36 



P.C. RASMUSSEN AND N.J. COLLAR 



the two. PCR examined all three of Sane's birds in the space of two 

 days and concluded that they lacked salient differences, all being 

 apparent krameri x cyanocephala hybrids. 



Immature Specimen 



AMNH 62 1545, considered to be a femaleintermedia by Rothschild 

 (1907), but thought by others to be an immature himalayana (Biswas 

 1959, 1990, Forshaw 1973), shows no relevant plumage differences 

 from the series of immature himalayana at AMNH with which we 

 directly compared it, nor from others at USNM and other museums 

 with which photos of it were compared. In a PCA of several 

 measurements, AMNH 621545 falls within the 95% confidence 

 limits of immature himalayana of both sexes (Figure 1 , component 

 loadings in Table 3, summary statistics in Appendix). In body 

 plumage 621545 resembles himalayana in being generally cooler 

 green and less yellowish than immature finschii, although some 

 juveniles of the two species overlap in this. AMNH 621545 is unlike 

 juvenile hybrid cyanocephala x himalayana (see below) in its 

 bigger, duskier maxilla, brighter green nape, lack of yellowish 

 collar, bluer-green overall body colour, and especially in its bright 

 green upper tail surface with a bright yellow tip. Thus, on the basis 

 of both plumage and measurements, all evidence supports the 

 hypothesis that AMNH 621545 is an immature himalayana, and we 

 therefore exclude this specimen from further analyses. 



BMNH Specimen 



Comparison of photos and measurements of BMNH 1980.3.1 with 

 those of the adult male AMNH intermedia showed that the former 

 has several differences from all other intermedia, although it is part 

 of the Rothschild series, and despite the seemingly inexplicable fact 

 that the BMNH specimen was the one upon which Husain (1959) 

 based his conclusion that intermedia was a hybrid himalayana x 

 cyanocephala. This bird was therefore lent to AMNH for our 

 comparisons, where we confirmed (Table 1, Appendix) that it has a 

 smaller maxilla with only a slight reddish tinge basally (less than in 

 all AMNH birds except 62 1 542, the specimen said by Biswas [ 1 959] 

 to be completing post-juvenile moult); it has a slightly duller head 

 with paler reddish-purple on the face (washed yellowish in front of 

 the eye) and paler greyish-blue on the crown and nape; and it totally 

 lacks maroon shoulder patches. Its P3 is narrower and less squared 

 at the tip than in all adult AMNH specimens except 62 1 544. The tail 

 is greener at the base, more turquoise for most of its length, and has 

 the pale tip whiter and shorter. This specimen is the only one of the 

 Rothschild series that has a measurable, fully grown tail, so its 

 rectrix length cannot be directly compared with the otherintermedia. 

 The salient differences between BMNH 1980.3.1 and typical roseata 

 are: the former lacks reddish shoulder patches; it has an entirely pale 

 lower mandible (though this is nearly all-pale in a few roseata; 

 Table 4); its hindneck has a turquoise tinge; it has a slightly broader 

 tail tip; the front of its face is slightly redder; and its P3 tip is broader 

 (Table 1). In most statistical analyses, BMNH 1980.3.1 falls within 

 the roseata and cyanocephala groups (Figures 2-5). 



Remaining intermedia specimens 



The other five AMNH specimens attributed to intermedia (including 

 the holotype), and also Bhargava's three specimens, are quite similar 

 to one another. However, although most previous authors (Rothschild 

 1907, Hartert 1924, Husain 1959) have treated the first five under 

 one description as if they were identical, they are in fact variable in 

 most of the characters that separate them from any of the putative 

 parental species (Table 1). Only Biswas (1959) mentioned variation 

 among these five, but even he called them 'exceedingly similar'. All 

 have fairly large bills with varying amounts of orange at the base of 

 the maxilla. All have nearly or entirely pale lower mandibles, 



although AMNH 621544 has a broad black stripe down one side of 

 the lower mandible. Each has the front of the face bright purplish- to 

 deep pink, grading into the duller grey-blue crown, nape, and lower 

 portions of the face. All have a pale blue-green collar, but this is 

 highly variable in breadth and prominence, even allowing for differ- 

 ences in preparation. In addition, all have a bluish wash of variable 

 strength on the wing coverts and/or rump. Only three of the speci- 

 mens now have the central rectrices present, and in none of these 

 (contra Biswas 1959) are they fully grown (this cannot now be 

 determined in AMNH 621540, the fourth intermedia that once had 

 central rectrices, but the fact that they are now missing from this 

 specimen suggests they were loosely attached and thus moulting). 

 Thus, original tail lengths presented in previous treatments - Biswas 

 (1959): 185, 202, 221mm; Husain (1959): about 220 mm; Forshaw 

 (1973): 167-195 mm (mean = 180.7, n = 3), 206 (n = 1) - would be 

 expected to be too short. However, we measured the central rectrices 

 of the three specimens in which they are now mostly grown as 157, 

 200 and 170 mm, and the now-missing rectrices of AMNH 621540 

 were estimated at ca. 200 mm. The central rectrices of all four of 

 these birds show (or showed) long, at least moderately broad, pale to 

 pure yellow tips and dark or royal blue upper tail surfaces. The 

 breadth and length of the yellow tail tip are quite variable, however, 

 and the length of the yellow Rl tip of AMNH 621540 is estimated to 

 have been 48 mm, compared with a mean tip length of 41 .6 mm for 

 the others (Appendix). From the photograph of Bhargava's two 

 specimens for which the central rectrices are present (both photo- 

 graphed next to a cm rule), these rectrices appear to be ca. 234 and 

 229 mm (although it cannot be determined from the photos whether 

 these rectrices are full-grown), while the yellowish tips are ca. 51 

 and 54 mm, respectively. 



We found no external qualitative characters in AMNH intermedia 

 or Bhargava's specimens that differ from those exhibited by at least 

 one member of one of the two species groups (roseata/cyanocephala 

 and finschii/lumalayana), or that are not intermediate between them 

 (Table 4). Among the potential parental species, roseata exhibits the 

 most plumage features incompatible with the AMNH intermedia 

 phenotype, vj\\\\t finschii also has a few characters inconsistent with 

 intermedia, mostly in tail shape and colour. Neither cyanocephala 

 nor himalayana shows any plumage features incompatible with their 

 being parental species of AMNH intermedia, and a combination of 

 the former two readily explains all plumage features of the latter. 



Statistical results 



Summary statistics for measurements of the putative intermedia 

 specimens (with the BMNH specimen treated separately), 

 Sedgemore's hybrids, Sane's specimens, and comparative samples 

 of the five putative parental species are given in the Appendix. For 

 almost all measures, the AMNH intermedia are intermediate between 

 cyanocephala and adulthimalayana, and in many cases also between 

 others of the putative parental species. Bivariate scatter plots of 

 selected measurements overwhelmingly demonstrate this pattern, 

 e.g. Figure 2A showing culmen length from cere vs. culmen width, 

 in which allAMNHintermedia and Sedgemore's hybrids fall between 

 the cyanocephala/roseata pair and the himalayana/finschii pair. 

 Furthermore, krameri is larger than, and Sane's two specimens are as 

 large as, the himalayana/finschii pair. A slightly different pattern is 

 shown in Figure 2B (wing length vs. culmen length): here again, 

 AMNH intermedia and Sedgemore's hybrids fall between 

 cyanocephala/roseata and himalayana but, because of the shorter 

 wing of finschii compared with himalayana, there is slight overlap 

 between finschii and intermedia. Psittacula krameri is similar in 

 wing length to himalayana but is bigger-billed, and Sane's birds fall 

 between cyanocephala/roseata and krameri, being considerably 



