Behavioral
Ecology
The official journal of the
ISBE
International Society for Behavioral Ecology
Behavioral Ecology (2016), 27(5), 1424–1431. doi:10.1093/beheco/arw048
Original Article
Linking consistent individual differences in
web structure and behavior in black widow
spiders
Nicholas DiRienzoa,b,* and Pierre-Olivier Montiglioc,d,e,*
aDepartment of Ecology and Evolutionary Biology, University of Arizona, PO Box 210088, Tucson, AZ
85721-0088, USA, bDepartment of Neurobiology, Physiology and Behavior, University of California at
Davis, One Shields Avenue, Davis, CA 95616, USA, cDepartment of Environmental Science and Policy,
University of California at Davis, One Shields Avenue, Davis, CA 95616, USA, dDepartment of Biology,
McGill University, 1205 Dr Penfield Avenue, Montreal, Quebec H3A 1B1, Canada, and eRedpath
Museum, McGill University, 1205 Dr Penfield Avenue, Montreal, Quebec H3A 1B1, Canada
Received 15 November 2015; revised 22 February 2016; accepted 8 March 2016; Advance Access publication 22 April 2016.
Individuals may vary consistently in their architectural makeup of the constructions they build, which might have tremendous implications for their ecology and fitness. In particular, the relationship between individual differences in architectural constructions and
behavior (i.e., animal personalities) is largely unexplored. Individual black widow spiders, Latrodectus hesperus, build a 3D cobweb
made up of distinctive components serving to support foraging (gumfooted lines) or antipredator protection (structural lines). To
explore the relationship between individual differences in behavior and architectural constructions, we quantified 1) the level of consistent individual variation in several elements of web structure and 2) the level of consistent individual variation in foraging behavior
and its relationship with web structure. We controlled for condition-dependent or environmental effects by satiating all spiders prior
to assays and maintaining them in standardized conditions. Spiders exhibited consistent differences in the number of gumfooted lines
they built for capturing preys and overall web weight, but not in the number of structural lines. Individuals also varied consistently in
their tendency to attack a prey cue, despite all spiders being satiated. Finally, spiders producing more gumfooted lines exhibited a
higher tendency to attack the prey cue. Our results suggest that the architectural constructions may impact the expression of individual behavioral differences (or animal personalities) and suggest that individual behavior and extended phenotype may be part of
alternative foraging strategies in L. hesperus.
Key words: animal architecture, animal personality, behavioral syndrome, extended phenotype, foraging, individual differences,
Latrodectus hesperus.
INTRODUCTION
Many animals create architectural constructions (Hansell 2005),
such as bird nests, mammal burrows, or spider webs (Blamires 2010;
Kelley and Endler 2012; Smith et al. 2015). Animal constructions
are important for many aspects of the animal’s ecology, such as foraging, mating interactions, reproduction, safety against predation
or parasitism (Whitehouse and Jaffe 1996; Blackledge and Wenzel
2001; Pinter-Wollman et al. 2012; Smith et al. 2015), and can vary
as a function of ecological conditions (Tso 2004; Zevenbergen
et al. 2008; Doerr and Endler 2015). Constructions have also been
Address correspondence to N. DiRienzo. E-mail: [email protected].
*These authors contributed equally to this manuscript.
© The Author 2016. Published by Oxford University Press on behalf of
the International Society for Behavioral Ecology. All rights reserved. For
permissions, please e-mail: [email protected]
shown to vary consistently among individuals (Rushbrook et al.
2008; Doerr and Endler 2015), often as a result of individual differences in resource availability or condition (Anotaux et al. 2012;
Japoshvili et al. 2012; Doerr and Endler 2015). Furthermore,
constructions are often coexpressed with behaviors (Borgia 1995;
Zevenbergen et al. 2008; Niemelä et al. 2015) as they need to work
in concert to be effective. However, few studies have investigated
whether an individual’s behavior and construction covary because
of environmental conditions or because of fixed or genetic link (but
see Blackledge and Zevenbergen 2007; Niemelä et al. 2015). This
gap limits our understanding of the ecological and evolutionary
role of animal constructions. For example, individual behavior and
construction could be linked because each of these traits is adjusted
independently in response to environment conditions or changes in
DiRienzo and Montiglio • Individual differences in web structure and behavior
observe spiders with webs structured for prey capture and a higher
tendency to attack prey cues, whereas other spiders might instead
produce webs offering increased protection and express a lower
tendency to attack prey cues.
MATERIALS AND METHODS
Subjects
We used field-caught mature female L. hesperus black widow spiders
(N = 17). All spiders were captured in Davis, CA, in August 2014.
Once in the laboratory, all spiders were given an individual container (6 cm high × 8.5 cm diameter) and were held at a 12:12 h
light:dark cycle at 24 ± 1 °C. The spiders were provided with ad
libitum food in the form of 3 large (0.3–0.5 g) Acheta domesticus crickets per week. The spiders were maintained in the laboratory for
approximately 3 months before beginning this experiment. Thus,
given the large amount of food they were provided, the spiders
likely had maximized their fat stores.
Spider web assessment
elt
er
In order to assess variation in web structure, we provided spiders
with a standardized structure on which to build. Five days after
being fed, each spider was transferred to a skeletonized cardboard
box (L 27.5 × W 21 × H 14 cm). The box had 3 walls and all but
3 cm of the top removed, thus leaving a rectangular cardboard
frame with the back, bottom, and the portion of the top walls
remaining (Figure 1). The bottom and back walls were covered
in black paper in order to ease the counting of web components.
This structure provided a shelter along with a frame on which to
build their web. This box was placed inside a larger plastic Sterilite
container (L 42.5 × W 27.75 × H 16.25 cm), and the spider was
allowed to build its web for 7 days. Spiders were weighed before
being introduced to the container. These methods are similar to
those used by other researchers in order to assess web structure
(Blackledge and Zevenbergen 2007; Zevenbergen et al. 2008).
After 7 days, we removed the box and counted the total number
of structural lines connected to the floor of the box and the total
number of gumfooted lines connected to the floor. Gumfooted lines
are conspicuous because they have several millimeters of a glue-like
substance starting at the ground moving up the line. Structural lines
Sh
individual condition. Such a relationship could still allow traits to
evolve independently from each other. Alternatively, there could be
a stable (i.e., genetic) relationship between an individual’s behavior
and its construction. In such a case, selection acting on behavior
would generate an indirect evolutionary response on extended phenotypes and vice versa (Niemelä et al. 2015).
Spiders provide an ideal system to study individual differences in
animal constructions and how these differences are associated with
individual behavior. Indeed, individuals can build multiple webs,
and their behavior while on these webs can be measured repeatedly.
In black widows, Latrodectus hesperus, spiders build a dense, semipermanent 3D cobweb (Benjamin and Zschokke 2003; Zschokke et al.
2006). The web contains different components each serving different functions. Gumfooted lines are long, vertical lines with a “glue”like substance that support foraging behavior by acting as snares
and providing the spider with information on prey (Argintean et al.
2006; Blackledge and Zevenbergen 2007). Structural lines are firmly
attached to the ground and provide increased safety by supporting
a dense 3D tangle web acting as a physical barrier between potential predators or parasitoids and the spider’s retreat (Blackledge
et al. 2003; Blackledge and Zevenbergen 2007; Zevenbergen et al.
2008). Silk is energetically costly (Tanaka 1989; Jakob 1991) and
thus may limit overall web size as well as how many lines a spider can build. Other trade-offs may occur if both gumfooted and
structural lines were more effective when built close to the spider’s
retreat. Potentially, it would require a large number of structural
lines at distance to be dense enough to defend from predators,
while prey caught in distant gumfooted lines may be able to escape
before being reached by the spider. Thus, spiders likely cannot build
webs with both a large number of gumfooted lines and many structural lines. Moreover, spiders are known to alter the structure of
their web, emphasizing either foraging efficiency or safety, depending on their condition. Food-deprived spiders will produce more
gumfooted lines that allow them to attack and capture more prey,
whereas satiated spiders instead invest predominantly in structural
lines and denser webs (Blackledge and Zevenbergen 2007). But, it is
unknown whether individuals differ consistently in the structure of
their web or in their foraging behavior.
Here, we have 2 specific goals. First, we aim to quantify consistent differences among individuals in the structure of their web.
We satiated field-caught black widow spiders and let them build
webs multiple times over the course of several months, which is
a substantial portion of a typical widow spider’s 1–2 year lifespan
(DiRienzo N, Montiglio PO, personal observation). If variation
in web structure comes from aspects of an individual’s condition
(e.g., body weight, current energetic reserves) or environmental
conditions, we expect very little variation in web structure among
individuals once all spiders are satiated and held under standardized conditions. Alternatively, some spiders might consistently build
webs offering increased prey capture (with a greater number of
gumfooted lines), whereas other spiders might consistently build
webs offering increased protection (denser with more structural
lines), irrespective of their body weight or conditions. Second, we
investigated if individual spiders differ in their tendency to attack
a prey cue repeatedly while on their webs, and if this behavior is
related to the structure of their web. If such a relationship arises
because of individual differences in body condition, then we should
not see any relationship between web structure and behavior once
all spiders are satiated, and body weight and size are accounted
for. We also hypothesize that individuals will vary in their level of
risk aversion irrespective of their condition, in which case we could
1425
27.5 cm
Figure 1
Diagram of skeletonized cardboard box provided as a web-building
structure. The shelter represents the remaining 3-cm portion of the top of
the box where the spiders would naturally use as their hide. The gray areas
represent the remaining walls that were covered in black paper to facilitate
the counting of structural and gumfooted lines. The structure was placed
inside a larger, clear Sterilite container to prevent the spider from escaping.
Behavioral Ecology
1426
were considered to be all lines connected to the ground that were
not gumfooted. After removing the spider, we measured its weight
to the nearest milligram before transferring it back to its home container. We then gathered the web by winding it onto a plastic rod.
The webs were placed in a desiccator for 48 h and then weighed
using a Mettler Toledo MT26 microbalance. In this species, web
weight is highly correlated (R2 = 0.8) with web density (measured
as the amount of reflectance from an illuminated web), and thus
provide a measure as to the level of overall web investment (denser
vs. sparse webs; Blackledge and Zevenbergen 2007).
Two days after being removed from their web and placed in
their home container, they were again satiated with 3 large crickets,
given 5 days to feed, weighed again, and then put in a new box
to build a new web. We discarded data from 2 web assessments:
1 spider built one of its webs outside of the cardboard box and 2
spiders did not feed on the crickets and thus were not satiated. All
other spiders built 3 webs in total, giving a total of 48 webs built by
17 individuals.
Spider foraging aggression assay
On the final day of the third web-building trial, we assessed spider
aggression toward a prey cue. In order to remove any influence
of prey behavior (DiRienzo et al. 2013), we used a standardized
vibrating mechanism (Classical Silicone Vibrator, Liler, Shenzhen,
China) (Keiser and Pruitt 2014a, 2014b). Attached to the vibrating
mechanism was a 10-cm-long plastic cable tie, which allowed us to
apply the vibratory cue on a single silk thread in specific locations
of the web. The cable tie also prevented any involuntary damage to the web, while also reducing the intensity of the vibrations.
The vibrator provided 1-s-long pulses at 100 cycles/s separated
by approximately half second long periods of reduced frequency.
This vibratory pattern and frequency is within the range of vibrations produced by houseflies caught in a spiderweb (Walcott 1963).
This frequency elicits a response in other species of spiders (Parry
1965).
Spiders respond to the prey cue as they would a prey trapped in
their web: They rapidly ran toward the source of the vibration and
attempted to trap it by covering it with silk. Spiders that did not
attack generally did not respond to the prey cue, although several
either moved toward the prey but stayed out of reach or oriented
in the direction of the cue but stayed in their shelter. The prey cue
was applied 3 times, once near the shelter (within 2 cm), once at the
end farthest from the shelter, and once in between. The cue was
held to the web for 15 s at each location, with 10 s separating each
application. We recorded if spiders did or did not attack the prey
cue in the 15-s interval as a binary response. Pilot data indicated
that if spiders did not attack within 15 s, they generally would not
attack. The cue application was applied in a random order (e.g.,
near, far, between; far, between, near; etc.). This assay was repeated
again 24 and 48 h later. Thus, we assayed each spider 3 times on 3
different days (n = 9 assays per individuals).
Spider morphology
We assessed spider body size by measuring the total length of the
front femur and patella. Given the femur–patella length does not
change as adults, it provides a reliable method to measure body size.
We lightly anesthetized each spider using CO2 and photographed
the front limb using a Canon EOS M digital camera. Images
included a ruler as reference, enabling us to measure femur–patella
length using the program ImageJ (Schneider et al. 2012).
Statistical methods
First, we quantified individual variation in web structure and tendency to attack the prey cue. We built generalized mixed models
to analyze each aspect of web structure (the number of gumfooted
lines, the number of structural lines, and the web weight) and the
tendency to attack prey cue. These models all included spider body
weight, femur–patella length, and the web trial number as fixed
effects. We also included individual identity as a categorical random
effect. The model analyzing spiders’ tendency to attack the prey cue
also included fixed effects of the distance between the retreat and
the prey cue, trial number to control for the effect of day, and subtrial order to account for randomized order or distances in which
the cue was presented. All explanatory variables were standardized
to a mean of zero and standard deviation of one prior to fitting
the models. The number of gumfooted and structural lines were
modeled with a Poisson error distribution, whereas web weight was
modeled with a Gaussian distribution. The tendency to attack prey
was modeled following a binomial distribution.
Second, we investigated the relationships 1) between the different components of web structure and 2) between web structure and
tendency to attack the prey cue. We calculated a Spearman rank
correlation coefficient between the number of gumfooted lines and
the number of structural lines, and between each of these components and web weight. The relationship between web structure and
the tendency to attack the prey cue was assessed using generalized
linear mixed models and a model comparison approach. A total of
4 models analyzing the probability of attacking the stimuli (response
variable) were created. Each model contained the main fixed effects
of distance between the prey cue and the retreat, spider weight,
spider femur–patella length, trial number, and subtrial order. Each
model varied in that they only contained one aspect of web structure as an additional fixed effect (i.e., number of gumfooted lines,
number of structural lines, web weight). All web structure variables
were standardized to a mean of zero and standard deviation of one
prior to fitting the models. The null model contained no aspect of
web structure. Individual identity was included as a categorical random effect in all models. We compared these models using Akaike
information criteria (AIC; Akaike 1987), which computes the difference in AIC between the models (ΔAIC) and the Akaike weights
(ωi; Burnham and Anderson 2002). A ΔAIC greater than 2 units
between 2 models suggests that the model with the lower AIC provides a better fit to the data (Richards 2005).
We used the statistical software package R version 3.1.1 for all
analyzes (R Core Team 2015). All models containing random effects
were fit with the lme4 package (Bates et al. 2013), whereas those
without random effects were fit with the BBMLE package (Bolker
2010). We tested for significance consistent individual differences
in web structure and aggressive behavior through likelihood ratio
tests (1 degree of freedom [df], Pinheiro and Bates 2000) comparing the fully parametrized models with the individual random effect
to a model without it while keeping the fixed effect structure constant. Repeatability of each component of web structure as well as
aggressive behavior was calculated using the rptR package, which
accommodates both Gaussian and non-Gaussian data (Schielzeth
and Nakagawa 2011). Note that this package only allows to calculate repeatability unadjusted for covariates. Hence, the repeatability
estimates we report do not account for any covariate (e.g., spider’s
body weight, see Results for details). In additional analyzes we do
not present here, we also ascertained that these repeatabilities were
not overly affected by accounting for individual body weight. We
DiRienzo and Montiglio • Individual differences in web structure and behavior
computed repeatabilities adjusted for body weight and size following the methods prescribed in Nakagawa and Schielzeth (2010).
The adjusted repeatabilities were 0.01–0.06 higher than the unadjusted repeatabilities. We thus present the unadjusted repeatability
values only. We calculated the proportion of variance described
by the fixed effects within the models (marginal R2), as well as the
proportion of variance described by both the fixed and random
effects (conditional R2) within the models following the methods of
Nakagawa and Schielzeth (2013).
RESULTS
Individual variation in web structure and tendency
to attack prey cue
We detected significant individual variation in web structure. The
web weight and the number of gumfooted lines built by individuals exhibited relatively high repeatabilities (repeatability = 0.476,
standard error [SE] 0.151, P = 0.002; repeatability = 0.422,
SE 0.206, P = 0.015, respectively). In contrast, the number of
structural lines exhibited a much lower repeatability (repeatability = 0.217, SE 0.164, P = 0.071). Our analysis of web structure
using generalized mixed models (see Table 1) showed that heavier
spiders built more structural lines (β = 0.252, SE 0.073; Figure 2a)
and produced heavier webs (β = 0.808, SE 0.257; Figure 2b), but
built fewer gumfooted lines (β = −0.540, SE 0.068; Figure 2c).
Spiders with a longer femur–patella produced webs with more
gumfooted lines (β = 0.586, SE 0.225). The number of gumfooted lines also increased between the first and the third web
assay (β = 0.111, SE 0.050). Taken together, spider weight, size,
and trial order explained 15%, 23%, and 32% of the variance
(as indicated by the marginal R2 values) in the number of structural lines, the number of gumfooted lines, and in web weights,
respectively. Even after controlling for spider size, weight, and
web number, individuals still differed consistently in the number
of structural lines they produced (χ2 = 119.48, df = 1, P < 0.001),
in the number of gumfooted lines they produced (χ2 = 343.49,
df = 1, P < 0.001) and in the weight of their web (χ2 = 5.84,
df = 1, P = 0.016; Table 1).
1427
Tendency to attack prey cue
Spiders’ tendency to attack the prey cue exhibited a similar repeatability as some web components (repeatability = 0.446, SE 0.12,
P = 0.001). The generalized mixed model (see Table 2) showed
that spiders were less likely to attack the cue when it was presented
further from their refuge (β = −1.440, SE 0.387; Figure 3a). Trial
number, subtrial order, body size, and body weight did not significantly predict the probability of attacking the prey cue (all P > 0.1;
Table 2). Even after accounting for these effects, we still detected
consistent differences in tendency to attack among individuals
(χ2 = 31.25, df = 1, P < 0.001).
Behavior/web structure relationship
There were strong correlations between the different aspects of
web structure. Webs with more gumfooted lines had fewer structural lines (ρ = −0.397, n = 48, P = 0.005), but the number of
gumfooted lines was not related to web weight (ρ = 0.104, n = 48,
P = 0.483). Yet, webs with more structural lines were also heavier
(ρ = 0.308, n = 48, P = 0.033). The best model (ΔAIC = 3.00,
weight = 0.65; Table 3) identified the number of gumfooted lines
as the best predictor of individual tendency to attack prey cue
(Figure 3b). Indeed, even after correcting for individual weight,
individuals producing more gumfooted lines in their web were still
more likely to attack the prey cue (β = 1.340, SE 0.587; Table 4,
Figure 3b).
DISCUSSION
Here, we determined if spiders exhibit consistent individual differences in several elements of web structure and investigated how
these differences related to individual differences in foraging behavior. Individual differed substantially and consistently in the number of
gumfooted lines they produced and in the silk weight of their webs.
In contrast, individuals exhibited lower consistent variation in the
number of structural lines they produced. Individuals also expressed
substantial differences in their tendency to attack a prey cue. Even
after satiating all individuals, we still observed that some individuals
consistently attacked the prey cue, whereas others failed to respond
Table 1
Determinants of the number of structural lines, the number of gumfooted lines, and web weight in black widows (N = 48 webs from
17 individuals
Structural lines
Gumfooted lines
Web weight
Random effects
Variance
LRT
P
Variance
LRT
P
Variance
LRT
P
ID
Residual
0.213
—
119.48
<0.001
0.756
343.49
—
0.001
1.067
1.085
5.84
0.016
Fixed effects
β
SE
z
P
β
SE
z
P
β
SE
t
P
Intercept
Body weight
Femur–patella length
Web number
2.830
0.252
−0.116
0.018
0.150
0.073
0.124
0.045
18.977
3.475
−0.937
0.403
<0.001
<0.001
0.349
0.687
2.061
−0.540
0.568
0.111
0.243
0.068
0.225
0.050
8.481
−7.893
2.519
2.194
<0.001
<0.001
0.012
0.028
2.452
0.808
0.164
0.374
0.480
0.257
0.307
0.195
5.101
3.150
0.503
1.920
<0.001
<0.001
0.593
0.055
Marginal R2
Conditional R2
AIC
Deviance
Residual df
0.150
0.825
468.6
458.6
43
Bold values have a P < 0.05. LRT, likelihood ratio test.
0.235
0.917
577.8
567.8
43
0.351
0.653
170.4
158.4
43
Behavioral Ecology
1428
(b)
8
6
0
2
4
Web weight (mg)
40
30
20
10
Structural lines
50
(a)
–1
0
1
2
Spider weight
3
–1
–1
0
1
2
Spider weight
0
1
2
Spider weight
3
40
30
20
0
10
Gumfooted lines
50
(c)
3
Figure 2
Relationship between spider weight (standardized) and (a) the number of structural lines, (b) the web weight, and (c) the number of gumfooted lines in black
widows (N = 48 webs from 17 individuals).
Table 2
Determinants of spider aggression toward a simulated prey cue
(144 behavioral assays on 17 individuals with 3 trials per day
over a 3-day period)
Random effects
Variance
LRT
P
ID
3.925
31.25
<0.001
Fixed effects
β
SE
Intercept
Trial
Distance
Trial order
Femur–patella length
Body weight
−2.783
0.369
−1.440
−0.202
1.025
0.328
Marginal R2
Conditional R2
AIC
Deviance
Residual df
0.268
0.666
148.3
134.3
137
1.288
0.305
0.387
0.324
0.641
0.644
z
P
2.161
1.208
−3.722
−0.625
1.599
0.510
0.031
0.227
<0.001
0.532
0.110
0.610
Bold values have a P < 0.05.
to it. Interestingly, we also observed clear relationships between different components of web structures, and between web structure
and behavior. In particular, some individuals consistently built webs
that contained a larger number of gumfooted lines and attacked the
prey cue more frequently, whereas others produced heavier webs with
fewer gumfooted lines and potentially more structural lines, and were
less responsive to prey cues. Such important differences were observed
in spite of all spiders being fully satiated and held in standardized
laboratory conditions.
We found that spiders exhibited consistent differences in the type
of webs they built. Some spiders produced webs containing more
gumfooted lines, whereas other spiders invested in denser webs
with more structural lines. Given that webs have multiple functions
(Benjamin and Zschokke 2003; Blackledge et al. 2003), consistent
differences in web structure among individuals are likely to have
important implications for foraging success and predator avoidance
(Zevenbergen et al. 2008). Webs with more gumfooted lines are
more effective at capturing prey, whereas denser webs with more
structural lines potentially offer greater protection against predation (Blackledge et al. 2003; Zevenbergen et al. 2008). Spiders who
build webs with more gumfooted lines may have increased foraging
success, but could also increase their risk of predation. Conversely,
spiders with denser webs and more structural lines may reduce
predation risk, but may also experience lower foraging success.
This trade-off in structural elements could arise if both structural
lines and gumfooted lines are more effective when built closer to
the retreat. Furthermore, extensive intermingling line types may
reduce foraging efficacy either by scaring off prey that encounter
a nonsticky structural line before a gumfooted line or by reducing the snare action of the gumfooted lines. The variation in web
structure could also be based on resource availability and energetic
constraints (Japoshvili et al. 2012; Doerr and Endler 2015) or originate from plastic responses to environmental conditions (Blackledge
DiRienzo and Montiglio • Individual differences in web structure and behavior
1.0
1.2
(b)
0.8
0.6
0.0
0.2
0.4
Attack
0.6
0.2 0.4
–0.2 0.0
Attack
0.8 1.0
(a)
1429
1.0
1.5
2.0
2.5
3.0
–1.0
Distance from refuge (cm)
–0.5
0.0
0.5
1.0
1.5
2.0
Gumfooted lines
Figure 3
Relationship between spider foraging aggressiveness and (a) the distance between the prey stimulus from the refuge and (b) the number of gumfooted
lines (standardized) on the web in black widows (N = 48 webs from 17 individuals). A jitter function was added to both plots in order to separate the
overlapping values.
Table 3
AIC rankings of models predicting the probability of attacking
a simulated prey cue as a function of distinct web structures
Web component
AIC
ΔAIC
k
AIC weight
Gumfooted lines
Structural lines
Null
Web weight
138.5
141.6
141.6
142.8
0
3.0
3.1
4.3
8
8
7
8
0.646
0.142
0.137
0.075
and Zevenbergen 2007; Walter et al. 2008; Blamires 2010; Nakata
2012; Bengston and Dornhaus 2014; Smith et al. 2015). Here, we
report consistent individual variation in web structure over many
months after satiating all individuals, while holding them in standardized conditions, and accounting for differences in size and
weight. Hence, such individual variation likely results from differences in genetics or permanent developmental effects. More generally, our study adds to the growing number of studies reporting
consistent individual variation in architectural constructions (e.g.,
Rushbrook et al. 2008; Walsh et al. 2011; Japoshvili et al. 2012;
Kelley and Endler 2012). Future studies will investigate the fitness
consequences of individual variation through foraging trials and
staged predator encounters.
We also observed consistent individual variation in foraging
aggressiveness. Some spiders attacked the prey cue much more
readily than others even after accounting for individual differences
in body size, weight, or satiation level. Such consistent individual
variation in behavior is observed in many systems (Bell et al. 2009).
Individual variation in tendency to attack could arise if individuals
differed in their tendency to perceive the prey cue or if they differed
instead in their willingness to attack it once it is detected. Although
we did not collect data allowing to distinguish these options, we did
occasionally observe spiders either orienting toward the source of
the vibration or leaving their retreat and moving toward the prey
cue, but staying at a distance from it (DiRienzo N, Montiglio PO,
personal observation). Hence, at least some of the individuals with
a lower tendency to attack the prey cue seemed to have detected
it. For a spider, leaving its retreat to inspect and attack a prey cue
may be costly. Black widow spiders can capture prey that are much
larger than them, but doing so may carry risk of injury. Hence,
the individual differences in tendency to attack the prey cue may
Table 4
The best-fit model predicting the probability to attack as
a function of web structure (144 behavioral assays on 17
individuals with 3 trials per day over a 3-day period)
Random effects
Variance
ID
3.439
Fixed effects
β
SE
z
P
Intercept
Trial
Distance
Trial order
Femur–patella length
Body weight
Gumfooted lines
0.562
0.342
−1.388
−0.385
0.793
0.702
1.340
1.023
0.318
0.357
0.350
0.662
0.697
0.587
2.751
1.075
−3.885
−1.098
1.006
1.199
2.283
0.583
0.282
<0.001
0.272
0.231
0.315
0.022
Marginal R2
Conditional R2
AIC
Deviance
Residual df
0.409
0.711
138.5
122.5
136
Bold values have a P < 0.05.
reflect stable differences in risk aversion among individuals (Sih
et al. 2004; Réale et al. 2010). Investigating this idea further would
require measuring the mortality or injury risks associated with leaving the retreat and attacking prey in nature.
Interestingly, individual behavior was related with web structure.
Spiders that built webs with more gumfooted lines were also more
likely to attack the prey cue. A previous study in this system reported
that webs with more gumfooted lines allow spiders to attack and
capture prey more often, and that fasted spiders produce more
gumfooted lines than satiated spiders (Blackledge and Zevenbergen
2007). Hence, our results confirm such previous findings, but also
suggest that consistent individual differences in web structure could
generate stable individual differences in behavior, a possibility that
has been rarely investigated (but see Niemelä et al. 2015). In our
case, such a relationship between web structure and behavior could
lead to individual differences in web structure that favor higher foraging success in spiders by allowing them to express a higher foraging aggressiveness or an increased responsiveness toward prey cues.
1430
Alternatively, individual differences in web structure could interact
with individual differences in behavior to determine foraging success, in which case correlated evolution between web structure and
behavior could potentially form alternative foraging tactics specializing on different types of prey. Aggressive spiders producing more
gumfooted lines could be more effective at briefly capturing larger
and higher value prey items, as the gumfooted lines may snare animals that would otherwise be strong enough to escape (Argintean
et al. 2006; Zschokke et al. 2006). Less aggressive spiders with
denser webs might be more effective at capturing a large number of
small prey given the increased density of the web. A similar pattern
has been reported in orb weavers, which utilize a 2D web to capture prey. Larger orb webs are more effective at capturing a large
number of prey items, but webs with narrow spacing between lines
are better at capturing higher value prey (Blackledge and Eliason
2007). Alternatively, less aggressive spiders that build denser webs
may indeed suffer reduced foraging efficacy, but instead experience
a longer lifespan due to the increased protection from predators
and/or parasites present in the environment.
In conclusion, we report consistent individual variation in
behavior and web structure in black widow spiders. Such individual differences in web structures persisted even when controlling
for body size and weight, and most likely represent stable variation arising from genetic or developmental effects. Irrespective
of their source, individual variation in web structure may have
important fitness implications, as web structure impacts multiple
aspects of black widow spider’s ecology (Zevenbergen et al. 2008).
Spiders with a large number of gumfooted lines in their webs
were more likely to attack the prey cue than those with less gumfooted lines. This variation could reflect potential constraints on
aggressive behavior due to web structure. For example, particular web structures might limit spider movement or fail to transmit
information about prey location and profitability. Alternatively,
this variation might represent alternative foraging tactics. Future
studies should investigate the fitness consequences of the individual differences we report here by monitoring the foraging success
and diet of individuals with known behavioral and web structure
phenotypes in nature.
FUNDING
This work was supported by a Sigma Xi Grants in Aid of Research
grant to N.D. (grant no. G20141015696446). N.D. was supported
by a University of California, Davis Dissertation Year Fellowship.
P.-O.M. was supported by a Fonds Québécois Recherche Nature et
Technologie Postdoctoral Fellowship.
We thank A.V. Hedrick for kindly lending us some of her lab space and
L.H. Yang for allowing us to use a microbalance.
Handling editor: Glauco Machado
REFERENCES
Akaike H. 1987. Factor analysis and AIC. Psychometrika. 52:317–332.
Anotaux M, Marchal J, Châline N, Desquilbet L, Leborgne R, Gilbert C,
Pasquet A. 2012. Ageing alters spider orb-web construction. Anim Behav.
84:1113–1121.
Argintean S, Chen J, Kim M, Moore A. 2006. Resilient silk captures prey in
black widow cobwebs. Appl Phys A Mater A. 82:235–241.
Bates D, Maechler M, Bolker B, Walker S. 2013. lme4: linear mixed-effects
models using Eigen and S4. R package version 1. Available from: https://
cran.r-project.org/web/packages/lme4/index.html.
Behavioral Ecology
Bell AM, Hankison SJ, Laskowski KL. 2009. The repeatability of behaviour: a meta-analysis. Anim Behav. 77:771–783.
Bengston S, Dornhaus A. 2014. Be meek or be bold? A colony-level behavioural syndrome in ants. Proc Biol Sci. 281:20140518.
Benjamin SP, Zschokke S. 2003. Webs of theridiid spiders: construction,
structure and evolution. Biol J Linn Soc. 78:293–305.
Blackledge TA, Coddington JA, Gillespie RG. 2003. Are three-dimensional
spider webs defensive adaptations? Ecol Lett. 6:13–18.
Blackledge TA, Eliason CM. 2007. Functionally independent components
of prey capture are architecturally constrained in spider orb webs. Biol
Lett. 3:456–458.
Blackledge TA, Wenzel JW. 2001. Silk mediated defense by an orb web spider against predatory mud-dauber wasps. Behaviour. 138:155–171.
Blackledge TA, Zevenbergen JM. 2007. Condition-dependent spider web
architecture in the western black widow, Latrodectus hesperus. Anim Behav.
73:855–864.
Blamires SJ. 2010. Plasticity in extended phenotypes: orb web architectural
responses to variations in prey parameters. J Exp Biol. 213:3207–3212.
Bolker B. 2010. bbmle: tools for general maximum likelihood estimation.
R package version 0.9. Available from: https://cran.r-project.org/web/
packages/bbmle/index.html.
Borgia G. 1995. Threat reduction as a cause of differences in bower architecture, bower decoration and male display in two closely related bowerbirds Chlamydera nuchalis and C. maculata. Emu. 95:1–12.
Burnham K, Anderson D. 2002. Model selection and multimodel inference:
a practical information-theoretic approach. New York: Springer.
DiRienzo N, Pruitt JN, Hedrick AV. 2013. The combined behavioural tendencies of predator and prey mediate the outcome of their interaction.
Anim Behav. 86:317–322.
Doerr NR, Endler JA. 2015. Illusions vary because of the types of decorations at bowers, not male skill at arranging them, in great bowerbirds.
Anim Behav. 99:73–82.
Hansell M. 2005. Animal architecture. New York: Oxford University Press.
Jakob EM. 1991. Costs and benefits of group living for pholcid spiderlings:
losing food, saving silk. Anim Behav. 41:711–722.
Japoshvili B, Lehtonen TK, Wong BB, Lindström K. 2012. Repeatability
of nest size choice and nest building in sand gobies. Anim Behav.
84:913–917.
Keiser CN, Pruitt JN. 2014a. Personality composition is more important
than group size in determining collective foraging behaviour in the wild.
Proc Biol Sci. 281:20141424.
Keiser CN, Pruitt JN. 2014b. Spider aggressiveness determines the bidirectional consequences of host–inquiline interactions. Behav Ecol.
25:142–151.
Kelley LA, Endler JA. 2012. Male great bowerbirds create forced perspective illusions with consistently different individual quality. Proc Natl Acad
Sci USA. 109:20980–20985.
Nakagawa S, Schielzeth H. 2010. Repeatability for Gaussian and nonGaussian data: a practical guide for biologists. Biol Rev. 85:935–956.
Nakagawa S, Schielzeth H. 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol Evol.
4:133–142.
Nakata K. 2012. Plasticity in an extended phenotype and reversed up-down
asymmetry of spider orb webs. Anim Behav. 83:821–826.
Niemelä PT, Lattenkamp EZ, Dingemanse NJ. 2015. Personalityrelated survival and sampling bias in wild cricket nymphs. Behav Ecol.
26:936–946.
Parry DA. 1965. The signal generated by an insect in a spider’s web. J Exp
Biol. 43:185–192.
Pinheiro JC, Bates DM. 2000. Linear mixed-effects models: basic concepts
and examples. Mixed-effects models in S and S-Plus. p. 3–56. Available
from: https://books.google.com/books/about/Mixed_Effects_Models_
in_S_and_S_PLUS.html?id=N3WeyHFbHLQC.
Pinter-Wollman N, Gordon DM, Holmes S. 2012. Nest site and
weather affect the personality of harvester ant colonies. Behav Ecol.
23:1022–1029.
R Core Team. 2015. R: a language and environment for statistical computing. Available from: https://cran.r-project.org/.
Réale D, Garant D, Humphries MM, Bergeron P, Careau V, Montiglio
PO. 2010. Personality and the emergence of the pace-of-life syndrome
concept at the population level. Philos Trans R Soc Lond B Biol Sci.
365:4051–4063.
Richards SA. 2005. Testing ecological theory using the information-theoretic approach: examples and cautionary results. Ecology. 86:2805–2814.
DiRienzo and Montiglio • Individual differences in web structure and behavior
Rushbrook B, Dingemanse NJ, Barber I. 2008. Repeatability in nest construction by male three-spined sticklebacks. Anim Behav. 75:547–553.
Schielzeth H, Nakagawa S. 2011. rptR: repeatability for Gaussian and nonGaussian data. R package version 0.6.404/r36. Available from: http://
rptr.r-forge.r-project.org/.
Schneider CA, Rasband WS, Eliceiri KW. 2012. NIH Image to ImageJ:
25 years of image analysis. Nat Methods. 9:671–675.
Sih A, Bell A, Johnson JC. 2004. Behavioral syndromes: an ecological and
evolutionary overview. Trends Ecol Evol. 19:372–378.
Smith ML, Ostwald MM, Seeley TD. 2015. Adaptive tuning of an extended
phenotype: honeybees seasonally shift their honey storage to optimize
male production. Anim Behav. 103:29–33.
Tanaka K. 1989. Energetic cost of web construction and its effect on web relocation in the web-building spider Agelena limbata. Oecologia. 81:459–464.
Tso I-M. 2004. The effect of food and silk reserve manipulation on decoration-building of Argiope aetheroides. Behaviour. 141:603–616.
1431
Walcott C. 1963. The effect of the web on vibration sensitivity in the spider,
Achaearanea tepidariorum (Koch). J Exp Biol. 40:595–611.
Walsh PT, Hansell M, Borello WD, Healy SD. 2011. Individuality in nest
building: do southern masked weaver (Ploceus velatus) males vary in their
nest-building behaviour? Behav Processes. 88:1–6.
Walter A, Elgar MA, Bliss P, Moritz RF. 2008. Wrap attack activates webdecorating behavior in Argiope spiders. Behav Ecol. 19:799–804.
Whitehouse ME, Jaffe K. 1996. Ant wars: combat strategies, territory
and nest defence in the leaf-cutting ant Atta laevigata. Anim Behav.
51:1207–1217.
Zevenbergen JM, Schneider NK, Blackledge TA. 2008. Fine dining or fortress? Functional shifts in spider web architecture by the western black
widow Latrodectus hesperus. Anim Behav. 76:823–829.
Zschokke S, Hénaut Y, Benjamin SP, García-Ballinas JA. 2006. Preycapture strategies in sympatric web-building spiders. Can J Zool.
84:964–973.