Effectiveness of wildlife underpasses and culverts in connecting elephant habitats: a case study of new railway through Kenya’s Tsavo National Parks

Transportation networks can be a major impediment to wildlife movements. We assessed the use of underpasses and culverts along newly constructed railway in Kenya's Tsavo National Parks by African elephants (L. africana). collared ten with GPS satellite transmitters within 20 km March 2016 analysed their movement data 2019. Eight used although one did not cross adjacent highway. The remaining two neither nor crossed highway despite ranging vicinity. Their median speed significantly increased 0.65 km/hr from 0.45 before crossing railway, then slowed 0.32 after crossing. Females family groups moved faster than lone bulls when using underpasses. Seventy-eight per cent all crossings made were at night. fast speeds nocturnal patterns are behavioural responses risky landscapes or under stress. Disturbance vehicles traffic on developed human settlements may have limited Wildlife structures, signage bumps highway; relocation illegal settlements; inter-agency coordination requisites for enhancing Tsavos' elephant habitat connectivity. Les réseaux de transport peuvent représenter un obstacle majeur aux mouvements la faune. Nous avons évalué l’utilisation passages souterrains et ponceaux pour faune le long d’une voie ferrée récemment construite dans les parcs nationaux Tsavo, au Kenya, par éléphants d’Afrique équipé dix avec des émetteurs rayon inférieur à en mars analysé leurs données mouvement jusqu'en Huit ont emprunté bien que l'un d'eux n'ait pas traversé l'autoroute adjacente. deux autres n'ont ni malgré distances proximité. Leur vitesse médiane augmenté manière significative 0,45 km/h 0,65 avant qu’ils ne traversent ferrée, puis diminué atteindre 0,32 après traversée. femelles groupes familiaux se déplaçaient plus vite mâles adultes solitaires lorsqu'elles utilisaient souterrains. Soixante-dix-huit toutes traversées effectuées sont déroulées nuit. vitesses rapides schémas nocturnes réponses comportementales environnements risque ou situations L’utilisation peut être relativement limitée zones perturbations causées circulation véhicules sur route adjacente établissements humains développés. Des structures passage faune, une signalisation adéquate, mise place dos d'ânes l'autoroute, déplacement illégaux interinstitutions est nécessaire améliorer connectivité l'habitat Tsavos. Linear infrastructure developments main activities that threaten biodiversity (e.g. Sala al., 2000; Sanderson 2002). They drive loss fragmentation, alter ecosystem dynamics facilitate invasions exotic species (Fahrig & Rytwinski, 2009). These infrastructures interfere animal (Eftestøl 2014; Frair 2005), lead changes distribution, resource selection population density Fahrig 2009; Zeller 2019) increase mortality through vehicle collisions (Neumann 2012; 2018). Limiting dispersal individuals isolated patches habitats also reduction gene flow among populations (Rico detrimental impacts set as roads, rails urban areas continue expand rapidly each year (Dulac, 2013; Seto 2012). greatest impact will occur primarily tropics where there high levels pristine ecosystems (Laurance ,2009, Torres 2016). Elephants particularly sensitive disturbances due wide-ranging nature. is non-random (Loarie Wittemyer 2008) driven need resources such food, water minerals (Chamaille-Jammes 2007; Harris 2008; Murwira Skidmore, 2005; 2007). travel vast scarce spatio-temporal behaviour mostly related vegetation greenness (Bohrer Bolger Cushman 2005). In densities, adopt risk avoidance strategies travelling night moving these areas, increasing busy roads (Blake Douglas-Hamilton Graham Research increasingly highlighting linear act barriers movement. Movement 28 Congo basin showed only single individual road during entire study period, this once 2008). A similar Central Africa all, some studied would even range close (Granados observed Malaysia reduced over 70% (Wadey Nonetheless, attracted several ecological factors structure food availability roadside Kenya country rich 35,000 flora fauna across varied marine, mountains, grasslands, forests savannahs (Ojwang 2017). However, faces significant challenges balancing infrastructural development conservation. There ongoing efforts government promote Vision 2030 programme launched 2008 (GoK, This aims help transform into middle-income industrialised economy 2030. One flagship projects was new 487 Standard Gauge Railway stretching Mombasa Nairobi cities. Its construction took between 2014 2017 without doubt most project since building original metre-gauge early 20th century. section (133 km) raised an embankment fenced either side. Thus, cutting key Conservation Area. poses conservation challenge provisions railway. study, we effectiveness enabling connectivity africana) tested hypothesis railway's analysing tracking elephants. Specifically, looked at: (1) periods (i.e. differences before, and; (2) temporal seasonal group levels. Our results guide future management decisions development. site encompassed cuts Area (TCA), largest protected area complex covering ca. 42,000 km2. three area; East, West Chyulu Hills number large community-groups ranches, which important (Figure 1). TCA 70 250 Kenyan coastline altitude ranges 200 1000 m. comprises semi-arid bushland acacia-savannah forest home approximately 12,843 (Ngene 2017) 16,681 (Lamprey 2020). Rainfall vary considerably month month. general wet seasons October December May average monthly rainfall 75 mm. rest months generally dry hot 15 mm (Leuthold Leuthold, 1975; Tyrrell Coe, 1974). communities living ranches practice nomadic pastoralism, sedentary livestock keeping small-scale farming. Commercial occurs Galana Kulalu. Small-scale farming dominates outside large-scale west central parts ecosystem. bound Northwest densely populated Ukambani; Southwest Kilimanjaro, Pare Usambara mountains; Southeast moderately coastal hinterland. borders Mkomazi Park Tanzania Southeast, trans-frontiers movements region. (except crossings) fenced-off mitigate collision. drainage 2–7 m Bachuma Mtito Andei stations. decision fence off 112 try reduce encroachment mainly cattle grazing. largely unusable people. six (70 6 high) connect East Park. addition underpasses, nine bridges available use. include the: river bridge (1980 × 9 m); Kenani rail inter-crossing (520 x 10 Kanga (210 12 Maungu (180 7.3 Manyani (20 6.5 Ndii oil pipeline (25 6.8 (60 1 4 2 (95 4.7 m) 2a). resulted being positioned every 9.5 km. placements guided best practices elsewhere FHWA, 2011) joint assessment report (KWS 2014) highlighted sections biological relevance. spacing equal determined engineering requirements Next five other types mega-infrastructure: old 100 years old; Mombasa-Nairobi highway, marks border Parks. Along no volume speed; (3) voltage power gridlines; (4) pipelines (5) Mzima pipeline. crisscross places 2b). left 10% animals corridor posing both safety challenges. For example, according Service's unpublished data, 32 train/vehicle–elephant recorded 2015 2017. occurred Mombasa–Nairobi panicked got trapped embankment, 256% total 2012 began. Three vehicle-elephant zero train 2018 July 2020. 67% decline compared period. Similar declining trends train/vehicle became operational data). At least fatalities, injuries economic losses linked reported collisions. Ten collars fitted adult male female Group size ranged 16 individuals. specifically (not included study) another study. order our sample size, additional who (Table deployed AWT (African Tracking) (https://awt.co.za/) while Henrik GL200 (http://www.savannahtracking.com/). All acquire fixes 60-min interval. Selection capture locations designed maximise chances sides enable us sexual dry/wet transitional March. sampling side railway) collaring >1500 km2 historical Sale, 1973). Capture handling followed procedures described KWS (2016), Okita-Ouma al. (2016) (2008). Elephant retrieved ‘STE Downloader’ programme. Downloader connects ESRI service internet protocol. Here transferred via Internet updates locally stored geodatabase server Animal Tracking database (Wall, 2015). interface lets users conduct advanced filtering operations extract further analyses. collected therefore filtered upper, biologically based threshold 7 (elephant hourly do exceed glean out erroneous caused error. spatial analyses map creation carried ArcMap projected Universal Transverse Mercator (UTM) WGS-84 reference system (Zone 37S). tracks visually explored determine identify potential those cross. Dry long-term already downloaded paths calculated, complete straight-line throughout Further manipulation 10.5.1 (Esri Inc. ArcGIS 10.5.1. Redlands, CA: Esri Individual trajectories extracted query tools ArcMap. Shapiro–Wilk tests normally distributed non-parametric paired Wilcoxon signed rank test performed speeds. Variations males females sum test, Kruskal–Wallis test. Statistical R 3.6.3 /RStudio (R Core Team, times segment midpoint data. time date, year. Specific computed VB Script field calculator aligned sunrise, sunset zenith. Additionally, examined possible relation calculating frequency (January–February June–September) (October–December March–May) track times. conducted examine season During three-year 176,184 (median = 25,670, 2,694–30,357) 315 uses designated structures. Ndara (M) Kamboyo (F) came less but NP. Visual observations hard boundaries seemed restrict 4a). NP 2). frequencies (36 month) June 2016, dominated (71%), Kanga, River Bridge. still construction. Unfortunately, poached 2016. sharp decrease <2 August Crossings suddenly 43 September 2017, drop seven More regular averaging thereafter up February 2019 3). her (the bridge). approached underpass (within 50 it She 85 4b, inset). probably confined 4c) having roamed 19 Water Bridge 2, 74 occasions totalling 18 hours, On numerous occasions, (F), Taita Rukinga Bull paced prior successful Sagalla Ngutuni sections. hr (W 0.83, df 943, p < 0.05). km/hr, respectively 5). hour higher (V 16236, Z ?5.3, 0.025, one-tailed) 15327, ?5.8, one-tailed). variation sexes 6). 0.26 1.36 difference (X2 46.957, 0.005, 7). Median 0.73 1.19 13394, highest midnight Figure A1) trains stop running. proportion daytime (54.3%), (24.1%). relatively evenly spread season, tended peak decreased sunrise diel hours different 0.868, 0.35, (40%) A2). (Ndara – (F)) culverts. 16, particular represented 30% infrastructures, them Previous studies shown move strongly influenced distribution 1973; Ojwang 2017; Wato highway's volumes transient restricted elephants’ Anecdotal information indicates never (A. Mwazo 2020 Pers. Comm). highlights avoid completely Granados 1:3 supporting assertion (Koskei 2018; more frequently they because favourable changes, plant moisture roadsides likely line crossings. completion barrier had been created leaving points explains low very few started run 31 noise machinery workmen could contributed period sudden regularised possibly learnt Low January. coincided human–elephant conflict cases community, located 1.5 kilometres crop raiding incidents previous (King leading uproars affected (KNA, Mkanyika, Over 500 community including (M), known raider, blocked accessing (K. Hellyer It 3 August/September completed start making bridges. positive result, China, Tibetan antelope (Pantholops hodgsonii) learn Qinghai–Tibet (Ruan Xia learning indictive elephant's intelligence has documented Mount began Nanyuki–Isiolo weeks its (G. Chege 2020, longer attributed wide construct down normal fastest respectively, 1.1 (2005) strip-like ‘safe’ areas. (0.65 km/hr) 0.7 (Ihwagi, 2019; Ihwagi (0.45 (0.32 little (Douglas-Hamilton Wall 2013). increases walking expressions behaviours, railways (Barrientos 2019). perceive access vital resources. travelled males. invest heavily offspring protective members. perceived females, quickly ensure members propensity risk-taking variance reproductive success, led pressures favouring (Chiyo 2011; King easily injury death reward highly nutritious gains body mating success 2011). perception does account willingness cross, physiological cost survival (Stevens 2006; Also, unknown inferred. supported results. found predominately night, strategy, active day hours. Travelling poaching (Ihwagi exception called Maungu, day. numbers theory she (3–40 individuals). gradually block if managed quickly. monitor any change Maungu's utilisation strategy future. crossings, season. balance search food/water meet energy demands, avoiding ambient temperatures risks seasons, favourable. Therefore, nighttime rather cool. explain slight skew towards part (sunset midnight), whereas sunrise). When low, utilisation, black bears (Ursus americanus) (Zeller today's society. comes environment. faced needs Trying find compromise requires systematic research partnerships conservationists developers. lend support Management Plan, emphasises stricter limits effective methods managing limiting (KWS, progressive retrofit bumps, warning signs wayleaves; sustaining novel first investigate connecting Africa. Understanding animal's crucial planning sustainable Thirty Crossing clearly elephants, patterns. must gain take risk. isolate workers initial phases limitations stem lack accounting forms vicinity, recommend collect immediate implications TCA. dynamics, migration home-range patterns, sources (Ito Olson van der Ree, 2015; Ruan (Yu 2017), bottlenecks life-history adaptation calf (Bolger 2008), deterioration health (See review Iosif (2012) Rautsaw (2018) human–wildlife co-existence strategies. findings present opportunity design develop national heritage intact. Partnerships developers lose chance influence how work together. Director General Service (KWS) CEO Save (STE) thanked authorising collaborative research. Many staff departments KWS, STE, Trust, Works Zoological Society London participated Jeremiah Poghon Veterinary unit Wainaina Kimani STE logistical ensured smooth exercise. Chris Thouless Frank Pope provided useful comments drafting article. Alex Dennis Kibara helped collection. Patrick Omondi organised presentation author senior publication. financial technical assistance partnership Trust. anonymous reviewers improved authors declare interest work. BO-O, LK ID-H initially thought idea effects coordinated FL RM. MK curated monitoring system, produced graphs, maps undertook statistical consultation RA LT. BO-O writing manuscript contributions MK, guidance input alongside LK, RM FL. read approved final version manuscript. Findings, opinions conclusion recommendation expressed material author(s) funding agencies accept liability regard. request Elephants. Data especially shared permission sensitivity surrounding information.