Mechanical loading of primate fingers on vertical rock surfaces

AFFILIATIONS: 1Department of Anthropology, Dartmouth College, Hanover, New Hampshire, USA 2Department of Archaeology, Simon Fraser University, Burnaby, British Columbia, Canada 3Centre for the Exploration of the Deep Human Journey, University of the Witwatersrand, Johannesburg, South Africa 4Mammal Research Institute, University of Pretoria, Pretoria, South Africa 5Department of Anthropology, University of Pennsylvania, Philadelphia, Pennsylvania, USA 6Skeletal Biology Research Centre, School of Anthropology and Conservation, University of Kent, Canterbury, UK 7Evolutionary Studies Institute, University of the Witwatersrand, Johannesburg, South Africa


Introduction
Bone is a dynamic tissue, and repeated mechanical loading can induce changes to its density and angular curvature. For example, loading stresses from vertical climbing and suspensory locomotion can increase the curvature of primate finger bones (phalanges) during growth and development. [1][2][3] Greater curvature is advantageous to arboreal primates because it dissipates forces and dampens recruitment of muscles during finger flexion, thus economising energetic costs and simultaneously lowering the risk of falling. Phalangeal curvature is therefore widely viewed as a measure of arboreality among primates -it is a classic form-functional trait that informs our interpretations of behaviour and ecology in the fossil record [1][2][3][4][5][6][7] (but see Wallace et al. 8

for a counterexample).
Upsetting this orthodoxy is the hand of Homo naledi, recovered from Rising Star Cave, South Africa, in 2013-14 and described in 2015. 9 The degree of phalangeal curvature is astounding, with an included angle (θ) that far exceeds that of modern humans ( Figure 1). Such curvature would normally and unequivocally suggest an arboreal milieu 3 , but there are at least two reasons to doubt such an inference for H. naledi. First, the phalanges are relatively short, resulting in humanlike hand proportions. 9 No arboreal primate has phalanges that are both short and curved; it is an aberration that confounds conventional interpretation. Second, existing specimens of H. naledi are dated to the Middle Pleistocene between 335 kya and 241 kya. 12 Nearby faunal assemblages of comparable age -those of Gladysvale 13 and Lincoln Cave, Sterkfontein 14 -favour habitat conditions that resemble those of today, i.e. a mix of acacia woodland and grassland, neither of which would incentivise a strongly arboreal ecology. 15 So, what was H. naledi doing with its hands? Voisin et al. 16 examined the shoulder girdle of H. naledi and reported morphological traits associated with vertical climbing and suspension. But given the inferred habitat conditions, they discounted arboreal activities and argued instead for 'movement across and climbing on rocky walls' 16(p.2) . Voisin et al.'s hypothesis is intriguing given that rock climbing puts high forces on finger flexor tendons, especially during the 'crimp' position ( Figure 2). [17][18][19][20] This position puts extreme stress on the flexor tendon pulley system, a series of five annular (A1-A5) and three cruciate ligaments (C1-C3) that resist bowstringing of the tendons. In other words, the pulley system holds the tendons close to the bone, effectively converting linear force into torque that produces flexion at the metacarpophalangeal and interphalangeal joints (Supplementary figure 1).
Ruptures of A2 17 and other severe pulley injuries are common among human rock climbers 21 , and a testament to the mechanical loading of phalanges. Indeed, frequent rock climbing is known to cause geometric and cortical thickening of the phalanges 22 and it may induce greater curvature, at least hypothetically. Tan et al. 18 calculated the benefits of greater phalangeal curvature for reducing tendon and pulley stresses of fingers in the crimp position; however, affirming increased curvature among rock climbers is difficult because it is challenging to measure angles from radiographs of living subjects. Another model system is needed. Some populations of baboons and chimpanzees enter cave systems to avoid predators and/or regulate body temperature. [23][24][25][26][27][28][29][30][31][32][33][34] Some caves are only accessible by climbing sheer cliffs. 28 For example, Marais 23 described a group of chacma baboons (Papio ursinus) and their daily ascent to a cliffside cave opening ≈150 m high. His vivid account speaks to the mechanical loading of fingers: 'they had to go more than a mile hanging only by their fingers to the [cliff] ledge; their hind feet against the smooth surface' 23(p.63) . In some cases, baboons must negotiate vertical rock surfaces within a cave. For example, at Misgrot Cave in Thabazimbi, South Africa, baboons face a perilous 17-m vertical descent from ground level ( Figure 3) 32 , whereas those entering Dronkvlei Cave, De Hoop Nature Reserve, South Africa, descend 7 m to access the interior ( Figure 4). Baboons use both caves regularly as overnight sleeping sites; Dronkvlei is used on 28% of nights. 30

Study aims and design
Dronkvlei Cave is an appealing model system for exploring the topography of a vertical rock surface and its potential to load primate fingers. Here we ask a basic question: what proportion of a vertical rock wall would, when climbed, force the distal phalanges of P. ursinus (and, hypothetically, H. naledi) into a crimp or slope grip position? It is a thought experiment that leaves formal comparative measures of phalangeal curvature in P. ursinus as a priority for future research. To answer our question, we used three-dimensional (3D) photogrammetry to detect and quantify concavities in the surface that range from a minimum graspable depth (determined as the length of the third distal phalanx) to a minimum crimpable depth (determined as the sum length of the third intermediate and distal phalanx) -i.e. depths of 11-25 mm for P. ursinus and 14-37 mm for H. naledi (Table 1).

Data acquisition
Photogrammetry is a low-cost method for extracting accurate measurements from photographs. It is well suited for producing 3D models of irregular surfaces, such as cave walls. 36 To obtain photographs of the rock surface climbed by baboons, we used metal irrigation piping to construct a temporary T-shaped scaffold. The cross pipe was mounted   at the cave opening and fixed to a vertical pipe that extended to the shaft floor (Figure 4a), where it was weighted to minimise lateral excursions. We hitched a digital camera to this pipe so that photography occurred from a fixed central axis, and we mounted ground control points to the rock surface. We abseiled into the cave to operate the camera, aiming for 80% overlap between serial photos in the horizontal and vertical planes.

Photogrammetry
We processed the image set (n=354) with Agisoft Metashape Pro (formerly Photoscan Pro), a common package used in archaeological research. [37][38][39] Due to logistical constraints in the field, we did not record spatial data for each ground control point, so the model was scaled to real-world dimensions by using the targets as scale bars, and manually oriented. Manual orientation is suboptimal for any calculations that rely on the slope of the wall, but it should not affect the local depth calculations described below. After alignment, dense cloud construction, and texture application, we trimmed the 3D model to the area used for climbing ( Figure 4b) and exported it as an orthophoto (Figure 5a) and digital elevation model (DEM) with an arbitrary local coordinate system. We brought the resulting geotiffs, both with sub-millimetre resolution, into ArcGIS and SAGA GIS for further processing and visualisation.

Depth processing
In ArcGIS, the DEM raster treats the surface of the rock wall as vertical elevation data. To calculate the local relative depth of each raster cell in the rock wall, we used focal statistics with a small neighborhood, and maximum elevation as the statistics type, to create a raster of the local maximum height of the wall, and then smoothed that raster again with focal statistics over a wider neighborhood, creating a smoothed idealised model of the maximum height of the local surface of the cave. We subtracted the original DEM from this idealised surface to find the local difference between each cell and the average nearest 'maximum' value. Finally, we queried this raster of local depths for cells in two ranges (11-25 mm for P. ursinus and 14-37 mm for H. naledi) and used this value to calculate the total area of each range.

Results and discussion
The entrance to Dronkvlei Cave is a useful model for exploring potential stresses on the flexor tendon pulley system and phalanges of primates.
We focused our analysis on the surface area climbed by baboonsrepresenting 5.98 m 2 (Figure 5a) -and we show that crimp-and slopeinducing handholds are distributed relatively evenly across a nearly vertical plane (Figure 5b). However, the total area of such holds differed between the two species.  (Figure 3c). Thus, 6-8% of the total surface area, when gripped, is expected to generate high forces on the flexor tendon pulley system and severely load the phalanges of both primate species. Such an estimate is conservative, however, as a large proportion of the rock surface is smooth and unclimbable. When we subtract surface depths of <11 mm and <14 mm from the total, the total graspable surface area is reduced to 0.602 m 2 and 0.390 m 2 , respectively, meaning that as much as 82% (P. ursinus) or 91% (H. naledi) of usable handholds would produce forces that would favour greater phalangeal curvature.
The significance of our finding is a matter of speculation -even at 8% of the total surface area, the rock wall of Dronkvlei Cave is a non-trivial source of phalangeal loading when extrapolated over a lifetime. The average life expectancy of a female baboon that survives to adulthood is 12.1 years in Amboseli, Kenya, and 19.7 years in Gombe, Tanzania (with record longevities of 27 years at both sites). 40 If we use these parameters to estimate an average life expectancy of ≈16 years for P. ursinus, and given that animals climb 12 m to enter and exit Dronkvlei Cave on 28% of days 30 , an individual would encounter ≈1.6 km of crimp-and slopegrip-inducing surfaces over its lifetime [(16 years × 365 days × 0.28) × (12 m × 0.08 m)]. It is a crude estimate of mechanical 'loading' on the flexor tendon pulley system and distal phalanges, but it invites an analysis of phalangeal curvature in this population.
More instructive perhaps are the mummified/skeletal remains of baboons in Misgrot Cave. 34 These individuals climb greater distances than those at Dronkvlei (Figure 3b), but the frequency of overnight use is unknown. Still, each climbing bout carries great risk. Falling from heights as high as 17 m is likely to cause severe morbidity or outright mortality 15 , although the rugosity of the dolomitic surface may differ from the limestone surface of Dronkvlei Cave in crucial ways, affecting risk. Misgrot Cave is a promising site for studying the natural history of why, and the mechanics of how, baboons climb vertical rock surfaces, as well as the effects on their finger bones. Another promising study system lies in the lava tubes of Mount Suswa, Kenya, the site of an oft-cited analysis by Simons 24 , who described the skeletal remains of baboons (P. anubis) and leopards, sometimes commingled, at the base of vertical shafts. Remains of baboons continue to accumulate there, serving as a tourist attraction.
So, did H. naledi climb vertical rock surfaces regularly? Answering this question is beyond reach at present, but their bodies, alive or dead, negotiated at least two steep walls of dolomite to arrive in the Dinaledi Chamber of Rising Star Cave: an entry point through the roof of Postbox Chamber and a 12-m fissure ('The Chute'). 12 Our results suggest that climbing these surfaces would have compelled the use of crimp or slope grips to some extent, and Tan  A limitation of our model system approach is that H. naledi (weighing ~40 kg 41 ) was heavier than P. ursinus is now and probably longerlived, factors that would load its fingers to a greater extent. Yet the pedal phalanges of H. naledi express the same curvature as those of Papio, which is '[possibly] indicative of elevated pedal grasping ability' 42(p.4) . Thus, both species are equipped to grip vertical rock surfaces with their feet -a prudent means of distributing body mass and mitigating forces on the fingers during climbing.
We conclude by acknowledging that rock-climbing baboons are an imperfect model for interpreting the curious anatomy of H. naledi, but we would argue that detailed studies of their functional anatomy and climbing kinematics, together with systematic measures of the rock surfaces used, are promising areas of future research.