Herbert Meier & Ingo Schaefer to confirm this idea (Meier & Schaefer 2003a, b). The type of illumination, and UV light in particular, therefore only play subordinate roles. A contrasting point of view was promoted by Pawley (1969) who described the positive effects of UV radiation on a Chelus fimbriata. His paper unfortunately leaves a detailed description of the husbandry conditions largely wanting, but as all his turtles only lived to ages of one or two years at best, it is safe to assume that they were far from adequate. The turtles leave the water only for the purpose of oviposition. Basking on the water’s surface is also unknown from Chelus fimbriata (Meier & Schaefer 2003a, b). A natural murkiness of the water and partly very dense vegetation on the banks largely prevent solar radiation from reaching into the water. Even if the water is crystal clear, the shade provided by vegetation eliminates all visibility from a depth of about 1.5 m and creates an atmosphere of duskiness between 1 and 2 m deep even in bright daylight. Assuming that the turtles have adapted to their natural environment, their diurnal activity in a well-structured and dimly lit aquaterrarium is easily explained. Their main activity periods in captivity fall into the hours of early morning and late afternoon to evening hours; they spend the night sleeping. Until 2006, the matamatas of the senior author’s were housed together with some river turtles, Podocnemis spp. (see the following chapter). For their benefit, a localized basking site was brightly illuminated with two 150 W HQI floodlights at a distance of 45 cm from the water surface. This installation now also provided some light to the spot under the suspended terrestrial part that had previously served as a hiding place to the matamatas. As a consequence, the turtles now largely avoided it, but rather sheltered between the logs lying in the water or beneath a planted floating island. On the other hand, Chelus fimbriata are usually found late in the day at advanced dusk during the dry season. Their resting places are then typically situated in sections that are exposed to sunlight, however (own obs.). Joint housing The matamatas were kept in the company of 2.1 Northern Australian snake-necked turtles, Macrochelodina rugosa, and four semiadult Orinoco river turtles, Podocnemis vogli, until December of 2002. Shortly before Christmas of 2002, the Podocnemis vogli were replaced with 3.2 adult Terekay river turtles, Podocnemis unifilis. The artificial pond is also used to temporarily accommodate individual females of the red-bellied short-necked turtle, Emydura subglobosa subglobosa, which are ready to lay their eggs in the terrestrial section. While some large goldfish lived in this basin earlier, these were replaced with mouthbrooding cichlids (Tilapia sp.) in 2001. Owing to their method of parental care, they manage to reproduce in spite of the presence of turtles. The population of these fish virtually exploded with their first filial generation, but the number of mature fish has by now stabilized at around ten. Young fish rarely live to grow larger than 3 cm as they represent a popular source of live food for both the snake-necked turtles and the matamatas. Joining the Podocnemis unifilis with the matamatas turned out to be a grave mistake. While Chelus fimbriata ranks amongst the laid-back turtles despite all activity, Podocnemis unifilis are typically lively (except during periods spent basking). Actually, the former did not show signs of being stressed, continued to feed, and did not appear to hide more strictly than normal. However, the extent of their courting and mating activities decreased notably. They still laid eggs during their first year in the company of the Podocnemis unifilis, but there was no second clutch, and fertilization rates were greatly reduced as compared to earlier times. In 2004, only one female produced a clutch that proved infertile, and no oviposition at all took place in 2005. Considering them a potential cause of the problems, the Podocne- 22 RADIATA 17 (3), 2008
The Matamata, Chelus fimbriata mis unifilis were removed in January of 2006. The matamatas responded with producing a clutch in the same year, even though the eggs were dropped in the water. They were also insufficiently shelled, being encased in a parchment-like, very soft membrane only; it was not even tried to incubate them. It then took until 2007 before two females each produced one regular clutch of eggs, but both proved infertile. This year, mating activity has increased substantially by comparison to the previous years, so that there is hope for 2008 producing fertile clutches once more. It is particularly promising that a new male has now taken an intense interest in the third female, which has so far been largely ignored. Temperatures Recommended water temperatures are often given as 28-30 °C (comp. Meier & Schaefer 2003a, b). This range is not generally off the mark, but data gathered from naturally inhabited waters usually showed a much lower temperature range of only 24-26 °C, measured at depths between 30 and 100 cm in slowly flowing to stagnant bodies of water. While no differences were measurable at various depths in moderate to strong currents, temperatures at one metre deep in stagnant water were often less than 24 °C, in some cases even as low as around 22 °C. Klingelhöffer & Scherpner (1959) stated that matamatas would continue to feed even at 17 °C. Although the animals will tolerate these low temperatures for a few hours, they should not be forced to exist at water temperatures of less than 22 °C, and values of 24-25 °C can be regarded as optimal (Meier & Schaefer 2003a, b). Heating the water of the tank by means of an underfloor heating system, which has proven very effective in other set-ups, caused severe problems with the matamatas. With the turtles spending long periods of time lying on the ground, this type of heating created a perfect microclimate for bacteria between the bottom of the tank and the plastron of the animals. It resulted in shell necroses that healed only very slowly. It took a switchover to other means of heating to reduce this problem to insignificance (Meier & Schaefer 2003a, b). The senior author has meanwhile been experimenting with higher water temperatures of 28 to a maximum of 30 °C. As was to be expected, these values accelerated the turtles' metabolism, which became particularly obvious from increased activity prior to feeding. Long before feeding was due, the turtles could be seen searching for food, and the population of company fish was substantially reduced in the process. It did not lead to an increase in reproduction-related activity, however. Rather, it saw the old problem of shell necroses revert to the extent that had been experienced during the time when underfloor heating was used. Temperatures were therefore once more returned to 24-25 °C. The water of rivers naturally inhabited by Chelus fimbriata is usually rather acidic, soft, and nutrient-deficient. Krause (1998) indicated pH-values of 3.7-6.1 for various rivers in the Amazon region, and measurements around pH 7-8 exist for the Río Pastaza (Ecuador). We were unfortunately unable to add data from our excursion because the pH meter we were carrying with us in a backpack was damaged beyond repair. As to how far water quality plays a role for adult Chelus fimbriata can only be speculated about. What is a fact, though, is that the turtles will sometimes reduce their food intake if their water has not been renewed for a longer period of time. This is particularly obvious in juveniles: three juvenile specimens of carapace lengths of about 7 cm were accommodated together in an 80 cm-tank with a water level of 12 cm. They were not used to feeding on unanimated food for which reason there were always some live topmouth gudgeons (Pseudorasbora parva) or sunbleak (Leucaspius delineatus) present in their tank. Their water was exchanged once a week and kept reasonably clean by means of a small internal filter. Fol- RADIATA 17 (3), 2008 23
