Evolution, 48(2), 1994, pp. 287-296 EVOLUTIONARY SHIFTS IN THE SPECTRAL PROPERTIES OF SPIDER SILKS CATHERINE L. CRAIG,' GARY D. BERNARD,2 AND JONATHAN A. CODDINGTON3 ' Department of Biology, Osborn Memorial Laboratories, Yale University. New Haven, Connecticut 06511 2Department of Electrical Engineering, University of Washington, Seattle, Washington 98195 ^Department of Entomology, Smithsonian Institution, 105 NHB, 10th and Constitution Avenue, Washington, D.C. 20560 Abstract. ? We measured the reflectance properties of unpigmented silks spun by a systematic array of primitive (Deinopoidea) and derived (Araneoidea) aerial, web-spinning spiders, as well as silks spun by Araneomorphae and Mygalomorphae spiders that do not spin aerial webs. Our data show that all of the primitive aerial web spinners produce catching silks with a spectral peak in the ultraviolet (UV), and cladistic analysis suggests that high UV reflection is the primitive character state for silk spectral properties. In contrast, all of the derived aerial web spinners produce silks that are spectrally flat or characterized by reduced reflectance in the UV. Correlated with the evolution of these catching silks is a 37-fold increase in species number and apparent habitat expansion. This suggests that the unique silk proteins spun by the araneoids have been important to their ecological and evolutionary diversity. Key words.? Spectral properties, spiders, spider silks. Received April 2, 1991. Accepted May 6, 1993. One basis of evolutionary change is the achievement of an adaptation or complex of ad- aptations that allows species to make better use of resources than their ancestors (Wright 1978). Numerous examples in the literature illustrate how molecular variations in proteins, shaped by natural selection, have allowed organisms to adapt to diverse habitats or microenvironments (reviewed in Gillespie 1991). Silks, a diverse group of crystalline proteins spun by all spiders, may be an example of a molecule that has been fundamental to the evolution of a predator's for- aging ecology. Although it has been suggested that the chem- ical composition of silks spun by spiders is much less variable than that of silks produced by larval insects (Rudall and Kenchington 1971), almost all research on the molecular, structural, and me- chanical properties of spider silk has been com- pleted on the support and viscid threads spun by fewer than 15 species of orb spinners (Lucas et al. 1955;Zemlin 1968; Denny 1976; Wainwright et al. 1976; Work 1977, 1981, 1985; Gosline et al. 1984; Craig 1987; Vollrath and Edmonds 1989; Xu and Lewis 1990; Dong et al. 1991; Beckwith 1991; Hinman and Lewis 1992). How- ever, the orb-spinning spiders and their close rel- atives (the Araneoidea) include only about one- third of all described spiders. Thus, there exists a broad diversity in the types of silks spun by aerial web-spinning spiders, in addition to silks spun by all other spider species, yet to be ex- plored. We propose that if the physical properties of silk proteins have been important to spider spe- ciation and the evolution of spider prey-capture methods, then silks spun by phylogenetically primitive and phylogenetically derived, aerial, web-spinning spiders will differ systematically. We tested this hypothesis by comparing the spec- tral properties of unpigmented silks spun by 16 species of spiders in the superfamilies Deino- poidea and Araneoidea. For outgroup compar- ison, we also examined the spectral properties of silks spun by five spiders that spin sheet webs and the retreat silks of three nonweb-spinning spiders. MATERIALS AND METHODS Spiders from Which Silks Were Sampled Silks were collected from spiders in the in- fraorders Mygalomorphae and Araneomorphae. Silks sampled from Mygalomorphae include those spun by Aphonopelma sp. (Theraphosi- dae), Ishnothele sp. (Dipluridae), and Hexurella sp. (Mecicobothriidae). Among the Araneomor- phae, the true spiders, we sampled silks spun by the most primitive spider in the group, Hypo- chilus sp. (Hypochilidae; Platnick 1977), and silks spun by a variety of more derived spiders in- 287 ) 1994 The Society for the Study of Evolution. All rights reserved. 288 CATHERINE L. CRAIG ET AL. eluding those in the families Eresidae (Stegody- phus sp.), Diguetidae (Diguetia), Salticidae (Phiddipus otiosus), Lycosidae {Sossipus sp.), and Filistatidae (Kukulcania sp.). The spectral properties of silks spun by spiders in the phylogenetically primitive, aerial web- spinning spiders, the Deinopoidea, and its sister taxon, the phylogenetically derived aerial web- spinning spiders, the Araneoidea (Coddington 1986), were examined in more detail. Among the Deinopoidea, silks were collected from webs spun by Deinopis sp. (Deinopidae), Uloborus glomosus (Walckenaer), Hyptiotes cavatus (Hentz), Philo- ponella tingena (Chamberlin and Ivie), and Mia- grammopes animotus (Chickering) (Uloboridae). Among the Araneoidea, we measured the spec- tral properties of viscid silks produced by 11 species in 10 genera of spiders and four families (Araneidae, Tetragnathidae, Theridiosomatidae, and Theridiidae). They include Argiope argen- tata (Fabricius), Micrathena schreibersi (Perty), Mangora pia (Chamberlin and Ivie), Eustala an- astera (Walckenaer), Eustala sp., and Neoscona domiciliorum (Hentz) (Araneidae), Nephila cla- vipes Linnaeus, Leucauge sp., Tetragnatha sp. (Tetragnathidae), Epilineules globosus (O. Pick- ard-Cambridge) (Theridiosomatidae), and Lat- rodectus mactans (Fabricius) (Theridiidae). Definition of Light Environments We classified the light environments in which the sampled spiders were foraging into three gen- eral groups. Diurnal, nonforest habitats were classified as "bright." Diurnal, forest habitats were classified as "dim," and nocturnal habitats were classified as "dark." Silk Measurement Techniques Most spiders spin silks that are unpigmented and most silks appear flat white, translucent, or iridescent. These differences in the reflectance properties of silks are apparent at casual obser- vation and suggest that silk proteins differ in their physical structure. To capture this range of vari- ation, we focused our analysis on quantifying silk spectral variations spun by an array of phylo- genetically primitive and derived spiders. Nev- ertheless, to make sure that any variation in spec- tral properties of silks spun by individuals of the same species was minor relative to variations in spectral properties of silks spun by different spe- cies, we collected and compared samples of sticky silks spun by five adult Argiope argentata (Ara- neidae). Silks were collected on notched cards (Craig 1987) from the catching surface of sheet webs, the catching threads of three-dimensional space webs, the sticky spiral thread of webs spun by aerial web weavers and from the retreat silks spun by nonweb-spinning, hunting spiders. No data are currently available to indicate the degree of homology among these silks. When silks are stressed their molecular con- figuration may change (Dong et al. 1991), and hence their spectral properties as well. To stan- dardize our data, we collected silk at the tensions at which the spiders laid them. In most cases, silks were collected and analyzed on the same day they were produced. Each sample was illuminated with a beam of monochromatic light and the light reflected back was measured in a UV-visible(350 nm-700 nm), incident-light microspectrophotometer. When the spectral properties of sticky silks spun were mea- sured, the light was directed at one viscid droplet (araneoid catching thread) or combed, silk puff (deinopoid catching thread) and the thread on which they adhered. The microspectrophotom- eter is based on a Leitz Ortholux microscope with MPV Pol-Opak illuminator, Nikon CF Fluor 20 x/0.75 or 40x/0.85 objectives, andZeiss UV- projecti ve coupled to a Hammamatsu R928 pho- tomultiplier (Bernard 1982). The field stop of the Epi-illuminator was reduced to illuminate a sin- gle strand. The aperture stop was reduced to make the illumination as unidirectional as possible, and the incident angle was made oblique to minimize instrumental stray light. Light reflected from the strand was collected over the full numerical ap- erture of the microscope objective. The instru- mental baseline was calibrated with a National Bureau of Standards SRM-20003D spectral re- flectance standard. We measured the spectral properties of silks over a range of 350 nm to 700 nm at 10-nm intervals. It is not possible to relate microspec- trophotometer absolute spectra to field absolute spectra. Therefore, the normalized spectral re- flectance functions were calculated by dividing the reflectance at each wavelength by the maxi- mum reflectance for that sample. This allows us to compare the shape of silk reflectance spectra and the variation in reflectance as wavelength changes. Phylogenetic Techniques To infer the ancestral character states for silk reflectance properties and foraging environ- SPECTRAL PROPERTIES OF SPIDER SILKS 289 CD CD A\ O -Q O CD CD -I CD CD T3 A? CD CD ?Q CD CD T3 ^Jg. ^J|iJ| CD CD A* ul CD CD luu I? ro CD CD "O JJ'8 o A a en ?a "CL o cz Im a CD ,T3 i^ ?" CD ? t LC It t Araneoidea Paleocribellatae Neocribellatae Mygalomorphae Araneomorphae L^tZ Light Environment ? Dark Dim C Bright Silk Relectance th High UV tZ Flat \L Low UV FIG. 1. Spectral properties of silks and spider foraging environments mapped onto a cladogram of the order Araneae redrawn from Coddington and Levi (1991). This figure illustrates the phylogenetic distribution of spiders that spin silks characterized by high reflectance in the ultraviolet, silks that are spectrally flat, and silks that are characterized by reduced reflectance in the UV. Light environments, in which the sampled spiders forage, also are illustrated. They include nocturnal, dim, and bright habitats. When the ancestral character state of silk reflectance or light environment is unambiguous, it is illustrated by a single icon at a node. Multiple icons at node illustrate features whose ancestral character state remains unresolved. ments, we plotted the data on the most recently proposed cladistic structure of the Araneae. The cladogram in figure 1 has been redrawn from Coddington and Levi (1991), and taxa for which we did not have data have been eliminated. Some of the taxa in figure 1 include spiders that spin silks characterized by more than one spectral property or spiders that forage in more than one light environment. Thus, in effect, they are "polymorphic" for these characters. For the data and terminal taxa examined here, poly- morphisms arise in the Uloboridae, Araneidae, and Tetragnathidae. Nixon and Davis (1991) suggest that all character states for polymorphic taxa should be encoded and included in analyses to infer correctly ancestral states. Therefore, all of the combinations of silk spectral patterns and spider foraging environments, illustrated on the cladogram at terminal taxa, were encoded. Given the character states illustrated at the tips of the branches, ancestral character states were inferred using the DOS EQUIS procedure with all char- acters treated nonadditively (Farris 1988). Whenever the results of the optimization tech- nique suggested more than one possible character state at an internal node in the cladogram, all possible results are illustrated. RESULTS General Patterns We found three different types of spectra, il- lustrated in figure 2, that characterize the silks spun by spiders: (1) retreat silks such as those spun by Latrodeclus mactans (Theridiidae) that arc spectrally flat and reflect between 80% to 290 CATHERINE L. CRAIG ET AL. TABLE 1. Silk spectral properties, light habitat, and foraging mode for a diverse array of primitive and derived spiders. Information on web type and light habitat are from Shear 1986 or observations made at silk collection. Phylogenetic group Family Genus/species Light habitat Spectra Web type Theraphosidae Dipluridae Mecicobothridae Hypochilidae Eresidae Lycosidae Diguetidae Filistatidae Salticidac Deinopidae Uloboridae Uloboridae Uloboridae Uloboridae Theridiidae Theridiosomatidae Tetragnathidae Tetragnathidae Tetragnathidae Araneidae Araneidae Araneidae Araneidae Araneidae Araneidae Aphonopelma sp. Nocturnal Ishnothele sp. Nocturnal Hexurella sp. Nocturnal Hypochilius sp. Dim Stegodyphus sp. Nocturnal Sossipus sp. Nocturnal Diguetia sp. Nocturnal Kukulkania Nocturnal Phiddipus otiosus Bright Deinopis Nocturnal Miagrammopes animotus Nocturnal Hyptiotes cavatus Dim Uloborus glomosus Dim Philoponella tingena Dim Latrodectus mactans Dim Epilneutes globosus Dim Leucauge sp. Bright, dim Tetragnatha sp. Bright, dim Nephila clavipes Bright, dim Eustala sp. Nocturnal Eustala anastera Nocturnal Neoscona domiciliorum Nocturnal Mangora pia Dim Argiope argentata Bright Micrathena schreibersi Bright UVpeak UVpeak UVpeak UV peak Cut on at 550 nm Flat Flat UV broad Cut on at 550 nm UV broad Peaks in UV, blue, green, orange Peaks in UV, violet, green, red Peaks in U V, blue green Peaks in UV, blue, orange Flat Cutoff at 390 nm Flat Flat Cutoff at 370 nm Flat Flat Flat Flat Cutoff at 350 nm Cutoff at 410 nm None None Funnel and sheet Vertical sheet Sheet Funnel and sheet Sheet Sheet None Modified orb Space Reduced orb Orb Orb Space Orb Orb Orb Orb Orb Orb Orb Orb Orb Orb m O z 1 1.0 0.8 UJ EC Q UJ N -J < s or. O Po??Cbc .BE Q JOQ -,o? 0.61- ? o 0.4 r^O 3*- ????? Y 02 - ? Hypochilus sp.Ci B Latrodectus mactans ? o Argiope argentata \L 0.0 350 400 450 500 550 600 650 700 WAVELENGTH (nm) FIG. 2. Types of reflectance spectra observed. Retreat silks spun by Latrodectus mactans (Theridiidae) are spectrally flat and reflect 80% to 100% light across all wavelengths; silks spun by Hypochilus sp. (Hypochil- idae) are characterized by a spectral peak where at least 30% more light is reflected across a region of about 30 nm than adjacent regions; and silks spun by Argiope argentata (Araneidae) are characterized by a spectral cutoff where reflectance shifts by more than 40% from low to high reflectance (other species may spin silks that are characterized by a spectral cuton where re- flectance shifts from high to low reflectance across an analogous region). 100% across all wavelengths, (2) catching silks such as those spun by Hypochilus sp. (Hypo- chilidae) that are characterized by spectral peak or reflect at least 30% more light across a region of about 30 nm than at adjacent regions of the spectrum, and (3) viscid silks such as those spun by Argiope argentata (Araneidae) characterized by a spectral cutoff of at least 40% and reflectance gap at adjacent wavelengths. All of the data are tabulated in table 1. A subset of these that illus- trate the reflectance patterns noted in table 1 are depicted graphically in figure 3A-F. Intraspecific Variation in the Spectral Properties of Silks Comparison of silks spun by five different A. argentata show they are characterized by almost identical spectra (fig. 3F) and indicate that the molecular composition of catching silks spun by spiders of the same species is similar. Therefore, differences in the reflectance patterns of silks spun by different spider species may reveal alternative structures for spider silk proteins. This study fo- cuses on outlining the spectral and hence poten- 1.0*3 0.8 0.6 0.4 0.2 0.0 ) ?oooo0, lc ?o^coo00oo-o - ? Aphonopelma sp. a Ischnothele sp. o Hexurella, sp. 350 400 450 500 550 600 650 700 0.8 HO ??? tr Q UJ0.4 " a 3 O O O Oo !!?. '? ?o B oo OO B o ? BH es o, o 1 0 2 - ? Deln?P's a Hyptlotes sp. o Uloborus sp. ?oOo?o o _l_ _) 0.0 350 400 450 500 550 600 650 700 1.0 r? ??"???? Q0 ?o o.8 0-^^u,;,.p^ - ^ oOn oo ? ^aasasSiaSiKis81 0.6 0.40- 0.2 0.0 - ? Tetragnatha sp. a Mangorapia o Eustala anastera 350 400 450 500 550 600 650 700 WAVELENGTH (nm) 1.0 0.8 a H ?0 0.6 0.4 0.2 0.0 SiiE3 ?BH ???l ? Hypochilus sp. ? Sfegodyphus sp. O Kulkulkanla sp. ooo>BBaBaraB3? B 350 400 450 500 550 600 650 700 0.4 0.2 0.0 ?OoOO a Diguetla sp. a Sosslpus sp. o Phiddipus sp. _i 350 400 450 500 550 600 650 700 1.0 __^w??a8g 0.8 0.6 * n 0.4? F O Oo 0.2 - a Nephlla clavlpes a Argiope argentata F 00 o Micrathena schrelbersl i i i i i 350 400 450 500 550 600 650 700 WAVELENGTH (nm) FIG. 3. Silk spectra spun by spiders in the Mygalomorphae (A) and Araneomorphae (B-F), with cribellum (B-C) and ecribellate (D-F) spiders, primitive aerial web weavers (C), and derived (E-F) aerial web weavers. A. Spectral properties of silks spun by the mygalomorph spiders Aphonopelma sp., Ischnothele sp., and Hexurella sp. B. Spectral properties of silks spun by the araneomorph spiders with cribellum Hypochilus, Kulkulkania, Stegodyphus. C. Spectral properties of silks spun by primitive aerial web weavers with a cribellum, Deinopis, Hyptiotes, Uloborus. D. Silks spun by ecribellate, nonweb-spinning spiders Sossipus, Phiddipus, Diguetia. E. Spectral properties of silks spun by ecribellate, derived aerial web weavers Mangora pia, Eustala anastera, Tetragnatha. F. Spectral properties of silks spun by ecribellate, derived aerial-web weavers, unmasked silks spun by Nephila clavipes, silks spun by Argiope argentata, and Micrathena schreibersi. Reflectance measures for silks spun by A. argentata indicate mean and standard deviation for five different individuals of approximately the same size. These data show that silks spun by mygalomorph spiders (A) and cribellate spiders are characterized by a spectral peak in the ultraviolet region (B-C, silks spun by Deinopis are characterized by a broader band reflectance that includes both the ultraviolet (UV) and blue). All of these spiders forage at night or in dim, diurnal sites. The ecribellate spiders including those that do not spin webs (D), as well as web spinners (E-F), produce silks that are spectrally flat or are characterized by reduced reflectance in the ultraviolet region. These spiders forage at night, as well as in a variety of diurnal light environments. 292 CATHERINE L. CRAIG ET AL. tial molecular variation among silks spun by dif- ferent species of phylogenetically primitive and derived orb spinners and compares these results to the spectral properties of silks collected from the sheet webs and retreats spun by nonweb- spinning spiders. Spectral Properties of Silks Spun by the Mygalomorphae The phylogenetically primitive spiders Apho- nopelma sp., Ischnothele sp., and Hexurella sp. (Infraorder Mygalomorphae; Platnick 1977) all produce silks that peak in reflectance at 350 nm (fig. 3 A) but that differ at other wavelengths. Silks spun by Ischnothele sp. range from 60% to 80% reflectance from 400 nm to 500 nm, whereas the silks spun by Aphonopelma sp. drop to only 40% reflectance at 400 nm but have a secondary peak at 500 nm. In contrast, the silks spun by Hex- urella sp. peak in reflectance at 370 nm and drop to about 45% reflectance at all other wavelengths between 500 nm and 700 nm. None of these spiders spin aerial webs to intercept prey, and all are most active at night (fig. 1; table 1). ARANEOMORPHAE Sheet- Web and Nonweb Spinning Spiders Comparison of the spectra in figure 3B-F shows that the reflectance properties of silks produced by the araneomorph spiders are diverse. Silks spun by Hypochilus (sheet web), the most prim- itive of the araneomorph spiders (Platnick 1977), are characterized by a spectral peak in the UV region at 380 nm that drops to 60%-40% reflec- tance at wavelengths between 400 nm to 700 nm (fig. 3B), similar to the spectra of silk spun by Hexurella. The spider Stegodyphus sp. (sheet web) produces silks characterized by broad-band re- flectance in the UV and blue with a cut-on point at about 550 nm (fig. 3B), whereas Kukulcania sp. (sheet web) produces silks with similar spec- tral properties to those spun by Hypochilus. The sheet-web weaving spiders are active primarily at night or in dim light. The spiders Diguetia (sheet web) and Sossipus (sheet web) produce silks that are spectrally flat (fig. 3D). Phiddipus, a nonweb-spinning spider, produces silks char- acterized by a cuton point at about 550 nm and is active during the day. Phylogenetically Primitive, Aerial, Web- Weaving Spiders (Deinopoidea) The property of ultraviolet reflection is marked among the catching silks spun by the primitive, aerial web-weaving spiders in the superfamily Deinopoidea. Deinopis sp. spin catching silks that broadly reflect wavelengths in the UV and blue regions (fig. 3C). The phylogenetically primitive, aerial web weavers Hyptiotes cavatus, Uloborus glomosus, Miagrammopes animotus, and Phil- oponella tingea (fig. 3C and/or table 1), spin silks characterized by multiple spectral peaks in the ultraviolet (350 nm^lOO nm), green (500 nm- 550 nm), and yellow (550 nm-600 nm) regions of the spectrum. Thus, although the spectral properties of silks spun by the deinopoids are diverse in the region of the spectrum above 400 nm, they all retain a reflectance peak in the ul- traviolet. All of the deinopoids forage at night or in dim light (table 1). Phylogenetically Derived, Aerial, Web- Weaving Spiders (Araneoidea) Unlike the phylogenetically, primitive, aerial web-weaving spiders, none of the derived spiders produce catching silks with a peak in the UV region of the spectrum. Many of the derived, web weavers, such as Eustala anastera, Eustala sp., Mangora pia, Leucauge sp., Tetragnatha, Neos- cona domiciliorum, and L. mactans, produce silks that are spectrally flat or white. All of the spiders sampled to date that spin white silks forage at night or in dim light (table 1 and/or fig. 3E). Alternatively, the diumal araneoids that spin silks characterized by reduced reflectance in the UV region forage in either bright or dim light. These include the silks spun by Nephila clavipes that forage in the forest understory, silks spun by A. argentata (table 1, fig. 3F), and silks spun by Epilineutes globosus (table 1). The catching silks spun by these spiders are characterized by a broad-band cut off at about 370 nm. None of the phylogenetically primitive, aerial web weavers, the Deinopoidea, spin silks characterized by re- duced reflectance in the UV region or are found foraging in the brightest light environments where araneoids are found. Phylogenetic Analysis of Silk Reflectance and Light Environments The phylogenetic analysis shows that the prim- itive character state for silk spectral properties is high UV reflectance, and the primitive foraging environment for spiders is nocturnal (fig. 1). Fur- thermore, spiders that do not spin webs or that produce sheet webs, as well as the phylogeneti- cally primitive, orb-spinning spiders, produce silks that are spectrally flat or that are characterized SPECTRAL PROPERTIES OF SPIDER SILKS 293 by a UV peak. Only spiders in the superfamily Araneoidea produce silks characterized by re- duced reflectance in the UV. Although these data are preliminary, our current findings show that the araneoid spiders are the only aerial web spin- ners that forage in bright environments. Summary In summary, spiders in the primitive and de- rived families of aerial web weavers spin at least three different types of silks that are used for prey capture alone: high-UV reflecting silks, low-UV reflecting silks, and spectrally flat silks. Although these data represent only a small subset of silks spun by spiders, the cladistic analysis suggests that the property of UV reflection is a primitive character state for silk proteins. In addition, all of the deinopoid spiders produce catching silks with a UV spectral peak but none of the derived araneoid spiders produce silks with a UV peak. The ability to produce silks that reflect white light is a derived feature of spiders that evolved fre- quently throughout the Araneae and among both web-spinning and nonweb-spinning species. Our current information, although limited, suggests that only the derived, web-spinning spiders pro- duce silks characterized by reduced UV reflec- tance. DISCUSSION Spider silks contain two kinds of protein sub- units: crystalline regions that give the silks strength and noncrystalline regions that give silks extensibility (Gosline et al. 1986). Although dif- ferences in the amino-acid sequences in the two regions could account for some of the silks' me- chanical properties, it is known that the molec- ular ordering of the crystalline regions occurs during the flow of liquid silk from the silk glands through the spinnerets (Denny 1980; Kerkam et al. 1991). Therefore, the variations in silk struc- tural properties that have been revealed in this study may have resulted from selection on the protein itself (the amino-acid sequences in the different protein subunits (i.e., Xu and Lewis 1990; Hinman and Lewis 1992) as well as the conditions under which the protein is produced (Kerkam et al. 1991). Rudall and Kenchington (1971) dismissed the importance of variations in silks produced by insects reasoning that because they were used outside the body, silks are not subject to as stringent selection pressures as are proteins used for intracellular purposes. In the case of spiders, we propose instead that the dif- ferences in silks result from selection for func- tional properties specific to the ecological pur- poses for which they are used. The simplest silk production system is found in spiders in the genus Antrodiaetus (Mygalo- morphae: Antrodiaetidae; Kovoor 1987). Antro- diaetus has four functional spinnerets and one kind of silk gland that produces the two different types of protein that comprise their silk threads (Palmer et al. 1982). Spiders, like Antrodiaetus that line subterranean burrows with silks (i.e., Coyle 1986) or spin silks to encase their eggs, need to produce only fibers that are strong or dense enough to protect the spiders from the walls of its burrow and to protect their egg cases from tearing, parasitism, and desiccation. The most complex silk producing systems are found among the araneoid spiders. The ara- neoids have six functional spinnerets and six dif- ferent kinds of silk glands each producing one to three types of protein (Kovoor 1987; Coddington 1989). Two of these glands, the flagelliform and aggregate glands, are unique to the araneoids and from these the viscid catching threads are drawn. The complete dependence of the araneoids on silks for prey capture and the functional spe- cialization of their silk-producing system sug- gests that selection pressures affecting the evo- lution of araneoid silks differ from those affecting the evolution of other silk proteins. Below we compare and discuss the importance of the me- chanical properties and spectral properties of prey-capture silks to the foraging ecology of dei- nopoid and araneoid web-spinning spiders. Comparison of the Mechanical Properties of Silks Spun by Primitive and Derived Web Spinners Spiders spin aerial webs to intercept flying in- sects by absorbing their kinetic energy. The en- ergy-absorbing properties of the silks are a func- tion of silk tensile strength, elasticity, and thread size (Denny 1976; Gosline et al. 1992; Craig 1987). Although there have been no systematic comparisons of the mechanical properties of dei- nopoid and araneoid silks, preliminary data for catching threads spun by Uloborus walckenaerius suggests that they are much less elastic, although equally strong, compared with the catching threads spun by Araneus diadematus (Vollrath pers. comm. 1993). Like the silks spun by Bom- byx mori (Gosline et al. 1992), the inelasticity of U. walckenaerius silks may result from a high proportion of crystalline subunits in the protein. 294 CATHERINE L. CRAIG ET AL. In general, the viscid catching silks spun by ara- neoid spiders exhibit extreme elasticity. For ex- ample, the spiral thread produced by Mangora pia extends to 6.2 times its resting length before breaking (Craig 1987). Through high tensile strength and greater elasticity the araneoids can capture larger and/or faster flying prey than the deinopoids. Spectral Properties of Web Silks and Their Importance to Prey Capture Web visibility is an important component of insect-spider interactions. Spiders forage in di- verse light environments and even at night in- sects can see webs and avoid them (Craig and Freeman 1990). The webs' catching silks spun by the deinopoids are characterized by reflec- tance peaks in the UV and blue regions of the spectrum and are drawn from the pseudoflagel- liform, paracribellar, and cribellar glands (Ko- voor 1987). The capture silks spun by the ara- neoids are derived from the flagelliform and aggregate glands and may be homologous in or- igin to deinopoid silks. None of the araneoids spin prey-capture silks with spectral peaks in the UV or blue regions of the spectrum, but some araneoids spin silks that are characterized by re- duced UV reflectance. Correlated with this difference in spectral properties between silks spun by araneoid and deinopoid spiders is a 37-fold increase in species number and an apparent expansion of the de- rived, aerial web-weaving spiders into new light environments. Whereas the deinopoids forage exclusively in shaded sites, such as the forest understory and at night, the araneoids forage in shaded sites, at night, and in open sites where the web is brightly illuminated throughout the spiders' foraging period. We propose that where increased levels of illumination enhance web vis- ibility, insect visual response to silk colors and brightness are an additional factor affecting the evolution of silk molecular organization. Craig and Bernard (1990) explored the behav- ior of Drosophila sp. to webs spun by Uloborus glomosus, a primitive, orb-weaving spider, that spins catching silks characterized by spectral peaks in both the UV and blue regions of the spectrum (fig. 3C). Drosophila were introduced into the enclosed baseleg of a Y-shaped choice number and UV-reflecting webs spun by U. glo- mosus were fixed to either chamber arm. One web was illuminated with white light containing a UV component and the second web was illu- minated with white light from which the UV component had been filtered. Drosophila were then released from the baseleg and flew to either chamber arm. In all of the trials, more flies flew to webs that reflected UV+ white light than webs illuminated with white light from which the UV light had been filtered. In a second study, Craig and Lesch (unpubl. data) further explored this result by videotaping Drosophila as they flew to- wards webs spun by U. glomosus. When the web was illuminated with white light from which the UV component had been filtered, the flies turned away from it before they were intercepted. When the same web was illuminated with white light containing a UV component, the flies were in- tercepted. Thus, even though Drosophila could see and avoid the web when it did not reflect UV light, they were unable to avoid interception when the web reflected UV. These studies suggest that UV-reflecting silks stimulate wavelength-depen- dent behavior or insect motor responses to par- ticular wavelengths that cannot be modified by learning. Therefore, if spiders evolved to spin silks that elicit wavelength-dependent behavior, insects could be drawn to webs that they other- wise would recognize and avoid if spun from different colors of silk. If the spectral properties of araneoid silks have been important to habitat expansion by web- spinning spiders, two alternative scenarios re- garding their ecological role could be envisioned: (1) silks may camouflage the web outline, (2) silks may attract prey to the web area. In the first case, spiders could have evolved to spin silks that are characterized by reduced brightness or spectral properties that match background vegetation. Both of these factors would make the web dif- ficult for insects to see. In the second case, spiders could have evolved to spin silks characterized by enhanced brightness or spectral properties that contrast with background vegetation. Thus, webs would be more visible to prey and hence easier for insects to see and avoid. However, because spiders spin webs from more than one type of silk protein, some silk components of the web may contrast with background vegetation and attract insects to them, whereas other compo- nents of webs may match background vegetation and camouflage the web outline. Argiope argentata, a derived, aerial web-spin- ning spider, seems to have adapted the later for- aging behavior. The web's spiral thread is char- acterized by reduced reflectance in the UV. It would appear white to approaching insects (Men- SPECTRAL PROPERTIES OF SPIDER SILKS 295 zel 1979; see fig. 3F), making webs difficult for insects to see. In contrast, the spider also pro- duces decorative silks characterized by enhanced reflectance in the UV. By manipulating the pres- ence and absence of web decorations in the field, Craig and Bernard (1990) found that more in- sects were intercepted at decorated webs than at webs from which the decorations had been re- moved. They proposed that UV-reflecting web decorations attract prey to the web area. This hypothesis was tested by comparing the rates that insects were intercepted at webs spun by spiders foraging in groups, some of whom decorated their webs and others that did not (Craig 1991). Spi- ders that decorated their webs attracted more prey than spiders that did not (Craig and Bernard 1990); spiders that foraged in groups of two or more (the spider's body also reflects UV light) captured more prey than solitary foragers. Group foraging spiders that spun web decorations drew prey to their webs and away from undecorated webs spun by neighboring individuals (Craig 1991). In conclusion, the data presented in this paper represent only a first step to quantifying the di- versity of spider silk proteins. Nevertheless, cla- distic analysis of our findings suggests that (1) the primitive foraging environment of spiders was nocturnal, (2) the primitive character state for silk spectral properties is high U V reflectance, and (3) the evolution of unique silk glands and proteins may have enabled the araneoids to use new habitats and led to a 37-fold increase in the number of aerial web-spinning species. ACKNOWLEDGMENTS We thank our colleagues, M. Stowe, G. Lowe, R. Jackson, P. Cushing, B. Opell, and J. Palmer, who sent us samples of silks and spiders, and R. Weber, D. Engleman, and C. Hayashi for com- ments on previous drafts of the manuscript. We thank the Instituto de Recoursos Naturales Re- novable for permission to export spiders to the United States and the Smithsonian Tropical Re- search Institute for use of their field facilities. This work was supported by grants from the Whitehall Foundation, the National Geographic Society, the National Science Foundation (BSR 8720093, BNS-9109468), and Yale University (support to C.L.C.), and by the United States Army Research Office, the Connecticut Lions Eye Research Foundation, and Research to Prevent Blindness (support to G.D.B.). LITERATURE CITED Beckwith, T. 1991. Amplification and analysis of spi- der silk genes using polymerase chain-reaction. American Zoologist 31:A30. Bernard, G. D. 1982. Non-invasive optical tech- niques for probing insect photoreceptors. Pp. 752- 759 in L. Parker, ed. Methods in enzymology. Ac- ademic Press, New York. Coddington, J. 1986. The monophyletic origin of the orb web. Pp. 319-363 in W. A. Shear, ed. Spiders: webs, behavior and evolution. Stanford University Press, Stanford, Calif. . 1989. Spinneret silk spigot morphology: ev- idence for the monophyly of orb-weaving spiders, Cyrtophorinae (Araneidae), and the group Theri- diidae plus Nesticidae. Journal of Arachnology 17: 71-95. Coddington, J. A., and H.W.Levi. 1991. Systematics and evolution of spiders (Araneae). Annual Review of Ecology and Systematics 22:565-592. Coyle, F. 1986. The role of silk in prey capture by nonaraneomorph spiders. Pp. 269-305 in W. A. Shear, ed. Spiders: webs, behavior and evolution. Stanford University Press, Stanford, Calif. Craig, C. L. 1987. The ecological and evolutionary interdependence between web architecture and web silks spun by orb-web-weaving spiders. Biological Journal of the Linnean Society 30:135-162. . 1991. Physical constraints on group foraging and social evolution: observations on web-spinning spiders. Functional Ecology 5:649-654. Craig, C. L., and G. D. Bernard. 1990. Insect attrac- tion to ultraviolet-reflecting spiders and web dec- orations. Ecology 71:616-623. Craig, C. L., and C. Freeman. 1991. Effects of pred- ator visibility on prey encounter: a case study on aerial web weaving spiders. Behavioral Ecology and Sociobiology 29:249-254. Denny, M. 1976. The physical properties of spider's silk and thejr role in the design of orb webs. Journal of Experimental Biology 65:483-506. . 1980. Silks?their properties and functions. Pp. 247-363 in J. F. V. Vincent and J. D. Currey, eds. The mechanical properties of biological ma- terials. Society for Experimental Biology Sympo- sium XXXIV. Cambridge University Press, Cam- bridge. Dong, Z., R. V. Lewis, and C. R. Middaugh. 1991. Molecular mechanism of spider silk elasticity. Ar- chives of Biochemistry and Biophysics 284:53?57. Farris, J. F. 1988. Hennig86, Version 1.5. [Micro- computer program available from author: 41 Ad- miral Street Port Jefferson Station, N. Y. 11776.] Gillespie, J. H. 1991. The causes of molecular evo- lution. Oxford University Press, New York. Gosline, J. M., M. Denny, and M. E. Demont. 1984. Spider silk as a rubber. Nature 309:551-552. Gosline, J. M., M. E. Demont, and M. Denny. 1992. The structure and properties of spider silks. En- deavour 10:37-43. Hinman, M. B., and R. V. Lewis. 1992. Isolation of a clone encoding a second dragline silk fibroin. Journal of Biological Chemistry 267:19320-19324. Kerkam, K., C. Viney, D. Kaplan, and S. Lombard). 1991. Liquid crystallinity of natural silk secretions. Nature 349:596-598. 296 CATHERINE L. CRAIG ET AL. Kovoor, J. 1987. Comparative structure and histo- chemistry of silk-producing organs in arachnids. Pp. 160-186 in W. Nentwig, ed. Ecophysiology of spiders. Springer, New York. Lucas, F., J. T. B. Shaw, and S. G. Smith. 1955. Chemical constitution of some silk fibroins and its bearing on their physical properties. Journal of the Textile Institute 46:440-452. Menzel, E. 1979. Spectral sensitivity and colour vi- sion in invertebrates. Pp. 504-580 in H. Autrum, ed. Comparative physiology and evolution of vi- sion in invertebrates. Invertebrate visual centers and behavior. Springer, New York. Nixon, K. C, and J. I. Davis. 1991. Polymorphic taxa, missing values, and cladistic analysis. Cladis- tics 7:233-241. Palmer, J. M., F. A. Coyle, and F. W. Harrison. 1982. Structure and cytochemistry of silk glands of the mygalomorph spider Antrodiaelus unicolor (Ara- neae, Antrodiaetidae). Journal of Morphology 174: 269-274. Platnick, N. I. 1977. The hypochiloid spiders: a cla- distic analysis, with notes on the Atypoidea (Arach- nida, Araneae). Novitates 2627:1-23. Rudall, K. M., and W. Kenchington. 1971. Arthro- pod silks: the problem of fibrous proteins in animal tissue. Annual Review of Entomology 16:73-96. Shear, W. A. 1986. Spiders: webs, behavior and evo- lution. Stanford University Press, Stanford, Calif. Vollrath, F., and D. T. Edmonds. 1989. Modulation of the mechanical properties of spider silk by coat- ing with water. Nature 340:305-307. Wainwright, S. A., W. D. Biggs, J. D. Currey, and J. M. Gosline. 1976. Mechanical design in organ- isms. Princeton University Press, Princeton, N.J. Work, R. W. 1977. Dimensions, birefringences, and force-elongation behavior of major and minor am- pullate silk fibres from orb web-spinning spiders? the effects of wetting on these properties. Textile Research Journal 47:650-662. . 1981. A comparative study of the supercon- traction of major ampul late silk fibers of orb-web- building spiders (Araneae). Journal of Arachnology 9:299-308. . 1985. Viscoelastic behaviour and web su- percontraction of major ampullate silk fibres of cer- tain orb-web-building spiders (Araneae). Journal of Experimental Biology 118:379-404. Wright, S. 1978. Variability within and among nat- ural populations. IV. University of Chicago Press, Chicago. Xu, M., and R. V. Lewis. 1990. Structure of a protein superfiber: spider dragline silk. Proceedings of the National Academy of Sciences, USA 87:7120-7124. Zemlin, J. C. 1968. A study of the mechanical be- havior of spider silks. Technical Report 69-29-CM, AD68433. United States Army Natick Laborato- ries, Natick, Mass. Corresponding Editor: G. Vermeij