The contribution of axial fiber extensibility to the adhesion of viscous capture threads spun by orb-weaving spiders

Viscous capture threads are highly evolved adhesive delivery systems found in orb-webs produced by over 4000 spider species(Fig. 1). They make crucial contributions to the operation of these webs by retaining insects, thereby giving spiders more time to locate, run to and subdue prey that their webs have intercepted (Chacón and Eberhard, 1980; Eberhard,1986; Eberhard,1989; Eberhard,1990). Viscous threads are spun from the spigots of two adjacent silk glands (Foelix, 1996). The flagelliform glands produce a pair of supporting axial fibers, and the aggregate glands coat these fibers with a viscous, aqueous solution that quickly forms into droplets (Peters,1986; Peters,1995; Vollrath et al.,1990). The glycoprotein granules that coalesce inside each droplet contribute to thread adhesion (Vollrath and Tillinghast, 1991; Tillinghast et al., 1993) and the hydrophilic compounds in the surrounding fluid attract atmospheric moisture to prevent droplets from drying(Townley, 1990; Vollrath et al., 1990; Townly et al., 1991).

Together with the cob- and sheet-web weaving species descended from them,these orb-weaving spiders comprise the Araneoidea clade, which includes 27% of the 40 024 living spider species(Platnick, 2008). The viscous threads produced by members of this clade replaced the cribellar prey capture threads spun by members of their sister clade, the Deinopoidea(Coddington, 1986; Coddington, 1989; Griswold et al., 1998; Garb et al., 2006). Cribellar capture threads are also supported by a pair of axial fibers. However, these fibers are covered by an outer sheath of fine, dry, looped protein fibrils(Peters, 1984; Peters, 1986; Peters, 1992; Eberhard and Pereira, 1993; Opell, 1999) that are drawn from the spigots of an oval spinning plate, termed the cribellum, by a spider's calamistrum, a setal comb on the metatarsus of each of its fourth legs (Eberhard, 1988; Opell, 2001). Rhythmic adductions of the median spinnerets press the fibril sheath around the supporting strands to produce a thread that has a complex, but often regular,surface configuration (Peters,1986).

Relative to the volume of material invested in a mm of thread, viscous threads achieved an average of 13 times more stickiness than cribellar thread(Opell, 1998). A factor contributing to the efficiency of viscous thread is its ability to recruit adhesion from multiple droplets using what Opell and Hendricks have described as a suspension bridge mechanism (SBM)(Opell and Hendricks, 2007). Together, the extensibility of a thread's axial fibers and the plasticity of its droplets allow it to bow as the thread is pulled away from a contacting surface. This configuration divides the loading force into perpendicular and parallel vectors, the latter being responsible for recruiting adhesion for droplets that lie interior to the edges of a thread's contact with a surface. As the axial fibers of viscous threads are more extensible than those of cribellar threads (Blackledge and Hayashi,2006), viscous threads appear better equipped to implement the SBM than do cribellar threads. Documentation of this comes from the observation that viscous thread spans of increasing length register increasing stickiness(Opell and Hendricks, 2007)whereas there is no change in the stickiness of cribellar thread spans of increasing length (Hawthorn and Opell,2003; Opell and Schwend, in press).

The present study examines more precisely the contribution of axial fiber extensibility to viscous thread adhesion as it tests the hypothesis that reducing the extensibility of a thread's axial fibers reduces its expressed stickiness. It does so by examining the viscous threads of five araneoid species that have droplet profiles that range from small, closely spaced droplets to large widely spaced droplets(Fig. 1). The stickiness of these threads was first measured under their native tensions and again after they were stretched to two different lengths to reduce the extensibility of their axial fibers. Stretching threads also increases the distance between their droplets (Fig. 2). To maintain the number of droplets that contributed to the stickiness of a thread span, we measured the stickiness of stretched threads with contact plates whose widths were increased in proportion to the degree of thread elongation.

We collected web samples from orb-webs constructed by adult females of five species of the family Araneidae [Araneus marmoreus Clerck, Argiope aurantia Lucas, Micrathena gracilis (Walckenaer), Verrucosa arenata (Walckenaer) and Cyclosa turbinata(Walckenaer)] from sites near Blacksburg, Montgomery Co., VA, USA. Orb-web sectors were collected in the morning, a few hours after webs were spun, using 18 cm-diameter aluminum rings with a 5 mm wide bar across their centers. Double-sided tape on the rim and center bar of each ring held the threads securely. We photographed and measured the stickiness of threads from each web sample in the laboratory under the same relative humidity and temperature within 6.5 h after web samples were collected.

We collected unstretched capture threads from web-sampling rings using a microscope slide sampler, made by gluing 4.8 mm square brass supports to microscope slides at 4.8 mm intervals. Double-sided Scotch® tape (Tape 665; 3M Co., St Paul, MN, USA) on these supports held the threads securely and maintained their native tensions. Before collecting thread samples, we placed brass bars with double-sided tape on one surface across the collecting ring's rim and center bar to isolate web regions. This permitted us to collect a thread sample from one region of the sampling ring without disturbing threads in other regions. Next, a set of 4–8 threads (depending on the spacing of a species' capture spirals) was collected between two 5 mm-wide bars that were attached to the jaws of a digital caliper in preparation for thread elongation. Double-sided carbon tape (used for mounting specimens to be examined with a scanning electron microscope) secured threads to bars. To hold these threads even more securely, we applied Kores® mimeograph correction fluid (Ink Technology Corp., Tenafly, NJ, USA) along the length of thread spans that contacted the tape. This red fluid is a fast-drying paint whose principal solvent appears to be ether. It immediately adhered to the double-sided tape and, when dry, formed a thin seal on the tape's surface. We then slowly separated the jaws of the caliper at a speed of approximately 232μms^–1 to elongate threads and then collected stretched threads on a microscope slide thread sampler. Threads were elongated to lengths that corresponded to the widths of the contact plates used to measure thread stickiness (Fig. 2). Unstretched threads were measured with contact plates that were 963 μm wide. One set of threads was stretched 2.215 times their native lengths and measured with 2133 μm-wide contact plates. Another set of threads was stretched 3.345 times their native lengths and measured with 3222 μm-wide contact plates. For simplicity, we refer to these elongations as 1×,2× and 3×. We examined each of these preparations under a dissecting microscope so that we could remove damaged threads, eliminate capture thread spans through which radial threads passed and, in species whose capture threads were very closely spaced, remove threads to achieve spacing appropriate for our stickiness measurement procedures. In the process, we were able to confirm that neither the mimeograph correction fluid or its solvent bled onto the suspended threads, as the spacing and features of the droplets near the edges of these threads were indistinguishable from those at the centers of the threads.

As illustrated previously (Opell and Hendricks, 2007), the instrument used to measure thread stickiness allowed us to align a microscope slide sampler so that a thread span was perpendicular to the length of a contact plate. A linear actuator moved thread spans relative to the contact plate, and a sensitive load cell, to which a contact plate was connected by a lever system, recorded the force of adhesion generated as a thread span was pulled from a contact plate. A thread was first pressed against a contact plate at a speed of 0.06 mm s^–1until a force of 25 μN was generated and was then immediately withdrawn at the same speed until the thread pulled free of the contact plate. The maximum force registered by a thread was recorded as its stickiness. For each thread elongation, we measured the stickiness of three thread sectors using contact plates of an appropriate width (963, 2133 or 3222 μm) and recorded the mean of these three measurements as a thread's stickiness value for that thread elongation. The contact plates were covered with Scotch Magic® tape (Tape 810; 3M Co.), which provided a smooth acetate surface that maximized thread contact and eliminated the possibility that threads with different droplet profiles might respond differently to a textured surface. This was the same material from the same roll of tape used by Opell and Hendricks to document the operation of the SBM (Opell and Hendricks, 2007). This acetate was replaced frequently and care was taken to ensure that each stickiness measurement was made with an unused sector of a contact plate. Immediately before taking each series of three stickiness measurements, we recorded laboratory temperature, humidity and barometric pressure. All measurements of one individual's threads were completed before measurements of another individual's threads were begun.

We assessed the range of droplet volumes for an individual spider's threads by first subtracting the mean droplet volume of the thread strand (1×,2× or 3×) with the smallest mean droplet volume from the thread strand with the greatest mean droplet volume. Next, we divided this difference by the mean droplet volume of the individual's three thread strands at the three elongations and multiplied this value by 100 to obtain an index that we term the range of intraindividual droplet volume(Table 1).

Deviations from the intended 2× and 3× thread elongation would compromise our experimental design. Therefore, we evaluated the thread elongation that we achieved by first computing the number of droplets per millimeter thread length (DPMM) from images of each individual's threads at their native lengths and at each of the two elongations. We then divided the DPMM of an individual's unstretched threads by the DPMM of its 2× and 3× stretched thread to determine the elongation that was achieved. As Fig. 3 shows, we achieved less elongation than intended. Our attempts to secure threads to the bars on the caliper's jaws by using a pliable adhesive tape followed by the addition of fast-drying paint may not have prevented the axial fibers of these threads from being pulled through these attachment media. Alternatively, threads were initially elongated as intended but slipped through the adhesive of the double-sided Scotch® tape as they were transferred to the supporting bars of the microscope slide samplers. Although we did achieve progressive thread elongation for all species (Fig. 3), our failure to fully elongate threads resulted in more droplets contacting plates used to measure stretched threads than those used to measure unstretched threads. To correct this problem, we computed stickiness per thread droplet, as explained more fully in the following section.

We first divide the stickiness registered by contact plates of each width by the number of droplets contacting the plate. Droplet number was computed by multiplying the DPMM for 1×, 2× and 3× elongated threads by the width of the 0.963, 2.133 and 3.222 mm-wide contact plates, respectively. Although this mean stickiness per droplet accounted for most of the effects of incomplete thread elongation, a minor additional adjustment was necessary to account fully for the operation of the SBM. In thread spans of increasing lengths, each additional pair of droplets contributes successively less adhesion, as less adhesion is recruited from interior droplets than from edge droplets (Opell and Hendricks,2007). Consequently, although increasing the number of droplets in a strand increases the strand's stickiness, it also results in a slight reduction in the mean stickiness per droplet(Fig. 4). Thus, our failure to adequately stretch threads increased the number of droplets that contributed to a strand's stickiness and this, in turn, slightly reduced the stickiness per droplet of 2× and 3× elongated threads. Fig. 4 illustrates this for threads of C. turbinata, using previous data(Opell and Hendricks, 2007) to compute the mean stickiness per droplet (SPD) for unstretched threads measured with contact plates of 963, 1230, 1613, and 2133 μm widths. It shows that,as the number of droplets contacting plates of greater widths increases, the thread span's mean SPD decreases.

Consequently, it was necessary to correct the SPD values of under-stretched threads. The relationship between droplet number and SPD described above and illustrated in Fig. 4 provides a mechanism for doing so. By knowing how many additional droplets contacted the plates used to measure the stickiness of 2× and 3× stretched threads than the plate used to measure unstretched thread, it was possible to restore the SPD lost to increases in the number of contacting droplets. In the example of C. turbinata (Fig. 4), insufficient thread elongation resulted in five additional droplets contributing to the stickiness of the stretched thread. Multiplying these five additional droplets by the slope of the regression line and adding this product to the measured mean SPD for the stretched thread yields a value that we term adjusted SPD (ASPD), which corrects for the slight reduction in SPD due to the increased number of droplets. We generated similar regressions for A. marmoreus from data used in Opell and Hendricks(Opell and Hendricks, 2007)and for A. aurantia, M. gracilis and V. arenata from data that were gathered for a broader survey (B.D.O. and M.L.H., unpublished observations). The regressions for these additional species are: y=0.0002x+5.1774, y=0.0003x+12.4666, y=0.0006x+1.8627, and y=0.0011x+11.7829,respectively, where y is the value added to the measured SPD, and x is the number of additional droplets contacting a plate used to measure the stickiness of stretched threads.

As the native extensibility of viscous threads differs among species(Opell and Bond, 2001), we judged that our thread elongation procedure probably did not affect the extensibility of each species' axial fibers in the same way. To evaluate this,we measured the breaking lengths of capture threads relative to their native lengths, using the same caliper apparatus described above, the same methods for affixing threads to the bars on this caliper's jaws, and the same rate of elongation to measure the breaking lengths of capture threads. We collected a series of 3 mm-long thread spans from each web, extended these threads and recorded the breaking length of each strand. We then computed breaking factor by dividing a thread's initial length by its length at rupture.

Differences in the breaking factors of the five species' threads(Table 1) indicate that our elongations did affect their threads differently. When a viscous capture thread is strained (elongated), its stress initially increases gradually and then in a more pronounced manner as it enters the stress hardened phase of its stress–strain curve prior to rupture(Köhler and Vollrath,1995; Blackledge and Hayshi,2006). Thus, at elongations of 3× stretched, the threads of V. arenata were much nearer their rupture values and were much stiffer than the 3× stretched threads of the other species. By contrast,at an elongation of 3×, the threads of M. gracilis were still quite extensible. Because the analysis of our controlled thread elongations indicated that some thread slippage occurred during the stretching procedure,the breaking factors that we report may be inflated. However, if most of this slippage occurred when stretched threads were transferred from the caliper to the microscope slide samplers, then these breaking factors are not greatly inflated. Whichever scenario is correct, we believe that breaking factors are useful in assessing differences in the residual extensibility of viscous threads in the five species' orb-webs.

The standard index of a fiber's extensibility is its Young's modulus, with higher values indicating stiffer fibers. Young's modulus is computed by dividing stress in MPa by strain, expressed as a percentage of a fiber's initial length. Thus, a fiber's Young's modulus can be computed at any elongation from its stress–strain curve. Although we were not equipped to generate stress–strain curves for threads of the species we studied,we were able to compute an index that we term `relative Young's modulus' (RYM)for each species' threads at each of their realized elongations. We based these values on the stress–strain curve for the viscous capture threads of Araneus diadematus(Köhler and Vollrath,1995). From this curve, we determined the Young's Modulus at the full range of thread stresses up to and including the thread's rupture value. We then plotted these values as RYM, where Young's modulus at rupture=1,against relative thread elongation, where elongation at rupture=1, and mathematically described this relationship(Fig. 5). For each elongation of each individual's threads we computed a relative thread elongation ratio by dividing the achieved thread elongation by the mean breaking elongation of its species. We then used the regression formula shown in Fig. 5 to assign RYM values to these achieved thread elongations (Table 1, Fig. 6). Although RYM is more appropriate than thread elongation for describing the amount of residual extensibility in a thread's axial fibers, it is based on the thread of a species not included in this study and, therefore, it only approximates residual thread extensibility. The regression model assigned a RYM value of 0.083843 to unstretched threads. Although this base value might be regarded as an artifact of the modeling process, it is probably a reasonable estimate because threads were already under some tension in the orb-web and because threads were slightly elongated as their stickiness was being measured.

For each individual's threads at each elongation we determined DV, ASPD and RYM. Within a species, droplet volume is directly related to the stickiness of viscous threads (Opell, 2002; Opell and Schwend, 2007),although this relationship may not be as strong among species(Opell and Schwend, 2008). Moreover, our hypothesis predicted that RYM should contribute negatively to stickiness, as larger values of RYM indicate stiffer threads. We used the SAS statistical package (SAS Inc., Cary, NC, USA) to test the normality of droplet volumes, to compare the droplet volumes of threads stretched to different lengths and to generate regression models that tested the hypothesized contribution of DV and RYM to ASPD in each of the five species. Data were considered normally distributed if P>0.05 for a Shapiro–Wilk W-statistic test. We examined normally distributed data with one-way analyses of variance (ANOVA) and t-tests (T). Data that were not normally distributed were compared with Kruskal–Wallis χ² tests(KW). We considered regression models with 0.10⩾P>0.05 to provide weak support for the hypothesis and P⩽0.05 to provide strong support for the hypothesis.

Thread elongation clearly altered the RYM of an individual's thread samples(Fig. 6). Moreover, the range of intraindividual droplet volume of the 1×, 2× and 3×threads was considerable, from 31 to 61%, and averaged 51% of mean individual droplet volume (Table 1). Given these differences, the separate measurements of the RYM, DV and ASPD that we obtained for each individual's 1×, 2× and 3× threads were largely independent of one another. However, there is still the possibility of a spider-specific effect among the three factors (RYM, DV, individual) that contributed to ASPD. Therefore, we report two sets of P values for each species' regression model: P (EDF1), whose F value was computed using an error degree of freedom (EDF) based on the number of individuals sampled, and P (EDF2), whose EDF was based on a sample size reduced by one-third to account for any effect of measuring the threads of individuals at three elongations. Thus, EDF2 provides a more conservative test of the hypothesis by diminishing the individual component through reduced F values and increased P values.

Tables 1 and 2 report the thread features and stickiness values for native and elongated threads of the five species. If thread elongation resulted in more viscous material being withdrawn from droplets and distributed along interdroplet regions, then a species' droplet volume should decrease as threads are elongated. In fact, it appears as though there might be a tendency for droplet volume to increase as threads are elongated (Table 1). However,this is not supported by comparisons of either the droplet volumes of threads at their three elongations or of the droplet volumes of the unstretched and 3× threads. For A. aurantia, A. marmoreus, M. gracilis, V. arenata and C. turbinata, the results of these tests were: KW P=0.3897, KW P=0.2482; ANOVA P=0.2441, T P=0.0944; KW P=0.8208, KW P=0.6272; KW P=0.6977, KW P=0.3258; ANOVA P=0.8749, T P=0.9397, respectively.

In M. gracilis, V. arenata and C. turbinata, DV and RYM jointly explained ASPD in models judged on both full and reduced EDF values(Table 3). In these three species, DV was a significant and positive contributor to ASPD under both full and reduced EDF values. In all species but A. marmoreus, RYM was a significant and negative contributor to ASPD under full EDF values. That is,as a thread's extensibility was reduced, its per droplet stickiness also decreased. When EDF2 values were considered, strong support for a negative contribution of RYM remained in M. gracilis and V. arenatabut there was only weak support for a negative contribution of RYM in A. aurantia and C. turbinata.

Our failure to find a strong support for models of ASPD based on DV and RYM in A. aurantia and A. marmoreus may result from high DV variance in these species or to a low correlation between the DV and ASPD in the individuals that we studied. The relationships between DV and ASPD for the unstretched threads of A. aurantia and A. marmoreus were not significant (P=0.1327 and 0.1859, respectively) whereas this relationship was significant for M. gracilis, V. arenata and C. turbinata (P=0.0038, P=0.0061 and 0.0303,respectively).

We used the significant regression models for M. gracilis, V. arenata and C. turbinata to illustrate graphically the contributions of RYM and DV to ASPD (Fig. 7). These models showed that, as threads are elongated, the adhesion attributed to droplet volume alone increasingly exceeded measured ASPD whereas increasing amounts of potential adhesion are lost to reduced thread extensibility.

Axial fibers contribute indirectly to thread stickiness by recruiting the adhesion of multiple droplets. To assess this contribution we used the regression models of M. gracilis, V. arenata and C. turbinata to estimate the percentage of an unstretchced thread's stickiness that can be attributed to the extensibility of its axial fibers. We express this contribution as percent elastic component (PEC), the reduced stickiness of a 3× elongated thread attributed to its lost extensibility divided by the stickiness of an unstretched thread. The PECs for M. gracilis, V. arenata and C. turbinata threads were 50.0%, 29.2%and 22.5%, respectively, yielding a mean PEC of 33.9%.

When a polymer is elongated and maintained in this strained condition, the resulting stress can cause the fiber to lengthen, thereby reducing this stress. This behavior is known as stress relaxation and was demonstrated(Denny, 1976) to occur in the viscous threads of Araneus sericatus. When these threads were strained to 262% of their initial length (87% of their breaking elongations),they registered a stress of 1^8.65 Nm^–2. Within 10 min, this stress diminished to 1^8.37 Nm^–2 (52% of their initial stress) and after 38 min they achieved stress equilibrium at a value that was only slightly less [fig. 11 in Denny(Denny, 1976)]. When interpreted in light of the stress–strain curve of this species' viscous threads [fig. 9 in Denny (Denny,1976)], this shows that, after undergoing stress relaxation, the effective strain of these threads was 227% rather than the initial 262% that they experienced. Thus, if the stress–strain curve of the stress-relaxed thread was unaltered, it appears that stress relaxation restored 13% of the thread's residual elongation and reduced the thread's Young's modulus. The Young's modulus of threads at rupture was 0.367, and at an elongation of 262%it was 0.179. If the stress–strain curve of the stress-relaxed thread was unaltered, the Young's modulus would be 0.115. These values translate into RYM values of 1.00, 0.49 and 0.31, respectively. However, the stress–strain curves of stress-relaxed threads almost certainly have steeper slopes than those of native threads.

The time required to screen and photograph threads before measuring their stickiness allowed all of the threads used in this study to reach their stress relaxation equilibriums. This means that the residual extensibility of all of the stretched threads was probably greater than our indices of relative Young's modulus indicate and was probably proportionately greater for threads that were stretched to a higher percentage of their breaking elongations. The mean realized 3× extensions of the threads of M. gracilis, A. marmoreus, A. trifasciata, C. turbinata and V. arenata,expressed as a percentage of their mean breaking extensions, were 30%, 43%,47%, 47% and 75%, respectively. Thus, stress relaxation should have had the most pronounced effect on stretched threads of V. arenata.

To assess this effect, we performed an additional regression for V. arenata threads using reduced RYM values computed from the data on A. sericatus. Relative to breaking elongation, the 3× elongation of V. arenata threads was 86% that of A. sericatus. As the stress-relaxed RYM of A. sericatus threads was 37% less than their elongated RYM, this translates to a 32% reduction in the RYM values of 3× V. arenata threads. The mean realized 2× extension of V. arenata threads was 53% of their mean breaking extensions. If the RYM of a stress-relaxed thread decreases in proportion to its elongation, then the values of a 2× thread should be 71% that of a 3× thread or 23%that of thread that has not undergone stress relaxation. The regression model based on these modified RYM values is significant (Model PEDF1=0.0001, DV P EDF1=0.0001, RYM DV P EDF1=0.0125) and shows that ASPD is directly related to DV and inversely related to RYM(ASPD=0.001DV–8.923RYM+13.529).

Our results support the hypothesized positive contribution of a viscous thread's axial fiber extensibility to its stickiness. When threads are stretched and their extensibility is reduced, their per droplet stickiness decreases. We observed this response both in threads of M. gracilis,which had the greatest native extensibility, and in threads of V. arenata, which had the least. We also observed it in threads whose droplet volumes differed by a factor as great as 152 and whose droplet spacing differed by a factor as great as 6. These findings are consistent with the operation of a SBM that enhances the stickiness of viscous capture threads by recruiting the adhesion of droplets interior to the edges of a thread's contact with a surface (Opell and Hendricks, 2007).

Our attempt to evaluate the effect of stress relaxation on stretched threads used the conservative assumption that the slope of the stress–strain curve of stress-relaxed threads is identical to that of threads that have not undergone stress relaxation. Nonetheless, it confirmed that thread extensibility contributes positively to the stickiness of V. arenata threads, which were elongated to a much greater percentage of their breaking lengths than were threads of the other species. Consequently,we believe that stress relaxation did not confound the broader conclusions of our study. A complete explanation of viscous thread performance must incorporate this phenomenon, although the preliminary calculations that we present suggest that this will be challenging.

The molecular structure of silk affects its mechanical properties(Hayashi et al., 1999; Hayashi et al., 2001; Hayashi et al.,2004; Hayashi and Lewis,2001; Craig, 2003; Ayoub et al., 2007) and appears to explain intraspecific (Blackledge and Hayashi, 2006) and interspecific(Swanson et al., 2006a; Swanson et al., 2006b)differences in thread properties. However, the observed 2.5-fold difference among the breaking factors of the five species' viscous threads(Table 1) cannot be attributed solely to differences in the molecular composition of their axial fibers. As we measured threads at their native, in-web tensions and did not standardize their tensions prior to measuring their breaking factors, we were unable to factor out the contribution that differences in web construction behavior may have made to the observed differences in thread breaking factors. Members of some species may stretch their capture threads more than others before they attach them to the web's radial lines. If they do, then a greater portion of the potential extensibility of these threads would have been expended, leaving them with less usable extensibility.

The mean 33.9% lost adhesion that can be attributed to reduced extensibility in the 3× stretched threads documents the important contribution that thread extensibility makes to thread stickiness. Threads of all species could be elongated more than the realized 3× extensions on which this estimate was based. However, this 33.9% is probably a reasonable estimate of the typical contribution of axial fiber extensibility to thread adhesion because features of orb-web architecture constrain the elongation that viscous threads undergo when intercepting and retaining prey. Insects usually strike multiple spiral turns, thereby distributing impact forces and struggling stresses over several thread spans. Moreover, aerodynamic dampening helps vertical orb-webs absorb the forces of prey impact as webs flex through the air (Lin et al., 1995). Even this web flexibility is constrained by the combined extensibility of the web's radial and capture lines (Craig,1987; Craig,2003).

Intraspecific differences in droplet volume, droplet spacing and maximum thread extensibility made it challenging to evaluate the contribution of axial fiber extensibility to thread adhesion. Only by accounting for each of these variables was it possible to document the role of axial fiber extensibility in thread adhesion. Interspecific differences in maximum thread extensibility and stickiness per droplet volume made it impossible to develop a more general model that describes the performance of the five species' threads. This may indicate that each species' capture threads comprise a unique and highly tuned system whose performance integrates the axial fiber's native extensibility,the extensibility realized after the thread is deposited in a web, the thread's droplet spacing, and the adhesion and plasticity of individual thread droplets.

LIST OF ABBREVIATIONS

 
• ASPD

  adjusted stickiness per droplet

 
• DPMM

  droplets per millimeter thread length

 
• DV

  droplet volume

 
• EDF

  error degree of freedom

 
• PEC

  percent elastic component

 
• RYM

  relative Young's modulus

 
• SBM

  suspension bridge mechanism

 
• SPD

  stickiness per droplet

Caitlin Flora and Genine Lipkey assisted with fieldwork. Harry Schwend helped compute RYM values and provided suggestions for improving figures. National Science Foundation grant IOB-0445137 supported this research.