3 2 2010. The Journal ofArachnology 38:183-191 Webs in vitro and in vivo: spiders alter their orb-web spinning behavior in the laboratory Andrew Sensenig1 , Ingi Agnarsson1 , Taylor M. Gondek1 and Todd A. Blackledge1: ‘Department of Biology and Integrated Bioscience Program, University of Akron, Akron, Ohio 44325-3908, USA; 2Department of Biology, University of Puerto Rico, PO Box 23360, San Juan, Puerto Rico 00931-3360, USA Abstract. Many studies of the elegant architectures of orb webs are conducted in controlled laboratory environments that remove environmental variability. The degree to which spider behavior in these circumstances resembles that of spiders in the wild is largely unknown. We compared web architecture and silk investment of furrowed orb weavers Larinioides cornutus (Clerck 1757) building webs in laboratory cages and spinning webs on fences in the field and found significant differences. Thevolume ofmajorampullate silk in radii was 53% lower in cage webs, primarily because the silk was 50% thinner, but also because spiders tended to spin 14% fewer radii than in fence webs. Cage spiders also invested about 40% less flagelliform silk and aggregate glue in the capture spiral, although the difference was not statistically significant, a trend primarily driven by a decrease in the length of the glue-coated capture spiral. These patterns were consistent with spiders reducing silk investment when building at new web sites while they assessed insect abundance. Differencesinthetypeofsubstrateforwebattachment,amount ofavailablespace,andcondition mayalsohaveinfluenced web architecture. Cage webs were more symmetrical than fence webs, which displayed an unusual horizontal asymmetry that may have maximized their capture areas within the constraints of the available fence-railing attachment sites. Our findingssuggest usingcaution when generalizing the properties oflaboratory-spun webs to more natural conditions. More importantly, they demonstrate that orb spiders actively modify their behaviors when spinning webs under different conditions. Keywords: Foraging, silk investment, behavioral plasticity, silk thread size, web architecture The silk that orb spiders invest in webs is critical for absorbing threads necessarily trade offwith large capture area determining energetic gain through captured insect prey or a fine mesh width. The important role oftradeoffs between (Sherman 1994; Blackledge & Eliason 2007). Therefore, the web architecture and silk structure has largely been ignored by rich variation in sizes and shapes oforb webs among different studies that focus solely on changes in web architecture, which species ofspiders may potentially be explained by the relative represent almost all previous work in this area (e.g. Sherman costs and benefits ofsilk investment and web architecture for 1994; Herberstein et al. 1997; Nakata & Ushimaru 1999; particular environments (Shear 1986; Eberhard 1990; Higgins Heiling & Herberstein 2000; Nakata 2007). 1995). For instance, spiders should invest more silk in Orb webs are spun largely using two very different types of environments that are least likely to damage the web or that fibrous silk. The potential for a web to stop the flight of are most likely to yield prey (Higgins & Buskirk 1992; different insectsdepends on the amount and placement ofeach Blackledge & Wenzel 2001; Segoli et al. 2004). Spiders modify silk within the web. Spiders first produce an outer framework web-spinning behavior in response to environmental factors and supporting radial threads using dry dragline silk from the that include wind (Eberhard 1971; Henschel & Lubin 1992), major ampullate (MA) silk glands (Foelix 1996). They then prey abundance (Pasquetet al. 1994; Higgins 1995; Blackledge spin a spiral of elastic, adhesive silk onto the radial threads 1998; Herberstein et al. 2000; Blackledge & Zevenbergen using a combination of fibrous flagelliform (Flag) silk and 2007), prey taxon (Sandoval 1994; Tso et al. 2007), competi- gluey aggregate (Ag) silk (Foelix 1996). Together these tion with other individuals (Leborgne & Pasquet 1987; Ward different silks compose a planar orb (Fig. 1) that functions & Lubin 1992), frequency of damage by non-prey animals in first intercepting insects, then absorbing the kinetic energy (Chmiel et al. 2000), and size of the vegetation scaffold on oftheir flight, and finally adhering to the insects long enough which the web is constructed (Lubin et al. 1991). Typically, for the spider to capture them (Blackledge & Hayashi 2006). such studies document changes in specific web properties Many studies oforb webs are based on laboratory-confined including size ofthe capture area as well as the total length or spiders due to experimental convenience and control over the spacing ofthreads, and then attempt to relate these changes to environment (Zschokke & Herberstein 2005). However, such foraging investment underdifferent environmental conditions. spiders spin webs in an environment that is fundamentally However, accurately assessing foraging investment is different from their natural habitat. This could potentially difficult, as it may be possible to divert available silk resources influence many features of webs (Brown 1981; Gillespie & in multiple ways to achieve equivalent foraging success. For Caraco 1987; Higgins et al. 2001). Here, we compare the example, a web with closely spaced threads is very good at difference in web investment between spiders spinning in an stoppingand retaining large insects, but a largerweb increases outdoor setting along fence railings that impose size con- the number of interceptions (Eberhard 1986; ap Rhisiart & straints on webs but is otherwise natural and a laboratory Vollrath 1994; Nakata & Ushimaru 2004; Blackledge & environment that imposes not just size constraints but also Eliason 2007). For a given volume of silk, thick, high energy changes in prey cues, web supports and weather, using the furrowed orb spider Larionioides cornutus (Clerck 1757) 3Corresponding author. E-mail: [email protected] (Araneae: Araneidae). We test the prediction that the 183 184 THE JOURNAL OF ARACHNOLOGY Hh made structures such as bridges (Burgess & Uetz 1982; seealso & Heiling 1999; Heiling Herberstein 1999). These spiders often spin webs at unusually high population densities for orb- weaving spiders. All of the spiders in this study came from a 100-m longbridgeoverthe Cuyahoga River in Akron, Summit County, Ohio, USA. Multiple generations of spiders coex- isted, and webswere incloseproximity but not interconnected. These spiders usually initiate web construction at dusk and then hunt at the hub during the night while remaining in a retreat located above or to the side ofthe web duringdaylight. — Fence spiders. We sampled the webs of 1 1 sexually mature females [132 ± 64 mg, (mean ± SD) and 4.3 ± 0.6 mm carapace length] on the fence railings of the bridge in August and September 2008. This population was also the source of the specimens used for the cage treatment. Fence webs were spun between the vertical confines of wooden railings, 18- 25 cm apart; or metal railings, 23-24cm apart (Fig. 2). Spiders were collected from the webs and web measurements made between 21:00 and 23:00 h, shortly after webs were spun. We — recognized freshly spun webs by their intact spirals and radial Figure 1. Web architecture parameters measured: Vertical dia- spokes as well as the lack of insects or detritus. — meter(dv)ofthecapturearea, horizontal diameterofthecapturearea Caged spiders. Thirteen sexually mature female L. cornu- (dh), upper radial length (ru), lower radial length (r,), upper free zone tus (131 mg ± 63 mg and 4.7 mm ± 0.6 mm carapace length) (dleeHlnvig)mt,hita(endHdrbuy)f,rtleheoewezoruotnfeerreemhozosortinzesotnlitecanklgythsdpii(raHarmlies).t,efrree(Hzoh)n.eCvaerpttiucarlediaarmeeateirs 4w0ereXca1p0tucremd iinn Atuhgeusltab2o0r0a7t,orpyl,acaenddinaclalgoewsedmetaosusrpiinngw40ebXs (Agnarsson & Blackledge 2009). The cages were composed of metal frames, with clear plexiglass on the 40 X 40 cm faces, laboratory environment will affect silk investment and web and insect screen on the remaining four 10-cm-wide edges. architecture and find that some measures of investment are Spiders experienced minimal temperature variation (22 ± 3 reduced in the laboratory environment. °C), airflow and insect cues, as well as a constant 15:9 h L:D cycle in the laboratory. The cages also provided stiff vertical METHODS supports for the sides of webs, which were generally lacking — Study species. We compared silk investment and architec- for one or both sides of fence webs. Spiders typically ture ofwebs spun by a single population of L. cornutus in the constructed webs within 1-5 days of captivity and were not field versus laboratory. L. cornutus is a common nocturnal fed prior to web spinning. orb-weaving spider, a cosmopolitan species known for Fence (n = 11) and caged spiders (n = 13) were equivalent building webs above or near water, particularly on human- in body mass and carapace width but differed in body cage fence Figure 2.—Sketchsummarizingtheimportantdifferencesbetweencageand fencewebs. Therelativesizesofthesubstratesreflect thedifferent shapesofspaceavailabletospidersinthelaboratoryand field. Cagesmeasured40 X 40 X 10cm,whilefencerailingswereseparated by 18-25cm but allowed semi-unconstrained widths ofwebs. Only 1/3 ofthe typical spiral threads are shown, but radii count is similar to real webs. — SENSENIG ET AL.—SPIDERS ALTER WEBS IN THE LABORATORY 185 spiTdaebrlseus1e.d—iBnotdheycapgaera(mne=ter1s3) a(nmeddfieannc,e(mni=ni1m1u)mw,ebmcaoxmipmaruims)onso.f thoyproetahcehsiaznedintsoecdteccaruegashettihnethteimleowneercecsastacrhyrfeogriaonhe(aMvaystseprisde&r Weightandcarapacewidthdid notdifferin medianbetweencageand Moffet 1983; Nentwig 1985; Herberstein & Heiling 1999; fencespiders(Mann-Whitney Utestbyranks, P = 1.00and P= 0.28, Heiling 2004). Alternately, the larger lower capture area may respectively). Body condition was significantly lower in cage spiders reduce the high metabolic cost of raising the abdomen above (Mann-Whitney U test by ranks, P = 0.01). the body when spiders spin the capture spiral in the upper region of the web (Herberstein & Heiling 1999). Hub (medCiaagne, smpiind,ermsax) (meFdieannc,emsipnid,ermsax) asymmetry is defined in Blackledge & Gillespie (2002) as; Weight (mg) 102, 64, 265 117, 42, 213 Hub asymmetry= 1 —ru/q. Carapace width Most orb webs are not round, but rather elliptical. The (mm) 3.7, 3.2, 5 3.6, 2.6, 4.1 exaggerated vertical axis of webs may facilitate prey capture Body condition -0.12, -0.33, 0.27 0.07, -0.07, 0.16 by taking advantage of the tendency of prey to tumble downward as they struggle. Web asymmetry describes the condition, with fence spiders having higher body condition relationship between height and width of the orb and is & (Table 1). For each spider, we measured aspects of silk defined in Blackledge Gillespie (2002) as; structure and orb-web architecture that played potentially Web asymmetry= 1—dh/dv. critical roles in prey capture and allowed quantification of total silk investment by spiders. Perfectlyroundwebshaveawebasymmetryvalueof0. Aweb Web architecture measurements. For caged spiders, web asymmetry < 0 indicates a web that is wider than it is tall, and architecture was measured from digital photographs taken web asymmetry >—0 indicates a web that is taller than wide. from webs placed in front of a shadow box and illuminated Silk structure. Finally, investment of silk in a web can be from the sides by fluorescent lighting (Langer & Eberhard quantified by measuring the structural sizes of threads and 1969; Zschokke & Herberstein 2005). For fenced spiders, it glue droplets. For each web, four radial threads, one from was difficult to obtain clear photographs ofentirewebs so that each cardinal axis, were collected onto cardboard holders architecture was instead measured directly in the field using described in AgnarssMon & Blackledge (2009). Cyanoacrylmatme techniques described by Blackledge et al. (2003). For both adhesive (Superglue1 ) was applied to two sides of a 16 types of webs, we measured the vertical diameter of the hole in thecenter ofthe card, which was then pressed against a capture area (dv), horizontal diameter ofthe capture area (dh), radius. After drying for several seconds, the radial thread was upper radial length (ru), lower radial length (q), upper free cut on either end of the card using a portable soldering iron, vzeornteicallendgitahme(tHerru(),Hvl),owaenrd ffrreeee zzoonneeholreingztohnta(lHqd)i,amferteeerz(oHnhe) relTeoasimnegatshuerethrspeiardalfrtohrmeatdhedwieabm.eter, the silk was collected directly onto a glass microscope slide by placing the slide (Fig. 1). Capture area is delimited by the outermost sticky spirals behind the four outermost strands and then gently pressing the and determines the number of insects that a web potentially slide against the web. The spiral threads were then cut along intercepts. Capture area was measured directly from photo- the edges ofthe slide with a portable soldering iron. The glass graphs of laboratory webs using Image J (Rasband 1997— slide caused the glue droplets to adhere and flatten, thereby 2R0a0d9i)i-aHnudbwfaosrmcuallacu(lBaltaecdklfeodrgefi&eldGiwlelbesspiuesi2n0g02)t:he Adjusted stehcrueraidngvistihbele.thWreeadussetdo tthhee sslaimdee apnrdocemdaukriengastAhgencaorrsesoanxia&l Blackledge (2009) and Blackledge et al. (2005) to measure Capture area= \}/2n rau2—V2TC(Hru)2j rmaidciraolscaonpdy astpi1r0al00Xthrmeaagdnifdiicaamtieotne.rs using polarized light + [1/2Ttrai2-1/27t (Hq)2], calTcoutlaaltevdoblyumfeirsotfdfeltaegremlilniifonrgmtshielktoitnalthleencgatphtuorfethsepicraalptwuarse where rau=(ru+dh/2)/2 and ra!=(r,+- -dh/2)/2. s(pSihraelr,matynpi1c9a9l4l)y: designated as capture thread length (CTL) The density of threads within webs influences the sizes of CTL=7i (average # spirals along the 4 web axes) insects likely to be intercepted, but more importantly affects [(ru+dh+q)/4—(Hru+Hh+Hq)/4]. both the ability ofwebs to absorb the kinetic energy offlying insects and whether webs retain insects long enough to be The factor in the brackets represents the average width ofthe captured by spiders (Blackledge & Zevenbergen 2006). There- capture area and is estimated by subtracting the average free fore we counted the number ofradii and the number ofspiral zone radius along the four cardinal axes from the average threads along four axes (top, bottom, left, and right). Mesh capture area radius along the four cardinal axes. width, the distance between adjacent capture spirals, was Volume was then computed as: calculated along each axis and then averaged (Herberstein & = Tso 2000). total spiral thread volume Spiders usually place the web hub above the center of the (CTL) n (hypothetical spiral diameter/2)2 . orb, resulting in a web with a larger lower capture region (Krink & Vollrath 1997, 2000). This hub asymmetry typically The diameter of a hypothetical thread that would be increases with spider size (Kuntner et al. 2008) and is equivalent in cross-sectional area to the two strands of P P 186 THE JOURNAL OF ARACHNOLOGY spiral Fibers that typically compose capture spirals was STATISTICAL METHODS — calculated as: Comparing spider body condition. Because spiders were hypothetical spiral thread diameter=2 [2n rss2]°'5Jn, caoultluemctne,d wferofimrstthdeetesrammienedbrwihdgeetheornthceonsspeicduetrisvefryoemarfsencien webs and cage webs differed in body size or condition. Body where rss was the measured radius of a single strand, condition was calculated as the residuals of the regression of assuming equal radius of each strand. Hypothetical radial thread diameter was calculated in an identical manner, also l1o9g96)w.eiMgahntn-onWthoitlnoegycUaratpesatcsewewriedtuhse(dTatboltees1t)fo(rJadkioffbereetncaels. assuming that the radial thread was composed of two between weight, carapace width, and spider body condition, equally sized strands. In the rare instances in which we because these variables were not normally distributed. These observed four-stranded radial or capture threads, all four tests found that spiderscollected from fence and cage websdid softforramJnu2dlsawwaessrheoiwnacnslsuuadmbeeoddvei.tno Ttbheheeofhayevpqeoutrahalegtediicahamyleptoettrhh,reetaaidncadldaiatfmhaercteteaodrr bnuott ddiifdfedrififnerwienigbhotdy(Pco=ndLitOiOo)no(r c=ara0.p0a2c)e, wwiitdhthfe(nPce=sp0i.d2e8r)s, diameter for a specific web and specific silk type was then having higher body condition (Table 1). The most likely reason for this difference is that the cohort associated with calculated as the average of the four collected thread each year experienced slightly different foraging histories. samples. Thus, location is confounded with body condition, so that Total radial thread volume was calculated as: some differences between fence and cage webs could be due to Total radial thread volume difference in body c—ondition. =(average single radial length) Comparing webs. The effect of captivity on web architec- ture was then tested using Multivariate Analysis of Variance (# radial threads)(average cross-section area) =(nru(+hydpoht+hertui)c/a4l(rraaddiiailctohurneta)d diameter/2)2 (vaMriAaNblOesVAin)thiemmpoldeemlen(tTeadblein2)S.taCtriosstsi-csaec6t.1i.onWalearienacloufdreaddia13l . (MA) and spiral (Flag) threads, volume of capture silk, To measure glue droplet volume on the capture spiral silk, volume ofglue, and glue droplet spacing were logtransformed we suspended threads between 3 mm diameter wooden to meet assumptions of the normal distribution (Shapiro- supports secured to a microscope slide. The four outermost Wilks Wtest, P > 0.05) and homogeneous variance (Levene’s capture spiral rows, adjacent to those already captured by Test for Homogeneity ofVariances, P > 0.05). Radial volume direct adhesion to glass, were collected simultaneously by was transformed by a power exponent of 0.25 to achieve pressing this slide against the web so that the threads bridged normality. Radii number and web asymmetry were normally the gap between the parallel supports (Agnarsson & Black- distributed but could not be successfully transformed to ledge 2009). Glue droplets were photographed at 10 or 100X achieve homogenous variance between cage and fence webs. magnification, and the length and width ofthe first and third There is no readily available non-parametric equivalent to droplet from the left edge of the photo were measured. This MANOVA, hence the MANOVA was also performed without avoided experimenter bias toward measuring large or small these two variables to confirm that their inclusion did not droplets. We did not measure the much smaller secondary affect the results. Post-hoc unequal n honest significant droplets that often occur between primary droplets because difference (HSD) mean comparison tests were used to detect secondary droplets contribute minimally to total glue volume which variable means differed between webs on fences and in (Opell & Hendricks 2007). The volume ofa single droplet of cages. Because radialcount and webasymmetrydid not satisfy glue (SDV) was calculated as: all assumptions of parametric tests, and radial volume did so only marginally even after transformation, we also performed SDV=(droplet width)2(droplet length)/15. univariate Mann-Whitney U tests by ranks for these three variables as raw variables. This formula accounts for the anisotropy ofthe droplet shape, which tends to be longer than wide as it adheres to the spiral thread (Opell et al. 2008). The average distance between glue There was a significant difference in web properties between vdroolpulmetesofwagsluemewiatshuirnedtheacwreobsswa1s0 tahdejnacceanltculdartoepdleatss:. Total c8.a3g7e,anPd f=enc0e.0s0p8i)d.ersUn(eMqAuaNlOVnA,HSWiDlksteFsts—i0n.d0i5c2a,te/dq3it6ha=t Total glue volume= laboratory webs had shorter CTL (P = 0.008) (Fig. 3D), (SDV)(CTL)/average distance between droplets. smaller radial cross-sectional area ( = 0.03), and lower asymmetry (P = 0.006) than field webs (Table 2). The Droplet size can increase with the relative humidity of the significant effect of web location in the overall model and air due to the hygroscopicity of glue silk (Mark et al. 1991; the significant post-hoc mean differences of CTL and radial Opell & Schwend 2008). Laboratory relative humidity was cross-sectional area were not changed by excluding the non- between 40-60% R.H. during measurement ofall threads and parametric variables, web asymmetry and radial count, from MANOVA thus was lower than the occasional high humidity (90%) that the model. occurred in the field. For all spiders, droplets were measured The univariate Mann-Whitney U tests identified lower within approximately 2 h ofcollection, and hence 2-4 h after median web asymmetry in fence webs (fence median -0.27, web production, to minimize any effect of drying or swelling range —0.8 to 0.5, cage median 0.11, range —0.18 to 0.4, on droplet volume. Mann-Whitney U — 19, P = 0.003, Fig. 3G). The Mann- SENSENIG ET AL.—SPIDERS ALTER WEBS IN THE LABORATORY 187 Table 2—Webarchitecture(mean ± SD,ri)compared betweencageand fencespiders. Forthevariablesthat met thenormaldistribution and homogeneous variance assumptions ofMANOVA, significant post-hoc unequal n HSD mean differences between cage and field are indicated witha * (P< 0.05). Significant median differences identified by the Mann-Whitney Utestsare indicated with a f (P < 0.05). Sample size forthe slue measurements was smaller than for the other samples due to accidental destruction prior to measurement. Capture thread length is abbreviated as CTL. % Cage webs Fence webs difference Web architecture Capture area (cnr) 462 ± 213 (13) 593 ± 247 (11) 22 Number ofradii 17.8 ± 2.0 (13) 21 ± 3.2 (11) 14t CTL (cm) 555 ± 217 (13) 963 ± 277 (11) 42*f Mesh width (mm) 4.6 ± 0.8 (13) 4.2 ± 0.8 (11) -10 Web asymmetry 0.1 ± 0.1 (13) -0.3 ± 0.3 (11) 136f Hub asymmetry 0.3 ± 0.2 (13) 0.2 ± 0.2 (11) —50 Silk structure Radial (MA) cross-section area(pnr) 5.6 ± 3.6 (13) 11.3 ± 5.8 (11) 50*| Capture spiral (Flag) cross-section area (pm2) 6.7 ± 1.9 (13) 4.5 ± 1.0 (11) —33 G#lugeluseindgrloepdlretosp/vmoml.(pm3) 708177 ±± 65(06122)(12) 780106 ±± 655(992) (8) -180 Silk investment Radial volume (mm3 0.016 ± 0.008 (13) 0.034 ± 0.019 (11) 53f ) Spiral volume (mm3) 0.026 ± 0.020 (13) 0.045 ± 0.029 (11) 41 Glue volume (mm3 0.6 ± 0.5 (12) 1.0 ± 0.7 (8) 38 ) Whitney U tests also identified a greater median number of cage. Our study primarily addresses the concerted changes of radii in fencewebs (fence median 22, range 15-25, cage median silk and web architecture. Secondarily, we speculate on the 18, range 15-22, Mann-Whitney U = 35, P - 0.037, Fig. 3B) adaptive significance of those changes. and greater median radial volume in fence webs (fence median Decreased investment in silk likely has implications for web 0.033 mm3 range 0.006 to 0.07, cage median 0.013 mm3 range performance. In other web systems, recent investigations of , , 0.004-0.03, Mann-Whitney U = 28, P = 0.01, Fig. 3C, the tradeoffs inherent in modifying web architectures revealed Table 2, Fig. 4A). Because of the non-homogeneous and that theeffectivecapture oflarger preydepends more upon the marginally homogeneous variance of these two variables increased energy absorption and stickinesss supplied by a (Levene’s Test for Homogeneity of Variances, P — 0.03, P concentrated capture spiral than on increased capture area = 0.08), respectively, the Mann-Whitney U tests may have (Blackledge & Zevenbergen 2006; Blackledge & Eliason 2007). offered greater power to reject the null hypothesis than the Highcapture area remains, however, as an effective strategy of MANOVA. increasing interception rate of all prey sizes and successful capture ofsmall prey. Larger mesh width, in conjunction with DISCUSSION greatercapturearea, has been reported as a typical response to Most research on energetic investment associated with larger prey (Herberstein & Heiling 1998; Schneider & Vollrath construction of orb webs focuses on total thread length (e.g. 1998). However, all of these studies largely assume that size Turnbull 1964; Nakata & Ushimaru 2004; Kawamoto & and mechanical properties of silk threads are invariant in Japyassu 2008). However, this ignores the important con- different spinning scenarios. We show here that, at least in this tribution of thread diameter to web function and total silk species, the common simplification of invariant thread size is investment. Our study quantifies this potentially important not valid for radial threads, but the data are consistent with parameter and directly compares web architecture and silk bigger webs as “better” webs. For L. cornutuscaged webs, the investment between fence and cage webs to gain a more reduced capture area, increased mesh width, and decreased accurate estimate of the material investment of spiders in silk volume all predict that these webs should function poorly webs. On average, cage webs were smaller and rounder than at intercepting, stopping and retaining prey compared to field fence webs, contained shorter lengths of capture spirals, and webs. The reduction ofradial thread cross-sectional area in L. were supported by fewer, as well as thinner, radii. Thus, cornutus caged webs thus accompanies the reduction in many spiders invested significantly less radial (MA) silk in the cage other parameters associated with high prey energy absorption, webs. There was also aconsistent, albeit non-significant, trend suggesting a concerted decline in web investment rather than a toward lowervolumes offlagelliform capture spiral silk (Flag) compensatory effect. and aggregate glue (Ag). These differences suggest that spiders Spiders in laboratory cages built architecturally different may initially reduce silk investment in webs when moved into webs from those on fences. Such shape variation may result in cages. However, several other factors, particularly spider part from the reduced silk investment in cage webs, but also condition, cohort, and available web frame size may also likely relates to thecharacteristics ofthe available supports for contribute to the observed differences in investment. These webs in cages versus fences. In both environments, web factors were confounded with the transition from fence to spinning was constrained by the rigid dimensions of the 188 THE JOURNAL OF ARACHNOLOGY cage fence > 0.04 Q-rsm cage fence cage fence cage fence — Figure 3. Web architecture and silk investment in laboratory cage and fence webs (median indicated by small box, 25-75 percentiles indicated by large box, and range indicated by capped bars). Non-parametricvaluesofmedian, minimum, and maximum areshown becausethe parametric parameters of mean and standard deviation are presented in Table 2. Statistically significant differences in MANOVA post-hoc unequaln HSD testsare indicated by * (P< 0.05). Significant median differences identified by the Mann-Whitney Utestsare indicated with at (P < 0.05). A. Capture area (MANOVA post-hoc unequal n HSD test, P = 0.52); B. Number ofradii (Mann-Whitney U= 35, P = 0.037); C. Radial volume (Mann-Whitney U = 35, P = 0.037); D. Capture thread length, CTL (MANOVA post-hoc unequal n HSD test, P = 0.008); E. Mesh width (MANOVA post-hoc unequal n HSD test, P = 0.21); F. Total spiral volume (MANOVA post-hoc unequal n HSD test, P = 0.31); G. Webasymmetry(Mann-Whitney U = 19, P = 0.003); H. Hubasymmetry(MANOVA post-hoc unequaln HSD test, P = 0.13); I. Totalglue volume (MANOVA post-hoc unequal n HSD test, P = 0.18). SENSENIG ET AL.—SPIDERS ALTER WEBS IN THE LABORATORY 189 26 18 fit between the available fence rails. Therefore, they expanded 24 laterally and ultimately produced capture areas similar to cage <<ri6 22 webs. This is similar to a study on A. diadematus where orb- 20 314 web shape changed to optimally fill the space available in $ small, irregularly shaped cages, but only after overall web size 18 was drastically reduced (Krink & Vollrath 1997, 2000). This 16 suggests that spiders do have the capacity to assess the 10 14 | microhabitat available for webs and subsequently adjust the 12 1 8 shapes ofwebs to maintain or maximize overall sizes. 10 ai Most research on orb-web function primarily focuses on 8 % 6 changes in the shapes of webs and lengths of threads. | 6 ! 4 However, there is growing evidence that spiders actively vary to 4 2 the diameters of silk threads within and between webs tS5 2 tao 2 (Blackle&dge et al. 2005; Blackledge & Zevenbergen 2007; Boutry Blackledge 2008). In general spiders increase silk 0 diameter with body size, particularly given the role of cage fence cage fence draglines in suspending falling and hanging spiders (Brand- co wood 1985; Osaki 1996; Ortlepp & Gosline 2008). Some j|l6000 90 spiders even control silk diameter in response to different types ofprey (Boutry & Blackledge 2008). The cross-sectional areas ©14000 80 of silk threads directly influence important mechanical Q. |12000 properties such as the total loading and energy-absorbing 70 capabilities ofwebs. In our study the greater volume ofradial O§)10000 ® 60 (MA) silk was driven primarily by the 50% greater cross- © 8000 © sectional areas of radii in fence webs, which would greatly CD 150 increase the kinetic energy these webs could absorb from wind i 6000 o and prey impacts. In contrast, the cross-sectional areas of ©40 capture spirals were relatively similar between laboratory and S 4000 fence webs, although the total length ofthe capture spiral was CD 1 2000 30 42% shorter in cage webs. o > In summary, we found that caged spiders in the laboratory 0 20 cage fence cage fence invested less material in their webs than did fence spiders in the field. Several factors may explain the lower investment in — Figure 4. Thread properties of cage and fence webs (median webs by cage spiders: 1) spiders may first test foraging sites indicated by small box, 25-75 percentiles indicated by large box, and before building more substantial webs in new locations range indicated by capped bars). Statistically significant differences in (Riechert & Gillespie 1986; Nakata & Ushimaru 1999); 2) MANOVA post-hoc unequal n HSD test indicated by *(P < 0.05). caged spiders were exposed to fewer insect cues, such as wing Significant median differences identified by the Mann-Whitney Utests vibrations and odor, and may have altered web spinning in are indicated with a f (P< 0.05). A. Radialcrosssectional area ofthe response to perceptions of a poor foraging environment daroeuabloef-stthreadnodeudblrea-dsitarla(ndMeAd)cfaipbteurr(eP(F=la0g.)03f)i;beBr.(SPpir=al0.c7r4o)s;s-Cs.ecStiinongalle (Pasquet et al. 1994 Nakata & Ushimaru 2004); 3) the stiff gluedroplet volume (P = 0.79); D. Gluedroplet spacing(P = 0.77). supports ofthe laboratory cages provided structures on which mechanically effective webs could be built using less material (Wirth& Barth 1992)and reducedinvestmentmayreflect lossof surroundings. Cage spiders could, and did, attach frame threads above, below, and to either side ofthe web, resulting silk resources when a spider is removed from its old web in the field without being allowed to recycle the silk (Zschokke 1997). in webs supported along both vertical and horizontal axes. In Our study was not intended to distinguish among these factors, contrast, the frame threads of webs on fences were usually but only to determine the general effects ofcaptivity on spider attached only to upper and lower fence railings with no lateral webs. Controlling for such factors in future studies of orb support, resulting in webs with tension derived solely from the spinning could reveal the relative importance ofeach for spider atolponagndtwbootatxoems (rFaitgh.er2).thWahniloenespemcualyatiivmep,rsouvpepofrutnicntgiownebisn bbeohdayvioofr.evRiedgeanrcdeletshsa,t sopuirdesrtsumdoydiuslcaotnesiwsetbe-ntspwiintnhintghebeghraovwiionrgs mechanical systems such as orb webs that must distribute the in response to changing environments. energy from impacts of flying prey and wind (Vollrath 1992; Lin et al. 1995). In other words, spiders may have reacted to ACKNOWLEDGMENTS the better support available to webs in laboratory cages by This research was funded by awards DEB-0516038, DBI- reducing the total number of radii in webs, while still 0521261 and IOS-0745379 from the National Science Founda- maintaining the same effective degree of overall stiffness and tion. function. Cagewebs wereelongated vertically (web asymmetry = 0.1) while fence webs were wider than tall (web asymmetry LITERATURE CITED = -0.3). This shape is unusual for orb webs, but fence spiders Agnarsson, I. & T.A. Blackledge. 2009. Can a spider web be too expended relatively large volumes ofsilk into webs that had to sticky?Tensilemechanicsconstrains theevolution ofcapturespiral 190 THE JOURNAL OF ARACHNOLOGY stickiness in orb weaving spiders. Journal of Zoology (London) Henschel, J.R. & Y.D. Lubin. 1992. Environmental factors affecting 278:134-140. the web and activity of a psammophilous spider in the Namib ap Rhisiart, A. & F. Vollrath. 1994. Design featuresoftheorb webof Desert. Journal ofArid Environments 22:173-189. the spider, Araneus diadematus. Behavioral Ecology 5:280-287. Herberstein, M.E., C.L. Craig & M.A. Elgar. 1997. Optimal foraging Blackledge, T.A. 1998. Stabilimentum variation and foraging success behaviour: web investment and web decorations in Argiope in Argiope aurantia and Argiope trifasciata (Araneae: Araneidae). keyserlingi. [Abstract], Advances in Ethology 32:185. Journal ofZoology (London) 246:21-27. Herberstein, M.E., A.C. Gaskett, D. Glencross, S. Hart, S. Jaensch & Blackledge, T.A., G.J. Binford & R.G. Gillespie. 2003. Resource use M.A. Elgar. 2000. Does the presence ofpotential prey affect web within a community of Hawaiian spiders (Araneae: Tetragnathi- design in Argiope keyserlingi (Araneae, Araneidae)? Journal of dae). Annales Zoologici Fennici 40:293-303. Arachnology 28:346-350. Blackledge, T.A., R.A. Cardullo & C.Y. Hayashi. 2005. Polarized Herberstein, M.E. & A.M. Heiling. 1998. Does mesh height influence light microscopy, variability in spider silk diameters, and the prey length in orb-web spiders (Araneae)? European Journal of mechanical characterization of spider silk. Invertebrate Biology Entomology 95:367-371. 124:165-173. Herberstein, M.E. & A.M. Heiling. 1999. Asymmetry in spider orb Blackledge, T.A. & C.M. Eliason. 2007. Functionally independent webs: a result of physical constraints? Animal Behaviour components of prey capture are architecturally constrained in 58:1241-1246. spider orb webs. Biology Letters 3:456-458. Herberstein, M.E. & i.M. Tso. 2000. Evaluation of formulae to Blackledge, T.A. & R.G. Gillespie. 2002. Estimation ofcapture areas estimate the capture area and mesh height of orb webs of spider webs in relation to web asymmetry. Journal of (Araneoidea, Araneae). Journal ofArachnology 28:180-184. Arachnology 30:70-77. Higgins, L.E. 1995. Direct evidence for trade-offs between foraging Blackledge,T.A. &C.Y. Hayashi. 2006. Silken toolkits: biomechanics and growth in ajuvenile spider. Journal ofArachnology 23:37-43. ofsilk fibersspun bytheorbwebspiderArgiopeargentata. Journal Higgins, L.E. & R.E. Buskirk. 1992. Atrap-buildingpredatorexhibits ofExperimental Biology 209:2452-2461. different tactics for different aspects offoraging behavior. Animal Blackledge, T.A. & J.W. Wenzel. 2001. State-determinate foraging Behaviour 44:485-499. decisions and web architecture in the spider Dictyna volucripes Higgins, L.E., M.A. Townley, E.K. Tillinghast& M.A. Rankin. 2001. (Araneae, Dictynidae). Ethology Ecology& Evolution 13:105-113. Variation in the chemical composition of orb webs built by the Blackledge, T.A. & J.M. Zevenbergen. 2006. Mesh width influences spider Nephila clavipes (Araneae, Tetragnathidae). Journal of prey retention in spider orb webs. Ethology 112:1 194-1201. Arachnology 29:82-94. Blackledge, T.A. & J.M. Zevenbergen. 2007. Condition dependent Jakob, E.M., S.D. Marshall & G.W. Uetz. 1996. Estimating fitness: a spider web architecture in the western black widow Latrodectus comparison ofbody condition indices. Oikos 77:61-67. hesperus. Animal Behaviour 73:855-864. Kawamoto, T.H. & H.F. Japyassu. 2008. Tenacity and silk Boutry, C. &T.A. Blackledge. 2008. Thecommon house spider alters investment of two orb weavers: considerations about diversifica- the material and mechanical properties ofcobweb silk in response tion ofthe Araneoidea. Journal ofArachnology 36:418-424. to different prey. Journal ofExperimental Zoology 309A:542-552. Krink, T. & F. Vollrath. 1997. Analysing spider web-building Brandwood, A. 1985. Mechanical properties and factors ofsafety of behaviour with rule-based simulations and genetic algorithms. spider drag-lines. Journal of Experimental Biology 116:141-151. Journal ofTheoretical Biology 185:321-331. Brown, K.M. 1981. Foraging ecology and niche partitioning in orb- Krink, T. & F. Vollrath. 2000. Optimal area use in orb webs ofthe weaving spiders. Oecologia 50:380-385. spider Araneus diadematus. Naturwissenschaften 87:90-93. Burgess, J.W. &G.W. Uetz. 1982. Social spacingstrategies inspiders. Kuntner, M., J.A. Coddington & G. Hormiga. 2008. Phylogeny of Pp. 317-351. In Spider Communication: Mechanisms and Ecolo- extant nephilid orb-weaving spiders (Araneae, Nephilidae): testing gical Significance. (P.N. Witt & J.S. Rovner, eds.). Princeton morphological and ethological homologies. Cladistics 24:147-217. University Press, Princeton, New Jersey. Langer, R.M. & W. Eberhard. 1969. Laboratory photography of Chmiel, K., M.E. Herberstein & M.A. Elgar. 2000. Web damage and spider silk. American Zoologist 9:97-101. feedingexperience influenceweb site tenacity in theorb-web spider Leborgne, R. & A. Pasquet. 1987. Influences ofaggregative behavior Argiope keyserlingi Karsch. Animal Behaviour 60:821-826. on space occupation in the spider Zygiella x-notata (Clerck). Eberhard, W.G. 1971. The ecology of the web of Uloborus diversus Behavioral Ecology and Sociobiology 20:203-208. (Araneae-Uloboridae). Oecologia 6:328-342. Lin, L.H., D.T. Edmonds& F. Vollrath. 1995. Structural engineering Eberhard, W.G. 1986. Web-building behavior of anapid, symphy- ofan orb-spider’s web. Nature 373:146-148. tognathid and mysmenid spiders (Araneae). Journal ofArachnol- Lubin, Y., M. Kotzman & S. Ellner. 1991. Ontogenetic and seasonal ogy 14:339-356. changes in webs and websites ofa desert widow spider. Journal of Eberhard, W.G. 1990. Function and phylogeny of spider webs. Arachnology 19:40^18. Annual Review of Ecology and Systematics 21:341-372. Mark, A.T., T.B. David, S.G. Kathleen & K.T. Edward. 1991. Foelix, R.F. 1996. Biology of Spiders. Second edition. Oxford Comparative study oforb web hygroscopicity and adhesive spiral University Press, New York. composition in three araneid spiders. Journal of Experimental Gillespie, R.G. & T. Caraco. 1987. Risk-sensitive foraging strategies Zoology 259:154-165. oftwo spider populations. Ecology 68:887-899. Masters, W.M. & A.J.M. Moffet. 1983. A functional explanation of Heiling, A.M. 1999. Why do nocturnal orb-web spiders (Araneidae) top-bottom asymmetry in vertical orb webs. Animal Behaviour search for light? Behavioral Ecology and Sociobiology 46:43-49. 31:1043-1046. Heiling, A.M. 2004. Effect ofspider position on prey capture success Nakata,K.2007.Preydetectionwithoutsuccessfulcaptureaffectsspider’s and orb-web design. Acta Zoologica Sinica 50:554-565. orb-webbuildingbehaviour. Naturwissenschaften94:853-857. Heiling, A.M. & M.E. Herberstein. 1999. The importance of being Nakata, K. & A. Ushimaru. 1999. Feeding experience affects web larger: Intraspecific competition for prime web sites in orb-web relocation and investment in web threads in an orb-web spider, spiders (Araneae, Araneidae). Behaviour 136:669-677. Cyclosa argenteoalba. Animal Behaviour 57:1251-1255. Heiling, A.M. & M.E. Herberstein. 2000. Interpretations oforb-web Nakata, K. & A. Ushimaru. 2004. Difference in web construction variability: A reviewofpast andcurrent ideas. Ekologia-Bratislava behaviorat newly occupied web sites between two Cyclosaspecies. 19:97-106. Ethology 110:397-411. SENSENIG ET AL.—SPIDERS ALTER WEBS IN THE LABORATORY 191 Nentwig, W. 1985. Top-bottom asymmetry in vertical orbwebs: A the sheet-web spider, Frontinellina cf. frutetorum (Araneae functional explanation and attendant complications. Oecologia Linyphiidae). Ethology Ecology & Evolution 16:291-298. 67:111-112. Shear, W.A. (ed.). 1986. Spiders: Webs, Behavior, and Evolution. Opell, B.D. & M.L. Hendricks. 2007. Adhesive recruitment by the Stanford University Press, Stanford, California. viscous capture threads ofaraneoid orb-weaving spiders. Journal Sherman, P.M. 1994. The orb web - an energetic and behavioral ofExperimental Biology 210:553-560. estimator of a spiders dynamic foraging and reproductive Opell, B.D., B.J. Markley, C.D. Hannum & M.L. Hendricks. 2008. strategies. Animal Behaviour 48:19-34. The contribution of axial fiber extensibility to the adhesion of Tso, I.M., S.Y. Jiang & T.A. Blackledge. 2007. Does the giant wood viscous capture threads spun by orb-weaving spiders. Journal of spider Nephilapilipes respond to prey variation by altering web or Experimental Biology 211:2243-2251. silk properties? Ethology 113:324-333. Opell, B.D. & H.S. Schwend. 2008. Persistent stickiness of viscous Turnbull, A.L. 1964. The search for prey by a web-building spider capture threads produced by araneoid orb-weaving spiders. Achaearanea tepidarionon (C. L. Koch) (Araneae, Theridiidae). Journal ofExperimental Zoology 309A:11-16. Canadian Entomologist 96:568-579. Ortlepp, C. & J.M. Gosline. 2008. The scaling of safety factor in Vollrath, F. 1992. Spider webs and silks. Scientific American spider draglines. Journal ofExperimental Biology 211:2832-2840. 266:70-76. Osaki, S. 1996. Spider silk as mechanical lifeline. Nature 384:419. Ward, D. & Y. Lubin. 1992. Temporal and spatial segregation of Pasquet, A., A. Ridwan & R. Leborgne. 1994. Presence ofpotential web-building in a community of orb-weaving spiders. Journal of prey affects web building in an orb-weaving spider Zygiella x- Arachnology 20:73-87. notata. Animal Behaviour 47:477^480. Wirth, E. & F.G. Barth. 1992. Forces in the spider orb web. Journal Rasband, W.S. 1997-2009. ImageJ. U.S. National Institutes of ofComparative Physiology A 171:359-371. Health, Bethesda, Maryland. Zevenbergen, J.M., N.K. Schneider & T.A. Blackledge. 2008. Fine Riechert, S.E. & R.G. Gillespie. 1986. Habitat choice and utilization dining or fortress? Functional shifts in spider web architecture by in web-building spiders. Pp. 23^48. In Spiders: Webs, Behavior, the western black widow Latrodectus Hesperus. Animal Behaviour and Evolution. (W.A. Shear, ed.). Stanford University Press, 76:823-829. Stanford, California. Zschokke, S. 1997. Factors influencing the size of the orb web in Sandoval, C.P. 1994. Plasticity in web design in the spider Parawixia Araneus diadematus. Pp. 329-334. In Proceedings of the 16th bistriata - a response to variable prey type. Functional Ecology European Colloquium of Arachnology. (M. Zabka, ed.). Wyzsza 8:701-707. Szkola Rolnicko-Pedagogiczna, Siedlce, Poland. Schneider, J.M. & F. Vollrath. 1998. The effect of prey type on the Zschokke, S. & M.E. Herberstein. 2005. Laboratory methods for geometry of the capture web of Araneus diadematus. Naturwis- maintaining and studying web-building spiders. Journal of senschaften 85:391-394. Arachnology 33:205-213. Segoli, M.,A. Maklakov, E. Gavish, I.Tsurim&Y. Lubin. 2004.The effect of previous foraging success on web-building behaviour in Manuscript received 7 April2009. revised6 December 2009.