RESULTS

The climate was maritime since the colony was located in a disturbed dune ecosystem within 300 meters of the Pacific Ocean. The nest, approximately 0·5 meters in diameter and 0·3 meters in height, consisted of plant debris (broken stems and leaves, seeds) and the remains of various arthropods. Dominant introduced plants surrounding the colony were the European dune grass (Ammophila arenaria (L.) Link) and scotch broom (Cytisus scoparius (L.) Link). Native plants included the coast strawberry (Fragaria chiloensis (L.) Duch), shore pine (Pinus contorta Loud.), beach knotweed (Polygonum paronychia C. & S.) and seashore lupine (Lupinus littoralis Dougl.).

Adult wasps appeared at the nest around the first of May, after the winter storms, and populations built up during the dry summer. Peak activity for the wasps was in July and early August, when up to 30 wasps could be seen hovering over the nest in the afternoon sun. Rates of parasitism in worker ants were highest in July and early August, then began to taper off, although adult wasps continued flying into mid-October before cold and wet weather terminated further activity for the season.

Wasp flight activity was dependent in large part on temperature, which was correlated with the rays of the sun reaching the colony. The wasps rarely appeared before the sun reached the ant nest and would retreat when clouds blocked the sun or the nest was shaded. Thus, in the early morning, evening and on cloudy days, few adult wasps appeared. After the sun reached the nest, wasp activity was high and that is when most worker ants retreated into the nest or to the shaded part of the nest. The adult female wasps would harass the ants at the boundary between sun and shade. Most of the male wasps remained around the edge of the nest, attempting to mate with newly emerged females. However, some males followed females over the nest as the latter were chasing ants.

The female wasps pursued the ants continuously during sunny periods, resting only occasionally for several seconds to several minutes on soil particles or grass stems. During pursuit, the wasps continuously advanced and retreated, always attempting to orientate themselves near the rear of the ant and alight on their abdomens. The ants were quite aware of the wasps and would turn about and lash out at them, occasionally catching them in their mandibles (Fig. 4C). Other workers in the vicinity of the ant being chased would often attempt to catch the wasps. The wasps would sporadically make actual physical contact (strikes) with the ants, but actual ovipositon in the abdomen of an ant was only observed twice. In each case, the wasp actually remained on the ant's abdomen for 1–2 sec. The stung specimens exhibited great agitation and, when later dissected, each contained a wasp egg in early stages of embryonic development. During all of the dissections, only in one case, were two wasp larvae in a single ant and one was dead. Otherwise, all infected ants contained only a single wasp egg or larva.

Wasp developmental stages were always found in the abdomen of the host, never in the thorax or head. The newly deposited eggs were very difficult to find since they were placed just under the body wall of the ants' gaster and hidden by a thin layer of host tissue. Each ovary of the female wasps contained from 30–50 broadly elongate eggs measuring 93–120 μm in length, 14–18 μm in greatest width and tapering at the ends (Fig. 1A). When first deposited, they appeared to adhere to the ants' body wall but became free as embryonic development proceeded. The earliest parasite stage found in an ant was an embryo developing within a trophamnion (a layer of cells derived from the serosal membrane of the egg), which was 345 μm long and 170 μm wide (Figs 1B and 7A). The largest trophamnion encountered was 426 μm long and 216 μm wide (Figs 1C and 7B). The trophamnion contained cells with distinct nuclei and nucleoli (Fig. 1B) and a small appendage at the posterior end (Figs 1A,B and 7A,B). Development of the embryonic head capsule in the trophamnion involved the formation of tissue lobes, beginning with a single lobe which then passed through a 3- and 5-lobed stage, until the first instar head capsule was formed (Figs 1B,C and 7A,B).

Fig. 1. Development of Elasmosoma sp. in Formica obscuriventris clivia. (A) An ovary containing a group of elongate ova. (B) Early embryonic development inside the trophamnion. Note nuclei and nucleoli in trophamnion cells. (C) Later embryonic development. Note lobes of tissue (arrow) forming head capsule.

Two distinct larval instars were commonly encountered, the mandibulate or polypodeiform first instar (Figs 2A,B and 7C) and the typical hymenopteriform third instar (Fig. 3B). Another rare instar was also discovered which did not resemble either the first or third instars (Fig. 3A). This instar is tentatively considered the second instar, making 3 instars in the development of Elasmosoma sp.

Fig. 2. Development of first instar larva of Elasmosoma sp. in Formica obscuriventris clivia. (A) Note large head capsule, ventral body lobes, elongate tail and dorsal anus (arrow). (B) Ventral view of head capsule showing paired, simple, falcate mandibles.

Fig. 3. Development of Elasmosoma sp. in Formica obscuriventris clivia. (A) ‘Unspecialized’ second instar larva. (B) Hymenopteroid third instar larva. Note coiled salivary silk glands filling most of body cavity.

When the first larval instar left the trophamnion, it contained 12 body segments, 11 of which contained paired distinct ventral lobes (Figs 2A and 7C). The head capsule was quite pronounced and the mandibles were large and sickle-shaped (Fig. 2B). There was no indication of eyes, antennae or spiracles. The tail was elongate and provided with small retrose spines, especially noticeable on the ventral surface, near the tip. The well-formed gut contained a single layer of columnar cells and the plugged anus opened on the dorsal side of the body. It was possible to determine if an ant was infected with a first instar larva at the time of dissection from the state of its fat body. In non-parasitized hosts, the fat body was compact, whereas in parasitized workers, the fat body was broken up into small particles.

Second instar wasp larvae, which varied from 1·4 to 1·6 mm in length, were much less specialized than the first instar. The head capsule had been shed and replaced with a greatly reduced oral opening (Fig. 3A). There were no eyes, antennae or spiracles and the body segments were, for the most part, indistinct. The anus appeared terminal. The long tail of the first instar was replaced with a small obtuse terminus containing a few spines.

Third instar larvae, which varied from 1·7 to 1·9 mm in length, possessed a small, but distinct head capsule, but lacked a tail (Figs 3B and 5C). The body, composed of 13 segments, was uniformly covered with microscopic setae. Spiracles, absent from the first and second instars, were present on the first thoracic and first 6 abdominal segments. The paired labial silk glands were enlarged and nearly filled the body cavity in mature larvae, imparting them with a light yellowish colour. This instar was a typical hymenopteriform larva.

Exit of mature third instar larvae from ant hosts usually occurred via the anus (Fig. 4A), but occasionally from between the abdominal segments. At the time of emergence, the host ants were usually dead or dying. The exiting parasite larvae were quite active, quickly moving away from the host and attempting to burrow into the substrate by vertical movements of both the anterior and posterior ends. In some cases, the larvae could not completely exit due to the constricted opening and eventually died. Other worker ants ignored the newly emerged wasp larvae.

Fig. 4. (A) Mature larva of Elasmosoma sp. emerging from the anus of Formica obscuriventris clivia. (B) Mature larva (arrow) of a neoneurine wasp emerging from a worker ant in 40 million-year-old Baltic amber. (C) A major worker of F. obscuriventris clivia holding a female wasp in its mandibles. Note extruded ovipositor of wasp (arrow).

After making a shallow depression in the substrate (sand), the newly emerged larvae would proceed to attach sand grains to the future cocoons. The method by which this was accomplished is as follows. The larva first moved around in a circle, touching its head and emitting a strand of silk to some dozen sand grains, thus uniting them all by a long, single silk strand. The larva then rotated and twisted the string of sand grains around its body. It then proceeded to spin a cocoon from its protected site inside the sand grains (Fig. 5A). Newly emerged larvae that were placed on plastic surfaces did not form a cocoon and although they would enter a ‘prepupal’ stage, they eventually died before reaching the pupal stage.

Fig. 5. Development of Elasmosoma sp. (A) Wasp cocoon with sand grains attached to the outer surface. (B) Same cocoon as in (A) with sand grains removed. Same magnification as (A). (C) Mature larva (left) and ‘prepupae’ (right) removed from cocoons.

All of the cocoons were covered with sand in the observation dishes as well as around the periphery of the ant nest. Some cocoons found in the vicinity of the nest had seeds (of scotch broom), bits of rootlets and soil particles attached to them. Since cocoons found inside the ant nest were also surrounded by sand grains, and the nest was composed mostly of plant debris, they were probably formed in the sandy area around the nest and carried in by worker ants. The cocoons were white, smooth, and oval with a small constriction at one end, and measured 2·1–2·3 mm in length (Fig. 5B). Under natural conditions, most cocoons were found in the upper 1 cm of sand within a radius of 500 cm from the nest.

After forming the cocoon, the larva entered what appeared to be a prepupal stage. This stage, identified by the swollen thoracic segments, lasted for at least 48 h before the pupal stage appeared (Figs 5C and 6A). The developing eyes of the adult wasps could often be seen in the older prepupal stage and always in the early pupal stage (Fig. 6A). At the time of emergence from the cocoon (Fig. 6B), the adult wasp used its mandibles to cut a neat circular opening at one end of the cocoon. The cap was completely cut away.

Fig. 6. Development of Elasmosoma sp. (A) Cocoon cut open to reveal developing pupa. Note dark eye spot (arrow) and remains of meconium smeared on abdominal segments. (B) Male adult wasp emerging from its cocoon (cocoon was cut open with a micro-scissors). (C) Adult female wasp. Note extruded ovipositor (arrow).

It was difficult to determine the actual rate of parasitism in the colony, since the percentage of infected ants on or around the colony varied over a 12-h period (Table 1). In July and August, when the wasps were especially abundant, ants harbouring third stage wasp larvae would gather along the edge of the nest in early morning. These parasitized ants received food by returning foragers. This collection of parasitized ants, which facilitated collection, was termed a ‘host assembly’. Apparently a shift in ant behaviour occurred when the parasites reached the third stage. At this time, the ants halted normal activity and became part of the ‘host assembly’. After receiving nourishment, these infected ants either retreated into the nest or wandered off by themselves, the latter possibly indicating that the parasites were ready to emerge. The average developmental time from oviposition to adult emergence was estimated to be 5–6 weeks, which would indicate that some 4–5 generations occurred per summer.

Table 1. Parasitism of Formica workers by Elasmosoma sp. during the late summer of 2003 (All samples taken at exactly the same location adjacent to the ant nest.)

During the course of this study, no other internal parasites were found, nor were any other types of wasps noticed flying around the nest. Occasionally some dauer stages of diplogasterid nematodes were recovered from the head glands of the ants, but they were uncommon and their effect on the ant appeared to be negligible. Some workers had their body cavities filled with iridescent fat globules, a condition suggesting the presence of an iridescent virus.