In general, deeper water cavities had a positive significant effect on the occupancy of E. mysteriosus . The need of water for amphibian survival makes them dependent on water availability. The depth of the bromeliads' water cavities were positively correlated with the width of the bromeliads, consistent with Fragoso and Rojas-Fernández (1996). Additionally, Zotz and Thomas (1999) report that larger bromeliads contain their water and humidity for longer periods than smaller bromeliads does in Panama. However, juveniles were not more frequent in bromeliads with deep water cavities in Bromeliad Species 1. This may be due to adult frogs using them for reproduction and therefore were more frequently present in bromeliads containing sufficient water for breeding whilst the juveniles were using them for feeding and therefore not dependent on how deep the water cavities are, only on water availability.

There were a higher presence of frogs in Bromeliad Species 1 receiving lower sun exposure (except tadpole- and adult presence), than in specimens with higher sun exposure. Higher sun exposure increases the UV-radiation, affecting the frogs negatively (e.g. Beebee & Griffiths 2005), by causing abnormalities in the skin affecting their possibility to breathe through their skin. However, Krügel and Richter (1995) performed a study on Syncope antenori (Microhylidae) in east-central Peru, where the presence of eggs and tadpoles were higher in bromeliads well exposed to sunlight (75%) compared to bromeliads in shade. In the present study, sun exposure was not found to have an effect on tadpole release nor on adult presence. Moreover, da Silva et al. (2011) found nine anurans using bromeliads in Brazil where bromeliads located at edges (exposed to sun) were more frequently occupied by anurans than bromeliads in the center of forests (in the shade). The difference to E. mysteriosus could be due to climate differences, since they are living in well-known dry habitats, making them less prone to “sunbathing” as species from rainforests tend to engage more in (da Silva et al. 2011). In the Bromeliad Species 2 sun exposure did not affect the frog presence, which could be explained by the Bromeliad Species 2 mainly growing in the D-area consisting of remnant dense forest and therefore had low variance of sun exposure.

In addition to the bromeliad quality variables the altitudes were also monitored for the individual bromeliads. Significantly higher presence of frogs were found on lower altitudes in Bromeliad Species 1 (except for presence all 6 times and tadpole presence) than on higher altitudes. However, this was not the case in Bromeliad Species 2. A possible explanation would be that the area where Bromeliad Species 1 grows are located at higher altitude than the area dominated by Bromeliad Species 2. That would be the probable reason for the lack of effect of altitude on occupancy in the Bromeliad Species 2 since they were already growing on an optimal altitude. The reason for the positive correlation in altitude and sun exposure could be due to the different vegetation types at the different altitudes. Higher up, bromeliads were mostly found on rock walls with low vegetation resulting in higher sun exposure compared to the lower altitudes mainly consisting of remnant forests with denser vegetation, where little sun perforate through the three canopy.

Another factor recorded was the density of bromeliads, which had no significant effect on the presence of E. mysteriosus in neither bromeliad species. It might be due to an overall low variation in bromeliad density in the areas, since they were mainly found in high numbers (clusters) and few were isolated. However, Donnelly (1989b) found that when increasing the bromeliad density the females and males increased in number, later on breeding increased as well as the number of juveniles in the poison dart frog Dendrobates pumilio .

Dispersal, feeding and breeding behaviour

The majority (72 %) of E. mysteriosus individuals stayed in the same bromeliad during the study of their dispersals. The mean dispersal distance of the territorial Dendrobates pumilio was found to be 20 m (Smith & Green 2005). The present study was performed during dry  months in Peru and the drought may make the frogs dependent on the bromeliads with water cavities more than during rainy months. However, it was discovered that the frogs could move up to 113 m which opens up possibilities of future habitat corridors, connecting suitable areas with wider bromeliads and plants containing deep water cavities for the frog species.

Caldwell (1997) stated that adult poison dart frogs primarily feed on ants, which is consistent with my observations.

There are numerous studies reporting on bromeliads being especially important in the reproduction, used for oviposition, tadpole development (Schiesari et al. 1996; Lannoo et al. 1987) and to complete metamorphosis (Donnelly 1989) in amphibians. Although, regarding the comparisons of tadpole releasing behaviour, not much articles have been found describing this in detail. The behaviour in E. mysteriosus does at least differ from one other Peruvian poison dart frog, who releases all their tadpoles at the same time in the same water cavity (pers. comm., Rainer Schulte, INIBICO, Peru). Nevertheless, the releasing of one tadpole per water cavity have been found in other poison dart frog species (Brown et al. 2008). The behaviour could be explained by space limitation in the water cavity, less competition of nutrients, decreased predation risk and avoiding of cannibalism (Brown et al. 2008). In terms of the breeding behaviour the positive correlation between depth of water cavities and bromeliad width are especially relevant since the adult frogs preferred to release their tadpoles in bromeliads that were bigger, which would then give tadpoles more water to develop in.

Conservation implications

The results show that the A-area had the most individuals, but the D-area had the highest occupancy rate in bromeliads. This might be due to the higher number of individuals encountered on ground in the A-area than in the D-area where most of the frogs were found inside the bromeliads. These two areas together with the B-area are core areas for the E. mysteriosus and should be the first targets for conservation efforts. The A-area is protected as reserve, which should be a priority for the other two as well.

But despite the protection as reserves slash-and-burn agricultural fires that run-away are common around the areas and most likely the biggest reason for possible extinction in the E- area and the great declines in the F-area. The A-area has also been affected by fires but are still harboring a large number of frogs. These fires need urgent controlling through conservation management of the areas.

Another big threat facing E. mysteriosus is deforestation. Both the frog-rich areas B and D are today threatened with total destruction giving way for plantations or grazing pastures. Species living in areas where the landscape is very fragmented have a higher risk of population isolation (Beebee & Griffiths 2005), which is happening in the areas right now due to the high deforestation pressure. If no conservation managements target these deforestations in the near future the possibility to create habitat corridors for the E. mysteriosus could get lost. Wider bromeliads (preferred by the E. mysteriosus ) are older and since it takes years for them to grow big they are very sensitive for deforestation, forcing E. mysteriosus to use smaller bromeliads with higher risk of population declines. Galindo-Leal et al. (2003) found that bromeliads in Mexico were completely absent from young forests (10-40 years old). Various bromeliad species are old-growth forest specialists (Benzing 1980, cited by Galindo- Leal et al. 2003) and therefore sensitive to human activities eliminating forest structures. Excidobates mysteriosus is only found in old-growth forests and if the rest of the remnant forests are cut down the species will be confined to rock walls solely.