[Federal Register Volume 60, Number 58 (Monday, March 27, 1995)]
[Rules and Regulations]
[Pages 15693-15699]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 95-7549]



=======================================================================
-----------------------------------------------------------------------

DEPARTMENT OF THE INTERIOR

Fish and Wildlife Service

50 CFR Part 17


Endangered and Threatened Wildlife and Plants; Notice of 
Determination To Retain the Threatened Status for the Coastal 
California Gnatcatcher Under the Endangered Species Act

AGENCY: Fish and Wildlife Service, Interior.

ACTION: Notice of determination.

-----------------------------------------------------------------------

SUMMARY: The Fish and Wildlife Service (Service), announces a 
determination affirming its earlier conclusion (March 30, 1993; 58 FR 
16742) that the coastal California gnatcatcher (Polioptila californica 
californica), a small, insectivorous songbird, is a distinct subspecies 
and, thus, meets the definition of a ``species'' pursuant to the 
Endangered Species Act of 1973, as amended (Act). In addition, the 
Service affirms its earlier conclusion (58 FR 16742) that the southern 
limit of this subspecies extends to about 30 deg. north latitude near 
the vicinity of El Rosario, Baja California, Mexico. Based on these 
determinations, the Service concludes that its March 30, 1993, decision 
that the coastal California gnatcatcher is a threatened species was 
correct. Federal protection for the coastal California gnatcatcher is 
thus continued.

EFFECTIVE DATE: March 23, 1995.

ADDRESSES: The complete administrative records and files for this 
determination and all related rule promulgations and notices are 
available for inspection, by appointment, during normal business hours 
at the Fish and Wildlife Service Carlsbad Field Office, 2730 Loker 
Avenue West, Carlsbad, California 92008.

FOR FURTHER INFORMATION CONTACT: Mr. Gail C. Kobetich, Field 
Supervisor, at the above address (telephone 619/431-9440).

SUPPLEMENTARY INFORMATION:

Background

    The coastal California gnatcatcher (Polioptila californica 
californica), a subspecies of the California gnatcatcher, is a small, 
long-tailed member of the thrush family Muscicapidae. The subspecies is 
restricted to California and Baja California, Mexico, and is an 
obligate resident of coastal sage scrub, which is one of the most 
depleted habitat types in the United States (58 FR 16742). The plumage 
color of the species is dark blue-gray above and grayish-white below. 
The tail is mostly black above and below. This subspecies is 
distinguished from the other subspecies by its darker body plumage, 
less extensive white on tail feathers (rectrices 5 and 6), and longer 
tail (Atwood 1991). The male has a distinctive black cap that is absent 
during the winter. Both sexes have a distinctive white eye-ring. 
Vocalizations of this species include a call consisting of a rising and 
falling series of three kitten-like mew notes (National Geographic 
Society 1983).
    The California gnatcatcher was originally described as a distinct 
species (Polioptila californica) by Brewster (1881) based on specimens 
collected by Stephens in 1878. Later taxonomic treatments (e.g., Coues 
1903 and Chapman 1903) reflected Brewster's (1881) conclusions. 
Grinnell (1926), however, later concluded that the species was a form 
of the black-tailed gnatcatcher (Polioptila melanura), which inhabits 
the Sonoran and Chihuahuan Deserts of the southwestern United States 
and northwestern Mexico. Subsequent scientific publications (American 
Ornithologists' Union 1931, Grinnell and Miller 1944, Friedmann 1957, 
American Ornithologists' Union 1957) adhered to the species limits as 
defined by Grinnell (1926). Three subspecies of the black-tailed 
gnatcatcher were recognized for southwestern California and western 
Baja California, Mexico: P. m. californica (ranging from Los Angeles 
County, California (formerly northward to Ventura County), south to 
about 30 deg. north latitude in Baja California, Mexico), P. m. 
pontilis (resident in central Baja California), and P. m. margaritae 
(ranging from about 27 deg. north latitude south to the Cape region of 
Baja California) (American Ornithologists' Union 1957).
    Based on identified differences in ecology and behavior that were 
elucidated as a result of specimen study and statistical analysis, 
Atwood (1988) proposed that Polioptila californica was specifically 
distinct from P. melanura. This finding was subsequently formally 
adopted by the American Ornithologists' Union Committee on 
Classification and Nomenclature (American Ornithologists' Union 1989), 
thus affirming Brewster's (1881) original taxonomic placement with 
respect to species. The American Ornithologists' Union 1989 publication 
did not address subspecies other than to refer the reader to the 
American Ornithologists' Union 1957 checklist of North American birds.
    The coastal California gnatcatcher, Polioptila californica 
(=melanura) californica, has been recognized as a distinct race or 
subspecies since Grinnell's (1926) publication (e.g., American 
Ornithologists' Union 1931, Grinnell and Miller 1944, Friedmann 1957, 
American Ornithologists' Union 1957, Garrett and Dunn 1981, Unitt 1984, 
Phillips 1991, Atwood 1991). As indicated above, this subspecies occurs 
from Los Angeles County (and, formerly, Ventura County) south to about 
30 deg. north latitude in Baja California, Mexico. Although Atwood 
(1988) proposed merging P. californica californica with a more 
southerly subspecies of P. californica, he later (1991) retracted this 
conclusion.
    On March 30, 1993, the Service published a final rule determining 
the coastal California gnatcatcher (Polioptila californica californica) 
to be a threatened species (58 FR 16741). In making this determination, 
the Service relied, in part, on taxonomic studies conducted by Dr. 
Jonathan Atwood of the Manomet Bird Observatory. As is standard 
practice in the scientific community, the Service did not request, nor 
was it offered, the data collected and utilized by Atwood in reaching 
his conclusions. Instead, the Service cited the conclusions presented 
by Atwood in a peer reviewed, published scientific article pertaining 
to the subspecific taxonomy of the California gnatcatcher (Atwood 
1991).
    The Endangered Species Committee of the Building Industry 
Association of Southern California and other plaintiffs subsequently 
filed a suit challenging the listing on several grounds. In a 
Memorandum Opinion and Order filed in the United States District Court 
for the District of Columbia on May 2, 1994, the Court vacated the 
listing determination, holding that the [[Page 15694]] Secretary of 
Interior (Secretary) should have made available the underlying data 
that formed the basis of the Atwood (1988) report in light of the 
controversy surrounding inconsistent conclusions reached by Atwood in 
his 1988 and 1991 studies.
    Following the Court's decision, Atwood released his data to the 
Service. These data were, in turn, made available to the public for 
review and comment on June 2, 1994 (59 FR 28508). By order of June 16, 
1994, the Court reinstated threatened status for the coastal California 
gnatcatcher pending a determination by the Secretary whether the 
listing should be revised or revoked in light of his review of the 
subject data and public comments received during public comment 
periods. This notice constitutes the Service's determination in 
response to the Court's June 16, 1994, order.

Summary of Comments and Recommendations

    A proposed rule to list the gnatcatcher as endangered was published 
on September 17, 1991 (56 FR 47053). Public comments were solicited and 
two public hearings were held on the proposed rule. Notification of the 
hearings was published in the Federal Register on February 7, 1992 (57 
FR 4747). A legal notice announcing the hearings and inviting general 
public comment on the proposal was also published on February 7, 1992 
in the Los Angeles Daily News, Los Angeles Times, Riverside Press-
Enterprise, and the San Diego Union-Tribune. Public hearings were 
conducted in Anaheim, California, on February 25, 1992, and in San 
Diego, California, on February 27, 1992. A notice of extension and 
reopening of the comment period for 30 days to obtain additional 
information on gnatcatcher taxonomy was published on September 22, 1992 
(57 FR 43688). On February 11, 1993, the Service published a notice 
announcing the reopening of the public comment period on the proposed 
rule for 20 days and the availability of a report prepared by Service 
taxonomists on the taxonomic validity of P. c. californica (58 FR 
8032). On March 30, 1993, the Service published a final rule 
determining the coastal California gnatcatcher to be a threatened 
species (58 FR 16741). That same day, a proposed special rule pursuant 
to section 4(d) of the Act was published (58 FR 16758). The final 
special rule was published on December 10, 1993 (58 FR 65088).
    Following the Court's Order of May 2, 1994, and receipt of Atwood's 
data, the Service announced the availability of these data and the 
opening of a public comment period on June 2, 1994 (58 FR 28508). 
Atwood's data were sent to 15 parties upon request. With the approval 
of the Court, the public comment period was extended to December 1, 
1994 (59 FR 53628), to allow the public additional time to receive and 
then comment upon the raw data and methodology utilized by Atwood.
    During this 6-month public comment period, Dr. William Link and Mr. 
Grey Pendleton of the National Biological Service, Department of the 
Interior (Department), conducted a new and independent analysis of 
Atwood's data (Link and Pendleton in litt. 1994). To assure that the 
Service utilizes the best scientific information available in the 
implementing the Act, it is policy (59 FR 34270) to seek independent 
review of the scientific basis for listing and recovery actions. 
Consistent with this policy, the Service solicited comments on the 
National Biological Service document and all other public comments 
received by December 1, 1994, from the general public, including 
scientists with an expertise in avian taxonomy. A new 30-day public 
comment period (59 FR 66509) was opened on December 27, 1994, to allow 
the public to review and comment on these documents. This final comment 
period closed on January 26, 1995.
    A total of 31 comments pertaining to either (1) Atwood's data, 
methodology, or results, or (2) the taxonomy of the gnatcatcher was 
received during the final two comment periods. This total includes 21 
comments received prior to December 1, 1994, and an additional 10 
comments received during the final comment period. Included among the 
comments were three new, independent analyses of Atwood's data. After a 
review and consideration of all such comments, five relevant issues 
have been identified and are discussed below. The five issues encompass 
all substantive comments pertaining specifically to Atwood's data, 
analyses, and conclusions regarding the taxonomy and geographic range 
of the coastal California gnatcatcher.
    Issue 1: Several commenters noted that Atwood's apparent discarding 
of raw data precludes an appropriate analysis of his conclusions. One 
commenter in particular was disturbed that ``Atwood no longer has the 
raw data used in his original analyses.'' Another commenter noted that 
Atwood admitted to discarding computer programs used in the analysis of 
the data subsequently analyzed and reported in his 1991 publication. 
Some stated that differences existed between the data sets used in 
Atwood (1988), Atwood (1991) and that provided to the Service and the 
public (Atwood in litt. 1994a).
    Service Response: Atwood (in litt. 1994b) has stated that the 
measurements provided to the general public following the May 2, 1994, 
Court Order ``represent the total and unmodified data set that formed 
the basis for my 1988 and 1991 publications on gnatcatcher 
morphology.'' Atwood (in litt. 1994b) also indicated that only one 
difference existed between the computer file data set transmitted to 
the public and the data on the original paper forms that he discarded 
after entering the data into a computer file, the sex of a single 
specimen from sample area SI29 was corrected. Atwood (in litt. 1994b) 
further indicated that he verified (in 1985) the data on the computer 
by comparing it with the hand-written information on the paper forms. 
The Service concludes that there is no reason to doubt the veracity of 
Atwood in this regard.
    Because data on paper forms cannot readily be subjected to 
statistical analysis, the data transferred to a computer or computer 
disk are, in essence, the raw data at issue. The Service, therefore, 
rejects the contention that Atwood discarded his raw data, thereby 
precluding reanalysis of the data.
    After providing his data, Atwood realized that discrepancies in 
sample size existed between data reported in his dissertation, his 
monograph (Atwood 1988), his subspecies paper (Atwood 1991), and data 
provided to the Service. He noted that for site SI29 there was a 
discrepancy with respect to one female and one male specimen and 
concluded that he had corrected the sex for one individual. His 
dissertation revealed 14 specimens for sites PP28 and MA30, whereas the 
Service data includes 13 specimens for site PP28 and 15 for site MA30. 
Atwood believes that this discrepancy was the result of correctly 
placing one specimen in site MA30 rather than PP28. These two 
corrections resulted in apparent discrepancies. Atwood was unable to 
explain an additional discrepancy, in his dissertation he reported 19 
female specimens for site SD24, whereas the data provided to the 
Service indicates 20 female specimens for site SD24; Atwood suggested 
that a typographical error had occurred.
    Atwood discovered numerous discrepancies between the sample sizes 
for his monograph (Atwood 1988) and the data given to the Service 
(amounting to 15 more male specimens and 7 fewer female specimens 
reported in the Service data set). Atwood could not conclusively 
explain these [[Page 15695]] discrepancies, but suspected that they 
were associated with the differing lengths of data set lines that may 
have caused the SAS program to skip lines or combine lines of data. He 
suggested that this problem may also have affected his analyses of the 
data presented in the 1988 monograph. He indicated that because of 
these potential problems, he felt that it is inappropriate to rely upon 
the 1988 monograph with respect to subspecific conclusions, although 
the conclusions with respect to species were unaffected, and are valid. 
However, programming errors would not have affected the original data 
set.
    Atwood (in litt. 1994a,b) has acknowledged that one of the 213 
samples in the data set provided to the public was not used in his 1991 
study or in his previous, unpublished status review of the gnatcatcher. 
Atwood (in litt. 1994b) believes that the excluded specimen was that 
designated YP2717, an aberrant specimen (possibly a black-tailed 
gnatcatcher or interspecies hybrid) collected in 1885. In Atwood's 
subsequent reanalysis of the original data set, specimen YP2717 was 
excluded from the data set because it differed from its sample area 
mean by more than 3 standard deviations (Atwood in litt. 1994b).
    The sample size discrepancies for all reports, except the 1988 
monograph, are very minor, and would not have affected the overall 
conclusions of the authors. Atwood (1994b) has characterized the 
analysis of his 1988 monograph as being ``seriously flawed'' with 
respect to data processing. The sample size discrepancies between 
Atwood's other reports, and the 1988 monograph are likely due to these 
data processing problems, and not the result of changes made to the 
data set. The Service, therefore, concludes that the data set provided 
by Atwood to the Service adequately duplicates the data originally 
written on paper forms.
    Issue 2: One commenter noted that two of Atwood's publications 
(1988 and 1991) were contradictory in that they proposed different 
geographic ranges for the taxon of California gnatcatchers occurring in 
the United States. This same commenter suggested that Atwood's (1991) 
retraction of his original (1988) conclusions pertaining to the 
subspecies taxonomy of the California gnatcatcher was prompted by his 
desire to affect the listing of the species.
    Service Response: While the record indicates that Atwood believes 
that the listing of the coastal California gnatcatcher is warranted, 
the record also indicates that Dr. Atwood's revised conclusion about 
the subspecific geographic limits of Polioptila californica californica 
resulted from his 1991 reanalysis of the data cited in his 1988 
monograph. The (1988) monograph had received peer review critical of 
its findings.
    The Service receives dozens of petitions to list or delist species 
each year. The Act requires the Service to conduct an independent 
review of each of these petitions, and to make final decisions on the 
basis of the best scientific data available. The motives of the 
petitioners, as with commenters, are not relevant to the Service's 
decisions on these issues.
    Issue 3: Several commenters alleged fundamental flaws in the data 
used by Atwood (1991) in generating his conclusions. In particular, 
commenters suggested or concluded that the data appeared to be 
incomplete, or non-random (i.e., ``censored''). Several commenters were 
concerned that the variables were ``confounded'' (i.e., the effects of 
two or more factors on a response variable could not be separated) due 
to the age or condition of certain specimens. These commenters 
indicated that for the northern sites nearly all specimens were 
collected prior to 1940, and none of the specimens from the remaining 
sites were collected prior to 1920. One commenter noted that a 
potential exists for serious bias in the data due to specimen 
``foxing'' (i.e., browning with age). Another commenter noted, citing 
relevant published scientific literature, that body size and plumage 
brilliance and iridescence can reflect variation in specimen condition. 
Some of these commenters suggested that differences in characters among 
sites may be the result of the age of the collection, and not the site 
from which they were collected.
    Service Response: On behalf of the Service, the National Biological 
Service independently conducted a new analysis of Atwood's data (Link 
and Pendleton in litt. 1994). Three additional independent analyses of 
the data were also submitted during the comment period.
    In response to one commenter's concern that the data appeared to be 
a non-random sample of California gnatcatchers, the National Biological 
Service (Newton, in litt. 1995) replied that although these are valid 
concerns, they are not proof, as acknowledged by the commenter, that 
Atwood's data are not representative. One commenting ornithologist who 
was largely critical of Atwood's (1991) analyses nevertheless concluded 
that ``[t]he data set gathered by Atwood was quite comprehensive and 
included measurements from a large number of specimens throughout the 
range of the species.'' In the Memorandum Opinion and Order filed May 
2, 1994, the Court declared, citing the declaration of this 
ornithologist, that ``it is not disputed that Atwood's means of 
collecting data were proper.''
    After noting the possible problem of the age of the specimen being 
confounded with the collection site, the authors attempted to adjust 
the data for year or month the data was collected. McDonald et al. (in 
litt. 1994) removed specimens collected from May to September and thus 
avoided problems associated with feather wear. Link and Pendleton (in 
litt. 1994) adjusted several characters for month and year based on the 
results of regression analyses. Messer (in litt. 1994) conducted two of 
her analyses by limiting the specimens to those collected between 1920 
and 1940, and 1980 to 1984. Link and Pendleton (in litt. 1994) were 
cautious and indicated that they may not have removed all of the 
confounding effects; however, they also indicated that they may have 
over adjusted the data and removed differences due to sites.
    After adjusting the data for year and month of collection, Link and 
Pendleton (in litt. 1994) obtained results similar to the unadjusted 
data. Messer (in litt. 1994) was able to classify the northern birds 
from the southern birds using specimens collected from 1980 to 1984 
correctly in 84 percent of the cases, and using birds collected from 
1920 to 1940 in 94 percent of the cases. The results of McDonald et al. 
(in litt. 1994) also yielded weak evidence of a break at 30 deg. north 
latitude, even though they had removed birds collected during certain 
months of the year. Atwood (in litt. 1994b) also had similar results 
before and after he excluded the variable brightness of breast plumage 
(a variable that would have changed as a specimen aged) from his 
analysis.
    Given the above considerations and results, the Service finds no 
justification or cause for concluding that Atwood's data were 
incomplete, censored, or otherwise inadequate. Further, the Service 
concludes that the available information does not support the 
hypothesis that the confounding of variables is responsible for 
erroneous conclusions regarding perceived breaks in the morphology of 
the coastal California gnatcatcher. The Service concludes that the 
analysts took adequate care to remove the possible effects of 
confounding of age of specimen and collection area.
    Issue 4: The Service received four significant analyses and a 
number of critiques of each of the analyses of Atwood's data. Each 
commenter [[Page 15696]] attempted to answer a different question, and 
consequently, each of the analyses used somewhat different statistical 
techniques, and drew somewhat different conclusions. Some of the 
commenters concluded that the clinal nature of the data would argue 
against subspecies; or that a primary break occurs further south and 
would argue that if there are subspecies, the boundary line should be 
drawn further south in Baja California. Others argued that the data are 
clumped (consistent with a subspecific break); or that the birds north 
of 30 deg. north latitude are different from the birds south of 30 deg. 
north latitude. The Service analyzed these reports to draw a conclusion 
regarding whether the data support Atwood's 1991 conclusions.
    Service Response: Several commenters produced new analyses of the 
data provided by Atwood. Atwood (in litt. 1994b) also provided an 
additional taxonomic analysis of the data. With the exception of Atwood 
(in litt. 1994b), all of the authors (Messer in litt. 1994, Link and 
Pendleton in litt. 1994, and McDonald et al. in litt. 1994) explicitly 
stated that their expertise is in statistics, and that taxonomic 
conclusions should be left to taxonomists. The Service has carefully 
reviewed each of these analyses and critiques to examine the strengths 
and the weaknesses of each approach. A summary of these analyses 
follows.
    Atwood (in litt. 1994b) presented a reanalysis of his data using 
log10 transformations of 6 variables (bill length, tarsus length, 
wing length, tail length, length of white spot on a tail feather 
(retrix 6), and brightness of breast plumage). In one analysis, he 
excluded the variable ``brightness of breast plumage'' because Mellink 
and Rea (1994) found readings inconsistent, even when resampling a 
single specimen. Atwood used a Tukey-Kramer method to conduct pairwise 
comparisons of the sample area means. He also conducted a principal 
components analysis (a method of determining how the data are 
intercorrelated, and reducing intercorrelated data to a principal 
component score) of the data and performed a cluster analysis on the 
first two principal component scores as well as on the original 
variables. Tail length, tail spot length, and brightness of breast 
plumage varied significantly among sample areas (all P<0.001), and 
multiple comparison tests revealed a grouping, or ``step,'' at 30 deg. 
north latitude. The cluster analyses grouped sites north of 30 deg. 
north latitude together, and variously grouped sites to the south. 
Atwood's methods show that regional means may be clumped, but do not 
show whether individual birds can be placed correctly into these 
groups.
    Link and Pendleton (in litt. 1994) used regression analysis of mean 
latitudes of Atwood's (1991) nine sample areas against 25 characters. 
They determined that the data vary along a geographic gradient. Link 
and Pendleton (in litt. 1994) then conducted a series of tests to 
determine if the characters were representative of gradual change or of 
groupings. They used multivariate analysis of variance (MANOVA) to 
place the original 9 sites into the best groupings of 8 sites, 7 sites, 
6 sites, 5 sites, 4 sites, 3 sites and 2 sites. Abbott et al. (1985), 
in their book on taxonomic analysis, recommended the use of canonical 
variate analysis (MANOVAs) for delineation of subspecies, where the 
data are continuous and the data are preclassified into postulated 
groups. Akaike's information criterion (AIC) was used by Link and 
Pendleton (in litt. 1994) to determine which grouping best fit the 
data. Link and Pendleton (in litt. 1994) conducted discriminant 
function analysis to determine if they could correctly classify birds 
into groups. Hotelling's t2 test was used to test the significance 
of the results. Cluster analysis and discriminant coordinates were 
computed on the individual specimens to see how the data was clumped. 
Finally, they attempted to adjust the data for time effects (see issue 
3 above).
    Link and Pendleton (in litt. 1994) obtained similar results in each 
of these tests. They concluded that the changes in the characters are 
more representative of groupings than of gradual change. They 
determined that, at least one break occurs north of site 5 (mean 
latitude of site 5 is 29.5 deg. north latitude) and at least one break 
occurs south of site 5. The use of MANOVA would reduce the likelihood 
of Type 1 error (reporting differences that do not exist) that would 
occur if you looked at each variable separately. The AIC is not prone 
to overfitting, and can be used to determine the model that best fits 
the data. The AIC does not have an associated statistical test for 
significance, and therefore, the groups identified in this manner may 
not represent actual groupings (Newton in litt. 1995). Though Newton 
(in litt. 1995) also indicated that Atwood's (in litt. 1994b) cluster 
analysis would have been more useful if he had used individual 
specimens rather than group means, Link and Pendleton's (in litt. 1994) 
cluster analysis did use individual specimens and yielded groups 
similar to their MANOVA results, creating a stronger basis for their 
conclusions.
    Messer (in litt. 1994) examined whether the birds north of 30 deg. 
north latitude can be distinguished from the birds south of 30 deg. 
north latitude. She used multivariate discriminant analysis to classify 
birds into northern and southern subgroups with the boundary set at 
30 deg. north latitude. Discriminant analysis is used when one is 
examining a categorical dependent variable (e.g., north or south of 
30 deg. north or one of 9 sites) and metric independent variables 
(e.g., measurements of gnatcatcher characteristics). Discriminant 
analysis would test whether the means among groups are equal. Using 
several subsets of the data (e.g., limiting years of collection to 
remove time effects, or in developing a model with one set of data and 
another to test the model), Messer (in litt. 1994) concluded that one 
could correctly classify the birds as being from the northern or 
southern areas with 86 to 92 percent accuracy.
    McDonald et al. (in litt. 1994) conducted their analyses using 
individual specimens, and estimated the latitude based upon the 
locality description given by Atwood in his original data set. They 
removed specimens collected from May through September to avoid data 
problems due to feather wear and molting, and attempted to adjust some 
data for year of collection. To examine how the data are 
intercorrelated, they conducted principal components analysis on size, 
color, and pattern variables separately. McDonald et al. (in litt. 
1994) conducted Gabriel's sum of squares simultaneous test procedure on 
the first principal component scores and on the original variables. 
McDonald et al. (in litt. 1994) conducted an intervention analysis to 
look for steps or breaks in the trends in means. In addition, they 
conducted a discriminant function analysis to determine whether birds 
could be correctly classified at various latitudes.
    The results from the Gabriel's test indicated that there were 
significant differences in means of the first principal component at 
24 deg. north latitude, and that for some of the size variables there 
was weak evidence for a trend in means at 28 deg. north latitude or 
27 deg. north latitude. There was weak evidence for difference in the 
means at 30.5 deg. north latitude for the first principal component for 
color variables. The intervention analysis revealed a significant rate 
of change for 4 of the 16 individual size variables (page 6) at 30 deg. 
north latitude (P< 0.10). The discriminant function analysis revealed 
that the lowest misclassification rate was at 24 deg. north latitude (4 
percent). The misclassification rate at 30 deg. north latitude was 13 
percent (a 25 percent [[Page 15697]] misclassification rate is 
generally acceptable for many subspecific groups).
    McDonald et al. (in litt. 1994) used principal components analysis 
(a method to reduce intercorrelated data to a single principal 
component score) on size, color, and pattern variables separately. This 
analysis may have been done to group data by measurement type (e.g., 
units of length, weight, etc.), as is recommended in some statistics 
books (Newton in litt. 1995). Other statistics texts (e.g., Hair et al. 
1995) apparently do not recommend grouping like measurements. A more 
exhaustive approach to principal components analysis would have been to 
do the analysis on all variables simultaneously, then exclude size 
variables, then pattern variables, and so forth (Newton pers. comm. 
1995). In that manner, McDonald et al. may have detected additional 
intercorrelations among gnatcatcher characteristics.
    McDonald et al. (in litt. 1994) presented a stronger case for 
breaks in characters south of 30 deg. north latitude than they did for 
characters found at 30 deg. north latitude. Atwood (1991, in litt. 
1994), and Link and Pendleton (in litt. 1994) also found breaks south 
of 30 deg. north latitude. The evidence of at least one break south of 
30 deg. north is supportive of Atwood's (1991) conclusion of an 
additional subspecific break. McDonald et al. (in litt. 1994) provided 
the strongest evidence against Atwood's (1991) conclusions. 
Nonetheless, they found weak statistical results supporting a break at 
30 deg. north latitude (Gabriel's SS-STP and intervention analysis). 
They also acknowledged that Gabriel's test may not have detected 
differences in the critical region, where Atwood concluded changes 
occur, because this test is sensitive to small sample sizes (i.e., an 
investigator needs a large number of individual records before the test 
will detect differences). Thus, in this portion of the analysis of 
McDonald et al., the possibility of a Type 2 error or accepting the 
null hypothesis when it should be rejected (i.e., believing that there 
is no break in characters when in fact one does occur) was higher than 
the possibility of a Type 1 error or rejecting the null hypothesis when 
it should be accepted (i.e., believing that there is a break in 
characters, when in fact no break exists).
    McDonald et al. (in litt. 1994) also used ``intervention 
analysis'', a procedure normally used when an experimenter intervenes 
in some way (i.e., provides medical treatment) and wants to evaluate 
whether changes in behavior or performance are statistically 
significant (Edgington 1987). McDonald et al. (in litt. 1994) 
apparently used this approach to see if changes at various latitudes 
resulted in a sharp step. McDonald et al. (in litt. 1994) provided 
limited details of this method, which they modified and ``tested using 
data from the literature.'' Therefore, the Service was unable to fully 
evaluate this method, which apparently is not commonly used. Messer (in 
litt. 1995), however, indicated that the technique is a ``nonparametric 
(and thus less powerful) version of linear regression analysis.''
    McDonald et al. (in litt. 1994) did find weak statistical evidence 
for a break in characters at 30 deg. north latitude, and were able to 
distinguish the birds north and south of this line with a 13 percent 
error rate. In evaluating their techniques, the Service notes that 
McDonald et al. (in litt. 1994) used techniques that were less 
exhaustive, or that were less well known, or that may have been more 
likely to result in a Type 2 than in a Type 1 error than techniques 
used by the other authors. The techniques of McDonald et al. (in litt. 
1994) appeared more likely to accept the null hypothesis (e.g., there 
is no subspecific break in gnatcatchers at about 30 deg. north 
latitude). Given the selection of statistical techniques by McDonald et 
al. (in litt. 1994), and that Atwood (in litt. 1994), Link and 
Pendleton (in litt 1994), and Messer (in litt. 1994) found evidence for 
a break at 30 deg. north latitude, the Service concludes that the weak 
statistical evidence of a break at 30 deg. north latitude presented by 
McDonald et al. should be given greater credence.
    In summary, the MANOVA conducted by Link and Pendleton (in litt. 
1994) and cluster analysis conducted by Atwood (in litt. 1994) are 
supportive of groupings of birds rather than a cline. Use of cluster 
analysis by Link and Pendleton (in litt. 1994) on individual specimens 
provides stronger evidence that groups or ``steps'' exist in 
characters. In addition, efforts by McDonald et al. (in litt. 1994) and 
Messer (in litt. 1994) to determine correct classification rates 
provide further evidence that gnatcatcher variance along a geographic 
gradient is more indicative of groupings than of a gradual cline. The 
misclassification rates at 30 deg. north latitude were well within the 
range acceptable for subspecies.
    Each author utilized different statistical methods to analyze the 
data and draw conclusions. As a first step, the authors investigated 
whether they could separate the means among various groupings of the 
data. Atwood (in litt. 1994b) used a Tukey-Kramer multiple comparison 
procedure to determine if the means of individual variables among 
previously selected groups could be separated. Link and Pendleton (in 
litt. 1994) used Hotelling's t2 on the groupings identified in 
their MANOVA analysis to determine if the means could be separated. 
McDonald et al. (in litt. 1994) used Gabriel's method to determine 
differences in means at selected latitudes. Each of these approaches 
was successful in separating means among groups of gnatcatchers.
    The investigators next examined whether there might be steps in 
these changes, or whether one could correctly classify (or place) the 
birds within these groups. Messer (in litt. 1994) conducted a 
multivariate discriminant analysis and found that the birds could be 
classified into a groups north and south of 30 deg. north latitude with 
an error rate of about 10 percent. Link and Pendleton (in litt. 1994) 
conducted a clustering analysis to group individual specimens into 
clusters and examined the overlap between the clusters and the 
groupings identified in the MANOVA. McDonald et al. (in litt. 1994) 
conducted a discriminant function analysis to identify latitudes that 
separate the range of the bird into 2 groups with minimal 
misclassification rates. Each of these approaches showed a break in the 
characters at 30 deg. north latitude, and was supportive of Atwood's 
(1991) conclusions.
    In a statistically pure sense, these methods are exploratory in 
nature and were useful in identifying hypotheses that could be tested 
with respect to the gnatcatcher. To formally test these hypotheses, an 
investigator would need to make similar measurements on newly gathered 
gnatcatcher specimens. Issue 5 below discusses the Service's response 
to this point. However, it is important to understand that statistics 
are a tool used to assist an investigator in drawing conclusions in 
that they can help quantify uncertainties with respect to those 
conclusions (Newton pers. comm. 1995). The investigator still needs to 
evaluate the practical significance of results, and should not focus 
exclusively on statistical significance (Abbott et al. 1985, Hair et 
al. 1995, Mayr et al. 1953). Statistics do not remove or supplant the 
need to make informed decisions with respect to any data set. Messer 
(in litt 1994), Link and Pendleton (in litt. 1994), and McDonald et al. 
(in litt. 1994) all explicitly recognized that taxonomic decisions 
should be made by taxonomists.
    The misclassification rates identified by Messer (in litt. 1994) 
and McDonald et al. (in litt. 1994), and the overlap in many of the 
characters show that these [[Page 15698]] groupings of gnatcatchers are 
not entirely discrete. Abbott et al. (1985) noted that taxonomists 
expect ``variation within species to involve either a continuum or at 
least some continuity or overlapping between forms.'' If the groupings 
of California gnatcatcher were entirely discrete, avian taxonomists 
likely would have assigned these groupings to separate species. Mayr 
(1970) defined subspecies as ``an aggregate of phenotypically similar 
populations of a species inhabiting a geographic subdivision of the 
range of a species and differing taxonomically from other populations 
of the species.'' Mayr (1970) concluded that the magnitude of taxonomic 
difference necessary to appropriately decide when subspecies should be 
delimited ``can be determined only by agreement among working 
taxonomists.''
    Grinnell (1926), Phillips (1991), and Atwood (1991) identified 
30 deg. north latitude as a boundary between Polioptila californica 
(=melanura) subspecies. Recent work suggests that the southern boundary 
of P. c. californica may be further north, near the international 
boundary between the United States and Mexico (Mellink and Rea 1994). 
Mellink and Rea (1994) placed the birds between the international 
border and 30 deg. north latitude in a new subspecies. Atwood 
identified another subspecific break south of 30 deg. north. McDonald 
et al. (in litt. 1994) and Link and Pendleton (in litt. 1994) also 
noted a break south of 30 deg. north latitude, consistent with Atwood's 
(1991) conclusion of an additional subspecific break. The consensus 
among working taxonomists supports recognition of P. c. californica, 
albeit its range may be more restricted than that proposed by Atwood 
(1991). Therefore, the Service concludes that a finding that 30 deg. 
north latitude as the southern specific boundary of P. c. californica 
is supported by the available scientific evidence. Until additional 
taxonomic work is published and accepted by the ornithological 
community, the Service will recognize 30 deg. north latitude as the 
southern subspecific boundary of P. c. californica.
    Issue 5: Several commenters stated that analyses of a newly 
collected independent data sets should be done to clarify gnatcatcher 
taxonomy or resolve differences of opinion among the various 
commenters. One commenter urged the Service to ``dismiss the subspecies 
issue for gnatcatchers (pending further study) and focus on the 
management of U.S. populations.'' Another commenter concluded that ``a 
rigorous analysis of both morphometric, reflectance, genetic, and other 
chemical data are required to address the problem in the clearest 
possible manner.'' Other commenters added that the gnatcatcher should 
not be listed until the perceived taxonomic controversy is resolved.
    Service Response: The Service fully endorses and encourages efforts 
to assess and refine the taxonomic status of all species, including the 
coastal California gnatcatcher, provided that any collection of 
specimens associated with such efforts does not result in unacceptable 
mortality or other impacts. However, in making listing determinations, 
section 4(b) of the Act requires the Service to make its listing 
decisions within set timeframes and requires the Service to base its 
listing decisions on the best scientific and commercial data available 
at the time of the decision. The Service is not authorized to delay 
listing decisions until all studies of arguable utility are completed, 
until scientific debate is exhausted, or until complete consensus 
occurs. The Service cannot await the ``next study,'' which may or may 
not occur and which may or may not be affirmed by the scientific 
community through the appropriate peer review process.
    Efforts to conduct further analysis on the taxonomy and subspecific 
limits of the California gnatcatcher would be costly and time 
consuming. One could seek additional museum records not analyzed by 
Atwood, or could collect new specimens. Collecting new specimens could 
result in unacceptably high mortality. Moreover, collecting new field 
specimens prior to making a final decision on this issue is not 
practical. Alternatively, investigators could capture birds in mist 
nets and obtain these measurements from live individuals, which would 
then be released. However, additional researchers would be unable to 
verify the results by visiting a museum and repeating the measurements. 
As stated above under issue 4, the Service was charged with evaluating 
whether Atwood's data supported his conclusions, and not with carrying 
out additional studies to remove any and all controversy surrounding 
the taxonomy of the Polioptila californica subspecies.

Conclusion

    The Service has been charged with scrutinizing data and conclusions 
rendered by Atwood, and determining if his data support his 
conclusions. The Act provides that the Service must render its 
determination on the basis of the best scientific and commercial data 
available. The Service has made a concerted effort to obtain and 
accurately assess the best scientific and best commercial information 
available regarding the taxonomy and range of the coastal California 
gnatcatcher. As an integral part of this process, the Service's 
statutory mandates and standard scientific protocol require that we 
recognize and act in accordance with the concepts, conventions, and 
practices of the scientific method. To this end, the Service must seek 
and seriously consider (1) data and analysis published in peer 
reviewed, scientific journals, (2) the opinions of recognized experts 
in given scientific disciplines, and (3) the input of the interested 
public.
    In this effort the Service has reviewed the analyses of the data 
used by Atwood in his 1988 and 1991 papers. The Service finds that the 
conclusions reached by Atwood (1991) are reasonable, and are generally 
supported by the additional analyses received.
    Under any circumstances that pertain to the taxonomy of North 
American bird species, the Service actively seeks the publications, 
input and expert opinion of the American Ornithologists' Union (AOU) 
and its constituent Committee on Classification and Nomenclature 
(Committee). The Committee and its publication (Check-list of North 
American Birds) are recognized by the Service, scientists, and 
scientific organizations throughout the world as authorities on avian 
taxonomy in North America. Although the AOU has formally published its 
positions on the taxonomy of the California gnatcatcher and coastal 
California gnatcatcher (American Ornithologists' Union 1957, American 
Ornithologists' Union 1989), the Service, nonetheless, made a concerted 
effort to solicit and receive the recent, unequivocal, expert opinion 
of the Committee and its members. During a past, prescribed public 
comment period, the Service received responses from four members of the 
Committee (including the Committee chair). The Committee members were 
unanimous in acknowledging that Polioptila californica californica is 
currently accepted as a distinct subspecies and that its southern 
distributional limit occurs at 30 deg. north latitude.
    In addition to independently seeking and reviewing the best 
scientific information available from expert sources pertaining to the 
taxonomic status of coastal California gnatcatcher, the Service also 
repeatedly solicited comments or suggestions from the public, other 
concerned governmental agencies, the scientific community, industry, 
and any other party interested in gnatcatcher taxonomy and all other 
aspects of the listing decision. In [[Page 15699]] response to these 
requests for comments, the Service received a wide variety of public 
comments and opinions, which are discussed earlier in this notice.
    The Service has carefully considered all public comments received, 
separate and independent analyses of Atwood's data, the National 
Biological Service's (Link and Pendleton in litt. 1994) analysis of the 
data, subsequent review of all technical submittals from the National 
Biological Service (Newton in litt. 1995) and other interested parties, 
the existing scientific literature, and the information presented in 
the final listing rule designating the gnatcatcher as threatened (58 FR 
16742). As a result, the Service concludes that the taxonomy and 
geographic limits of the coastal California gnatcatcher are as provided 
by Grinnell (1926, 1928) van Rossem (1931), American Ornithologists' 
Union (1931), Grinnell and Miller (1944), Friedmann (1957), American 
Ornithologists' Union (1957), Paynter (1964), Garrett and Dunn (1981), 
Atwood (1991), and Phillips (1991). All of these scientific, peer 
reviewed, publications present conclusions or affirmations that the 
gnatcatcher (Polioptila californica californica) is restricted to 
coastal southern California and northwestern Baja California, Mexico, 
from Los Angeles County (and formerly Ventura County) south to the 
vicinity of El Rosario at about 30 deg. north latitude.
    The Service determines that the coastal California gnatcatcher is a 
distinct taxon and that its geographic range is that described and 
considered in the final listing rule for the coastal California 
gnatcatcher (58 FR 16742). Therefore, the coastal California 
gnatcatcher shall remain classified as a threatened species for reasons 
that are stated in the final rule to list the species (58 FR 16742).

References Cited

    A complete list of all references cited herein is available upon 
request from the U.S. Fish and Wildlife Service, Carlsbad Field Office 
(see ADDRESSES above).
List of Subjects in 50 CFR Part 17

    Endangered and threatened species, Exports, Imports, Reporting and 
recordkeeping requirements, and Transportation.

    Authority: The authority for this action is the Endangered 
Species Act of 1973, as amended (16 U.S.C. 1531 et seq.).

Authors

    The primary authors of this notice are Loren R. Hays of the 
Carlsbad Field Office (see ADDRESSES section), and Karla J. Kramer of 
the U.S. Fish and Wildlife Service, Portland Regional Office, 911 
Northeast 11th Ave., Portland, Oregon 97232-4181 (telephone 503/231-
6131).

    Dated: March 22, 1995.
Mollie H. Beattie,
Director, U.S. Fish and Wildlife Service.
[FR Doc. 95-7549 Filed 3-23-95; 10:39 am]
BILLING CODE 4310-55-P