Markerless tracking suggests a tactile sensing role for forelegs ofDolomedesspiders during locomotion

Abstract

statement: 11We developed a machine vision technique for markerless tracking of locomotion in the spider 12 Dolomedes aquaticus. Gait analysis suggests that each pair of legs plays a specific role in 13 locomotion. 14 15Abstract 16Because of their rigid exoskeleton with relatively simple joint mechanics, arthropods can 17 provide useful models for studying the sensory-neural and mechanical design principles of 18 agile animal locomotion. Gait analysis usually requires attaching markers or manually 19 identifying reference points in video frames, which can be time consuming and inaccurate, 20 especially with small animals. Here we describe a markerless motion capture technique and its 21 application to gait analysis in the New Zealand semi-aquatic hunting spider, Dolomedes 22 1985; Booster et al.prey strikes -Zeng and Crews, 67 2018). 68The present study focuses on the stepping patterns and kinematics involved with normal 69 spider locomotion. Similar analyses mentioned above generally used one of two approaches. 70Marker-based techniques, which involve the placement of markers (e.g. paint, nail polish, paper 71 or tape) onto the centre of a joint or tip of a limb. The marker is then tracked in video frames, 72 usually through the aid of commercial software packages. Alternatively, the researcher 73 manually digitized points of interest in each frame through the use of a stop-motion projector. 74Both approaches are laborious in nature and require careful attention, the first at the beginning 75 of the experiment to ensure the correct placement of markers and the second during processing 76 to accurately locate points of interest. These problems are compounded when working with 77 smaller animals, where positioning of markers, sensors or points of interest is more critical in 78 an accurate description of locomotor kinematics (Gambaretto and Corazza, 2009). This limits 79 the researcher's ability to examine locomotion, as markers can interfere with the animals 80 normal movement and behaviour (Bender et al., 2010). Additionally, although maker-based 81 approaches are widely accepted, they come with intrinsic problems e.g. ghost markers, 82 occlusions of markers, markers can move or become completely removed during locomotion 83 Materials and methods 126Animals 127The experimental procedure consisted of filming the locomotor patterns of unrestrained 128Dolomedes aquaticus Goyen. This species is commonly found near water on or under 129 surrounding objects such as plants, stones and pieces of wood. D. aquaticus sense insects and 130 small vertebrates both on water and land by detecting vibrations using the two front pairs of 131 legs (Forster and Forster, 1999). They then chase down their prey with rapid bursts of speed. 132Only relatively large, adult, female spiders possessing all eight legs were used during filming 133 (mass = 1.194 g  0.376; mean ± S.D.). 134 135