Strong and frequent rain is essentialfor the existence of tropical rain for-ests. Heavy showers, however, pose athreat especially for arboreal insectswhich can be washed down fromtrees or even be drowned in theirshelters. Queens of the plant-antgenus Cladomyrma, which must bitenest chambers into their host plantstems for colony foundation, fre-quently disappear during heavy rainand are sometimes drowned whenseeking refuge in their unfinished,still unclosed chambers due to runoffwater which collects in the hollow in-ternodes via the entrance holes. Ma-ture colonies of these ant species facethe same problem of intruding rain-water endangering ant brood and tro-phobiotic mealybugs, but they re-spond with an unusual behaviour onlyonce previously reported for ants: theworkers ingest the liquid and regurgi-tate it to the outside until no standingwater remains. This behaviour can bereadily provoked by injecting smallamounts of liquid into inhabited inter-nodes. The ability of Cladomyrma toremove water from its nest site ap-pears to be an adaptation to their obli-gate colonization of myrmecophytes(ant-plants).Cladomyrma is a formicine genuswhose species live exclusively in avariety of trees and vines from Bor-neo and the Malay Peninsula. To ourpresent knowledge Cladomyrma inha-bits 18 host plant species from 9 gen-era and 8families ([1–4], unpublishedresults). These ants protect their hostplants against herbivores ([4, 5], un-published results) and prune (at leastin part) encroaching vegetation. Theants gnaw open the host plant inter-nodes and hollow them out by remov-ing the pith. With both increasingplant and colony size other internodesare colonized successively. Eventuallythe nest chambers run through stem,branches and every twig of the host.These hollows contain brood and tro-phobiotic pseudococcids which arecultivated by the ants. As a rule onenest entrance per internode is found.In regard to nest structure the hostNeonauclea Merr. (Rubiaceae) differsfrom other host plants of Cladomyr-ma. Here only internodes which areswollen ( = myrmecodomatia) are hol-lowed out and colonized by the ants.The number of entrance holes per do-matium varies greatly, on averageabout 4–6 are found, depending onboth domatium and colony size.In all host plants the entrances are si-tuated more or less at the lateral sidesof the internodes. Thus the risk ofgetting runoff-water into the nestchambers is reduced (exept for thevertical stem). However, rainwatermay still be able to collect in the nest.We dissected longitudinally several in-habited internodes of the trees Sara-ca, Ryparosa, Neonauclea, Cryptero-nia, and the woody climber Spatholo-bus. By watering the outer walls ofthe internodes we could check frominside that small amounts of water dointrude into the nest. Most of the run-off water, however, passes the en-trances due to very small callus ridgesaround the holes. Rain water mayalso penetrate into the nest hollowsthrough openings of damaged twigs,the latter being torn open by predatorsof ants and brood.To test how Cladomyrma ants react tothe intrusion of water we injected be-tween 0.2 and 6 ml artificially co-loured liquid into nest entrances witha syringe. Table 1 summarizes the re-sults. In total we performed 16 testswith nine colonies of three Cladomyr-ma species, C. near andrei (Emery; 1colony, 3 tests), C. maschwitzi Agosti(4 colonies, 6 tests), C. petalae Agosti(4 colonies, 7 tests); and in nine hostplants of five various other species:Saraca thaipingensis Cantley ex Prain(n = 2), Ryparosa fasciculata King(n = 1), Spatholobus bracteolatusPrain (n = 1), Crypteronia griffithiiClarke (n = 4), and Neonauclea gigan-tea (Valeton) Merr. (n = 1). On eachplant we fixed the experimental twigsin a horizontal position to prevent theinjected water from draining away in-side the nest channels. Prior to the in-jection of coloured liquid we fixed asmall filter paper (4 ×4 cm) beneatheach experimental entrance hole(n = 16). Within minutes after flood-ing the internodes (1–15 min, median4, mean 5.38 ± 4.76) the number ofworkers increased noticeably (at leastthree-fold the activity level prior to in-jection) around the affected areas.Shortly afterwards the first workerswith distended gasters, containing co-loured liquid (tested by squashingworkers on filter paper), appeared onthe plant surface (3–65 min, median5, mean 12.08 ± 17.52). There alwayspassed a considerable period of timebefore the first appearance of thewater-ejection behaviour (3–41 min,median 27, mean 24.92 ± 13.18).The results are highly variable andmay be due to (a) different amountused of injected liquid, (b) the num-ber of workers involved, i.e., colonysize, (c) incomparability of risk levels,i.e., different number of larvae orpseudococcids inside the affected in-ternodes, and (d) species-specific dif-ferences in behavioural response.
Read full abstract