at SciVerse ScienceDirect

Arthropod Structure & Development 42 (2013) 209e217

Contents lists available

Arthropod Structure & Development

journal homepage: www.elsevier .com/locate/asd

Alleged silk spigots on tarantula feet: Electron microscopy revealssensory innervation, no silk

Rainer Foelix a,*, Bruno Erb b, Bastian Rast c

aNeue Kantonsschule Aarau, Biology Department, Electron Microscopy Unit, Schanzmättelistr. 32, CH-5000 Aarau, SwitzerlandbKilbigstr. 18, CH-5018 Erlinsbach, SwitzerlandcGrütweg 8, CH-5522 Tägerig, Switzerland

a r t i c l e i n f o

Article history:Received 20 January 2013Accepted 17 February 2013

Keywords:SpidersTarantulasScopulaAdhesive hairsSilk spigotsChemoreceptors

* Corresponding author. Tel.: þ41 628245240.E-mail addresses: r.foelix@gmx.ch, rainer.foelix@a

1467-8039/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.asd.2013.02.005

a b s t r a c t

Several studies on tarantulas have claimed that their tarsi could secrete fine silk threads which wouldprovide additional safety lines for maintaining a secure foot-hold on smooth vertical surfaces. Thisinterpretation was seriously questioned by behavioral experiments, and more recently morphologicalevidence indicated that the alleged spigots (“ribbed hairs”) were not secretory but most likely sensoryhairs (chemoreceptors). However, since fine structural studies were lacking, the sensory nature was notproven convincingly. By using transmission electron microscopy we here present clear evidence thatthese “ribbed hairs” contain many dendrites inside the hair lumen e as is the case in the well-knowncontact chemoreceptors of spiders and insects. For comparison, we also studied the fine structure ofregular silk spigots on the spinnerets and found them distinctly different from sensory hairs. Finally,histological studies of a tarantula tarsus did not reveal any silk glands, which, by contrast, are easilyfound within the spinnerets. In conclusion, the alleged presence of silk spigots on tarantula feet isrefuted.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

It was rather surprising when some reports claimed that ta-rantula spiders (Theraphosidae) possess tarsal silk glands thatproduce fine threads which would aid in the adhesion to smoothvertical surfaces (Gorb et al., 2006; Rind et al., 2011). However,these findings were soon contradicted by behavioral experiments(Pérez-Miles et al., 2009) and more recently by morphological andfunctional arguments (Foelix et al., 2011, 2012a). Although therewas good morphological evidence why the alleged tarsal silkspigots (“ribbed hairs“) were chemoreceptors rather than silk-secreting spigots, one decisive point was lacking, namely theproof of sensory innervation of those ribbed hairs. Unfortunately,the sensory fibers (dendrites) of chemosensitive hairs in spiders arevery thin (0.1e0.2 mm in diameter) and cannot be detected in thelight microscope (LM). Therefore, we now used transmission elec-tronmicroscopy (TEM) in order to reveal those delicate dendrites inthin sections of those tarsal “ribbed hairs”. Another open questionwe addressed was, whether tarantulas have any silk glands insidetheir tarsi, comparable to the regular silk glands inside the

g.ch (R. Foelix).

All rights reserved.

abdominal spinnerets. No silk glands are known in the legs of anyspider, but there are some insects such as the webspinners(Embioptera) and the dance flies (Diptera, Empididae), which dohave silk glands in their basitarsi (Alberti and Storch, 1976;Nagashima et al., 1991; Young and Merritt, 2003). However, silkthreads from those glands are not used for adhesion but forbuilding delicate shelters or for wrapping captured prey.

2. Materials and methods

We used mostly molted skins (exuvia) but also some alcoholmaterial of the genus Brachypelma (B. albiceps, B. albopilosum,B. smithi). All our material came from spiders bred in the laboratoryby one of the authors (B. R.).

2.1. Light microscopy (LM)

Tarsi and spinnerets were first bleached in lactic acid orhydrogen peroxide for several hours to render them more trans-parent. They were then transferred to 70% ethanol, furtherdissected into small, flat pieces and inspected under a Leitz Diaplanmicroscope. Bright field, dark field, phase contrast and polarizationwere employed. For histological studies material embedded inEpon was sectioned with glass knives at 1e2 mm and stained with

R. Foelix et al. / Arthropod Structure & Development 42 (2013) 209e217210

toluidine blue/methylene blue. Digital photographs were takenwith a Nikon Coolpix or a Motic camera.

2.2. Scanning electron microscopy (SEM)

Selected pieces of tarsi and spinnerets were dehydrated in alcoholseries and acetone, then transferred to hexamethyldisilazane (HMDS)to minimize shrinkage, and after 15e30min air-dried on filter paper.Specimens were carefully oriented and attached to carbon-coatedaluminum stubs before sputtering them with gold. Thereafterthey were inspected in a Zeiss DSM-950 SEM; digital pictures weretakenwithaPentaxK20Dcameraatmagnifications from20to5000�.

2.3. Transmission electron microscopy (TEM)

A male Brachypelma albiceps was anesthetized with CO2 andafter 5e10 min a tarsus and a posterior spinneret were clipped off.They were immediately immersed in ice-cold, cacodylate-buffered5% glutaraldehyde; further small incisions weremadewith amicro-scalpel to ensure better penetration of the fixative. After 12 h, andafter several rinses in buffer (2 h), a post-fixation in 1% OsO4 fol-lowed (3 h). Dehydrationwas achieved in a graded series of ethanoland acetone, before an oriented flat embedding in Epon and a 2 daypolymerization at 65 �C. Thin sections were cut on a Reichert ul-tramicrotome using a diamond knife; silver and golden sectionswere picked up on Formvar-coated slot grids and staining was donewith uranyl acetate and lead citrate for 20 min each. A Zeiss EM-9Swas used for inspecting those sections and a Zeiss EM-10 for takingdigital pictures.

3. Results

Our first objective was to determine if there was sensoryinnervation of those alleged spigots (“ribbed hairs”) and for thispurpose transmission electron microscopy (TEM) seemed to be thebest choice. In order to optimize this search we made grazinghorizontal sections of the distal tarsal scopulae. This yielded cross-sections of hundreds of flat scopula hairs and very few “ribbedhairs”, but due to their circular cross-section they would stand out(Fig. 1a). Thus, in semi-thin sections those few “ribbed hairs” couldbe mapped under the light microscope and the Epon blocks couldafterwards be trimmed down to the desired area. Although this

Fig. 1. (a) Grazing horizontal section of the distal tarsal scopula in Brachypelma smithi. Aencountered. Note the typical double lumen of the hair shaft. (b) “Ribbed hair” from a transpNote that the cuticular tube ends above the hair base (arrowhead). (c) A broken shaft of a “ribNote the fine ribs on the surface of the hair wall, forming a chevron pattern. (d) Tip region ofJ. Kottsieper).

approach worked reasonably well, it still takes patience to relocatethe few cross-sectioned “ribbed hairs” in the TEM.

The second objective was to study the structure of the “real”spigots on the abdominal spinnerets, in order to see if any typicalmorphological characters of the spigots were also present in the“ribbed hairs”. Although spigots have been extensively pictured inthe spider literature using LM and SEM, there are hardly any finestructural studies that would reveal their internal structure.

3.1. Morphology of the “ribbed hairs”

The morphological characters of the “ribbed hairs” have alreadybeen described in detail in a previous publication (Foelix et al.,2012a). The main features are: (1) an S-shaped hair shaft arisingsteeply (70e80�) from a socket (Fig. 1b), (2) a relatively thin hairwall (1e2 mm) bearing fine cuticular ribs along the hair shaft(Figs. 1c and 2a), (3) a blunt tip with a small, subterminal poreopening (Fig.1d), and (4) a double lumen consisting of a hollow hairshaft with a smaller cuticular tube inside; this tube starts from thedistal pore and ends shortly before reaching the hair base (Fig. 1b).All these morphological features are typical for contact chemore-ceptors in spiders (Foelix, 1970, 1985; Foelix and Chu-Wang, 1973;Harris and Mill, 1973, 1977).

3.2. Innervation of the “ribbed hairs”

The decisive feature for chemoreception is the presence of manyfine dendrites inside that cuticular tube and their exposure to theenvironment at the terminal pore (Foelix and Chu-Wang, 1973;Foelix, 2011). So far this had not been demonstrated for the “ribbedhairs” in tarantula feet. Our recent micrographs of cross-sectioned“ribbed hairs” show that sensory fibers are clearly present. Theyalways occupy the lumen of the cuticular tube (Fig. 2b) and mea-sure only 0.1e0.2 mm in diameter. In cross-sections near the hair tip,the cuticular tube may appear empty, probably because the den-drites do not always reach all the way up to the terminal pore(Fig. 2a). Cross-sections of the dendrites showbetween one and fivemicrotubules, but no other cell organelles (Fig. 2c). The outer hairlumen surrounding the cuticular tube does not contain any cellularelements but is filled with some fluid that is commonly called“receptor lymph” (or “sensillum lymph”). This fluid is apparentlyalso bathing the dendrites inside the cuticular tube and may ooze

mong hundreds of flattened adhesive hairs only very few “ribbed hairs”(center) arearent wholemount showing the distinct socket and the double lumen of the hair shaft.bed hair” as seen in the SEM. A solid cuticular tube traverses the hair shaft eccentrically.a “ribbed hair” (Aphonopelma seemanni) showing a distinct subterminal pore (photo by

Fig. 2. Cross-sections of “ribbed hairs” as seen in the TEM. (a) Close to the tip the central canal (cc) appears empty; note the prominent ribs on the outer wall of the hair shaft. (b)More proximally, small nerve fibers appear inside the central canal and the outer wall becomes two-layered. (c) At high magnification, the central canal exhibits about a dozendendrites (d) in cross-section, each containing a few microtubules. The large structure in the center is probably an inflated dendrite (fixation artifact?).

R. Foelix et al. / Arthropod Structure & Development 42 (2013) 209e217 211

out through the pore at the hair tip. In SEM pictures the terminalpore is usually clogged by coagulated receptor lymph and appearsonly indistinctly. In rare cases, however, an unobstructed poreopening of 0.5e1 mm is clearly visible in these “ribbed hairs”(Fig. 1d; Rind et al., 2011; Kottsieper, 2012).

3.3. Morphology of abdominal silk spigots

If the “ribbed hairs”were actually secretory spigots, they shouldexhibit very similar features as the regular silk spigots located onthe abdominal spinnerets. We therefore had a closer look at thosespigots in order to establish the typical morphology of a silk-emitting spigot. We used whole-mounts of bleached spinneretsfor the inspection with the light microscope and dried exuvia forscanning electron microscopy.

In tarantulas, hundreds of spigots occupy the ventral side of thethree distal spinneret segments. This in contrast to araneomorphspiders, where the spigots are restricted to the distal end of the lastspinneret segment. Each spigot consists of two parts (Figs. 3 and 5):a large basal bulb and a long shaft with a terminal pore. Both partsare connected by a joint membrane, which means that the shaft ismovable with respect to the bulbous base. A second joint mem-brane connects the bulb to the surface of the spinneret. Interest-ingly, the bulb is not equipped with a deep socket as in sensoryhairs, but sits flush on the spinneret cuticle. We found two types ofspigots, some with relatively large bulbs (about 70 mm high and60 mmwide) and others with smaller bulbs (about 55� 40 mm); theoverall spigot length is about 350 mm in both types.

3.3.1. Spigot shaftThe shaft is about 300 mm long and 10e12 mm in diameter at its

base, tapering down to 2 mmnear the tip (Fig. 4a). On average spigotshafts measure 6e7 mm, which is only slightly thinner than themany hairs present on the spinnerets. The shaft is usually straight orslightly curved, the surface bears many small scales with longitu-dinal striations, and the terminal pore (1e2 mm in diameter) is al-ways located centrally. Cross-sections show a double-layeredcuticular wall (1e2 mm thick), often with small cavities forming aring in the middle of the wall (Fig. 4bed). The spigot lumen is tra-versed by a delicate silk canal (Fig. 4bee); its cuticular sheath is

electron-lucent andmaybe extensible.We conclude this fromcross-sections of the proximal shaft, where the sheath does not exhibit aregular circle but is irregular in outline (Fig. 4d). Often this silk canalis filled with a silk thread, which usually appears homogenous(Fig. 4c and e), but sometimes may have a dense outer layer. Theentire spigot shaft is acellular, i.e. there are no cells or cell processesinside; instead, the lumen seems to be filled with some fluid.

As can be seen in transparent whole-mounts, the shaft connectsvia a joint membrane to an elaborate socket of the distal bulb(Fig. 3a). The spigot shaft is thus similarly suspended as the hairshaft of (sensory or non-sensory) setae (Figs. 1b and 8).

3.3.2. BulbThe cuticular parts are best seen in whole-mounts of bleached

exuvia. The bulb has a massive cuticular ring at its base and anotherless developed ring at the distal end, where the shaft is articulated(Figs. 3 and 6). Both rings are connected by a rather thin cuticularmantle (1 mm), which is reinforced by several longitudinal ribs. Theentire cuticular framework is thus reminiscent of the wire cageholding the cork in a champagne bottle. In general, the bulb ap-pears round and smooth like a balloon, but sometimes it isshrunken, with the ribs protruding (Figs. 5b and 6a). We assumethat this is due to changes in the internal pressure, but this needsphysiological studies for confirmation. It is interesting that the silkcanal, after entering the upper part of the bulb region, widens into afunnel that consists of extracellular fibers (Figs. 4e and 6c), and isenclosed by thin epithelial cells. This unusual, highly localizedstructure is difficult to explain in physiological terms, but it may beinvolved in the regulation of thread thickness, since it looks similarto control valves found in spigots of araneomorph spiders (Wilson,1962a, 1962b). In the midregion of the bulb most of the lumen isfilled by cells (Fig. 5d) encircling the central silk canal. Inspectingexuvia from the inside, this canal is seen to exit the bulb base(Fig. 6b), but often it is torn and thus does not show its origin.

3.4. Histology of abdominal spinnerets and silk spigots

In cross-sectioned posterior spinnerets, we could see that thesilk canal exiting from the spigot base enters the lumen of thespinneret and eventually ends in a small silk gland. There, the

Fig. 3. Light microscopy of spigots. (a) Each spigot consists of a distal shaft (S) which is movably articulated (arrow) to a proximal bulb (B). The bulb arises from a cuticular ring (cr)at the base. (b) Cross-sections of spigot shafts at different levels. Note the silk canal (arrowhead) in the shaft lumen. The lowest picture is taken at the junction of the shaft and thebulb (see arrow in a); the central silk canal is surrounded by several cells.

R. Foelix et al. / Arthropod Structure & Development 42 (2013) 209e217212

cuticular lining of the silk canal flares into a funnel and apparentlycollects the liquid silk that is accumulating in the lumen of the silkgland (Fig. 7; Palmer et al., 1982; Palmer, 1985). Large, circularcross-sections within the gland lumen can easily be distinguishedfrom much smaller silk ducts, which contain intensely stained silk.Since we encountered many more of the smaller cross-sectionsthan of the large ones, we assume that the long silk-containingducts form several loops before entering the spigots.

3.5. Histology of the distal tarsus

We also did cross-sections of the distal tarsus and inspectedstained 1 mm sections for the presence of any possible silk glands. In

contrast to the spinnerets we found only sensory nerve fibers,tendons of the claw muscles, and many blood cells (hemocytes)within the tarsal lumen, but no silk glands. Tarantulas do have smalldermal glands (probably unicellular), but those do not extend fromthe hypodermis into the leg lumen and they terminate in tiny slits(about 5 mm long, 1 mm wide) on the leg surface (Foelix, 2011).

4. Discussion

When it was first reported that tarantulas could secrete “silk”from their tarsi (Gorb et al., 2006), the obvious question was:Where do those silk-like threads come from? A first morphologicalsurvey pointed to certain “ribbed hairs”, which were protruding

Fig. 4. Electron microscopy of the spigot shaft. (a) The spigot wall bears fine cuticular scales with longitudinal ridges. The silk canal opens through a central terminal pore to theoutside. (b) Spigot shaft sectioned close to the tip with the silk canal (sc) in the lumen. Scale bar 2 mm. (c) Silk canal with silk thread (S) inside. Note the double-layered shaft wall.Scale bar 1 mm. (d) Silk canal (sc) without a silk thread, but with an irregular outline. Scale bar 1 mm. (e) Silk canal with a silk thread (S) and surrounded by cuticular fibers (cf); cross-section of the upper bulb region. Scale bar 2 mm.

R. Foelix et al. / Arthropod Structure & Development 42 (2013) 209e217 213

from the many thousand adhesive hairs covering the ventraltarsus. Since these hairs also exhibited a small pore at the tip, theyseemed to be the right candidates to act as silk spigots. However,such a terminal pore is not necessarily secretory, but could as wellbe a sensory pore, e.g. from a contact chemoreceptor. In fact, thepublished SEM-pictures (Gorb et al., 2006; Rind et al., 2011)looked very much like the contact chemoreceptors known frommany spider species (Foelix, 1970; Foelix and Chu-Wang, 1973;Harris and Mill, 1973). Yet, the interpretation of those “ribbedhairs” being chemoreceptors was never considered. Our recentstudies (LM, SEM) provided strong evidence that the allegedspigots (“ribbed hairs”) were in fact contact chemoreceptors(Foelix et al., 2011, 2012a). However, one decisive argument wasstill lacking, namely that these “ribbed hairs” are innervated. Thereason why no investigator had examined the innervation before

is two-fold: firstly, TEM inherently means much more work thanLM or SEM, and secondly, finding the few “ribbed hairs” amongthe thousands of adhesive scopula hairs is like searching for theproverbial needle in the haystack. Using horizontal sections of thetarsal adhesive pads we could localize some of those “ribbedhairs” and we found many dendrites inside their central lumen e

just as is the case in known chemoreceptors of spiders and otherarachnids (Foelix, 1985; Foelix et al., 2010). Since all the cuticularcomponents (hair shaft, hair socket, inner cuticular tube) are alsothe same, our conclusion is that the “ribbed hairs” are actuallycontact chemoreceptors.

We also have good morphological evidence that the “ribbedhairs” are not silk spigots, as claimed by Gorb et al. (2006) and Rindet al. (2011). First of all, no silk thread was ever seen under themicroscope exiting through the terminal pore of a “ribbed hair”,

Fig. 5. Fine structure of spigot bulbs. (a) Lateral view of the junction (arrow) between spigot shaft (S) and the bulb (B). The basal part of the bulb is formed by a solid cuticular ring(cr), which is attached to the surface of the spinneret by a joint membrane (asterisk). (b) Dorsal view of a bulb with the shaft (S) broken off at its base. The silk canal in the center ismarked by an arrowhead. (c) Semi-thin section just below the shaft/bulb junction. The silk canal (arrowhead) is surrounded by two different layers of extracellular material (cf). (d)A more proximal cross-section of the bulb with the silk canal (arrowhead) in the center, enclosed by cells. Note the varying thickness of the bulb wall (B).

R. Foelix et al. / Arthropod Structure & Development 42 (2013) 209e217214

although this is commonly observed in abdominal spigots (Foelixet al., 2012a). Secondly, no silk glands or ducts were found histo-logically in the tarsi, quite in contrast to the spinnerets. And, sincetarantulas with sealed spinnerets did not leave any silk threadsbehindwhile climbing vertical glass plates, there is no physiologicalevidence for a tarsal silk production (Pérez-Miles et al., 2009;Pérez-Miles and Ortiz-Villatoro, 2012).

However, although there is no secretion of silk, some other kindof secretion seems to exude from the “ribbed hairs”, which formsdelicate short “trails” rather than threads (Peattie et al., 2011).Nothing is known about the chemical composition or the functionof this substance, but it has been suggested that it may be receptorlymph oozing out of the tip of a “ribbed hair” (Foelix et al., 2012a).This is still an open question and needs further study. Consideringthe small quantities of these trails and the fact that they snap whenthe tarsus is lifted off the substrate, it is highly unlikely that theyplay any role in adhesion (Peattie et al., 2011).

It has been suggested that spider spigots evolved from hairs(Bristowe, 1932), or more specifically, from chemosensory hairs

(Palmer, 1990; Bond, 1994). Similar claims exist for silk glands(including the spigots) in certain insects (Empididae), in which thebasitarsal silk glands are considered as “specialized secretoryspines” (Young and Merritt, 2003; Merritt, 2006). This idea ofcorrespondence is based on many structural and developmentalfeatures that are shared between sensory hairs and glands(Quennedey, 1998). Glands and sense organs are apparently ho-mologous, having been derived from a common epidermis-associated lineage. In some insects (dance flies) there is actuallyevidence that contact chemoreceptors have been transformed intosecretory units. Apparently the hair shaft was retained to allow silkapplication while the modified chemoreceptor lost its sensorycomponent (Young and Merritt, 2003).

Typical silk spigots on the abdominal spinnerets in tarantulasare quite similar to cuticular hairs with respect to the distal parts,i.e. the spigot shaft and the hair shaft. In both cases we have acuticular tube with a double lumen, and a movable articulation in asocket. However, the proximal parts differ in several aspects(Fig. 8): (1) there is a large and complex bulb region in spigots that

Fig. 6. (a) Side view of a spigot arising from the spinneret cuticle (Cu). The base forms a distinct bulb (B) that continues distally into a long and slender shaft (S). Both parts seem tohave pliable junctions (arrows). (b) Inside view of a spigot (exuvium) showing the silk canal (sc) exiting centrally from the bulb (B). Two hair sockets are seen in the lower part of thepicture. (c) Similar view as in (b) but with the silk canal (sc) having been broken off; however, its entrance into the funnel (F) of the bulb (B) is apparent. Inset: Lateral view of ableached spigot showing the silk canal (sc) traversing the center of the bulb and widening apically into a cuticular funnel (F). The arrow indicates the site where the silk canal hasbeen torn off in Fig. 6c.

Fig. 7. Cross-section of a posterior spinneret in Brachypelma. (a) Several silk glands can be identified within the hemolymph space (HL). Note the large circles (asterisks) whichrepresent cuticular funnels collecting the liquid silk inside the gland lumen. The smaller circles (arrowheads) correspond to the silk canals (ductules) which will lead into thespigots. Cu ¼ cuticle; hy ¼ hypodermis. Scale bar 15 mm. (b) Higher magnification of the large funnels (asterisks) in the gland lumen and the much smaller ductules (arrowheads)with silk threads (stained dark blue) inside. Scale bar 10 mm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of thisarticle.)

R. Foelix et al. / Arthropod Structure & Development 42 (2013) 209e217 215

Fig. 8. Comparison of a tarsal “ribbed hair”(a) and an abdominal silk spigot (b). (a) Left: Diagram of a longitudinal section of a “ribbed hair” (chemoreceptor). The subterminal poreis marked by an arrowhead. Right: Cross-section of the shaft of a “ribbed hair” showing the typical surface ripples of the hair wall and the eccentric central canal (cc) containingmany fine nerve fibers (d). (b) Left: Diagram of a longitudinal section of a spigot, showing the basal bulb (B) and the distal shaft (S). The distal silk pore is marked by an arrowhead.Right: Cross-section of a spigot shaft with a multi-layered, scaly wall and in the center a delicate silk canal (sc) with a silk thread.

R. Foelix et al. / Arthropod Structure & Development 42 (2013) 209e217216

is completely lacking in sensory hairs, (2) the delicate cuticularcanal of a spigot contains a silk thread, whereas the comparablecuticular tube of a chemosensory hair has many dendrites inside,and (3) a silk glandwith a long duct is only associated with a spigot;sensory hairs have several primary neurons instead.

Coming back to the alleged tarsal “silk spigots” (or “ribbedhairs”) in tarantulas: they have all the morphological features ofchemosensory hairs rather than those of the typical silk spigotsfound on the abdominal spinnerets. Furthermore, the distributionof the “ribbed hairs” also speaks for chemoreception and not for silksecretion: these hairs are not restricted to the ventral tarsi, butoccur on all extremities (including mouthparts and spinnerets) onthe tarsus, metatarsus, tibia and patella, on ventral and dorsal sides.Anatomically and physiologically, there is no evidence that any silkthreads originate from those “ribbed hairs” (rather, fine “threads” ofan unknown substance, perhaps receptor lymph). Behaviorally, ithas been shown repeatedly that tarantulas with sealed spinneretsdo not leave any silk threads behind after they had walked overclean glass plates (Pérez-Miles et al., 2009; Pérez-Miles and Ortiz-Villatoro, 2012). It thus seems inevitable to dismiss the idea of ta-rantulas “shooting silk from their feet”; and consequently it is veryunlikely that tarantulas have an adhesive system in addition totheir claw tufts that would “help these huge spiders to avoidcatastrophic falls” (Gorb et al., 2006). Indeed, we have recentlydetermined the number of adhesive hairs in the distal scopulae(claw tufts) of the tarantula Poecilotheria ornata and extrapolatedthat the claw tufts of a single tarsus can make 50 million contactpoints (Foelix et al., 2012b). Since all extremities are equipped withclaw tufts, a tarantula has certainly plenty of attachment forcesavailable when walking sure-footedly up a vertical glass plate.

Acknowledgments

We thank Karin Boucke for patiently taking digital pictures atthe Zeiss EM-10 electron microscope, Jan de Vries, Peter Wagner

and Dieter Ertinger for technical support with our electron micro-scopes, Johanna Kottsieper for providing an excellent micrograph(Fig. 1d), Leslie Brunetta and Rolf Thieleczek for help with literaturesearch, David Merritt for enlightening discussions, and JeromeRovner for critically reading our manuscript. We are also grateful tothe Neue Kantonsschule Aarau for the use of their EM-facilities.

References

Alberti, G., Storch, V., 1976. Transmissions- und rasterelektronenmikroskopischeUntersuchungen der Spinndrüsen von Embien (Embioptera, Insecta). Zoolog-ischer Anzeiger, Jena 197, 179e186.

Bond, J.E., 1994. Seta-spigot homology and silk production in first instar Antro-diaetus unicolor spiderlings (Araneae: Antrodiaetidae). The Journal of Arach-nology 22, 19e22.

Bristowe,W.S.,1932. The liphistiid spiders,with anappendixon their internal anatomyby J. Millot. Proceedings of the Zoological Society London 103, 1015e1057.

Foelix, R.F.,1970. Chemosensitive hairs in spiders. Journal ofMorphology 132, 313e334.Foelix, R.F., 1985. Mechano- and chemoreceptive sensilla. In: Barth, F.G. (Ed.),

Neurobiology of Arachnids. Springer Verlag, Berlin, pp. 118e137.Foelix, R.F., 2011. Biology of Spiders. Oxford University Press, New York.Foelix, R.F., Chu-Wang, I.-W., 1973. The morphology of spider sensilla. II. Chemo-

receptors. Tissue and Cell 5, 461e468.Foelix, R., Erb, B., Michalik, P., 2010. Scopulate hairs in Liphistius spiders: probable

contact chemoreceptors. Journal of Arachnology 38, 599e603.Foelix, R.F., Rast, B., Peattie, A.M., 2012a. Silk secretion from tarantula feet revisited:

alleged spigots are probably chemoreceptors. Journal of Experimental Biology215, 1084e1089.

Foelix, R., Rast, B., Erb, B., 2012b. Hafthaare bei Vogelspinnen:Vergleich einerbodenlebenden Brachypelmamit einer baumlebenden Poecilotheria. Arachne 17,16e23.

Foelix, R., Rast, B., Erb, B., Wullschleger, B., 2011. Spinnspulen auf den Tarsen vonVogelspinnen? Eine Gegendarstellung. Arachne 16 (4), 4e9.

Gorb, S.N., Niederegger, S., Hayashi, C.Y., Summers, A.P., Vötsch, W., Walther, P.,2006. Biomaterials: silk-like secretion from tarantula feet. Nature 443, 407.

Harris, D.J., Mill, P.J., 1973. The ultrastructure of chemoreceptor sensilla in Ciniflo(Araneida, Arachnida). Tissue and Cell 5, 679e689.

Harris, D.J., Mill, P.J., 1977. Observations on the leg receptors of Ciniflo (Araneida,Arachnida). II. Chemoreceptors. Journal of Comparative Physiology 119, 55e62.

Kottsieper, J., 2012. Functional Morphology and Biomechanic of the AttachmentSystem in Mygalomorph Spiders. Diploma thesis, University of Kiel, Germany.

Merritt, D.J., 2006. The organule concept of insect organs: sensory transduction andorganule evolution. Advances in Insect Physiology 33, 192e241.

R. Foelix et al. / Arthropod Structure & Development 42 (2013) 209e217 217

Nagashima, T., Niwa, N., Okajiwa, S., Nowaka, T., 1991. Ultrastructure of silk gland ofwebspinners, Oligotoma japanica (Insecta, Embioptera). Cytologia 56, 679e685.

Palmer, J.M., 1985. The silk and silk production system of the funnel-web myga-lomorph spider Euagrus (Araneae, Dipluridae). Journal of Morphology 186,195e207.

Palmer, J.,1990. ComparativeMorphology of the External Silk ProductionApparatus of“Primitive” Spiders. Ph. D. thesis, Harvard University, Cambridge, Massachusetts.

Palmer, J.M., Coyle, F.A., Harrison, F.W., 1982. Structure and cytochemistry of the silkglands of the mygalomorph spider Antrodiaetus unicolor (Araneae, Antrodiae-tidae). Journal of Morphology 174, 269e274.

Peattie, A.M., Dirks, J.H., Henriques, S., Federle, W., 2011. Arachnids secrete a fluidover their adhesive pads. PLoS ONE 6, e20485.

Pérez-Miles, F., Ortiz-Villatoro, D., 2012. Tarantulas do not shoot silk from their legs:evidence in four species of New World tarantulas. Journal of ExperimentalBiology 215, 1749e1752.

Pérez-Miles, F., Panzera, A., Ortiz-Villatoro, D., Perdomo, C., 2009. Silk productionfrom tarantula feet questioned. Nature 461, E9eE10.

Quennedey, A., 1998. Insect epidermal gland cells: ultrastructure and morphogen-esis. In: Harrison, F.W., Locke, M. (Eds.), 1998. Microscopic Anatomy of In-vertebrates, Insects, vol. 11A. Wiley-Liss, London, pp. 177e207.

Rind, F.C., Birkett, C.L., Duncan, B.J., Ranken, A.J., 2011. Tarantulas cling to smoothvertical surfaces by secreting silk from their feet. Journal of ExperimentalBiology 214, 1874e1879.

Wilson, R.S., 1962a. The structure of the dragline control valves in the garden spider.Quarterly Journal of Microscopical Science 103, 549e555.

Wilson, R.S., 1962b. The control of dragline spinning in the garden spider. QuarterlyJournal of Microscopical Science 103, 557e571.

Young, J.H., Merritt, D.J., 2003. The ultrastructure and function of the silk-producingbasitarsus in the Hilarini (Diptera: Empididae). Arthropod Structure & Devel-opment 32, 157e165.