M. E. Hauber, Department of Integrative Biology, 301 VLSB, University of California, Berkeley, CA 94720-3140, U.S.A. E-mail: email@example.com
Abstract 1. Conspicuous body colouration is counter-intuitive in stationary predators because sit-and-wait tactics frequently rely on concealed traps to capture prey. Consequently, bright colours and contrasting patterns should be rare in predators using traps as they may alert potential prey. Yet, some orb-weaving spiders are brightly coloured and contrastingly patterned. How can conspicuousness of trap-building sit-and-wait predators be favoured by natural selection?
2. Observations of spiny spiders Gasteracantha fornicata in north-eastern Australia showed that the size of spiders relative to their orb webs correlated positively with relative prey numbers already captured in their webs. A possible explanation is that the relatively larger appearance of the yellow–black striped dorsal surface of this spider attracts more visually oriented prey items. Prey attracted to webs may get trapped, thereby increasing the spiders' foraging success.
3. To test this hypothesis for the function of conspicuous body colouration, a field experiment was conducted that documented the prey capture rates of spiny spiders after manipulating or sham-manipulating their appearance.
4. As predicted, spiders that were dyed black on their striped dorsal surface caught relatively fewer prey items than did control spiders. Thus, conspicuous dorsal body colouration may be adaptive in spiny spiders because it increases foraging success and, presumably, survival rates and reproductive outputs. Overall, these data support the colour-as-prey-attractant hypothesis in a stationary, trap-building predator.
There are several hypotheses about how conspicuousness in general, and bright or contrasting body colouration in particular, may be adaptive in stationary or slow-moving taxa (Cott, 1940). One possibility is that colouration and contrasting visual patterns of stationary predators and/or their traps [(e.g. the body colouration of Argiope argentata spiders (Craig & Ebert, 1994) and their bright zigzag patterned web decorations or stabilimenta (Craig & Bernard, 1990)] are perceived to be attractive by potential prey items. According to this prey-attractant hypothesis, visually oriented insects are lured actively to the vicinity of the web (Tso, 1998; Watanabe, 1999), where the spiders increase their prey-encounter and prey-capture rates and, hence, their foraging success (Tso, 1996; Hauber, 1998; Herberstein, 2000).
An alternative explanation is that the conspicuousness of the sit-and-wait predator and/or its trap (e.g. the stabilimenta of Argiope; Eisner & Nowicki, 1983) warns fast-moving and large animals that are unsuitable as prey not to hit the trap (Horton, 1980). According to this warning hypothesis, bright colouration of the spider and/or its trap is beneficial because it reduces the rate of unprofitable web damage (Eisner & Nowicki, 1983; Kerr, 1993; Blackledge & Wenzel, 1999).
Yet another function of conspicuous colour in sit-and-wait predators may be unrelated to their predatory life histories. As seen in several other brightly coloured organisms, conspicuous colour patterning may be an aposematic signal (Brodie, 1993) that reduces the rate of attack by potential predators of the trap builders themselves (Schoener & Spiller, 1992). Also, unrelated to predatory foraging strategies is the possibility that bright colours evolved as sexually selected signals of mate choice or intra-sexual competition. Finally, despite the apparent conspicuousness of colouration to the human eye, predators' and their traps' bright colours and disruptive colouration [e.g. ventral and dorsal colouration of Argiope spiders (Nentwig & Rogg, 1988) or the bright silk colour of the webs of Nephila clavipes (Craig et al., 1996)] may actually be cryptic when viewed by different visual systems (such as those of insects or birds) and in the natural habitat of the predator (Blackledge, 1998).
Previous observational and experimental studies regarding the function of bright body and trap-colouration in stationary predators, provided some support for the prey-attractant, warning, aposematic signal, and cryptic hypotheses but not for the sexually selected signals hypothesis, sometimes even in the same taxa (see above for references, also reviewed for web stabilimenta by Herberstein et al., 2000a). These findings suggest that one or more functional roles may be served by the same signal yet, to date, comparative data from evolutionarily independent taxa are too sparse to predict to a finer degree which function would be most likely to evolve in a specific set of ecological and life-historical contexts. Hence the need for further empirical studies.
Spiny spiders Gasteracantha fornicata are orb-weavers of the north-eastern tropical rainforest in Australia where they frequently inhabit the boundaries between forests and clearings (Mascord, 1970). Although the web silk of this spider appears inconspicuous to the human eye, the spider itself, sitting in the orb's hub, is coloured with contrasting white–yellow and dark dorsal stripes (Mascord, 1970; M. E. Hauber, pers. obs.; Fig. 1). In contrast, its ventral side is darker overall and finely spotted with small yellow dots. In support of the aposematic signal hypothesis, spiny spiders have hard exoskeletons and sharp spines that may protect them from avian predators (Mascord, 1970).
Conspicuousness is also found in other Gasteracantha species (Edmunds & Edmunds, 1986; M. E. Hauber, pers. obs.). In G. curvispina, adults but not early instar spiderlings have distinct colour-morphs that are found at varying densities in different microhabitats (i.e. open vs. under trees) and their survival rates in either of these habitats depend on their own morph colouration (Edmunds & Edmunds, 1986). This suggests that aposematism is not the sole function of bright (and variable) colouration in some spiny spider relatives.
Despite the widespread occurrence of spiny spiders near human-modified habitats (e.g. artificial clearings; Mascord, 1970), no study has examined the potential function of bright body colouration in these sit-and-wait predators. To start addressing this question, the visual appearance of spiny spiders was manipulated in a field experiment to test predictions of the colour-as-prey-attractant hypothesis. This hypothesis predicted that decreasing the conspicuousness of spiders would result in a decrease in the foraging success of manipulated spiny spiders.
Materials and methods
The study was conducted on spiny spiders near Daintree National Park, Queensland, Australia (16°9′S, 145°27′E) in December 1999. By walking around marked trails of the coastal rainforest, spiny spiders (n = 21) were located at heights < 3 m. All spiders located between 07.00 and 15.00 hours that had a spinetip-to-spinetip (i.e. size; Fig. 1) measurement of ≥10 mm (an arbitrary threshold to limit observations to mature female spiders, M. E. Hauber, pers. obs.) were included in the study. Because spiny spiders take their webs down at night and construct new webs in the morning, all observations were limited to daylight hours.
The following measurements were collected for each spider: size (to the nearest mm), length of vertical and horizontal radii of the orb-web (cm), number of silk threads crossing an imaginary line drawn from the hub to the upper edge of the orb, and number of prey items trapped by the silk of the web. Most prey items trapped in spiny spiders' webs were small (≤1 mm) dipterans (≥90%; M. E. Hauber, pers. obs.). An examination of the mouthparts of each spider to determine whether they were feeding actively at the time of discovery revealed that these dipteran prey were indeed consumed by these spiders (M. E. Hauber, pers. obs.), suggesting that small insects trapped in the web contributed to the spiders' diet.
In addition to prey numbers, the number of damage patterns visible in each web was also determined. A damage pattern was defined as any contiguous visible break or hole in the silk structure of the orb (following Craig, 1989; Hauber, 1998). To control for the effect of the size of the spider's trap, total mesh length was calculated using radii and thread number as: 0.25 × (upper radius + lower radius + left radius + right radius) × upper thread number × 3.14 (modified from Sherman, 1994; Herberstein & Tso, 2000).
After measurements were taken, each spider was assigned randomly by the toss of a coin into one of two experimental groups. Dyed spiders (n = 11) were removed from their webs and painted black using a magic marker on their dorsal (striped) surface. Control spiders (n = 10) were sham-manipulated by removing and dyeing them black on the ventral (spotted) surface. There was no difference between the body sizes, mesh lengths, and the times of day (to the nearest minute) of first visits of dyed and control spiders (Mann–Whitney tests, all P > 0.25).
For dyeing, the spiders were restrained using soft mesh for ≤ 2 min. Dyeing seemed to be a successful way of changing the appearance of spiny spiders for ≥ 24 h, as observed by the human eye during repeated visits to the same webs (M. E. Hauber, pers. obs.; Fig. 1). After treatment, subjects were returned immediately to their respective webs. Following an initial period of immobility (≤ 3min), all spiders included in this study returned to the hub of their respective webs (M. E. Hauber, pers. obs.).
Each web was revisited after 75.2 ± 6.32 min (mean ± SE) but spiders could not be relocated even after extensive searches of the nearby vegetation at n = 2 dyed and n = 1 control sites. During these second visits, the number of prey items and damage patterns (holes) present in the web was counted again. There was no difference in the duration of time elapsed from the first to the second visits between treatment groups (Mann–Whitney test, P > 0.96).
The prey-capture success of spiders prior to manipulation was analysed using multiple linear regressions to examine the relationship between spider size/mesh length and time of day vs. prey number (at the time of the first visit)/mesh length and damage patterns/mesh length. These measures were devised a priori to incorporate the potential effects of web size and insect activity (Herberstein et al., 2000b). To compare the effects of manipulations, relative prey rates and damage rates were calculated as: (number of prey or damage patterns at the second visit – number of prey or damage patterns at the first visit)/(time passed between visits × web mesh length) respectively. Proportional variables were log (x + 1) transformed for linear regressions while non-transformed data were used for non-parametric comparisons and the illustrations.
To collect additional information on the position and orientation of spiders within the hub of their webs, an additional n = 19 spiders (i.e. not included in the experiment) were located and the distance between the web hub and the nearest vegetation or the ground was also measured.
Observational data on the webs of spiny spiders during the first visits indicated that the relative number of prey trapped in the web was related positively to both the relative size of the spider and the time of observation (linear regression on log-transformed values: Prelative spider size = 0.026, Ptime of day = 0.017, overall R2 = 0.40, n = 21) (Fig. 2). There was also a positive relationship between the relative damage patterns and the relative size of the spider but not the time of observation (linear regression on log-transformed values: Prelative spider size < 0.001, Ptime of day = 0.14, overall R2 = 0.61, n = 21).
As predicted by the colour-as-prey-attractant hypothesis, relative prey capture rates were lower for experimentally dyed spiders (ndyed = 9) than for control spiders (ncontrol = 9; Mann–Whitney test, P = 0.027; Fig. 3). Relative damage rates were similar between the two treatment groups (dyed: 1.6 × 10−5 ± 9.0 × 106, ndyed = 9; control: 4.6 × 10−6 ± 1.1 × 10−5, ncontrol = 9; Mann–Whitney test, P = 0.45).
The observations on the orientation of additional spiny spiders (n = 19) within their webs revealed that the striped dorsal surface was always slanted towards the ground (100%; binomial test, P < 0.001) and tended to face towards nearby vegetation and away from openings or more distant vegetation (74%; binomial test, P = 0.064, random expectations: 50%).
Bright body colouration and contrasting patterns of the dorsal surface of spiny spiders make them conspicuous to human observers and, perhaps, to other visually oriented animals with different sensory systems, including insects and birds. Yet, spiny spiders, stationed at the hub of their inconspicuous orb-webs, are locally abundant and clearly successful at avoiding recognition by potential prey items and detection by their own predators. Indeed, in this observational data set, a relative measure of prey capture rate was correlated positively with spider size when web area and time of day were controlled statistically (Fig. 1). This suggests that body colour does not warn potential prey items to avoid spiders' traps. Instead, conspicuous colouration may be attractive to visually oriented prey (Craig & Ebert, 1994), so that when the conspicuous body surface (spider size) was larger relative to trap size (web area, approximated here by mesh length), the observed prey capture success (number of prey items/mesh length) was also greater.
Statistically significant relationships from observational data sets, however, can also be epiphenomenonal if they are the outcome of differences between relatively large and small spiders due to causes other than relative conspicuousness (e.g. variation in habitat selection, size-dependent ability to subdue prey, etc.). Therefore, the colour-as-prey-attractant hypothesis was also tested using an experimental manipulation of the stationary predators' contrasting colour patterns. As predicted, black-dyed spiders had lower relative prey capture rates than did control, sham-manipulated spiders. The inference is that the colouration pattern of spiny spiders functions to increase the foraging success of these stationary predators.
A common criticism of the prey attraction function of both conspicuous web structures and body patterns of spiders is that stabilimenta are not constructed of sticky silk that can effectively trap prey and spiders are typically stationed in the hub of their web where sticky silk is absent (Eisner & Nowicki, 1983; Blackledge, 1998). How can prey attracted to a functionally non-trap location be captured? The simplest possibility is that increasing the number of prey in the proximity of webs alone can lead to greater rates at which prey touch and become entangled in the otherwise inconspicuous sticky web silk (Craig & Bernard, 1990). Alternative scenarios may involve perceptual and cognitive interference between the responses of approaching colourful and patterned objects and avoiding orb-webs (Craig, 1994). Careful observations on the movements of individually marked prey items in the vicinity of traps will be required to determine which explanation fits spiny spiders and their prey best.
In several previous studies (Craig, 1989; Craig & Bernard, 1990; Tso, 1996; Hauber, 1998; Craig et al., 2001) damage pattern numbers (defined similarly to the method used here) were validated and used as approximate measures of prey capture success. Assuming the validity of this measure, the finding that there was a positive relationship between relative damage patterns and relative spider size at unmanipulated webs (i.e. at first visits; see Results) also supports the prey-attractant hypothesis. Yet, there was no significant relationship between experimental colour treatment and subsequently recorded relative damage rates in webs of spiny spiders. Because most observed prey were small dipterans trapped by a single thread of silk, it is possible that damage patterns in spiny spider webs are not caused primarily by the entanglement of the prey but rather created when the spider moves about to remove trapped prey. Perhaps the differences in prey capture rates are not reflected by the relative rates at which new damage patterns appear until the spider moves about to collect its prey.
To test predictions of the colour-as-warning hypothesis for conspicuousness in stationary predators, damage patterns or damage rates have also been used as measures of the rates of web deterioration caused by birds or fast-moving insects (Eisner & Nowicki, 1983; Blackledge, 1998; Blackledge & Wenzel, 1999); however these authors used different criteria for web damage than those used here. Assuming that damage rates do reflect web deterioration caused by unsuitable prey travelling with high kinetic energies (Craig, 1989; Hauber, 1998), the present data do not lend support to the colour-as-warning hypothesis because there was a negative relationship between relative spider size and initial hole numbers. This is opposite of what is predicted by the warning hypothesis (i.e. more conspicuous spiders should have fewer damage patterns). Also, contrary to these latter predictions, there was no statistical relationship between the dyeing treatment and subsequent damage rates. Nevertheless, direct observations using video cameras or measurements based on the overall area of webs before and after treatments may describe more accurately the extent, relevance, and importance of web damage caused by fast-moving or large objects in spiny spiders.
Although the presented data support the colour-as-prey-attractant hypothesis, it is possible that the control, sham treatment of dyeing the ventral surface of spiny spiders black actually increased the attractiveness of the spider to visually oriented prey, resulting in a differential outcome of prey capture rates between treatment and control spiders that is consistent with the prey-attractant hypothesis. This was unlikely, however, because the net change in prey numbers within the same webs was consistently positive between visits for control (Wilcoxon signed rank test, P < 0.011) but not different from zero for dyed spiders (Wilcoxon test, P > 0.48). Assuming that webs trap an increasing number of prey throughout daylight hours, the implication is that prey capture was reduced for dyed spiders rather than increased for control spiders due to treatment effects.
Why should prey items be attracted to bright body colouration of spiny spiders? One possibility is that the colours and patterns of some spiders' dorsal surface mimic the colours of food items (e.g. flowers) of visually oriented prey species (e.g. Trigona stingless bees are attracted to flower-like patterns of Argiope argentata spiders; Craig, 1994). More generally, bright body colouration may be perceived by the visual systems of potential prey items as similar to any class of attractive objects or locations. For example, it is possible that some portions of the dorsal surface of spiny spiders are highly reflective in the ultraviolet spectrum (although yellow dorsal colours of A. argentata do not reflect ultraviolet light; Craig & Ebert, 1994). Many dipterans, in turn, are attracted to bright light and light rich in ultraviolet spectra because these indicate the presence of clearings in dense forests (Craig & Bernard, 1990; Tso, 1998; Watanabe, 1999). The observations reported here on the orientation of unmanipulated spiny spiders foraging in their webs showed that the striped dorsal surface always faced towards the ground and tended to be oriented towards the ground or nearby vegetation and away from openings or more distant vegetation. Diurnal web building and hunting in spiny spiders, together with such selective placement and directional orientation of the spider and its web, are also consistent with a foraging strategy that aims to attract small potential prey items travelling from darker to lighter patches (i.e. the colour-as-prey-attraction hypothesis) rather than a functional role of advertising the location of the spider's web to fast-moving insects and birds heading towards vegetation (i.e. the colour-as-warning hypothesis) (Elgar et al., 1996).
Even though under these scenarios attraction by potential prey items to bright colour patches will be maladaptive at some frequency (i.e. because occasionally they get trapped by spiny spiders), overall the attraction to bright colours by prey species can remain an evolutionary stable orienting response. This stability will be favoured when predator densities [i.e. trap encounter rates: acceptance errors sensuReeve's (1989) optimal acceptance threshold theory] are lower and the benefits associated with arrival at bright patches (i.e. foraging or mating opportunities: rejection errors sensuReeve, 1989) are greater. Unfortunately, ecological and behavioural data on spiny spiders and their potential prey items are not available to evaluate these predictions.
Overall, the data reported here are consistent with the possibility that spiny spiders actively select microhabitats and web orientations that maximise the visibility or attractiveness of their dorsal colouration. Future work should, therefore, measure the reflectance spectral properties of both the dorsal and ventral surfaces of these spiders within their foraging microhabitats. In addition, to evaluate the adaptive benefits of spider colours fully, the composition of actual prey items in the diet of spiny spiders should be examined to determine whether, as predicted, most of these predators' prey items are visually oriented and attracted to the spiders' conspicuous colouration. Finally, further experimental tests will be required to evaluate the applicability of other possible, and not necessarily mutually exclusive, functions of bright colouration in spiny spiders.
Many thanks are due to W. Shields, B. Hager, and D. Ardia for advice and equipment, and SUNY-CESF, Syracuse, for the opportunity to carry out my research in Australia. The manuscript benefited from constructive discussions with and comments by M. C. B. Andrade, J. C. Biesmeijer, C. L. Craig, T. Eisner, M. E. Herberstein, J. M. Hernandez, S. R. Leather, S. M. Murphy, N. Pierce, P. T. Starks, D. F. Westneat, and anonymous reviewers. This research was supported by a Howard Hughes Medical Institute Predoctoral Fellowship.