Introduction

Conservation of a species requires knowledge of its taxonomy, distribution, and density, combined with evaluation of threats, and finding solutions. In East Africa, mountains have become forest islands isolated by dry savannas, semidesert, and agricultural landscape (Lowett & Wasser 1993; Hemp & Hemp, 2018). The biogeography of mammals living on mountaintops is analogous to that of species living in islands (Brown, 1971). Forest-dependent species cannot cross these dry areas. For this reason, conservation needs for each isolated population should be evaluated. Limited knowledge of the distribution of species, and sometimes unsolved taxonomy, also may lead to local extinctions that go unnoticed by scientists.

Modern techniques may help to solve research questions without the disturbance caused by human observers (Piel et al., 2022). Passive acoustic monitoring (PAM) allows the collection of a large body of data simultaneously from several locations, and over extended periods of time, and is particularly useful in conservation (Marques et al., 2013; Hill et al., 2018). Studies using PAM in primates have examined a variety of research questions. For example, a study used PAM with an acoustic location system (ALS) to detect the very loud long distance calls of male orangutans (Pongo pygmaeus), triangulate their positions with mean accuracy of 58 m and study the species’ social organization (Spillmann et al., 2015). Another study of the occurrence of three species of primate in Côte d’Ivoire concluded that PAM could be used as early-warning system for poaching activity (Kalan et al., 2015). A study of chimpanzees (Pan troglodytes) in Tanzania found that chimpanzees were acoustically active during the night and that these calls were social, reflecting chimpanzee grouping patterns (Piel, 2018). A study using 57 PAM locations in Vietnam estimated the population size of southern, yellow-cheeked, crested gibbon (Nomascus gabbrielae) (Vu & Tran, 2019) and concluded that method was reliable and less laborious, cheaper, and less dependent on skilled field surveyors than other methods. PAM also has been used to estimate the population density of pale fork-marked lemurs (Phaner pallescens) (Markolf et al., 2022). There is growing interest in the use of artificial intelligence (AI)-based methods to identify vocalizations of study species from large datasets. For example, a convolutional neural network was used successfully to identify individual indri (Indri indri) from songs with more than 90% accuracy (Ravaglia et al., 2023).

It is increasingly common to use airborne light detection and ranging (LiDAR) data on forest and canopy structure in biodiversity studies on birds, mammals, reptiles, amphibians, invertebrates, bryophytes, lichens, and fungi, because it provides much needed information for conservation management (McLean et al., 2016; Singh et al., 2018; Acebes et al., 2021). Airborne LiDAR systems send numerous laser pulses a fraction of which can penetrate the forest canopy all the way to the ground and can hence provide information on both canopy height and the vertical and horizontal distribution of plant biomass (Adhikari et al., 2020; Simonson et al., 2014). One benefit of airborne LiDAR is that it can be used to obtain accurate data on many different attributes of forest structure without time-consuming ground surveys (Adhikari et al., 2020). In primates, a study using LiDAR revealed that bald-faced saki monkey (Pithecia irrorate) home ranges occur in the tallest and most uniform canopies in Peruvian Amazon, whereas the species was missing from areas with lower and heterogenous canopies (Palminteri et al., 2012). A study combining LiDAR with field surveys found that the movement patterns of three arboreal primate species in the tropical forests of Panama correlated with canopy height and distance to gaps (McLean et al., 2016). A study combining visual land survey and LiDAR in orangutans concluded that they preferred closed canopy, tall trees with uniform height, and avoided canopy gaps and trees with high crowns (Davies et al., 2017). As study of red langurs (Presbytis rubicunda) and Bornean agile gibbons (Hylobates albibarbis) in Kalimantan, Indonesia, revealed that of several different variables anthropogenic disturbance was most important variable in explaining habitat suitability for both species, although there was only limited overlap in their habitats (Singh et al., 2018).

The number of galago species has long been underestimated. However, by the 1980s and 1990s, primatologists had become aware that several galago species were in fact cryptic species complexes (Groves, 2001; Grubb et al., 2003; Nash et al., 1989). Even in recent years, galago species that are new to science have been described (Ambrose, 2003, 2013; Svensson et al., 2018). Galago species and genera have typical vocalizations, by which they can be identified in the field (Bearder et al., 1995; Groves, 2001). For example, the largest galagos, of the genus Otolemur, are characterized by trailing calls, while the lesser galagos of the Senegal galago (Galago senegalensis) group use repetitive calls. The vocalizations of eastern dwarf galagos are either rolling or incremental calls, depending on the species, whereas the western dwarf galagos, which are only distantly related to them, use crescendo calls. Galago species’ vocal repertoires can be subdivided into different categories (e.g., alarm calls and courtship calls), which may be species-specific too. These differences in vocalizations presumably play an important part in mate recognition and selection in galagos (Bearder et al., 1995; Honess, 1996; Honess & Bearder, 1996; Butynski et al., 1998). In areas where the distributions of different galago species overlap, acoustic differences are likely to play an important part in preventing interspecific mating (Bettridge et al., 2019; Butynski et al., 2006; Courtenay & Bearder, 1989; Génin, 2021; Masters, 1998; Masters & Couette, 2015; Masters et al., 2017; Rovero et al., 2009; Zimmermann, 1990).

The development of molecular biology has added new insights into the phylogenetic relationships and taxonomy of galagos (Masters et al., 2007; Pozzi, 2016; Pozzi et al., 2014, 2015, 2020). These studies suggest that some extant galago lineages diverged during the Miocene, 10–20 Ma, or even during the Oligocene, more than 30 Ma (Masters et al., 2017). This information, together with increased knowledge of galago biogeography, genetics, acoustics, and morphology, has resulted in revisions of the genus-level taxonomy of Galagidae. Most researchers today recognize several different galago genera. One of the newly recognized genera is the eastern dwarf galago genus Paragalago, the members of which were previously included in Galagoides (Masters et al., 2017). The genus Paragalago includes several taxa that all live in East Africa. Most researchers recognize the following species: Kenya coast dwarf galago (P. cocos), Mozambique dwarf galago (P. granti), mountain dwarf galago (P. orinus), Rondo dwarf galago (P. rondoensis), and Zanzibar dwarf galago (P. zanzibaricus). The last species is separated into two subspecies, P. z. zanzibaricus and P. z. udzungwensis (Honess et al., 2013). The different Paragalago species are very similar in appearance, but there are slight differences between them in body size and morphology (Butynski et al., 2006; Harcourt & Bearder, 1989; Harcourt & Nash, 1986; Harcourt & Perkin, 2013; Perkin, 2007).

Some Paragalago species, such as the mountain dwarf galago, are found in Afromontane forests, such as the Uluguru Mountains in Tanzania, where this species was first described by scientists (Lawrence & Washburn, 1936). Much later, Perkin et al. (2002) discovered dwarf galagos in the Mbololo and Ngangao forests of the Taita Hills in Kenya at elevations of 1,550–1,900 m a.s.l. Perkin et al. (2002) were unable to determine the taxonomic status of these galagos and suggested that the Taita Hills Paragalago population may either belong to the Kenya coast galago, the mountain dwarf galago, or be a species new to science. The Taita dwarf galagos then received no scientific attention until Rosti et al. (2020b) found small populations of them in the same forest fragments. The species-level identity of the Taita dwarf galagos has remained unclear, as no voucher specimens of the animals have been deposited in museum collections. However, analyses and comparisons of vocalizations suggest an affinity with Kenya coast dwarf galago (Butynski et al., 2006; Harcourt & Perkin, 2013; Rosti et al., 2020a), and the Taita Hills populations may thus be relics of a once wider distribution of this species.

The Taita Hills are part of the Eastern Arc Mountains, which have a diverse fauna and flora with many endemic species (Burgess et al., 1998, 2007). With the coastal forests of Kenya and Tanzania, they represent one of the biodiversity hot-spot areas in Africa (Myers et al., 2000). The Taita Hills rise from the surrounding plains at ca. 600–1,000 m a.s.l. to a series of mountain ridges, which reach 2,208 m a.s.l. at their highest peak, Vuria (Pellikka et al., 2009). The mountain ridges of the Taita Hills are isolated from each other by deep, dry valleys. The montane forests appear to have been continuous in the Pleistocene, but dryer climate-induced forest fragmentation set in circa 0.93 Ma, when Mt. Sagalla forest became isolated from the rest (Measey & Tolley, 2011). Subsequently, Mt. Kasigau became isolated from the Mbololo–Ngangao forest bloc at circa 0.76 Ma, and finally, contact between Ngangao and Mbololo forests was severed at circa 0.6 Ma (Measey & Tolley, 2011). The Taita Hills are approximately 160 km from the Indian Ocean coast of Kenya. The area between Taita Hills and coastal forests is dry lowland shrub land and savanna, which presumably represents an insurmountable dispersal barrier for forest-dependent dwarf galagos.

In this study, we describe the distributions and estimate the population density of Taita mountain dwarf galago populations in Mbololo (Fig. 1a) and Ngangao (Fig. 1b) forests in the Taita Hills, Kenya. Both forests, and therefore the animals, are under serious threat from human activity including illegal exploitation of the natural forests (Teucher et al., 2020; Wekesa et al., 2020, 2021; Kung’u et al., 2023). We measured forest size, forest height, canopy cover, canopy density, and distance to the forest edge by using airborne LiDAR and analyze their effects on the presence of Taita dwarf galagos, indicated by calling activity per hour. We describe field observations from Mbololo and Ngangao forests and compare our findings to the closest Kenya coast dwarf galago populations, which inhabit natural forests in Diani (Fig. 1c), and Shimba Hills (Fig. 1d) on the Kenyan coast. As the phylogenetic relationships between the dwarf galagos in the Taita Hills and these Kenyan coast populations are still unresolved, the possibility remains that they are taxonomically distinct. Thus, we call our study populations “Taita dwarf galagos” to distinguish them from other Paragalago populations.

Fig. 1
figure 1

Dwarf galagos from genus Paragalago from Kenya. A) Taita dwarf galago from Mbololo forest. B) Taita dwarf galago from Ngangao forest. C) Kenya coast dwarf galago from Diani. D) Kenya coast dwarf galago from Shimba Hills. Photos Hanna Rosti 2021.

Methods

Study Sites

We conducted field work in the Taita Hills (03 220 S, 38 200 E), southeastern Kenya (Fig. 2). Our main research areas were two largely intact indigenous montane forests: Mbololo (185 ha) and Ngangao (120 ha). Both forests have multilayered canopy formed by many different tree species and patches of exotic trees (Thijs, 2015). Mbololo has an elevation of 1,550–1,700 m a.s.l., is far from other mountains, and is rainy, windy, and wet. The very steep slopes of Mbololo Ridge have helped to protect the forest from timber harvesting. Ngangao forest (120 ha) is at an elevation of 1,700–1,870 m a.s.l., surrounded by agricultural land and dense human population, and easier to access than Mbololo (Omoro et al., 2010).

Fig. 2
figure 2

Study locations. A) Location of forests in Taita Hills, Shimba Hills, and Diani in Kenya. B) Ngangao forest in Taita Hills. C) Mbololo forest in Taita Hills. Sampling sites in B and C are locations of AudioMoth recorders.

The mean annual temperature in Wundanyi, the largest city in the area, is 19.7 °C (Pellikka et al., 2009; Stam, 2020). The coldest month is July with a mean temperature of 17.3 °C, and the warmest month is March with a mean temperature of 21.8 °C. Mean annual rainfall is 1,140 mm. The indigenous cloud forests in the Taita Hills have suffered substantial decline in size and increasing degradation over several centuries due to agricultural expansion (Pellikka et al., 2013). Sykes’ monkeys (Cercopithecus mitis) from the forest forage in farms (Siljander et al., 2020). Small-eared greater galagos (Otolemur garnettii lasiotis) are frequently heard in the agricultural landscape and feed commonly on cultivated bananas in gardens and hunt insects that are attracted by artificial lighting (Pihlström et al., 2021). There is also dense tree hyrax (Dendrohyrax sp.) population in Mbololo and Ngangao forests (Rosti et al. 2023a). 

Fieldwork and Recordings

We conducted fieldwork in Ngangao and Mbololo forests from August 2018 until March 2022. Our research focused on Taita dwarf galagos, small-eared greater galagos, and tree hyraxes (Dendrohyrax sp.) During fieldwork, we observed galagos with red flashlights and by using a thermal imaging camera Pulsar Helion 2 XP50 (Pulsar, Vilnius, Lithuania). We mainly focused on the Ngangao forest (455 nights) and spent less time in Mbololo forest (30 nights). We did regular walks calculating animal’s encounter rate/hour, during which we also observed animal behavior. We also conducted these walks during the automated recordings.

We used Passive AudioMoth automatic recorders (v1.1.0 Open Acoustics Devices, Southampton, UK) in Ngangao and Mbololo forests in January and February 2021. We combined data collected in recordings with airborne LiDAR data for six locations in Mbololo forest (131 h) and four locations in Ngangao forest (99 h).

We acquired LiDAR data for the forests in January–February 2014 and February 2015 using an aircraft-mounted Leica ALS60 sensor (Leica Geosystems AG, Heerbrugg, Switzerland) (Adhikari et al., 2017). We classified LiDAR points as ground points and nonground points to compute a digital terrain model and a canopy height model (CHM) at 1-m resolution. We normalized point clouds for the ground elevation to derive heights from the ground level. We extracted normalized point clouds for circular areas of 0.5 ha (39.9-m radius) around AudioMoth sites. We acquired the following data: maximum canopy height (m), canopy coverage at 20-m height (%), canopy density at 75% return height (%), and distance to the forest edge (m) (Heiskanen et al., 2015). As all sites are situated in tropical montane forest, they have high vegetation cover close to the ground (below 3 m) unless there are major gaps in the canopy. For detailed description of airborne LiDAR (laser scanning) methods, see Heiskanen et al. (2015) and Rosti et al. (2022). We calculated all canopy structure metrics were using lidR package (Roussel et al., 2020) in R (R Core Team, 2020).

In addition to Ngangao and Mbololo forests, we searched other forests for dwarf galagos by using AudioMoths and on foot, including Kasigau (203 ha), Vuria (96 ha), Chawia (85 ha), Fururu (8 ha), Yale (2 ha), and Sagalla (2 ha) (Rosti et al., 2022). The populations in Ngangao and Mbololo forests are isolated from each other and from the distant Paragalago populations on the Kenya coast.

We configured AudioMoths with a sample rate of 48 kHz, recording length of 60 s and medium gain. Because saving each recording in AudioMoths takes 5 s, the total recording period during each hour was 55 min. We analyzed 231 recording hours and calculated the calling density for each hour between 19:00 and 06:00. We included loud contact and advertising calls. Loud advertising calls are species-specific in Paragalago species and thus likely to be an important part of mate recognition (Bearder et al., 1995; Groves, 2001; Masters et al., 2017). We did not use warning and mobbing calls, because these are more conserved across taxa (Masters et al., 2017).

We verified all calls by listening with RAVEN PRO 1.6 (Cornell University, Ithaca, NY) using the following spectrogram parameters: DFT size 512, 50% overlap,hann 86.1 Hz. We inspected recordings visually from spectrograms in a window frame of 70 s. We did not use automatic detection, because dwarf galago calls are weaker than the acoustic signals of other fauna, and most calls would not be recognized.

We also visited the indigenous natural forests in Diani and Shimba Hills National Reserve, where Kenya coast dwarf galagos are found (Harcourt & Perkin, 2013). In both locations, we photographed dwarf galagos and recorded their calls with three AudioMoths. We visited Diani twice, where we studied one dwarf galago population in an isolated forest fragment in the beach hotel area. Our first visit to Diani was on 21–23 September 2021, with 66 h of recordings, and the second was on 13–14 September 2022, with 22 h of recordings. In Shimba Hills, we recorded dwarf galagos on 27–28 January 2022, collecting 66 h of recordings. We compared the photographs and spectrograms of Kenyan coastal dwarf galagos with those of Taita dwarf galagos.

Statistical Methods

We performed statistical analyses in RStudio 2022.07.01 (Rstudio Team, 2020) with R version 4.1.2 (R Core Team, 2020). We used packages “glmmTMB” (Brooks et al., 2017) and “performance” (Lüdecke et al., 2021) and generated plots with ggplot2 (Wickham, 2016). We explored data according to Zuur and Ieno (2016). We used DHARMa package in model validation (Hartig, 2022).

We applied a zero-inflated negative binomial general linear mixed model (ZINB GLMM), for data consisting of number of calls/hour and forest structure and size. The response variable was calls per hour to measure the relative density of calling animals at each site. We used recording site as a random intercept to avoid pseudo-replication. The data included a high number of zeros (55%), making a zero-inflated model necessary. We used drop function for likelihood ratio test to estimate the significance of each predictor variable. We checked model assumptions by plotting residuals versus fitted values. Model validation indicated no problems, and we retained all variables in the model.

Explanatory variables were forest size (ha) and maximum tree height (m), canopy cover (%), and canopy density (%) extracted from airborne LiDAR. We defined maximum tree height as 99% percentile return height considering only data point returns above 3 m in height. We defined canopy cover as canopy cover at 75% height from maximum tree height. We measured canopy density at 20-m height (%). We measured distance to the forest edge from LiDAR data. There was only weak collinearity between these variables. We used Akaike information criteria (AIC) to test the model with drop 1 function.

Results

Statistical Analysis

Mbololo forest is larger (185 ha) and hosts a larger population of Taita dwarf galagos than Ngangao forest (χ2 = 13.4, degrees of freedom (df) = 1, p < 0.001) (Fig. 3; Table I). Higher canopy cover increased Taita dwarf galago presence (χ2 = 13.4, df = 1, p < 0.001) (Fig. 4a). Taita dwarf galagos preferred sites with lower forest height (20–30 m) (χ2 = 9.7, df = 1, p < 0.002) (Fig. 4b). Also, higher canopy density had positive effect on dwarf galago presence (χ2 = 3.7, df = 1, p < 0.008) (Fig. 4c). The conditional R2 was 0.69, and the marginal R2 (without random effects) was 0.68, meaning that model explained 68% of where Taita dwarf galago were calling in Mbololo and Ngangao. The intraclass coefficient (ICC) was 0.029, meaning that correlation within sites (random intercept) in the model (AudioMoth site) was small.

Fig. 3
figure 3

Taita dwarf galago calls/ hour recorded by AudioMoth recorders in Mbololo and Ngangao forests, Kenya, in January and February 2021 (N = 231 h, Mbololo 131 h; Ngangao 99 h). A) Contact calls per hour; each dot represents an hour. Horizontal line represents median, box represents upper and lower quartile, and whisker represents scores outside middle 50% of the scores. B) Calling rate per hour during the night between 19:00 and 06:00. The line is the mean, the smoother with coloring, indicates conditional means created by ggplot2, to aid visualization of the graph (Wickham, 2016). Dots represents call rate per hour, value for each hour is marked to the end of the hour (e.g., hour between 19:00 and 20:00; value is found from 20:00).

Table I Results of a zero inflated negative binomial generalized linear mixed model (calls per hour ~ forest size + canopy cover + tree height + canopy density) from Taita dwarf galago calls per hour in Mbololo and Ngangao forests, Taita Hills, Kenya from January and February 2021
Fig. 4
figure 4

Analyzed forest characteristics with Taita dwarf galago calls from sample points in Mbololo and Ngangao forests from Kenya. Dots indicate calls per hour in the location; line is regression line. Mbololo forest is with black dots and Ngangao forest with grey dots. A) Canopy density (%) measured from 20 m height. B) Maximum tree height (m). C) Canopy cover (%) measured from 75% height from maximum tree height.

We recorded 412 calls in 131 h in Mbololo forest and 16 calls in 99 h in Ngangao forest. The mean calling rates were 3.14 calls per hour in Mbololo forest and 0.16 calls per hour in Ngangao (Fig. 3).

Observations

We observed acoustical differences between Taita dwarf galagos from Mbololo and Ngangao forests (Table II; Fig. 5ab, Sound S1 and S2). Individuals give a contact call in Ngangao and Mbololo forests approximately every 20 min, and this is answered in approximately 50% of cases. In the incremental call, units are typically combined towards the end of the call (Butynski et al., 2006); however, this pattern is mostly missing from dwarf galago calls in Taita Hills. We also heard this call ending at Diani (Fig. 5c, Sound S3) and Shimba Hills (Fig. 5d, Sound S4). However, we also recorded short contact calls without an incremental ending from Kenya coast dwarf galago in Diani.

Table II Taita dwarf galago and Kenya coast dwarf galago call measurements easurements from contact calls presented in Fig. 5 measured from different forests in Kenya, recorded in January and February 2021
Fig. 5
figure 5

Spectrograms of dwarf galago contact calls from forests in Kenya recorded in 2021 and 2022. A) Contact call from Mbololo. B) Contact call from Ngangao forest. C) Contact call from Diani. D) Incremental call from Shimba Hills.

We often observed dwarf galagos hunting insects close to the ground and occasionally saw them walking awkwardly on the ground. Taita dwarf galagos in Mbololo were very shy, and almost all individuals hid in the canopy immediately as we encountered them.

In 2019, we heard the Taita dwarf galago morning assembly chorus each morning just before 06:00 h in Ngangao forest. We no longer heard this chorusing behavior in 2020. Calculating encounter rate per hour is not meaningful as Taita dwarf galagos are so few in number. Based on the decrease of individuals observed, reduced number of calls heard and changes in vocalization behavior, we estimate that the population in Ngangao has decreased during the past 4 years and is currently fewer than ten individuals.

In Mbololo forest, the encounter rate was 0.4 animals per hour. We did not observe changes over time in Taita dwarf galago encounters. In our Diani study site, the density of the dwarf galagos was very high compared with other study sites, with an encounter rate of 4.0 animals per hour. In Shimba Hills, we found dwarf galagos in only one location, at Sable Bandas, although we also placed AudioMoth recorders in two other locations in the indigenous forest.

In Ngangao, we followed a dwarf galago group to several different sleeping sites inside hollow trees in January and February of 2019 and 2021, and occasionally between 2019–2022, for total of 37 times. The group changed sleeping site every few days. We observed them as they first gathered at dawn. Then, group members greeted each other and went into the tree hole. We also observed them occasionally meeting and greeting one another and then separating to use two different daytime sleeping sites.

Discussion

Taita dwarf galagos are more common in Mbololo than in Ngangao forest. This is most likely the result of larger forest size (185 ha). This is expected, as typically in montane islands number of species is closely correlated with size of the island (Brown, 1971). In Mbololo, there is less disturbance, because human population density around remote Mbololo forest is lower than around Ngangao forest (Omoro et al., 2010). Our study shows that Taita dwarf galagos prefer habitats where tree height is lower. Because dwarf galagos are mainly insectivorous (Harcourt & Perkin, 2013), tall trees may not be necessary for their foraging behavior. Based on LiDAR analysis, Taita dwarf galagos seem to prefer dense canopy coverage and canopy density. To fully understand the habitat characteristic preferences on arboreal primates, GPS collars or tags should be used to monitor movement patterns of the study species (McLean et al., 2016).

Based on phylogenetic studies, Zanzibar dwarf galago and Kenya coast dwarf galago split approximately 3 million years ago (Pozzi, 2016; Pozzi et al., 2019, 2020). At that time (Plio-Pleistocene) climatic fluctuations caused forest contraction in East Africa and populations became separated (Pozzi, 2016). Different habitats may influence the evolution of Paragalago loud calls (Génin, 2021). Slight differences in contact calls between populations of Taita dwarf galagos may be caused by their long isolation from each other (Baker et al., 2006; Campbell et al., 2010; Grant & Grant, 1996; Hoskin et al., 2005; Pang-Ching et al., 2018; Wich et al., 2008) and by the different acoustic environments of their habitats (Rosti et al., 2020a, b). The Taita dwarf galagos in Mbololo and Ngangao may have been isolated from each other for more than 0.5 million years, based on a genetic study of the endemic Taita blade-horned chameleon (Kinyongia boehmei) (Measey & Tolley, 2011). There also is variation in mountain dwarf galago vocalizations among populations in Tanzania (Perkin et al., 2013).

Based on acoustics, Taita dwarf galagos sound like Kenya coast dwarf galagos (Rosti et al., 2020a, b, 2022). At the Kenyan coast, Kenya coast dwarf galago lives in dry, mixed coastal forests (Harcourt & Perkin, 2013). In Taita Hills, dwarf galagos live in moist mountain cloud forest with mostly different tree species. Miller et al. (2023) studied climatic niches (temperature, precipitation, and seasonality characteristics) of three Paragalago species, Kenya coast dwarf galago, Zanzibar dwarf galago, and Mozambique dwarf galago. In the geographic projection of the climatic niche model Kenya coast dwarf galago had quite small range at the coast of Kenya, mostly overlapping with Zanzibar dwarf galago (Miller et al., 2023). This model showed optimal predictive ability for Kenya coast dwarf galago based on the species current known range. In the East Usambara Mountains in Tanzania Kenya the coast dwarf galago is found at altitudes of up to 350 m a.s.l. (Butynski & De Jong, 2016; Butynski et al., 2006). In the Taita Hills, however, the dwarf galagos live at altitudes of 1750–1900 m a.s.l. If Taita dwarf galagos are Kenya coast dwarf galagos, they must be relict population of once much wider distribution of Kenya coast dwarf galago.

There are behavioral differences between Paragalago populations from Taita Hills and coastal forests. Taita dwarf galagos are very shy and difficult to approach. Local people are unaware of their existence (personal observations HR, JR), probably because dwarf galagos do not venture out of the forests into gardens and agricultural lands. The Kenya coast dwarf galago is common in some remaining patches of indigenous forest in Diani. These animals are not shy and are quite used to people. Kenya coast dwarf galago also can be found in gardens and in coconut groves in Diani (Harcourt & Perkin, 2013).

Perkin et al. (2002) observed or heard 0.4 dwarf galagos/hour in Mbololo and Ngangao forests. Our observations match theirs in Mbololo, suggesting that the population in Mbololo has remained stable since 2002. The extremely steep slopes of Mbololo forest impede human access to many parts of the forest, which is likely to explain why it hosts a large population of dwarf galagos. In the Ngangao forest, Taita dwarf galagos are close to extinction, and we estimate that there are fewer than ten individuals left. Unfortunately, in a population this small, inbreeding, disease, predators, or a combination of these may cause extinction in the near future. Persistent logging and firewood collection still cause disturbance in both forests, and this has a negative effect on the abundance of arthopods (Kung’u et al., 2023). Kung’u et al. (2023) concluded that this may have a negative effect on insectivorous birds. Taita dwarf galagos also are insectivorous, and continuing forest degradation will most likely have a negative effect on dwarf galagos as well.

Dwarf galagos are missing from several other remaining fragments of indigenous moist montane forest in Taita Hills, namely, Mt. Kasigau, Vuria, Chawia, Fururu, Yale, and Sagalla. Of these, the largest and best-preserved forest is Kasigau forest. The apparent absence of dwarf galagos from Mt. Kasigau may be the result of extensive drought and forest contraction in East Africa during the Pleistocene. The latest such dry period started circa 0.9 Ma (Trauth et al., 2007) and is likely to have affected the present-day distribution of several forest-dependent East African taxa (Measey & Tolley, 2011). Mt. Kasigau is quite isolated from other mountain blocks in the Taita Region and surrounded by very dry, semiarid plains. The absence of dwarf galagos in Vuria, Chawia, Fururu, Yale, and Sagalla may be linked to human disturbance, as all of these forest fragments are very small and surrounded by dense human populations.

Ngangao forest is narrow, only 700-m wide at its widest point. The future of this forest is uncertain, because it is surrounded by increasingly dry deforested landscape. Several endemic bird species in Taita Hills also are at the brink of extinction due to habitat loss (Obunga et al., 2022). Reforestation and strict protection of remaining forests from illegal logging and firewood harvesting is urgently needed to prevent extinctions.

Taita dwarf galagos need a more effective conservation strategy. They can be considered as small flagship, or Cinderella species (Nekaris et al., 2015), as they are cute, with large round eyes, and there is no conflict with local people. We have proposed that Taita dwarf galago are possible species for sustainable ecotourism in Taita Hills (Rosti et al., 2023b). We also have initiated a conservation project with a local school and a school in Finland. This project has already resulted in the planting of more than 600 trees at the edges of Ngangao forest. The project gives students an opportunity to learn about the forest and its animals and to visit the forests and conservation areas. There also are other ongoing reforestation projects in Taita Hills. However, it is unknown how successful these projects will be in the long run, because the growth of natural indigenous trees is very slow.

Arboreal nocturnal habits, cryptic appearance, and shy behavior make dwarf galagos difficult to study. Current ranges of many dwarf galagos remain poorly known (Harcourt & Perkin, 2013; Honess et al., 2013; Perkin & Honess, 2013; Perkin et al., 2013). Of 47,100 articles published on primates, listed in the Thomson Reuters’ Web of Science between January 1965 and March 2016, only 1% focused on galagos (Estrada et al., 2017). Moreover, there are no studies on primates in many African countries, and only 17.6% of all scientific articles on primates focused on conservation, while most focussed on other topics (Bezanson & McNamara, 2019). Considering that currently 75% of the primate populations have declining population and approximately 30% are Critically Endangered (Estrada et al., 2017; Fernández et al., 2022), focusing on gaps of current knowledge in primate conservation is crucially important. However, without up-to-date knowledge of the species range conservation is difficult and uncertain. IUCN (2022) category “Least Concern” means that extinction risk for the species is lower than in other threat categories. However, species may still need conservation (IUCN, 2022). Criteria of listing species from Least Concern to increased threat levels has been criticized for the difficulty of providing solid evidence of a decrease in population size or the area of the occupancy or extent of occurrence (Fernández et al., 2022). Providing this evidence may be particularly difficult with small, cryptic, and nocturnal primates. The Difficulty of assessing the threat level may reduce conservation attention and measures from species that are in desperate need of them. The Taita dwarf galago is a good example of this, because without verified species and conservation status, effective conservation measures are difficult to implement.

Conclusions

The Taita dwarf galagos may represent a relict population of Kenya coast dwarf galago—a species that is otherwise restricted to the Kenyan Coast. However, there are acoustical and behavioral differences between dwarf galagos in Taita Hills and the coastal dwarf galagos. Molecular analyses are needed to resolve the taxonomic status of the Taita animals. The two-remaining dwarf galago populations in Taita Hills are in great danger of going extinct, because the total area of their home forests is less than 3 km2. The Taita dwarf galago population in Ngangao, with only about 10 remaining individuals, is at imminent risk of extinction.