I ”VIM fl
. .III' IIIUUIIII I II'II'II ‘IIIIIII'
I}, IIIEI' I” [mum
"INII‘ "

  

J:

A
:m
—..._:_’—M

:II"'

II“; 3!
.I’I

“I"

-,
_—~

_‘::‘: =¢=
:.
a ‘z—gci‘ '22:: ,_ a
j— “M H
“‘-
%:_ g: :“QiN _
“a: 32*“:m
:W
:b'

~

)

3' III"); ""
II, ”III I'II'II: III’III’II I "If I
.‘ H'IIIII 'fI'II'III 'II'IIII III I II I“ ,‘IIh'llthI

 

III ,‘m’ I [113W
a? II IIIIIII
I. III'II .IIII.

I w

" I I 'I'I
I II I
IIIIII I ‘

II

“III I

'I'IIII'

III J'I'I‘IIII’

III'

[I

 
  
 

III "-IIII

I I III ‘HI'II III"'I'I'I'.I I‘I'I’IIII", 'I'I' III II .'I.IIIII III'I"_""'."I'II'IIII" II

II" I',I

«fir-*3;

III 'III'IIIII

III III

I'u

HIIII
I'll}! “I33 “H {1"

I'II‘IIIIIII'III’:
I I

  
 
 

I""II|(:|.‘
IIII‘IHI'I1

    
  
  
 

"My" " IL...” I

IIIII' IIII
I'III'II'I

  

I'I"|'I III
3'I'.'II" 33
I
II “I III
' ‘III IIIII I'III IIIIIIII I'IIIIII‘I

'I'III'I‘ ‘II' I‘ll IIIIIl "I IIIIIIIIIIII
'IIIIIIIII’I II"

II|'IIHII I‘|.IIIIII:"II. .IIIII
l I'l . U

"I
IMW‘I

        

III I

v ' I "I

I v (I: 'I'I'II' I

III! I”

III!" I'II'I‘II

I'IIII IM‘IIIIII II?“

 
 
  

   
   
  
 
  

      

 
  
  

      
  
 

  
    
 
 
   
  
    

 

 
 
 
   
     
  
 

312 1639

m uunuwwwmxmmmu111ml ‘13“ x y '

 

This is to certify that the

thesis entitled

FORAGING SPECIALIZATIONS IN ORB-WEAVING SPIDERS

presented by

Cader Wesley Olive

has been accepted towards fulfillment
of the requirements for

Meme in 2mg?

W/Gafl

Major professor

 

Dateozs/Afimfi/j

O-

FORAGING SPECIALIZATIONS IN ORB-WEAVING SPIDERS

By

Cader Wesley Olive

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Zoology

1979

ABSTRACT

FORAGING SPECIALIZATIONS IN ORB—WEAVING SPIDERS

By

Cader Wesley Olive

Previous studies indicate prey-specific attack behaviors in
orb-weaving spiders. This study explored functional relationships
between the morphology of contrasting orb—weaver species and their

prey-specific attack capabilities. Araneus trifolium and Argiope

 

trifasciata were observed attacking Acridid Orthoptera, Calyptrate

 

Diptera, and Lepidoptera in the laboratory. Diets and habitat use of
spiders in the field were determined by surveys. Distributions of
prey types were determined by trapping.

Araneus captured Orthoptera more quickly than did Argiope,
but Argiope was more successful in capturing this prey type. There
were no attack time differences between spider species on Dipterans
and Lepidopterans, but Araneus captured these types more successfully.
Attack times do not appear to limit prey intake rates in the field.
Spider morphology and web design appear to interact in causing the
above behavioral differences.

Araneus tends to build its web high in herbaceous vegetation.

Argiope builds in low sites in both grassy and herbaceous vegetation.

Cader Wesley Olive

Prey distributions are such that each spider is found most often in
sites where its more successfully captured prey type is relatively

more abundant. Predation and physiological stress may also select

for these differences in habitat use.

If these results are extended to other species, the following
predictions arise. Spiders with relatively short, stout legs, large
fangs, and high, simple, open—meshed webs (such as Araneus) should
specialize on innocuous, rapidly escaping prey types (Diptera, Lepi—‘
doptera). Spiders with relatively long legs, small fangs and low,
dense—meshed, ornamented webs should specialize on dangerous, slowly
escaping insects (Orthoptera, Homoptera, Hymenoptera).

Morphological, web design, and habitat use data collected from
a field survey in southern Michigan on 8 common old—field orb-weaving
species were analyzed for patterns of prey specialization using factor
analysis, discriminant analysis, and a model predicting net energy
return rates for spiders of a given size and morphology in different
habitats and seasons. This model was based on the data from the
previous 2-species comparison.

The original morphological and behavioral dichotomy contrasting
Araneus and Argiope did not adequately describe patterns for all the

species. The marked exception was Gea heptagon, which combined long

 

legs, short fangs, and a low, open-meshed web. A model allowing fang
and leg length to sort independently is necessary to describe this
suite of characters. The prey specialization indicated for this com—

bination by the previous study would be one for weak, slowly escaping

Cader Wesley Olive

insects. Data on habitat use and diet was generally insufficient
to test the original hypotheses for these species but predictions

are made for each individual species.

ACKNOWLEDGEMENTS

This research was supported by an NSF pre—doctoral fellowship,
NSF Grant DEB 78-11202, and the Zoology Department of Michigan State
University. Donald Beaver, Bill Cooper, Earl Werner, Gary Mittelbach,
Don Hall, Peter Witt, Allen Brady, Leni Wilsmann, Judy Soule. Kay
Gross, and Jim Gilliam all supplied advice and encouragement through»

out research and writing. My wife, Sara, put up with my dreams.

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

INTRODUCTION

CHAPTER 1: MECHANISMS OF FORAGING SPECIALIZATION IN ORBrWEAVERS

Prey Capture Behavior in Orb—Weavers
Methods
Prey attack proficiency
Web site characteristics
Diet of the spiders in the field
Prey availability
Results
Capture proficiency
Prey escape
Web sites and design
Diet
Prey availability
Spider phenologies
Discussion
Foraging strategy differences between Araneus

trifolium and Argiope trifasciata

 

CHAPTER II: PATTERNS OF FORAGING SPECIALIZATION IN A GUILD OF
OLD—FIELD ORB-WEAVERS

Mechanisms of Attack Differences

ii

page

iv

vi

11

12

12

13

13

19

20

23

26

33

36

36

43

43

Patterns of Specialization

Optimal Phenology and Habitat Use

Foraging Strategies of Individual Species
Assumptions and Qualifications of These Predictions
Evolutionary Considerations

LITERATURE CITED

iii

47

55

69

74

75

77

LIST OF TABLES

Table page

Spider species effects on attack times after correcting for

prey size and spider cephalothorax width ..... ......... ....... 14
Frequency of initial attack behaviors on 3 different prey

types for Argiope and Araneus ................. . ........ ..... l6
Capture success differences between Argiope and Araneus on

three different prey types. Relative prey length ranges

were selected for similarity of sample frequency distribution
between the two spider species, using Pi'.19 in a Kolmogorov-
Smirnov 2-sample test as the criterion ...................... 18
Mean web heights in cm (:_s.e.) for Araneus and Argiope in

early and late summer, 1978 .................................. 21
Relative frequencies of jumping vs. flying insects in the

diets of Araneus and Argiope before August 8, 1978 ..... ..... 25
Accumulation rates of numbers and biomass of prey by Argiope

and Araneus webs in herbaceous habitat in late summer, 1978 .. 29
Summary of characteristics of 8 old-field orb-weaver species

from Michigan ................................................. 49
Varimax rotated factor matrix after rotation with Kaiser
normalization for morphological and web variables from 8

Old-fiEId orb-weaver SpECiES oooooooooooooeooooooooo0000000000 51

iv

Table

9.

10.

11.

12.

Standardized discriminant function co—efficients for Spring

vs. Fall phenologies and hub vs. retreat behaviors in 8 old-
field orb-weaver species ...................................
Regression parameters for the equation: handling time =

b1 + b2 (relative prey length)b3 , for different prey types.
Handling time is in hundredths of minutes...................

Regression parameters for capture success by Argiope and

Araneus on different prey types. Regression equation is:

arcsin z capture success = a + b(relative prey length).
Parameters designated by asterisks were used in the model...
Modifications fo capture success vs. relative prey length
regressions used for various spider vs. prey combinations
in the net energy return model. Where indicated, the esti-

mate of the y-intercept was changed by 15%..................

page

56

59

6O

62

LIST OF FIGURES

Figure page

Schematic comparison of the web design and morphology of

Araneus trifolium and Argiopg trifasciata. Morphological

 

 

comparisons are between individuals of the same cephalo—

thorax width ............................................... 8
Prey-size distribution of Argiope and Araneus diets in herb-
aceous vegetation after August 8, 1978. Abscissas are body

length of prey in mm observed entering webs. Stippled areas

are captured insects, un-stippled are total insects entering

webs ........................................................ 27
Biomass in mg dry weight vs. insect body length in mm for

window pane trap samples from Summer of 1977. In each row

from left to right graphs represent sites that were: (60%:

40%), (30%:7OZ), (1%:99Z), and (1%:99Z) herbaceouszgrassy
vegetation. Each column represents different heights, insect
types, and dates for the same site. Jumpers include Homop-

tera and Orthoptera. Fliers are Hymenoptera, Diptera, Lepi-
doptera, and Coleoptera ...................................... 31

Cephalothorax width in mm vs. day of year collected for

 

 

Araneus trifolium and Argiope trifasciata for 1977-1978.
Curves represent estimated cohort boundaries. Horizontal lines

bound mature female size ...................................... 35

vi

Figure page

5.

Population distributions in factor space. Ordinate is factor

one score, abscissa factor two score. Centroids are bound by

95% confidence ellipses. X's are low scores, O's high scores,

and dots intermediate scores on factor three (web area). Size

of type represents typical relative size of mature females.
Underlined species are spring maturing, others mature in fall.
Quadrants are labeled with variables that are highly correlated
with factors one and two ....................................... 54
Per cent capture success vs. relative prey length for Argiope

(x's) and Araneus (dots) on three prey types. Regression lines

are from Tables 11 and 12. Sample sizes for points are small

(1-5) at extreme relative prey lengths and large (15-25) for

the intermediate range ......................................... 64
Net energy return rates (kcal) in different habitats based on
"wrapper" (dotted lines) and "biter" (solid lines) models. Dates
are May 31, July 19, August 30, and September 28, 1977. Results
for spiders of four different cephalothorax widths are shown.... 66

Cephalothorax width vs. day of year collected for Argiope

aurantia (x's) and Argiope trifasciata (dots) for 1977.......... 72

 

vii

INTRODUCTION

From studies of animal foraging behavior, ecologists hope to predict
environmental factors determining changes in diet over ecological time
and the effects of dietary changes on population density and
distribution, and thus on community structure. The general nature of
these dietary responses are currently being explored under the principle
of optimization or maximization of energy or nutrient intake. However,
the specific behavioral response of a consumer will be determined by the
interaction of its own morphology and phenology with a variety of complex
factors. These include not only capture efficiency and availability of
various prey types, but habitat structure, predators, competitors,
physical conditions, and the interactions between them.

This study explored the basic morphological and life history
constraints on feeding behaviors of two orb-weaving spiders: Araneus

trifolium (Hentz) and Argiope trifasciata (Forskal). The comparative

 

approach of the study exposed ranges of behavior and their possible
associations with the factors mentioned above. I chose these two species
for contrasting morphological characters which might logically affect
their relative abilities to subdue certain prey types. I also examined
selected life history characters to explore relationships between
morphological constraints on prey capture proficiency and behavior
influencing encounter rates with different prey types. For brevity in
further discussion, I will refer to relationships of this type which lead

to increased overall capture efficiency as specialization on the prey

2

type in question. Using the patterns of specialization, indicated by
this two-species comparison, I then examined the morphology, behavior and

life-histories of six other orb-weaver species for similar patterns.

Chapter I

MECHANISMS OF FORAGING SPECIALIZATION IN ORB-WEAVERS

Prey Capture Behavior in Orb-Weavers
The following background will clarify the reasons for choosing

Argiope trifasciata and Araneus trifolium. Several studies of

 

 

orb-weavers suggest possible modes of specialization. Robinson's series
of studies (Robinson, 1975; Robinson and Mirick, 1971; Robinson, §£_§l.,
1969; Robinson and Olazarri, 1971) show a set of prey-specific attack
behaviors in orb-weavers. To summarize, the two basic characteristics
distinguishing functional prey types are speed of escape from the web and
ability to injure the spider. The simplest measurable characters that
best predict these capabilities of the prey are taxonomic order and size
of the prey relative to the spider. Rapidly escaping types include all
Lepidoptera, Diptera, and Hymenoptera strong enough to fly free of the
web and any other insect large enough to rip through the web by its own
weight. The more dangerous prey include Orthopterans, Homopterans, and
some Coleopterans of sufficiently large size to damage the spider by
kicking or biting, as well as stinging Hymenoptera and noxious Coleoptera
and Hemiptera.

Robinson (1975) discusses three basic orb-weaver attack modes. In

"pluck-out, the spider simply plucks the prey from the web with its
chelicerae. The prey is subdued immediately and the web only serves to

detain the prey until the spider can reach it. In "bite-attack," the

4
spider grasps the prey with its legs and makes a venomous bite. Until
the prey is sufficiently paralyzed by the venom, the web helps to subdue
the prey and the spider does not attempt to remove it from the web.
After the venom takes effect, the spider carries the prey to hub or
retreat for feeding. In "wrap attack," the spider throws silk over the
prey while it is entangled in the web. When the prey's movement is
sufficiently restricted, the spider makes a short venomous bite. Then,
either immediately or after the venom takes effect, the spider carries
the prey to hub or retreat for feeding.

Pluck-out attacks are restricted to relatively small, innocuous prey
less than 1/20 the weight of the spider. Bite attacks are consistently
used on relatively large Lepidoptera and Diptera. Wrap atacks are used
on relatively large Orthoptera, Homoptera, Coleoptera, stinging
Hymenoptera, and all sizes of noxious insects. Robinson's explanation of
this prey type specificity in attack behavior is simple. Lepidoptera
shed their scales into the adhesive of the orb-web and escape rapidly
(Eisner, E£.Elf’ 1964) unless attacked. Strong flying Diptera (i.e.,
most Brachycera and Schizophora) appear to vibrate themselves free of the
adhesive (Robinson, 1975). If these prey are not quickly subdued they
will escape. The hold and bite attack either quickly subdues the prey or
the prey escapes quickly. The wrap attack takes more time to subdue
these prey types.

Weak flyers and non-flyers of sufficiently small size are retained
by the web for a longer time. Insects with long, spiny appendages that
become entangled in many threads are especially well detained. The
resilient silk of the sticky spiral threads is well-designed to retain
insects without breaking yet not afford a resistance against which to

crawl (Denny, 1976). The wrap-attack allows orb-weavers to subdue these

5

prey while remaining at a distance determined by the length of the
posterior, wrapping legs. These legs have spines like crochet hooks
(sustentacula) that aid in throwing the silk. The wrapping distance plus
the spider's ability to choose the range and direction of attack often
enable the spider to avoid kicking legs, biting mandibles, and stings.
For these prey types, the wrap attack is faster and more effective than a
bite attack of equivalent safety (Robinson, 1975).

Although these previous studies make clear the consistency,
distinctness, and universality of these prey-specfic attack modes, no
quantitative tests have been made of capture success and efficiency
differences between morphologically distinct orb-weaver species. The
legs and biting mouthparts of the spiders play distinctly different roles
in the different attack modes, and prey types fall into taxonomic and
ecological groups that might lead to their spatial and temporal
segregation. These distinctions suggested two possible patterns of prey
specialization. I will call the first pattern the "biter" for
simplicity. Such a specialist would have stout, powerful legs and large
chelicerae for achieving a quick, effective hold on even large, rapidly
escaping flyers. In contrast, the "wrapper" would have longer legs to
maintain a greater distance from the prey and more dense sustentacula to
cast more silk per throw.

From previous field surveys in old-fields near East Lansing,
Michigan, I chose two orb-weaver species that fit the morphological types
above but were otherwise similar. Both species live in old-fields, hatch
in the spring, mature in mid-summer, and lay eggs in the fall. They are
among the largest (10-20 mm body length for mature females) and most
abundant orb-weavers in this area. In all discussions below, spider body

size will mean the greatest transverse width of the cephalothorax. I use

6

this parameter because: 1) unlike weight or total body length it is
constant within an instar, 2) all leg and mouthpart dimensions are
strongly correlated with it (r > .99), 3) the cephalothorax contains the
muscles for the basal portions of legs and chelicerae, and 4) it is
easily and accurately measured. The body weight can vary ten-fold
between individuals of the same size, though Argiope tends to be heavier
for a given size. The fourth (wrapping) legs of Argiope are 50% longer
than those of Araneus of the same size while the mouthparts (paturon and
fang) of Araneus are 25-33% longer and thicker than Argiope's. These
values are the ratios of estimated slopes of regressions of the parts
against cephalothorax width. F-tests for differences in regression
co-efficients are significant at P < .01. Araneus' legs are much thicker
for their length than those of Argiope. Mean sustentacula density on
tarsus IV of Argiope is three times greater than for Araneus (Mann
Whitney U test, p << .001). See Figure 1.

Given these morphological differences, I hypothesized that Araneus
trifolium individuals would capture a larger percentage of rapidly
escaping, flying prey introduced into their webs, with shorter attack

times and fewer injuries than would individuals of Argiope trifasciata.

 

Conversely, I hypothesized that Argiope would be more proficient (using
the same criteria) than would Araneus on slowly escaping, more dangerous
prey. I also hypothesized that differences in habitat use, web
placement, and web design might exist which caused the prey types they

captured more proficiently to enter their webs relatively more often.

Figure 1. Schematic comparison of the web design and morphology

of Araneus trifolium and Argiope trifasciata. Morph-

 

ological comparisons are between individuals of the

same cephalothorax width.

Araneus trifolium rg_9_p_e_ trifosciatg

£8 RETREAT 4, 5’,

r
1 —-r
- '- ‘13 I Q.
' ‘20???" x-
\c / hiss/[:3 9“ «I

 

 

 

 

   

/ ‘fli"\.' "
v Marga , -
l

SPIDER
‘SIGNAL
THREAD

. . BARRIER
(aide yiew) WEB

  

sustentacula

density %

 

Methods

Prey Attack Proficiency

I conducted behavioral studies in the laboratory to control feeding
history of spiders and the entry of prey items. Spiders lived in
individual cages similar to those described by Witt (1971). Spiders
could build normal webs and were easily accessible for maintenance and
experimental feeding. Cages were ventilated with outdoor air at night to
stimulate daily web renewal (Witt, 1963). Maintenance diets consisted of
a variety of insects collected from local old-fields. Food intake rates
could not be strictly controlled due to erratic web-building and prey
utilization efficiency (Haynes and Sisojevic, 1966), but spiders received
about their own weight in prey daily.» Spiders drank daily from webs
watered with an atomizer. The experimental period lasted from July 2 to
September 24, 1978. Many individuals survived the entire period, and
some produced eggs. Spiders renewed webs every one or two days, except
for longer periods during molting and reproduction.

Prey types were grasshoppers (Acrididae), adult Lepidoptera, and
adult flies (Calyptratae) collected from local old-fields, at lights, and
at bait-traps. All sizes of prey were used. Insects were kept at 8°C
until used but never longer than two days. These prey types were 1)
available to the spiders naturally as prey, 2) easy to catch and handle,
and 3) consistent in behavior and morphology within types.

For reasons dictated by statistical design I tried to maximize the
number of different prey and spider size combinations. Thus, prey to
spider assignments were not random, but depended on previous size
combinations as well as on insects available on that particular day. I

used only fresh webs for tests. Each spider received only one capture

10

test on a given day, though I used the same spiders repeatedly throughout
the summer. I measured body lengths of insects (tip of head to tip of
abdomen) with vernier calipers to the nearest 0.1 mm while the insects
were still cold and inactive. While insects warmed to room temperature
(25° - 28°C) I placed spiders with fresh webs in a row on a table and
removed the cage fronts. I recorded vertical and horizontal diameters,
number of radii, mean number of spirals per mm in the vertical segment of
the bottom half of the web, and spider location for each web. Spiders
were left undisturbed for 30 minutes after web measurements.

I released insects into the center of the lower half of the web from
a distance of a few centimeters. A light behind the web induced the prey
to enter the web. If grasshoppers did not jump into webs I introduced
them by hand. Prey entry into the web marked time zero. I recorded
starting and ending time to within 0.01 minute on a Datamyte
(Electro-General Corp.) for orienting toward the prey and plucking
(Robinson and Olazzari, 1971), approaching the prey, biting, wrapping,
plucking-out, simultaneous biting and wrapping, carrying prey to hub or
retreat, feeding, or escape. Following each bout, the cephalothorax
width of the spider was measured to the nearest 0.1 mm with vernier
calipers. Representative sequences were filmed with a Canon 514 XL Super
8 mm Movie Camera using two 500 watt floodlights for illumination.

Total attack time will be defined as the time from first attack to
the beginning of the carry to hub or retreat (Robinson, 1975). These
behaviors are consistently and clearly recognizable and mark the period
during which the spider actually subdues the prey. Robinson (1975)
asserts that selection for a rapid return to the hub (or retreat) to
continue monitoring for prey or predators may be strong. This would

account for the consistent sequence of these behaviors. In contrast, the

ll

behaviors leading to feeding vary greatly in length and sequence. Also,
venomous, enzyme injection, and feeding bites are indistinguishable.
Total time biting and wrapping were summed separately for each bout.

Total time to escape is from time zero to escape.

Web Site Characteristics

I conducted surveys one morning a week from July 13 to September 29,
1978, in old fields near East Lansing, Michigan. A variety of herbaceous
and grassy vegetation was included in each survey. Patches of Timothy,
foxtail, and quack grass and several species of Solidago dominated most
sites, with some dominated by sedges. .Astgr spp., Erigeron spp., Daucus

carota, Potentilla spp., Hieracium spp., Fragraria spp., Rubus spp., and

 

 

Cornus spp. were common. From a chosen starting point I ran one transect
in a random direction 40-60 meters for each survey. Survey times were
7:00 to 10:00 A.M. when webs were fresh, dew-covered, and easily
observed. Both spider species rebuild webs at dawn. For every web in a
one meter wide swath I recorded spider species, time of day, height of
hub above ground, vertical and horizontal diameter, mean spirals per
millimeter in the vertical segment of the lower half of the web, genus of
all plants to which the web was attached, percent cover of different
vegetation types within two meters, and openness of the web site.
Openness was a subjective estimate of the relative proximity and density
of vegetation in the cylinder projected by the web in each direction for
two meters. The measure ranged from 0 (completely closed) to 5
(completely open). I used four different 4-10 acre old fields for the

surveys throughout the summer.

12

Diet of the Spiders in the Field

To eliminate biases against rapidly processed prey, I watched webs
for entering insects continuously for 1-4 hour periods instead of
recording prey already present in webs as I encountered them during a
survey. Observation periods were chosen between 11:00 A.M. and 5:00 P.M.
(DST) on days without excessive (> 20 mph) wind or rain since initial
observations showed these to be peak insect activity times. Strictly
crepuscular insects may be under-represented as a result, but this
protocol was necessary to achieve minimal sample sizes. Observations of
both spider species indicated that spiders were inactive below 10°C, webs
were often eaten (Peakall, 1971) shortly after sunset, and many webs were
destroyed by insects and wind before sunset. Choice of sites where the
most webs were observable was also necessary to generate a sufficient
sample size. From 1-8 webs were observed simultaneously from a minimum
distance of one meter, but all webs were close enough to observe the
entry of 2 mm long Diptera (no more than 2.5 meters). For each insect I
recorded time of entry (before or during observation period), spider's
response, its fate by the end of the observation period, estimated body
length to the nearest mm, taxonomic order and (if possible) superfamily
or family. All of the above spider and web measurements as well as
length of observation period were recorded for each web. Observations
were made from June 28 to September 24, 1978, in the same fields surveyed
for web site characteristics. The webs in this sample were in a variety

of vegetation types.

Prey Availability
I collected insect samples from May to October, 1977, in old-field

sites. Traps were open one-gallon cans with a 15 x 15 cm pane of glass

l3

protruding vertically from the top and contained one liter of ethylene
glycol as a killing agent and preservative. Each sample collected
represents one week of continuous trap-exposure. Traps were in four
sites ranging from (60% herbaceous: 40% grass) to 100% grass cover. At
each site one trap was on the ground and was one meter above the ground.
The high traps were at the top of surrounding vegetation. These traps
mimic spider webs only crudely. Smaller insects blow around, large
insects may escape, some crawl in, and the liquid may attract ovipositing
and drinking insects. They do catch larger insects than sticky traps and

reflect activity better than sweeping (See Kajak, 1965).

Results

Capture Proficiency

I used analysis of covariance to examine effects of prey length,
spider size, and spider species on the total attack time for captures of
each prey type (Table 1). In verifying that assumptions were met, I
found that prey length cubed was better correlated with attack time than
prey length for grasshoppers and Lepidoptera. Prey size and spider size
had strong independent effects (positive and negative, respectively) for
all prey types. There were no significant effects due to spider species
for Lepidoptera and Diptera. On grasshoppers, Araneus' times were
shorter than Argiope's (P < .011). The two spiders spent similar amounts
of time biting and wrapping. Most of the difference was in idle time
near the prey or at the hub. Araneus spiders typically wrapped steadily,
bit while the grasshopper was relatively loosely wrapped, and immediately
hauled it to the retreat. In contrast, Argiope spiders typically wrapped

sporadically, backing away from the prey in interludes, bit only after

14

Table 1. Spider species effects on attack times after correcting for

prey size and spider cephalothorax width.

 

 

Prey Type Factor DF MS F P

 

Orthoptera (prey length)3 1 6,749,301 7.99 .006
cephal. width 1 7,918,440 9.38 .003
spider species 1 5,664,307 6.71 .011

residual 87 844,021

Lepidoptera (prey length)3 1 17,808,754 141.9 .001

cephal. width 1 2,629,637 21.0 .001
spider species 1 323,276 2.6 .116
residual 43 125,495

Diptera (prey length)3 1 761,350 6.39 .013
cephal. width 1 1,153,715 9.67 .003
spider species 1 20,021 0.17 .683

residual 90 119,234

15

the prey was relatively securely bound (often after several aborted bite
attempts followed by more wrapping), and might return to the hub for a
period before carrying the prey. Argiope's attack times on grasshoppers
were also much more variable and showed much less correlation with prey
length than those for Araneus. One consistent behavior not previously
recorded was the laying of several swaths of silk below large
grasshoppers in their escape path before the actual attack began.

Usually the spider directed its attack from a direction out of the range
of movement of kicking legs. Initial bites were usually to the thorax or
the femurs of the third pair of legs.

One striking qualitative difference between spider species was
evident from observing bouts and analysis of films of bouts. In
attacking Lepidoptera Argiope was frequently observed to pursue
relatively large insects around the web, repeatedly attempting to close
and bite. In Araneus webs the same relative size prey was usually well
retained, and if the spider reached the prey it closed quickly and bit in
the thorax or wing base. In several bouts, mature female Araneus spiders

would poise above large Monarch butterflies (Danaus plexippus) and

 

suddenly drop onto the thorax when the flapping wings were spread.
Patterns of initial mode of attack followed those noted by Robinson
(1975), as shown in Table 2. A significant discrepancy is the relatively
frequent wrap attack on Diptera by Argiope spiders. The relative size
(hereafter: prey length/spider cephalothorax width) of these wrapped
flies was significantly greater than that of flies attacked by biting
(Mann-Witney U test, P < .01). The Araneus sample of attacked Diptera
included even larger relative size prey, and almost all of these were

attacked with a bite.

16

Table 2. Frequency of initial attack behaviors on 3 different prey

types for Argiope and Araneus.

 

 

 

Number of Number of
Spider species Prey type bite attacks wrap attacks
Orthoptera O 57
Araneus Lepidoptera 32 l
Diptera 39 4
Orthoptera 2 45
Argiope Lepidoptera 16 2

Diptera 20 12

17

Capture success is the number of captures relative to the number of
that prey type introduced into webs. Due to experimental design for
analysis of covariance and the restrictions imposed by concurrent
availabilities of prey and spider sizes, samples of different relative
prey sizes were insufficient to use a 4-way chi square test (capture vs.
escape, spider species, spider size, prey size). Consequently I selected
sub-samples of the same range of relative prey sizes for each prey type
using two criteria. First, a Kolmogorov-Smirnov test for differences in
cumulative frequency distributions of relative prey sizes from the two
spider species sub-samples had to show lack of statistical significance
(P > .2). For example, the sub-sample of grasshoppers introduced into
Araneus webs could not contain relatively more large or small prey than
the same sub- sample from Argiope and grasshoppers. This corrected for
relative prey size effects in the test. Second, each cell of the
chi-square tests had to contain at least five observations. Using these
criteria, the relative prey size ranges in Table l were selected for each
prey type. These relative prey size ranges include the bulk of the total
samples. An assumption of this use of relative prey size is that prey
and spider size effects on capture success are fairly linear and do not
interact strongly. The 2 x 2 Chi-square tests using these sub-samples
(Table 3) show Araneus catches relatively more Diptera and Lepidoptera
than Argiope does (P < .02 and P < .056, respectively) but that Argiope
catches relatively more Orthoptera (P < .013). Samples are too small for
tests with other relative prey size ranges, but for all prey the same
trend remains for larger relative prey sizes and becomes less distinct
for smaller sizes.

No detectable injuries were sustained by any spiders. Injuries to

spider's legs are usually obvious since the leg is either inactivated or

18

 

 

 

Table 3. Capture success differences between Argiope and Araneus on
three different prey types. Relative prey length ranges
were selected for similarity of sample frequency distribu—
tion between the two spider species, using P > .19 in a
Kolmogorov-Smirnov 2-sample test as the criterion.

Relative
prey length P
range of Captures/escapes Captures/escapes of

Prey type sample for Araneus for Argiope )CZ

Acridid

Orthoptera 5.0 to 8.0 17/43 29/26 .013

Lepidoptera 2.25 to 5.75 22/46 11/55 .056

Calyptrate

Diptera 1.0 to 4.0 47/29 35/48 .02

19

removed by the spider (Gertsch, 1949). On a few occasions grasshoppers

placed solid kicks which dislodged spiders from webs.

Prey Escape

Grasshopper escapes and spider attacks are extremely variable in
duration. Often the spider would not attack for several minutes, during
which time the grasshopper would remain motionless in the web. Usually
the grasshopper began struggling at the approach of the spider. This
variability makes comparison of escape times difficult, and the data do
not indicate significant differences in escape time between spider
species.

Diptera escapes were much more rapid and uniform, most insects
flying free in less than a second and the longest unattacked detention
being about four seconds. For the same relative size range of flies, the
longest detention on an Argiope web was less than two seconds, while
several flies were held in Araneus webs for three or four seconds.
Overall, Argiope detentions were significantly shorter (Mann-Whitney U
test, P < .03).

The results for Lepidoptera are more complex. Within a given
relative size range the smaller Lepidoptera escaped much more frequently
and more rapidly. These smaller Lepidoptera were mostly moths. The
larger insects were butterflies. Many of the moths appeared to have much
thicker coatings of scales. Escape times for most of these moths were
less than one second, with a few ranging to four seconds. The larger
butterflies were detained for as long as two minutes. Capture success
for the moths was less than 20%, but that of butterflies was 60% for
Araneus. Even within the same relative size range of moths there was a

strong trend towards increased capture success for larger prey and

20

spiders. These trends were similar for both spider species, and escape

times did not differ significantly between spider species.

Web Sites and Design

I will now examine web designs and habitat use of these two species
in light of the habitat distribution of the prey types discussed above.
Though individuals of both spider species may occur in similar vegetation
at the same heights, there are consistent differences in overall web site
characteristics. Araneus webs tend to be higher than Argiope's
(Mann-Witney U test, P < .002). As shown in Table 4, these height
differences are about 20 cm. Several rather distinct changes in web
placement and diet occurred at the first week in August. Before August
8, all immature Araneus spiders built webs between two leaves of single
Solidago or éfiEEE plants near the top, and webs were less than 10-15 cm
in diameter. During the same period Argiope spiders built in both solid
grass stands and herbaceous vegetation, with webs from ground level to
about 1/2 - 2/3 the height of the vegetation.

After August 8, Araneus webs increased in size and began to span
wider gaps between plants. One bridge thread spanned a 2 meter free gap.
Araneus spiders continued to build in herbaceous and broad sedge
vegetation. Only one individual of 150 built a web with most attachments
to grass spikes, but even then had its retreat in the head of a lone
thistle plant. Due either to differential survival or selective
individual movement the distribution of Argiope shifted to herbaceous
vegetation in the latter part of the summer. The overall result was a
convergence between the two species in habitat use patterns. In the
sites surveyed there were no Argiope spiders in extensive stands of solid

grass during this period. There were Argiope spiders in August and

21

Table 4. Mean web heights in cm (:_s.e.) for Araneus and Argiope in

early and late summer, 1978.

 

 

Spider species Before August 8, 1978 After August 8, 1978

 

Araneus 60 + 7.2 81.8 + 4.6

Argiope 42.4 :_ 3.5 59.2 + 3.68

22

September, 1978 with webs entirely attached to grass, but these were
small clumps of grass interspersed with herbaceous vegetation. This
distribution is in contrast to the one observed in similar surveys in the
same fields the previous year. During the summer of 1977, Argiope
spiders occurred in grassy fields in densities as high as 0.5

spiders/m2. In the same areas in 1978, the highest densities observed
were one spider per hectare, and these individuals were near the margins
of the grassy areas.

The web sites of the two species also differed in relative openness.
Again, despite considerable overlap, Araneus sites tended to be more open
to the rear (Mann Whitney U test, P < .01), though there was no
significant difference in front openness. The rear side is the one
tipped towards the ground, on which Argiope rests and both spiders move.
This openness difference is reflected in Argiope's shorter and more
numerous web attachments to plants, especially those of the rear barrier
web (Figure 1).

There are several consistent, qualitative differences in web design
between the two spider species (Figure l). Argiope spins a dense mesh on
the hub, where it waits for prey. Dense zig-zag swaths of silk
(stabilimenta) are often vertically above and below the hub (Robinson and
Robinson, 1970). Irregular barrier webs are often attached several
centimeters from the same side of the web the spider rests on, and less
frequently on both sides of the web (Tolbert, 1975). In contrast, all
Araneus webs are simple and empty. The spider waits for prey in a retreat
of rolled leaves several centimeters from the edge of the web (usually
near an upper corner) and monitors the web via a signal thread attached to
the hub and held taut by one leg (Gertsch, 1949). The hub consists of a
few threads and there are no accessory webs or adornments. These web

design differences are invariable.

23

Web design also differs quantitatively, with some overlap. In both
species, web diameter and distance betweeen spiral threads increase with
spider size. For a given size spider, Araneus webs are smaller with
larger spaces between spirals (less dense spirals). Individual spiral
threads of Araneus appeared to be thicker, though difficulties of
measuring thread thickness and strength precluded quantification in this
study (see Witt, 1963; and Denny, 1976, for problems). Finally, Argiope
webs have more radii (X = 35.4, 95% C.I. 34.3 to 36.5) than Araneus webs

(X = 20, 95% C.I. 19.33 to 20.71).

Diet

Eighty hours were spent watching 6O webs, for a total of 200
individual web-hours. Observation periods were usually for two hours,
but varied from 0.5 to 4.6 hours. For both spider species, samples
consisting of prey already in the web at the start of observation were
biased toward very small (< 3 mm) and very large (> 20 mm) prey relative
to those prey entering during the period. Spiders ignored most prey less
than 3 mm when they entered. In general, prey less than 3 mm were
Diptera and Hymenoptera, 3-6 mm prey were Homoptera, 7-8 mm prey were
mostly Calyptrate Diptera plus a few Coleoptera and Hymenoptera. Prey
that were 10-16 mm were almost all flying pollinators: Hymenoptera
(Apinae and Vespinae) and Cantharid beetles. Prey greater than 20 mm
were all Acridid Orthoptera.

Diptera were by far the most numerous prey captured, followed by
Hymenoptera, Coleoptera, and Homoptera, with very few Orthoptera,
Lepidoptera, or Hemiptera. The size-frequency distribution of prey
(either entering or especially those already in the web) is strongly

truncated or log normal. However, in terms of biomass the numerous

24

Diptera and Hymenoptera less than 3 mm contribute less than 15% even to
diets of small spiders unable to catch prey larger than 7 mm, and are
insignificant in the diet of larger spiders. Dry weight biomass for all
prey was computed from lengths using an exponential relation derived
empirically from a sample of Brachyceran and Schizoceran Diptera (Beaver
and Baldwin, 1975). Although body shape is a factor, a large part of the
weight difference between Diptera and more armoured insects is in
indigestible exoskeleton. Using this estimate of available biomass, the
exponential decrease in abundance of insects with increasing length
appears to be roughly balanced by exponentially increasing biomass at
about 8-9 mm body length. That is, for prey smaller than 8-9 mm, biomass
contributed to the diet increases with prey length. For insects larger
than this, biomass contributed by increasing size classes remains the
same or increases slightly. This is a very crude estimate, averaged over
several months and a variety of habitats and web sites. It does indicate
that large insects contribute substantially to the diet despite their
rarity. For spiders after August 8 (> 2 mm cephalothorax width) prey
less than 7 mm contribute less than 20% to total biomass entering webs.
This contribution is even less if larger Orthopteran prey already in webs
were included.

A distinct change in diet difference occurred from early to late
summer, paralleling the change in habitat differences. Before August 8,
relatively more Homoptera and Orthoptera entered and were captured in
Argiope webs than flying insects (Diptera, Hymenoptera, Coleoptera, and
Lepidoptera) as compared to Araneus webs (Chi-square, P < .005, see Table
5). During this early period Argiope webs were in grassy and herbaceous
sites. After August 8, when web site characteristics of the two species

converged, dietary differences diminished. In terms of size-frequency

Table 5.

25

Relative frequencies of jumping vs. flying insects in the

diets of Araneus and Argiope before August 8, 1978.

 

 

Insects

Captured

Total

Insects

 

Spider Jumping Flying Chi-
Species Insects Insects square P
Araneus 2 13

8.99 .005
Argiope 16 10
Araneus 6 27

8.07 .005

Argiope 17 16

26

distributions and prey types the diets were remarkably similar (see
Figure 2). The major difference in prey type was that in the

8-16 mm range Argiope took relatively more Hymenoptera and Coleoptera
while Araneus took more Diptera. The major size-frequency difference was
that Argiope trapped more flying insects smaller than 3 mm.

If the two web types are identical passive filters then insect entry
rate should be directly proportional to web area and hours exposed. The
sum for all webs of (area x hrs) is three times greater for Argiope than
Araneus in this sample, due to more and larger Argiope webs. Despite
this 3-fold exposure difference, total biomass of insects accumulated by
webs is almost equal for the two spider species (see Table 6).
(Differences in number are due to more small, ignored prey less than 3 mm
trapped by Argiope webs). These differences are the same for all insects
entering webs as well as insects captured. This indicates that Araneus
webs accumulate prey more rapidly under these conditions. Time of day
observed, vegetation type and weather during observations were very

similar for both samples.

Prey Availability

I analyzed trap samples for weeks ending May 31, July 19, August 30,
and September 28, 1977, for numbers, lengths, and taxonomic orders of
insects. Only prey types included in the 1978 spider web samples were
considered. I calculated available biomass as for web samples. Total
dry weight for each prey size class is shown in Figure 3. These data do
not indicate population levels as much as activity levels determined by
the particular sites, weather conditions, and trap types used. Given
this, two striking trends are apparent. First, for both flying

(Coleoptera, Diptera, Hymenoptera, Lepidoptera) and jumping (Orthoptera,

Figure 2.

27

Prey—size distribution of Argiope and Araneus diets
in herbaceous vegetation after August 8, 1978.
Abscissas are body length of prey in mm observed
entering webs. Stippled areas are captured insects,

un—stippled are total insects entering webs.

28

 

Araneus

11! ‘ a I. o
O O 0 A o I
_ bakery... «a. .

 

 

 

 

 

:—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

~..qfi.-d1.4
o I

 

 

20 '1 Araneus

 

 

 

 

 

 

29

Table 6. Accumulation rates of numbers and biomass of prey by

Argiope and Araneus webs in herbaceous habitat in late

summer, 1978.

 

 

number of webs observed

total web—hours observed

total web area x hours observed

total no. of insects entering webs

total mg dry weight of insects
entering webs

total number of insects captured

total mg dry weight of insects captured

 

Argiope Araneus
36 20

61.5 37.2
30,602 10,898
53 42

73 83

40 24

45.4 46.5

 

 

30

Figure 3. Biomass in mg dry weight vs. insect body length in mm
for window—pane trap samples from Summer of 1977. In
each row from left to right graphs represent sites
that were: (60% : 40%), (30% : 70%), (1% : 99%), and
(I%I: 99%) herbaceous : grassy vegetation. Each col-
umn represents different heights, insect types, and
dates for the same site. Jumpers include Homoptera
and Orthoptera. Fliers are Hymenoptera, Diptera,

Lepidoptera, and Coleoptera.

 

31

MAY
FLIERS
H... L- L

T Y

fié‘wiflfiggL—Lmjfi

 

 

 

 

 

 

 

 

 

 

MAY

1011195125104. . ‘ 'fiL r . - L L

HIGH ' ' ' ' I ' ' ‘
MAY 7°

JUMPERS

LOW IO [I . A- . .

JULY 30

FLIERS a: 3

HIGH ‘°i-|~—I4———.L-I-LLw—: . . , e-l. . .
FL! .

Low 20L! r v v L1 7 r v L‘f Ir r w JJr- 1
JULY

JUMPERS

HIGH 10‘1“, T . 1 4.2;, . r g L-lv -L_-

 

 

2|0

JULY

JUMPERS

LOW
'° '2 r . .
150

AUGUST

FLIERS

HIGH
10

110605130
mans ,,

LOW 10203010111T'4

 

 

 

32

I20

AUGUST

JUMPERS

HIGH
mrl_r

I70

AUGUST
JUMPERS
LOW

I0 I l
SEPT 30 '
mass 4 3 j
HIGH '° , r r ‘ f . 1 . T r T ,

SEPT

FLIERSi’gi—T‘ t f “fig, . 1 51.;

LOW

I:
FIE;

SEPT

JUMPERS
H'GH '04 ' T f W L r r r 1 -L r r T 4‘;

SEPT '5°
JUMPERS

LOVV
IO A I 3 ll

10 2'0 30 4'0

 

q

33

Homoptera) insects, the biomass contributed by larger size classes
increases through the summer. This trend is especially striking for
jumping types and reaches a peak in August. Second, jumping types are
more available in low grassy sites than high herbaceous sites. This is
largely due to Orthoptera (> 8 mm). Homoptera show the vertical
stratification but not the vegetative one. In contrast there is a much
weaker trend in the opposite direction for flying types. That is, there
is as much biomass available in high herbaceous sites as in low grassy
sites, if not more. The large spikes of biomass in 13-16 mm flying
insects in the high herbaceous site in August are Cantharid beetles and
Hymenoptera (Apinae and Vespinae), all of which are pollen and nectar
feeders. At this time the dense Solidago surrounding this trap was in

bloom.

Spider Phenologies

Mature females of Araneus trifolium were present from May to October

 

in both 1977 and 1978, while mature Argiope trifasciata females were

 

never present before August. The estimated phenologies of the two
species are depicted in Figure 4 using data from all observations in 1977
and 1978. The cohorts of both species that hatch in the same year grow
similarly. The full reproductive schedule of Araneus females is not
known. No mature Araneus males were observed until August 21 in 1978.
They were frequently observed wandering through the vegetation from

August 21 to September 15, 1978.

34

Figure 4. Cephalothorax width in mm vs. day of year collected

for Araneus trifolium and Argiope trifasciata for
1977-1978. Curves represent estimated cohort bound—

aries. Horizontal lines bound mature female size.

35

Argiope Araneus
d T
spider j _ mat 99
ceph. '1 ~ .L
-l -1
\Vidth .1 .-
(m m) '1 ‘1

 

 

 

 

   

36

Discussion

 

Foraging Strategy Differences Between Araneus trifolium and Argiope
trifasciata

 

 

From all available information (Levi, 1968, 1971a; Katson, 1948;

Tolbert, 1976; Phelps, 1967; Brown, 1979), Argiope trifasciata builds in

 

grass or herbaceous vegetation but not in the high, open sites that
Araneus can. Araneus can build its web at any height or density but not
in solid grass stands. Considering only the insect types used in the
laboratory studies, both species can build webs in sites where their more
easily caught prey are most available, but are more or less restricted
in building where the prey they catch with low success are most
available. However, before discussing the causal relationship between
morphology and web design on the one hand, and diet and foraging behavior
observed, on the other, I must consider the effects of several more
factors that constrain morphology, web design, and web placement. These
are predation, physiological homeostasis, web visibility to prey, and the
relevance of the experimental prey types.

First, there is considerable evidence that Hymenopteran predation is
a significant mortality factor for orb-weavers (Muma and Jeffers, 1945;
Dorris, 1970; Kurczewski and Kurczewski, 1968; Tolbert, 1976 and Coville,
1976). The predator avoidance tactics of these two spider species are
strikingly different. Tolbert (1975) gives behavioral evidence that the
dense hub and barrier webs of Argiope serve as screens and warning
devices. The spider's ultimate defense is escape by dropping into the
vegetation on a safety line. Effective use of this strategy requires the
visible accessory structures as well as suitable attachment sites behind

(and often in front of) the web for barrier webs. Use of high, open web

37

sites, where attachments for these structures are not available, may
involve higher predation risks.

The key element in Araneus' Hymenopteran defense appears to be the
retreat (Eberhard, 1970). When attacked the spider backs to the rear
corner until the attacker faces only legs and fangs. The spider can be
forced from the retreat by its destruction or if the attacker is immune
to its bite (e.g., a bird beak or ball-point pen), at which point it
drops into the vegetation beneath. This defense allows less visible webs
in higher, open sites, but may require vegetation with structures that
provide large, impenetrable retreats. This restriction appears to
prevent Araneus from building in sparse, thin grass blades even though
trap samples indicate flying insects are available here.

A second factor is physiological homeostasis. Argiope trifasciata

 

spiders have been shown to orient their elongated abdomen so as to
minimize heat input from the sun (Tolbert, 1976). Also its dorsum is
shiny silver and white and its venter black, facilitating reflection of
solar heat and radiation of heat to the ground. Furthermore, water
availability may be limiting in the dry habitats lrgigpg_uses. I
observed the stabilimentum to collect and hold drops of water sprayed on
the web in the lab. The spider drank these drops and the same may occur
with rain and dew. In contrast Araneus' abdomen is spherical in shape
and uniformly colored (usually dark brown). The shade and transpiration
of its leaf retreat greatly reduce the heat and dessication stress it
would experience in the open positions Argiope assumes.

Third, the fact that prey encounter rates of pollinator prey types
with Argiope webs were lower than those of Araneus webs in the same
macro-habitat (i.e., herbaceous vegetation) suggests that web visibility

and/or micro-habitat differences affect the probability of these flying

38

pollinators entering a web. The accessory structures, denser spirals,
and presence of the spider in the hub may make Argiope webs more
avoidable to these insects having greater visual acuity and
maneuverability, but may have little effect on avoidance by non-flyers
that first perceive the web in mid-jump. On numerous occasions while
monitoring webs I observed Diptera and Lepidoptera that were approaching
webs to execute abrupt 90° turns, thereby avoiding the web. The ability
of insects to avoid different web types should be testable under
controlled laboratory conditions. Open spaces in vegetation, especially
at the level of flower-heads being visited, should be high activity areas
for flying pollinators compared to lower, denser sites. Araneus' ability
to span larger gaps in the vegetation makes more of these sites available
to these spiders.
Finally, of the prey types tested in the laboratory, Lepidoptera are
insignificant in the diet. Diptera are significant contributors to the
diets of both spiders in this study and previous ones, and Orthopterans
are important in the diets of Argiope in grassy habitats in previous
studies. The unimportance of Orthoptera in this study is due to the lack
of Argiope webs in grassy sites. Several studies and habitat information
from taxonomic collections indicate that the absence of Argiope

trifasciata from grassy habitats is not typical (Levi, 1968; Kaston,

 

1948; Tolbert, 1976; Phelps, 1967; and Brown, 1979). In examining webs

of Argiope trifasciata and other orb-weavers Brown (1979) found

 

Orthoptera and Homptera to be relatively more frequent in grassy fields
than herbaceous ones and flying pollinators to be more frequent in

herbaceous than grassy fields. Levi (1968) frequently observes Argiope
webs "loaded" with grasshoppers. Several other prey types not examined

in the laboratory are common in the spiders' diets. Jumping Homoptera

39

are important in immature Argiope diets and are probably analogous to
Orthoptera relative to capture success. In a comparison of the diets of
two European Araneus species Kajak (1965) found the highest densities of
Homoptera in low, grassy sites. Cantharid, Chrysomelid, and Mordellid
beetles were found in Araneus and Argiope diets by Brown (1979), Phelps
(1967), Bilsing (1920), and Tolbert (1976). From attacks I observed in
the field these beetles escape much more slowly than Diptera yet are less
dangerous than Orthoptera. Both spiders appear to capture beetles of
their own weight with no difficulty. Stinging Hymenoptera are also noted
in spider diets in the same studies as above. These insects escape
somewhat more rapidly than Orthoptera and the above Coleoptera and they
are dangerous to the spiders. They are attacked with a wrap attack
(Robinson and Olazarri, 1971, and my own field observations). The small
sample of captures in the field indicate that Argiope is at least as
successful in capturing them as Araneus, if not more so.

The interaction of these factors (predation, homeostasis, web
visibility, and additional prey types) with morphological and behavioral
ones directly explored by this study could explain the differences in
foraging strategies of the two spiders. Argiope appears to be able to
use a broader range of vegetative habitats than Araneus, with the
literature indicating a more frequent use of dry, grassy habitats. Its
predator avoidance and thermo-regulatory tactics function equally well in
any plant type, and some prey types that it is capable of capturing with
relatively high success are available in both vegetation types. However,
Argiope's range of effectively used web heights appears constrained by
web design, and this restriction in turn affects encounter rates in
herbaceous vegetation. In contrast, Araneus' web design allows it to use

a wider range of plant densities and web heights, but restricts its

40
its effective range of vegetative habitats to those including some
herbaceous plants. In short, morphological and behavioral mechanisms
affecting predation risk, physiological homeostasis, and foraging success
appear to be tightly co-evolved and strongly interacting.

These arguments generate the following predictions. Given
sufficiently high prey availability in all habitats, Araneus should
survive and reproduce better in high, herbaceous sites and Argiope should
do so in low, dense, grassy sites. If the availability of flying prey
should be drastically reduced in a season, Araneus abundance should shift
to lower web heights rather than grassy sites. Should jumping,
herbivorous prey in grassy sites be reduced in availability Argiope
abundance should shift to herbaceous vegetation rather than higher sites.
Such a shift may have occurred (either by movement of differential
survival) in Argiope populations I studied in 1978. If drought-induced
decreases in primary and/or secondary productivity in drier, grassy areas
reduce prey availability there, then lower, moister herbaceous sites
certainly represent a more likely refuge than higher sites in grassy
habitats. If herbaceous sites do represent such a refuge, this might
explain Argiope's greater flexibility in vegetative habitat use. The
predominant prey in these sites (Hymenoptera, Diptera, and Coleoptera)
forage more widely than jumping herbivores (Orthoptera and Homoptera)
that tend to mature as cohorts in smaller areas. Consequently, the
availability of these flying prey within the sampling area of a spider
may not fluctuate as severely with climatic extremes as those of the
jumping prey. As a result, the frequency of seasons in which selection
favors a shift from grassy to herbaceous sites may be much higher than
that for seasons in which the move from herbaceous to grassy sites is
profitable. This could result in differences in foraging specialization

between the two species (Wiens, 1977).

41

The ability of Araneus females to survive through two seasons may
also be linked to the prey specializations discussed above. For both
species in the latter part of the summer, prey greater than 7 mm make up
over 80% of diet biomass. Although the same rate of intake may not be
necessary for maintenance as for reproduction, the relative abundance of
large prey items of the right type may be a limiting factor in early
spring for mature spiders. Many of the abundant grasshopper species of
mesic old-fields hatch in spring and mature in late summer (Cantrall,
1943), whereas large social Hymenoptera and many Coleoptera overwinter as
adults, and large Diptera have relatively short life cycles. All of
these fliers may constitute the most available source of large prey in
the spring. A variety of other factors could be invoked to explain this
life-history difference. Levi (1968 and 1971a) indicates that Araneus is
more of a temperate species and Argiope is distributed more tropically.
Patterns of life-histories relative to latitude and local prey
availability schedules would help determine the roles of season length
and prey specialization in shaping phenology.

There is evidence for a phylogenetic basis to the differences in
foraging strategy between the two species. The genus Araneus in North
America consists almost entirely of species preferring forests, shrubs,

and buildings. Araneus trifolium and Araneus marmoreus are the two major

 

 

exceptions. These two species typically build webs in old-fields, with
‘A, marmoreus preferring older, shrubbier fields (Levi, 1971a). Most
Araneus spiders build retreats. In contrast, Argiope species prefer
grassier fields and earlier successional stages. They all have

morphology, behavior, and web design similar to A. trifasciata (Levi,

 

1968). The role of interspecific competition in shaping the divergent

evolution of these two genera or in determining present structures of

42

orb-weaver guilds cannot adequately be assessed with information
currently available (Enders, 1975). More extensive field surveys
comparing life histories of entire guilds of orbdweavers and field
manipulations examining mechanisms of inter-specific interactions are
needed.

Understanding the mechanisms of foraging specialization and their
conflicts and interactions with other life history mechanisms is a
pre-requisite to predicting the potential distribution of a species as
well as the response of populations to changes in habitat, climate, and
prey availability. Such mechanistic information will allow more
effective investigation and application of general, reductionist models
of the effects of isolated factors such as competition and predation on
abundance and distribution of populations. This comparative analysis of
two species cannot constitute evidence of these foraging specialization
patterns in all orb-weavers, but it can give more precise direction to
broader behavioral and ecological analyses in the future. Several
testable predictions of population level responses are also generated by

this qualitative foraging model.

Chapter II
PATTERNS OF FORAGING SPECIALIZATION IN A GUILD OF OLD-FIELD ORB-WEAVERS

Mechanisms of Attack Differences

The extension of the results from the comparison of these two
species to the prediction of diets and foraging behavior of other
orb-weavers requires an understanding of the mechanisms linking
morphology and web design to capture success and probability of trapping
prey. These data suggest several such mechanisms as well as ways to test
their validity. .

The laboratory results support hypothesized capture success
differences between spider species, but not attack time differences. The
best explanation of these prey capture differences involves both spider
morphology and web design. The web serves two related functions in prey
capture: detaining the prey until the spider can reach it and
restraining the prey during the attack. The relative importance of these
functions differs for different combinations of prey types and spider

species. In attacking Orthoptera, the more robust legs and
mouthparts might allow Araneus to bite after a shorter wrap time due to a
greater number of potential biting sites. On similar prey with very hard
exoskeletons (some Coleoptera, Hemiptera, and Orthoptera) both spider
species were frequently observed to make repeated futile attempts to bite
until membranous joints, eyes, or mouthparts were encountered. Argiope's

small chelicerae may force it to search longer for a more restricted set

43

44

of suitable sites, yet the denser mesh of its web may better restrain the
grasshopper during this time. A more controlled analysis of web
retention differences for Orthopterans is needed to determine the causes
of differences in length and variability of attack times.

In attacking relatively small Diptera, if spiders of either species
reached the insect they could subdue it. For the range used, this
applies to all sizes of flies for Araneus (including a 24 mm Tabanid).
However, Argiope's tendency to wrap the largest flies it captured may
indicate it had reached the effective limit of the attack capabilities of
its chelicerae. The greater prey retention capabilities of Araneus webs
probably also contribute strongly to differences in capture success.

Yet, since Argiope is already in the hub when a fly enters, while Araneus
has to descend from its retreat first, it would seem that Argiope would
have some time advantage in reaching a prey. However, approach is too
rapid in both species to be analyzed for time differences with these
data. A large sample of approach sequences filmed at high speed might be
able to detect such differences.

For Lepidoptera, the lack of significant differences in escape times
between spider species rules this factor out as a cause of the capture
success difference. Although escape times for moths and Diptera are very
similar, capture success is much lower for moths. This may be due to the
fact that moths are usually gliding through the web during their escape,
but flies remain in one spot until they break free. Such a moving target
may be more difficult to close with accurately. Similarly, the greater
range and rapidity of movement of butterflies through Argiope webs may
play a role in its lower capture success. For relatively large
Lepidoptera, Argiope could reach the prey but was unable to close and

bite as effectively as Araneus, as described above.

45

Thus a synergism between morpholgoy and web design appears to cause
the observed attack differences, especially for relatively large prey.
Web detention properties alone may determine capture success for prey
small enough to be easily handled by both spider species. These patterns
suggest that capture success may decrease with increasing relative prey
size at different rates for the two spider species, so that capture
success differences on larger prey may be even greater than those
testable with those data. This argument also suggests that the largest
prey effectively available to the two spider species may also be
different.

Consideration of web design differences and prey escape behaviors
suggests possible bases for prey retention differences. The flying
insects pull away from the web by the force of flight, aided by scales or
high vibration frequencies to break the adhesive bond. Both escapes tend
to be perpendicular to the plane of the web such that the insect does not
encounter and accumulate more threads of silk. However, non-flying
insects tend to rip down vertically through the web, accumulating more
and more resilient, binding threads. Given these escape patterns,
grasshopper retention might be increased by an increase in the number of
spirals accumulated, via a bigger web, denser spirals, or both.
Lepidoptera and Diptera might be better retained by stronger spirals or
contact with a greater surface area of adhesive. The latter might be
accomplished by denser spirals, a thicker adhesive coating, or both. A
proper test of this argument would entail mechanical tests in conjunction
with tests of prey retention capabilities of empty webs. Robinson (1975)
gives comparative behavioral evidence that more closely spaced spirals
and stickier spirals retain flying prey better, giving the same

mechanical reasons as above. The effects of web visibility on encounter

46

rates with flying prey may cause selection against increased mesh density
as a mode of increasing capture success on this prey type. Also,
trade-offs between thickness of thread (or adhesive layer) and length of
thread are caused by limited available silk (Witt, 1963). Both of these
factors may contribute to Araneus' more widely spaced spirals.

The many constraints on attack time do not indicate a simple, net
positive selection for minimizing this parameter. Since attacks require
energy, take place away from the hub or retreat where the spider can
sense predators or other prey, and may expose the spider to continued
risk of injury, it would seem there would always be strong advantages to
minimizing attack time (Robinson, 1975). On the other hand, certain
combinations of web design and spider morphology make the slower of a
possible choice of attacks both safer and more effective. Certainly
the attack times observed in the lab and field rarely cause the spider to
lose any prey entering the web while a prior insect is being attacked.

In short, attack time will not necessarily be an accurate predictor of
capture success or long-term foraging efficiency. In summary, a variety
of objectively measurable morphological and web design characters may be
used to predict the following prey specializations in other orb-weaver
species. Overall body size will, of course, restrict the size range of
prey available. Relative fang and leg sizes should be related to bite
and wrap attack capabilities. Stout-legged, large-fanged spiders would
be expected to specialize on relatively innocuous prey, including rapidly
escaping insects. Longer legs, denser sustentacula, and smaller fangs
should be associated with more slowly escaping and more dangerous prey.
Widely spaced spirals should be associated with prey more capable of
detecting and avoiding webs. More dense spirals should indicate

specialization on less agile but more dangerous prey types.

47

I will examine these hypotheses in the light of data available on
prey-type distributions and patterns of morphology, web design, and
life-history in an assemblage of old-field orb-weaving spider species.

To begin, the combinations of morphology, web design, and web placement
in eight species of orb-weavers will be searched for patterns supporting
the specializations discussed above. Then, I will examine the effects of
spider body size and phenology on relative foraging success in different
seasons and habitats. Finally I will discuss each of the species
separately and make predictions of their diets and habitat use. As
pointed out in the discussion above, any spider character is no doubt
subject to many different selective forces. The approach here will be to
examine patterns of association of existing characters in light of their

potential effects on foraging.

Patterns of Specialization

Data for this study were collected in the summer of 1977 in 14
old-fields of the East Lansing, Michigan area. These fields ranged from
1-20 hectares in area and were generally mesic to xeric. Vegetation
ranged from almost solid stands of quack grass to almost solid stands of
Solidago spp. In general fields were mosaics of the vegetation types
described on p. 8. From a selected starting point I ran a transect in a
random direction for 20-100 m. A11 orb-webs within a l m swath were
included. The following data were collected for each web: height of hub
above ground, vertical and horizontal diameters, mean distance between
spirals in lower vertical sector, position of spider in web, percent area
of web un-damaged, per cent cover of vegetation types in a 5 m radius,
vegetation types web was attached to, time and date. Each spider was

preserved in alcohol and the following were measured for each using a

48

100x binocular microscope: greatest transverse cephalothorax width,
length of leg IV, paturon and fang length, and the number of sustentacula
on tarsus IV. Spiders were identified to species if possible. Immatures
were classified only if a clear series of coloring and morphology leading
to adults was evident. The eight species shown in Table 7 were chosen
for this analysis because there was sufficient sample sizes of each and
because the taxonomic literature indicates they are typically found in
old-fields (Levi, 1968, 1971a, 1971b, 1973, 1975, 1976; Kaston, 1948).
There were small numbers of several other species, and large numbers of
immature Araneus spp. that were probably forest canopy types (Levi,
1973).

These data were used in a principal components factor analysis of
the relationships between the following variables: length of leg IV,
paturon length, fang length, mean distance between spirals, web area, and
web height. A logarithmic transformation was made on all variables to
meet the assumption of normality in the factor analysis. All variables
except web height were divided by the logarithm of cephalothorax width to
eliminate biases due to size-frequency distribution differences between
species samples. Thus, these variables are size-free, relative measures
of the morphologies and web designs. Web area was calculated using the
mean of vertical and horizontal diameters. Absolute web heights were
used because the real height of prey-type distributions will be compared
to them. The SPSS Factor Analysis routine with default control
parameters and varimax rotation was used (Nie, et al., 1970).

The correlation coefficients of the six variables on the three
factors are shown in Table 8. Factor 1, which explained 53% of the
variance in the data, is strongly correlated with fang length and more

weakly correlated with paturon length and web height. The association

49

 

 

 

 

 

 

 

 

 

Anssv Apszv £63 GM
as: as: as: as: amouumu as:
ammuuou umwuumu cowuawoa umvwnm
wash 0 uamm uwnouoo Honouoo Honouoo museum QM\m .on2 ca monEom
umsws<
1mm am: INN sass Image 1G3 «can IoN sass HHm Imz\m spasms mo magma
wcfiwam Hamw Ham“ wcfiwmm wafiuam Hamw HHMM Mama hwoaoaoza
masomuaoumam
o mIo N N wlo colmfi nmlm anoH
. we Hanan:
monEwm musuma
m.~lo.~ m.m1m.~ m.fi1~.fi *~.m1m m.~1w.~ m.m|m.~ m.o1m.~ o.m1m.~ «6 auvws
xmuosuoamnaou
mwmcmummm. mfimcoummm. mmouonnww. mandamum :owmumo: maucmusm Eafiaomfiwu wumwommwfinu
msocmw< mcoomooz muowdwz mufiomusuamo< moo umdawu< mamamu< omofiwu<

 

 

.amwwnowz scum mowooam Ho>mmzlnuo wamfimlvao w mo mofiumfiumuomnmnu mo humasam .n magma

50

.owcmu .mofiomam mam mo uuma Gauguuo: mam Ga Howuma mum moamaom

 

 

 

 

 

em mm m - 0 mm

:oEEou coaaou
:oEEoo :oEEooca mama :oEEoo

mum> muo>

oumuomsou .Gaou mumumaEmu oauuum oumuonamu
oumummsou oumumGEou oumuoasou

Iamofiaouu I.aouu Iamowaouu Ioumumasou Iamofiaouu
mfimzmumum mfimcmummm amoumnaww mumHHoum commune: mwucmusm sswaomwuu mumaommmwuu
mamcwu< «woman: whammoaucmu< mamcmu< oaowwu<

¥

zvsum mfisu a“

swam maaamm

 

cmmfigoaz as

.pasnm o>HumHmu

 

owcmu

canamumoow

.A.e.ucoov s magma

51

Table 8. Varimax rotated factor matrix after rotation with Kaiser
normalization for morphological and web variables from

8 old-field orb—weaver species.

 

 

 

Variable Factor 1 Factor 2 Factor 3
Leg length - .18 - .52 .25
Paturon length .70 .18 .09
Fang length .99 - .05 - .17
Web height .52 - .26 .22
Mesh width - .17 .72 .17

Web area .08 .00 .75

52

between the two cheliceral parts reflects their mechanical and functional
linkage. The possible correlation between fang size and web height is
weak, but is consistent with the hypothesis generated by the

Araneus-Argiope comparison above that prey types most effectively

 

captured by bite attack (Diptera, Lepidoptera, some Coleoptera) are more
available in high sites than are Orthoptera and Homoptera. Such a
prey-type distribution might select for this pair of traits in
combination.

Factor 2, accounting for 26% of the variance, is most strongly
correlated with mesh width, and a weak negative correlation exists with
leg length. The association of these two traits is consistent with
another hypothesis from the previous study: long legs and dense spiral
meshes act synergistically to subdue dangerous prey types that are too
awkward to avoid the more visible, high-density mesh. Finally, the third
factor accounts for 22% of the variance and is most strongly correlated
with web area. Figure 5 describes the attributes of the eight species in
terms of these three factors.

Two relationships are suggested by the distribution of these species
in the factor space. First, the retreat-utilizing species (Araneus

trifolium, Acanthepeira, and Araneus pratensis) all tend to have large

 

 

 

fangs, short legs, and higher webs with more open mesh. This whole suite
of attributes might simultaneously enhance encounter rates and capture
success of agile, flying types as discussed above (low web visibility,
webs in flight paths and effective bite attack). Second, spring maturing
species tend to have small fangs, short legs, and low webs with open
mesh. The low web heights may be largely a reflection of lower
vegetation heights in early spring. However, the whole group of
attributes may be related to a specialization on weak, slowly escaping

prey. This hypothesis will be further discussed below.

Figure 5.

53

Population distributions in factor space. Ordinate

is factor one score, abscissa is factor two score.
Centroids are bound by 95% confidence ellipses. X's
are low scores, 0's high scores, and dots intermediate
scores on factor three (web area). Size of type rep-
resents typical relative size of mature females. Un-
derlined species are spring maturing, others mature

in fall. Quadrants are labeled with variables that

are highly correlated with factors one and two.

 

54

FACTOR ONE

LARGE FANG LARGE FANG

 

227.235.“... AR ANEUS 21:27.22.
LONG LEG 2 TR' FOLIUM SHORT LEG

     

ARGIOPE
AU RANTIA

ACANTHEPEIRA

ARGIOPE

TRIFASC IATA
-I U v I 2

MW

 

 

FACTOR
TWO
0
MANGORA
-|
smu FANG SMALL FANG
LOW was LOW wen
DENSE MESH open MESH

LONG LEG SHORT LEG

55

Both of these patterns in the factor space distribution of the
species are corroborated by discriminant analysis. For the retreat vs.
hub comparison, a single discriminant function with factor score
coefficients as shown in Table 9 correctly identifies 89% of the
individual spiders into the correct category. As the coefficients
indicate, most of the discrimination is achieved by leg length (shorter
in retreat species), but mesh width and paturon length are also
important. For the spring vs. fall comparison, a single discriminant
function predicts the groups membership of 89% of the cases correctly.
Again the variables important in discrimination are among the same as
those indicated in the factor analysis: largely leg length and web
height, with mesh width secondarily important.

Thus, the factor and discriminant analyses tend to confirm the
functional relationship between high webs and large fangs, between long
legs and dense mesh, and between retreat-use and short legs, open mesh,
and large fangs. However the unified suite of long legs, short fangs,
low web, dense mesh, and hub use (and its converse) is not supported.
This decomposition of the originally hypothesized "biter-wrapper"
dichotomy will be discussed below when I deal with the exception to it:

Gea heptagon.

 

Optimal Phenology and Habitat Use

Using models developed from the previous study, I will estimate the
effects of spider phenology and morphology on energy return rates for
different habitats and seasons. The model (in its biologically
meaningful form) is simply net energy = available energy - prey handling
costs - search costs. Since no comparisons will be made between

different size spiders, search costs will be omitted.

56

Table 9. Standardized discriminant function coefficients for Spring
vs. Fall phenologies and hub vs. retreat behaviors in 8

old-field orb-weaver species.

 

 

Hub vs. Spring

 

Variable Retreat vs. Fall Group Centroids
Leg length .95 .97

Spring — 1.8
Mesh width - .58 - .51

Fall .8
Paturon length - .47 - .31
Web height .25 .83

Hub .7
Fang length - .24 - .14

Retreat - 1.8

Web area - .02 - .12

57
n n
E = (eij aij Cij - aij cij hij m)
i=1 j=l

where E is net available energy in calories, i_is prey type, j is prey
length, £_is total energy potentially available per prey, a_is relative
availability of that size and type prey, g is capture success, h is
handling time, and.giis spider metabolic rate. Estimates of each
parameter for different relative size and types of spiders and prey are
derived below. §_is calculated from length-dry weight regressions for
Calyptrate Diptera, as justified in Chapter 1. These weight values are
multiplied by 5150 gm-cal/gm dry weight (Connell, unpub. data). Relative
availability is numbers of insects caught in window-pane traps described
in the previous study. Web area and visibility will affect availability
rates, but since I have no quantitative data for visibility effects and
since these two attributes seem to vary inversely, they will be omitted.
Spider metabolic rate was taken from Anderson's (1970) study, using

the estimate of 356 :_33 1 Oz/g-hr for the web spider Achaeranea

 

tepidarioram. This basal metabolic rate was multiplied by 3 for an

 

estimate of active metabolic rate during an attack. The final value of
4.8 cal/g-hr was obtained using the standard conversion 1 metabolic 02
= 4.8 kcal. Spider weight was calculated from a highly significant
regression of cephalothorax width in millimeters vs. fresh weight in mg:
fresh weight .00319 (cephw) 3‘36 (r2 = .92). This was derived using

lab-reared Araneus trifolium spiders. Finally, metabolic rate was

 

calculated as: m = (4.8 cal/g-hr) (weight)0'76 (Ford, 1977).
Handling time and capture success were estimated from the lab data
in the previous chapter. Handling times were calculated from power

functions fit to attack time vs. relative prey length for each spider

58

are given in Table 10. .bl‘values are asymptotes of plots rather than
actual regression parameters. Argiope regressions were either very
similar to those for Araneus (Orthoptera and Lepidoptera) or unusable.
The model does not incorporate metabolic costs for attacks ending in
escapes because these always involved minimal or zero handling times.

Capture success was estimated for linear regressions of arcsin

 

“‘1% capture success vs. relative prey length. This transformation was
used to correct for truncated variance at high and low % values. This
regression gave a better fit than an untransformed linear or negative
exponential or a transformed negative exponential plot. There were no
significant differences between slopes or intercepts of regressions for
Araneus and Argiope on any prey types, so the regression with the
greatest r2 was used for each prey type (Table 11). In the model,
arcsin % capture success was converted to % capture success. Since
probability of capture success is multiplied directly by prey
availability, the model effectively allows the spider to eat a fraction
of each prey item proportional to its capture success probability.
Obviously in nature each insect is either captured or escapes as a unit.
Thus, this model must be interpreted as an average of a population's
foraging behavior, not as one incorporating the effects of discrete
capture events. For large spiders using larger and relatively rarer
prey, the actual effects of discrete capture events could greatly
increase the variance in foraging success between individuals relative to
smaller spiders.

Two sets of comparisons of net energy return rates between habitats
and seasons were made. In the first set, the goal was to determine the
optimal habitat usage pattern for spiders of given morphology, size, and

phenology. Morphological effects on capture success were incorporated

Table 10.

59

Regression parameters for the equation:

handling time = b1 + b2(relative prey length)b3 ,

for different prey types.

hundredths of minutes.

Handling time is in

 

 

spider x prey

used for these

 

combination prey types in

derived from b1 b2 b3 the model

Araneus x Diptera 1.0 .00156 8.62 Diptera

Araneus x Orthoptera 6.0 7.869 1.95 Orthoptera, Homoptera,
Coleoptera, Hemiptera

Araneus x Lepidoptera 6.0 54.85 1.76 Lepidoptera,

Hymenoptera

60

Table 11. Regression parameters for capture success by Argiope and

Araneus on different prey types. Regression equation is:

 

arcsinV% capture success = a + b(relative prey length).
Parameters designated by asterisks were used in the

model.

 

 

spider x prey

 

combination a j; s e b i s e r2
*Argiope x Lepidoptera 47.1 :17.7 -5.4 i 1.3 .65
Araneus x Lepidoptera 20.2 i 13.5 2.6 i 2.7 .15
Argiope x Orthoptera 341.6 1:261 -70 i140 .26
*Araneus x Orthoptera 103.63: 7.0 -10.9;t 1.1 .92
Argiope x Diptera. 49.2 t 9.6 -6.1 1:1.9 .63
*Araneus x Diptera 65.3 1:5.3 -7.1 1:1.1 .92

61

into the model according to the hypotheses developed from the Argiope-
Araneus comparison. Since the imprecision of the capture success-prey
size regressions did not allow direct parameterization of these differ-
ences, a constant and arbitrary difference of 15% was chosen for the
effect of spider morphology on capture success of different prey types.
The justifications for this are as follows. There are no indications
from the data that a more complex transformation would be any more
accurate at this stage. Also, the mean difference in capture success on
all 3 prey types using the sub-samples selected for the chi-square tests
in Chapter One was 19%. (These sub-samples are from the mid-range of
relative prey length where sample sizes are highest). This crude
adjustment was accomplished by using the above regression parameters and
their modifications for different spider-prey types combinations as shown
in Table 12.

Thus, the "wrapper" and "biter" are equally successful on Coleoptera
and Hemiptera, the "biter" is more successful on Lepidoptera and Diptera,
and the "wrapper” more successful on Orthoptera, Homoptera, and
Hymenoptera. This combination of regressions used the data as accurately
as possible and also follows the qualitative predictions of the
hypotheses generated by the previous study. The relationships may be
much more complex in reality, but the imprecision of the trap sample data
alone does not warrant further refinement. Also, only relative
comparisons will be made using this model. Plots of original data with
the regressions in Table 12 superimposed are shown in Figure 6.

The results of this model using the trap samples described above for
prey availabilities are shown in Figure 7. The first pattern that
emerges is that the "wrapper" type does as well or better than the

"biter" in almost all habitats and seasons (Sign test, P < .05). Given

62

Table 12. Modifications of capture success vs. relative prey length

regressions used for various spider vs. prey combinations

in the net energy return model.

Where indicated, the

estimate of the y—intercept was changed by 15%.

 

 

spider x prey type in model

regression used

 

"biter" attacking Diptera
"biter" attacking Lepidoptera and
Hymenoptera
"biter" attacking Orthoptera,
Homoptera, Coleoptera, & Hemiptera
"wrapper" attacking Diptera
"wrapper" attacking Lepidoptera
"wrapper" attacking Hymenoptera
"wrapper" attacking Orthoptera and
Homoptera
"wrapper" attacking Coleoptera and

Hemiptera

(Araneus

(Argiope

(Araneus

(Araneus

(Argiope
(Argiope

(Araneus

(Araneus

Diptera)

Lepidoptera)+15%

Orthoptera)

Diptera)-15%

Lepidoptera)

Lepidoptera)+15%

Orthoptera)+15%

Orthoptera)

63

Figure 6. Per cent capture success vs. relative prey length for
Argiope (x's) and Araneus (dots) on three prey types.
Regression lines are from Tables 11 and 12. Sample
sizes for points are small (1—5) at extreme relative
prey lengths and large (15—25) for the intermediate

range.

643

X X
(ARGIOPE-LEPID)

IOO

1111111111

3

 

 

(ARANEUS-ORTHOP)

  
 
 

(ARANEUS-ORTHOPI*I5 %

100 '1

% .1
CAPTURE -
success j
1

IO ‘1

 

 

I 5 l0 I2
RELATIVE PREY LENGTH

65

Figure 7. Net energy return rates (kcal) in different habitats
based on "wrapper" (dotted lines) and "biter" (solid
lines) models. Dates are May 31, July 19, August 30,
and September 28, 1977. Results for spiders of four

different cephalothorax widths are shown.

66

 

 

 

 

 

 

 

 

 

 

 

 

 

290 ,m. r;
200 =5
#0
5‘
100
10
I70 .-.
g":
" 5 3mn1

 

 

a 8
11111111111111111

 

     

 

 

 

 

KCAL 1222

.-~.---

IO -. , ' E -. . ..
MJAS MJAs MIAs MJAS MIAs MJAs MJAs MJAS
grass of”: mixod hubs gross gran mixed herbs

LOW

 

 

HIGH

67

the nature of the model, however, it is not possible to test the
significance of the quantitative differences in net energy return between
habitats. However, the ranking of energy return rates by habitat does
correspond qualitatively with the usage patterns observed for Argiope and
Araneus in the previous study. That is, low grassy and mixed herbaceous
sites are most productive followed by high herbaceous sites, which are
equal to or followed by low herbaceous sites, with high grassy and mixed
sites least productive. As noted above, Argiope prefers low grassy sites
but does use low mixed and herbaceous sites, while Araneus prefers high
herbaceous site but does use low herbaceous and mixed sites. Neither
species uses high grassy sites. The apparently greater productivity of
Argiope's preferred site is consistent with its greater density relative
to Araneus. Finally, the fact that the "wrapper" does at least as well
in all habitats may reflect the possibility that some prey types that it
is equally or more successful on are relatively abundant in every
habitat. In contrast, prey types captured more successfully by the
”biter" are relatively less abundant in low grassy habitats. However,
all of these habitat comparisons depend especially on the accuracy with
which these traps reflect the relative encounter rates of different prey
types with webs. Because of this assumption and others, such conclusions
should be regarded cautiously. At this point, I can only say that the
model results do not contradict the hypotheses about habitat use
generated by the previous study.

The major change in these patterns as spider size is decreased is
that the return rate of low herbaceous sites become increasingly greater
relative to other habitats. This is due to the large number of small
prey in this sample (mostly Homoptera), as compared to the predominance

of large prey (mostly Coleoptera and Orthoptera) in the low grass and

68

high herbaceous samples. Finally when the smallest spiders are examined,
return rates for all high sites become very small relative to those for
low sites. This pattern may indicate prey availability as a partial
cause of the trend for young spiders to start building low webs (Enders,
1974, data from Chapter 1), in addition to the fact that vertical
orientation and increasing web size physically force this result to some
degree.

In the 2nd set of comparisons using this model, the goal was to
determine the optimal phenology for a spider of given mature size and
habitat preference. This was done by simply examining the seasonal
pattern of return rates within given habitats for given spider sizes.

For larger spiders (3-5 mm cephw), maximum return rates occur in August
for all habitats except low herbaceous sites, where the peak is in June
and July. For medium-sized spiders (cephw 2 mm), some of the peaks shift
so that in all low sites greatest return rates occur in May and July.
Peak return rates remain in August for high sites. Again significance of
quantitative differences is not estimable. For small spiders, all peaks
are in early summer. These patterns hold true for both spider types.

The late summer peaks for large spiders reflect the higher
availability of large insects at this time. The early summer peaks for
medium size spiders indicate their inability to capture these larger
insects as well as the preponderance of medium-sized (5-8 mm) Homoptera
in the low herbaceous site. In summary, fall maturing phenologies are
predicted to be optimal for 3-5 mm cephw spiders in all habitats except
low herbaceous sites, where a spring maturing phenology is optimal.
Spring phenologies are optimal for 2 mm spiders in low sites, whereas
fall maturing is optimal in high sites. For 1 mm spiders, optimal

maturation time should be early summer.

69

Foraging Strategies of Individual Species

The examination of each of the eight species for which I have data
will have 2 parts. First, I will briefly summarize their characteristics
in terms of the factor analysis model. Second, I will attempt to
interpret each spider's set of characteristics in terms of the hypotheses

generated by the ArgiopeeAraneus comparison. Since these hypotheses

 

linked morphology and web design to foraging behavior, this interpreta-
tion will lead to specific predictions about habitat use and diet. Where
possible, I will compare these predictions to actual behavior. In the
cases where the data are available, I will also compare actual habitat
use and phenology to that predicted by the net-energy- return model. For
cases where data are currently lacking, the predictions will stand as
hypotheses to be tested in future studies.

Araneus trifolium: Araneus is a large (3-6 mm) fall maturing

 

species that can overwinter as adult females. It is a retreat builder
with large fangs and high, small webs. It has an intermediate density of
sustentacula. Interpretation of this suite of characters would be highly
circular. However, one valid point can be made concerning its phenology
and habitat use relative to prey availability. As described above, it
fits the optimal phenology for a large spider. In addition, mature
individuals overwintering and building webs in thier second spring also
appear to use the only habitat which had an optimal spring phenology for
large spiders: the low herbaceous habitat. This is based entirely on a
few casual observations of mature Araneus females in the spring. Of
course, most vegetation is shorter in the spring and would tend to
physically constrain web heights at this time.

Argiope trifasciata: Argiope is also a large (cephw 3-5 mm), fall

 

maturing species. However, mature females never overwinter. It is one

70

of the most common and locally abundant species. These spiders have long
legs with very dense sustentacula, intermediate size fangs, and sit in
the hub of their medium-sized, medium height, dense meshed web. As for
Araneus, no predictions need be made.

Argiope aurantia: .éf aurantia is very similar to A. trifasciata.

 

 

It is a large (3-6 mm) fall maturing species that never overwinters as

mature females. It is usually less abundant than A. trifasciata. It

 

sits in the hub of its web, which is intermediate in mesh density and

height to Araneus' and A. trifasciata's. This height difference is

 

supported by Enders (1974), Phelps (1967), and Tolbert (1976), though

Brown (1979) gives evidence for A. trifasciata having higher webs. .A.

 

aurantia has a high density of sustentacula and fairly large fangs. (I
should note that, as the 95% confidence ellipse indicates, these
characterizations are the weakest of all 8 species). A? aurantia also
builds webs with stabilimenta and barrier webs. From its almost exactly

intermediate position between Araneus and A. trifasciata in the factor

 

space diagram, I would predict that it would have an intermediate diet
and habitat use. Certainly the habitat descriptions from my sample,
taxonomic collections (Levi, 1968, Kaston, 1948), and previous
comparisons of the two Argiope species (Tolbert, 1976, Phelps, 1967)
support this prediction. All describe A. aurantia as building in

moister, more herbaceous habitats than A. trifasciata but still in

 

predominately grassy fields. The only available dietary information
(Phelps, 1967, and Brown, 1979) does not indicate striking differences in

prey-type frequency distributions from A. trifasciata.

 

A very interesting phenological difference exists between the two
Argiope species. .A. aurantia is consistently 2-3 weeks ahead of A.

trifasciata in emergence, maturation, and reproduction (Tolbert, 1976,

 

71

and Figure 8). Tolbert attributes this to higher, more open egg-laying
sites in A. aurantia where eggs receive more insolation early in the
spring and thus metabolize at a faster rate. This earlier phenology in
conjunction with use of moister, higher habitats may lead A. aurantia to
have more large, flying prey in its diet.

Acanthepeira stellata: Acanthepeira is a medium sized spider that

 

matures in the spring. Some females may survive throughout the summer.
It is very uncommon but locally abundant (Levi, 1976). It has very short
legs with 2 sustentacula, and very open-meshed, small webs placed fairly
high. It has fairly large fangs. This species is fairly unusual in
having a hard abdomen with short, obtuse spines. Individuals usually
wait for prey in retreats but occasionally sit in the hub. This species

may be nocturnal (building its web at dusk). Acanthepeira does have the

 

optimal spring phenology for its body size. The taxonomic literature
(Levi, 1976, Comstock, 1948) only indicates a general meadow and field
habitat use for this species. My own data indicates a tendency to use
drier sites of mixed grass and herbs, though the sample size (n=7) is
small. From its morphology and web design I would predict that it would
typically occur in herbaceous vegetation and that its diet would be
predominately rapidly escaping, harmless, flying insects.

Gea heptagon: This species is also medium-sized (2-3 mm cephw),

 

matures in the spring, and can survive through the summer as mature
females. It, too, is rare but locally abundant. It has short legs with
few sustentacula and small fangs. It has a very low, small web with

relatively open mesh. Gea's abdomen is similar to Acanthepeira's in

 

being obtusely spined, though it is not as hard. These spiders sit in
the hub, but build no stabilimenta or barriers. The maturation in spring

is optimal for this size spider. Gea confounds the morphological and

72

Figure 8. Cephalothorax width vs. day of year collected for

Argiope aurantia (x's) and Argiope trifasciata (dots)

 

 

for 1977.

o H o o
x x O O. ”O.
3.4..
00000
omuuoalo
3 coco-mo
7 .0
xx .....
O
0..

 

 

SPIDER
CEPHW.

74

web-design dichotomy on which I based the "biter - "wrapper" model by
having relatively small fangs and open mesh combined with short legs.
The only consistent dietary prediction I could make from such a
combination would be that it preyed mainly on weak, slowly-escaping
insects. This excludes most of the prey types discussed above, but does
include relatively small prey and Nematoceran Diptera. These types are
certainly common in low dense vegetation in the spring. The prey
availability data above do not allow a precise prediction of habitat use
based on this speculated prey specialization. However, my small sample
(n=12) does indicate a preference for grassy or mixed vegetation and a
consistent placement of the web right at ground level.

Neoscona pratensis: This medium-sized (2-3 mm) species matures in

 

the fall. It is common and abundant. It is intermediate in all
characters as denoted by its central position in the cloud of species in
the factor space diagram. It sits in the hub of its simple web.
Concommitant with its generalized morphology and web design I would
predict it would be generalized and flexible in its habitat use and diet.
No dietary information is available, but taxonomic descriptions (Levi,
1971) locate it in a range of habitats from swamps to xeric old-fields.
Neoscona does not fit the optimal phenology predicted for its size.

Araneus pratensis: Araneus (as described in Levi, 1973) is a small

 

spider (1-1.5 mm). It matures in the spring and is abundant and
widespread. Araneus has relatively short legs with very few sustentacula
and intermediate-sized fangs. It builds its open mesh webs at medium
heights. It sits either in the hub or on vegetation attached to the web.
Its rather intermediate position in the factor space indicates a foraging

strategy more generalized than that of Acanthepeira, but still

 

predominately using herbaceous habitats and flying insects. Again, there

75

is no diet information, but the habitats used tend to be "moister' (Levi,
1973).

Mangora gibberosa: Mangora is one of the most specialized of these

 

species. It is small (l-l.5 mm), common, abundant, matures in mid to
late summer, and overwinters as eggs. It has long legs with 2
sustentacula and a very large, dense-meshed, low web. Its fangs are
small. The hypotheses above would clearly predict it to build mostly in
dry, grassy sites and to prey predominately on Coleoptera, Orthoptera,
Homoptera, and/or Hymenoptera. No data on diet are available, and the
taxonomic literature indicates only that it prefers fields. My own data

does indicate a preference for dry, grassy and mixed herbaceous sites.

Assumptions and Qualifications of these Predictions

Very little dietary information is available for any of these
species. The phenologies predicted by the net-return-rate model are
consistently corroborated only for the large species. For the
phenological predictions, besides the many assumptions in deriving and
applying the model, there is the additional question of what "size" class
some of these species actually fall into (see ranges in Table 8).. The
question is especially crucial for those species at the size where the
switch occurs from optimal spring to optimal fall phenologies, i.e., 2-3
mm. The cephw's used in the above discussion are based on both widely
scattered taxonomic samples (Levi, 1968, 19713, 1971b, 1973, 1975, 1976)
and my own local and small samples. If larger females are more
successful reproductively than smaller females, mean size may not be an
appropriate measure for these evolutionarily based arguments. Also, the
smallest spiders discussed here are at the extreme of the size range used

in deriving the capture success parameters.

76

Very little precise data on habitat use are available for any of
these species. The few for which descriptions exist seem to fit the
predictions generated by the above model, but even these are based on
extremely general habitat descriptions for small samples. Given the

example of Araneus trifolium and A, trifasciata, such "typical" habitat

 

 

descriptions may be of little use in determining detailed mechanisms of
foraging strategies. Specific habitat and web site selections in
response to specific changes in prey availability will be much more

valuable.

Evolutionary Considerations

 

In conclusion, the foraging specialization model as originally
conceived struck upon some morphological and web design factors important
in shaping foraging success. However, a more precise understanding of
their mechanisms of action is needed in order to indicate other possible
traits that may be linked with them or other functions for which may be
important and constrained by. Furthermore, the additional life-history
constraints affecting prediction of habitat use from morphology
(predation, physiological homeostasis) need to be carefully explored for
these species. At this point, controlled manipulations are needed to
determine functional relationships where possible (e.g. behavioral
changes in web site and design in response to different prey types and
availability), as well as more precise comparative surveys incorporating
information on diets, morphology, behavior and geographical variation.

Certainly the Gea heptagon case indicates that all modes of

 

specialization are not accounted for by the original, one-dimensional
"biter" - "wrapper" dichotomy for even old-field orb-weavers.

Consideration of species from other biomes will be likely to expose

77

further suites of co-adapted characters to the extent that patterns of
interactions between prey types and habitat type exist. The initial goal
of exploring the "biter" - wrapper” model was to determine some simple
correlates between morphology and behavior. The fact that foraging
adaptations are not this simple does not mean a working understanding of
the system cannot be developed. It merely indicates that a more complex
system of spider trait suites and prey-habitat "niches" will have to be
considered. In the case of this set of species, incorporating ESE
haptagon in the scheme requires a 2-dimensional model allowing a suite of
short fangs, short legs, open mesh and low web to be associated with a
weak, slowly escaping prey type. If this modification suffices to
adequately describe orb-weaver foraging types, the system may still prove
tractable enough to begin examining community structure patterns, species
assembly rules, and niche shifting via manipulations and surveys designed
around such a model. The strong interactions of habitat type, height,
and functional prey types may produce some isolated ”adaptive peaks" of
foraging adaptations, as suggested by the higher net energy return rates
in particular height-vegetation combinations. However, this interaction
does not preclude specialization of species on the lower net energy
return "niches.” Consideration of this aspect of the determination of
community structure requires the additional factors of population size
and density as well as dispersal traits to be incorporated into the model
(Diamond, 1975). In short, the data available at this point do not allow
realistic prediction of community structures or assembly rules, but
rather point to the need for further exposition and clarification of the
nature of foraging adaptations and specializations and their potential as

determinants of species interactions and distribution patterns.

LITERATURE CITED

Anderson, J.F. 1970. Metabolic rates of spiders. Comp. Biochem.
Physiol. 33:51-72.

 

Bilsing, S.W. 1920. Quantitative studies in the food of spiders. Ohio
Jo SCio 20:215-600

Brown, K.M. 1979. The foraging ecology of Argiopid spiders. In prep.
Cantrall, I.J. 1943. The ecology of the Orthoptera and Dermaptera of

the George Reserve, Michigan. Misc. Publ. Mus. Zool. Univ. of
MCI]. N00 54 182 pp.

 

 

Comstock, J.H. 1948. The Spider Book. Comstock Publ. Co., Inc.
Ithaca.
740 p.

 

Coville, R.E. 1976. Predatory behavior of the spider wasp, Chalybion
californicum (Hymenoptera: Sphecidae). Pan-Pac. Entom. 52:229-33.

 

 

Denny, M. 1976. The physical properties of spider's silk and their role
in the design of orb-webs. J. Exp. Biol. 65:483-506.

 

Diamond, J.M. 1975. Assembly of species communities. in Ecology and
Evolution of Communities. Ed. by M.L. Cody and J.M. Diamond.
Belknap Press. Cambridge, Mass. pp. 342-444.

 

 

Dorris, P.R. 1970. Spiders collected from mud-dauber nests in
Mississippi. ‘J. Kansas Entomol. Soc. 43:10-11.

 

Eberhard, W. 1970. The predatory behavior of two wasps. Argenoideus
humilis (Pompilidae) and Sceliphron caementarium (Sphecidae) on the
orb-weaving spider Araneus cornutus (Araneidae). Psyche.
77:243-51.

 

 

 

 

Eisner, T., R. Alsop, and G. Ettershank. 1964. Adhesiveness of spider
silk. Science 146:1058-62.

Enders, F. 1974. Vertical stratification in orb-web spiders (Araneidae,
Araneae) and a consideration of other methods of co-existence.

Ecology 55:317-28.

Enders, F. 1975. The influence of hunting manner on prey size,
particularly in spiders with long attack distances (Araneidae,
Linyphiidae, and Salticidae). Am. Nat. 109(970):737-63.

78

79

Ford, M.J. 1977. Energy costs of the predation strategy of a
web-spinning spider (Linyphiidae). Oecologia 28(4):341-9.

Gertsch, W.J. 1949. American Spiders. Van Nostrand. N.Y., N.Y., 285
PP-

 

Haynes, D.L. and P. Sisojevic. 1966. Predatory behavior of Philodromus
rufus Walckenaer. (Araneae: Thomisidae). Can. Entomol.
985113-330

 

 

Kajak, A. 1965. An analysis of food relations between the spiders
Araneus cornutus Clerck and Araneus quadratus Clerck and their prey
in meadows. Ekol. Polska Ser. A 13:717-64.

 

 

 

Kaston, B.J. 1948. Spiders of Connecticut. Conn. Geol. Nat. Hist.
Surv. Bull. 70:1-874.

 

 

,Kurczewski, F.E. and E.J.Kurczewski. 1968. Host records for some North
American Pompilidae (Hymenoptera) with discussion of factors in prey
selection. 'J. Kansas Entomol. Soc. 41:1-33.

 

Levi, H.W. 1968. The spider genera Gea and Argiope in America (Araneae:
Araneidae). Bull. Mus. Comp. Zool. 136:319-52.

 

Levi, H.W. 1971a. The Diadematus group of the orb-weaver genus Araneus
north of Mexico (Araneae: Araneidae). Bull. Mus. Comp. Zool.
141(1):131-79.

 

Levi, H.W. 1971b. The orb-weaver genus Neoscona in North America
(Araneae: Araneidae). Bull. Mus. Comp. Zool. l4l(8):465-500.

 

Levi, H.W. 1973. Small orb-weavers of the genus Araneus North of
Mexico. Bull. Mus. Comp. Zool. 145(9):473-552.

 

Levi, H.W. 1975. The American orb-weaver genera Larinia, Cercidia, and
Mangora North of Mexico. Bull. Mus. Comp. Zool. 147(3):101-135.

 

 

Levi, H.W. 1976. the orb-weaver genera Verrucosa, Acanthepeira,
Wagneriana, Acacesia, Wixia, Scoloderus, and Alpaida north of
Mexico. Bull. Mus. Comp. Zool. l47(8):351-391.

 

 

 

 

Muma, M.H. and W.F. Jeffers. 1945. Studies of the spider prey of
several mud-dauber wasps. Ann. Entomol. Soc. Amer. 38:245-55.

 

Nie, N.H., C.H. Hull, J.G. Jenkins, K. Steinbrenner, and D.H. Bent.
1975. Statistical Package for the Social Sciences. McGraw-Hill.
N.Y., N.Yo 670 pp.

 

Peakall, D.B. 1971. Conservations of web proteins in the spider Araneus
diadematus. J. Exp. Zool. 176:257-64.

 

 

Phelps, L. 1967. On the nature of predation by the spiders Argiope
aurantia Lucas and Argiope trifasciata Forskal. unpub. M.A. thesis.
U. of Wisconsin. Madison, Wis.

 

80

Robinson, M.H. 1975. The evolution of predatory behavior in araneid
spiders. in Function and Evolution in Behavior. Ed. G. Baerends,
Clarendon Press. pp. 292-312.

 

Robinson, M.H. and H. Mirick. 1971. The predatory behavior of the
golden web spider Nephila calvipes (Araneae: Araneidae). Psyche.
78:123-39.

 

Robinson, M.A., H. Mirick, and O. Turner. 1969. The predatory behavior
of some araneid spiders and the origin of immobilization wrapping.
PSYCheo 76: 487-5010

Robinson, M.H. and J. Olazarri. 1971. Units of behavior and complex
sequences in the predatory behavior of Argiope argentata (Fabricius)
(Araneae: Araneidae). Smithson. Contr. Zool. 65:1-36.

 

Robinson, M.H. and B. Robinson. 1970. The stabilimentum of the orb-web
spider Argiope argentata: an improbable defense against predators.
Canad. Entomol. 102:641-55.

 

 

Tolbert, W.W. 1975. Predator avoidance and behaviors and web defensive
structures in the orb weavers Ar iope aurantia and Argiope
trifasciata (Araneae: Araneidae . Psyche. 82(1):29-52.

 

 

Tolbert, W.W. 1976. Population dynamics of the orb-weaving spiders
Argiope trifasciata and Argiope aurantia (Araneus: Araneidae):
Density changes associated with mortality, natality, and migrations.
Ph.D. Thesis. Univ. of Tennessee. Knoxville, Tenn. 172 pp.

 

 

Wiens, J.A. 1977. On competition and variable environments. Am.
Scientist. 65(5):590-7.

Witt, P.N. 1963. Environment in relation to the behavior of spiders.
Arch. Environ. Health. 7:4-12.

 

Witt, P.N. 1971. Instructions for working with web-building spiders in
the laboratory. Bioscience 21(1):23-5.

 

 

MICHIGAN sran UNIV. LIBRRRIES
llWWW!IIIHII”WIWWWIIIIWIHIWIIIHIIWHI
31293100639644