Comments on the Cape Wind Energy DEIS-DEIR: Assessment of potential effects on birds.

USACE NAE-20040338-1

 

Submitted by Ian C.T. Nisbet, Ph.D.

150 Alder Lane, North Falmouth, MA 02556

[email protected]

 

Submitted 29 January, 2005.

 

 

1.  Summary Comments

 

            These comments are limited to parts of the DEIS-DEIR that address potential effects of the project on birds: specifically, Section 5.7, Appendices 5.7-A through 5.7-N, and Sections 3.4.3.2.1 (Alternatives Analysis) and 6.3.3.4 (Post-Construction Monitoring).  I have reviewed descriptive material in Sections 3.0 and 4.0 to the extent necessary to evaluate the cited sections on potential impacts on birds.

 

            I am a professional environmental scientist who has studied bird movements around Nantucket Sound since 1958.  I have been a member of the Recovery Team for the endangered Roseate Tern (Northeastern Population) since 1988.  I submitted detailed comments to USACE on the scoping process for the avian studies in April 2002, and I submitted comments on the agency draft of the DEIS in August 2004.  Otherwise, I have not taken part in the studies or in any reviews or assessments.  I have discussed issues with several government agencies and parties who have taken positions on the proposal, but I have not endorsed any of these positions nor taken any position myself.  These comments are made on my own behalf as an independent expert and are not made on behalf of the Recovery Team or any other agency or institution.

 

            Although the facility is the first in what is likely to be a large number of offshore wind energy projects in the USA, the DEIS-DEIR does not consider cumulative impacts.  Because of the potential importance of offshore wind energy projects and because they pose similar and generalizable risks to birds and to other environmental resources, a Programmatic Environmental Impact Statement should be completed before evaluating or permitting any single project.

 

            The avian studies reported in the DEI-DEIR do not meet the minimal standards set out by myself in April 2002 or by the U.S. Fish and Wildlife Service (USFWS) and other professional ornithologists.  The aerial and boat surveys were well conducted and provided useful information on the distribution of terns, sea ducks, and other waterbirds in Nantucket Sound.  However, they were conducted for only two years and did not provide information on the circumstances (evening and morning movements, bad weather, etc.) in which these birds are most at risk.  The radar surveys were poorly designed, were conducted for only a few weeks in one year, and were totally inadequate to assess risks.  Virtually no information was generated or cited on the migrations of seabirds through Nantucket Sound.  The data presented on bird “traffic rates”, directions of flight and heights of flight are subject to considerable bias and should be recalculated.  The estimates presented of the potential numbers of birds that might be killed by collision with the turbines are largely fanciful and almost totally baseless. 

 

            The DEIS-DEIR states that the offshore structures will be “bird-proofed”, but does not give details.  Depending on how the “bird-proofing” will be achieved, it may present significant hazards to birds. 

 

            The emphasis in the DEIS-DEIR on evaluating whether bird casualties would be sufficiently large to affect populations is inappropriate, given the fact that state and federal laws afford protection to individual birds.

 

            No new information was generated about bird movements through areas designated as Alternative sites.  The information cited is incomplete and insufficient.  Although evidence outside the DEIS-DEIR suggests that the marine Alternative sites would be more risky to birds than the proposed site and the terrestrial site less risky than the proposed site, the information presented in the DEIS-DEIR provides no basis whatsoever for evaluating alternatives. 

 

            Instead of making its own evaluation of the material generated by the Applicant, the Corps has adopted many of the Applicant’s conclusions, in some cases verbatim.  This not only gives the impression of bias: it means that the Corps has assumed responsibility for the poor design, inadequate execution, and flawed analysis and interpretation of the results. 

 

            At least in regard to risks to birds, the DEIS-DEIR is inadequate as the basis for conclusions.  Any final EIS-EIR will be indefensible without substantial new information and independent analysis.  Although the Applicant and its consultants bear primary responsibility for the inadequacy of these sections, the Corps shares responsibility because it rejected the advice of USFWS and other professionals in the scoping and study design phases.  The DEIS-DEIR should be withdrawn and new studies must be initiated to remedy the deficiencies of the present draft.  Because the proponent and the Corps did not seek review of the study designs or of early results, the opportunity to make mid-course corrections has been lost.  Some of the required studies will take several more years to conduct. 

 

 

2.  Scope of these comments.  

 

            These comments are submitted by Ian C. T. Nisbet, Ph.D.  My qualifications and experience were summarized in my letter dated 26 March, 2002, addressed to Karen M. Adams of USACE during the scoping process for the Cape Wind Energy DEIS-DEIR; that letter is incorporated herein by reference.  Briefly, I have studied the occurrence and movements of birds over and around Nantucket Sound and adjoining areas since 1958.  These studies included radar studies of migration (1958-1968), and studies of terns nesting at several sites: at Bird and Ram Islands in Buzzards Bay (1970-2004), Monomoy and Tern Islands, Chatham (1972-80), Harding’s Beach, Chatham (1975-76), Dead Neck/ Sampson’s Island, Cotuit (1980-81), Penikese Island, Gosnold (1998-99) and Muskeget Island, Nantucket (2000-01).  Thus, I have studied breeding terns all around Nantucket Sound.  I have also studied staging and feeding of terns around Nantucket Sound in August-September, and was the principal author of a paper describing their distribution and behavior (Trull et al. 1999).  I have surveyed several parts of Nantucket Sound by boat or aircraft, but I have not visited Horseshoe Shoals.  I have been a member of the Recovery Team for the Roseate Tern (Northeastern Population) (hereafter, RTRT) since 1988; however, these comments are submitted in my capacity as an individual expert and not as a representative of RTRT.

 

            These comments are limited to the sections of the DEIS-DEIR that address possible risks to birds resulting from the Cape Wind Energy Project: specifically, Section 5.7 and Appendices 5.7-A through -N.  I have reviewed other sections of the DEIS-DEIR, including Section 3.4.3.2.1 (Alternatives Analysis), Section 6.3.3.4 (Post-Construction Monitoring) and descriptive material in Sections 3.0 and 4.0 to the extent necessary to evaluate the cited sections on potential impacts on birds.

 

 

3.  Authorship and Completeness. 

 

            Appendices 5.7-A through 5.7-M are authored by consultants to the Applicant, The relevant parts of the text of the DEIS-DEIR also appear to have been written or drafted by the Applicant or its consultants, and rely entirely on the data and analysis in Appendices 5.7-A through 5.7-M, without evidence of critical review.  Some conclusory statements in the text appear to have been copied verbatim from the Appendices or from early drafts that were prepared by the Applicant.  Although it is appropriate for the field work and initial analysis to be conducted by the Applicant or its consultants, and for the Applicant’s reports to be appended to or referenced in the DEIS-DEIR, the Corps should not rely uncritically on these reports.  The current draft DEIS-DEIR relies so heavily on the studies and assessments by the Applicant that it suggests that the Corps is no more than an agent of the Applicant.  The Corps should conduct an independent, critical review of the data, analysis and conclusions (either itself or through hiring an independent consultant or expert panel), and should make clear what conclusions are its own and in what ways they differ from the Applicant’s.  The final EIS-EIR should be updated to include the latest available information.

 

            Appendix 5.7-H is not only authored by consultants to the Applicant, but is marked “Internal Review Only”.  It is not clear why this uncompleted draft has been included in the DEIS-DEIR, nor why the Corps has relied upon it. 

 

 

4.  Need for a Programmatic EIS.

 

            The proposed project would be the first wind power project to be located in inshore marine waters of the United States, but already many more have been planned or proposed.  These projects pose new and in some respects unique environmental hazards, many of which are generic to the offshore environment and will recur with each project that is proposed.  This DEIS-DEIR addresses several of these generic issues, but does not resolve any of them in an acceptable way.  As I will show in succeeding comments, the Applicant in this proposed project (Cape Wind) has done a very poor job in planning, conducting and reporting studies of the affected environment and the environmental resources at risk.  This is partly because the Corps (itself with no experience of the issues) did a poor job in scoping the studies and (apparently) did a poor job in overseeing them and reviewing the reported results.  The Corps appears to have developed a somewhat adversarial relationship with other government agencies with expertise in marine environmental resources and with individual experts in the scientific community.  Specifically, the Corps did not allow these agencies and experts to comment on its Scoping document, and did not release (or require the Applicant to release) interim reports on the study until the DEIS-DEIR was nearly complete.  The result (as detailed in my comments below) is that parts of this DEIS-DEIR are incomplete and inadequate, so that it cannot be used as the basis for a defensible decision.  Lengthy additional studies will now be needed. 

 

            Many of these deficiencies could have been averted if the Corps had conducted a generic review at the outset, with input from resource agencies and from the scientific community.  This could have identified the resources at greatest risk, the studies needed to assess these risks in specific local circumstances, and appropriate mitigation measures.  This could have been achieved if the Corps had prepared a Programmatic Environmental Impact Statement at the outset of the process.  I recommend that the Corps should now do so, before specifying the additional studies that will need to be carried out in this case. 

 

 

5.  Terminology.

 

            The DEIS-DEIR frequently uses the term “Wind Farm” or “Wind Park” to describe the proposed project, or other groups of turbines used to capture power from the wind.  This proposed project (like others referred to in the DEIS-DEIR) is neither a “Farm” nor a “Park”.  It is an engineering project designed to convert power from the wind into electrical power.  The terms “Wind Farm” or “Wind Park” are misleading metaphors that appear to have been coined to create a favorable impression of the project by referring to it in terms that imply rural or recreational values it will not provide.  I presume that these terms were written into the DEIS-DEIR by the Applicant.  The Corps should not display bias in favor of the project by including these terms in a document that is nominally its own work product.  I suggest using the neutral terms “Wind Power Project” or “Turbine Array”. 

 

            As an agency that is supposed to have expertise in engineering, the Corps should also avoid using the term “Energy” in contexts where the correct term is “Power”. 

 

 

6.  Scope of the Evaluation. 

 

            Appendix 5.7-H states “This Evaluation …. has been prepared in accordance with the USACE Scope for the Cape Wind DEIS … which was developed in consultation with USFWS, MassWildlife, and Massachusetts Audubon Society…”.  The citation of these institutions may be intended to imply that they approved of the USACE Scope.  In fact, although these institutions (and I, as an independent expert) submitted suggestions to the Corps during the scoping process, we were not made aware of the contents of the USACE Scope, and were not given any opportunity to review and comment on the scope or design of the Applicant’s field studies.  As detailed below, the scope and design of the Applicant’s field studies, as described in Appendices 5.7-A through M, ignored a number of suggestions made by these institutions and by myself.  Hence, the assessments are substantially incomplete or deficient in several respects.  This outcome could have been avoided if either the Corps or the Applicant had submitted drafts of the Scope or the study designs for advance comment, or even if they had released draft reports on the first year’s work.  I repeatedly asked the Corps for the opportunity to review and comment on the scope and design of these studies, and I offered to provide constructive comments without fee.  However, the Corps declined this offer and sent me nothing to review until June, 2004, when it sent me an incomplete draft of the evaluation of the Roseate Tern and Piping Plover in Appendix 5.7-H.  Although I sent some comments on this Appendix, I was unable to review it fully because it was incomplete and referred to the other documents that I was asked not to review.  If I had been able to review either the Scoping Document or interim reports at earlier dates, my comments could have been more constructive.

 

            Since receiving the DEIS-DEIR and writing the previous paragraph, I have seen an undated document entitled “Environmental Impact Statement -- Scope of Work” which starts “This is the Corps of Engineers scope of work….”.  Assuming that this is in fact the document referred to as the “USACE Scope” and was issued to the Applicant at an early stage in planning the field studies, it is pathetically vague and inadequate.  It contains only one short paragraph on Avian Impacts, which includes only three short sentences specifying new field studies.  These lay out in very general terms what data should be collected, ignoring many of the recommendations made by myself and by the wildlife agencies.  Although the studies conducted by the Applicant did not meet even the modest requirements set out in this document, the Corps bears some of the responsibility for the inadequacy of the studies because it failed to specify the requirements in sufficient detail, because it failed to incorporate recommendations made by experts, and because it did not make this document available for review. 

 

            The field studies included radar studies (Appendices 5.7-E and J) and aerial and boat surveys (Appendices 5.7-C, F, K, L M and N).  I assess the scope and design of these studies in relation to the specifications in the “USACE Scope”, and also in relation to my recommendations in the scoping process (my letter dated 26 March 2002):

 

     “To assess potential risks to migrating landbirds, fuller and more precise information is needed on numbers, timing, and heights of flight of birds passing through the project area, especially at night during September, October, and early November….

 

     “To assess potential risks to migrating waterbirds, fuller and more precise information is needed on numbers, locations, timing, and heights of flight of birds passing through the project area, especially during evenings and at night during April-May and September-November….

 

     “To assess potential risks to terns (including the federally Endangered Roseate Tern and the state-listed Common Tern), fuller and more precise information is needed on numbers, timing, and heights of flight of birds passing through the project area, especially by day in May-September and in late evenings and early mornings in August and September.

 

     “To assess potential risks to resident and wintering waterbirds, fuller and more precise information is needed on numbers, timing, and heights of flight of birds passing through the project area, especially in evenings and at night, throughout the year….

 

     “To assess potential risks to wintering sea-ducks, fuller and more precise information is needed on their distribution and movements within Nantucket Sound, including heights of flight of any birds that may pass through the project area, especially at night….

 

     “To assess potential risks to birds flying over Nantucket Sound at night, whether migrating or in transit to roosts or nesting areas, information is needed on the extent to which the lights proposed for installation on the project towers would attract birds under different weather conditions, and expose them to risk of collision with the towers or rotating blades….

 

      “A common feature of most of these scenarios is that birds would be primarily at risk when traversing the project area at night.  Hence, studies of the numbers, distribution and movements of birds through the project area during daylight hours will not be sufficient to address these questions.

 

     “In my opinion, it will be essential to study the movements of birds through the project area at night (including late evenings and early mornings), and radar is the only feasible method for obtaining the required information.  Accordingly, I recommend that a comprehensive radar study be required as part of the Environmental Impact Statement for this project. …

 

     “At a minimum, the station should be operated throughout each night (from early evening to mid-morning) during the migration seasons (mid-April to late May and late August to mid-November), and in evenings and mornings at other times of year (daily from late July through August and periodically at other times of year).  One year’s sampling will not be sufficient: I recommend at least a three-year study to ensure that critical but infrequent weather conditions are encountered.

 

“In addition, I recommend that focused radar studies should be conducted during the migration seasons to investigate the response of migrating birds to the tower lights during nights with low overcast, mist, fog, or rain.”   (All emphases in original).

 

Similar or identical recommendations were made by agencies with Trustee responsibility for migratory birds and with expertise in avian risk assessment, including the U.S. Fish and Wildlife Service (USFWS) and the Massachusetts Division of Fisheries and Wildlife (MDFW).

 

            VIRTUALLY NONE of these recommendations was followed in the studies conducted by the Applicant and reported in Appendices to the DEIS-DEIR.  The Corps appears to have negotiated with the Applicant behind closed doors and has made decisions about the scope of studies that would be acceptable, without soliciting comments on these decisions either from experts or the general public.  By including the Applicant’s flawed studies in the DEIS-DEIR and by parroting the Applicant’s conclusions in the text, the Corps is thumbing its nose at expert opinion and is assuming full responsibility for the inadequate design, execution and interpretation of the field studies. 

 

            Although page 13 of Appendix 5.7-H asserts that field data for five years are available, the DEIS-DEIR presents systematic data for the project area for only two years, 2002 and 2003; the surveys in 1991 and 1992 were outside the project area, and the surveys in 2001 were preliminary (pilot) studies only.  Both USFWS (letter dated 8 May 2002) and Dr. Nisbet (letter dated 26 March 2002) advised USACE that “at least” three years’ data would be required; MassWildlife recommended “several years of careful work” (letter dated 20 December 2001).  These recommendations for three years’ data collection were intended to be understood as minima (“at least” three years), rather than maxima; they were intended as compromises between the desirability to monitor for long periods to detect infrequent events and the costs of longer-term studies and of delays in approval and construction.  In response, the Corps merely specified “Information derived from other studies, which provides a three-year baseline data set, should be included if available” (emphasis added).  The phrase “if available” is ambiguous, but evidently encouraged the Applicant to limit its studies to two years; this must have been approved by the Corps in private discussions with the Applicant.    To my knowledge, the Corps has never explained why it would regard two years’ data as sufficient, nor has it offered USFWS, MDFW or myself any opportunity to comment on this decision.  I repeat that at least three years’ data will be required before defensible decisions about risk could be made. 

 

            Appendix 5.7-H (page 13) indicates that a third year’s field data (2004) have been collected by Mass Audubon, but does not refer to these data, even though the DEIS-DEIR was dated November 2004, long after the field studies were completed.  I do not know whether the design or execution of Mass Audubon’s studies will prove sufficient to provide the third year’s data that the wildlife agencies and I regard as minimal.  However, I understand that Mass Audubon’s studies were limited to the summer months (May-September), so they will not help to fill the gaps in information on migrating and wintering birds that occur in other months.  At a minimum, the final EIS-EIR must include the data from Mass Audubon’s 2004 studies, as well as independent evaluation of all the studies. 

 

            The aerial and boat surveys reported in the DEIS-DEIR provide information on the numbers and heights of flights of terns and other waterbirds within the project area during daylight hours in May-September.  However, they provide virtually no information of any kind about movements of waterbirds through the project area in May-September, nor about “movements, timing, and heights of flight of birds passing through the project area ….. in late evenings and early mornings in August and September”.   Likewise, they provide information on the distribution of sea-ducks and other wintering waterbirds during daylight hours in November-March, but virtually no information of any kind about movements of these waterbirds through the project, nor about “numbers, timing, and heights of flight of birds passing through the project area ….. at night.”  Although radar studies were conducted during parts of May and September in one year, only minimal information is presented on the results of these studies, and this does not address any of the information needs listed above except for summary data on bird densities and heights of flight during those months.  The minimal information on these topics that is referred to in the DEIS-DEIR is discussed in the next two sections of these comments. 

 

 

7.  Aerial and Boat Surveys

 

            Aerial and boat surveys are reported in Appendices 5.7-C (July-September 2001), 5.7-D (December 2001 and March-April 2002), 5.7-F (May-August 2002), 5.7-G (September 2002- February 2003), 5.7-K (March-June 2003), 5.7-L (June-August 2003), 5.7-M (September 2003 – February 2004), and 5.7-N (May-September 2002-2003). The surveys in 2001 were reported as pilot studies and are not considered further here.  Some of the remaining surveys overlap the natural divisions in the annual cycle of waterbirds (spring migration, breeding season, autumn migration, winter season).  For a breakdown that is more biologically meaningful, I divide the surveys into three groups:

 

            (a) Breeding season (May-September).  The Applicant reported 17 aerial and 10 boat surveys during this period in 2002 and 2003 (Appendices 5.7-F, -G, -K, -L and -M).  In addition, Mass Audubon reported 9 aerial and 3 boat surveys during this period in 2002, and 3 aerial and 13 boat surveys during this period in 2003 (Appendix 5.7-N).  Most of the birds registered during these surveys were terns, gulls and cormorants.  This period includes spring and autumn migrations for terns, as well as the breeding season, post-breeding dispersal and staging for terns, gulls and cormorants.  As stated above, I regard the two-year coverage as inadequate, although Mass Audubon’s surveys in 2004 will probably fill this gap.  The coverage within years is minimally adequate, given the marked seasonal differences in tern distribution.

 

            Appendix 5.7-F also includes summaries of birds seen from a boat on 13 days in 2002 (9 days on Horseshoe Shoals in May-June and 4 days near Cape Poge in September).  These observations were described as intended to serve as “ground-truthing” for the radar surveys.  However, they were apparently not used for that or any other purpose and hence are of no value for risk assessment.     

 

            (b) Migration periods (April and October-November).  The Applicant reported 11 boat surveys and 5 boat surveys during these periods.  These periods include the arrival and departure of wintering sea-ducks, departure and arrival of Common Terns and some gulls, and passage migration of transient waterbird species such as loons, gannets and cormorants.  For most of these species, larger numbers were recorded during these periods in one or both of the years studied than in the summer or winter periods.  Because most of these birds were in transit through the area during these periods, considerable fluctuations in numbers are expected from day to day and even from hour to hour.  Accordingly, the 11 aerial surveys and 5 boat surveys during two years are seriously inadequate to characterize the numbers of any of these species at risk from the project.

 

            (c) Winter period (November-March).  The Applicant reported 22 aerial surveys but no boat surveys during this period, including 4 surveys during November that have been included in the previous paragraph of these comments.  As these surveys demonstrate, this is the period in which sea-ducks, grebes and alcids occur in largest numbers, as well as important numbers of some species such as loons and gulls that are present in larger numbers during migration.  For these species, the surveys were minimally adequate to characterize the numbers and distribution during daylight hours in the two years covered.  However, they were inadequate to characterize the late evening and early morning movements of Long-tailed Ducks and other species that were identified in my comments and in those of the wildlife agencies as critical for risk evaluation.  Also, despite the arguments in Appendix 5.7-B, two years’ data are not sufficient to characterize the variability in numbers of birds within the project area.   

 

            Limitations of aerial and boat surveys.  Subject to the limitations discussed below, the Applicant’s aerial and boat surveys were well designed and appear to have been well conducted and reported.  Apart from the fact that the quantitative information on the occurrence of waterbirds in the study area is based on only 45 aerial and 28 boat surveys in two years (see above), these field studies have two important limitations.  The first, acknowledged by the Applicant, is that they were restricted to daylight hours (0500-2000) and to good weather.  In spite of the statements by myself and others that observations during bad weather and in early mornings and late evenings would be essential, virtually no information was obtained about the occurrence or movements of waterbirds in these circumstances.

 

            The second limitation, not acknowledged either by the Applicant or by the Corps, is that the information presented is on the occurrence and density of waterbirds present on Horseshoe Shoals and other parts of Nantucket Sound, not their movements through the project area.  In the winter months, birds such as eiders, scoters, grebes and alcids that frequented Horseshoe Shoals probably spent much of their time on the water and would only have been at risk from collision with turbine blades during their arrival, departure, and occasional flights within the area.  However, the field studies conducted by the Applicant and by Mass Audubon indicate that these species at other times of the year, and most other birds throughout the year, probably spent little time resting or feeding within Horseshoe Shoals.  Instead, most of the observations probably referred to birds on transit through the area; each transit by each individual bird would have placed it at risk.  For example, the observation of a single cormorant or gannet in the air during an aerial transect in April or October might have represented more risk than that of a flock of hundreds of scoters on the water in December.  Because the Applicant’s field surveys did not attempt to measure the numbers of birds flying through the project area at any time of year, they are seriously deficient as the basis for any risk assessments (see further discussion below under species and groups of species). 

 

           

8. Radar Studies. 

 

            (a)  Scope.  The Applicant also reported observations of flying birds using two radars during the periods 8 May – 7 June and 3-30 September 2002, on a “jack-up” barge on Horseshoe Shoals and on a nearby land site at Cape Poge, Martha’s Vineyard, respectively (Appendices 5.7-E and J).  These periods of operation fell far short of the multi-year studies called for as minimal in my comments during the scoping period: “each night (from early evening to mid-morning) during the migration seasons (mid-April to late May and late August to mid-November), and in evenings and mornings at other times of year (daily from late July through August and periodically at other times of year)”.  In particular, they overlapped by only 13 days with the mid-July to mid-September period identified as the period of greatest potential risk to Roseate Terns by myself and Dr. Hatch, and did not cover at all the period from mid-July to mid-August when juvenile terns are learning to fly and to forage.   Even within the 60-day periods of operation, equipment malfunctions limited the data obtained to only 59% of available hours (26 days and 26 nights) in spring and 60% of available hours (24 days and 25 nights) in autumn.  Presenting these meager observations as though they were representative of the movements of birds through the area is unwarranted.   

 

            Appendix 5.7-F also includes summaries of birds seen from a boat on 13 days in 2002.  These observations were intended to serve as “ground-truthing” for the radar surveys.  However, they were apparently not used for that purpose and are not mentioned in the reports on the radar studies, except for two tables which were added at the end of Appendix 5.7-J.  No useful findings are mentioned in the reports on the boat surveys, either.

 

            The USACE Scope specified the following: “Data on use throughout the year, especially through November for migratory species, and under a range of conditions should be collected.  Data collection methods should include remote sensing through radar and direct observations through aerial reconnaissance and boat-based surveys.  Data gathered through radar should be validated with direct observations” (emphases added).  The Applicant’s studies did not meet the specifications italicized, and hence are unacceptable even by the Corps’ lax standards. 

 

            (b) Classification of Targets.  Without any serious attempt at “ground-truthing”, the radar observations are of little use for evaluating numbers or species of birds passing through the project area.  Most of the data presented are numbers of “targets” detected by the radars, with crude breakdowns by season, time of day, ground speed, direction of flight and magnitude of target. 

 

            Data were obtained using two radars: TracScan to plot the distribution and tracks of targets in the horizontal plane, and VerCat to measure heights of flight.  Targets detected by the TracScan radar were divided into two categories based on speed: “Slow” (< 27 knots), or “Fast” (> 27 knots).  These were ground speeds (Appendix 5.7-E, p. 5).  However, ground speeds are of little or no value for classifying flying birds because they depend very strongly on wind velocity.  For example, a small bird with air speed of 20 knots (34 km/hr) would have ground speed of only 10 knots (17 km/hr) when flying into a head wind of 10 knots, but ground speed of 30 knots (52 km/hr) when flying with a tail wind of 10 knots.  Thus, the ground speed of a target gives no information whatsoever about the size or other characteristics of the birds(s) unless the air speed is calculated.  All the data in Appendices 5.7-E and 5.7-J must be recalculated in terms of air speed before they will have any value for biological inferences.  When this is done, the targets should be classified into more than two categories: “Fast” and “Slow” provide little meaningful information about the nature and composition of targets. 

 

            Targets detected by the VerCat radar were divided into four categories based on inferred size: “Small” (< 80 g), “Medium” (80-800 g), “Large” (> 800 g) and “Flock” (Significantly > 800 g).  This classification is of limited value given the expected range of targets, from small landbirds flying singly (10-15 g) to flocks of cormorants or other large birds (10-100 kg).  If targets could be classified reliably according to total mass, then many more categories should have been distinguished.  If (as I suspect), targets cannot be classified reliably on the basis of mass or other measures of body size or flock size, then this fact should have been acknowledged and the data should have been reported accordingly.  Because the strength of a radar echo (and hence the inferred size of a target) varies with the 4th power of distance, it would have been more meaningful to report and classify targets according to their radar cross-sections than to have attempted to classify them by mass. 

 

            (c) Estimates of Target Density.  Table 3 in Appendix 5.7-E presents seasonal averages of “Tracks per hour”.  This metric is stated to be “roughly equal to birds tracked within the 8 nautical mile (14.8 km) area within which the TracScan radar registered birds (out to about a 4 nautical miles [7.4 km] from the radar” (emphasis added).  It appears to have been assumed (a) that targets can be equated with birds; and (b) that all birds within the 7.4 km circle were detected.  The first assumption is known to be wrong: many of the birds detected and reported in the aerial and boat surveys were in flocks, each of which would have been detected as a single target.  The second assumption is very unlikely to be true, for three reasons:

 

            (i) the power density in the beam declines with the angle off-axis, so that targets at angles between 3^o and 12.5^o above the horizontal are progressively less likely to be detected; this would limit the detection of small targets at close range (3^o and 12.5^o above the horizontal correspond to altitudes of 52 and 222 m at a range of 1 km); this would also be true for birds below the beam axis (e.g., all birds flying at rotor height at ranges greater than about 2.8 km);

 

            (ii) targets above the upper margin of the radar beam (12.5^o above horizontal) would not be detected at all;

 

            (iii) even close to the axis of the beam, distant targets would not be detected because they return echoes too weak to be registered.

 

            To relate the numbers of targets detected to the numbers of targets in the air as functions of radar cross-section, altitude and range, would require calibrated information on the performance of the radar equipment, including polar diagrams of the dependence of the minimum detectable radar cross-section on the off-axis angle.  Appendices 5.7-E and –J present no useful performance information except the statement that “TracScan’s operational detection range for large birds (i.e. geese) is over 10 nmi (18.5 km)”.  Assuming that detection range scales approximately as mass^0.25, the “operational detection range” for a 10-g bird would be about 4.1 km, and that for a 100-g bird would be about 7.4 km.  Detection ranges would be smaller for targets off-axis.  This rough calculation suggests that many small and some medium-sized birds would have been missed in the outer parts of the scanned circle. 

 

            For all these reasons, the numbers listed as “Tracks per hour” in Table 3 are likely to be substantial underestimates.  These numbers cannot be used as estimates of the numbers of birds passing per hour without correction for the geometrical biases pointed out above, and factoring in data on average flock size.

 

            (d) Calculation of Traffic Rates.  Appendix 5.7-E also presents calculations of “traffic rates”, defined as “numbers of radar tracks per hour per kilometer of front.  These numbers appear to have been calculated using the formula:

 

            Traffic rate (targets/km/hr)  =  tracks per hour / diameter of scanned circle (km).

 

For example, for “Slow” Tracks in spring, 663 Tracks per hour in Table 3 was divided by 14.8 km to obtain the traffic rate of 44.8 targets/km/hr listed in Table 4.  This would be correct if all birds passing through the circle in each hour were detected.  However, because many distant birds are likely to have been missed (see previous section), the effective length of the “front” putatively crossed by the birds is likely to have been substantially less than 14.8 km.

 

            (e) Heights of flight.  Summary statistics on the altitudes at which targets were detected by the VerCat radar are presented in Attachments 3 and 4 of Appendix 5.7-E.  Because the data presented are simple counts of targets falling into various categories, these data evidently have not been converted to target densities or otherwise corrected for differential detectability.  However, because the radar beam spreads linearly, birds flying low are less likely to be detected than birds flying at moderate altitudes; birds flying very  high are also less likely to be detected because the “typical operation range” is only 1,375 m downrange and 2,750 m vertically. 

 

            The limited technical information given in Appendix 5.7-J defies rational interpretation.  Page 6 states that a minimum of 3 registrations were required to record a “track”.  A bird flying at 40 km/hr (ground speed) will travel 56 m in 2 revolutions of the VerCat beam (5 sec).  At a nominal 1^o beam width, the beam does not reach 56 m width until 3,180 m from the source, far beyond the typical operating range.  Thus, according to information given, the VerCat radar could not detect any tracks of birds passing overhead except for birds flying in the plane swept out by the radar beam (stated to be east-west).  At oblique angles, detectability will be higher.  However, even at a 15^o angle, the beam does not reach 56 m width until 3,070 m horizontally from the source (820 m altitude); again, no birds would be detected except those flying in or near the east-west plane.  At all angles, the VerCat radar will detect few birds flying north-south and far more birds flying east-west than NW-SE or NE-SW; at all angles, it will detect far more birds flying high than flying low, up to the heights where high-flying birds are lost because they are out of range.  Because of these geometric properties of the VerCat radar, the data presented in Attachments 3 and 4 cannot be used to make inferences about the proportions or numbers of birds flying in the rotor-swept-zone, unless and until they have been corrected for the large variations in detectability according to target size, height, and ground speed.

 

            (f)  Directions of flight.  Data on directions of flight are summarized into 16 compass-rose diagrams, representing various combinations of spring/fall, day/night, fast/slow targets, and clear/nonclear weather.  These broad classifications are of little use for interpretation or assessments of possible risks.  Compounding this problem, the data are reported as headings, calculated from the observed tracks using unspecified data on wind velocities.  Because the targets were divided into two broad categories (faster or slower than 27 knots), the calculated headings are subject to substantial error.  Tracks are more relevant to risk assessment than headings, and should have been reported.

 

            (g) Spatial gradients.  All data from the TracScan radar are reported as composites for all parts of the 7.4 km circle that was nominally covered.  Hence, all information about spatial gradients in bird numbers has been lost.  This is of particular concern for the birds tracked from Cape Poge in September, many of which would have been overland.  Birds flying over land have little or no relevance for risk assessment for the marine site and should have been omitted from the summaries.  Even the data for birds flying over water should have been analyzed to identify which birds were likely to have flown through the project area.  Most seabirds avoid flying over land, so it is likely that the conformation of the land areas around Nantucket Sound would have led to marked differences in flight patterns.  In spring, for example, it is to be expected that seabirds migrating northeastwards would be diverted east by the south shore of Martha’s Vineyard, and hence would be concentrated into a stream running northeast from the southeast point (Wasque).  Depending on the flight directions of birds in this stream (NNE, NE or ENE), they may or may not cross the project area.  This is important information and should have been reported in a way that is useful for risk assessment. 

 

           

9.  Discrepancies between radar and aerial survey data. 

 

            The radar data were not interpreted and were not compared with the visual observations in any way, despite the reported intention to “ground-truth” the radar data and despite that Corps’ specification that “Data gathered through radar should be validated with direct observations”.  For example, terns are probably the seabird species at greatest risk in September.  During 24 days of daytime operation in September, a total of 174,113 “Slow” and 128,861 “Fast” targets were tracked by the TracScan radar.  These were calculated to correspond to average traffic rates of 46.7 and 34.6 targets per km of front per hour, respectively.  Using the VerCat radar, 76% of daytime targets in September were above rotor height, 24% were in the rotor-swept zone (23-126 m above sea level), and only 0.3% were below rotor height.              There is a complete disconnect between the visual record of 356 individual birds within the study area on 25 September 2002, mostly cormorants, seaducks, gulls and terns flying at altitudes less than 12 m above the water (Appendix 5.7-F, Table A), and the 11,156 targets detected by the radar on that day, mostly small and medium-sized targets flying higher than 23 m (Appendix 5.7-J, Table 12).  The Applicant’s reports made no attempt to relate the two or to explain the discrepancy.  Obviously, the radar data showing “targets” flying within the rotor-swept zone are directly relevant to risk assessment, but the Applicant’s Evaluation dismissed the radar data in one sentence (Appendix 5.7-H, p. 22). 

 

            The Applicant’s reports acknowledge that the radars were not configured to detect birds flying close to the water surface, but the failure of the aerial and boat surveys to detect the targets higher than 23 m requires explanation.  The most likely explanation is that the observers in aircraft or boats simply missed the high-flying birds.  Except for large, dark birds such as cormorants, it is difficult to see high-flying birds against a bright sky, especially from a moving boat.  The aerial surveys were conducted by looking downward between calibrated markers from an aircraft flying at 75 m altitude.  In these circumstances, no birds flying higher than 75 m would have been observed, and birds flying between 23 and 75 m would have been difficult to detect (a) because the observers’ attention was directed to the sea surface, (b) because the width of the transect declined linearly with distance below the aircraft, and (c) because they would be seen more fleetingly than birds further below.  For these reasons, the data on heights of flight reported in the aerial and boat surveys should not be used to infer lack of risk.

 

 

10.  Roseate and Common Terns.

 

            Potential risks to Roseate and Common Terns are addressed in Appendices 5.7-H and 5.7-I, respectively.  The relevant field studies were those carried out during the approximately 170-day period when these terns are present in the Nantucket Sound area (20 April – 30 September for Common Terns, 1 May – 16 September for Roseate Terns).  The Applicant reported 7 aerial and 9 boat surveys during this period in 2002 (Appendices 5.7-D and -F), and 11 aerial and 4 boat surveys during this period in 2003 (Appendices 5.7-K, L and M).  In addition, Mass Audubon reported 9 aerial and 3 boat surveys during this period in 2002, and 3 aerial and 13 boat surveys during this period in 2003 (Appendix 5.7-N).  For reasons given above, the two-year coverage is inadequate, although Mass Audubon’s surveys in 2004 may fill this gap.  The coverage within years is minimally adequate, given the marked seasonal differences in tern distribution.

 

            (a) Results of aerial and boat surveys. The reported results of aerial and boat surveys indicated that (a) most terns observed in the systematic surveys were close to the shorelines of either Cape Cod or Monomoy, Nantucket, Tuckernuck or Muskeget Islands; (b) substantial numbers of terns were also observed over shoals immediately to the north of the project area of Horseshoe Shoals (HSS); (c) terns were observed regularly over HSS, but in much smaller numbers; and (d) observed numbers were larger in May and July-September than in June.  Mass Audubon’s field studies gave generally concordant results, although their aerial surveys in 2002 were not designed to yield quantitative estimates of bird density. 

 

            The Applicant’s aerial surveys were designed to yield quantitative estimates of bird density through use of calibrated transects.  Densities of terns (pooling Roseate and Common Terns) were reported as totals over surveys both inside and outside the period when Roseate Terns were present.  Reanalyzing the data to yield mean densities over all the surveys that were conducted within this period, the average densities on HSS (Alternative 1) were about 0.77 terns km-2 in May-August 2002 (6 surveys), 3.95 terns km-2 in May – early June 2003 (2 surveys), and 0.10 terns km-2 from mid June – mid September 2003 (7 surveys).  Mass Audubon’s boat surveys similarly showed much larger numbers in May 2003 than later in that year.  The wide variations in density within and between years illustrate the need to continue the surveys for at least one more year.  The densities cited above correspond* to average numbers of about 85, 435 and 11 terns, respectively, within the project area of 110 km².

 

            Estimating the relative numbers of Roseate and Common Terns is made difficult by the fact that only about 40% of terns could be identified to species.  Restricting attention to terns observed over the study area, the proportion of Roseates among the terns so identified was 209/2,508 (8%) from May – August 2002, 94/397 (24%) in May – early June 2003, and 23/185 (12%) from mid June – mid September 2003. 

 

            The Applicant’s observations during aerial and boat surveys indicated that almost all (>98%) of the terns observed in the study area were below rotor height (23-127 m), although two flocks of terns were observed at rotor height outside the study area and 47 terns were seen at rotor height within the study area in 2002 (Table 10 in Appendix 5.7-F).  However, the reliability of these findings is questionable, for reasons stated in Section 9 of these comments.  

 

            (b) Limitations of field studies.  Apart from the fact that the quantitative information on the occurrence of terns in the study area is based on only 21 aerial and 13 boat surveys in two years (see above), these field studies have two important limitations.  The first, acknowledged by the Applicant, is that they were restricted to daylight hours (0500-2000) and to good weather.  In spite of the statements by myself and others that observations during bad weather and in early mornings and late evenings would be essential, virtually no information was obtained about the occurrence or movements of terns in these circumstances.

 

            The second limitation, not acknowledged by the Applicant or by the Corps, is that the information presented is on the occurrence and density of terns present on Horseshoe Shoals, not their movements through the project area.  Unlike sea-ducks, terns infrequently rest on the water; when observed at sea, they are usually flying, either while foraging or en route to somewhere else.  Terns are at little risk of collision with turbine rotors when they are foraging, because they are then usually within 10-15 m of the sea surface; the main risk of collision is when they are commuting, when they sometimes fly higher. The only information that would throw light on this is the report by Mass Audubon (Appendix 5.7-N, Table 2), that 62% of the Common Terns and 100% of the Roseate Terns observed on HSS in 2003 were traveling.  Depending on the direction of flight and the wind velocity, a tern would travel through the entire project area in about 10-15 minutes.  Hence, even if high-flying terns had not been overlooked (see below), the estimated number of 435 terns in the area at any one time in May-early June 2003 would correspond to 1,700-2,500 transits per hour.  If these numbers were maintained throughout a 14-hour day and throughout the period from early May to early June, this would correspond to roughly 800,000 – 1,200,000 transits (24% Roseates).  These estimates are derived by extrapolation from a very small data set and are obviously extremely rough, but since the observations were designed to yield systematic samples, it should be possible for the Applicant or the Corps to derive statistically-based estimates and ranges.  The point made here is that a “small” number of birds observed on a set of transects through the area does not mean that the number of birds at risk is small, unless statistically valid methods of estimation are used.

 

            These two limitations are combined when attempting to consider the commuting of terns to and from their daytime staging areas and nocturnal roosts around Nantucket Sound in July-September.  These commuting flights were identified as the principal risk to terns from the project both by myself (my letter of 26 March 2002) and by the Applicant’s consultant Dr. Hatch (Appendix 5.7-C).  However, the field surveys were not designed to yield information on these flights and did not do so.  The reports include only a few anecdotal observations that are relevant.

 

            The main nocturnal roost of Common and Roseate Terns, at least in late August and early-mid September, is at South Beach in Chatham (Trull et al. 1999).  Most terns arrive at this roost late in the evening (some after dark); I know of no observations of the times of departure.  At roosts in winter quarters in Brazil, almost all Roseate Terns arrive after dark and depart before first light (H. Hays, unpubl. obs.) and I think it likely that behavior is similar at South Beach.  Both Dr. Hatch (Appendix 5.7-H, p. 16) and I (unpubl. obs.) have seen and heard flocks of Common and Roseate Terns arriving at South Beach after sunset, descending from heights of 37-60 m (or higher).  The Applicant’s Evaluation of the Roseate Tern (Appendix 5.7-H, p. 16) argues that Roseate Terns would only fly high in this way when crossing land, but this argument is unconvincing without more data on birds commuting over water.  The Evaluation claims (p. 16) that terns of both species flying towards their overnight roost at Fernando’s Fetch on 1 and 15 August 2002 were below 9 m elevation.  However, the original report (Appendix 5.7-F, p. 10) stated that these terns were flying higher than 18 m.  I consider it likely that terns habitually fly high on commuting flights, especially after sunset and before sunrise, and especially when they are flying downwind.  Specific evidence to the contrary would be required before this source of risk could be discounted.

 

            The project area at HSS does not lie directly between the roosting area at South Beach and either known daytime resting areas (Trull et al. 1999) or major feeding areas (as documented in the surveys by the Applicant or Mass Audubon).  However, neither Trull et al. (1999), nor the Applicant, nor Mass Audubon surveyed the parts of Martha’s Vineyard from which terns would fly through the project area on a direct course to South Beach.  Systematic observations will be required before it can be concluded that numbers of terns making commuting flights from these areas are low.  These observations could easily have been made already if either I or the wildlife agencies had been given the opportunity to review the design of the field studies.

 

            Of greater immediate concern is the tern roost at “Fernando’s Fetch” (Appendix 5.7- F, p. 10).  The project area at HSS lies directly between this roost and major feeding areas to the north and northeast, and indeed the Applicant reported seeing terns flying (some at over 18 m height) through the southern part of the study area towards Fernando’s Fetch after dark on 15 August 2002.  This anecdote resulted from a single unplanned observation; systematic observations and assessments of risk are needed.  Fernando’s Fetch is a recently-formed island, and if it continues to grow in size, I predict that it will become more important as a tern roost and might even displace South Beach in importance, because it is more secure from nocturnal human disturbance and nocturnal predation.  If so, risks to commuting terns would increase. 

 

            (c) Radar studies.   The radar studies conducted in September overlapped by only 13 days with the mid-July-mid-September period identified as the period of greatest potential risk by myself and Dr. Hatch, and did not cover at all the period from mid-July to mid-August when juvenile Roseate Terns are learning to fly and to forage.  Although the May-June radar studies were at the time of year when the greatest numbers of terns were seen on HSS in 2003, there were virtually no field surveys on HSS at this period in 2002 (one aerial survey on 22 May and no boat surveys).  Without any attempt at “ground-truthing”, the radar observations are of little use for evaluating numbers of terns passing through the project area.  Most of the data presented are numbers of “targets” detected by the radars, with crude breakdown by season, time of day, ground speed, direction of flight and magnitude of target.  Single terns or small groups would have fallen into the “Medium” target category (80-800 g), but flocks of more than 6 birds would have fallen into the “Large” or “Flock” categories (> 800 g).  Foraging terns would probably have fallen into the category of “Slow” targets (< 27 knots), but commuting terns would probably have fallen into the “Fast” category, at least when flying downwind.  Average traffic rates of 46.7 and 34.6 targets per km of front per hour were reported for “Fast” and “Slow” targets, respectively.  Using the VerCat radar, 24% of daytime targets in September were in the rotor-swept zone (23-126 m above sea level).  These data are not specific to terns, but are consistent with substantial traffic of tern-sized targets through the rotor-swept zone. 

 

            (d) Evaluation of Potential Collision Losses.  The Applicant’s “Evaluation“ of the Roseate Tern (Appendix 5.7-H) summarized the field studies referred to above, but did not mention the radar studies except to dismiss them in one sentence (p. 22).  It then (p. 23) proceeded to discard all the data from the field studies as well (!), and instead based its risk assessment (pp. 23-24) on a report on Common Terns found dead at a terrestrial turbine array in Belgium.  The Applicant’s risk assessment referenced an unpublished report (Everaert et al. 2002) that is not available for review, as well as additional unpublished information from the first author (J. Everaert).  After reviewing the published paper on this study (Everaert 2004), I believe that it could provide a partial basis for risk assessment for terns at HSS (specifically, because it gives an empirical value for the proportion of Common Terns flying in the rotor-swept-zone at the Belgian site that were killed).  However, use of this value would require better site-specific data on the numbers of terns flying in the rotor-swept-zone at HSS, as well as scaling factors to take account of the greater number and larger size of the rotors at HSS.  The Applicant’s risk assessment (Appendix 5.7-H, p. 24) includes scaling factors, but its assumptions about the numbers of birds at risk ignore its own field studies and appear to be based largely on guesswork.  The Applicant’s Biological Review of the Common Tern (Appendix 5.7-I) cited some of the same data, but concluded that “No biologically realistic or precise prediction of number killed is possible with the available data.”  The text of the DEIS-DEIR (p. 5-135) similarly makes numerical estimates of possible deaths of Roseate Terns, but not of Common Terns.  It is not clear why the Applicant (and by adoption, the Corps) is willing to make numerical estimates for one species but not for the other.  The assessments of risk in the cited sections of the DEIS-DEIR are unsound and unacceptable.  I recommend that risk assessment should be based on the site-specific data for Roseate and Common Terns in Nantucket Sound presented in other Appendices, combined with an explicit model of collision risks.  .

 

Such an assessment will require:

 

    (i) reconciliation of the visual and radar studies.  This will require, at a minimum, further analysis of the “ground-truthing” data collected in 2002 and careful collation of these data with simultaneous radar echoes.  It will also require re-analysis of the 2002 radar data to focus on specific times of day and types of echo that might yield information on terns.  It will probably require additional radar studies, both to correct defects in the 2002 surveys and to expand coverage to additional months and additional years;

 

    (ii) additional field surveys to collect data on birds arriving at and departing from the known roosting aggregations;

 

    (iii) either focused field studies to investigate the behavior of terns in bad weather at lighted structures at sea, or a search for more relevant literature;

 

    (iv) de novo modeling of collision risks (the volume-based model outlined on p. 23 is unsound). 

 

            (e) Other Hazards Potentially Posed by the Project.  Section 4.2.1.2 of Appendix 5.7-H considers potential effects of project infrastructure used for perching/roosting.  The statement (p. 19) that migrating Peregrines “typically travel along shorelines” is incorrect (it is well known that they also travel over the sea, often stopping over on offshore islands, rocks, ships, etc.).  It is likely that there will be little temporal overlap between migrating Peregrines and pre-migratory Roseate Terns in Nantucket Sound.  However, there will be some overlap between migrant Peregrines and pre-migratory Common Terns in late September and early October.  At this time, it is very likely that Peregrines will perch on the project infrastructure and prey on passing terns.  This issue needs to be addressed properly and an assessment should be added to Section 5.7.3.4.

 

            I also have concerns about the intended use of “avian deterrent systems” on the WTGs and ESP.  The system described for the WTGs, with a chain-link fence, a solid panel, and a stainless steel wire on top of the railing, is likely to be effective in discouraging Roseate Terns from perching on the decks.  However, if these systems deteriorate under marine conditions (e.g., by corrosion of the fence or breakage of the wires) they might allow terns (and other birds) to perch on the decks and might pose hazards during take-off or landing.  More information needs to be given about scheduled inspections and maintenance of these deterrent systems.

 

            The ESP is even more problematic.  The heliport deck will be large (60 m x 30 m) with a perimeter railing.  It will have fences, stainless steel wires and solid panels to deter birds from perching on the railing (p. 20), but there will be no way to prevent birds from perching on the deck without preventing helicopters from landing there also.  The arrival of helicopters will cause such birds to flee in panic, when the fences and wires will pose hazards to them.  The Evaluation makes clear that “A final, complete deterrent design” has not yet been made (p. 20).  The final EIS-EIR must present a complete design, along with a data-based evaluation of potential risks to Roseate Terns and other birds.

 

            I agree with the assessments in Section 5.7.3 that other potential hazards associated with the project (disturbance, displacement, habitat modification, vessel traffic, indirect impacts, onshore construction) are unlikely to be significant for terns.

 

            (f)  Population Viability Analysis (PVA).  Appendix 5.7-H includes a “Population Viability Analysis” (PVA) for Roseate Terns.  The Roseate Tern Recovery Team (RTRT) has previously taken the position that it would be premature to conduct a PVA or to construct any detailed population model for Roseate Terns, because of incomplete information on the key demographic parameters.  After reviewing the PVA presented in Appendix 5.7-H, I consider that it is a useful first step towards a future metapopulation model, but provides insufficient basis for risk assessment for this project.  Specifically, the demographic parameters assumed for this PVA were derived from a period (1988-1998) when the regional population was increasing fairly steadily, at a rate of about 2% per year.  These demographic parameters (and hence the rate of population growth) were then assumed to remain constant, subject to stochastic variations.  Following the methods of PVA, population changes were projected forward for up to 100 years.  Not surprisingly, given the assumptions, the probability of extinction was calculated to be very low.  The problem is that in recent years, total numbers of Roseate Terns in the northeastern region have shown marked fluctuations; the latest census data (2004) are 15% lower than those in 2001.  Hence, it is not reasonable to assume that the demographic parameters will remain constant for even 5 years, let alone 100 years. Recent studies (e.g. Fieberg and Ellner 2000) have shown that predictions of extinction probability are extremely sensitive to the choice of initial demographic parameters, and that predictions become unreliable when projections beyond 10-20% of the period from which the parameters were derived.  In the case of Roseate Terns, this would limit projections to no more than 2 years. 

 

            Accordingly, I think that this PVA is not an appropriate basis for evaluating the population consequences of losses from this population.  The significance of the potential “take” of Roseate Terns by this or other projects should be weighed in the usual way, on an individual basis in the context of a currently declining population.

 

 

11.  Piping Plovers.

 

            The Applicant’s Evaluation of the Piping Plover is closely parallel to that of the Roseate Tern, both in Appendix 5.7-H.  It is bizarre that the predicted number of fatalities due to the project is 400 times higher for Piping Plovers than for Roseate Terns (0.08 vs 0.00002), despite the fact that the Roseate Tern is 10 times more numerous and crosses Nantucket Sound on a daily basis (versus the assumed twice per year for the Piping Plover: pp. 24-25).  For reasons stated in section 10 of these comments, both assessments are unsound and need to be revised.   However, I recommend that primary attention be given to the Roseate Tern, because even a superficial comparative analysis would show that risks to plovers are much smaller. 

 

 

12.  Wintering Sea-Ducks.

 

            Sea-ducks (eiders, scoters, Long-tailed Ducks, mergansers, and other species) were the most numerous species detected in the Applicant’s aerial and boat surveys in Nantucket Sound during the winter months (November-March).  These included 4 aerial surveys in March 2002 (Appendix 5.7-D), 9 aerial surveys from November 2002 to March 2003 (Appendices 5.7-G and -K), and 8 aerial surveys from November 2003 to February 2004 (Appendix 5.7-M).  No boat surveys were conducted during these periods.  Although partial data were obtained from three winters, the surveys spanned less than two complete 12-month periods, and so fell short of the period of three years specified as minimal by myself and the wildlife agencies.  The aerial surveys gave adequate coverage during the mid-winter period December-February, but not in November or March (4 and 3 surveys, respectively) when birds were expected to be arriving in, passing through and departing from the area.

 

            The data presented in the cited Appendices are confusing because densities were averaged over surveys inside and outside the periods when sea-ducks were present in largest numbers. In addition, the metric reported as “Density” appears to be the sum of densities over all the surveys, rather than the mean.  In the Horseshoe Shoals area (Alternative 1 in the tables) recalculated mean densities during November-March surveys were approximately as follows: Common Eider, 10/km² in 2002-2003, 20/km² in 2003-2004; Long-tailed Duck, 7/km² in 2002-2003, 8/km² in 2003-2004; combined scoters, 15/km² in 2002-2003, 55/km² in 2003-2004.  Combining all five species and all surveys during November-March, the average number of sea-ducks present in the project area was about 2,000.  All species displayed marked fluctuations in numbers both within and between the two winter seasons.  In 2002/2003, numbers of Long-tailed Ducks were higher in October-November than in December-February, while numbers of eiders and scoters were similar.  In 2003/2004, in contrast, numbers of scoters were higher in October-November than in December-February, while numbers of Long-tailed Ducks were similar and numbers of eiders were lower.  These differences evidently reflect frequent movements of large numbers of sea-ducks into and out of the area.  Large but similarly variable numbers of sea-ducks were recorded in other parts of Nantucket Sound.

 

            As in the case of terns, these surveys provide a reasonable picture of the distribution of sea-ducks in the project area, but they provide no useful information about the movements of sea-ducks into, through and out of the project area.  They are further limited by the fact that they were all conducted in good weather and during daylight hours.  Although the USACE Scope specified collection of data, including radar studies, throughout the year, no radar studies were conducted during the period when sea-ducks are present.  Although the evening and morning flights of Long-tailed Ducks into and out of Nantucket Sound were identified as a major concern in my letter of April 26, 2002, the Applicant did not make any attempt to study or document these flights or to assess risks.  That letter made the following recommendation:

 

“To assess potential risks to wintering sea-ducks, fuller and more precise information is needed on their distribution and movements within Nantucket Sound, including heights of flight of any birds that may pass through the project area, especially at night….

 

Except for information on distribution, none of this information was generated and the Applicant appears to have made no effort to do so.  Accordingly, the information presented in the DEIS-DEIR is inadequate to draw conclusions about risks.

           

            The Applicant’s assessment of risks to sea-ducks (adopted by the Corps in Section 5.7.3.2.1) places weight on the fact that only 54 of 377,432 sea-ducks seen during the aerial and boat surveys were observed flying at rotor height.  No boat surveys were conducted during the period when most sea-ducks were present, so this conclusion depends primarily on data collected during the aerial survey.  Since the aerial surveys were conducted only in daylight hours and only in good weather, this does not provide a meaningful basis for extrapolation to other conditions.  As pointed out in my section 9 of my comments, the discrepancy between results of the aerial surveys and the radar studies during September indicates that high-flying birds of other species were largely overlooked during the aerial surveys.  Radar studies could have helped to resolve the issue of the heights at which sea-ducks fly at night and/or in bad weather, but no radar studies were conducted during November-March when sea-ducks are present.

 

            The Applicant (p. 5-126) somehow extrapolated from the 54 sea-ducks recorded flying at rotor height to estimate that “14,645 sea-ducks might be at rotor height”, and then discussed the potential significance of “fatalities in single digits per turbine per year”.  No basis was given for these extrapolations and there does not appear to be any such basis.  Even if there were, “fatalities in single digits per turbine per year” might result in more than 1,000 deaths each year. 

 

 

13.  Other Seabirds.

 

            In addition to terns and sea-ducks, other seabirds recorded in significant numbers during the aerial and boat surveys included grebes, loons, Northern Gannets, cormorants gulls and alcids.  Grebes and alcids were seen in largest numbers during the winter months (December-March), but gulls were seen throughout the year and loons, gannets and cormorants were seen in largest numbers from mid-October to mid-November and in late March and April.  For example, loons were present at average densities of about 11/km2 during three surveys in November-December 2002, about 9/km2 during three surveys in November-December 2002, and about 21/km2 during three surveys in March-April 2003.  [Average densities presented in the Applicant’s reports were lower because they include surveys outside the main periods of occurrence.]  These and other data in the Applicant’s reports indicate that loons, Northern Gannets and Razorbills were sometimes present in thousands on Horseshoe Shoals, gulls in hundreds, and grebes in tens.  Marked fluctuations in numbers from survey to survey within these periods indicate large movements into and out of the area.  However, the reports on the surveys are limited to numbers observed and do not provide any information on movements.  No radar studies were conducted during these periods, despite the Corps’ specification that radar studies should be conducted throughout the year.   

 

            It has long been known that many of these species pass through Massachusetts in large numbers in October-December and in April, with smaller numbers present in mid-winter (e.g., Veit and Petersen 1993).  Gannets and loons in particular are known to fly habitually at heights of 30-150 m above the water, gannets when feeding, traveling between fishing areas, or migrating, and loons when migrating.  Scoters and other seabird species commonly fly at these heights when migrating in late afternoons or evenings, even though they usually fly low over the water at other times.  My letter dated 26 April 2002 drew attention to this and identified these movements as a major source of risk from the project; I specifically drew attention to major hazards to these birds while migrating at night: 

 

“My radar studies detected migration of waterbirds through the area at several different times of year.  Waterbirds were distinguished from landbirds on radar by several different characteristics, including echo intensity, wing-beat frequency, flock size and coherence, and temporal patterns (Nisbet and Drury 1967).  I have also observed waterbirds migrating through the area during my field work at sites including Monomoy Island, Muskeget Island, Penikese Island, and Woods Hole (see below).  The most conspicuous movements were: (a) large numbers of birds flying towards the SE or ESE (about 125o) from mid-October through November.  These were probably mainly waterfowl such as oldsquaws, scoters, and loons arriving at the wintering grounds from staging areas in the Great Lakes.  They usually arrived on a broad front in the later hours of the night at low altitudes and continued SE through Nantucket Sound and past Monomoy and Nantucket.  (b) Large numbers of birds departing towards the SE (about 130o) from July to early October.  These were probably shorebirds traveling towards winter quarters in South America.  They usually took off in flocks from staging areas in late afternoon and ascended rapidly to high altitudes (up to 20,000 feet or 7,000 m) before departing towards the SE at high airspeeds (Nisbet and Drury 1967).  (c) Variable numbers of birds flying through the area towards the SW or WSW in September-November and NE or NNE in April-May.  These include gulls, terns, sea-ducks, gannets, loons, etc.  In daylight hours, these fly in complex spatial and temporal patterns through Nantucket Sound, being diverted by coastlines and islands.  Some of them take off from staging sites in late afternoon and continue flying through the night.  I have observed birds such as gannets and white-winged scoters passing both ends of Muskeget Island in large numbers and flying towards the NE in May, but I have not spent time at locations such as the eastern shore of Martha’s Vineyard in spring or Point Gammon in fall, from which birds would fly over Horseshoe Shoals. At night, waterbirds fly on a broader front, but I have not studied them in detail because my attention was focused on migrating landbirds.

 

“In my opinion, the birds most at risk from flying through the project area would be those in group (a) in the above paragraph, but those in group (c) also need further study. 

 

To assess potential risks to migrating waterbirds, fuller and more precise information is needed on numbers, locations, timing, and heights of flight of birds passing through the project area, especially during evenings and at night during April-May and September-November.” (Emphasis in the original).

 

Despite these recommendations, the Applicant made no attempt to study timing or heights of flight of birds passing through the project area, nor to study movements during evenings and at night.  Accordingly, the data presented in the DEIS-DEIR are totally inadequate to assess risks to these birds. 

 

            The Applicant’s risk assessments (pp. 5-125 to 5-130) acknowledged some risk to migrating waterbirds.  On p. 5-128, it is estimated that 164 grebes, 1,350 loons, 4,091 northern gannets, 8,767 cormorants, 658 scoters, 18,629 gulls, and 10,958 terns “may” fly through the Project area at rotor height in a given year.  However, these numbers were derived by extrapolating from the numbers of birds actually seen flying at rotor height.  For reasons stated in earlier sections of these comments, these numbers are not acceptable as reliable measures of the proportions of birds flying at these heights, even in daylight hours in good weather.  The Applicant’s risk assessments then proceed to discuss the radar data and conclude (p. 5-130) that “it appears that over 600,000 birds could be flying at rotor height in any given year”.  This number also is totally fanciful, given that the radar studies were limited to 26 days in May and 24 days in September in one year, and that the numbers reported were of “targets” (i.e., birds and flocks) rather than birds.  After discussing these findings, the Applicant (and, by adoption, the Corps) rejects all the data generated in the aerial surveys and radar studies (!), and bases its risk assessment (p. 5-129) on extrapolation from “literature [that] suggests that collision-related mortality of birds within wind farms is a relatively rare event; an estimate of 0 to 2.8 fatalities/turbine/yr is low compared to other causes of mortality, such as collisions with buildings (Erickson et al., 2001).  [The cited reference (Erickson et al., 2001) is a paper published by the National Wind Coordinating Committee, summarizing data on birds killed and found at terrestrial wind power projects in the United States.  Data from terrestrial turbines cannot be used without modification to predict mortality from offshore turbines, because seabirds fly around much more than landbirds.]  This range of 0 to 2.8 fatalities/turbine/yr is then (p. 5-130) used as the basis for the final estimate of numbers of birds killed: “conservatively indicating that an estimated 364 birds could be killed each year.”  The Corps demeans itself by publishing such arrant nonsense. 

 

 

14.  Migrating Land Birds.

 

            The only information on migrating land birds generated by the Applicant is from the radar studies conducted in May and September, 2002.  I have addressed the limitations of these studies and the deficiencies in reporting them in Section 8 of these comments.  Although May is the most important period of spring migration of land birds through the region, significant migrations also take place in April.  September is only the beginning of migrations of land birds through the region; much larger movements occur in October and early November.  I pointed this out in my letter dated 26 April 2002: 

 

“My studies revealed that tens of millions of landbirds fly through the area every year, primarily between late August and early November.  Most of these birds passed at night, flying on a broad front without detectable gaps or concentrations at coastlines or islands.  We identified several discrete groups of birds flying in different directions, including SSE (about 170o), SSW (about 215o), and WSW (about 245o) (Drury and Nisbet 1964).  The largest numbers passing over Nantucket Sound were those flying SSE and WSW.  The former were small birds departing from southeastern Massachusetts towards winter quarters in South America, mainly from early September to mid-October.  They took off en masse about 40 minutes after sunset, ascended slowly to heights of several thousand feet, and passed through the area during the first few hours of each night.  The latter were small and medium-sized birds arriving over water from the direction of Nova Scotia, mainly in October and early November.  They arrived after midnight and passed through the area at lower altitudes during the later part of the night.  Birds were also seen migrating in many other directions at various times of the year, but usually in much smaller numbers.  Most landbirds migrating in spring (April to early June) passed well to the northwest of Nantucket Sound.

 

“To assess potential risks to migrating landbirds, fuller and more precise information is needed on numbers, timing, and heights of flight of birds passing through the project area, especially at night during September, October, and early November.” (Emphasis in original).

 

Despite these statements and recommendations, the Applicant’s radar studies were limited to May and September, and the data were reported in such an aggregated way that it is impossible to discern what data refer to the specific groups of birds referred to in my letter.  The Applicant stated that “Night migrating songbirds, for the most part, are likely to fly at altitudes well above the turbine rotors and are not at great risk of collision (Kerlinger, 1995; Kerlinger and Moore, 1989; Able, 1970).”  The Applicant did not explain how songbirds could reach these altitudes without climbing and descending through lower altitudes, however.  The Applicant acknowledged (p. 5-127) that 127,697 targets detected by the radars were flying in the rotor-swept zone, including 44,614 that did so at night. It then proceeded to reject its own data (!) and relied instead on compilations of birds killed at onshore wind power facilities: “The highest fatality rates at onshore wind power facilities in the United States have been about 3 to 7 night migrating songbirds killed per turbine per year (Kearns and Kerlinger 2004; Nicholson 2003)” .  [The first of these references is not in the bibliography; the second refers to a single mountaintop site.]  The Applicant (and, by adoption, the Corps) then rejected even these data (!), and based its final risk assessment (p. 5-129) on extrapolation from “literature [that] suggests that collision-related mortality of birds within wind farms is a relatively rare event; an estimate of 0 to 2.8 fatalities/turbine/yr is low compared to other causes of mortality, such as collisions with buildings (Erickson et al., 2001).  This range of 0 to 2.8 fatalities/turbine/yr is then (p. 5-130) used as the basis for the final estimate of numbers of birds killed: “conservatively indicating that an estimated 364 birds could be killed each year.”   This number apparently is intended to include landbirds as well as waterbirds.  I repeat: The Corps demeans itself by publishing such arrant nonsense

 

 

 

15.  Alternatives.

 

            Section 3.4 outlined alternatives to the proposed project, of which four alternative sites for wind turbines were included in the Detailed Analysis of Alternatives in Section 3.4.3.  Despite the title, the analysis in this section is far from “Detailed”.  The information provided about avian resources at the four alternative sites is a superficial review of a few published sources and consultations with MNHESP and three unnamed experts.  Although new data on the occurrence of seabirds within the Nantucket Sound Alternative sites were generated during the Applicant’s surveys, neither these data nor any newly-generated data of any kind were used.  No information whatsoever is given about movements of birds through the Alternative Site areas, including the daily movements to and from breeding or roosting sites that present the greatest risks to Common and Roseate Terns (see above).  Most of this section is restricted to Federally-Listed or State-Listed species or to State-Listed Habitats, despite the fact that all migratory birds (not only listed species) are protected under federal and state laws.  The “Comparative Summary of Existing Resources” is limited to comparisons of “avian diversity”, listed habitats and/or listed species.  It contains exactly five sentences on seabirds (pp. 3-48 and 3-49). 

 

            Although this section makes clear that the terrestrial site is the only one of the Alternative sites where Roseate Terns and other seabirds are not known to occur, it affords no way to distinguish among or to rank the four marine sites.  This section is totally inadequate to serve as the basis for decision-making.  At a minimum, it should be revised and expanded to reflect available information and knowledge about these sites, including the Applicant’s own data.  Rational comparative assessment of the five sites will require additional, focused field studies to establish their relative risks to birds.  If either the Corps or the Applicant had sought advice from federal or state wildlife agencies, or from individual experts such as myself, at the time the Alternative sites were selected, these studies could have been carried out already.   

 

            Based on my knowledge of these areas (but without detailed review), I believe that all three of the offshore Alternative sites would probably present greater hazards to seabirds than the Horseshoe Shoals site, although the South of Tuckernuck site would probably present lower hazards specifically to Roseate Terns.  Only the Terrestrial (MMR) site would clearly present low hazards to seabirds, but detailed studies would be needed to evaluate whether this would be offset by greater hazards to land birds.

 

 

16.  Standards for Assessing Potential Impacts.     

 

            Section 5.7.3 on “Analysis of Impacts” repeatedly attempts to weigh potential impacts of the proposed project in terms of effects on populations, or against other sources of mortality.  For example, the Population Viability Analyses in Appendix 5.7-H set out to determine whether hypothetical mortality to Roseate Terns or Piping Plovers would significantly increase their probabilities of extinction, or significantly reduce their probabilities of recovery.  Page 5-122 compares the numbers of fatalities for wind turbines with those for collisions with glass windows, hunting activities, and collisions with communications towers.  Page 5-126 compares hypothetical numbers of eiders and scoters that might be killed by collision with turbines with those shot legally each year.  The same page compares numbers of Double-crested Cormorants that might be killed by collision with turbines with those killed under depredation permits.  Table 9 in Appendix 5.7-G compares numbers of waterbirds recorded in the project area with those wintering in Massachusetts and in the Atlantic flyway, and with annual harvests of hunted species.

 

            All these comparisons are misleading and inappropriate.  Federal and state laws protect all migratory birds, not only those listed as Endangered, Threatened, or Special Concern.  Furthermore, these laws protect individual birds, not populations.  Mortality due to other causes is relevant only insofar as it places stress on wild populations, making them more sensitive to additional mortality from other causes.  Thus, any fatalities that may be caused by this project should be considered as cumulative with other causes of death, not as alternatives to them.  In particular, hunted species are monitored and managed by federal and state agencies; hunting regulations are established to maintain populations at harvestable levels.  Under this regulatory system, additional mortality due to this or other projects would be significant because it would reduce the numbers available for harvest.  Mortality of wild birds due to windows, communications towers, cats, etc., occurs and is regrettable, but that does not mean that additional mortality can be trivialized.  To the contrary, it makes any additional mortality more serious. 

 

 

17.  Data Requirements: Terrestrial vs Offshore Sites.

 

            My comments have pointed out severe limitations and deficiencies in the Applicant’s field studies and have specified additional work that will be needed to characterize risks to Roseate Terns, both from the proposed project and from projects at Alternative sites.  I recognize that field studies in offshore areas are difficult and expensive to conduct, and that there is little previous information from the offshore sites from which to draw inferences about risk or lack of risk.  I also understand that the Applicant’s field studies have been limited in scope and rigor because of these difficulties and costs.  I am concerned that the Corps, by issuing this DEIS-DEIR in its present form, appears to be endorsing inadequate field studies, superficial analyses, and inappropriate risk assessment.  The Applicant has proposed to construct this project in offshore waters, presumably because of other advantages that would result from doing so (I assume that these advantages are considered elsewhere in the DEIS-DEIR, in sections that I have not reviewed).  Compared to siting the project on land, this decision obviously results in additional costs in construction, maintenance, and decommissioning.  As environmental scientists and managers, I point out that this decision also leads to additional environmental costs, both in conducting field studies and in assessing risks to environmental resources.  If this project had been proposed for a terrestrial site such as the MMR, the field studies and assessments presented here would have been dismissed as laughably inadequate.  This project poses risks to many important resources, including the Roseate Tern and other listed species that would not be affected by a similar project at a terrestrial site.  An inadequate assessment of such a project should not be accepted simply because an adequate assessment would be expensive or burdensome to the Applicant who chose a difficult site. 

 

 

18.  References.

 

Everaert, J. 2004.  Wind turbines and birds in Flanders: Preliminary study results and recommendations.  Natuur Oriolus 69: 143-155. 

 

Fieberg, J. and S.P. Ellner. 2000. When is it meaningful to estimate an extinction probability? Ecology 81(7):2040-2047.