Prepared for: Prepared by: Prime Contractor David C. Smith and Associates, Inc SE Tacoma, Portland, OR (503)

Transcription

1 Enumeration and Monitoring Surveys of Double-Crested Cormorants in the Lower Columbia River Estuary for the 2015 Nesting Season Contract No. W9127N-13-D-0005, Task Order No. 011 Prepared for: CENWP U.S. Army Corps of Engineers Portland District Prepared by: Prime Contractor David C. Smith and Associates, Inc SE Tacoma, Portland, OR (503) Project Team Subcontractors:

2 ii Contents Executive Summary... iv 1.0 Introduction Goals and Objectives Overview of Key Tasks Methodology Detail of Key Tasks Quality Control Summary of Results Overall Counts for ESI Summary of Overall ESI Nesting Chronology and Field Observations Accuracy Assessment, Quality Assurance and Comparison with Past Methods ESI Analysis and Results by Region Western Analysis Region Central Analysis Region Eastern Analysis Region Other Colony Sites in the CRE Other Colony Sites Enumeration Results and Findings Project Deliverables Weekly Reporting Deliverables Final Deliverables Conclusions ESI Enumeration Enumeration and Monitoring of Other CRE Colonies Recommendations ESI Enumeration Enumeration and Monitoring of Other CRE Colonies...8-1

3 iii MAP FOLDOUTS (ATTACHMENTS) 2015 Monitoring and Analysis Regions... Map No Columbia R. Estuary Monitored Colonies and Loafing Areas Overview... Map No. 02a Detail Map... Map No. 02b Western Analysis Region Select Nest Distributions... Map No. 03 Central Analysis Region Select Nest Distributions... Map No. 04 Eastern Analysis Region Select Nest Distributions... Map No. 05 APPENDIX A. Weekly Biologist Reports APPENDIX B. Selected Photographs of Other CRE Colonies List of Tables Table 1-1. Summary of Task Completion Dates for Each Study Week Table 3-1. ESI DCCO Totals Table 3-2. ESI BRAC Totals Table 3-3. Overall Nesting Chronology Statistics (Central Analysis Region Only), Mean Value for 166 Nests Monitored in Weeks Table 3-4 Accuracy Assessment Results for Test Plot A and Test Plot B Table 4-1. DCCO and BRAC Totals for the Western Analysis Region Table 4-2. DCCO and BRAC Totals for the Central Analysis Region Table 4-3. DCCO and BRAC Totals for the Eastern Analysis Region Table 5-1. Summary of Weekly Nest Counts for Other CRE Colonies (excluding ESI) Table 5-2. PECO and DCCO nests observed at the Astoria-Megler Bridge Table 5-3. Comparison of Aerial Recon Photo and Boat Observation Counts List of Figures Figure 1-1. Estimated number of DCCO breeding pairs on ESI for , Bar Graph... v Figure 3-1. ESI DCCO Totals Bar Graph Figure 3-2. ESI BRAC Totals Bar Graph Figure 3-3. ESI DCCO and BRAC Nesting Pairs Totals, Combined Bar Graph Figure 3-4. ESI Nesting Chronology Chart, Comparison with Past Years Figure 3-5. Comparison of 2D and 3D Digitizing Results for Test Plot A Figure 4-1. DCCO and BRAC Totals Bar Graph, Western Analysis Region Figure 4-2. DCCO and BRAC Totals Bar Graph, Central Analysis Region Figure 4-3. DCCO and BRAC Totals Bar Graph, Eastern Analysis Region Figure 5-1. Summary of Weekly Nest Counts for Other CRE Colonies (excluding ESI)...5-3

4 iv Executive Summary Objective This study supported efforts by the U.S. Army Corps of Engineers, Portland District (CENWP) to evaluate the status of the Columbia River Estuary population of double-crested cormorants (Phalacrocoraxauritus) (DCCO). It supports the CENWP Management Plan to Reduce Predation of Juvenile Salmonids in the Columbia River Estuary. Study Area This project was divided into two primary study areas: 1) Enumeration and monitoring of DCCO and Brandt s cormorants (Phalacrocoraxpenicillatus) (BRAC) on East Sand Island (ESI), and 2) Enumeration and monitoring of DCCO at other colony sites throughout the Columbia River Estuary (CRE), and exclusive of ESI. The ESI study area is shown on Map 01. The other colony sites throughout the CRE that were surveyed and monitored are shown on Map 02a (entire study area) and Map 02b (detail area of highest concentration of colonies). Methods The enumeration and monitoring work on ESI relied primarily on photogrammetric surveys utilizing high resolution stereo imagery and field observations. General reconnaissance by boat and aircraft were used to supplement field and photogrammetry surveys in areas of limited visibility. The enumeration and monitoring work for other colony sites in the CRE utilized aerial and boat-based reconnaissance and observation. Both boat and aerial reconnaissance were conducted to identify key loafing and roosting areas, and to identify new nesting areas (five or more active nests in one contiguous location). Where more than five active DCCO nests were detected in one contiguous location, detailed counts were obtained by observation from boat. Oblique aerial imagery was used to verify boat counts; in most cases the counts from boat observation were used for reporting. Required metrics for the ESI work, reported separately for DCCO and BRAC, included: number of nesting DCCO/ BRAC number of breeding pairs nest density and distribution total area (acreage) occupied by DCCO/BRAC characteristics of habitat use nesting chronology (limited to what could be observed from blinds in the time allotted) and any other relevant observational information pertaining to nesting success and failures and colony productivity (e.g., site conditions, evidence of predators, etc.) Evaluation of nesting success was not a focus of the study and could not be fully determined within the scope of the tasks authorized under this contract. All CRE colony sites detected and monitored (other than ESI) were located on bridge structures or channel markers. Nest density and total area were not reported. Nesting chronology observations were limited to the observation of birds on nests and audible or visible chicks or fledglings. Eggs in nests could not be observed.

5 v Results, ESI The largest observed population ( 12,150 DCCO breeding pairs and 2,071 BRAC breeding pairs) was observed on ESI during the Week 10 (July 1, 2015) aerial imagery counts. This observation followed a three week gap in observations due to contracting delays. The Week 10 counts were the highest observed value and are assumed to represent the peak cormorant population on ESI for the 2015 breeding season. This observed peak DCCO population was about 10 percent less than the peak population reported for the 2014 breeding season. A comparison of the 2015 results to past years is shown in Figure 1-1. Figure 0-1. Estimated number of DCCO breeding pairs on ESI for , Bar Graph * Data in Figure 1-1 for 2014 and prior years was provided by the USACE from results published in prior annual reports on the research and monitoring of avian predation in the Lower Columbia River. Error bars for 2014 and prior years are presumed to represent 95% confidence intervals based on multiple counts; error bars from 2015 represent an estimated range of variability as outlined in Section 3.3.

6 vi Results, Comparison with Past Methodologies A comparison with past 2D aerial image counting methods was performed. The stereophotogrammetry results fell within the overall error range of the 2D methodology used in past studies, but had a higher level of confidence for total nest counts and species delineation. Results, other CRE colonies An estimated 1,022 additional DCCO nests were observed at other colony sites in the CRE. The largest of these colonies was found on the Astoria-Megler bridge. Other notable colonies include marine navigation markers between river miles 20 and 30, Longview Bridge, and transmission towers near Troutdale/Camas. Conclusions and Recommendations The overall approach and methodology proved to be a successful means to efficiently enumerate DCCO and BRAC nesting pairs on ESI and other colonies throughout the CRE. Stereophotogrammetry methods used for ESI produced results that are consistent with past methods and have the advantage of establishing a more accurate baseline and more repeatable methodology for future work. Use of GIS technologies for georeferenced deliverables will allow the study data to better facilitate future evaluation of changes in population trends. Field observations provided critical and necessary information, though observations of nesting chronology were limited due to lack of access to blind 4 and contracting delays. Recommendations for future efforts include addressing contracting timing issues to better track nesting chronology (if desired), and to ensure observations capture the peak colony population. Additional site preparation is also recommended in order to achieve more value from field observations.

7 Introduction 1.1 Goals and Objectives The primary goals and objectives that the adopted methodology sought to achieve included: 1) Establish a new and efficient methodology of using remote sensing to estimate nesting counts, including distinction between BRAC and DCCO on ESI; 2) Utilize existing field survey capability (boats, crews, and resources) to accomplish field missions; 3) Incorporate Geographic Information Systems (GIS) technologies for field work and reporting; 4) Ensure data collected with these new methods meet project needs and criteria as established by past studies and identify differences between current methodologies and past efforts so there can be a meaningful comparison and analysis of results; 5) Evaluate options for improved technologies for future counting efforts, particularly as new technologies become available. 1.2 Overview of Key Tasks Four primary tasks were performed to complete the survey: 1) Aerial reconnaissance of the entire CRE study area to identify new colonies or areas requiring boat or field observations; 2) Field observation at ESI; 3) Aerial photogrammetry at ESI for DCCO and BRAC counts; and, 4) Boat observation/survey of identified nesting colonies and reconnaissance to identify non-nesting areas of high use (greater than 50 loafing/roosting birds). Overall project management, reporting and coordination of subcontractors was the responsibility of prime contractor David C. Smith and Associates, Inc. (DSA). Statewide Land Surveying, Inc. (SWLS) provided: field survey personnel for boat and field observations; GIS field tablets for data recording; survey boats; and coordination of daily activities for boat and field survey missions. Harris Environmental Group, Inc. (HEG) provided an avian biologist with past experience observing and studying cormorant nesting behavior and habitat. Photogrammetry work and final integration of all field data was performed by DSA. Monitoring and enumeration tasks were performed for 10 study weeks during the 15 week period from April 27, 2015 and August 7, Field, boat, and flight missions were completed consecutively for Weeks 1 through 6 under the initial contract. A contract amendment was executed to extend the work to include Weeks 10, 11, 13 and 15. Monitoring or enumeration was not conducted for three weeks during the peak of the breeding season, between June 4, 2015 and July 1, A schedule of dates for completed tasks is summarized in table 1-1.

8 1-2 Table 1-1. Summary of Task Completion Dates for Each Study Week Study Week Aerial recon flight ESI field visit Date of photography ESI Aerial Photography Provider and camera Flight Alt (MSL) Flight GSD (m) Boat Missions Standard boat missions Addtl. recon, Longview to Troutdale Week #1 4/27/2015 4/29 4/27 4/28 GPS Surveying, Inc. ione m /28-30; 5/1 x Week #2 5/4/2015 Week #3 5/11/2015 Week #4 5/18/2015 5/5 5/4 5/4 5/11 5/12 5/11 5/18 5/18 GPS Surveying, Inc. ione 85 Bergman Photographic Ultracam XP Stereo imagery not acquired; handheld aerial recon imagery used 340 m /5-5/7 5/7 370 m n/a n/a 5/11; /19; x x Week #5 5/25/2015 5/27 5/27 5/27 GPS Surveying, Inc. ione m /26; 5/28-29 x Week #6 6/1/2015 6/4 6/4 6/4 GPS Surveying, Inc. ione m /1-2;6/5 6/3 Week #10 6/29/2015 6/29 6/30 7/1 Bergman Photographic Ultracam XP 370 m /29; 7/1-2 x Week #11 7/6/2015 7/7 7/6 7/6 GPS Surveying, Inc. Jena LMK (film) 205 m /7-8;7/10 7/9 Week #13 7/20/2015 7/23 7/23 7/23 GPS Surveying, Inc. Jena LMK (film) 190 m / /24 Week #15 8/3/2015 8/3 8/3 8/3 GPS Surveying, Inc. Jena LMK (film) 200 m /4-6 8/7

9 Methodology 2.1 Detail of Key Tasks Aerial Reconnaissance Aerial reconnaissance missions were scheduled as early in the week as possible given considerations for weather, aircraft availability and scheduling of other project tasks. The primary purpose of the aerial reconnaissance flights was to monitor the CRE study area for other DCCO colonies and to direct boat and field crews to sites that may require field observation. Aerial recon flights were also used for the purpose of acquiring oblique imagery of other nesting colonies when possible and practical.not all sites could be photographed on every mission. When suitable, aerial recon imagery was used as a secondary check to validate the primary bird counts acquired from boat or field observation. In most cases, the boat observations were more reliable and were used as the reported value. The initial aerial reconnaissance mission in Study Week #1 included the HEG avian biologist and the SWLS survey crew manager. This provided the biologist with an overview of the study area and the opportunity to provide input on what habitat could potentially support DCCO nesting colonies. Information gathered during the first flight was used to structure and plan remaining flight missions. The SWLS survey crew manager was responsible for identifying and recording potential new nesting and colony areas that may require field reconnaissance. SWLS utilized GIS enabled tablets to record areas with colonies. The GIS information was used to assist in planning boat and field survey missions. Oblique aerial photography was obtained with a handheld Cannon DSLR camera and a 300mm zoom lens. For Study Week #2 and subsequent flights, the SWLS survey crew manager was accompanied by DSA s project manager. DSA s project manager coordinated and scheduled the flight missions and was responsible for acquiring aerial oblique imagery used to document other colony sites and validate bird counts from boat observation. On several missions, aerial oblique imagery was also acquired for ESI as general reference imagery to assist in the aerial photogrammetry interpretation of nesting vs. non-nesting birds. For Study Week #4, this reconnaissance imagery was used for final bird counts, as stereo coverage imagery with the aerial mapping camera could not be acquired due to cloud cover and weather. ESI Field Observation and Survey ESI field observations coincided as closely as possible with the aerial photography flights. In most cases, the field and aerial photography missions occurred on the same day; however weather and scheduling presented some conflicts. In Week #1 and Week #10 the field observations were made the day before the aerial photography flight and in Week #3 the field observations occurred the day after. The ESI field missions were performed by SWLS survey crews accompanied by the HEG avian biologist. Survey crews located and delineated approximate areas of BRAC colonies on the island, noted predator activity, obtained photographs of nesting DCCO and BRAC, and made nest counts to ground truth and validate the aerial photogrammetry results. The HEG biologist provided overall input/supervision of the field observations and recorded detailed field notes for reporting habitat use, nesting chronology, and other relevant information related to bird behavior and colony productivity. Survey crews utilized GIS tablets to record findings, delineate BRAC colony areas, identify ground truth areas, note predator activity, and document nesting chronology information.

10 2-2 Field observations were made from Blinds 1, 2, and 3. Only portions of Monitoring Regions A-M (see Map No. 01) could be seen from the blinds. In the Western Analysis Region, most cormorants were nesting outside areas visible from the blind or boat; field observations were limited to about 50% of the nesting sub-colony in Region B, and less than 10% in Region P. Regions A and C-E were visible, but had only intermittent nesting activity. In the Central Analysis Region, Regions F, G, J and L were nearly completely visible from the blinds; however, terrain limited visibility of regions H and I to about 70% to 80% of those nesting sub-colonies. Blind 4 could not be accessed without disturbing nesting birds, limiting what could be observed in regions K, M, N and O. Field observations were limited to approximately 40% of region K, 25% of region M, and less than 10% of regions N and O. Nesting attempts by 218 nesting pairs of DCCO in nine separate plots were monitored for chronology and productivity. Sampling plots were selected for good visibility from the blinds and to be as representative as possible of the monitoring region. Individual nests were assigned unique letters and labeled on digital photographs. Observations of nest contents were recorded during each field visit in order to estimate nesting chronology. Breeding success (number of chicks fledged per nest) could not be accurately determined because of a three week gap in monitoring at the time when chicks began to fledge. Similar to nesting chronology sampling areas, several ground truth areas were established each week for the purposes of validating the interpretation of nesting vs. non-nesting birds from aerial photography. Because of the limited visibility from the blinds and the difficulty in correlating areas observed in the field to the same area on the rectified imagery, it proved difficult to establish large enough areas to be of statistical significance. The ground truth areas were used for overall validation of the aerial photography interpretation methods, but were not used for statistical error analysis. ESI field observations are summarized in the narrative, charts and tables in Sections 2 and 3 of this report. Detailed field data (as recorded on the GIS tablets) and weekly biologist reports on nesting activity and bird behavior were provided as weekly interim reports and data deliverables. Aerial Photogrammetric Counts ESI nesting chronology field observation: Example log book notes (top) and sample plot identification (bottom). ESI bird counts and enumeration metrics were determined using high resolution stereo imagery and softcopy photogrammetry 3D mapping technologies.with the exception of Week 4, stereo coverage aerial photography was acquired for each study week to support both softcopy stereo photogrammetry bird counts and secondary digital orthophotography. In Week 4, stereo coverage imagery could not be acquired due to cloud cover and weather; oblique aerial reconnaissance photos were used for final counts during this week. Digital orthophotography was prepared each week for report maps, general reference, and quality control of stereo-photogrammetric bird counts. The mapping targeted relative accuracies needed to

11 2-3 support DCCO enumeration and monitoring tasks only. Processes were not intended to achieve published map accuracy standards. Detailed metadata was provided with the final orthophoto and bird count point file GIS deliverables. The metadata provides additional detail on control methodologies, Digital Elevation Model (DEM) sources used, and limitations regarding each week s imagery. The digital orthophotography was used to generate an initial point coverage of cormorant locations from an unsupervised GIS image classification. This classification was based on pixel color and cluster size. Given the dark color of the cormorants and generally even, light colored terrain within nesting areas this process was very successful for generating an initial point coverage. In rocky, rough terrain the automation was less practical. Final counts were done by manual review in a 3D softcopy stereo photogrammetry environment. The automated point coverage was superimposed on the stereo imagery and thoroughly reviewed and edited to remove false positives, digitize missed cormorants, and classify the points as either nesting or nonnesting. Stereo 3D was used instead of 2D methods as 3D made it much easier to DCCO (orange) and BRAC (blue) points digitized on nests, distinguish individual stationary superimposed on orthophoto imagery. birds from shadows or moving birds, identify clusters of birds and identify mounded nests. During initial nesting stages (in particular Week 1), distinguishing nesting vs. non-nesting birds was difficult. Many nests were poorly formed with nesting birds were not tending eggs and a high level of nest building and courtship activity intermixed with established nests. During Week 1, stationary birds at even spacing were assumed to be nesting. The Week 1 counts are expected to overestimate nesting birds. As the nests became more developed and the birds began tending eggs (Week 2 and later), the distinction between nesting and non-nesting became more discernable. For Week 2 and later, a bird point was classified as an active nest primarily if the bird was on or near visible nesting material and/or visible mounds. In some cases nesting birds obscured the nesting material and it was more difficult to make this distinction. In these cases, additional criteria evaluated included behavior (stationary/sitting vs. standing/moving, or tight clusters indicating nesting pairs or adults with chicks), whether or not the bird was in the same location on overlapping flights and location relative to nesting locations digitized from past flights.

12 2-4 Distinguishing between BRAC and DCCO nests proved more challenging than simply identifying active cormorant nests. Particularly since Blind 4 could not be accessed, field observations could only provide limited observations, counts, or ground truthing for delineating BRAC sub-colonies. BRAC delineation for Weeks 1 and 2 was based solely on field observations. Because of limited Aerial recon Week 4 oblique photo of ESI used for DCCO and BRAC counts, Week 4 oblique photo, close-up showing BRAC vs. DCCO nests visibility, BRAC delineations for these first two weeks may significantly underestimate the total counts. To address this problem, methodologies were changed in Week 3. Criteria were established using nest height and material type so that species delineations could be made from stereo imagery. Field observations continued to be used to identify general areas with nesting BRAC. Focusing on these areas, stereo photogrammetry was used to delineate between species based on nest characteristics. Stereo imagery was not available for Week 4; Week 4 counts were achieved by mosaicing oblique imagery into a series of mosaics. In most cases, oblique imagery was available from both sides of the island, minimizing the amount of area obscured by blinds and tunnels and allowing for a clear view of birds to be counted. Nesting material was often visible and similar behavior criteria as used from stereo imagery could be used to help distinguish nesting vs. non-nesting birds. Areal delineation was not possible since the imagery was not to scale. This methodology appeared to be generally consistent with counts from stereo coverage imagery, though of slightly lower confidence since quality control processes were more difficult to implement. Furthermore, there is a possibility of some overlap in imagery as well as some areas of limited visibility. A final point file was prepared containing attributes data for each bird identified on the photo. Bird points were assigned the following attributes: DCCO_NEST: DCCO: BRAC_NEST: BRAC : Occupied DCCO nest Other DCCO, including loafing or roosting birds as well as any additional birds in a nest, when occupied by more than one bird Occupied BRAC nest Other BRAC, including loafing or roosting birds as well as any additional birds in a nest, when occupied by more than one bird The attributed point files were intersected with polygon boundaries for each management unit to quantify nesting and non-nesting BRAC and DCCO counts for each region. These values were reported for each monitoring region in the weekly interim reports and then aggregated and summarized according to the three primary analysis areas for this final report (the Western, Central, and Eastern subcolonies). Boat observation and survey Boat observation/survey was relied on for bird counts at other colonies in the CRE and for reconnaissance to identify non-nesting areas of high use (greater than 50 loafing/roosting birds).

13 2-5 Weekly boat surveys counted and monitored all identified DCCO colonies consisting of 5 or more nests as shown on Map No. 02a and 02b. Reconnaissance was performed on stretches of river en-route to survey areas, resulting in weekly monitoring by boat (supplementing the work done by aircraft) of most of the Columbia River below Longview. Boat observation was the most efficient way to locate and monitor groups of loafing birds greater than 50 in number. Additional boat missions were mobilized periodically throughout the study period to monitor other stretches of river not along the standard routes or not readily visible from the aircraft. All of the Columbia River, Willamette River and Multnomah Channel reaches within the CRE study area were observed by boat at least once during the study period. The stretch of river between Longview and Government Island was monitored by boat for five of the ten study weeks. The Willamette River, Scappoose Bay and Multnomah Channel were observed once. No birds were observed and these areas appeared to have very low potential to support nesting habitat. While boat missions for those areas were not repeated, monitoring by aircraft continued within limits of visibility. Additional field observation At project startup, field visits were made to many of the historic colony sites noted from previous studies. It was initially anticipated that additional field visits or boat missions may be required to monitor these sites. However, it was concluded that all of those secondary sites had very low potential for DCCO colonies and could be more efficiently monitored by either boat or aircraft. All historic sites were observed at least once during the breeding season, either in the field, by boat, or aircraft. Sites that showed higher potential for nesting colonies were monitored more frequently by aircraft and boat at regular intervals throughout the study period. For the Longview Bridge, Week 1 counts were observed from the bank on the north shore. For subsequent weeks, counts from boat observation were found to be more efficient. 2.2 Quality Control Quality control of nest counts and species delineation on ESI were especially important to this study because weekly results were used by CENWP and other agencies to inform decisions related to management of DCCO without disturbing BRAC. Although timelines for processing and reporting results during the breeding season were very short, each step of the weekly reporting process had a built-in quality review. Counts of the total number of cormorants on the island were obtained through GIS image classification followed by a manual photogrammetric review to refine the total. Selected ground truthing plots were used to compare field counts to aerial image counts to verify the methodology. Delineation of nesting areas for each species were approximated by the field crew and digitized on a data collection tablet. Those approximate delineations were reviewed and refined by photogrammetric interpretation of the nest type and characteristics indicative of each species for the final delineation of the nesting areas of each species. After the breeding season, once the date of photography with the largest observed ESI DCCO population was identified, results from that particular week underwent further refinement and quality assessment (QA), including a comparison to past 2D photo interpretation methods in order to ensure

14 continuity between past and future monitoring efforts. See section 3.3 for detailed discussion of the final QA procedures, accuracy assessment and adjustments made to the final results. 2-6

15 Summary of Results 3.1 Overall Counts for ESI Aerial photography was used to determine the following final counts and metrics for DCCO and BRAC on ESI: Number on colony (observed value): Total number of birds observed on the photography at time of flight. Number nesting on colony(derived/computed value): It is assumed that the number of nesting birds is exactly equal to two times the number of active nests observed. The reported value is a derived number computed as two times the number of active nest points digitized. Number of nesting pairs on colony (observed value): Assumed to be equal to the number of active nests (nests with either one or two birds present) as interpreted and digitized from the aerial photography. Acres occupied by nests (derived/computed value): Areas occupied by nesting birds were derived using a concave hull algorithm to generate polygons around points interpreted and digitized as active nests. An approximate separation value of 3 meters was used as the basis for clustering points. The reported acres occupied value only includes the area occupied by active nests. (Note: The areas provided in the interim weekly report deliverables used a digitized polygon around visual clusters of birds. This process was improved upon for the final reporting provided here. The final reporting values are intended to supersede and replace those from the interim reports). Areal nest density (derived/computed value): Reported as the number of nests per acres. Reported value was computed as the number of active nest points divided by the acres occupied value. Metrics from field observations of ESI and photogrammetric counts were prepared for each study week for each of the 16 individual monitoring regions shown on Map No. 01. For this final report, those metrics have been aggregated into the three primary Analysis Regions (Western Subcolony Analysis Region, Central Subcolony Analysis Region, and Eastern Subcolony Analysis Region) as shown on Map 01. The following tables and charts summarize the overall counts and results for the entire ESI colony. Maps 03, 04 and 05 show how the areal distribution of nesting cormorants fluctuated over the span of the study period. These maps show the nest distribution for each of the three different portions of the island for four selected aerial photography observation periods (April 28, June 4, July 1, and August 3). A more detailed breakdown of the counts and metrics for each Analysis Region is presented in Section 3.

16 3-2 Table 3-1. ESI DCCO Totals East Sand Island DCCO Enumeration Metrics Metric Number on colony (as observed on imagery) Number nesting on colony (computed as 2x # nests) Nesting pairs on colony (# nests observed) Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 10 Week 11 Week 13 Week 15 (4/28/2015) (5/4/2015) (5/11/2015) (5/18/2015) (5/27/2015) (6/4/2015) (7/1/2015) (7/6/2015) (7/23/2015) (8/3/2015) 10,779 15,206 19,378 13,979 12,705 22,380 29,217 23,464 15,249 27,786 16,016 21,044 22,278 18,370 18,294 19,220 24,300 23,052 12,340 8,708 8,008 10,522 11,139 9,185 9,147 9,610 12,150 11,526 6,170 4,354 Acres occupied by nests n/a Areal nest density (# nests / acre) 4,110 4,636 4,335 n/a 4,368 3,796 4,092 3,990 2,810 2,371 Figure 3-1. ESI DCCO Totals Bar Graph

17 3-3 Table 3-2. ESI BRAC Totals East Sand Island BRAC Enumeration Metrics Metric Number on colony (as observed on imagery) Number nesting on colony (computed as 2x # nests) Nesting pairs on colony (# nests observed) Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 10 Week 11 Week 13 Week 15 (4/28/2015) (5/4/2015) (5/11/2015) (5/18/2015) (5/27/2015) (6/4/2015) (7/1/2015) (7/6/2015) (7/23/2015) (8/3/2015) ,059 2,145 3,558 4,577 4,959 5,980 3,377 5, ,110 2,552 2,200 3,272 3,994 4,142 3,490 3, ,555 1,276 1,100 1,636 1,997 2,071 1,745 1,712 Acres occupied by nests Intermixed n/a Areal nest density (# nests / acre) n/a 1,287 3,022 n/a 3,453 2,768 3,483 3,483 3,432 3,301 Figure 3-2. ESI BRAC Totals Bar Graph

18 3-4 Figure 3-3. ESI DCCO and BRAC Nesting Pairs Totals, Combined Bar Graph All enumeration counts for ESI were from stereo aerial photography with the following exceptions: Week 1 and Week 2 delineation of BRAC colonies was done in the field and not from the aerial photography. Since field delineation of BRAC polygons could not be directly correlated to the imagery, the value reported in the tables for BRAC in Week 1 and 2 is the field observation value. BRAC points identified in the GIS deliverables for those weeks are for visual representation only and do not directly correlate the reported value. As discussed in the methodology section, in Week 3 and subsequent weeks, BRAC delineation was done from aerial photography; BRAC GIS points in those weeks directly correlate to reported counts. Oblique aerial recon photography was used for Week 4 counts as discussed in the methodology discussion. Acres occupied and areal nesting density could not be reported for Week 4. Counts were obtained by tabulating points digitized on oblique imagery mosaics; these points are not available as a GIS deliverable as the imagery was not at a uniform scale.

19 Summary of Overall ESI Nesting Chronology and Field Observations DCCO Nesting Observations DCCO nesting began prior to the initial field visit on April 27, Approximately 50% of the nests in the Central Analysis Region were observed to contain eggs during the April 27 observations. By Week 3 (May 12 observations) the percentage of nests containing eggs ranged from85% to 100%, depending on monitoring region. DCCO chicks were first observed in Week 5; however, the following three weeks between Week 6 (week of June 1) and Week 10 (week of June 29) were not monitored. The oldest chicks were estimated to have reached fledging age (28 days after hatching) during the Week of June 22, a week in which monitoring did not occur. During Week 10, crews observed some older chicks moving about the breeding area, making it difficult to assign chicks to a particular nest. Because the three week gap in monitoring occurred during a period when the chicks began leaving the nest, nesting success could not be accurately determined. Nesting chronology was documented beginning in Week 3. Nesting chronology statistics could only be attained for the Central Analysis Region. Nests in the Western Analysis Region that were visible from the blinds and established in Week 3 were no longer occupied in Week 4 and all subsequent weeks. Nests in other parts of the Western Analysis Region and throughout the entire Eastern Analysis Region were not visible from the blinds, preventing field crews from assessing nesting chronology in these areas. Nesting chronology varied in those monitoring regions where it could be observed.nest chronology in Monitoring Regions I and K were generally the most advanced. Monitoring Region H was the least advanced as a result of many of the nests being vacated during week 5, including nearly all of the selected nests within the sampling area. Some nests in the sampling area in Monitoring Region H were re-established in subsequent weeks, but nesting chronology remained delayed relative to other Monitoring Regions. Nesting chronology statistics for the 166 nests monitored in the Central Analysis Region are summarized in Table 3-3 and compared with past years (Figure 3-3). Because of the timing of arrival on the island and gap in observations, a direct comparison with past years is not possible for all metrics. Table 3-3. Overall Nesting Chronology Statistics (Central Analysis Region Only), Mean Value for 166 Nests Monitored in Weeks Observation Date Nesting Chronology Week 3 Week 4 Week 5 Week 6 Week 10 Week 11 Week 13 Week 15 Metric (5/11/2015) (5/18/2015) (5/27/2015) (6/4/2015) (7/1/2015) (7/6/2015) (7/23/2015) (8/3/2015) Number of nests monitored Percent abandoned 0.4% 0.4% 17.5% 8.8% 10.2% 8.9% 67.3% 78.9% Percent empty, but active 20.7% 5.3% 5.7% 20.8% 3.4% 3.7% 7.6% 2.7% Percent containing eggs 78.9% 82.7% 41.5% 20.9% 17.6% 18.7% 1.1% 0.5% Percent containing chicks 0.0% 0.0% 25.8% 45.8% 63.5% 63.8% 22.9% 17.9% Percent Unknown * 0.0% 11.6% 9.1% 3.7% 5.4% 4.9% 1.1% 0.0% *Nest contents were recorded as unknown if a nesting bird remained still on its nest throughout the monitoring period, thereby preventing surveyors from viewing nest contents. These nests are presumed to contain eggs.

20 Figure 3-4. ESI Nesting Chronology Chart, Comparison with Past Years. 3-6

21 3-7 BRAC Observations. BRAC were first observed nesting during Week 2 (week of May 4). BRAC nesting was most advanced in the Central Analysis Region, particularly in Monitoring Regions I, K, L, M and N. BRAC nests established in the Eastern Analysis Region in Week 3 (week of May 11) were no longer in use after Week 6 (week of June 1). BRAC were first observed establishing nests in Monitoring Region P within the Western Analysis Region in Week 5 with 65 nesting pairs; the number of nests in that area continued to increase for the remainder of the study period. BRAC eggs were first observed during Week 5 (week of May 25). BRAC chicks were first observed during Week 10 (week of June 29), after a three week gap in monitoring. Other Observations During Week 13 (week of July 20), crews observed approximately dead DCCO juveniles in monitoring zones I, K, L, and M. A similar number of DCCO juveniles exhibited symptoms consistent with Newcastle Disease (ND). During Week 15 (week of August 3) crews observed approximately 75 DCCO that were dead or exhibited symptoms consistent with ND. The disease appeared to only be affecting large, adult-sized DCCO juveniles. No other age classes or species were observed exhibiting symptoms. The outbreak appears to be centered around Blind 3, with the largest concentrations of dead or symptomatic birds in monitoring regions I, J, K, L, and M. Very few dead and no symptomatic DCCO were observed in other monitoring regions. Additional dead or symptomatic DCCO were observed along the northern shore of ESI when crews entered and exited the island. Based on crew observations, the disease may have impacted roughly DCCO on ESI. 3.3 Accuracy Assessment, Quality Assurance and Comparison with Past Methods This study utilized stereo photogrammetry in order to increase the accuracy of nest counts and nest type delineation compared to a more traditional two dimensional image interpretation. To assess the accuracy of enumeration and identification results in terms of the repeatability of the measurements, the ratio of range to the mean was calculated. Repeatability and accuracy were assessed using the imagery and counts collected on July 1st, the flight date when the peak population was observed on ESI. This assessment was used to adjust and finalize the preliminary nest counts for the peak week and to estimate a range of uncertainty to be applied to the final DCCO peak nesting count. The assessment focused on three key aspects of the image interpretation: 1)The ability to identify cormorants on the imagery,2) The ability to distinguish active nests (defined as being occupied by one or more birds), and, 3) The ability to further categorize an active nest as a BRAC or DCCO nest. Assessments were done using both 2D imagery (for comparison to with past methods) and the stereo 3D imagery used for the current enumeration work. To compare the stereo 3D methodologies to past 2D methods, an initial test plot was selected and first analyzed in 2D by experienced GIS professionals. Next, the same area was evaluated in stereo 3D by both professional stereo compilers and GIS professionals. Four metrics were evaluated: 1) total number of cormorants in the scene, 2) total number of cormorant nests in the scene, 3) number of DCCO nests in the scene, and 4) number of BRAC nests in the scene. After evaluation of the first test area (Plot A) in both 2D and 3D a second test area was selected and evaluated in stereo only, to

22 3-8 further validate the estimated range of uncertainty achieved by stereo 3D image interpretation.compiled results of all the participants results are shown in Table 3-4 and Figure 3-5 below. Table 3-4. Accuracy Assessment Results for Test Plot A and Test Plot B COMPARISON OF 2D GIS/ORTHO (NO STEREO) DIGITIZING RESULTS - TEST PLOT A n min max range mean range/ mean Total Cormorants Total Cormorant NESTS Total DCCO NESTS TOTAL BRAC NESTS COMPARISON OF 3D STEREO DIGITIZING RESULTS - TEST PLOT A n min max range mean range/ mean Total Cormorants Total Cormorant NESTS Total DCCO NESTS TOTAL BRAC NESTS D STEREO DIGITIZING RESULTS - TEST PLOT B n min max range mean range/ mean Total Cormorants Total Cormorant NESTS Total DCCO NESTS TOTAL BRAC NESTS

23 3-9 Figure 3-5. Comparison of 2D and 3D Digitizing Results for Test Plot A. Comparison of 2D and 3D Digitizing Results - Test Plot A 2D 3D Total Cormorants Total Cormorant NESTS Total DCCO NESTS TOTAL BRAC NESTS Error bars in Figure 3.5 represent the total range of measurements, not standard error. Comparison of current 3D methods with past 2D methodologies for nest determination and total bird counts. For Plot A, as shown in Table 3-4 and Figure 3-5, analysts digitizing DCCO nests in 2D imagery agreed on the number nests in the scene within a range of about 9% of the mean value; analysts using 3D stereo methods agreed on the number of nests within a scene within a range of slightly over 1% of the mean for both Plot A and Plot B. Image interpretation to distinguish between BRAC and DCCO nests resulted in a wider range of uncertainty. The 2D measurements produced a range of greater than 40% of the mean and 3D measurements produced a range of approximately 10% to and 20% of the mean. Measurements made of nesting vs. non-nesting birds with the stereo 3D methodology fall within the overall range of measurements made with the 2D approach, but achieve a significantly higher range of precision. This indicates that the methodology is consistent with past methods in terms of overall accuracy for purposes of comparison, but can also establish a more precise baseline for use in future enumeration efforts. For the determination of the total number of cormorants observed (including nesting and loafing birds) the stereo results were only slightly better than the 2D results. This is due to the fact that it

24 3-10 was very difficult to distinguish and count individual birds when the birds were tightly clustered together, no matter which method was used. Since the number of loafing birds is constantly changing, this range of error has minimal impact on the primary objectives of the enumeration. Adjustment of BRAC vs. DCCO nest counts for final Week 10 counts Based on the findings from the accuracy assessments of Test Plot A and Test Plot B, further evaluation of the BRAC vs. DCCO estimate was performed to refine the Week 10 DCCO count. As the range of uncertainty for the BRAC vs. DCCO delineation is relatively small compared to the overall DCCO nest count, only the Week 10 count (the week with the highest nest count) was re-evaluated. Larger error bars were assigned to the other weekly counts. The main uncertainty in distinguishing BRAC nests was at the edges of the sub-colonies where the buffer between DCCO and BRAC sub-colonies was not always clear. The percent error was smaller for large contiguous or more isolated BRAC sub-colonies and larger for smaller sub-colonies where the two species in much closer proximity making it more difficult to delineate between the two nest types. In order to narrow the range of uncertainty in the BRAC count as it affects the Week 10 DCCO peak nest count, the following methodology was used: 1) BRAC nest counts were repeated by two stereo operators for all of the BRAC sub-colonies in the Central Analysis Region for Week 10. BRAC nesting in the Western Analysis Region were an isolated sub-colony and easily distinguished with minimal uncertainty and were not re-evaluated. 2) The total range of measurements for the photogrammetry estimate of BRAC nests for Week 10 was approximately 20% of mean BRAC nest count for the Central Analysis Region. This equates to a range of uncertainty of about 1.6% of the reported peak DCCO nest count. 3) The original BRAC delineation and the re-counts were provided to the field crew for review and refinement to narrow the range of uncertainty. The field crew overlaid the original BRAC delineation boundary with the two point sets created during the stereo 3D recounts. This data was compared to photos taken in the field, field notes and overall field observations. Based on this information, the field crews determined a final boundary as the best estimate delineating BRAC and DCCO nests. 4) The final boundary was used to refine the nest points for Week 10 and to adjust the final Week 10 DCCO and BRAC nest counts. A relatively minor adjustment of 78 nests was made to the preliminary Week 10 DCCO nest count for a final peak observed DCCO nest count of 12,150.

25 4.0 ESI Analysis and Results by Region 4-1 Field observations and aerial photography counts were initially performed and reported for each Monitoring Region (A P). The initial monitoring regions were delineated to correspond to boundaries used for potential management activities. For final reporting, individual monitoring regions in which the cormorants exhibited similar behavior and nesting chronology were aggregated together to form final Analysis Regions consisting of the Western, Central, and Eastern portions of the island. These final Analysis Regions represent the range of nesting chronology and activity observed on the island. Final metrics and a synopsis of nesting behavior for each analysis region are presented in Sections 4-1 through 4-3. Final results may vary from those reported in the interim weekly reports, reflecting modifications or updates that were made in finalizing the data. The counts and area data below supersede and replace the counts in the prior weekly reports. Weekly biology reports were compiled for each study week. These include more detailed nesting observations and nesting chronology statistics broken out by individual monitoring region. The weekly biology reports are included in Appendix A. 4.1 Western Analysis Region The Western Analysis Region contains Monitoring Regions A-E and P. Nesting activity was erratic and less advanced on this portion of the island. Many of the DCCO nests that had appeared to be established nests in Week 1 were not occupied in Week 2. Nesting in the Western Analysis Region continued to be erratic throughout subsequent weeks, particularly in the area bounded by tunnels in Monitoring Regions A-E. By Week 10, nearly all of the nests inside the area bounded by the tunnels were no longer occupied; no additional nesting activity was observed in subsequent weeks. These DCCO may have relocated to other portions of the island, possibly due to high predation pressure from bald eagles (Haliaeetusleucocephalus) and gulls. While few specific depredation events were observed, field crews frequently observed numerous bald eagles flying or loafing in the Western Analysis Region; number of eagles observed daily ranged from 0 to a maximum of 20 individuals on May 12. In areas outside the area bounded by tunnels, more stable nesting behavior for both BRAC and DCCO began to occur in Week 6 and was well established in Week 10 and subsequent weeks. Most of this occurred in the western portion of the island (primarily BRAC) and in the dissuasion areas (DCCO). Map 03 shows the distribution of DCCO and BRAC nests within the Western Analysis Region for four selected aerial photography observation periods (April 28, June 4, July 1 and August 3). Table 4-1 and Figure 4-1 summarize the counts and metrics for the Western Analysis Region for each study week in the study period.

26 Table 4-1. DCCO and BRAC Totals for the Western Analysis Region 4-2 Metric Number on colony (as observed on imagery) Number nesting on colony (computed as 2x # nests) Nesting pairs on colony (# nests observed) Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 10 Week 11 Week 13 Week 15 (4/28/2015) (5/4/2015) (5/11/2015) (5/18/2015) (5/27/2015) (6/4/2015) (7/1/2015) (7/6/2015) (7/23/2015) (8/3/2015) ,403 1,406 1,252 10,129 7,928 4,453 2,462 9, , ,784 8,822 7,790 2,210 1, , ,892 4,411 3,895 1, Acres occupied by nests n/a Areal nest density (# nests / acre) Number on colony (as observed on imagery) Number nesting on colony (computed as 2x # nests) Nesting pairs on colony (# nests observed) Western Subcolony Analysis Region DCCO 3,152 n/a 3,085 n/a 2,232 2,483 3,816 3,541 2,278 2,079 BRAC ,037 3,567 3,564 4,625 1,871 3, ,978 1,968 2,242 1,852 1, , Acres occupied by nests n/a Areal nest density (# nests / acre) n/a 1,683 2,604 2,893 3,106 3,234 3,160 Figure 4-1. DCCO and BRAC Totals Bar Graph, Western Analysis Region

27 4.2 Central Analysis Region 4-3 The Central Analysis Region contains Monitoring Regions F-N. Nests in these monitoring regions were well established prior to the Week 1 (week of April 27) field and aerial photography observations. Good visibility from the blinds for most of the monitoring regions in this area allowed for relatively detailed nesting chronology statistics. Unlike the Western Analysis Region, the distribution and density of DCCO nesting attempts in the Central Analysis Region was relatively stable from week to week. Nesting chronology statistics were compiled for this analysis region and presented in Section 2. These statistics are not necessarily representative of the nesting activity on ESI as a whole. Nesting in The Western Analysis Region was delayed compared to the Central Analysis Region. Activity in the Eastern Analysis Region initially appeared similar to the Central Analysis Region. However, since the nests in that region were not visible from the blinds, it could not be determined how many chicks were produced. Based on reported management activities in the Eastern Analysis Region in Week 5, very few chicks were likely produced in that region. Map 04 shows the distribution of DCCO and BRAC nests within the Central Analysis Region for four selected aerial photography observation periods (April 28, June 4, July 1 and August 3). Table 4-2 and Figure 4-2 summarize the Central Analysis Region counts and metrics for each study week. Table 4-2. DCCO and BRAC Totals for the Central Analysis Region Central Subcolony Analysis Region Metric Number on colony (as observed on imagery) Number nesting on colony (computed as 2x # nests) Nesting pairs on colony (# nests observed) Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 10 Week 11 Week 13 Week 15 (4/28/2015) (5/4/2015) (5/11/2015) (5/18/2015) (5/27/2015) (6/4/2015) (7/1/2015) (7/6/2015) (7/23/2015) (8/3/2015) DCCO 7,749 11,197 11,712 9,088 8,011 9,512 16,356 14,437 10,654 9,687 14,058 16,086 14,832 13,148 12,758 12,864 13,146 12,938 7,988 5,326 7,029 8,043 7,416 6,574 6,379 6,432 6,573 6,469 3,994 2,663 Acres occupied by nests n/a Areal nest density (# nests / acre) Number on colony (as observed on imagery) Number nesting on colony (computed as 2x # nests) Nesting pairs on colony (# nests observed) 4,173 4,706 4,788 n/a 4,481 4,479 4,386 4,375 2,842 2,174 BRAC ,265 1,418 1, ,395 1,355 1,506 1, ,530 1,786 1,746 1,038 2,026 1,900 1,638 1, , , Acres occupied by nests n/a Areal nest density (# nests / acre) 0 1,287 3,038 n/a 3,711 3,251 4,392 4,109 3,729 3,537

28 4-4 Figure 4-2. DCCO and BRAC Totals Bar Graph, Central Analysis Region 4.3 Eastern Analysis Region The Eastern Analysis Region consists of Monitoring Region O. This area was not used by cormorants in past years due to hazing. This year nests were well established throughout the study period. Blind 4 had good visibility of this region, but could not be accessed without disturbing nesting cormorants. DCCO nests were observed in this region during the Week 1 (week of April 27) field and aerial photography observations. The number of nests increased significantly in Week 2 and remained fairly constant until declining in Week 6. A population of BRAC was observed nesting in this Region between Weeks 3 and 6; all BRAC had left by Week 10 and no BRAC were observed in Week 10 or subsequent weeks. In Week 15 a large number of loafing and roosting DCCO were observed in this region and along the shoreline continuing east to areas occupied by pelicans and other bird species. Only minimal field observations could be made for this region due to limited visibility from the remaining blinds that could be accessed.nesting chronology statistics could not be compiled for this analysis region. Nesting chronology appeared to follow a similar timeline as the Central Analysis Region, until management activities occurred on May 23. Map 05 shows the distribution of DCCO and BRAC nests within the Eastern Analysis Region for four selected aerial photography observation periods (April 28, June 4, July 1 and August 3). Table 4-3 and Figure 4-3 summarize the counts and metrics for the Eastern Analysis Region for each study week in the study period.

29 4-5 Table 4-3. DCCO and BRAC Totals for the Eastern Analysis Region Metric Number on colony (as observed on imagery) Number nesting on colony (computed as 2x # nests) Nesting pairs on colony (# nests observed) Eastern Subcolony Analysis Region Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 10 Week 11 Week 13 Week 15 (4/28/2015) (5/4/2015) (5/11/2015) (5/18/2015) (5/27/2015) (6/4/2015) (7/1/2015) (7/6/2015) (7/23/2015) (8/3/2015) DCCO 2,899 3,962 5,263 3,485 3,442 2,739 4,933 4,574 2,133 8,885 1,840 4,958 5,194 4,802 4,858 2,572 2,332 2,324 2,142 2, ,479 2,597 2,401 2,429 1,286 1,166 1,162 1,071 1,052 Acres occupied by nests n/a Areal nest density (# nests / acre) Number on colony (as observed on imagery) Number nesting on colony (computed as 2x # nests) Nesting pairs on colony (# nests observed) 3,748 4,421 3,960 n/a 4,682 3,860 3,708 3,748 3,513 3,461 BRAC Acres occupied by nests n/a Areal nest density (# nests / acre) n/a n/a 2,956 n/a 3,626 2,479 n/a n/a n/a n/a Figure 4-3. DCCO and BRAC Totals Bar Graph, Eastern Analysis Region