338 Bull.Br.arachnol.Soc.(1982) 5 (8),338-343

Cribellum, and ventral comb swathsit as it is held and rotated by the secondand ontogeny in Hyptiotes covatus (Hentz) third legs(Eberhard, 1967; Marples, 1962). (Araneae:) Membersof the genusHyptiotes produceaunique vertical triangle-web, consideredby many arachnol- ogiststo representa reorientedsector of the family's Brent D. Opell common and primitive orb-web form (Opell, 1979). Departmentof Biology, The Hyptiotes web consists of only four radii, Virginia PolytechnicInstitute and State University, between which cribellar threads extend. When Hyp- Blacksburg,V A 24061,U.S.A. tiotes csvatus (Hentz) spiderlings emerge from the eegsacas second instars they lack a and calamistrum and produce only horizontal or diagonal Summary restinglines. Third instarspossess a cribellum, cala- Values and growth rates of fourth leg articles, mistrum and ventral comb and construct small calamistrum, ventral comb and cribellum were triangle-webs.Nearly all laboratory-rearedspecimens greater in female than in male Hyptiotes cavatus matured as sixth instars, at which time maleslost the (Hentz). plus growth Thesedifferences disparity in cribellum and ceased constructing capture webs. rates of the fourth leg articles and similarity in cribellum, calamistrum and ventral comb values Females continued to produce webs throughout and growth rates appear necessaryto allow the adulthood(Opell, in press). cribellum, calamistrum and perhaps the ventral The purpose of this study is to examinedevelop- comb to manipulate cribellar and wrapping silk. ment of the cribellum, calamistrumand ventral comb The two silk-handlingcombs must be able to pass to evaluatehow growth of these structuresand in proper orientation over the calamistrumand/or and spinneretsand must have lengths that will accom- the fourth leg articles responsiblefor their alignment modate silk strandsissuing from thesestructures. and use facilitates successfulweb constructionarrd Cribellum size and the spacingof setaein the cala- prey capture during each of the three male and four mistrum and ventral comb increasedduring devel- female instarsin which a capture web is produced. opment, but cribellar spinning spigot density remained constant. This resulted in adult females Answers to these questions will provide insight into having4.1 times more cribellum spigotsthan third developmental relationships of three functionally instars and producing webs with about 14 times associatedstructures responsiblefor prey capture. more cribellarsilk. Methods and Materials Spiderswere reared from eggsacscollected on 25 Introduction November1978 from the lower limbs of a singlehem- Like other cribellate (Berland, l9l3), lock tree, Tsuga canadensis(L.), near Newport, members of the largely orb-weaving family Ulobori- Giles County, Virginia.Eggsacs were refrigeratedand, dae use a calamistntm on the proximal, dorsal surface as specimenswere needed, incubated at 23-25"C. of each fourth metatarsus to comb silk from the Each was kept in a plastic box (measuring cribellum. In addition, members of this family and either30 x 18 x 8.5 cm or 34.5 x25.5 x 16.5cm) those of the Dinopidae have another comb (Opell, into which wooden rods were cementedas attach- 1979), composed of straight, stout setae, found on ment sites for webs. All containerswere kept in a the ventral surface of the fourth metatarsus and chamber where conditions of 23-25"C, S5-95% tarsus (Plate I b, c; Fig. l). In Dinopidae this ventral relativehumidity and a 10:14 hour light:dark cycle comb is poorly developed, but in Uloboridae it is were maintained.Laboratory studieswere conductec prominent and appears to be used to draw silk from from 14 January to I July 1979. Except for a few the cribellum and/or during prey capture. occasions,spiders were observeddaily and fed one When wrapping a struggling prey, the spider faces wild type Drosophila melanogoster for each web away from it and, in rapid succession,uses alternate produced. Developmentaltimes were similar to those fourth legs to throw silk onto it from a distance. observed in the field (Opell, in press). and After the prey is encumbered the spider thoroughly adults were placed into individually labelled alcohol B. D. Opell 339

vials for later study. Male third, fourth and fifth instar measurementsare each based on six to eight specimens;male sixth instar measurementson four or five specimens.Female third, fourth and fifth instar measurementsare eachbased on 1l to 14 specimens; female sixth instar measurementson eight or nine specimens. Carapacelength and width of each specimenwas s measuredwith a dissectingmicroscope equipped with an ocular micrometer. The cribellum and left fourth leg were removed,mounted on a microscopeslide in Hoyer's medium, and photographedwith a differen- tial interference contrast (Nomarski) compound microscope (Plate 1). A11counts and measurements were made from projected colour slides using cali brations obtained from a projected slide of an ocular micrometer. Spacing of setae on both the calamis- trum and ventral comb was determined by dividing the length of each structure by the number of its setae.Cribellum spigot spacingwas determinedby the number of spigots circumscribed by four trianpf,es placed at random on the projected cribellum. The slide of each cribellum was projected at l75O x, at Plate 1: A Cribellum of third instar female Hyptiotes covatus which mapgrificationeach triangle encompassedan showing spinning spigot bases. B. Prolateral surface areaequivalent to 100 pm2.Cribellum surfacearea of left fourth leg metatarsus and tarsus of a third instar female, showing calamistrum (upper comb) is reported as the mean value of areasdetermined by and ventral comb. C. Prolateral surface of left fourth formulae for a rectanpfe and for half an ellipse. leg metatarsus and tarsus of a sixth instar female, The total number of cribellum spinning spigotswas showing fully developed calamistrum and ventral determined by multiplying cribellum surface areaby comb. Scale lines = 50 ptm. spigot density.

Results Female leg article, calamistrum and cribellum lengthsand cribellum width exceededthose of males except during the third instar when values were similar (Iable 1). This is the result of greaterfemale growth ratesshown in Figure2.The rank order of the leg article growth rates was the same in males anc females.Within each sex both the values and growth rates of the cribellum. calamistrum and ventral comb were similar. Increases in the number of calamistrum and Fig. 1: Lateral view of the cribellar thread spinning posture ventral comb setae did not parallel the growth of of a Hyptiotes cavatus female, showing the right leg these structures (Table 1). This resulted in subse- just passed after its calamistrum has over the quent instars having more widely spaced setae, a cribellum. The cribellum is visible on the abdomen pronounced just anterior to the fourth leg, the ventral comb on trend that was more for the ventral comb the fourth leg's leading surface,and the calamistrum than for the calamistrum. In femalesthe increasein on its trailing surface. ventral comb setae was due largely to an increasein 340 Ontogenyin Hyptiotescavonts tarsal setae from a mean of 2.0 in third instars to a greater increasesin female femur and patella lengths mean of 5.6 in sixth instars (Plate I b, c). Both probably optimally facilitate greaterposterior exten- spinning spigot density and the cribellum width/ sion of the fourth leg and accommodationof a more length ratio remained fairly constant during develop- rotund abdomen. ment, the former showing an eiglrt percent increase A second case of apparent functional linking of only in sixth instar females and the latter a three growth rates is seenin parallel increasesin cribellum percent decreasein adult females. Except in sixth width and calamistrum length and perhaps also instar males,cribellum growth resulted in eachinstar ventral comb length. The association of cribellum havingabout 1.6 times more spigotsthan the previous and calamistrum is clear and their alignment is one. apparently facilitated by the tip of the leg whose calamistrum is being used resting on the metatarsus Discussion of the opposite fourth leg (Gertsch, 1979; Opell, Differencesin growth rates of male and femaleleg 1979; Fig. 1). This association is reflected by a articles appearrelated to abdominal growth. Basedon cribellum width/calamistrum length ratio for both measurementsof nine third instars and six male and males (through the fifth instar) and females which six female sixth instars, the slope of the ca:apace remainedbetween 0.90 and 0.95 throughout develop- length/pedicel to cribellum distance regressionline was 0.06 for malesand 0.12 for females.The slopeof lrd INSTAR 4th INSTAR 5th INSTAR 6th INSTAR the carapacelength/maximum abdominal width line X (SD) X (SD) X (SD) X (5D) (*) was 0.07 for malesand 0.15 for females.This means cARAPAct d tr, 678 ())) 810 (28) 928 (t0) LENGTH q 598 (122) 685 (5t) 867(89) io27 (66) that in order for the calamistrumand ventral comb to FEMUR d t68 (t1) 44) (12) ,46 (54) 660 (22) (27 (57 continue to pass across the cribellum (and/or LENGTH E t70 ) 464 ) 616 (81) 762 (135) spinnerets, Fig. l) the growth rates of female leg PATELLA T 208(17) 252(11 ) ?58(41) tt1 (54) (26) (37) (56) articles must exceed those of males. In addition to LENGTH 9 212 271 ttr6 4O9(6?) TIBIA d 2t6 (16) tot (15) t65 (to) 460 (8) abrupt changeswhich occur at moulting, there are (24) (12) (78) LENGTH E 2t3 t24 (41) 4'.t5 511 gradual volume that occur also increasesin abdomen T.fETATARSUS Lf 2r9 (2O) r08 (11) )58 ()e) 455(8) within each instar. The latter may not greatly influ- LTNGTH O 261(271 t42 (4J) 416 (58) 522 (81) ence the pedicel to cribellum distance, but it does TARSUS { 184(l) 210t8) 7J4 (17) ?67('t7) require the fourth femur of the leg whose calamis- LENGTH q 19t (e) 224(15) 251(21) ll0 (84) greaterdis- cALAMISTRUM N 184(11 ) 22?(12) Z7t (22) )20 (54) trum is being used to extend laterally a (15) (2e) (55) (91) LENGTH 9 181 ?4? 322 427 tance, thereby reducing its effective retraction. vENTRALcol4B ( 182(5) 211(10) 271(41) 3)t (50) The greatest disparity between growth rates of LENGTH O 1e4 (2?) 22t (1) 296 (66) 42t (141) male and female leg articles (Fig. 2) occurs between cRIBELLUM d 16t (t) ?06 (4) 251 (24) WIDTH (8) (27) (46) (50) femur and patella and the least between tibia and 9 156 226 292 352 metatarsus.The former two articleshave the greatest cALAMrsTRUr4 d 11(1) 14(1) 16(1) 15(3) SETAI NUMBER 11(1) 14(1) 17(?) 2t(5) flexibility, both in literal and functional senses.In 9 CALAMISTRUM ( 17(1) 16(1) 17(1) ?1(t) (2) (1) (2) (t) conjunction with coxa and trochanter articulations SETALSPACING I 15 17 18 1e the femur establishesboth the amount of lateral vENTRALcoMB ( 7(0) 7(0) 8(1) 8(1) movement and retraction of the fourth leg. Patellar SETAENUMETR I 7(0) 7(0) 8(1) 10(2) flexion further determinesthe effective length of the VENTRALCOMB N 26 (1) r0 (1) t5 (5) 42 (6) SETALSPACTNC 27 (2) t? (t) J8 (5) 4t (6) remaining leg articles, brings the metatarsusto the I sPrcoT DTNSITY d 12(0) 1r (0) 1r (0) abdomen's midline, ffid provides for tibial rotation Q/'roa ynz) q 12rc) 1l (0) 1r (0) 14 (0)

so that the metatarsus can be held parallel to and ToTALsPrcoTS ( o.1e(.12) 1.46(,17) 2."t0(.r4) (xtol) (,51) (.46) (.80) 4.50(.e7) against the abdomen. Flexion of the metatarsus ! 1.10 1.61 2.68 allows the calamistrum to become parallel to the Table 1: Developmental changesin the Hyptiotes cavatus probably very limited, cribellum. Tarsal flexion is carapace, leg IV articles, calamistrum, ventral but may be important for establishingcontact with comb and cribellum, All measurements are in the opposite fourth leg. Therefore, proportionately pm. B. D. Opell 341

ment. It is not known whether uloborids use cribellar If all or a constantproportion of cribellar spigots silk for wrapping prey, but the cribellum width/ spin silk, female fourth instars produce 1.5 times, ventral comb length was not constant, increasing female fifth instarc2.4 times, and femalesixth instars from 0.88 to 0.98 in malesand from 0.81 to l.11in 4.1 times as many cribellar strandsas do female third females. During prey wrapping the ventral comb is instars. These ratios may be even greaterif the para- not aligred in a manner similar to the calamistrum. cribellar spigotsdescribed by Peters& Kovoor (1980) lnstead, it appearsto move acrossthe cribellum urd increasein a similar manner. Meanlengths of cribellar spinnerets at an angle with both the sagittal and threads in field-produced I{ cavatus female third, frontal body planes. Either this apparent lack of fourth, fifth and sixth instar webs were 41, 53, 113 precision or the fact that cribellar silk is not used in and 145 cm respectively(Opell, in prep.).Therefore, prey wrapping may explain changesin the cribellum female fourth instars probably produce 1.9 times, width/ventral comb length ratio. fifth instars 6.6 times, and sixth instars 14.5 times More difficult to explain are changesin the spacing more cribellar silk per web than do third instars. of calamistrum and ventral comb setae(Table 1), par- Light and electron microscope photographs ticularly in view of the constant cribellar spigot (Comstock, l9l3; Friedrick & Langer, 1967; density. It may be that these changesare associated Kullmann, 1968; Opell, 1979) show that cribellar with the maintenanceof a surface area whose resis- strands are convoluted and drawn into tufts so that tance is just sufficient to allow silk to be drawn from individual cribellar strands in each cribellar thread the spigots, that they reduce tangflingtendencies in are at least 1.5 times the length of the thread. Using older individuals forced to handle more threads, or these data and the total number of cribellar spigots that they allow a certain amount of slippagewhich (Table 1), one can calculate that third instar webs may be necessaryto establish the kinky nature of each contains 0.68 km of cribellar strands, fourth cribellar strands. However, there are no studies of instar webs 1.28 km, fifth instar webs 4.54 km, and the finer details of cribellar and wrapping silk sixth instarwebs 9.79 km. handling upon which such conclusions might rest. The cribellate web thus seemsto be an extremelv

FEMUR .9O

FEMUR .8I

TtBtA.62 - METATARSUS .57 f- TtBtA .61 z METATARSUS .cJ z CALAN,4ISTRUM .55 I I PATELLA .45 F F tn VENI. COMA .52 z z CRIBELLUM .44 LU VENT,COMB LU J PATELLA ,32 J '38 LIJ uJ TARSUS .30 J t7J O t F cc cc

(n tiJ tu ) J

CA VC CR

57( 600 700 A CARAPACE LENGTH IN UM B CARAPACE LENGTH IN PM

Fig.2: Growth rates of fourth leg articles (solid lines) and cribellum, calamistrum and ventral comb (dotted lines) of Hyptiotes cavatus males (A) ana females (B). Stopes are given after the identification of each line. Both horizontal and vertical axesare logarithmic. Although malesdo not have a functional cribellum upon reachingmaturity (indicatedby //), this line hasbeen extended for comparativepurposes. 342 Ontogenyin Hyptiotesccvstus costly method of prey capture, both in terms of silk I-Iloboridaewhere cribellar strands are laid down in output and labour required to comb and deposit parallel rows and in spiralsrespectively. The rangein cribellar silk. This study does not allow determina- cribellum size and spigot number relative to the tion of the labour investment,but does permit esti- spider's size (Berland, 1913; Foelix & Jung, 1978; mation of material investment.Based on the previous Kullmann, 1968) raisesquestions about the relative information, coupled with Friedrick & Langer's efficiency and importance of the cribellum and its (1967) finding that Uoborus diverszsMarx cribellar silk among the cribellate groups. For example, did strandswere about 0.025 pm in diameter, a 10 trrm the cribellate orb-web evolve becauseuloborids had length of adult femalecribellar thread should contain a proportionately larger cribellum than other cribel- approximately 33.1 pm3 of cribellar silk. When late groups (e.g. filistatids) and were, therefore, similar estimatesof the adhesivematerial containedin subjected to more intensiveselection for economical the capture threadsof ecribellatespiders are obtainec use of cribellar silk, or do uloborids have a larger it will be possible to compare more precisely the cribellum becausetheir cribellar silk is more impor- materialcosts of the two web types. tant for prey retention than that of filistatids? Despite its production cost, cribellar silk must Most groups which have lost the cribellum and its originally have enhancedprey capture by allowing a associatedcalamistrum rely on hunting, densestreet web to retain prey rather than simply signaltng its webs, or webswith adhesivethreads for prey capture. presence or delaying its passage.This would have When cribellar and adhesivethreads of orb-websare been particularly important to spiders which compared,cribellar silk appearsto be superiorin two constructedaerial webs and capturedflying prey. The respects.Although less elastic than adhesivethreads similarity between cribellar silk production and prey (Lubin, in press), cribellar threads are able to give wrapping suggests that the cribellum may have when stressed.This is afforded by the sliding connec- evolved from the anterior median spinneretsas a tions of spiral and radial web elements(Eberhard, source of dense prey wrapping silk. This silk could 1976; Opell, 1979) and by the ability of the cribellar also have been placed on the web to retard prey mat to pull free from its supporting axial threads escape.The more effective that cribellar silk became (Eberhard, 1976). Additionally, cribellar silk appears for holding prey in the web the lessneed there would to retain its holding propertieslonger and to be less have been to use it for wrapping. The result would affected by drying than adhesivethread (Eberhard, then have been a spider which produced an aerial 1980). These properties suggest that there are web with cribellar capture elements,but no longer featuresof cribellar silk that may compensatefor the used wrapping silk. This prediction fits well with the cost of producing it and may help explain the persis- biology of Hypochilus (Shear, 1969) which Platnick tence of the cribellum and calamistrum. (1977) has shown to be the most primitive araneo- morph. Although it produces an extensive web, Acknowledgements Hypochilus gertschi Hoffman respondsonly to prey I thank Teresa L. Thompson for assistingwith that comesinto contact with the cribellar silk covered specimen photography and measurement, and lampshaderegion of the web and there subduesprey Herbert W. Lrvi, Cecile Villars, and William G. only by biting. This is also consistentwith Eberhard's Eberhard for providing useful comments and sugges- (1967) conclusion that primitive web spidersdid not tions on earlierdrafts of this paper. subduetheir prey by wrapping. After cribellar silk became a component of the References araneomorphweb, the cost of producingit may have BERLAND,J. 1913:Note pr6liminaire sur le cribellumetle intensifiedselection for its more efficient (restricted) calamistrumdes araigndes cribellates et surles moeurs placementor for its loss. Evidenceof the former is de cesaraign6es. Archs Zool.exp.gdn5l(Z)' 2341, seen in severalDtctyna species(Comstock, 1913; COMSTOCK,J. H. l9I3: TheSpider Book, 1st ed., L-72L. GardenCity, New York, Doubleday, Page. Wiehle, 1953) and Eresus (Wiehle, 1953) where EBERHARD,W. G. L967: Attack behaviorof diguetid cribellar silk is deposited only on peripheral threads spidersand the origin of prey wrappingin spiders. extending from the retreat, and in Dinopidae and Psyche, Camb. 7 4{2) : I 73-1g1. B. D. Opell 343

EBERHARD, W. G. 1976: Physicalproperties of sticky OPELL, B. D. 1979: Revisionof the generaand tropical spirals and their connections:sliding connectionsin American speciesof the spider family Uloboridae. orb webs.J.nat.Hist. l0: 481488, Bull.Mus.comp.Zool.Harv. 148(tO): 443-549. EBERHARD, W. G. 1980: Persistentstickiness of cribellum OPELL, B. D. (in press): Post-hatchingdevelopment and stlk. J.Aracnnol. 8Q\ : 283. web production of Hyptiotes cavatus (Hentz) FOELIX, R. F. & JUNG, H. 1978: Someanatomicai aspects (Araneae: Uloboridae). J.Arach n ol. of with special reference to the OPELL, B. D. (in preparation): Ontogeneticweb changesin calamistrum and cribelltm. Symp. z o ol. Soc. Lond. 42: the trianglespider Hyptiotes cavatus(Hentz). 4I7422. PETERS,H. M. & KOVOOR,J. 1980: Un compl6menti FRIEDRICK, V. L., Jr. & LANGER, R. M. 1969: Fine struc- I'appareils6ricigdne des llloboridae (Araneae):le para- ture of cribellate spider sr7k.Am.ZooI. 9: 9I-96. cribellum et sesglandes. Zoomorphologie 96:9I-t02. GERTSCH,W. J. 1979: American Spiders,Znd ed., L-274. PLATNICK, N. L 19?7: The hypochiloidspiders: a cladistic New York, Van Nostrand. analysis, with notes on the Atypoidea (Arachnida, KULLMANN, E. 1968: Das Cribellum zweier Stegodyphus- Araneae).Am.Mus. Novit. 2627: I-23. Arten im elektronenoptischen Bild (Arachnida: SHEAR,W. A. 1969: Observationson the predatorybehavior Araneae: Eresidae).Senckenberg.biol. 49: 451460. of the spider Hypochilus gertschi Hoffman (Hypo- LUBIN, Y. D. (in press): Web function and prey capture chilidae).Psyche, Camb. 76G) : 407-4I7 . behaviorin Uloboridae. WIEHLE, H. 1953: Spinnentiereoder Arachnoidea(Araneae) MARPLES, B. J. 1962: Notes on spiders of the family IX : Ortho$natha-Cribellatae-, Uloboridae.Ann.Zool., Agra 4(1): 1-11. (Pholcidae,Zodaridae, Oxyopidae,Mimetidae, Nesti- cidae).Tierwelt Dtl. 42:1-150.