Crematogaster hova

Blaimer, Bonnie B., 2010, Taxonomy and Natural History of the Crematogaster (Decacrema) - group (Hymenoptera: Formicidae) in Madagascar, Zootaxa 2714, pp. 1-39 : 17-22

Crematogaster hova -complex

( Fig. 44–53 )

Crematogaster hova Forel, 1887:387 . Worker syntypes from Madagascar: Bois de l’Ivondro près de Tamatave (Dr.C.Keller) [MHNG, examined;]. One worker syntype (CASENT0101804, top specimen of 3w on one pin, image on Antweb) hereby designated the LECTOTYPE.

Crematogaster hova Forel, Forel,1910b:9 . Combination in C. ( Decacrema ).

= Crematogaster hova latinoda Forel, 1892:535 . Worker syntypes from Madagascar, Mangoroufer, Est Imerina (M.Sikora) [MHNG, examined]; Wheeler, 1922b:1026. Combination in C. ( Decacrema ). NEW SYNONYMY

Crematogaster schenki Forel,1891:182 . Workers syntypes from Madagascar: Imerina: Antananarivo, Andrangoloaka (Rév. Père Camboué, Hildebrandt, Sikora) [MHNG, examined]. One worker syntype (CASENT0101748, top specimen of 2w on one pin, image on Antweb) hereby designated the LECTOTYPE.

Crematogaster schenki Forel, Forel,1910a:18 and Forel, 1910b:9. Combination in C. ( Decacrema ).

Crematogaster schencki as justified emendation, Emery, 1922c:138.

Crematogaster (Decacrema) ensifera Forel,1910a:18 . Worker syntypes from Madagascar: Forêts vierge de Sahana [MHNG, examined]. One worker syntype (CASENT0101790, middle specimen of 3w on one pin, image on Antweb) hereby designated the LECTOTYPE.

Material examined. ( BBBC, CASC, PSWC, MHNG, NHMB, ZMBH, ZSM) MADAGASCAR: Antsiranana: Sakalava Beach: -12.26278, 49.39750, 10m (R. Harin'Hala); 3km W of Sakalava beach: - 12.28617, 49.36667, 40m (R. Harin'Hala); Montaigne Français: -12.32500, 49.33333, 150m (R. Harin'Hala); 5km SW Ambohitra [Joffreville]: -12.53330, 49.16670, 1100m (P.S.Ward); 7 km N Joffreville: -12.33333, 49.25000, 360m (R. Harin'Hala); R.S. Ambre:-12.46889, 49.24217, 325m (B.L.Fisher et al.); P.N. Montagne d’Ambre: -12.53444, 49.17950, 925m; -12.51444, 49.18139, 960m; -12.51667, 49.18333, 975m; -12.58132, 49.13368, 1110m; -12.52028, 49.17917, 1125m; -12.58500, 49.15147, 1250m (B.L.Fisher et al.); P.N. Montagne d’Ambre: -12.51670, 49.16670, 1100m (D.M.Olson); R.S. Manongarivo: -13.93145, 48.45143, 360m; -13.96167, 48.43333, 400m; -13.97667, 48.42333, 780m; -13.99833, 48.42833, 1175m; -14.02167, 48.41833, 1580m; -14.04667, 48.40167, 1860m (B.L.Fisher et al.); R.S. Manongarivo: -13.93145, 48.45183, 360m; -13.93153, 48.45213, 370m; -13.93188, 48.4492333, 470m; -13.93155, 48.449333, 520m; (B.B.Blaimer); Ampasindava, Ambilanivy: -13.79861, 48.16167, 600m (B.L.Fisher et al.); F Andavakoera: - 13.11833, 49.23000, 425m (B.L.Fisher et al.); F Antsahabe: -13.21167, 49.55667, 550m (B.L.Fisher et al.); Rés. Analamerana: -12.80467, 49.37383, 225m (B.L.Fisher et al.); F Binara: -13.25500, 49.61667, 375m; - 13.26333, 49.60333, 650-800m; -13.26333, 49.60000, 1100m (B.L.Fisher et al.); Montagne d’Akirindro: - 15.28833, 49.54833, 600m (B.L.Fisher et al.); 6.9 km NE Ambanizana: -15.56667, 50.00000, 825m (B.L.Fisher et al.); Nosy Mangabe: -15.5000, 49.7667, 300m (P.S.Ward); Rés. Anjanaharibe-Sud: -14.75000, 49.50000, 875m; -14.75000, 49.46667, 1180-1280m; -14.75000, 49.45000, 1480-1585m (B.L.Fisher et al.); Montagne d’Anjanaharibe: -15.18833, 49.61500, 470-1100m (B.L.Fisher et al.); P.N. Marojejy: -14.43333, 49.78333, 450m; -14.43817, 49.77400, 488m; -14.43500, 49.76000, 775m; -14.44333, 49.74333, 1325m; - 14.44500, 49.74167, 1575m; -14.44500, 49.73500, 2000m (B.L.Fisher et al.); P.N. Marojejy [Manantenina]:

-14.43667, 49.77500, 450m; -14.43333, 49.76667, 750m; -14.43333, 49.75000, 1225m; -14.45000, 49.73333, 1875m (B.L.Fisher et al.); F Betaolana: -14.52996, 49.44039, 880m (B.L.Fisher et al.); Antananarivo: NE Andranomay: -18.47333, 47.96000, 1300m (B.L.Fisher et al.); F Analamazoatra: -18.38466, 48.41270, 980m (B.L.Fisher et al.); R.S. Ambohitantely: -18.19800, 47.28150, 700m; -18.22528, 47.28683, 1410m (B.L.Fisher et al.); R.S. Ambohitantely: -18.21656, 47.28275, 1520m; -18.21722, 47.28470, 1550m; - 18.19473, 47.28642, 1560m; -18.19442, 47.28623, 1590m; -18.17272, 47.28023, 1610m (B.B.Blaimer); 25km NNE Ankazobe: -18.10000, 47.18330, 1500m (P.S.Ward); Kaloy: -18.58998, 47.65102, 1420m (B.L.Fisher et al.); Antananarivo: -18.91667, 47.51667 (Camboué); Andrangoloaka: -19.03333, 47.91667 (Sikora); Fianarantsoa: F Atsirakambiaty: -20.59333, 46.56333, 1550m (B.L.Fisher et al.); P.C. Ankazomivady: -20.77667, 47.16500, 1780m (B.L.Fisher); P.N. Isalo: -22.48167, 45.46167, 725m; - 22.62667, 45.35817, 700-825m (B.L.Fisher et al.); 9km NNW Ranohira, P.N. Isalo: -22.48330, 45.38330, 800m (P.S.Ward); F Analalava: -22.59167, 45.12833, 700m (B.L.Fisher et al.); P.N. Ranomafana: -21.26650, 47.42017, 1020m; -21.29000, 47.43333, 1100m; -21.23333, 47.36667, 1110m; -21.25000, 47.36667, 1130m; -21.23667, 47.39667, 1150m (B.L.Fisher et al.); 3km W Ranomafana, nr Ifanadiana: -21.25000, 47.41670, 950m (P.S.Ward);); 3km WNW Ranomafana, nr Ifanadiana: -21.25000, 46.43330, 840m (P.S.Ward); F.C. Vatovavy: -21.40000, 47.94000, 175m (B.L.Fisher et al.); 2km W Andrambovato: -21.51167, 47.41000, 1075m (B.L.Fisher et al.); Tsaranoro: -22.08317, 46.77400, 975m (B.L.Fisher et al.); 45km S Ambalavao, Rés. Andringitra [=P.N. Andringitra]: -22.21667, 47.01667, 720-785m (B.L.Fisher); 40km S Ambalavao, Rés. Andringitra [=P.N. Andringitra]: -22.21667, 46.96667, 1275m (B.L.Fisher); P.N. Andringitra: -22.21167, 46.84500, 1800m (S. Razafimandimby) P.N. Andringitra: -22.22350, 47.01176, 780m; -22.22806, 47.00335, 890m; -22.21495, 46.96890, 1400m (B.B.Blaimer); R.S. Ivohibe: -22.47000, 46.96000, 900m; -22.42167, 46.89833, 1200m (B.L.Fisher et al.); F Vevembe: -22.79100, 47.18183, 600m; Mahajanga: F Telomirahavavy: -18.12167, 47.20627, 1520m (B.L.Fisher et al.); F Andranorovitra: -18.11243, 47.19757, 1490m (B.L.Fisher et al.); F Sohisaka: -18.10322, 47.18692, 1460m (B.L.Fisher et al.); Rés. Marotandrano: - 16.28322, 48.81443, 865m (B.L.Fisher et al.); P.N. Namoroka: -16.37667, 45.32667, 100m (B.L.Fisher et al.); Toamasina: P.N. Mananara-Nord: -16.45500, 49.78750, 225m (B.L.Fisher et al.); Rés. Ambodiriana: - 16.67233, 49.70117, 125m (B.L.Fisher et al.); Forêt vierge de Sahana: -16.08333, 49.63333 (unknown); RS Ambatovaky: -16.81739, 49.29402, 360m; -16.81620, 49.29202, 425m; -16.77550, 49.26427, 430m; - 16.77274, 49.26551, 450m; -16.81209, 49.29216, 460m; -16.77020, 49.26638, 470m; -16.76912, 49.26704, 475m; -16.80561, 49.29507, 480m; -16.76330, 49.26692, 520m (B.L.Fisher et al); P.N. Andasibe: -18.77071, 48.43164, 995m; -18.92639, 48.40783, 1025m (B.L.Fisher et al.); 1km SSW Andasibe [Périnet]: -18.93330, 48.41670, 920m (P.S.Ward); 6km ESE Andasibe [Périnet]: -18.95000, 48.46670, 900m (P.S.Ward); 8km ESE Andasibe [Périnet]: -18.95000, 48.50000, 800m (P.S.Ward); Rés. Perinet-Analamazoatra: -18.93330, 48.43330, 950m (D.M.Olson); F Torotorofotsy: -18.87000, 48.34670, 1070m (B.L.Fisher et al.); F Didy: - 18.19833, 48.57833, 960m (B.L.Fisher et al.); Manakambahiny: -17.75000, 48.71667 (B.L.Fisher et al.); F Ambatovy: -18.84770, 48.29600, 1000m; -18.84950, 48.29470, 1010m; -18.85800, 48.28480, 1040m; - 18.85083, 48.32000, 1075m; -18.83930, 48.83930, 1080m (B.L.Fisher et al.); F Analamay: -18.80623, 48.33707, 1068m (B.L.Fisher et al.); F.C. Sandranantitra: -18.04833, 49.09167, 450m (B.L.Fisher et al.); P.N. Mantadia: -18.79167, 48.42667, 875m; -18.79167, 48.42667, 895m (B.L.Fisher et al.); P.N. Zahamena: - 17.75244, 48.85320, 760m; -17.75257, 48.85725, 765m; -17.75908, 48.85468, 780m (B.L.Fisher et al.); Rés. Betampona: -17.92400, 49.19967, 390m; -17.91801, 49.20074, 500m; -17.88667, 49.20250, 520m (B.L.Fisher et al.); Est-Imerina: Mangoro-ufer (Sikora); Bois de l’Ivondro près de Tamatave: -18.23333, 49.36667 (Keller); Toliara: R.S. Ambohijanahary: -18.26667, 45.40667, 1050m (B.L.Fisher et al.); F Vohimena: -22.68330, 44.83330, 730m (B.L.Fisher et al.); P.N. Zombitse: -22.84333, 44.71000, 770m; - 22.84050, 44.73117, 825m; -22.88650, 44.69217, 840m (B.L.Fisher et al.); 15km E Sakaraha: -22.90000, 44.68330 760m, (P.S.Ward); F Analavelona: -22.67500, 44.18667, 1050m; -22.67500, 44.19000, 1100m; - 22.64333, 44.17167, 1300m (B.L.Fisher et al.); Fiherenana: -23.23528, 43.87083, 50m; -23.17694, 43.96083, 100m (B.L.Fisher et al.); Mahafaly Plateau: -24.65361, 43.99667, 80m (B.L.Fisher et al.); P.N. Befotaka- Midongy: -23.84080, 46.95750, 1250m (B.L.Fisher et al.); P.N. Andohahela: -24.75850, 46.85370, 275m; and 24.76389, 46.75167, 900m (B.L.Fisher et al.); 10km NW Enakara, Rés. Andohahela: -24.56667, 46.81667, 430m; 13km NW Enakara, Rés. Andohahela: -24.55000, 46.80000, 1150-1250m (B.L.Fisher); P.N. Andohahela: -24.73732, 46.83950, 550m (B.B.Blaimer); 5km NNW Isaka-Ivondro, Rés. Andohahela: - 24.75000, 46.85000, 280m (P.S.Ward); 6km SSW Eminiminy, Rés. Andohahela: -24.75000, 46.78330, 500m (P.S.Ward); 9km SSW Eminiminy, Rés. Andohahela: -24.73330, 46.80000, 330m (P.S.Ward); Col de Manangotry: -24.73330, 46.85000, 1500-1800m (D.M.Olson); F Ivohibe: -24.56170, 47.20020, 650m (B.L.Fisher et al.); Grand Lavasoa: -25.08770, 46.74900, 450m (B.L.Fisher et al.); province unknown: Central Madagascar (Hildebrandt); Südcentral Madagascar (unknown).

Worker measurements (n=58). HW 0.74–1.19; HL 0.69–1.08; EL 0.15–0.27; SL 0.65–0.93; WL 0.78– 1.22; SPL 0.19–0.44; PTH 0.15–0.24; PTL 0.22–0.38; PTW 0.22–0.39; PPL 0.14–0.23; PPW 0.20–0.33; LHT 0.64–0.97; CI 1.04–1.19; OI 0.18–0.29; SI 0.83–0.99; SPI 0.19–0.40; PTHI 0.52–0.86; PTWI 0.83–1.39; PPI 1.24–1.79; LBI 1.17–1.56.

Worker description. Size highly variable, small to very large (HW 0.74–1.19, WL 0.78–1.22), commonly medium to large size.

Masticatory margin of mandibles with 4 teeth in smaller individuals, 5 teeth in larger specimens; posterior margin of head straight but sometimes medially depressed, laterally forming round or subangular corners; antennae with scapes reaching or surpassing posterior margin; midline of eyes situated at midline of head in full face view.

Pronotum laterally with distinct rounded or subangular shoulders; promesonotal suture incomplete, laterally impressed; outline of promesonotum in lateral view more or less rounded; dorsal face of mesonotum either flat or weakly convex, sometimes medially slightly raised over pronotum and rarely forming a small median tubercle anteriorly; lateral margin of mesonotum from rounded to slightly angular, sometimes bearing postero-lateral angular tubercules; length and divergence of propodeal spines variable; petiole highly variable; shape in dorsal view from weakly to strongly lobed, trapezoidal, hexagonal or suboval, in lateral view petiole usually tapering in height from posterior to anterior; subpetiolar process variable, either narrow to broad protuberance, with more or less angular tooth or absent, postpetiole in dorsal view from weakly to much wider than long (PPI 1.24–1.79).

Head sculpture variable (see below); sculpture on mesosoma following mostly that on head, aciculate, reticulate, costulate or areolate; dorsal face of petiole mostly shiny, ventrally coarsely reticulate; postpetiole dorsally and laterally rugulose or reticulate, helcium carinulate; face with 4–6, sometimes up to 10 erect setae; 0–2, sometimes 4, rarely 6 erect-suberect humeral setae present on promesonotum; petiole and postpetiole devoid of long erect pilosity, sometimes shorter, suberect setae present on postpetiole.

Colour light to dark brown or black, sometimes dorsum of 4th abdominal segment yellow and gaster thereby bi-coloured.

Queen measurements (n=16) HW 1.60–1.97, HL 1.37–1.63, EL 0.43–0.53, SL 0.99–1.19, MSNW 1.22– 1.60, MSNL 2.36–2.67, SPL 0.02–0.13, WL 2.68–3.05, PTH 0.38–0.45, PTL 0.47–0.59, PTW 0.53–0.69, PPL 0.40–0.47, PPW 0.54–0.68, LHT 1.08–1.36, CI 1.14–1.25, OI 0.29–0.37, SI 0.69–0.78, MSNI 0.52–0.66, SPI 0.01–0.05, PTHI 0.70–0.91, PTWI 1.00–1.36, PPI 1.28–1.69, LBI 2.13–2.32.

Queen description. Large to very large (HW 1.60–1.97, WL 2.68–3.05). With worker characters, except as described below.

Antennal scapes usually not surpassing posterior margin of head; midline of eyes situated at or slightly below midline of head in full face view; posterior margin of head straight or medially depressed.

Mesosoma broad and long (MSNI 0.52–0.66, WL 2.68–3.05); mesoscutum in dorsal view oval, less than half as wide as long; mesopleuron with episternal groove carinulate and usually ending in small pit just short of margin; in lateral view mesepisternum meeting pronotum at an oblique angle; postscutellum with posterior face distinctly flattened; propodeal suture weak, laterally not reaching level of propodeal spiracle; dorsal face of propodeum short, about one third of the size of posterior face; propodeal spines reduced, ranging from tubercule to denticles to very sharp points; petiole shape usually triangular, tapering from anterior to posterior margin, antero-ventral subpetiolar tooth absent.

Propodeum with horizontal carinulae on dorsal face; petiole dorsally more or less shiny, laterally and ventrally reticulate to costulate; postpetiole aciculate throughout; face with 10–20 short, erect setae, and otherwise abundant appressed pilosity; mesonotum with scattered erect pilosity, usually>20 setae, and abundant appressed pubescence; petiole laterally with short suberect pilosity, dorso-posterior erect setae absent; postpetiole with dorso-posterior erect setae present, further scattered suberect pilosity may be present; petiole and postpetiole with appressed pilosity throughout.

Colour light to dark brown, or black; metasoma usually lighter coloured.

Variation. I distinguish 5 morphotypes of workers within the C. hova -complex on the basis of sculpture, body size and propodeal spines. Designation of these morphotypes is somewhat arbitrary as they are not entirely distinct from each other, but have intermediate forms – and further intermediate forms of these ‘intermediates’ exist. They should therefore not be perceived as distinct units, but rather as samples from an overall gradient. For the purpose of distribution mapping some ambiguous assignments to morphotypes were therefore made. The following distinction however simplifies the description of morphological variation greatly, and allows for its analysis for possible geographic patterns. Moreover, this morphotype concept presents a starting hypothesis for a much needed study on the population genetic level of the C. hova - complex.

Morphotype 1 (mt1) ( Fig. 44 & 49 ). Small to large size (HW 0.83–1.10, WL 0.87–1.14). Head sculpture reduced aciculate to superficially areolate; propodeal spines medium size, at most as long as distance between their bases (SPI 0.21–0.28), in dorsal view only weakly diverging (<20º). Colour light brown to dark brown, less common black, metasoma often lighter coloured.

This morphotype conforms most closely with the C. hova and C. hova latinoda type specimens. C. hova latinoda type material has a petiole that is strongly broadened anteriorly with respect to C. hova .

Morphotype 2 (mt2) ( Fig. 45 & 50 ). Medium to large size (HW 0.92–1.19, WL 0.93–1.22). Head sculpture reticulate, reticulate-areolate to weakly areolate; propodeal spines medium size, at most as long as distance between their bases (SPI 0.22–0.26), in dorsal view only weakly diverging (<20º). Colour light brown to black, metasoma often lighter coloured in brown form, 4th abdominal segment or entire gaster sometimes of yellow coloration (may be due to parasitism rather than inherent pigmentation).

This morphotype is intermediate in regard to mt1 and mt4.

Morphotype 3 (mt3) ( Fig. 46 & 51 ). Small to large size (HW 0.74–1.13, WL 0.78–1.19). Head sculpture from scattered aciculate to reticulate to costulate-areolate or areolate; propodeal spines large and robust, often “swollen” at base, at least as long or longer than distance between their bases (SPI 0.29–0.40), in dorsal view strongly diverging, ca.>30º from lateral margin of propodeum. Colour brown to black.

This morphotype conforms best with the C. ensifera type material.

Morphotype 4 (mt4) ( Fig. 47 & 52 ). Medium to large size (HW 0.97–1.18, WL 1.01–1.21). Head sculpture deeply costulate to costulate-areolate; propodeal spines medium size (SPI 0.24–0.29), about as long as distance between bases or shorter, in dorsal view weakly or moderately diverging (<30º). Colour brown to black, more often black, sometimes gaster shows bi-coloration described for mt2.

This morphotype conforms closely to the C. schencki type material. Together with mt2 it is the most common morphotype of the hova -complex.

Morphotype 5 (mt5) ( Fig. 48 & 53 ). Medium to large size (HW 0.96–1.13, WL 1.01–1.17). As mt4, except as follows. Head sculpture deeply areolate. Colour black.

Comments. The question whether the C. hova- complex consists of one single, highly variable species, or multiple, very closely related species was thoroughly investigated in this study, with molecular data adding an important second line of evidence. Both nuclear and mitochondrial data were inconclusive in regard to the status of the involved taxon or taxa. Phylogenetic analyses of sequence data show two distinct “clusters” among specimens sampled within the complex. These form a clade together with C. sabatra , but in consequence of the exclusion of the latter from the C. hova -complex (see species description of C. sabatra ) are a paraphyletic grouping by themselves. The two clusters further only broadly correspond with the characteristics of the morphotypes, in that mt1 + mt5 group together in one cluster, and mt3 + mt4 group together in the other. Morphotype 2 however appears in both clusters. A geographic pattern, as would be expected if the two clusters represented geographically separated regions with reduced gene flow, also could not be discerned. The underlying cause of this problematic species-complex may be one of incipient speciation where complete reproductive isolation is still lacking between diverging populations within a species, or alternatively one of hybridization between previously well separated species. Cases like this are discovered more and more in ants (e.g. Feldhaar, 2008; Korczynska et al., 2010), and hybridization has already been posited to occur in members of the subgenus Crematogaster (Decacrema) from Asia ( Feldhaar et al., 2010).

Consequently three described species are here placed under the C. hova -complex: C. hova , C. schencki and C. ensifera . The identities of these species have been thoroughly investigated, including examining the type specimens in the MHNG collection. Among the material examined for this study, specimens close to the type localities also closely resemble Forel’s type specimens – material from the vast remainder of the distribution range however presents a variety of intermediate morphological forms. Crematogaster hova latinoda is synonymised under C. hova since its syntype material very closely resembles that of C. hova in all but the width of the petiole and postpetiole, which are – the name implies – wider in C. hova latinoda than in C. hova .

Forel (1887) described C. hova first, with a very detailed species description. When he subsequently described C. schencki ( Forel, 1891, here originally misspelled as C. schenki ) he separated this new species from C. hova solely by size and sculpture – however, he already suspected a strong relatedness between the two species and concluded with the thought C. schencki might only be a “race” of C. hova . Two decades later Forel described C. ensifera as being very close to C. hova and C. schencki , distinguishing it from both by petiole shape and greater spine length, from C. hova by greater size and robustness and from C. schencki by sculpture and colour (Forel, 1910).

In conclusion, Forel’s taxonomic decisions were based upon a set of highly variable morphological characters – and an incomplete sample of specimens from very few localities in Madagascar. Inevitably, extensive sampling and more sophisticated taxonomic methods here have presented a much more complex picture. These advances have certainly been able to better define this taxonomic problem, even though a definite solution on the species level is not attained at this point in time. For the time being the names C. hova , C. schencki and C. ensifera should thus be treated as part of and referred to as the C. hova-complex. An alternative possibility could be to use C. ( hova ) as a “species aggregate” (Article 6.2, ICZN, 1999) for the three names, as has been proposed by Taylor (1989 and 1991) for similar cases in ants.

Distribution and biology. The C. hova -complex has a wide distribution throughout the humid and transition forests of northern, central and southeastern Madagascar. It is generally absent from the western dry deciduous and spiny forests occupied by C. grevei , but present in isolated pockets of humid or transition forest remaining at a few locations in the west, such as in e.g. P.N. Zombitse or P.N. Isalo (see Fig. 60 ). From the very northern tip to the extreme south of the island these ants are a prominent element of the rainforest canopy fauna, inhabiting all altitude levels, with a highest record of 2000m in the Marojejy massif. The C. hova - complex occurs in sympatry with C. malala , C. sabatra , C. sisa , C. mahery and C. nosibeensis , and a narrow sympatry or parapatry exists with C. grevei (see details under description of C. grevei ).

The distribution of the five morphotypes can be characterized as follows. Mt1 is found predominantly in the far north, along the east coast and throughout the above mentioned remnant rainforests in the west of Madagascar ( Fig. 61 ). This morphotype co-occurs mostly with mt2, and only rarely with the other morphotypes. Mt2 occurs over the entire range of the complex, but with an emphasis on the northern and central highland localities ( Fig. 62 ), and further widely co-exists with all other morphotypes. Mt3 has a distinct distribution along the east coast and eastern central highland localities, with isolated distributions in the Sambirano Region of the northwest and in the very far north ( Fig. 63 ). This morphotype co-occurs widely with mt2, but is mostly allopatric with mt4. Mt4 is the most abundant form of the complex and is found throughout the north, the highlands of the Central Region and the southeast of Madagascar, with an isolated occurrence on the west coast ( Fig. 64 ). Mt5 is less common and shows a more patchy distribution ( Fig. 65 ) than the other four morphotypes; it is sympatric with mt4 at some and with mt1 and mt2 at only a few localities.

Within its distribution range, the C. hova -complex is the most abundant and dominant representative of the Malagasy Decacrema . Although these ants possess an arboreal lifestyle, they are nonetheless highly conspicuous on the forest floor when foraging for food and nesting material, and could hardly be missed in an inventory. Fisher et al. collected workers of these ants innumerable times from leaf litter and beating samples, as well as pitfall, yellow pan and malaise traps. Prevailing nesting behaviour of the C. hova -complex however is the construction of arboreal carton nests. I have found these nests in various heights in the canopy: from as low as 1.20 m, attached to vines or branches of young trees, to 20m or more high up in the crowns of canopylevel trees. The sizes of carton nests inhabited by workers, the reproductive queen and brood range from as small as 6 x 4cm in diameter to as large as 25 x 20 cm. Nests housing only workers are frequently encountered and can be smaller in size. It is unclear whether these queenless colonies represent satellite nests of larger nests that are hard to locate, or aggregations of workers whose colony underwent disturbances that resulted in the loss of their reproductive queen and brood. Polydomy in these carton-nesting colonies seems to be quite common, and two or more nests often can be found very close to each other on the same tree, or located on adjacent trees. Experimental or genetic studies would be needed to fully explore details of colony structure. In 17 carton nests out of a total of 30 dissected nests I found one dealate queen together with workers and brood, sometimes with alate queens and males as well. The remaining 13 nests contained only workers, sometimes together with brood. Carton nest colonies further housed staphylinid beetles (6 colonies), and in one case scale insects of an undetermined genus of the family Monophlebidae , possibly Gigantococcus (P.J. Gullan, pers. comm.).

There are some records in the CASC of C. hova -complex workers (in some cases with dealate queens) collected from dead twigs or branches, and in a single instance I have found workers with brood in a dead twig myself. It seems likely that colony founding within the species-complex takes place in dead twigs or branches and the construction of carton nests subsequently commences after a certain colony size is reached, i.e. when the colony can afford to send workers to forage for nest material rather than food. The ultimate domicile may always be the carton nest, since this can support much larger numbers of individuals and can be extended in size as the colony grows.

Etymology. The C. hova -complex is named for the oldest available species name that belongs under the complex, C. hova Forel. The Malagasy word “ hova ” denotes the inhabitants of the region Imerina on the east coast of Madagascar, where the type locality of C. hova lies.

Figures

FIGURES 44 – 48. The Crematogaster (Decacrema) hova - complex; a: full-face view, b: lateral view, c: dorsal view. 44: morphotype 1; 45: morphotype 2; 46: morphotype 3; 47: morphotype 4; 48: morphotype 5.

FIGURES 49 – 53. The Crematogaster (Decacrema) hova - complex; a: petiole & postpetiole in dorsal view, b: petiole & postpetiole in lateral view. 49: morphotype 1; 50: morphotype 2; 51: morphotype 3; 52: morphotype 4; 53: morphotype 5.

FIGURES 58 – 61. Distributions of Malagasy Crematogaster (Decacrema). 58: C. sabatra; 59: C. sisa; 60: C. hovacomplex, entire distribution; 61: C. hova- complex morphotype 1.

FIGURES 62 – 65. Distributions of Malagasy Crematogaster (Decacrema). 62: C. hova- complex morphotype 2; 63: C. hova- complex morphotype 3; 64: C. hova- complex morphotype 4; 65: Crematogaster hova- complex morphotype 5.