Collins New Naturalist Library. Philip Chapman

carried on a sinking stream, tree trunks hurled into a pothole by violent floods, or well-rotted humus quietly inched down the cracks of a limestone pavement. Usually by the time such materials reach the depths of the cave where the true cave faunas live, they have been thoroughly pulverised by water and rock, tenderised by bacteria and fungi, and often enough, passed through a series of invertebrate guts.

      This was brought home to me during my first visit into the higher levels of GB Cave in the Mendip Hills, several years ago. The upper levels of the cave contain a series of narrow rifts which leak water from the overlying land whenever it rains. I was hunting for tiny cave beasts in these upper levels, and looking out for patches of fresh mud which I guessed would be organically richer and so would contain more life. I soon began to realize that each and every sediment cone emanating from each and every leaky crack was made up of thousands of tiny mud pellets, like scaled-down grains of rice. They were arthropod droppings – hundreds of millions of them, forming a deposit which over the centuries was gradually filling up the upper dry passages, and probably the lower reaches beneath the water table too. Fortunately, most cavers are blissfully unaware of the true nature of the substance which they spend each weekend crawling through and, no doubt, liberally ingesting themselves.

      Biologists used to studying ecosystems based on living plants too often treat detritivores – animals which eat dead organic material – as a single dietary category. In the cave ecosystem, all the first-level consumers are ‘detritivores’ and it is surprising just how many different specializations they manage. This is most clearly seen in guano communities, where chains of very specialized organisms use different components of the insect remains which form the bulk of bat droppings. Stuart Hill has studied the ecology of one such community in a bat cave in Trinidad. Guano of the funnel-eared bat, Natalus tumidirostris (notorious, incidentally, as a carrier of the human diseases, relapsing fever and blastomycosis), was eaten as soon as it fell by a cockroach, Eublabarus distanti, which removed most of its fat and some of its protein (much of the unbound protein having been already stripped out in the gut of the bat). The cockroach droppings, still rich in chitin, were decomposed by a fungus, Penicillium janthinellum. This in turn was fed on by the mite Rostrozetes foveolatus and various other tiny arthropods. The mites in turn were eaten by other arachnid predators. Similar sorts of food chains probably operate in British bat caves, but the research has not yet been done.

      In our present ignorance, we really do not know exactly what many of our British cave detritivores get out of their ‘junk food’ diets. There must be some degree of specialization, because certain types of materials are largely ignored by certain cave animals, while others attract them strongly.

      One way to study dietary specialization is to put down baits in the cave and see what species come to them. The baits should be very small or they will provide an un-natural boost to the numbers of certain species and so upset the balance of the cave community.

      I tried a baiting programme in the Ogof Ffynnon Ddu system in Wales, during the course of a wider survey of the fauna. I already knew, from a number of earlier collecting visits, that at least six species of fly were common in the upper level passages (a dung fly, a coffin fly, a winter gnat, and three or more fungus gnats), together with eleven springtail species, a millipede, a woodlouse and various beetles and mites. All were associated with detritus in one form or another, so when I put down small pieces of raw liver or cheese in various choice spots, I expected to gather quite a collection of invertebrates. Over the next couple of weeks I came back regularly to check the baits and was disappointed to find nothing on them. By now the liver was beginning to get a bit unsavoury and the cheese had gone slimy. Two and a half weeks after placement, I found every single liver and cheese bait crawling with the slender maggots of the winter gnat, Trichocera maculipennis, but still nothing else came near them. Eventually, the larvae crawled away to pupate, the bacterial smell subsided, and only then did a few Folsomia springtails gather in the area. So, far from being unspecialised scavengers, these detritivores showed themselves to be quite a discerning bunch.

      One of the key factors in determining how the cave fauna will respond to a particular source of detritus seems to be whether it is first attacked by fungi or bacteria. Many fungi exclude bacteria from their chosen food by secreting antibiotics, and avoid other foods of the wrong pH which are decomposed by bacteria. Sometimes the dominance of either one or the other is determined by the size of the potential food source. In the course of a study of lava cave invertebrates on Kilauea volcano in Hawaii, I put out baits of supermarket white bread, to see what would be attracted. Some baits were in the form of crumbs spread along the rough cave wall, others were big chunks which I kneaded into golf-ball-sized lumps. The crumbs quickly went soggy and grew a thin gelatinous slime of yeast-like fungi which attracted cave-specialised millipedes and crickets; while the chunks became bacterial stink bombs infested by scuttle-fly maggots, but shunned by all the true cave fauna.

      Several European biospeleologists have noticed a similar specialization between bacterial- and fungal-feeders. In general it seems that ‘surface grazing’ arthropods, such as millipedes, isopods and springtails, tend to be fungus-eaters; while ‘gulpers’ and burrowers, such as fly larvae and earthworms tend to be bacterial feeders. However, even closely related species within the same group can specialize to different foods to avoid competition. An example is found in the springtails Tomocerus minor and its congener T. problematicus, which co-exist in similar habitats in the Grotte de Sainte-Catherine, in the Ariège region of France. The former mainly munches fungi, while the latter feeds largely on bacteria-rich silt.

      In general it seems that microfungi, which occur on a variety of substrates, including wood, animal faeces and plant detritus, form the most important dietary component for most terrestrial cavernicoles. In a study of several Virginia caves, Dickson and Kirk (1976) found that the abundance of the cavernicolous invertebrates was correlated with the abundance of microfungi and with high fungal-bacterial ratios, but not with abundance of bacteria or actinomycetes.

      Friedrich, Smart and Hobbs (1982) summarized the literature on bacterial counts for cave sediments and waters. As might be expected, heterotrophic bacteria (which feed on inwashed organic material) are far more numerous than autotrophic bacteria (which utilise the oxidation of inorganic compounds) – but there are very wide variations. Sediments give consistently higher bacterial counts, generally of the order of 10 x 10⁶ g^-1, but with large variations either side; while cave waters give very much lower counts, generally of the order of 100 ml^-1. These authors’ own figures for different types of water inputs in Mendip caves are interesting. Swallets feeding directly into open cave passages give the highest figures (500 ml^-1). Percolation recharge which is integrated into mesocavernous fissures and conduits has an intermediate bacterial count (50–300 ml^-1), while diffuse flow waters in phreatic microcaverns (sampled via boreholes) give a very low count (2 ml^-1). One of their conclusions is that water flow through the limestone aquifers feeding the major springs used as drinking supplies around Mendip does not provide any significant filtration of microbial input. While bacterial counts at the major Mendip risings in general do not give cause for public health concern, this may not be the case in other springs, particularly in tropical countries, used as an untreated drinking supply on the supposition that water which bubbles out of the ground must be pure. Sinkholes are widely used as rubbish dumps, attracting disease-carrying rodents, and the water which enters them may reappear many miles away as a spring, perhaps in a different valley. The late Dr Oliver Lloyd, a well-known Mendip caver, contracted Weil’s Disease from Leptospira bacteria present in Longwood Cave on Mendip, while several members of expeditions to Borneo in 1980 and 1984 (myself included) contracted a similar form of leptospirosis in the famous Mulu caves, and nearly died as a result.

      In slow-moving underground streams with mud bottoms, and in drip- and seep-fed pools isolated from swallet-fed streamways, microfungi are more common than bacteria and are correlated with the abundance of macroscopic cavernicoles. Microfungi and bacteria may be utilized directly by such cavernicoles, or may be eaten by Protozoa, nematodes and other micro-fauna which are in turn eaten by Crustacea (isopods and amphipods) and aquatic insect larvae.

      I have tried so far to paint a picture of the cave community as a fairly structured world, where, despite the absence of green plants, the lowly animals