1The Neotropical spider Cupiennius salei Keyserling (Ctenidae) selects prey in a manner consistent with the amount of venom available in its venom glands. It distinguishes the venom sensitivity of different prey species, and uses its venom economically (according to the venom-optimization hypothesis).
2A prey-choice experiment was performed to test whether spiders use olfactory cues to detect prey and select prey items that are appropriate for their amount of available venom.
3The spider could choose between two similar prey dummies made of agar. We added the odour of two prey species, either by adding minced insects to an agar block or by offering it on filter paper which had previously been exposed to the living prey. Cupiennius salei spiders had either full or emptied venom glands.
4Two insects of distinctive venom sensitivity, but high acceptance, were tested: a sensitive cricket and a less sensitive cockroach.
5Using video surveillance, we found an attraction effect of prey odour in the prey-capture behaviour of C. salei. Spiders preferred agar pieces with minced insects or insect odour on filter paper over non-smelling items. Reaction frequency and attack rates were equal for spiders with full venom glands if they had to choose between cricket and cockroach odour. When the venom glands were empty, however, C. salei significantly preferred the venom-sensitive cricket over the venom-insensitive cockroach.
6We showed for the first time that C. salei uses its olfactory sense to detect prey items, and distinguishes between prey species with low and high sensitivity to spider venom. This study supports the venom-optimization hypothesis.
How C. salei obtains the information needed to differentiate between prey items is not fully understood. Our previous results suggest that C. salei receives important information via odour molecules emitted by the prey. This could enable the spider to distinguish prey species and to make further decisions within its behavioural sequence of prey catching. To evaluate the importance of odour and olfaction, we tested discrimination between agar pieces with minced insects or filter paper and with an insect smell against a control. In a second step, we tested to see if the spider made its prey-catching decision according to this information. The spiders had the choice between the odour of insects with a high (cricket) or low (cockroach) sensitivity towards its venom. These experiments were carried out using spiders with full or empty venom glands. Our results clearly show that C. salei uses olfactory information to make economical use of its venom.
Materials and methods
The experiments were conducted with the night-active wandering spider C. salei from our permanent breeding stock (Malli, Imboden & Kuhn-Nentwig 1998; Kuhn-Nentwig et al. 1998). The spiders were housed individually in 2-l glass jars at room temperature (0–25°C) with a light : dark regime of 12 : 12 h. Sub-adult (8-month-old) female and male spiders were used throughout the study. Each individual was used only once for one test, and prior to testing it was starved for 2 weeks. Spiders were raised exclusively on adult crickets (500–600 mg), Acheta domesticus (L.) (Gryllidae, Saltatoria) purchased from Grigfarm (Basel, Switzerland). In the experiment, the cockroach Nauphoeta cinerea (Blattodea) from a private breed was offered as alternative prey (we used only sub-adult specimens, body weight 500–600 mg); C. salei had no experience with other insects and was naive to the cockroach. The smell of these two insects is different in human perception, and they also differ considerably in their sensitivity to the venom of C. salei: A. domesticus is venom-sensitive with a low LD50 (2·1 nl mg−1); C. cinerea is venom-insensitive with a high LD50 (17·5 nl mg−1, Wigger et al. 2002; Wullschleger & Nentwig 2002).
We performed two prey-choice experiments: one with minced insects in agar, and one with insect smell on filter paper (Fig. 1). For the first experiment, we used 180 spiders and randomly divided them into six groups of 30 (15 females, 15 males). The first group received either an agar block with minced cricket or without cricket (control); the second group received agar with minced cockroach or without cockroach (control); and the third group received agar with minced cricket or minced cockroach. Spiders of the remaining three groups received the same prey choices, but underwent a procedure to empty the venom glands 24 h before the experiment started. For this experiment, spiders were anaesthetized with carbon dioxide and the venom was collected by electrical stimulation as described by Friedel & Nentwig (1989). Previous studies indicated that the spider C. salei does not suffer from such treatment (Boevéet al. 1995). The second experiment, also with 180 spiders, was identical to the first, except that the odour came from filter paper with an insect smell.
agar with minced cricket and cockroach
Agar pieces with minced insects served as prey items in the first prey-choice experiment. An equivalent of 5 g cricket or cockroach was frozen in the refrigerator for 2 h. Then the insects were cut into small pieces and mixed together with three droplets of distilled water in a homogenizer (Polytron-Aggregate, Kinematica, (Lucerne, Switzerland) until a consistent mass was obtained. Agar microgranulate (2·4 g, Merck, Darmstadt, Germany) and distilled water (60 ml) were boiled for 1 min in the microwave and subsequently stained with three droplets of black food dye (brilliant black PN) to prevent visual inspection of the agar block (Werna, Wollerau, Switzerland). Semi-fluid agar at 60 °C was added to the homogenized insect in a Petri dish (5 cm diameter × 1 cm height) and stirred intensively. Another Petri dish was filled only with semi-fluid agar and served as the control. We let the agar blocks harden for 3 h before we cut out agar pieces of 5 mm diameter × 7 mm height using a cylindrical stamping tool. The centre of the agar block was pierced with an iron pin (7 mm long). Despite their different content, all agar pieces had the same size, colour and form. Agar blocks were used immediately or stored in a refrigerator for 2 days; otherwise new preparations were made every 3 days.
agar blocks with insect odour on filter paper
In this second prey-choice experiment, only insect odour was offered. Round filter papers (2 cm diameter) from the laboratory storage (control) or with insect odour were glued with bonding foil under the agar blocks. Insect-odour filter papers were placed for 36 h in a box (60 × 40 × 40 cm) where 40 crickets or cockroaches (body weight 500–600 mg each) were stored. Temperature was 20–25 °C, air humidity 70–80%, and no food or water was supplied. The adult insects were in constant contact with these filter papers. After removing the papers, the paper smell could clearly be distinguished by humans. The agar blocks were prepared as described above, but did not contain minced insects.
One experimental set-up contained two magnetic stirrers (IKA-Combimag RCH, Janke & Kunkel, Staufen, Germany), side-by-side, contact-free and surrounded by a wooden frame. The magnetic stirrers were used to generate movements of the agar block to simulate prey movement. The frame supported a plastic box (32 cm long × 22 cm wide × 9 cm high, closed by a lid and with rectangular corners), which served as prey-choice arena. A 2-mm space between the upper ridge of the magnetic stirrer and the bottom of the plastic box guaranteed that there was no direct contact between stirrer and box. Every day the interior of the box was covered with clean white filter paper. A hole with a diameter of 5·5 cm was cut in the centre of one of the walls of the plastic box and a plastic tube (5·3 cm diameter × 10 cm length) was inserted in this opening. Between 4 and 5 pm, a spider was anaesthetized with CO2 and released in the tube, which was closed behind the spider. Two agar pieces were placed on the bottom of the box with 19 cm distance between them. Due to the iron pin, both agar pieces turned with the same speed when the magnetic stirrer was activated. As soon as the spider gained consciousness, it could leave its retreat tube and enter the prey-choice arena. The distance from the exit of the tube to both prey items was 12 cm. We performed the feeding experiments simultaneously with six of these devices.
From 5 pm, after starting the experiment, the activity patterns of C. salei (a nocturnal species) were monitored for 12 h using infrared video cameras (Conrad Electronics, Solothurn, Switzerland), a time-lapse video-cassette recorder (VCR, Panasonic AG-6124, Lucerne, Switzerland), and a multiplexer (Sony YS-SX310P, Schlieren, Switzerland). Details of the video equipment and installation are described by Schenk & Bacher (2002). Tapes were analysed for predation events in the laboratory on an additional VCR-multiplexer set connected to a video monitor (Sony PVM-14N6E). We recorded movement of the spider after it had left the plastic tube and entered the prey-choice arena as a reaction. This behaviour includes the following steps: in the opening of the plastic tube the spider orients its forelegs, then the body towards one side of the arena, it walks 5 cm in this direction and waits on the vertical wall until it decides to attack prey. The approach to a prey item was regarded as attack when the spider inserted its chelicerae in the target. We evaluated reaction and attack rates, choice frequency, and sex of the attacking spider using χ2 tests. The latency, time span (min) between onset of the experiment and first reaction, and attack on a prey item were analysed using the non-parametric Kruskall–Wallis test (SPSS).
The majority of the spiders reacted towards the prey (χ2 = 54·444 insects, χ2 = 64·187 filter paper, P < 0·001 for both experiments, N = 90) and attacked it (χ2 = 12·844, P < 0·001 for agar with insects; χ2 = 4·444, P = 0·035 for filter paper, N = 90) when their venom glands were full (Table 1). After the venom glands had been emptied, spiders still showed interest in the prey items (χ2 = 17·778 insects, χ2 = 14·400 filter paper, P < 0·001 for both experiments, N = 90) but attacked significantly less often (χ2 = 11·378, P = 0·001 for agar with insects; χ2 = 10·000, P = 0·002 for filter paper, N = 90). Comparing the proportions of reaction and attack rates of spiders with full and empty venom glands, we found that spiders with empty venom glands were less active, more reserved, and reacted less often to the presented object. Spiders reacted and attacked less often in both experiments with empty venom glands (reaction: χ2 = 7·980, P = 0·004 for agar with insects, χ2 = 14·500, P < 0·001 for filter paper; attack: χ2 = 24·200 insects, χ2 = 13·930 filter paper, P < 0·001 for both experiments, contingency table, full vs empty venom glands). On average, spiders attacked 3–6 h (medians of box plots, graphs not shown) after the experiment had started. The latency (time span in min until the first attack occurred) was not different between treatments (P > 0·5, Kruskall–Wallis non-parametric test). The attack rate of female (n = 92) and male (n = 84) juveniles was equal (χ2 = 0·364, P = 0·546).
Table 1. Number of spiders (Cupiennius salei) showing a reaction toward the prey item and subsequent feeding behaviour with full and empty venom glands (contingency table, full vs empty venom glands)
Venom glands (numbers yes/no)
Agar with insect
Filter paper with odour
Agar with insect
Filter paper with odour
importance of insect odour
Cupiennius salei showed a significant preference for prey items with insect odour from agar blocks or from filter paper, as indicated by reaction and attack rates (P < 0·05, χ2 tests for all treatments except filter paper control/cockroach: χ2 = 0·400, P > 0·05; Table 2). Additionally, we observed that the spider made its choice straightforwardly and almost never switched between targets (two switches between targets out of 176 attacks). The spider always attacked the same item to which it reacted previously (χ2 = 168·091, P < 0·001). This leads us to suggest that the smell of a prey item can be perceived by a spider over a distance of at least 12 cm under our test conditions.
Table 2. Prey-choice experiments with Cupiennius salei spiders
Agar giving off insect odour is significantly preferred to agar without odour, independently of whether venom glands are full or empty. However, overall reaction and attack rates are higher for spiders with full venom glands (χ2 tests of independence). Thirty spiders were tested in each group; spiders that did not react at all are excluded from the analysis.
Agar with insect
Agar with insect
influence of venom sensitivity
Spiders with full venom glands did not discriminate between prey items with different venom sensibilities. In both experiments, the reaction rates did not differ when the spider had to choose between agar with insects and filter paper dispersing the odour of insects (Fig. 2a,b). With empty venom glands, spiders showed a preference (reaction rates) for crickets (low LD50) (minced cricket in agar, χ2 = 12·462, P < 0·001; filter paper with cricket smell, χ2 = 5·000, P = 0·025). We obtained the same results for attack rates (minced cricket in agar, χ2 = 8·333, P = 0·004; for filter paper with cricket smell, χ2 = 6·400, P = 0·011) (Fig. 3a,b).
We performed this prey-choice study to investigate the importance of olfaction in the prey-capture behaviour of the spider C. salei. There is evidence that the spider selects prey items consistent with the amount of venom available in its venom glands (Wigger et al. 2002; Wullschleger & Nentwig 2002). Our results confirm these suggestions: after depletion of the venom glands, the spiders are more selective. Analyses of video recordings showed that, after their venom glands had been emptied electrically, spiders were oriented towards prey less often and displayed decreased attack rates (Table 1).
The results of this study support our hypothesis that the spider C. salei (1) uses olfactory orientation when catching prey, and (2) uses this information to select appropriate prey according to the amount of venom available. Odour is a guide to locate the prey. We demonstrated that the spider can sense odour at a distance of at least 12 cm when offering an agar piece with minced insects or a pure agar piece on filter paper with an insect smell. The spiders preferred the agar pieces dispersing the odour of crickets/cockroaches, and moved straight towards the selected prey without switching to another target. They attacked these prey objects without any previous direct contact, providing clear evidence that C. salei is able to perceive odour molecules through the air. The agar piece with minced insects appears to emit odour molecules. The spider's reaction after biting into a pure agar block showed that it realized the prey (in the filter paper experiment) had no energetic content, as it dropped the agar piece. In contrast, items with minced crickets or cockroaches were carried away to the retreat inside the tube or into a corner of the box. Spiders chewed on the agar block with minced insects, probably ingested some parts, and returned to the site where they were before the attack.
According to the venom-optimization theory (Wigger et al. 2002), C. salei uses its venom as economically as possible. The spider is able to inject the precise amount of venom needed to paralyse its prey. Consequently, it is necessary for the spiders to identify the prey type or distinguish its venom sensitivity in order to save venom. Our findings suggest that the odour of the prey is an indicator of the prey's susceptibility to the venom of C. salei. The spider selected the agar piece with insects and the agar piece on filter paper dispersing the odour of a cricket (indicating high venom sensitivity) with empty venom glands. The agar pieces offered as prey were identical except for the odour; all other aspects, such size, weight, form, colour and speed of turning, were kept constant for all experiments.
Some reports neglect the possibility that C. salei uses its olfactory sense to detect and locate prey. One explanation could be that motionless prey, such as dead flies, were offered (Seyfarth & Barth 1972). Usually C. salei does not attack immobilized prey. This leads us to suggest that airborne signals or substrate vibrations are the first information in a sequence, indicating the occurrence of prey in the closer environment. Additional information, such as odour (or view in some species), provides conditional factors for the spider in deciding whether to attack or not. Previous experiments focusing on the sense of smell demonstrated that C. salei recognizes conspecifics with pheromones attached to their dragline (Tichy et al. 2001). Behavioural studies indicate that the desert spider Agelenopsis aperta responds with courtship to a volatile pheromone emitted by females (Papke et al. 2001).
This is the first report showing that C. salei is olfactorily attracted to prey, and distinguishes venom sensitivities of different prey species on the basis of its odour. Prior arena experiments by Persons et al. (2001) reported that the wolf spider Pardosa milvina showed a significantly reduced activity level when placed on filter paper previously exposed to faecal material from a natural predator. Punzo & Preskhar (2002) demonstrated in feeding experiments that lycosid spiderlings learn to respond to odour associated with three different nymphs of Gryllidae. All spiders in our experiment were used only once, thus learning can be excluded. The fact that spiders with full venom glands in both experiments showed no preference for the crickets (initial diet) confirms that the nature of selection is not acquired. Chemical analysis of the substances emitted by crickets and cockroaches could be helpful for understanding which substances affect the spider, and how this information is used.
Previous investigations into prey size, activity pattern, defensive behaviour, thickness of chitinization of prey, risk to the predator, and degree of hunger of a spider have demonstrated the importance of these parameters within the prey-capture behaviour of the model spider C. salei. We have shown that olfactory input is also very important and has to be taken into account in understanding these complex predator–prey interactions.
We are grateful to L. Kuhn-Nentwig and B. Wullschleger for support in the laboratory and for many helpful hints. We thank E. Jutzi for technical aid, B. Tschanz and S. Bacher for providing the video surveillance system, and J.-P. Airoldi for statistical advice. This work has been supported by the Swiss National Science Foundation.