Voucher material (total = 2): Orosa (AMNH 74113), Río Manití (FMNH 112563).

Other interfluvial records: Río Yavarí- Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: Both specimens of nakedtailed armadillos that we examined from the Yavarí-Ucayali interfluve conform to the diagnosis of Cabassous unicinctus in Wetzel's (1980) revision of Cabassous. The Orosa specimen (AMNH 74113) was among those that Wetzel (1980) examined and assigned to the nominotypical subspecies (with type locality “Surinam”). According to this authority, C. u. unicinctus intergrades with a smaller subspecies, C. u. squamicaudis (Lund, 1843; type locality Minas Gerais, Brazil) in “the Amazon-Solimões River area in Brazil and Peru.” The material at hand is entirely inadequate to assess geographic variation in this species, but it is noteworthy that cranial measurements of AMNH 74113 (table 3) are much smaller than the mean values for C. u. unicinctus tabulated by Wetzel (1980: table 1), but they are close to the corresponding means for C. u. squamicaudis. However, the number of scutes on the cephalic shield of this specimen (N = 36) is much closer to the mean value tabulated for C. u. unicinctus (34.6 ± 2.2 SD; Wetzel, 1980: table 4) than it is to the mean value for C. u. squamicaudis (54.0 ± 5.5). The taxonomic significance of these conflicting indications is unclear, but the mean differences between Wetzel's (1980) subspecies are substantial, and a fresh assessment of the material he allocated to these nominal taxa would seem to be appropriate in any future taxonomic study of Cabassous.

The second known specimen of Cabassous unicinctus from the Yavarí-Ucayali interfluve (FMNH 112563) consists only of a live-mounted skin of a specimen “caught swimming in the river” (P. Soini's notation on skin tag); this specimen has 33 scutes on the cephalic shield. The skull, apparently, is lost.

Ethnobiology: The naked-tailed armadillo has only one name, menkudu, which is not analyzable and has no cognate in other Panoan languages, but which may contain the adjective kudu (“grayish” or “light-colored”). No subtypes are named by the Matses, although some informants claim that there is more than one kind.

This species is not eaten by the Matses, and is of no economic importance. Very rarely a hunter will kill one out of curiosity.

If a hunter encounters a naked-tailed armadillo, it may cause his children to fall ill, in which case a medicine man will try several different medicinal plants, some of which are said to be specific to contagions caused by this species (other plants are said to be general remedies for contagions caused by any type of armadillo). Adults can become constipated due to exposure to a naked-tailed armadillo, in which case there is a specific plant to cure this. As is generally the case with rarely encountered animals, coming across a naked-tailed armadillo is interpreted as an omen that someone in the (extended) family of the person who sees it will fall ill and die soon thereafter.

Matses natural history: The naked-tailed armadillo is like a small giant armadillo. It is small, the size of a nine-banded long-nosed armadillo (Dasypus novemcinctus). It is lighter in color than other armadillos. Its front claws are large, the central ones as large as tablespoons (an obvious exaggeration). Its back feet are like those of a long-nosed armadillo. They are hairier than other armadillos. The tail is hard, but not ringed.

Naked-tailed armadillos are seldom seen and apparently very rare. Many Matses have never seen one, but burrows are somewhat frequently found. Their burrows are usually found in primary forest.

Naked-tailed armadillos sleep in burrows. The burrow has a small diameter, barely wide enough for the animal to fit. The burrow is deep and points straight down, unlike those of other armadillos, and one can recognize unquestionably that a burrow belongs to a naked-tailed armadillo. One informant found a naked-tailed armadillo sleeping in a hollow log. No leaf beds have been found in burrows of this species.

The naked-tailed armadillo is nocturnal. It roots at the bases of trees and digs into rotten logs. It walks around on the surface of the ground but also travels underground. It is very strong and can dig very quickly and very deep. It escapes by digging a deep hole in the ground, covering itself with excavated soil as it digs. It expels the dirt when it surfaces, as if it had dug up to the surface from another location. It does not make or use paths (as dasypodid armadillos are said to do; see below).

One informant saw three individuals of this species traveling together.

There are no flies at the entrances to burrows of this species (as there are at the burrow entrances of other sympatric armadillos; see below).

Naked-tailed armadillos are said to grunt by some informants, but others say they make no sounds.

The naked-tailed armadillo eats earthworms, millipedes, grubs that live in the ground, and other invertebrates. (Most informants had no idea what it eats, but they assumed it eats the same foods as long-nosed armadillos.)

Remarks: Although sparse, Matses information about Cabassous unicinctus includes a few novel observations, notably that this species travels underground as well as on the surface.¹ The nocturnal activity reported by the Matses agrees with other rainforest observations (Emmons, 1997), but not with the diurnal activity reported from Brazilian savanna habitats (Bonato et al., 2008). The perpendicular orientation of burrows attributed to this species by the Matses is consistent with quantitative data on burrow characteristics summarized by Trovati (2015).

Voucher material (total = 1): Nuevo San Juan (MUSM 11091).

Other interfluvial records: Nuevo San Juan (camera-trap photograph), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: Adult giant armadillos cannot be confused with any other mammal, although juveniles might be mistaken for Cabassous unicinctus, which they superficially resemble. Our single voucher, of which only the skull is preserved, is a young adult male with a completely co-ossified occiput but unfused preoccipital sutures.

The only traits in which our specimen differs from the diagnosis of Priodontes maximus provided by Wetzel (1985b: 31) concerns the dentition, which he reports as consisting of “18/19 pairs [of teeth], which are very flattened transversely.” In MUSM 11091, by contrast, there are alveoli for only 16 upper teeth (T) on the left side, of which the first (T1) seems to have been a tiny cylindrical peg; T2–T10 are sharp, flat blades; and T11–T16 are worn, peglike elements. There are alveoli for 17 upper teeth on the right side of this specimen, of which T1 and T17 are missing but were obviously tiny pegs; T2–T10 are sharp, flat blades like their counterparts on the left side; and T11–T16 are worn and peglike. Therefore, the symmetrical dental complement of this specimen includes 16 paired teeth, of which only nine on each side conform to the flattened morphology described by Wetzel.

Ethnobiology: The principal name for the giant armadillo is tsawesamë, a term that can be analyzed as a compound of tsawes (“armadillo”) and amë (“father” or “large”). The archaic synonym is panu, the pan-Panoan term for “giant armadillo.” No subtypes are recognized by the Matses.

The giant armadillo is not eaten or killed by the Matses, and it is of no economic importance. However, members of the Mayú tribe (whom the Matses exterminated and from whom they captured many women) formerly ate giant armadillos.

The Matses believe that it is a death omen to encounter a giant armadillo as it is foraging during the day, or to find giant armadillo diggings in a swidden; such omens are interpreted to mean that someone (other than the observer) will soon die. Additionally, if one steps on dirt that an armadillo has dug up, one may become weak and thin. If a hunter sees or touches a giant armadillo, his children may fall ill. To cure the child's illness, a medicine man will try several different medicinal plants, some of which are specific to contagion by giant armadillos and others are general for contagion caused by any type of armadillo. A baby that suffers from giant armadillo sickness will dig its nails into its mother's flesh, and it may die from this condition.

Matses natural history: The giant armadillo is similar to other armadillos, but it is huge. It has very large claws on its front feet, like spoons. It bends its claws inward when it walks. The tracks of its hind feet look like the footprints of a (small) tapir. It has a large head similar to that of an armadillo. It has a very hard carapace. Its carapace has round bumps on the neck portion. It has less distinct bands and has lightly haired undersides.

Giant armadillos prefer primary upland forest, especially the area of headwater gullies, but their tracks can also be found in floodplain forest. (Not every Matses has seen one, though their diggings and tracks are commonly encountered.)

They make many very large holes. They do not make dens with leaf nests, but simply sleep in the large holes they have dug, usually in a stream headwater gully. They may fix up an old hole to sleep in for a night.

The giant armadillo is almost exclusively nocturnal. It does not make or use paths, although it may follow the same route occasionally. It travels far, visiting different streams. When it forages, it digs large holes, toppling saplings and removing roots. It leaves behind large piles of dirt and clay where it forages. It also forages by digging into rotten logs and at the base of swamp palms (Mauritia flexuosa [Arecaceae]), where there is black soil. Giant armadillos escape by digging into the ground, so that they are difficult to kill with a club.

Giant armadillos are solitary.

There are many white flies (probably phlebotomine psychodids) in the holes where a giant armadillo has slept.

The giant armadillo grunts when one touches it with a stick when it is in its burrow.

The giant armadillo roots in the ground to eat earthworms and grubs that live in the ground. It digs up rotten logs and stumps to eat armored millipedes (Barydesmus sp. [Platyrhacidae]) and termite larvae. It digs up termite nests to eat the larvae. It digs out beehives in hollow trees to eat their larvae and drink their honey. It digs out the pith of fallen rotting swamp-palms to eat beetle grubs. It takes apart the crowns of fallen isan palms (Oenocarpus bataua [Arecaceae]) to eat the beetle grubs that eat the heart of the palm. Giant armadillos eat isan palm fruits and swamp palm fruits, chewing up the whole fruit, including the seed.

Remarks: Matses observations about this species are not very extensive—understandably so, since giant armadillos are not hunted for food and are seldom encountered while active at night—but they include some information that does not appear in the scientific literature (reviewed by Carter et al., 2016), notably about foraging behavior and diet. By comparison with Matses accounts of sympatric dasypodid armadillos (see below), their remarks that Priodontes maximus does not follow paths or make leaf nests seem noteworthy.

Voucher material (total = 6): Nuevo San Juan (AMNH 268229, 268230, 268231; MUSM 11088, 11089, 11090).

Other interfluvial records: Actiamë (Amanzo, 2006), Itia Tëbu (Amanzo, 2006), Jenaro Herrera (Pavlinov, 1994), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: Our voucher material conforms to the qualitative descriptions of Dasypus novemcinctus provided by Wetzel and Mondolfi (1979) and Wetzel (1985b). Additionally, measurements of our voucher material (table 3) fall within the observed range of variation among Amazonian specimens of D. novemcinctus tabulated by Wetzel and Mondolfi (1979: table 1). Our single preserved skin (MUSM 11088) has eight moveable bands, of which the fourth has 56 scutes; both counts are well within the range of meristic variation for the species (Wetzel and Mondolfi, 1979).

Wetzel et al. (2008) recognized several subspecies of Dasypus novemcinctus but provided no phenotypic criteria for distinguishing them. In the absence of any critical revision of this implausibly widespread taxon (which ranges from the southern United States to Uruguay), it seems pointless to use trinomial nomenclature or to speculate about the validity of any nominal forms currently treated as subspecies or synonyms. However, it is noteworthy that (1) Amazonian specimens seem to be substantially larger than specimens from other South American landscapes (e.g., the Brazilian highlands and northern Venezuela; Wetzel and Mondolfi, 1979: table 1), (2) sequence variation at the mitochondrial ND1 locus among specimens identified as D. novemcinctus appears to be highly structured geographically (Loughry and McDonough, 2013: fig. 7.2), and (3) phylogenetic analyses of mitogenomes do not recover D. novemcinctus as a monophyletic taxon (Gibb et al., 2015; Mitchell et al., 2016). Unfortunately, western Amazonian populations of D. novemcinctus are not represented in any published molecular analysis.

Dasypus novemcinctus is easily distinguished from its sympatric congener D. pastasae by its smaller adult size (typically <6 kg, versus >8 kg in D. pastasae); by lacking a vestigial fifth digit on the forefoot (a tiny fifth digit is almost always present in D. pastasae); by the absence of enlarged, spurlike scales on the knee (present in D. pastasae); and by having rounded lateral palatine margins (fig. 4A).

Ethnobiology: The only general name for the nine-banded long-nosed armadillo is sedudi (an unanalyzable term with no other meaning), a word that is not found in other Panoan languages and which has no archaic or ceremonial synonyms. Three subtypes of the nine-banded longnosed armadillo are recognized by Matses hunters: sedudimpi (“small nine-banded long-nosed armadillo”), sedudidapa (“large nine-banded longnosed armadillo”), and akte tsawes (“water/stream/river armadillo”). It is notable that the term for the third subtype contains the term tsawes, which is also the name for the great longnosed armadillo (D. pastasae), but in this case the term tsawes can be interpreted as a general term for “armadillo” (the Matses are steadfast in their classification of this variety as a type of sedudi). In addition to being smaller, sedudimpi is said differ from the other varieties by having a darker back, a grayish yellow underside, and a tail with stripes along the edges of the bands. Sedudidapa, in addition to being larger, is said to be lighter in color than other varieties. Akte tsawes is said to be characterized by living in or adjacent to floodplain forest along large streams and rivers.

The only economic importance of ninebanded long-nosed armadillos for the Matses is as food, but unlike the greater long-nosed armadillo it is not a preferred game species and, due to the dietary taboo limiting its consumption to old people, it is not frequently hunted.

When a nine-banded long-nosed armadillo is encountered in a burrow it may be flooded out in the same manner as described below for the great long-nosed armadillo, but this species often nests on the surface under piled-up leaves, typically in floodplain forest and on levee islands. A hunter may come upon armadillo spoor and follow it, or he may simply happen upon such a leaf nest. When he finds the leaf nest, he lightly introduces a palm frond into the entrance of the nest to see if it is inhabited. If the armadillo growls, he cuts saplings into stakes and makes a circular fence around the nest. Then he enters the circle, takes apart the nest, and kills the trapped armadillo with a machete.

Now that the Matses have flashlights, they hunt at night by walking along forest paths. The primary motivation for night-hunting is to kill pacas, which are common in secondary forest near villages, especially when peach palm (Bactris gasipaes) fruits are ripe (from January to March), but nine-banded long-tailed armadillos also frequent secondary forest and are sometimes shot if encountered on a night hunt.

Young people do not eat the nine-banded longnosed armadillo lest they become weak and thin or begin to eat clay. This armadillo is also believed to make children ill, causing a high fever. Therefore, it is primarily eaten by the elderly.

Matses natural history: The nine-banded long-nosed armadillo is similar to the greater long-nosed armadillo, but it is smaller, has a thinner tail and head, has a bump on its head, has a yellower underside, narrower bands on its carapace, and a stronger and fouler smell. Both species have a “branched” penis.

Nine-banded long-nosed armadillos are found in upland forest, in the floodplains of large streams and rivers, and in both primary and secondary forest. They are frequently found in abandoned swiddens and old blowdowns. They are also common in fresh blowdowns and on levee islands along rivers. (Thus, nine-banded long-nosed armadillos use a wider range of habitats than greater long-nosed armadillos.) They are common and, unlike greater long-nosed armadillos, do not tend to get hunted out (at least in part because they are not a preferred game species).

Nine-banded long-nosed armadillos make their nests in underground burrows, on the surface under leaf piles, and inside hollow logs. Burrows are especially common in blowdowns. These armadillos have more than one nest and typically sleep in a different one each night. Burrows may be on hilltops or hillsides, in floodplain forest, or in secondary forest, but not in stream headwater gullies (the preferred site of great long-nosed armadillo burrows). The typical burrow entrance is angled straight down. If the burrow is in floodplain forest or on a levee island it goes straight down for about 30 cm and then becomes horizontal and extends only a short distance (in which case the armadillo can be easily dug out). If the burrow is in a hillside in upland forest, it will become horizontal for a short distance and then angle upward (in which case it cannot be flooded out). The burrows have a bed made of dead leaves. Nests in hollow logs also have a leaf bed. Leaf-pile nests are usually made next to a log or a buttress root. The leaf litter is piled very high and the armadillo sleeps under the leaves, not on them. Leaf-pile nests are made in floodplain forest.

The nine-banded long-nosed armadillo is nocturnal. It forages all night long. It has clear and wide paths that it follows as it forages. It leaves its path to root for earthworms at the base of hills and other places. Sometimes it returns to the same place to forage, and at other times it does not. It follows streams, rooting in the soft soil for earthworms. It roots in places that are close to each other, leaving areas clear of leaf litter where it roots. It makes a lot of noise as it travels quickly along its path, with its tail up in the air. It runs very quickly along its path to escape from humans, smacking its tail on the ground as it runs. Where there is no path it cannot run as quickly, having to jump over or go around obstacles. It can be grabbed when it is not on its path. Armadillos are safer from predation when they are on their paths, but they must leave their paths to forage. (One could infer the same for greater long-nosed armadillos, but because those are more rarely encountered at night, Matses informants did not comment on this.)

Nine-banded long-nosed armadillos are solitary. The female gives birth to two or three offspring. Larger females give birth to three. The young follow the mother when they are little.

White flies (small biting flies that look like light-colored mosquitoes; probably phlebotomine psychodids) live with and follow ninebanded long-nosed armadillos. Jaguars, pumas, bush dogs, and caimans eat nine-banded longnosed armadillos.

They make a low grunting growl when disturbed.

Nine-banded long-nosed armadillos dig into rotten logs to eat armored millipedes (Barydesmus spp. [Platyrhacidae]) and other invertebrates. They root in the ground to eat earthworms and grubs that live in the ground. They also eat mole crickets and other insects. They eat the mesocarp of isan palms (Oenocarpus bataua [Arecaceae]), kuëbun isan palms (O. mapora) and swamp palms (Mauritia flexuosa [Arecaceae]). They also eat some dicot tree fruits and the seeds of tonnad trees (an unidentified species of Myristicaceae). They eat maggots that they find in rotten meat, but do not eat the meat itself.

Remarks: The nine-banded long-nosed armadillo has been the subject of numerous field studies at temperate latitudes, especially as an invasive species in the United States, but little has been published about its ecology or habits in tropical rainforest. Information about this species (or species complex; see above) obtained from Matses interviews is notable for observations about several behaviors not or seldom mentioned in the literature, including use of well-worn pathways (mentioned only by Neck [1976] and Emmons [1997] among the references we consulted), tail-slapping when pursued by predators, the use of surface nests rather than burrows as diurnal refugia in floodplain habitats (Layne and Waggener, 1984; Platt and Rainwater, 2003), and frugivory (Emmons, 1997). Also of interest is the alleged association of this species with small biting flies (?Brumptomyia spp. [Psychodidae]; Lainson et al., 1979), and observations about predators (Dasypus novemcinctus leads an almost predation-free existence in the partially defaunated habitats of temperate North America; Loughry and McDonough, 2013). By contrast, Matses observations about litter size are anomalous. Whereas the Matses report litters of only two to three offspring in this species, 95% of D. novemcinctus litters in North America consist of genetically identical quadruplets (Newman, 1913; Prodöhl et al., 1996), and similar observations have been reported from South America (e.g., Noss et al., 2003: table 3). Although we could offer several ad hoc explanations for this discrepancy, none are supported by actual evidence, so we cannot discount the probability that the Matses are simply wrong.

Voucher material (total = 4): Nuevo San Juan (AMNH 268227, 268228; MUSM 11081, 11083).

Other interfluvial records: Río Yavarí- Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: We follow Feijó and Cordeiro- Estrela (2016) in recognizing three distinct species among the nominal taxa formerly synonymized with Dasypus kappleri Krauss, 1862. Of these, D. kappleri (sensu stricto) is restricted to the Guiana Region (north of the Amazon and east of the Rio Negro/Orinoco), whereas D. beniensis Lönnberg, 1942, occurs in southeast Amazonia (south of the Amazon and east of the Rio Madeira), and D. pastasae is widespread in western Amazonia (on both banks of the upper Amazon west of the Negro and Madeira rivers). Together, these three taxa belong to the subgenus Hyperoambon Peters, 1864, whereas the ninebanded species is referred to the nominotypical subgenus (Wetzel and Mondolfi, 1979).

Our voucher material exhibits most of the distinguishing features attributed to Dasypus pastasae by Feijó and Cordeiro-Estrela (2016), of which the most consistently useful seem to be (1) the high relief of the central scale of each scalerosette on the pelvic shield (giving this part of the carapace a diagnostically bumpy texture), and (2) the relatively low and uninflated lateral palatine keels (fig. 6B). Measurements of our two adult vouchers (table 3) are within the range of morphometric variation for D. pastasae as quantified in their study.

Ethnobiology: The principal Matses name for the greater long-nosed armadillo, tsawes, is not analyzable and has no other meaning, except that it can be used a general term for all armadillos. It has one archaic synonym, yosh, also not analyzable but cognate with the term for “armadillo” in many other Panoan languages. In the language formerly used in the Matses' komok ceremony (Romanoff et al., 2004), the greater long-nosed armadillo is called şhëdëk-şhëdëk, a term that refers to its many bands (literally “wrinkles”).

Three subtypes of the greater long-nosed armadillo are recognized by Matses hunters: tsawes çhëşhe (“black/dark-colored armadillo”), tsawes uşhu (“white/light-colored armadillo”), and tsawes piu (“yellowish armadillo”) or tsawes takpiu (“yellow- bellied armadillo”). The latter type is said to be characterized by a yellow-gray venter, and the dark variety is said to be smaller than the others. While there is some consensus with respect to these color and size distinctions, there is much variation among speakers—who often directly contradict one another—with respect to the habitat preferences of these subtypes.

The only economic importance of greater long-nosed armadillos for the Matses is as food (fig. 5). It is one of the most appreciated game species, and it is the most favored meat for some Matses. Occasionally, when a female with young is killed, the young are kept as pets.

Armadillos are hunted principally by flooding them out of their burrows. A hunter may decide to hunt greater long-nosed armadillos after having seen fresh armadillo spoor the day before, or he may come across fresh tracks when it is still early in the day. When his wife or children request it, or during the period when armadillos have much fat (in April and May), a hunter may decide to search primarily for armadillos, even without having found any tracks (in which case much more search time is invested). This species, along with two-toed sloths and caimans, were formerly the most important species to be hunted during the komok ceremony (which is no longer practiced; Romanoff et al., 2004).

When a hunter finds armadillo tracks he inspects them and then searches for burrows in the vicinity. Generally the hunter does not attempt to follow armadillo tracks, because armadillo trackways are not continuously visible, and because armadillos tend to forage going in circles. Rather, the hunter searches nearby places that might be suitable for burrows; specifically, stream headwater gullies and stream banks. When he finds a fresh burrow, he looks for fresh tracks and white flies at the burrow entrance and sniffs the burrow for the armadillo's scent. If these signs are present, he cuts a palm frond, knots the leaflets at the tip of the frond into a ball, and removes the rest of the leaflets. He then introduces the frond into the burrow and listens. The armadillo generally growls if disturbed in this manner.

If the armadillo growls or rustles around in its leaf bed, the hunter will stop up the hole with dry or rotting pieces of logs to keep the armadillo from running out while he goes to cut a digging stick and stakes to make a fence. Once he has gathered these materials, he begins to dig inward toward the sleeping chamber, either by enlarging the burrow entrance or by opening a new hole from above; meanwhile, the armadillo goes into its retreat tunnel (a blind, narrow, horizontal tunnel adjacent to its sleeping chamber; fig. 6). The hunter continues to dig until he reaches the sleeping chamber and has room enough to stand upright in it. After removing the bed of dry leaves, the hunter blocks the entrance tunnel with excavated clay, forming a funnel so that water poured into the excavation will fill the retreat tunnel. To keep the armadillo from escaping, a fence is made of stakes to block the exit from the retreat tunnel.

The next step is to make a watertight basket by weaving a palm frond, lining it with wild banana leaves, and reinforcing it with a vine. The hunter will then make several trips back and forth to a nearby stream, fetching water to flood the hole. The hunter's wife often accompanies her husband on the hunt, in which case she will help dig, make the basket, and haul water. It may take 10 or more baskets of water to fill the retreat tunnel. Once it is flooded, the hunter waits quietly beside the excavation. If properly flooded, the armadillo will not be able to breathe and will try to exit the retreat tunnel. The first sign of the armadillo's exit is bubbles of air, then churning water; finally, the armadillo bumps into the fence. When the armadillo emerges, the hunter quickly introduces several sticks through the fence to block the armadillo from going back into the retreat tunnel. Once he has blocked its retreat, the armadillo is hopelessly trapped. The hunter simply waits for the armadillo to eventually stick its head between the sticks that form the fence and clubs it, breaking its skull or neck. With particularly deep holes, it may take several hours to flood the burrow to obtain this highly prized game. Sometimes water leaks into the ground, and the burrow will not flood. In such cases, the armadillo cannot be killed.

A less frequently used method for extracting an armadillo from its burrow is to smoke it out. This is done simply by lighting a fire at the entrance and fanning the smoke into the burrow. The armadillo comes out with its eyes closed and the hunter kills it with a machete or a stick. Although this method requires much less work than flooding, it is less effective, and hunters seldom carry matches, fire drills, or other means of starting a fire.

Nowadays the Matses hunt at night by walking along forest paths with a flashlight and shotgun. Because greater long-nosed armadillos do not live in the secondary forest near villages and are usually hunted out from adjacent primary forest, it is rare for one to be killed in this manner. (By contrast, nine-banded long-nosed armadillos are more commonly killed by hunting at night; see above.)

The Matses formerly hung the pelvic shields of armadillo carapaces on the horizontal poles of their longhouses as hunting trophies and to keep track of how many armadillos had been killed locally. Today this is still done by a few old men.

Although all Matses eat greater long-nosed armadillos, there are several partial dietary taboos. Young people do not eat greater longnosed armadillo fat lest their teeth rot. Young men also do not eat young armadillos lest they become cowards. Young men don't eat the tail, lest they grow thin. Old people can eat the young, and the fat, and the tail. No one eats the lungs. Greater long-nosed armadillos can make children ill, causing a high fever. When several people are flooding out an armadillo, they cannot say out loud “the armadillo is coming out of its burrow,” lest it not come out; instead, they whistle softly to announce that it is starting to exit its burrow. Additionally, one should not throw around pieces of clay that are dug out of the burrow (as small boys are often tempted to do), lest the armadillo not come out.

The tail is sometimes burned, letting the smoke enter the hunter's eyes, which is believed to help him find armadillos in the future.

Matses natural history: Greater longnosed armadillos prefer primary upland forest. They make their burrows in the headwaters of streams and along small streams. They forage in the dry floodplains of streams and in palm swamps, where the earth is softer, but they also root around on hilltops and hillsides. They are common in upland forest where they have not been hunted out.

Greater long-nosed armadillos always nest in burrows that they dig in the ground. Each armadillo has several active burrows and sleeps in a different one each night. There are also abandoned burrows in the vicinity of active burrows. Burrows in stream headwater gullies are deeper than burrows in stream floodplains. Each burrow has a large sleeping chamber, where the armadillo has its leaf bed. The leaf bed smells like armadillo urine. The burrow also has a long, narrow, blind retreat tunnel adjacent to the sleeping chamber. The retreat tunnel is generally somewhat horizontal and has a few centimeters of water on the floor. Greater long-nosed armadillo burrows have only one entrance.

The greater long-nosed armadillo is nocturnal. During the day it sleeps in its burrow. Before dusk it is awake in its burrow, rustling the dry leaves in its leaf bed, waiting for it to get dark. Right at dusk it rushes out of its burrow and then begins to travel noisily along one of its paths, which are primarily along hilltops. It stops along its path to forage, rooting for worms and grubs in soft dirt and digging into rotten logs for armored millipedes and other invertebrates. It leaves its path to root in lower ground, in the floodplains of streams or in palm swamps. If these are flooded, it roots at the edges of the flooded area. It sniffs the ground as it roots for earthworms. It follows streams as it forages, often crossing one or more streams, and then circles back to its path. It swims across deep streams. It bathes in mud holes, where collared peccaries may also bathe (during the day). It may come across one of its other burrows and check it out, but it will not sleep there if it is not yet late. When it is near dawn (between 05:00 and 05:30 in northeastern Peru) it finds its path and follows it to its nearest burrow. Once it finds its burrow, it collects fresh leaf litter to add to its bed. It leaves an area clear of leaf litter near its burrow where it does this. It rolls in its leaf bed to pack it down and may be awake in its burrow rustling the leaves in its bed for a short time after the day dawns. It sometimes walks around during the day in a heavy rain.

Greater long-nosed armadillos are solitary. Males do not sleep with females in their burrows. They copulate when they find each other while foraging at night. The female gives birth to two offspring inside its burrow. The female eats the placenta. The young follow the mother when they are little.

White flies (small biting flies that look like light-colored mosquitoes; probably phlebotomine psychodids) live with greater long-nosed armadillos. They are always present at the entrance of active burrows. When the armadillo leaves its burrow, some follow it while others remain at the burrow. A burrow that is inhabited will have more white flies during the day than other, uninhabited but active burrows.

Jaguars eat greater long-nosed armadillos while hunting at night. They may pounce on an armadillo from above as the armadillo passes by. They remove the carapace, and often stash a portion of the armadillo to eat later. Pumas also kill armadillos. Bush dogs kill armadillos by entering the burrow and following them into the (blind) retreat burrow. They pull the armadillo out and eat it at the entrance of the burrow. Tayras that hunt in trios can also kill an armadillo. Black caimans and anacondas catch armadillos as they swim across large streams.

Greater long-nosed armadillos make a low rumbling growl when disturbed. They growl loudly when a predator grabs them. Newborns whine inside the burrow.

Greater long-nosed armadillos find armored millipedes (Barydesmus sp. [Platyrhacidae]), round millipedes (Neocricus sp. [Rhinocricidae]), centipedes, beetles, and beetle grubs in rotten logs. They root in the ground for earthworms and grubs that live in the ground. They eat any invertebrate they find. They are also very fond of isan palm (Oenocarpus bataua [Arecaceae]) fruits. They eat the mesocarp of ripe isan palm fruits that fall to the ground. They also eat chukë ants that feed on the isan fruits. While eating isan palm fruit they also root in the vicinity for earthworms. They also eat the mesocarp of fallen swamp palm (Mauritia flexuosa [Arecaceae]) fruits. They eat insect larvae that they find in rotten echo tree (Jacaratia sp. [Caricaceae]) fruits.

Remarks: In a recent publication (Fleck and Voss, 2016), we compared Matses natural history information about Dasypus pastasae item by item with the scientific literature on D. kappleri (the name by which this species was formerly known; see above). Briefly, almost 80% of what the Matses have to say about D. pastasae is new information, and most of the rest essentially agrees with the literature. The single point of disagreement between our interview results and the literature concerns burrow construction, for which it seems likely that the Matses account is correct.

Matses observations about Dasypus pastasae, a primary game species, are more detailed than those about D. novemcinctus, which is much less often hunted and consumed. To the extent that information about the two species overlaps, it would seem that these sympatric congeners are ecologically and behaviorally similar, with the noteworthy exception that D. pastasae seems invariably to dig its burrows in the sides of streams and stream headwater gullies, whereas D. novemcinctus digs burrows in different places and sometimes also uses surface nests. In the absence of other evidence for niche divergence, this difference in use of diurnal refugia is perhaps significant for species coexistence.

Sloths (Bradypodidae and Megalonychidae)

The taxonomy of Recent sloths remains unrevised, and the current application of names is based on decades-old conventions that are now being challenged by molecular analyses. Morphological examination of available specimens from the Yavarí-Ucayali interfluve suggest that only a single species each of Bradypus (threetoed sloths, Bradypodidae) and Choloepus (twotoed sloths, Megalonychidae) occur here, although a second species of Choloepus might also be expected.

Voucher material (total = 4): Nuevo San Juan (MUSM 11075, 23811), Orosa (AMNH 73758, 73759).

Other interfluvial records: Actiamë (Amanzo, 2006), Jenaro Herrera (Pavlinov, 1994), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: Four living species of bradypodid sloths are currently recognized, of which one (B. pygmaeus Anderson and Handley, 2001) is an insular endemic, another (B. tridactylus Linnaeus, 1758) is restricted to northeastern Amazonia (east of the Rio Negro and north of the lower Amazon), a third (B. torquatus Illiger, 1811) is endemic to the Atlantic forest of southeastern Brazil, and the fourth (B. variegatus) is thought to range throughout most of Central America and tropical South America (Anderson and Handley, 2001; Gardner, 2008).

Although Bradypus variegatus (the “brownthroated three-toed sloth” of English usage) and B. tridactylus (the “pale-throated three-toed sloth”) are the only currently recognized species of Amazonian bradypodids (Wetzel, 1985a; Anderson and Handley, 2001; Gardner, 2008), Amazonian threetoed sloths have received no modern revisionary attention, and there is little compelling evidence to support current taxonomic usage. In particular, the extensive distribution of B. variegatus (from Honduras to northern Argentina), substantial geographic variation in morphology (see below), and preliminary genetic evidence from gene-sequencing studies (e.g., Moraes-Barros and Arteaga, 2015; Ruiz-García et al., 2017) all suggest that brownthroated three-toed sloths include several distinct taxa. Seven South American subspecies of B. variegatus were recognized by Gardner (2008), but no explicit justification for sloth trinomial nomenclature has yet been provided by any author.

The type locality of Bradypus variegatus is assumed to be somewhere in the Atlantic Forest of southeastern Brazil (Wetzel and Kock, 1973), and specimens that are morphologically similar to Atlantic Forest material have been collected throughout southeastern Amazonia, from the vicinity of Belém westward to the left (west) bank of the Tapajós. Sequence data analyzed by Moraes-Barros and Arteaga (2015) and Ruiz-García et al. (2017) likewise suggest that southeastern Amazonian and Atlantic Forest populations of brown-throated three-toed sloths are closely related. Additionally, photographs of the holotype skull (SMF 4313; available online from the Senkenberg Naturmuseum database) closely match the craniodental morphology of southeastern Amazonian specimens that we examined. In the paragraphs that follow, we assume that southeastern Amazonian three-toed sloths represent the nominotypical form, B. variegatus variegatus.

Three-toed sloths from the Yavarí-Ucayali interfluve and others that we examined from western Amazonia fit the description of Bradypus variegatus in the inclusive sense that this binomen is currently applied (e.g., by Wetzel and Avila-Pires, 1980; Wetzel, 1985a; Anderson and Handley, 2001; Gardner, 2008). Traits that support the identification of western Amazonian three-toed sloths as belonging to the B. variegatus complex include (1) the presence of a speculum in adult males (a speculum is absent in B. torquatus); (2) absence of a mane of long black hair (present in B. torquatus); (3) the consistently brownish coloration of the throat (the gular fur is whitish or yellowish in B. tridactylus); (4) the absence of large foramina in the anterior part of the mesopterygoid fossa (present in B. tridactylus and B. torquatus); and (5) large size (B. pygmaeus is much smaller). However, other phenotypic characters distinguish western Amazonian material from southeastern Amazonian specimens of B. variegatus.

In western Amazonian specimens of brownthroated three-toed sloths, the anteriormost maxillary teeth (the “pseudo-incisors” of Thomas, 1917) are large and procumbent; the rostrum is correspondingly broad; the nasal bones are short, often exposing the nasal orifice and the projecting pseudoincisors in dorsal view; and the postorbital process of the jugal is well developed (figs. 7A, 7B, 8A). The co-ossified mandibles are likewise distinctive, entirely lacking any trace of the median anterior process (symphysial spout) seen in other sloths, but with a deep symphysis that often extends posteriorly between the third pair of teeth (fig. 9A). By contrast, the pseudoincisors of southeastern Amazonian specimens are much smaller and nonprocumbent; the rostrum is narrower; longer nasal bones conceal the nasal orifice from dorsal view but extend the orifice such that it is visible from below; and a postorbital process of the jugal is absent or indistinct in most examined specimens (figs. 7C, 7D, 8B). Additionally, the mandibles of all examined specimens from southeastern Amazonia have a small but distinct symphysial spout, and the symphysis only extends posteriorly between or just beyond the second pair of teeth (fig. 9B). On average, skulls from western Amazonia are substantially larger than those from southeastern Amazonian in all measured dimensions (table 4), but most measurements exhibit overlapping variation; an exception is rostral breadth, which is diagnostically greater in western Amazonian material than in specimens from southeastern Amazonia. We have not been able to discover any pelage or other external differences between three-toed sloths from western and southeastern Amazonia, which both seem to be highly variable (e.g., in pelage coloration).

Western Amazonian three-toed sloths are usually associated with the epithet infuscatus (e.g., by Thomas, 1928; Gardner, 2008), a convention that we follow despite misgivings about the application of this name. According to Wagler (1831: 611), Bradypus infuscatus was based on a single specimen collected by “Herr von Spix” (= Johann Baptist Ritter von Spix) on the border between Brazil and Peru (“Brasilia versus Peru”). Based on Spix's known itinerary (Vanzolini, 1981; Hershkovitz, 1987), it seems reasonably certain that the holotype (by monotypy, a specimen in the Zoologische Staatssammlung München, ZSM 1162; Wetzel and Kock, 1973) was collected at or near Tabatinga, on the left (north) bank of the upper Amazon, just across the river from Peru and the mouth of the Yavarí (fig. 1).² We have not examined this specimen, which consists only of a mounted skin (A. van Heteren, personal commun., 2016); the skull was said to be missing in the original description (Wagler, 1831) and has not been recovered. Based on the geographic distribution of examined material with the morphological traits described above, it is plausible that B. v. infuscatus is the proper trinomen for this form, but without confirmatory evidence (e.g., DNA sequence data obtained from the holotype skin) it is hard to be sure. Possible junior synonyms include brachydactylus Wagner, 1855; macrodon Thomas, 1917; codajazensis Lönnberg, 1942; and subjuruanus Lönnberg, 1942 (for type localities, see Gardner, 2008). Of these nominal taxa, we have only examined the holotype of macrodon (BMNH 80.5.6.56), which exhibits all the diagnostic craniodental traits of the western Amazonian phenotype as described above.

Given the distinct morphologies associated with western Amazonian specimens on the one hand and southeastern Amazonian specimens on the other, we would be inclined to call these distinct species. However, we are provisionally using trinomial nomenclature because there are intermediate phenotypes, suggestive of intergradation, in central Amazonia. Based on specimens we examined with the morphological distinctions described and illustrated above, Bradypus variegatus variegatus occurs along the right (south) bank of the Amazon from the vicinity of Belém westward to the left (west) bank of the Tapajós, whereas B. v. infuscatus occurs in eastern Ecuador, eastern Peru, and southern Venezuela (fig. 10). Only a few specimens are available from the wide (ca. 1300 km) central-Amazonian gap between these morphologically diagnosed taxa, but those we examined are difficult to assign with certainty to either form. For example, three BMNH specimens from Codajás (on the north bank of the upper Amazon west of Manaus; fig. 10: locality 8a) have small pseudoincisors, long nasals, narrow rostrums, and short mandibular symphyses (like B. v. variegatus), but they have well-developed postorbital jugal processes and lack symphyseal spouts (like B. v. infuscatus). Pending a much-needed revision of the brownthroated three-toed sloth complex, trinomial nomenclature seems like the appropriately conservative option.

Ethnobiology: The three-toed sloth has only one name, mëinkançhuşh. It is not analyzable and there are no cognates in other Panoan languages. No subtypes are recognized by the Matses. The Matses do not hunt three-toed sloths, eat them, or keep them as pets. The Matses generally will not even look directly at them, because if a hunter looks at a three-toed sloth it can make his children ill. Therefore, the Matses know very little about the natural history of this species.

Matses natural history: The three-toed sloth is similar to the two-toed sloth, but it has three claws on its front feet, a short tail, a smaller head, a striped face, and a spotted back. Its face looks like that of a little person.

Three-toed sloths are found mostly beside rivers in stands of cecropia trees (Cecropia spp. [Moraceae]). They are rarely encountered in Matses territory (where large stands of cecropia trees are absent).

It makes no nest. It is nocturnal and diurnal. It swims slowly but effectively across rivers. It climbs along lianas and branches upside-down. It sits in thickets. It is solitary. It eats mostly cecropia tree leaves and also eats fig-tree (Ficus sp. [Moraceae]) fruits.

Remarks: The Matses do not have a lot to say about three-toed sloths—doubtless because this species is seldom seen and never hunted for food—and their few observations seem intended primarily to distinguish this species from the superficially similar two-toed sloth (a primary game species). This intent is explicit in the comparative phrasing of their morphological descriptions, but diagnostic comparisons are also implied by their remarks that three-toed sloths are both nocturnal and diurnal (two-toed sloths are exclusively nocturnal), and that three-toed sloths swim (whereas two-toed sloths do not, at least according to the Matses; see below).

The widespread notion that three-toed sloths have a close relationship with trees of the genus Cecropia has been called a myth by Montgomery (1983), who averred that Bradypus are simply easier to see in the relatively open, sunlit crowns of cecropia trees than in the denser crowns of other tree species. However, it should be noted that Montgomery's research on sloths (e.g., Montgomery and Sunquist, 1978) was carried out on an island with very little early-successional vegetation (the preferred seral stage of Cecropia spp.), where an essentially captive population of sloths may have had no choice but to forage in habitats that would have been avoided by unconstrained animals. Subsequent studies from mainland habitats seem to agree with earlier reports that young cecropia leaves are a preferred forage for Bradypus, perhaps due to their low fiber content, high nutrient value, and lack of defensive secondary compounds (Urbani and Bosque, 2007). Therefore, the Matses observation that riparian stands of cecropia trees are the primary habitat of threetoed sloths merits credibility despite Montgomery's (1983) dismissal of essentially similar observations in the older literature.

Voucher material (total = 10): Nuevo San Juan (AMNH 268225, 268226, 273184; MUSM 5072, 11077, 11079, 11080, 15346), Orosa (AMNH 73760, 73761).

Other interfluvial records³: Actiamë (Amanzo, 2006), Choncó (Amanzo, 2006), Estación Biológica Quebrada Blanco (Heymann et al., 2011), Itia Tëbu (Amanzo, 2006), Jenaro Herrera (Pavlinov, 1994), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: The genus Choloepus has received no modern revisionary attention. The current recognition of two valid species and the application of their names are largely based on diagnoses provided by Wetzel and Avila-Pires (1980) that were subsequently incorporated in dichotomous keys by Wetzel (1985a) and Gardner and Naples (2008). According to these sources, C. didactylus (with type locality in Surinam) is a widespread Amazonian species, whereas C. hoffmanni (with type locality in Costa Rica) occurs in Central America, trans-Andean South America, and western Amazonia. Mapped geographic ranges (in Wetzel, 1985a; Gardner and Naples, 2008) suggest that both species occur in northeastern Peru, so either C. didactylus or C. hoffmanni might occur in the Yavarí- Ucayali interfluve; alternatively, the two species might be sympatric in our region.

According to the literature cited above, Choloepus didactylus and C. hoffmanni can be distinguished by several cranial characters (table 5): (1) On the dorsal surface of the rostrum, just anterior to the orbit, either the maxillary bones contact the frontals, or the lacrimals contact the nasals; because most cranial sutures are fused in fully adult sloths, this is a character that can only be scored from immature specimens. (2) In the rear of the mesopterygoid (“interpterygoid”) fossa, a pair of large foramina—one on each side—that communicate with the pterygoid sinuses is either present or absent; unlike the preceding character, this feature can be scored from both immature and adult specimens. (3) The ratio between the anterior (widest) and posterior (narrowest) transverse dimensions of the mesopterygoid fossa is said to be taxonomically diagnostic. Additionally, pelage differences are said to distinguish C. didactylus from C. hoffmanni, but we experienced considerable difficulty in evaluating pelage traits, none of which appear to offer an unambiguous basis for character scoring, so we do not consider them further here.

In order to evaluate the allegedly diagnostic cranial traits of Choloepus didactylus and C. hoffmanni, we examined series of specimens from regions where these species occur allopatrically (table 6). Of the three characters described above, only the presence/absence of posterior mesopterygoid foramina consistently distinguishes eastern Amazonian specimens (C. didactylus) from Central American material (C. hoffmanni). Although the mean difference in computed mesopterygoid ratios between these samples is obviously significant, the observed ranges overlap, so specimens cannot be sorted consistently by this criterion. Similarly, although most eastern Amazonian specimens exhibit maxillary-frontal contact, over 20% exhibit lacrimal- nasal contact (which appears to be a fixed trait in Central American material). Nevertheless, these results provide compelling evidence of phenotypic divergence, and they are consistent with the current recognition of two species of two-toed sloths, albeit with somewhat less distinct recognition criteria than suggested by the literature cited above.

The ten voucher specimens we examined from the Yavarí-Ucayali interfluve are all referable to Choloepus hoffmanni as that species is currently recognized by these criteria. All the immature specimens (juveniles and subadults: AMNH 73761, 268225, 273184; MUSM 11077, 11080) exhibit lacrimal-nasal contact, and all the specimens that preserve an intact mesopterygoid region (N = 9) have paired posterior foramina that open into the pterygoid sinuses. All of the four adults we measured (table 7) have mesopterygoid ratios <2.0.

Interestingly, almost all the two-toed sloths we examined from the north bank of the Amazon (directly opposite the Yavarí-Ucayali interfluve) are referable to Choloepus didactylus by the criteria discussed above, as are all the specimens we examined from eastern Ecuador. By contrast, both C. didactylus and C. hoffmanni occur south of the Amazon in Peru, but they occur at different localities. Therefore, although the ranges of these species overlap, we have yet to find any place where they are actually sympatric. Possibly they occur in different forest types, a hypothesis that merits testing in the field.

Ethnobiology: The principal name for the two-toed sloth is şhuinte, which is not analyzable or found in other Panoan languages. It has three archaic synonyms: nai, posën, and tabidiate. The first two are not analyzable but, unlike the principal term, they do occur as sloth names in other Panoan languages; the third term is a nominalization meaning “one for tying up its feet” (when the Matses bring a sloth back to the village alive, they tie their claws to their hand so that they cannot harm the person carrying it; this was the standard way of bringing back sloths during the komok ceremony, and is still done sometimes today). In the language used in the komok ceremony, the two-toed sloth is called uşhtud kudu; the first word seems to include the verb uşh ‘sleep’ (the Matses associate two-toed sloths with sleepiness and laziness) and the second term is an archaic adjective meaning “grayish” or “light-colored.” Four subtypes of two-toed sloth are recognized by Matses hunters: şhuinte uşhu (“white/light-colored two-toed sloth”), şhuinte piu (“red/reddishbrown two-toed sloth”), şhuinte poçhëşh (“black-/dark-bellied two-toed sloth”), and chompish. The last name is not analyzable, but it is also the name of a small bird that purportedly reveals the presence of two-toed sloths. Chompish and the other three subtypes seem to occupy different levels in the Matses classification, in the sense that chompish is considered more distinct from the other three types. This could be represented as follows:

The chompish variety is said to be much rarer than the others and to be the size of a young twotoed sloth, but to have head, claws, and teeth the size of a full-grown adult. It has a short back, rostrum, and limbs. Its body is reddish, while its head is light colored. Informants never fail to mention that its meat is very hard and takes a long time to cook. The other three varieties are generally associated with different sizes and habitat preferences, but there is much inconsistency and even direct contradictions among informants.

This is a primary game species that is much appreciated by the Matses. Informants emphasize that every part of the sloth is eaten, including its vulva, bladder, and viscera. This species, along with the greater long-nosed armadillo and the spectacled caiman, were the most important species to be hunted during the Matses komok ceremony. Two-toed sloths are considered very desirable and hardy pets. When a hunter kills a mother that is carrying its young, the hunter brings it home for his children to raise (fig. 11).

Two-toed sloths are usually hunted by climbing trees and clubbing or strangling them.⁴ A hunter may be tipped off as to the presence of two-toed sloth by hearing its urine drip from the treetops or by hearing the call of a bird called chompish (a type of flycatcher). Otherwise the hunter simply scans the canopy as he walks down a path. The sloth is typically curled up asleep in the crotch of a branch, in the crown of a bottle palm (Iriartea deltoidea [Arecaceae]) or in a vine tangle. When he spots a sleeping sloth, the hunter makes a climbing ring (a loop of epiphyte stems or a palm frond, to loop between his feet as he shinnies up the trunk) and judges whether he can get within arm's reach of the sloth. If he can get close to it, he cuts a hard stick to use as a club, climbs up next to the sloth, wakes it, and then clubs it on the head until it dies or falls out of the tree. Twotoed sloths are very resistant, and do not succumb quickly when they are clubbed. While the hunter is clubbing a sloth, it will become fierce and try to bite the hunter. If he cannot get close to it (for example, if it is in the crown of a bottle palm), he prepares a noose from epiphyte stems and attaches it to a stick 1.5 to 2 m long. He then climbs an adjacent tree, taps the sloth lightly with the stick (so that it wakes up and sticks out its head), snares it around the neck, and then pulls hard to yank it off its perch. Next, he throws down the sloth with the noose still attached. He may then club it to death on the ground or decide to take it back home alive. The latter option is preferred if the hunter is far from home, to keep the meat from spoiling.

There is a particular way to pack up a sloth for carrying it home. First, the hunter presses on its abdomen to make it defecate and urinate. Then, he breaks all of its limb bones with a club, ties the hind feet together, tucks the tied-up legs together, and lashes all in place to make a roughly spherical bundle. If the sloth is still alive, its front and back claws are tied tightly to its arms and feet, so that it cannot grab the carrier. Then a tumpline (a carrying strap worn across the forehead, made from the bark of certain trees) is attached to the sloth and the hunter carries it home on his back.

Hunters typically leave hunting signs where they kill certain game, usually next to the path that is closest to the kill site. Sloth kills are marked with a meter-long stick jammed into the ground; a slot is cut in the top of the stick, and a tuft of the sloth's fur is wedged in the slot. If the sloth was captured by noosing, the noose stick is jammed in the ground next to the stick with fur; if the sloth was captured by clubbing, the club and the climbing ring are placed next to the trophy stick.

The Matses used to hang two-toed sloth mandibles on one the horizontal poles of their longhouses, as hunting trophies and to keep track of how many sloths had been killed at that locality. Today this is still done by a few old men.

The Matses believe that a certain type of owl hoots at night near a Matses house to announce human visitors or the presence of game species, including two-toed sloths. After killing a sloth, a hunter may remove the sloth's anal scent gland and rub it across his eyes. This is believed to improve his ability to find two-toed sloths in the future. A similar effect is thought to be achieved by burning a sloth forearm bone and letting the smoke enter one's eyes.

Children do not eat the jaw meat, lest their jaws swell up. Young men do not eat young sloths, lest they not wake up early. Young men do not eat the ball of fat that is found in the twotoed sloth's abdominal cavity, lest they become unable to spot sloths. The hunter who killed the sloth does not eat the intestines, lest he not find more sloths. Young men and the hunter who killed the sloth do not eat the liver, lest they do not find more sloths.

When a sloth is killed, the hunter's children (or any other child that looks upon the dead sloth) may be made ill by the sloth's spirit. The symptoms of contagion by sloths include oversleeping and fever.

Matses natural history: Two-toed sloths are very hairy. They have no tail. They have sharp black teeth, small ears, and hairless noses. They have two long claws on their front feet and three on their hind feet. Their backs are lighter-colored than their undersides. Their meat is hard and takes a long time to cook. They have the most fat during the rainy season (December to April), especially in February.

Two-toed sloths are found in any type of primary forest, including upland forest, and flooded or dry floodplain forest. They are also found in very old (>25 years old) abandoned swiddens. They are relatively common, especially in areas that have not been hunted.

Two-toed sloths do not make nests. They sleep during the day in vine tangles, under the cover of large-leafed epiphytes, in crotches of trees high in the canopy, or in the crown of bottle palms (Iriartea deltoidea [Arecaceae]). They sleep curled up, with their head tucked in. They usually perch high up to sleep, but they perch lower when they are at the edge of a river or when it is raining.

The two-toed sloth is nocturnal. After sleeping all day, it wakes up at dusk and stays in its perch looking around until it is fully dark. It sets out traveling along the undersides of branches and horizontal lianas looking for leaves and fruits to eat. It avoids dead branches. It drinks water from holes in trees in the canopy, where frogs have laid eggs, or water trapped in palm-tree crowns. It also licks rainwater from leaves. It climbs down to the ground to drink water when there is none in the treetops. It also climbs down to the ground to eat clay that has been dug out by an armadillo, to eat rotten meat, and to defecate. It climbs back up quickly after having defecated, drunk water, or eaten clay or rotten meat. After eating a lot it stops to sleep for a while before setting out to forage again. Unlike three-toed sloths, two-toed sloths do not swim.

Two-toed sloths are usually solitary but are sometimes found in male-female pairs perched near each other. They give birth to a single young, which clings to its mother's venter.

Two-toed sloths are followed by a small swarm of little black flies and sweat bees (tabanid flies that have black-and-yellow striped abdomens). A bird called chompish (a type of flycatcher) follows sloths (perhaps to feed on its flies?). Large eagles and hawks kill and eat twotoed sloths, even fully grown adults. Jaguars also occasionally kill them.

Two-toed sloths do not make any noise while perched or while feeding. They huff when they are being killed or when they are defending themselves.

Two-toed sloths eat the young leaves of almost any tree. They are particularly fond of the leaves of tote trees (Eschweilera spp., Lecythis spp., or Cariniana spp. [Lecythidaceae]). Other favored foods are the ripe fruits and leaves of trees in the cacao (Sterculiaceae) family, particularly tonkodo (Theobroma sp.) and senad dëbiate (Theobroma subincanum). They eat the ripe fruits of a few other types of dicot trees, including mamuin (Garcinia longifolia [Guttiferae]), which they eat after splitting open the rind; piuşh bëchi (Helicostylis tomentosa [Moraceae]); tonnad (a general term for trees in the Myristicaceae); and mannan tsipuis (Inga spp. [Leguminosae]). They also eat the fruits of some epiphytes and the heart and young leaves of okodonte epiphytes (Philodendron [Araceae]). They feed much more on leaves than on fruits, and they do not eat old/mature leaves. They descend to the ground to eat rotten meat.

Remarks: Matses interviews include many original observations about the behavior and diet of two-toed sloths, notably including their use of water sources in the canopy, geophagy, occasional carrion eating, food plants, and a possibly mutualistic interaction with an unidentified species of bird. To the extent that Matses observations overlap with previously published results of scientific research on Choloepus, there is good agreement (e.g., about nocturnality, reproduction, and predation), but one discrepancy merits comment. Two literature reviews—one about C. didactylus and the other about C. hoffmanni— both claim that two-toed sloths can swim (Adam, 1999; Hayssen, 2011a), but none of the references cited in either review explicitly report swimming behavior in these species.⁵ Most mammals, of course, can swim when necessary, but the absence of any published observation of swimming by Choloepus spp. tends to support the Matses' claim that two-toed sloths do not often do so, by contrast with three-toed sloths, which are frequently found swimming across rivers and lakes.

Voucher material (total = 2): Nuevo San Juan (AMNH 268232; MUSM 11093).

Other records: None.

Identification: The pygmy (or “silky”) anteaters currently recognized as Cyclopes didactylus cannot be confused with any other mammal, although it seems highly probable that they comprise a species complex rather than a single species (Coimbra et al., 2017). Among the numerous epithets currently treated as synonyms or subspecies of C. didactylus (see Gardner, 2008) are several that might apply to the population that occurs in the Yavarí-Ucayali interfluve, but available museum specimens are too few and widely scattered to attempt a revision for this report.

Our two voucher specimens have co-ossified occiputs but unfused braincase and facial sutures. AMNH 268232 consists only of a skull, but MUSM 11093 includes a skin. The pelage of the latter specimen is silvery grayish-brown dorsally and lacks any distinct markings, although the fur is noticeably browner over the head, shoulders, and upper back, whereas the lower back and rump are grayer; the ventral fur is abruptly self-beige over the chest and abdomen, but brownish on the throat. Measurements of MUSM 11093 are: headand- body length, 182 mm; length of tail, 229 mm; hind foot, 43 mm; ear, 16 mm; condylonasal length, 50.1 mm; nasal length, 14.0 mm; least interorbital breadth, 9.4 mm (measured just anterior to small postorbital processes); anterior zygomatic breadth, 14.2 mm; posterior zygomatic breadth, 22.5 mm; breadth of braincase, 22.9 mm. This specimen weighed 286 g.

Ethnobiology: The pygmy anteater is generally called tsipud, which is not analyzable and has no cognates in other Panoan languages. In some villages the term tsekeded is used instead. Tsekeded is an onomatopoetic representation of the “purr” that the pygmy anteater makes when disturbed. A few Matses consider tsekeded to be a synonym of tsipud. No subtypes are recognized. Some Matses consider it to be a type of sloth, while others consider it to be more closely related to tamanduas.

The pygmy anteater is not hunted, eaten, or kept as a pet.

It is a death omen to come across a pygmy anteater. If a person comes in contact with one, his children are very likely to fall ill with excessive sleepiness and fever (symptoms similar to those caused by two-toed sloths). Men with children will not even look at them.

Matses natural history: The pygmy anteater is like a baby two-toed sloth, but has a tail. Its tail has a hairless patch (the ventral prehensile surface). The soles of its feet are hairless and reddish. Its eyes are small and its nose is pink. It is pretty and its fur is like cotton.

Pygmy anteaters are usually found perched on thin vines in open forest. They can be found high or low in the canopy. They sit on branches that hang over large streams.

Pygmy anteaters are nocturnal. During the day they forage along vines and thin trees. They stop to rest frequently, perching in the same position as two-toed sloths.

Pygmy anteaters are generally solitary, but occasionally two are found near each other.

A pygmy anteater may curl up and cover its face with its front paws when it is scared.

When knocked off its perch or poked with a stick, a pygmy anteater purrs loudly, saying “tsequededededed.”

Its diet is not known by the Matses.

Remarks: Most of what little the Matses have to say about the pygmy anteater agrees with the sparse natural history literature on this species (reviewed by Hayssen et al., 2012), although no vocalizations seem to have been reported previously.

Voucher material (total = 3): Boca Rio Yaquerana (FMNH 88890), Nuevo San Juan (MUSM 11094), Quebrada Esperanza (FMNH 88891).

Other interfluvial records: Actiamë (Amanzo, 2006), Choncó (Amanzo, 2006), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: Only three specimens of the giant anteater seem to have been collected in the Yavarí-Ucayali interfluve. Two of them, both females, are preserved as skins and skulls. Of these, FMNH 88890 is obviously the younger animal, with a co-ossified occiput but otherwise unfused cranial sutures; the skull is lightly built, and partially disarticulated. The second, FMNH 88891, is much larger and has a more heavily ossified skull, but all of the cranial sutures (except those of the occiput) are still visible. Although the latter specimen is seemingly immature according to the age classification that Wetzel (1975) proposed for tamanduas, measurements of FMNH 88891 are within the range of variation that Wetzel (1985a) subsequently reported for adult Myrmecophaga. Both skins exhibit all of the usual diagnostic external traits of M. tridactyla (e.g., those described by Husson, 1978; Emmons, 1997), with no apparent pigmental or other pelage difference between the younger and older individual. Selected measurements of FMNH 88891 are: head-and-body length, 1202 mm; length of tail, 687 mm; hind foot, 157 mm; ear, 51 mm; condylonasal length, 330.7 mm; nasal length, 151.3 mm; least interorbital breadth, 45.5 mm; posterior zygomatic breadth, 65.6 mm; breadth of braincase, 63.0 mm.

Our specimen from Nuevo San Juan (MUSM 11094) was killed by a Matses hunter to protect his dogs, which were fighting with it. This is a fully adult female (braincase elements are coossified) with the following measurements: headand- body length, 1150 mm; length of tail, 710 mm; hind foot, 153 mm; ear, 49 mm; condylonasal length, 353.0 mm; nasal length, 178.4 mm; least interorbital breadth, 45.9 mm; anterior zygomatic breadth, 61.2 mm; posterior zygomatic breadth, 68.5 mm; breadth of braincase, 63.5 mm. This specimen weighed 32 kg.

Ethnobiology: The giant anteater has only one name, ʂhaë. It is not analyzable, but it is a common term for this species in other Panoan languages. A few informants suggested that there is a large and a small subtype, but most Matses do not recognize any subtypes.

Giant anteaters are not eaten or kept as pets. Although giant anteaters are not hunted, the Matses sometimes club them to death when they fight with dogs. Giant anteaters sometimes kill dogs if a hunter does not arrive at the scene quickly enough. The Matses are quite careful in approaching giant anteaters, knowing that they could kill a person with their claws.

People don't look at giant anteaters, lest their children fall ill.

Matses natural history: Giant anteaters are very large and have a big, bushy tail with long hairs, reminiscent of the flowers of arrow cane (Gynerium sagittatum [Gramineae]). They wag their tails back and forth as they walk. They have large claws on their front feet, which they tuck in when they walk around. Their hind footprints look like a human child's. They have a very elongated snout and a very long and thin tongue. They have a stripe around their neck. Their eyes and ears are small.

Giant anteaters use all habitats: floodplain forest, upland forest, primary forest, and secondary forest (including abandoned swiddens).

Giant anteaters do not make a nest. Instead, they sleep lying in hollows in the ground, all curled up. They also sleep between buttress roots.

Giant anteaters are diurnal and nocturnal. They are strictly terrestrial.

Giant anteaters are solitary. The young ride on the mother's neck.

Giant anteaters are a favorite food of jaguars.

Giant anteaters can roar loudly, like a jaguar.

Giant anteaters eat bullet ants (Dinoponera spp., Ectatomma [Formicidae]). They eat bullet ants by sticking their nose in the nest to make many come out, and then lick them up with their tongue. They dig into leaf-cutter ant (Atta sp.) nests and feed there for a long time. They also eat other ants, including army ants (Eciton spp.), eu ants (tiny biting ants), and masioko ants (small biting ants). They dig into hives of stingless bees (Apidae: Meliponini) that are at the base of trees and stick their snouts in to eat bee larvae and lick up the honey. They do not eat isan palm (Oenocarpus bataua [Arecaceae]) fruits or other fruits.

Remarks: Matses interviews about giant anteaters are of particular interest because most of what is known about the natural history of this magnificent species is based on fieldwork in savanna habitats (e.g., Redford, 1985; Shaw et al., 1987; Medri et al., 2003). Although Matses information is sparse and generally agrees with the scientific literature, their observation that giant anteaters are a preferred prey of jaguars is noteworthy because the high frequency of jaguar attacks on Myrmecophaga has only recently been documented (Cavalcanti and Gese, 2010; Sollmann et al., 2013).

Giant anteaters are predators of social insects, and they are widely believed to feed almost exclusively on termites and ants (Redford, 1985, 1986). However, Matses observations suggest that stingless bees might be an important alternative food resource in Amazonia. In upland savannas (e.g., the Cerrado; Redford, 1985), giant anteaters are said to feed mostly on termites, whereas populations in seasonally inundated grasslands (e.g., the Llanos and Pantanal; Medri et al., 2003) appear to eat mostly ants. Judging from Matses observations, Amazonian populations of Myrmecophaga feed primarily on ants, perhaps because most rainforest termitaria are arboreal (Constantino, 1992). The feeding bouts of giant anteaters are said to be very brief (usually less than a minute; Redford, 1985), so it is interesting that the Matses say they feed for a long time at leaf-cutter ant nests.

Voucher material (total = 7): Nuevo San Juan (AMNH 268233; MUSM 11095, 11096), Orosa (AMNH 73752, 74114, 74115), Santa Cecilia (FMNH 86892).

Other interfluvial records: Actiamë (Amanzo, 2006), Choncó (Amanzo, 2006), Divisor (Jorge and Velazco, 2006), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006).

Identification: The seven tamandua specimens we examined from the Yavarí-Ucayali interfluve exhibit all the pelage color variants that Wetzel (1975) reported for the species, from entirely blond (FMNH 86892, MUSM 11095) to completely blackish (AMNH 74115, 268233) and vested (with dark-brownish midbody, shoulders and nape but pale legs, face, and tail; FMNH 74115, MUSM 11096). As noted by Wetzel (1975), these phenotypes appear to be uncorrelated with age or sex and seem to represent genuine coat-color polymorphisms. All of our specimens are females, of which four could be called subadults (Wetzel's age class 1; with co-ossified occiputs but otherwise unfused cranial sutures) and three are young adults (with fused midfrontal sutures). Measurements of the latter specimens (AMNH 74114, FMNH 86892, MUSM 11095) are all within the range of variation for homologous dimensions of Tamandua tetradactyla provided by Wetzel (1985b). All of our specimens have three foramina in the back of the orbital fossa, a trait that is said to reliably distinguish T. tetradactyla from its trans-Andean congener T. mexicana (see Wetzel, 1975: fig. 3). Selected measurements of FMNH 86892 and MUSM 11095 (the specimens with the most complete data in our series) are, respectively: head-andbody length, 605 and 538 mm; length of tail, 515 and 497 mm; hind foot, 102 and 111 mm; ear, 51 and 56 mm; condylonasal length, 129.3 and 121.7 mm; nasal length, 45.9 and 38.2 mm; least interorbital breadth, 22.7 and 24.0 mm; posterior zygomatic breadth, 39.6 and 44.2 mm; breadth of braincase, 40.1 and 44.9 mm. The weight of MUSM 11095 (our only adult specimen accompanied by this datum) was 5.3 kg.

Ethnobiology: The collared tamandua has only one name bëwi. It is not analyzable but it is a common term for this species in other Panoan languages. The Matses are aware of different color morphs (black, blond, and vested), but these are not named or considered subtypes. Some informants describe the coat variation as analogous to that found among domestic dogs. Informants agree that it is similar and closely related to the giant anteater.

The tamandua is not hunted, eaten, or kept as a pet by the Matses. Tamanduas sometimes injure dogs, when dogs attack them. When such a fight is in progress, a hunter will try to kill the tamandua with a stick. Tamanduas may try to claw people if approached too closely.

People don't look at tamanduas, lest their children become ill.

Matses natural history: Tamanduas have a tail with a bare patch (the ventral prehensile surface), an elongated snout, a long thin tongue, small eyes, small ears, and very strong forelegs. The claws of its forefeet are similar to those of a two-toed sloth. Some tamanduas are black, others are tan, and other have a black body with light-colored head and limbs. They have a strong, bad smell, like a termite nest.

Tamanduas are found in all habitats: upland forest, floodplain forest, and along rivers and streams, in primary forest, and in abandoned swiddens.

Tamanduas sleep in holes in trees and also in abandoned paca and armadillo burrows.

Tamanduas are diurnal and nocturnal. They forage up in trees and on the ground. They travel long distances while foraging. They rest curled up on branches.

Tamanduas are usually solitary, although sometimes more than one are found travelling together.

When dogs bark at a tamandua, it rears up on its haunches and spreads its arms. Tamanduas are preyed upon by jaguars, pumas, and large raptors.

Tamanduas dig into rotten logs and arboreal termite nests to eat termites. They also dig into binsan (black arboreal ants) nests and eat the ants and their larvae. (Some informants say tamanduas dig into stingless-bee hives to lick up the honey, but other informants deny this.)

Remarks: Matses observations are completely consistent with published natural history information about tamanduas (Emmons, 1997; Hayssen, 2011b; Navarrete and Ortega, 2011). Among the few novel items mentioned by our interviewees, the Matses add pumas (Puma concolor) to the very short list of predators known to attack this species.

Our three specimens from Nuevo San Juan were all killed by Matses hunters in primary floodplain forest.

Voucher material: None.

Other interfluvial records: Quebrada Pobreza (Escobedo-Torres, 2015), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: The short-eared dog is unmistakable in external appearance (Emmons, 1997), so sight records from competent observers are reliable evidence for local occurrence. There is, additionally, a published camera-trap photograph of this species from Quebrada Pobreza (Pitman et al., 2015: fig. 10S).

Ethnobiology: Few Matses have ever seen the short-eared dog, which is either called mayanën opa (“demon's dog”) or nimëduk opa (“jungle dog”). These are never considered synonyms; rather, there is intervillage variation whereby one of the names is used and the other is considered incorrect. Short-eared dogs are sometimes confused with jaguarundis, some informants believing them to be the same species. Some Matses comment that they would seem to make nice pets, but they are never raised as such.

Touching or even looking at a short-eared dog is likely to make one's children ill with high fever and intense thirst. As with most very rare animals, an encounter with a short-eared dog is interpreted as an omen that someone in the person's village or a close relative will soon die.

Matses natural history: Short-eared dogs look very much like domestic dogs, but with small ears. One can easily mistake one for a domestic dog. Their footprints are like a dog's footprints.

Short-eared dogs are found in all habitats. They frequent mineral licks to look for prey, and frequent palm swamps to eat swamp-palm (Mauritia flexuosa [Arecaceae]) fruits, and to look for game. They den in hollow logs, holes in the ground, or undercut banks of streams. They are diurnal and are most commonly encountered in pairs or groups of three, but they are also frequently solitary. They give birth to as many as three pups, which may be found abandoned in their dens (in hollow logs or holes in the ground), presumably while the mother is out hunting. (According to one informant, the inside of the hollow log in which pups were found was scraped smooth.) They whine and snarl. They do not bark like domestic dogs.

Short-eared dogs eat pacas, agoutis, acouchies, and spiny rats, which they chase down or dig out of their burrows or nests. They also eat grounddwelling birds like white-throated tinamous, wood-quails, and trumpeters. They dig jungle frogs out of their burrows. They eat bata tree (Pseudolmedia spp. and ?Maquira spp. [Moraceae]) fruits and the mesocarp of swamp-palm (Mauritia flexuosa) fruits.

Remarks: Short-eared dogs are among the least known of all canid species, and even the Matses have little to say about them. Most of the natural history observations in the interviews we compiled about this species agree with the literature reviewed by Pitman and Williams (2004), notably with respect to diurnality, denning sites, and diet. However, whereas those authors reported fish to be the most frequent item found in short-eared dog scat from their study site in southern Peru, the Matses do not mention fish among the items consumed by Atelocynus in the Yavarí-Ucayali interfluve, where only terrestrial prey and fruit are said to be eaten. The Matses observation that short-eared dogs eat the fruit of Mauritia flexuosa (not mentioned as a food plant by Pitman and Williams, 2004) is corroborated by a recent report based on a camera-trap survey at a Colombian locality by Acevedo-Quintero and Zamora-Abrego (2016).

Voucher material: None.

Other interfluvial records: Choncó (Amanzo, 2006), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: Bush dogs are unmistakable in external appearance (Emmons, 1997), so sight records from competent observers are reliable evidence for local occurrence.

Ethnobiology: The bush dog has only one name, achu kamun. The term achu (“red howler monkey”) is used as a modifier of the term kamun (“feline/canine”) based on the similarity in pelage coloration of bush dogs and howler monkeys.

Bush dogs are not eaten. The Matses often comment that they would seem to make nice pets, but they almost never raise them due to the belief that they will make children fall ill. One informant knew of a single case in which a woman found a bush dog pup and raised it; when it became an adult it hunted together with Matses hunting dogs.

The spirits of bush dogs can make children ill when a Matses looks at a bush dog or kills one. If a hunter does so, he will collect medicinal plants to bathe his children with, to prevent them from becoming ill. Among the symptoms of being made ill by a bush dog (or a feline or other wild canine) are a high fever and intense thirst.

Matses natural history: The head and forequarters of bush dogs are orangeish. Their tails and hindquarters are black or blackish. Bush dogs are smaller and fatter than domestic dogs.

Bush dogs are more commonly encountered than short-eared dogs (but there are many Matses who have never seen one). They are found in all types of habitats.

They den together in hollow logs or in large holes in hillsides. Sometimes they dig depressions in the ground to sleep there one night.

Bush dogs are diurnal. They hunt as a group searching for the spoor of their prey. They search for pacas, sniffing for their scent along large or small streams. When they find paca spoor, all the members of the pack begin to follow it. When one dog finds the paca's burrow, it calls the other dogs. A large male dog enters the burrow. The paca then dashes out of one of the other exits of its burrow, and all the dogs chase it down, barking. The paca typically goes to a stream, follows it to a deep bend, plunges in, and holds its breath underwater. The bush dogs arrive and the leader barks out orders for the others in the pack to surround the deep stream bend ready to pounce on the paca when it comes out. While some wait upstream, others downstream, and some on the bank above the stream bend, one or two of the bush dogs plunge into the water and feel around for the paca underwater. When one of the dogs touches it underwater, the paca emerges and flees to a shallow section of the stream, where the waiting dogs pounce on it. They kill it together, biting its neck and other vital parts, and they eat it together after pulling it to the bank. They eat every part of it.

When bush dogs find a greater long-nosed armadillo in its burrow, one of them enters the burrow, follows the armadillo into its retreat tunnel, kills it, and drags it out of the burrow to eat it with the other bush dogs. When they find a nine-banded long-nosed armadillo in a leaf nest on the ground, they surround the nest, and one of the bush dogs jumps on it (to make the armadillo come out). The other dogs then pounce on the armadillo and kill it, or they may have to chase it down to kill it.

When bush dogs find an agouti, they chase it down barking until it seeks refuge in a hollow log or a hole in the ground. Then, one of the dogs goes into the hole while the others wait at the opening. If the log or the hole in the ground has more than one opening, one dog goes into each hole. Then they kill the agouti in the log or hole, or they kill it as it exits. Bush dogs rest after eating, lying in a dry spot for a while, before hunting again. Or they may first drink water at a stream.

Bush dogs travel in packs of three to eight individuals; five is the typical size of a pack. The pack has a male leader. Both males and females hunt.

(The Matses do not know of any animal that eats bush dogs, although they imagine a jaguar would do so.) Bush dog barks are more high-pitched than those of domestic dogs.

Bush dogs eat pacas, long-nosed armadillos, agoutis, acouchies, and spiny rats. They do not eat larger mammals like peccaries. They also eat whitethroated tinamous, smaller species of tinamous, wood-quails, and other terrestrial birds. They dig jungle frogs out of their burrows to eat them.

Remarks: Bush dogs have long been something of a zoological enigma, with their small size, highly developed social behavior, hypercarnivorous dentition, absurdly short legs, partially webbed feet, diurnal activity, and a remarkable ability to swim underwater (Sheldon, 1992; Beisiegel and Zuercher, 2005). This odd combination of traits seems all the more extraordinary by comparison with those of closely related Chrysocyon brachyurus (the maned wolf; Perini et al., 2010), a much larger, solitary, omnivorous, long-legged, crepuscular/nocturnal, and strictly nonaquatic species. Attempts to explain bush dog morphobehavioral traits have included seemingly implausible suggestions that packs of these diminutive, dachshundlike animals can run down and kill much larger prey (e.g., peccaries, deer, and even tapirs; Zuercher et al., 2004); that their short legs and hypercarnivorous dentition are the nonadaptive consequence of phyletic dwarfing (Wayne and O'Brien, 1987); that their partially webbed feet are somehow useful for walking on soft soil near streams (Beisiegel and Zuercher, 2005); and that they might cache their food underwater (Kleiman, 1972). A defining aspect of the literature on this species is that most information about diet and behavior is derived from captive studies; few biologists have seen bush dogs alive in the wild for more than a few minutes at a time (e.g., Deutsch, 1983; Peres, 1991; Strahl et al., 1992; Aquino and Puertas, 1997).

Matses observations about bush dogs—which agree strikingly with reports by Tate (1931) and Cabrera and Yepes (1940) that were also derived from indigenous sources—convincingly account for many unusual aspects of the bush dog phenotype. In particular, their dachshundlike morphology⁶ is clearly adaptive for entering burrows or hollow logs to drag or flush their inhabitants (armadillos, pacas, agoutis) to the surface. Their cooperative social behavior and swimming abilities may be especially important for hunting pacas, whose streamside burrows have multiple exits, and whose evasive behavior often includes hiding underwater (Tate, 1931; Cabrera and Yepes, 1940; personal obs.). Although Matses accounts of bush dog hunting behavior include obvious anthropomorphisms, the ambush tactics they describe are plausible in the context of hunting behavior previously reported for other social canids (e.g., wolves and African hunting dogs). From these accounts, and from previously published anecdotes and captive observations, Speothos venaticus seems best characterized as a pack-hunting diurnal predator anatomically and behaviorally specialized to extract medium-sized (ca. 3–12 kg) mammalian prey from burrows, and to pursue escaped prey (especially pacas) into water.

Voucher material (total = 5): Boca Río Yaquerana (FMNH 88887), Nuevo San Juan (MUSM 11170, 13150), Orosa (AMNH 73762), Quebrada Esperanza (FMNH 88888).

Other interfluvial records: Jenaro Herrera (Pavlinov, 1994), Quebrada Pobreza (Escobedo- Torres, 2015), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: The ocelot (Leopardus pardalis) and the margay (L. wiedii) are gaudily streaked-and-spotted small cats with reversed nuchal fur (the hairs of the nape pointing forward rather than backward; Pocock, 1941).⁷ Ocelot specimens from northeastern Peru are consistently larger than margays in most measured dimensions (table 12), and these species can also be distinguished by external and cranial proportions (see the account for L. wiedii, below).

Numerous subspecies of the ocelot are currently recognized as valid (e.g., by Wozencraft, 2005), but it is not known whether any represent taxonomically meaningful subdivisions. The last specimen-based revision was Pocock's (1941), who assigned all the Peruvian material he examined to the subspecies L. p. aequatorialis (Mearns, 1902), the type locality of which is in the Pacific lowlands of northern Ecuador. Although Eizirik et al. (1998) suggested that several phylogeographic partitions are present within L. pardalis, their study did not include any western Amazonian sequence data, so the assignment of our material to any of the phylogroups they recognized is problematic. In the absence of any compelling reason for trinomial nomenclature, it seems pointless to speculate about the subspecific assignment of our material.

Ethnobiology: The Matses name for the ocelot is bëdimpi, the term for jaguar/feline with the diminutive suffix mpi. It has no other names and no varieties are distinguished by the Matses. The term bëdimpi can also be a more general term that includes the ocelot, the margay, the jaguarundi, and the house cat.

The ocelot is of no economic importance to the Matses. It is not eaten or kept as a pet. The Matses are not afraid of ocelots, because they are too small to attack humans. However, ocelots enter villages to eat chickens in their coops at night, and they prowl around near villages in the daytime to attack free-ranging chickens on the outskirts of clearings. Once an ocelot kills and eats a chicken, it keeps coming back to get more. When an ocelot becomes a pest in this way, the Matses hunt it down with dogs.

Matses with young children avoid having any contact with or even looking at ocelots, lest the ocelot's spirit make their children ill (see the ethnobiology entry for Puma concolor for details on symptoms and treatment of contagion by felids).

Matses natural history: The ocelot is small and spotted.

The ocelot is found in any type of habitat, including upland and floodplain forest, and primary and secondary forest. It mainly walks on the ground, but also often climbs trees.

The ocelot is diurnal and nocturnal. It walks following streams, sniffing as it hunts. Or it lies in wait for prey on the ground or sitting up in a tree. It lies on fallen trees in blowdowns to warm itself in the sun. It sleeps on trees that lean somewhat horizontally. It defecates in habitats called “demon's swiddens” that have an open understory.⁸

The ocelot is solitary. Sometimes two are seen together, perhaps male and female. The ocelot gives birth to two kittens in a hollow log or in a hole in the ground.

During the day ocelots kill agoutis and acouchies, and at night they kill pacas. As an ocelot walks along a small stream at night it may find and catch a paca that is eating aquatic snails. Then it drags the paca to dry land to eat it. It may dig into an acouchy burrow when it chases one into its burrow. The ocelot stalks its prey crouching, as it slowly advances, and then pounces on the quarry, grabs it with its claws, and bites its head. Ocelots go to drink water repeatedly while eating. An ocelot will stash part of the kill, if it is a large animal.

The ocelot growls when it is taking prey.

Ocelots eat pacas, agoutis, acouchies, spiny rats, opossums, tinamous, other terrestrial birds, lizards, and jungle frogs (Leptodactylus spp. [Leptodactylidae]).

Remarks: Matses observations about ocelots largely overlap with the scientific literature on this common and widespread species, notably agreeing with the results of radio-tracking studies in rainforest habitats (Emmons, 1988; Aliaga- Rossel et al., 2006) with respect to diel activity pattern, killing behavior, and caching of large prey. Curiously, Matses observations suggest that twinning is common for ocelots, whereas most captive litters consist of a single young (Havlanová and Gardiánová, 2013).

The Matses list of ocelot prey closely resembles that obtained by analyzing scat at another Peruvian rainforest site (Emmons, 1987), but not with known ocelot diets from other rainforested regions. In particular, the Matses list omits sloths, which are said to be commonly eaten by Central American ocelots (e.g., by Moreno et al., 2006), and primates, which are often eaten by ocelots in southeastern Brazil (Bianchi and Mendes, 2007). To the best of our knowledge, no previous dietary study has reported that ocelots eat frogs. The omission of any mention of ocelot frugivory by our informants seems noteworthy by contrast with lists of fruits eaten by jaguars, pumas, and jaguarundis in Matses accounts of those species (see below).

Voucher material (total = 1): Boca Río Yaquerana (FMNH 88889).

Other interfluvial records: Nuevo San Juan (this report), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: The only available margay specimen from the Yavarí-Ucayali interfluve consists of the skin and skull of a young adult male (FMNH 88889). Although margays are much smaller than ocelots on average, large specimens of margays are sometimes confused with small specimens of ocelots; fortunately, these species are readily distinguished by tail length and cranial proportions (Pocock, 1941). Based on collectors' measurements (table 12), the ratio LT/HBL × 100 equals 65% for our margay voucher versus 44%–48% for three adult ocelot vouchers.⁹ Additionally, the tanned skin of FMNH 88889 can be folded to show that the tail is substantially longer than the hind leg (a useful field character mentioned by Emmons, 1997), whereas the tail is substantially shorter than the hind leg on the ocelot skins that we examined.

The postorbital constriction is much wider than the interorbital constriction in Leopardus wiedii by contrast with L. pardalis, whose postorbital and interorbital constrictions are more nearly equal (Pocock, 1941). For FMNH 88889, the postorbital constriction is approximately twice as wide as the interorbital constriction (LPB/LIB × 100 = 204%), whereas this ratio ranges from 118% to 137% among our four adult ocelots. In dorsal view, margay skulls have larger orbital fossae than temporal fossae, whereas ocelots have larger temporal than orbital fossae. Lastly, margay skulls usually lack a sagittal crest, whereas most fully adult ocelots have well-developed sagittal crests. These cranial differences are visually conspicuous (fig. 14).

The last comprehensive revision of Leopardus wiedii was Pocock's (1941), which restricted the nominotypical form to southeastern Brazil, Argentina, and Paraguay; in his classification, western Amazonian margays were referred to L. w. pirrensis (Goldman, 1920), with type locality in eastern Panama. However, amazonicus Cabrera, 1917, based on a specimen from Tabatinga, Brazil, would appear to be the appropriate name if western Amazonian populations were judged to be taxonomically distinct from other margays (Oliveira, 1998b).

Ethnobiology: The margay is called tëstuk mawekid, which literally means “one that lays under epiphytes” owing to its habit of lying on tree limbs under the cover of large-leaved arboreal plants. The margay is sometimes called bëdimpi (ocelot) by observers unfamiliar with the species, but more knowledgeable Matses hunters say that this usage is incorrect.

The margay is of no economic importance to the Matses. Only rarely does one approach the outskirts of a Matses village to stalk chickens in the daytime. Unlike ocelots, margays do not raid chicken coops at night.

Matses with young children avoid having any contact with or even looking at margays, lest the margay's spirit make their children ill (see the ethnobiology entry for the puma for details on symptoms and treatment of contagion by felids).

Matses natural history: The margay is small and spotted. It has a long tail.

The margay is found in any habitat, including floodplain and upland forest. It is more frequently found in primary forest than in secondary forest (e.g., sites of abandoned swiddens). It is more rarely encountered than the ocelot.

The margay spends much of its time lying up in the trees, on tree branches or on upward-spiraling lianas. It walks up inclined trees and lies on the inclined trunk waiting for prey to pass by underneath. As it lies on a branch, tree, or liana, it hides under epiphytes or thick vegetation. It also lies in the open on branches or inclined tree trunks to rest after eating and to sleep. It also hunts by searching for prey on the ground, but it does not lie down on the ground.

The margay is solitary. It gives birth to two kittens in a hollow log on the ground or in a burrow, not up in the trees.

A margay may pounce on a tinamou that passes under the tree where the cat is waiting. Margays walking on the ground also kill tinamous, pouncing on them from far away. At night margays find tinamous sleeping on low perches. Margays pluck the feathers from tinamous before eating them. Margays kill other animals in the same ways (from ambush and by active diurnal and nocturnal hunting).

The margay growls when it is taking prey.

The margay eats pacas, agoutis, acouchies, spiny rats, other rats and mice, squirrels, common opossums, four-eyed opossums, and mouse opossums. It also eats white-throated tinamous (Tinamus guttatus), great tinamous (T. major), smaller tinamous (Crypterellus spp.), other terrestrial birds, and small arboreal birds. It also eats lizards, tree frogs, and jungle frogs (Leptodactylus spp.).

Remarks: Matses observations broadly agree with the scattered scientific literature on this small cat (reviewed by Oliveira, 1998b), notably with respect to its arboreal habits, denning behavior, and the wide range of prey taken. Matses accounts of arboreal ambushing versus active terrestrial searching, however, suggest a characteristic foraging strategy that is not described as such in the literature, nor does the literature describe several other details of margay predatory and feeding behaviors (e.g., feather-plucking) mentioned by our informants.

Voucher material: None.

Other interfluvial records: Actiamë (Amanzo, 2006), Divisor (Jorge and Velazco, 2006), Nuevo San Juan (this report), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006).

Identification: Jaguars are unmistakble externally (Emmons, 1997), and their skulls can be distinguished cranially from those of pumas (the only other large Amazonian cat) by nonoverlapping measurements (e.g., condylobasal length, zygomatic width; Husson, 1978); therefore, identification is not problematic. Rumors have long existed, however, of large Amazonian cats thought (by some) to represent one or more undescribed species of Panthera. Recent morphometric analyses of two skulls from eastern Peru alleged to represent such cryptic taxa suggest that they are simply jaguars, albeit perhaps with unusual coat-color phenotypes (Naish et al., 2014). Size variation among the eastern Peruvian material we examined is bracketed by a small female from the Río Cenepa (AMNH 98679) and a large individual (sex unrecorded but probably male) from the Río Aguaytía (AMNH 147513): condlyobasal length, 190.5–240.5 mm; least interorbital breadth, 39.8–50.3 mm; least postorbital breadth, 42.4–50.4 mm; zygomatic breadth, 148.9–174.1 mm; length P4, 25.6–30.2 mm.

Numerous subspecies of the jaguar have long been recognized (e.g., by Wozencraft, 2005). The western Amazonian population was referred to the nominotypical form in the last specimenbased revision (Pocock, 1939), but jaguar “subspecies” seem not to correspond either to mtDNA phylogeographic units (Eizirik et al., 2001) nor to geographic variation in cranial measurements (Hoogesteijn and Mondolfi, 1996), so it is not clear that any purpose is served by a trinomial nomenclature of these cats.

Ethnobiology: In addition to the term bëdi, the Matses have three archaic synonyms that refer to the jaguar. The first, kamun (an unanalyzable term commonly encountered among other Panoan languages), like bëdi, can also refer collectively to members of the category that includes all felids, wild canids, and the tayra. A second archaic synonym, winsad, specific to the jaguar, also means “frightening.” The third archaic synonym, chuisad, also specific to the jaguar, is not analyzable and has no other current meaning. In the language used in the Matses' komok ceremony, the jaguar is called mëndu, a monomorphemic terms that also designates the domestic dog.

Because bëdi can refer to any felid, or can be used generically, the term bëdidapa (literally “large jaguar/felid”) is used to specify the jaguar, although it can also refer to a Matses-recognized variety of the jaguar. In fact, there are two over differentiated varieties: bëdidapa and wispan kamun (“jaguar of the stars,” in reference to its particular spotted pattern). The wispan kamun variety is said to be fiercer and to have a larger head and a smaller body than the typical variety. Interestingly, melanistic jaguars seem to be absent in Matses territory. Men who have travelled outside Matses territory have heard of them, but the Matses have no names for them.

Jaguars are not currently of any economic importance to the Matses. They are not eaten, nor are they kept as pets. Several decades ago some Matses hunters participated in the illegal fur trade, trapping jaguars and other cats in box traps made from split palm logs and baited with capuchin monkey meat.

All Matses (especially women and children) fear jaguars, which often stalk people without attacking them. Although jaguars usually run away when they are encountered, occasionally a jaguar will bound toward a person. When they do so, the animal can sometimes be scared off by yelling loudly or by pounding on a buttress root. However, these tactics do not always work, and many Matses hunters tell of having to defend themselves with shotguns, arrows, or sticks. There have been a few isolated cases of jaguars killing men, women, and children in the forest, and there have been two instances within the last 40 years of a jaguar entering a Matses village and attacking people. Jaguars often kill Matses hunting dogs.

The Matses believe that if one kills, touches, or even looks at a jaguar, the spirit of the jaguar can make their child ill. The principal symptoms of jaguar-induced illness are a fever and constant thirst for water (jaguars are characterized by the Matses as drinking water constantly while eating meat). Several plants known as “jaguar/felid” medicine can be used to cure jaguar sickness. The child is bathed with an infusion of their leaves. If hunter has had contact with or seen a jaguar, he will collect these medicinal plants to treat his children prophylactically.

The Matses have two clans, the jaguar clan and caterpillar clan. If a jaguar whines within hearing range of a Matses household, the Matses say that it is announcing that someone of the jaguar clan will soon die.

Matses natural history: Jaguars are large, spotted, and have large feet and large canines.

Jaguars are mostly terrestrial, but they also climb trees. Although jaguars can be found in any habitat, including upland forest and floodplain forest, they are most often encountered along rivers and streams. Another favorite habitat is forest whose understory is dominated by the stemless palm Attalea racemosa (according to one informant, the palms provide cover, and this habitat is otherwise free of obstructive undergrowth). Jaguars are rarely encountered, but their tracks on river beaches and claw marks on trees are seen frequently. They are especially common on the banks of the Yaquerana (upper Río Yavarí).

Jaguars are nocturnal and diurnal. They travel very far, to other large streams. They spend much of their time walking slowly, searching for prey, or waiting to ambush prey. They sleep in hollow logs, hollow trees, cavities in the banks of stream headwater gullies, or between buttress roots, but not in the open. A jaguar does not sleep in the same place on consecutive nights, unless it is a female that has just given birth to cubs. However, a jaguar may return to a sleeping place after several days or more. Jaguars rest up in trees, in a sunny spot, or in or beside a treefall. Jaguars scratch trees and fallen logs, leaving claw marks that the Matses often find. Tonnad trees (species of Myristicaceae) are those most frequently found with jaguar claw marks.

Jaguars are solitary. The female gives birth to two cubs in a hole in a stream headwater gully or in a hollow log. At the entrance of the hollow log where it gives birth, the ground is swept clear. The female brings meat back to the den for the cubs to eat.

Jaguars wait beside tapir trails and follow tapir paths. When the tapir comes walking by, the jaguar pounces on it, grabs it with its claws, bites the back of its head and eats it. If the tapir runs, the jaguar chases it and grabs it. The pursued tapir may plunge into a stream, but the jaguar can kill it while it is still submerged, and will eat it right in the streambed. Jaguars also wait for tapirs and other prey at mineral licks. The jaguar also walks where there are no paths. It stalks collared peccaries without rustling the leaves as it walks. It captures the peccary by grabbing it with its claws and biting its head. It does likewise with deer. It ambushes monkeys up in trees. It kills most of its prey by biting the head.

After killing an animal, the jaguar begins to eat it right away. It eats the intestines first and laps up the blood. If the prey is not a small animal, the jaguar will hide part of the kill, covering it with leaf litter at the base of a tree. It may hide the meat in more than one stash. It drinks a lot of water after eating its fill of meat. After drinking, it lies down to rest near its stash. After resting it eats some more. Before it is all done, it hunts again. If it does not kill anything else, it comes back to eat more. It may return to eat the bones last.

The jaguar growls when it attacks. It has a very loud whining roar that can be heard from far away, day or night.

The jaguar eats tapirs, deer, collared and white-lipped peccaries, giant anteaters, tamanduas, giant armadillos, long-nosed armadillos, sloths, capybaras, pacas, pacaranas, agoutis, porcupines, coatis, monkeys, common opossums, tinamous, tortoises, river turtles, river turtle eggs, caimans, and fish. It commonly eats the mesocarp of isan palm (Oenocarpus bataua [Arecaceae]) fruits, but it does not eat most dicot tree fruits. Two types of dicot tree fruit that it does eat are those of the bata tree (Pseudolmedia spp. [Moraceae]) and the dadain tree (Clarisa racemosa [Moraceae]).

Remarks: Matses observations about jaguars are consistent with an emerging consensus about the distinctive trophic role of these top predators in Amazonian ecosystems. Whereas jaguar diets in subtropical and semiarid landscapes broadly overlap those of sympatric pumas (Taber et al., 1997; Núñez et al., 2000), Amazonian jaguars often eat armored reptiles and large mammals that are seldom taken by pumas. Such prey are thought to be uniquely vulnerable to jaguars because these cats have adaptations for durophagy that allow them to bite through reptilian armor and heavily ossified mammalian skulls (Schaller and Vasconcelos, 1978; Emmons, 1989; Miranda et al., 2016). Other prey commonly taken by Amazonian jaguars, but not (or less often) by other sympatric cats, include aquatic and semiaquatic taxa that jaguars hunt by patrolling river beaches and lakeshores (Emmons, 1987; Silveira et al., 2010).

The prey species that Matses interviewees report to be eaten by jaguars but not by pumas—notably giant armadillos, giant anteaters, tapirs, adult white-lipped peccaries, capybaras, tortoises, river-turtle eggs, and fish (table 11)—exemplify both aspects of this hypothetically distinctive niche (defined by durophagy and riparian/aquatic foraging). However, Matses descriptions of jaguar behavior also extend our knowledge of the remarkable predatory abilities of this species, which has not previously been reported to kill tapirs under water. That jaguars routinely patrol trails is well known, but the Matses say that jaguars also hunt in trailless parts of the forest (contra Weckel et al., 2006). An apparently novel observation is occasional frugivory, which seems not to have been reported in any previous analysis of jaguar diets.

Voucher material: None.

Other interfluvial records: Río Yavarí- Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999). Identification: Pumas are externally distinctive (Emmons, 1997), so sightings by competent observers are not problematic; cranial measurements of Amazonian specimens are provided in Husson (1978: table 51). Although the nominal taxon borbensis Nelson and Goldman, 1933, is based on a western Amazonian type, all tropical South American pumas are now referred to the nominotypical subspecies (Puma concolor concolor) following Culver et al.'s (2000) analysis of mtDNA sequence data.

Ethnobiology: The Matses name for the puma, bëdi piu, literally means “red(dish) jaguar/feline” (the color term piu can refer to red, orange, pink, yellow, or reddish brown). All Matses hunters recognize at least two subtypes of puma, one of which is known by the same term as the superordinate category, bëdi piu (an example of multilevel polysemy). The second subtype of puma is called sipidin. The sipidin subtype is characterized principally by having a white muzzle. Although the term sipidin is not completely analyzable, it appears to contain the term sipi (“tamarin”), perhaps because tamarins have white muzzles. The sipidin variety is said to be fierce and to attack people, whereas the bëdi piu variety runs away when it sees people. Some speakers recognize a third variety they call bëdi piudapa (“large puma”), which they say is the most rarely encountered variety.

Pumas are of no economic importance to the Matses. They are not eaten or kept as pets. Despite the purported aggression of the sipidin variety, in the last 40 years there has been only one instance of a Matses being attacked by a puma; although the victim was able to fight the cat off with a stick, he suffered serious wounds. Pumas often kill Matses hunting dogs.

The Matses believe that if one kills, touches, or even looks at a puma, the spirit of the puma can make one's child ill. A child can also be made ill by seeing or touching a puma. The principal symptoms of puma sickness are identical to those of jaguar sickness: a high fever and constant thirst for water. The cure for such illness is several plants known as “puma medicine,” and along with these plants, others known as “jaguar/felid” medicine are also collected. The child is bathed with an infusion of the leaves. If a hunter has had contact with or seen a puma, he will collect these medicinal plants to treat his children prophylactically. Contagion and symptoms are the same for all the wild felids and wild canines, with the exception that jaguars and pumas are more likely to make children ill, and the illness is more likely to result in death. In any case, the spirits of the smaller wild felids and canids are more likely to make children ill (when the animal is touched or seen) than are the spirits of game animals. Their propensity to cause illness is the main reason why wild felids and canids are not kept as pets. Treatment for contagion by smaller felids and the canids is the same as for the puma: one collects medicinal plants specific to species suspected to have caused the illness along with “jaguar/felid” medicinal plants.

Matses natural history: The puma is reddish, the color of a red brocket deer. Its undersides are light colored. It is smaller than a jaguar and has a large tail. Its spoor can be distinguished from that of the jaguar because the marks made by the toes are further from the impression of the central paw pad in puma tracks than in jaguar tracks.

The puma uses all rainforest habitats, including upland forest and floodplain forest. It is rarely seen.

The puma is nocturnal and diurnal. It walks slowly looking for prey with its tail hanging down close to the ground. It also sits up in trees, on large spiraling lianas, or on the ground waiting for prey to pass by. It waits at mineral licks for deer and peccaries. The puma sleeps day or night in cavities in the ground or in hollow trees or other sheltered places. It does not sleep in same place every night. If hunting dogs chase it, the puma may climb up a tree. Or, especially if there is only one dog, the puma may kill a dog. It does not kill a dog quickly, but rather keeps it alive for a while, making it whine.

The puma is solitary. Occasionally two are seen together. It gives birth to two cubs in holes in stream headwater gullies and in hollow logs. It goes out to hunt and eat and then comes back to the den to suckle its young. When they get older, it brings meat back to the den for them. Once they get strong, it walks around with the cubs. Before they are fully grown they leave the mother, one by one, and begin to hunt for themselves.

The puma stalks and pounces on deer, collared peccaries, or agoutis from a distance, grabs the prey with its claws, and bites the back of its head. It eats the liver and the rest of the viscera first. It continually goes to drink water while it eats. When it gets full, it stashes part of its prey, covering it with leaves, and lies down to rest nearby. It hunts again while there is still stashed meat left. Sometimes vultures find its stashed prey and finish it off. If the puma does not kill another animal, it comes back and eats the bones.

The puma vocalizes differently than the jaguar. Its roar/whine sounds a bit like the way one Matses calls out to another Matses in the forest, saying “ooo.”

The puma eats deer, pacas, agoutis, collared peccaries, young white-lipped peccaries, river turtles, caimans, and tree frogs. It does not eat tapirs. It eats the mesocarp of isan palms (Oenocarpus bataua [Arecaceae]) and a very few types of dicot tree fruits, such as those of the bata tree (Pseudolmedia spp. [Moraceae]).

Remarks: Although pumas are very widely distributed (from Canada to Chile) and have often been studied by wildlife biologists at high latitudes and in open habitats, the biology of this species in lowland tropical rainforest is not well documented. Matses observations are largely consistent with known aspects of puma behavior reported in previous studies, but their description of frugivory is apparently unique. Additionally, Matses assertions that pumas eat caimans and turtles are noteworthy given the absence of these taxa from previous accounts of rainforest puma diets (e.g., Emmons, 1987; Novack et al., 2005; Moreno et al., 2006).

Voucher material: None.

Other interfluvial records: Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: Jaguarundis are externally distinctive (Emmons, 1997), although inexperienced observers sometimes confuse them with tayras. Several subspecies are currently recognized (Wozencraft, 2005), of which the local form is melantho Thomas, 1914, with type locality at 800 m in the upper Ucayali drainage (Pasco department, Peru). Recent phylogeographic analyses of mtDNA sequence data, however, do not support the taxonomic recognition of any geographic subdivisions of this widespread species (Ruiz-Garia and Pinedo Castro, 2013). Thomas's (1914) measurements of an adult male and an adult female appear to be the only published morphometric data from western Amazonian specimens of Puma yagouaroundi.

Ethnobiology: The most common name for the jaguarundi is bëdi çhëşhe (“black feline/canine”). A variant of this term is bëdi wisu, in which the archaic term for “black” (wisu) is substituted for the regular term for “black” (çhëşhe). In some villages it is called şhododon, which is an onomatopoetic form imitating the jaguarundi's loud whining call. Some speakers who use şhododon for the jaguarundi use the term bëdi çhëşhe or bëdi wisu to refer to the short-eared dog. Others consider bëdi çhëşhe, bëdi wisu, and şhododon to be synonyms. A minority consider the short-eared dog and the jaguarundi to be the same animal.

The jaguarundi is of no economic importance. Is not eaten by the Matses, and there only is only one known case of one being kept as a pet.

When the jaguarundi whines loudly at night near a village or longhouse, saying “şhon şhon şhon,” it is interpreted as a death omen, that is, that someone in the village will soon die.

Matses with young children avoid having any contact with or even looking at a jaguarundi, lest its spirit make their children ill (see the ethnobiology entry for Puma concolor for details on symptoms and treatment of contagion by felids).

Matses natural history: The jaguarundi is black or grayish black. It has a very long tail.

Jaguarundis can be found in any rainforest habitat, including upland and floodplain forest. They sometimes hunt along the margins of Matses swiddens where rodents (spiny rats, agoutis, and acouchis) that feed on Matses crops (especially manioc) are abundant.

Jaguarundis hunt during the day. At bata trees (Pseudolmedia spp. [Moraceae]), where ripe fruits have fallen to the ground, the jaguarundi waits for animals that come to eat the fruits, and it also eats the fallen fruits. It waits for prey sitting on or under fallen trees.

Jaguarundis do not travel in large groups. The female gives birth to two young in a hollow log or a hole in the ground. When the young are very small and still have their eyes closed, the mother hunts and brings back meat for them. When their eyes open and they are a bit stronger, they travel with the mother and the mother kills prey and gives them some to eat. Eventually they leave the mother and begin to hunt for themselves.

The jaguarundi chases down lizards that are sunning themselves. It does not eat the tails of lizards. It kills geckos by jumping up on a tree trunk as they are climbing up. It covers its kill with dry leaves after eating a part and then goes to drink water. It walks on logs to catch tinamous.

The jaguarundi makes a whining sound (the Matses imitate the whine as “sho-do-doon”).

The jaguarundi has a very varied diet. It eats terrestrial birds, including white-throated tinamous (Tinamus guttatus), great tinamous (T. major), smaller species of tinamous (Crypterellus spp.), wood quails (Odontophorus spp.), ground doves (multiple species) and striated antthrushes (Chamaeza nobilis). It also eats agoutis, acouchies, spiny rats, and other small rodents, but not pacas or other large animals. It also eats short-tailed opossums (Monodelphis spp.), iguanas, golden tegus, other large lizards, and geckos. It also eats bata tree fruits, ripe plantains, and fallen echo tree (Jacaratia sp. [Caricaceae]) fruits. It eats the eggs of Spix's guan (Penelope jacquacu) and those of other birds that nest close to the ground. It eats meat that a jaguar has stashed and the guts of armadillos that Matses have butchered, dragging the guts into the forest from the edge of the village.

Remarks: Most Matses observations about jaguarundis agree with the literature reviewed by Oliveira (1998a) and Giordano (2016), especially with respect to diurnal activity and terrestrial habits (the latter is implied rather than stated directly in Matses interviews). Their observation that “jaguarundis do not travel in large groups” contrasts with unambiguous statements that other felids are solitary (see Matses accounts for ocelots, margays, jaguars, and pumas; above) and is probably explained by the fact that these small cats are sometimes observed travelling in pairs (Giordano, 2016). The repeated mention of lizards and birds as prey by Matses interviewees contrasts with an emphasis on mammals (especially rodents) in much of the jaguarundi dietary literature; because the latter is mostly based on observations from Central America and southeastern Brazil, it is possible that Amazonian populations are distinctive in this respect. Frugivory— not previously mentioned in the jaguarundi literature—is another dietary trait described by the Matses that may also be distinctively Amazonian. Alternatively, if jaguarundis consume only fruit pulp (without swallowing seeds), this dietary component may have been missed in published studies based on analyses of scat. Jaguarundis are the only cats that the Matses report to eat birds' eggs and carrion.

Voucher material (total = 3): Nuevo San Juan (MUSM 11171, 13149), Orosa (AMNH 74116).

Other interfluvial records: Actiamë (Amanzo, 2006), Choncó (Amanzo, 2006), Divisor (Jorge and Velazco, 2006), Jenaro Herrera (Pavlinov, 1994), Quebrada Pobreza (Escobedo-Torres, 2015), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006).

Identification: Of the three tayra specimens known to have been collected in the Yavarí-Ucayali interfluve, one (AMNH 74116) is a juvenile, and another (MUSM 11171) cannot now be located. Fortunately, the remaining specimen (MUSM 13149) consists of the wellpreserved skin and skull of a fully adult individual that exhibits all the diagnostic external and craniodental traits attributed to the species by authors (e.g., Husson, 1978). Like many other tayra skins from northeastern Peru, MUSM 13149 has a grizzled-brownish head that does not contrast abruptly in coloration with the fur of the shoulders and middle back, the brownish tones of these regions darkening posteriorly and laterally to merge with the blackish pigmentation of the limbs, flanks, hindquarters, and tail. There is a small selfcream marking on the throat.

Tayras exhibit geographic variation in pelage color that is reflected in the description of numerous subspecies; Wozencraft (2005), for example, recognized eight, some of which have multiple synonyms. Our material perhaps represents the nominal form that Lönnberg (1913) called Eira barbara peruana Tschudi, 1844, but analyses of mtDNA sequence data (Ruiz-García et al., 2013) suggest an almost complete lack of phylogeographic structure in this species. Although a trinomial nomenclature of tayras seems pointless at the present time, we note that genetic data are currently lacking from Central American and Atlantic Forest populations, which might yet be shown to be taxonomically distinct. Morphometric data from our adult female voucher (table 13) compare closely with homologous dimensions of Surinamese specimens (Husson, 1978: table 45) and reinforce our impression that Amazonian tayras comprise a single, genetically cohesive, undifferentiated species.

Ethnobiology: The Matses name for the tayra, batachued (“one that likes sweet food”), derives from their observation that tayras eat sweet wild tree fruits and ripe plantains and papayas from Matses swiddens. There are no archaic synonyms or overdifferentiated varieties, and the Matses term differs from the name for tayra in other Panoan languages. Some Matses include the tayra in the category bëdi, which otherwise designates felids and wild canids.

The tayra is not eaten by the Matses, who consider it a pest and often call it “thief “because it frequently feeds on plantains and papayas in Matses swiddens. Tayras also eat chickens that are ranging at the edge of the village during the day, but they do not raid coops. When dogs pursue a tayra, the tayra sometimes bites the dogs. Tayras are not kept as pets.

Matses with young children avoid having any contact with or even looking at tayras, lest the tayra's spirit make their children ill. Symptoms of tayra sickness include a high fever (but not constant thirst, as is case with contagions induced by felids). To treat this ailment, certain medicinal plants (“tayra medicine”) are collected, and the sick child is bathed with an infusion of their leaves. It is noteworthy that, unlike felid-induced contagions, tayra sickness is not treated with jaguar medicine, suggesting that the folk-taxonomic association of tayras with other members of the bëdi category is not strong.

Matses natural history: The tayra has a dark body and a light-colored head and neck. It has the shape of a dog with long neck. It has a furry tail. It has a distinctive but not strong smell.

Tayras are found in all habitat types, including upland and floodplain forest, and in primary and secondary forest. They are frequently encountered in secondary forest while walking to swiddens, in primary forest while hunting, and on the banks of rivers and streams while traveling by canoe or motorized boat.

The tayra is diurnal. It walks on the ground and also climbs high up in trees. It comes to swiddens to eat ripe plantains, bananas, and papayas. It stashes plantains at the base of a tree and covers them with leaves. It eats fallen fruits on the ground and up in trees. It can be heard rustling branches as it climbs through trees. It often walks on the trunks of fallen trees and defecates on the fallen trees. Its feces often have many seeds of fruits. It climbs high up in trees when it sees people. It sleeps in the same hole in a tree every night.

Tayras are usually solitary, but they also travel in pairs, trios, or sometimes larger groups. They give birth to two young in a den in a hole in a tree.

Jaguars and pumas eat tayras.

The tayra snarls.

Tayras eat all types of sweet things, including ripe bananas, plantains, papayas, and wild dicot tree fruits, such as those of diden këku (Couma macrocarpa [Apocynaceae]), bata (Pseudolmedia spp. and/or Maquira spp. [Moraceae]), and këku (Parahancornia peruviana [Apocynaceae]). They also eat the fruits of cecropia trees (Cecropia spp. [Moraceae]). They drink honey from beehives. They also eat meat, particularly agoutis, acouchies, spiny rats, lizards, tinamous, bird eggs, and hatchlings. (One informant said he saw a large group of tayras chase a gray brocket deer, kill it, and start eating it.)

Remarks: Matses interviews about tayras include many of the salient facts about this versatile diurnal omnivore mentioned in the literature reviewed by Presley (2000), including its use of every forest stratum from ground level to canopy. Additionally, Matses observations confirm the tayra's curious habit of caching fruit stolen from gardens (Soley and Alvaro-Díaz, 2011), and they provide novel information about predation on this species by jaguars and pumas. However, perhaps the most interesting aspect of tayra biology contained in these accounts and in Matses interviews about Dasypus pastasae (see above) are the suggestions that tayras hunting in groups can kill larger prey than solitary tayras can subdue. Although the notion of tayras attacking ungulates seems implausible, this behavior was previously reported by Villa (1948), who witnessed a solitary individual chasing a deer; if one of our informants is to be believed, tayras hunting deer cooperatively are sometimes successful.

Voucher material (total = 1): Nuevo San Juan (MUSM 15157).

Other interfluvial records: Río Yavarí- Mirím (Salovaara et al., 2003).

Identification: Our single grison specimen from the Yavarí-Ucayali interfluve consists of the skin and skull of a young adult female. The pelage markings of this specimen correspond exactly with those of topotypical (Surinamese) material described by Husson (1978: 291–292), and most of its measurements (table 13) fall within the range of morphometric variation reported for females of the species by Bornholdt et al. (2013: table 4).¹⁰ Among other (nonmetrical) traits that distinguish Galictis vittata from its southern congener (G. cuja), MUSM 15157 has a well-developed metaconid on the lower first molar. Several subspecies of G. vittata are recognized as valid by some authors (e.g., Yensen and Tarifa, 2003; Wozencraft, 2005), but no compelling empirical basis for a trinomial nomenclature seems to have been published.

Ethnobiology: Although the grison's Matses name is bosen uşhu (literally, “white otter”), the Matses do not consider grisons to be a type of otter. There are no archaic synonyms or overdifferentiated varieties. Many Matses have never seen a grison, and only a limited number of Matses are familiar with its name. Some know of it through the accounts of those who have seen it.

The grison is of no economic importance to the Matses.

Because grisons are so rarely encountered, the Matses have no specific beliefs regarding them and no special medicinal plants for contagions by a grison spirit. However, Matses medicine men say that grisons could make a child ill if touched or looked upon and that “Neotropical otter medicine” would probably be effective.

Matses natural history: The grison is black with a light-colored back. Its teeth are small. They are found near streams, but they are not aquatic like otters. They can swim. (One interviewed Matses hunter said that he saw one catch and eat an agouti).

Remarks: Matses informants had little to say about this seldom-encountered species, but its pursuit of agoutis is corroborated by the literature cited in Yensen and Tarifa's (2003) review.

Our voucher was shot by a Matses hunter, who encountered a group of three individuals fighting over a dead spiny rat (Proechimys sp.) in upland primary forest near a muddy mineral lick during the day.

Voucher material: None.

Other records: This report (Matses observations).

Identification: No other Amazonian mammal resembles this species, which we judge to be present in the Yavarí-Ucayali interfluve on the basis of unambiguous Matses observations. The local form is Mustela africana stolzmanni Taczanowski, 1881, which was described from a specimen collected at Yurimaguas, a lowland site about 200 km west of our region. Technical descriptions and measurements of the holotype and other referred material were provided by Hall (1951).

Ethnobiology: Only three Matses men whom we interviewed have seen an Amazon weasel, and each had encountered them only once. As such, the name we give here, mayanën opampi (“demon's little dog”) is far from established. However, it has some currency among people who had heard of it from those who have seen it. One informant called it bosenëmpi, the term for the Neotropical otter with the diminutive suffix -mpi (i.e., “little otter”).

The Amazon weasel is of no economic or cultural importance to the Matses. Since most Matses are not familiar with the species, they have no established beliefs about it.

Natural history: Amazon weasels are small, the size of an acouchy. They are brown, the color of an otter. They have an extremely strong and foul smell. They run very fast. One informant said that his hunting dogs were not able to catch one. Another informant said he found three together in a burrow at the base of a tree.

Remarks: This exceptionally rare (or elusive) species is seldom encountered, even by field researchers with many years of Amazonian experience, so it is not surprising that only a few Matses have seen one.

Voucher material (total = 1): Nuevo San Juan (MUSM 11172).

Other interfluvial records: Anguila (Escobedo- Torres, 2015), Choncó (Amanzo, 2006), Río Yavarí (Salovaara et al., 2003), Río Yavarí- Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006).

Identification: Following van Zyll de Jong's (1972) revision of the Nearctic and Neotropical river otters (previously treated as congeneric with Old World Lutra), three subspecies of Lontra longicaudis have been consistently recognized: L. l. annectans (Major, 1897) from Central America and trans-Andean South America, L. l. enudris (Cuvier, 1823) from Amazonia, and L. l. longicaudis (Olfers, 1818) from southeastern Brazil and the La Plata drainage. Analyses of mtDNA sequence data (Trinca et al., 2012) suggest that annectans should perhaps be recognized as a distinct species, but there is evidence of genetic exchange between Amazonian populations (referred to enudris) and La Platan populations (referred to longicaudis). Although Feijó and Langguth (2013) recently suggested that all the South American subspecies of L. longicaudis (sensu van Zyll de Jong, 1972) be recognized as full species, we prefer to maintain current usage until the reproductive isolation and/or diagnosability of these nominal taxa can be more convincingly established. Cranial measurements of our voucher specimen (table 13) compare closely with those of almost-topotypical specimens of L. l. enudris (from Surinam; Husson, 1978: table 46).

Ethnobiology: The name for the Neotropical otter is bosen, which is monomorphemic and widespread in other Panoan languages. There are no archaic synonyms or overdifferentiated varieties.

Otters are not eaten by the Matses, nor are they usually kept as pets.

The spirit of a Neotropical otter can make children ill if their parent looks at one. Contagion by an otter spirit causes high fever, which is treated with certain medicinal plants (“Neotropical otter medicine”). Other medicinal plants effective against sickness caused by giant otters (“giant otter medicine”) are also said to cure contagion by Neotropical otters, suggesting that, despite the lack of any linguistically labeled category for otters and the complete dissimilarity between the Matses names for Lontra longicaudis and Pteronura brasiliensis, the Matses recognize an association between these taxa at some level.

Matses natural history: The Neotropical otter has a smooth, slick body, with a somewhat flat tail and a head like a dog's. Its fur is shiny and gray when wet. The underside of its neck is light-colored. It has thick whiskers. It has short legs and webbed feet with which it cannot run quickly on land.

The Neotropical otter lives along rivers, large and small streams, streams with muddy bottoms or sandy bottoms, and in lakes and flooded forest.

The Neotropical otter is diurnal. It spends much of the day swimming along streams looking for fish. It travels short distances overland to search for fish at other streams. It makes the water in small streams turbid as it chases fish and looks for wolffishes (Hoplias spp. [Erythrinidae]) lying in the streambed. It feels inside submerged hollow logs and under submerged logs for armored catfishes and tëpuşh fish (Erythrinus erythrinus [Erythrinidae]). When it catches a fish, it eats it sitting on a log that is in the water, or on the bank. It makes a loud gnawing sound as it eats. The otter sleeps in the undercut banks of streams, in holes made in the bank by motmots (birds in the family Momotidae), or other sheltered places along the bank. It does not sleep in the same place every night, and it does not make burrows to sleep in (as it does to give birth). When it sees people it flees quickly, swimming underwater.

Neotropical otters are usually solitary, but they are also found in groups of two or three. When two or three are together, they play by chasing each other and taking fish from each other. The female gives birth to two young in a shallow burrow dug into the bank of a stream or river. She continually goes to catch fish, eats them, and then comes back to suckle her young. When the young get older, she brings fish to the den for them to eat.

No predators eat Neotropical otters.

The Neotropical otter whines, saying “weee weee weee.” It barks saying “kuesak kuesak kuesak.” The Neotropical otter eats all kinds of fish, including armored catfish, large pimelodid catfishes, tëpuşh (Erythrinus erythrinus [Erythrinidae]), pone (Hoplias spp. [Erythrinidae]), and bëdichued (Leporinus spp. and/or Schizodon spp. [Anostomidae]). It also eats crabs and shrimp.

Remarks: Matses interviews about Neotropical otters are consistent with the results of most published studies of this widespread taxon (e.g., Kasper et al., 2008; Silva, 2010), notably with respect to its fondness for slowmoving benthic prey (such as armored catfish and erythrinids), but Amazonian field studies are almost nonexistent, so these observations also provide novel details of habitat use and behavior that may be peculiar to the local subspecies (Lontra longicaudis enudris; see above). Matses accounts that Neotropical otters make transient and opportunistic use of makeshift shelters (including burrows made by other species) is interesting by comparison with the fixed den sites of sympatric giant otters (see below). Another noteworthy point of comparison based on Matses observations of these species is that female Neotropical otters with newborn young in nursery burrows are said to forage for themselves, whereas female giant otters with nursing young are said to be provisioned by the male.

A published range map (Larivière, 1999: fig. 3) that shows Lontra longicaudis as absent throughout most of western Amazonia is inconsistent with the documented presence of this species in the Yavarí-Ucayali interfluve and at many other western Amazonian inventory sites (e.g., Balta, Cocha Cashu/Pakitza; Voss and Emmons, 1996). An accompanying statement that “Lontra longicaudis favors clear, fast-flowing rivers and streams and may be absent or rare from sluggish, silt-laden lowland rivers,” (Larivière, 1999: 2) is likewise impossible to reconcile with the presence of Neotropical river otters throughout the flat, sedimentary landscapes of western Amazonia where clear, fastflowing rivers and streams are virtually nonexistent, and where this species is known to forage in lentic habitats (e.g., oxbow lakes and flooded forest).

Voucher material (total = 1): Nuevo San Juan (MUSM 11173).

Other interfluvial records: Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999).

Identification: Giant otters are externally and cranially unmistakable (Husson, 1978; Emmons, 1997) and no conspicuous morphological differences have been reported among the Amazonian populations traditionally referred to Pteronura brasiliensis brasiliensis. Apparently, the same mtDNA phylogroup—the “Amazon/Orinoco/Guianas” clade of Pickles et al. (2011)— extends from western Amazonia to French Guiana (the type locality; Husson, 1978), and craniodental measurements of our single voucher specimen (table 13) compare closely to those of nearly topotypic material from Surinam (Husson, 1978: table 47).

Ethnobiology: The Matses name for the giant otter is onina. It is probably onomatopoetic and does not occur in other Panoan languages. There are no archaic synonyms or overdifferentiated varieties.

The giant otter is of no economic importance to the Matses. They are never kept as pets.

The spirit of a giant otter can make children ill if their parent looks at one. Matses with children formerly made great efforts to avoid seeing giant otters, but now that the Matses frequently travel by boat, it is almost impossible to avoid seeing them. Giant otter sickness causes high fever (as does contagion by Neotropical otters), which can be treated with particular medicinal plants (“giant otter medicine”). Other plants used to treat illness caused by Neotropical otters are also used to treat contagion by giant otters.

Matses natural history: The giant otter has a head like a paca's and teeth like a jaguar's. It has a light-colored patch on the front of its neck. It has thick whiskers, a flat tail, short legs, and webbed feet. It is much larger than the Neotropical otter.

Giant otters are always near water, in ox-bow lakes, rivers, and large streams, but not in small streams, except near their mouths. They also catch fish in flooded forest.

Giant otters are diurnal. They sleep at night in an undercut bank or some other sheltered place along a river, stream, or lake. They spend much of the day chasing fish, swimming very quickly. They travel far, swimming along rivers and streams. They can swim underwater and are always poking their heads out of the water. They fish for a long time in deep river curves. They play in the water chasing each other.

They make clearings on riverbanks and lakeshores where they eat the fish they have caught. Such clearings are free of all vegetation and look as if they had been swept. Giant otters come back to the same clearings to eat. One clearing is close to their den, and others are further off, often on the bank of a deep curve of a river, or at the mouth of a stream. Such clearings are littered with fish bones and scales and smell of rotting fish.

When giant otters see people, they dive and swim away underwater. If they have young, they become fierce when they see people, baring their teeth and growling. When one imitates their call, they come calling.

Giant otters live in packs of five to 20 individuals. The females give birth to two young in a burrow, the entrance of which is in the undercut bank of a stream. Males catch fish and take them to feed the young while the female stays with the young. The den stinks like rotten fish and has many flies. Only the female with young sleeps in the den. When the young are strong enough to swim, they abandon the den and sleep in other places (not burrows).

Giant otters fish in the presence of dolphins. No predators kill giant otters.

Giant otters have a loud squealing call that the Matses imitate as “waa waa waa.”

Giant otters eat mostly fish, all types of fish including inchishchued (Brycon spp. [Characidae]), bëdichued (Leporinus spp., Schizodon spp. [Anostomidae]), wolffishes (Hoplias spp. [Ery thrinidae]), armored catfish, and large pimelodid catfishes. They also eat crabs.

Remarks: Matses interviews about giant otters include most of the essential natural history facts about this remarkable species, including its diurnal activity, piscivorous diet, highly social behavior, almost predator-free existence, construction of vegetation-free campsites on river banks, exclusive use of a birthing den by females and newborn young, and aggressive defense of family groups against human intruders (Duplaix, 1980; Carter and Rosas, 1997; Duplaix et al., 2015). The Matses observation that males provision females with nursing young is not reported in the literature we consulted. Their interesting observation that giant otters fish in the presence of dolphins hints at, but does not explicitly confirm, the possibly cooperative association between Pteronura and Inia suggested by Defler (1983).

Voucher material (total = 2): Nuevo San Juan (AMNH 268247; MUSM 11174).

Other interfluvial records: Anguila (Escobedo-Torres, 2015).

Identification: Our two voucher specimens conform qualitatively to Helgen et al.'s (2013) diagnosis of Bassaricyon alleni, the only olingo species known to occur in Amazonia. Additionally, the external and craniodental measurements of our specimens (table 14) fall within the range of morphometric variation in B. alleni tabulated in that study. No subspecies of B. alleni are currently recognized, and the trivial genetic distance between sequenced specimens from Guyana and Peru (ca. 1.3% at the cytochrome- b locus; Helgen et al., 2013) suggests that even widely separated Amazonian populations are not significantly differentiated.

Ethnobiology: The Matses name for the olingo is şhëmën, a monomorphemic term that is common in other Panoan languages as a name for the olingo and/or the kinkajou. Only a small number of Matses are aware that olingos and kinkajous are different animals. Those who recognize them as distinct note the nonprehensile, ringed tail of the olingo and its slightly different vocalization. Those who are not aware that these are two species consider the name for the kinkajou, kuichikkekid, to be a synonym of şhëmën.

The olingo is of no economic importance to the Matses.

Contagion by an olingo spirit causes a very high fever in children (like the illness caused by a kinkajou spirit).

Matses natural history: The olingo is like a kinkajou, but has a nonprehensile and ringed tail and a smaller head. Its call is very similar to but softer than that of the kinkajou. The olingo's call is heard less frequently than the kinkajou's, and is seldom heard in secondary forest. (The remaining natural history information that Matses interviewees provided for the olingo is essentially the same as that provided for the kinkajou.)

Remarks: Both of our specimens were shot at night in trees (at estimated heights of 15 and 35 m above the ground) in primary upland forest.

Voucher material (total = 8): Boca Rio Yaquerana (FMNH 88877, 88878), Nuevo San Juan (AMNH 268248; MUSM 11176, 11178), Quebrada Esperanza (FMNH 88879–88881).

Other interfluvial records: Choncó (Amanzo, 2006), Itia Tëbu (Amanzo, 2006), Quebrada Pobreza (Escobedo-Torres, 2015), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006), Wiswincho (Escobedo-Torres, 2015).

Identification: Measurements of our voucher material (table 15) fall within the range of variation for Nasua nasua tabulated by Decker (1991), at least to the extent that her unexplained measurement abbreviations can be deciphered,¹¹ but morphometric variation in this species broadly overlaps with that of the somewhat larger white-nosed species (N. narica), so quantitative comparisons are of limited diagnostic value. Of the qualitative characters said to distinguish N. nasua from N. narica in Decker's revision, our material lacks a whitish patch of postrhinarial fur (present in N. narica), the anterior alveolar foramen is visible anterior to the infraorbital foramen (the anterior alveolar foramen is concealed inside the infraorbital canal of N. narica), and a well-developed postorbital process of the jugal is present (this process is absent or indistinct in N. narica). By contrast, we were not consistently able to distinguish our material from N. narica using the other craniodental and pelage characters alleged to diagnose these species.

Although 41 nominal taxa are currently regarded as synonyms or subspecies of Nasua nasua (see Wozencraft, 2005), the empirical basis for a subspecies classification has never been established. Following Cabrera (1958), western Amazonian coatis (which tend to have very dark pelage) are usually referred to Nasua nasua dorsalis Gray, 1866, but the type locality of dorsalis is effectively unknown; among the nominal taxa commonly listed as synonyms of dorsalis, the oldest that might apply to any population from western Amazonia is juruana Ihering, 1911. In the absence of any assessment of geographic variation in phenotypic or molecular traits, however, we are not persuaded of the need for a trinomial classification of N. nasua. A comparison of our measurement data (table 15) with measurements of Surinamese specimens identified as N. nasua vittata (in Husson, 1978: tables 41, 42), for example, does not suggest any substantial morphometric divergence between populations from opposite sides of Amazonia.

Most coati skins from the Yavarí-Ucayali interfluve are rich reddish brown lined with black, but the pelt of FMNH 88881 (an old adult male) is predominantly blackish, the black almost obscuring the banding pattern on the tail. As usual for this species, old male skulls have tall sagittal crests, widely flaring zygomatic arches, and huge canines by comparison with female skulls (which lack sagittal crests and have narrower zygomatic arches and much smaller canines; fig. 16C).

Ethnobiology: The coati is called tsise, a monomorphemic term common in the Panoan family. The Matses recognize two named varieties: tsisedapa (“big coati”) and tsisempi (“small coati”). The large variety occurs in smaller packs (up to about 10) and the small variety is darker and runs in larger packs (up to about 15). The Matses have no archaic synonyms for the coati.

The coati is a game animal of secondary importance. Traditionally only old people ate coatis. Today, after having seen that non-Indians eat them, some younger Matses eat them, but only roasted and if they have a lot of fat. Sometimes coatis injure dogs by biting them when chased. Coatis are considered very good pets to keep, not only because young animals readily become tame and are fun to play with, but because pet coatis warn people about pitvipers (e.g., Bothrops spp.) near the village. Coatis give a warning call (“tsa tsa tsa”) when they find a venomous snake, and then someone goes and kills it.

Young people do not eat coatis lest they become lethargic. Coati spirits sometimes make children ill, causing them to have a high fever, with is treated by bathing the sick child with certain medicinal plants (“coati medicine”).

Matses natural history: The coati has a ringed tail that it carries raised up vertically as it walks on the ground. It has a dark-colored body and a light-colored face. It has a long snout, small ears, and big claws. Seasonally it has a lot of fat. It has a very strong smell, such that one can easily know that coatis have recently passed by.

Coatis are found in all types of habitats, including upland and floodplain forest, and in primary and secondary forest. They are encountered frequently while hunting.

Coatis are strictly diurnal. They sleep together up in trees at night. They lie together in trees to rest when it is dry. They eat fruits up in the trees and also forage for fallen fruits on the ground. They search the ground for earthworms, and when they find a place with many earthworms they root there for a long time. They eat beetle grubs that feed on the rotting pith of fallen palm trees. They dig into rotten logs with their noses to search for invertebrates.

When they see people from far off, they yell, saying “kosh,” drop to the ground, and flee running on the ground. If they are on the ground when they see or hear people, they climb partway up a tree to get a good look at the person(s), yelling “kosh, kosh, kosh,” and then drop to the ground and flee running on the ground.

Coatis live in packs of up to about 15 individuals. Sometimes only two or three are seen traveling together. To give birth, coatis make big nests by breaking off many small branches with the leaves still attached and weaving them together in the crotch of a tree branch. Each female that is going to give birth makes a separate nest in the same tree. They give birth and suckle their young in the nest. Once they get stronger, the females take their young down to the ground to forage with the rest of the pack, and then at dusk they carry them back up to the nest. Once the young are strong enough to grasp tightly, they begin to come down to the ground on their own, and eventually the nests are abandoned. The nests are often made in a fruiting tree.

Jaguars and pumas eat coatis.

Coatis bark saying “tsat tsat tsat tsat,” and hiss saying, “tse, tse, tse.”

Coatis eat all sorts of things. They eat dicot tree fruits, including those of bata (Pseudolmedia spp. [Moraceae]). They eat the mesocarp of the fruits of isan palms (Oenocarpus bataua [Arecaceae]) and swamp palms (Mauritia flexuosa) that have ripened and fallen to the ground. They eat earthworms, armored millipedes, round millipedes, scorpions, beetle grubs that feed on palm pith, and grubs that live in the soil.

Remarks: Matses observations about Nasua nasua agree in most essential details (e.g., diurnal activity, sociality, omnivory, construction of arboreal nursery nests, escape behavior) with Kaufmann's (1962) classic study of the Central American species (N. narica), and with the scattered natural history literature on N. nasua (reviewed by Gompper and Decker, 1998). A significant omission from our interviews is any mention of solitary males.¹² Additionally, these accounts describe nursery-nesting behavior in greater detail than in any previous report about Nasua spp. (including Olifiers et al., 2009), and they provide new information about fruit species and invertebrate taxa consumed in Amazonia, where the foraging habits of coatis have not previously been studied.

Voucher material (total = 3): Nuevo San Juan (AMNH 268249; MUSM 11179) Orosa (AMNH 73765).

Other interfluvial records: Actiamë (Amanzo, 2006), Choncó (Amanzo, 2006), Itia Tëbu (Amanzo, 2006), Quebrada Pobreza (Escobedo- Torres, 2015), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006).

Identification: Kinkajou specimens collected in the Yavarí-Ucayali interfluve conform closely to Husson's (1978: 285–287) description of topotypical material from Surinam, and measurements of our vouchers (table 14) broadly overlap the range of variation in homologous dimensions reported from kinkajous collected in the Guianas (Husson, 1978; Voss et al., 2001). Side-by-side comparisons of crania from eastern Peru and French Guiana suggest that the former have somewhat larger auditory bullae, but no other consistent differences are apparent. Therefore, based on the phenotypic evidence at hand, we are quite confident of this identification and would even assign our vouchers to the nominotypical subspecies if a trinomial identification were deemed necessary. Remarkably, however, DNA sequence data suggest that western Amazonian and Guianan kinkajous differ by as much as 7%–9% at the mitochondrial cytochrome-b locus according to Nascimento et al. (2016). Those authors correctly point out that such high levels of sequence divergence are often found between full species, and they reasonably suggest that additional studies based on other genetic loci are needed to assess the possibility that several valid taxa are represented among the nominal forms currently treated as synonyms or subspecies of P. flavus. In this context, our morphological comparisons of Peruvian and Guianan specimens are inconclusive, but if a different name were eventually needed for our material, the geographically closest nominal taxon is chapadensis Allen, 1885, based on a type from Mato Grosso, Brazil.

Ethnobiology: The Matses name for the kinkajou is kuichikkekid, which can be analyzed as meaning “one that says ‘kuichik’” (“kuichik” is the Matses rendition of the vocalization that kinkajous are often heard to make from the treetops at night). The name is often shortened to kuichik. It has no archaic synonyms or named overdifferentiated varieties (but see the Ethnobiology entry for olingos, above).

The Matses do not kill or eat kinkajous, they do not raise them as pets, and they have no other interest in them. Although most Matses have never seen a kinkajou, most have heard kinkajous vocalizing in the treetops at night. One of the few occasions when the Matses get a close look at a kinkajou is when they find one when felling trees for a swidden.

Contagion by a kinkajou spirit causes a very high fever in children.

Matses natural history: The kinkajou is like a small dog, but with a prehensile tail and larger eyes. It has a short rostrum, large eyes, and ears like a jaguar's. Its body is reddish gray.

Kinkajous are arboreal. They almost never come down to the ground. They can be found in all types of habitat, including floodplain and upland forest, and primary and secondary forest.

Kinkajous sleep in dicot tree holes and holes in the trunks of bottle palms (Iriartea deltoidea [Arecaceae]).

The kinkajou is nocturnal. It is almost never active in the daytime. It climbs around on the branches of trees looking for fruits, calling out “kuichik.” As it moves around up in the trees it rustles the branches lightly. Kinkajous come out of their holes during the day when a hunter climbs up a tree (to kill a sloth, retrieve an arrow, recover a killed monkey, etc.) and may try to bite him.

Kinkajous are solitary.

Kinkajous call out at night repeatedly saying “kuichik.”

Kinkajous eat mostly dicot tree fruits, especially those of këku (Parahancornia peruviana [Apocynaceae]) and bata (Pseudolmedia spp. [Moraceae]). They also eat bottle palm (Iriartea deltoidea) fruits. They also eat the eggs of toucanets (Selenidera sp. [Rhamphastidae]) and other birds, baby birds, and adult passerines. They also eat katydids.

Remarks: Matses observations about kinkajous are mostly consistent with published field studies of this species (e.g., Julien-Laferrière, 1993; Kays, 1999; Kays and Gittleman, 2001)—notably with respect to its exclusively nocturnal-arboreal activity and predominantly solitary lifestyle—but they are notably discrepant in one respect. Whereas published dietary studies suggest that Potos flavus is entirely frugivorous (Julien-Laferrière, 1999; Kays, 1999) or partially insectivorous (Bisbal, 1986; Redford et al., 1989), the Matses claim that it also eats bird eggs, nestlings, and adult birds. Given that captive kinkajous are known to eat meat and eggs (Ford and Hoffmann, 1988), Matses observations are not implausible, but the discrepancy is of interest. Although Matses hunters could have mistaken olingos for kinkajous, olingos are also thought to be frugivorous (Kays, 2000), so either the Matses are wrong, or there is still more to be learned about the diets of arboreal procyonids.

Voucher material (total = 1): Boca Rio Yaquerana (FMNH 88876).

Other interfluvial records: San Pedro (Valqui, 1999)

Identification: The single specimen of Procyon that we examined from the Yavarí- Ucayali interfluve (FMNH 88876) consists of the well-preserved skin and skull of a young adult female (with fully erupted permanent dentition but unworn molars). These elements exhibit all of the diagnostic qualitative traits of the subgenus Euprocyon and its single referred species, P. cancrivorus, including reversed fur on the nape of the neck, lack of a distinct lingual accessory cusp on I3, the isolated hypocone of P4, and massively developed molars with robust cingula and rounded-coniform cusps (Hollister, 1915).

The external and craniodental dimensions of FMNH 88876 compare closely to those of almosttopotypical (Surinamese) specimens of Procyon cancrivorous tabulated by Husson (1978) but are slightly smaller for least interorbital breadth and zygomatic breadth and slightly larger for breadth of M1. The differences are not large enough, however, to suggest any substantial geographic variation among Amazonian populations of this species, all of which could be referred to the nominotypical race if a trinomial nomenclature were adopted. Selected measurements of FMNH 88876 are: headand- body length, 494 mm; length of tail, 307 mm; hind foot, 143 mm; ear, 55 mm; condylobasal length, 127.0 mm; nasal length, 36.0 mm; least interorbital breadth, 25.2 mm; least postorbital breadth, 27.0 mm; zygomatic breadth, 81.8 mm; breadth of braincase, 56.8 mm; maxillary toothrow (C1–M2), 52.2 mm; breadth of M1, 13.8 mm.

Ethnobiology: The crab-eating raccoon is called tsisebiekkid (“one that is like the coati”). Although the name indicates a perceived similarity to the coati, the raccoon is not considered a type of coati. Few Matses have seen a crab-eating raccoon, and not all are familiar with its name. Some of our informants had heard of them, but none had seen one. The raccoon is of no economic importance to the Matses.

Since raccoons are so rarely encountered, the Matses have no specific beliefs regarding them and no special medicinal plants for contagions by a raccoon spirit. However, Matses medicine men say that they could make a child ill if touched or looked upon, and certain medicinal plants (“coati otter medicine”) would be expected to be effective.

Matses natural history: The raccoon is similar to the coati, but bigger and darker colored.

Voucher material (total = 5): Boca Rio Yaquerana (FMNH 88794), Nuevo San Juan (MUSM 11181), Orosa (AMNH 73766, 74118, 74119).

Other interfluvial records: Divisor (Jorge and Velazco, 2006), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006), Wiswincho (Escobedo- Torres, 2015).

Identification: Tapir specimens collected in the Yavarí-Ucayali interfluvial region conform to the typical morphology of Tapirus terrestris described by Hershkovitz (1954) and Husson (1978), and they do not include any examples of the unusual cranial phenotype described by Hagmann (1908) and Cozzuol et al. (2013). Measurements of our material (table 16) are all within a few millimeters of homologous values obtained from Surinamese specimens (Husson, 1978: table 55), suggesting little geographic variation in cranial dimensions across vast Amazonian landscapes despite modest mtDNA heterogeneity in this species (Thoisy et al., 2010; Ruiz-García et al., 2016).

Ethnobiology: The principal name for the tapir is nëishame, which can be analyzed as meaning “large game animal.” There are three archaic synonyms: awad (a monomorphemic pan-Panoan term), wisu (an archaic term that also means “black”), and danchish (synchronically unanalyzable, but seems to include the prefix dan-, meaning “knee”). The tapir is the only animal with pet vocative¹³ terms: dampiada and choada (both synchronically unanalyzable, but the first also seems to include the prefix dan-). In the language used in the Matses' komok ceremony, the tapir is called dëpachi, a term that means “soft snout.”

The Matses recognize three types of tapirs: nëishamedapa (“large tapir”), nëishame çhëşhe (“black tapir”), and nëishame mebëdi (“stripedforeleg tapir”). According to Matses hunters, the type with striped forelegs is the smallest of the three varieties, prefers upland forest (as opposed to floodplain forest), and when chased by dogs runs without tiring out. The large variety is found along large rivers and is the type that most readily takes refuge in the water when chased by dogs.

The tapir is a principal game animal for the Matses, although tapirs are killed infrequently compared to other game species. In addition to providing a bonanza of meat, a butchered tapir is much appreciated for its fat, which is carefully rendered for frying manioc and plantains, and to make an oily broth thickened with grated manioc. Tapirs are desirable pets that quickly become tame when captured as juveniles. Even as adults, pet tapirs roam the village peacefully.

Tapirs are killed by the Matses in various ways. A hunter may happen upon a tapir as it sleeps on the forest floor and shoot it with a shotgun (formerly it would have been shot with a bow and arrows). Or, a hunter may find tapir spoor and track the animal to where it is sleeping or feeding, and then shoot it. Hunters also encounter tapirs at mineral licks. When a tapir has been scared off or runs off when shot and injured, the hunter whistles, imitating the tapir's call. Interestingly, the tapir often replies with a whistle or comes to where the hunter has whistled, even if it has been shot. Often a shot tapir escapes, and killed tapirs are sometimes found to have healed shotgun wounds.

Tapirs are also hunted with dogs. The best hunting dogs will chase a tapir, nipping at its legs. If the dogs do this, or if the tapir tires out, it may take refuge by submerging itself in a deep bend of a stream. (If the dogs do not follow the tapir closely or far enough, the tapir will just keep running and the hunter will not be able to catch up.) Once the hunter reaches the place where the tapir is submerged, he kills it. In the past the tapir would be killed in such a situation with a spear as it lay underwater. If the stream is too deep, the hunter will try to get the tapir to move away from the deep river bend and shoot it (with a shotgun or, formerly, with arrows) when it emerges into a shallower part of the stream.

Now that the Matses have acquired flashlights, they sometimes wait for tapirs at mineral licks at night. A hunter may visit a mineral lick during the day and find fresh tapir tracks. Since tapirs often return to the same mineral licks, the hunter builds a platform about 2 m off the ground and returns at dusk to wait, sitting on the platform.

An additional method for killing tapirs, which is not employed frequently anymore, is to build a trap along a tapir path, usually at the edge of a mineral lick or where tracks reveal that a tapir comes every night to feed on fallen fruits. A pole some three meters long is cut from a sapling and lashed with epiphyte stems to a tree that is right next to the tapir path; the pole is lashed parallel to the tree, with the lashed end at the top and the bottom end about 40 cm above the ground (about tapir-chest height). Next, a daggerlike bamboo blade about 50 cm long is lashed to the pole (pointing toward the tree). Thus armed, the pole is bent away from the tree and held in place by an ingenious trigger mechanism actuated by a trip wire (figs. 17, 18). When the tapir walks by, it hits the trip wire with its foreleg, releasing the spring-loaded pole with the bamboo blade. If the bamboo blade stabs it in the chest, the tapir will die close to where the trap was set. Often, however, the tapir is not mortally wounded. The trap is checked every three days or so, and if the trap has been sprung, the hunter follows the blood trail to find the dead or injured tapir. The tapir may need to be finished off with a club or by shooting.

Because an adult tapir is too large to be carried by a single person, a successful tapir hunter will return to his village to recruit other Matses, especially women, to come to the kill site and help butcher the carcass. The tapir is skinned and butchered by the women, after which everyone carries back a portion of the carcass to his or her own household. The hunter who killed the tapir takes home the ribs and other choice portions and then invites others to eat at his house. In this way, everyone in the village partakes of the tapir.

The Matses believe that, while packing tapir meat back to the village, one must not look back over one's shoulder toward the butchering site, lest someone in the family die. Women and young men cannot eat the tail or the part of the rump near the tail, lest they begin to walk bent over like an old man. While waiting at home for a tapir trap to be sprung, the hunter who set the trap follows several dietary restrictions (e.g., he does not eat tortoise or spider monkey meat) and must abstain from sexual intercourse, lest the trap not be sprung or the tapir not be mortally injured. Additionally, men who regularly set tapir traps do not eat the liver and intestines of tapirs. Pregnant women do not eat young tapirs (or young game animals in general), lest they grow weak while giving birth. Young men likewise do not eat young tapirs (or young animals in general), lest they become cowards. Thus, only old people may safely eat immature tapirs.

When a hunter kills, eats, or sees a tapir, the spirit of the tapir may cause one of his young children to fall ill. Occasionally a tapir's spirit makes a child ill even if there has been no contact with a tapir. The symptoms for contagion by tapir spirits are a high fever and the rolling of eyes into the back of the head. When a child exhibits these symptoms after the father has eaten or had contact with a tapir, a medicine man will collect medicinal plants known as “tapir medicine” and bathe the child with an infusion of the leaves.

Matses natural history: The tapir is dark colored. It has large ears with white tips. Its snout is soft and flexible and can be curled upward. Its feet are flat. It has much fat and a thick hide. The male has a large penis.

The tapir uses all types of habitats, including floodplain and upland forest, palm swamps, primary forest, secondary forest from blowdowns, and secondary forest from abandoned swiddens. Tapirs are especially common in the floodplains of rivers and streams. They come to Matses swiddens to eat manioc leaves, but only in swiddens that people do not visit regularly, such as those made at hunting camps.

The tapir is mostly nocturnal. It is more wary when the moon is bright. It sleeps during the day on the ground, often on dry hilltops or at the edge of a treefall. It does not sleep in the same spot every night. It does not walk around in the late morning if the sun is shining, but does when it is cloudy or raining. It travels far, crossing streams and rivers. It has many paths in the forest. It leaves its path, foraging in a large circle, and return to the same place where it had left the path.

It visits mineral licks between 7 PM and midnight, and between 2 AM and dawn (as noted by hunters that wait for tapirs at mineral licks). The tapir eats mud and drinks muddy water at mineral licks (small areas in the forest with poor drainage where minerals collect and the activity of animals make the area muddy). It always returns to the same mineral licks to drink the muddy water; that is, it visits several different mineral licks, but returns to the same ones. There is often a well-worn tapir path leading to a mineral lick. Sometimes several tapirs congregate at mineral licks and a tapir may use a mineral lick together with deer.

The tapir walks around constantly eating the leaves or succulent stems of understory plants. It bends down saplings to eat the leaves. In secondary forest it frequently bends over small Cecropia trees to eat the leaves (but not the stems). It pulls down certain types of soft vines and chews and sucks on them. It drinks water in deep bends of large streams. It defecates in the deep parts of little streams.

The tapir is usually solitary. It gives birth to a single large young. The tapir gives birth in the open, at the edge of a blowdown. It goes to eat without going far and comes back to suckle its young repeatedly. It suckles its young while lying on its side. When the young tapir grows strong, the mother takes it to forage for fruits at night, and the young tapir eats the fruits too. Then the mother takes the young tapir to a good place to sleep during the day. The young tapir leaves the mother when it becomes black (i.e., loses its stripes).

Tapirs are eaten by jaguars, but not by pumas. Large anacondas occasionally capture and eat tapirs. Large tabanid flies are always biting tapirs.

The tapir whistles loudly, saying “pin.” When a Black Caracara (a vulturelike falcon, Daptrius ater) calls out, the tapir answers with this same whistling call, and goes to where the caracara is calling. The young tapir whines saying “chee-oo chee-oo chee-oo” when its mother leaves it. It snorts and stamps its feet. It travels through the forest making a lot of noise as it crushes through the vegetation.

The tapir eats the young leaves of dicot plants and small trees, including those of shuişhën chete (?Alchornea sp. [Euphorbiaceae]) and cecropias (Cecropia spp. [Moraceae]). It also sucks on and eats the stems of soft vines and the leaves of harder vines that it pulls down to the ground. It eats the leaves of epiphytes that grow close to the ground. It also eats the stems (and leaves) of succulent plants.

It eats many types of fallen dicot tree fruits, including those of the rubber tree (Castilla [Moraceae]), dadain (Clarisa racemosa [Moraceae]), figs (Ficus spp. [Moraceae]), diden këku (Parahancornia peruviana [Apocynaceae]), nuëkkid neste tree (Bellucia sp. [Melastomataceae]), nëishame naëşh (unidentified), echo (Jacaratia sp. [Caricaceae]), and pënkad (an unidentified large tree with large fruits that the Matses also eat). It eats the mesocarp of the fruits of swamp palms (Mauritia flexuosa [Arecaceae]) and isan palms (Oenocarpus bataua). It also eats fallen fruits of some types of epiphytes in the family Araceae.

Remarks: Matses interviews about tapirs include many familiar aspects of the biology of this culturally important species, including its marked preference for floodplain habitats, use of paths, nocturnal visitation of mineral licks, solitary behavior, mixed diet of browse and fruit, and curious habit of defecating in water (Salas, 1996; Salas and Fuller, 1996; Henry et al., 2000; Tober et al., 2009; Link et al., 2012). Additionally, several food-plant taxa mentioned by our informants have previously been reported in tapir dietary studies (e.g., Bodmer, 1990; Salas and Fuller, 1996; Henry et al., 2000). However, the Matses state unequivocally that tapirs are killed and eaten by jaguars and large anacondas, whereas Salas (1996) claimed that adult tapirs are immune from predation. Because jaguars in the Pantanal and Cerrado are definitely known to kill tapirs as well as cattle (which are substantially larger than adult tapirs; Cavalcanti and Gese, 2010; Sollmann et al., 2013), and because green anacondas (Eunectes murinus) are also known to prey on tapirs (Martins and Oliveira, 1999), we do not doubt that the Matses are correct, although their accounts do not explicitly report the age of tapirs taken by these formidable predators.

Voucher material (total = 9): Boca Rio Yaquerana (FMNH 88799–88802, 89176, 89177), Nuevo San Juan (MUSM 11182, 11183), Quebrada Esperanza (FMNH 88803).

Other interfluvial records: Actiamë (Amanzo, 2006), Choncó (Amanzo, 2006), Divisor (Jorge and Velazco, 2006), Itia Tëbu (Amanzo, 2006), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006).

Identification: Specimens of Pecari tajacu collected in the Yavarí-Ucayali interfluve externally resemble the widespread Amazonian phenotype (Husson, 1978; Emmons 1997), and craniodental measurements of our material (table 17) broadly overlap those of Surinamese specimens (Husson, 1978: table 58) for most dimensions. Although skulls of P. tajacu from the Yavarí-Ucayali interfluve seem to average a bit larger than conspecific Surinamese specimens, they are much smaller than the skulls of P. maximus measured by Roosmalen et al. (2007: table 1). For example, whereas the mean and standard deviation for condyloincisive length in our material (N = 7, males and females combined) is 211.3 ± 6.0 mm, the homologous dimensions of two specimens of P. “maximus” (sex unknown) are reported as 260 and 262 mm (Roosmalen et al., 2007). Although sample size and sample composition are obviously problematic in this comparison, the estimated mean difference between our material and Roosmalen et al.'s (about 50 mm, equivalent to eight standard deviations) is too large to be easily dismissed as intraspecific variation. If P. tajacu populations sometimes include “extremely large” individuals— equivalent in size to Roosmalen et al.'s material, as Gongora et al. (2011) suggest—we have not seen any evidence of it.

Analyses of mtDNA control region sequences (Gongora et al., 2006, 2011) suggest the existence of two moderately well-supported phylogroups of Pecari tajacu, consisting of predominantly North American and Central American sequences on the one hand and of South American sequences on the other. Unfortunately, the relevance of this analytic result for assessing the plethora of currently recognized subspecies (Grubb, 2005) is unclear, because no sequence data are available from the type localities of several key nominal taxa, including the nominotypical form (restricted by convention to Pernambuco, Brazil; Cabrera, 1961; Hershkovitz, 1963). At the moment, no trinomial classification of P. tajacu seems justified by the analytic results in hand, although Amazonian material is often refered to P. t. patira Kerr, 1792.

Ethnobiology: The principal term for the collared peccary is şhëkten, a monomorphemic term that is not found in any other Panoan language. The collared peccary has two archaic synonyms: unkin and matoşh, both of which are monomorphemic terms found in other Mayoruna languages, but not in other Panoan languages.¹⁴ In the language used in the Matses' komok ceremony, the collared peccary is called pani tuku, a term that is not synchronically analyzable.

Two varieties of collared peccaries are recognized: şhëkten çhëşhe (“black” or “dark-colored” collared peccary) and şhëkten uşhu (“white” or “light-colored” collared peccary). The dark variety is said to be larger. The light colored variety does not enter holes quickly when chased by dogs and its hide dries more quickly than that of the darker variety.

The collared peccary is a primary game animal for the Matses, who encounter them in various ways. Hunters hear collared peccaries crunching up palm nuts and then approach quietly and shoot them with a shotgun (formerly with arrows). Hunters also visit mineral licks during the day to see if peccaries or other game animals are there. When one collared peccary is killed, the rest of the herd will run off, but if the hunter remains hidden and quiet, the rest of the herd often returns (even if a shotgun was used), and the hunter can kill a second peccary. The herd may even return a third and fourth time.

Hunters also track collared peccaries after finding their spoor. Collared peccaries often sweep away leaf litter, root, defecate, rub their scent gland on saplings, and leave tracks right on Matses paths. Hunters follow spoor until they can hear the animals, or until their hunting dogs pick up the scent.

One hunting method that, as far as we know, is employed only by the Matses, is to chase down a collared peccary with dogs until it enters a hole in a headwater gully or a hollow log, and then to strangle it with a noose on a stick. When Matses hunting dogs find a herd of collared peccaries, they chase them as the hunter follows, encouraging the dogs. If the dogs follow a peccary closely enough, the peccary (usually one, but occasionally two or three) may seek refuge in a hole in a stream headwater gully. These holes are cavities in the bank that have been formed over time by erosion; often the roof of such a hole will be close to the surface of the ground overhead. Alternatively, the peccary may enter a hollow log. When the hunter catches up to the dogs and the cornered peccary, he blocks the entrance to the hole or log with any dry or rotten woody debris that he can find nearby. Then he prepares a noose from the hard vinelike stems of an epiphyte called ayaşh (Heteropsis spp. [Araceae]) and attaches it to the end of a stick about 1 m long. Next, the hunter pokes a small hole in the roof of the hole or hollow trunk, introduces the noose, works it around the peccary's neck using the stick, and garrotes the animal.

Slain peccaries—both collared and whitelipped— are prepared for packing home in a particular way (figs. 19, 20). First, the lower jaw is tied to the ankles of the forelegs with an epiphyte stem (the same stem used to make the noose, if the peccary was killed in this manner). Then the ankles of the hind legs are tied to those of the front legs. Next, a tumpline is fashioned from the inner bark of certain trees, one end of which is tied to the upper jaw and the other end to the rump. The lower tusks keep the lashing from slipping, and the upper tusks keep the tumpline from slipping. If the animal was killed far from the village, the carcass is gutted to lighten the load. Before carrying the peccary, the hunter rubs his dogs' noses on the peccary's caudal scent gland so that they will follow peccaries readily in the future.

The carcass is skinned at home by a woman. Other women will come to where the skinner is working and ask for a leg or part of the viscera, and she will give it to them. Different cuts are given or fed to different people, for example, an old woman typically eats the intestines, visiting men the ribs, a young woman is given the leg, etc. If the peccary was killed by dogs, the dogs will be fed some of the meat.

The Matses raise collared peccaries as pets. The young are often left behind when the adult peccaries flee, and these can be chased down and caught. Tame young peccaries are allowed to roam the village, but peccaries become aggressive when they are older and must kept in a pen. Peccaries are not raised in sufficient numbers or in conditions that would allow them to reproduce in captivity, and the Matses do not eat their pets.

Collared peccary hides can be legally sold, but currently the hides are not worth much, so the Matses only occasionally prepare the skins of peccaries they have killed. Some Matses smoke peccary meat for sale to non-Indians at Colonia Angamos and Requena. However, it is mainly those who live close to these markets who do so regularly, because the money earned by selling smoked meat barely covers the cost of the gasoline needed for canoe travel from more distant Matses villages.

Men do not eat the intestines of peccaries, lest they scrape themselves with a rough-barked vine while they are chasing animals during a hunt. Hunters do not eat peccary spleens, lest their spleens hurt while chasing game animals.

Matses natural history: The collared peccary has a white stripe around its neck. It has large ears and a flat-tipped nose. It has two hooves on each foot. It has a tiny tail. Even newborns have a caudal scent gland, which emits a very strong smell that is different from the scent of whitelipped peccaries. Its feces are seed-shaped pellets.

The collared peccary is found in all habitats, including floodplain and upland forest, and primary and secondary forest. They are especially commonly found along small and medium-sized streams. They seem to be more abundant in areas that are near abandoned Matses villages, where there are large stretches of secondary forest growing in abandoned swiddens. They also come to active Matses swiddens to eat manioc tubers and cush-cush yams.

Collared peccaries are diurnal. They travel far looking for fruits to eat. They chew loudly on palm nuts. They follow streams, rooting for earthworms in the rich soil of the stream floodplain and digging in the streambed for aquatic snails. They leave the water turbid where they have foraged in the streambed, and in areas where they have rooted the leaf litter is swept away. They also root on hilltops, leaving the ground swept clear.

Collared peccaries drink muddy water, eat mud, and bathe at mineral licks. They also root for earthworms at the edge of mineral licks and palm swamps. They come to swiddens where they root in the ground with their noses to expose manioc tubers.

Collared peccaries take mud baths in small muddy depressions in the ground. A depression used for this purpose is often created when a tree is blown over and its roots are uplifted. Peccaries return again and again to bathe in the same mud holes, and are often caked in mud. As they travel through the forest they leave their scent on saplings by rubbing their scent glands on them.

Collared peccaries sleep on the ground wherever they are when it becomes night. They sleep on hilltops, stream valleys, and even on hillsides if the slope is not too steep. They sweep a small patch of ground (but do not dig a depression, as they do to give birth) and sleep on it. They sleep near each other, but not touching, with a space of about 1 m separating one from another.

Collared peccaries live in herds of 5 to about 15 individuals. Occasionally one finds a solitary peccary, or a pair. Collared peccaries are fattest at the end of the rainy season (May). The female gives birth to a single young during the rainy season after digging a depression in level ground. It suckles its young right in the same place where it gave birth. During the first day, the female leaves her young to eat fruits and then comes back to suckle it again. By the second day the newborn peccary begins to travel, very slowly, with the herd. During the first day the rest of the herd stays around the area where the female gave birth, and after that they walk slowly so the newborn can keep up (the female does not separate herself from the herd to give birth).

Collared peccaries are eaten by jaguars, pumas, and (less frequently) by anacondas.

Adults make groanlike grunts. They clack their teeth when they become aggressive. The young also grunt, saying “wek wek wek.”

Peccaries crunch the hard nuts of pinchuk palms (Astrocaryum murumuru, A. chambira, and A. jauari [Arecaceae]) to eat the endosperm. They also eat the endosperm of şhuinte mapi (Attalea tessmanii [Arecaceae]) nuts and the mesocarp and endosperm of swamp palm (Mauritia flexuosa [Arecaceae]) fruits and isan (Oenocarpus bataua [Arecaceae]) fruits. Among the dicot tree fruits they eat are kuëte ise (unidentified), poshton tonte (?Macoubea guianensis [Apocynaceae]), tonnad (a general term for trees in the family Myristicaceae), kuëte mëdiad (an unidentified tree with starchy fruits), and tote (Eschweilera spp. and Lecythis spp. [Lecythidaceae]). Collared peccaries also eat the new unrolled leaves of wild banana plants. They gnaw on the pith of fallen budëd palms (Attalea butyracea [Arecaceae]).

They also eat invertebrates, including aquatic snails, clams, crabs, freshwater shrimp, and earthworms. They occasionally find and eat rotten meat.

Remarks: Matses observations about collared peccaries are richly detailed and suggest long and intimate familiarity with this primary game species. All the salient facts about collared peccary natural history documented in the literature are reported by the Matses, including diurnality, small herd size, use of wallows and mineral licks, scent-marking, feline predators, and a mostly frugivorous/granivorous diet supplemented by invertebrates and browse (Kiltie, 1981, 1982; Kiltie and Terborgh, 1983; Byers, 1985; Bodmer, 1989; Tobler et al., 2009; Blake et al., 2012). Matses observations confirm the fondness of this species for the very hard, golf-ball-size nuts of Astrocaryum spp., the coconutlike endosperm of which is an important trophic resource otherwise accessible only to capuchin monkeys (Terborgh, 1983; Voss and Fleck, 2011), white-lipped peccaries (Kiltie, 1982; see below), and rodents (Emmons, 1997; Voss and Fleck, in prep.). Interestingly, the Matses claim that collared peccaries consume the dense, ivory-hard endosperm of Mauritia flexuosa seeds, which Kiltie (1982) believed to be eaten only by whitelipped peccaries. Other noteworthy dietary items are aquatic mollusks and crustaceans, both seemingly improbable food resources, but snail opercula were reported from peccary stomachs by Kiltie (1981). Many other behavioral details (e.g., of nocturnal bivouacking) are not described in the literature we consulted.

Voucher material (total = 5): Boca Rio Yaquerana (FMNH 88795–88798), Nuevo San Juan (MUSM 11184).

Other interfluvial records: Choncó (Amanzo, 2006), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006), Wiswincho (Escobedo-Torres, 2015).

Identification: Specimens of Tayassu pecari collected in the Yavarí-Ucayali interfluve agree in all qualitative respects with near-topotypical material described by Husson (1978), and measurements of our material (table 18) overlap broadly with Husson's (1978: tables 56, 57). As usual, the current subspecific classification (Grubb, 2005) is difficult to reconcile with the results of analyzing mtDNA sequence data (Ruiz- García et al., 2015).

Ethnobiology: The Matses term for the white-lipped peccary is şhëktename, analyzable as meaning “large collared peccary.” As with the term şhëkten, the term şhëktename is not found in other languages. The archaic term for the white-lipped peccary, chede, is monomorphemic and found in some other Mayoruna languages. In the language used in the Matses' komok ceremony, the white-lipped peccary is called pashankid, a term whose meaning seems to contain the nominalizing suffix -kid, but the meaning of pashan is not clear.

As with the collared peccary, two varieties are recognized, şhëktename çhëşhe (“black” or “dark-colored” white-lipped peccary) and şhëktename uşhu (“white” or “light-colored” white-lipped peccary). The dark variety is said to be larger and has a very white jaw, while the lighter variety is said to have a grayish, less contrastingly colored jaw. An additional name is panchu, which uniquely designates the leader of a white-lipped peccary herd.¹⁵

White-lipped peccaries are a primary game species for the Matses. The Matses find them while hunting in the forest by smell, or when dogs pick up their scent, or when they are heard grunting or crunching palm nuts. Hunters also find their tracks, or see muddied water flowing downstream from where a herd has foraged in the streambed. White-lipped peccaries are sometimes encountered by canoe travellers, who find herds crossing rivers, see their tracks on the bank where they have crossed, or smell or hear them eating in the forest near the banks. Unless he encounters white-lipped peccaries far from the village, a hunter is expected to refrain from killing them and to return to the village to recruit other men to come and hunt them collectively. If the herd is found at the end of the day, the hunt will begin at dawn the following day.

Matses hunters usually kill white-lipped peccaries with shotguns or arrows. Since there are seldom enough shotguns to arm all the men in a village, bows and arrows are still used in these collective hunts. Also, because there is often a shortage of ammunition, spears are often made on the spot from the trunks of sinnad palms (Bactris spp. [Arecaceae]), oninan siante (Iriartella stenocarpa [Arecaceae]), or from the petioles of the stemless budëd uşhu (Attalea microcarpa [Arecaceae]). If it is a very large herd, the peccaries may not run off when hunters start to kill them, defending themselves by trying to bite the hunters. In such cases the Matses can easily kill many individuals, although hunters may have to climb trees if they are attacked. If the peccaries run off, as is most frequently the case, the hunters will chase after them. Because white-lipped peccaries do not run quickly and can be headed off, they are easily killed, and one can even get right next to them while running and kill them with a spear or club. When the leader of the herd is killed, the peccaries run aimlessly, often circling back toward the hunters. When clubbed on the head, right where its ears are, the peccary dies immediately.

Dogs are also used to hunt white-lipped peccaries. Dogs can help tire out the peccaries, giving the hunters a chance to catch up, if the herd had a good head start. Additionally, a white-lipped peccary that is chased by a dog may stop, turn to face the dog, and try to bite it. If the hunter can catch up in time, he kills the peccary. However, dogs are often bitten during such hunts.

White-lipped peccaries are considered potentially dangerous, and any children present during a hunt are told to climb trees in case the peccaries become fierce. Baby whitelipped peccaries are often captured during hunts and kept as pets, although captive animals become fierce as adults and must be kept in a pen. White-lipped peccary hides could formerly be sold legally, so the Matses used to prepare them for sale, but only collared peccary hides are purchased now. Some hunters smoke white-lipped peccary meat for sale at nearby non-Indian towns.

Men do not eat the heart, lest they lose their endurance while running after game.

Matses natural history: White-lipped peccaries are larger than collared peccaries. Their lower cheeks and jaws are white. Their feet have two hooves. They emit a strong, foul smell, different from that of collared peccaries.

White-lipped peccaries walk and sleep in all types of habitats, including upland and floodplain forest, and palm swamps.

White-lipped peccaries are diurnal. They sleep at night on the ground and set out at dawn. They travel very far, swimming across rivers, and come back after a long time. When the Matses kill some of them, the rest of the herd travels very far away and comes back only after a very long time. They travel slowly when there are many recently born young. They go around looking for fallen tree fruits and fallen palm nuts that they crunch loudly. They follow streams rooting in the floodplain earth for earthworms and digging into the streambed looking for mollusks and crustaceans. They make the water turbid where they dig into streambeds. They leave a wide path where they travel and leave large cleared areas where they have rooted. They raise the hair on their back and clack their teeth facing upward when they become aggressive.

White-lipped peccaries frequent mineral licks, where they drink the muddy water, eat mud, and bathe. They root beside mineral licks for earthworms.

White-lipped peccaries live in large herds (of up to 200 individuals) or in smaller herds (of 20 or 30 individuals). A large old male leads them.

White-lipped peccaries are fattest at the end of the rainy season (May). The female gives birth to a single young during the rainy season where the herd stops to sleep for the night. The first day the young does not walk, and the mother suckles it while the rest of the herd forages in the vicinity, without leaving the new mother. The next day, while its umbilical cord is still hanging, she takes it traveling around with the rest of the herd, moving slowly. The rest of the herd also travels slowly to let the newborn(s) keep up. The female eats and lies down frequently to suckle her young.

Jaguars and pumas eat white-lipped peccaries.¹⁶

White-lipped peccaries grunt, snort, scream, and clack their teeth. The young grunt saying “wek wek.” They travel through the forest making a lot of noise.

While-lipped peccaries eat aquatic snails, clams, crabs, freshwater shrimp, earthworms, snakes (including pitvipers), and rotten meat. They eat many types of dicot tree fruits. They are especially fond of pinchuk (Astrocaryum spp. [Arecaceae]) nuts. They eat the mesocarp and endosperm of swamp palm (Mauritia flexuosa [Arecaceae]) fruits. They eat the new leaves of wild banana plants.

Remarks: Matses interviews about whitelipped peccaries are much less informative than those about collared peccaries, presumably because this species is encountered at infrequent intervals. Although broadly consistent with the scattered literature (e.g., Kiltie and Terborgh, 1983; Fragoso, 1998, 1999; Tobler et al., 2009) in most respects, Matses observations do not suggest that this species routinely eats any item not also eaten by collared peccaries (contra Kiltie, 1982; see above), with the possible exception of venomous snakes.

Voucher material (total = 6): Boca Rio Yaquerana (FMNH 88806–88808), Nuevo San Juan (MUSM 11185), Orosa (AMNH 74117), Santa Cecelia (FMNH 86900).

Other interfluvial records: Choncó (Amanzo, 2006), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006).

Identification: Several of our voucher specimens of red brockets (FMNH 86900, 88807, 88808) are immature individuals that retain their milk premolars (dP2–dP4), but only one of these (FMNH 88807) has the immature markings of a fawn. Of our three adult specimens, two are antlerless females (AMNH 74117, MUSM 11185) and the third (FMNH 88806) is a very young male with tiny antlers that measure less than 20 mm (not including the bony pedicel). In pelage characters (the FMNH specimens are accompanied by well-preserved skins) and adult cranial measurements (table 19), this material agrees closely with Husson's (1978) descriptions and measurements of almost-topotypical (Surinamese) material of Mazama americana.

Although the identification of our material does not seem problematic based on morphological criteria, karyological and molecular studies of Mazama americana suggest that specimens sharing the red brocket phenotype are genetically heterogeneous. In fact, phylogenetic analyses of cytochrome-b sequence data do not support the monophyly of Mazama americana, which was recovered as two weakly supported haplogroups in an unresolved polytomy with Odocoileus hemionus, O. virginianus, and two other species of Mazama by Duarte et al. (2008). Subsequently, Abril et al. (2010) recovered two haplogroups of M. americana with stronger support by sequencing the mitochondrial control region and by omitting sequences from some of the other odocoileine taxa included in Duarte et al.'s (2008) analysis. Abril et al. (2010) referred to these haplogroups as morphologically cryptic “species,” but this conclusion is not supported by their karyological data.¹⁷ In effect, the taxonomic interpretation of the cytogenetic and molecular data at hand is not straightforward, nor is it known which (if either) of the two haplogroups recovered in published analyses of mtDNA sequence data occurs in French Guiana (the type locality of M. americana).

Grubb (2005) recognized numerous allegedly valid subspecies of Mazama americana for which (as usual) he cited no supporting revisionary study. According to Cabrera (1961), the western Amazonian form is M. a. zamora Allen, 1915, the type locality of which is in the southeastern Andean foothills of Ecuador. In light of molecular studies cited above, a comprehensive taxonomic revision of the many nominal taxa currently treated as synonyms or subspecies of M. americana is clearly needed.

Ethnobiology: The red brocket is called senad piu (“reddish deer”). The Matses recognize and name two varieties of this species, senad maçhëşh (“black-headed deer”) and senad bëdimpi (“little spotted deer”). The latter would seem to designate young individuals, yet the Matses insist that this variety does not grow any larger, does not lose its spots, and runs as fast as an adult. Additionally, a third variety is simply called senad piu.¹⁸ The spotted variety is rarely encountered and is only found in upland forest, the senad piu variety is the most commonly encountered, and the black-headed variety is the largest.

The red brocket is a primary game animal for the Matses. Deer fat sticks to the mouth when one eats it, so the Matses are not fond of deer fat. However, lean meat is appreciated, and a red brocket carcass has a lot of meat on it. The Matses sometimes keep fawns as pets (fig. 21), but tame deer wander away when they become adults.

Deer are difficult to kill because they almost always run off before one can shoot them. The Matses often kill deer when they happen upon one that is sleeping in the daytime and can be approached before it wakes up. The Matses often visit mineral licks during the day to look for deer and other game species that visit mineral licks.

Now that the Matses have flashlights, they hunt at night by walking down forest paths. The intention of night hunting is primarily to kill pacas, which are common in secondary forest near villages, especially when peach-palm (Bactris gasipaes [Arecaceae]) fruits are ripe (from January to March). Hunters also occasionally encounter red brockets that come to secondary forest near villages at night, and they occasionally kill deer while waiting at night for game at a mineral lick.

Slain deer are prepared for packing home in a certain way (fig. 22). The front legs are lashed to the neck with epiphyte stems, and the hind legs are lashed to the front legs. Then a strip of the inner bark of certain types of trees is tied to the neck and rump to make a tumpline. If the deer is large and is killed far from the village, it is gutted before being tied up for carrying. A hunter might also skin and butcher the deer and carry it back in a palm leaf basket that is woven on the spot.

Young men and women do not eat the head, lest they stab themselves in the thigh with a sharp (antlerlike) stick when chasing after an animal while hunting (women often help men chase down animals, especially while guiding hunting dogs). Old people, however, may eat the head.

When a hunter kills, eats, or sees a deer, the deer's spirit may make one of his children ill. Occasionally the deer's spirit makes a child ill even if there has been no contact with a deer. The symptoms for contagion by deer spirits are the same as those by tapir spirits: high fever and rolling of eyes into the back of the head. When a child exhibits these symptoms after the father has had contact with a deer, a medicine man will collect certain medicinal plants (“deer medicine”) and bathe the child with an infusion of the leaves. Interestingly, “tapir medicine” is also used to treat ailments caused by deer spirits, suggesting that the Matses perceive an affinity between these two ungulates.

Matses natural history: Red brockets are reddish, the color of some dogs. Males have antlers, but females do not. The antlers are very hard. Red brockets have a white tail, large ears, and large nostrils. Their hindquarters have a lot of meat. They have thin lower legs and ankles, and two parallel hooves, similar to those of a collared peccary, but smaller. The young are spotted.

Red brockets walk around in all habitats, including upland and floodplain forest, primary and secondary forest, and along streams and rivers. They come to the edges of Matses swiddens and sometimes enter Matses swiddens to eat manioc leaves.

Red brockets do not make nests. They clear a small patch of ground and lie down to sleep, curled up on their sides like dogs. In the daytime they often sleep in forest with an open canopy. They do not sleep in the same place twice.

Red brockets are mostly nocturnal. They sleep during the day. Sometimes they walk around during the day, but not when it is dry. They walk around in the rain by day or night. They travel far, looking for leaves and fruits to eat. The places where they eat fruits are swept clear of leaf litter. After eating their fill, they lie down to rest, often in a sunny spot in a treefall, in streamside forest on high ground where a stream bends, at the foot of a hill, or on a hilltop.

Red brockets visit several different mineral licks by day or at night, and they always return to the same mineral licks. They make the water in the mineral lick turbid with their feet and then slurp up the muddied water. Unlike tapirs and other animals that use mineral licks, red brocket deer do not eat the mud if the mineral lick is not waterlogged. Red brockets do not usually make paths, but one can find deer paths that lead to a mineral lick.

Red brockets are generally solitary. Very rarely two adults may be encountered together. The male does not live with the female. They mate when they encounter each other. The deer gives birth to a single young under the shelter of a stemless palm with large simple leaves, or at the edge of a blowdown. It suckles its newborn at the same place where it gave birth to it, lying on its side, like a dog does. It goes to drink muddied water at a mineral lick and then returns to suckle its young. Then it goes to eat ripe fruits and comes back to suckle its young again. The young deer stands up after two days and starts to walk around and forage with the mother. Once it is large and strong, it leaves its mother.

Red brockets are a favorite prey of jaguars and pumas. Large anacondas and black caimans occasionally capture deer.

Red brockets call out saying “mia” (a sort of high-pitched whine). Females call out saying “ooo” (a sort of howl) when they are in heat. They stamp their feet when they see a person at a distance (if they are not wary).

Red brockets eat the fruits of many types of dicot trees—particularly sweet fruits—including those of şhannëd (?Brosimum [Moraceae]), figs (Ficus spp. [Moraceae]), şhëşhun (Spondias mombin [Anacardiaceae]), taëpa (unidentified), and piuşh bëchi (Helicostylis tomentosa [Moraceae]). They don't just eat the pulp of the fruits, but also swallow the seeds. Their favorite dicot tree fruit is that of echo (Jacaratia sp. [Caricaceae]), which is like a wild papaya. They also eat papayas that have fallen to the ground in Matses swiddens. They also eat the seeds of some palms, including those of bottle palms (Iriartea deltoidea [Arecaceae]) and stilt palms (Socratea exorrhiza [Arecaceae]). Red brockets also eat the leaves of many dicot trees and understory plants, including cecropia trees (Cecropia sp. [Moraceae]), which grow in secondary forest. They eat manioc leaves and papaya leaves when they come to Matses swiddens. Pet deer are fed papaya, echo fruits, and the otherwise discarded pulp of strained plantain beverages.

Remarks: Matses interviews about red brockets include most of the well-established natural history facts about this widespread species, including its use of floodplain habitats and secondary vegetation (avoided by gray brockets); solitary habits; feline predators; mixed diet of fruit, seeds, and browse; and propensity for visiting mineral licks (e.g., Bodmer, 1989, 1991; Gayot et al., 2004; Tobler et al., 2009; Blake et al., 2013). Many other behavioral details, however, are not reported in the literature we consulted, including information about daily movements, geophagy, reptilian predators, and sexual vocalizations. Although the Matses say red brockets are mostly nocturnal, this is perhaps a result of local hunting pressure; at unhunted western Amazonian sites red brockets are often active by day (Gómez et al., 2005; Blake et al., 2013).

Voucher material (total = 3): Nuevo San Juan (MUSM 11186, 13148), Santa Cecilia (FMNH 86898).

Other interfluvial records: Anguila (Escobedo-Torres, 2015), Quebrada Pobreza (Escobedo-Torres, 2015), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Tapiche (Jorge and Velazco, 2006), Wiswincho (Escobedo-Torres, 2015).

Identification: The voucher material we examined corresponds closely to the qualitative description of Mazama nemorivaga provided by Rossi et al. (2010), who recognized the Amazonian brown brocket as a distinct species from M. gouazoubira, with which it was formerly synonymized (e.g., by Grubb, 2005). Among other diagnostic craniodental traits of M. nemorivaga, the premaxillary does not contact the nasal, from which it is widely separated on each side by a dorsolateral process of the maxillary. Unusually, one of our vouchers (FMNH 86898) retains welldeveloped canine teeth.

Measurement data that we obtained from adult specimens collected in the Yavarí-Ucayali interfluve (table 19) broadly overlap those previously reported from Amazonian brown brockets by Husson (1978: table 61) and Bisbal (1991: table II), although an adult male from Santa Cecilia (FMNH 86898) is somewhat larger than the topotypical specimens measured by Voss et al. (2001: table 16).

The recognition of Mazama nemorivaga as a species distinct from M. gouazoubira by recent authors is consistent with molecular evidence that these are not sister taxa (Duarte et al., 2008), but the same analyses also suggest that Mazama is not monophyletic. Since the type species of Mazama is M. americana, it is plausible that M. nemorivaga will eventually be referred to another genus if Duarte et al.'s results are confirmed by additional research. However, few of the relevant nodes in Duarte et al.'s phylogeny are strongly supported, so it is not clear what nomenclatural solution is appropriate. Additionally, few morphological characters apart from size and pelage color appear to distinguish M. nemorivaga from M. americana, so describing new genera based only on molecular results is likely to cause problems for generic assignments of fossil odocoileines.

Ethnobiology: The gray brocket is called senad tanun (“gray deer”). A few Matses recognize a small and large variety of this species, but most do not.

The gray brocket is a game animal of secondary importance (due to its small size). It is hunted in the same manner as the red brocket.

All the Matses beliefs associated with deer are the same for red and gray brockets.

Matses natural history: The gray brocket is gray, has light-colored undersides, and is smaller than the red brocket. The male's antlers are shorter than those of the red brocket. The young are spotted and thin.

The gray brocket prefers upland forest, away from rivers and large streams. It does not come near Matses villages, nor does it enter secondary forest growing in abandoned swiddens. It is often found on hilltops where the understory is dominated by thatch palms (Lepidocaryum tenue [Arecaceae]). The gray brocket is encountered much less frequently than the red brocket.

(The rest of the natural history information for this species is the same as for the red brocket, except for the foods eaten in abandoned and active swiddens.)

Remarks: The only ecological distinction that the Matses report between the gray and red brockets, that Mazama nemorivaga is an upland species that avoids river floodplains, is corroborated by field research (Bodmer, 1991; Tobler et al., 2009). However, the Matses' failure to distinguish these sympatric cervids in other ecobehavioral respects is hard to reconcile with published evidence that gray brockets are primarily diurnal and do not visit mineral licks, whereas red brockets are often active at night and commonly visit mineral licks (Tobler et al., 2009; Blake et al., 2012, 2013).

Voucher material: None.

Other interfluvial records: Actiamë (Amanzo, 2006), Jenaro Herrera (Pavlinov, 1994), Nuevo San Juan (this report), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Wiswincho (Escobedo-Torres, 2015).

Identification: Identifications of pink Amazon river dolphins are not problematic.

Ethnobiology: The pink Amazon river dolphin is named chishkan piu (“reddish dolphin”) by the Matses.

Dolphins are of no economic importance to the Matses, but they are sometimes a nuisance. The Matses now purchase nylon gill nets, which they set up in rivers, streams, and lakes near their homes. Dolphins, especially Amazon pink river dolphins often steal fish caught in the gill nets, tearing holes in the nets when they do so.

The Matses traditionally believed that pink Amazon river dolphins are spirits. Pink Amazon river dolphins (but not gray dolphins) tend to follow the Matses when they travel in canoes, which make the Matses nervous.

Matses natural history: The pink Amazon river dolphin is reddish. It has a very long snout with pointy teeth and a ball-shaped lump on its forehead. Its tail is similar to that of a fish, but oriented laterally. The dorsal fin is like the keel of a canoe. They have nipples.

Pink Amazon river dolphins are found in rivers and large streams, and in ox-bow lakes and flooded forest during the high-water season. They are especially fond of deep river bends. They congregate at the mouths of streams.

Pink Amazon river dolphins swim following the river looking for fish. The may stay a long time in deep river bends and at the mouths of streams, catching fish. They chase fish down and eat them underwater. They especially chase fish close to the bank and near river beaches. When Matses travel by canoe, they follow the canoe and exhale loudly through their blowholes to scare Matses.

Pink Amazon river dolphins are most frequently seen alone, but it is also common to see a pair or a group of three. Larger numbers can be seen where large streams feed into the Rio Yaquerana. They copulate on river beaches.

They eat all types of fish, especially large catfishes.

Remarks: Matses observations about Inia geoffrensis are limited in scope but agree in several details with the scientific literature on this species, notably about their their numerical abundance at stream confluences, seasonal use of flooded forests, predominantly piscivorous diet, and uncanny habit of following canoes and frightening people (Martin and da Silva, 2004; Martin et al., 2004; Gómez-Salazar et al., 2012; Paschoal et al., 2013). We have not seen any explicit description of the copulatory behavior of this species in the wild, but the common Amazonian superstition that pink dolphins have sexual relationships with people may have something to do with mating on or near river beaches, as alleged by the Matses.

Voucher material: None.

Other interfluvial records: Nuevo San Juan (this report), Río Yavarí (Salovaara et al., 2003), Río Yavarí-Mirím (Salovaara et al., 2003), San Pedro (Valqui, 1999), Wiswincho (Escobedo- Torres, 2015).

Identification: Identifications of gray dolphins are not problematic.

Ethnobiology: The gray dolphin is called chishkan uşhu (“white” or “light-colored dolphin”). Some Matses recognize a darker variety that they call chishkan umu (“blue dolphin”), although most speakers consider the color variation to be continuous.

Dolphins are of no economic importance to the Matses.

The Matses are not afraid of gray dolphins (as they are of pink Amazon river dolphins).

Matses natural history: Gray dolphins are light gray to dark gray.

Gray dolphins are found in rivers and large streams, but not in smaller streams or flooded forest. They are also found in ox-bow lakes when they are flooded.

Gray dolphins travel through the rivers and streams chasing down fish and eating them. Unlike pink Amazon river dolphins, they do not follow canoes, and they can jump high out of the water.

Gray dolphins are solitary and often travel in pairs. They feed in groups of up to 5.

Gray dolphins exhale audibly when they surface, but not as loudly as pink Amazon river dolphins do.

Gray dolphins eat fish.

Remarks: The Matses do not have much to say about gray dolphins, but they accurately note several behavioral differences from pink Amazon river dolphins, notably their fondness for acrobatic aerial displays (Inia seldom jumps clear of the water) and their avoidance of flooded forest (an important seasonal habitat of Inia; da Silva and Best, 1996; Emmons, 1997; Martin and da Silva, 2004). The benign attitude of the Matses toward gray dolphins is similar to those of other native Amazonians, who do not fear this species as they do pink Amazon river dolphins (McGuire, 2010; Paschoal et al., 2013).

Voucher material: Río Tapiche (AMNH 98691).

Other interfluvial records: Jenaro Herrera (Tovar, 2011), Reserva Comunal Tamshiyacu- Tahuayo (Bodmer, 1994), Rio Orosa (Mármol, 1995), Río Tapiche (Reeves et al., 1996), Río Yavarí (Mármol, 1995).

Identification: Only a single species of manatee is known to occur in the upper Amazon Basin. Our voucher consists of the skull of a mature adult of unknown sex. Selected cranial measurements of AMNH 98691 are all within the range of variation for Trichechus inunguis tabulated by Domning and Hayek (1986): condylobasal length, 383.1 mm; breadth across postorbital processes of frontals, 130.1 mm; least postorbital breadth, 54.1 mm; zygomatic breadth, 214.1 mm.

Remarks: The holotype skull of an allegedly new species of “dwarf” manatee recently described by Roosmalen (2015: figs. 20, 21) from the Rio Aripuanã basin of Amazonas, Brazil, is an obviously immature individual with open basicranial and occipital sutures.