Temperature mediates the effect of humidity on the viscoelasticity of glycoprotein glue within the droplets of an orb-weaving spider's prey capture threads

All animals produce glycoproteins, but few incorporate them as adhesive in traps that capture prey. Onychophorans squirt a sticky substance supported by proteinaceous threads onto nearby prey; however, this material is not extruded prior to prey encounter (Betz and Kölsch, 2004). Marine gastropods in the genus Dendropoma deploy sticky mucus nets that gather suspended plankton and detritus (Kappner et al., 2000). Modern orb-weaving spiders incorporate viscoelastic glycoprotein glue in the spiral capture threads of their webs (Fig. 1). This system is unique in several ways; the adhesive is a crucial part of an aerial filtering system and its production anticipates future prey. Unlike onychophoran adhesive, which is produced when needed, or gastropod mucus, which remains in an osmotically and thermally stable marine environment, spider thread glycoproteins must function over the course of a day where ambient temperature and humidity oscillate, often dramatically (Fig. 2).

Three spinning spigots on each of an orb-weaving spider's paired posterior lateral spinnerets produce its composite viscous capture threads. A single flagelliform spigot produces a supporting protein axial line (Sekiguchi, 1952), which is coated by material from two adjacent aggregate glands as it emerges (Apstein, 1889). The two coated lines from each spigot merge to produce a cylindrical, proto viscous thread (Warburton, 1890). Low molecular weight molecules and inorganic salts in the fluid surrounding the axial lines rapidly attracts atmospheric moisture, increasing the cylinder's volume and surface tension, setting the stage for Plateau–Rayleigh instability to form the cylinder into droplets (Plateau, 1873; Boys, 1889; Rayleigh, 1892; Edmonds and Vollrath, 1992). Within each droplet, an adhesive glycoprotein core forms and remains covered by a hygroscopic aqueous outer layer that also covers axial lines in the thread's inter-droplet region (Opell and Hendricks, 2009) (Fig. 1C).

Thread adhesion is enhanced by the extensibility of both the glycoprotein cores within droplets (Sahni et al., 2010) and the axial lines (Opell et al., 2008). As force is applied to a thread that has contacted a surface, these droplets extend, permitting the axial line to sum their adhesive forces (Opell and Hendricks, 2007; Opell and Hendricks, 2009). The contribution of glycoprotein to this suspension bridge mechanism depends on a combination of viscosity that absorbs energy through hysteresis (Gosline et al., 1986) or dissipates energy upon droplet pull off (Sahni et al., 2010) and the reversible elastic deformations that retain prey in the web after interception (Sahni et al., 2010). These qualities, which contribute to the overall prey capture success of the web, have the potential to be altered by ambient conditions.

The hygroscopic aqueous material that coats the glycoprotein and axial lines of viscous threads absorbs atmospheric moisture and transfers this moisture to the glycoprotein glue (Opell et al., 2011b). Water absorbed by this outer layer during periods of high humidity [greater than ~60% relative humidity (RH)] causes over-lubrication (over-hydration) of the glue in Argiope aurantia Lucas 1833, reducing adhesion (the energy required to pull a droplet from a surface) through the disruption of hydrogen bonding and electrostatic interactions of the glycoprotein molecules, as Sahni and colleagues (Sahni et al., 2011) also show for Larinioides cornutus. However, threads of the nocturnal Neoscona crucifera, found in humid habitats, do not exhibit an upper humidity threshold (Opell et al., 2013). This suggests that the glycoprotein is an adaptable material, whose properties are selected to suit a species' habitat.

Argiope aurantia is an active daytime predator that prefers open, grassy habitats and, in our area of southwestern Virginia, USA,

matures in late summer. Their webs are available to capture prey in the early morning as soon as enough webbing is produced for interception. Webs of adult females are usually recycled daily; however, individuals may occasionally wait several days before replacing a web (Reed et al., 1969). Females continuously occupy the center of the web and only leave to capture prey, seek shelter during hazardous weather, and to deposit egg sacs. During the early morning, webs experience low, late summer and early autumn temperatures (15–20°C) and high relative humidities (90–100% RH). As the day progresses, temperatures rise as high as 35–40°C, while RH often drops to 45% or below (Fig. 2). This normal cycling of temperature and humidity is sometimes interrupted by periods of rainy weather, resulting in unusually high daytime humidity (Fig. 2, 5–7 August). Although A. aurantia forage from morning to evening, the bulk of their prey is captured during the afternoon, when large orthopterans are the most active and likely to be captured (Harwood, 1974).

As A. aurantia webs function over the course of a day when temperature and humidity vary greatly, the goal of this study is to examine the combined effect of these environmental factors on the properties of their glycoprotein adhesive. When temperature drops, viscous materials become stiffer, requiring either more force or more time to undergo deformation. Therefore, low temperatures may increase the viscosity of a spider web's adhesive droplets, resulting in stiffer glue and increased energy absorption during glycoprotein extension. However, the temperature increase that causes glycoprotein to become more pliable is usually accompanied by a decrease in humidity, which reduces glycoprotein hydration, extensibility and the potential for over-lubrication. Therefore, we hypothesize that temperature and humidity act in an antagonistic fashion, which tends to stabilize glycoprotein function over the course of a day. We test this by measuring the performance of viscous droplets from the webs of A. aurantia under conditions that simulate cool and humid mornings, hot and dry afternoons, and hot and humid afternoons.

In contrast to previous studies that measured total droplet extension (Opell et al., 2011b; Opell et al., 2013), we focused on the time during which an extending thread droplet was under tension, reasoning that non-loaded glycoprotein filaments that occasionally persist after the filament is no longer under an observable tension contributed neither to thread adhesion nor to energy dissipation. We used angular deflection of a thread's axial line (Fig. 3) to quantify the tension on droplets that remained consolidated before extending (pre-extension phase, Fig. 4) and during extension (droplet extension phase), with a 180 deg angle denoting a non-loaded condition. This angle also provided a means of comparing the impact of temperature and humidity on glycoprotein viscosity, as axial line deflection at the initiation of droplet extension is inversely proportional to glycoprotein viscosity. Plotting this angular index (y-axis) against the time in seconds at which each angle was measured produces a plot (Fig. 4) that is similar to a stress–strain curve, with the area under the line being proportional to the energy required to extend a droplet. These curves provide a convenient way to compare droplet performance under the three experimental conditions.

Table 1 presents mean droplet dimensions (length, width and volume) for droplets in the three treatment groups. The first set of values averages the means of five droplets from each of the 13 individuals and includes the two extended droplets. The second set includes only the averages of each individual's two extended droplets. When log transformed, the droplet length, width and volume for each treatment were normally distributed; however, the trend for these dimensions to increase with humidity was not significant (ANOVA, P>0.36).

Table 2 compares the duration of total loaded time (TLT), pre-extension (PE) and droplet extension (DE) phases and the axial line deflection angles at five intervals during DE for the three experimental conditions. The TLT for droplets exposed to afternoon conditions was 29.3 s, which is 28% longer than morning conditions (22.9 s) and 67% longer than combined hot and humid (hereafter ‘hot & humid’) conditions (17.5 s) (Student's t, P=0.0233 and <0.0001, respectively). Thus, a 10°C rise in temperature increased TLT by 5.4 s (24%) whereas a 35% decrease in RH increased TLT by 11.8 s (40%). When an individual's mean droplet volume prior to extension was averaged across the three temperature/humidity treatments, droplet volume was directly related to the angle at the initiation of droplet pull off, inversely related to TLT and unrelated to DE (P=0.0320, 0.0317 and 0.5306, respectively). This indicates that as a droplet incorporates more atmospheric moisture, the glycoprotein becomes more dilute, less force is required to initiate filament extension and the filament more quickly attains a non-loaded length.

The pre-extension phase differed between the afternoon and hot & humid conditions, while the morning condition did not differ from either the afternoon or hot & humid conditions (Student's t, P=0.0085, 0.1079 and 0.2631, respectively). Under afternoon conditions, droplet pre-extension comprised 71% of the total loaded time (Fig. 5). This increased to 75% under morning conditions and 85% under hot & humid conditions. During the afternoon condition, the extension phase was longer than during the morning and hot & humid conditions; however, the latter two conditions did not differ (Student's t, P=0.0219, 0.0035 and 0.4729, respectively).

Axial line deflection at the initiation of droplet extension (DE time 0) decreased from hot & humid conditions to morning conditions to afternoon conditions (ANOVA, P=0.0328), denoting a progressive increase in the tension on the glycoprotein core required to initiate filament extension as glycoprotein viscosity increased (Table 2, Fig. 6). This was associated with a progressive increase in the duration of the pre-extension phase (ANOVA, P=0.0287), as more time was required to generate the increased tension on axial lines necessary to initiate droplet extension (Table 2). Under all three conditions, axial line deflection decreased during the DE phase (ANOVA or Wilcoxon, P<0.008; Table 2, Fig. 6). As extension progressed, deflections at 25% and 50% under afternoon conditions differed from morning and hot & humid conditions, and at 75% and 99% differed from hot & humid conditions, although morning and hot & humid conditions did not differ (Wilcoxon each pair; Table 2). The total change in angular deflection was remarkably similar among treatments (ANOVA, P=0.9432; Table 2, Fig. 6), exhibiting only an offset in the initiating angle. Under afternoon conditions, the rate of angular change was linear, and under morning and hot & humid conditions most of the change occurred during the first half of loaded droplet extension.

As axial line deflection is proportionate to the tension on an extending droplet filament, the cumulative product of axial line deflection and the duration of loaded droplet extension is proportional to the energy required to extend a droplet. Plots of these values show that 25% more energy is required to extend a droplet under morning than under hot & humid conditions, and 58% more energy is required to extend a droplet under afternoon conditions than under morning conditions (Fig. 6).

Viscous thread performance has been measured by the force required to pull a thread from a surface (Agnarsson and Blackledge, 2009; Opell and Hendricks, 2009) and by the energy required to do so (Sahni et al., 2010; Sahni et al., 2011). Our results allow us to gauge the affect of environmental conditions on both measures of droplet performance. Pre-extension time is directly related to the force required to initiate droplet extension and axial line deflection at the initiation of droplet extension is inversely related to this force. The area under the axial line deflection/loaded extension time curve represents the energy required to complete loaded droplet extension (Fig. 6). As hypothesized, environmental conditions impact both measures of A. aurantia droplet performance. Like increased humidity, increased temperature reduced glycoprotein viscosity and, thereby, decreased the force required to initiate droplet extension, the duration of droplet extension times, the tension on extending droplet filaments and the energy required to extend droplets.

Under hot and dry afternoon conditions, initiation of droplet extension required the greatest force and droplet extension required the most energy (Table 2, Fig. 6). Under cool and humid morning conditions, less force and less energy were required. Although the least force and energy was required under hot & humid afternoon conditions, these values were more similar to those of morning conditions than the latter was to afternoon conditions. Thus, under afternoon conditions, droplet extension has the potential to dissipate the greatest amount of energy from struggling prey (Fig. 6), suggesting that the viscous glue droplets function optimally during this time. Although the ability to dissipate energy is lower under morning conditions (Fig. 6), lower temperatures help to stabilize droplet and thread function and facilitate prey capture during this time.

The effect of humidity, which has been documented previously (Opell et al., 2011b; Sahni et al., 2011; Opell et al., 2013), was greater than that of temperature. A 35% increase in RH produced a 14.2 deg increase in axial line deflection at the initiation of droplet extension, whereas a 10°C increase in temperature produced only a 6.1 deg increase in axial line deflection at the initiation of droplet extension (Table 2). This is also clearly seen in the progression of axial line deflection during the droplet extension phase (Fig. 6), where the difference between afternoon and hot & humid conditions is the result of a 35% RH difference and the difference between morning and hot & humid conditions is the result of a 10°C difference. Thus, decreasing humidity and increasing temperature experienced by webs during the afternoon affect viscous threads in opposite ways, but these changes are not strictly compensatory.

Spiders need time to assess, locate and subdue prey intercepted by their webs. Large orb-weavers such as A. aurantia rely extensively on large orthopteran prey captured during the afternoon (Harwood, 1974), and these prey can quickly escape from the web. A study examining the escape rates of insects in webs demonstrated that of the large, energetic insects that were intercepted by A. aurantia webs, 18% escaped in less than a second (Blackledge and Zevenbergen, 2006). Consequently, the length of time that thread droplets can maintain a load while being extended is crucial. Moreover, it is during this extension phase that energy from a struggling prey is dissipated through hysteresis, which occurs when droplets extend and contract.

The rate of droplet strain that we used (69.6 μm s⁻¹) was in the upper range of values used in other studies [1, 10, 50 and 100 μm s⁻¹ (Sahni et al., 2010; Sahni et al., 2011)]. In both of these studies, conducted at a 25°C, increased strain rates increased the stress and strain on extending droplets. In the latter study, the maximum stress and strain achieved for each strain rate increased when humidity increased from 15% to 40%, and then decreased from 40% to 90%; however, the relative response to each strain rate remained consistent. Therefore, we believe that, similar to humidity, temperature acts to shift but not reorder the response of viscous droplets to strain rate; however, the strain rate at which viscous threads have been selected to function has not been evaluated.

The molecular structure of spider glue glycoprotein from Nephila clavipes is thought to incorporate components that are similar to those of mucin glycoprotein (Tillinghast et al., 1992; Choresh et al., 2009), which is highly O-glycosylated (Rhodes and Ching, 1993). These carbohydrate side chains of glycoproteins are very hydrophilic and usually have all possible hydrogen bonds fulfilled with either water molecules or neighboring saccharides (Cambillau, 1995). One to three water molecules can act as bridges between a saccharide and a protein, establishing a network of viscous material in and around the glycoprotein (Cambillau, 1995). Opell and colleagues (Opell et al., 2013) suggested that over-lubrication in high humidity is due to water molecules providing alternate bonding sites, and disrupting secondary molecular structure of the glycoprotein glue. At optimum humidity, water molecules may keep some carbohydrate moieties ‘open’ by establishing bonds with them, while still allowing intramolecular structural stability. When a web intercepts an insect, these open carbohydrates may dissociate from the water to form bonds with polar cuticle and cuticular protrusions such as setae (Opell and Schwend, 2007). It may be that at lower temperatures, intramolecular hydrogen bonding within the glycoprotein becomes more stable, decreasing the bonding potential of water molecules with the carbohydrate side chains, and helping to stabilize the glycoprotein glue as its water content increases.

Orb-webs constructed by large spiders such as A. aurantia have been selected to withstand the impacts of large, often fast-flying, highly profitable prey (Blackledge and Eliason, 2007; Sensenig et al., 2010; Harmer et al., 2011; Kelly et al., 2011; Sensenig et al., 2013). This is achieved principally through the energy absorbing toughness of the web's non-sticky radial lines, with viscous threads making only a minor contribution (Sensenig et al., 2012). Consequently, these viscous threads have been largely selected for prey retention (Blackledge and Eliason, 2007). The hygroscopicity that is responsible for viscous droplet formation continues to impact the performance of viscous treads over the course of a day and has the potential to adapt a species' web to its environment. A previous study (Opell et al., 2013) showed that the high hygroscopicity of A. aurantia threads adapts this species' threads to the low humidity of their exposed habitats. Our study suggests that these threads have also been selected to function optimally at times of the day when they have the greatest chance of intercepting large prey, with low afternoon humidity restoring some of the glycoprotein's viscosity lost to the typically higher temperature at this time of the day.

We sampled the orb-webs constructed by 13 adult female A. aurantia found near Blacksburg, Montgomery County, VA, USA, from 31 August to 18 October 2012. Samples were collected between 05:30 and 09:30 h and analyzed by 16:00 h of the same day. One sector of each spider's web was collected on a 17 cm diameter aluminium ring with a bar across the center. The 5 mm wide ring and bar were covered with Scotch® double-sided tissue tape (Tape 4101T; 3M, St Paul, MN, USA), which adhered to web sectors, maintaining the threads at their native tension. When web sectors were collected, threads extending from the ring were cut with scissors to avoid distorting the sample when the ring was withdrawn from the web. We placed web-sampling rings in a closed plastic container for transport to the laboratory. Each web location was marked with flagging tape to prevent an individual's web from being resampled.

From each spider's web sector we transferred two viscous thread strands to a microscope slide sampler used for each of the three temperature/humidity treatments. We measured the extension of one droplet on each strand and the length (dimension parallel to axial line) and width of these two droplets plus three additional droplets per individual. An individual's extension and droplet volume measurements were averaged to give an operational sample size of 13 for each temperature/humidity treatment. Before collecting thread strands, we placed 4 mm wide brass bars that were covered with double-sided carbon tape (product 77816, Electron Microscopy Sciences, Hatfield, PA, USA) between the ring's rim and center bar along web radii. This further isolated and stabilized the web sample and ensured that the tension of viscous threads adjacent to the ones being collected was not altered. Forceps, which were blocked open at a distance to accommodate the supports on which thread strands would be suspended, were used to collect individual capture spiral threads and transfer them to a microscope slide sampler. Double-sided carbon tape on the forceps' tips held each thread strand securely when the thread was burned free using a hot wire probe. Five U-shaped brass struts were epoxied at 4.8 mm intervals to microscope slides with their free ends extending upward and covered with double-sided carbon tape [fig. 3 in Opell et al. (Opell et al., 2011a)]. The forceps' tips were inserted into the Us of adjacent supports, allowing threads to be secured to the tops of their tape-covered supports.

To ensure that the probe used to extend droplets only contacted a single droplet at a time, we used a minuten insect pin moistened with distilled water to move away droplets that were adjacent to the test droplet located at the center of the thread strand. This process retained the aqueous coating of the strand's axial fibers, as demonstrated by the formation of small droplets similar to those often present between the large primary droplets of many viscous threads.

Thread samples were placed in a glass-covered aluminum observation chamber whose humidity was controlled as described previously [see fig. 4 in Opell et al. (Opell et al., 2011b)] and monitored with a Fisher Scientific Instant Digital Hygrometer. A Pletier thermo-electric unit mounted on the rear wall of this chamber and controlled by a thermostat permitted the temperature within the chamber to be controlled to ±1°C. Small auxiliary heating or cooling blocks were placed on the observation window to prevent the condensation of water vapor on the glass's inner surface.

We chose 20°C to represent early to mid-morning temperatures and 30°C to represent mid-day and afternoon temperatures (Fig. 2). At humidities above 90%, threads tended to pull free from the tape on the slide; therefore, morning humidity was simulated by 85% RH. Afternoon humidity was simulated by 50% RH, as a previous study suggests that optimal humidity for A. aurantia is below 60% (Opell et al., 2013). Hot & humid conditions were simulated by 30°C and 85% RH. Although realistic, these values are probably a conservative portrayal of morning and afternoon environmental differences.

Prior to extension, we photographed an isolated suspended droplet and determined its volume as described below. A steel probe was then inserted through a port in the side of the test chamber and its 413-μm-wide polished tip aligned and brought into contact with the focal droplet. To ensure full droplet adhesion, the probe was pressed against the droplet until the thread was deflected by a distance of 500 μm. A 60 frames s⁻¹ video then recorded the probe's withdrawal at a velocity of 69.6 μm s⁻¹ as a stepping motor moved the microscope stage, on which the chamber rested, pulling the droplet away from the stationary probe.

The total loaded time (TLT) of droplet extension begins when the axial line is deflected from its initial non-loaded, 180 deg configuration and ends when the droplet returns to a non-loaded condition. TLT was divided into two phases; the pre-extension (PE) phase, during which a droplet exhibited tensive axial line deflection, but did not extend, thus holding the axial line in contact with the probe, and the droplet extension (DE) phase, which began when a droplet filament started to form and ended when the axial line returned to its 180 deg configuration at the end of TLT (Fig. 4).

During DE, we measured five axial line deflection angles: the angle at the start initiation of this phase, and the angles at 25, 50, 75 and 99% of the total duration time of DE, using iMovie '11 (Apple Inc. 2010) and ImageJ. Subtracting the angle of axial line deflection from 180 deg produces an index that is directly proportional to the tension on the extending droplet filament.

We analyzed the data for this study using JMP (SAS Institute, Cary, NC, USA) and considered comparisons with P≤0.05 as significant. Normally distributed values (as confirmed by Shapiro–Wilk W-tests with P≥0.05) were compared using ANOVA and t-tests. If the data were not normal, they were log-transformed, and once again tested for normality, and their transformed values were then compared using ANOVA. The Wilcoxon signed-rank test was used when values were not normally distributed.

Carlyle C. Brewster assisted with statistical analyses.