Cribellate and the production of their capture threads

Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der RWTH Aachen University zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften genehmigte Dissertation

vorgelegt von

Anna-Christin Joel, M.Sc. aus Neuss in Nordrhein-Westfalen, Deutschland

Berichter: Universitätsprofessor Dr. tech. Werner Baumgartner Universitätsprofessor Dr. rer. nat. Peter Bräunig

Tag der mündlichen Prüfung: 15.07.2016

Diese Dissertation ist auf den Internetseiten der Universitätsbibliothek online verfügbar.

Abstract

Spider silk production has been studied intensively during the last years due to their mechanical properties. However, spiders do not only produce silk with outstanding mechanical properties: Some of them, the cribellate spiders, are specialized in producing nanofibres as their capture threads. In general, these threads are highly interesting, because their production involves not only a controlled arrangement of three types of silks with one being nanofibres (cribellate fibres), but also a special comb-like structure on the metatarsus of their fourth legs (). There are several hypotheses about how this thread assembly takes place, but none of them is able to explain all shapes in which capture threads of different occur. Hence, the cribellate thread production has to be examined more closely to establish a model describing the process in general. Evaluating the structure of the capture thread, I found the cribellate fibres organized as a mat, forming the typical puffy structure of cribellate threads. Due to this shape, they enclosed two larger parallel fibres (axial fibres). Mat and axial fibres were linked to each other between two puffs presumably by the action of the median . This linkage alone cannot lead to the typical puffy shape of a cribellate thread. After removing the calamistrum, a functional capture thread was still produced, but the puffs of the thread were not shaped anymore. Therefore, the calamistrum is not necessary for the extraction or combination of fibres, but for further processing of the cribellate fibres. Using data from plumipes I was able to develop a model of the cribellate thread production, connecting morphological data with the movement of the single limbs and determining the influence of this action on the structure of the thread. Although I was not able to determine, how cribellate fibres are processed by the calamistrum to form the puffy structure of the capture thread, I was able to refute some hypotheses, for example that Coulomb-forces are involved in keeping the nanofibres separated. The main features of the cribellate thread production were found conserved between species of very distant related families, suggesting the model to be generally valid. Differences in the morphology of the calamistrum were found to be an adaption to the web producing behaviour, not influencing the processing of the cribellate fibres. A teeth-like structure for example might help picking up fibres for transportation of threads after production in non-orb-weaving spiders. Only the asynchronous movement of the posterior spinnerets observed for Kukulcania hibernalis has to have an impact on the thread structuring according to the previously established model. Indeed their threads showed a pronounced looping structure, probably established by fewer linkages between axial fibres and the cribellate mat.

Zusammenfassung

Die bemerkenswerte mechanische Eigenschaft von Spinnenseide hat dazu geführt, dass die Produktion dieser Seide in den letzten Jahren intensiv erforscht wurde. Betrachtet man allerdings nicht nur die Eigenschaften der Spinnenseide, sondern auch deren Produktion, entdeckt man, dass einige Spinnen, nämlich die cribellaten Spinnen, sogar Nanofasern als Fangfäden herstellen. Für die Herstellung dieser speziellen Fäden werden die Nanofasern (cribellate Fasern) mit zwei weiteren Seidenarten verwoben. Weiterhin ist eine Art Kamm am Metatarsus des vierten Beines, das Calamistrum, an der Produktion beteiligt. Es gibt verschiedene Hypothesen, wie diese Fäden genau hergestellt werden und wie sich die besondere Struktur dieser Fäden ausprägt. Es gibt jedoch bisher keine These, welche alle Strukturunterschiede erklären könnte, die bei den Fäden verschiedener Spinnen vorkommen. Um diese Lücke zu schließen muss die cribellate Fangfadenproduktion näher untersucht und ein Modell erstellt werden, welches den Prozess im Allgemein beschreiben kann. Im Rahmen dieser Arbeit konnte gezeigt werden, dass die cribellaten Fasern sich als umhüllende Matte um zwei größere, parallel zueinander laufende Fäden, den Axialfasern, anordnen. Die Matte und die Axialfasern sind miteinander verbunden, vermutlich durch die Bewegung der medianen Spinnwarzen. Diese Verbindung alleine sorgt allerdings nicht für die typische puffige Struktur der cribellaten Fangfäden. Wenn man das Calamistrum entfernt, wird wider Erwarten ein funktionierender Fangfaden erzeugt, jedoch ohne die charakteristische Struktur. Das Calamistrum ist dementsprechend nicht notwendig, um die cribellaten Fasern zu extrahieren, sondern sorgt durch eine Kräuselung dieser für die puffige Struktur. Mit Hilfe der Spinne konnte ein Model des cribellaten Spinnprozesses erstellt werden, welches morphologische Daten und Bewegungsanalysen vereint und damit erklärt, wie die Struktur des Fadens erzeugt wird. Da die charakteristischen Merkmale des Spinnprozesses von U. plumipes auch in weit entfernt verwandten Spinnen entdeckt wurden, kann davon ausgegangen werden, dass das Model allgemeingültig ist. Einige Unterschiede, wie z.B. die zeitversetze Bewegung der posterioren Spinnwarzen bei Kukulcania hibernalis können mit dem Model zusammen die schleifenförmige Struktur von deren Fangfäden erklären, da hier weniger Verbindungen zwischen cribellater Matte und Axialfasern entstehen. Unterschiede in der Morphologie des Calamistrums hingegen scheinen keinen Einfluss auf die Struktur des Fadens zu haben. Eine Zahn-artige Struktur am Calamistrum zum Beispiel scheint erst nach Fertigstellung des Fadens zu helfen, den Faden zu packen und unabhängig von den Spinnwarzen zum gewünschten Zielort zu transportieren.

Table of contents

List of abbreviations and nomenclature ...... v

List of figures ...... vii

List of tables ...... ix

1. Introduction ...... 1

1.1. Spiders and their silk ...... 1

1.2. A ’s web ...... 2

1.3. Capture threads ...... 5

1.4. Cribellate capture threads ...... 5

1.5. Differences in the production of the cribellate thread ...... 9

1.6. Aim of this thesis ...... 11

2. Material and Methods ...... 13

2.1. Study ...... 13

2.1.1. Keeping and breeding spiders for in vivo experiments ...... 13

2.1.2. Preserved specimens ...... 14

2.1.3. Thread samples ...... 14

2.2. Behavioural experiments ...... 14

2.2.1. Observations in the natural habitat and the lab colony ...... 14

2.2.2. Standardized interactions ...... 15

2.2.3. Characterization of vibrational patterns ...... 17

2.2.4. Statistical analysis ...... 17

2.3. Recording cribellate spinning process ...... 18

2.3.1. housing for recordings ...... 18

2.3.2. Nocturnal observations ...... 18

2.3.3. Recordings with low speed camera ...... 19

2.3.4. Recordings with high speed camera ...... 20

2.3.5. Using freely available clips ...... 20

i

2.3.6. Data acquisition and statistical analysis ...... 21

2.4. Microscopic analysis ...... 23

2.4.1. Scanning electron microscopy (SEM) ...... 23

2.4.2. Focused ion beam treatment (FIB) and tomography ...... 24

2.4.3. Transmission electron microscopy (TEM) ...... 25

2.4.4. Light microscopy ...... 26

2.5. Manipulation of spiders ...... 27

2.5.1. Removing setae rows ...... 27

2.5.2. Conglutination of the calamistrum ...... 27

2.5.3. Shock freezing experiments ...... 28

2.6. Adhesion measurements ...... 28

2.6.1. Prey adhesion ...... 28

2.6.2. Standardized adhesion surface ...... 29

2.7. Reaction to high humidity ...... 30

2.8. Electrostatic charging ...... 30

2.8.1. Checking the thread’s charge ...... 30

2.8.2. Recharging ...... 31

3. Results ...... 33

3.1. Uloborus plumipes ...... 33

3.1.1. ...... 33

3.1.2. Intraspecific fighting ...... 35

3.2. Cribellate thread production ...... 43

3.2.1. Thread structure of U. plumipes ...... 43

3.2.2. Thread structure of other cribellate spiders ...... 47

3.2.3. and paracribellum ...... 48

3.2.4. Analysis of the limb movements of U. plumipes ...... 50

3.2.5. Limb movement during the cribellate spinning process of other spiders ...... 53

3.3. The function of the calamistrum ...... 58 ii

3.3.1. Morphology ...... 58

3.3.2. Contact between calamistrum and cribellate fibres ...... 61

3.3.3. Transfer of defined contact between calamistrum and thread to other spiders ...... 65

3.3.4. Removing the nearby second row of setae ...... 69

3.3.5. Removing the calamistrum ...... 70

3.3.6. Electrostatic charging ...... 75

4. Discussion ...... 77

4.1. Uloborus plumipes: from the natural habitat to a lab colony ...... 77

4.2. Capture thread structure ...... 80

4.3. The capture thread’s structure during the developmental process ...... 83

4.4. The calamistrum ...... 84

4.4.1. The extraction of the cribellate fibres...... 84

4.4.2. The function of the calamistrum ...... 85

4.4.3. Influence of thread structure on the function of the capture thread ...... 85

4.4.4. Puff formation ...... 86

4.4.5. Contact between calamistrum and cribellate fibres ...... 90

4.4.6. Morphological differences of the calamistra ...... 92

4.5. Model of the cribellate thread production ...... 93

4.5.1. Establishing a model of the capture thread production for U. plumipes ...... 93

4.5.2. Validation of the model ...... 95

4.6. Summary ...... 98

4.7. Outlook ...... 99

Literature ...... 101

Acknowledgements ...... 111

Curriculum vitae ...... 113

Supplement ...... 115

iii

iv

List of abbreviations and nomenclature

α angle between metatarsus and cribellum parallel to the ventral side β angle between and produced capture thread CI confidence interval FIB focused ion beam combined with a scanning electron microscope fps frames per second LED light emitting diode n number of specimen / individual trials n1 number of data points of one specimen / one trial min minute SD standard deviation SEM scanning electron microscope TEM transmission electron microscope

v

vi

List of figures

Fig. 1.1: Overview over a spider and its silk production...... 1 Fig. 1.2: Organization of an orb web...... 3 Fig. 1.3: Different shapes of a spider’s web...... 4 Fig. 1.4: Cribellate capture thread and the according spinnerets of ...... 6 Fig. 1.5: The calamistrum...... 7 Fig. 1.6: Different thread structuring depending on the species...... 8 Fig. 1.7: Different combing postures...... 9 Fig. 1.8: Degree of relation between the cribellate spiders used in this study...... 12

Fig. 2.1: Arena for standardized interaction...... 16 Fig. 2.2: Setup for low-speed recording...... 19 Fig. 2.3: Setup for modification of U. plumipes...... 27 Fig. 2.4: Setup for adhesion force measurement...... 29

Fig. 3.1: Uloborus plumipes...... 33 Fig. 3.2: Thread pulling and abdominal trembling...... 37 Fig. 3.3: Web shaking...... 38 Fig. 3.4: Behavioural and morphological parameters associated with winning the contest. ... 39 Fig. 3.5: No correlation between web shaking parameters and the size of the spider...... 42 Fig. 3.6: The alignment of axial fibres in a capture thread of U. plumipes...... 43 Fig. 3.7: Cribellate capture thread of U. plumipes...... 44 Fig. 3.8: Linkages between cribellate fibres and axial fibres in the intermediate zones...... 45 Fig. 3.9: Capture thread during the development of U. plumipes...... 45 Fig. 3.10: Structure conserved in the capture threads of Z. geniculata...... 47 Fig. 3.11: Collapsing threads after coating with gold...... 48 Fig. 3.12: Spinnerets during capture thread production...... 49 Fig. 3.13: No cribellum or paracribellum developed in 2nd instar of U. plumipes...... 49 Fig. 3.14: Combing position of U. plumipes...... 50 Fig. 3.15: Sequence of limb movement during capture thread production...... 52 Fig. 3.16: geniculata...... 54 Fig. 3.17: Deinopis sp...... 54 Fig. 3.19: Cribellate spiders during capture thread production...... 55 Fig. 3.18: Kukulcania hibernalis...... 56 vii

Fig. 3.20: Morphology of the calamistra of Uloboridae...... 58 Fig. 3.21: Development of the calamistrum...... 59 Fig. 3.22: Structures on the calamistrum of U. plumipes...... 60 Fig. 3.23: The socket...... 60 Fig. 3.24: Evaluating the jamming-hypothesis...... 61 Fig. 3.25: Cribellate fibres do not pass between the setae of the calamistrum...... 62 Fig. 3.26: Reconstructing the contact between calamistrum and cribellate fibres...... 64 Fig. 3.27: Behavioural adaption enable correct contact between calamistrum and thread...... 65 Fig. 3.28: The calamistra of ...... 66 Fig. 3.29: The calamistra of more distant related cribellate spiders...... 67 Fig. 3.30: Displaced interaction between calamistrum and cribellate fibres in Deinopidae. ... 68 Fig. 3.31: Changed structure of cribellate thread after removing both calamistra...... 71 Fig. 3.32: No changed thread stability...... 72 Fig. 3.33: Clotting of metatarsus after removal of the calamistrum...... 73 Fig. 3.34: Adhesion of cribellate thread on prey...... 74 Fig. 3.36: Adhesion properties of processed and non-processed threads...... 75 Fig. 3.35: Removing electrostatic charge from the capture thread...... 76

Fig. 4.1: Spigot placement and thread formation...... 81 Fig. 4.2: Nubbly structure of cribellate fibres...... 89 Fig. 4.3: Model of the cribellate thread production shown at three different points in time.... 94 Fig. 4.4: Lateral overview of the model of the cribellate spinning process...... 95 Fig. 4.5: The influence on and of the combing frequency...... 97

viii

List of tables

Tab. 2.1: Links to recordings of cribellate spiders...... 20

Tab. 3.1: Behavioural patterns during intraspecific competitions...... 35 Tab. 3.2: Independent behavioural patterns from morphological traits...... 40 Tab. 3.3: Thread development of U. plumipes...... 46 Tab. 3.4: Differences in thread structure of two Uloboridae...... 47 Tab. 3.5: Development of the cribellum of U. plumipes...... 50 Tab. 3.6: Characterizing the cribellate capture thread production of U. plumipes...... 53 Tab. 3.7: Characterization of the cribellate capture thread production in different species..... 57 Tab. 3.8: No difference in thread structure due to conglutination of the calamistrum...... 63 Tab. 3.9: Contact area on the calamistrum in different species...... 69 Tab. 3.10: Removing the nearby second row of setae has no influence on thread structure. .. 70 Tab. 3.11: Structural change of the cribellate thread after removal of the calamistrum...... 71

ix x

1. Introduction

1.1. Spiders and their silk Most spiders lead a solitary existence and react strongly to invaders of any kind within their territory. Such behaviour leads to a reputation of being an aggressive species with cannibalistic tendencies, making them very unpopular in the general public. However, spiders have quite fascinating qualities. Especially their highly adaptive silk with one of the toughest fibres not only found in nature but also in technical applications has fascinated humans for a long time now (Bourzac 2015; Eisoldt et al. 2011; Heim et al. 2009; Mortimer et al. 2015; Rising et al. 2015; Romer et al. 2008; Silva et al. 2013; Spiess et al. 2010; Vollrath 2000).

Fig. 1.1: Overview over a spider and its silk production. A spider’s body is divided in two halves: prosoma and opisthosoma. The silk producing system is located in the opisthosoma. It consists of a gland with its tail for protein production and storage, a duct for protein assembly and a spigot, placed on the spinnerets, from which the silk is extracted. The colour code within the silk production system indicates an increasing acidification, needed for correct protein align ment. Modified figure after Foelix (2011), Rising et al. (2015) and Heim et al. (2009).

The silk is produced in the tail of glands located in the opisthosoma of a spider (Fig. 1.1). It consists of filamentous proteins (called spidroins) as a core and a skin of lipids and glycoproteins (Blackledge 2012). The spidroins consist mainly of two amino acids, alanine and glycine, arranged in motives forming the secondary structure of the protein like crystalline β- sheets or amorphous β-turns (Eisoldt et al. 2011; Hinman et al. 2000). Those motives are again replicated in repetitive sequences, which finally determine the properties of the finished silk. Within the glands the spidroins are stored until use. To extract finally the silk from the spigots located at the spinnerets at the posterior end of the opisthosoma, the silk has to pass a duct connecting the gland and the spigot. There the final spidroin conformation is established by ion 1 exchange, acidification, shear forces and dehydration (Fig. 1.1) (Askarieh et al. 2010; Eisoldt et al. 2011). Although there are muscles within the opisthosoma and a muscular valve at the spigot, fibres have to be extracted mechanically rather than being pressed out. Spiders can use different strategies to extract silk: The spinnerets can e.g. be pushed towards a surface and the silk is attached to it. As the spider moves away from the surface, the silk is extracted. Other spiders use their legs to extract silk e.g. by gripping a thread with the claws of their tarsus. Some spiders, like and Uloboridae have specialized combs on their most posterior (fourth) leg, which is used to extract silk to wrap prey (Foelix et al. 2015b; Huber et al. 2008). The extraction speed has an impact on the spidroin alignment and hence on the fibre’s properties as well. Many experimental approaches aim to technically replicate the spider’s silk, because direct silk harvesting from the spider has its limitations. Scientist are already able to produce small amounts of spidroins and crystalize them in films or fibres, but those materials do not reach the same mechanical properties as the ones found in nature (Rising et al. 2015; Romer et al. 2008).

1.2. A spider’s web Also spiders know the value of silk. Most spiders only repair damaged webs or eat their old ones as a source of nutrition, previous to building a new one. Other spiders, like lanyuensis, are kleptoparasites and steal silk from their hosts’ webs to consume it (Tso et al. 1998). Not only single silk strands can be taken from another spider: Whole webs can be conquered by conspecifics (Buskirk 1975; Ferretti et al. 2011; Hodge 1987; Hodge et al. 1995; Hoffmaster 1986; Riechert 1978; Wise 1983). As illustrated by the well-known phenomenon of post-copulatory cannibalism, spiders tread a thin line between cooperation and aggression (Andrade et al. 2005; Polis 1981; Schneider et al. 2001; Schüch et al. 1990; Schwartz et al. 2014; Wise 2006). A different kind of intraspecific interaction arises from conflicts of interest, when e.g. food resources or mating partners are a limiting factor (Elias et al. 2008; Miyashita 1992, 2001; Taylor et al. 2001; Wells 1988; Wise 1975, 2006; Wise et al. 1992). The web conquer attempts of female spiders also belong to such competitions. These aggressive contests were attributed to the competition for territory, for the orb webs themselves or for prey, rather than cannibalism (Wise 1983, 2006). Not only aggressive interaction between female spiders occurs: E.g. within the Uloboridae different kinds of socialization have developed. Some spiders establish colonies, others are at least very

2 tolerant to neighbouring spiders (Bradoo 1986; Eberhard 1971; Lahmann et al. 1979; Masumoto 1998; Opell 1979; Peaslee et al. 1983; Spiller 1992). The outcome of an aggressive competition is determined either by the social position of being resident or intruder with respect to the contested territory (Hodge 1987; Hoffmaster 1986; Kasumovic et al. 2011; Rovner 1968) or the relative size of the contestants (Horton et al. 1983; Riechert 1978, 1981). Although there are notable exceptions, the behavioural patterns that characterise aggressive female-female interactions have received little attention (Buskirk 1975; Ferretti et al. 2011; Hodge 1987; Hodge et al. 1995; Hoffmaster 1986; Riechert 1978; Wise 1983). A comparison between the behavioural programs of different sexes of jumping spiders (Phidippus clarus) revealed that ritualized displays are rarer between females (Elias et al. 2010). For male-male contests it has already been shown that such displays and even the outcome of the last contest can influence the outcome of current contests (Elias et al. 2008; Fowler-Finn et al. 2006). Comparable studies for female-female interactions are missing.

Fig. 1.2: Organization of an orb web. Left half of a typical orb web including the terms of the structures found. From Foelix (2011).

Although web accessions can be used as a strategy to save time and energy, spiders build usually a web of their own. Uloboridae, living in a conventional radial orb web, start building a web by producing the frame and the radius (Fig. 1.2). During web construction the radial threads are combined in the centre to form the hub. This hub will be the resting place of the spider after

3 finishing the web. Between the radial threads, first an auxiliary spiral is spun. This spiral is replaced later by sticky threads, catching the prey. The most familiar web type is the radial orb web, but there are several other shapes of webs, like funnel or sheet webs (Fig. 1.3). They all have in common that they substitute the environment for the inhabitant (Krafft et al. 2012; Nentwig et al. 1987). Vibratory signals transmitted via the orb web have consistently been found to be important means of communication between spiders and their environments (Baurecht et al. 1993; Elias et al. 2008; Elias et al. 2003; Girard et al. 2011; Klarner et al. 1982; Maklakov et al. 2003; Sivalinghem et al. 2010; Suter 1978; Vibert et al. 2014; Wignall et al. 2013a, b).

Fig. 1.3: Different shapes of a spider’s web. Adapted to their environment, spiders build different levels of sophisticated webs. Starting at the roots of the tree in this drawing, web architecture gets more complex with each level. In the crown finally, reduced or modified orb webs are d epicted. Note that distant ancestors like mygalomorphs and scorpions only use silk to line burrows or construct trip lines rather than a web. From Vollrath et al. (2007).

4

1.3. Capture threads There are two types of capture threads used as a sticky spiral, both employing a different mechanism to capture prey. Ecribellate spiders have capture threads consisting of two core fibres out of flagelliform silk and additional glue droplets to capture prey (Peters 1995b; Vollrath et al. 1990). Cribellate spiders use a capture wool instead of glue droplets (Hawthorn et al. 2003; Opell 1994a). This wool consists of nanofibres, called cribellate fibres, with a diameter of 10 to 30 nm (Friedrich et al. 1969; Lehmensick et al. 1957; Opell 1995). Some species’ fibres bear knots, but these knots have no function in restraining prey (Lehmensick et al. 1957; Opell 1994b, 1995). Prey is restrained by entangling their surface irregularities (like setae) in a Velcro-like mechanism in the cribellate fibres (Opell 1994a). Furthermore, the adhesion force is enlarged by van-der-Waals forces and in some species also by capillary action (Hawthorn et al. 2002, 2003; Opell 1994a). Of about 40,000 described spider species only roughly 9% are cribellate spiders (Foelix 2011; Sahni et al. 2011). This discrepancy leads to the question, whether the ecribellate system might be superior in the evolutional context. It is assumed that the cribellate threads are the primordial type of capture threads from which the ecribellate threads once evolved (Bond et al. 2014; Fernandez et al. 2014). Cribellate threads show a higher material efficiency, measured by evaluating the stickiness per material used, but their production is much slower than that for gluey capture threads (Opell et al. 2009; Zschokke et al. 1995). Additionally, gluey capture threads have less UV-light reflection and therefore should be less visible to prey (Craig et al. 1990; Zschokke 2002). However, the orientation of the web in space has more influence on the capturing of prey, than the type of capture thread (Opell et al. 2006). The main advantage of cribellate capture threads is hypothesized to be a dehydration resistance, because gluey droplets depend on fluids (Peters 1987). Indeed the adhesion force of cribellate threads is maintained over longer periods of time, but this factor has never been proven to influence the biological competition, and the occurrence of cribellate species is not limited to dry areas (Eberhard 1980; Foelix et al. 1978; Opell 1993; Opell et al. 2000; Opell et al. 2008; Reukauf 1930).

1.4. Cribellate capture threads Even though and its mechanical and adhesive properties have been studied intensively, the process of thread production itself has not been described sufficiently to rebuild fibres with the same properties. This is especially true for the cribellate capture thread, although these threads are highly interesting from the biomimetic point of view (Bai et al. 2012; Gong et

5 al. 2015; Hou et al. 2012; Kronenberger et al. 2015; Vollrath 2006; Zheng et al. 2010). Cribellate spiders have evolved a way to produce, process and handle nanofibres to form complex capture threads, which are a composite consisting of up to three different silk types. Typical threads of spiders consist of only two fibres, due to the bilateral symmetry of their spinnerets.

Fig. 1.4: Cribellate capture thread and the according spinnerets of Uloboridae. A) A cribellate thread of Uloborus plumipes in polarized light, showing both axial fibres (ax) consisting of pseudoflagelliform silk, the paracribellate fibres forming a substructure and the cribellate fibres (cf) forming puffs and intermediate zones (iz). B) Overview over the spinnerets of , showing the typical pairwise arrangement of the spinnerets. Only the cribellum (cr), where the cribellate fibres emerge, is a single spinning plate in this species. By removing one anterior (as), the median spinnerets (ms) bearing the spigots of the paracribellum (spc) become visible. They produce the paracribellate substructure. The axial fibres emerge from the spigots of the glandulae pseudoflagelliformes on the posterior spinnerets (ps). C) Close up of the cribellum of U. plumipes. D) The elongated spigots of the paracribellum (spc) on the median spinnerets of sp.. B to D) SEM images, samples coated with gold.

Basic cribellate capture threads of e.g. Uloboridae have two large axial fibres, comparable to typical threads. These are surrounded by thousands of cribellate nanofibres and a substructure consisting of a network of paracribellate fibres (Fig. 1.4 A) (Eberhard et al. 1993; Opell 2013; Peters 1983). The cribellate fibres shape the outside of the cribellate thread, subdividing it into two regions: a puff with a larger diameter than the intermediate zones, between two puffs (Opell 1989; Peters 1984). The fibres originate from the eponymous cribellum, a spinning plate anterior to the spinnerets with up to 40,000 spigots, in Uloboridae about 5400 spigots (Fig. 1.4 B, C) (Bertkau 1882; Foelix et al. 1978; Peters 1982; Wilder 1874). The paracribellate fibres emerge from the paracribellum, a structure of elongated spigots on the median spinnerets with very similar glands to the ones of the cribellum (Fig. 1.4 B, D) (Kovoor 1978; Peters 1983;

6

Peters et al. 1980). The dominant occurrence of the paracribellate fibres within the thread is a substructure in the intermediate zones, only visible in polarized light (Fig. 1.4 A) (Peters 1984). The larger axial fibres are extracted from the spigots of the glandulae pseudoflagelliformes, one spigot of those situated on each posterior spinneret, probably by the movement of the whole spider during thread production (Eberhard et al. 1993; Kovoor 1978; Peters 1983; Peters et al. 1980). In contrast, cribellate fibres are suggested to be actively pressed out in a viscous state and transformed into fibres by the traction forces exerted upon them by the calamistrum (Kronenberger et al. 2015; Peters 1984). The calamistrum is a comb-like row of setae on the metatarsus of the fourth legs (Fig. 1.5 A). It is expected to be mainly involved in the extraction of the cribellate fibres and their transportation to the axial fibres (Bertkau 1882; Kullmann et al. 1971; Opell 1989, 2002; Peters 1983, 1992c). The setae of the calamistrum are twisted and curved, with the tips of the setae pointing towards the tarsus and away from the opisthosoma (Fig. 1.5) (Peters 1984). Due to this structuring it is hypothesised that facilitated by an angular change the calamistrum traps the fibres between its setae (Kovoor et al. 1988; Opell 2013; Peters 1984).

Fig. 1.5: The calamistrum. A) Ventral overview over the fourth leg of Uloborus plumipes. On the metatarsus, you can see the calamistrum, a specialised row of setae, used during the cribellate thread production. B) Close-up of the tips of the calamistrum (ca). On this side, a second row of similar structured setae (2 nd) is present in U. plumipes. Its function is unknown. C) Close-up of the side view, facing the spinnerets during thread production. D) Single setae of the calamistrum, separated by removing the others with a piece of freshly broken glass. SEM images, samples coated with gold.

If such trapping facilitates the longer extraction of cribellate fibres compared to the axial ones, this could lead to an accumulation of cribellate fibres, probably forming the puffy structure (Eberhard et al. 1993; Peters 1992a). The calamistrum might as well loop the cribellate fibres, by either twisting the fibres on the curved setae or by having the tension removed when fibres are freed from the extracting calamistrum (Opell 1989). Furthermore, it is commonly assumed 7 that the calamistrum somehow charges the cribellate fibres electrically, leading to a repulsion of the single fibres and therefore to the establishment of the puff (Elettro et al. 2015; Kronenberger et al. 2015; Opell 1982a, 1994c; Peters 1984; Sahni et al. 2011; Vollrath 2006). Such repulsion would mean that the intermediate zones are established by some kind of constriction, bringing the cribellate fibres closer together. Because of that it is hypothesized that cribellate fibres are jammed between the axial fibres in these regions (Peters 1984). This jamming-hypothesis is supported by a rhythmical ab- and adduction of the posterior spinnerets. Although there are several hypotheses about the thread structure establishment, these have never been examined more closely and/or corroborated. Hence, it is not evident how thread structure is and stays established. Without this knowledge, one is not able to explain the different thread shapes occurring in different species (Fig. 1.6). Only the looped structure of capture threads of Filistatidae (Fig. 1.6 E) is explained by suggesting a fixation of the capture thread to a previous laid supporting line (Opell 1999, 2002).

Fig. 1.6: Different thread structuring depending on the species. A) Thread of the Uloboridae waitakerensis. From Opell (1999). B) Thread of the Uloboridae Miagrammopes animotus. From Opell (1995). C) Thread of the . From Opell (1999). D, E) Thread of the Mexitlia trivittata (D) and Filistatidae Kukulcania hibernalis (E). From Opell (2002). F) Thread of the Hickmania troglodytes. From Griswold et al. (2005).

Beside the insufficient description of thread structuring, there is little proven knowledge about the production itself. To brush out the fibres with the calamistrum and with this action structure the thread structure, Uloboridae establish a unit with both fourth legs, one being the combing leg, placing the tarsus on the metatarsus of the other fourth leg, called supporting leg (Fig. 1.7 B) (Eberhard 1972, 1988). The calamistrum of the combing leg is moved from the anterior to the posterior of the spinnerets and back again, probably coming in close contact with 8 the spigots of the cribellum, because the concavity of the metatarsus bearing the calamistrum matches the convexity of the cribellum (Peters 1984). This way the calamistrum of brushes out the fibres with at least 300 cycles per minute, respectively 480 when belonging to Uloborus plumipes (Kronenberger et al. 2015; Wiehle 1927; Wilder 1874).

Fig. 1.7: Different combing postures. Cribellate spiders show two different combing postures, depending on the supporting leg (s) used. If they use the third leg (sIII), like e.g. Filistatidae, this posture is called “Type I” (A , dorsal view on the spider). If using both forth legs for combing (sIV), like e.g. Uloboridae, this posture is called “Type II” (B, ventral view on the spider). The combing leg is always the fourth one, bearing the calamistrum (cIV, shaded in grey). After Eberhard (1988).

1.5. Differences in the production of the cribellate thread In other spiders the process can differ from the one observed in Uloboridae. For example the frequency of combing is estimated with 120 cycles per minute in Acanthoctenus spinipes and 200 cycles per minute in Deinopis subrufa (Clyne 1967; Wiehle 1931). Furthermore, not all spiders use both fourth legs for combing. Some spiders like the Filistatidae use their third leg as supporting leg. This combing position is called Type I, whereas using both fourth legs is Type II (Fig. 1.7) (Eberhard 1988). Not only behaviour can differ, but also the morphology of the calamistrum and spinnerets vary between spider genera (Foelix 2011, 2015; Griswold et al. 2005). In many species, e.g. the Filistatidae Kukulcania hibernalis, the Psechrus sp., the Dictynidae Mallos pallidus or the Eresidae lineatus or henscheli, the cribellum is divided in two plates (Bond et al. 1997; Griswold et al. 2005; Kullmann et al. 1971; Miller et al. 2012; Opell 2002; Peters 1983, 1987, 1992a, b). Some spiders even have a cribellum divided in four parts, like spinimana and the Eresidae Dressus sp. (Foelix et al. 2015a; Griswold et al. 2005). The division of the cribellum in two parts is known to affect capture thread structuring. These spiders build two single strands of cribellate threads during combing, which 9 can be separated after thread production (Griswold et al. 2005; Lopardo et al. 2007; Peters 1987, 1992b). Only one axial fibre is present in these finished threads, but this procedure has no influence on thread performance (Bond et al. 1997). Not all species have a paracribellum, e.g. Psechrus sp. has not (Griswold et al. 2005). Many other spiders have only single spigots instead of an array, producing a paracribellate substructure called “undulating fibres” or “reserve wrap” (Griswold et al. 2005; Lehmensick et al. 1957; Miller et al. 2012; Peters 1983, 1992a, b). Its function is assumed to be a supporting one (Peters 1983, 1992c). Furthermore, spigots of the paracribellum not have to be necessarily placed on the median spinnerets as in Uloboridae, but can also occur on the posterior spinnerets or on both (Griswold et al. 2005). It is not described how these differences influence thread structure or performance. Finally, the calamistrum of different species can be built of e.g. one or several rows of setae and sometimes bear a teeth-like substructure (Foelix 2011; Griswold et al. 2005). In U. plumipes only one row of setae belongs to the calamistrum (Peters 1984). A second row nearby with a structure similar to the calamistrum might help extracting paracribellate fibres, although its function is still under discussion (Peters 1984). Eresidae on the contrary have several rows of setae. In addition, they have a teeth-like substructure on their setae, which is suggested to facilitate extracting cribellate fibres. Although Uloboridae are lacking these teeth, their close relative sp. from the Deinopidae again bears such teeth (Griswold et al. 2005). Already these supplemental information about other species show that the hypotheses established for thread shape formation by studying basically one single species have their limitations. E.g. splitting the thread in two and leaving each single strand with only one axial fibre inhibits a jamming of cribellate fibres to build the intermediate zone. These spiders nevertheless have a puffy structured thread (Griswold et al. 2005; Lehmensick et al. 1957; Opell 2002; Peters 1983, 1992a). Beside the influence of the divided cribellum on the structure of capture threads, differences in morphological features and thread structures have not been correlated to one another.

10

1.6. Aim of this thesis Although the current literature proposes several hypotheses to explain the cribellate thread production, the previous performed single species studies have their limitations. A complete model, validated with the observations of also distant related cribellate spiders, is missing. Furthermore, single species observations do not allow any abstraction or estimation about possible factors involved in handling or processing the nanofibres. The aim of this thesis is therefore to analyse the morphological and behavioural differences or similarities during capture thread production and to evaluate their impact on the thread structure. Finally, a broadly valid model for the cribellate thread production will be proposed. For the primary establishment of a model, the cribellate species U. plumipes (Uloboridae) was chosen, whose capture thread building behaviour has already been well described. To validate the model afterwards, spiders from very different families were incorporated (Fig. 1.8). The nearest related species was the likewise orb-weaving Uloboridae Zosis geniculata. There is no data available about the thread structuring as well as spinneret and calamistrum morphology of this species. Also, by including the Deinopidae, spiders building modified orb webs could be analysed. These spiders are net-casting, catching their prey by throwing themselves and their web onto approaching prey (Clyne 1967; Coddington et al. 1987). Their web-building behaviour differs from typical orb-weaving spiders for example by keeping their bodies on spot during web construction (Clyne 1967). Spinnerets are similar to the ones observed in Uloboridae, but their calamistra bear teeth (Griswold et al. 2005; Peters 1992c). Other spiders included in this study are e.g. the Eresidae. They do not build classical radial webs, but rather unkempt ones. Furthermore, they split the cribellate thread into two strings after production (Peters 1983, 1992a, b). Their calamistrum has multiple rows of setae and bears teeth, as already previously stated. The most unusual cribellate spider included in this study is . This spider is a hunting spider and although it builds cribellate threads, it does not use them as catching devices (Foelix et al. 2015a). Their function is unknown. With such a diversity in spider species, observable differences in the capture thread production should reflect different needs during web construction, differences in the calamistra morphology and/or explain variations in thread structuring, whereas similarities hint to preserved features of the cribellate spinning process.

11

Fig. 1.8: Degree of relation between the cribellate spiders used in this study. The information needed to develop this scheme are extracted from Bond et al. (20 14) and Fernandez et al. (2014). Data missing otherwise were taken from other sources. Note that although phylogenetic relations are presented, this scheme is not a phylogeneti c tree. Phylogenetic relations are continuously actualized. E.g. at the moment it is discussed whether Uloboridae and Deinopidae actually belong to the same superfamily of Deinopoidae (Garrison et al. 2016).

12

2. Material and Methods

2.1. Study animals

2.1.1. Keeping and breeding spiders for in vivo experiments

2.1.1.1. Ethics The species used in the experiments (Uloborus plumipes and Zosis geniculata) are not endangered or protected species. Special permits were not required. All applicable international, national and institutional guidelines for the care and use of animals were followed.

2.1.1.2. Uloborus plumipes

Uloborus plumipes (LUCAS, 1846) of undefined age and sex were captured in different flower selling shops in Aachen, Essen and Rostock (Germany) in 2013 and raised in the lab colony under room temperature (≈ 21 °C), room humidity (≈ 30%) and northern European diurnal rhythm. They were kept either separated in petri dishes (adult: Ø 9 cm, juvenile: Ø 4-6 cm) with roughened surface or in larger mixed colonies in several terrariums with egg carton as structured ground and wood branches as web support. Cocoons were collected directly after the mother stopped protecting it. Freshly hatched spiders were separated from one another to inhibit cannibalism. Once to twice a week spiders were fed with Drosophila melanogaster, juvenile Acheta domestica, bean weevils or aphids. Juvenile spiders were fed more often than adult ones. Water was provided once to twice per month by sprinkling the web. From the spiders which were separated after hatching the date of the cocoon laying, the date of hatching as well as the one of each moult was logged to analyse the duration until reaching adulthood. Furthermore, the survival rate as well as the partition between males and females was logged. Spiders not surviving until adulthood were neither counted as males nor females.

2.1.1.3. Zosis geniculata

Adult spiders of Zosis geniculata (OLIVIER, 1789) were raised from the hatchlings of two cocoons. They were kept as larger colonies in several terrariums under the same circumstances already described for U. plumipes (chapter 2.1.1.2.).

13

2.1.2. Preserved specimens Preserved specimens (70% ethanol) of undefined age and gender of one Badumna longinqua

(L. KOCH, 1867), several Z. geniculata as well as several Deinopis subrufa (L. KOCH, 1879) were provided by Dr. Barbara Baehr, Dr. Robert Raven and Robert Whyte from the Queensland Museum in Brisbane, Australia. In addition, two B. longinqua, one D. subrufa, one Hickmania troglodytes (GERTSCH, 1958), one Miagrammopes sp. and one Menneus sp. were borrowed from Graham Milledge from the Australian Museum in Sydney, Australia.

2.1.3. Thread samples Samples of B. longinqua’s thread were provided by Barbara Baehr, Robert Raven and Robert Whyte from the Queensland Museum in Brisbane, Australia. They were stored dark and protected against dust at room temperature and room humidity in a box until use. Samples of Kukulcania hibernalis’ threads were taken from a colony of Dipl-Biol. Thomas Vinmann and used directly after collection in the experiments.

2.2. Behavioural experiments

2.2.1. Observations in the natural habitat and the lab colony Observations of wild populations of U. plumipes were made in different flower selling shops in Aachen and Essen in 2013 to 2015. A more detailed study including density measurements, data for web sizes or the occurrence of male spiders was performed at a flower shop, where no population-minimizing actions (e.g. webs destroyed due to cleaning) had been performed during the last months (Florenia e.K., Essen). Population densities were measured at eight spots with an average size of 0.66 m2 and a height of ≈ 2 m. Spots not easy-to-reach were chosen to minimize the influence of fluctuation of plants due to the removal by customers. We excluded juvenile spiders smaller than 2 mm, because freshly hatched spiders remain together for a few days without competing, thereby increasing density. As they grow, spiders spread out over the area and build individual webs. Seasonal variations were not monitored.

14

To investigate social interactions under controlled conditions, a small colony of adult female virgin spiders (last moult in captivity) was kept in a terrarium of a size of 47.0 × 31.5 × 25.5 cm (length × width × height), decorated like described previously (chapter 2.1.1.2.). To prevent escaping, the top was covered with gauze. Population density in this terrarium was increased stepwise starting with four individuals and adding one every two weeks up to finally six individuals (40.5 individuals/m2). This experiment was carried out over a period of six weeks by Anne Habedank during her bachelor thesis in 2014. Since U. plumipes is nocturnal, position and size of the webs as well as the location of all individuals were logged every morning, excluding weekends.

2.2.2. Standardized interactions For the standardized interactions, ten adult virgin female U. plumipes (last moult in captivity) were kept separately in 9 × 9 cm wide arenas with eight 7 cm long stakes (tooth picks) at all sides as attachment points for their webs (Fig. 2.1). After some days of acclimatisation to the new environment all spiders spun horizontal webs on their individual platform. To start an interaction, two arenas were placed adjacent to each other and one spider (intruder) was motivated with an air puff to enter the web of the other spider (defender). When starting a contest, it was controlled that the defender itself was not sitting at the hub of its web, because the occupation of the hub in competitions has been proven to be beneficial for some spiders (Buskirk 1975; Christenson et al. 1979). After 20 min prey in form of one A. domestica was added to the web. This enabled the experimenter to define the individual capturing the prey as owner of the web. Duration and winner of web contests, as well as behavioural patterns during the interactions were logged by Anne Habedank during her bachelor thesis in 2014. Additionally, she recorded videos (50 fps; Canon EOS 550D, Canon, Tokyo, Japan) using one camera from above and one camera at an angle of 90°. The last one was attached with an adapter to a binocular microscope with 70x magnification (Olympus, Tokyo). This helped to reconstruct details.

15

Fig. 2.1: Arena for standardized interaction. * An arena was designed, where the spider was motivated to b uild its web at the tips of the toothpicks. A yarn was used to connect the toothpicks with each other , that way reducing the walk for the spider. Because web-building spiders are not able to walk easily on glass, a petri dish was used to prevent escaping from the area.

These web contests were performed once a week for every spider. Dividing the spiders in two separated groups (four spiders from Rostock and six spider from Aachen), each individual interacted with every other in its group once as a defender and once as an intruder, all in all performing 44 experiments. Note that this rule was violated in one case, because one spider did not build a functional web. Furthermore two contest-constellations were repeated to control the reproducibility of the experiments. At any time, all individuals had experienced the same number of contests as their contestant since reaching adulthood. Drawing of the contestants was otherwise done randomly. The weight of the spiders was determined either one day before the contest or directly after the experiment (no feeding occurred between). Since spiders were fed once per week, we considered hunger levels to be similar for all spiders. Furthermore, the feeding caused a linear increase of body weight (about 5.6% per week) for all individuals during the course of our experiments. At the end of the experiments, spiders were anesthetized with diethyl ether (≥ 99% stabilized with 5-7 ppm butylated hydroxytoluene; Carl Roth GmbH & Co., Karlsruhe, Germany) to measure body size (pro- and opisthosoma length) and length of the first legs. Body size and leg length do not change after the final moult.

* Pictures taken by Anne Habedank during her bachelor thesis “Kampf ums Netz – Intraspezifisches Konkurrenzverhalten von Weibchen der Federfußspinne Uloborus plumipes“ in 2014. 16

2.2.3. Characterization of vibrational patterns A detailed characterization of three vibrational behavioural patterns was performed with the help of high speed microscopic recording (500 fps; Keyence VW-600C, Keyence Cooperation, Osaka, Japan). The arenas of the previous described experiment (chapter 2.2.2.) were used and vibration provoked by placing two spiders in one web. Combining optical recordings from the top and from the side as well as an integrated calibration tool permitted to distinguish and measure horizontal and vertical movement vectors. The side view recordings were analysed with a custom-made tool implemented in MATLAB R2014b (Version 8.0.0.783, The MathWorks Inc., Natick, USA) for manual point-tracking. The position of the tarsus of one of the first legs in contact with the web was selected to track vibratory movements.

2.2.4. Statistical analysis Data are presented as mean ± SD (standard deviation), if not stated otherwise. “n” specifies the number of individual trials. For data characterizing the vibrational patterns, all available data for one spider was used to calculate the individual mean before calculating the general one. Data introduced in percentage values are shown as mean and the according 95% confidence interval (CI). To evaluate the different occurrence of behaviours, a two-tailed non-parametric Wilcoxon- Mann-Whitney U-test was used, as data were not distributed normally (after Kolmogorov- Smirnov, α = 0.05, two-tailed). To evaluate the relationship between being intruder or defender and winning a contest, a two-tailed Chi-Square-Test was used (α = 0.05 and one degree of freedom). Significance was assumed if p < 0.05. p-values are indicated in figures and text. These calculations were performed by Anne Habedank during her bachelor thesis in 2014, using SPSS (Version 21; IBM, Armonk, USA) or Microsoft Office Excel 2013 (Version 15.0.4603.100; Microsoft Cooperation, Redmond, USA). Correlations between web shaking parameters and morphological traits were carried out by Prof. Dr. Jörg Mey using a simple linear regression analysis with the least square approach and coefficient of determination (r2), determining whether the slopes were significantly different from zero. To evaluate factors influencing competitions, all factors which were independent from the experimental setup were used in a binary logistic regression model calculated by Jonas Hausen, M.Sc. from the Institute for Environmental Research (Biology V) at the RWTH Aachen. To check the chosen explanatory variables for collinearity, pairwise correlation coefficient (Spearman, polychoric or polyserial) was used. For pairs with high correlation, one of the variables was removed from analysis. The selected variables were used to fit the model to the

17 data. Afterwards, Akaike’s information criterion (AIC) was used in variable selection to determine the best set of explanatory variables. The final model was tested for significance against the null model using a Chi-Square-Test. To exclude autocorrelation between the results of the individual spiders, the same analysis was performed using a mixed effects logistic regression model with a random effect between the spiders included. The result of the model was nearly the same with slightly changed significance p-values of the variables used, and the variance of the random effect (spiders) was nearly zero (1.811*10-17). Therefore, only data from the binary logistic regression model are presented within the results. The binary logistic regression and mixed effects logistic regression as well as the necessary statistical analysis for model selection were performed using the packages arm, lme4 and polychor for the R statistical package (Version 3.1.2).

2.3. Recording cribellate spinning process

2.3.1. Animal housing for recordings Adult female specimens of U. plumipes and Z. geniculata were taken from our laboratory colony and raised separately. U. plumipes was placed in smaller boxes (6 × 5 × 2 cm) with black cardboard as background and web support. The lid and at least one side of the boxes were made of acrylic glass for undisturbed observation. Z. geniculata was placed into petri dishes (Ø 9 cm) with dark adhesive tape as background and web support. One side was cut open at about 1/4 of the petri dish diameter and sealed again with a microscope slide for undisturbed observation. For easier observation, their diurnal rhythm was shifted 12 h: white light was turned on from 06:00 pm to 07:00 am. Spiders were illuminated from 07:00 am to 06:00 pm using red LED (Paulmann Licht GmbH, Springe, Germany).

2.3.2. Nocturnal observations To visualize nocturnal movements of the spiders in the terrarium, video recordings of U. plumipes were made using a webcam (Logitech HD Webcam C270, Apples, ; software: Logitech Webcam Software Version 2.5.1) and a red light source (gooseneck lamp with red light filter). Because the resolution is not very high with this technique, it is only possible to observe the general movement of a spider and determine, if e.g. the leg movement during cribellate thread production is performed at all. This experiment was used to control spiders modified according to chapter 2.5.1.

18

2.3.3. Recordings with low speed camera

Fig. 2.2: Setup for low-speed recording. *

In the “low-speed”-setup recordings could be performed under red light with a modified gooseneck lamp (compare to chapter 2.3.2.). Spiders were observed from top and side view, using two binocular microscopes and 10× to 70× magnification (Fig. 2.2). The top view resulted in an observation of the ventral side of the spider, because the spider hangs upside down in its webs during thread production. The microscopes could be moved almost vibration-free due to PTFE (Teflon) sliding feet. A digital reflex camera (EOS 550D) was attached to both binocular microscopes and recording was performed like described in chapter 2.2.2. The data were used to analyse leg and spinneret movement during the cribellate spinning process. Video recordings of U. plumipes were mainly performed by Peter Kappel during his bachelor thesis in 2014 and by Linda Orth for Z. geniculata during her bachelor thesis in 2015.

* Picture taken from Peter Kappel’s bachelor thesis “Analyse des Spinnprozesses cribellater Fasern der Federfußspinne (Uloborus plumipes)” in 2014. 19

2.3.4. Recordings with high speed camera The “high speed”-setup required white light during recordings. Spiders were brought into focus of a high-speed video recording microscope (VW-9000C) using red light. During production of capture threads, white light of a gooseneck lamp was activated, leading to short video sequences, in which spiders were still producing thread before stopping due to disturbance by the white light. Such procedure is only possible when spiders are close to finishing their web. Recordings were performed with 500 to 1000 fps and at about 100× magnification. This experiment was performed only with U. plumipes by Peter Kappel during his bachelor thesis in 2014.

2.3.5. Using freely available clips Online available video recordings of cribellate spiders were used accordingly to Nelson et al. (2013) (Tab. 2.1). Special care was taken for determining the frequency by only evaluating recordings were speed manipulation can be excluded.

Tab. 2.1: Links to recordings of cribellate spiders. Link Date accessed Spider Special data extracted

https://www.youtube.com/ Deinopis angle between body & thread (β), 13/08/2015 watch?v=GWY5cBueiHk subrufa elongated leg movement

https://www.youtube.com/ angle between body & thread (β), 08/06/2015 Deinopis sp. watch?v=_NthIs56OkY frequency

https://www.youtube.com/ 08/06/2015 Deinopis sp. frequency watch?v=1H4ayAN1u7o

https://www.youtube.com/ 13/08/2015 Deinopis sp. none watch?v=e4MhaEgrDV4

https://www.youtube.com/ 13/08/2015 Deinopis sp. none watch?v=gryaUTpPgGU

http://footage.framepool.com/de/bin/12 no. 640-469-842 & 809-396-662: 10/06/2015 Deinopidae 250,deinopidae,spinnennetz,australien/ elongated leg movement

https://www.youtube.com/ Zoropsis 16/01/2016 frequency watch?v=n-f2tJ0pYfE spinimana

https://www.youtube.com/ Zoropsis 16/01/2016 frequency watch?v=Z1J2pB45T8k spinimana

20

Link Date accessed Spider Special data extracted

https://www.youtube.com/ Zoropsis 16/01/2016 frequency watch?v=dXe9zjyRXGg spinimana

https://www.youtube.com/ Zoropsis 16/01/2016 frequency watch?v=uhmm3yBy6Zw spinimana

https://www.youtube.com/ Kukulcania 16/01/2016 frequency watch?v=HP8KkSKBYYc hibernalis

https://www.youtube.com/ Kukulcania 16/01/2016 frequency watch?v=HvcNdInCCr0 hibernalis

https://www.youtube.com/ Stegodyphus 16/01/2016 frequency watch?v=8AwxVYHUtXg mirandus

https://www.youtube.com/ 16/01/2016 Psechrus sp. frequency watch?v=02RZe0c_VHo

https://www.youtube.com/ 16/01/2016 Psechrus sp. frequency watch?v=gwG2-YG4KXc

https://www.youtube.com/ Eresus 16/01/2016 frequency watch?v=5gYrYq_qL5c walckenaeri

https://www.youtube.com/ Eresus 16/01/2016 frequency watch?v=X4TDDR8sbGI walckenaeri

2.3.6. Data acquisition and statistical analysis Selected videos were analysed using Keyence VW-9000 Motion Analyser (Version 1.4.0.0; Keyence Cooperation, Osaka) or a custom made tool implemented in MATLAB R2014b either for point-tracking or for angle measurements. The following data were extracted from the recordings: 1) The frequency of the movement of the combing leg was determined by counting the number of strokes over time. More than one combing movement was incorporated to evaluate one data point. This analysis was performed by Peter Kappel during his bachelor thesis in 2014 for U. plumipes. For Z. geniculata, this analysis was performed by Linda Orth during her bachelor thesis in 2015. 2) If the movement of the posterior spinnerets was not synchronously to the movement of the combing leg, the frequency of the movement of the posterior spinnerets was determined by choosing one posterior spinneret and by counting how often it is abducted over time. More than one movement was incorporated to evaluate one data point. 21

3) If a high speed recording was available, the movement of the combing leg was divided into of three periods: from anterior to median position, from median to posterior position and from posterior to again anterior position (definition of the positions see chapter 3.2.4.). The velocity is given by the quotient of the distance between two positions and the duration of the movement in-between. This evaluation was performed by Peter Kappel during his bachelor thesis in 2014. If only the difference between posterior and anterior leg movement could be observed due to low temporal resolution, the difference was determined by counting the number of frames necessary for the combing leg to reach the posterior respectively the anterior position again. Afterwards, the ratio of both numbers of frames was calculated. More than one movement was incorporated for evaluating one data point. 4) To identify the number of puffs produced during one stroke of the combing leg, samples of freshly produced capture threads were taken and puffs between two radial threads as well as the number of stroking during the matching sequence in a recording were counted. Small variations which could not be explained by the capture thread production itself were neglected. This analysis was performed by Peter Kappel during his bachelor thesis in 2014 for U. plumipes. For Z. geniculata, this analysis was performed by Linda Orth during her bachelor thesis in 2015. 5) To evaluate the elongated leg movement observable in Deinopis sp., the metatarsus of the combing leg was tracked from the lateral perspective with the help of the point-tracking tool. By tracking two cycles previous to the elongated one, the mean length of a normal stroke was calculated. Due to a missing scale bar, only the ratio of the elongated one to a normal stroke can be determined. 6) The point-tracking tool was also used to calculate the lowering of the opisthosoma of U. plumipes, Z. geniculata and Deinopis sp. (data not shown for Deinopis sp.) by tracking the position of the spinnerets from the lateral perspective after the spider fixed the capture thread at the radial frame. Given that low-speed recordings have no scale bar incorporated, the scale bar was estimated by the known dimension of the spider (typically the calamistrum length). For Z. geniculata, the evaluation was performed by Linda Orth during her bachelor thesis in 2015. 7) With the help of an angle measurement tool, the angle between combing leg and cribellum (called α) of U. plumipes and Z. geniculata was determined. This evaluation for Z. geniculata was performed by Linda Orth during her bachelor thesis in 2015. Her raw data were re-evaluated for this thesis.

22

8) To determine the angle variation between the different segments of a spider’s leg, recordings of U. plumipes were analysed by Peter Kappel during his bachelor thesis in 2014 using the Keyence VW-9000 Motion Analyser. Only the difference between anterior and posterior position (definition of the positions see chapter 3.2.3.) was determined. 9) The angle between opisthosoma and produced capture thread (β) was determined with the Keyence VW-9000 Motion Analyser. Therefore, single pictures of a recording of a spider from the lateral perspective were selected, where spider and thread were visible. Special focus was put on the exclusion of optical distortions due to incorrect orientation of the spider. Hence, only U. plumipes and Deinopis sp. could be evaluated. Data are presented if possible as mean ± SD. “n” indicates the number of different spiders used for these data and “SD” is the standard deviation between the individuals. The mean of each spider was calculated individually before calculating a general one. If data of only one spider were available (noted as “n1”), the SD refers to the variance of this data sheet and the value of n1 indicates the number of data points.

2.4. Microscopic analysis

2.4.1. Scanning electron microscopy (SEM)

2.4.1.1. Thread preparation Two stripes of conductive foil (PLANO Leit-Tabs; Plano GmbH, Wetzlar, Germany) were put onto an aluminium stub (Stiftprobenteller; Plano GmbH) with a distance of less than 5 mm. Thread samples were spun between those stripes by approaching slowly the web from above or below. If the background of the sample had to be more homogenous than the simple stub, a piece of aluminium foil was placed between the conductive foils and the stub. The samples could be either observed without any further preparation, after coating with carbon or after coating with a 10 nm gold layer (Hummer; Technics Inc., Alexandria, USA) before examination in a scanning electron microscope (SEM 525 M; Philips AG, Amsterdam, Netherlands). Carbon coating was performed by Dr. Moritz to Baben from the Institute for Material Chemistry, Lehrstuhl für Werkstoffchemie at the RWTH Aachen. One sample was carbon coated by Dr. Alexander Schwedt and Ruth Harscheidt from the central facility for electron microscopy of the RWTH Aachen. Data generated with this sample are marked with an according footnote. The axial fibre distance of native threads as well as the maximal thread diameter (typically the puff) were measured using the Keyence VW-9000 Album. Data are presented if possible as 23 mean ± SD. “n” indicates the number of different spiders used for these data and “SD” is the standard deviation between the individuals. The mean of each spider was calculated individually before calculating the general one. To evaluate differences between spiders, a two-tailed T-test was performed. Significance was assumed, if p < 0.05. p-values are indicated in the text or in tables. Calculations were performed with Microsoft Office Excel 2013.

2.4.1.2. Specimen preparation The preserved spiders of chapter 2.1.3 and likewise preserved specimen of U. plumipes of defined aged were gradually (80%, 90%, 95%, 99%) transferred to 100% ethanol, each step for 30 min. Afterwards they were dried using hexamethyldisilazane (Merck, Darmstadt, Germany) in ascending concentration (3:1, 1:1, 1:3, 0:1 with ethanol) for each 1 h. Either the whole spider or parts of it were placed on an aluminium stub with conductive foil. If only parts of the spider were placed on the stub, it was taken care that the interesting spots are clearly visible by positioning the sample using a binocular microscope and, if necessary, removing all parts blocking the view. Specimens were sputter coated with a 10 nm gold layer before examination in a SEM. The spigot number on the cribellum of U. plumipes was counted. After the 4th shedding, spigot number was determined by counting the spigots in a close-up of the middle of the cribellum. The number of spigots on the complete cribellum was extrapolated by measuring the area (length × width) of the cribellum. Data are presented if possible as mean ± SD. “n” indicates the number of different spiders used for these data and “SD” is the standard deviation between the individuals. If data of only one spider was available, no SD is given.

2.4.2. Focused ion beam treatment (FIB) and tomography Adult females of U. plumipes were fixed in 1% glutaraldehyde (≈ 25% in water (2.6 M); Fluka Chemie GmbH, Buchs, Switzerland) with 70% ethanol. Afterwards they were dried with ethanol and hexamethyldisilazane as described in chapter 2.4.1.2. The fourth leg of the spider was removed and positioned on an aluminium stub with conductive foil. In some samples, setae of the calamistrum were removed as described in chapter 2.5.1. The dropped off setae were left on the conductive foil. Afterwards the samples were coated with a 10 nm gold layer.

24

These samples were now either examined using a focused ion beam (FIB) in a SEM with Martina Schiffers and Dr. Daesung Park from the central facility for electron microscopy of the RWTH Aachen. Otherwise, they were likewise examined using a FIB but pictures were assembled to a 3D model by Dr. Wolf-Alexander Heiß from the Forschungsinstitut Edelmetalle + Metallchemie in Schwäbisch Gmünd.

2.4.3. Transmission electron microscopy (TEM)

2.4.3.1. Thread preparation Samples were taken by picking up a piece of thread with a finder-grid (400er mesh, 3.5 mm, Plano GmbH). Without any further treatment, the samples were observed with the transmission electron microscope (EM 10; Carl Zeiss AG, Oberkochen, Germany). With the help of Keyence VW-9000 Album, the diameter of the cribellate fibres as well as the diameter of the knots on the cribellate fibres were measured. Furthermore, the distance of two knots was determined. Data are presented as mean ± SD. “n” indicates the number of different spiders used for these data and “SD” is the standard deviation between the individuals. The mean of each spider was calculated individually before calculating a general one. If data of only one spider were available (noted as “n1”), the SD refers to the variance of this data sheet and the value of n1 indicates the number of data points. The data for juvenile spiders of U. plumipes was collected by Margret Weißbach during an internship in 2015. To evaluate differences between spiders a two-tailed T-test was performed. Significance was assumed, if p < 0.05. p-values are indicated in the text or in tables. Calculations were performed with Microsoft Office Excel 2013.

2.4.3.2. Specimen preparation Two females of U. plumipes were fixed in 1% glutaraldehyde with 70% ethanol. Afterwards, they were transferred to 100% ethanol and stored in there for 15 min. The ethanol was removed by washing the samples twice with propylene oxide (Serva, Heidelberg, Germany) for each 30 min. The samples were stored overnight in a mixture of 1:1 propylene oxide and epon resin (Epon; Serva). The propylene oxide evaporated overnight and samples were washed twice in epon resin for each 2 h. Samples were transferred to embedding moulds, filled with epon resin, and the epon resin polymerised for 48 h at 65 °C. The block was trimmed and afterwards slices were cut off with a diamante knife and an ultra-microtome (Reichert OmU3; Reichert GmbH, Wien, Austria). Samples were transferred to finder grids and contrasted with heavy metal for TEM observations. Therefore, a drop of 2% uranyl acetate (Sigma-Aldrich, München, 25

Germany) in 70% ethanol was placed on a parafilm piece. The grid with the sample was placed onto the drop, the sample facing downwards. This was incubated dark at room temperature for 20 min. Afterwards samples were washed with degassed double distilled water 5 times, by dipping the grids into the water. The grids were dried by taking off supernatant water with filtering paper. The same procedure was repeated with 0.2% lead citrate (Sigma-Aldrich) instead of uranyl acetate. This was incubated only 7 min, before washing the samples with water. Afterwards samples were finally dried for at least 30 min previous to observation in the TEM.

2.4.4. Light microscopy

2.4.4.1. Thread observations To observe thread structure under native conditions, threads were picked up with two arms of a metal filament and as a whole placed on black cardboard for better contrast. Pictures were taken with the help of a microscope (VW-9000C) at 100× to 1000× magnification. To illuminate the paracribellate substructure, polarized light was used.

2.4.4.2. Specimen preparation Specimen of U. plumipes were prepared as described in chapter 2.4.3.2. After epon embedding, slices were cut of the block with a glass knife instead of a diamante knife. The samples were placed onto cover slips and dried at 60 °C for 2 h. Afterwards, they were polychromatically stained: For 45 s they were placed in a solution, consisting of 12.5 ml carbol methylene blue

(2% methylene blueaq and 0.5% phenoleaq), 12.5 ml carbol gentian violet (Chroma – Waldeck, Münster, Germany), 10.0 ml 96% ethanol, 12.5 ml distilled water and 2.5 ml pyridine (Sigma- Aldrich). The sample was washed in distilled water for 45 s and afterwards covered with a solution consisting of 5 g paraosaniline powder (Chroma – Waldeck), 97 ml distilled water, 1 ml glacial acetic acid and 2 ml 5% phenol (Carl Roth GmbH & Co.). After 45 s the solution was replaced by distilled water, in which the sample was washed for another 30 s. Samples were dried at room temperature previous to embedding the slices in Euparal (Chroma – Waldeck) on a microscopic slide. This slide was afterwards used for light microscopic observations of the sample.

26

2.5. Manipulation of spiders

2.5.1. Removing setae rows Adult female spiders of U. plumipes were anaesthetised with diethyl ether and fixed by confining the spider with needles on a petri dish (Fig. 2.3 A, B). Using a freshly broken piece of glass cover slip, the calamistrum was removed by rubbing the ridge of the glass against the setae of the calamistrum (Fig. 2.3 C). The same method was used to remove the second row of setae nearby the calamistrum. Spiders with non-removed calamistrum were taken as a control group. For the examination of capture threads produced by such modified spiders, samples of freshly produced threads were taken (compare chapter 2.4.1.1., 2.4.3.1., 2.4.4.1. and 2.6.). As a control of successful modification, the spiders were frozen at -20 °C and dried for a week at room temperature using drying pellets. Specimens were sputter coated before examination in a scanning electron microscope. Some spiders were modified by Hana Adamova during her bachelor thesis in 2015.

Fig. 2.3: Setup for modification of U. plumipes.

2.5.2. Conglutination of the calamistrum Adult female spiders of U. plumipes were anaesthetised with diethyl ether and fixed by confining the spider with needles on a petri dish. With the help of a needle, a small droplet of superglue (UHU SEKUNDENKLEBER blitzschnell SUPERGEL; UHU GmbH & Co. KG, Brühl, Germany) was picked up and brought near to the gap between calamistrum and the second row of setae nearby. The glue was sucked up by this gap, gluing the setae of the calamistrum together. For observation of the capture threads produced by such a modified spider, samples of freshly produced threads were taken and examined (compare chapter 2.4.1.1., 2.4.3.1. and 2.4.4.1.). As a control of successful modification, the spiders were frozen at -20 °C and dried for a week at room temperature using drying pellets. Specimens were sputter coated before examination in a scanning electron microscope. 27

2.5.3. Shock freezing experiments 25 adult female spiders of U. plumipes were transferred into petri dishes (Ø 9 cm) with roughened surfaces for web support. Their diurnal rhythm was shifted 12 h (compare to chapter 2.3.1.). They were shock frozen with cold spray (TC-KC400C; Toolcraft, Hirschau, Germany) during production of cribellate threads, immediately transferred to -20 °C and stored for two weeks on drying pellets. After another two weeks at +4 °C, they were transferred to room temperature. Fresh drying pellets were added, if necessary. Specimens were sputter coated with a 10 nm layer of gold before examination using a SEM.

2.6. Adhesion measurements

2.6.1. Prey adhesion Adhesion measurements were performed with living and dead D. melanogaster. To kill the fruit flies, they were frozen at -20 °C. The dead flies were afterwards dried at room temperature. By placing these samples onto capture threads, the adhesion area between prey and different types of capture threads could be determined without any further drying treatment. Samples were coated with gold and observed with the SEM. This experiment was performed by Hana Adamova during her bachelor thesis in 2015. To determine the retention time and not directly the adhesion area, fruit flies hatched that very day were taken. By gently picking them up with a featherweight forceps, they were placed in a web, from which the spider was removed. If squeezed too much and this way curtailing the fruit flies’ movement, fruit flies were discarded to eliminate errors due to distorted behaviour. The time which the fruit flies took to escape the capture thread was measured afterwards. U. plumipes react typically directly to living prey within the web, wrapping it up tightly. Therefore, the biological relevant time scale is “< 2 min” (prey might be able to escape) and “< 5 min” (capturing prey is most likely). Everything above this relevant time is counted as “> 5 min”, as prey would be definitely wrapped by the spider. To evaluate differences in the retention time, a G-test with two degrees of freedom was performed. Significance was assumed, if p < 0.05. Calculations were performed with Microsoft Office Excel 2013.

28

2.6.2. Standardized adhesion surface To quantify the adhesion force within different structured capture threads, pull of forces were measured with a micro balance (JB1603/C-FACT; Mettler Toledo AG, Greifensee, Switzerland). Threads were placed on a 2 cm wide aluminium U-beam. From above, an acetone cleaned tip of a ballpoint pen was brought near the thread with the help of a micromanipulator (M3301R; Märzhauser Wetzlar GmbH & Co. KG, Wetzlar, Germany). The whole process was monitored with the high-speed recording microscope (VW-9000C) from the lateral perspective (Fig. 2.4). After pressing slightly the tip of the ballpoint pen against the capture thread, the moving direction was reversed and the adhesion force could be measured with the micro balance. To calculate the adhesion force, the adhesion surface has to be determined. This is rather difficult with a wool like structure of nanofibres. Many studies on cribellate threads hence skipped the determination. Within this study, the difference between the width of the capture threads after adhering to standardized glass beads (Silibeads Typ S 300 – 400 µm; Sigmund Lindner GmbH, Warmensteinach, Germany) was determined with SEM (comparable to “dead prey adhesion” in chapter 2.6.1.). Because no significant difference was observed, adhesion surface was assumed to be equal in all tested threads. It was by default determined as the area of the spherical calotte of the ballpoint pen in contact with the thread, estimated by the depth of immersion. This experiment was conducted by Hana Adamova during her bachelor thesis in 2015.

Fig. 2.4: Setup for adhesion force measurement. * 1: sample on sample holder, 2: ballpoint pen, 3: micromanipulator, 4: micro balance, 5: microscope, 6: light source, 7: monitor to control position of ballpoint pen via microscope.

* Picture taken from Hana Adamova’s bachelor thesis “Der Einfluss vom Calamistrum auf die Struktur und die Funktion cribellater Fangfäden” in 2015. 29

To evaluate differences between the adhesion forces a two-tailed T-test was performed. Significance was assumed, if p < 0.05. p-values are indicated in the text or in tables. Calculations were performed with Microsoft Office Excel 2013.

2.7. Reaction to high humidity Threads of U. plumipes were spun between two arms of a metal filament and thread shape was examined with light microscopy (compare chapter 2.4.4.1.). Afterwards, the thread was placed in a climate chamber provided by the colleague Dipl.-Biol. Florian Hischen. Temperature was 28 °C and humidity was raised to 80% either by letting tap water evaporate from a tissue or by controlling humidity with the help of a nebulizer (Super Fog Nano, Lucky Reptile; Import Export Peter Hoch GmbH, Waldkirch, Germany). After an hour, the samples were taken from the climate chamber and again examined with light microscopy. To observe, if thread shape changes after drying the thread again, the samples were stored in a desiccator for a week, checking regularly the shape of the threads.

2.8. Electrostatic charging

2.8.1. Checking the thread’s charge Whole webs of U. plumipes, including the spider at the hub, were taken as a sample. After rubbing a glass rod against silk and therefore charging the rod positive, it was brought near to single capture threads without touching them. Due to the elasticity of these threads, any deformation can be easily seen with the naked eye. The same procedure was repeated with a piece of foamed plastic rubbed against cellulose tissue. The plastic is described to be charged negatively afterwards. The actual charge of the objects was not tested.

30

2.8.2. Recharging To check whether missing puffs (either due to the removal of the calamistra (compare chapter 2.5.1.) or due to the exposure to fine mist (compare chapter 2.7.)) can be re-established, a sample of the thread in question was placed on an aluminium stub as described in chapter 2.4.1.1. and transferred without coating into the SEM. Each sample was exposed directly to the electron beam for at least 5 minutes, observing thread structure and possible changes in the structure, before shifting to another sample. Although this procedure should be able to (re)charge the fibres, one has to keep in mind that there might be several forces acting together to keep the fibres apart. In this case, the removal of one force, e.g. removing the electrostatic charge, can already have an irreversible effect on the thread’s assembly.

31

32

3. Results

3.1. Uloborus plumipes

3.1.1. Ecology To establish a lab colony and perform experiments with living animals of the chosen organism U. plumipes (Fig. 3.1), its ecology such as distribution and behavioural patterns within the natural habitat needed to be analysed and afterwards surveyed in the lab colony. In its natural habitat of Western Europe (flower shops and garden centres) densities could rise over 100 spiders/m2. There was a large variation in density with a median of 14 spiders/m2 (n = 8) ranging from 4 spiders/m2 to 125 spiders/m2. Such high variation is probably influences by different factors like prey availability, structuring of the habitat as well as the removal of pot plants by customers. At night, U. plumipes built typically horizontal radial webs with an area of 708 ± 266 cm2 (n = 5) after reaching preferably the highest spot (Fig. 3.1 B).

Fig. 3.1: Uloborus plumipes. A) Female in its resting positon. B) Ventral view of a spider resting upside down in the hub of its horizontal web. C) Ventral overview over an adult male in its resting position. Note the missing tufts of hair on its elongated first legs, the enlarged pedip and its slim opisthosoma. D) Dorsal view of different coloured opisthosoma enabling the experimenter to discriminate individual spiders.

 Parts of this chapter have been included in the manuscript “Joel, A.-C., Habedank, A., Hausen, J., Mey, J.. Fighting for the web: Intraspecific competition among female feather-legged spiders Uloborus plumipes“. Writing of the manuscript was part of this thesis. 33

Although U. plumipes does not belong to the social spiders within the Uloboridae, different forms of interactions occurred: Freshly hatched spiders were observed sharing their web, and also juvenile spiders could connect their webs to one another by thin threads. At least between freshly hatched spiders competition for food did not occur. They were never observed feeding at all. Likewise, adult males did not build their own webs, but lived at the margin of webs of females. The female did not show cannibalistic tendencies towards the male. The males were even observed catching prey of the female’s web (own observations and Jonsson (1998)). In contrast to juvenile or sub-adult males, adults could be easily distinguished from females. Their opisthosoma shape was slimmer and the characteristic tufts of hair on their elongated first legs vanished with the final moult (Fig. 3.1 C). Only 10% (n = 8; 95% CI [21, 2]) of all encountered spiders in the natural habitat were adult males. This low density in adult males reflected in a low hatching rate of males within the lab colony (9% males (n = 13; 95% CI [18, 0]) with 9 ± 4 spiders/cocoon and a general survival rate of 85% (n = 14; 95% CI [24, 5])). Although parthenogenesis is suggested to happen in this species (Dawson 2001), reproduction without males was never observed in the lab colony. Two females were observed even sharing a web. The spider with the much larger opisthosoma occupied the hub, whereas the smaller spider was sitting at the margin of the web. When prey entered the web, the smaller spider tried regularly to steal the prey. Once a spider without a web (floater) was observed trying unsuccessfully to conquer an already occupied web in the natural habitat. Such conquering attempts could be observed regularly within the lab colony. To quantify these interactions, spiders in a natural-structured terrarium were observed during a period of 42 days*. The individual identification of spiders was supported by different patterns and shadings of the opisthosoma (Fig. 3.1 D). Starting with four adult females in a terrarium (27 individuals/m2), competition took place and changes of web occupancy occurred usually overnight. When raising the density to 40 individuals/m2, spiders were more often inhibited to build a web at all. The probability of observing a floater raised from 0.12 floaters/day to 0.58 floaters/day. Not only floaters were observed conquering webs during the lab-experiment. Also females which were already in possession of a web occupied new ones, sometimes even without abandoning their old one (observed three times). All incorporated webs were defended against intruders. Such defensive behaviour could be triggered by e.g. adding prey to a web where a

* Experiment was performed under my supervision by Anne Habedank during her bachelor thesis “Kampf ums Netz – Intraspezifisches Konkurrenzverhalten von Weibchen der Federfußspinne Uloborus plumipes“ in 2014. Her raw data were re-evaluated for this thesis. 34 floater was in the vicinity. When the floater tried to steal the prey, the new owner reacted strongly to the invading spider by emitting vibrations and attacking the intruder. Although prey can be used as a trigger for intraspecific contests, they occurred frequently in the absence of prey. Cannibalism was never observed in the course of experiments and rarely within the lab colony at all. Therefore, the territories and/or webs themselves have to be objectives of the competition. Getting the prey, however, can be used as a reliable indicator for web ownership.

3.1.2. Intraspecific fighting To characterize behavioural patterns during this intraspecific competition as well as the factors influencing the contest, standardized experiments were performed*. An experiment was started by placing two occupied webs adjacent to each other and one spider entering the web of its conspecific. The addition of prey after 20 min triggered activity and was given also in the 25% of all experiments where contests were already decided earlier, to follow one standard protocol.

Tab. 3.1: Behavioural patterns during intraspecific competitions. In standardized competitions different behavio ural patterns could be identified. Although no stereotypical sequence could be determined, some patterns are related to each other e.g. by being a reaction to one another. Furthermore, not all patterns were performed equally often by the later winner/loser, respectively by the intruder/defender. Behaviour Description Occurrence

Threads of the web were grasped and due to a Occurred continuously during intraspecific Thread pulling contraction of the legs, a longitudinal tension interactions, but also as reaction to other (Fig. 3.2 A-C) was caused in the threads. The performance of stimuli (e.g. prey). It was often combined with contraction could vary. abdominal trembling.

The opisthosoma was swinging vertically Occurred continuously during intraspecific (main component) and horizontally, supported Abdominal trembling interactions, but also as reaction to other by leg contractions. This lead to vibrations (Fig. 3.2 D-F) stimuli (e.g. prey). It was often combined with within the web. The performance of the thread pulling. opisthosoma movement could vary. The 1st legs were contracted and with the help of a swinging opisthosoma the spider got into Occurred in almost all intraspecific contests a pendulum movement. The vibration in the (86%) and only during or shortly after these. Web shaking (Fig. 3.3) web provoked by this movement appeared to In most contests (70%) the later winner shook be a lashing of the web, often causing the the web more often than the loser. opponent to loose footing.

* Experiments were performed under my supervision by Anne Habedank during her bachelor thesis “Kampf ums Netz – Intraspezifisches Konkurrenzverhalten von Weibchen der Federfußspinne Uloborus plumipes“ in 2014. Her experimental setup did not include the characterization of the vibrational patterns. Most of her raw data were re-evaluated for this thesis. 35

Behaviour Description Occurrence

Often the first reaction in a contest (68%). Reaching hub 1st Being the first one in the centre of the web. More often performed by the later winner of the contest (68%).

Touching the antagonist with the 1st leg Occurred after attacking, but also Leg contact (duration < 1 s). spontaneously.

Physical interaction employing the 1st legs Fighting (duration ≈ 1 s) and/or the 4th legs comparable Occurred after attacking. to wrapping prey (duration > 1 s). In 76% of all contests attacking was more often performed by the later winner of the web. Attacking Running towards the competitor. About half of the time an attack led to fighting (44%). Otherwise the opponent escaped directly or after leg contact.

Reaction to attacking, leg contact or fighting. In 97% of all contests this behaviour was more often performed by the later loser. Escaping Running away from the competitor. Keep in mind, this behaviour was also used to define the loser of a contest. Hence, this pattern is not independent. Grasping the drag line of the antagonist and Only occurred when the antagonist was Drag line pulling coiling it between one’s legs. This behaviour escaping. could draw the antagonist towards the grasper. During escaping, spiders destroyed the threads Often a reaction to drag line pulling. 91% of Cutting threads between themselves and the contesting spider. all observed thread cuttings were performed This could cause the web to collapse. by web owner.

The spider lost its grip on the threads and Often a reaction to web shaking of the Loosing footing dropped from the web. opponent.

Occurred continuously not only during movement Rubbing the against each other. interactions. Not performed more often by one of the spiders.

Contests had a duration of 25 ± 14 min (n = 44; normal distributed after Kolmogorov-Smirnov, α = 0.05, two-tailed, p = 0.555) before the winning spider could be determined by the experimenter, by either observing a permanent escape of one of the spiders (loser) or by defining the winner as the one consuming the prey. As a first reaction in 68% of all contests, one of the spiders occupied the hub (“reaching hub 1st”). The subsequently following behavioural program showed stereotypical modules of behaviour (Tab. 3.1). Although no stereotypical sequence could be determined, some modules were associated to one another. E.g. “thread pulling” depended on an escaping opponent and “losing footing” was often triggered 36 by “web shaking”. When attacking, the contestants frequently provoked physical contact between the first legs. This contact could lead to a grabbling movement with the first legs or a movement similar to wrapping prey with the fourth ones (both defined as “fighting”). Although all behaviours cause vibration within the web, three behavioural patterns can be mainly associated with producing vibration: “thread pulling”, “abdominal trembling” and “web shaking”. The predominant vibration caused by thread pulling (Fig. 3.2 A-C) was in the longitudinal direction, whereas abdominal trembling (Fig. 3.2 D-F) and web shaking (Fig. 3.3) resulted mainly in transverse movements of the threads. Thread pulling had a much lower frequency (1.1 ± 0.7 Hz, n = 3) than abdominal trembling (9.0 ± 1.2 Hz, n = 3). Both behaviours were also observed in the absence of a competitor, e.g. as reaction to prey.

Fig. 3.2: Thread pulling and abdominal trembling. A) Schematic drawing of the limb movement during thread pulling depicting both extreme positions occurring (most extended and most contracted legs). The red arrows indicate the limb movements. B) Stills of a recording reflecting both positions already described in (A). The dashed line indicates the same level. C) Tracking the tarsus of a first leg visualized the vibrations emitted by a spider in vertical and horizontal direction. Both movements influence the longitudinal tension within the web. D-F) Same data as presented in (A-C), but for abdominal trembling. Beside an unincisive leg contraction, a pronounced movement of the opisthosoma (op) occurred. Note the different time scales in (C) and (F).

37

In contrast, web shaking formed a prominent part in agonistic interactions and was displayed in short bursts (495 ± 176 ms, n = 8) during most of the observed contests (86%). The spider bent and extended its first legs with a frequency of 23.3 ± 2.4 Hz (n = 8), swinging its opisthosoma up (ventral direction) during leg contraction, and down during leg extension (Fig. 3.3 A, B). Due to this behaviour, the spider fell into a pendulum movement. This caused vibrations consisting of a horizontal component (displacement: 2.9 ± 1.1 mm, n = 8) and a vertical component (displacement: 5.0 ± 0.8 mm, n = 8), the latter being the predominant feature of the behaviour (Fig. 3.3 C). Such strong vibrations displaced the tarsus, respectively the web, in the range of the spider’s body dimension (5.08 ± 0.31 mm, n = 10). It appeared to be a lashing of the web, often causing the opponent to loose footing.

Fig. 3.3: Web shaking. A) Schematic drawing of the limb movement during web shaking depicting the start position and both extreme positions (most contracted and most extended le gs). The red arrows indicate the limb movements. B) Stills of a recording reflecting the positions already described in (A). The dashed line indicates the same level. C) Tracking the tarsus of a first leg visualized of the vibrations emitted by the spider in vertical and horizontal direction.

38

Web shaking was significantly more often performed by the later winner of the web (2.3 ± 1.7 vs 1.3 ± 1.8 times/encounter; Z = -3.16, p < 0.01). Likewise, other behavioural patterns were not distributed equally between winning and losing spiders (Fig. 3.4, Tab. 3.1). The winning spider, for example, was the first one reaching the hub in 68% of all contests. Furthermore, the typical contest related behaviour “attacking” could also be used to characterize the winning spider.

Fig. 3.4: Behavioural and morphological parameters associated with winning the contest. Different parameters were measured during standardized contests and their relation to “winning the contest” was calculated. Note that besides being the larger one (measured either by the correlated factors weight, leg length or body size ( Tab. 3.2); marked light grey), also typical behaviours of the winning spider can be found. Error bars indicate the 95% confidence interval.

Finding a different behaviour for winning or losing spiders, the question arose, if such patterns influence the contest and/or if they are used for communicating the determining factors. Being intruder or defender had no significant impact on winning the contest (χ2 = 0.09; p < 0.77). Being the larger spider, measured either by weight, first leg length or body size, seemed to be beneficial, as they were more often winning the contest (Fig. 3.4). To determine the factors influencing the outcome of the contest, a binary logistic regression was performed*. The different occurrence of behavioural patterns (i.e. “web shaking”, “attacking”, “reaching hub 1st” and as a negative control “pedipalp movement”) as well as differences in morphological traits (i.e. length of the first pair of leg, body size and weight) and the experience in winning/losing the previous contest were included in the calculation. The difference in body weight was

* By Jonas Hausen, M.Sc. from the Institute for Environmental Research (Biology V) at the RWTH Aachen. 39 correlated with the difference in first leg length (ρ = 0.75, p < 0.001) and body size (ρ = 0.69, p < 0.001; Tab. 3.2). The difference in first leg length and the difference in body size was correlated as well (ρ = 0.69, p < 0.001). To avoid multicollinearity, only the weight difference was included in the model. All other factors were independent of each other.

Tab. 3.2: Independent behavioural patterns from morphological traits. All parameters which should be used in a binary logistic regression were tested for their correlation by different methods (Polyc: polychoric; Polys: polyserial; Spear: Spearman). Correlations above 0.6 and p < 0.05 are shaded in grey. For correlated parameters, only one parameter (i.e. “weight”) was included in the binary logistic regression model. 1st leg Reaching Web Pedipalp Difference in Weight length Body size hub 1st shaking movement Attacking Experience

Weight Spear Spear Polys Spear Spear Spear Polys

1st leg length 0.75 Spear Polys Spear Spear Spear Polys

Body size 0.69 0.69 Polys Spear Spear Spear Polys

Reaching hub 1st 0.12 0.00 -0.02 Polys Polys Polys Polyc

Web shaking 0.31 0.26 0.16 0.41 Spear Spear Polys Pedipalp movement -0.26 -0.25 -0.23 -0.24 -0.29 Spear Polys

Attacking 0.26 0.20 0.13 -0.00 0.18 0.13 Polys

Experience 0.59 0.50 0.34 0.30 0.31 -0.60 -0.01

In the final model (n = 44, df = 40, p < 0.001), body weight (Estimate = 6.53, Std. Error = 2.05, p = 0.00149), reaching the hub first (Estimate = 2.65, Std. Error = 1.25, p = 0.0342) and attacks (Estimate = 1.07, Std. Error = 0.6, p = 0.0739) predicted the contest outcome. Differences in body weight were the main predictor and explained most of the variance, while “reaching the hub first” and different frequency of attacking were subordinate predictors. Web shaking activity, pedipalp movement as well as a different experience in winning or losing the previous contest did not significantly influence the outcome of the contest. None of the interactions between the explanatory variables was significant.

40

Finding weight difference the main factor influencing the contest outcome, one can conclude heavier, respectively larger spiders are more likely to win contests. Nevertheless, physically inferior spiders should be able increase their chance of winning by reaching the hub first or attacking more often. Within the standardized experiments, physically inferior spiders winning a contest indeed showed a higher tendency for aggressive behaviour, including attacking and web shaking. Such conspicuous behaviour might be a behavioural adaption of the spider. However, to adapt someone’s behaviour, some form of communication between the spiders has to occur. The behaviour emitting the strongest vibrations, i.e. web shaking was characterized mainly through a swinging opisthosoma and a contraction of the elongated first legs. Both factors are characterizing likewise the winning spider. Based on these observations, we proposed that web shaking, in addition to being a direct aggressive action, might also communicate physical properties, e.g. the body weight of the spider that displays the behaviour. Because no correlation between the frequency of web shaking occurrence and the morphological properties of a spider could be detected (Tab. 3.2), amplitude and frequency of the web movements caused by spiders with different body weight were analysed (Fig. 3.5 A, B). There was a correlation between body weight and vertical amplitude, but it was low (Spearman correlation coefficient of 0.45) and not significant*. There was also no correlation between body weight and the frequency of the vibration or between leg length and any web shaking parameter (Fig. 3.5 C, D)*. Direct communication of the spider’s morphological traits by web shaking can therefore be excluded.

* Correlations calculated by Prof. Dr. Jörg Mey. 41

Fig. 3.5: No correlation between web shaking parameters and the size of the spider. A, C) The mean horizontal displacement of the thread during web shaking is plotted against the weight (A) respectively the first leg length (C) of the spider performing the behaviour. B, D) The mean frequency of the pendulum movement of one spider is plotted against the weight (B) respectively the first leg length (D) of this spider.

Characterizing the ecology of U. plumipes respecting e.g. space, mating behaviour or nocturnal activity, the demands for establishing a lab colony could be specified and setups for following behavioural experiments can be customized to the specific needs of the spiders. The observation that female spiders conquer web of conspecifics lead to the necessity to separate spiders, when it is important to observe each individual’s thread construction behaviour. Although our experiments did not provide deeper insight in the vibrational communication of spiders, they nevertheless indicated that vibrations influence the spider’s behaviour. External vibrations had to be excluded for undisturbed observations.

42

3.2. Cribellate thread production

3.2.1. Thread structure of U. plumipes To validate the structural differences of capture threads of cribellate spiders, first the structure had to be revised. U. plumipes’ capture thread consisted of two axial fibres, aligned parallel in one plane with a distance of 22 ± 7 μm (n = 6) to each other (Fig. 3.6 A). This parallel alignment and consequently also the distance was not established directly when extracted from the spinnerets, but several millimetre after the spider’s production site (Fig. 3.6 B).

Fig. 3.6: The alignment of axial fibres in a capture thread of U. plumipes. A) Capture thread showing the typical structure with puffs and intermediate zones (iz). Furthermore, both axial fibres (ax), which are enclosed by cribellate fibres (cf), can be observed. SEM-image, sample not coated. B) Ventral overview over the spinnerets of U. plumipes with emerging axial fibres (ax) visible. The axial fibre’s distance to each other is larger near by the spinnerets and they merge when getting more distant to the spider.

Around both axial fibres a sheet-like mat of cribellate fibres was curled, building a hollow structure which wrapped the axial fibres completely (Fig. 3.7 B). Taking a thread sample from a finished web, the seam of this mat was helical twisted, suggesting the spider somehow twists the thread during production#. The mat itself was organized in puffs and intermediate zones, with 12 ± 3 of these structure combinations per millimetre (n = 7) and a puff length of 69 ± 1 μm + (n1 = 3) . The puffs could not only be distinguished from the intermediate zones by an increased thread’s diameter (Fig. 3.6 A, Fig. 3.7 A-C), they also showed an irregular alignment of cribellate fibres within one puff (Fig. 3.7 C, D). In the intermediate zones, these fibres’ alignment was almost parallel to the thread itself (Fig. 3.7 C, E). Fibre density within a puff was non-uniform with more strongly convoluted cribellate fibres in-line on one side of the puff.

 Parts of this chapter have been included in the paper “Joel, A.-C., Kappel, P., Adamova, H., Baumgartner, W., Scholz, I., 2015. Cribellate thread production in spiders: Complex processing of nano-fibres into a functional capture thread. Structure & Development 44, 568-573“. Submission was part of this thesis. # Observed by Hana Adamova during her bachelor thesis “Der Einfluss vom Calamistrum auf die Struktur und die Funktion cribellater Fangfäden” in 2015. Thesis was supervised by me. + Data collected under my supervision by Peter Kappel during his bachelor thesis “Analyse des Spinnprozesses cribellater Fasern der Federfußspinne (Uloborus plumipes)” in 2014. 43

Due to this structuring, it was possible to determine the moving direction of the spider during thread production, because this line was always at the “posterior” end of the puff* (Fig. 3.7 C).

Fig. 3.7: Cribellate capture thread of U. plumipes. A) Overview over capture thread. cf: cribellate fibres, iz: intermediate zone. B) Capture thread broke open at its seam, revealing that cribellate fibres (cf) are organized as a mat, wrapping3 the axial fibres (ax). iz: intermediate zone. C) Coating with gold enabled one to determine cribellate fibre (cf) alignment: irregular within a puff and almost parallel to the thread itself in the intermediate zones (iz). Note the non-uniform cribellate fibre density within the puff. The spider was moving from left to right during capture thread production. D) Close-up of irregular cribellate fibre alignment in a puff. E) Close-up of more regular cribellate fibre alignment in an intermediate zone. SEM-images, A and B: samples coated with carbon, C to E: samples coated with gold.

Note that the shape of the thread changed after coating it with gold: An untreated thread showed a puffy structure looking like beads on a string with a maximal diameter of about 170 μm (Fig. 3.6 A). Gold coating led to a collapse and a more tooth shaped structure of the puffs (Fig. 3.7 C). Here, cribellate fibre alignment was determinable. However, the gold coating led to a covering of the knots on the cribellate fibres and enhanced the diameter from 20 ± 5 nm, respectively 30 ± 6 nm for the knots (n = 4), to about 100 nm (n = 7). In contrast, coating a thread with carbon, the thread’s structure resembled the one in the native state (Fig. 3.7 A). Opening the mat at its seam, one could see linkages between the cribellate mat and the axial fibres in the intermediate zones (Fig. 3.8 A, B). The axial fibres were covered over their entire length with nanofibres (Fig. 3.8 C). Although such opening of the cribellate mat should open the view on the fibres forming the paracribellate substructure, no structural deviant fibres were observable.

* Observed by Peter Kappel during his bachelor thesis “Analyse des Spinnprozesses cribellater Fasern der Federfußspinne (Uloborus plumipes)” in 2014. Thesis was supervised by me.

44

Fig. 3.8: Linkages between cribellate fibres and axial fibres in the intermediate zones. A) Mat of cribellate fibres (cf) opened at its seam. Linkages between cribellate mat and axial fibre (ax) are visible (arrows). B) Close up of a linkage (arrow) between cribellate fibres (cf) and axial fibre (ax). C) Axial fibres (ax) surrounded by nanofibres (nf).* SEM images, samples coated with carbon.

The final structure of the capture thread developed with each shedding of U. plumipes (Fig. 3.9, Tab. 3.3). Spiderlings matured 23 ± 3 days (n = 9) in the cocoon before hatching. They shed 5.7 ± 0.4 times (n = 9) (excluding the one inside the cocoon (Eberhard 1977; Opell 1979)) during a period of 131 ± 19 days (n = 14) before reaching sexual maturity. Although the 2nd instar was already able to build webs, capture threads were missing there (Fig. 3.9 A). Nevertheless, some nanofibres spun between the radial frames bore knots (Fig. 3.9 B). These structured fibres did not differ significantly in any aspect from the cribellate fibres observed in adult U. plumipes (fibre diameter: 30 ± 1 nm, p = 0.05; knot diameter: 42 ± 4 nm, p = 0.07; distance between two knots: 179 ± 37 nm, p = 0.34; n1 = 2; data for adult spiders are given in Tab. 3.3).

Fig. 3.9: Capture thread during the development of U. plumipes. A) Single nanofibers are observable in 2 nd instar spiders. No capture thread formation visible. B) Nanofibres of (A) can be divided into smooth fibres and fibres bearing knots. C) Capture thread of a 3 rd instar showed already the cribellate fibres (cf) arrangement in puff and intermediate zone (iz). A, C: SEM, samples coated with gold. B: TEM, sample not treated .

* Picture taken by Dr. Alexander Schwedt and Ruth Harscheidt from the central facility for electron microscopy of the RWTH Aachen. 45

In general, the number of puffs per millimetre decreased linearly with each moult (f(x) = -3.1395 puffs/mm * x + 36.953 puffs/mm; r2 = 0.8562; x = {3, 4, 5, 6, 7, 8}; with “f” as function of the spatial frequency and “x” the developmental stage), whereas the diameter of the cribellate thread increased exponentially (g(x) = 9.8414 µm * e0.338x ; r2 = 0.954; x = {3, 4, 5, 6, 7, 8}; with “g” the function of the diameter and “x” the developmental stage). Although the shape of the capture thread assimilated with each moult to the one of the adult spider (Tab. 3.3), the structure of the cribellate fibres themselves stayed roughly the same.

Tab. 3.3: Thread development of U. plumipes. Native threads of spiders with different developmental status were observed with a SEM, respectively for the cribellate fibres a TEM. Axial fibre distance, the maximal thread diameter (puff) as well as the diameter of cribellate fibres and its knots as well as the distance between knots were measured and compared to the data for adult spiders, using the two tailed T -test for statistical analysis. If p < 0.05, significance was assumed (grey shaded). Data are presented as mean ± SD. “n” indicates the number of individuals, for each one the mean was calculated individually. n1 means data were available for only one spider. Thread of 3rd instar 4th instar 5th instar 7th instar adult (8th)

30 ± 11 24 ± 7 17 ± 2 17 ± 0 12 ± 3 Spatial frequency [puffs/mm] n = 2 n = 3 n = 2 n = 2 n = 7 26 ± 7 38 ± 12 62 ± 10 83 ± 1 167 ± 14 Diameter of cribellate thread [µm] n = 2 n = 2 n = 2 n = 2 n = 6 28 ± 3 25 ± 1 31 ± 10 27 ± 6 20 ± 5 Cribellate fibre diameter [nm] n1 = 7 n1 = 5 n1 = 3 n1 = 4 n = 4

38 ± 4 35 ± 2 40 ± 12 38 ± 6 30 ± 6 Cribellate fibre knot diameter [nm] n1 = 7 n1 = 5 n1 = 3 n1 = 4 n = 4 99 ± 23 127 ± 21 174 ± 64 153 ± 8 147 ± 34 Cribellate fibre knot distance [nm] n1 = 7 n1 = 5 n1 = 3 n1 = 4 n = 4

46

3.2.2. Thread structure of other cribellate spiders* The examination of the Uloboridae Z. geniculata revealed similarities as well as differences regarding its thread structure (Fig. 3.10, Tab. 3.4). The thread of Z. geniculata showed a puffy structure with intermediate zones, too, but only seven of these structure combinations per millimetre were found. The diameter of the puffs was significantly lower than observed for threads of U. plumipes. Nevertheless, the principle structure was conserved: Two axial fibres, aligned during thread production, were surrounded by a mat of cribellate fibres and linked to them in the intermediate zones (Fig. 3.10 B). Likewise, the thread collapsed after coating with gold, not after coating with carbon.

Fig. 3.10: Structure conserved in the capture threads of Z. geniculata. A) Uncoated capture thread showing the characteristic puffy structure. ax : axial fibres, cf: cribellate fibres, iz: intermediate zone. B) Capture thread coated with gold. The structure has collapsed, but cribellate fibre (cf) alignment is determinable. They are more highly curled in the puffs than in the intermediate zones (iz). Furthermore, a seam can be detected, indicating the fibres are arranged in a mat likewise. SEM images.

Tab. 3.4: Differences in thread structure of two Uloboridae. Native threads of two different species were observed with a SEM. Axial fibre distance as well as the maximal thread diameter (puff) were measured and compared, using a two -tailed T-test for statistical analysis. If p < 0.05, significance is assumed (grey shaded). Data are presented as mean ± SD. “n” indicates the number of individuals, for each one the mean was calculated individually. Thread of U. plumipes Z. geniculata p-value

12 ± 3 7 ± 1 Spatial frequency [puffs/mm] p = 0.02 n = 7 n = 4 22 ± 7 17 ± 5 Distance of axial fibres [µm] p = 0.31 n = 6 n = 3

167 ± 14 140 ± 14 Diameter of cribellate thread [µm] p = 0.03 n = 6 n = 3

* Parts of this chapter have been included in the paper “Joel, A.-C., Scholz, I., Orth, L., Kappel, P., Baumgartner, W., 2016. Morphological adaptation of the calamistrum to the cribellate spinning process in Deinopoidae (Uloboridae, Deinopidae). Royal Society Open Science 3“. Submission was part of this thesis. 47

The thread’s assembly is not restricted to Uloboridae: In recordings of the Deinopidae Deinopis sp. the same alignment of axial fibres could be observed. Furthermore, coating threads of the Desidae Badumna longinqua and the Filistatidae Kukulcania hibernalis showed the same reaction as described for Uloboridae: a collapsing after coating with gold (Fig. 3.11). B. longinqua’s thread additionally did not collapse after coating it with carbon. This has not been tested for K. hibernalis.

Fig. 3.11: Collapsing threads after coating with gold. A, B) Threads of B. longinqua. In (A) the native thread is shown while in (B) the same spot is shown after coating the thread with gold. The cribellate fibres (cf) have collapsed. Furthermore, the paracribellate fibres (pcf) are not visible anymore. C, D) Same as described for (A, B) but for a thread of K. hibernalis.

3.2.3. Cribellum and paracribellum All three involved silk types emerge from different spigots, each situated on an individual spinneret (Fig. 3.12 A). By fixing U. plumipes during capture thread production, one could replicate the observation that both axial fibres are extruded from the spigots of the glandulae pseudoflagelliformes on the posterior spinnerets (Fig. 3.12 B). The paracribellum on the median spinnerets was covered with fibres. However, fibres actually emerging from the spigots were not visible (Fig. 3.12 C). The same shock freezing experiments though confirmed fibres being extruded from the spigots of the cribellum. In all 25 experiments a fluid film of silk covering the cribellum could not be replicated (Fig. 3.12 D).

 Parts of this chapter have been included in the paper “Joel, A.-C., Kappel, P., Adamova, H., Baumgartner, W., Scholz, I., 2015. Cribellate thread production in spiders: Complex processing of nano-fibres into a functional capture thread. Arthropod Structure & Development 44, 568-573“. Submission as part of this thesis. 48

Fig. 3.12: Spinnerets during capture thread production. A) Overview of the spinnerets of U. plumipes during cribellate thread production including an emerging capture thread (ct). as: anterior spinnerets, cr: cribellum, ms: median spinnerets, ps: posterior spinnerets B) Close-up of a posterior spinneret showing a spigot of glandulae pseudoflagelliformes (sgp) and an emerging axial fibre (af). C) Overview over the median spinnerets (ms), showing their elongated spigots of the paracribellum (spc) covered in thin fibres. D) Close-up of several spigots of the cribellum (cr) including emerging cribellate fibrils (cf). SEM images, coated with gold.

The spigots involved in capture thread synthesis were not fully developed in freshly hatched U. plumipes. In the 2nd instar neither cribellum nor paracribellum was detectable (n = 3) (Fig. 3.13). There were two smooth patches were the cribellum should be situated. The 3rd instar already had first spigots of the paracribellum situated on the median spinnerets (n = 3). Furthermore, one single spinning plate forming the cribellum could be observed, although bearing much less spigots than the adult one (Tab. 3.5). The size of the cribellum and therefore also the number of spigots increased exponentially with each moult (h(x) = 17.527 * e0.7141x; r2 = 0.997; x = {3, 4, 5, 6, 7, 8}; with “h” staying for the number of spigots and “x” for developmental stage).

Fig. 3.13: No cribellum or paracribellum developed in 2 nd instar of U. plumipes. A) Overview over two flat patches where the cribellum (cr) should be situated. B) Median spinnerets (ms) without paracribellum. SEM images, sputtered with gold.

49

Tab. 3.5: Development of the cribellum of U. plumipes. Gold coated specimens of different developmental status were examined using the SEM. Length, width and spigot number was surveyed. With these data, the mean spigot density of the cribellum could be determined. Data are presented as mean ± SD. “n” indicates the number of individuals. SD is only given, if n = 1.

Cribellum of 3rd instar 4th instar 5th instar Adult

73 ± 7 115 ± 1 175 393 Length [µm] n = 3 n = 3 n = 1 n = 1 15 ± 3 21 ± 2 33 79 Width [µm] n = 3 n = 3 n = 1 n = 1 148 ± 26 280 ± 16 708 5126 Number of spigots [-] n = 3 n = 3 n = 1 n = 1

0.14 ± 0.04 0.12 ± 0.02 0.12 0.16 Spigot’s density [1/µm2] n = 3 n = 3 n = 1 n = 1

3.2.4. Analysis of the limb movements of U. plumipes

Fig. 3.14: Combing position of U. plumipes. A) Ventral overview of U. plumipes during capture thread production. The fourth legs are classified into supporting leg (sl) and combing leg (cl) with the calamistrum of the combing leg used for cribellate thread production. The arrow indicates the moving di rection of the spider during thread production. sp: spinnerets. B) Caudal close up of the tarsus of combing leg (cl) placed near the joint between tarsus (ta) and metatarsus (mt) of the supporting leg (sl). C) Caudal close up, showing the parallel orientation of the combing leg (cl) respectively its metatarsus (mt) to the ventral site of the opisthosoma. Anterior and posterior spinnerets (as, ps) are visible. D) Vertical overview over the opisthosoma, showing the parallel alignment of the supporting leg (sl) to the ventral site of the opisthosoma. For easier detection, dotted lines indicate the supporting leg. sp: spinnerets

 Parts of this chapter have been included in the paper “Joel, A.-C., Kappel, P., Adamova, H., Baumgartner, W., Scholz, I., 2015. Cribellate thread production in spiders: Complex processing of nano-fibres into a functional capture thread. Arthropod Structure & Development 44, 568-573“. Submission was part of this thesis. 50

During the extraction of the fibres, the spinnerets as well as both fourth legs perform a continuous oscillating movement with a frequency of 10.36 ± 0.07 Hz (n = 2)*. The combing leg positioned its tarsus tip near the joint between tarsus and metatarsus of the supporting leg, forming a rigid unit sweeping over the spinnerets (Fig. 3.14 A, B; peak angle difference between tarsus of the combing leg and metatarsus of the supporting one: 0 ± 1°; n = 2*). The posterior movement of the combing leg could be characterized by a stretching of the joint between femur and patella, respectively tibia (the patella could not be identified in the recordings), showing an increasing angle of 45 ± 13° (n = 3)*. Furthermore, the joint between tibia and metatarsus was contracted by 23 ± 9° (n = 3)*. The joint between metatarsus and tarsus was not involved in this movement (0 ± 0°; n = 2)*. In contrast to the combing leg, which was aligned almost parallel to the ventral site of the opisthosoma (Fig. 3.14 A, C), the femur of the supporting leg was lowered, leading to a leg- orientation almost parallel to the lateral site of the opisthosoma (Fig. 3.14 A, B, D). Although optical distortion did not allow to measure the correct angle between the members of the supporting leg, the approximate principle resembled the one of the combing leg, showing a stretching of the joint between femur and patella and a contraction of the joint between tibia and metatarsus during the posterior movement*. U. plumipes moved the metatarsus of its combing leg in an oval manner over its spinnerets, lowering the metatarsus of the combing leg during its posterior movement. One such stroke * produced one puffy structure (n1 = 5) . In this way, it produced 0.63 ± 0.03 mm thread per second (n = 2). This fast production of cribellate thread requires a highly synchronized movement of both fourth legs as well as all three involved spinnerets: posterior spinnerets, median spinnerets and the cribellum (Fig. 3.15).

* Data collected under my supervision by Peter Kappel during his bachelor thesis “Analyse des Spinnprozesses cribellater Fasern der Federfußspinne (Uloborus plumipes)” in 2014. 51

Fig. 3.15: Sequence of limb movement during capture thread production. The directional change of a movement is indicated by arrows. Note the encircled spigots of the paracribellum (spc), visible as a pale glint overlapping the median spinnerets (ms) in the pictures 83 ms to 93 ms. a: anus, as: anterior spinnerets, cl: combing leg, cr: cribellum, ps: posterior spinnerets

Starting anterior of the spinnerets (“anterior position”, Fig. 3.15: 0 ms), the combing leg was * moved posterior with a velocity of 39 ± 4 mm/s (n1 = 13) , whereas the posterior spinnerets were still retracting from the previous cycle. The metatarsus of the combing leg lowered itself and moved over the cribellum (“median position”), where the movement was slowed down to * 10 ± 1 mm/s (n1 = 16) . Moving further posterior, the metatarsus of the combing leg reached the retracting posterior spinnerets after 15 ± 1 ms (n1 = 3). The posterior spinnerets stopped their retraction and moved synchronously backwards with the combing leg. At the same time the cribellum started moving in the opposite direction. While combing leg and posterior spinnerets moved back, the previously hidden median spinnerets became visible, not moving at this time. Shortly before the metatarsus of the combing leg reached the “posterior position”, * 727 ± 64 µm (n1 = 16) posterior of the median position, the median spinnerets started spreading apart (74 ± 0 ms, n1 = 4). During this movement, the long spigots of the paracribellum even overlapped the anterior spinnerets (88 ± 1 ms, n1= 3), covering an area as wide as the width of the cribellum. As soon as the combing leg reached the posterior position, all limbs except the median spinnerets reversed their movement, returning to their starting position

(83 ± 0 ms, n1 = 4). Note the depth blurriness of the combing leg during its anterior movement, indicating the lifting of this leg (88 and 93 ms). Observing the spider from the lateral perspective, this oval movement was even more obvious. The leg’s velocity again increased to * 56 ± 6 mm/s (n1 = 19) , making the anterior movement 3.7 ± 1.2 times faster than the posterior movement (n = 3). Before the metatarsus of the combing leg reached the anterior position, the median spinnerets stopped spreading and closed (93 ± 1 ms, n1 = 4). When interrupting the spinning process (for example to attach the thread to the radius or to react to external

* Data collected under my supervision by Peter Kappel during his bachelor thesis “Analyse des Spinnprozesses cribellater Fasern der Federfußspinne (Uloborus plumipes)” in 2014. 52 disturbance), no return to the starting point occurred and the spider stopped the cycle at the posterior position. The key data extractable from this detailed description are presented in Tab. 3.6.

Tab. 3.6: Characterizing the cribellate capture thread production of U. plumipes. Data are collected of several video sequences, each sequence including several combing cycles. “n” is the number of different specimens used. AP: anterior position, mt: metatarsus, PP: posterior position, ps: posterior spinnerets, t: tarsus. Family Uloboridae Species Uloborus plumipes

10.4 ± 0.1 Frequency [Hz] n = 2

Combing position after Eberhard (1988) Type 2

Touching point Tip of t near joint mt/t

Relation of velocity of posterior leg movement 3.7 ± 1.2 × to anterior leg movement [times slower] (n = 3)

Start process AP

Stop process PP

Lowering of metatarsus Posterior leg movement

Abduction ps Posterior leg movement

Abduction ms Anterior leg movement

3.2.5. Limb movement during the cribellate spinning process of other spiders For a generalization of the cribellate spinning process, the observations made for U. plumipes need to be compared to the process performed by other cribellate spiders. Therefore, the spinning process of seven species from four different superfamilies (respectively six families) were analysed (Tab. 3.7), extending from a spider of the same family like U. plumipes, Z. geniculata (Uloboridae; Fig. 3.16), to distantly related spiders like K. hibernalis (Filistatidae; Fig. 3.19) with very different web building behaviour. For this reason, differences and/or similarities observed in cribellate thread production should reflect differences or similarities within the web-building behaviour, thread shape, or reveal family specific patterns respectively conserved features within the cribellate thread production.

 Parts of this chapter have been included in the paper “Joel, A.-C., Scholz, I., Orth, L., Kappel, P., Baumgartner, W., 2016. Morphological adaptation of the calamistrum to the cribellate spinning process in Deinopoidae (Uloboridae, Deinopidae). Royal Society Open Science 3“. Submission was part of this thesis. 53

Fig. 3.16: Zosis geniculata. A) Overview over an adult female in its resting positon during the day. B) Ventral overview over a spider during capture thread production. Both fourth legs establish a unit with one being the combing leg (cl) and the other the supporting leg (sl). The calamistrum of the combing leg is placed over the spinnerets (sp; encircled) of the spider. op: opisthosoma. C) Caudal close-up of the parallel orientation of combing leg (cl), respectively its metatarsus to the ventral site of the opisthosoma. Anterior and posterior spinnerets (as, ps) are visible. Note the light reflection on the metatarsus indicating the position of the calamistrum. op: opisthosoma, sl: supporting leg.

Fig. 3.17: Deinopis sp. A) Spider during capture thread production. B) Close up of the spinnerets in the posterior position during capture thread production. The tarsus of the combing leg (cl) is resting loose on the tarsus (ta) of the supporting leg (sl). The posterior spinnerets (ps) as well as the anterior spinnerets (as) are abducted. The median spinnerets (ms; encircled in grey) are adducted. op: opisthosoma. C) Same scenario as in (B), although now the combing leg is already retracting to the anterior position. Within the encircled area, the spreading of the median spinnerets can be observed. All pictures are extracted from a youtube -video. The scale bar is therefore estimated based on the mean body size of D. subrufa.

54

Fig. 3.18: Cribellate spiders during capture thread production. A, B) Eresus walckenaeri from the dorsal (A) and caudal (B) perspective. ct: cribellate thread, cl: combing leg, sl: supporting leg. C, D) Stegodyphus mirandus from the ventral (C) and lateral (D) perspective. ct: cribellate thread, cl: combing leg, sl: supporting leg. E) Psechrus sp. from the ventral perspective. cl: combing leg, sl: supporting leg. F) Zoropsis spinimana from the ventral perspective. This perspective is enabled by filming the spider through a piece of glas s, on which the spider is walking. ct: cribellate thread, cl: combing leg, sl: supporting leg. All pictures are extracted from youtube-videos. Scale bars are not included, because data of the mean body size of these spiders are missing.

For starting the combing process, the observed species positioned their fourth legs anterior of the spinnerets by placing the metatarsus of the combing leg parallel to the ventral side of their opisthosoma (Fig. 3.16 C; Fig. 3.17 B, C; Fig. 3.19; Fig. 3.18 B, D, F; Tab. 3.7). To establish a unit with its fourth legs, e.g. Z. geniculata positioned the tip of the tarsus of its combing leg near the joint between metatarsus and tarsus of the supporting leg (Fig. 3.16 B, C). Although using also both fourth legs, Deinopis sp. placed the tarsus of its combing leg loose near the joint between tarsus and metatarsus of its supporting leg (Fig. 3.17 B, C). In the other observed species (except for K. hibernalis which used also the third leg as supporting leg; Fig. 3.19), both fourth legs formed likewise a unit by placing their tarsus (or also the metatarsus for 55

Psechrus sp.; Fig. 3.18 E) in the region of tarsus or metatarsus of the supporting leg (except for K. hibernalis which could also use the tibia; Fig. 3.19 B, C). With a frequency ranging between 4 Hz (Z. spinimana) and 12 Hz (K. hibernalis), the combing leg was moved in an oval manner over the spinnerets, lowering the leg during its posterior movement. The previously adducted posterior spinnerets abducted simultaneously to this movement of the combing leg in Deinopoidae and Lycosoidae. After reaching the posterior position, the leg reversed its movement and enhanced its velocity about 3 times. While posterior spinnerets started adducting, the median spinnerets spread apart in Deinopoidae (Fig. 3.17 B, C). Thread production always stopped with the combing leg in the posterior positon. This way, longer pieces of capture threads were produced previous to attachment. One single oval combing cycle produced one puffy structure in Z. geniculata (n1 = 2).

Fig. 3.19: Kukulcania hibernalis. A) Spider during capture thread production. Tarsus (ta) of the combing leg (cl) is placed loose near the joint between tibia and metatarsus of the supporting leg (sl). ct : capture thread, mt: metatarsus, op: opisthosoma. B) Ventral overview, showing another resting position of the metatarsus (mt) of the combing leg (cl) on the tibia (ti) of the supporting leg (sl). Furthermore , you can see the shifted positions of posterior spinneret (ps). op: opisthosoma, ta: tarsus C) Caudal perspective, showing an emerging capture thread (ct) as well as the shifted positions of the posterior spinnerets (ps). cl: combing leg, mt: metatarsus, op: opisthosoma. All pictures are extracted from youtube-videos. A scale bar is not included, because data of the mean body size of these spiders are missing.

Although the characterization of the spinning process of these species is not as detailed as for U. plumipes, a comparison revealed that the main features characterizing the thread production are the same within the examined cribellate spiders (compare Tab. 3.6 and Tab. 3.7). It can be assumed that the basic principle of the cribellate spinning process is conserved. Only minor difference like a variable frequency were determined. Furthermore, the posterior spinnerets of K. hibernalis showed an aberrant movement. They moved independently of the leg movement (12 Hz) with a frequency of 1 Hz. Their movement was also asynchronous and time shifted against each other. This aberrant movement has to have an impact on the structure of the capture thread, if in general the position of the posterior spinnerets is important.

56

Tab. 3.7: Characterization of the cribellate capture thread production in different species. Data are collected of one or several sequences, each sequence including several combing cycles. AP: anterior position, mt: metatarsus, n 1: the same spider was visible in “n” available recordin gs, n.d.: not determinable with the available recordings, PP: posterior position, ps: posterior spinnerets, t: tarsus, ti: tibia. Superfamily Deinopoidae Eresoidae Filistatoidae Lycosoidae

Family Deinopidae Uloboridae Eresidae Filistatidae Psechridae Zoropsidae

Zosis Eresus Stegodyphus Kukulcania Zoropsis Species Deinopis sp. Psechrus sp. geniculata walckenaeri mirandus hibernalis spinimana

Fig. Fig. 3.17 Fig. 3.16 Fig. 3.18 A, B Fig. 3.18 C, D Fig. 3.19 Fig. 3.18 E Fig. 3.18 F

Frequency 8.4 ± 1.1 8.7 ± 0.4 4.6 ± 0.1 8.2 12.0 ± 0.1 4.4 ± 0.3 4.0 ± 0.1

[Hz] n = 2 n = 3 n1 = 2 n1 = 1 n = 2 n1 = 2 n1 = 4

Combing Type 2 Type 2 Type 2 Type 2 Type 1 Type 2 Type 2 position

Touching t loose near Tip t near t loose near t loose at mt t or mt loose t loose near n.d. point joint mt/t joint mt/t joint mt/t* or ti at t or mt joint mt/t

Relation of velocity of posterior leg 3.0 ± 0.3 × 3.1 ± 0.9 × 3.4 ± 1.4 × 1.8 × 1.5 × 3.2 ± 0.2 × 2.9 ± 0.0 × movement to n = 2 n = 3 n1 = 2 n1 = 1 n1 = 1 n1 = 2 n1 = 2 anterior leg movement [times slower]

Start process AP AP AP AP AP AP AP

Stop process PP PP n.d. PP PP PP PP

Lowering of Posterior leg Posterior leg Posterior leg Posterior leg Posterior leg Posterior leg n.d. metatarsus movement movement movement* movement* movement* movement*

Posterior leg Posterior leg 1.14 ± 0.02 Posterior leg Posterior leg Abduction ps n.d. n.d. movement movement n = 2# movement movement

Anterior leg Anterior leg Abduction ms n.d. n.d. n.d. n.d. n.d. movement movement

: after Eberhard (1988) *: data presented here were extracted from an online available recording with low quality and therefore should be handled with care #: see text for further information

57

3.3. The function of the calamistrum

3.3.1. Morphology Proving one cycle of leg movement is producing one puff and that the calamistrum comes probably in contact with fibres during the posterior movement, does not enable one to characterize the function of the calamistrum during the capture thread production. So what happens exactly, when the calamistrum comes in contact with the fibres? Prior to such investigations, the calamistrum morphology needs some re-evaluation and specification.

Fig. 3.20: Morphology of the calamistra of Uloboridae. A to C) Overview over the calamistra (ca) on the metatarsus (mt) of the right fourth leg of U. plumipes (A), Z. geniculata (B) and Miagrammopes sp. (C). The arrow point towards the continuous area on the calamistrum observed in all three species. Side visible is facing the spinnerets during capture thread production. ta: tarsus. D to F) Detail of the calamistrum, showing the setae emerging the cuticle. Between the setae of U. plumipes (D), Z. geniculata (E) and Miagrammopes sp. (F) large gaps are observable. G to I) Detail of the calamistrum showing the tip of the setae orientated parallel to the roots of the setae. The tips of the setae of U. plumipes (G), Z. geniculata (H) and Miagrammopes sp. (I) are forming a continous area though overlapping setae. SEM images, samples sputtered with gold.

The calamistra of the Uloboridae U. plumipes, Miagrammopes sp. and Z. geniculata showed similarities regarding their morphology (Fig. 3.20). Single setae were easily distinguishable at the root, forming one row with gaps between single setae (Fig. 3.20 A-F). Tracing a setae to its tip, the setae bent approximately 90° while enlarging and flattening its breadth (Fig. 3.20 G-I).

 Parts of this chapter have been included in the paper “Joel, A.-C., Scholz, I., Orth, L., Kappel, P., Baumgartner, W., 2016. Morphological adaptation of the calamistrum to the cribellate spinning process in Deinopoidae (Uloboridae, Deinopidae). Royal Society Open Science 3“. Submission was part of this thesis. 58

This led to single setae overlapping the adjacent ones. There were no more gaps visible between two setae, thereby mimicking a continuous and surface-like area parallel to the roots of the setae within the calamistrum. The setae were completely covered with grooves, teeth were not visible in either species. The calamistrum was not established as soon as the 2nd instar of U. plumipes. Only a concave impression without any setae was visible (Fig. 3.21 A). As soon as the spinnerets developed in the 3rd instar (compare to chapter 3.2.3.) a calamistrum was formed, not differing structurally from the adult one, but bearing less setae (Fig. 3.21 B). With each moult, the number of setae increased, thus enlarging the calamistrum stepwise (Fig. 3.21 C).

Fig. 3.21: Development of the calamistrum. Overview over the calamistrum (ca) on the metatarsus (mt) of the right fourth leg of the 2 nd instar (A), 3rd instar (B) and 4 th instar (C) of U. plumipes. Side visible would be facing the spinnerets during capture thread production. Note that the arrow indicating the calamistrum in th e 3rd and 4th instar is pointing towards a setae free zone in the 2 nd instar. ta: tarsus. SEM images, samples coated with gold.

Characterizing one single setae of the calamistrum, it was a hollow structure (Fig. 3.22 C-D). No special inner structures were observable. Ripples, ridges and extensions on the outer cuticle thickened it (Fig. 3.22). The socket was flexible in only one direction (Fig. 3.23). As blocking any movement by conglutinating the setae still enabled a structural and functional normal thread, any potential flexibility of the setae or its base was not necessary for the production of a thread (for further information see chapter 3.3.2.).

59

Fig. 3.22: Structures on the calamistrum of U. plumipes. A) Overview of the surface features from the tip (bottom) to the continuous area (top) of a setae of the calamistrum. Three different structural components of different size are visible: ripples, ridges and extensions (small to large scale). * B) Close-up of the ripples of the calamistrum. TEM-image. C) Cutting away the continuous area, the setae reveals itself as a hollow structure. The extensions lay in parallel orientation to the continuous area. * D) Several slices of a setae from its tip to its continuous area. One can detect the thickening of the cuticle by different surface features. * FIB images (all but (B)).

Fig. 3.23: The socket. A) Section through the calamistrum near its socket showing a movement restriction on the left side. Picture taken from a FIB-Tomography-3D-Model of the calamistrum. # B) Section through a root performed with a FIB.* C) Same section as in (A), but after epon embedding and light microscopic staining.

* Picture taken in cooperation with Dr. Daesung Park and Martina Schiffers from the central facility for electron microscopy of the RWTH Aachen. # FIB tomography performed by Dr. Wolf-Alexander Heiß from the Forschungsinstitut Edelmetalle + Metallchemie in Schwäbisch Gmünd. 60

3.3.2. Contact between calamistrum and cribellate fibres* To characterize the function of the calamistrum during the spinning process, the contact area between calamistrum and cribellate thread has to be defined first. Therefore, one has to determine, whether cribellate fibres are actually snapped between the setae of the calamistrum as suggested by Peters (1984) or whether the fibres are passing across the calamistrum. If fibres are picked up and trapped between the setae, it is likely that the angle (α) in which the metatarsus brushes over the cribellum changes. This would facilitate fibres slipping through the gaps at the tip of the bended setae first and afterwards closing the gap to prevent emerging fibres during the combing process (Opell 2013).

Fig. 3.24: Evaluating the jamming-hypothesis. A) The angle α measured in different positions during the combing process (x -axis). The light grey lines are representing the mean combing movements of three single spiders. The black line represents the mean angle difference of all three. Error bars are the standard deviation betw een the individuals. B) Same as in (A) but for four spiders of Z. geniculata.

In U. plumipes the angle α was 8 ± 1° (n = 3) at the anterior position and changed to 10 ± 1° (n = 3) during the movement between median and posterior position (Fig. 3.24 A). A more detailed analysis of leg orientation showed that α could reach a maximum of over 12°. An angle difference of 4° might already indicate a closing gap to inhibit fibres emerging. However, the angle difference is still very small and the trend within the graphs of the individual spiders are not consistent (Fig. 3.24 A). Observing Z. geniculata the observations likewise points towards no jamming of fibres between the setae during the cribellate spinning process#. The angle α was

* Parts of this chapter have been included in the paper “Joel, A.-C., Scholz, I., Orth, L., Kappel, P., Baumgartner, W., 2016. Morphological adaptation of the calamistrum to the cribellate spinning process in Deinopoidae (Uloboridae, Deinopidae). Royal Society Open Science 3“. Submission was part of this thesis. # The raw data for Z. geniculata were collected by Linda Orth during her bachelor thesis “Die Produktion cribellater Fasern von Zosis geniculatus” in 2015. 61 nearly constant during the whole process and much lower than observed for U. plumipes (4 ± 1°, n = 4) (Fig. 3.24 B). Because the calamistrum morphology is identical in both species, such difference in the angle α cannot be explained by a morphological peculiarity. Indeed, the observation could be corroborated when fixing U. plumipes by shock-freezing them during capture thread production. No fibres or residuals of fibres were found between the setae, but a mat of fibres covering the continuous area of the calamistrum was often observable (Fig. 3.25 A). These result indicate fibres are passing across the calamistrum without passing between the setae.

Fig. 3.25: Cribellate fibres do not pass between the setae of the calamistrum. A) Shock freezing experiments revealed a covering of the setae near their tip with a mat of nanofibres. No fibres or residuals of fibres were found between the setae. B) Conglutinated calamistrum, inhibiting a passing of fibres between the setae. C) Cribellate thread of the spider shown in (B). ax: axial fibre, cf: cribellate fibres, iz: intermediate zone. SEM images, A and B: coated with gold, C: no treatment.

To substantiate these findings, the setae of the calamistrum were conglutinated and that way a passing through of fibres was inhibited (Fig. 3.25 B). Such modified spiders were still able to produce normal cribellate threads (Fig. 3.25 C) with no significant difference in axial fibre distance, cribellate fibre dimensions or capture thread diameter (Tab. 3.8). Likewise, prey was still captured. Therefore, no jamming of cribellate fibres can occur, but fibres have to pass across the calamistrum.

62

Tab. 3.8: No difference in thread structure due to conglutination of the calamistrum. Native threads of control spiders and spiders with conglutinated calamistrum were observed with a SEM, respectively the cribellate fibres with a TEM. Axial fibre distance, the maximal thread diameter (puff) as well as diameter of the cribellate fibre with and without knot and the distance between knots were measured and compared, using the T -test for statistical analysis. If p < 0.05, significance is assumed. Data are presented as mean ± SD. “n” indicates the number of individuals, for each one the mean was calculated individually previous to calculating the general one. Thread of Control spider Modified spider p-value

22 ± 7 23 ± 5 Distance of axial fibres [µm] p = 0.90 n = 6 n = 2 167 ± 14 178 ± 8 Diameter of cribellate thread [µm] p = 0.34 n = 6 n = 2 20 ± 5 20 ± 3 Diameter of cribellate fibres [nm] p = 0.88 n = 4 n = 2 30 ± 6 30 ± 3 Diameter of knots on cribellate fibres [nm] p = 0.98 n = 4 n = 2 147 ± 34 128 ± 21 Distance of knots on cribellate fibres [nm] p = 0.53 n = 4 n = 2

If no jamming of fibres occur, contact between calamistrum and cribellate fibres takes place most probably at the continuous area of the calamistrum. To support this hypothesis, the contact between calamistrum and thread was reconstructed by measuring the angle between opisthosoma and produced capture thread, called “β” (Fig. 3.26). With about 100°, the angle β was remarkably constant in U. plumipes. As already earlier described, U. plumipes kept the metatarsus of its combing leg almost parallel to the ventral side of its opisthosoma, respectively parallel to the spinnerets. The setae of the calamistrum though were not emerging in a parallel orientation to the spinnerets during capture thread production, but were slightly bended dorsal towards the spinnerets*. Including the orientation of the calamistrum towards the opisthosoma, the morphology of the calamistrum and the understanding that fibres are passing across the calamistrum, β can be used to reconstruct the contact between thread and calamistrum. Such a reconstruction proved that an angle β of 100° leads to contact between cribellate fibres and the continuous area near the tips of the calamistrum (Fig. 3.26). This result confirms the previously proposed hypothesized.

* Observed by Peter Kappel during his bachelor thesis “Analyse des Spinnprozesses cribellater Fasern der Federfußspinne (Uloborus plumipes)” in 2014. Similar information are found in Bertkau (1882). 63

Fig. 3.26: Reconstructing the contact between calamistrum and cribellate fibres. Reconstruction of the contact between calamistrum on the metatarsus (mt) and cribellate thread (ct) with an angle (β) of 100° between thread and opisthosoma (op). The continuous area on the calamistrum is highlighted in dark. Calamistrum morphology adapted to specimen, not true to scale. ps: posterior spinnerets.

Enlarging the angle β would lead to contact shifted towards the base of the setae. In contrast, reducing the angle would lead to contact only with the outmost tips of the calamistrum. Having the angle β determining the area of contact between the fibres and the calamistrum, the spider needs permanently to countervail an elongating thread. Such an elongating thread would otherwise lead automatically to a continuous increasing angle β and hence to a displacement of the area where contact occurs. To compensate this effect, the Uloboridae Z. geniculata and U. plumipes showed an adapting movement of their opisthosoma during laying the capture thread spiral: After fixing the capture thread at the radial frame, they lowered their opisthosoma once initially without performing any combing movement (Fig. 3.27 B, C)*. While one capture thread was spun between two radial threads, an oscillating, but overall lowering movement of the opisthosoma started. Trigonometric calculations revealed, this lowering can compensate the effect of the elongating thread and help keep β constant (Fig. 3.27 A, B). Furthermore, the oscillation within the lowering reflected the movement of the combing leg. The spider lifted slightly its body each time the combing leg would otherwise lower β during its posterior movement. This effect was not as obvious for Z. geniculata as for U. plumipes. Z. geniculata emitted strong vibrations during the thread production, resembling the behaviour “abdominal trembling” described for U. plumipes, although being much slower (≈ 2 Hz in Fig. 3.27 C; compare to chapter 3.1.2.). This vibration was heterodyning the adaption of the position of the opisthosoma. Both species lifted the opisthosoma again to fix the thread at the radial frame after finishing one capture thread.

* Experiments with Z. geniculata were performed under my supervision by Linda Orth during her bachelor thesis “Die Produktion cribellater Fasern von Zosis geniculatus” in 2015. 64

Fig. 3.27: Behavioural adaption enable correct contact between calamistrum and thread. A) Trigonometric calculation, resolving the necessity of lowering the opisthosoma to k eep the angle (β) between opisthosoma and cribellate thread constant despite elongation of the thread. lct: length of cribellate thread, l lw: lowering of the opisthosoma, l b: length between gripping point on thread and spinnerets (simplified as the body size of the spider). B) Example for U. plumipes’ lowering of the opisthosoma over time, equivalent to the elongation of the thread (green dotted line). The process started with an initial lowering, where no combing movement took place. Each combing cycle showed a lifting of the opisthosoma during the p osterior movement of the combing leg. The red rectangles are the calculated data with the formula presented in (A) with l b = 5 mm (average body size of U. plumipes), β = 100° and an elongation of the thread of 0.63 mm/s (average production speed in U. plumipes), including an initial lowering of 0.7 mm within 0.3 s after attachment. C) Same as in (B) but for Z. geniculata. Due to a heterodyning vibration emitted by the spider during thread production, mathematical calculations were not performed for this specie.

3.3.3. Transfer of defined contact between calamistrum and thread to other spiders Changing the focus of the Uloboridae to other cribellate spiders, the morphology of their calamistra differed remarkably. Spiders of the Deinopidae (here: Deinopis subrufa and Menneus sp.) had a completely different shape, although belonging to the same superfamily of Deinopoidae (Fig. 3.28 A, E). Directly after emerging from the cuticle surface, their setae flattened and enlarged their breadth. This led to a covering of the adjacent setae and that way to a formation of a continuous area this time orthogonal to the roots of the setae (Fig. 3.28 B, F). The setae also bend approximately 90° and reduced their breadth again, resulting in large

 Parts of this chapter have been included in the paper “Joel, A.-C., Scholz, I., Orth, L., Kappel, P., Baumgartner, W., 2016. Morphological adaptation of the calamistrum to the cribellate spinning process in Deinopoidae (Uloboridae, Deinopidae). Royal Society Open Science 3“. Submission was part of this thesis. 65 spaces between the setae tips in a parallel plane to the roots (Fig. 3.28 C, G). The complete setae were covered with grooves. Furthermore, the thinner tips of the setae bore one irregular row of teeth (Fig. 3.28 D, H).

Fig. 3.28: The calamistra of Deinopidae. A, E) Overview of the calamistra (ca) on the metatarsus (mt) of D. subrufa (A) and Menneus sp. (E). The arrow points towards the continuous area on the calamistrum observed in both species. The tarsus would be on the right side. B, F) Detail of the calamistrum, showing the setae emerging the cuticle (root (r)). The setae of the calamistrum of D. subrufa (B) as well as Menneus sp. (F) overlap each other, building a continuous area before bending. C, G) Detail of the calamistrum showing the tips of the setae orientated parallel to the roots of the setae. This area on the calamistrum of D. subrufa (C) as well as Menneus sp. (G) is not building a continuous area but shows wide gaps between the single setae. On the edge of each seta (indicated by an arrow), one row of irregular teeth can be observed. D, H) Close-up of the teeth (arrow) on the edge of each seta. SEM images, samples coated with gold.

Calamistra of more distant related spiders, i.e. Hickmania troglodytes (Austrochilidae, Austrochiloidea) and Badumna longinqua (Desidae, Dictynoidae), had a shape all in all more comparable to the shape of calamistra of Uloboridae (Fig. 3.29). Gaps between single setae emerging from the cuticle were visible (Fig. 3.29 B, F). In contrast to the calamistra of Uloboridae, these species had no pronounced continuous area near the tip of their setae (Fig. 3.29 C, G). Their calamistra were covered with grooves and showed teeth-like structures after bending 90° (Fig. 3.29 C, D, G, H). The setae of B. longinqua were furthermore covered with fine bristles at their basis and had not a single row, but an array of teeth (Fig. 3.29 B, D).

66

Fig. 3.29: The calamistra of more distant related cribellate spiders. A, E) Overview of the calamistra (ca) on the metatarsus (mt) of B. longinqua (A) and H. troglodytes (E). The tarsus would be on the right side. B, F) Detail of the calamistrum, showing the setae emerging the cuticle (root (r)). The setae of B. longinqua (B) were partly covered with fine bristles (arrow), which were not observable in H. troglodytes (F). C, G) Detail of the calamistrum showing the tip of the setae orientated parallel to the roots of the setae. The tips of B. longinqua (C) as well as H. troglodytes (G) are not building a distinct continuous area. On the edge of each seta (indicated by an arrow), teeth can be observed. D, H) Close-up of the teeth (arrow) on the edge of each seta. SEM images, samples coated with gold.

The morphological differences in the shape and structuring of the calamistra could reflect in differences within the cribellate spinning process. For this reason, the available recordings of the cribellate thread production of Deinopis sp. were analysed like previously described for U. plumipes (chapter 3.3.2.). Also Deinopis sp. kept a remarkably constant angle β between opisthosoma and thread, although it was higher (120°). Transferring the reconstruction-process proposed for Uloboridae to Deinopis sp., a difference of 20° already lead to a displacement of the contact area between setae and cribellate fibres: The higher (respectively more obtuse) angle β shifted the contact more towards the basis of the calamistrum in Deinopis sp. (Fig. 3.30). This result fits perfectly to the observation of the displaced continuous area on the calamistra of Deinopidae. A lowering of the opisthosoma comparable to the one described for Uloboridae could not be observed. Deinopis sp. kept its body on spot during the whole thread production process, changing only the orientation of its spinneret to the left or right side. Furthermore, no initial lowering of the opisthosoma was observed and cribellate thread production started immediately after fixation at the radial frame. Deinopis sp. seemed to have another strategy to keep the angle β constant: It griped the web with its III legs and adjusted angle β by moving the whole web.

67

Fig. 3.30: Displaced interaction between calamistrum and cribellate fibres in Deinopidae. Reconstruction of the contact between calamistrum on the metatarsus (mt) and cribellate thread (ct) with an angle β of 120° between thread and opisthosoma (op). Contact is shifted toward s the base of the setae by enlarging the angle. This fits to the observation of a changed position of the continuous area (highlighted in dark). Calamistrum morphology adapted to specimen, not true to scale. ps: posterior spinnerets

Comparable results were obtained when analysing the recordings of the other cribellate spiders (except presumably Psechrus sp.). Eresus walckenaeri, Stegodyphus mirandus, K. hibernalis and Z. spinimana did not show an adaption of their opisthosoma during the spinning process (Tab. 3.9). However, most of these spiders showed a movement of their opisthosoma, which could indicate an initial extraction of fibres before the combing started. However, with the available recordings it was not determinable, whether this movement is necessary to obtain correct contact between fibres and calamistrum, comparable to the behaviour observed in Uloboridae. Although the calamistra of the observed specimen from B. longinqua and H. troglodytes showed a continuous area, it remains in order to prove, whether all cribellate spiders have a constant contact area on the calamistrum at all. The Eresidae E. walckenaeri and S. mirandus as well as the Zoropsidae Z. spinimana showed an accumulating of the thread behind their opisthosoma during thread production, which is most likely to inhibit a constant contact area on the calamistrum. Interestingly Deinopis sp. showed an 1.7 ± 0.2 times longer posterior movement of its fourth leg in the last cycle before attaching the finished thread to the radial frame (n = 3). After such an elongated movement, the cribellate thread sticked to the calamistrum in contrast to what was typically observed. This enabled the spider to move the thread independent of its spinnerets. None of the other observed spiders showed this elongated movement within the available recordings. Their cribellate fibres were also never sticking to their calamistra. If movement of the thread independent of the spinnerets occurred, the spiders picked it up with their tarsi instead.

68

Tab. 3.9: Contact area on the calamistrum in different specie s. Data are collected of one or several sequences of online available recordings (other than Uloboridae), each sequence including several combing cycles. Keep in mind, number of available recordings aside Uloboridae was limited and the behaviour observed m ight mark the exception. Superfamily Deinopoidae Eresoidae Filistatoidae Lycosoidae

Family Deinopidae Uloboridae Eresidae Filistatidae Psechridae Zoropsidae

U. plumipes Eresus S. Kukulcania Zoropsis Species Deinopis sp. Psechrus sp. Z. geniculata walckenaeri mirandus hibernalis spinimana

Body adaption No Yes No° No° No Yes° No Direct start of Yes No Yes No° No° No No° combing Tension in Yes Yes No No Yes° Yes No thread

Elongated last Yes No No No No No No leg movement

Picking up of Yes, with Yes, with Yes, with Yes, with No No No thread calamistrum tarsus tarsus tarsus

°: data presented here were extracted from an online available recording with low quality and therefore should be handled with care.

3.3.4. Removing the nearby second row of setae Having characterized the area, where contact between calamistrum and thread occurs, one can now specify, what is happening with the fibres at this contact area. There are several hypotheses about the function of the calamistrum within the cribellate spinning process, which will be examined consecutively more closely. There are two similar structured rows of setae in Uloboridae, one being the calamistrum and the second one building a parallel structure to the calamistrum. Its function is not resolved yet. Due to its close proximity and its similar surface’s structure, it is suggested that it plays a role in cribellate thread production as well (Peters 1984). However, our reconstruction of the contact area indicates no contact between this second row of setae and fibres during capture thread production. Indeed, spiders of which this row of setae was removed produced a capture thread, which did differ neither in shape nor function from threads of control spiders (Tab. 3.10). Therefore, this nearby second row of setae has no relevant function within the cribellate thread production.

69

Tab. 3.10: Removing the nearby second row of setae has no influence on thread structure. Native threads of control spiders and spiders with the second row of setae removed were evaluated with a SEM or in case of the cribellate fibres a TEM. Axial fibre distance, the maximal thread diameter (puff) and the dimensions of single cribellate fibres were measured and compared, using a two-tailed T-test for statistical analysis. If p < 0.05, significance is assumed. Data are presented as mean ± SD. “n” indicates the number of individuals, for each one the mean was calculated individually previously. Thread of Control spider Modified spider p-value

22 ± 7 25 ± 1 Distance of axial fibres [µm] p = 0.66 n = 6 n = 2

167 ± 14 172 ± 13 Diameter of cribellate thread [µm] p = 0.63 n = 6 n = 2 20 ± 5 19 ± 1 Diameter of cribellate fibres [nm] p = 0.83 n = 4 n = 2 30 ± 6 29 ± 0 Diameter of knots on cribellate fibres [nm] p = 0.85 n = 4 n = 2 147 ± 34 153 ± 55 Distance of knots on cribellate fibres [nm] p = 0.53 n = 4 n = 2

3.3.5. Removing the calamistrum One general accepted hypothesis is that the calamistrum is involved in cribellate fibre extraction. After removing the calamistrum of both metatarsi, U. plumipes was still able to extract cribellate fibres and produce a functional capture thread. This contradicts the current hypothesis. Note that the shape of these capture threads differed from threads of not modified spiders, as the puffy shape of the cribellate mat was missing (Fig. 3.31). With a diameter of about 200 μm, the thread of a modified spider (“non-processed thread”) did not differ significantly from the diameter of the puff of a capture thread of a control spider (“processed thread”; about 170 μm; Tab. 3.11). Indeed, the regions where the puffs should have been established were determinable due to a slackly alignment of cribellate fibres (Fig. 3.31 B, C). This slackly alignment vanished after coating the thread with gold. Here, one could determine the cribellate fibre’s alignment. This alignment ascertained the puff itself missing, because the cribellate fibre alignment within a non-processed thread resembled the parallel alignment of cribellate fibres described for the intermediate zones of processed threads (Fig. 3.31 A, E). The enhanced diameter of a non-processed thread could be traced back to a non-closure of the cribellate mat, exposing partially the axial fibres (Fig. 3.31 A). These fibres were not aligned

 Parts of this chapter have been included in the paper “Joel, A.-C., Kappel, P., Adamova, H., Baumgartner, W., Scholz, I., 2015. Cribellate thread production in spiders: Complex processing of nano-fibres into a functional capture thread. Arthropod Structure & Development 44, 568-573“. Submission was part of this thesis. 70 parallel anymore. Their averaged distance was significantly higher (≈ 40 μm) than in processed threads (≈ 20 μm; Tab. 3.11).

Fig. 3.31: Changed structure of cribellate thread after removing both calamistra. A) After coating with gold the thread can no further be divided into puffs and intermediates zones. Furthermore, there is no non-uniform fibre density or irregular arrangement of cribellate fibres, because the puff itself is missing. The seam of the cribellate mat is still observable, not closed over the entire length of the thread. B) Carbon coated thread, showing the cribellate fibres (cf) still lying fluffy around the axial fibres (ax). This treatment enabled one to determine where the puffs should have been. C) Uncoated thread, showing the not parallel alignment of the axial fibres (ax). cf: cribellate fibres. D, E) An axial fibre (ax) not covered with cribellate fibres (cf), allowing one to see the linkages between axial fibres and the cribellate mat (arrows). SEM images, A, E: coated with gold, B: coated with carbon, C, D: no treatment.

Tab. 3.11: Structural change of the cribellate thread after removal of the calamistrum. Native threads of control spiders and spiders with removed calamistra were observed with a SEM, respectively the cribellate fibres with a TEM. Axial fibre distance as well as the maximal thread diameter (in threads of control spiders, this is the puff) were measured and compared, using a two-tailed T-test for statistical analysis. If p < 0.05, significance is assumed. Data are presented as mean ± SD. “n” indicates the number of individuals, for each one the mean was calculated individually previously. Thread of Control spider Modified spider p-value

22 ± 7 40 ± 17 Distance of axial fibres [µm] p = 0.04 n = 6 n = 6

167 ± 14 195 ± 59 Diameter of cribellate thread [µm] p = 0.29 n = 6 n = 6 20 ± 5 18 ± 2 Diameter of cribellate fibres [nm] p = 0.59 n = 4 n = 6 30 ± 6 29 ± 2 Diameter of knots on cribellate fibres [nm] p = 0.82 n = 4 n = 6

147 ± 34 138 ± 29 Distance of knots on cribellate fibres [nm] p = 0.68 n = 4 n = 6

71

The linkages between cribellate mat and axial fibres were still observable (Fig. 3.31 D, E). This result is supported by observations made within adhesion measurements.* Pulling at the cribellate fibres in orthogonal direction with an adhered spherical item revealed no significant difference in thread stability, indicated by a separation of cribellate fibres from axial fibres (Fig. 3.32; p = 0.51). Such an instability would have been a hint for a missing or declined linkage between axial thread and cribellate mat. Finding these linkages not curtailed in non- processed threads, it can be ruled out that the linkages are produced by the action of the calamistrum and are the reason for the puffy structure of the thread exclusively.

Fig. 3.32: No changed thread stability. Pulling an adhered spherical item in orthogonal direction away from the capture thread, the capture thread can frazzle, separating the cribellate fibres from the axial ones. This frazzling behaviour was not significantly changed comparing processed threads of a control spider (n = 3) with the non-processed ones of a spider were the calamistra were removed (n = 5) (two-tailed T- test; p = 0.51).

* Raw data were collected by Hana Adamova during her bachelor thesis “Der Einfluss vom Calamistrum auf die Struktur und die Funktion cribellater Fangfäden” in 2015. Her thesis was supervised by me. 72

Although the calamistrum is not necessary for fibre extraction, the typical combing movement was still performed by spiders with removed calamistrum*. Indeed, contact between leg and fibres still has to occur, because in all cases, where the calamistrum was removed, the metatarsus was covered with < 200 nm thick fibres, clotting the metatarsus (Fig. 3.33). Such a clotting was never observed on metatarsi of unmodified spiders or on metatarsi of spiders of which only the nearby second row of setae was removed.

Fig. 3.33: Clotting of metatarsus after removal of the calamistrum. A) Overview of the metatarsus (mt) showing the removal of the calamistrum, without removing the setae row nearby (“2 nd row”). B) Close up of the same metatarsus, showing fibres clotting the roots and stubs of the calamistrum. In the background you can see s etae of the row nearby (“2 nd row”). SEM images, sputtered with gold.

Here we demonstrated that the calamistrum is neither necessary for brushing out the cribellate fibres nor for linking them to the axial ones, but is necessary for producing the puffs themselves. Removing the calamistrum leads also to a clotted metatarsus. As a consequence, the calamistrum has to have also a function in keeping the metatarsus clean. The nonexistence of the puffs seems to have an impact on the correct alignment of the fibres, as the cribellate mat was not able to enclose the axial fibres over the entire length and the axial fibre alignment was disturbed.

* Note that nocturnal observations of the spiders’ behaviour were performed shortly after the removal of the calamistrum. This was done to control, whether spiders try and potentially only fail to produce a thread. It was not controlled, whether spiders can adapt their behaviour to a missing calamistrum. 73

By proving the calamistrum not necessary for the extraction of the cribellate fibres, but for processing those, the question arose, what function the processing, respectively the resulting puff has. The influence of the thread’s structure on its performance has already been often investigated. Until now, scientist were only able to use different shaped threads of again different spider species. Number of fibres as well as protein-composition might differ between those spiders, curtailing the findings. Removing the calamistrum enables one to determine the function of the puff without these restrictions, because the same species can be modified to produce different structured threads. Comparing now capture threads with and without puff, one could observe prey is captured with both of them. The adhesion itself did not look different in the samples. In several samples of processed and non-processed threads, a coalescence of fibres seems to happen (Fig. 3.34). Likewise, there was no significant difference in the restraining time of prey (p = 0.82) (Fig. 3.35 A). Similar results were obtained, when measuring the adhesion force with an artificial spherical item (p = 0.74) (Fig. 3.35 B)#. The puffy structure therefore has no significant influence on the adhesion of the cribellate capture thread.

Fig. 3.34: Adhesion of cribellate thread on prey. A) Drosophila melanogaster entangled in a non-processed capture thread. During adhesion, the cribellate mat can open its structure, enhancing the adhesion area. B) Single cribellate fibres (cf) adhering to and entangling the setae of a fruit fly. C) Beside the observation of single cribellate fibres (cf), there were areas, in which the cribellate fibres seemed to fuse to one single layer (cf*). This layer coats the features of the fruit fly, but also the axial fibres (ax) do impress the layer. SEM-images, coated with gold.

 Data were collected under my supervision by Hana Adamova during her bachelor thesis “Der Einfluss vom Calamistrum auf die Struktur und die Funktion cribellater Fangfäden” in 2015. 74

Fig. 3.35: Adhesion properties of processed and non-processed threads. A) Retention time of Drosophila melanogaster in webs built by spiders with removed calamistra (non-processed threads) and by not modified control spiders (processed threads). The data were categorized in a biological sensible cluster, because spiders would definitely catch any prey, which is restrained more than 5 minutes and might have trouble catching prey restrained below 2 minutes. The differences were not significant, following a G-test with two degrees of freedom and p < 0.05 defined as significant. n = 6 for non-processed threads and n = 12 for processed threads. B) Measuring the adhesion force with a standardized globular adhesion object with an area of 0.23 mm2, no significant differences in the adhesion force could be determined (two - tailed T-test, p > 0.05, n = 5 for non-processed threads and n = 3 for processed ones). Adhesion area was roughly determined to be identical in all samples.

3.3.6. Electrostatic charging Because the calamistrum has been proven not to extract the cribellate fibres, but to process them, one can now has a closer look at the potential processing event. It is a common hypothesis that cribellate fibres are electrostatically charged by the calamistrum, keeping the adhesive cribellate fibres separated within a puff. To validate this hypothesis, one can discharge the fibres for example by exposing them to high humidity. Within the experiments, the thread lost its puffy structure when raising the humidity to 80% with the help of a nebuliser (n = 3). The same phenomenon, however, could not be replicated by letting water simply evaporate, although reaching the same humidity (n = 3) (Fig. 3.36). The collapsing after exposure to fine water droplets was irreversible also when drying the thread on silica gel and/or exposing it to an electron beam afterwards. The beam however should be able to recharge the fibres. Interestingly, the alignment of cribellate fibres of such modified threads resembled the alignment observed in non-processed threads, respectively the one in intermediate zones (Fig. 3.36 C, D). The puff-less non-processed threads could not be re-shaped likewise by charging these threads with the help of an electron beam. The puffy structure therefore is not maintained only by electrostatic forces.

75

Fig. 3.36: Removing electrostatic charge from the capture threa d. A) Thread after exposure to high humidity shows no structural change compared to untreated threads. ax: axial fibre, cf: cribellate fibres, iz: intermediate zone. B) Thread after exposure to fine mist to enhance humidity shows collapsing of the puffy st ructure. ax: axial fibre, cf: cribellate fibres, iz: intermediate zone. C) Thread after exposure to fine mist (compare (B)). SEM image, coated with gold. D) Thread after exposure to water mist. SEM image, coated with carbon.

To quantify, if the thread is charges at all, one can approach the thread with a charged object and determine, whether it reacts to the approaching object and if it does so, whether it is drawn towards it or repelled. When approaching a normal puffy capture thread with a positive charged glass rod, the thread was drawn towards the rod. The thread was drawn also to negative charged foamed plastics. It can therefore be excluded that the complete capture thread itself is charged at all. It acts like a dipole.

76

4. Discussion

4.1. Uloborus plumipes: from the natural habitat to a lab colony Typically occurring in southern Europe, India, China, south-east Asia, Africa or islands in the Atlantic (Marples 1962), U. plumipes was introduced into northern European garden centres and flower shops probably during the 1990s, where it built autarkic populations (Dawson 2001; Jonsson 1998; Klein et al. 1995; Suvák 2013). First detection in Germany was near cologne in 1989 (Klein et al. 1995). Already then, high population densities were described, most certainly favoured by the missing of its natural predators. We observed densities over 100 spiders/m2 in the natural habitat, making the maintenance of larger colonies of these spiders in principle very easy. Such high population densities are probably favoured by the fact that Uloboridae are lacking venom glands and its only possibility to make prey is direct confrontation and wrapping the opponent (Foelix et al. 2015b; Jonsson 1998; Marples 1962; Peters 1982). This lowers the risk of cannibalization, because spiders typically not adhere to their own sticky silk and therefore an escaping of the opponent is likely (Briceno et al. 2012; Kropf et al. 2012). Matching this description, we found no distinct cannibalistic tendencies in nature, the lab colony or during the experiments. Although cannibalism can be excluded as driving force, we found females of U. plumipes competing on a regular basis. The winner of the competition was always occupying the web afterwards. Such competitions were already described earlier for spiders of different families (Horton et al. 1983; Riechert 1978, 1981). Cannibalism as driving force could be excluded likewise (Wise 2006). Competition was rather associated to e.g. the acquisition of a larger territory or nutrition. Fighting was not only induced by floaters of U. plumipes, but also by spiders already owing a web. These spiders enlarged their webs by merging several. Web conquests may therefore reflect a strategy to save the time and investment required for silk production, similar to silk-stealing kleptoparasites (Tso et al. 1998). Because we observed a higher occurrence of floaters when raising the density in the lab colony, U. plumipes might not only compete for the woven web itself, but also for the territory when space gets scarce. Similar observations were made for other spider species (Hodge 1987; Riechert 1981; Rypstra 1983). For instance, Agelenopsis aperta not only compete for webs but actively patrol within the vicinity of their web sites. They attack neighbouring spiders and thus establish territories larger than their sheet webs (Riechert 1981).

77

The outcome of a competition between two female U. plumipes is not influenced by the resident status, as described for A. aperta and others (Hodge 1987; Riechert 1981; Rovner 1968). An intruder/defender-bias is suggested to be a side-effect of extreme habitats where finding a suitable spot is difficult. In the cases where an intruder/defender-bias can be excluded, typically the size of the spider, measured either with weight, leg length or size of the body, determines the outcome not only of female-female contests, but also of male-male ones (Bridge et al. 2000; Christenson et al. 1979; Enders 1974; Faber et al. 1993; Hack et al. 1997; Hodge et al. 1995; Riechert 1978, 1981; Taylor et al. 2001; Taylor et al. 2003; Wells 1988; Whitehouse 1997; Wise 1983). This phenomenon is not restricted to spiders, but typically influence the outcome of contests in the animal kingdom (Dodson et al. 2001; Hack 1997; Lopez et al. 2001; Neat et al. 1998; Turner 1994). This observation applies also to U. plumipes. In all these exemplary named cases, the contests were not always won by the bigger one. For instance, larger females of Metepeira labyrinthea won only 76% of the disputes (Wise 1983). Likewise, in U. plumipes 19% of the contests were won by the lighter spider. This indicates that physical advantage does not always translate into competitive success (Draud et al. 2004; Kasumovic et al. 2011). When analysing contest related activities, we found a positive association of several behaviours with winning. “Attacking”, “reaching the hub first” as well as “web shaking” were more often performed by the later winner of the contest. Similar association can be found in males of the web building spider Argyrodes antipodiana. These spiders showed typical behavioural patterns for winner or loser (Whitehouse 1997). Although one study suggest larger females of the Grammostola schulzei tend to be more aggressive, little is known whether morphological differences of the competing spiders influence their behaviour and/or whether the behaviour influences the outcome of the contest (Ferretti et al. 2011). In other animals, the behaviour and/or the experience in fighting has been proven to effect the contest (Dodson et al. 2001; Hsu et al. 1999; Jennings et al. 2004; Lopez et al. 2001; Olsson 1994; Stuart-Fox 2006). Furthermore, morphological traits could be linked to the performance of a spider (McGinley et al. 2013). We found the frequency of aggressive activities to be independent of the morphological traits for U. plumipes. Individuals that won the contests despite an inferiority showed a higher tendency for aggressive behaviour, probably as a compensatory demeanour. In addition, the behaviours “attacking” as well as “reaching hub first” were determined as factors influencing the outcome of a contest. This stands to reason that other factors beside the physical advantage, e.g. level of commitment, might influence the outcome of the contest as well. There might be competition without any communication and on that account competition outcome would be a

78 mixture of physical fitness and motivation. As suggested in cumulative and sequential assessment models, continuing escalation in the conflict may be based on rival-dependent cost accumulation and individual thresholds without the need to assess the potential of the opponent (Arnott et al. 2009; Enquist et al. 1983; Payne 1998). In contrast to those strategies stands the finding that the spider occupying the hub first is more likely to win the contest. The occupation of the hub is well-known behaviour, giving the spider best vibrational perception within the web. Most web-building spiders, including U. plumipes, have poor vision and do not appear to detect acoustic signals. Their main sensory environment used for communication and signalling consists of chemical and tactile stimuli, including substrate-born vibrations (Barth 1982, 1986, 1997, 1998; Baurecht et al. 1993; Elias et al. 2008; Elias et al. 2003; Girard et al. 2011; Maklakov et al. 2003; Masters et al. 1981; Schüch et al. 1990; Wignall et al. 2013b). “Reaching the hub first” has been proven an advantageous behaviour in combats of other spiders as well (Buskirk 1975; Christenson et al. 1979). Because “reaching hub first” is not an (at least) obvious aggressive display, the behaviour itself might not influence the outcome of the contest. “Reaching the hub first” though should give the spider the advantage to get more precise information about location and traits of the opponent. Such information could enable the spider to adapt its (signalling) behaviour. The ability to adapt its behaviour would be favoured by (or explain) the flexibility of the sequence: While the distinct activities of the contests can be combined to stereotypical modules of behaviour, there were no fixed sequences of the different modules. This contrasts descriptions of courtship displays, however this flexibility in female-female contest has been already described earlier (Buskirk 1975; Elias et al. 2010; Elias et al. 2003; Riechert 1978; Wignall et al. 2013a). Having a closer look at potential signalling behaviours, we characterized three vibrational patterns. The behaviours “thread pulling” and “abdominal trembling” are not restricted to intraspecific competition, but occurred regularly as a reaction to web vibrations. It has been suggested that they are used for detecting and locating prey (Bradoo 1986; Jackson 1992; Jonsson 1998; Lubin 1986; Marples 1962; Opell 1979). “Web shaking” on the other hand is a previously unknown pattern of behaviour, which we observed only in intraspecific competitions. A similar behaviour has been observed in the Pholcidae Holocnemus pluchei as an effective defence mechanism to predators within the web (Jackson 1992). Because its occurrence and likewise its intensity (i.e. amplitude or frequency) are not correlated to the morphological traits of a spider, this behaviour might be adaptable by the inferior spider. Web shaking therefore could serve two functions: 1) It dislodges the antagonist from the web, providing the aggressor a strategical advantage; 2) The movements communicate the aggressive

79 potential of the contestant via substrate-bound vibrations. Nevertheless, also “abdominal trembling” and “thread pulling” might not only be used for detection, but also for communication. Buskirk (1975) already suggested an aggressive form of communication between females of Metabus ocellatus, finding food-locating web movements having probably a communicatory function in this species. Further analysis has to determine, whether information about the spider itself or the opponent can be communicated through these vibrations and if so, whether this form of communication is widely distributed among spiders. These intraspecific fighting, however, had no crucial impact on establishing a lab colony. It influenced the experimental setup though by making it necessary to separate spiders after e.g. modifying the calamistrum. The weak point of the lab colony was the low male hatching rate of 10%, reflecting in equally low densities of adult males in the natural habitat (own observation as well as Jonsson (1998)). In addition, their survival rate as adult males was rather low and similar to the one observed in the Uloboridae , although sexual cannibalism as described for Nephila plumipes or Dolomedes tenebrosus is not pronounced in this spider (Opell 1982b; Schneider et al. 2001; Schwartz et al. 2014). Male cribellate spiders are not able to build their own web, as they lose cribellum and calamistrum with the final moult (Coddington et al. 1987; Eberhard 1977; Foelix 2015; Griswold et al. 2005; Kovoor et al. 1988; Miller et al. 2012; Opell 1979, 1982b, 1989; Opell et al. 1995; Peters 1992c). Stealing prey of the female was only tried occasionally by males of U. plumipes and seldom successful. Hence, male spiders starved most of the time. This made the breeding of this species rather difficult.

4.2. Capture thread structure Changing now the focus from the ecology of U. plumipes to its prey capturing devise, new insight in the structure of the cribellate capture thread can be provided. As often previously assumed without any proof, this study confirmed that one structural unit (consisting of a puff and an intermediate zone) of a cribellate thread is produced by one stroke of the calamistrum, not only for U. plumipes, but also for Z. geniculata (Kovoor et al. 1988; Kullmann et al. 1975; Opell 1979, 1989; Peters 1987, 1992a, c). Furthermore, a sheet like arrangement of the cribellate fibres, called “cribellate mat”, could be confirmed (Eberhard et al. 1993; Opell 1989, 2013; Peters 1983). The cribellate mat built a hollow structure around the larger axial fibres, establishing a seam where the structure closes. Although publications are not explicitly referring to such a seam, it can be observed in sputter coated threads of many cribellate spiders, e.g. Desidae (B. longinqua), Austrochilidae (H. troglodytes or Thaida peculiaris), Deinopidae and other Uloboridae (Bond et al. 2014; Fernandez et al. 2014; Foelix 2011, 2015; Griswold et 80 al. 2005; Hawthorn et al. 2002, 2003; Huang et al. 2009; Kullmann et al. 1975; Opell 1979, 1999, 2002, 2013; Peters 1984, 1992c, 1995a). Such an fibre alignment is consistent with the placing of the spigots and the spinnerets of a spider (Fig. 4.1 A) (Eberhard 2010).

Fig. 4.1: Spigot placement and thread formation. A) Ventral overview over the cribellate spinning process. Cribellate fibres are extracted as a mat from the cribellum (cr) anterior of the median spinnerets (ms) and posterior spinnerets (ps). This mat can easily be transported over the spinnerets and afterw ards surround the axial fibres (ax). This is probably facilitated by the puffy structure (compare to (B)). iz: intermediate zone, op: opisthosoma. B) Model of the mat’s enclosure. Cribellate fibres (cf) are linked to the axial fibre (linkage) and afterwards extracted uniformly from the cribellum (cr) (left picture). When a puff is produced, the curling fibres take up more space, stretching the outer cribellate fibres (right picture). To relax fibres again and due to a twisting of the complete thread by the spider, the mat of cribellate fibres enclose around the axial fibres.

Earlier studies characterized the capture thread as a relatively robust unit, although consisting of three types of silk (Opell 1989; Peters 1983, 1992a, c). For that reason, some kind of connection between axial fibres and cribellate fibres was suggested. This hypothesis could be verified by finding linkages between cribellate mat and axial fibres in the intermediate zones. Although the authors did not specifically refer to these, linkages are also depicted in literature (Elettro et al. 2015; Kronenberger et al. 2015; Kullmann et al. 1975; Kullmann 1972; Peters 1992c). Beside the linkage, fine fibres covering the axial fibres over their entire length were observed. Such a cover of fine fibres is also described for the axial fibre of Eresidae (Peters 1992a). The linkage as well as the fine fibres covering of axial fibres actually might be no cribellate fibres but paracribellate fibres, because their occurrence fit to the paracribellate substructure described by Peters (1984). It could not be determined within this study, whether the linkages are produced by simple entangling of fibres or whether fibres are linked with the help of a glue-like secretion comparable to the one of pyriform glands (Apstein 1889; Gorb et al. 1998; Saffre et al. 1999; Schutt 1996; Townley et al. 2003; Wolff et al. 2015). As the pyriform glands’ secretion is ejected from the anterior spinnerets, which are not involved in the cribellate spinning process, Peters (1984) suggested that fibres might be connected to each other during their hardening process.

81

Having now a more detailed insight in the structuring of the cribellate thread, the question how the fibres are assembled arises. The mat aligns presumably autonomously around the axial fibres: Within a puff, cribellate fibres are convoluted, taking up more space per single fibre compared to the parallel aligned ones in the intermediate zone. Nevertheless, cribellate fibres are probably extracted uniformly from the cribellum with similar length. By being linked to the axial fibres, tension is generated in the outer cribellate fibres of the sheet when being pressed aside due to convolution (Fig. 4.1 B). To minimize the tension again, fibres at the margin of the sheet probably fold over the inner fibres. The final enclosure is facilitated by twisting the thread during production, as observed when trying to track the seam. This hypothesis of self-assembly is supported by two observations: 1) Axial fibres are not directly aligned parallel after being pulled from the spigots of the glandulae pseudoflagelliformes. They align later in the thread production process, probably by being pressed together by the cribellate mat. This kind of assembly process is not only observable in the Uloboridae U. plumipes, Z. geniculata and the Deinopidae Deinopis sp., but also visible in a picture of the Austrochilidae Austrochilus forsteri (Lopardo et al. 2004). 2) In a non-processed thread without any puffy structure, the axial fibres are not aligned parallel anymore and have a significant larger distance to one another. Furthermore, the mat is not able to enclose the axial fibres along the whole length of the thread. Although twisting of the thread still occurs, the tension generated by the convoluted cribellate fibres is missing, not facilitating the closure of the cribellate mat. Keep in mind that though these findings support the proposed hypothesis of self-assembly, it is possible that the calamistrum influence the cribellate mat closure in other ways than only crimping the cribellate fibres. By characterizing the cribellate capture thread structure in more detail and finding this structure not limited to Uloboridae, the first structural variation visible in capture threads of different species can be traced back to the cribellate fibre alignment building a mat. Without the knowledge that cribellate fibres are arranged as a mat whose closure establishes a seam, this fibre arrangement feigns different shapes of capture threads when changing the angular field (compare again Fig. 1.6 A, B, F). Nevertheless, some differences in thread shape cannot be explained by observing only the structural components. E.g. the width of the thread can change, the number of puffs per millimetre is not constant, and especially the looped structure of the capture threads of the Filistatidae K. hibernalis and the Dictynidae Mexitlia trivittata needs further evaluation (Griswold et al. 2005; Opell 2002).

82

4.3. The capture thread’s structure during the developmental process It is well established that 2nd instar spiders do not have a functional cribellum, paracribellum and calamistrum and build webs without any capture threads (Eberhard 1977; Hajer 1991; Kovoor et al. 1988; Marples 1962; Opell 1982b; Peaslee et al. 1983; Peters et al. 1980; Wiehle 1927). Own observations of 2nd instar of U. plumipes confirmed these studies. After the second moult (3rd instar) a cribellum, paracribellum and calamistrum could be detected. Although the size of all three is increasing with each moult, 3rd instar spiders can already produce functional capture threads. However, these first threads lack the pronounced puffy structure. Not only for U. plumipes, but also for other cribellate spiders, like the Uloboridae and Z. geniculata or for Filistatidae and Eresidae, more protruding structures within the thread develop with each moult (Kullmann et al. 1971; Lopardo et al. 2007; Opell 1979, 1982a, b, 1989; Wiehle 1927). Such observations go hand in hand with a previous study proving an increasing cribellum spigot number (measured as “cribellum width”) as best predictor for cribellate puff width for threads of Miagrammopes animotus (Opell 1989). The capture thread width therefore is suggested to depend on the spigot number of the cribellum, explaining the second structural difference of capture threads. Even though those results do not influence our knowledge about the thread structuring, one should not disregard the finding that although 2nd instar U. plumipes lack a cribellum, paracribellum and calamistrum, they are able to produce 1) nanofibres at all and 2) nanofibres with knots, resembling the cribellate fibres. The production of nanofibres in spiders was always associated to the existence of the cribellum as a specialized system comparable to electrospinning in technical applications (Kronenberger et al. 2015). This phenomenon has not been described previously and the origin of these nanofibres could not be determined within this study. More investigations focusing on this aspect are needed to give an eligible explanation. This applies also to the function of these threads, as 2nd instar spiderlings were never observed catching or feeding on prey. Hence, these fibres cannot serve the same function as cribellate fibres.

83

4.4. The calamistrum

4.4.1. The extraction of the cribellate fibres To determine the influence of the calamistrum on the cribellate thread shape, the function of the calamistrum had to be more closely examined. The primary function of the calamistrum was assumed to be brushing out the cribellate fibres (Bertkau 1882; Eberhard 2010; Foelix 2011; Kronenberger et al. 2015; Opell 1982a, 2013; Peters 1983, 1984, 1987, 1992a, c; Sahni et al. 2011). Finding still a functional cribellate thread after the removal of the calamistrum is contradicting this hypothesis, proving the calamistrum is not necessary to extract the fibres. Eberhard (2010) already deliberated (and negotiated) whether there is any possibility to extract cribellate fibres without the help of the calamistrum: Although there are muscles which could help eject silk from the spigots of the cribellum, the common process of spider silk extraction involves pulling forces (Bertkau 1882). The extraction of cribellate fibres by pulling forces due to the movement of the spider would be possible, once the cribellate fibres were attached to the axial fibres. This process would be comparable to dragline extraction in Nephila clavipes or argentata. However, spigot placement inhibits contact of the cribellum with the emerging axial fibres even when the posterior spinnerets are retracted. For this reason, cribellate fibres have to be extracted at least initially. There are two possibilities, how cribellate fibres could be extracted without calamistrum: This study proved a motion of the median spinnerets during the cribellate capture thread production. Their abduction might enable contact between the paracribellate fibres and the cribellate ones, connecting both. When adducting again, they come close to the likewise adducted posterior spinnerets, possibly connecting the paracribellate fibres (themselves connected to the cribellate fibres) to the axial ones. This would facilitate the extraction of the cribellate fibres. Yet, contact between the spigots of the cribellum and paracribellum was not observed in this study. Therefore, this hypothesis is rather unlikely. Secondly, modified spiders are still performing the same stroking behaviour described for spiders with calamistrum. Furthermore, residuals of nanofibres were clotting the metatarsi of spiders with removed calamistra. It is most likely that the metatarsi of modified spiders still come in contact with the cribellate fibres, thus taking over the assistance of extracting those. In contrast to the hypothesis that the extraction is due to the action of the paracribellate spigots, the assistance of the metatarsus during extraction of cribellate fibres would be a compensatory behaviour. If this second hypothesis is true, the calamistrum does indeed help extracting fibres, despite not being necessary.

84

Although a changed behaviour of the spider due to a missing calamistrum was not detectable within this study, it is most likely that the spider notice the difference and can adapt its behaviour to the loss of the calamistrum. The setae of the calamistrum probably serve as mechano receptors, because they are triple innervated (Foelix et al. 1978).

4.4.2. The function of the calamistrum Whatever behaviour the spider uses to extract the cribellate fibres, this behaviour is not able to replace the function of the calamistrum completely. A clotting of the metatarsus with fibres was never observed in spiders still bearing a calamistrum. The calamistrum therefore needs antiadhesive properties, which make either the removal of fibres simpler (spiders show extensive cleaning of the leg after web construction (Eberhard 1972)) or which inhibit an adhesion of fibres in the first place. Similar observations were already described previously, not only for the calamistrum, but also for spinnerets and tarsi of cribellate spiders (Opell 2013). The antiadhesive properties might be established due to a coating, like described for ecribellate spiders (Briceno et al. 2012; Kropf et al. 2012). Furthermore, the structuring of the calamistrum could have an impact of nanofibres adhesion. The general assumption that the calamistrum is involved in the production of the puff itself, could be verified (Coddington et al. 1987; Kronenberger et al. 2015; Peters 1992a, c; Sahni et al. 2011): The removal of the calamistrum results in a lack of puffs. Linkages between cribellate mat and axial fibres still existed. The calamistrum hence is neither necessary for fibre extraction nor for thread assembly in terms of forming linkages, but contact between cribellate fibres and calamistrum leads to the formation of the puffy structure within the cribellate capture thread. These results indicate that the calamistrum has two functions: 1) the formation of the puffy structure and 2) keeping a distance between the adhesive cribellate fibres and the metatarsus and therefore inhibiting a clotting of the metatarsus.

4.4.3. Influence of thread structure on the function of the capture thread By showing the calamistrum is not necessary for fibre extraction, but for formation of a puffy structure within the thread, the question of the function of this puffy structure is provoked and should be evaluated shortly. Earlier studies proved more protruding structures of capture threads increasing the adhesion properties of the thread, probably by increasing the contact area (Opell 1999, 2002). Furthermore, prey should be entangled more easily in wider and more curly structures (Opell 1994a). Following these hypotheses, the formation of the puffy structures should have a positive influence on the adhesion of the cribellate thread by increasing the contact area. This hypothesis was not confirmed. Neither a decreased ability to restrain prey 85 nor a decrease in the adhesion force on smooth artificial surfaces could be detected in non- processed threads. These results supports the observations that the number of cribellate fibres is mainly determining the adhesion properties (Opell 1994c, 1995, 1999; Opell et al. 2009). Because this study did not include the measurement of the adhesion force on natural surfaces like the integument of prey, any potential differences here remain undetected. Proving the function of the thread not impaired in non-processed threads, one should assume the calamistrum would degenerate to a structure inhibiting only a clotting of the metatarsus with cribellate fibres. It is of course possible that there is no negative selective pressure driving the reduction of the calamistrum. It is equally likely that not the puff itself, but the hand-in-hand going disturbed thread assembly influences negatively the fitness by compromising the adhesion: With a non-closure of the cribellate mat as observed in non-processed threads, parts of the non-sticky axial fibres are not covered with the adhesive cribellate fibres. Any insect getting in touch with this part of the thread does not get caught. Although the adhesion measurements tried to incorporate this aspect, the twisting of the thread by the spider inhibits any prediction, in which position the seam can be found. It is not possible to evaluate this aspect with current methods, as the seam becomes only visible after coating the thread with gold.

4.4.4. Puff formation Although the biological function of the puff remains unclear, finding the calamistrum controlling the puff formation enables one to investigate different hypotheses about how the puffy structure of capture threads is established: Jamming between axial fibres Peters (1984) suggested that the intermediate zones are a product of jamming the bundle of cribellate fibres between both axial fibres. There are several reasons, why this hypothesis can be excluded as an explanation for capture thread structuring: In the introduction it was already pointed out that several spiders split their capture thread during production, which leads to only one stabilizing axial thread in each half strand (Eberhard et al. 1993; Kullmann et al. 1971; Lehmensick et al. 1957; Opell 2002; Peters 1983, 1992a, b). Additionally, K. hibernalis showed a deferred posterior spinneret movement in this study, disabling a jamming of cribellate fibres between axial fibres by the action of the posterior spinnerets. Nevertheless, their capture threads show a puffy structure (Griswold et al. 2005; Opell 2002). Likewise, a jamming could neither be observed in the finished thread of Uloboridae, nor does this jamming-hypothesis fit to the observation of the missing puffs when removing the calamistrum. Jamming of cribellate fibres between the axial ones can therefore 86 be excluded. What Peters (1984) suggested to be the constriction of an axial fibre might be the seam of the cribellate mat. Electrostatic charge The most common hypothesis is that an electrostatic charge is exerted on the cribellate fibres by the calamistrum, keeping the cribellate fibres apart within a puff (Elettro et al. 2015; Kronenberger et al. 2015; Opell 1982a, 1994c; Peters 1984; Sahni et al. 2011; Vollrath 2006). With the calamistrum removed, this charge would be missing and the puff collapse. This hypothesis is supported by the observation that a suggested crimping of cribellate fibres due to post-draw loading is missing in non-processed threads of spiders (Kronenberger et al. 2015). The observed collapse of puffs after coating them with gold or after contact with fine mist could also be explained with the removal of electrostatic charge. These observations are restricted not only to U. plumipes, Z. geniculata, B. longinqua and K. hibernalis, but such collapsing can also be found comparing literature about other Uloboridae and Deinopidae (Opell 1989, 1990; Peters 1992c). Looking closer, this hypothesis has some serious drawbacks: Although one could suggest the collapsing of the threads after gold-coating is due to the removal of the electrostatic charge, puff and intermediate zones are still discriminable in these samples. On the contrary, after exposing the threads to fine mist, the puffs and intermediate zones are not distinguishable anymore. If both methods were equally removing electrostatic charge, threads should look alike after both treatments. In addition, after coating a thread with carbon, thread shape resembled the one of the native thread. As carbon is also used to make non-conductive samples conductible, fibres should be discharged due to this treatment, too. Coating could lead to artefacts by e.g. covering the cribellate fibres with too much material and/or conserving the structure. This would influence negatively the drawn conclusion. If only electrostatic forces are keeping the fibres apart within a puff, one could 1) easily re-establish the puffy structure of non-processed threads or threads after contact with mist by exposing them to an electron beam, 2) remove the charge of a processed thread by raising the humidity below saturation (so no fluid water occurs) and 3) observe the thread is drawn towards, respectively repulsed from a positive or negative charged object. All those three objectives were not confirmed in our experiments, excluding Coulomb-forces as main component keeping the fibres separated in the finished thread. This observation fits to the description that although some cribellate spiders live in deserts, like , many other cribellate spiders, like U. plumipes, sinensis, Stegodyphus pacificus or Waitkera waitakerensis, live in areas with 60% up to 90% relative 87 humidity (Eberhard 1971, 1972; Kullmann et al. 1971; Kumhof et al. 1992; Opell et al. 2000). Furthermore, threads with cribellate fibres bearing knots show a raising of the adhesion force when being exposed to humidity up to 99% (Hawthorn et al. 2002, 2003; Opell 2013). Such high humidity should be able to reduce charge within the fibres and, if they were kept apart only by electrostatic forces, lead to a collapse of the puffs. A collapsing of puffs due to wetting is described to annihilate the adhesion force (Elettro et al. 2015). Therefore, fluid water has to have another impact on the thread than high humidity. Already Opell (1993) observed no collapsing of puffs when raising humidity by evaporation. He suggested already no electrostatic forces involved in the maintenance of the puffy structure, but in newer literature this work has been neglected. Compression of cribellate fibres It has been hypothesised that the calamistrum hook the cribellate fibres for transportation by bending the twisted and curved setae of the calamistrum and enclosing the fibres in-between (Kronenberger et al. 2015; Peters 1984). This hooking and releasing of the cribellate fibres might be assisted by an angular change of the spider’s fourth leg (Opell 2013). Furthermore, such tight jamming between the setae was suggested to enable the spider pushing the cribellate fibres together, forming one puff. Although U. plumipes showed a slightly changing angle between cribellum and calamistrum (α) during one stroke, the same holds not true for the related species Z. geniculata. Conglutinating the setae of the calamistrum finally proved no jamming of cribellate fibres. This modification had no impact on the capture thread structure at all. Cribellate fibres do not pass between the setae and therefore cannot be jammed in-between. It remains in order to prove, if the exclusion of this hypothesis directly rules out a compression of the nanofibres. Pulling out longer cribellate fibres than axial fibres Although a jamming of cribellate fibres between the setae can be excluded, cribellate fibres can nevertheless be extracted in longer length by the action of the calamistrum and therefore undulate around the axial fibres between the linkages in the intermediate zones, as previously suggested (Coddington et al. 1987; Eberhard et al. 1993; Opell 1989; Peters 1992a). The stroke length of U. plumipes is about 10 times longer than one single puff, which matches this hypothesis. This hypothesis is at least partly true, because threads of spiders with removed calamistra still show a fluffy alignment of cribellate fibres between two linkages. Because these spiders still perform the extraction movement of the combing leg, cribellate fibres are probably extracted longer than axial fibres by the combing leg even without the calamistrum.

88

Processing Although some part of the puffy structure is probably established by extracting longer cribellate fibres than axial fibres, such length difference cannot explain the overall structure. The crimping of the cribellate fibres forming an asymmetric puff structure in gold coated threads has to have another origin. If not compressed, the cribellate fibres have to be processed by the calamistrum somehow. Although the socket of the setae would allow flexibility in one direction, any flexibility is not necessary in the process, because conglutination has no impact on thread structure. Therefore, one can assume the calamistrum as a rigid and stiff unit. The ridges and ripples might work as sharp tools, modifying the surface of the cribellate fibre by shifting parts of it, comparable to a curling ribbon (Fig. 4.2 B). To evaluate whether such mechanical processing of cribellate fibres with the ridges or ripples of the calamistrum is likely, focus was put onto the knobs within the cribellate fibres: Their origin still has to be determined and they might represent the ripples of a curling ribbon (Fig. 4.2). Comparing the cribellate fibres of processed and non-processed threads, no significant difference in the diameter or occurrence of the knots could be determined. Therefore, their occurrence has nothing to do with the action of the calamistrum or the puffy structure of a cribellate thread and cannot be validated as hint for mechanical processing.

Fig. 4.2: Nubbly structure of cribellate fibres. A) Native cribellate fibres showing knots (arrow). TEM image. B) Displacement (arrow) on a nylon thread after contact with a sharp knife. SEM image, coated with gold. C) Knots on cribellate fibres (arrow) look like displacements occur red after contact with a sharp ripple (compare to (B)). SEM image, coated with carbon. *

Processing not necessarily needs to be a mechanical deformation of the surface: Spider silk consists of spidroins, fibrous proteins, which are able to react to e.g. water or to the extraction speed with a conformational change, changing the properties of spider silk (Blamires et al.

* Picture taken by Dr. Alexander Schwedt and Ruth Harscheidt from the central facility for electron microscopy of the RWTH Aachen. 89

2012). Processing could occur on protein level. This hypothesis is most feasible, because the raising of the humidity with the help of a nebulizer lead to a collapse of the puff. Afterwards, cribellate fibre alignment reflected the one in the intermediate zones, respectively the one of non-processed threads. Because removing solely a potential electrostatic charge does not lead to a changing shape, the fibres have to react to the liquid water directly. For spidroins in general it has been described that the proteins deform after contact with water into their “native state”. Water is therefore suggested to remove any post-secretion effect on the protein level (Blamires et al. 2012). Such a protein conformational change would also explain the irreversibility of the structural change, not only observed in this study (Peters 1987). However, if contact with water only results in a structural change of the spidroin conformation, the impaired adhesion force after contact with water does not fit to the described adhesion forces employed by the cribellate thread, i.e. entangling as well as hygroscopic and van-der-Waals forces. Evaluating the possible impact of the calamistrum on the cribellate fibres, it can be excluded that the puffs are formed through a jamming between the axial fibres and that cribellate fibres are kept apart only by electrostatic forces. The puff formation itself is probably enabled by extracting longer cribellate than axial fibres and either a compression of these during combing or a processing of the fibres, leading to autonomous curling.

4.4.5. Contact between calamistrum and cribellate fibres Although it remains unclear, how the cribellate nanofibres are processed in detail, we were able to characterize, where processing of cribellate fibres on the calamistrum occurs. In the Uloboridae U. plumipes as well as the Deinopidae Deinopis sp. cribellate fibres were passing over a similar structured continuous area on the calamistrum. The contact area could be displaced comparing different species, from the tip of the setae as observed in U. plumipes towards the roots of the setae in D. subrufa. This changed the overall morphology of the calamistrum strikingly. The variation of the position of the continuous area was accompanied by a correspondingly changed angle β (angle between opisthosoma, reflecting the position of the calamistrum, and thread). Reconstruction proved a higher, respectively more obtuse angle β leads to contact closer to the basis of the setae. In contrast, a lower and therefore more acute angle β leads to contact closer to the tips of the setae. We observed Deinopis sp. using their third pair of legs to move the whole web during thread production, this way probably controlling the angle β for correct processing of fibres. This matches to an earlier description of D. subrufa’s web building behaviour, entitling the third legs to move constantly through the web during web production, while first and second legs have fixed positions (Clyne 1967).

90

Instead, Uloboridae lower constantly their opisthosoma in addition to adapting their body position during each single combing cycle. The changed morphology of the calamistrum as well as the different strategies for controlling the contact area are not restricted to the studied species. Comparing the calamistra of other Deinopidae like Menneus sp. or Deinopis spinosa (own observations; Coddington et al. 2012; Griswold et al. 2005) with calamistra of other Uloboridae like Z. geniculata, M. animotus, Miagrammopes sp., H. cavatus or producta (own observations; Opell 1982a, 1995, 2001; Peters 1984), it can be observed that the continuous area on the calamistra of Deinopidae is always close to the basis of the setae while the area on the calamistra of Uloboridae is always closer to the tip. Besides Z. geniculata and U. plumipes, an abdominal lowering is described also for other Uloboridae like and U. walckenaerius (Wiehle 1927; Wilder 1874). Due to such an adaptation, spiders loose several micrometre capture thread by having to lower initially their opisthosoma before thread production can start with the correct contact between thread and calamistrum ensured. Peters (1984) already described for U. plumipes that about 600 μm of the following capture thread was not covered with cribellate fibres without any further explanation. Similar observations were made for P. producta (Peters 1995a). In contrast to Uloboridae, Deinopidae are moving the web/thread instead of their whole body. Hence, they do not need such initial lowering of their body and capture thread production starts immediately after fixation. The complete capture thread can be covered with cribellate fibres, enhancing the sticky area of one web (Coddington et al. 1987; Peters 1992c). We suggest this β-dependent adaptation of the calamistrum morphology as well as the strategy of controlling the angle β are conserved within each family and probably reflect different needs during the construction of the web. There are, however, no indications that the differences between the calamistra of Uloboridae and Deinopidae influence the processing of cribellate fibres. In addition, the results give further insight in the importance of correct contact between cribellate fibres and calamistrum for successful processing. With the available data, it is not possible to estimate the general validity of a conserved contact area between cribellate fibres and calamistrum. In other species like the here investigated B. longinqua or H. troglodytes, but also in many other species, a pronounced continuous area is not determinable (Griswold et al. 2005). Furthermore, not all spiders seem to keep up tension within their produced capture thread. For the Filistatidae Filistata and Kukulcania, the Zoropsidae radiata and the Eresidae Stegodyphus it is described that the cribellate fibres rather accumulate behind the spider and sack down to the substrate (Eberhard 1988; Eberhard et al. 1993). Our observations confirmed this accumulation of thread for Eresidae and 91

Zoropsidae. It remains in order to prove, if a fixed contact area on the calamistrum is conserved within the cribellate spiders.

4.4.6. Morphological differences of the calamistra Beside the displaced continuous area on the calamistrum, there were other morphological peculiarities of the calamistrum, like teeth near the tips of the setae of D. subrufa’s, B. longinqua’s and H. troglodytes’ calamistra. Such teeth were suggested to pull out the cribellate fibres, although not all cribellate spiders do have teeth on their calamistra (e.g. Uloboridae (own observation and Foelix (2011)) and the calamistrum itself has proven not to be necessary for fibre extraction. In addition, contact between teeth and cribellate thread during capture thread production should be inhibited at least in Deinopidae, because contact between fibres and calamistrum occurs only at the continuous area. It is unlikely that the teeth have a purpose during the cribellate thread production. Deinopis sp. showed a conspicuous elongated posterior leg movement in the last combing cycle, after which fibres were attached to the calamistrum. Coddington (1986) already described a fixation of the cribellate thread at the fourth leg previous to radial attachment. As Uloboridae adapt their opisthosoma position during each single combing movement to prevent a changed contact, this last elongated leg movement of Deinopis sp. has to have an impact on the area where contact between calamistrum and fibres occur. This elongated movement should lower the angle β, which should in turn change the contact from the continuous area near the basis of the calamistrum to the thinner tips of the setae. This would lead to contact comparable to the one observed in Uloboridae. In contrast to the Uloboridae’s calamistra, the tips of the calamistra of Deinopidae bear teeth. Hence, we suggest these teeth come in contact with the cribellate fibres after the last elongated leg movement, and therefore help picking up the thread and handling it independently of the spinnerets. Our hypothesis is supported comparing the occurrence of such teeth in cribellate spiders: Uloboridae like U. plumipes and Z. geniculata do not have teeth. They do not need to move the thread independently of their spinnerets, because they move their whole body due to their web building behaviour to the attachment site anyway (Eberhard 1972; Peters 1995a; Wiehle 1927; Wilder 1874). For other spider families like the Deinopidae, , Austrochilidae, Eresidae, Hypochilidae or Zoropsidae the occurrence of teeth is described (Foelix et al. 2015a; Foelix et al. 1978; Griswold et al. 2005; Kullmann et al. 1971). These spiders have in common that they do not move constantly (or at all) during thread production (own observations; Clyne 1967; Eberhard 1988; Lopardo et al. 2007; Peters 1992b). Consequently, they have to move the

92 produced thread independently of their spinnerets to its final destination. Admittedly, Eresidae and Zoropsidae, both bearing teeth on their calamistrum, did neither show a picking up of the cribellate thread with the calamistrum, nor an elongated last leg movement (within the available recordings). Furthermore, B. longinqua, H. troglodytes and other species have teeth within the continuous area, so on/near the potential contact area on their calamistrum (Griswold et al. 2005). These contradictions show that although we can provide a plausible hypothesis for the purpose of the teeth, more evidence is needed to verify their function. One should keep in mind that the purpose of these teeth needs not be identical in all cribellate species.

4.5. Model of the cribellate thread production

4.5.1. Establishing a model of the capture thread production for U. plumipes With a frequency of 10 Hz, the rate of stroking for U. plumipes was faster than previously proposed (Kronenberger et al. 2015; Peters 1984; Wilder 1874). The combing leg was moved in an oval manner over the spinnerets, which lead to an uplifting of the calamistrum during anterior movement. This uplifting is likely to prevent contact of the calamistrum with the spinnerets during this phase. Earlier studies already suggested the posterior movement of the combing leg to be the combing movement, where contact with the cribellate fibres occur (Eberhard 1988; Kovoor et al. 1988; Peters 1984, 1992a, c). This study supports this hypothesis. We were able to further specify the combing movement by confirming a deceleration of the combing leg when reaching the cribellum. Such deceleration could indicate the uptake of the cribellate fibres at this position. The combing movement is therefore defined as the period during which the metatarsus of the combing leg moves from the median to the posterior position. This matches our observation that spiders leave the last cycle of the stroke unfinished when interrupting the spinning process. One can assume thread production has already taken place when reaching the posterior position and, as no further thread will be produced, an anterior movement is not necessary any more. We found a linkage between axial and cribellate fibres in the intermediate zones, probably fixing the cribellate mat to the axial fibres. This linkage does not depend on the calamistrum, revealing a major task in highly coordinated spinneret movement. As the puff itself is produced during the combing movement (because the cribellate fibres have to be in contact with the calamistrum) (Fig. 4.3; Fig. 4.4), we had a closer look at the non-combing movement where the intermediate zones are produced. We found a shifted movement of the median spinnerets to the common combing movement. They showed a spreading during the non-combing movement, 93 covering an area as wide as the width of the cribellum. With its elongated spigots, the paracribellum should be able to infiltrate the freshly synthetized cribellate mat when spread apart (Fig. 4.3; Fig. 4.4). During retraction, the spigots of the paracribellum should likewise be able to pull down the fibres towards the axial ones, their spinnerets retracting during this leg movement, too. By bringing all fibres close together, the fibres should be linked to each other (compare chapter 4.2.).

Fig. 4.3: Model of the cribellate thread production shown at three different points in time.* The spinnerets are shown from ventral side of the opisthosoma including a freshly produced thread. For simplification the anterior spinnerets are not shown here. At the “median position” the metatarsus is covering the cribellum, coming in contact with the cribellate fibres. During posterior movement of the metatarsus the cribellate fibres are squeezed or modified otherwise, building one puff (“posterior position”). Moving away from the posterior position, the combing leg is lifted and moved anterior (leg no t shown in this model). During this period, the intermediate zones are produced. The median spinnerets spread and infiltrate the cribellate mat with the elongated spigots of the paracribellum (“spreading of the median spinnerets”). When retracting the median spinnerets, the cribellate mat is tighten and fixated at the axial fibres and the production of one puff is finished (not shown here).

* Picture included in the paper “Joel, A.-C., Kappel, P., Adamova, H., Baumgartner, W., Scholz, I., 2015. Cribellate thread production in spiders: Complex processing of nano-fibres into a functional capture thread. Arthropod Structure & Development 44, 568-573“. Submission was part of the thesis. 94

Fig. 4.4: Lateral overview of the model of the cribellate spinning process. * During the production of the cribellate thread, the metatarsus (mt) bearing the calamistrum performs an oval movement over the spinnerets (arrows), coming in contact with the cribellate fibres (cf) during the posterior movement (combing movement; a to b). Due to this, cribellate fibres are processed and a puff is formed ( b). During the anterior movement (c) of the metatarsus, it is lifted and contact between calamistrum and fibres inhibited. That way, an intermediate zone (iz) is produced. The median spinnerets (ms) show a spreading, extending their elongated spigots of the paracribellum in the extracted cribellate mat. This infiltration with paracribellate fibres (in hatched red) and the synchronously adducting posterior spinnerets (ps) enable the production of linkages in the intermediate zones, finally linking cribellate mat and axial fibres (ax). op: opisthosoma.

4.5.2. Validation of the model Due to similarities in the capture thread structure, for example a puffy structure, the occurrence of linkages and the sheet-like organization of cribellate fibres, it can be assumed that the principle of the capture thread production is conserved in cribellate species. This hypothesis can be supported by finding similar limb movements in other Uloboridae but also Deinopidae, Eresidae, Filistatidae, Psechridae and Zoropsidae: Combing took place during the posterior movement of the fourth legs, indicated by a lowering and deceleration of the metatarsus. For this reason, thread production always started at the anterior position and no retracting anterior leg movement occurred, when capture thread production was interrupted. Furthermore, the median spinnerets showed a spreading during the non-combing movement (observed for Uloboridae and Deinopidae), producing the linkages between cribellate mat and axial fibres according to the model. Likewise, the same posterior spinneret movement as in U. plumipes could be observed in all other species, except for the Filistatidae which showed an aberrant movement and the Eresidae, for which the movement was not determinable due to low resolution. For the Eresidae the abduction of the posterior spinnerets has already been described (Peters 1992a).

* Picture included in the paper “Joel, A.-C., Scholz, I., Orth, L., Kappel, P., Baumgartner, W., 2016. Morphological adaptation of the calamistrum to the cribellate spinning process in Deinopoidae (Uloboridae, Deinopidae). Royal Society Open Science 3“. Submission was part of this thesis. 95

Finding the main features characterizing the cribellate spinning process of U. plumipes conserved in other Uloboridae, but also Deinopidae, Eresidae, Psechridae, Zoropsidae and (if excluding the posterior spinneret movement) Filistatidae, the previously proposed model can be assumed to be valid in all cribellate species. This finding backs up the hypothesis that the production of cribellate threads is a primordial and conserved principle of some spiders (Bond et al. 2014; Fernandez et al. 2014). Nevertheless, some differences in thread structure, e.g. the looped structure of the capture thread of Filistatidae, cannot be explained if assuming no differences in thread production. Filistatidae though show an asynchronous posterior spinneret movement. According to the model, linkages between cribellate mat an axial fibres can only occur when posterior spinnerets and median spinnerets are brought together in close proximity. Therefore, a twelve times slowed down movement with only one of the posterior spinneret in close proximity to the median spinnerets at a time leads to less linkages. Such reduced number linkages would lead to a more loosely mat surrounding the axial fibres and probably looping around those. Depicted cribellate threads with looped structure replicate exactly these features (Opell 2002). Because the assumption that the thread is fixated at a supporting line to establish a looping structure, is not consistent with the observable behaviour within the recordings, the hypothesis should be corrected to fewer linkages between axial fibres and cribellate mat (Opell 1999, 2002). Beside the asynchronous posterior spinneret movement of Filistatidae, smaller differences, like the number of cribellate spigots or the frequency of combing, were observable between species. For adult U. plumipes the extraction speed was approximately constant during the production of the capture spiral. The other two phases of one stroke showed a higher variability. Because less than 20% of the time of one stroke is needed for these two phases, the frequency is mainly determined by the velocity of the combing movement. One can assume that the extraction of nanofibres needs a lot of energy. Extracting more fibres would mean more effort and consequentially lower the extraction speed and therefore also the frequency. This theoretical correlation can be reproduced, plotting literature data for spigot number against the frequency (Fig. 4.5 A). Keep in mind that frequency acquisition was performed with online available recordings and number of spigots increase with each moult not only in Uloboridae but also in Filistatidae and Eresidae (Kullmann et al. 1971; Lopardo et al. 2007; Opell 1989, 1995). Neither the number of spigots nor the recordings are matched to a developmental status.

96

Fig. 4.5: The influence on and of the combing frequency. A) Frequency of different species (Deinopis sp., E. walckenaeri, K. hibernalis, S. mirandus, U. plumipes, Z. spinimana, Z. geniculata) validated from online available recordings (except for U. plumipes and Z. geniculata) compared to the spigot number of their cribellum. A higher number of spigots means lower frequency. Note the light grey indicates individuals were excluded from the line of best fit. Data for these extracted from: Foelix et al. 2015a; Foelix et al. 1978; Kullmann 1968; Kullmann et al. 1971; Peters 1982, 1992c . Note that data for K. hibernalis are taken from a Filistata, because no data for K. hibernalis could be found. Furthermore, data used for S. mirandus is the mean of Stegodyphus sarasinorum and Stegodyphus pacificus. B) The influence on the frequency of the number of puffs per millimetre is depicted. A higher frequency means more puffs per millimetre. Data were extracted by measuring pictures from: Griswold et al. 2005; Opell 2002; Peters 1992a, c.

During the development of Uloboridae, spigot number was already associated with puff width and length (Opell 1989) (compare chapter 4.3.). As the frequency seems to depend on spigot number as well, the frequency should also be linked to the thread shape. Such correlation could be detected, when comparing the frequency with the number of puffs per millimetre. A higher frequency was associated with a higher number of puffs per millimetre (Fig. 4.5 B). If every spider would move with the same velocity during thread extraction, there should be a linear relationship. Hence, although a correlation can be detected, one should keep in mind that other factors beside the frequency have to influence the number of puffs per millimetre as well. Although most features characterizing the cribellate thread can be explained with the newly established model, there are distinctions within the thread production which are not yet incorporated. The undulating fibres (paracribellate fibres) of e.g. Desidae, as well as morphological differences in the paracribellum need to be brought into the context of the model in further studies.

97

4.6. Summary Finding the main features of the cribellate thread production conserved within the observed spider species, the results of this thesis can be presented best by summarizing the whole cribellate thread production process in an up to date model and influence on thread structure. Keep in mind that some aspects already integrated in this model need further evaluation to validate this model. All aspects of this model except the influence of cribellum’s spigot numbers on the thread width are a result of this thesis. Starting the cribellate capture thread production, the spider places the calamistrum of the combing leg anterior of the spinnerets, stabilizing the leg by forming a unit with the other fourth leg serving as supporting one. A stroke starts by moving the calamistrum posterior. When reaching the cribellum, contact between cribellate fibres and calamistrum occurs at the continuous area of the calamistrum. The contact occurs always in this area, entailing a behavioural adaption of the spider by e.g. adapting the body position to prevail an elongating thread. The number of spigot influences the highest possible extraction speed, decreasing the speed with increasing number of spigots. Because 80% of the time needed for one stroke is spend during the combing movement (i.e. fibre extraction), the number of spigots influences the frequency of combing in each species. Although the calamistrum helps also extracting the fibres, contact with the calamistrum modifies mainly the cribellate fibres to form the puff of the cribellate thread. The puffs are not established by electrostatically charging the nanofibres, but probably by extracting cribellate fibres longer than the supporting axial fibres and perhaps also by influencing the spidroin conformation, enabling a curling of fibres. The puffs width is influenced by the number of fibres extracted. After reaching the most posterior position, the movement of the calamistrum is altered into the anterior direction, inhibiting contact with the fibres by lifting the whole combing unit. Due to this inhibited contact, the intermediate zones are produced. Within this movement, the median spinnerets start spreading, infiltrating the cribellate mat with paracribellate fibres due to their elongated spigots. When retracting them again, the posterior spinnerets are retracted as well. The resulting proximity of paracribellate fibres and axial fibres enables a linkage of both, linking finally the cribellate mat to the axial fibres. If the posterior spinnerets are not retracted during this phase and therefore are not in proximity to the median spinnerets, fewer linkages are established and the mat loops arbitrarily around the axial fibre. Having the mat of cribellate fibres extracted, puffed and linked to the axial fibres, the cribellate fibres enclose the axial fibres. This enclosure is facilitated by tension generated in the outer convoluted cribellate fibres of a puff and a twisting of the thread by the spider prior to attaching it. The process of capture thread production can now start over again, 98 producing the next puff. Spiders with higher combing frequency produce more puffs per millimetre than spiders with lower frequency. On the contrary, a higher number of spigots leads to wider puffs. After thread production is finished, some spiders transport the capture thread to the target location using their legs. Although some spiders use the tarsus for this transportation, other spiders can pick up the cribellate thread with their calamistrum supported by a specialized teeth- like structure.

4.7. Outlook Although the model of the cribellate thread production presented within this thesis has incorporated the influence of most morphological differences, a more detailed study including a broader range of cribellate species should be performed to validate the model. For example the relation between frequency, cribellate spigot number and thread shape (i.e. puff per millimetre and puff width) should be evaluated by comparing data of the three categories taken from adult spiders. Likewise, the proposed hypothesis for the influence of the asynchronous posterior spinneret movement of K. hibernalis on its thread structure needs further investigation to be verified or falsified. Another focus should be on the morphological differences of spinnerets and their influence on thread structuring, because this aspect has not been included in the model so far. Future studies should also focus on the processing of the cribellate fibres with the calamistrum. It is not yet determined how the puffy structure is (and stays) established. Related to these studies, one should examine the general production and extraction of the nanofibres in cribellate spiders. Technical nanofibre production is mainly linked to high tension within the electrospinning (Huang et al. 2003; Teo et al. 2006), but a spider is able to produce nanofibres without any high tension. This biological model could be an inspiration for future technical trends. Not only the production of the nanofibres in adult spiders can be examined, but also the nanofibre production of 2nd instars should be explored. The origin as well as the function of these fibres are not resolved yet, because they are lacking a cribellum, a paracribellum, the calamistrum and do neither catch nor consume prey. The idea to biomimetically abstract the spinning process itself and to transfer the fibre production to an artificial process is not new (Kronenberger et al. 2015). Beside the production of nanofibres, the newly described processing of nanofibres as well as the handling of

99 nanofibres with the calamistrum are likewise very interesting for a biomimetic reproduction.* Within technical applications, the handling of nanofibres is a highly investigated field. Nevertheless, solutions for occurring problems like the generally good adhesion properties of nanofibres, which interfere with the handling of fibres, are not yet at hand (Baji et al. 2015). To simplify such handling, technical nanofibres are already transferred to larger supporting fibres, which are then used as transportation device. This is similar to the process observed in cribellate spiders. However, the technical nanofibres often adhere better to the tools than to the supporting fibres. Cribellate spiders overcome this problem with two different mechanisms: First, they link the cribellate fibres to the axial ones with a third type of fibres. Secondly, their calamistrum has antiadhesive properties towards nanofibres. To link stably the nanofibres to a supporting fibre comparable to the process observed in the biological model, the technique employed by the spiders needs further evaluation. From this point of view, the spinneret movement as well as the involved spigots and glands should be more closely analysed. Also the antiadhesive properties of the calamistrum need further investigation previous to a transfer to a technical tool. It has to be determined whether the structure of the calamistrum and/or a coating lead to antiadhesive properties and/or whether this combination is only working with natural or also with technical nanofibres. Such “non grip” between tools and nanofibres would also entail some disadvantages e.g. for spooling fibres one has to grip them first. Even in this case, cribellate spiders could be likewise used as an example to solve this problem. A teeth-like structure is suggested to enable picking up and transportation of fibres, but function of these teeth has not been proven so far and might differ between species. Therefore, the teeth’s function has to be corroborated. Although cribellate threads can be used for biomimetic inspirations, there are also several mainly biological features non-described yet. Especially the thread’s adhesive mechanism should be examined more closely, because the assembly of the cribellate fibres to one single layer has not been described previously. This might serve another not yet identified adhesion mechanism of cribellate fibres.

* The ideas about the biomimetic transfer were discussed and/or developed with Magnus Kruse, M.Sc. from the Intitut für Textiltechnik of the RWTH Aachen. 100

Literature

Andrade, M.C.B., Kasumovic, M.M., 2005. Terminal investment strategies and male mate choice: Extreme tests of Bateman. Integrative and Comparative Biology 45, 838-847. Apstein, C., 1889. Bau und Function der Spinndrüsen der Araneida. Archiv für Naturgeschichte 55, 29-74. Arnott, G., Elwood, R.W., 2009. Assessment of fighting ability in animal contests. Animal Behaviour 77, 991-1004. Askarieh, G., Hedhammar, M., Nordling, K., Saenz, A., Casals, C., Rising, A., Johansson, J., Knight, S.D., 2010. Self-assembly of spider silk proteins is controlled by a pH-sensitive relay. Nature 465, 236-238. Bai, H., Ju, J., Zheng, Y.M., Jiang, L., 2012. Functional fibers with unique wettability inspired by spider silks. Advanced Materials 24, 2786-2791. Baji, A., Zhou, L.M., Mai, Y.W., Yang, Z.F., Yao, H.M., 2015. On the Adhesion performance of a single electrospun fiber. Applied Physics a-Materials Science & Processing 118, 51-56. Barth, F.G., 1982. Spiders and vibratory signals: sensory reception and behavioural signigicance, Spider communication: Mechanisms and ecological significance. Princeton Univ. Pr., Princeton, NJ, pp. 67-122. Barth, F.G., 1986. Vibration sense and vibratory umwelt of spiders. Naturwissenschaften 73, 519-530. Barth, F.G., 1997. Vibratory communication in spiders: Adaptation and compromise at many levels, In: Lehrer, M. (Ed.), Orientation and Communication in . Birkhäuser Basel, pp. 247-272. Barth, F.G., 1998. The vibrational sense of spiders, In: Hoy, R., Popper, A., Fay, R. (Eds.), Comparative Hearing: Insects. Springer New York, pp. 228-278. Baurecht, D., Barth, F.G., 1993. Vibratory communication in spiders II Representation of parameters contained in synthetic male courtship signals by female vibration receptor. Journal of Comparative Physiology A 173, 309-319. Bertkau, P., 1882. Über das Cribellum und Calamistrum: Ein Beitrag zur Histiologie, Biologie und Systematik der Spinnen. Archiv für Naturgeschichte 48, 316-362. Blackledge, T.A., 2012. Spider silk: a brief review and prospectus on research linking biomechanics and ecology in draglines and orb webs. Journal of 40, 1-12. Blamires, S.J., Wu, C.-L., Blackledge, T.A., Tso, I.-M., 2012. Post-secretion processing influences spider silk performance. Journal of the Royal Society Interface 9, 2479-2487. Bond, J.E., Garrison, N.L., Hamilton, C.A., Godwin, R.L., Hedin, M., Agnarsson, I., 2014. Phylogenomics resolves a spider backbone phylogeny and rejects a prevailing paradigm for orb web evolution. Current Biology 24, 1765-1771. Bond, J.E., Opell, B.D., 1997. The functional significance of medially divided cribella in the spider Mallos (Araneae, Dictynidae). Bulletin of the British Arachnological Society 10, 239-241. Bourzac, K., 2015. Web of intrigue. Nature 519, S4-S6. Bradoo, B.L., 1986. Feeding behaviour of a non-poisonous spider Uloborus ferokus Bradoo (Araneae: Uloboridae). Zoologischer Anzeiger 217, 75-88. Briceno, R.D., Eberhard, W.G., 2012. Spiders avoid sticking to their webs: clever leg movements, branched drip-tip setae, and anti-adhesive surfaces. Naturwissenschaften 99, 337- 341. Bridge, A.P., Elwood, R.W., Dick, J.T.A., 2000. Imperfect assessment and limited information preclude optimal strategies in male-male fights in the orb-weaving spider Metellina mengei. Proceedings of the Royal Society B: Biological Sciences 267, 273-279.

101

Buskirk, R.E., 1975. Aggressive display and orb defence in a colonial spider, Metabus gravidus. Animal Behaviour 23, 560-567. Christenson, T.E., Goist, K.C., 1979. Cost and benefits of male-male competitions in the orb weaving spider, Nephila clavipes. Behavioral Ecology and Sociobiology 5, 87-92. Clyne, D., 1967. Notes on the construction of the net and sperm-web of a cribellate spider Dinopis subrufus (Koch) (Aranieda: Dinopidae). Australian Zoologist 14, 189-198. Coddington, J.A., 1986. Orb webs in "non-orb-weaving" ogre-faces spiders (Araneae: Dinopidae): A question of Geneoalogy. Cladistics 2, 53-67. Coddington, J.A., Kuntner, M., Opell, B.D., 2012. Systematics of the spider family Deinopidae with a revision of the genus Menneus. Smithsonian Contributions to Zoology, 1- 61. Coddington, J.A., Sobrevila, C., 1987. Web manipulation and two stereotypes attack behaviors in the ogre-faced spider Deinopis spinosus Marx (Araneae, Deinopidae). Journal of Arachnology 15, 213-225. Craig, C.L., Bernard, G.D., 1990. Insect attraction to ultraviolet-reflecting spider webs and web decorations. Ecology 71, 616-623. Dawson, I., 2001. Finding Uloborus plumipes. British Arachnological Society 40, 2-3. Dodson, G.N., Schwaab, A.T., 2001. Body size, leg autotomy, and prior experience as factors in the fighting success of male crab spiders, Misumenoides formosipes. Journal of Insect Behavior 14, 841-855. Draud, M., Macias-Ordonez, R., Verga, J., Itzkowitz, M., 2004. Female and male Texas cichlids (Herichthys cyanoguttatum) do not fight by the same rules. Behavioral Ecology 15, 102-108. Eberhard, W.G., 1971. Ecology of the web of Uloborus diversus (Araneae: Uloboridae). Oecologia 6, 328-342. Eberhard, W.G., 1972. Web of Uloborus diversus (Araneae, Uloboridae). Journal of Zoology 166, 417-465. Eberhard, W.G., 1977. Webs of newly emerged Uloborus diversus and of a male Uloborus sp. (Araneae: Uloboridae). Journal of Arachnology 4, 201-206. Eberhard, W.G., 1980. Persistent stickiness of cribellum silk. Journal of Arachnology 8, 283- 283. Eberhard, W.G., 1988. Combing and sticky silk attachment behaviour by cribellate spiders and its taxonomic implications. Bulletin of the British Arachnological Society 7, 247-251. Eberhard, W.G., 2010. Possible functional significance of spigot placement on the spinnerets of spiders. Journal of Arachnology 38, 407-414. Eberhard, W.G., Pereira, F., 1993. Ultrastructure of cribellate silk of nine species in eight families and possible taxonomic implications (Araneae: Amaurobiidae, Deinopidae, Desidae, Dictynidae, Filistatidae, Hypochilidae, , ). Journal of Arachnology 21, 161-174. Eisoldt, L., Smith, A., Scheibel, T., 2011. Decoding the secrets of spider silk. Materials Today 14, 80-86. Elettro, H., Neukirch, S., Antkowiak, A., Vollrath, F., 2015. Adhesion of dry and wet electrostatic capture silk of uloborid spider. Naturwissenschaften 102, 41. Elias, D.O., Botero, C.A., Andrade, M.C.B., Mason, A.C., Kasumovic, M.M., 2010. High resource valuation fuels "desperado" fighting tactics in female jumping spiders. Behavioral Ecology 21, 868-875. Elias, D.O., Kasumovic, M.M., Punzalan, D., Andrade, M.C.B., Mason, A.C., 2008. Assessment during aggressive contests between male jumping spiders. Animal Behaviour 76, 901-910. Elias, D.O., Mason, A.C., Maddison, W.P., Hoy, R.R., 2003. Seismic signals in a courting male (Araneae : Salticidae). Journal of Experimental Biology 206, 4029-4039. 102

Enders, F., 1974. Vertical stratification in orb-web spiders (Araneidae, Araneae) and a consideration of other methods of coexistence. Ecology 55, 317-328. Enquist, M., Leimar, O., 1983. Evolution of fighting behaviour: Decision rules and assessment of relative strength. Journal of Theoretical Biology 102, 387-410. Faber, D.B., Baylis, J.R., 1993. Effects of body size on agonistic encounters between male jumping spiders (Araneae: Salicidae). Animal Behaviour 45, 289-299. Fernandez, R., Hormiga, G., Giribet, G., 2014. Phylogenomic analysis of spiders reveals nonmonophyly of orb weavers. Current Biology 24, 1772-1777. Ferretti, N.E., Perez-Miles, F., 2011. Intraspecific non-sexual interactions of Grammostola schulzei (Araneae: Theraphosidae) under laboratory conditions. Revista De Biologia Tropical 59, 1173-1182. Foelix, R.F., 2011. Biology of spiders, 3rd ed. Oxford University Press, New York. Foelix, R.F., 2015. Biologie der Spinne, 3rd ed. Chimaira Buchhandelgesellschaft mbH, Frankfurt am Main. Foelix, R.F., Erb, B., Eggs, B., 2015a. Morphologische Besonderheiten der Kräuseljagdspinne Zoropsis spinimana. Arachne 3, 4-14. Foelix, R.F., Joel, A.-C., Erb, B., 2015b. Wie die Kräuselradnetzspinne Uloborus ihre Beute einwickelt und verdaut. Arachne 2, 16-23. Foelix, R.F., Jung, H., 1978. Some anatomical aspects of with special reference to the calamistrum and cribellum, Symposia of the Zoological Society of London. Academic Press, pp. 417-422. Fowler-Finn, K.D., Hebets, E.A., 2006. An examination of agonistic interactions in the whip spider Phrynus marginemaculatus (Arachnida, Amblypygi). Journal of Arachnology 34, 62-76. Friedrich, V., Langer, R.M., 1969. Fine structure of cribellate spider silk. American Zoologist 9, 91-96. Garrison, N.L., Rodriguez, J., Agnarsson, I., Coddington, J.A., Griswold, C.E., Hamilton, C.A., Hedin, M., Kocot, K.M., Ledford, J.M., Bond, J.E., 2016. Spider phylogenomics: untangling the Spider Tree of Life. PeerJ 4, e1719. Girard, M.B., Kasumovic, M.M., Elias, D.O., 2011. Multi-modal courtship in the peacock spider, Maratus volans (OP-Cambridge, 1874). Plos One 6, 1-10. Gong, G.M., Zhou, C., Wu, J.T., Jin, X., Jiang, L., 2015. Nanofibrous adhesion: The twin of gecko adhesion. Acs Nano 9, 3721-3727. Gorb, S.N., Landolfa, M.A., Barth, F.G., 1998. Dragline-associated behaviour of the orb web spider Nephila clavipes (Araneoidea, Tetragnathidae). Journal of Zoology 244, 323-330. Griswold, C.E., Ramirez, M.J., Coddington, J.A., Platnick, N.I., 2005. Atlas of phylogenetic data for entelegyne spiders (Araneae: : ) with comments on their phylogeny. Proceedings of the California Academy of Sciences 56, 1. Hack, M.A., 1997. Assessment strategies in the contests of male crickets, Acheta domesticus (L). Animal Behaviour 53, 733-747. Hack, M.A., Thompson, D.J., Fernandes, D.M., 1997. Fighting in males of the autumn spider, : Effects of relative body size, prior residency and female value on contest outcome and duration. Ethology 103, 488-498. Hajer, J., 1991. Notes on the spinning apparatus of the spiders Hyptiotes paradoxus C.L.K., 1834, and Uloborus walckenaerius Latr., 1806 (Araneae: Uloboridae). Bulletin de la Société neuchâteloise des sciences naturelles 116, 99-103. Hawthorn, A.C., Opell, B.D., 2002. Evolution of adhesive mechanisms in cribellar spider prey capture thread: evidence for van der Waals and hygroscopic forces. Biological Journal of the Linnean Society 77, 1-8. Hawthorn, A.C., Opell, B.D., 2003. van der Waals and hygroscopic forces of adhesion generated by spider capture threads. Journal of Experimental Biology 206, 3905-3911.

103

Heim, M., Keerl, D., Scheibel, T., 2009. Spider silk: From soluble protein to extraordinary fiber. Angewandte Chemie-International Edition 48, 3584-3596. Hinman, M.B., Jones, J.A., Lewis, R.V., 2000. Synthetic spider silk: a modular fiber. Trends in Biotechnology 18, 374-379. Hodge, M.A., 1987. Agonistic interactions between female bowl and doily spiders (Araneae, ): owner biased outcomes. Journal of Arachnology 15, 241-247. Hodge, M.A., Uetz, G.W., 1995. A comparison of agonistic behavior of colonial web-building spiders from desert and tropical habitats. Animal Behaviour 50, 963-972. Hoffmaster, D.K., 1986. Aggression in tropical orb-weaving spiders a quest for food? Ethology 72, 265-276. Horton, C.C., Wise, D.H., 1983. The experimental analysis of competition between two syntopic species of orb-web spiders (Araneae: Araneidae). Ecology 64, 929-944. Hou, Y.P., Chen, Y., Xue, Y., Zheng, Y.M., Jiang, L., 2012. Water Collection Behavior and Hanging Ability of Bioinspired Fiber. Langmuir 28, 4737-4743. Hsu, Y.Y., Wolf, L.L., 1999. The winner and loser effect: integrating multiple experiences. Animal Behaviour 57, 903-910. Huang, Z., Liao, X., Yin, G., Kang, Y., Yao, Y., 2009. Segmented nanofibrils of spiral silk in Uloborus walckenaerius spider. The Journal of Physical Chemistry B 113, 5092-5097. Huang, Z.M., Zhang, Y.Z., Kotaki, M., Ramakrishna, S., 2003. A review on polymer nanofibers by electrospinning and their applications in nanocomposites. Composites Science and Technology 63, 2223-2253. Huber, B.A., Fleckenstein, N., 2008. Comb-hairs on the fourth tarsi in pholcid spiders (Araneae, Pholcidae). Journal of Arachnology 36, 232-240. Jackson, R.R., 1992. Predator-prey interactions between web-invading jumping spiders a web- building spider, Holocnemus pluchei (Araneae, Pholcidae). Journal of Zoology 228, 589-594. Jennings, D.J., Gammell, M.P., Carlin, C.M., Hayden, T.J., 2004. Effect of body weight, antler length, resource value and experience on fight duration and intensity in fallow deer. Animal Behaviour 68, 213-221. Jonsson, L., 1998. Toftspideln-en spidelart som invaderat Europas växthus [The feather-legged spider, Uloborus plumipes-invading European greenhouses]. Fauna och Flora 93, 119-124. Kasumovic, M.M., Mason, A.C., Andrade, M.C.B., Elias, D.O., 2011. The relative importance of RHP and resource quality in contests with ownership asymmetries. Behavioral Ecology 22, 39-45. Klarner, D., Barth, F.G., 1982. Vibratory signals and prey capture in orb-weaving spiders (Zygiella x-notata, Nephila clavipes, Araneidae). Journal of Comparative Physiology 148, 445- 455. Klein, W., Stock, M., Wunderlich, J., 1995. Zwei nach Deutschland eingeschleppte Spinnenarten (Araneae)–Uloborus plumipes (Lucas) und Eperigone eschatologica (Bishop)–als Gegenspieler der weißen Fliege im geschützten Zierpflanzenbau. Beiträge zur Araneologie 4, 301-306. Kovoor, J., 1978. Silk glands in genus Uloborus latreille (Araneae, Uloboridae) 2. Histochemical data. Annales Des Sciences Naturelles-Zoologie Et Biologie Animale 20, 3-25. Kovoor, J., Peters, H.M., 1988. The spinning apparatus of Polenecia producta (Araneae, Uloboridae): Structure and histochemistry. Zoomorphology 108, 47-59. Krafft, B., Cookson, L.J., 2012. The Role of Silk in the Behaviour and of Spiders. Psyche 2012, 1-25. Kronenberger, K., Vollrath, F., 2015. Spiders spinning electrically charged nano-fibres. Biol Lett 11, 20140813. Kropf, C., Bauer, D., Schlappi, T., Jacob, A., 2012. An organic coating keeps orb-weaving spiders (Araneae, Araneoidea, Araneidae) from sticking to their own capture threads. Journal of Zoological Systematics and Evolutionary Research 50, 14-18. 104

Kullmann, E., 1968. Social phenomena in spiders. Insectes Sociaux 15, 289-298. Kullmann, E., Bach, K., Burkhardt, B., Graefe, R., Raccanello, R., 1975. The production and function of spider threads and spider webs. Netze in Natur und Technik, 318-387. Kullmann, E., Nawabi, S., Zimmermann, W., 1971. Neue Ergebnisse zur Brutbiologie cribellater Spinnen aus und der Serengeti (Araneae, Eresidae). Zeitschrift des Kölner Zoos 14, 87-108. Kullmann, E.J., 1972. The convergent development of orb-webs in cribellate and ecribellate spiders. American Zoologist 12, 395-405. Kumhof, E., Sengonca, C., Montagne, E., Deut Gesell, A., Angewandte, E., 1992. Laboratory studies on the development and feeding-activity of Uloborus plumipes Lucas (Araneae, Uloboridae). Mitteilungen der Deutschen Gesellschaft für Allgemeine und Angewandte Entomologie 8, 204-208. Lahmann, E.J., Eberhard, W.G., 1979. Selective factors that affect the tendency to form groups in the colonial spider semiplumosa (Araneae, Uloboridae). Revista De Biologia Tropical 27, 231-240. Lehmensick, R., Kullmann, E., 1957. Über den Feinbau der Fäden einiger Spinnen. Zoologischer Anzeiger, Supplementband 20, 123-129. Lopardo, L., Ramirez, M.J., 2007. The combing of cribellar silk by the prithine Misionella mendensis, with notes on other filistatid spiders (Araneae : Filistatidae). American Museum Novitates, 1-14. Lopardo, L., Ramirez, M.J., Grismado, C., Compagnucci, L.A., 2004. Web building behavior and the phylogeny of austrochiline spiders. Journal of Arachnology 32, 42-54. Lopez, P., Martin, J., 2001. Fighting rules and rival recognition reduce costs of aggression in male lizards, Podarcis hispanica. Behavioral Ecology and Sociobiology 49, 111-116. Lubin, Y.D., 1986. Web building and prey capture in the Uloboridae. Stanford University Press, Stanford, USA. Maklakov, A.A., Bilde, T., Lubin, Y., 2003. Vibratory courtship in a web-building spider: signalling quality or stimulating the female? Animal Behaviour 66, 623-630. Marples, B., 1962. Notes on spiders of the family Uloboridae. Annals of Zoology (Agra) 4, 1- 11. Masters, W.M., Markl, H., 1981. Vibration signal transmission in spider orb webs. Science 213, 363-365. Masumoto, T., 1998. Cooperative prey capture in the communal web spider, Philoponella raffrayi (Araneae, Uloboridae). Journal of Arachnology 26, 392-396. McGinley, R.H., Prenter, J., Taylor, P.W., 2013. Whole-organism performance in a jumping spider, Servaea incana (Araneae: Salticidae): links with morphology and between performance traits. Biological Journal of the Linnean Society 110, 644-657. Miller, J.A., Griswold, C.E., Scharff, N., Rezac, M., Szuts, T., Marhabaie, M., 2012. The velvet spiders: an atlas of the Eresidae (Arachnida, Araneae). Zookeys, 1-144. Miyashita, T., 1992. Food limitation of population density in the orb-web spider, Nephila clavata. Researches on Population Ecology 34, 143-153. Miyashita, T., 2001. Competition for a limited space in kleptoparasitic Argyrodes spiders revealed by field experiments. Population Ecology 43, 97-103. Mortimer, Vollrath, F., 2015. Diversity and properties of key spider silks and webs. Research & Knowledge 1, 32-43. Neat, F.C., Huntingford, F.A., Beveridge, M.M.C., 1998. Fighting and assessment in male cichlid fish: the effects of asymmetries in gonadal state and body size. Animal Behaviour 55, 883-891. Nelson, X.J., Fijn, N., 2013. The use of visual media as a tool for investigating animal behaviour. Animal Behaviour 85, 525-536.

105

Nentwig, W., Heimer, S., 1987. Ecological Aspects of Spider Webs, In: Nentwig, W. (Ed.), Ecophysiology of Spiders. Springer-Verlag, Berlin Heidelberg Olsson, M., 1994. Rival recognition affects male contest behavior in sand lizards (Lacerta agilis). Behavioral Ecology and Sociobiology 35, 249-252. Opell, B.D., 1979. Revision of the genera and tropical american species of the spider family Uloboridae. Bulletin of the Museum of Comparative Zoology 148, 443-549. Opell, B.D., 1982a. Cribellum, calamistrum, and ventral comb ontogeny in Hyptiotes cavatus (Hentz)(Araneae: Uloboridae). Bulletin of the British Arachnological Society 5, 338-343. Opell, B.D., 1982b. Post-hatching development and web production of Hyptiotes cavatus (Hentz) (Araneae, Uloboridae). Journal of Arachnology 10, 185-191. Opell, B.D., 1989. Functional associations between the cribellum spinning plate and capture threads of Miagrammopes animotus (Araneida, Uloboridae). Zoomorphology 108, 263-267. Opell, B.D., 1990. Material investment and prey capture potential of reduced spider webs. Behavioral Ecology and Sociobiology 26, 375-381. Opell, B.D., 1993. What forces are responsible for the stickiness of spider cribellar threads? Journal of Experimental Zoology 265, 469-476. Opell, B.D., 1994a. The ability of spider cribellar prey capture thread to hold insects with different surface-features. Functional Ecology 8, 145-150. Opell, B.D., 1994b. Factors affecting the diameter of axial fibres in cribellar threads of the spider family Uloboridae. Journal of Arachnology 22, 12-18. Opell, B.D., 1994c. Factors governing the stickiness of cribellar prey capture threads in the spider family Uloboridae. Journal of Morphology 221, 111-119. Opell, B.D., 1995. Ontogenetic changes in cribellum ppigot number and cribellar prey capture thread stickiness in the spider family Uloboridae. Journal of Morphology 224, 47-56. Opell, B.D., 1999. Changes in spinning anatomy and thread stickiness associated with the origin of orb-weaving spiders. Biological Journal of the Linnean Society 68, 593-612. Opell, B.D., 2001. Cribellum and calamistrum ontogeny in the spider family uloboridae: Linking functionally related but separate silk spinning features. Journal of Arachnology 29, 220-226. Opell, B.D., 2002. How and thread configuration shape the stickiness of cribellar prey capture threads. Journal of Arachnology 30, 10-19. Opell, B.D., 2013. Cribellar thread, In: Nentwig, W. (Ed.), Spider Ecophysiology. Springer- Verlag Berlin Heidelberg, pp. 303-315. Opell, B.D., Bond, J.E., 2000. Capture thread extensibility of orb-weaving spiders: testing punctuated and associative explanations of character evolution. Biological Journal of the Linnean Society 70, 107-120. Opell, B.D., Bond, J.E., Warner, D.A., 2006. The effects of capture spiral composition and orb-web orientation on prey interception. Zoology (Jena) 109, 339-345. Opell, B.D., Schoener, T.W., Keen, S.L., Davies, V.T., 1995. The new species Purumitra australiensis (Araneae, Uloboridae) with notes on its natural history. Journal of Arachnology 23, 127-129. Opell, B.D., Schwend, H.S., 2008. Persistent stickiness of viscous capture threads produced by araneoid orb-weaving spiders. Journal of Experimental Zoology Part a-Ecological Genetics and Physiology 309A, 11-16. Opell, B.D., Schwend, H.S., 2009. Adhesive efficiency of spider prey capture threads. Zoology (Jena) 112, 16-26. Payne, R.J.H., 1998. Gradually escalating fights and displays: the cumulative assessment model. Animal Behaviour 56, 651-662. Peaslee, J.E., Peck, W.B., 1983. The biology of Octonoba octonarius (Muma) (Araneae, Uloboridae). Journal of Arachnology 11, 51-67.

106

Peters, H.M., 1982. Wie Spinnen der Familie Uloboridae ihre Beute einspinnen und verzehren. Verhandlungen des Naturwissenschaftlichen Vereins in Hamburg 25, 147-167. Peters, H.M., 1983. Structure and Formation of the Tentacles of cribellate spiders arachnida araneae. Verhandlungen des Naturwissenschaftlichen Vereins Hamburg, 241-254. Peters, H.M., 1984. The spinning apparatus of Uloboridae in relation to the structure and construction of capture threads (Arachnida, Araneida). Zoomorphology 104, 96-104. Peters, H.M., 1987. Fine structure and function of capture threads, In: Nentwig, W. (Ed.), Ecophysiology of Spiders. Springer-Verlag Berlin Heidelberg, pp. 187-202. Peters, H.M., 1992a. On structure and construction of capture threads by cribellate spiders of the Eresidae family (Arachnida, Araneae). Verhandlungen des Naturwissenschaftlichen Vereins in Hamburg 33, 213-227. Peters, H.M., 1992b. On the burrowing behaviour and the production and use of silk in Seothyra, a sand-inhabiting spider from the Namib Desert (Arachnida, Araneae, Eresidae). Verhandlungen des Naturwissenschaftlichen Vereins in Hamburg 33, 191-211. Peters, H.M., 1992c. On the spinning apparatus and the structure of the capture threads of Deinopis subrufus (Araneae, Deinopidae). Zoomorphology 112, 27-37. Peters, H.M., 1995a. Polencia producta and its web - structure and evolution (Araneae, Uloboridae). Zoomorphology 115, 1-9. Peters, H.M., 1995b. Ultrastructure of orb spiders' gluey capture threads. Naturwissenschaften 82, 380-382. Peters, H.M., Kovoor, J., 1980. A complement to the spinning apparatus in Uloboridae (Araneae): the paracribellum and its glands. Zoomorphology 96, 91-102. Polis, G.A., 1981. The evolution and dynamics of intraspecific . Annual Review of Ecology and Systematics 12, 225-251. Reukauf, E., 1930. Zur Biologie von Hyptiotes paradoxus. Zeitschrift für Morphologie und Ökologie der Tiere 21, 691-701. Riechert, S.E., 1978. Games spider play: behavioral variability in territorial disputes. Behavioral Ecology and Sociobiology 3, 135-162. Riechert, S.E., 1981. The consequences of being territorial: spiders, a case study. American Naturalist 117, 871-892. Rising, A., Johansson, J., 2015. Toward spinning artificial spider silk. Nature Chemical Biology 11, 309-315. Romer, L., Scheibel, T., 2008. The elaborate structure of spider silk Structure and function of a natural high performance fiber. Prion 2, 154-161. Rovner, J.S., 1968. Territoriality in the sheet-web spider triangularis (Clerck) (Aranease, Linyphiidae). Zeitschrift fur Tierpsychologie 25, 232-242. Rypstra, A.L., 1983. The importance of food and space in limiting web-spider densities; a test using field enclosures. Oecologia 59, 312-316. Saffre, F., Mailleux, A.C., Deneubourg, J.L., 1999. Dragline attachment pattern in the neotropical social spider Anelosimus eximius (Araneae : ). Journal of Insect Behavior 12, 277-282. Sahni, V., Blackledge, T.A., Dhinojwala, A., 2011. A review on spider silk adhesion. The Journal of Adhesion 87, 595-614. Schneider, J.M., Elgar, M.A., 2001. Sexual cannibalism and sperm competition in the golden orb-web spider Nephila plumipes (Araneoidea): female and male perspectives. Behavioral Ecology 12, 547-552. Schüch, W., Barth, F.G., 1990. Vibratory communication in a spider: female responses to synthetic male vibrations. Journal of Comparative Physiology a-Sensory Neural and Behavioral Physiology 166, 817-826. Schutt, K., 1996. Wie Spinnen ihre Netze befestigen. Mikrokosmos 85, 274-278.

107

Schwartz, S.K., Wagner, W.E., Hebets, E.A., 2014. Obligate male death and sexual cannibalism in dark fishing spiders. Animal Behaviour 93, 151-156. Silva, L.P., Rech, E.L., 2013. Unravelling the of nanoscale signatures of spider silk fibres. Nature Communications 4, 1-9. Sivalinghem, S., Kasumovic, M.M., Mason, A.C., Andrade, M.C.B., Elias, D.O., 2010. Vibratory communication in the jumping spider Phidippus clarus: polyandry, male courtship signals, and mating success. Behavioral Ecology 21, 1308-1314. Spiess, K., Lammel, A., Scheibel, T., 2010. Recombinant Spider Silk Proteins for Applications in Biomaterials. Macromolecular Bioscience 10, 998-1007. Spiller, D.A., 1992. Relationship between prey consumption and colony size in an orb spider. Oecologia 90, 457-466. Stuart-Fox, D., 2006. Testing game theory models: fighting ability and decision rules in chameleon contests. Proceedings of the Royal Society B-Biological Sciences 273, 1555-1561. Suter, R.B., 1978. Cyclosa turbinata (Araneae, Araneidae): Prey discrimination via web- bourne vibrations. Behavioral Ecology and Sociobiology 3, 283-296. Suvák, M., 2013. Invasive spider Uloborus plumipes Lucas, 1846 (Araneae: Uloboridae), new to Slovakia. Folia Faunistica Slovaca 18. Taylor, P.W., Hasson, O., Clark, D.L., 2001. Initiation and resolution of jumping spider contests: roles for size, proximity, and early detection of rivals. Behavioral Ecology and Sociobiology 50, 403-413. Taylor, P.W., Jackson, R.R., 2003. Interacting effects of size and prior injury in jumping spider conflicts. Animal Behaviour 65, 787-794. Teo, W.E., Ramakrishna, S., 2006. A review on electrospinning design and nanofibre assemblies. Nanotechnology 17, R89. Townley, M.A., Tillinghast, E.K., 2003. On the use of ampullate gland silks by wolf spiders (araneae, lycosidae) for attaching the egg sac to the spinnerets and a proposal for defining nubbins and tartipores. Journal of Arachnology 31, 209-245. Tso, I.M., Severinghaus, L.L., 1998. Silk stealing by Argyrodes lanyuensis (Araneae : Theridiidae): a unique form of . Animal Behaviour 56, 219-225. Turner, G.F., 1994. The fighting tactics of male mouthbrooding cichlids: the effects of size and residency. Animal Behaviour 47, 655-662. Vibert, S., Scott, C., Gries, G., 2014. A meal or a male: the 'whispers' of black widow males do not trigger a predatory response in females. Frontiers in Zoology 11, 4. Vollrath, F., 2000. Strength and structure of spiders’ silks. Molecular Biotechnology 74, 67- 83. Vollrath, F., 2006. Spider silk: thousands of nano-filaments and dollops of sticky glue. Current Biology 16, R925-927. Vollrath, F., Fairbrother, W.J., Williams, R.J.P., Tillinghast, E.K., Bernstein, D.T., Gallagher, K.S., Townley, M.A., 1990. Compounds in the droplets of the orb spider's viscid spiral. Nature 345, 526-528. Vollrath, F., Selden, P., 2007. The Role of Behavior in the , Silks, and Webs. Annual Review of Ecology, Evolution, and Systematics 38, 819-846. Wells, M.S., 1988. Effects of body size and resource value on fighting behaviour in a jumping spider. Animal Behaviour 36, 321-326. Whitehouse, M.E.A., 1997. Experience influences male–male contests in the spider Argyrodes antipodiana (Theridiidae: Araneae). Animal Behaviour 53, 913-923. Wiehle, H., 1927. Structure of the orb-webs of the Epeiridae, Tetragnathidae, and Uloboridae. Zeitschrift für Morphologie und Ökologie der Tiere 8, 468-537. Wiehle, H., 1931. Neue Beiträge zur Kenntnis des Fanggewebes der Spinnen aus den Familien Argiopidae, Uloboridae und Theridiinae. Zeitschrift für Morphologie und Ökologie der Tiere 22, 349-400. 108

Wignall, A.E., Herberstein, M.E., 2013a. The influence of vibratory courtship on female mating behaviour in orb-web spiders (, Karsch 1878). Plos One 8, 1-9. Wignall, A.E., Herberstein, M.E., 2013b. Male courtship vibrations delay predatory behaviour in female spiders. Scientific Reports 3, 1-9. Wilder, B., 1874. The nets of Epeira, Nephila and Hyptiotes (Mithras). Proceedings American Association for the Advancement of Science 22, 264-274. Wise, D.H., 1975. Food limitation of the spider Linyphia marginata: Experimental field studies. Ecology 56, 637-646. Wise, D.H., 1983. Competitive mechanisms in a food-limited species: relative importance of interference and exploitative interactions among labyrinth spiders (Araneae: Araneidae). Oecologia 58, 1-9. Wise, D.H., 2006. Cannibalism, food limitation, intraspecific competition and the regulation of spider populations. Annual Review of Entomology 51, 441-465. Wise, D.H., Wagner, J.D., 1992. Evidence of exploitative competition among young stages of the Schizocosa ocreata. Oecologia 91, 7-13. Wolff, J.O., Grawe, I., Wirth, M., Karstedt, A., Gorb, S.N., 2015. Spider's super-glue: thread anchors are composite adhesives with synergistic hierarchical organization. Soft Matter 11, 2394-2403. Zheng, Y., Bai, H., Huang, Z., Tian, X., Nie, F.Q., Zhao, Y., Zhai, J., Jiang, L., 2010. Directional water collection on wetted spider silk. Nature 463, 640-643. Zschokke, S., 2002. Ultraviolet reflectance of spiders and their webs. Journal of Arachnology 30, 246-254. Zschokke, S., Vollrath, F., 1995. Unfreezing the behavior of two orb spiders. Physiology & Behavior 58, 1167-1173.

109

110

Acknowledgements

Ich danke meinem Doktorvater Prof. Werner Baumgartner, dass er mir sein Vertrauen entgegengebracht hat und ich an diesem sehr spannenden Thema arbeiten durfte. Insbesondere danke ich ihm für die gewährten Freiheiten, die sehr inspirierenden und motivierenden Diskussionen und seinen immerwährenden Optimismus. Vielen Dank an meinen Zweitgutachter Prof. Peter Bräunig, der diese Arbeit in so vielerlei Hinsicht unterstützt und möglich gemacht hat. Danke für die tollen Anregungen zu meinen geplanten Versuchen und die guten Gespräche. Ich danke Prof. Björn Kampa, der mich mit seiner herzlichen Art in seine Arbeitsgruppe aufgenommen hat. Ebenso danke ich Prof. Jörg Mey für seinen Enthusiasmus sowie für die kritischen und anregenden Auseinandersetzungen zum Thema „Spinnenkämpfe“. Prof. Herman Wagner danke ich für seine Unterstützung meiner Arbeit. Vielen Dank an alle ehemaligen und aktuellen Mitglieder des Instituts der Biologie II und an meine Arbeitsgruppe ZNB bzw. nun MSN für ihre mir entgegengebrachte Hilfsbereitschaft bei Fragen und Problemen und die freundschaftliche Arbeitsatmosphäre. Insbesondere möchte ich hier Christine Böhme, Dipl.-Biol. Philipp Comanns, Claas Halfmann, M.Sc., Dipl.-Biol. Florian Hischen, Dr. Martin Singheiser, Dipl.-Biol. Isabell Weiß, Maria Bugaro, Susanne Schotthöfer, M.Sc sowie meinen ehemaligen Bacheloranden Peter Kappel, Anne Habedank, Hana Adamova und Linda Orth danken, die alle auf ihre ganz eigene Art mich und meine Forschung unterstützt haben. Dr. Ingo Scholz möchte ich für die schöne gemeinsame Zeit in einem Büro danken und dafür, dass er sich immer die Zeit genommen hat, mit mir meine Probleme und Ideen zu besprechen. Für ihre Unterstützung danke ich meinen Kooperationspartnern Dr. Daesung Park und Dr. Wolf-Alexander Heiß, Magnus Kruse, M.Sc, Dr. Helga Krieger und Dipl.-Biol. Sabrina Jandrey sowie Jonas Hausen, M.Sc.. Dr. Barbara Baehr, Dr. Robert Raven, Robert Whyte und Graham Milledge danke ich für Ihre Mühen, mir Spinnen und Spinnfäden zur Verfügung gestellt zu haben (thank you for taking the trouble collecting spiders and thread samples). Vielen Ebenso vielen Dank an Dr. Moritz to Baben, bei dem ich regelmäßig meine Proben mit Kohle bedampfen durfte. Stellvertretend für alle netten E-Mail-Kontakte, welche über die vergangenen Jahre entstanden sind, möchte ich Dr. Rainer Foelix nennen und ihm dafür danken, dass er mit seiner direkten und freundlichen Art mir bei allen Fragen jeglicher Art sofort geholfen hat. Ich möchte abschließend meinen Freunden, meiner Familie, meinen Großeltern, meinen Eltern Marie-Helen Joel und Heribert Feckler, meiner Schwester Marie Joel, sowie Philipp Hartmann und seiner Familie danken, dass sie mich in den letzten 3 ½ Jahren (und ein paar mehr) immer mit Rat, Tat und viel Herz unterstützt haben. Philipp und Heribert möchte ich weiterhin danken, dass sie diese Arbeit Korrektur gelesen haben. Ihr seid die Besten!

111

112

Curriculum vitae

Persönliche Daten: Name: Joel Vorname: Anna-Christin Geburtstag: 06.12.1987 Geburtsort: Neuss Staatsangehörigkeit: deutsch

Qualifikationen: 2007 Abitur 2007 - 2012 Studium der Biologie, RWTH Aachen University 2010 B.Sc. Abschluss 2012 M.Sc. Abschluss 2012 - 2016 Promotion am Institut für Biologie II

Veröffentlichungen: 2015 Foelix, R.F.; Joel, A.-C.; Erb, B. Wie die Kräuselradnetzspinne Uloborus ihre Beute einwickelt und verdaut. Arachne 2, 16-23. Joel, A.-C., Kappel, P.; Adamova, H.; Baumgartner, W.; Scholz, I. Cribellate thread production in spiders: Complex processing of nano- fibres into a functional capture thread. Arthropod Structure and Development 44 (6 Pt A), 568-573. Foelix, R.F.; Rechenberg, I.; Erb, B.; Joel, A.-C.

Zum Sandtransport der „Radlerspinne“ Cebrennus rechenbergi JÄGER, 2014. Arachne 6, 4-13. 2016 Joel, A.-C.; Scholz, I.; Orth, L.; Kappel, P.; Baumgartner, W. Morphological adaptation of the calamistrum to the cribellate spinning process in Deinopoidae (Uloboridae, Deinopidae). Royal Society Open Science 3 (2).

113

2016 Foelix, R.F.; Rechenberg, I.; Erb, B.; Joel, A.-C. Über den Bau der Wohnröhren bei wüstenlebenden Spinnen. Arachne 1, 4-17. Eingereicht Joel A.-C., Habedank A., Hausen J., Mey J. Fighting for the web: Intraspecific competition among female feather- legged spiders Uloborus plumipes.

Patente: 2015 Baumgartner, W.; Comanns, P.; Gries, T.; Jandrey, S.; Joel, A.-C.; Krieger, H.; Scholz, I.. Passiver gerichteter Flüssigkeitstransport senkrecht zu einer Oberfläche. DPMA, No. 10 2015 001 461.7

Konferenzen: 2015 108th Annual Meeting of the DZG, Graz, Österreich, Vortrag 2. Bocholter Bionik-Workshop, Bocholt, Deutschland, 2 Poster

114

Supplement

S1: Incomplete first data pooling.

Family:

Spider Spinnerets Calamistrum Capture thread Other data

Mahura sp. [1]

Neoramia sana [2]

Family: Amaurobiidae

Spider Spinnerets Calamistrum Capture thread Other data

Amaurobius sp. [2] [2] Type II combing Amaurobius Teeth posture [3] fenestralis

Amaurobius Type II combing Teeth similis posture [3] Callobius gauchama [2]

Pimus sp. [2]

Retiro sp. [2] Poaka graminicola [2] [2]

Family: Amphinectidae

Spider Spinnerets Calamistrum Capture thread Other data

Metaltella simoni [2] [2]

Family: Austrochilidae

Spider Spinnerets Calamistrum Capture thread Other data

Austrochilus forsteri Type II combing posture [2, 4] Austrochilus franckei

Hickmania troglodytes [2] [2] [2] Thaida sp. Pictures in [5]

Thaida peculiaris [2] [2] [2]

Family: Ctenidae

Spider Spinnerets Calamistrum Capture thread Other data

Acanthoctenus sp. [2] Acanthoctenus spinipes [2]

Family: Deinopidae

Spider Spinnerets Calamistrum Capture thread Other data

Avella sp. [1]

Type II combing Deinopis spinosa posture [3] [6] [6]

Deinopis subrufa [7] [7] [7]

Menneus sp. [2]

Menneus camelus [2]

Family: Desidae

Spider Spinnerets Calamistrum Capture thread Other data

Badumna sp. [1] Badumna longinqua [8] [8] [2]

Matachia australis [2] Matachia livor

Paramatachia decorata [1]

Family: Dictynidae

Spider Spinnerets Calamistrum Capture thread Other data

Aebutina bionata [2] [2] Type II combing Dictyna sp. [1] posture [3]

Dictyna arundinacea [2] [2]

Lathys humilis [2]

Mallos sp. No teeth [2]

Mallos dugesi [9] Type II combing Mallos gregalis Pictures in [5] posture [3]

Mallos niveurs [9]

Mexitlia trivittata [10] [10]

Nigma linsdalei [2]

Family: Eresidae

Spider Spinnerets Calamistrum Capture thread Other data

Eresus cf. Teeth cinnaberinus [2]

Eresus Teeth sandaliatus [2]

Stegodyphus sp.

Stegodyphus Pictures in [5] lineatus

Stegodyphus mimosarum [2] [2] Stegodyphus Type II combing sarasinorum posture [3]

Family: Filistatidae

Spider Spinnerets Calamistrum Capture thread Other data Filistata No teeth arizonicus [11] Filistata No teeth insidiatrix [2]

Filistatinella sp. [2]

Kukulcania sp. [2]

Kukulcania Type I combing [10] hibernalis posture [3] [2]; More pictures in [10] [2] More pictures in [1] and [4]

Misionella mendensis Type I combing [4] [4] posture [4]

Pritha nana [4]

Family: Hypochilidae

Spider Spinnerets Calamistrum Capture thread Other data

Hypochilus pococki [2] [2] Teeth [12]

Hypochilus Type I combing Picture in [13] thorelli [13]; More posture [3] informations in [13] [1]; More pictures in [13]

Family:

Spider Spinnerets Calamistrum Capture thread Other data

Megadictyna thilenii [2] [2]

Family:

Spider Spinnerets Calamistrum Capture thread Other data Oecobius Type II combing annulipes posture [3]

Oecobius navus [2]

Family: Phyxelididae

Spider Spinnerets Calamistrum Capture thread Other data Phyxelida tanganensis [2] [2]

Xevioso amica [2]

Family: Progradungula

Spider Spinnerets Calamistrum Capture thread Other data Progradungula carraiensis

Family: Psechridae

Spider Spinnerets Calamistrum Capture thread Other data

Type II combing Psechrus sp. posture [3]

Psechrus himalayanus [2]

Family: Stiphidiidae

Spider Spinnerets Calamistrum Capture thread Other data

Pillara griswoldi [2]

Stiphidium sp. [1]

Family:

Spider Spinnerets Calamistrum Capture thread Other data

Titanoeca americana [2]

Titanoeca nigrella [2]

Family: Uloboridae

Spider Spinnerets Calamistrum Capture thread Other data

Hyptiotes sp. [2]

Hyptiotes cavatus [14] Type II combing [12] [14] [12] posture [3]

Hyptiotes Type II combing paradoxus [15]; More pictures in posture [3] [5] [15] Miagrammopes sp. [16]

Miagrammopes animotus [18] More [17] [17] pictures in [17] Philoponella Type II combing tingena posture [3] Type II combing Philoponella vicia posture [3]

Polenecia Type II combing producta [19] [20] [19] [20] [19] posture [3] More pictures in [19] More pictures in [5] More pictures in [5]

Tangaroa beattyi [18]

Uloborus sp. No teeth [2] Type II combing Uloborus diversus No teeth posture [3] [11] Uloborus Type II combing glomosus posture [3] No teeth Uloborus Type II combing plumipes [20] [20]; No teeth [20] posture [3]

Uloborus Pictures in [5] No teeth [20] walckenaerius [20] [21] Type II combing More pictures in [5, 22] posture [3] Waitkera waitakerensis [16] [8]

Family: Zorocratidae

Spider Spinnerets Calamistrum Capture thread Other data

Raecius jocjuei [2]

Raecius scharffi [2] Zorocrates mistus [2] [2]

Family: Zoropsidae

Spider Spinnerets Calamistrum Capture thread Other data

Zoropsis rufipes [2]; Teeth

Zoropsis spinimana [2] [2]; Teeth [2]

1: Eberhard et al. 1993; 2: Griswold et al. 2005; 3: Eberhard 1988; 4: Lopardo et al. 2007; 5: Peters 1983; 6: Coddington et al. 2012; 7: Peters 1992c; 8: Opell 1999; 9. Bond et al. 1997; 10: Opell 2002; 11: Friedrich et al. 1969; 12: Hawthorn et al. 2002; 13: Foelix et al. 1978; 14: Opell 1982a; 15: Reukauf 1930; 16: Opell 1994b; 17: Opell 1995; 18: Opell 1989; 19: Peters 1995a; 20: Peters 1984; 21: Zheng et al. 2010; 22: Liao et al. 2011.