CAN SIMPLE EXPERIMENTAL ELECTRONICS SIMULATE THE DISPERSAL PHASE OF SPIDER BALLOONERS?

2005·10.2 MB·English

by  James R Bell

2005. The Journal of Arachnology 33:523-532
CAN SIMPLE EXPERIMENTAL ELECTRONICS SIMULATE
THE DISPERSAL PHASE OF SPIDER BALLOONERS?
James R* Bell: Warwick HRI, Wellesbourne, Warwickshire CV35 9EF. E-mail:
@
.nbell warwick.ac.uk
j
David A« Bohan and Richard Le Fevre: Rothamsted Research, Harpenden,
Hertfordshire AL5 2JQ
Gabriel S« Weyman: Syngenta, Jealott’s Hill International Research Centre,
Bracknell, Berkshire RG42 6EY
ABSTRACT. Here we describe the structure of a fall speed chamber designed to measure, with low
experimental error, the terminal velocities (fall speeds) of spiders ofknown weight and a given length of
silk. We also describe the construction ofa simulated individual (SI) which could laterbeused to estimate
the distance travelled by ballooning spiders in the field. Our data and analysis suggest that Oedothorax
spp. (Linyphiidae) and Pachygnatha degeeri (Tetragnathidae) individuals have fall speeds that can be
described by their silk length and mass. Of the observed deviance in the fall speeds, 73.7% could be
explained by a GLM model common to both species groups. Overlaying the SI fall speed data on this
GLM surface suggests thatthe Sis have similarfall speedbehaviors to spiders. However, furtherestimation
is necessary before Sis could be considered valid models for evaluating spider ballooning distances.
Keywords? Dispersal, ballooning, schottky diodes, silk, fall speed chamber
Ballooning research has faced a seemingly flying insects up to 1 km above ground level
intractable question for over 300 years: how (Chapman et al. 2003), as yet ballooning spi-
far do ballooning spiders disperse once air- ders cannot be uniquely identified. The VLR
borne? While there have been attempts to ob- fails to resolve ballooeers because spiders
serve ballooning distances visually, which lack distinctive allometric ratios and tend to
suggest that spiders move no more than a few have masses near or below the critical thresh-
hundred metres in any one attempt (e.g. old for the radar (Chapman et al. 2003; Jason
MacCook, 1877; Folleer & Klareeberg 1995; Chapman pers. comm.).
Schneider et al. 2001), it also may be inferred Recently, indirect molecular genetic tech-
from anecdotal evidence that spiders also niques have been employed as an alternative
make journeys of several hundred kilometres to measuring airborne spiders directly (Good-
(Yoshimoto & Gressitt 1960; Okuma & Kisi- acre 2004). This approach was designed to de-
moto 1981). However, such visual and anec- tect the effect ofdispersal rates on the genetic
dotal data are rare and have yet to yield any diversity of a number of key linyphiid popu-
significant data for the great majority of bai- lations across the British mainland and its is-
looners, including the linyphiids (Bell et al, lands. The research has shown that popula-
2005). Although models of ballooning dis- tions on islands have lower genetic diversity
tance have been constructed (e.g. Thomas et than those found on the mainland, implying
al 2003), the predicted distances have yet to changes in gene flow with isolation distance
be verified. and island size. It should be noted however,
The lack of progress is perhaps surprising that these findings were not independent of
given recent advances in radar technology Wolbachia infections which confounded ob-
(Chapman et ai. 2003). Although Rothamsted served gene flow measurements. Othermolec-
Research’s vertical looking radar (VLR) can ular studies, which indirectly estimate bal-
measure the horizontal speed, displacement looning distance using gene flow, have been
direction, body alignment, mass and shape of conducted (as reviewed by Bell et al. 2005)
523
—
524 THE JOURNAL OF ARACHNOLOGY
inductive loop to which a
schottky diode is attached
—
Figure 1. A simulated individual, which shows a 16mm dipol antenna attached to an inductive loop.
The 0.3mm schottky diode (not visible) is fixed to the inductive loop using gold spatter technology. The
complete unit weighs 8mg.
i
but have not yielded estimates for the distanc- times (see world catalog in Bell et al. 2005).
es travelled by individual spiders. These species were used as model ballooners
We present an alternative approach, based for comparison with a simulated individual
upon synthetic models for spiders that we (SI). While the properties of an SI are yet to j
term simulated individuals (SI). Sis show be established, the desirable traits should be
great promise because they are traceable and that it: i) is structurally similar to a spider,
biologically inert, thus resolving problems of consisting of a body and an associated silk
airborne detection and Wolbachia infection. component; ii) is able to generate its own drag
However, while we have begun to understand to enable it to become airborne; iii) is trace-
the physical properties ofspiders and theirim- able, producing an automated signal of its lo-
plications for ballooning (Suter 1991, 1992, cation; iv) allows manipulation of the silk
1999), the properties of Sis are unknown and component to known levels ofdrag; and, last-
their comparative behavior remains untested. ly v) is a 'nuir spider with no behavior which
In this paper, a description of the technology minimizes drag variability (i.e. absence ofbit-
and data used to compare SI properties against ing and reeling of the silk line and reduced
spiders is presented, concluding with a dis- body posture modification). Schottky diodes,
cussion of future research prospects for bal- mounted onto an inductive loop with a dipole
looning. antenna (referred to as ‘diodes’ hereafter)
METHODS have the potential to offerthese properties, de-
spite having none of the physical attributes of
Simulated individuals and spiders. spiders (Fig. 1). We used 8 mg diodes in the
Males and females ofOedothorax spp. (mixed following experiments.
apicatus,fuscus and retusus species: Linyphi- Spiders create drag with single or multiple
idae) and Pachygnatha degeeri (Tetragnathi- silk lines that may account for 75% of the
dae) have been recorded ballooning many total drag ofthe spider (Humphrey 1987). For
BELL ET AL.—SIMULATION OF BALLOONING DISPERSAL PHASE 525
I the diodes, simulated silk was adopted initial- stage at terminal velocity when timing started
ly as a replacement for natural silk. Titanium- (Fig. 2). Timing was stopped, and the fall
coated fibre glass was identified as a possible speed computed, when the individual passed
solution and responded positively to very light through the third detector stage.
convection currents (i.e. < 1 m/s). However, The hardware environment behind the fall
it proved to be fragile despite being fourtimes speed chamber measurements is based on the
the diameter (400 nm) ofnatural linyphiid silk simple principle that when an object breaks a
(e.g. Tenuiphantes tenuis 100 nm). In light of light beam, a passive record can be logged at
these flaws, we used natural silk. Although a given point in time. Technically, the cham-
linyphiid silk is too fine to manipulate easily, ber included its own microprocessorcontroller
it was possible to attach the drag line silk of based on a PIC16F876 running at 20 MHz and
immature Araneus diadematus (Araneidae) to programmed using CCS PICC compiler (Fig.
; the diodes. For both spiders and diodes, all 4). This controller was connected to a PC run-
individuals were weighed before being intro- ning dedicated software through aRS232 port,
duced to the Rothamsted fall speed chamber which allowed the user to control the light
described below. In total, 38 spiders (Oedoth- source and silk release mechanism (i.e. auto-
orax spp. n ~ 13; P. degeeri n = 25) and 4 matic/manual release) remotely. All control
diodes were dropped attached to silk lengths outputs were by opto-isolated open drain mos-
within the range of 0-2.3 m. — fet drivers. The hardware detected falling ob-
I Rothamsted fall speed chamber. The jects through the use of a medium area photo
physical structure ofthe 9 m vertical chamber diode (41.3 mm^) connected to a two stage
was relatively simple and included three de- high gain amplifier. A first order bandpass fil-
tector stages and a hotwire (Fig. 2). The hot- ter was used to remove unwanted signals be-
;
wire was used as a silk-shearing mechanism low 300 Hz and above 5 KHz. The photo di-
to allow suspended spiders to be dropped ode was mounted in a black box, with one end
without human intervention. The first detector cut off, to help prevent ambient light interfer-
j stage was used to manipulate the silk length, ing with the source light. The initial design
m
between 0.11 and 2.3 m, at which a sus- for the detection system was to incorporate a
pended spider or diode entered free fall and laser diode with line generatorlens as the light
;
the two remaining stages measured the fall source. However, the tested lasers were found
speed of each individual having reached ter- to have a small but significant fluctuation in
minal velocity. As a precursor to entering the their output which made it impossible to dis-
chamber, spiders were first prompted to drop tinguish the object signal from noise when
down on a drag line from an oscillating probe. used in conjunction with the detection circuit.
I Having produced a dragline of >10 cm, spi- The circuit will need to be redesigned before
'
ders were then fixed to the hotwire and al- lasers can be used in this application.
I lowed to pay out more silk until triggering the As an alternative, high power quartz halo-
!' first detector stage (Fig. 3). In separate exper- gen bulbs (60 W) were used in conjunction
mm mm
iments, the diodes were suspended on fixed with two 0.8 slits spaced at about 160
lengths of silk placed on the hotwire. The silk apart so that a reasonably fine line beam could
was sheared by one of two methods: either a) be produced (Fig. 5). To focus the light onto
j the spider broke the first detector stage light the photo diode, Fresnel lenses (—300 mm
I beam which automatically triggered the ho- wide, cut from a 280 mm square lens along
! &
twire (Figs. 2 3); or, b) if shorter lengths of the diagonal, 50 groves per inch and a focal
! silk (i.e. < 0.11 m), silkless drops or fixed length of234 mm) were employed. The signal
drops with diodes were required, a PC-oper- from the photo diode was then amplified and
I
ated drop mechanism which manually trig- filtered before being applied to the single in-
gered both the three detector stage lightbeams put channel of the analogue switch driven by
' and the hotwire to an ‘on’ position was used. a free running 3 KHz quartz clock (Fig, 4).
After either the hotwire or manual drop had The two output channels of this switch were
I been triggered, the spiderordiode enteredfree then applied to the inputs of the voltage com-
i fall for at least 5.4 m (i.e. depending on the parator. Any low frequency variation of the
first detector stage height) until it was mea- input signal due to amplifier drift or ambient
sured passing through the second detector light falling on the photo diode was ignored
526 THE JOURNAL OF ARACHNOLOGY
air movement due to chimney effect
—
Figure 2. Side view of the Rothamsted fall speed chamber.
by the comparator. However, any object pass- as the trigger input to the microprocessorcon-
ing through the light beam produced a much troller. The inherent precision of the micro-
faster change in signal level which trigged the processor quartz clock ensured that the accu-
comparator. The comparator output was used racy of the system fell within at least ± 1 ms
BELL ET AL.—SIMULATION OF BALLOONING DISPERSAL PHASE 527
—
Figure 3. Top of Rothamsted fall speed chamber showing the hotwire release mechanism to which a
spider is suspended on a silken line.
GLM
(accuracy checked against acalibrated Systron 73.7% of the deviance observed (Fig.
Donner counter timer type 6250A) and rep- 6). The data were found to be underdis-
resented a fall time recorder error of the spi- persed, suggesting that the data were more
ders sampled between—0.0046-0.055%. regularly distributed than expected for data
Statistical analysis. Fall speeds were an- conforming to the Normal distribution. An
alyzed using a Generalized Linear Model empirical scale parameter was used to adjust
(GLM), the Normal distribution and the log- the model fitted estimates of error to account
link function in Genstat (version 6, VSN in- for this underdispersion (see McCullagh &
ternational, Oxford, UK; McCullagh & Nelder Nelder 1989).
1989). Logio(Silk Length + 1) was fitted in Spider fall speeds were found to decrease
the model as the explanatory variable, with with increasing silk length (fi^g = 2.87, P =
spider species and logio(Spider Mass) as cov- 0 004 and increase with an increase in spider
. )
ariates. The model fit was checked for over- body mass “ 3.25, P < 0.001). There
&
dispersion in the data (McCullagh Nelder was no interaction between silk length and
1989). The model’s standardized residuals spider body mass (tj35 = 0.28, P = 0.78). No
wmatoetgreeemnpectihtewycakse(dMmcafCdourellltiaongefhairti&tay,GNlLeelMvdeerrtaog1et9h8ea9n)pd.rohvNoio-- dasicpfteficeoirneesnwcaesifno~tuhne0d.G7b0Le,tMPwew=eans0s.f4po9idu),enrdanswdpietcnhioessipniatdenerdr-
sional data for the diodes. silk length (ti3^ = 0.62, P = 0.53) nor spider
RESULTS weight (^136 = 1.31, P = 0.19). Thus, a com-
The GLM was found to fit the spider fall mon GLM was applicable to both Oedothorax
speed data extremely well, explaining some spp. and P. degeeri:
528 THE JOURNAL OF ARACHNOLOGY
—
Figure 4. Block diagram detailing system electronics.
logio(Fall Speed) = 3.83 with respect to weight (females =2.4 ± 0.3
mg; males = 0.8 ± 0.004; n = 50.14, P <
- 0.951og,o (Silk Length +1) 0.001), yielding sex specific fall speeds for a
We
+ l.lllogio (Spider Mass) given silk length in this species group.
plotted the fall speeds for diodes over the
Spider sex was a non-significant model GLM in Fig. 6. The overlaid data suggests
covariate — 0.06, P = 0.95). However, that diodes behave in a manner that is analo-
the Oedothorax spp. are sexually dimorphic gous to the spiders.
—
—
BELL ET AL. SIMULATION OF BALLOONING DISPERSAL PHASE 529
Photo diode
and enclosure
Fresnel lens - 300 x 50mm
(L X H) and FL = 234mm
Halogen bulb and first
mm
1 slit in metal
enclosure
\
n Senso'
Light beam - Second 1mm slit
detection zone
0.29m 1.05m
Plan view
—
Figure 5. Diagramatic view ofthe filtered light source (4=-) producing a large detection zone which is
constantly monitored by the photo diode.
DISCUSSION have the potential to develop our understand-
Rothamsted fall speed measurements.- ing of spider ballooning far beyond our pre-
This experiment unequivocally demonstrates sent knowledge.
that natural spider silk can be attached to di- Spider ballooning research is limited, al-
odes and that drag, and consequently fall though scientists are aware of the importance
speeds, can be systematically manipulated of silk in ballooning (Bell et ah 2005). For
through the length of the silk line. The ob- example, the effect of silk length on the fall
served positive relationship between drag and speeds of spiders (Suter 1991), moth larvae
silk length for Sis was analogous, but not (Lepidoptera) (Batzer 1968; Barel 1973;
identical, to spiders in free fall. Despite the Mitchell 1979; McManus & Mason 1983; Ra-
limits of the provisional data presented, our machaedrae 1987) and spider mites (Tetran-
&
results are supportive and imply that these di- ychidae) (Jung Croft 2001) has already
odes represent a simple, yet viable paradigm been demonstrated. Of these, Suter’s (1991)
of real spiders. Encouragingly, these diodes seminal research attempted to evaluate fall
530 THE JOURNAL OF ARACHNOLOGY
—
Figure 6. Observed fall speed data for individuals of Pachygnatha degeeri (•), Oedothorax spp. (o)
and diode () against silk length and spider mass. The hatched surface represents the GLM fitted model
for fall speeds with silk length and spider mass.
times independently of human error. Even so, influence where they land (Jung & Croft
fall speeds still had to be estimated by extrap- 2001). Such postural control could accountfor
olation because of the short fall distances in some ofthe unexplained variation in ourGLM
Suter’s experimental chamber. The advantages and may be estimated by placing digital cam-
ofthe Rothamsted fall speed chamber are that eras inside the Rothamsted fall speed cham-
measurements may be taken in near-still air ber. The posture of the photographed spiders,
conditions and when they have reached their once categorized by shape, might then be in-
terminal velocity, after individuals have fallen cluded as a third covariate within the GLM.
at least 5.4 m. Despite this, the results are sub- However, this behavior might only be expect-
ject to error due to spiderbehaviors when fall- ed to account for a maximum of 25% of the
ing. Here no attempt was made to control for observed variation (deviance) in the fall speed
postural variation, such as spreading or with- data (Suter 1991; see also Humphrey 1987).
drawing legs, which has been estimated to Allowing spiders to reach their terminal ve-
have up to a 10 fold effect on body drag (Su- locities over a comparatively large distance
ter 1992). Postural control may also be im- simplifies the mathematics of calculating ter-
portant in mites which manipulate drag in a minal fall speeds, but also has the potential to
similar fashion to spiders and may be able to be biologically erroneous. Purely from obser-
BELL ET AL.—SIMULATION OF BALLOONING DISPERSAL PHASE 531
vatioe during handling, both species tested 1987); although much heavier spiders can be
were able to pay-out silk at a rate of >1 m/s, found ballooning. Reducing schottky diode
particularly when individuals adopted "es- mass by at least 75% would affect the drag
I cape’ behaviors. If this payieg^out of silk oc- dramatically and could require models for Sis
curred during freefall, thee the lightest indi- that differ significantly from that estimated
viduals could produce several meters of here for spiders. Only after completion ofthis
‘undetected’ silk length during their fall. In diode model estimation phase of the project
practice this is unlikely, given that the residual could radar-based fieldwork follow.
variation was relatively small. While it is im- ACKNOWLEDGMENTS
portant to highlight posture and silk reeling as
sources of error, they are inescapable covar- Thanks are due to Alan Smith, Jason Chap-
iates of spider ballooning. Posture variation man, Kelvin Conrad (RR) and Andrew Mead
might be standardized, though not removed, (Warwick HRI) for their helpful discussions
' by anaesthetising individuals with carbon di- and advice. We also thank Ian Denholm (RR),
oxide before entering the chamber (Jung & David Skirvin and Rosemary Collier (War-
Croft 2001). However, this would have an im- wick HRI) for proof reading the MS. We are
pact on an individual’s ability to produce silk. grateful for the assistance of Soeren Toft as
The solution to evaluating the effects of pos- ecology editor and thank the two peer review-
ture and variation in silk length during freefall ers for their comments. This paper is dedicat-
can only be to increase the number of obser- ed to the memory of Julian Haughtoe, natu-
vations (replicates). ralist and friend.
Suter (1991, 1992) recognized the impor-
LITERATURE CITED
tance ofspider mass, which served to increase
the fail speeds at a given length of silk. Math- Bare!, CJ.A. 1973. Studies on dispersal ofAdoxo-
ematical models which seek to determine the phyes orana F.V.R. in relation to the population
probability of dispersal based on a species by sterilization technique. MededelingenLandbhoo-
: species account, also need to parameterize gesch, Wageningen 73:1-107.
mass and consider sex as a covariate where Batzer, H.O. 1968. Hibernation site and dispersal of
i spruce budworm larvae as related to damage of
obvious differences in males and females oc-
I cur. As far as we are aware, this has been ig- 2sa1p6l-i2n2g0.balsam fir. Economic Entomology 61:
'j noted to date. Bell, J.R., D.A Bohan., E.M. Shaw & G.S. Wey-
! The future ofschottky diodes to simulate man. 2005. Ballooning dispersal using silk:
the dispersal phase of spider ballooeers.— world fauna, phylogenies, genetics and models.
This research has shown that simulating bal- Bulletin of Entomological Research 95:69-114.
looeers has potential. Natural silk attached to Chapman, J.W., D.R. Reynolds, & A.D. Smith,
I diode bodies produces drag in a manner di- 2003. Vertical-looking radar: A new tool for
rectly analogous to a spider. Scanning har- monitoring high-altitude insect migration. Bio-
monic radar has been shown to be effective in science 5&3:503-511.
m Follner, K. A.J. Klarenberg. 1995. Aeronauticbe-
tracking diode-tagged bees for up to 900
haviour in the wasp-like spider Argiope bruen-
from the radar station (Osborne et al. 1999). nichi (Scopoli) (Araneae. Argiopidae). In: V.
To follow Sis, the use of a similar scanning Ruzicka (ed.) Proceedings of the 15th European
radar set-up is planned. Using this technology Colloquium ofArachnology. Ceske, Budejovice.
i we can explore unanswered questions includ- pp. 66-72.
i ing; how far do ballooeers travel; and, what Goodacre, S. 2004. Population structure and evolu-
is the pattern ofdispersal ofballooners within tionary relationships among linypMid spiders: a
a 1-2 km range? However, our research is at molecular analysis. In: Scientific abstracts of the
an early stage. While releasing diodes in the 16* International Congress ofArachnology, Gent,
I field is the ultimate objective, several aspects rBuegl.gaicu.mbe2/-'7^jApumgauesltfa2/0A0b4s.trp.ac5t6s.%2ht0tLpe:/z/ianllgseernv%.
of SI behavior need further estimation before
I SI ballooning data can be captured. Notably, 20(all).pdf
I Greenstone M.H., C.E. Morgan, A.L. Hultsch, R.A.
the dependence of fall speeds on diode mass Farrow & J.E. Dowse. 1987. Ballooning spiders
requires evaluation because the 8 mg diodes in Missouri, USA, and New South Wales, Aus-
used do not represent the majority ofballooe- tralia: family and mass distributions. Journal of
ers, which are under 2 mg (Greenstone et al. Arachnology 15:163-170.
532 THE JOURNAL OF ARACHNOLOGY
Humphrey, J.A.C, 1987. Fluid mechanical con- harmonic radar. Journal of Applied Ecology 36:
straints on spider ballooning. Oecologia73:469- 519-533.
477. Ramachandran, R. 1987, Terminal velocity of the
&
Jung, C. B.A. Croft. 2001. Aerial dispersal of first instar Ectropis excursia (Guenee) (Lepidop-
phytoseiid mites (Acari: Phytoseiidae): estimat- tera: Geometridae). Proceedings of the Indian
ing falling speed and dispersal distance of adult Academy of Sciences (Animal Science) 96:673-
females. Oikos 94:82-190. 678.
MacCook, H.C. 1877. The aeronautic flight of spi- Schneider, J.M., J. Roos, Y, Lubin & J.R. Henschel
ders. Proceedings ofthe Academy ofSciencesof 2001. Dispersal of Stegodyphus dumicola (Ara-
Philadelphia 1877:308-312. neae, Eresidae): they do balloon afterall! Journal
McCullagh, P. & J.A. Nelder. 1989. Generalized of Arachnology 29:114-116.
Linear Models, 2nd edition ed., Chapman and Suter, R.B. 1991. Ballooning in spiders: results of
wind tunnel experiments. Ethology, Ecology &
Hall.
McManus, M.L. & C.L. Mason, 1983, Determina- Evolution 3:13-25.
tion of the settling velocity and its significance Suter, R.B. 1992. Ballooning: data from spiders in
to larval dispersal of the gypsy moth (Lepidop- freefall indicate the importance ofposture. Jour-
tera: Lymantridae). Environmental Entomology nal of Arachnology 20:107-113.
12 pp 270-272. Suter, R.B. 1999. An aerial lottery: the physics of
ballooning in a chaotic atmosphere. Journal of
Mitchell, R.G. 1979, Dispersal of early instars of
Arachnology 27:281-293.
tthoemoDloouggilcaasl-fSiorciteutsysoocfkAmmoetrhi.cAann7a2:l2s9o1f-2t9h7e.En- Thomas, C.EG., P. Brain & PC. Jepson. 2003. Ae-
Okuma, C, & R, Kisimoto. 1981. Air borne spiders rial activity of linyphiid spiders: modelling dis-
persal distances from meteorology and behav-
collected overthe EastChinaSea, JapaneseJour-
iour. Journal of Applied Ecology 40:912-927.
nal of Applied Entomology and Zoology 25: Yoshimoto, C.M. & J.C. Gressitt. 1960. Trapping
296-298.
of air borne insects on ships in the Pacific (part
Osbloiranmse,,JJ..RL..,RiSl.eJy., CAl.aDr.k,SmRi.tJh.,MDo.rRr.is,ReyI.nHo.ldWsil&- 3). Pacific Insects 2:239-243.
A.S. Edwards. 1999. A landscape-scale study of Manuscript received 16November2004, revised10
bumble bee foraging range and constancy, using July 2005.