29% 29%

Table 4. Site Specific Observation and Diversity. 

Occupancy Modeling

Occupancy modeling is used to estimate the geographic range, habitat relationships, resource selection, and population dynamics of species, and to monitor species across large areas (Mackenzie, et al.). Models can be developed for single season/single species estimates as well as multiple season/multiple species estimates. Additionally, models can incorporate variation between observers, between observation sites, and other parameters like human interaction. For this extension of my research, I use relatively simple models to establish a baseline set of estimates for the presence of selected species in the Ryker Lake observational area. 

Total Site Analysis   Yr1   Yr2   Yr3  
Species Type Model Type Psi Std error Psi Std error Psi Std error
Warblers Naïve Estimate 1.00   1.00   1.00  
  Constant P 1.00 0.0 1.00 0.0 1.00 0.0
  Survey-Specific P 1.00 0.0 1.00 0.0 1.00 0.0
Swallows Naïve Estimate 0.50   0.25   0.75  
  Constant P 0.5234 0.2639 0.2509 0.2173 0.7508 0.2167
  Survey-Specific P 0.5000 0.2500 0.2500 0.2165 0.7408 0.2187
Flycatchers Naïve Estimate 1.00   1.00   0.75  
  Constant P 1.00 0.0 1.00 0.0 0.8098 0.2432
  Survey-Specific P 1.00 0.0 1.00 0.0 0.7696 0.2241

Table 5:  Occupancy Modeling 

I focused my occupancy modeling analysis on the three families of insectivorous passerines most likely to be affected by creation of successional habitat: warblers (12 species), swallows (3 species) and flycatchers (4 species). Table 5 shows the naïve occupancy estimate and occupancy modeling statistics for each of the three years for each family of birds. For each site and observation period, the detection or non-detection of the species is marked, and the matrix is used as input for the presence models. This data representation does not take into account abundance (the number of birds sighted per site or per observation period) and assumes each site is equivalent in terms of potential occupancy. The “naïve estimate” is calculated by taking the number of times an observation of a species occurred divided by the number of observations made. This estimate assumes an ideal detector (which I’m not sure I am), and also that the species is present at the time that the detector is detecting (which might not be true in point counts for ten minutes at a site). For the presence models, I looked at whether there was no difference in the probability of me detecting the birds at each site (constant P) versus a difference due to the detector doing the survey (survey specific).

The naïve estimate tended to underestimate the presence probability of the species, and the standard error of the model showed the probability that on any given day or site, one might detect the bird type. Swallows tended to be less present than warblers and flycatchers.

In Table 6 I compared two types of models: a constant probability (p) model, and a survey-specific probability model.

Season 1            
Total Warblers:            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par. -2*LogLike
1 group, Constant P 105.96 13.33 0.0013 0.0013 2 101.96
1 group, Survey-specific P 92.63 0 0.9987 1 20 52.63
Total Flycatchers:            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par.



1 group, Constant P 50.72 0 0.9945 1 2 46.72
1 group, Survey-specific P 61.13 10.41 0.0055 0.0055 20 21.13
Total Swallows:            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par. '-2*LogLike
1 group, Constant P 42.52 0 0.9991 1 2 38.52
1 group, Survey-specific P 56.64 14.12 0.0009 0.0009 20 16.64
Season 2            
Total Warblers:            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par. '-2*LogLike
1 group, Constant P 70.31 0 0.9483 1 2 66.31
1 group, Survey-specific P 76.13 5.82 0.0517 0.0545 23 30.13
Total Flycatchers:            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par. "-2*LogLike
1 group, Constant P 57.62 0 1 1 2 53.62
1 group, Survey-specific P 78.54 20.92 0 0 23 32.54
Total Swallows:            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par. '-2*LogLike
1 group, Constant P 32.07 0 0.9999 1 2 28.07
1 group, Survey-specific P 50.5 18.43 0.0001 0.0001 23 4.5
Season 3            
Total Warblers:            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par. '-2*LogLike
1 group, Constant P 107.35 0 0.8943 1 2 103.35
1 group, Survey-Specific P 111.62 4.27 0.1057 0.1182 25 61.62
Total Flycatchers:            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par. '-2*LogLike
1 group, Constant P 58.32 0 0.9999 1 2 54.32
1 group, Survey-specific P 77.27 18.95 0.0001 0.0001 25 27.27
Total Swallows            
Model AIC deltaAIC AIC wgt Model Likelihood no.Par. '-2*LogLike
1 group, Constant P 79.4 0 0.9987 1 2 75.4
1 group, Survey-specific P 92.68 13.28 0.0013 0.0013 25 42.68

Table 6. Occupancy Model Comparisons

Out of the nine cases, in only one case is the model with a survey-specific P value a better fit, based on lower AIC value and model likelihood, than one with a constant P-value. It therefore seems likely that a constant P-value creates a more accurate model than a survey-specific P-value. This means that the probability of detection was likely uniform across the sites and observation periods for each of the species. This makes sense from the standpoint that the same detector (me) was detecting at each site, but also indicates that influences that might affect my ability to detect (sound, distractions) did not seem to play a significant role.

Site 1 Analysis   Yr1   Yr2   Yr3  
Species Type Model Type Psi stderror Psi stderror Psi stderror
All sites.   Yr 1   Yr 2   Yr 3  
Warblers Naïve Estimate 1 0 1 0 1 0
  Constant P 1 0 1 0 1 0
  Survey-Specific P 1 0 1 0 1 0
Swallows Naïve Estimate .5 0 .25 0 .75 0
  Constant P 1 0 1 0 1 0
  Survey-Specific P 1 0 1 0 1 0
Flycatchers Naïve Estimate 1 0 1 0 .75 0
  Constant P 1 0 1 0 1 0
  Survey-Specific P 1 0 1 0 1 0
Site 2 Analysis   Yr!   Yr2   Yr3  
Species Type Model Type Psi stderror Psi stderror Psi stderror
Warblers Naïve Estimate 1 0 1 0 1 0
  Constant P 1 0 1 0 1 0
  Survey-Specific P 1 0 1 0 1 0
Swallows Naïve Estimate .5 0 .25 0 .75 0
  Constant P .0469 .9127 .00112 .0034 .00331 438.1535
  Survey-Specific P 0 0 0 0 .00331 438.1535
Flycatchers Naïve Estimate 1 0 1 0 .75 0
  Constant P 1 0 1 0 .2390 .3866
  Survey-Specific P 1 0 1 0 .078 650
Site 3 Analysis   Yr1   Yr2   Yr3  
Species Type Model Type Psi stderror Psi stderror Psi stderror
Warblers Naïve Estimate 1 0 1 0 1 0
  Constant P 1 0 1 0 1 0
  Survey-Specific P 1 0 1 0 1 0
Swallows Naïve Estimate .5 0 .25 0 .75 0
  Constant P .0469 .1827 .0012 .0034 1 0
  Survey-Specific P 0 34.8224 0 0 1 0
Flycatcher Naïve Estimate 1 0 1 0 .75 0
  Constant P 1 0 1 0 1 0
  Survey-Specific P 1 0 1 0 0.0785266 0.1754
Site 4 Analysis   Yr 1   Yr 2   Yr 3  
  Model Type Psi value Std error Psi value Std error Psi value Std error
Warblers Naive 1   1   1  
  Constant P 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000
  Survey-specific P 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000
Swallos Naive .5000   .2500   .7500  
  Constant P 1.0000 0.0000 0.0012 0.0081 1.0000 0.0000
  Survey-specific P 1.0000 0.0000 0.000 0.0000 1.0000 0.0000
Flycatchers Naive 1   1   .7500  
  Constant P 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000
  Survey-specific P 1.0000 0.0000 1.0000 0.0000 1.0000 0.0000

Table 7. Site Specific Occupancy Model Output

Over all sites, warblers were most successful overall, with constant occupancy. In Year 2 (2012), swallow presence dropped, likely because of lake eutrophication due to sudden runoff pollution that interfered with insect breeding. However, swallow presence climbed again as flycatcher occupancy dropped in 2013, when pollution declined and the habitat experienced much tree destruction as a result of several large storms and early tree-clearing for golden-winged warbler habitat expansion. These results seemed to support my hypothesis that swallows would be less affected by creation of successional habitat compared to flycatchers and other warblers. The lake-edge habitat of Site 1 provided consistent habitat for all groups, but Site 2 (deeper in the forest) provided no habitat for swallows (as expected), as did the slightly disrupted deep-forest habitat of Site 3. Flycatcher presence dropped noticeably in Year 3, as the bare, open branches that they prefer for hunting perches were destroyed by storms in late 2012 and early 2013.

Yr 3 Analysis of Yellow-Rumped Warblers Site # Naïve Estimate Psi Stderror Mode 1 AIC
Constant P 1 1 1.0000 0.0000 59.07
  2 1 1.0000 0.0000  
  3 1 1.0000 0.0000  
  4 1 1.0000 0.0000  
Survey-specific P 1 1 1.0000 0.0000  
  2 1 1.0000 0.0000  
  3 1 1.0000 0.0000  
  4 1 1.0000 0.0000  

Table 8. Yellow Rumped Warblers Occupy the Ryker Lake Area

The data in Table 8 gives a baseline for occupancy statistics for the yellow-rumped warbler, the predominant warbler species at Ryker Lake and the most common forest warbler in the eastern United States. This data should provide a solid baseline for comparison as habitat modification continues in the coming years. Although the yellow-rumped warbler is a common bird, it should not be driven out of its habitat when the golden-winged warblers move in.

Limitations of the Study

The surveys were all performed at the same approximate time of day (although the exact start time varied greatly from week to week), and were not performed on the same day every week. As a result, the survey favorably detected common, diurnal birds with long morning active periods. Also, some weeks were missed entirely due to my ever-increasing school work load, unusually bad weather (such as Hurricane Sandy), or to my father (who served as transportation and bookkeeping) being out of town. I am an avid but admittedly amateur birder and ornithologist. Due to my developing audio identification skills and the fact that birds tend to resort to flight just as I have the binoculars trained on them, not all birds detected were identified and counted as observations. (This holds for species totals only—the graph of total daily bird numbers over the course of the survey period contains unidentified birds.)

Also, I found that since I was treating each observation as equivalent, this might introduce error into the presence model. My observations spanned multiple months and seasons, and migratory birds would be present only during the spring, summer, and fall months. For modeling purposes, seasonal variation might need to be taken into account.

Conclusions and Next Steps 

Conclusion 1. The use of occupancy modeling for establishing and monitoring the presence of bird species was shown to be more accurate and better than naïve estimations that assume perfect detectors and detection methods.

Conclusion 2. The baseline occupancy of warblers and flycatchers in the area intended for golden-winged warbler habitat creation is 100%. Swallows have slightly lower occupancy, especially in the specific area of successional habitat creation. Conservationists will need to be careful not to cause another species to become endangered as they try to protect the golden-winged warbler.

While I think that my hypothesis about feeding strategies would indicate that flycatchers and other warbler species might be more affected by creation of golden-winged warbler habitat, I would need to follow the population through the tree-clearing process to be sure. There are a number of next steps for me to follow up on with this research. First, I have to continue to conduct the presence analysis by species so that a full set of baseline data exists. Second, I will want to consider potential variables in the modeling like temperature, sky cover, wind, time of day, and time of year. Third, I will want to conduct the surveys after the girdled trees are felled so I can detect the maintenance of the resident species as well as new occupancy by golden-winged warblers. This might be a challenge as I am going to college in the fall, but hopefully someone else in the Citizen Science programs with New Jersey Audubon and New Jersey Fish and Wildlife can continue this important work and not let the area decline (Bouchal & Scruton).


I would like to thank my dad for driving, providing food, writing down data, showing me how to use Excel, and generally supporting me.

I would also like to thank Dr. Nellie Tsipoura and Dr. Kristin Munafo at the Citizen Science Program at New Jersey Audubon for protocol training, introducing me to the Presence software, and mentoring my research. 

Appendix A: Listing of Bird Species Observed in Survey Area During Observation Period

Common Name Scientific Name
American Black Duck Anas rubripes
American Coot Fulica americana
American Crow Corvus americanus
American Goldfinch Cardelius tristis
American Redstart Setophaga ruticilla
American Robin Turdus migratorius
Baltimore Oriole Icterus galbula
Barn Swallow Hirundo rustica
Barrow's Goldeneye Bucephala islandica
Belted Kingfisher Megaceryle alcyon
Black Vulture Coragyps atratus
Blackburnian Warbler Setophaga fusca (formerly Dendroica fusca)
Black-capped Chickadee Poecile atricapilla
Black-throated Blue Warbler Setophaga caerulescens
Black-throated GreenWarbler Setophaga virens
Blue Jay Cyanocitta cristata
Blue Winged Teal Anas discors
Brown Creeper Certhia Americana
Bufflehead Bucephala abeola
Cackling Goose Branta hutchinsii
Canada Goose Banta Canadensis
Cardinal  Cardinalis cardinalis
Carolina Chickadee Poecile carolinensis
Cedar Waxwing Bobmycilla cedorum
Chestnut-sided Warbler Setophaga penslyvanica
Chipping Sparrow Spizella passerina
Cliff Swallow Petrochelidon pyrrhonota
Common Grackle Quiscalus quiscala
Common Raven Cornus corax
Common Yellowthroat Geothylpis trichas
Coopers Hawk Accipiter cooperi
Dark-eyed Junco Junoco hyemalis
Double-crested Cormorant Phalcrocorax auritus
Downy Woodpecker Picoides pubescens
Eastern Kingbird Tyrannus tyrannus
Eastern Phoebe Sayornis phobe
Eastern Towhee Pipillo erythrophthalmus
Eastern Wood Pewee Contopus virens
Fish Crow Corvus ossifraus
Great Blue Heron Ardea Herodias
Greater Scaup Aythya marila
Green Heron Butoroides virescens
Grey Catbird Dumetella carolinensis
Hairy Woodpecker Picoides vilosus
House Finch Carpodacus mexicanus
Louisiana Water Thrush Seiurus motacilla
Magnolia Warbler Setophaga magnolia
Mallard Anas platyrynchos
Mourning Dove Zenaida macroura
Mute Swan Cygnus olor
Northern Flicker Colaptes auratus
Northern Mockingbird Mimus polyglottos
Northern Water Thrush Seiurus novaboracensis
Osprey Pandion haliaetus
Ovenbird Seiurus aurocapillus
Palm Warbler Setophaga palmarum
Pileated Woodpecker Drycopus pileatus
Prothonatory Warbler Protonotaria citrea
Red Shouldered Hawk Buteo lineatus
Red-bellied Woodpecker Melanerpes carolinus
Red-eyed Vireo Vireo olivaceus
Red-Tailed Hawk Butwo jamaicencis
Red-winged Blackbird Agelaius phoeniceus
Ring neck duck Aythya collaris
Ring-billed Gull Larus delawarensis
Ruby ThroatedHummingbird Archilochus colubris
Ruby-crowned Kinglet Regulus calendula
Rusty Blackbird Euphagus carolinus
Savannah Sparrow Passerculus  sandwichensis
Scarlet Tanager Piragna olivacea
Song Sparrow Melospiza melodia
Tennessee Warbler Vermivora peregrine
Tree Swallow Tachycineta bicolor
Tufted Titmouse Beaopholus sicolor
Turkey Vulture Cathrates aura
Veery Cathrus fuscescens
White-breasted Nuthatch Sitta carolinensis
White-Throated Sparrow Zonotrichia albicollis
Wild Turkey Meleagris gallopavo
Willow flycatcher Expidonax trailii
Wood Duck Aix spons
Yellow bellied sapsucker Sphyrapicus varius
Yellow-rumped Warbler Setophaga coronata
Yellow-throated Warbler Setophaga dominica
Yellow Warbler Setophaga petechia


Bouchal, Lyndsay Cayetana, and Bruce A. Scruton. “Reward posted in vandalism of wildlife area equipment. ” New Jersey Herald, 14 Feb 2013. Web. 28 Feb 2013.   http://www.kwwl.com/story/21203384/reward-posted-in-vandalism-of-wildlife-area-equipment.

DeFalco, Sharon, and Amanda Dey. “Golden-winged Warbler Reproductive Success and Habitat Assessment on Sparta Mountain Wildlife Management Area. Progress Report, 2003.” New Jersey Department of Fish and Wildlife Archives.    Web. 24 Feb 2013. http://www.state.nj.us/dep/fgw/ensp/pdf/literature/gw-warbler_repro-success_spartamtn.pdf. 

Donovan, T.M., and J. Hines. Exercises in Occupancy Estimation and Modeling. University of Vermont, online course, 2007. Web. http://www.uvm.edu/rsenr/vtcfwru/spreadsheets/?Page=occupancy/occupancy.htm.

“Important Bird and Birding Areas: Sparta Mountain Wildlife Management Area.” New    Jersey Audubon. Web. 28 Feb 2013. http://www.njaudubon.org/SectionIBBA/IBBASiteGuide/tabid/1258/sk/3061/Version/1b/Default.aspx.

Kaufman, Kenn. Kaufman Field Guide to Advanced Birding. Houghton Mifflin Harcourt, 2011.

MacKenzie, D., et al. Occupancy Estimation and Modeling, Inferring Patterns and Dynamics of Species Occurrence. Academic Press, 2005.

Mattison, B.J., and M.R. Marshall. “Occupancy Modeling as a Framework for Designing Avian Monitoring Programs: A Case Study Along Appalachian    Streams in Southern West Virginia.” Proceedings of the Fourth International Partners in Flight Conference: Tundra to Tropics (2012): 617-632.

Pakonis, Kristen. “Saving songbirds: Sparta Mountain project an effort to lure golden-winged warblers.” The Daily Record, 4 Jan 2012. Web. 28 Feb 2013. http://www.dailyrecord.com/article/20120105/GRASSROOTS/301050006/      

Sibley, David Allen. The Sibley Field Guide to Birds of Eastern North America. New York: Alfred A. Knopf, 2003.

“Sparta Mountain.” New Jersey Audubon. Web. 28 Feb 2013. http://www.njaudubon.org/SectionConservation/NJAUnstaffedWildlifeSanctuaries/SpartaMountain.asp

“Sparta Mountain Sanctuary/Sparta Mountain Wildlife Management Area.” Birding and Wildlife Trails. New Jersey Audubon, n.d.. Web. 28 Feb 2013. http://www.njwildlifetrails.org/SkylandsTrails/Sites/tabid/445/Scope/site/Guide/ SKYLANDS/Site/160/Default.aspx.