Cyphomyrmex longiscapus

Nesting in abundance on stream embankments in the wet forests of Panama, the fungus-growing ant Cyphomyrmex longiscapus sensu lato has become a model organism for the study of behavior, ecology, mating frequency, cultivar specificity, pathogenesis, and social parasitism in the attine agricultural symbiosis.

Identification
Schultz et al. (2002) - Originally described by Weber (1940), C. longiscapus is unique within the genus. Of its distinctness, Kempf (1966, p. 167) cited, in addition to the uniquely elongate scapes (which surpass the occipital corners), the characters of the weakly expanded frontal lobes, elongate mandibles, and neck-like collar on the occiput. Unfortunately, all of these character states are shared with the newly described Cyphomyrmex muelleri, and thus both species run to C. longiscapus in the keys of Weber (1940), Kempf (1966), and Snelling and Longino (1992). The information provided here supplements information provided by these authors and is sufficient for distinguishing C. longiscapus from Cyphomyrmex muelleri.

Workers Possessing 11 antennal segments and palpal formula 4, 2 (features plesiomorphic for, and widespread in, the Attini). Color ranging from yellow to testaceous to fuscous brown. Head and alitrunk uniformly foveate, each fovea usually surrounded by a circlet of whitish “bloom” that resembles the attine actinomycete symbiont (Currie et al., 1999a), the extent of this bloom highly variable across individuals: It may be entirely absent; it may be present as small, separate circlets, creating a discontinuous pattern; or it may be present as a continuous integumental blanket formed of larger, overlapping circlets. Pilosity inconspicuous, fine, thin, silvery, and decumbent. The following characters reliably separate C. longiscapus from its cryptic sister species (described below): Vertexal carinae (i. e., paired carinae on the vertex, running parallel to and on either side of the mid-line) strongly produced (Fig. 1a, VC). Metanotal groove (“mesoepinotal impression” of Kempf) deep, clearly interrupting the continuity of the alitrunk in lateral view (Fig. 2a, MG). Third intersegmental groove of the thorax, separating the mesopleuron from the metapleuron, complete, extending from the metanotal groove to between the coxae, although obscured in some specimens by the whitish integumental “bloom” (the attine actinomycete symbiont?) (Fig. 2a, IG). Posterior tubercles on the postpetiolar dorsum forming low, rounded tumuli (Fig. 3a, PT); in dorsal view, the postpetiole only slightly emarginate and broadly and shallowly impressed posteriorly (Fig. 3a). Hind femur lacking ventral carinae and a ventral lobe (Fig. 4a, arrow).

The following characters are generally useful for distinguishing C. longiscapus from the new species, but, because a minority of specimens in both species possess intermediate states, they are less reliable than the characters listed above and, therefore, should be used for identification only with caution: Whitish integumental “bloom” (actinomycete symbiont?), when present, rarely occurring within the antennal scrobe. Frontal triangle usually broad, usually forming a finger-shaped area between the frontal lobes that is rounded posteriorly. Posterior mesonotal tubercles usually strongly produced (Fig. 2a, PT), but variable, more weakly produced in some specimens. Propodeal angle usually present, with dorsal and declivous faces usually separated by a distinct shoulder.

Queens Like the workers, possessing 11 antennal segments, palpal formula 4, 2, and the unique C. longiscapus s.l. characters of the elongate antennal scapes and weakly expanded frontal lobes. Color, microsculpture, integumental bloom, and pilosity as in the workers.

C. longiscapus gynes generally differ from those of the new species in the same character states as the workers except, obviously, for those of the alitrunk. The most reliable characters include: Vertexal carinae strongly produced (Fig. 5a, VC). Posterior tubercles on the postpetiole forming low, rounded tumuli (as in worker, Fig. 3a, PT); in dorsal view, the postpetiole only slightly emarginate and broadly and shallowly impressed posteriorly (as in worker, Fig. 3a). Hind femur lacking ventral carinae and lobe (as in worker, Fig. 4a, arrow).

Somewhat less reliable, more continuously varying characters include: Frontal triangle usually broad, usually forming a finger-shaped area between the two lobes that is rounded posteriorly. Whitish integumental “bloom” (actinomycete symbiont?) rarely present within the antennal scrobe.

Males Possessing 13 antennal segments and palpal formula 4, 2. Mandibles with four or five teeth, the basal (fifth) tooth sometimes reduced to a rounded basal angle. As in the workers and gynes, head and alitrunk rather uniformly foveate, the foveae occasionally surrounded by a whitish “bloom”; pilosity as in the other castes. Based on three dissections (one male from each of three Panamanian nests), male genitalia conform closely to the plesiomorphic attine pattern: parameres simple, forming short, rounded, concave lobes; aedeagus simple, forming a broad rounded lobe with minute teeth along the ventral edge; and volsella with digitus simple, long, narrow, and strongly recurved. The only departure from the plesiomorphic attine pattern (in which the volsellar cuspis is absent) is that the cuspis is present as a short, simple, rounded lobe.

In general, males of C. longiscapus are difficult to distinguish from males of the new species, but the most reliable characters include: Postpetiole in dorsal view weakly emarginate posteriorly (Fig. 7a); and the propodeal spines short, the width at the base of the spine exceeding the total spine length (Fig. 6a, PS). In all C. longiscapus males examined, the head and usually the dorsum of the alitrunk are more darkly pigmented (testaceous to fuscous brown) than the rest of the body, which is yellow, a pattern suggesting a day-flying species.

Distribution
Wet forests of Colombia, Panama, and Costa Rica.

Distribution based on Regional Taxon Lists
Neotropical Region: Colombia, Costa Rica, Panama.

Biology
Schultz et al. (2002) - What little knowledge we have of the ecology of Cyphomyrmex longiscapus in Colombia is consistent with the ecology of the far better studied Panamanian Cyphomyrmex longiscapus and Cyphomyrmex muelleri (Mueller and Weislo 1998; UGM, unpublished). Although no ecological data are reported in Weber’s original (1940) description of C. longiscapus, a subset of the paratype labels indicates that the colony was taken at 1020 m elevation (another label indicates 3200 ft); one paratype label specifies “nesting in rain forest”; and Weber’s notes indicate a collection location of 6°40'N, 75°10'W. Weber (1972) supplies a verbal description and a photograph of the Colombian C. longiscapus type series nest, and provides the additional information that the nest was collected in “a deep, moist ravine in the Andes of Colombia” (p. 57), consistent with Mueller and Weislo (1998), who surveyed 203 nests of C. longiscapus s.l. in Panama and documented a nesting preference for steep embankments along permanent streams. Label data accompanying Brown’s 1971 Colombian collection of C. longsicapus indicate that the nest was encountered at 2000 m elevation in rain forest. The only known Costa Rican specimen, a lone worker, was found inside a dead stick on the ground of an old alluvial terrace close to a river (J. Longino, pers. comm.). Colombian C. longiscapus specimens have been identified from the stomach contents of two poison-dart frogs, Phyllobates aurotaenia and Dendrobates histrionicus (Snelling and Longino, 1992). The only known non-Panamanian C. muelleri specimen was taken in wet forest at sardine bait in Ecuador (C.R.F. Brandão, pers. comm.).

Both species are monogynous and perennial (Mueller and Weislo, 1998; Villesen et al., unpubl.), queens are singly mated (Villesen et al., 1999; Adams et al., unpubl.; Villesen, unpubl.), and both cultivate mycelium gardens (Mueller and Weislo, 1998; Mueller et al., 1998). Nests consist of a single garden chamber and are usually constructed in steep embankments under the shelter of overhangs or set back underneath the shelter of a rock or root. Nest entrance architecture is complex and characteristic (Fig. 8). In Panama, C. longiscapus constructs nests of two distinct, but intergraded, morphologies: (1) the hanging, baglike “swallow’s nest” type (Fig. 8a), sometimes suspended from rock faces; and (2) the “auricle” type (Fig. 8b). Both of these nest types have pronounced, vertically elongate auricle-shaped entrances, but, whereas the “swallow’s nest” type is suspended and the garden chamber is surrounded at the sides by thin (approximately 2 to 5 mm thick) walls constructed by the ants from clay, the simpler “auricle” type nest is set into the soil, the lateral walls of the excavated garden chamber are natural, and only the front walls surrounding the auricle are constructed by the ants. The single described non-Panamanian C. longiscapus nest, from Colombia, was of the swallow’s-nest type (Weber, 1972; Mueller and Weislo, 1998). C. muelleri constructs nests of the auricle type only, but significant differences separate the auricle-type nests of the two species: Only C. longiscapus constructs large, flaring nest auricles that are generally longer in the vertical than in the horizontal dimension (Fig. 8 b; see also Fig. 1 in Mueller and Weislo, 1998). C. muelleri, in contrast, constructs “mouthlike” auricles that possess swollen or thickened rather than flared rims and that are generally longer in the horizontal rather than vertical dimension (Fig. 8 c). Neighboring nests of the two species almost invariably maintain these species-specific features, indicating that differences in auricle architecture are not microhabitat-dependent and confirming that the two species have diverged with respect to nest construction behavior.

Mueller and Wislo (1998) report an average colony size of 29.4 workers for a mixed sample of C. longiscapus and C. muelleri. However, when the colony sizes of these nests are recalculated by species, C. muelleri nests are found to contain twice as many workers (average 43.8 ± 27.57 s.d. workers; range 6-109; N=106) than C. longiscapus nests (average 22.7 ± 12.33 s.d. workers; range 4–58; N=67). This difference in observed nest size may be due in part to sampling bias: Because C. muelleri nests possess less conspicuous entrance auricles than do C. longiscapus nests (Figs. 8b and 8c), smaller colonies of C. muelleri may be less frequently noticed and collected in the field relative to smaller colonies of C. longiscapus. However, in a more recent survey of three populations in central Panama where both species occur sympatrically, conducted in June 1998, a special effort was made to locate and collect smaller nests. In this case the average number of workers per nest was 14.6 ± 9.73 s.d. (range 0–40; N=76) for C. longiscapus and 29.4 ± 24.24 s.d. (range 4–117; N=42) for C. muelleri. Thus, when the problem of size-biased sampling error is addressed and when nests of all sizes are sampled, sympatrically occurring C. muelleri nests are found to contain about twice as many workers as C. longiscapus nests.

For the subset of colonies with alates reported in Mueller and Weislo (1998), C. muelleri averaged 12.4 alates per nest (N=30), whereas C. longiscapus averaged only 7.8 alates per nest (N=56), a pattern paralleling the average worker colony-size differences between the two species. At locations where both species occur in mixed aggregations, alates were found in nests of both species during the early dry season (December to February) of 1996, and also during the wet season in July and August of 1997, suggesting temporal overlap of alate production between the two species at these times. However, collections of both species taken at the same sites in June 1998, at the start of the wet season, yielded only a single alate (a male) from 29 nests of C. muelleri and 101 alates of both sexes from 72 nests of C. longiscapus (Villesen et al., unpubl.). In addition, late dry-season collections from April 2001 yielded only six males from 34 nests of C. muelleri and 111 alates of both sexes from 32 nests of C. longiscapus (UGM, unpubl.). Although these data are inadequate for drawing firm conclusions, they suggest a scenario of partial reproductive isolation in which both C. longiscapus and C. muelleri produce alates during the wet season and early dry season (July to February), but in which only C. longiscapus (and not C. muelleri) produces sexuals during the late dry season and early rainy season (March to June). Additional data are needed from more extensive nest surveys conducted throughout the year, as well as from observations of mating flight times in both species. Pigmentation differences between males of the two species, noted above, may indicate time-of-day separation in alate flight times, a phenomenon known to occur between sympatric, closely related species of Atta (Mariconi, 1970; Weber, 1972; TRS and UGM, pers. obs.). Specifically, the lighter pigmentation in males of C. muelleri suggests nocturnal mating flights, whereas the darker pigmentation in males of C. longiscapus suggests diurnal mating flights.

C. longiscapus and C. muelleri are obviously very closely related, and are remarkably similar in terms of ecology, behavior, and morphology. The primary morphological differences separating these species suggest a common evolutionary pattern: Relative to C. longiscapus, C. muelleri appears to be more specialized for cryptic defense. Specifically, the surfaces of the head and alitrunk of C. muelleri are smoother and more rounded than are those of C. longiscapus. Carinae and tubercles are more reduced and the dorsal profile is less interrupted by sutures and grooves (Fig. 2b). This “streamlined” morphology in C. muelleri plausibly reduces the available points of purchase for the mandibles or grasping organs of an attacking, similarly-sized predator (e. g., another arthropod). In contrast to this general trend toward reduced sculpture, but in agreement with the general trend toward more efficient cryptic defense, sculpturing in two features in C. muelleri is increased over that found in C. longiscapus: the posterior postpetiolar tubercles are produced into teeth (Fig. 3b) and the hind femur is equipped with a pair of ventral carinae, forming a ventral groove, and with a ventral lobe (Fig. 4 b). These features, which occur independently in other Cyphomyrmex species (Kempf, 1966), serve to protect vulnerable body parts that are commonly attacked by arthropod predators, particularly other ants; specifically, the postpetiolar tubercles protect the point of articulation between the postpetiole and the gaster; the metafemoral groove receives the tarsus and tibia when the leg is folded in the cryptic-defensive posture; and the metafemoral lobe protects the point of articulation between metatibia and metatarsus.

Morphological features of Cyphomyrmex spp. In general and of C. muelleri in particular suggest adaptations to predation pressure from arthropod-sized predators. Such predation pressure, at least from above-ground (rather than subterranean) hunters, is also suggested by the unusual “auricle” nest-entrance morphology of C. longiscapus and C. muelleri, which may serve as a partial physical or even chemically protected barrier to surface-raiding arthropods, particular predatory ants. Likely examples of such predators include army ants in the subfamily Ecitoninae, which are known to significantly impact Neotropical ant colonies in general (Schneirla, 1971; Rettenmeyer, 1983; Gotwald, 1995; Kaspari, 1996). Published records of army ant predation specifically on fungus-growing ants other than Atta spp. are rare and include no raids on Cyphomyrmex spp. (Cole, 1939; Weber, 1945; Schneirla, 1958, 1971; Fowler, 1977; Mirenda et al., 1980; summarized in LaPolla et al., 2002). That the auricle nest entrance could serve to deter the entry of surface-raiding army (and other) ants is suggested by a single observation in Panama in 1996 in which a Neivamyrmex sp. raiding column, consisting of many thousands of workers, swarmed past the entrances of two C. longiscapus nests. Although many dozens of army ant workers climbed up the outside (i.e., ground-facing) surfaces of the auricles, none ventured onto the auricle rim or onto the frontal (outward) face adjacent to the nest opening (UGM, pers. obs.). If nest-entrance auricle morphology of C. longiscapus (Figs. 8a and 8b) is more efficient at repelling army ants than the auricle morphology of C. muelleri (Fig. 8 c), this could account for stronger selection for the seemingly more effective anti-predatory body morphology in the latter species.

Another known predator of both C. longiscapus and possibly of C. muelleri is the semi-nomadic, socially parasitic, agropredatory ant species Megalomyrmex sp. nov. (Formicidae: Solenopsidini) (Adams et al., 2000b). Based on field and laboratory data, Megalomyrmex sp. nov. colonies aggressively raid C. longiscapus and C. muelleri nests, biting and stinging host-species workers. The raiders eject the resident Cyphomyrmex colony and then occupy and consume the fungus garden over a period of weeks or months, depending on garden size. Unfortunately, this species is known from only a few collections (5 colonies of Megalomyrmex sp. nov. from 344 C. longiscapus/C. muelleri nests collected during 1999 and 2001) (Adams et al., 2000b), and so, without more research, it is impossible to accurately assess the (possibly differential?) predation pressure exerted by this species on C. longiscapus and C. muelleri.

Perhaps the most remarkable difference between C. longiscapus and C. muelleri is that, even though these two species are quite similar biologically and even though they occur sympatrically in the same microhabitats, they consistently employ two very different, distantly related fungal cultivar species. Each of these fungal cultivars is also employed by other, distantly related attine ants that occupy different microhabitats and that are otherwise quite dissimilar biologically (Mueller et al., 1998; Green et al. 2002). Specifically, C. longiscapus shares a narrow group of cultivars of the “Clade 1” type (Mueller et al., 1998) with the sympatric fungus-growing ant Apterostigma auriculatum; molecular data indicate that in one case a cultivar clone has been transferred recently between nests of these two ant species (Mueller et al., 1998). Similarily, C. muelleri shares a narrow group of cultivars of the “Clade 2” type with the sympatric Cyphomyrmex costatus, and multiple cultivar exchanges have occurred between these two ant species (Green et al. 2002). Both A. auriculatum and C. costatus are commonly encountered under logs and rocks on the rain forest floor (UGM, pers. obs.), a very different microhabitat from the embankments preferred by C. longiscapus and C. muelleri. Based on these microhabitat differences, we might expect cultivar exchanges between C. longiscapus and A. auriculatum, or between C. muelleri and C. costatus, to occur at very low frequencies relative to exchanges between C. longiscapus and C. muelleri, which often occur in mixed aggregations. Yet such across-microhabitat exchanges between the more distantly related attine species are well documented (Mueller et al., 1998; Green et al. 2002). In contrast, in over 400 nest collections in which cultivar species could be identified, C. longiscapus was invariably associated with its own Clade 1 type cultivar, and C. muelleri was invariably associated with its own Clade 2 type cultivar. Thus, cultivar exchanges apparently do not occur between nests of the closely related and physically more proximate C. longiscapus and C. muelleri.

Nesting Biology
Schultz et al. (2002) - Nest Architecture (diagnosis): Nest entrance architecture differs enough between C. longiscapus and Cyphomyrmex muelleri to constitute a generally useful character for distinguishing between nests of the two species in the field. C. longiscapus constructs nest entrances of two kinds: the swallow’s-nest type, reported from Colombia and Panama (Fig. 8a; Weber, 1972; Mueller and Wcislo, 1998), and the auricle type, reported from Panama (Fig. 8b). Auricle dimensions are summarized in Table 3. In C. longiscapus, nest entrance auricles are usually higher than wide (i. e., longer in the vertical than in the horizontal dimension), and auricles are strikingly flared, i. e., they are much broader across the rim than across the base.

Nomenclature

 *  longiscapus. Cyphomyrmex longiscapus Weber, 1940a: 410 (diagnosis in key) (w.) COLOMBIA. Schultz, Solomon, Mueller, et al. 2002: 333 (q.m.). See also: Kempf, 1966: 165.

Worker
Kempf (1966) - The present species, still known only from the type series, is highly distinctive both in general aspect (elongate head mandibles and scape) and in several features not shared by any other known species of the group: the long mandibles, the scarcely expanded frontal lobes, the neck-like collar on occiput, and above all the configuration of the antennal scrobe which, although lacking the posterolateral limit of a postocular carina, is nevertheless similar to that of the strigatus-group. Fore the preocular carina does not curve mesad above eyes, and the postero-lateral limit of the scrobe is indicated by a difference in sculpture. The dentition of the mandibles, however, is of the rimosus-group, i. e. consisting of only five teeth.

There is no close relative for this species, even though Cyphomyrmex wheeleri and Cyphomyrmex costatus have the same reticulate-punctate sculpture of the antennal scrobe, and their preocular carina likewise does not curve mesad above eyes; wheeleri, in addition, even possesses the reticulate-punctate sculpture of the integument of body and appendages.



Kempf (1966) - (lectotype). - Total length 3.7 mm; head length 0.91 mm; head width 0.67 mm; thorax length 1.23 mm; hind femur length 1.17 mm. Fuscous brown; appendages lighter. Integument of body and appendages densely and sharply reticulate-punctate; body with densely distributed foveolae, which are rather difficult to see on account of the microsculpture, extending equally over both pits and intervals.

Head (fig 2). Mandibles longitudinally striolate, strikingly elongate. Clypeus with prominent and strongly convex, mesially vestigially notched, anterior apron; postero-mesially, in front of each anterior corner of frontal lobes, a vestigial tooth. Frontal lobes very little expanded laterad, continued posteriorly by the nearly straight, subparallel, strongly carinate, frontal carinae, which attain the triangular occipital lobes. Vertex with a pair of strong carinae. Preocular carinae fading out at level of eyes, not curving mesad nor prolonged caudad. Antennal scrobe neatly differentiated in its posterior half from the cheeks by the absence of coarse pits. No supraocular tumulus. Eyes strongly convex with about 7-8 facets across their greatest diameter. Cheeks immarginate below. Occiput prolonged into a low, collar-like prominence around foramen. Scapes elongate-clavate, projecting beyond the occipital lobes by a distance exceeding twice their maximum width. Funicular segment I very long, subequal to II-IV combined; segments II-IX distinctly elongate.

Thorax (fig 18). Pronotum completely unarmed, lacking also lateral carinae on disc; antero-inferior corner obtusely dentate. Mesonotum with two pairs of rather strong tubercles. Mesoepinotal constriction present. Epinotum completely unarmed; spiracle on a slight prominence. Legs rather long and thin; hind femora (fig 45). not dilated ventrally on basal third; posteroventral border vestigially carinulate.

Pedicel (fig 18, 29). Petiolar node, as seen from above, one and a half times as broad as long, its anterior corners broadly rounded; lacking dorsal carinae and tumuli. Postpetiole with an oblique anterior face nearly at right angle to the dorsal face; the latter with a pair of approximated small tubercles on anterior corners, another pair, blunter, stronger and more widely separated, on posterior corners; posterior border between tubercles slightly excised. Anterior border of first gastric tergum and sternum marginate; sides of tergum I practically immarginate.

Pilosity inconspicuous, consisting solely of fine, thin, silvery, decumbent hairs.

Type Material
Kempf (1966) - Several workers, collected at an altitude of 1020 m (3400') by Prof. N. A. Weber; 3 specimens seen.

Schultz et al. (2002) - LECTOTYPE (reexamined by TRS): Worker. Designated by Kempf (1966). Colombia: Rio Porce; 1020 m (3200’); nesting in rain forest; Weber No. 1088; 3 Aug 1938; N.A. Weber, collector. Based on a handwritten note accompanying Weber-collected vials in the MCZ, the type series was collected at 6°40’N, 75°10’W. Measurements (in mm, remeasured by TRS): HL=0.90; HW=0.64; WL=1.26; SL=0.79; maximum diameter of eye=0.15; hind femur length=1.14. Note that in the original description, Weber (1940, p. 407) reports a collection time period of June-August 1938 and that Kempf (1966) mistakenly reports the collection date as “Nov. 1938.”

Paratypes examined (10 total): 2 workers, 4 workers (MCZ), 1 worker (, W.W. Kempf Collection), 1 alate gyne (MCZ), 1 dealate gyne (MCZ), 1 male (MCZ): Colombia: Rio Porce; 1020 m; nesting in rain forest; No. 1088; 3-viii-1938; N.A. Weber, coll.