The biology and relationships of the Torrent PAUL A. JOHNSGARD 1 Summary On the basis of field observations of five of the six described forms of Torrent , it is con­ cluded that only a single of Merganetta should be recognized, and that probably no more than three exist, armata armata, colombiana and leucogenis; turneri, garleppi and berlepschi are believed to represent variants of leucogenis. The Merganetta should either be placed in the perching-duck tribe Cairinini or given separate tribal status and placed adjacent to the Cairinini. Although the Merganetta behavioral and ecological findings of the armata occupies a vast geographic range trip; the taxonomic problems involved are extending some 5000 miles from the too great to allow anything more than cordilleras of northwestern to preliminary conclusions at this time. This Tierra del Fuego, the species’ ecological paper follows the taxonomic interpretation distribution is limited to the torrential of Delacour and Mayr (1945) and Delacour streams that originate in the high Andes (1954), which assumes a single species of and flow downward into the major river Merganetta. It should be noted, however, systems of both the Atlantic and Pacific that earlier authors accepted as many as drainages. Its altitudinal distribution varies five (Phillips 1926) or six (Salvadori 1895) with latitude and local conditions, but the species. Conover (1943), who undertook upper limits appear to occur at elevations the only thorough specimen analysis to where the streams become too small to date, concluded that three species (colom­ provide adequate foraging opportunities, biana, leucogenis and armata) and three ad­ while the lower limits are probably deter­ ditional subspecies (of leucogenis) should mined by diminishing stream gradients, be recognized. He based these conclusions since the larger and slower rivers lack the on the considerable differences in the male- falls, rapids, and emergent rocks which are patterns and on the well-marked the heart of the Torrent Duck’s habitat geographic separation of the three major requirements. It is not surprising, there­ population groups, although the latter fact fore, that few biologists have seriously would today be regarded as an argument in attempted to study this remarkable , favor of considering the forms as sub- for in most parts of its species. Phillips and Salvadori regarded favored habitats can be reached only with the Torrent Duck as a unique form, difficulty, and even under the best of comprising the subfamily Merganettinae conditions the are never abundant together with the genera Hymenolaimus and are usually extremely wary. As a and Salvadorina (= Anas waigiuensis). result, much uncertainty has existed Delacour and Mayr (1945) separated the concerning the biology of the Torrent genus in a monotypic tribe Merganettini, Duck, not only as to its ecology and but Niethammer’s (1952) anatomical study behavior, but also regarding the evolu­ convinced Delacour (1954) that Merganetta tionary relationships of Merganetta to other should be included in the dabbling-duck and even as to the number of tribe Anatini. More recently, however, taxonomic subunits (species or sub- Woolfenden (1961) has urged that a tribal species) within the genus. distinction for Merganetta be retained. Because of these several uncertainties, a trip to South America was made for the Appearance primary purpose of investigating as many In life the Torrent Duck is a singularly as possible of the six described populations beautiful bird, without close similarity to of Merganetta, comparing details of any other anatid with the possible exception appearance, behavior and ecology, and of the Salvadori’s Duck Anas waigiuensis. attempting to resolve at least some of the Both species are slim-bodied and long­ more obvious deficiencies in our know­ tailed but, while the Torrent Duck’s ledge of the Torrent Duck. This trip, from rectrices are distinctly stiffened and slightly 4th July to 28th August, 1965, included decurved and are used for support when stops in , , and Chile, climbing on slippery rocks, this adaptation where five of the six populations (all but is lacking in the Salvadori’s Duck. How­ the Argentine Torrent Duck) were ob­ ever, both species doubtless use their long served. This paper is a summary of the tails for underwater maneuvering.

1 Studies (No. 374) from the Department of Zoology and Physiology, University of Nebraska, Lincoln. Drawings by the author. Photographs in Photographic Section I.

66 THE WILDFOWL TRUST Another unusual feature of the Torrent Peruvian race (M. a. leucogenis) is more Duck which is barely indicated in Salva­ distinctly striped below, although at least dor’s Duck is the presence of a bony spur in eastern Peru near Lake Junín there is at the carpal joint, or ‘wrist’. Adults of distinct variation in the degree of mantle both sexes exhibit this feature, although the and underpart patterning (Phillips 1926). spurs are longer on adult males. Unlike In the Cuzcan Andes the Turner’s Torrent the similar spurs of the Spur-winged Duck (M. a. turneri) reputedly has the most Plectropterus gambensis they are not exten­ uniformly dark body of all the races, with sions of the radial bone (Rand 1954), but black flanks and breast, and black mantle rather are somewhat flexibly attached to feathers that are brown-edged. However, the first metacarpal. The use to which of at least eleven different males seen by these spurs are put is still questionable, me in the Cuzco area (three skins, eight or since through its habitat preference the more different wild males observed closely), Torrent Duck is well protected from most only one bird was almost as dark as the type terrestrial predators, and no intraspecific specimen (illustrated in Phillips 1926). fighting among males has been noted to my The remaining birds varied greatly, but knowledge. There is no evidence that the most of them could not be readily dis­ spurs are used as an aid in climbing rocks, tinguished from leucogenis males (see as has been suggested. photo section). These variations in bodily Females of all the populations are similar melanism must certainly have minor in appearance, and are a rich ochraceous genetic significance, since the darkest male brown below the gray above, totally observed was found in Huarocondo canyon, lacking the barring of typical female where at least four other lighter-colored dabbling ducks. The Colombian race males were resident in a six-kilometer river differs from the others slightly in that the distance. In Bolivia (M. a. garleppi) the gray of the head and neck does not extend males likewise very closely approach typical so far downward, and the body color may leucogenis, supposedly differing from it in be slightly paler. When swimming, females having white, rather than brownish, edging are often very difficult to see, since only the to the mantle feathers. However, this gray dorsal surface is visible, but when distinction did not apply to four males perched on a rock the richly toned brown observed closely (one of which was col­ flanks and underparts are most con­ lected) on the Río Zongo. Although I did spicuous. As in males, the bright-red bill is not observe the Argentine race (M. a. also highly conspicuous. berlepschi), it is known that males are highly Males, unlike the females, differ variable in their underpart coloration markedly in different populations, and (Dabbene 1927). This race is also reported these plumage variations have been the to possess white edging on the mantle basis for taxonomic discriminations. In all feathers, which, if true, would help to populations, however, the head patterning distinguish it from leucogenis, and suggest is very similar; a black ocular stripe some affinities with the Chilean race (M. a. extends back from the eye to the occipital armata), although there is a considerable region where it branches and continues geographic gap between these populations down the sides and back of the neck. (Johnson 1963). This last race is the most This stripe does not actually merge with distinctive of all, with definitely white- the crown stripe that extends back from edged mantle feathers, brown flanks, and the base of the bill (see photo section), black underparts extending up the throat although in the Chilean Torrent Duck the to the bill and eyes. In short, it would black striping extends down from the eyes appear that there exists a light-colored and merges with a black throat and neck. and fairly uniform northern population This general increase in melanism toward (colombiana), a melanistic and fairly uniform the south is accompanied by a darkening population (armata), and several highly of the legs , from a bright carmine similar variable intermediate populations with to that of the bill in the Colombian race to varying degrees of underpart melanism a dusky red, and a blackish culmen area and white to brown mantle edging. Con­ also appears on the red bill in the more over (1943) recognized this intermediate southerly forms. group as a distinct species (leucogenis) con­ Body of the males differ sisting of four subspecies, but considering greatly between populations, and have the individual variation I observed in the caused much taxonomic confusion. The Colombian race (M. a. colombiana) exhibits Cuzco area and reported elsewhere by the least melanism, having the breast and Dabbene and Phillips, I would suggest that underparts nearly a pure white, with only only a single race, leucogenis, should be light-gray striping, and the edging of the recognized. Besides the marked gap sep­ mantle feathers is a light brown. The arating this central group from the Chilean

TORRENT DUCK 67 race, there is probably also a gap in north­ rent Ducks were found between the ern Peru and Ecuador separating it from elevations of 1500 and 2500 meters on colombiana (Conover 1943). the Río Cauca, Río Chisbar, and Río This unusual degree of individual and Grande, near Popayán. However, Dr. F. subspecific male–plumage variation is Lehmann (pers. comm.) has observed unique among ducks, and deserves some them as low as 200 meters on the Pacific attention. It would seem that the absence slope of the Farallones mountains, near of closely related, sympatric forms would Cali. In Peru we observed M. a. leucogenis reduce selective pressures for male uni­ on the Río Lurín above Santa Cruz de formity to ensure species recognition by Luya, at approximately 2000 meters, and females, and would provide at least a partial Phillips (1953) found them on the nearby explanation for this variability. More Río Cañete at 2500 meters. Dr. Maria important, the fact that Torrent Ducks are Koepche (in litt.) has observed Peruvian confined in their movements within river Torrent Ducks at altitudes between 800 systems, and may be completely isolated meters (Río Chancay) and 3600 meters from adjoining populations by lack of (Quebrada Llnganuco, Cordillera Blanca). connecting river drainages, would tend to Near Cuzco we observed M. a. turneri result in much local genetic variation. along the Vilcanota, Urubamba River Finally, Torrent Ducks appear to be highly system from a point just above Yaucat sedentary and monogamous, with relatively (near Tinta, the type locality of turneri), at permanent pair bonds, further reducing 3386 meters, to a few kilometers above gene flow and increasing variability even Machu Picchu, at about 2040 meters. within potentially interbreeding popula­ Chapman (1921) reports Torrent Ducks in tions. the Urubamba valley at elevations of 9100 feet (2270 meters), 10 000 feet Environmental Conditions (3050 meters), and 11 000 feet (3350 As suggested above, water conditions are of meters). In Bolivia on the Río Zongo we critical importance in determining Torrent observed M. a. garleppi from a short Duck distribution. In every location where distance above Cuticucho station (2697 Torrent Ducks were found, the river could meters) to a point two kilometers above the be characterized as having rapids and falls lowest accessible point, Chururaqui station, (up to 20 feet high), interspersed with at 1830 meters. Niethammer (1953) states stretches of more placid water. The width that they are found between 1200 and of the river appeared to be of little im­ 3600 meters in Bolivia, and Phillips (1925) portance; ducks were observed on high reports that specimens have been taken as mountain streams no more than six feet high as 15 000 feet (4500 meters). In across (e.g. near Písac, Peru), as well as Chile, we found M. a. armata on the Río such streams as the Río Petrohué, which is Teno above Los Queñes, at 1200 meters, over 200 feet wide. All, however, had and also on the upper Río Petrohué, at numerous slightly submerged and emergent 160 meters. Mr. A.W. Johnson (pers. rocks ,which produced the falls, rapids, and comm.) informs me that there is a definite pools that appear to be of crucial importance latitudinal, altitudinal gradient of Torrent for Torrent Ducks. The surrounding Duck distribution in Chile, and that in the vegetation appeared to be of no signifi­ extreme north M. a. turneri occurs near cance; Torrent Ducks were observed in Arica (lat. 18°S) only at elevations above such diverse vegetational zones as the 4000 meters. The Chilean race occurs at moist alpine páramo of Colombia, the dry similar heights as far north as latitude 27°S, alpine puna zone of Peru, upper and lower but that from that point south the birds montane rainforests of Colombia, coastal occur at gradually lower altitudes, even­ xerophytic desert of Peru, and temperate tually reaching close to sea level in the rainforest of Chile. provinces of Aisen and Magallanes. All of the rivers supporting Torrent As an indication of the typical stream Ducks were distinctly cold (the Río gradients that support Torrent Ducks, a Chisbar had an average temperature of few examples might be mentioned. In 12°C) and thus the water was capable of Peru, the Huarocondo canyon enters the retaining much of the oxygen to which it Urubamba valley above Ollantaytambo, and is exposed when passing over falls and the stream which flows through this canyon rapids. Such high oxygen tensions would supports a good population of Torrent clearly be of importance in supporting the Ducks. From the village of Huarocondo aquatic life upon which the Tor­ (3320 meters) to Pachar station (2,800 rent Duck is dependent. meters), a distance of 19 kilometers, there The altitudinal distribution of Mer­ is an average stream gradient of 27 meters ganetta appears to be as varied as its per kilometer. The middle third of the ecological distribution. In Colombia Tor­ canyon, from a point six kilometers below

68 THE WILDFOWL TRUST Huarocondo to approximately six kilo­ While swimming, the body is often totally meters above Pachar station, supported at submerged, with only the head and neck least five and possibly six pairs of Torrent visible as the bird moves up or downstream. Ducks during July of 1965, suggesting a When swimming upstream in very rapid maximum population of about one pair per water, both adults and young sometimes kilometer. Mr. John O’Neill (in litt.) ‘run’ over the water surface in the manner observed eleven Torrent Ducks in the of frightened fledgling ducklings. When lower thirteen kilometers of this canyon frightened, a typical maneuver is to swim in January, 1965, indicating a similar 10 or 20 yards upstream from the observer, population at that time of year. then dive and be swept downstream by In Bolivia, the Río Zongo provided the current some 50 yards or more, finally another basis for estimating stream gradi­ emerging and flying away or continuing ents suitable for Torrent Ducks. Between downstream with only the head above Cuticucho station and Chururaqui water. Although this is a typical adult station there is a linear distance of nine response, we also saw two groups of downy kilometers, with an average stream gradient young, no more than two weeks old, of 96 meters per kilometer, or four times as perform the same tactic. In one case they steep as the Huarocondo location. In this maneuvered about 30 yards downstream distance two pairs plus three additional through two stretches of rapids before males were seen, but owing to the im­ emerging again. Although I was not able possibility of an adequate inventory of this to ascertain this point positively, adults precipitous and thickly vegetated canyon, appeared to hold their wings open slightly many additional birds may have gone when swimming under water, probably for unnoticed. steering purposes rather than propulsion. Torrent Ducks are particularly abundant In general, birds took flight when frightened in the lake district of southern Chile only as a last resort, usually flying no more (Johnson 1963). We observed at least four than a few hundred yards before landing males in a two-kilometer stretch of the Río in the water again and generally hiding Petrohué, between Lake Esmeralda and among the shoreline rocks. In three cases the first extensive rapids. The Río Petro­ where pairs with downy young were hué drops from an initial height of 180 found, the male conspicuously flew away meters to sea level in approximately 40 from the group, often calling loudly. The kilometers, with an average gradient of females always stayed near the young and 4.5 meters per kilometer. However, Tor­ led them to shore where they hid among rent Ducks are probably restricted to the the rocks. upper portion of the river, where the One remarkable and previously un­ gradient is considerably higher than this. noticed escape device was observed that Mr. Johnson tells me (pers. comm.) that warrants description. In the Río Chongo, the largest number of Torrent Ducks he above Pisac, Peru, a pair had been under has ever seen was at Los Queñes, where in observation for several days and it was the month of August he observed seven believed that the female might be nesting. males and four females in an approximate On one afternoon the male was observed ten-kilometer river distance. foraging alone in the river several hundred In summary, it would appear that Tor­ yards above the presumed nest site. At our rent Duck populations rarely exceed one approach he began to swim downstream, pair per kilometer, and usually would be then turned and, swimming underwater, much less than this. went rapidly back upstream to the base of a small waterfall about 2 feet high, and General behavior disappeared from view in the spray of the By far the most impressive and unforget­ fall. The river was sufficiently shallow to table aspect of Torrent Duck behavior is see that he had not gone back downstream, their incredible ability to negotiate the and we concluded that he must be hiding most impossible rapids, making headway at the base of the fall. To find out we upstream against an overpowering current, threw several large rocks into the water or turning and careening downstream near the point where he disappeared, but through the tumultuous rapids, barely to no effect. We were about to abandon the avoiding the rocks and nearly disappearing search when the male suddenly appeared from sight amid the spray and froth. I have on a rock at the base of the fall, saw us, observed adults dropping over falls several and disappeared again. This time we feet high, particularly when they are approached the fall, threw several more attempting to escape danger. When not rocks into the spray, and even reached disturbed, however, they usually stop back into the recesses of the rocks as far as short of such falls, climb out on a rock, then possible. Still failing to dislodge the bird, fly down to the next stretch of calmer water. we began to walk away, whereupon the

TORRENT DUCK 69 male emerged flying from the very spot we diving the bird usually drops headfirst had searched, moving downstream some directly into the current in front of or 150 yards to the next waterfall of similar beside the rock, and completely disappears size. It landed just above the brink, swam from view until it suddenly appears over it, and again disappeared in the white scrambling up the smooth rock surface water below. Later, in Bolivia, we observed near the point where it entered the water. a female with young hide in exactly the In Chile on the Río Petrohué some same manner after the young bird had measurements of diving and resting times retreated from us to the rocks on the oppo­ were taken of a male foraging in the fastest site shore. In this case she remained rapids of this then-raging river. During a hidden for over half an hour before she period of 3 minutes and 22 seconds the finally emerged and flew downstream to male made seven dives, ranging in length her mate, who had remained in sight dur­ from 12 to 18 seconds (average 16.0), ing most of this time. and rested on a rock for seven periods One difference in general behavior of ranging from 9 to 18 seconds (average the Torrent Duck that sets it apart from 12.6). I estimated the surface velocity of such stream-dwelling species as the Salva­ the river at that point to be approximately dori’s Duck and the Harlequin Duck 10 feet per second; thus, a bird simply Histrionicus histrionicus is the fact that maintaining its submerged position in the Torrent Ducks do not ‘nod’ the head water for 16 seconds would have to per­ when swimming or walking as these form the equivalent effort of swimming species typically do. Furthermore, unlike 160 feet underwater! The repeated per­ the Salvadori’s Duck, the Torrent Duck formance of this feat would appear to be rarely raises its tail from the water and almost impossible, and thus I believe that never cocks it vertically during normal during these dives the bird must go directly swimming. When standing on a rock and to the bottom, foraging upstream along watching some object in the distance or the river floor where the water velocity is about to dive into the water, Torrent greatly reduced, and then returning to the Ducks do frequently move their heads up surface at nearly the exact point of entry. and down, or from front to back, possibly Three other methods of foraging were as a distance-estimating technique. These observed. At times the birds would swim may also serve as diving intention move­ on the surface, with only the head sub­ ments. merged and directed upstream, apparently Torrent Ducks take flight without any picking up food as it was swept past in the obvious pre-flight movements or calls, and current. An extension of this method was can spring out of a rapid torrent with the to be almost totally submerged except for agility of a teal. They fly low over the the tail, feeding in the pools and eddies water, rarely higher than 20 feet, and behind large rocks. Finally, a bird would methodically follow the course of the river. sometimes swim to the foot of small fall a While most flights are of short duration foot or two high, crawl behind the main and only long enough to put them out of flow of water, and probe among the rocks the observer’s view, on one occasion we with the bill as the water flowed over its flushed a pair that must have flown at least back and head in an almost unbroken sheet. three kilometers. This occurred on the Río It is clear that Torrent Ducks rely largely Lurín, near Lima, and as we followed the on food of animal origin, judging from birds upstream, we met several groups of published records. Niethammer (1952) natives who informed us that they had seen noted that three Bolivian and one Peruvian the birds fly past. Finally, some three Torrent Duck he examined contained kilometers upstream, we abandoned the almost entirely stonefly and caddisfly search. larvae. A nesting female Colombian The wingbeats of Torrent Ducks in Torrent Duck collected by Holman, near flight are unusually shallow and rapid, Cali, contained caddisfly larval cases, and although the speed of flight does not appear three Peruvian Torrent Ducks collected to be great. On the basis of some short by Lord W. Percy near Lake Junín con­ motion picture scenes, the rate of wing tained primarily larvae of caddisflies beats during sustained flight is about and mayflies, with only a trace of vegetable twelve per second. materials (Phillips 1926). Pemberton (in Phillips) reported that in southern Argen­ Foraging behavior and food tina the Torrent Ducks ate ‘mollusks The foraging of Torrent Ducks is no less including gastropods’, while Conover remarkable than their escape behavior. found that two collected in the same Typically the birds forage from a large, general area contained and their rounded rock near the middle of a stream, larvae. from which they repeatedly dive. When Although it has not been reported in the

70 THE WILDFOWL TRUST literature, it is possible that at least in the Cuzco area occurs during the dry some areas small fish may contribute to the season. diet of Torrent Ducks. Dr. F.C. Lehmann In Bolivia, Niethammer (1953) reported tells me that biologists in Bolivia report collecting a pair of Torrent Ducks with considerable damage to planted trout enlarged gonads near Pojo in early August, fingerlings by Torrent Ducks. In Peru leading him to conclude that the ducks we were told by a local hunter who knew breed prior to the rainy season, when the the birds well that one male he had shot in water is clear and not flowing too swiftly. March had a full crop of ‘sardine-sized’ Our observation of a pair on the Río fish. I examined the male we collected on Zongo with a single downy young (two the Río Zongo, and found that the gizzard or three weeks old) on 3rd August, supports contained only well-ground vegetable this view, although the male we collected matter and a few very small stones. only ten kilometers upstream had inactive It is worth noting that, except in Chile, testes. we never failed to find the White-capped It is clear that in Chile and southern Dipper Cinclus leucocephalus on every seasonal temperature variations stream where we found Torrent Ducks. are of greater importance than farther Indeed, the best ‘indicators’ of Torrent north, and that breeding occurs during Ducks proved to be the presence of Dippers the southern spring. Johnson (1963) found and white droppings on the larger rocks in a nest in an old burrow of the Southern the rivers. Since the quantity of droppings Ringed Kingfisher Megaceryle torquata produced by Merganetta is considerably stellata with three nearly fresh eggs at greater than that of Cinclus, it was a simple Llifén, on 11th November. He has also matter to distinguish them. In Chile, observed (pers. comm.) broods somewhat however, we were south of the Dipper’s later than this, usually numbering two or range, and the winter rains quickly washed three young, but as many as five (twice). away droppings on the rocks, so that we Therefore, the nesting season in Chile were not able to use these clues. In any occurs at the end of the wet winter, and case, it is probable that Dippers and young emerge during the mild and Torrent Ducks have nearly identical relatively dry spring. habitat requirements and probably are Judging from our observations of four dependent on the same sources of food. broods in Colombia, Peru and Bolivia, no specific brooding habitat requirements Breeding Seasons exist. However, in two cases the brooding Although very few Torrent Duck nests have area was bounded below by fairly steep been found (two of the Chilean race and canyons and high falls, which perhaps two of the Colombian race are known to represented the original nesting sites. me), it is possible to make a few observa­ Likewise, the presumed nesting area near tions about breeding periodicity. On the Pisac, Peru, was a steep-walled canyon Río Chisbar near Popayán, 11th July, we with falls and rapids both above and below. found a pair with two young less than a Both of the broods found in Huarocondo week old. We were told by natives that two canyon were seen on or near low, rocky other pairs with young of similar age were islands on which shrubs and low trees seen a few days previously on the Río were growing; presumably these islands Blanca, near Paletara. Conover (1943) served as relatively safe resting and sleeping reports downy young of this race taken in areas. October, eggs taken in November, and immatures taken in January, March, April Vocalizations and September. Thus, it is clear that the Previous accounts of Torrent Ducks have breeding season in Colombia is a very offered Little information regarding vocal­ extended one, as might be expected near izations, which is not surprising consider­ the equator. ing the problems of hearing calls above the In the Cuzco area of Peru, Chapman noises of rushing water. Scott (1954) (1921) reported a pair with two downy reported a call, described as a ‘keech’, young in Huarocondo canyon in late uttered by Bolivian Torrent Ducks re­ July. On 26th July we located two pairs peatedly when the birds were nervous, and with downy young in this same canyon. Niethammer (1952) mentioned a shrill One of the broods, with two young, was whistle uttered by the male and possibly estimated to be no more than two weeks also the female. old, while the other brood of three appeared We heard vocalizations from males of slightly older. We also had reason to believe the Colombia, Turner’s, Bolivian and that a female on the Río Chongo, near Chilean races. In all these races the calls Pisac, was nesting at this time. These appeared identical, and most commonly sightings would suggest that breeding in consisted of a sharp, clear whistle, dropping

TORRENT DUCK. 71 slightly in pitch toward the end. This was leucophrys, Brazilian Teal Amazonetta uttered at several second intervals by males brasiliensis and Chiloe Anas to other males, their mates, or their sibilatrix. families, apparently as a warning note. It There is little opportunity for contact was usually uttered as the bird stood on a between Torrent Ducks and other water­ rock, but was also produced by swimming fowl under most conditions. We observed and flying birds. On one occasion, on the at least three pairs of Sharp-winged Teal Río Chongo, the male swam up the canyon Anas flavirostris oxyptera in Huarocondo in which we believed his mate to be nesting, canyon and the adjacent Urubamba River, uttering a series of eight to ten soft and occupying relatively calmer stretches of rapidly repeated whistling notes. At this these rivers. In southern Chile there may time the female remained hidden, but on be occasional contacts with the Bronze­ the next day when the same call was given, winged Duck Anas specularis, although we the female emerged from the rocks above did not encounter this species. the river and joined the male, then both flew downstream together. Sexual behavior and displays We did not hear any definite calls from The published observations of Phillips females, but Mr. H. Lüthi (pers. comm.) (1953), Scott (1954) and Johnson (1963) heard an unusual sound from a group of comprise the sum total of previously four Torrent Ducks (two pairs) in May. available information of Torrent Duck This was a rough, guttural sound, pro­ displays, and these three accounts show longed and on the same pitch, which may remarkably little similarity. Undoubtedly, well have originated from the females. the fact that pair bonds are relatively Mrs. A.W. Johnson (pers. comm.) permanent largely accounts for the rarity remembers that when they captured a with which display has been observed in nesting female it uttered a high and strange this species, and thus it might be expected squealing note, quite different from the that most displays serve in maintaining, usual calls of female ducks, and that only rather than establishing, pair bonds. The once did it produce anything resembling a absence of sympatric, closely related species ‘quack’. Mr. Johnson also recalls hearing also reduces the need for elaborate pair­ a ‘weak quack’ from this bird. forming displays. The ‘lost’ calls of the downy young, Of the available accounts, Scott’s (1954) which we heard on several occasions, are appears to be the one which most probably single-noted peeping whistles, much like represents pair-forming display in Torrent the distress note of downy Mallards Anas Ducks. His observations of the Bolivian platyrhynchos. Torrent Duck, made in late March, were of a group of two females and three males. Pair bonds and social behavior The male displays he describes show no In contrast to most ducks, the Torrent similarities to those of typical dabbling Duck does not appear to be gregarious and ducks or any other anatids, but such has never been reported as occurring in singularity in male pair-forming displays flocks. This solitary nature may simply be might be expected. a reflection of limited food supplies and the Phillips’ (1953) accounts of display in sedentary tendencies of the species, al­ the Peruvian Torrent Duck were obtained though some altitudinal movements may during January on the Río Cañete. He occur at various seasons. We observed describes seeing a male on a rock ‘bowing birds almost exclusively in pairs or deeply and at the same time pushing the presumed pairs; the only exception occur­ tip of its tail upward to a height exceeding red in Chile, where no females were that of its vertically extended head and observed. However, according to Johnson neck. The entire action appeared to be (1963), female Chilean Torrent Ducks pendulum-like, with the feet serving as become much less conspicuous in winter inverted fulcrums. After each such four- or than males. five-second performance, it would resume There can be little doubt that Torrent its normal stance which was marked only Ducks are monogamous and have relatively by an occasional turn of the head. It would permanent pair bonds, which would then advance 6 to 12 inches and perform explain the greatly extended breeding again.’ A female was later found to be season in northern South America, and standing on part of the same large rock. might possibly allow for more than one Wright (1965) has recently reported on brood per year. This prolonged pair bond some apparently aggressive behavior he would also account for the lack of an observed in Peruvian Torrent Ducks. He ‘eclipse’ plumage in males. Other South describes two drakes facing one another in American species which are similar in this a stiffly upright posture, with the body, latter regard are the Callonetta neck and head all held nearly vertically.

72 THE WILDFOWL TRUST While in this upright posture the males On 19th August, on the Río Petrohué in rhythmically bobbed their heads and Chile, I observed what clearly was the remained about half a meter apart, in the same display. In this case a male per­ presence of a single female. One male formed the display apparently to another finally flew away, while the other remained male which had just flown off the rock. In with the female. this instance the male preceded the display My own observations of display were by standing very erect and nearly motion­ made on the Colombian and Chilean less for several seconds, and followed the races, and are remarkably similar to those of display with the same rapid wing-shaking Phillips. On 8th July on the Río Grande, movement that was observed in the we watched a pair of Torrent Ducks forag­ Colombian race. In both instances the ing some 100 yards away. The male soon distances were too great to photograph the climbed a large rounded rock, and was activity or to determine whether any calls shortly joined by the female who stood were associated with it. It is significant, within 2 feet of him and facing him. She however, that the same display should stood quite erect, with her tail propped on occur in the three populations which the rock, and began to perform rapid represent the most morphologically dis­ vertical bill movements toward him. The tinctive forms, indicating that the taxo­ male, also standing erect and facing the nomic interpretation of a single species female, suddenly began a series of in­ of Merganetta is a reasonable one. creasingly strong downward and forward Although no female was definitely thrusts with his head, each one more present in the last encounter mentioned, pronounced than the one before, until his I believe that this is basically a pair­ bill nearly touched the rock, and with each maintaining display, and thus is probably thrust raising his hindquarters and cocking a basic and important display in the his tail (see accompanying sketch). After species. It is similar in form to the mutual

about six of these fairly rapid, teeter-totter displays of such perching ducks genera as movements the bird assumed a normal Cairina and Pteronetta, although in these posture, appeared to shake his wings, and forms there is no tail-cocking associated repeated the sequence. At least four such with the head-thrusting movements (Johns- sequences were performed in a half minute gard 1965). In these species the display or less. The female responded to the male occurs in ‘triumph ceremony’ situations as with much less conspicuous vertical head well as during aggressive encounters. movements. Both birds then jumped into Of all aspects of behavior, none is more the water, swam downstream through uniform within taxonomic groups and several series of rapids, and resumed more indicative of evolutionary relation­ foraging, A few days later a less elaborate ships than that associated with copulation. version of the same display was observed Although I did not witness such behavior, in another pair. In this instance as a male a detailed account has been provided by approached a resting female on a large Johnson (1963) for the Chilean Torrent rock, he made several ‘bowing’ movements Duck. Precopulatory behavior consisted without cocking the tail, then sat down of the two birds swimming round each beside her and tucked his bill under the other, making repeated bill-snapping and scapulars. thrusting movements of the head, as if

TORRENT DUCK 73 catching flying insects, and mutually rising recognized I would favor including in the water in a grebe-like manner. Similar Merganetta in the perching-duck tribe mutual thrusting movements occur during Cairinini. precopulatory display in some perching The fact that at least the Chilean, ducks ( galericulata in particular), Peruvian and Colombian forms of Torrent but the grebe-like posturing appears to be Ducks possess identical head-thrusting and unique among the Anatidae. tail-cocking displays may be regarded as evidence supporting the view that only a Evolutionary relationships and single species of Merganetta should be taxonomic conclusions recognized. Further, because of the un­ It appears that Torrent Ducks have little usually great male-plumage variation or nothing in common with typical dabbling occurring in Peru, Bolivia and northern ducks in their sexual-behavior patterns, Argentina, I believe that the races turneri, and may more closely approach the perch­ garleppi and berlepschi should not be ing ducks in this regard. Some aspects of recognized, and that the entire population their general behavior, such as their from northern Peru to the Tucumán area hole-nesting, apparently reduced female of Argentina should be regarded as M. a. vocalizations, and obvious climbing and leucogenis until a more thorough study of perching abilities might also suggest these regions can be made. This pro­ affinities with the perching ducks, but the cedure would also avoid the present possibilities of evolutionary convergence inexplicable situation of four alternately must be carefully considered here. light-bodied and dark-bodied races geo­ Niethammer’s (1952) conclusion that graphically replacing one another in a the sternum and trachea of Merganetta north-south direction. indicate dabbling-duck affinities appears to be an oversimplification, since Woolfen- Acknowledgements den (1961) reports numerous unique This study was financed by a National skeletal features of the genus which are not Science Foundation grant (GB 1030) for suggestive of this relationship. The tracheal research on waterfowl systematics. Numer­ bulla has the same general configuration as ous persons provided me with invaluable that found in the dabbling ducks, perching assistance and facilitated what would ducks, and , but it is no more otherwise have been an impossible study. similar to that of Anas than, for example, I am especially indebted to the following: is that of Pteronetta hartlaubi. Sr. Hernando Acevedo Q., Mr. John If the distinct morphological similarities Ardell, Dr. Ismael Ceballos B., Dr. F.C. between Torrent Ducks and Salvadori’s Lehmann, Mr. A.W. Johnson, Dr. and Ducks are simply the result of con­ Mrs. H.W. Koepcke, Mr. H. Lüthi, and vergence, then inclusion of Merganetta in Dr. N.G. Smith. Finally, I wish to thank the tribe Anatini is certainly not warranted. Mr. Dale Funk for accompanying me as Unless a separate tribe Merganettini is interpreter and field assistant.

References Chapman, F.M. 1921. The distribution of bird life in the Urubamba Valley of Peru. US Nat. Mus. Bull., 117. 138 pp. Conover, B. 1943. A study of the Torrent Ducks. Zool. Series, FieldMus. Nat. Hist., 24:345–356. Dabbene, R. 1927. Distintas fases de coloración del plumaje en la Merganetta del noroeste de al Argentina. El Hornero, 4: 34–38. Delacour, J. 1954. The Waterfowl of the World. Vol. 2. Country Life, London. 232 pp. Delacour, J. and E. Mayr. 1945. The family Anatidae. Wilson Bull., 57: 30055. Johnsgard, P.A. 1965. Handbook of Waterfowl Behavior. Cornell University Press, 378 pp. Johnson, A.W, 1963. Notes on the distribution, reproduction and display ofthe Andean Torrent Duck, Merganetta armata. Ibis, 105: 114–116. Niethammer, G. 1952. Zur Anatomie und systematischen Stellung der Sturzbach-Ente (Mer­ ganetta armata). Jour. für Onr ithol., 93: 357–360. Niethammer, G. 1953. Zur Vogelwelt Boliviens. Bonn. Zool. Beiträge, 4: 195–303. Phillips, J.C. 1926. A Natural History of the Ducks. Vol. 4. Houghton-Mifflin Co.,Boston, 489 pp. Phillips, S. 1953. An incident concerning the Peruvian Torrent Duck. Avicult. Mag., 59: 134. Rand, A.L. 1954. On the spurs on birds’ wings. Wilson Bull., 66: 127–134. Scott, P. 1954. Behaviour of the Bolivian Torrent Duck. Wildfowl Trust Sixth Ann. Rep., 69–72. Woolfenden, G.E. 1961. Postcranial osteology of the waterfowl. Bull. Florida State Museum, Biological Sciences, 6(1): 129 pp. Wright, J.K. 1965. Observations of behavior of the Andean Torrent Duck. Condor, 67: 535.

74 THE WILDFOWL TRUST TORRENT DUCKS Merganetta armata (See pages 66 to 74). With the exception of the upper photograph on the next page, taken by and reproduced with the permis­ sion of H. Luthi, the photographs in this section were taken by Paul A. Johnsgard. Their poor technical quality is regretted. (Above) 31 male and (below) 32 female Colombian Torrent Ducks, Río Chisbar, Popayan. (Above) 33. Two male and a female Peruvian Torrent Duck, Río Lurín. (Below) Tur­ ner’s Torrent Duck, Huarocondo Canyon. (Left) 34. A light-colored male. (Right) 35. A downy young. (Above) 36. Male Bolivian Torrent Duck, Río Zongo. (Below) 37. Male Chilean Torrent Duck, about to dive, Río Petrohué. 38, 39. Chilean Torrent Ducks.