Biomechanical properties of fishing lines of the glowworm Arachnocampa luminosa (Diptera; Keroplatidae)

Animals use adhesive secretions in highly diverse ways, such as for settlement, egg anchorage, mating, active or passive defence, etc. One of the most interesting functions is the use of bioadhesives to capture prey, as the bonding has to be performed within milliseconds and often under unfavourable conditions. While much is understood about the adhesive and biomechanical properties of the threads of other hunters such as spiders, barely anything is documented about those of the New Zealand glowworm Arachnocampa luminosa. We analysed tensile properties of the fishing lines of the New Zealand glowworm Arachnocampa luminosa under natural and dry conditions and measured their adhesion energy to different surfaces. The capture system of A. luminosa is highly adapted to the prevailing conditions (13–15 °C, relative humidity of 98%) whereby the wet fishing lines only show a bonding ability at high relative humidity (>80%) with a mean adhesive energy from 20–45 N/m and a stronger adhesion to polar surfaces. Wet threads show a slightly higher breaking strain value than dried threads, whereas the tensile strength of wet threads was much lower. The analyses show that breaking stress and strain values in Arachnocampa luminosa were very low in comparison to related Arachnocampa species and spider silk threads but exhibit much higher adhesion energy values. While the mechanical differences between the threads of various Arachnocampa species might be consequence of the different sampling and handling of the threads prior to the tests, differences to spiders could be explained by habitat differences and differences in the material ultrastructure. Orb web spiders produce viscid silk consisting of β-pleated sheets, whereas Arachnocampa has cross-β–sheet crystallites within its silk. As a functional explanation, the low tear strength for A. luminosa comprises a safety mechanism and ensures the entire nest is not pulled down by prey which is too heavy.


Material and Methods
All collections and measurements other than the thread length observations were carried out from October to December 2014 within the Spellbound Cave (S38°19′549; E175°04′454), Waitomo, New Zealand 17 . This is a commercial tourist cave with a wide and easily-accessible entrance and with a large open stream that enters it.
Tensile and peeling strength measurements were taken using newly collected fishing lines in the cave (temperature 13-15 °C, relative humidity RH = 98%) at night after the tourist visits. Measurements on dried threads were taken during daytime in the laboratory (temperature 20 °C, ≈60% RH) in the Waitomo Cave Discovery Centre. Thread length under drying conditions was carried out on Arachnocampa flava at the University of Queensland, Australia.
Instrumental set-up. We constructed a simple, light-weight apparatus for taking strength measurements of fishing lines (Fig. 1), designed for easy transportation and installation at night in the cave. It was equipped with a 10 gram force transducer (Model FORT 10 g, World Precision Instruments, USA) above a motorized stage (M-168 Linear Actuator Stepper, Physik Instrumente GmbH, Germany). Both instruments were placed on a Swiss Boy lab jack, and the force transducer was mounted on an XY positioning table (KT2304-KL-PT1104, MM Engineering GmbH, Germany). The apparatus was completely enclosed within an acrylic glass cover to protect it against damage during transport and dripping water from the cave ceiling. It also reduced air movement during measurements. Precise alignment of both the apparatus and the force transducer in all directions was verified using a spirit level on the XY table.
Force measurements were made using an MP 100 data acquisition system (BIOPAC Systems Inc., USA), connected to a notebook computer and controlled by the software AcqKnowledge 3.7. A 12 V car battery was used as a power source for the linear motor and the BIOPAC data acquisition system, when taking measurements within the cave. Measurements in the laboratory used mains electricity.
Instrumental settings and calibration. The signal was fed into the digitizing system using a gain of 200 and acquisition rate was set to 5 kHz. The 10 g force transducer was adjusted with a standard comprising a pre-weighted (750 mg) piece of wire. This was weighed periodically to check that it remained clean and that its www.nature.com/scientificreports www.nature.com/scientificreports/ mass remained constant. The force transducer was calibrated at both the start, and rechecked and recalibrated at the end of each individual measurement.
Tensile strength measurements. Tensile strength tests were conducted with fishing lines fastened between a metal bar (diameter 1.51 mm, length 32.6 mm) which was screwed into the end of the force transducer (Fig. 2) and a thicker steel pin (diameter 3.45 mm, length 11.32 mm) screwed to the linear motor. The distance between the upper bar and lower pin was always set to 15 mm.
Only a single fishing line was collected from a glowworm nest, additional fishing lines were always collected from different nests. This minimized damage to nests and provided better statistical replicates than using multiple fishing lines from the same nest. Long single threads were detached from glowworm nests without destroying the hammocks and fastened to the transducer bar by wrapping them around it several times. The threads were then attached to the lower pin, by wrapping them around it and securing with a magnet. On the upper wire, no magnet was attached on the threads as the magnet appeared to be too heavy for the analyses. When measurements were taken using three fishing lines (subsequently named "triplets") of the same length and from the same nest, they were twisted by slow rotation around their axis and then fixed as bundle within the instrument as described above for the single thread.  Once fixed to the apparatus, tensile measurements were made after 30 s to allow the transducer and the threads to stabilize, then the lower pin was locked to the motor which then pulled at a constant speed of 0.1 mm/s until the thread broke off. The strain on the threads was calculated after recording the position of the motor at the start and when the threads broke. The transducer wire and motor pin were carefully cleaned with paper towels before new threads were attached.
Peeling strength measurements. Peeling strength was measured using the same force transducer and motorized stage as above, except that a u-shaped aluminium frame was attached to the transducer. A 16 mm gap in the frame allowed a probe to make contact with fishing lines when laid across the gap (Fig. 3A). Five different probes were used comprising: (i) a stainless steel metal bar (∅ 3.5 mm), (ii) a plain, epoxy resin block 3.4 mm wide (Agar 100 Resin, Agar Scientific, UK) (Fig. 3B), (iii) an epoxy-cast mold of the abdomen of the larval Zophobas (Coleoptera, Tenebrionidae) (Fig. 3C), (iv) a plain, PVS (polyvinyl siloxane) block 3 mm wide (Coltene/ Whaledent AG, Switzerland) ( Fig. 3D) and (v) the abdomen of a dead larval Zophobas. Each was attached with super glue to a Hitachi screwable stub. These probes were employed one at a time after screwing them into the lower motorized stage.
Newly collected fishing lines (singular or triplet) were wrapped around the two uprights of the frame, then the frame was screwed to the force transducer. Measurements were then taken after 30 s as for tensile testing above. After one of the five different probes were brought into contact with the fishing lines that were orientated horizontally within the u-shaped frame. The probe was then pulled down at a constant speed of 0.1 mm/sec, until contact was broken (Suppl. Video 1). As for the tensile strength measurements, the frame and probes were cleaned with paper towels and distilled water and dried after each measurement.
Each measurement of adhesion strength was simultaneously recorded with a video camera (SONY, Handycam HDR-SR5E). Afterwards still images were extracted from the recorder information with the free software Avidemux Version 2.6.12 were used to measure both thread lengths and contact angle on the different probes shortly before detachment, using the microscopic software cell D (Version 5.1, Olympus, Austria). The size of the u-shaped frame was used in all still images as a reference. www.nature.com/scientificreports www.nature.com/scientificreports/ Tensile and peeling strength tests with dry threads. Single or triplet fishing lines were collected as above for both tensile and peeling strength tests and fixed to the same wire or frame as above (Fig. 4). They were then placed in a desiccation cabinet at 24 °C and 60% RH filled with dry calcium chloride for a minimum of 24 h. Measurements of tensile and peeling strength was made as above in the laboratory (24 °C, 60% RH) immediately after removing the threads from the desiccation cabinet.
Thread length under desiccation. Larvae of Arachnocampa flava taken from the field were kept in the laboratory in artificial habitats carved from moistened phenolic foam blocks (floral foam) and stored in an aquarium with water covering the base. Individual threads were detached from a larval snare using fine forceps and mounted on a stand placed in a glass aquarium. The thread was imaged using a Sony A7S camera and 100 mm Canon macro lens, taking a photograph (4240 × 2384 pixels) every 20 sec. A ruler was photographed after each run to provide a scale reference. As soon as the thread was installed, the aquarium was sealed and dry ambient air was introduced into the aquarium through a plastic tube terminating in a bubbler stone to reduce turbulence. The air was dried by pumping it through a chamber filled with silica gel desiccating beads. For analysis, thread length was measured from the individual images using ImageJ.

Statistical evaluation.
Statistical evaluation of all peeling strength data and thread strength data were performed in MS Excel and Stata 14.2 and presented as boxplot graphs.
The skewness-curtosis test was used to ensure that the data did not differ significantly from normal distributions (P ≥ 0.05) before further analyses. Differences between two or multiple samples were tested with the unpaired t-test and one-way ANOVA tests followed by the post hoc Bonferroni test were used for multiple samples, P < 0.05 was regarded as significant, P < 0.01 as highly significant.
Calculation of the tensile strength were made using a mean thread diameter of 8 µm based on morphological data 17 and a local gravity value of 9.80 m/s 2 for Waitomo, New Zealand (based on the Gravity Information System of the Physikalisch-Technische Bundesanstalt Germany; https://www.ptb.de/cartoweb3/SISproject.php).
Measuring surface wettability of probes. The polar and non-polar components of the surface energy of all five probes (metal bar, resin block, epoxy Zophobas larva mould, PVS block, dead Zophobas larva), was measured using the 'sessile drop' method 34 with a OCAH 200 high-speed contact angle measuring system (DataPhysics Instruments GmbH, Filderstadt, Germany) equipped with silanized glass capillaries (tip diameter 0.15 mm) and two polar (distilled water and ethylene glycol) and one non-polar liquid (diiodomethane). Each measurement was taken after depositing 2 µl of one of the three fluids onto the probe and taking an image of the droplet once its shape had stablized. This was repeated at least five times (Table 1) with the contact angle being measured using the software SCA20 (DataPhysics Instruments). Measurement were taken under ambient conditions (25 °C and 60% RH).
Adhesion energy estimation. The adhesion energy in the peeling experiments was estimated from the energy balance at detachment of an incompressible neo-Hookean thread from a substrate in a way similar to that previously used by Kendall 35 . The line pull-off from the substrate takes place when the energy balance at its infinitesimal release from substrate Δz, (Fig. 5), is negative:

el s
where E el and E s are elastic and surface energies. We assume: where W is the adhesion energy, d is the contact width (it was assumed to be equal to the fishing line diameter, 8 µm). The elastic energy of a neo-Hookean thread was taken equal to Odgen 36 :  www.nature.com/scientificreports www.nature.com/scientificreports/ where μ is a Lamé coefficient (shear modulus), l 0 is the initial length, A 0 is the initial cross-section area. For the extension ratio λ, under assumption of an infinite friction between the thread and substrate, for experimental scheme shown in Fig. 5 we can write: where ϕ is a peeling angle. The change in elastic energy at infinitesimal pull-off is then given by . Taking into account Eq. 3 for Δz → 0, neglecting the terms with Δz k for k > 1, after simplification, one obtains: . The frame pulling force equals to: where α is a substrate slope at the peeling position and the traction force equals to Odgen 36 : , and from Δ = Δ E E el s we obtain the following expression for the adhesion energy: Additional data. Additionally, three individuals of Arachnocampa luminosa (one with a length around 20 mm and two with a length of 40 mm) and 100 individuals of the midge Anatopynia debilis, the main food source of Arachnocampa luminosa in Spellbound, were collected and its wet weight measured individually.

Results
Strain. Newly collected fishing lines tend to have not statistically significant higher strain (0.47) than dry ones (0.40) when pulled vertically during the tensile strength tests (Fig. 6A). In peeling experiments, wet threads show slightly, but not significantly higher, extensibility on all probes (0.55-0.69) than on metal bar (0.41) (Fig. 6B). Statistically significant differences could only be observed between the stainless steel metal bar and the resin block or the epoxy Zophobas larva mold (Table S1).  www.nature.com/scientificreports www.nature.com/scientificreports/ Tensile strength. Preliminary tests showed that newly collected fishing lines did not adhere well to the metal bar and had to be wrapped several times around it to ensure reasonable "clamping". However, these lines could not be wrapped around the lower bar without stretching them unintentionally so they were anchored using a magnet.
Measurements with single newly collected fishing lines gave a tensile strength of 1.9 MPa (maximum force = 0.1 mN) (Fig. 7) that should be sufficient to catch and support one or more prey items as Anatopynia debilis (weight between 1 and 10 mg). Although not tested in this study, it could be assumed that the thread type used to attach the "hammock" to the rocky substrate displays higher tensile strength values than the fishing lines used for prey capture. Weight measurements yield for the 20 mm long larva 30 mg and for the larger, 40 mm long ones a weight of 100 mg, ten times heavier than the maximum load derived from the tensile strength values for a single fishing line.
Triplet-combined fishing lines only doubled the tensile strength to 3.5 MPa (maximum force = 0.17 mN), indicating that these lines interact slightly and do not break simultaneously. However, as the moving prey during catch often comes in contact with several fishing lines and tangles them up, this multi-thread effect would provide weight relief for each thread and support larger or more prey items.
Time-lapse imaging of single threads freshly collected (Fig. 8A) and dried (Fig. 8B) showing that the lines shorten and the droplets leave a residue as they evaporate. When threads were removed from a snare and immediately exposed to dry air, they shortened to approx. 80% after 60 min and 59.1% of their original lengths after 100 min, respectively (Fig. 8C). As the threads dry and lose droplet mass, they tend to drift in the slightest air movement and could possibly become statically charged.
Dried fishing lines have a much higher tensile strength than wet ones, e.g., 4.6 MPa (maximum force = 0.23 mN) for single fishing lines and 9.8 MPa (maximum force = 0.49 mN) for three combined (P < 0.001) (Fig. 7B). Moreover, drying of the fishing lines under low humidity (<80% RH) resulted in dehydration of the droplet 17 and only a small remnant comprising the central core remained visible. The dried fishing lines also appeared stiffer  Table S1 for details).
www.nature.com/scientificreports www.nature.com/scientificreports/ and could not be smoothly wrapped around the upper bar in the same way as fresh threads. The silk thread of dry fishing lines also showed no adhesion to the probe holder, whereas the dry remnants of the droplets attached firmly to the metal surfaces ( Fig. 9) (Table S2). Statistically significant differences for the metal bar vs. the Zophobas mold (a: P = 0.007) and the PVS block vs.  In general, threads (singular or triplet arranged) dried in the desiccator cabinet (60% RH, 20 °C) display higher tensile strength values than the wet ones, however show no bonding ability. The single wet threads are able to hold the load up to two average prey items (mean weight of Anatopynia debilis = 5 mg), while more/or heavier preys could be held by tangled up threads. Differences between tensile strength for the triple-arranged threads (wet/dry) and the single ones (wet/dry) are highly significant (P < 0.001).  www.nature.com/scientificreports www.nature.com/scientificreports/ was about the triple (70.6 N/m) than that for a single fishing line (24.6 N/m). Zophobas larva probe also showed the widest range of adhesion energies from 20 N/m to 225 N/m. The calculated adhesion energy was statistically different between the metal bar and resin block (P = 0.001), the metal bar and Zophobas larva mold (P < 0.001) as well as the metal bar and native Zophobas (P = 0.001) (Table S2).
Peeling strength experiments could not be carried out on dry fishing lines as they showed no adhesion to any of the different probes.

Discussion
This study provides the first description of the biomechanical properties of the fishing lines of Arachnocampa luminosa larvae in situ and details on their bonding strength to materials with different surface energies.
The collected fishing line of A. luminosa displays a slightly, but not significantly, higher extensibility than dry threads; visual observations confirm that dehydration also results in a loss of flexibility, as mentioned previously 17,31,33 .
Comparison with data given for the fishing lines in Arachnocampa tasmaniensis show large differences in mechanical properties. The tensile strength in A. tasmaniensis is much higher (122.77 MPa at 90% RH; 159.04 at 30% RH) 33 than that measured for A. luminosa (from 1.9 MPa at RH = 100% to 4.6 MPa at RH = 60%). Also, the extensibility of the fishing lines is slightly higher under highly humid conditions (A. tasmaniensis = 0.6; A. luminosa = 0.47). Interestingly, at 60% RH, the threads still show an ability for extension (A. luminosa = 0.40), while at 30% RH it is rather low (A. tasmaniensis = 0.02) 33 .
The question arises, whether these measured differences are a consequence of the different sampling and handling of the threads prior to the tests. Fishing lines from A. tasmaniensis were affixed for air transport with wood working glue on a paper frame, stored for a long time under dry conditions and segments of these samples were re-hydrated for a short period at 90% RH before the test series 33 . The authors, however, assure that their sample handling had no influence on the mechanical properties 33 . In the present study in A. luminosa, in contrast, long freshly-collected segments (>4 cm) of the A. luminosa fishing lines were measured in situ with minimal handling and mechanical stress before the measurement.  www.nature.com/scientificreports www.nature.com/scientificreports/ However, it could also not be excluded that the silk composition in both species differs, leading to different mechanical properties. The element composition of the fishing line of A. tasmaniensis seems to contain a higher amount of sulphur beside carbon and calcium in relation to those measured for Arachnocampa luminosa 17 and quantified for A. richardsae 31 . Disulfide bonds are known to link in Bombyx mori silk two different fibroin chains (the hydrophobic heavy and the elastic hydrophilic light chain) 37,38 necessary to form fibroin in a normal ratio for the cocoon production 37 . Further studies are therefore planned to quantify the amino acid composition in both Arachnocampa species in detail and repeat the biomechanical tests in A. tasmaniensis as done for A. luminosa to confirm different mechanical properties as a consequence of its molecular composition.
In this regard, the Arachnocampa fishing line silk has also very different mechanical properties to the egg stalk threads of the green lacewing Chrysopa carnea, those silk also exhibits a cross-β-sheet structure. The egg stalks have much higher extensibility (4-6 at RH = 100% and 2.5 at RH = 65% 39,40 ) and tensile strength (232 MPa at RH = 100% to 70 MPa at RH = 30%) 39 than A. luminosa (present study) and Arachnocampa tasmaniensis 33 , probably because "glow-worm silk is likely to have a substantially lower crystalline fraction, compared to either silkworm silk or lacewing egg-stalks" 31 . However, the differences in the strain and tensile strength between hydrated and dry C. carnea egg stalks presumes that the increased extensibility and strength is largely influenced by the hydrogen bonds between the cross-β strands which results in a rearrangement of the cross-β strands into a parallel β-sheet conformation 39 with increasing humidity. In representatives of Arachnocampa, the threads are fully saturated with water and deformation is effected by "the amorphous fraction, or due to deformation of both the amorphous fraction and the cross-β-sheet crystallites" 31 . Absence of the hydrogen bond donor may hinder this rearrangement, leading to higher tensile strength and lower extensibility.
Meyer-Rochow 13 reported that glowworm fishing lines of the same diameter were capable of supporting a weight of 15 mg before they snapped, which agrees with our measurements. These results indicate that A. luminosa fishing lines perform poorly in terms of tensile strength. In orb web spiders, a tensile strength up to 1.1 GPa was reported for major ampullate silk and 0.5 GPa for viscid silk 42 . Also, the cocoon silk of Bombyx mori (0.6 GPa) 42 and the byssus thread of mussels  MPa, proximal to distal) 44 are much stronger and somewhat tougher. The majority of other biomaterials, such as tendon collagen (0.12 GPa) 49 , bone (0.16 GPa) 42 , or artificial materials, such as nylon fibre (0.95 GPa) 42 , Kevlar (3.6 GPa) 50 and carbon fibre (4 GPa) 50 , also have higher tensile strength properties than those of A. luminosa silk.
The reason for such relatively low extensibility and tensile strength of A. luminosa fishing lines could have a functional explanation, related to their prey and habitat. For example, orb web spiders require a web with high extensibility and energy-absorbance properties, able to resist the impact of any prey hitting the web at high velocity, be it small flies, heavy beetles or grasshoppers, and, at the same time, supporting the mass of the spider 51 . Moreover, spider web also has to withstand rapidly changing climate conditions (rain, winds, humidity and temperature fluctuations, UV-radiation) without breaking 51 . On the other hand, glowworms build their nests directly above rivers in caves (see Fig. 1 in 17 ) or under overhangs in sheltered humid environments such as adjacent to streams and these exhibit relatively minor varying climate conditions throughout the year 26 . Their prey comprises a high proportion of small flying dipterans (89%), about 1-4 mm long 25 and of relatively low mass as reported here for the prey Anatopynia debilis. Large heavy arthropod prey is rarely caught by A. luminosa 25 . As a consequence of the degree of prey specialization, glowworms may not necessarily need silk that is as resilient as those of orb web spiders.
It is also conceivable that the low strain and tensile strength values of the A. luminosa fishing lines serve as a safety mechanism. The web structure might hold prey that are small enough to be overcome and eaten, whereas larger prey that could destroy the snare and damage the larva readily break free. Also, it may limit the number of prey items collected and their total load, thereby preventing the nest from being detached from the substratum. Glowworms do occasionally fall from nests and survive if they can find an overhang to produce a new nest 13 .
While A. luminosa fishing line silk shows a low tensile strength compared to other silk-producing animals, its adhesive properties exhibit much higher values than those of A. tasmaniensis 33 and for example orb web spiders ( Table 2). In the present study, adhesive droplets of A. luminosa have a mean adhesion energy of 20-45 N/m, while the values for A. tasmaniensis and different spider species range from 0.02 N/m to 0.1 N/m 33,52-56 . The adhesive droplets of orb web spiders show similar bonding ability at both high (100%) and low (≈60%) relative humidity levels, while bonding in A. luminosa is highly dependent on humidity levels. Other biological adhesives have weaker (e.g. 2.5-17.5 N/m for the ascidian Botrylloides sp.) 57 or stronger surface energy values (97 N/m for Crucifix Toad Notaden bennettii) 58 than A. luminosa droplets, but the latter is comparable with some commercial products such as fibrin (20 N/m) and gelatin-resorcinol-formaldehyde (39 N/m) 58 . The bonding abilities of some biological adhesives also need to be treated with caution depending on the substrates the adhesives were tested with (e.g. sandpaper, glass, animal tissue) as the substrates can greatly influence adhesion energy.
The present investigation confirms an earlier observation that high humidity (>80% RH) is essential for A. luminosa to catch preys with its fishing lines 17 . Low humidity not only affects the properties of the silk fishing lines, but it also leads to dehydration of the adhesive droplets even when re-hydrated as in the case of A. tasmaniensis 33 . Such water loss, (water normally comprises 98% of droplets) 43 reduces the bonding ability of the glue, making it impossible to catch prey. Adhesion by the central droplet core after dehydration, which is presumably a gluing component, only becomes possible when the fishing line is under artificially produced tension. Under desiccating conditions, threads shorten and wave about more readily, whereas wet lines with large adhesive droplets are more resistant to tangling and maintain their sheet-like spacing. The droplets, however, are hygroscopic and can rehydrate at high humidity 17 . Thus, the prey capture system of A. luminosa can recover after a dry spell to a