2017. Journal of Arachnology 45:166 176 Natural history and courtship behavior in Tengella perfuga Dahl, 1901 (Araneae: Zoropsidae) Rachael E. Mallis and Kelly B. Miller: Museum of Southwestern Biology, Arthropods, 1 University of New Mexico, MSC03 2020, Albuquerque, New Mexico 87131: E-mail: [email protected] Abstract. Tengella perfuga Dahl. 1901 is a Nicaraguan cribellate zoropsid spider found in high altitude remnant cloud forest habitats bordering coffee plantations. Since its description in 1901, and its rediscovery in 2012, almost nothing is known of its natural history, life history, courtship or web spinning behavior. Observations were made in the field, as well as in the lab. Mature female T. perfuga occurred in funnel webs with several knockdown lines comprised of cribellate silk, and that were typically placed between buttress roots of strangler figs or other outcropping structures, while males abandoned their webs upon adulthood to search for females. Here, we describe the life history, growth, web ontogeny, courtship and reproductive behaviors, as well as silk use of this spider for the first time. There are 11-12 instars to reach adulthood and cribellate silk did not appear in juvenile webs until the eighth instar. Interestingly, orbicularian-like behaviors were observed in the initial appearance of cribellate silk lines in the juvenile web in a spiral-like pattern. Males exhibited positive allometric growth in Leg I from penultimate to adult instars, which likely plays an important role in courtship; this included strumming the sheet, stroking the female and depositing a thin 'bridal veil' of silk on the female. Virgin females had ‘mating plugs' prior to exposure to males. This suggests that T. perfuga may be an interesting species with which to further examine sexual evolution and female choice. Keywords: Web ontogeny, growth, cribellate silk, reproduction Tengella perfuga Dahl, 1901 is the type species for the genus resemble the ancestral web design or plesiomorphic traits in silk Tengella Dahl, 1901, which is the type genus for the recently use, so understanding web ontogeny can help reconstruct reassigned family Tengeliidae Dahl, 1908 (= Zoropsidae evolutionary history (Eberhard 1985, 1986; Barrantes & Bertkau, 1882 (Polotow et al. 2015)). They are medium to large Madrigal-Brenes 2008; Barrantes & Eberhard 2010). Given the cribellate spiders and little is known about their natural history, recent genomic evidence of paraphyly in the Orbiculariae and courtship behavior, feeding behavior and web ontogeny (Fig. 1, the placement of T. perfuga within the RTA clade, a web Platnick 2009; Leister et al 2013). While spiders of the genus ontogeny study becomes important as a way to behaviorally Tengella are distributed across various habitats in Centra! examine the evolution of silk use and orb web ancestry in this America, from caves in Mexico to tropical lowland forests in clade (Agnarsson et al. 2013; Bond et al. 2014; Fernandez et al. Panama, T. perfuga is limited to high altitude remnant cloud 2014). Orbicularian traits of silk use within the RTA clade have forest habitats bordering coffee plantations in northwestern only recently been examined once, where the phylogenetic Nicaragua and Honduras (Leister et al. 2013, S. Longhorn, pers. placement of the Psechridae, a family of cribellate pseudo-orb comm.). However, the potential sister species, Tengella radiata weavers, was moved from the Orbiculariae to the RTA clade (Kulczyhski, 1909), is widespread and found in various high and (Agnarsson et al. 2013). This study of T. perfuga is the first study low elevation tropical forests, coffee plantations and developed to specifically look for evidence of orbicularian behavioral traits sites from Honduras south to Panama (Leister et al. 2013). The in a member of the RTA clade. phylogenetic placement of T. perfuga has long been uncertain in Two previous studies were conducted on T. radiata and these large part due to its use of cribellate silk (Fig. 1). Cribellate silk is served as a basis for the studies reported here. Tengella radiata produced from glands that open on a plate-like structure, the develop to adulthood in 9-10 molts, with males having longer cribellum, derived from the ancestrally lost anterior median forelegs than females (Barrantes & Madrigal-Brenes 2008). spinnerets (Pechmann et al. 2010). Cribellate silk production Webs begin as basic sheets and more complex elements are also requires a comb (calamistrum) on the fourth leg which added with each instar; notably, cribellate silk does not appear primes the silk glands and combs out the loops of fibrils until the seventh instar (Barrantes & Madrigal-Brenes 2008). (Hawthorn & Opell 2002). Use of this type of silk in Tengella is Males court females with a series of plucking the web, rocking surprising, given the developmental investment and metabolic motions and tapping to induce the female to copulate; females costs to maintaining the cribellum and calamistrum (Blackledge show some aggressive reactions to potentially unsuitable et al. 2009a), and that this trait is typically found in older spider partners (Barrantes 2008). Males also exhibited a ‘flub' behavior lineages, whereas Tengella is part of a more recently evolved while attempting insertions, with repeated scraping motions, but clade, the majority of whose members have evolved the loss of successful insertion and single expansion of the hematodocha silk use as a foraging tool (Raven & Stumkat 2003, 2005; was extremely rapid when it occurred (Barrantes 2008). Griswold et al. 2005; Spagna & Gillespie 2008; Blackledge et al. The objective of this study was to learn more about these 2009a, b; Agnarsson et al. 2013; Polotow et al. 2015). As such, T. enigmatic spiders, particularly their use of cribellate silk, life history, perfuga is at the crux of important phylogenetic questions about behavior, and ontogenetic changes in web size and structure. the evolution of silk use in spiders, particularly within the RTA clade to which both Tengella (zoropsids) and their sister group METHODS Lycosoidea belong (Griswold et al. 2005; Polotow et al. 2015). One approach to understanding the evolutionary history of Field collection and specimen sources.—Live T. perfuga adult silk use is to observe web ontogeny. Early instar webs may females were collected in Nicaragua (Selva Negra, 12.9984°N, 166 MALLIS & MILLER—TENGELLA PERFUGA NATURAL HISTORY 167 Figure 1.—Adult male (right) and female (left) Tengellaperfuga, with cribellate silk from a female web (inset) (Spider whole body photos: M. Leister, with R. Mallis 2012). 85.9105°W) in May 2012 (permit: DGPN / DB - 09 - 2012), the containers in which they were housed (Gladware® storage and subsequently in May 2014 (permit: DGPN / DB - 006 - containers, 15.5 x 15.5 cm and approximately 3.5 cm high) and 2014) and allowed to mate and/or lay egg sacs, giving rise to a so web size was artificially limited. Webs for each instar were lab-reared spider colony which at one time numbered well over imaged using a Nikon Coolpix LI 10 camera. 500 individuals. Many of these reached adulthood and Colony Maintenance: Spider habitats were Gladware® square reproduced, allowing for observations of courtship behavior, plastic storage containers filled with a layer of EcoEarth®. growth and web ontogeny. Observations of web structure and Similar to a study of the closely related T. radiata, we provided feeding behavior were made in the field (n » 100). Most field 2-4 pieces of corkwood for web attachment, rather than rocks, encountered males were collected in female webs (in 2014, n = and a retreat option of a 2mL vial, rather than a rolled-up leaf 7; in 2012, n > 15) or as penultimate males in their own webs. as in a previous study (Barrantes & Madrigal-Brenes 2008). In All field-caught spiders and reared individuals are deposited in 2012, we collected two gravid females in the field, and used 25 the Museum of Southwestern Biology, Division of Arthropods spiderlings from one female and 27 from the other in our study. collection or teaching collection (MSBA 24980 — 24982, Three groups of six additional spiderlings each were set up for 24985, 24986, 29081, 29082, 30589, 30591, 30592, 30596 - observation as well to better understand conspecific tolerance, 30599, 30619, 30621, 30622, 30635, 30640, 30642, 30643). A as T. perfuga was sometimes found at higher densities in a single complete instar series, as well as SEM specimens, from second site in the field. Containers were spritzed with distilled water instar to adult male and female T. perfuga is also deposited at weekly to provide moisture, and spiders were fed a steady diet the California Academy of Sciences, Department of Entomol¬ of crickets ranging in size from pinheads to medium-large and ogy alcohol collection. These include a few of the specimens fruit flies. Natural history traits such as feeding behavior and used in the growth and web ontogeny studies. timing of molts were recorded. Containers were monitored Web ontogeny and life history.—A sub-group of approxi¬ daily for spider status and spiderlings fed twice weekly, but as mately 50 second-instar individuals, which had recently emerged they aged, feedings became once weekly to biweekly. Spiders from two different egg sacs, were removed from the maternal were maintained at ambient room temperatures in the webs and housed individually. Web ontogeny and spider growth laboratory and as close to a 10:14 lighLdark ratio as possible. at each instar stage were measured. Each stage is referred to by Individuals that died of natural causes were also preserved in its sequential number; for example, the second instar is called 70% EtOH with a leg placed in 95% EtOH for potential future ‘instar 2' and so forth. We measured the width at widest points molecular sequencing work. and the lengths of the cephalothorax, femur I, tibia I and the Measurements: After all spiders in the colonies had completed overall body length of recently molted individuals for each an instar and molted, three from a pool of the two mothers’ instar. Measurements were made using an Olympus SZ60 offspring were randomly selected and their containers placed in binocular dissecting scope equipped with a calibrated 10x a -20°C freezer to preserve the web for imaging. Culled spiders micrometer. Images of instars and silk were taken using a were immediately removed after webs were imaged. One spider Visionary Digital System (online at http://www.visionarydigital. was stored in 95% EtOH at -80°C to preserve genetic data, and com). We noted web dimensions and characteristics such as the the remaining two were placed in 70% EtOH and used for appearance of cribellate silk, but the webs of older instars filled morphological measurements. Webs were photographed dry 168 JOURNAL OF ARACHNOLOGY culled spiders for measurements, rather than rehydrating shed molts from the same individual spiders as they grew. Courtship observations.—Using adult spiders from later generations and other field collections, as well as the remaining adults from the web ontogeny study, non-related pairs were randomly assigned for mating (n = 35 documented observa¬ tions; 24 initial pairings, and 11 subsequent interactions). All females were virgin and well fed prior to introduction of the males. Males were placed onto female webs and courtship encounters were video recorded and behaviors noted. Males were removed after copulating once or after rejection by the female in order to reduce the chances of mortality (n = 24). They were secondarily introduced to the same or a different non-related female to propagate the spider colony, but allowed us to observe differences in courtship behavioral patterns and acceptance (n = 11). Females were later allowed to lay egg sacs following successful copulations. RESULTS Webs in the field.—T, perfuga sheet webs typically had a funnel retreat either at a corner or the middle of the sheet that Figure 2A C.—Webs in the field in Nicaragua, showing variation receded into the substrate (n >> 100) (Fig. 2). Sheets were in structure. A. Preferred habitat of strangler fig buttress roots. B. typically surrounded by a scaffolding of tangle or knockdown stacked webs of juveniles. C. adult female web. lines above and anchor lines below (Fig. 2; Leister et al. 2013). The sheets were lined with cribellate silk and the majority of knockdown lines had cribellate silk laid over them. Webs and subsequently wet after being lightly spritzed with water to typically occurred along stream embankments, tree trunks increase their visibility, and the presence or absence of cribellate (especially strangler figs) and between stones or wood beams silk and length, width and height (if applicable) measurements on structures at high elevation cloud forest sites associated were taken. Webs from eighth-instar spiders were inadvertently with shade coffee plantations in northern Nicaragua. Spiders damaged prior to measurements, but presence or absence of were observed in the retreat or just at the retreat opening; they cribellate silk observations were still recorded. ran out onto the sheet to capture prey and drag it back to the Averages and standard deviations of morphological mea¬ retreat, as we also recorded in Leister et al. (2013). surements for each instar were calculated from the pooled data A variety of web locations were noted in the field, with some per instar. To calculate the relative percent growth from instar adult female webs appearing in unexpected places, like a hole to instar, we used the equation as in Barrantes & Madrigal- in a trail sign on a tree or across the span of an empty bell Brenes (2008): [(Tibia I InstarN - Tibia I InstarN-l)/Tibia I tower of a stone chapel. Some webs had egg sacs or second or InstarN-1] * 100. While much of our approach for the natural third instar spiderlings in the retreat (Fig. 3A, B). The egg sacs history study was inspired by Barrantes & Madrigal-Brenes were covered with pieces of the surrounding substrate, from (2008), we had some differences, such as spritzing webs with bark to soil to leaves (Fig. 3B). Egg sac production was not water rather than corn starch and more importantly, using observed in the field. At two less disturbed sites, webs were Figure 3A B.—Egg sac construction and camouflage. A. Female silking over egg sac after depositing the eggs. B. Female guarding a camouflaged egg sac. MALLIS & MILLER—TENGELLA PERFUGA NATURAL HISTORY 169 Figure 4.—Growth from instar to instar of cephalothorax length and width, femur I and tibia I lengths. Comparison of the amount of growth in Leg I that occurs from second to twelfth instar (adulthood), as well as the aliometric growth of male during the final molt (inset). observed with commensal bugs and kleptoparasitic spiders, Ultimately, adult webs were comprised of a deep retreat into similar to reports for the closely related species, T. radiata the substrate, surrounded by a broad sheet with several (Eberhard et al. 1993). These web symbionts have not yet been knockdown lines or ‘scaffolding’ above the sheet and retreat identified. entrance. The majority (« ~ 25 adult webs observed) of the Observations in the laboratory.—Tengella perfuga reached structures were lined with cribellate silk (Figs. 2 & 6D). adulthood between 205 - 226 days and 11-12 molts after Life history: Eggs hatched in approximately 54 days (n = 2 emerging from the egg sac as a second instar nymph. Females egg sacs). It took five days for all spiderlings to leave the egg reached adulthood typically in 12, sometimes 11 instars, sac. In the groups of six spiders, each spider in the group had whereas most males reached adulthood in 10-11 instars. For its own retreat and shared use of the sheet. Minimal the morphological measurements and web ontogeny observa¬ cannibalism was observed and there appeared to be tolerance tions, we used males who reached adulthood in 12 instars. for con-specifics. Development time varied. Small males were Female T. perfuga tended to be more robust than males, observed after 9-10 molts. Some females developed with less however the color patterning is similar (Fig. 1). Males had time between each of the 12 molts. Other members in = 4) of longer legs than females, particularly leg I (Fig. 4 inset; Leister the groups appeared to have arrested development in earlier et al. 2013). The calamistrum did not appear to be a full oval instars and never reached adulthood during the nearly year¬ until instar 5, and the cribellum did not appear to be long observational period. functional until instars 7 and 8 (see Fig. 5). Cribellate silk Growth: Overall, from second to eleventh instar, T. perfuga did not appear in the webs until the eighth instar (Figs. 6 & 7). grew by 612% relative to the body length of a second instar After hatching from the egg sac, spiderlings remained with the (Table 1). Other body parts, such as tibia I and femur I, also mother in her retreat on a collectively spun ‘molting web.' grew by over 1400% and 1200%, respectively (Table 1; Fig. 4). Once molted to the third instar, spiderlings began to disperse. There appeared to be less growth between the fifth and sixth Most of the early instar webs exhibited features such as a instars and seventh and eighth instars (Table 1, Fig. 4). simple sheet and clear retreat tunnel similar to those of adult Because the sample size for each instar (« = 2-3) was small, webs, but were smaller and lacked cribellate silk (Fig. 6A, B). there was no power for statistical analyses. Second instar First, a tiny retreat was formed with some lines extending to spiders did not eat, as they still had yolk fat, and although they form the scaffolding for the subsequent sheet (instar 3), then a were able to readily walk on the mother's web, the legs sheet was filled in (instar 4). Prey capture was still successful, appeared short for the body size (approximately 4:1 ratio of despite a small capture surface, and these instars grabbed prey body length to tibia I length, vs. approximately 2:1 ratio of through the retreat or sheet or ran on top of the structure to body length to tibia I length in penultimates). Spiders began bite the prey. Knockdown lines were not observed until instars foraging on their webs in the third instar, and this was when 5 and 6 (Fig. 6A, B). When cribellate silk first appeared in the their overall appearance mirrored adult gestalt and pattern. webs, it was in an orb-like spiral laid out on the sheet (Fig. 7). Males had a greater increase in leg I length than females from 170 JOURNAL OF ARACHNOLOGY Figure 5A E.—Cribellum and calamistrum images from selected instars demonstrating increasing size and complexity of structures, as well as potential functionality. A. Instar 2, note the lack of a cribellar plate. B. Instar 2, leg IV, note the lack of the calamistrum (however, see next image). C. Instar 3, cribellum present. D. Instar 11, cribellum (penultimate male), and E. Instar 11, oval shaped calamistrum (penultimate male). Figure 6A D.- Web images from selected instars in the lab, demonstrating web growth and increasing complexity of structure and the presence of cribellate silk. A. Web of third instar spiderling. B. Web of fourth instar spiderling. C. Web of ninth instar spideriing. D. Web of adult female with male present (Instar 12). MALLIS & MILLER -TENGELLA PERFUGA NATURAL HISTORY 171 were observed in the web, with some seemingly thinner and other ‘combed out’ areas filling in the webs (Fig. 6C, D). The cribellum appeared functional (fully developed) in instar 7. The calamistrum appeared functional between instars 5 and 6, during which little body growth occurred (Table 1, Fig. 4). Second instar spiderlings did not possess a cribellar plate¬ like structure or any precursors to functional spigots, nor did they have a calamistrum or any type of modified setae on metatarsus IV (Mallis-Alfaro, Miller & Griswold, unpublished data). Third instar spiders possessed a single row of stout curved setae as a calamistrum and a small cribellum. Despite cribellar spigots present from third instar onward to adult (instars 11/12), and active expansion of the spigot field on the cribellum from instar to instar (Mallis-Alfaro, Miller & Griswold, unpublished data), functionality did not seem to occur until the appearance of cribellate silk in the eighth instar. Similarly, the calamistrum expanded from a single row of setae to an oval shaped patch in the sixth instar that Figure 7.—The orb-like pattern, indicated by the arrows, of appeared as in the adult. Under the dissecting microscope, the cribellate silk in the sheet of a juvenile, eighth instar T. perfuga. In this cribellum appeared as a pseudobipartite plate from the third image, focus was sharpened and contrast was enhanced in order to aid instar onward. There was a single cribellar plate, but two in observing the spiraling cribellate lines. separate spinning fields of cribellar spigots in third and subsequent instars (Mallis-Alfaro, Miller & Griswold, unpub¬ the penultimate to adult molt, going from an average of tibia I lished data). length of 7.17 mm to 11.19 mm in the male versus 7.17 mm to To comb the cribellate silk out, spiders crossed both the 7.63 mm in the female, suggesting allometric growth occurred ‘combing leg’ (leg IV) and the ‘supporting leg’ (other leg IV) (see Fig. 4 inset, adult length values previously published in and moved them synchronously as a single unit, using swift Leister et al. 2013). and sharp anterior to posterior movements. The tarsus of the combing leg IV rested on the lower half of the metatarsus on Ontogeny of the cribellum, calamistrum and cribellate silk the supporting fourth leg. The same combing legs were used to use: In mature individuals of both sexes, the cribellum was a complete each cribellate silk segment being laid down. Spiders pseudobipartite plate (Fig. 5D). There were two patches of tended to switch combing legs between cribellate lines. spigots in the female, a row of setae at the anterior margin and Ontogenetic changes in web structure: Second instar spider- a line of sclerotization at the posterior margin. The lings emerging from the egg sac did not construct individual calamistrum was an oval patch that extended one third the webs. Instead they remained in the maternal retreat on a length of metatarsus IV on the proximal half dorsoretroiat- molting web spun collectively by all spiderlings. The molting erally (Fig. 5E). The male cribellum was a featureless plate; web was comprised of thin drag lines with no adult silk however, he retained a calamistrum. Cribellate silk appeared contributions; however, movement onto and throughout the in the webs between instars 7 and 8, first lining the retreats or mother’s web was possible. Third instars began to disperse incorporated in the tangle above the sheet, and ultimately in from the maternal web to form individual webs. In three cases, the sheet itself in an orb-like spiral radiating out from the two in the field, one in the lab, some formed a second retreat, replacing main support lines, then filling in the sheet in collective web separated from the mother’s retreat in the subsequent instars (Figs. 6 & 7). Mostly thick cribellate lines tangle scaffolding. In the webs of third instar spiderlings, the Table 1.—Instar growth data, using the averages of 2 3 spiderlings and the standard measures of carapace length and width, tibia I length and width, femur I length and width and body length (all in mm). In order to calculate the relative percent growth from instar to instar the same equation as Barrantes & Madrigal-Brenes (2008) was used to calculate relative percent growth from instar to instar: [(Tibia I InstarN Tibia I InstarN-l)/Tibia I InstarN-1]* 100 (Standard deviations not listed here). Percentage Change Between Instars Instar Carapace Length Carapace Width Tibia I Length Tibia I Width Femur I Length Femur 1 Width Body Length 2 to 3 48.68 34.62 102.25 41.18 94.12 71.43 35.73 3 to 4 37.17 32.57 74.44 33.33 63.64 8.33 39.88 4 to 5 23.55 25.00 33.44 34.38 26.54 48.72 24.44 5 to 6 3.92 5.52 27.92 16.28 -0.49 0.00 12.87 6 to 7 24.37 24.18 21.83 10.00 32.11 18.97 25.10 7 to 8 8.28 22.89 11.64 9.09 30.80 42.03 19.90 8 to 9 57.46 30.62 33.88 65.00 32.20 54.08 35.47 9 to 10 12.68 22.62 17.32 20.20 18.56 25.17 9.84 10 to 11 17.67 13.77 25.24 25.21 23.98 16.40 19.98 172 JOURNAL OF ARACHNOLOGY Figure 8A C.—Courtship behaviors. A. Male strumming the web and stroking the female. B. Male stroking the female, female passive. C. Male stilting behavior. beginnings of basic elements of adult webs were apparent with while hanging upside down, directly below the bowl, deposited spiders forming small funnel retreats, followed by a very small eggs and fluids into the bowl structure. Afterwards, they laid sheet and tangle lines in the fourth instar (Fig. 6A, B). silk over the entire bowl, reinforcing it (Fig. 3A), and then Retreats were either located in the middle or at one side of the added cribellate silk lines that eventually covered the entire egg web. Web complexity and size increased from one instar to the sac, completely covering it section by section. Lastly, females next, including multiple retreat entrances, sheet expansion and took pieces of the substrate in their chelicerae and placed them additional tangle lines (Fig. 6). against the cribellum silk where they adhered (Fig. 3B), Cribellate silk did not appear in the web until instar 8. presumably serving as camouflage for the egg sac (Fig. 3B). Nearly simultaneously, cribellate lines were observed in the Egg sacs were constructed singly or in pairs, about 1-2 weeks retreat, along with an orb-like spiral in the sheet. This was apart. Virgin females in the lab occasionally constructed egg followed in later instars with heavy or thick cribellate lines in sacs and deposit unfertilized eggs (n >10). the tangle that eventually covered the majority of the web in Courtship behavior.—Courtship began with the male orient¬ the eleventh or penultimate instar, giving the webs a fuzzy ing toward the female. Typically, in the lab, orientation by the appearance. Of the three randomly selected spiders measured male was preceded by preening (n = 20), during which time the for morphological growth and web ontogeny, two were male cleaned his palps and first two, or sometimes three, pairs penultimate males and one penultimate female. These males of legs between his chelicerae. At this time, the majority of actively maintained their webs and laid down cribellate lines. setae on the legs were fully visible and erect (see video SI, The twelfth instar, or adult stage, saw changes in male web online at http://dx.doi.org/10.1636/JoA0S-15-004.Sl). Once use. Males were observed or collected outside of webs or in oriented toward the female, likely through vibratory cues, the female retreats in the field. Males in their lab containers laid male shook his abdomen while plucking the web with his first down a circular sheet-like web composed of dragline silk. and second pairs of legs (Fig. 8A, see also video S2, online at Females continued to lay down cribellate lines on their webs http://dx.doi.org/10.1636/JoA0S-15-004.S2). If receptive (n = and tangle scaffolding until egg sac production. In the 16), the female generally responded by tapping the web with laboratory, most adults took refuge under the sheet, instead her legs I and sometimes legs II. If not receptive, the female of maintaining a retreat. This was apparently due to the lifted the web around her with all four pairs of legs and artifact of the short square containers used to house the forcefully pushed the web downward, as though shaking out a spiders in the lab colony. Webs in the field (n > 100) had a rug. If the female tapped, the male paused, then approached much more vertical stratification and multidimensional her and strummed the web and stroked her carapace and structure compared to those in the lab. abdomen with his first pair of legs, which are much longer Egg sac construction: Egg sacs in the field (n >10) were than those of the female (Fig. 8B). He interrupted stroking for similar to those constiucted in the lab (// ^ 20) (Fig. 3). variable periods to shake his abdomen. Sometimes the female Females constructed egg sacs in a stereotypical pattern (n > tapped or plucked in response, sometimes repeatedly. Even¬ 20). First, they erected a hammock-like structure, with three to tually (n = 35), the male ‘stilted’ up, standing as tall as four attachments at the ends to the top and sides of the physically possible on all four pairs of legs, and shook his container (or retreat if in the field) (n > 10) (Fig. 3A). Next, abdomen, typically above the female carapace (Fig. 8C, see they added silk to form a much thicker central disc at the also video S2, online at http://dx.doi.org/10.1636/ center of the hammock, followed by a spherical bowl JoAOS-15-004.S2). He did this stilting and shaking sequence underneath this disc. All of this was done while hanging up to three to four times. If she remained still, he deposited a upside-down. They seemingly sealed the disc to the bowl, then ‘bridal veil' of silk across the female’s carapace and legs. This MALLIS & MILLER—TENGELLA PERFUGA NATURAL HISTORY 173 was not a restraint, as the female could easily break the lines. plugs with the same appearance as the epigynal plugs observed If the female was receptive (n = 16), she exposed her epigynum in virgin females. by laying nearly completely on her side while the male silked the bridal veil (see video S3, online at http://dx.doi.org/10. DISCUSSION 1636/JoAOS-15-004.S3). As the male continued to stroke her, Life history.—While T. perfuga took 9-12 molts to reach he gathered her legs in towards her body with his long first adulthood, smaller numbers can be seen in deinopoids, such as pair of legs. The male then copulated, using his left palp to 6-7 instars to reach maturity after emergence in Hyptiotes transfer sperm into the opening of the left spermatheca and cavatus (Hentz, 1847). Similar to T. perfuga, some Pardosa vice versa. He did this while leaning across and over the C.L. Koch, 1847 have multiple egg sacs, with at least 30+eggs female. Coupling lasted several seconds, and the male in each and follow an approximately two-year life cycle from appeared to hook the female’s epigynum with the R.TA or egg sac emergence to reproduction and death (Buddie 2000). median apophysis of his partially inflated palp. When the palp The purported sister species of T. perfuga, T. radiata reaches engaged, the hematodocha rapidly expanded once and then maturity in 9 instars (8 molts) for males and 10 instars (9 deflated, which took less than a second once the embolus was molts) for females. The time to reach adulthood was similar to engaged (see video S4, online at http://dx.doi.org/10.1636/ that observed in the lab for T. perfuga: approximately 187 JoAOS-15-004.S4). Typically, both parties immediately moved days for males and 229 days for females (Barrantes & rapidly apart. In a few instances (n = 3), the female and male Madrigal-Brenes 2008). slowly separated a short distance and then resumed courtship, Growth.—Whereas the overall growth from second instar to but the majority of interactions were characterized by a rapid, adulthood was over 1000% for some structures, the amount of dramatic separation. On several occasions, the bridal veil growth varied between different instars. Despite a small sequence was repeated three or more times before successful sample size and, therefore, a lack of statistical testing, there copulation occurred. The female slowly broke the silk veil and appears to be less relative growth or slower growth rate the male would then re-approach her with web strumming and between instars 5 and 6 than any other stage. This warrants carapace stroking. There was a total of 35 documented further study and may be due to more energy invested in interactions; 24 were initial pairings (n = 16 copulations with developing structures such as the calamistrum (instars 5 and 6) single spermatheca), 11 reintroductions or subsequent inter¬ and cribellum (instars 7 and 8) becoming functional (Table 1, actions (n — 6 copulations, 5 rejections). Two subsequent Fig. 4) than to morphological growth. Interestingly, Barrantes exposures led to multiple copulations (// = 4) with alternation & Madrigal-Brenes (2008) do not report any apparent slowing between right and left sides each insertion. An extended of growth between instars, particularly the seventh instar when courtship sequence which led to two copulations is presented cribellate silk first appears in the webs of T. radiata. in supplementary materials (see video S5, online at http://dx. Allometric growth occurs between the penultimate and doi.org/10.1636/JoAOS-15-004.S5). adult molts in leg I of males, as has also been observed in T. Males of T. perfuga did not possess epiandrous spigots radiata (Barrantes & Madrigal-Brenes 2008). Similarly, in (Mallis, unpublished SEM data) and, therefore, may load the Pisaurina mira (Walckenaer, 1837), mature males have longer palps with sperm and seminal fluid deposited on the web. No legs relative to the overall body size than adult females, observations were made of males constructing a sperm web or particularly the first pair of legs (Anderson & Hebets 2016). Anderson and Hebets (2016) attribute this to allometry priming the palps. While courting, however, both male palps potentially driven by sexual selection. This is similarly were partially expanded. The male did preen at times just after hypothesized by Barrantes & Madrigal-Brenes (2008) for T. or between copulations if left in with the female, cleaning the radiata, as in the field, males were collected on or near female palps and first, second and at times even the third pair of legs webs, suggesting males abandon their webs in search of with his chelicerae. No stridulatory mechanisms, such as a file, females, as we suspect for T. perfuga. They proposed that were observed on the abdomen or carapace of adult male T. longer legs lead to larger step sizes to bridge the distances perfuga specimens. between male and female webs or to escape cannibalistic In the initial 24 pairings, 16 females were receptive, 6 were females; however, they did not associate the longer pair of legs not receptive, and 4 males did not court. Subsequent with courtship behavior or explicitly with sexual selection as a exposures (n = 11) of females (both mated and not) using possible mechanism for the allometry in Leg I (Barrantes & the same (n = 7) or different males (n = 4) resulted in more Madrigal-Brenes 2008). successful courtships with some pairs (n = 6) copulating Ontogeny of cribellum, calamistrum and cribellate silk use. multiple times (up to four times before removal of the male or The combing behavior for production of cribellate silk lines is the female retreated or became non-receptive. Females with similar to that reported for the closely related T. radiata egg sacs (n = 2) were not receptive to courtship, and either (Eberhard 1988). Individual spiders varied in their favored use ignored the male (n = 1) or non-aggressively drove them from of the right or left leg for combing (Mallis, pers. obs.). Some the web (« = 1). All virginal females had “plugs” prior to switched combing legs between one line and the next. Despite courtship encounters and these plugs generally appeared soon the physical presence of a cribellum and a partial to full after molting to adulthood. One female was examined under calamistrum, cribellate silk does not appear until the eighth the dissecting scope, post-copulation, and had a plug on left instar. This is interesting, as many zoropsid spiders have side (non-insertion) and no plug on the right (successful varied use of cribellate silk throughout their life cycles copulation). However, within a day of copulation, a plug (Lehtinen 1967; Griswold et al. 2005). In the closely related appeared on the right side. All mated females had sclerotized T. radiata, the apparent non-functional status of the cribellum 174 JOURNAL OF ARACHNOLOGY and calamistrum in early instars is suggested by the lack of Deinopoidea (cribellate horizontal orb-weavers) are more cribellate silk in the web until the seventh instar (Barrantes & closely related to the RTA clade than to the Araneioidea Madrigal-Brenes 2008). This was speculated to be due to the (viscous silk orb-weavers, etc.), as previously thought, making energetic costs of producing cribellate silk, or a reemergence of the historical “Orbiculariae” paraphyletic (Bond et al. 2014; a plesiomorphic condition (Barrantes & Madrigal-Brenes Fernandez et al. 2014; Garrison et al. 2016). The orb web is 2008). In Hyptiotes cavatus, newly emerged second instars considered plesiomorphic for the deinopoid Uloboridae, and also lacked a functional cribellum and calamistrum and did the modified cribellate webs of Hyptiotes Walckenaer, 1837 not form a web, but simply hung by a single line until molting (triangular orb, spider as the hub) and Miagrammopes O.P.- to the third instar (Opell 1982). Cambridge, 1869 (single capture thread) as derived or Ontogenetic changes in web structure.—Silk played a role in apomorphic traits (Opell 1982). Given these recent phyloge¬ many facets of life for T. perfuga, from foraging and shelter, to netic discoveries in the evolutionary history of spiders, one constructing egg sacs and in courtship. Early instar webs had would expect to find remnants of orb-weaving behavior or silk many characteristics of adult webs, but on a smaller, simpler use in members of the RTA clade (Agnarsson et al. 2013). scale and without cribellate silk (Fig. 6). These are acquired in Tengella spiders are members of the RTA clade, but, as the following order: basic retreats; small sheets; and knock¬ demonstrated by the spiral pattern of cribellate silk, still down lines. At the eighth instar, cribellate silk appeared in an exhibit some deinopoid orb-weaving behavior. This is orb-like spiral in the sheet, and subsequently throughout the consistent with the recent studies reporting Deinopoidea web structure (Figs. 6C, D & 7). The lack of retreats in lab ancestor to the RTA clade, and the ecribellate “Orbiculariae” spiders was likely due to the artifact of being in the lab in a as sister to the Deinopoidea + RTA clade (Agnarsson et al. small container, as most field-caught or observed spiders 2013; Bond et al. 2014; Fernandez et al. 2014; Garrison et al. actively used retreats at all life stages (Mallis, pers. obs.). Male 2016). T. perfuga lose functionality of the cribellum in adulthood. The initial orb-like spiral of cribellate silk only occurred Females line the sheets, edges of their web and tangle with across instar 8 (Fig. 7). Without a web ontogeny study, these cribellate silk, and maintain the webs until egg sac production. behavioral and structural characters that reflect the deinopoid Cribellate silk is not only used for prey capture, but also likely and orb web ancestry would have been missed. Using T. plays a role in courtship, propagating male and female perfuga as a focal study system, it would be of particular acoustic signaling (see Courtship below). interest to move from a web ontogeny study to a comparative In the web ontogeny study of T. radiata, Barrantes & study of silk spigot ontogeny across cribellate silk users Madrigal-Brenes (2008) reported that in the field, second including the former Orbiculariae and the RTA clade. A instar spiderlings did construct a collective molting web inside complete spigot ontogeny dataset of T. perfuga is forthcoming the mother's retreat and dispersed after molting to the third and a phylogenetic comparative analysis of spigot ontogeny instar, as in T. perfuga. In the lab, second instars removed data for several species is ongoing. These studies can further immediately after emergence from the egg sac did not readily elucidate ancestral orbicularian traits, such as the cribellate spin a web individually (Barrantes & Madrigal-Brenes 2008). spiral reported here in Tengella, in both cribellate and non- Third instar T. radiata constructed a dense horizontal sheet cribellate silk using spiders from the RTA clade. with retreats either below or above the sheet covered by tangle Egg sac construction.—Camouflaging of the egg sac by the lines. Webs were expanded through subsequent instars. Most female and tolerance of second instars is recorded in numerous importantly, the seventh instar is when cribellate silk lines are spider clades, such as the tetrablemmid Monob/emma much- observed in the tangle and sheet of the web, but the authors do morei Shear, 1978 (Edwards & Edwards 2006). In T. radiata, not indicate in what pattern it was observed (Barrantes & both lab and field observations reported similar camouflage Madrigal-Brenes 2008). techniques and placement of the egg sac by the female in her Similar to T. perfuga, the first capture webs of Hyptiotes retreat (Barrantes & Madrigal-Brenes 2008). cavatus in the form of a horizontal orb are constructed in the Courtship behavior.—Tengella perfuga has similar mating third instar, and subsequently become a cribellate triangular behaviors to those reported for T. radiata (Barrantes 2008), slice of an orb held tautly by the spider as the hub itself (Opell such as the male abdomen shaking and approach, as well as 1982). Males also ceased web production or maintenance in the strumming of the female web. Females, as in T. perfuga, adulthood (Opell 1982). This has also been demonstrated in assumed a passive position on their sides, exposing the another uloborid, Uloborus diversus Marx, 1898, where second epigynum to the male (Barrantes 2008). Similar broader instar webs were horizontal orbs, without cribellate spiral silk descriptive phases could be identified as in Barrantes (2008). (Eberhard 1977). Adult male Uloborus lack a functional These include (1) male preening and orientation to female, (2) cribellum and their web structure, if any, was similar to a courtship performance, and (3) copulation. Whereas the basic second instar web (Eberhard 1977). In the ecribellate modified steps tended to follow the same order, there were some notable orb web araneioids, similar ontogenetic patterns are observed differences. Female responses to the initial courtship of the with early instars spinning vertical sticky orbs and adults using male involved plucking or strumming with legs I and II, modified webs (Eberhard 1985, 1986). In the communal whereas T. radiata females used their palps (Barrantes 2008). araneid, Cyrtophora moluccensis (Doleschall, 1857), second Tengella radiata males had a rocking behavior while stilting instar spiderlings formed a nursery web, similar to the (Barrantes 2008), whereas T. perfuga males remained still and, collective molting web of T. perfuga, while adults had rather than rocking, either shook their abdomen or strummed communal webs which consisted of individual orbs (Berry the web. The process of male courting, female strumming 1987). response, and male advancement towards the female described MALLIS & MILLER—TENGELLA PERFUGA NATURAL HISTORY 175 for T. radiata by Barrantes (2008) is similar to that observed the lab, had epigynal plugs prior to mating (Mallis, pers. obs.). here in T. perfuga. Whereas Barrantes (2008) observed female While this has never been reported before in a spider, it is not attack behavior to repel male suitors, that kind of aggression entirely surprising, as some studies, particularly with Leucauge was not observed in the T. perfuga females in the lab. If a White, 1841 spiders, have found that females participate in lunge had occurred, no contact was made with the male and producing a mating plug both during and after copulation typically legs I and chelicerae were not outstretched as though (Aisenberg & Barrantes 2011). Therefore, it would not be a attacking (Mallis, pers. obs.). The male position during stretch to consider females producing an epigynal plug prior to copulation was different than that reported for T. radiata, copulation, which is a very novel observation. It is possible where male and female ventral surfaces are positioned parallel that the consistent behavior of flubs or ‘scraping' by the male to each other and touching while facing in opposite directions palp prior to insertion was an effort to remove the plug and (Figs. 3 & 4 in Barrantes 2008). Tengella perfuga males instead that played a role in female choice. Given their relative ease of reach across and over the female dorsum to access the rearing and large size, combined with variable mating epigynum typically while facing nearly perpendicularly with behaviors and other attributes, spiders of the genus Tengella the female. At times, due to web constraints, they were lend themselves well as a model system for both sexual and silk positioned parallel to each other. Palpal insertion and “flubs,” use evolution in spiders. defined by Barrantes (2008) as rapid scraping motions of the palp or failed embolus insertion attempts in T. radiata, were ACKNOWLEDGMENTS similar in T. perfuga. For example, once the palp successfully The authors would like to thank two anonymous reviewers engaged, the hematodocha expanded once, lasting less than for their insightful input in managing and editing this one second (Barrantes 2008). Also as in T. radiata, if multiple manuscript. We thank Dr. Jean-Michel Maes for his assistance successful copulations were allowed to proceed, T. perfuga with travel to and permitting in Nicaragua and collecting live females would expose the alternating side for copulation spiders in order to establish our lab colony. We would also like (Barrantes 2008). Sometimes female T. perfuga did not accept to thank Dr. Sandra Brantley for her review and editorial further copulation attempts and males were removed from the suggestions for this manuscript. We thank Matthew Biggs and containers. Matthew Leister for their assistance with lab colony mainte¬ The most notable and obvious difference between these two nance, care and courtship observations. Ms. Mallis would like closely related species was the lack of a bridal veil in T. to thank Dr. Charles Griswold for his discussion and radiata, as Barrantes (2008) never made mention of this in his encouragement to re-examine the web ontogeny data and courtship study. It may be that males did employ this observations for evidence of orbicularian traits. Specimens behavior; it was just not reported in the two courtship were collected in Nicaragua under Autorizacion Especial encounters of the study. Anecdotally, the first author collected DGPN / DB - 09 - 2012 and DGPN / DB - 006 - 2014 from T. radiata in Nicaragua (2012) for similar purposes, but was Carlos Ramiro Mejia, Director of Biodiversity, MARENA of unable to get a viable colony established. She introduced a Nicaragua. Portions of this project were supported by NSF male T. perfuga to a female T. radiata who accepted his grant DEB-0845984 (K.B. Miller, PI), as well as Grove courtship advances, including a heavy bridal veil, and allowed Scholarship and Graduate Research Allocation Committee him to copulate. Although an egg sac was produced, it was grants through the University of New Mexico Department of non-viable (Mallis, pers. obs.). While female T. radiata were Biology (R.E. Mallis). not reported to end courtship and/or copulation by breaking out of the bridal veil as T. perfuga, the movements of the legs LITERATURE CITED to pull themselves back to standing on the sheet as reported by Barrantes (2008), were similar to those of breaking out of the Aisenberg, A. & G. Barrantes. 201!. Sexual behavior, cannibalism, veil to stand in T. perfuga. and mating plugs as sticky traps in the orb weaver Leucauge argyra Similar overall mating behaviors have been observed not (Telragnathidae). Naturwissenschaften 98:605 613. only in T. radiata, which was selected for study because it is a Agnarsson, I., J.A. Coddington & M. Kuntner. 2013. Systematics: cribellate relative of the Lycosoidea, but also in closely related Progress in the study of spider diversity and evolution. Pp. 58 111. In Spider Research in the 21st Century: Trends & Perspectives. (D. lycosoids and agelenoids (Stratton et al. 1996; Huber 1998; Penney, ed.). Siri Scientific Press, Manchester, U.K. Barrantes 2008). While courting a female, Pisaurina mira Anderson, A.G. & E.A. Hebets. 2016. Benefits of size dimorphism males used their legs to help wrap her with silk before and and copulatory silk wrapping in the sexually cannibalistic nursery during copulation (Anderson & Hebets 2016). Consequently, web spider, Pisaurina mira. Biology Letters 12:20150957. Online at males typically with longer forelegs (Leg I), who could wrap http://dx.doi.org/10.1098/rsbl. 2015.0957 females, had increased sperm transfer and a lower likelihood Barrantes, G. 2008. Courtship behavior and copulation in Tengella of falling prey to cannibalism, suggesting some form of sexual radiata (Araneae, Tengellidae). Journal of Arachnology 36:606- selection occurred (Anderson & Hebets 2016). Many araneid 608. males also employed a plucking or strumming of the web Barrantes, G. & R. Madrigal-Brenes. 2008. Ontogenetic changes in web architecture and growth rate of Tengella radiata (Araneae, behavior (i.e.. Berry 1987). Tengellidae). Journal of Arachnology 36:545 551. While further study is needed, the potential lack of Barrantes, G. & W.G. Eberhard. 2010. Ontogeny repeats phylogeny choosiness on the part of females was not very surprising in Steatoda and Latrodectus spiders. Journal of Arachnology because in the field, males must wander to find females and 28:484-494. encounter rates could be inherently low at some field sites. Berry, J.W. 1987. Notes on the life history and behavior of the Also, several virginal females, collected in the field or reared in communal spider Cyrtophora moluccensis (Doleschall) (Araneae: