Animal Behaviour 89 (2014) 107e114

Contents lists available at ScienceDirect

Animal Behaviour
journal homepage: www.elsevier.com/locate/anbehav

Poke but dont pinch: risk assessment and venom metering in the
western black widow spider, Latrodectus hesperus
David R. Nelsen*, Wayne Kelln, William K. Hayes
Department of Earth and Biological Sciences, Loma Linda University, Loma Linda, CA, U.S.A.

a r t i c l e i n f o
The capacity to assess threat, and modulate defensive behaviour accordingly, is widespread among an-
Article history: imals. Although numerous studies have demonstrated that venomous animals possess control of venom
Received 10 July 2013 expenditure in predatory contexts, few have examined venom metering in defensive contexts. Here, we
Initial acceptance 11 September 2013 investigated venom use in the context of threat assessment in the western widow spider, Latrodectus
Final acceptance 26 November 2013 hesperus, a synanthropic, medically relevant species found throughout western North America. In
Available online 21 January 2014 experiment 1, we subjected wild-caught adult females (N 43) to single (low threat) and repeated
MS. number: A13-00577R (medium threat) prods with a gelatin nger, and repeated pinches between two gelatin ngers (high
threat). Poking at low threat elicited primarily avoidance responses (move and retract), repeated
Keywords: prodding at moderate threat stimulated increased silk-icking, and pinching at high threat provoked
antipredatory behaviour
signicantly more biting. In experiment 2, we pinched spiders (N 20) to induce bites of three suc-
black widow
cessive targets (paralm-covered tubes) from which venom could be recovered. Spiders modulated
Latrodectus hesperus
silk use
venom expenditure at two levels. First, spiders chose whether to use venom, frequently delivering dry
threat sensitivity bites (>50% of interactions) independent of venom depletion. Second, spiders metered venom quantity,
venom metering delivering 1.8-fold more venom per bite when pinched on the body versus a leg, and 2.3-fold more
venom optimization venom when successive target presentations were separated by a lengthy (5 min) compared to a brief
(5 s) interval. These ndings suggest that spiders make decisions about silk and venom use for defence.
Because these weapons are metabolically costly to replace, they represent limited commodities that
should be used judiciously.
2014 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

The survival of animals represents a continual trade-off between
acquisition of resources, reproduction and avoidance of predation
(Bednekoff, 2007; Caro, 2005; Ferrari, Sih, & Chivers, 2009; Lima &
Bednekoff, 1999; Lima & Steury, 2005; Wisenden, 2000). While an
unsuccessful attempt at resource acquisition or reproduction may
be overcome by subsequent successes, failing to avoid predation is
the end. Thus, an organisms ability to perceive and respond to a
threat is under high selective pressure (Kats & Dill, 1998; Lima & Dill,
1990). Accordingly, risk assessment, also known as threat sensitivity,
offers an excellent opportunity to understand the complex re-
lationships between environmental cues and an organisms ability
to modulate its behaviours in response. Although many studies have
tested risk assessment in vertebrates and invertebrates, relatively
few studies have focused on venomous organisms, such as snakes
(Glaudas & Gibbons, 2005; Glaudas, Winne, & Fedewa, 2006), spi-
ders (Jackson, Brassington, & Rowe, 1990, 1992; Jackson, Rowe, &
* Correspondence: D. R. Nelsen, Department of Earth and Biological Sciences,
Loma Linda University, 11065 Campus St, Griggs Hall, Rm 101, Loma Linda, CA
92350, U.S.A.
E-mail address: dnelsen@llu.edu (D. R. Nelsen).
Wilcox, 1993; Lohrey, Clark, Gordon, & Uetz, 2009; Riechert &
Hedrick, 1990; Taylor, Persons, & Rypstra, 2005) and other arach-
nids (Grostal & Dicke, 2000; Nisani & Hayes, 2011).
Venom deployment is a ubiquitous strategy among organisms
and has evolved independently in multiple groups (Fry et al., 2009;
Nelsen et al., 2013). Venoms generally comprise a complex mixture
of proteins and/or nonproteinaceous compounds. Individual
venoms may be composed of hundreds, to thousands, of peptides
and proteins (Escoubas, Sollod, & King, 2006; Nascimento et al.,
2006). Venoms may even have redundancies in toxin activity,
with several different toxins all affecting the same receptor subtype
(Morgenstern & King, 2012). The complexity and redundancy all
result in a measurable metabolic cost to synthesize, store and
maintain venom (McCue, 2006; Nisani, Boskovic, Dunbar, Kelln, &
Hayes, 2012; Nisani, Dunbar, & Hayes, 2007; Pintor,
Krockenberger, & Seymour, 2010). The cost of venom can also be
inferred by examples of its loss as a result of shifts in diet (Li, Fry, &
Kini, 2005) or changes in prey capture behaviour (Hayes, 2008;
King, 2004; Wigger, Kuhn-Nentwig, & Nentwig, 2002).
Animals make choices about venom use on at least two levels:
whether to use venom, and how much to use. Dry bites and stings
have been reported in numerous species, with evidence suggesting

0003-3472/$38.00 2014 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.anbehav.2013.12.019
108 D. R. Nelsen et al. / Animal Behaviour 89 (2014) 107e114

involvement of decision making (Alves De Rezende et al., 1998; Colton, California (San Bernardino County). Spiders were housed in
Hayes, Herbert, Rehling, & Gennaro, 2002; Nisani & Hayes, 2011). 540 ml plastic deli cups at 22.2  C on a 12:12 h light:dark cycle. We
Venom metering (or venom optimization), the ability to actively provided spiders with a small stick that facilitated web construc-
control the amount of venom delivered during bites or stings, has tion, and offered house crickets (Acheta domestica) as food once
been studied across many species and in both defensive and every 2 weeks. No water was provided, as it was deemed
predatory contexts. Predatory venom metering has been observed unnecessary.
in several species of snakes and in the spider Cupiennius salei, and is
related to factors such as prey size (Boev, 1994; Hayes, Lavn-
Ethical Note
Murcio, & Kardong, 1995, 2002; Malli, Imboden, & Kuhn-Nentwig,
1998; Malli, Kuhn-Nentwig, Imboden, & Nentwig, 1999), prey
The Institutional Animal Care and Use Committee of our insti-
type (Boev, Kuhn-Nentwig, Keller, & Nentwig, 1995; Hayes, 1992;
tution (Loma Linda University) does not regulate the use of in-
Rodrguez-Robles & Leal, 1993), struggle intensity (Malli et al., 1999;
vertebrates in research; however, great care was taken to minimize
Rodrguez-Robles & Leal, 1993; Wigger et al., 2002), satiety (Hayes,
permanent harm and suffering experienced by the spiders used in
1993) and venom availability (Hostettler & Nentwig, 2006;
these experiments (N 45 and 20 in experiments 1 and 2,
Wullschleger & Nentwig, 2002). Defensive venom metering has
respectively). Spiders were subjected to only one of the two
been comparatively less studied, with experiments limited to
experimental procedures. Two of the 65 spiders were excluded
scorpions (Nisani & Hayes, 2011), ants (Haight, 2006; Obin &
from the nal analysis due to injuries sustained in experiment 1.
Vander Meer, 1985) and snakes (Hayes, 2008; Hayes et al., 2002,
One individual died following an abdominal puncture incurred
2008). Heretofore, risk assessment and defensive venom meter-
while removing it from its housing container the rst day, and the
ing have not been examined in spiders.
other died between observation days 2 and 3 with no visible signs
The western widow spider, Latrodectus hesperus, or western
of physical trauma. Spiders used in experiment 2 did not appear to
black widow, is a medically relevant, synanthropic species found in
suffer any ill effects, with all surviving and no leg autotomy
western North America, ranging from southern British Columbia
observed. After the conclusion of these experiments, spiders were
into Mexico. We chose to study Latrodectus hesperus because of its
maintained in the collection as described above, living 1e2 years
local abundance and its well-known reliance on venom for defence
(expected longevity) following the experiments.
against humans (Nordt, Clark, Lee, Berk, & Cantrell, 2012; Vetter &
Isbister, 2008) and presumably other mammals. Bites to humans
induce considerable pain, and onset can be immediate (Blair, 1934). Experiment 1: Risk Assessment
Natural predators of subadult and adult L. hesperus include two
wasps (Chalybion californicum and Tastiotenia festiva) and other Forty-ve adult female spiders were initially collected for this
opportunistic predators, such as rodents, mantids and centipedes experiment, but two individuals died before completion of the
(Irving & Hinman, 1935; Landes, Obin, Cady, & Hunt, 1987; Vetter, experiment, yielding a nal sample of N 43. We subjected each
1980). Defensive behaviours against rodent predators were spider to three threat conditions (low, medium, high) in a repeated
described by Vetter (1980), although not in relation to risk measures design. Each threat involved a simulated attack by gelatin
assessment, biting and venom expenditure. Vetter (1980) ngers. The ngers were made by adding four packets of Knox
concluded that silk icking, the act of attaching visibly viscous gelatin (28.8 g; Associated Brands, Medina, NY, U.S.A.) to 0.240 li-
silk from the spinnerets onto the threatening animal via arcing tres of water. Once all granules were moistened, the mixture was
extensions of one or both posterior-most legs, occurs only at the refrigerated for 2 h. The mixture was then melted using a generic
highest levels of threat in this species. Based on anecdotal evidence, stovetop double boiler, with temperature never exceeding 54.4  C.
dAmour, Becker, and Van Riper (1936) suggested that venom use Gelatin was poured into plastic containers and refrigerated for 24 h.
similarly occurs only at the highest levels of threat, and is under the Gelatin was removed from the containers and then sliced into
volition of the spider, but no study to date has investigated 6  3  2 cm (L  W  H) rectangles. The gelatin ngers were then
defensive venom use or venom metering within Latrodectus. mounted on 19.5 cm steel microspatulas (model number 702700,
The goal of this study was to investigate how the western widow Carolina Biological Supply Company, Burlington, NC, U.S.A.) before
modulates its defensive behaviour during different levels of threat, being used in experiments (Fig. 1). New gelatin ngers were used
emphasizing the decisions it makes in using its metabolically
expensive venom. We sought to experimentally answer two key
questions. (1) Does L. hesperus modulate its defensive behaviour at
different levels of threat? (2) Does L. hesperus show control over
venom expulsion (i.e. venom metering)? Afrmative answers
would suggest that L. hesperus is capable of risk assessment. The
rst of two experiments examined defensive behaviours of
L. hesperus at three levels of threat. The second experiment
considered specic factors that inuence biting and venom
expenditure, including location of contact on the spiders body by a
simulated predator, interval between successive predator attacks,
sequence of successive attacks and venom depletion from prior
venom use.

METHODS

Spider Collection, Housing and Care

We collected spiders in the spring and summer months Figure 1. Gelatin ngers mounted on steel microspatulas to pinch western widow,
(generally MayeSeptember) from Redlands, Loma Linda, and Latrodectus hesperus, spiders during the high-threat condition of experiment 1.
 D. R. Nelsen et al. / Animal Behaviour 89 (2014) 107e114 109

for each trial. Gelatin ngers enabled spiders to bite and pierce the
offender; however, we were unable to recover and measure
venom injected into the gelatin.
For low threat, spiders were tested in their home containers
while remaining within their web. The treatment consisted of a
single 1 s prod using a single gelatin nger randomly directed at the
cephalothorax, abdomen or leg of a spider; we targeted different
body parts to maximize variation in behavioural responses, but we
pooled the data since many presentations contacted more than one
body part. The medium threat, representing a more persistent
predator, also took place within the home container. This condition
followed the same procedure as the low threat, but consisted of 60
brief prods at one prod/s for a total observation time of 60 s. For
high threat, we coaxed the spider to climb onto a gelatin nger, and
then subjected the spider to a series of three gentle whole-body
pinches between two gelatin ngers, thereby modelling a preda-
tors grasp. Each individual pinch lasted 10 s, with a minimum of
10 s between successive pinches. The steel microspatula helped to
keep the spider on the gelatin nger, as the spider could not readily
climb along the spatula itself (Fig. 1).
We subjected each spider to one threat per day over three
successive days in a randomized sequence. We identied and
recorded ve primary defensive behaviours of L. hesperus, as
dened in Table 1. The experimenter (D.R.N.), who for consistency
conducted all trials, made an audio recording while narrating each
trial, and subsequently analysed the recordings for presence/
absence, frequency and duration of each behaviour observed. Ob-
servations were aided by a desk-mounted magnifying lamp
(L745BK, Luxo Corp, Elmsford, NY, U.S.A.), which facilitated the
observation of ne behaviours, such as biting.
Experiment 2: Venom Expenditure
To measure venom expenditure during biting, we prepared
targets to be bitten from which we could readily collect the venom.
The target comprised a 1.5 ml microcentrifuge plastic tube (model
05-402-27, Eppendorf North America Biotools, Hauppauge, NY,
U.S.A.) with a single sheet of Paralm (Bemis Company Inc.,
Neenah, WI, U.S.A.) stretched over the opening via gloved hands
(model 304362073, Handgards, El Paso, TX, U.S.A.) to avoid protein
contamination. We tested spiders (N 20) by presenting three
successive targets to bite within four randomly ordered conditions
combining presentation interval (2 levels) and pinch location
(2 levels) in a 3  2  2 repeated measures design. For each of the
four trials, spaced 2 weeks apart, we removed the spider from its
home container, either coaxing the spider to the top of its container
using 114 mm long forceps (model 4527, BioQuip Products, Rancho
Dominguez, CA, U.S.A.) and then pinching the leg with the same
forceps, or by grasping the spider by the abdomen with a gloved
hand. Once removed, and while still grasped by a leg or by the
abdomen, the Paralm surface of the target was positioned so
that the spiders chelicerae and fangs were in contact. We repeated
the target presentation two more times, either 5 s apart (brief in-
terval, modelling a single persistent encounter) or 5 min apart
(lengthy interval, presumably modelling three separate encoun-
ters). Spiders were allowed to interact with each target for 5 s. For
the brief intervals, we repositioned the spiders before presenting
the next target. For the lengthy intervals, we returned spiders to
their home container between each presentation. We observed the
spiders interaction with the target using a dissection microscope
(Nikon SMZ-10A, Nikon Instruments Inc., Melville, NY, U.S.A.) set to
7.5 magnication. Number of bites delivered during a given pinch
was determined subsequently by counting the number of puncture
marks on the Paralm. A bite consisted of two punctures from the
two fangs spaced closely together. If only one puncture was
observed, then the bite was scored as 0.5. After observing all spiders
on a given day, each spider was fed and not offered food again until
the next trial 2 weeks later.
Immediately after each trial, the Paralm stretched over each
tube was cut by razor along the internal diameter of the tube to
collect venom deposited on the surface. We retained only this small
portion of the membrane to reduce contamination from extraneous
proteins, especially silk. We placed the cut portion of the Paralm
within the snap-cap tube and added 150 ml of carbonate buffer
(50 mM). We then briey agitated the Paralm and carbonate
buffer mixture with a VWR Vortexer 2 (set to 8; VWR International
LLC., Radnor, PA, U.S.A.), and then immediately placed the sample
on ice until samples were permanently frozen at 80  C.
For this experiment, we obtained three dependent variables:
quantity of venom delivered, number of bites and quantity of
venom per bite. We measured venom expenditure (dry mass of
venom protein present) using an indirect enzyme-linked immu-
nosorbent assay (ELISA) on hydrophobic 96-well plates (Immulux
catalogue number 1000, Dynex, Chantilly, VA, U.S.A.). Antigen was
bound to the plate by adding 100 ml of the bite samples (unknown
quantity of antigen in carbonate buffer) and standard samples (0,
0.03, 0.06, 0.13, 0.25, 0.50 and 1.0 mg/ml of venom) to separate
wells. Two to three additional wells served as reagent controls. We
could not test samples in duplicate or triplicate because of the
minute venom amounts. We then covered the plates with scotch
tape and incubated them overnight at 4  C. The following day,
plates were warmed to room temperature and then washed three
times with 300 ml of phosphate-buffered saline (PBS; pH 7.4). Wells
were then blocked with 200 ml SuperBlock (number 37515, Pierce,
Rockford, IL, U.S.A.) for 3 min, icked to remove liquid, and repeated
two more times. The plates were again washed three times with
300 ml PBS-Tween (PBS 0.05% Tween-20, Sigma P6585 low
peroxide/carbonyl, St. Louis, MO, U.S.A.). Next, we added 100 ml of
primary antibody (Aracmyn Plus, Bioclon, Mexico City, Mexico),
diluted 1:25 000 in Diluent (10% blocking protein in PBS-Tween), to
each well. Aracmyn Plus comprises a polyclonal Fab2 IgG antibody
made in horse (Equus ferus) hyperimmunized with whole venom

Table 1
Denitions and contexts of western widow (Latrodectus hesperus) defensive behaviours

Behaviour Structure used Denition Threat

Move Whole body Spider moves entire body in response to stimulus, usually away from it Low
Silk-ick Legs IV and spinnerets Spider moves leg(s) IV towards spinnerets, grabs and holds a strand of Low and high
visible viscous silk, then extends leg(s) simultaneously or asynchronously
towards the threatening stimulus to transfer silk, upon contact, to the target
Bite Chelicerae and fangs Spider moves chelicerae close to gelatin nger or paralm target, distal Low and high
ends of chelicerae open laterally, and fangs are inserted as chelicerae close
medially
Retract leg(s) One or more leg(s) Spider moves (adducts) one or more legs medially from initial position Low
Play dead Whole body Spider drops from web, retracts (adducts) all legs medially against body, Low and high
and remains motionless 1 s
110 D. R. Nelsen et al. / Animal Behaviour 89 (2014) 107e114

from Latrodectus sp. (manufacturer does not indicate species; Assumptions of normality and homoscedasticity were often not
however, L. hesperus is native to the region). We covered the plate met, but because the multivariate assumption of sphericity was
with scotch tape and incubated it on a shaker for 1 h at room always met, and results of nonparametric tests (Wilcoxon tests and
temperature. After incubation, the plates were washed three times Friedmans ANOVAs) applied to the same data yielded identical
with 300 ml PBS-Tween. We then added 100 ml of secondary anti- conclusions for main effects, we report only the parametric results.
body (rabbit anti-horse IgG, ab6921, Abcam, Cambridge, MA, U.S.A., We computed partial h2 as a measure of effect size, indicating
diluted 1:1000 in Dilutent) to each well, covered the plate with approximate percentage of variance explained by main effects and
scotch tape and incubated it on a shaker for 1 h at room tempera- interactions, with values of w0.01, w0.06 and 0.14 loosely
ture. Dilutions of the primary and secondary antibodies were considered small, medium and large, respectively (Cohen, 1988).
determined previously using a checkerboard assay. After incuba- Although partial h2 values tend to be upward biased (Pierce, Block,
tion, the plates were again washed three times with 300 ml of PBS- & Aguinis, 2004), they never summed to more than 1.0 in our
Tween. Finally, we added 100 ml of substrate (Bio-Rad 172-1067, models. We also applied a one-way repeated measures ANOVA to
Hercules, CA, U.S.A.) to each well and incubated the plate uncov- the proportion of dry bites delivered to each of the three targets
ered for 20e30 min. Absorbance values of individual wells on the presented in succession, and relied on h2 for effect size. We used a
plates were read on a Dynex MRX II (Chantilly, VA, U.S.A.) at 630 nm Spearmans rank correlation (rs) to test the association between
and reference at 490 nm. Venom quantities from spider bites were number of bites delivered to a target and quantity of venom
calculated based on regression of the ve control samples plotted expended; we considered the parametric correlation unreliable for
on a quadratic curve. All estimates less than 0.06 mg, the low limit of this test due to extreme violation of assumptions and a result that
detectability, were assigned a value of 0 mg. The coefcients of differed substantially from the nonparametric test. We further
variation (r2) for standard curves ranged from 0.97 to 0.99, indi- calculated coefcients of variation (s/m) for venom per bite (for all
cating high reliability of venom measurement. Each plate had its targets pooled, N 240; 20 spiders  12 targets per animal), total
own standards made from the same stock on the same day. venom expended across all bites per individual (N 20) and venom
expended for all targets bitten a single time (N 67).
Analyses
RESULTS
We conducted all analyses using SPSS 20.0 for Windows (SPSS
Inc., Chicago, IL, U.S.A.) and alpha set at 0.05. We examined all data Experiment 1: Risk Assessment
prior to conducting parametric tests to determine whether they
met assumptions of normality and homoscedasticity. When data All ve defensive behaviours recorded differed signicantly
failed to meet assumptions, we conducted both nonparametric and among the three threat conditions (Table 2). Three variables were
parametric tests, reporting only the parametric tests if the two were signicantly greater for medium and high threat compared to low
in agreement. In the one case of disagreement, we provide the more threat, with higher proportions of spiders showing move and silk-
conservative nonparametric test. We further computed effect sizes, ick and delivering more silk-icks (Cochrans Q tests and one-way
which are independent of sample size (in contrast to statistical repeated measures ANOVA: all P  0.001; McNemars and LSD
signicance) and more readily compared among different data sets pairwise comparisons: P  0.016 and P < 0.001, respectively). Move
and different studies (Cohen, 1988; Nakagawa & Cuthill, 2007). usually involved retreat, but we did not record direction of move-
Following Nakagawa (2004), we chose not to adjust alpha for ment. The proportion of spiders showing bite and the number of
multiple tests. All measures of central tendency are reported as bites was similar for low and medium threat, but substantially
means  1 SE. greater at high threat, when spiders were pinched rather than
For experiment 1, we used Cochrans Q tests, followed by poked (Cochrans Q test and one-way repeated measures ANOVA:
McNemars tests for pairwise comparisons (Zar, 1996), to compare both P < 0.001; McNemars and LSD pairwise comparisons: both
the proportion of spiders that engaged in each behaviour across P < 0.001). Spiders seldom retracted legs during the trials, but
threat levels and across the series of pinches within the high threat while retract did not differ between the low- and medium-threat
level. For effect sizes, we computed Kendalls W, indicating strength conditions, we never observed this behaviour in high-threat trials.
of association (Green & Salkind, 2005), with values of w0.1, w0.3 Similarly, few spiders played dead, but the proportion of spiders
and 0.5 roughly corresponding to small, medium and large effects, doing so was signicantly greater for medium threat than for low
respectively (Cohen, 1988). We used repeated measures one-way
ANOVAs, followed by least signicant difference (LSD) tests for
pairwise comparisons (Zar, 1996), to compare the frequency of each Table 2
Comparison of spider behaviours across threat levels in experiment 1 (N 43)
behaviour across all threat levels and across pinches within high
threat. Because the multivariate assumption of sphericity was not Dependent variable Low threat Medium High threat P value* Effect
always met with these tests, we applied GreenhouseeGeisser ad- threat sizey
justments to the degrees of freedom when necessary (Green & Retract (%) 16a 16a 0b 0.016 0.16
Salkind, 2005), although doing so was inconsequential. We Move (%) 84a 100b 100b 0.001 0.16
computed eta-squared (h2) as a measure of effect size, indicating Play dead (%) 0a 16b 9 ab
0.016 0.10
Silk-ick (%) 5a 56b 44b <0.001 0.31
approximate percentage of variance explained, with values of Silk-icks 0.05  0.03a 3.72  0.84b 1.84  0.44b 0.001z 0.21
w0.01, w0.06 and 0.14 loosely considered small, medium and (mean  SE)
large, respectively (Cohen, 1988). We used Pearsons correlation (r; Bite (%) 0a 2a 60b <0.001 0.58
Zar, 1996) to test the effect of handling time on frequency of bites Bites (mean  SE) 0.00  0.00a 0.02  0.02a 2.70  0.55b <0.001z 0.37
within high threat, with values of w0.1, w0.3 and 0.5 deemed Different superscript letters indicate signicant pairwise differences; McNemars
small, medium and large effects, respectively (Cohen, 1988). test for percentage variables, least signicant difference test for number of silk-icks
For experiment 2, we conducted 3  2  2 (sequence of suc- and bites.
* Cochrans Q for percentage variables, one-way repeated measures ANOVA for
cessive pinches  pinch location  interval duration) repeated number of silk-icks and bites.
measures ANOVAs for each dependent variable (venom expended, y
Kendalls W for percentages, eta-squared (h2) for number of silk-icks and bites.
z
number of bites and venom expended per bite for each tube). GreenhouseeGeisser adjustment of degrees of freedom was applied.
 D. R. Nelsen et al. / Animal Behaviour 89 (2014) 107e114 111

threat (Cochrans Q test: P 0.016; McNemar pairwise comparison: Table 4

P 0.016). During the high-threat trials, we observed no signicant Venom expenditure, number of bites and venom per bite by spiders in experiment 2
(N 20)
decline or increase in silk-icks or bites among the three successive
pinches (Table 3). Independent Total venom (mg) Total bites Venom/bite (mg)
Spiders were removed from the home web during the high- variable

threat trials and manipulated between two gelatin ngers. Under Location
these conditions, the majority of spiders (68%) at some point Leg 10.1  1.7 2.1  0.2 5.3  1.0
Body 13.1  2.4 1.3  0.1 9.8  1.9
rapidly dropped off the gelatin ngers and onto the oor, a distance
F1,19 1.15, F1,19 14.49, F1,19 4.86,
of about 1 m. The drop was rapid but controlled, as spiders attached P 0.30 P < 0.001 P 0.040
a dragline to the gelatin ngers prior to dropping. The drops Partial h2 0.06 Partial h2 0.43 Partial h2 0.20
nevertheless were rapid enough that spiders sometimes bounced Interval
Brief 7.0  1.2 1.7  0.1 5.5  1.2
off the ground upon impact. Because of the difculty in handling,
Lengthy 16.2  2.8 1.7  0.1 9.6  1.9
high-threat observations lasted an average of 2.53 min (range F1,19 9.84, F1,19 0.19, F1,19 3.79,
1.45e3.97 min). We found no correlation of handling time with P 0.005 P 0.67 P 0.067
silk-icks (Pearsons correlation: r41 0.22, P 0.16) or number Partial h2 0.34 Partial h2 0.01 Partial h2 0.17
of bites (Pearsons correlation: r41 0.14, P 0.36). Sequence
Tube 1 10.3  2.6 1.6  0.1 6.4  1.8
Spiders usually showed more than one behaviour during a given
Tube 2 10.3  2.4 1.7  0.1 6.9  2.0
trial, often switching between alternative defensive tactics. At Tube 3 14.3  2.8 1.8  0.1 9.4  2.1
medium and high threat, for example, all spiders moved, but many F2,38 0.84, F2,38 1.13, F2,38 0.66,
paused to also engage in retract, play dead, silk-ick and/or bite. We P 0.44 P 0.33 P 0.52
did not record or analyse behavioural transitions. Partial h2 0.04 Partial h2 0.06 Partial h2 0.03

Values are means  SE. Spiders were presented three paralm-covered snap-cap
tubes to bite after being pinched in two locations (leg, body) and with two different
Experiment 2: Venom Expenditure intervals between successive presentations (5 s, 5 min). Main effects of 2  2  3
(location  interval  sequence) repeated measures ANOVA are indicated; see text
We examined three dependent variables in this experiment: for details on the one signicant interaction (location  sequence for total venom).

quantity of venom delivered, number of bites and quantity of

venom per bite (Table 4).
successive bites compared to brief intervals. No interactions were
Quantity of venom delivered was similar for the two pinch lo-
signicant for number of bites or venom per bite.
cations and within the sequence of three successive targets. How-
Spiders delivered a mean of 5.1  0.3 bites and 34.9  4.5 mg of
ever, the main effect of interval was signicant, as spiders delivered
venom during individual trials (N 80; quantities were summed
2.3-fold more venom when biting during the lengthy intervals
across the three targets). Spiders bit individual targets (N 240) a
between successive bites compared to the brief intervals (repeated
mean of 1.7  0.1 times (range 0e5), expending 11.0  1.4 mg of
measures ANOVA: P 0.005). A two-way interaction also existed
venom per bite (range 0e142.7 mg). The majority of these targets
between location and sequence (F2,38 4.31, P 0.021, partial
(65.4%) were below the threshold for venom detectability with our
h2 0.19). When a leg was pinched, more venom was expended in
assay. Targets bitten a single time (N 67) received a mean of
the rst pinch compared to subsequent pinches, but when the body
16.0  3.2 mg of venom (range 0e110.0 mg). Of these, 53.7% were
was pinched, venom expenditure increased with each successive
below the threshold for venom detectability and were potentially
pinch (Fig. 2). No other interactions were signicant (data not
dry bites. Some targets bitten up to ve times contained no
provided here).
measurable venom. When dry bites were excluded, targets bitten a
Number of bites was similar for the two intervals and the three
single time (N 31) received 34.6  5.3 mg of venom (range 12.3e
targets. However, the main effect of location was signicant, as
spiders delivered 1.6-fold more bites when a leg was pinched than
when the body was pinched (repeated measures ANOVA:
25
P < 0.001). Although total venom expended was similar when a leg Leg
or the body was pinched, the amount of venom expended per bite
Body
averaged 1.8-fold greater when the body was pinched (P 0.040).
20
The amount of venom per bite also approached signicance for
interval (P 0.067, partial h2 0.20; note large effect size), with
1.7-fold more venom injected per bite for lengthy intervals between
Venom (ug)

15

Table 3
Comparison of spider behaviour among pinches within the high-threat condition of
experiment 1 (N 43) 10

Dependent Pinch one Pinch two Pinch three P value* Effect

variable sizey
5
Silk-ick (%) 11 27 16 0.128 0.03
Silk-icks/pinch 0.33  0.17 0.91  0.29 0.60  0.29 0.284 0.03
(mean  SE)
Bite (%) 36 34 39 0.854 <0.01 0
Bites/pinch 0.60  0.14 0.65  0.16 0.84  0.20 0.314 0.03 1 2 3
(mean  SE)
Sequence of target presentations
* Cochrans Q for percentage variables, one-way repeated measures ANOVA for
number of silk-icks and bites. Figure 2. Mean  SE venom expended by western widow, Latrodectus hesperus, spi-
y
Kendalls W for percentage variables, eta-squared (h2) for number of silk-icks ders, illustrating the location  sequence interaction in experiment 2 (N 20 for each
and bites mean).
112 D. R. Nelsen et al. / Animal Behaviour 89 (2014) 107e114

110.0 mg). When all targets were pooled and treated as independent Persons, & Rypstra, 2002; Johnson & Sih, 2007; Lohrey et al.,
(N 240), there was a weak but signicant positive correlation 2009; Persons, Walker, & Rypstra, 2002). When prodded repeat-
between number of bites and total venom expended (Spearman edly under medium threat, a higher proportion of L. hesperus
rank correlation: rs 0.18, P 0.006). Coefcients of variation were moved (100%), played dead (16%) and icked silk (56%), but only
2.13 for venom per bite per target (N 203), 0.72 for total venom one attempted to bite (2%). The increase in silk icking indicated an
expended across all bites per individual (N 20) and 1.65 for escalation in defensiveness, but nearly half the spiders remained
venom per bite for targets bitten a single time (N 67). nonconfrontational. When pinched under high threat, defensive-
Dry bites could result from venom depletion or a decision not to ness escalated further, with the majority of spiders biting (60%).
release venom. When trials were pooled across intervals and Spiders could have retaliated more often with bites at lower threat
location, the proportion of dry bites did not decline among the levels (one actually did so), as some snakes and tarantulas are prone
three targets in succession (target 1: 0.65  0.07; target 2: to do (D. R. Nelsen, personal observation), but the spiders bit only as
0.56  0.06; target 3: 0.51  0.06; N 20 for each target; repeated- a measure of last resort when physically grasped. Although esca-
measures ANOVA: F2,38 1.24, P 0.30, partial h2 0.06). Among lated defensiveness in the high-threat condition could have resul-
all trials (N 80), three (3.8%) resulted in wet bites for all three ted from the longer duration of the trial, neither silk icking nor
targets bitten in succession, 22 (27.5%) resulted in dry bites for all biting were associated with duration of trial in the high-threat
three targets, and the remainder (68.8%) involved both wet and dry condition. Retract was not observed during high threat, probably
bites (Table 5). For the latter group, 23 (28.8%) could be interpreted due to the forceful nature by which spiders were removed from
as support for venom depletion (dry bite following wet bite) and 32 their containers.
(40.0%) could be interpreted as support for venom metering (wet Vetter (1980) similarly reported that mature females of
bite following dry bite; Table 5). Dryewet and wetedry sequences L. hesperus use silk icking for defence only at higher levels of
supporting each interpretation existed for both brief and lengthy perturbance. In the laboratory, females that silk-icked were better
trials (Table 5). We did not subject this comparison to statistical protected against rodent predators (Peromyscus spp.) than those
analysis because it involved pseudoreplication. One spider deliv- whose spinnerets were experimentally blocked. Mature females,
ered 142.7 mg of venom when biting the rst target, and 0 mg in the which likely suffer greater predation than other age groups due to
two subsequent targets. their greater exposure to predators (see Vetter, 1980), were most
likely to ick silk. Mature males, in contrast, lost their silk-icking
behaviour and may have had diminished capacity to produce silk.
DISCUSSION
Vetter (1980) also found that the silk provided a mechanical rather
than a chemical deterrent to the rodents. Defensive use of silk has
In this study, we sought to test whether L. hesperus modulates
been observed in other families of spiders (Blackledge & Wenzel,
defensive behaviour based on different levels of threat and makes
2001; McAlister, 1960), but the act of throwing or icking silk
corresponding decisions about venom use. Experiment 1, which
may be unique to family Theridiidae.
compared defensive behaviours at three levels of threat, revealed
Silk icking may be preferable to biting because the spider
that poking elicits primarily avoidance responses, whereas pinch-
maintains distance from the attacker. During a bite, the cephalo-
ing provokes more defensive responses including silk-icking and
thorax is brought into direct contact with the threat, resulting in
biting. Experiment 2, which examined the factors that inuence
sensitive organs being exposed to potential injury. The spiders
biting and venom expenditure, indicated that spiders modulate
short fangs and relatively weak jaws also impose constraints on
venom expenditure depending on which body part is pinched and
envenomation. When dealing with rodent predators, L. hesperus
the interval between successive simulated attacks.
never attempted to bite (Vetter, 1980). However, our ndings
Results from experiment 1 conrmed that L. hesperus shows a
indicate willingness to bite when physically pinched for an
range of defensive behaviours and uses them differentially
extended period of time.
depending on level of threat. Under low threat, spiders showed
Experiment 2 supported our hypothesis that L. hesperus pos-
primarily nonconfrontational behaviours, including move (84%)
sesses control over venom expenditure. The rst level of venom
and retract (16%). Only a few responded by icking silk (5%), and
control is whether to use venom. Dry bites and stings have been
none responded by biting. Changes in movement patterns,
reported during defensive encounters in snakes (Hayes et al., 2002)
including arrested or diminished movement, is a common anti-
and scorpions (Nisani & Hayes, 2011), respectively. Our results
predator response observed in other spider species (Barnes,
suggest that L. hesperus frequently delivers dry defensive bites as
well. The majority of single bites (53.7%) yielded no measurable
Table 5 venom, and a larger proportion of all targets (65.4%), many bitten
Sequences of venom usage (dry versus wet venom expended) by spiders when multiple times, lacked measurable venom. Although our ELISA
defensively biting three targets in succession (N 80 trials) could not reliably measure venom samples below 0.06 mg, this
Target 1 Target 2 Target 3 Frequency Plausible value was <1% of the mean mass of venom measured on targets
interpretation bitten a single time (16 mg), suggesting that most or all of these
Brief Lengthy
intervals intervals bites were, indeed, functionally dry. Two lines of evidence further
suggest that dry bites were the result of decision making rather
Dry Dry Dry 15 7
Dry Dry Wet 8 7 Venom metering*
than venom depletion. First, there was no decline in the proportion
Dry Wet Dry 7 6 Venom depletiony of dry bites among the three successive targets, which would be
Dry Wet Wet 1 9 Venom metering* expected if the venom supply was being depleted. Second, dry bites
Wet Dry Dry 5 3 Venom depletiony often preceded wet bites, indicating that venom was available but
Wet Dry Wet 3 4 Venom metering*
withheld during the bites. Venom depletion almost certainly
Wet Wet Dry 1 1 Venom depletiony
Wet Wet Wet 0 3 occurred in some cases: one spider, for example, appeared to empty
its venom gland in the rst of three targets (142.7 mg of venom), and
* Venom metering: dry bites preceded wet bites, indicating available venom and
decision making.
released 0 mg of venom in each of the two subsequent targets. Some
y
Venom depletion: dry bites resulted from prior venom use that depleted spiders delivered venom to all targets, whereas others never used
reserves. their venom, suggesting individual proclivities towards dry bites.
 D. R. Nelsen et al. / Animal Behaviour 89 (2014) 107e114
The second level of venom control consists of varying the
amount of venom ejected. Spiders expulsed anywhere from 0 to
110 mg of venom on targets bitten once, averaging 34.6 mg when dry
bites were excluded. Coefcients of variation were similar to those
reported for snakes (Hayes, 2008) and the scorpion P. transvaalicus
(Nisani & Hayes, 2011). Thus, we can reject any notion that
L. hesperus always ejects a consistent bolus of venom (i.e. the bullet
hypothesis; Hayes, 2008; Hayes et al., 1995). In the envenomating
(wet) bites we measured, spiders expended an average of 27%
(34.6 mg) and up to 87% (110 mg) of the available venom (126 mg; see
Table 8 in dAmour et al., 1936) in a single bite.
Spiders could vary the amount of venom expended in two ways:
by varying the number of envenomating (wet) bites or by varying
the amount of venom delivered in individual bites. For the rst
possibility, we could not quantify the number of wet bites on in-
dividual targets bitten multiple times; however, ve spiders
delivered wet bites to multiple targets presented during the brief
intervals only 5 s apart, suggesting spiders can meter venom during
a single defensive encounter through multiple bites. Several lines of
evidence supported the second possibility as well. Spiders bit more
frequently when a leg was pinched, but delivered 1.8-fold more
venom per bite when the body was pinched. This might be ex-
pected because a body pinch presents a greater threat to the spider.
If a leg is lost, the spider can survive and tness is not greatly
reduced (Brueseke, Rypstra, Walker, & Persons, 2001), whereas
pinches or bites to the body can be lethal (Vetter, 1980). Further-
more, spiders pinched by the leg expulsed the most venom with the
rst of three targets, whereas those pinched by the body ejected the
most venom with subsequent targets. Some snakes (Hayes et al.,
2002) and scorpions (Nisani & Hayes, 2011) similarly increase or
decrease the amounts of venom expended in successive bites or
stings. Considering the large effect size, spiders also likely delivered
more venom during the lengthy intervals (5 min) between suc-
cessive pinches than during the brief intervals (5 s). We suggest
that the spiders treated the brief intervals as a single predatory
encounter and the lengthy intervals as separate events requiring
additional venom for each new attacker.
The use of venom and silk in defence, although potentially effec-
tive (Siemens et al., 2006; Vetter, 1980), is also costly. The metabolic
cost of venom has been demonstrated in snakes (McCue, 2006;
Morgenstern & King, 2012; Pintor, Krockenberger, & Seymour, 2010)
and in the scorpion Parabuthus transvaalicus (Nisani et al., 2007, 2012).
Studies also suggest there is a metabolic cost associated with silk
production (Boutry & Blackledge, 2008; Craig, Hsu, Kaplan, & Pierce,
1999; Tanaka, 1989). Because these two products, venom and silk,
can be considered limited commodities (Hayes, 2008), spiders should
be judicious in dispensing them. Our results support the suggestions
of dAmour et al. (1936) and Vetter (1980) that widow spiders make
decisions about use of venom and silk, respectively, and deploy them
only at highest levels of threat.
One caveat regarding venom measurement bears mention.
Scorpion venom has well-documented heterogeneity; that is, the
rst venom expelled from the gland has a different chemical
composition than venom expulsed during later stings (Inceoglu
et al., 2003; Nisani & Hayes, 2011). Venom heterogeneity may
also exist in spiders, but only one published abstract has explored
this possibility (Morgenstern et al., 2012). If venom heterogeneity
exists in L. hesperus, venom differences across successive stings
could result in differential recognition by the antibodies of our
ELISA measurement, resulting in bias in the quantities of venom
measured. The independent variable most likely affected by venom
heterogeneity would be comparisons among the three successive
targets.
Our study was also unnatural in that we did not test defensive
behaviours against natural predators. Apart from the wasps
113
C. californicum and T. festiva, little is known about other major
predators of L. hesperus, although centipedes, mantids and rodents
are believed to be important in controlling black widow pop-
ulations (Irving & Hinman, 1935; Landes et al., 1987; Vetter, 1980).
Additionally, little to nothing is known about how predators of
L. hesperus vary through ontogeny, although young spiders enjoy
protection within their mothers web and conspecic predation
may exist (Vetter, 1980). Even with known predators, no research to
our knowledge has focused on defensive interactions of L. hesperus
in a natural setting.
In conclusion, we believe our ndings support the interpreta-
tion that black widow spiders make decisions about defensive be-
haviours and about venom expenditure in particular. Although
cognitive control, including higher-order cognitive processing
(Jackson & Cross, 2013; Menzel, 2013; Toates, 2006; Webb, 2012),
may be involved, more testing is necessary to explore this possi-
bility, as other more simple interpretations may explain our results.
Our ndings also support the growing body of literature showing
that a diverse range of animals have the ability to meter their
venom. This is the rst study to demonstrate that spiders actively
control venom expulsion during defensive interactions. Further
study should be undertaken to measure silk quantities expended by
spiders to determine whether this limited commodity is also
metered.
Acknowledgments
We express our thanks to Gerad Fox and Allen Cooper for their
input on designing the experiments and measuring the venom.
These individuals, along with Leonard Brand, Kevin Nick, Penelope
Duerksen-Hughes, Ernie Schwab, Zia Nisani and anonymous ref-
erees provided helpful comments on the manuscript.
References
Alves De Rezende, N., Maia Torres, F., Borges Dias, M., Campolina, D., Chavez-
Olortegui, C., & Faria Santos Amaral, C. (1998). South American rattlesnake bite
(Crotalus durissus sp) without envenoming: insights on diagnosis and treat-
ment. Toxicon, 36, 2029e2032.
dAmour, F. E., Becker, F. E., & Van Riper, W. (1936). The black widow spider.
Quarterly Review of Biology, 11, 123e160.
Barnes, M. C., Persons, M. H., & Rypstra, A. L. (2002). The effect of predator chemical
cue age on antipredator behavior in the wolf spider Pardosa milvina (Araneae:
Lycosidae). Journal of Insect Behavior, 15, 269e281.
Bednekoff, P. A. (2007). Foraging in the face of danger. In D. W. Stephens, J. S. Brown,
& R. C. Ydenberg (Eds.), Foraging behavior and ecology (pp. 305e329). Chicago,
IL, U.S.A.: University of Chicago Press.
Blackledge, T. A., & Wenzel, J. W. (2001). Silk mediated defense by an orb web spider
against predatory mud-dauber wasps. Behaviour, 138, 155e171.
Blair, A. W. (1934). Spider poisoning: experimental study of the effects of the
bite of the female Latrodectus mactans in man. Archives of Internal Medicine,
54, 831e843.
Boev, J. L. (1994). Injection of venom into an insect prey by the free hunting spider
Cupiennius salei (Araneae, Ctenidae). Journal of Zoology, 234, 165e175.
Boev, J. L., Kuhn-Nentwig, L., Keller, S., & Nentwig, W. (1995). Quantity and quality
of venom released by a spider (Cupiennius salei, Ctenidae). Toxicon, 33, 1347e
1357.
Boutry, C., & Blackledge, T. (2008). The common house spider alters the material
and mechanical properties of cobweb silk in response to different prey. Journal
of Experimental Zoology, 309, 542e552.
Brueseke, M. A., Rypstra, A. L., Walker, S. E., & Persons, M. H. (2001). Leg autotomy in
the wolf spider Pardosa milvina: a common phenomenon with few apparent
costs. American Midland Naturalist, 146, 153e160.
Caro, T. (2005). Antipredator defenses in birds and mammals. Chicago, IL, U.S.A.:
University of Chicago Press.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).
Hillsdale, NJ, U.S.A.: L. Erlbaum.
Craig, C. L., Hsu, M., Kaplan, D., & Pierce, N. E. (1999). A comparison of the
composition of silk proteins produced by spiders and insects. International
Journal of Biological Macromolecules, 24, 109e118.
Escoubas, P., Sollod, B., & King, G. F. (2006). Venom landscapes: mining the
complexity of spider venoms via a combined cDNA and mass spectrometric
approach. Toxicon, 47, 650e663.
114 D. R. Nelsen et al. / Animal Behaviour 89 (2014) 107e114
Ferrari, M. C., Sih, A., & Chivers, D. P. (2009). The paradox of risk allocation: a review
and prospectus. Animal Behaviour, 78, 579e585.
Fry, B. G., Roelants, K., Champagne, D. E., Scheib, H., Tyndall, J. D., King, G. F., et al.
(2009). The toxicogenomic multiverse: convergent recruitment of proteins into
animal venoms. Annual Review of Genomics and Human Genetics, 10, 483e511.
Glaudas, X., & Gibbons, J. W. (2005). Do thermal cues inuence the defensive strike
of cottonmouths (Agkistrodon piscivorus)? Amphibia Reptilia, 26, 264e267.
Glaudas, X., Winne, C. T., & Fedewa, L. A. (2006). Ontogeny of anti-predator
behavioral habituation in cottonmouths (Agkistrodon piscivorus). Ethology, 112,
608e615.
Green, S. B., & Salkind, N. J. (2005). Using SPSS for windows and macintosh: Analyzing
and understanding data (4th ed.). Upper Saddle River, NJ, U.S.A.: Pearson
Prentice Hall.
Grostal, P., & Dicke, M. (2000). Recognising ones enemies: a functional approach to
risk assessment by prey. Behavioral Ecology and Sociobiology, 47, 258e264.
Haight, K. L. (2006). Defensiveness of the re ant, Solenopsis invicta, is increased
during colony rafting. Insectes Sociaux, 53, 32e36.
Hayes, W. K. (1992). Prey-handling and envenomation strategies of prairie rattle-
snakes (Crotalus viridis viridis) feeding on mice and sparrows. Journal of Her-
petology, 26, 496e499.
Hayes, W. K. (1993). Effects of hunger on striking, prey-handling, and venom
expenditure of prairie rattlesnakes (Crotalus viridis viridis). Herpetologica, 49,
305e310.
Hayes, W. K. (2008). The snake venom-metering controversy: levels of analysis,
assumptions, and evidence. In W. K. Hayes, K. R. Beaman, M. D. Cardwell, &
S. P. Bush (Eds.), The biology of rattlesnakes (pp. 191e220). Loma Linda, CA,
U.S.A.: Loma Linda University Press.
Hayes, W. K., Herbert, S. S., Harrison, J. R., & Wiley, K. L. (2008). Spitting versus
biting: differential venom gland contraction regulates venom expenditure in
the black-necked spitting cobra, Naja nigricollis nigricollis. Journal of Herpetology,
42, 453e460.
Hayes, W. K., Herbert, S. S., Rehling, G. C., & Gennaro, J. F. (2002). Factors that in-
uence venom expenditure in viperids and other snake species during preda-
tory and defensive contexts. In G. W. Schuett, M. Hggren, M. E. Douglas, &
H. W. Greene (Eds.), Biology of the vipers (pp. 207e233). Eagle Mountain, UT,
U.S.A.: Eagle Mountain.
Hayes, W. K., Lavn-Murcio, P., & Kardong, K. V. (1995). Northern Pacic rattlesnakes
(Crotalus viridis oreganus) meter venom when feeding on prey of different sizes.
Copeia, 1995, 337e343.
Hostettler, S., & Nentwig, W. (2006). Olfactory information saves venom during
prey-capture of the hunting spider Cupiennius salei (Araneae: Ctenidae). Func-
tional Ecology, 20, 369e375.
Inceoglu, B., Lango, J., Jing, J., Chen, L., Doymaz, F., Pessah, I. N., et al. (2003). One
scorpion, two venoms: prevenom of Parabuthus transvaalicus acts as an alter-
native type of venom with distinct mechanism of action. Proceedings of the
National Academy of Sciences USA, 100, 922e927.
Irving, W. G., & Hinman, E. H. (1935). The blue mud-dauber as a predator of the
black widow spider. Science, 82, 395e396.
Jackson, R. R., Brassington, R. J., & Rowe, R. J. (1990). Anti-predator defences of
Pholcus phalangioides (Araneae, Pholcidae), a web-building and web-invading
spider. Journal of Zoology, 220, 543e552.
Jackson, R. R., & Cross, F. R. (2013). A cognitive perspective on aggressive mimicry.
Journal of Zoology, 290, 161e171.
Jackson, R. R., Rowe, R. J., & Campbell, G. E. (1992). Anti-predator defences of Psi-
lochorus sphaeroides and Smeringopus pallidus (Araneae, Pholcidae), tropical
web-building spiders. Journal of Zoology, 228, 227e232.
Jackson, R. R., Rowe, R. J., & Wilcox, R. S. (1993). Anti-predator defences of Argiope
appensa (Araneae, Araneidae), a tropical orb-weaving spider. Journal of Zoology,
229, 121e132.
Johnson, J. C., & Sih, A. (2007). Fear, food, sex and parental care: a syndrome of
boldness in the shing spider, Dolomedes triton. Animal Behaviour, 74, 1131e1138.
Kats, L. B., & Dill, L. M. (1998). The scent of death: chemosensory assessment of
predation risk by prey animals. Ecoscience, 5, 361e394.
King, G. F. (2004). The wonderful world of spiders: preface to the special Toxicon
issue on spider venoms. Toxicon, 43, 471e475.
Landes, D. A., Obin, M. S., Cady, A. B., & Hunt, J. H. (1987). Seasonal and latitudinal
variation in spider prey of the mud dauber Chalybion californicum (Hymenop-
tera, Sphecidae). Journal of Arachnology, 15, 249e256.
Li, M., Fry, B. G., & Kini, R. M. (2005). Eggs-only diet: its implications for the toxin
prole changes and ecology of the marbled sea snake (Aipysurus eydouxii).
Journal of Molecular Evolution, 60, 81e89.
Lima, S. L., & Bednekoff, P. A. (1999). Temporal variation in danger drives anti-
predator behavior: the predation risk allocation hypothesis. American Naturalist,
153, 649e659.
Lima, S. L., & Dill, L. M. (1990). Behavioral decisions made under the risk of pre-
dation: a review and prospectus. Canadian Journal of Zoology, 68, 619e640.
Lima, S. L., & Steury, T. D. (2005). Perception of predation risk: the foundation of
nonlethal predatoreprey interactions. In P. Barbosa, & I. Castellanos (Eds.),
Ecology of predatoreprey interactions (pp. 166e188). Oxford, U.K.: Oxford Uni-
versity Press.
Lohrey, A. K., Clark, D. L., Gordon, S. D., & Uetz, G. W. (2009). Antipredator responses
of wolf spiders (Araneae: Lycosidae) to sensory cues representing an avian
predator. Animal Behaviour, 77, 813e821.
Malli, H., Imboden, H., & Kuhn-Nentwig, L. (1998). Quantifying the venom dose of the
spider Cupiennius salei using monoclonal antibodies. Toxicon, 36, 1959e1969.
Malli, H., Kuhn-Nentwig, L., Imboden, H., & Nentwig, W. (1999). Effects of size,
motility and paralysation time of prey on the quantity of venom injected by the
hunting spider Cupiennius salei. Journal of Experimental Biology, 202, 2083e
2089.
McAlister, W. (1960). The spitting habit in the spider Scytodes intricata Banks
(family Scytodidae). Texas Journal of Science, 12, 17e20.
McCue, M. D. (2006). Cost of producing venom in three North American pitviper
species. Copeia, 2006, 818e825.
Menzel, R. (2013). Learning, memory, and cognition: animal perspectives. In
C. G. Galizia, & P. M. Lledo (Eds.), Neurosciences: From molecule to behavior: A
university textbook (pp. 629e653). New York, NY, U.S.A.: Springer.
Morgenstern, D., Hamilton, B., Sher, D., Jones, A., Mattius, G., Zlotkin, E., et al. (2012).
The bio-logic of venom complexity. Toxicon, 60, 241e242.
Morgenstern, D., & King, G. F. (2012). The venom optimization hypothesis revisited.
Toxicon, 63, 120e128.
Nakagawa, S. (2004). A farewell to Bonferroni: the problems of low statistical power
and publication bias. Behavioral Ecology, 15, 1044e1045.
Nakagawa, S., & Cuthill, I. C. (2007). Effect size, condence interval and statistical
signicance: a practical guide for biologists. Biological Reviews, 82, 591e605.
Nascimento, D. G., Rates, B., Santos, D. M., Verano-Braga, T., Barbosa-Silva, A.,
Dutra, A. A., et al. (2006). Moving pieces in a taxonomic puzzle: venom 2D-LC/
MS and data clustering analyses to infer phylogenetic relationships in some
scorpions from the Buthidae family (Scorpiones). Toxicon, 47, 628e639.
Nelsen, D. R., Nisani, Z., Cooper, A. M., Fox, G. A., Gren, E. C. K., Corbit, A. G., et al.
(2013). Poisons, toxungens, and venoms: redening and classifying toxic bio-
logical secretions and the organisms that employ them. Biological Reviews.
Advance online publication. http://dx.doi.org/10.1111/brv.12062.
Nisani, Z., Boskovic, D. S., Dunbar, S. G., Kelln, W., & Hayes, W. K. (2012). Investi-
gating the chemical prole of regenerated scorpion (Parabuthus transvaalicus)
venom in relation to metabolic cost and toxicity. Toxicon, 60, 315e323.
Nisani, Z., Dunbar, S. G., & Hayes, W. K. (2007). Cost of venom regeneration in
Parabuthus transvaalicus (Arachnida: Buthidae). Comparative Biochemistry and
Physiology Part A: Molecular & Integrative Physiology, 147, 509e513.
Nisani, Z., & Hayes, W. K. (2011). Defensive stinging by Parabuthus transvaalicus
scorpions: risk assessment and venom metering. Animal Behaviour, 81, 627e633.
Nordt, S. P., Clark, R. F., Lee, A., Berk, K., & Cantrell, F. L. (2012). Examination of
adverse events following black widow antivenom use in California. Clinical
Toxicology, 50, 70e73.
Obin, M. S., & Vander Meer, R. K. (1985). Gaster agging by re ants (Solenopsis
spp.): functional signicance of venom dispersal behavior. Journal of Chemical
Ecology, 11, 1757e1768.
Persons, M. H., Walker, S. E., & Rypstra, A. L. (2002). Fitness costs and benets of
antipredator behavior mediated by chemotactile cues in the wolf spider Pardosa
milvina (Araneae: Lycosidae). Behavioral Ecology, 13, 386e392.
Pierce, C. A., Block, R. A., & Aguinis, H. (2004). Cautionary note on reporting eta-
squared values from multifactor ANOVA designs. Educational and Psychologi-
cal Measurement, 64, 916e924.
Pintor, A. F., Krockenberger, A. K., & Seymour, J. E. (2010). Costs of venom production
in the common death adder (Acanthophis antarcticus). Toxicon, 56, 1035e1042.
Riechert, S. E., & Hedrick, A. V. (1990). Levels of predation and genetically based
anti-predator behaviour in the spider, Agelenopsis aperta. Animal Behaviour, 40,
679e687.
Rodrguez-Robles, J. A., & Leal, M. (1993). Effects of prey type on the feeding
behavior of Alsophis portoricensis (Serpentes: Colubridae). Journal of Herpetol-
ogy, 27, 163e168.
Siemens, J., Zhou, S., Piskorowski, R., Nikai, T., Lumpkin, E. A., Basbaum, A. I., et al.
(2006). Spider toxins activate the capsaicin receptor to produce inammatory
pain. Nature, 444, 208e212.
Tanaka, K. (1989). Energetic cost of web construction and its effect on web relo-
cation in the web-building spider Agelena limbata. Oecologia, 81, 459e464.
Taylor, A. R., Persons, M. H., & Rypstra, A. L. (2005). The effect of perceived predation
risk on male courtship and copulatory behavior in the wolf spider Pardosa
milvina (Araneae, Lycosidae). Journal of Arachnology, 33, 76e81.
Toates, F. (2006). A model of the hierarchy of behaviour, cognition, and con-
sciousness. Consciousness and Cognition, 15, 75e118.
Vetter, R. S. (1980). Defensive behavior of the black widow spider Latrodectus
hesperus (Araneae: Theridiidae). Behavioral Ecology and Sociobiology, 7,
187e193.
Vetter, R. S., & Isbister, G. K. (2008). Medical aspects of spider bites. Annual Review of
Entomology, 53, 409e429.
Webb, B. (2012). Cognition in insects. Philosophical Transactions of the Royal Society
B: Biological Sciences, 367, 2715e2722.
Wigger, E., Kuhn-Nentwig, L., & Nentwig, W. (2002). The venom optimisation hy-
pothesis: a spider injects large venom quantities only into difcult prey types.
Toxicon, 40, 749e752.
Wisenden, B. D. (2000). Olfactory assessment of predation risk in the aquatic
environment. Philosophical Transactions of the Royal Society of London, Series B:
Biological Sciences, 355, 1205e1208.
Wullschleger, B., & Nentwig, W. (2002). Inuence of venom availability on a spiders
prey: choice behaviour. Functional Ecology, 16, 802e807.
Zar, J. H. (1996). Biostatistical analysis (3rd ed.). Englewood Cliffs, NJ, U.S.A.: Prentice Hall.