Species identity and behavior of cave‐dwelling tree hyraxes of the Kenyan coast

Abstract We surveyed tree hyrax populations living in forests, limestone rocky formations, and caves in coastal Kenya to identify the species and estimate the threat‐level populations are in. Tree hyrax vocalizations were recorded in three different habitats with passive acoustic monitoring (PAM) for a total of 84 h in January and February 2022. We also observed tree hyrax behavior with thermal imaging camera and photographed individuals. Tree hyraxes in coastal Kenya are vocally active throughout the night, with most calls emitted between 23.00 and 04.00. We identified four different calls: snort, hac, hac ping‐pong, and wheeze. Their calling range is between 220 and 15,000 Hz. Calls of tree hyraxes from the coast of Kenya were compared with calls stored by the Oxford Brookes University's Nocturnal Primate Research Group and identified as eastern tree hyrax, previously recorded from Tanzania. Here, we present what are, to our knowledge, the first photographs of live D. validus from Kenya. These tree hyraxes live in social groups. Due to strong pressure from humans, conservation measures are necessary to prevent the extinction of these isolated D. validus populations in Kenya.

Tree hyraxes have short legs, and their ability to migrate between isolated forest fragments is poor. Lawes et al. (2000) studied tree hyrax patch occupancy and found that the probability of occupancy of a forest patch by tree hyraxes was zero if the distance between forest patches was more than 1.5 km.
The number of extant tree hyrax species is controversial.
However, the exact ranges of the different tree hyrax species are poorly known and many questions about their taxonomy are far from settled, suggesting that undescribed species may still await scientific discovery (Hoeck, 2017;Oates et al., 2022;Roberts et al., 2013).
The first account of tree hyraxes living in caves at the Kenya coast was published by Seibt et al. (1977), but this report has been largely overlooked. The discovery was fortuitous, as the authors were actually looking for fruit bats that roost in the caves. They observed tree hyraxes, recorded their vocalizations, and even collected some "mummified" remains and a complete skull of an adult male hyrax. Based on this material, Seibt et al. (1977) identified the cave-dwelling hyraxes as eastern tree hyrax Dendrohyrax validus but did not assign it to any subspecies. Their record was the first report of this species from Kenya; previously, eastern tree hyrax was only known from Tanzania. The eastern tree hyrax Dendrohyrax validus ( Figure 1a,b) has a short, smooth, dark brown coat. According to the IUCN's assessment (Hoeck et al., 2014), D. validus has a fragmented distribution in East Africa (Kingdon, 1971). As is the case of tree hyraxes generally, the subspecies-level taxonomy of the eastern tree hyrax taxonomy is controversial (see, e.g., Meschke, 2019, Rosti et al., 2020. Tree hyrax population density is usually estimated by analyzing the calling rate, as their sightings are rare (Rosti et al., 2022). The highest population densities of tree hyraxes have been found in large and intact natural forests (Topp-Jørgensen et al., 2008). Smaller forest size is usually linked with high levels of human disturbance such as the removal of trees and branches for timber and firewood, as well as poaching. When forest size decreases, calling patterns, and calls/hour change significantly, probably because tree hyraxes try to avoid being noticed by humans (Rosti et al., 2022).
Dendrohyrax validus population decline has been significant over the course of the last two decades due to hunting and habitat loss (Hoeck et al., 2014). In the Tanzanian Eastern Arc Mountains, tree hyraxes are threatened by hunting and snaring (Allen & Loveridge, 1927;Cordeiro et al., 2005). Even though the species is still locally abundant in some protected sites, sufficient data for accurately estimating population numbers are not available; therefore, D. validus is classified as Near Threatened by the IUCN (Hoeck et al., 2014).
The purpose of this study is taxonomical confirmation and disturbance-level estimation of previously unstudied populations of Dendrohyrax validus living in limestone caves and rocky outcrops on the coast of Kenya. We present photographs and videos ( Figure S1) of D. validus, describe their acoustic communication, calling patterns, and social behavior, and also estimate current conservation needs.

| Study sites
We were able to locate and visit three tree hyrax habitats on the

| Materials and methods
Tree hyrax species can be identified by their calls (Oates et al., 2022;Roberts et al., 2013). Calling activity and pattern of tree hyraxes correlate with population size and level of human disturbance (Rosti et al., 2022). For recordings, we used AudioMoth automatic record-

| Habitats and behavior
In Shimba Hills two of the three AudioMoths placed in mature natural forests did not record any tree hyrax calls, indicating that the animals are not common throughout the region. Two species of galagids (Otolemur garnettii and Paragalago cocos) were common at all three study sites. In Shimba Hills, we could hear tree hyraxes vocalizing while moving around in the forest, indicating that calling plays a role in group coordination.
In Vipingo, the tree hyraxes clearly live in groups, as we could repeatedly observe as many as nine individuals together. At this site, tree hyraxes were often seen walking on the ground. They used the caves as resting sites, and occasionally the cave-dwelling animals responded to the calls of others that were vocalizing in nearby trees.
In Vipingo, Otolemur garnettii was calling actively, but no calls of Paragalago cocos were recorded.
In Chasimba, tree hyraxes inhabit a rocky outcrop with vertical cliffs of up to 15 meters. They climb on the rocks and find safe daytime resting sites from the cliffs. At this site, tree hyraxes were the only nocturnal animals caught in our recordings (below 22 kHz), katydids and galagos were absent. Here, as at both other study sites, tree hyraxes defecate in middens.

| Calling activity
The data on calling activities were obtained during single recording nights in Shimba Hills and Chasimba, respectively, and two recording nights in Vipingo. At all three sites, tree hyraxes were called most actively between 23.00 and 04.00 ( Figure 4)

| Description and analysis of calls
Dendrohyrax validus calling range is 220-15,000 Hz ( Figure 5). Typical calls are short (duration 0.25-1.4 s), although some calls are longer.
We could recognize four different calls, even though elements of all of them could often be modified and combined, adding variation to the acoustic communication.
A Snort call has average duration of 0.4 s and maximum frequency of 3000 Hz (n = 61) (Figure 5a; Audio S1). Second pulse-like element has a curve that sounds like a chuck with a twist. This call was used at all sites, the example call was recorded from Vipingo. Here the element is repeated 17 times. Maximum frequency is 6100 Hz. The call has a variable number of harmonics from 3000 to 12,000 Hz. The call was usually combined with other elements and was recorded alone only once in Vipingo.

| DISCUSS ION
Dendrohyrax validus was first described by True (1890)  Southern tree hyraxes, D. arboreus are known to inhabit rocky terrain in the Ruwenzori Mountains, Uganda, and in this type of habitat, the animals have been reported to live in colonies (Kingdon, 1971).
In some parts of Usambara and Uluguru Mountains, Tanzania, tree hyraxes have been observed to retreat to boulders and rocky outcrops as local forests were cleared (Allen & Loveridge, 1927). Clearly, rock formations and caves can provide suitable resting places for tree hyraxes, and if at least some trees suitable for foraging remain in the area, they may have high value as refugia for dwindling tree hyrax populations.
While tree hyraxes of the Kenya coast are clearly not solitary animals, the social structure of their groups remains unknown. Recent research in the Taita Hills, Kenya, also suggests that tree hyraxes are more social than previously thought (Rosti et al., 2020(Rosti et al., , 2022. Tree hyraxes call most actively between 24.00 and 04.00 hr. This timing may reflect natural behavior, but it may also be a response to human disturbance (Gaynor et al., 2018). At the Kenyan coast, tree hyraxes sometimes, although rarely, are heard calling during the day. Dendrohyrax validus uses several different call types and different calls are commonly graded and combined. Call descriptions may have value for elucidating taxonomic relationships, as different tree hyrax species exhibit differences in vocal communication (Oates et al., 2022;Roberts, 2001;Roberts et al., 2013).
According to Seibt et al. (1977), local people of the Kenya coast, the Giriamas, consider some caves sacred and avoid disturbing bats and other animals that live in the caves (cf. Metcalfe et al., 2010). Seibt et al. (1977) hence suggested that the hyraxes may have received a level of protection from the sacred caves for quite some time, and might thus represent a relic population. We can confirm that the caves are still used by local people for cultural purposes and a population of tree hyraxes still exists around the caves. As the local coastal forests have been almost completely cleared, caves and sacred forest groves have become rare habitat islands, surrounded by open farmland or secondary shrub. Thus, such sites should be prioritized in conservation, not only for cultural reasons but also for their high value in conserving biodiversity (Brandt et al., 2013;Byers et al., 2001;Himberg, 2011;Ormsby & Bhagwat, 2010).

ACK N OWLED G M ENTS
We thank anonymous private landowners for their cooperation. We received research permits from the National Council for Science and Technology of Kenya, NACOSTI/P/21/9300. We are also grateful to Daniela Kalthoff at the Swedish Museum of Natural History, Stockholm, and Simon Musila at the National Museums of Kenya, Nairobi, for permission to study hyrax specimens in their care.

FU N D I N G I N FO R M ATI O N
The author HR was supported by the Kone Foundation Grant 202007604, Tuovinen Foundation, Ripaco, and by the University of Helsinki. The author HP was supported by grants from the Oskar Öflund Foundation and the Waldemar von Frenckell Foundation.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

O PE N R E S E A RCH BA D G E S
This article has earned an Open Data badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data is available at https://doi.org/10.5061/ dryad.sf7m0 cg99.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in Dryad at https://doi.org/10.5061/dryad.sf7m0 cg99.