Rekoa stagira Hewitson

Rekoa stagira Hewitson

NOMENCLATURE.—Thecla stagira Hewitson, 1867, and Thecla erenea Hewitson, 1867: Hewitson (1862–1878) described and illustrated Thecla stagira from three males, which he considered varieties of the same species, and simultaneously named the second specimen Thecla erenea, a variety of T. stagira. The specimens were in the collections of the British Museum (locality not given), H.W. Bates (Santarem, Amazon), and W.C. Hewitson (Rio de Janeiro). Godman and Salvin (1887–1901) and Druce (1905) were unable to locate the type specimens.

The nomenclature of Thecla stagira and T. erenea is a mess. Hewitson badly confused this species with phenotypically similar species, and admitted that T. stagira gave him trouble. Godman and Salvin (1887–1901) and Druce (1905) noted that Hewitson put T. stagira labels on specimens of other species. Hewitson described this species five times in seven years.

A male labelled “Santarem” was segregated in the BMNH type collection as the type of T. stagira. This specimen fits Hewitson's original figures of the first two type specimens, and I designate it lectotype of both T. stagira and T. erenea. I also accord priority to T. stagira over T. erenea because previous synonymies are unclear on this point. Hopefully, these actions will end nomenclatural confusion and promote stability.

Godman and Salvin (1887–1901) noted that the third type above is a different species from the other two. The illustration of this specimen is a good representation of Thecla falerina, a species that (1) Hewitson named in 1867 from the Amazon, (2) that does not belong in the Thereus Section of the Eumaeini, and (3) that does not occur in Rio de Janeiro, so far as I am aware. I (Robbins, 1987) listed its closest relatives. This species has not been associated with T. stagira, and my action above will prevent that from happening.

Thecla timaea Hewitson, 1869: Hewitson (1862–1878) described and illustrated Thecla timaea from a female in the collection of H.W. Bates from Para, Amazon. There is a specimen from Bates' Collection that is so labelled and that fits the original figure. This specimen had been segregated in the BMNH type collection as the type of T. timaea, and I designate it lectotype. Kirby (1871) made it a secondary homonym of Pseudolycaena timaeus Felder and Felder, a valid action. Thus, T. timaea is a homonym of T. timaeus and cannot be used.

Thecla volana Hewitson, 1869: Hewitson (1862–1878) described Thecla volana from a female in the collection of W.W. Saunders from the Amazon. An Amazonian female that was in the Saunders Collection and then the Grose Smith Collection before being transferred to the BMNH is labelled the type of T. volana. It fits the original figure and was segregated in the BMNH type collection. I designate it lectotype of T. volana.

Thecla lydia Kirby, 1871: Kirby (1871) proposed Thecla lydia as a substitute name for the secondary homonym Thecla timaea Hewitson. Its type is the lectotype of T. timaea.

Thecla thoana Hewitson, 1874: Hewitson (1862–1878) described and illustrated this species from a Nicaraguan female in his collection. There is one Nicaraguan female in the BMNH that fits the original figure, and I designate it lectotype of T. thoana. It had been segregated in the BMNH type collection as the type of T. thoana.

Thecla carioca Ebert, 1965: Ebert (1965) described T. carioca as a subspecies of T. spurina. He designated a male holotype from Brazil, Guanabara (now Rio de Janeiro), Pao de Acucar. It is deposited in Curitiba, Brazil.

SYNONYMS AND NEW COMBINATIONS.—Hewitson (1862–1878) made Thecla erenea a synonym of T. stagira. Druce (1905) considered T. stagira, T. volana, and T. timaea synonyms of T. spurina. Draudt (1919–1920) followed Druce's synonymy, but added T. lydia to the list and made T. erenea an “insignificant form” of T. spurina. He also synonymized T. thoana to T. brescia. Thecla spurina and T. brescia belong to the synonymy of T. marius. The placement of T. stagira and its synonyms in Rekoa is a new combination for each.

GEOGRAPHICAL VARIATION.—The ventral wing pattern of R. stagira varies over its range. The major variable element is the ground color distal of the postmedian line. Specimens from Mexico to Nicaragua have lighter ground color distal of the postmedian line than basal to it, giving specimens a “two-tone” appearance (Figure 109). In Panama, the contrast in ground color between the distal and basal parts is less in the majority of specimens than it is further north (Figure 106), and some specimens show no difference in color. Specimens from Venezuela, Trinidad, and the Guianas show little, if any difference between the basal and distal ground color (Figure 108), while those from south of the Amazon Basin show no difference (Figure 110). As a result of this coloration, specimens from northern Central America look different from Brazilian specimens, and it is undoubtedly the reason that R. thoana had not been associated with R. stagira.

IDENTIFICATION.—Rekoa stagira is most often confused with its close relative R. marius. The lack of pale blue scales in the ventral hindwing submarginal spot, as noted in the Key, differentiates R. stagira from the other unbanded Rekoa species. Those few specimens of R. stagira with pale blue scales in the spot have them restricted to the basal part of the spot.

GEOGRAPHIC DISTRIBUTION (Figure 111), TEMPORAL DISTRIBUTION, AND ELEVATION.—MEXICO: Tamaulipas: Gomes Farias (280–700 m, (Jan). San Luis Potosi: Valles (26 Jun, Jul, Oct), Palitla (Dec). Veracruz: Jicaltepec (Mar), Presidio (Jan, Mar, May–Jul, Sep, Oct), Feronapa (??, Jun). Colima: Colima (1600 ft, Mar). Oaxaca: S. Jose Chiltepec (May). Tabasco: 2–8 mi E La Venta (Jan). Guerrero: Acahuizotla (Oct), Dos Arroyos (1000 ft, Sep). Chiapas: Comitan–Sta. Rosa (Mar, Jun).

GUATEMALA: Peten: Tikal (200 m, May). Alta Verapaz: Polochic Valley. Izabal: Quirigua (Sep, Nov), Cayuga (Aug).

EL SALVADOR: La Libertad: La Libertad (Jan, Feb).

HONDURAS: Cortes: San Pedro Sula.

NICARAGUA: Managua: Pochomil (Jan, Jul), Within 20 km of Managua (Feb–Apr; 19, 29 May; Jul; 15–23 Aug; 15 Sep). Magalpa: Cd. Dario (Jul). Madriz: Somoto (2000 ft, Sep).

COSTA RICA: Limon: Guapiles (May). Heredia: La Selva (found as larva 18 Apr).

PANAMA: Chiriqui: Bugaba. Canal Area (formerly Canal Zone): Los Rios (9, 24, 29 Jan; 3, 7, 21 Feb; 13, 30, 31 Mar; 2 Feb), Paraiso (30 Jan; 22, 24 Apr; 3 Jun; 9 Jul), Cocoli (13, 26 Jun), La Pita (10 Jun), Rodman (4 Feb), Madden Forest (8, 25 Feb; 4 Mar), Madden Dam (18, 25 Apr), La Boca (21–23, 25 Jan; 2 Feb), Chiva-Chiva Rd., near Paraiso (23 Jul), Gamboa (10, 17 May; 4 Sep; 16 Dec), Gatun (18, 24 Apr; May; 8 Jun), Barro Colorado Island (14 Mar; 29 May; 29 Nov). Colon: Pina (200 m, Mar). Panama: Bayano (8 Feb), Cordillera de San Blas, N of El Llano (330 m, 1 Jan).

FRENCH GUIANA: Guyane: Cayenne (Mar, Dec), St Jean du Maroni (Jul), Nouveau Chantier.

SURINAM: Paramaribo: Paramaribo (Oct–Nov), no further locality data (Nov). Saramacca: Saramacca River. Afobaka (??, 7–12 Jan).

GUYANA: East Berbice-Courentyne: no further data. Maroni (a mistake; either French Guiana or Surinam).

TRINIDAD: St. Patrick: Pitch Lake (17 Apr). St. George East: Tacarigua (Nov), 4 mi E Arima (Jul). ??: Balaba Hill (??, Mar), no further data (Aug).

VENEZUELA: Merida: no further data. Zulia: Mision El Rosario (50 m, 12/13 Jan). Carabobo: Las Quiguas, Esteban Valley. Bolivar: Suapure (Jan). Delta Amacuro: Rio Acure.

COLOMBIA: Valle: Cali (3260 ft, 6, 20 Feb), Canas Gordas–Cali (1000 m, Sep, Nov). Antioquia: Casabe (across R. Magdalena from Barrancabermeja, 21 Jan; 28 Mar; 20 Apr; Nov). Meta: Villavicencio (500 m, 1800 ft, 3, 5, 6, 12, 16, 18, 19 Jul; 1 Aug). Cesar: Manaure. ??: Env. Bogota.

ECUADOR: Pichincha: Tinalandia (700 m, 26 May; 28 Aug). El Oro: No further locality data (Jun).

PERU: Junin: Satipo (Oct). Madre de Dios: Puerto Maldonado (290 m, 15 Oct).

BOLIVIA: Santa Cruz: Las Juntas, Sta. Cruz de la Sierra (450 m).

PARAGUAY: Central: Asuncion (Jul–Sep). Cordillera: Santisima Trinidad (Aug). Paraguari: Sapucay. ??: S. Paraguay, no further data (Mar).

ARGENTINA: Misiones: Goa (??).

BRAZIL: Para: No further data (female reared by A.M. Moss), Belem (12 Jan), Igarape-Acu (Jan, Feb, Dec), Santarem, Obidos (Aug), Tapajos. Amazonas: Igaribe (??), Tonantins, San Juan–Solimoes (??), Manaus, Tefe. Ceara: No further data (Aug). Mato Grosso: Chapada Campo (??), Colegio Buriti–Chapada Guimaraes (700 m, 25 May). Minas Gerais: Cabeceira do Corrego Leitao, 8 km from Belo Horizonte-Brasilia Hy on rd. to Curvelo (Apr). Espirito Santo: Conceicao da Barra (Jun), Linhares (Apr, May, Aug). Rio de Janeiro: Pao de Acucar (150 m, 5 Jul), Santa Teresa (100 m, 2 Apr), Restinga da Tijuca; the previous three localities from Ebert (1965). I believe that they are within the city of Rio de Janeiro, but cannot find them on maps. São Paulo: Mendes (??, There is one in Rio de Janeiro), Sorhuini (??), Araras (30 May). Parana: Londrina (4 Jun). Santa Catarina: Joinville (10–200 m, 22, 23 Apr), Massaranduba-Blumenau.

ECOLOGY AND BEHAVIOR.—Habitat: Rekoa stagira occurs throughout the wet and dry neotropical lowlands (except for western Peru) from Tamaulipas, Mexico, to Santa Catarina, Brazil, but differs ecologically from the other widespread Rekoa species. It is found in both disturbed and relatively undisturbed habitats up to 1100 m elevation, whereas R. marius and R. zebina are more restricted to disturbed habitats, but at a wider range of elevations. Rekoa stagira has been reared at La Selva, Costa Rica—an area that receives more than 300 cm annual precipitation (Coen, 1983; Hartshorn, 1983), whereas R. zebina rarely occurs in areas with more than 250 cm annual precipitation. As a rough measure of rarity, I have not seen more than three specimens of R. stagira in a day in contrast to the more common R. marius and R. palegon.

Dispersal: Rekoa stagira does not seem to be very dispersive; it has not been collected at wind dispersal sites in Panama (Robbins and Small, 1981) or Venezuela (Robbins, unpublished), and with the exception of Trinidad, is unknown on continental islands.

Larval Foodplants: Rekoa stagira has been reared three times. I identified the specimens except the record in D'Araujo e Silva et al. (1967–1968).

Malpighiaceae: vine with yellow flowers. Male. Costa Rica: Heredia Province: La Selva. Leg. P. DeVries. Ex. larva, 18 Apr 1972, pupated 22 Apr 1972, eclosed 1 May 1972. Larva feeding on young leaves. In NMNH with pupal case.

Leguminosae. Brazil: Rio de Janeiro: Campos. Observation of Aristoteles Silva. Published in D'Araujo e Silva et al. (1967–1968) under the name Thecla spurina.

Brazil: Para. Female. Leg. A.M. Moss. No foodplant mentioned on specimen or in Moss' notes in the BMNH library. In BMNH with pupal case.

Adult Foodplants: The only nectar feeding record for R. stagira is on Hamelia flowers (Rubiaceae) in Panama.

Territoriality and Pheromones: Males of R. stagira set up mating territories on hilltops from 1230–1400 h, unlike R. marius and R. palegon, which do not set up mating territories on hilltops. I barely perceived a faint unrecognizable odor in males from Panama. The following records of territorial behavior are my observations in Panama's Canal Area (formerly Canal Zone). Specimens are deposited in the NMNH Collection.

Two males. Top of unnamed hill in Gamboa. 10 May 1979.

1400 h. Perching height 1–2 m.

Male. Top of unnamed hill in Gamboa. 17 May 1979. 1330 h.

Perching height 4 m.

Male. Top of Cerro Pelado in Gamboa. 4 Sep 1979. 1320 h.

Male. Top of Cerro Pelado in Gamboa. 16 Dec 1979. 1230 h.

Male. Top of unnamed hill in Paraiso. 30 Jan 1980. 1400 h.

Wing Pattern and Predation: Of 6 specimens collected at Villavicencio, Colombia (Robbins, 1981), none showed symmetrical wing damage indicative of an unsuccessful predator attack to the hindwing anal angle.

Phylogenetic Analysis

The database for the phylogenetic analysis is the character matrix (Table 1), which contains 47 characters, 116 character states, seven species, and outgroup data. There are 15 multistate characters. I used Swofford's PAUP version 2.4 software for phylogenetic analysis on PC-compatible computers for most of the analysis.

The most parsimonious network for Rekoa using the Branch and Bound option (Figure 115) contains 78 steps, and has a consistency index of 0.885. Seven characters are homoplastic, five (3, 15, 32, 41, 44) requiring one extra step and two (33, 43) requiring two extra steps. Running the qualitative multistate characters (12, 29, 40) with the Unordered option did not change the network because these characters are not homoplastic.

The most parsimonious 78-step network is well supported. In the PAUP run, the node for the group R. meton and R. malina is supported by 12 character state changes; that for R. meton, R. malina, and R. palegon by 15 changes; that for R. zebina and R. bourkei by 6 changes; and that for R. marius and R. stagira by 7 changes (Figure 115). Because a number of characters can be placed on the tree in different, but equally parsimonious ways, there is no unique number of changes supporting each node. However, the large number of character changes supporting each node coupled with a high consistency index is an indication that the network is a well-supported phylogenetic hypothesis. Further, the second shortest network (calculated using the Bbsave option) requires 5 extra steps. The significance of the strong support for the 78-step tree is that I can confidently use it to test evolutionary hypotheses and to construct a classification.

Because quantitative characters have rarely been used in lycaenid phylogeny, I compare them with qualitative ones. Twenty two characters are quantitative measures of antennae, androconia, male genitalia, and female genitalia. The most parsimonious network for the quantitative characters is the same as the one for the entire data set, and has a consistency index of 0.848. The consistency and informativeness of the quantitative measures of antennae, labial palps, and androconia suggest that these characters should be used far more extensively by lepidopterists. The most inconsistent quantitative characters are genital measurements (characters 32, 33, and 43 are homoplastic), which appear to change “quickly” over evolutionary time. They are thus good specific distinguishing structures, but not particularly good phylogenetic ones.

The 25 qualitative characters are primarily wing-pattern characters. The most parsimonious network for the qualitative characters is the same as the one for the entire data set, and has a consistency index of 0.938. Because change in the majority of qualitative wing-pattern characters occurred between the banded and unbanded species, I suspect that many of these changes are not independent.

The root of the most parsimonious network (Figure 115) is between the immediate ancestor of R. meton and R. malina and that of R. palegon (Figure 116). The distance between the outgroups and the root is 0 steps, and is the only root location that is 0 steps. The lineage on the right side of the cladogram evolved (point C, Figure 116) a ventral, sclerotized, anterior pointing pouch between the female sixth and seventh abdominal segments (character 31) that does not occur in Thereus or Arawacus. The lineage on the left side of the cladogram evolved (point A) a ventral hindwing postmedian band in cell Sc+R1-Rs displaced so that it is basal to the end of the discal cell (character 29). This displaced spot does not occur in the outgroup genera. Thus, rooting the network beyond points C or A would require at least one change in characters 31 or 29. Although I would prefer to have more information than these two characters for rooting the network, it will probably not be possible to do better until Thereus and Arawacus are revised.

Evolution

In this section, I discuss Rekoa evolution by superimposing biological traits on the cladogram to the species (Figure 116) (Coddington, 1988). Using parsimony to determine the character states of ancestors at the nodes of the cladogram, I estimated those points where each trait evolved.

HABITAT.—Most Rekoa species are elevation generalists, occurring from sea level to at least 2000 m. There are two exceptions. Rekoa stagira occurs only rarely above 1100 m elevation (lowland specialist), while R. malina does not occur in the tropical lowlands (upland/temperate climate specialist). In both cases, habitat specialization evolved from the more general condition (at points G and B, respectively, on the cladogram, Figure 116).

Most Rekoa species are also rainfall generalists, occurring in both dry and wet habitats. The exceptions are R. zebina and R. bourkei, which are usually restricted to habitats with less than 250 cm annual precipitation. The specialization evidently evolved from the general condition once (designated F, Figure 116).

These instances of elevation and rainfall habitat specialization are noteworthy because in each case, the distribution of the specialist lineage overlaps that of the generalist sister clade. For example, although R. stagira is a lowland specialist, its sister species, R. marius, is sympatric with it in the lowlands. There is no evidence that habitat specialization affected speciation or coexistence.

DISPERSAL.—Available evidence indicates that the widespread Rekoa species (R. meton, R. palegon, R. marius, and R. stagira) have a marked propensity for dispersal. Each is sympatric with its sister clade, occurs on offshore islands where it is undifferentiated, and has been found at wind-mediated dispersal sites (except R. stagira).

BIOGEOGRAPHIC PATTERNS.—Although a majority of Rekoa species are distributed throughout the Neotropics, there are several distribution patterns within the genus that are repeated in other hairstreaks or animals.

The distribution of R. zebina—from Mexico and Guatemala south, primarily along the Pacific side of Central America, to Guanacaste Province, Costa Rica (Figure 91)—mirrors that of many animals, such as terrestrial vertebrates, birds in particular, and hawk moths that are also restricted to seasonally dry habitats (Slud, 1964; Mueller, 1973; Schreiber, 1978). Other hairstreaks with a similar distribution include Ministrymon clytie Edwards, Cyanophrys miserabilis Clench, C. goodsoni Clench, Strymon bebrycia Hewitson, S. cestri Reakirt, S. alea Godman and Salvin, a new species of Brangas, Arawacus sito Boisduval, and a new species near “Thecla” hesperitis Butler and Druce.

Rekoa meton and R. malina represent another distribution pattern among hairstreaks. Rekoa meton occurs from Mexico to southern Brazil (Figure 70), primarily in the lowlands. Rekoa malina, its sister species, is restricted to Brazil's central planalto, coastal moutains, and temperate areas to the south (Figure 83). Other presumed sister-species pairs among the hairstreaks with a similar distribution pattern are Arcas imperialis Cramer and A. ducalis Westwood, “Thecla” rustan Stoll and “T.” polama Schaus, Theritas mavors Hübner-T. triquetra Hewitson and T. drucei Lathy, “Thecla” ligurina Hewitson and “T.” species near ligurina, and Contrafacia imma Prittwitz and C. muattina Schaus. This pattern can be tested by determining whether these pairs are sister taxa. Further, I believe that other examples will be found as more eumaeine genera are revised.

Although the biogeographic history responsible for the R. meton–R. malina distribution pattern is unknown, I suggest that these pairs speciated in the southern neotropics and that the widespread species, in this case R. meton, then spread northward. The evidence is as follows: There are two geographcially variable wing pattern characters in R. meton. First, the ventral hindwing basal line is present in the southern part of its range, vestigial in central and northern South America, and absent in Central America. Because R. malina and R. palegon both have the ventral hindwing basal line, its loss in the northern half of the neotropics is a derived character state. Second, female dorsal ground color in R. meton is blue in southern Brazil, blue-tinted in Paraguay and central Brazil, and white everywhere else. The blue character state also occurs in R. malina, once again indicating that the northern populations of R. meton are derived from the southern ones. In general, these hypotheses can be tested when the widespread species is geographically variable (or composed of two or more parapatric species).

LARVAL FOODPLANTS AND MORPHOLOGY.—Larval foodplant data and morphology are recorded for four of the seven Rekoa species. Rekoa marius feeds on the flowers, fruits, and sometimes leaves of plants in seven families, but the Leguminosae and Malpighiaceae account for a majority (65%) of the records. Rekoa stagira and R. zebina are also recorded from these families, and this foodplant specificity probably evolved at cladogram point E (Figure 116). Rekoa palegon, on the other hand, eats plants in five families, but most of the records (65%) are Compositae. Rekoa marius is not recorded from Compositae, while R. palegon apparently does not eat Leguminosae or Malpighiaceae. Further, the unique spiny green larva of R. palegon is different from the unspined larvae of the other species (and other eumaeines), and apparently evolved at point D (Figure 116). The hypothesis that sympatric eumaeine larvae partition resources is falsified by the similarity in foodplants between R. marius and R. stagira and between these two species and R. zebina, whereas the differences between R. palegon and the others is consistent with it.

COURTSHIP BEHAVIOR.—There is little recorded information on Rekoa courtship behavior. Male territorial behavior is known only for R. stagira, but negative evidence indicates that its behavior is unique in the genus, and probably evolved at point G on the cladogram (Figure 116). This indicates that changes in courtship behavior might allow sister species to co-exist, as has been suggested in riodinid butterflies (Callaghan, 1983).

WING PATTERN AND PREDATION.—I (Robbins, 1981) placed R. meton and R. palegon—the banded species—in rank 2 (good false head wing patterns), and R. marius and R. stagira—the unbanded species—in rank 3 (average wing patterns). However, R. palegon is attacked at a rate equivalent to the unbanded species (less than 7% of specimens with predator inflicted hindwing damage), while R. meton is attacked almost four times as frequently (24%). Because R. malina has virtually the same ventral wing pattern as sympatric populations of R. meton, I infer that at point A on the cladogram (Figure 116), an effective “false head” wing pattern evolved. The wing pattern of R. palegon is evidently more cryptic than directive, contrary to my earlier assumption. The R. palegon wing pattern also appears to be the primitive one in Rekoa, and is very similar to that of Thereus pseudarcula and of some Arawacus, such as A. ellida Hewitson. The marked change from the banded wing pattern to the unbanded one occurred at point E on the cladogram (Figure 116), but there is no evidence that this change affected predator attacks, as there is for the change at point A.

The striking ventral wing pattern of R. meton (Figure 65) is unlike that of other eumaeines except R. malina. Particularly noteworthy is the eyespot at the base of hindwing cell Sc+R1-Rs. Eyespots or similar markings at the base of this hindwing cell have evolved repeatedly among the eumaeines (Robbins, unpublished). It is currently unclear whether these eyespots influence predator attacks. However, I hypothesize that in each case, the eyespot evolved from a piece of the postmedian line in cell Sc+R1-Rs that was displaced basally. Indeed, there is no postmedian line segment in this cell in Rekoa meton. Further, the intermediate stages in the displacement process are reasonably clear (R. palegon, R. malina, see discussion in the Morphology Section). If this hypothesis is correct, then it should shed some light on the “rules” of wing pattern development in eumaeines.

EVOLUTION OF SPECIES DIVERSITY.—Perhaps the most notable characteristic of Rekoa evolution is that ecological or behavioral differences evolved in at least one daughter species after almost every speciation event. At point A (Figure 116), an effective “false head” wing pattern that misdirects predator attacks evolved. At point B (Figure 116), the ancestor of R. malina became a habitat specialist in upland/subtropical climates. At point D (Figure 116), a spiny larva that feeds on composites evolved. At point F (Figure 116), the ancestor of R. zebina and R. bourkei became restricted to xeric habitats with a marked dry season. At point G, the ancestor of R. stagira evolved a mating system in which males set up territories on hilltops in the afternoon, and became a tropical lowland habitat specialist.

With only seven Rekoa species, it is not possible to say much about correlates of relative rates of increase in species diversity. However, outgroup genera Thereus and Arawacus have 27 and 19 species, respectively, so that a comparison of ecological factors in these genera may give insight into the correlates of variable increases in species diversity. Two factors are immediately evident. Although four of the seven Rekoa species are widespread from the Tropic of Cancer to the Tropic of Capricorn, only two Thereus and no Arawacus species have such a widespread distribution. Further, Arawacus larvae are restricted to feeding on the leaves of Solanaceae, in contrast to the usual flower feeding polyphagy among the Eumaeini. Evolution in these genera should be a revealing contrast to Rekoa.