Crematogaster pygmaea (Hymenoptera: Formicidae: Myrmicinae), a highly polygynous and polydomous Crematogaster from northeastern Brazil Author QUINET, Y. Author HAMIDI, R. Author RUIZ-GONZALEZ, M. X. Author de BISEAU, J. - C. Author LONGINO, J. T. text Zootaxa 2009 2075 45 54 journal article 22778 Crematogaster pygmaea Revised Status Fig. 1 (A -E) Crematogaster pygmaea Forel, 1904:37 . Syntype worker: Brazil, Ceara (Diaz da Rocha) [MHNG] (examined). Emery, Accepted by J. Pitts: 19 Mar. 2009; published: 17 Apr. 2009 45: 135: combination in C. (Orthocrema), Forel, 1909:259 : variety of brevispinosa . Forel, 1912:235 ; Emery, 1922:135 : revived status as species. Crematogaster abstinens Forel : Longino 2003:30 (incorrect synonymy). Crematogaster sp. prox. abstinens : Heredia et al. 2005 . Taxonomic conclusions are based on abundant collections and natural history observations from Fortaleza, State of Ceara , Brazil. These collections were compared directly with the types of C. pygmaea . These are the only collections of C. pygmaea known to date. Crematogaster pygmaea is very close to the widespread C. abstinens . Both species occur sympatrically in Fortaleza, where they are clearly distinct. FIGURE 1. Crematogaster pygmaea worker: (A) face view, (B) lateral view; queen: (C) face view, (D) lateral view. Imaged worker and queen from one colony, Fortaleza, unique pin code JTLC000003537. E. comparison of worker mesosoma of C. pygmaea (left) and C. abstinens (right). Note shorter propodeal spine and shinier lateral pronotum of C. pygmaea . Crematogaster pygmaea and C. abstinens are uniquely characterized by the combination in the worker of (1) shiny face, (2) subquadrate dorsal face of petiole, (3) appressed tibial pilosity, (4) very wide postpetiole, (5) abundant short stiff setae on face that curve toward the median axis, and (6) complete absence of an anteroventral petiolar tooth. Crematogaster pygmaea differs from C. abstinens in shorter propodeal spines. The propodeal spines of C. pygmaea are relatively triangular in shape, not spiniform, and slightly longer than the maximum diameter of the propodeal spiracle. The propodeal spines of C. abstinens are spiniform and about twice as long as the maximum diameter of the propodeal spiracle (Fig. 1G). Another difference in the zone of sympatry is that C. abstinens has a more heavily sculptured promesonotum, particularly near the juncture of dorsal and lateral surfaces. However, in other parts of the range of C. abstinens this character is variable and approaches the condition seen in C. pygmaea . Crematogaster obscurata is also similar to these species but has a punctate face. Measurements of a worker and a queen were made with a micrometer stage with accuracy to the nearest 0.01 mm. Worker: HL (head length; perpendicular distance from line tangent to rearmost points of vertex margin to line tangent to anteriormost projections of clypeus, in full face view) 0.565, HW (head width; maximum width of head in face view, including eyes if they project beyond the sides of the head) 0.604, SL (scape length; length of scape shaft from apex to basal flange, not including basal condyle and neck) 0.436, EL (eye length, measured along maximum diameter) 0.174, WL (Weber's length; viewing mesosoma in lateral profile, distance from approximate inflection point, where downward sloping pronotum curves into anteriorly projecting neck, to posteroventral propodeal lobes) 0.609, SPL (propodeal spine length; measured from tip of propodeal spine to closest point on outer rim of propodeal spiracle) 0.076, PTH (petiole height; viewed in lateral profile, perpendicular distance from ventral margin to highest point of posterolateral tubercles) 0.124, PTL (petiole length; viewed in lateral profile and measured in same plane as anterodorsal face, distance from inflection point marking juncture of posterolateral lobes and cylindrical posterior portion of segment to anterior inflection point where petiole curves up to condyle) 0.217, PTW (petiole width; maximum width of petiole in dorsal view) 0.217, PPL (postpetiole length; viewed dorsally, perpendicular distance from narrowest point of peduncle joining postpetiolar node and helcium, to line tangent to posteriormost lobes) 0.155, PPW (postpetiole width; maximum width of postpetiole, in same view as and perpendicular to postpetiole length) 0.201. Queen (previously undescribed): HL 0.998, HW 1.330, SL 0.671, EL 0.364, WL 2.402, PTH 0.399, PTL 0.538, PTW 0.574, PPL 0.431, PPW 0.643. Natural History Observations The state of Ceara belongs to the "caatinga" domain, a 750,000 km2 area of northeastern Brazil with a semi-arid climate. Temperature varies little, with an annual average of approximately 26oC, and rainfall is low (less than 750 mm/year). Rainfall is concentrated in three consecutive months during the southern hemisphere summer (November until June). Crematogaster pygmaea occurs in two phytogeographic zones: the littoral zone stricto sensu and the adjacent savana-like formation called "tabuleiro." The first is a narrow strip of coastal vegetation along the seashore, with dune and mangrove formations. The tabuleiro is a savanna-like formation that occurs close to the coast, on flat sandy plains of northeastern Brazil. The physiognomy of the "tabuleiro" is characterized by dense patches of trees and shrubs surrounded by a grassy cover with scattered low shrubs ( Oliveira-Filho 1993 ). All the observed C. pygmaea colonies were found in these two zones. Colonies could be found very near the sea, in completely sandy areas with only grassy cover. Crematogaster pygmaea was never found in the "caatinga" zone, a seasonal xerophilous thorn woodland/shrubland that prevails on the semi-arid lowlands and covers most of Ceara state. Detailed observations were made of four C. pygmaea colonies (I, II and III, IV) at sites situated on or near the campus of the State University of Ceara (3o 47' S - 38o 33' W), in Fortaleza (state of Ceara , northeastern Brazil), about 7 km from the coast. The campus itself is located in the tabuleiro zone. In the study sites the original vegetation was modified by human activities, being in the campus and Fortaleza urban areas. Colony I was located on the campus, in a sandy/clayey area with patches of mango or cashew trees surrounded by dense herbaceous cover. Very close to colony I and physically separated from it by an asphalt road, colony II was in a more open area. The area had rows of square beds with fodder plants (see Fig. 2) separated by sandy/clayey areas covered with sparse herbaceous vegetation. Patches of mango trees, sugar cane and dense herbaceous vegetation surrounded it. Colony III was located on the campus, some 800m from colonies I and II, in an open sandy/clayey area with only herbaceous cover. Colony IV was located outside, but close to the campus, in an open area covered with dense herbaceous vegetation and patches of low shrubs surrounding sports (football) grounds. Detailed maps of colonies I and II were made in July 1999 and October 2006, respectively. To facilitate mapping of colony I, a grid of 0.5m squares was marked out on the whole colony area, using string. The rows of square beds served as references to map colony II (see Fig. 2). All the nest entrances and trails of the two colonies, as well as the plants visited by C. pygmaea workers for food sources, were precisely mapped. Each plant visited by C. pygmaea workers was carefully inspected to identify the food sources explored by the ants. Colony I extended over an area about 26 by 10m (Fig. 2). It consisted of 36 nest entrances connected by nearly 104m of trails that also led the workers to plants where they explored liquid sugary food sources (nectar or honeydew). One hundred herbaceous plants were visited by ants. Nearly 50 % of them (N = 46) were Borreria verticillata G. Mey. plants ( Rubiaceae ) whose floral nectaries were explored by C. pygmaea workers. Some B. verticillata plants also had scale insects ( Coccidae ) colonies attended by ants for their honeydew production. The second and third most visited plants (N = 27 and 12 respectively) were two unknown Cyperaceae species where C. pygmaea workers attended scale insect colonies. On the remaining plants (N = 15) ( Commelina sp. - Commelinaceae ; Mimosa sensitiva L. - Mimosaceae ; Turnera subulata Smith - Turneraceae ; one unknown species of Papilionaceae ; one unknown species of Cyperaceae ; one species of unknown family) ants explored sugary secretions from scale insects, aphids ( Aphididae ), and/or extrafloral nectaries. Colony II had roughly the same size (about 30 by 8m) and consisted of 62 nest entrances connected by nearly 75m of trail network (Fig. 2). More than 50% of the 97 herbaceous plants visited by ants (N = 54) were T. subulata plants (white alder), a common ruderal plant in northeastern Brazil. The petiole of T. s ubu la ta leaves has two large apical extrafloral nectaries that were explored by C. pygmaea workers. The second most visited plant was B. verticillata (N = 27) whose floral nectaries and, sometimes, aphids or scale insect colonies were explored by ants. The remaining plants belonged to Asteraceae , Euphorbiaceae , Fabaceae , Plumbaginaceae and Poaceae families and had scale insect colonies attended by ants. In each colony, one or two groups of nest entrances were disconnected from the main group (Fig. 2). They probably represented colony disjunctions, since preliminary observations showed that the trail network of a C. pygmaea colony is an unstable structure that undergoes size and shape modifications with time. For example, during the dry season the trail network undergoes a great size reduction. In order to investigate nest architecture and composition, 72 C. pygmaea nests were excavated: 20 from colony I, in June and July 2002; 37 from colony II, in June 2003; and 25 from colony IV, in May 2006. For the nests of colony I and the first ten nest excavations of colony II, only queen number and brood presence (colony I), or number and depth of chambers per nest were recorded (colony II). More detailed observations were made with the remaining nests of colony II and those of colony IV: nest depth, number of chambers per nest, depth and size of chambers, number of queens per chamber and presence of brood in chambers. Nest excavation was carried out by digging a deep soil parallelepiped (20 x 20 x 40 cm), with the nest entrance in its center. The whole parallelepiped was carefully laid down on a plastic sheet, and then cut into thin slices with a spatula, starting from the side with the nest entrance. Depth and, when possible, size of each chamber was recorded, as well as the queen number and presence of brood. Thirteen of the 34 active nests (i.e. with at least workers) excavated in June 2003 were brought to the laboratory to count the queens and workers. Finally, the diameter of 52 nest entrances from 4 colonies (colony I, II, I V, and a fifth one) was measured in March 2003 and April 2008. Eleven to 40% of the excavated nests were inactive (Table 1), meaning that there was neither gallery nor chamber below the nest entrance, in spite of workers entering and exiting the nest entrance. Of the remaining, active, nests (i.e. with gallery, chambers, and at least workers in the chambers), almost all (92-100%) contained brood, and 47-71% contained at least one queen (Table 1). Colonies were clearly polygynous (Table 1), with some nests containing more than 10 queens. FIGURE 2. Map of two Crematogaster pygmaea colonies (colony I and II), showing the location of all nest entrances, trails and plants visited by ants for food sources (honeydew and/or nectar). TABLE 1. Number of queens per nest in three Crematogaster pygmaea colonies. Active nest: nest entrance with gallery, chamber(s) and at least workers in the chamber(s).

Number of excavated nest	Number (% of active nests	) Number of q active nest Mean + SD	ueens per Range	Proportion (%) of active nests with queen(s)	Proportion (%) of active nests with brood	Nests excavation period
Colony I	20	15 (75)	3.2 + 3.2	0-9	66.7	100	25.06-08.07.2002
Colony II	27	24 (89.8)	3.9 + 4.3	0-15	70.8	91.7	20-22.06.2003
Colony IV	25	15 (60)	0.7 + 0.9	0-3	46.6	93.3	22.05.2006

Nests had a maximum depth of nearly 30cm and each of them was formed by a single straight vertical gallery with 1 to 4 horizontal chambers (Table 2). The depth of the first chamber varied depending on the number of chambers. When there were multiple chambers, the first chamber tended to be shallower (Table 3). Chambers had a circular, or oval, shape with a mean thickness of 4.9 + 0.9 mm (N = 16) and a diameter of roughly 3 cm. Galleries always entered the center of a chamber from above and continued from the center of the chamber floor. Nest entrance had a mean diameter of 1.89 + 0.15 mm (N = 52) and the gallery had exactly the same diameter. It is therefore possible to describe the typical structure of a C. pygmaea nest as a straight vertical gallery, 20 cm in length and 1.9 mm in diameter, with a circular entrance 1.9 mm in diameter, and generally 3 circular or oval chambers that are 5 mm thick and 3 cm in diameter, giving a 6.9 cm3 total nest volume. Nests were measured in the rainy season, when the soil was humid. During the dry season, the soil was very hard, making it almost impossible to dig. However, a few attempts showed that the galleries and chambers were much deeper in the dry season. TABLE 2. Number of chambers per nest in two Crematogaster pygmaea colonies.

Proportion (%) of nests with: 1 chamber 2 chambers	3 chambers	4 chambers	Number of inspected nests
Colony II	11.8 35.3	50	2.9	34
Colony IV	6.7 33.3	60	0	15

TABLE 3. Mean depth (cm) of chambers in nests of two Crematogaster pygmaea colonies. N: number of nests with 1, 2, 3 or 4 chambers; Ch.: chamber.

Colony II
Number of chambers	N	Ch. 1	Ch. 2	Ch. 3	Ch. 4
1 chamber	4	5.4 + 3.8
2 chambers	12	8.5 + 2.4	16 + 3.5
3 chambers	17	3 + 1.3	8.5 + 2.7	16.1 + 2.2
4 chambers	1	1.5	5	10	14
Colony IV
Number of chambers	N	Ch. 1	Ch. 2	Ch. 3
1 chamber	1	5
2 chambers	5	6.6 + 3.1	14.4 + 4.1
3 chambers	9	4.8 + 1.3	9.8 + 2.3	22.8 + 5.3

Fifty queens and 4644 workers were found in the 13 nests analyzed in the laboratory, giving a queen/ worker ratio of nearly 1:100. Foraging activity of C. pygmaea was recorded during four 24 hours periods: the first in February 2000, the second in December 2000, the third in October 2001, and the fourth in September 2006. In each of the first three 24 hour recording periods, all carried out with colony I, activity was recorded at two nest entrances and on one foraging trail. In the fourth one, carried out with colony II, activity was recorded at three nest entrances and on three foraging trails. During a 24 hour period, all ants crossing a point on the selected trail(s), and all ants exiting or entering the selected nest entrances were recorded with a click counter during a 5 minute session, at 1-h intervals. Soil temperature was monitored at each recording point, for each activity measurement. C. pygmaea forages day and night (Fig. 3). However, foraging activity is temperature dependent since it stopped between 10:00 a.m. and midday, when soil temperature exceeded 40oC (Fig. 3). A slight reduction of foraging activity was also observed just before dawn, when soil temperature dropped below 30oC. Foraging activity tended to peak at sunrise, when soil temperature began to rise after low night temperatures (Fig. 3). Foraging activity is high between 24 and 32°C, above which it drops (Fig. 4). Reproductive phenology of C. pygmaea was investigated by excavating about ten nests of colony II and III each month, from March 2005 to March 2006. The number of queens, gynes and males found in each nest was recorded. Rainfall data were obtained from the Ceara state fundation of meteorology (FUNECE). The mean number of queens per nest was nearly constant (+ 2 queens/nest) during most part of the year (Fig. 5). At the beginning of the rainy season (January to March), it rose to nearly six queens per nest (Fig. 5). Gynes were observed at the beginning of the rainy season, with a mean number reaching up to 17 gynes per nest in February (Fig. 5). Males production began in the dry season, some three months before gyne production, and continued during the rainy season, reaching mean number of up to 9 per nest (Fig. 5). FIGURE 3. Daily cycle of activity of a Crematogaster pygmaea colony (colony II). All ants crossing a point on three trails (A) and exiting/entering three nest entrances (B) were counted during 5 minutes every hour, during a 24 h period (30.09-01.10.2006). Results were similar for the other three recording periods. Discussion The presence of several reproductive females in the same nest (polygyny) is frequently observed in ants, and in most cases, nests of polygynous species contain no more than ten queens ( Keller, 1993 ; Crozier & Pamilo, 1996 ). However, some species have a highly polygynous social structure characterized by a queen/worker ratio between 1:200 and 1:50. Moreover, most of the highly polygynous species so far studied are invasive/ tramp species that are also characterized by lack of hostility between colonies (unicoloniality), like in Linepithema humile , Monomorium pharaonis , Tapinoma melanocephalum , or Solenopsis invicta ( Passera, 1994 ; Tsutsui & Suarez, 2003 ). FIGURE 4. Relationship between soil surface temperature and ant activity measured on three trails and at three nest entrances of a Crematogaster pygmaea colony (colony II). Ant activity on the three trails and at the three nest entrances was combined. FIGURE 5. Reproductive phenology of Crematogaster pygmaea . The mean number of queens, gynes and males per nest was obtained by excavating about ten nests of colony II and III each month, during one year. Data for rainfall were obtained from FUNCEME ( Fundacao Cearense de Meteorologia e Recursos Hidricos ). This contrasts with the situation observed in C. pygmaea whose high polygynous social struture is not associated with invasive behavior. Moreover, field observations have shown that workers from different colonies interact in a hostile way (Hamidi et al., in prep.). High polygyny associated with non-invasive behavior and territoriality has also been found in Dolichoderus mariae ( Laskis & Tschinkel, 2008 ) , a North American species that shares many traits with C. pygmaea . Both are ground-dwelling species with highly polygynous and polydomous colonies whose nests are connected by an above ground trail network, and they heavily rely on sugared solutions collected from hemipteran colonies and/or nectaries. Moreover, in both species, nest location seems to be strongly related to the food source location and there is marked seasonal polydomy, with colony retraction during winter ( D. mariae ) or dry season ( C. pygmaea ) and colony expansion during spring-summer ( D. mariae ) or rainy season ( C. pygmaea ). It has been suggested that the seasonal polydomy of D. mariae represents a response to the seasonal fluctuations of food source location and abundance , allowing the ants to closely track hemipteran populations ( Laskis & Tschinkel, 2008 ). The polydomy of C. pygmaea could also represent an adaptation to exploit dispersed and rapidly changing food sources (nectaries and hemipteran colonies). The main elements of C. pygmaea nests are consistent with the nest architecture generally observed in ant species that excavate nests in soil, i.e. vertical tunnels connecting horizontal chambers ( Tschinkel, 2003 ). However, compared to other ant species with subterranean nests, C. pygmaea nests are small and very simple, the average nest being no more than a single 20 cm straight vertical gallery connecting three small horizontal chambers, at least during the rainy season. Most ant species whose subterranean nest architecture has been investigated have much more complex nests, like in Pogonomyrmex badius , Solenopsis invicta , Prenolepis imparis or Formica pallidifulva ( Tschinkel, 2003 ) . Pogonomyrmex badius , whose deep nests (up to 3.5 m) may contain four or five vertical tunnels connecting up to 150 chambers, is one of the best known examples of such complex nests ( Tschinkel, 2004 ). The smaller and shallower nests of Formica pallidifulva with its multiple branching tunnels are another example ( Mikheyev & Tschinkel, 2004 ). Although formed by a single shallow conical chamber, the nests of Dolichoderus mariae are much larger, with a 930 cm3 mean volume ( Laskis & Tschinkel, 2008 ), than those of C. pygmaea that are only about 7 cm3 in volume. The simple architecture of C. pygmaea nests could be related to the need to have a flexible colony structure, with nests and trail networks able to track food sources that frequently change location. In conclusion, the high polygyny found in C. pygmaea , combined with other prominent features (noninvasive behavior, high polydomy, strong worker/queen dimorphism, ground-nesting habits, simple nest architecture), make this species a particularly interesting model to investigate the selective pressures that allowed the evolution and the maintenance of a social structure in which workers rear genetically distant brood.