Webs from orb-weaving spiders (Tetragnatha) were collected in July 2015 at two sites in the Hawaiian archipelago: Waikamoi Nature Conservancy Preserve, Maui Island and Upper Waiakea Forest Reserve, Hawaii Island. Three species (Tetragnatha stela robusta Gillespie 1992, Tetragnatha trituberculata Gillespie 1992, and Tetragnatha hawaiensis Simon 1900) were commonly found at Waikamoi Preserve. Two other species (T. filiciphilia Gillespie 1992 and T. acuta Gillespie 1992) were found cohabiting in the same area but were not as abundant and thus were not included because of the low sample size. At Upper Waiakea Forest Reserve, only webs from T. hawaiensis were collected. With the assumption that no gene flow is occurring between Tetragnatha hawaiensis collected at Waikamoi Nature Conservancy Forest Preserve, Maui and T. hawaiensis collected at Upper Waiakea Forest Reserve, Hawaii Island (Roderick and Gillespie, 1998), these will be treated as distinct populations, respectively, called T. hawaiensis (Maui) and T. hawaiensis (Big Island) from this point forward. Although the Hawaiian Tetragnatha comprises more than these species, these were the only species found during fieldwork in these two locations. Dry weather during field season may have impacted the diversity and abundance of spiders found.

Hawaiian Tetragnatha are largely nocturnal and were located in the field from 20:00 until 02:00 hours. Whole webs were collected using plastic circle frames (15 cm in diameter) covered in double-sided tape. The frames were adhered to the webs, and the radial threads were then cut outside the frames using a butane lighter to retain the natural tension of the webs. On average, webs from the three species included in this study had diameters of 12–15 cm (Blackledge et al., 2003; Blackledge and Gillespie, 2004). When webs were larger than the circle frame, only the bottom half of the webs were collected because orb-web spiders can allocate different amounts of resources to upper and lower orb-halves (Blackledge et al., 2011). Webs were carried back to the field station using a waterproof box. Up to six samples of radii (major ampullate silk) and 20 samples of capture spiral were obtained from each web, suspended across 12.58 mm gaps in cardboard holders. Radial threads were secured with cyanoacrylate glue (Superglue©). Capture spiral threads were secured with water-soluble glue (Elmers©) because cyanoacrylate desiccates the glue droplets. Threads were kept in a dark box to prevent damage from UV light (Lai and Goh, 2015) until they were shipped to and tested at the University of Akron. The remains of the web after thread collection were gathered using a clean glass pipette for chemical composition analysis of the glue, with all webs pooled within each population. All insects or debris stuck to the web were removed prior to the collection of capture spirals. For some webs, capture threads were stuck together, and it was not possible to get good thread samples for mechanical testing, so these webs were only collected for chemical composition purposes.

Radial threads were tested 3 months after collection. Even after just a month, radial silk could show higher stiffness and less strain and stress due to aging (Lepore et al., 2016). However, because silk samples were collected in the field over the course of a month and mailed to the University of Akron before any testing could occur, this time effect could not be avoided. Stress–strain curves were generated for major ampullate radial silk using a Nano Bionix test system (Agilent Technologies, Oak Ridge, TN, USA) following previously described methods (Blackledge et al., 2005; Sensenig et al., 2010). Fibers were extended at a strain rate of 0.015 s⁻¹. A minimum of three threads per web were tested at room humidity [~40%–50% relative humidity (RH)]. Diameters of threads were measured from threads viewed under a Leica DM LB2 polarized light microscope (Leica Microsystems CMS GmbH, Wetzlar, Germany) at ×100 magnification. Photographs of the suspended radial threads were taken using a 5-Megapixel Olympus Q-Color5TM imaging system (Olympus America Inc., Melville, NY, USA) attached to the microscope. Measurements were taken at three points using Image-Pro Plus 6.2 software (Media Cybernetics, Inc., Bethesda, MD, USA) and averaged for one axial fiber per double-stranded thread. Material properties (tensile strength, extensibility, toughness, and stiffness) were obtained as previously described (Vollrath et al., 2001; Sensenig et al., 2010). Tensile strength, or true breaking stress, was calculated as the amount of force required to break a fiber relative to the instantaneous cross-sectional area. Extensibility, or true breaking strain, was calculated as the natural log of one plus the ratio of breaking length to the original length and refers to how much the material has stretched normalized to the instantaneous length. Toughness, or the energy absorbed by a fiber prior to rupture, was calculated as the area under the stress–strain curve. Stiffness, or Young’s modulus, was obtained from the initial slope of the stress–strain curve, in which the relationship of stress and strain was almost linear.

The adhesiveness of the capture spiral silk was measured 3 months after collection. Older capture spirals lose adhesion as they dry out (Agnarsson et al., 2008); however, adhesive testing was performed at high humidities, which rehydrates samples, and therefore little effect of aging is expected (Opell and Schwend, 2008). Capture silk threads were brought in contact with a clean, 5-mm-wide glass surface and pressed until they generated 15 µN. Samples were then held for 6 s and pulled off at an extension of 0.1 mm s⁻¹ using the Nano Bionix test system (Agilent Technologies, Inc., Santa Clara, CA, USA) to generate force-extension curves (Agnarsson and Blackledge, 2009). The work to release was calculated from the area under the force-extension curve and represents the energy required to separate the thread from the glass substrate. Peak load was the maximum force of adhesion achieved by the system, and extensibility was the distance the capture spiral displaced until reaching the maximum force of adhesion. This distance includes both the stretching of the flagelliform fiber and the stretching of the glue droplets themselves. Three capture spiral threads were tested per web. To control for humidity, adhesion tests were performed inside a humidity chamber at 85% RH to mimic the natural conditions of the spiders’ habitat. Samples were placed inside the humidity chamber and allowed to acclimate for 2–3 min prior to testing, enough time for droplets to swell or contract to equilibrium. Prior to testing, the glue droplet length and width were measured from images of the suspended capture silk under a Leica DM LB2 polarized light microscope (Leica Microsystems) at ×100 magnification and the space between glue droplets at ×10magnification. Photographs of the suspended capture silk were taken at room humidity (~ 46% RH) using a 5-Megapixel Olympus Q-Color5TM imaging system (Olympus America Inc., Melville, NY, USA) attached to the microscope. Glue morphology measurements were done using Image-Pro Plus 6.2 software (Media Cybernetics Inc., Bethesda, MD, USA). The glue droplet volume, the distance between the droplets, and the glue volume per millimeter were calculated as in Liao et al. (2015).

We used ¹H solution-state nuclear magnetic resonance (NMR) to compare the chemical composition of LMWCs present in the glue of each species (Sahni et al., 2014; Jain et al., 2015). The collated samples of webs for each species (T. hawaiensis (Big Island) ~ 22, T. hawaiensis (Maui) ~13, T. trituberculata ~ 9, and T. stelarobusta ~ 14) were first washed with deionized water (~10 mL) to remove the soluble LMWC components. Next, the soluble residue was lyophilized for about 10 h to remove the water and obtain the dried water-soluble LMWCs components. Finally, the dried extracts were dissolved in about 1 ml of 99.96% deuterated water (Sigma-Aldrich Solutions Merck KGaA, Darmstadt, Germany) and placed in 5-mm NMR glass tubes (Norell, Morganton, NC, USA). All experiments were performed on a Varian 300 MHz spectrometer at 298 K. The spectra were recorded with a scan size of ~ 512, acquisition time of ~3 s, a 900 pulse width of 15–16 µs, and a sweep width of 5 ppm in the ¹H dimension. Although there may be additional peaks beyond the 5-ppm chemical shift (Hu et al., 2007; Wolff et al., 2023), this was beyond the scope of our study due to the limitations of the NMR equipment, leading to lower resolution and sensitivity, and the limited silk material that we could collect from webs. Once the spectra were processed, different peaks were assigned to known LMWCs using data reported in the literature (Anderson and Tillinghast, 1980; Tillinghast and Christenson, 1984; Tillinghast et al., 1987; Townley et al., 1991; Townley and Tillinghast, 2013; Sahni et al., 2014; Jain et al., 2015, 2006, 2012; Vollrath et al., 1990) and an unpublished database provided by Townley. We compared the LMWC profiles of Hawaiian Tetragnatha spiders to those of a mainland species, Tetragnatha laboriosa Hentz, 1850, using data from Jain et al. (2018).

All data were analyzed, and graphs were built using JMP Pro 17 (JMP Statistical Discovery LLC., SAS Institute Inc., Cary, NC, USA). Values were averaged across all threads tested for an individual spider. The data were checked for normality by the Shapiro–Wilk’s test. While major ampullate silk followed a normal distribution, adhesion test results and capture silk morphology did not, so those values were log-transformed. A Wilk’s Lambda multivariate analysis of variance (MANOVA) was used to see if the mechanical properties of the major ampullate silk and the capture spiral threads differed between the populations. Individual ANOVAs were used to compare each of the material properties (strength, extensibility, toughness, stiffness, and silk diameter) of major ampullate threads, the adhesiveness of capture spiral threads (ln work of adhesion, ln peak force of adhesion, and ln extension), and the morphology of capture threads (volume per droplet, spacing between droplets, and glue volume per mm) between the four populations of Tetragnatha: T. stelarobusta, T. trituberculata, T. hawaiensis (Maui), and T. hawaiensis (Big Island). Tukey’s honest significant difference tests were used post hoc when needed to identify which pairwise comparisons were significant among the populations. Individual spiders were used as a random variable in all tests.

The material properties of radial threads differed among the species (F(15, 58.37) = 3.87, p < 0.0001; Wilk’s Λ = 0.15, Figure 1 ). One-way analysis of variance showed that T. stelarobusta had a significantly stiffer silk than T. hawaiensis (Big Island) (21.75 GPa ± 4.54 GPa and 10.42 GPa ± 1.55 GPa, respectively) (F(3, 25) = 4.24, p = 0.015; Figure 1D ). Even though no statistical differences were found for toughness (F(3, 25) = 2.45, p = 0.087, Figure 1A ), true stain (F(3, 25) = 2.08, p = 0.128, Figure 1B ), and true stress (F(3, 25) = 1.74, p = 0.185; Figure 1C ), a trend can be observed with T. trituberculata having stronger, stretchier, and tougher silk (785.58 MPa ± 120.76 MPa, 0.19 mm/mm ± 0.04 mm/mm, and 85.09 MPa ± 21.51 MPa, respectively) and T. hawaiensis (Big Island) having the least strong, least stretchy, and least tough radial silk (523.85 MPa ± 76.99 MPa, 0.16 mm/mm ± 0.01 mm/mm, and 45.81 MPa ± 6.77 MPa, respectively). Radial thread diameter differed between the four populations of Tetragnatha spiders, following the same trend as strength, extensibility, and toughness (F(3, 25) = 0.18.15, p < 0.01; Figure 1F ). The diameters of T. trituberculata silk were larger than T. hawaiensis (Big Island) (1.31 µm ± 0.05 µm vs. 1.05 µm ± 0.02 µm, respectively).

Adhesive properties of capture silk differed among the different populations of spiders (F(9, 68.30) = 3.24, p = 0.003; Wilk’s Λ = 0.42, Figure 2 ). Work of adhesion (F(3, 30) = 5.53, p = 0.004) and peak force (F(3, 30) = 4.79, p = 0.008) differed between species, while extension remained similar (F(3, 30) = 0.69, p = 0.56). Tetragnatha trituberculata capture silk had a higher work of adhesion (0.79 µJ ± 0.12 µJ) and higher force of adhesion (170.9 µN ± 23.6 µN), followed by T. stelarobusta (0.67 µJ ± 0.21 µJ and 157.64 µN ± 38.37 µN, respectively), while T. hawaiensis (Maui) and T. hawaiensis (Big Island) showed the lowest stickiness with a work of adhesion of 0.33 µJ ± 0.1 µJ and 0.27 µJ ± 0.06 µJ and a force of adhesion of 89.35 µN ± 19.91 µN and 77.89 µN ± 12.68 µN.

Capture thread stickiness can be affected by the size and spacing of the glue droplet (Opell and Hendricks, 2007), which varied among the four groups in this study (F(9, 78.03) = 2.91, p = 0.005; Wilk’s Λ = 0.50, Table 1 ). Each structural property measured for glue varied among the four groups (glue droplet volume: F(3, 34) = 10.66, p < 0.001; spacing: F(3, 34) = 6.29, p = 0.002; glue volume per mm: F(3, 34) = 8.14, p = 0.0003). Tetragnatha stelarobusta had on average the highest glue volume and the largest spacing between glue droplets, which means that glue is getting packed in larger droplets that are relatively farther apart from each other. Tetragnatha trituberculata had the second-largest volume of glue per mm of thread, followed by T. hawaiensis (Maui). Tetragnatha hawaiensis (Big Island) had, in contrast, the lowest glue volume and the least space in between the droplets ( Table 1 ).

Figure 3 shows the ¹H NMR spectra illustrating different small molecules present in the glues of Tetragnatha spiders. The major components include betaine, choline, glycine, and alanine. The quantitative comparison of LMWCs among species was challenging due to the small samples, which led to a poor signal-to-noise ratio. However, we compared the LMWC profiles of Hawaiian Tetragnatha spiders to those of mainland (Ohio, USA) Tetragnatha laboriosa and found no major differences except for the lack of N-acetyl taurine in Hawaiian spiders. While the absolute lack of N-acetyl taurine in Hawaiian species is not clear because the starred peaks around 2 ppm and 3 ppm in T. trituberculata and T. hawaiensis (Big Island and Maui) may represent low concentrations of the compound, these unidentified peaks are clearly absent in the better-resolved spectrum of the glue of T. stelarobusta. LMWCs found in the glues of other local US species, such as T. versicolor Walckenaer 1841, and T. elongata Walckenaer 1841 also shows the presence of N-acetyl taurine, and in some cases, proline (Townley unpublished data), in addition to our reported LMWCs for Tetragnatha from the Hawaiian archipelago.