The African golden cat Caracal aurata in Tanzania: first record and vulnerability assessment I LARIA G RECO and F RANCESCO R OVERO Abstract We report on the first population found in Tan- and snaring (Bahaa-El-Din et al., ). Although the pop- zania of the Vulnerable African golden cat Caracal aurata, ulation is believed to be decreasing (Bahaa-El-Din et al., extending its documented range c. km to the south and a), robust knowledge of the species is scant. The golden south-east. This is one of the least-known and truly forest- cat occurs in West and Central Africa, with an easternmost dependent felines in Africa, ranging across the Guinea– population recently confirmed in Kenya from a roadkill Congolian forest block. We recorded the new population in the Aberdare Mountains (Hatfield et al., ). Western in Minziro Nature Forest Reserve, north-west Tanzania, Tanzania has been regarded as potentially suitable for the during a -month survey in . We deployed camera species (Butynski et al., ), but no records have previous- traps on a regular grid and obtained detection events of ly been reported. Here, we present the first record and re- the golden cat at % of sites, with a minimum of indi- sults of habitat association modelling for a population of viduals across km . We estimated occupancy and de- golden cat found in Tanzania in the course of a biodiversity tection probability and modelled these in relation to the survey in Minziro Nature Forest Reserve. We used a camera- distance of sampling sites to the forest edge, which coincides trap survey, and occupancy analysis, to determine the dis- with both the Reserve boundary and proximity to human tribution of the golden cat and its association with habitat settlements surrounding the Reserve. Mean estimated occu- edges and human disturbance. pancy was . ± SE . (mean detectability = . ± SE .), Camera trapping was conducted in Minziro Nature with occupancy increasing significantly with distance from Forest Reserve, north-west Tanzania, during October– the forest edge. Detectability did not vary significantly with December . Established in and formerly a reserve distance from the forest edge, but was higher for camera with lower protection status, it comprises km of flat, models that had a shorter trigger time. Our findings add to moist forest at a mean altitude of , m (Rovero et al., the scant data available for this species. It appears threatened ). The northern boundary is the country border, mer- by human activity, which we recorded both outside and with- ging with the Sango Bay Forest Reserve in Uganda. All other in the Reserve, and the presence of the species indicates boundaries adjoin human-modified areas with settlements, Minziro Forest is an important site for its conservation. plantations and patches of drier woodland and grassland (Fig. ), with Kagera river and seasonal flooding to the Keywords African golden cat, camera trapping, Caracal east, and a paved road and several villages to the west. The aurata, detection probability, edge effect, Minziro, occu- Reserve is the easternmost extension in Tanzania of the pancy modelling, Tanzania Congo–Guinea forest, with fauna typical of West and Supplementary material for this article is available at Central Africa. The Reserve is heavily disturbed by illegal doi.org/./SX logging, burning, livestock grazing and fishing occurring in the area (Rovero et al., ). We found evidence of bush- meat hunting: we encountered hunters with dogs on one he African golden cat Caracal aurata is an elusive, occasion and found a number of snares, presumably set Tmedium-sized felid categorized as Vulnerable on the for ungulates but potentially a threat to the golden cat and IUCN Red List, regarded as the least-known wildcat in other species. Africa and one of the least-studied felines (Bahaa-El-Din We surveyed sites across the Reserve using a regular et al., b). As the only truly forest-dwelling feline in grid of – km cell size, firstly surveying camera sites Africa, it is threatened by deforestation, habitat destruction and then (Fig. ), for a minimum of days (mean days) each. We used Browning (Morgan, USA), Cuddleback (Green Bay, USA) and UOVision (Shenzhen, China) cameras, ILARIA GRECO (Corresponding author, orcid.org/0000-0002-6208-2465) with infrared flash and motion-activated sensor. The trigger Department of Biology, University of Florence, via Madonna del Piano 6, speed was . s for Browning, and – s for Cuddleback 50019 Sesto Fiorentino, Italy. E-mail [email protected] and UOVision. Cameras were unbaited and attached to FRANCESCO ROVERO*( orcid.org/0000-0001-6688-1494) Department of Biology, trees at c. cm from the ground, facing a presumed animal University of Florence, Sesto Fiorentino, Italy – trail. *Also at: Tropical Biodiversity Section, MUSE Museo delle Scienze, Trento, Italy The survey yielded , camera-days from camera Received December . Revision requested February . traps (two were stolen). We annotated images using Wild.ID Accepted May . First published online November . and ran analyses using R . (R Core Team, ). We calculated This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, Downloaded fromdistribution, https://www.cambridge.org/core and reproduction in any medium,. IP address: provided 170.106.35.229 the original work, on 29 is properly Sep 2021 cited. at 00:06:50, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/termsOryx, 2021, 55(2),. https://doi.org/10.1017/S003060532000040X 212–215 © The Author(s), 2020. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S003060532000040X African golden cat in Tanzania 213 border as a proxy for increasing habitat quality and decreas- ing human disturbance. We assumed that both ψ and p increase with distance to border. For detection probability we also considered that camera-trap model could potential- ly influence detection. Using unmarked (Fiske & Chandler, ), we built all model combinations and ranked them according to the Akaike information criterion (AIC). We considered statistically supported models with ΔAIC , . and used AICcmodavg (Mazerolle, )toaverageparam- eter estimates. After discarding , blank and set-up images, camera trapping yielded , photographs of species of medium to large wild mammals (Supplementary Table ), including images of golden cats. These corresponded to in- FIG. 1 Minziro Nature Forest Reserve, Tanzania, with camera- trap sites and records of the African golden cat Caracal aurata. dependent detection events of the species within -hour intervals at sites (naïve occupancy = .; Fig. ). Based on individual pelage marks, we identified a minimum of the number of golden cat detections per day, and used individuals. Pelage colouration varied from dark brown them to derive a relative abundance index. We plotted the to sandy and golden (Plate ). Animals were mainly active species’ daily activity pattern using hourly events. We then at night, with activity peaking at ., . and ., but used single-season occupancy modelling (MacKenzie et al., were also detected in daylight (Supplementary Fig. ). Based ) to estimate golden cat occupancy (ψ) and detection on occupancy model ranking (Table ), we averaged the two probability (p) and their association with presumed covari- most supported models, and the resultant mean estimated ates. We arranged golden cat detections/non-detections into occupancy was . ± SE .. Occupancy increased signifi- a site-by-sampling occasion matrix; we chose days as the cantly with distance to the Reserve border (. ± SE .; resolution of the occasions as with poorly detected species Table , Fig. ). Mean detection probability was . ± SE this provides the best estimation of detectability, and hence . and did not vary significantly with distance to the occupancy (Rovero & Spitale, ). We built parsimonious Reserve border but did vary between camera models, with models using the distance from each site to the closest Browning cameras having significantly higher values (Sup- Reserve border (i.e. forest edge) as a covariate. We derived plementary Table ). these distances using QGIS (QGIS Development Team, Our findings suggest that habitat intactness and human ) and a :, scale map. Based on our survey of disturbance affect golden cat occurrence, with increased oc- human disturbance, we considered greater distance to the cupancy with increasing distance from the Reserve border PLATE 1 African golden cat Caracal aurata photo-trapped in Minziro Nature Forest Reserve, Tanzania. The survey revealed golden/brown-reddish (a, c) and dark/light grey (b, d) morphotypes. Oryx, 2021, 55(2), 212–215 © The Author(s), 2020. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S003060532000040X Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.229, on 29 Sep 2021 at 00:06:50, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S003060532000040X 214 I. Greco and F. Rovero TABLE 1 Model ranking for occupancy (ψ) and detection probability (p) of the African golden cat Caracal aurata detected by camera trapping in Minziro Nature Forest Reserve (Fig. ), Tanzania. Models No. of parameters AIC1 Δ AIC AIC weight R2 p (Camera model) ψ (Distance from closest edge) 5 198.88 0.00 0.59 0.27 p (Camera model + Distance from closest edge) ψ 6 200.50 1.62 0.26 0.27 (Distance from closest edge) p (.) ψ (Distance from closest edge) 3 203.44 4.56 0.06 0.16 p (Distance from closest edge) ψ (Distance from closest edge) 4 204.18 5.31 0.04 0.18 p (Camera model) ψ (.) 4 204.83 5.95 0.03 0.17 p (Camera model + Distance from closest edge) ψ (.) 5 205.68 6.80 0.02 0.18 p (.) ψ (.) 2 212.82 13.95 , 0.01 0.00 p (Distance from closest edge) ψ (.) 3 213.16 14.28 , 0.01 0.03 Models were ranked according to the Akaike information criterion (AIC), with ΔAIC , considered supported.