Behavioral Characters for the Higher Classification of Orb-Weaving

Behavioral Characters for the Higher Classification of Orb-Weaving Spiders

Author(s): William G. Eberhard

Source:

Evolution,

Vol. 36, No. 5 (Sep., 1982), pp. 1067-1095

Published by: Society for the Study of Evolution

Stable URL: http://www.jstor.org/stable/2408084 .

Accessed: 11/02/2011 13:36

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .

http://www.jstor.org/action/showPublisher?publisherCode=ssevol. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

Society for the Study of Evolution is collaborating with JSTOR to digitize, preserve and extend access to

Evolution.

http://www.jstor.org

Evolution, 36(5), 1982,

pp.

1067-1095

BEHAVIORAL

CHARACTERS FOR THE

HIGHER CLASSIFICATION OF

ORB-WEAVING SPIDERS

WILLIAM

G.

EBERHARD

Smithsonian

Tropical

Research Institute

and

Escuela de

Biologia,

Universidad

de Costa

Rica,

Ciudad

Universitaria,

Costa

Rica

Received August 25, 1980. Revised June

22, 1981

A number

of studies

have

shown

that

behavior

patterns

can be useful

taxonomic

characters

(see

review

by

Mayr,

1958; also

Evans,

1966;

Crane,

1975;

Michener

et

al.,

1978;

Eickwort

and

Sakagami,

1979;

Greene,

1979).

The webs

and web-build-

ing

behavior

of

orb-weaving

spiders

are

complex,

apparently

stereotyped

charac-

ters,

and as such

offer

promise

of

being

useful

in

indicating

taxonomic

affinities.

As

Levi (1978a,

1978b) has

noted,

how-

ever,

this

promise

has not

been fulfilled.

The

gross,

relatively

superficial

web

char-

acters such

as the

presence

or absence

of

stabilimenta,

or

open

versus closed

hubs

which have been

studied

to date have not

proved

to

be useful

indicators of

subfa-

milial relations.

Webs are

directly

in-

volved

in

orb weavers' interactions

with

a number

of

aspects

of their environments

(particularly

prey),

and

relatively

minor

changes

in

environmental

factors

could

result

in selection

for changes

in

web

form.

Levi

argues

(echoing

Darwin,

1859)

that

at least some

aspects

of web

design

might

thus

be

expected

to be evolutionarily

non-

conservative

and

of little

use

in

indicating

higher

taxonomic

relationships.

This does

not,

however,

eliminate

the

possibility

that webs and web-building

behavior

may

be

useful

as taxonomic

characters.

It

is

possible

that some

details

of web

design

with

apparently

low

func-

tional significance

may

be

more conser-

vative

than other

more

obviously

func-

tional characters.

I

have

the

impression

that

many

web

characters

are

not scat-

tered randomly

among

the

webs

of ara-

neoid

species,

and that their patterns

of

occurrence will

be

of at least some use

in

systematics,

particularly

at

generic

and

tribal levels.

The

analysis

necessary

to

substantiate this has

yet to be performed,

although Risch (1977)

has

made a

prom-

ising start by finding

that two congeneric

species' webs were

more similar to each

other in

a number of

details

than to those

of two species

from

other genera.

It is also possible

that some details of

construction behavior

are

employed by

spiders

in

unchanged

form to construct

a

variety

of different web forms,

and that

these

behaviors

are

thus slow

to evolve

even while

the

webs themselves change

rapidly. This paper

describes several such

behaviors and shows that they are conser-

vative

enough

to

characterize

the classical

subfamily

and

family groupings

which

have been

based on

adult morphology

and

can

thus

be

used

to indicate relationships

between

them,

a topic on which there is

currently substantial

disagreement (e.g.,

Lehtinen, 1967, 1975;

Levi, 1978a, 1980;

Robinson and Robinson,

1978, 1980;

Opell, 1979).

Obviously

the more characters used

in

constructing

a

system

of

relationships

the

greater the likelihood

of the results being

correct.

A

great deal

is

known about the

morphology of orb

weavers, but no com-

prehensive surveys

of characters are yet

available

(but

see

Levi, 1980),

so

they

can-

not yet be included.

This paper is meant

to provide

useful

data

for

later syntheses,

not to give the final

word on the classifi-

cation of

orb

weavers.

Choice

of

Characters

When

one has

collected data

on

a

pre-

viously

untried character or set of char-

acters,

the first

step

in

their

analysis

is to

compare

their

distribution

with

schemes based on other char-

acters.

If

the new

characters

are function-

1067

1068

WILLIAM

G.

EBERHARD

ally

independent of

the older

ones-as

would

appear to be the

case here since

the

morphology of male

and female

genitalia,

eye

positions and

structures,

cheliceral

morphology,

cephalothorax form and

oth-

er

morphological

characters used by

other

workers

would seem to

have

little func-

tional

relation to the

details

of

leg move-

ments

and thread

manipulation

during

web

construction and

attack and

court-

ship

behavior-then there

are

several

pos-

sible

results.

If

the

previous schemes are

all

incorrect,

the

distribution of the

new

characters will not be

in

accord with

any

of

them,

and the new

characters can be

used to construct a new

set

of

relation-

ships.

If,

on

the other

hand, the previous

schemes

are at

least

approximately

cor-

rect,

then the degree

to which the

new

characters "fit" will

depend

on

the rates

of

evolution of these

characters. Those

which

have evolved

very slowly will be

uniform

over

many

different

groups,

those

which

have evolved

relatively rapidly

will

vary

even

within given

taxa,

and still oth-

ers will more or

less match the taxonomic

scheme. Since some of the

behavioral

characters

examined

in

this

study

do have

distributions similar

to

the

classical

group-

ings of

Simon (1892)

based on adult

mor-

phology, this scheme

is probably at

least

approximately

correct.

There were other

characters

which

were

constant

in

all

the

groups

studied

(e.g.,

starting sticky spiral

from the

edge

rather

than

the

center

or

any

other

part

of the

web),

and

others which varied within

giv-

en

subfamilies

(e.g.,

pulling

motions of

legs

IV on

sticky spiral

as

it

was pro-

duced). These behaviors are not included

here since the basic

objective of the

study

was,

after

testing

the

classical

scheme,

to

attempt to use the behavioral

characters

to

indicate

relationships

between

subfam-

ilies and families. Thus

only those char-

acters which

appear

to be

constant

or

nearly

constant

within

subfamilies or fam-

ilies and which

also

differ

between them

are

discussed.

Detailed

descriptions

of

these

characters,

the different states which

they

assume,

their

functional indepen-

dence,

and

the

most

probable

directions

of

transformation

are

given

in

Appen-

dix

1.

MATERIALS

AND

METHODS

Specimens

of

spiders

are

deposited

in

the

Museum

of

Comparative

Zoology,

Cambridge,

Mass.

02138.

The

specimen

numbers

mentioned

in

the

text

and

ap-

pendices

refer to

numbered

labels

includ-

ed

in

individual

spiders'

vials. At

present

it is

unfortunately

impossible to

identify

many

(most?)

orb-weaving

spiders

at

the

species

level, and

most

specimens

are

giv-

en

only

generic

names.

This situation

does

not

seem

likely

to

change

radically in

the

near

future,

and

it

thus

seemed

wise

to

proceed

with

the

presentation

of

the

data

in

this

paper

rather

than

wait for

more

complete

identifications.

Unless

otherwise

noted all

references

to

subfamilial

and

tribal

groupings are

based on

the monu-

mental

work

of

Simon

(1892).

The

ulob-

orid

names follow

Opell

(1979).

The

techniques

of

observation

and

their

limitations

are

described in

the

Appendi-

ces.

As

might

be

anticipated,

many

details

of

web-building behavior are

extremely

stereotyped

within

a

given

species

(see

Appendix

1).

This

uniformity

makes

web-

building

behavior

an

attractive

set of

characters

to

study because

relatively brief

observations

suffice to

characterize

a

species.

RESULTS AND

DISCUSSION

Observations of

at least

148

species

in

at

least

55

genera

are

presented

in

Appen-

dices

2

and

3

and

summarized

in

Table 1.

The

data

are

arranged

according to

the

groupings

of Simon

(1892). Both

substan-

tial

concordance within

subfamily

and

family

groups

with

respect

to a

number

of

characters and

clear

differences

between

subfamilies and

families are

evident.

These

patterns constitute

confirmation of

the

classical

groupings.

The one

distinction

which

is

not

confirmed is

that

between

Tetragnathinae and

Metinae,

as there

were

no

consistent

differences between

species

of

these

two

groups.

SPIDER BEHAVIOR AND

TAXONOMY

1069

TABLE 1. Summary of data in

Appendices

2

and

3

and Robinson

and Robinson, 1980. The data are

relatively

scarce for some characters for

Theridiosomatidae

and

Anapidae

and their characterization here is

tentative.

Parentheses indicate character

states thought to

be

secondarily

derived within the group;

in

all cases

fewer

than 10% of the species observed

for that group have the presumed

secondary

state,

and the species

involved

have all been

classified

without

question on morphological grounds

in the taxa in which they are placed

here.

Further

justifications

for considering

these as convergences rather

than synapomorphies are given in

the text

and Appendices

2

and

3. The numbers of

the

character states

refer to designations

in

Appendix

1 and the

text.

Behavior

Group

A B C D E

F G H I J

Araneinae

2

(4,

"3") 1

1 2

(1, 3)

1 1

(3, 4)

2

&

3

(1)

1

(3)

3

& it

2

(1)

Tetragnathinae-

Metinae

1

(2)

1

1 1

(3)

1

1 2

(1, 3)

1

(3)

2

?1tt

Nephilinae

3

1 1 1 & 2

1

2

1 1 1 1

Theridiosomatidae-

Anapidae

1

& 4

1

1 1 & 3

1

4

&

5

2

& 3

1 & 2 ?

Uloboridae

2

2 2 1 & 2 2

(1)ttt

4 1 1 2

?

t

Gasteracantheae, Micratheneae,

Mastophoreae, Cyrtarachneae,

and Celaenieae

only.

tt

Pers.

observ.

of

Leucauge sp

near venusta

and

Tetragnatha sp. (#0-19-1);

TR

also absent

in

Tetragnatha spp.

and

Pachygnatha

spp.-

see Bristowe, 1958

ttt Hyptiotes spp.

which

spin

reduced, presumably

derived orbs

(e.g., Marples

and

Marples, 1937; Comstock, 1940).

Relationship

between

Uloboridae

and

Araneoid

Orb Weavers

There

has

been

a long

unresolved

con-

troversy

concerning

the possibility

that

orb web

construction,

which

is known

in

six

different

spider

families,

evolved more

than once

(see

Kaston,

1964,

and

Kull-

mann,

1972,

for the

most recent

summa-

ries).

There

is general

agreement

that

the

five araneoid

orb-weaving

families (Ara-

neidae,

Theridiosomatidae,

Anapidae,

Mysmenidae,

and

Symphytognathidae

sensu

Forster

and

Platnick,

1977)

all

evolved

from a single

ancestor,

but

their

relationship

to the

sixth

family,

Ulobori-

dae, is

disputed.

Some

authors

have

thought

that

the

lack

of

clear

synapomor-

phies

shared

by

Uloboridae

and

other,

non-orb-weaving

taxa

plus

the

similarities

between

the

designs

of the

webs

and the

web-building

behavior

of uloborids

and

araneoids

are

so great

as

to make a

com-

mon derivation

inescapable

(e.g.,

Pe-

trunkevitch,

1926;

Wiehle,

1928;

Lehti-

nen,

1967;

Opell,

1979;

Levi,

1980b;

see

also

Brignoli,

1979).

But

in

other,

more

widely

used schemes

(e.g.,

Simon,

1892)

uloborids

are

widely

separated

from

ara-

neoids on the basis of their possession of

a pair

of

structures (cribellum and cal-

amistrum) involved in the production of

one type

of

sticky

silk. The

presence

or

absence

of these structures has

recently

been shown, however,

to be

unreliable

as

an

indicator

of

higher-level relationships

in

several other spider groups (e.g., Lehti-

nen, 1967; Forster, 1970; Kullmann and

Zimmermann, 1976).

The distributions of behavior patterns

in

Table

1

support the idea that Araneidae

is more closely related to Theridiosoma-

tidae

and Anapidae than to Uloboridae

(fragmentary data on Mysmenidae and

Symphytognathidae will be presented

elsewhere;

they suggest

close

ties between

these

families and Anapidae). They do

not

permit

one

to

decide whether

or

not all

four

groups

evolved

from

a common orb-

weaving ancestor,

but

they suggest that

they

did

not.

In

order

to

decide between

a

single

or a dual

origin

of

orb webs,

one

must compare the orb weavers' behavior

with

comparable behavior of non-orb-

weaving

araneoid

groups

such

as

therid-

iids (possible sister groups of araneoid orb

weavers) and non-orb-weaving cribellates

such as

Dictynidae (possible

sister

groups

1070

WILLIAM

G.

EBERHARD

THERIDIOSOMATIDAE

ANAPIDAE

TETRAGNATH

I

NAE

METINAE

GASTERACANTHEAE

AND

REST

OF

MICRATHENEAE

ARANEINAE

HUB

DESTRUCTION

#4&5

ATTACK

WRAP

#3 TEMPOR

&PIRAL

DETERMINE

STICKY

SPIRLL

ATTACHMENT

POINT

#2

MAINTAIN

CONTACT

WITH

TEMPORARY

SPIRAL A$

LAY

STICKY

SPIRAL

I2\

ATTACK

WRAP

#2

TR IN

MALE

COURTSHIP

#2

NEPHI

L

INAE/

DETERMINE

STICKY

SP#3AL

HUB

DESTRUCTION

#2 OR

3

ATTACH

STICKY

SPIRAL

ALL

RADII

ATTACHMENT

POINT

(DTRIESTCYSIA

RADIUS

CONSTRUCTION

#2

ATTACHMENT

POINT

LEGS

HOLD

RN+1

#1

PUSH

STICKY

SPIRAL

WITH

oIV

#1

MAINTAIN

CONTACT

WITH

TEMPORARY

/ ~~~~~~~~~~~~~SPIRAL

AS LAY

STICKY

SPIRAL

#

/ ~~~~~~~~~~~~~(RADIUS

CONSTRUCTION

#1)

/ ~~~~~~~~~~~~~~(HUB

DESTRUCTION

#1)

TEMPORARY

SPIRAL

FORM

#1

A

ATTACK

WRAP

#1

TR IN

MALE

COURTSHIP

#1

FIG.

1.

Schemes

of

relationships

between

araneoid

orb-weavers

favored

by

the

data

in

Table

1.

A)

Most

parsimonious

scheme

assuming

that

lack

of

attack

wrapping

(I1)

is primitive.

B)

Alternative

scheme

assuming

that

attack

wrapping

evolved

independently

in

Theridiosomatidae-Anapidae.

Proposed

synapomorphies

are

underlined,

autapomorphies

are

not.

The

placement

of male

courtship

behavior

is

tentative

since

no

data

are

available

for

some

groups.

The

probable

character

states

of the

ancestral

stock

are

given

for all

characters.

There

is

less

certainty

for some

of

these

than

others

(e.g.,

the

character

states

designated

as

autapomorphic

in A

for

Nephilinae

could

also

be plesiomorphies

and

the

alternative

states

synapomorphies

in the

line

which

of

Uloboridae).

Since

the behaviors ex-

amined here are

mostly

involved

with

orb

web

construction,

this is not

strictly pos-

sible. However some

non-orb weavers

do

attach

sticky

lines to

non-sticky lines, and

the positions of

their legs

III

and IV can

be compared with

characters

B

and C of

the orb weavers.

The following observa-

tions were made

in

an attempt to permit

such

comparisons.

The

mesh-weaving araneoid Achaear-

anea

tepidariorum

(Theridiidae) attached

a line with

sticky silk to a dry line with

the same movements

as those

of

all ara-

neoid

orb-weavers (ipsilateral legs

III

and

IV held the dry

line on either side of the

attachment,

and

the other

leg

IV held the

sticky line), thus

suggesting

a closer rela-

tion of araneoid orb

weavers to Theridi-

idae than

to

Uloboridae.

The

cribellate

mesh-weaver

Mallos gregalis

(Dictyni-

dae), on

the other hand,

did not

use either

of the

hind legs to

stretch -the

cribellum

silk as

it was attached,

thus

resembling

the uloborids.

Both

legs IV seized

the dry

silk

(probably

more than one line-I

could

not convince

myself

on this point),

but in-

stead of holding

one IV anterior to the

oth-

er, they

were both equally

to

the side and

posterior

to

the

spinnerets,

and these

were

then touched

closer

to

one

leg

and some-

what anterior to

it. Thus this species'

be-

havior

was similar

but clearly

not identi-

cal to that

of the

uloborids.

Another

unidentified

dictynid

species was

so small

and

moved

so quickly

that the

details of

its behavior

could

not be distinguished,

but

it

seemed

to behave as

did

M.

gregalis.

SPIDER BEHAVIOR AND TAXONOMY 1071

GASTERACANTHEAE

TETRAGNATHI

NAE METI NAE

AND

NEPH1LINAE

MICRATHENEAE

REST OF

ARANEINAE

\

\f/T~~~~~~~~~TTAC

K

WRAP

#3

\

DETERMINE

STICKY

SPIRAL#f

TACHMENTDEEMNSTCYPRA

POINT

ATTACHMENT

POINT 2

\

ATTACK WRAP #2

RADIUS

CONSTRUCTION

#2

MAINTAIN

CONTACT WITH

\TEMPORARY

SP

IRAL

AS

\ LAY

STICKY

SPIRAL

2

THERIDIOSOMATIDAE

ANAPIDAE

TR IN MALE COURTSHIP #2

K

\ /< ~~~~~~~~~HUB DESTRUCTI

ON

#2

OR

3

DETERMINE

STICKY SPIRAL

ATTACHMENT POINT

#1

\

/ ~~~~~~~~~LEGS

HOLD

R

N+1

#1

\

/

~~~~~~~~PUSH

STICKY

SPIRAL WITH

oIV #1

HUB DESTRUCTION

#4'5

MAINTAIN CONTACT

WI%TM

TEMPORARY

SPIRAL

AS

TEMPORARY

,PIRAL

LAY STICKY SPIRAL

FTORM

#2&

RADI US

CONSTRUCTION

#1

ATTACK WRAP

#2

/HUB DESTRUCTION

#1

FOR

TEMPORARY SPIRAL

FORM

#1

K

2ATTACK

WRAP

#1

/ ~~~~~~~~~~~TR

I N MALE COURTSH IP

#1

ATTACH

STICKY SPIRAL TO ALL RADII #1

B

split

from

Nephilinae);

this uncertainty

is indicated by parentheses.

Species of the araneine

groups

Masto-

phoreae, Cyrtarachneae,

and Celaenieae

are related on morphological

grounds to

Gasteracantheae (Levi,

1978) and also lack

attack wrapping behavior,

and should thus

probably be included

on this basis with

Gasteracantheae

and Micratheneae in the

diagrams. It is possible

however that they

lost attack wrapping

secondarily

as a result of specialization on

moth prey (see Appendix

1). Uloborids differ

from the hypothetical

araneoid ancestors

in

six

of

nine characters.

The attachment behavior of another

crib-

ellate, Tengella sp. (Tengellidae?),

was

identical

to that of

M.

gregalis,

but still

another cribellate,

Filistata

hibernalis,

behaved differently. It lowered both

IV's

to the substrate out of contact

with the

web, and pulled the dry line toward

its

body with one leg

III

and held it

there as

the sticky line was attached.

These limited data on characters

B

and

C

support the classic

dual

origin

hypoth-

esis (unless one makes

the

unlikely

as-

sumption

that theridiids or

dictynids

and

tengellids have secondarily lost

orb webs

[see however Levi,

1980b]).

Araneids are

closer to theridiids than to uloborids and

other cribellates,

and uloborids are closer

to dictynids and tengellids

than to ara-

neids or theridiids. The similarity

between

uloborids

and other

cribellates is only ap-

proximate, however. An

additional reason

to

doubt a close

relationship between

uloborids

and theridiids is

given

in

the

discussion of directions of

change

in

char-

acter

I in

Appendix

1.

It is

worth noting that

a convergent

evolution

of orb web construction

in

two

separate

lines may not be as

improbable

as some authors

suggest,

since an

orb-like

geometry

probably has several

advantages

over

other

web

designs

(Witt, 1965; Lang-

er, 1969;

Eberhard, 1972;

Denny, 1976).

Groups

within Araneoid Orb Weavers

The

data

in

Table

1

could be

interpret-

ed to

support

a

number

of alternative

sys-

tems of

relationships.

If

however one ac-

cepts that 11 is more

primitive

than

12

and

1072

WILLIAM

G.

EBERHARD

13

and that

only in

well-defined

conditions

will a

reversion

occur (see

Appendix

1),

much of

the

ambiguity

disappears. The

system

of

relationships illustrated

in

Fig.

la

seems to

be

favored. It

should be

noted

that

the lack

of

attack

wrapping

(I1) in

Gasteracantheae and

Micratheneae,

which

nevertheless

show

three

characters

which

may

be

synapomorphies

with the

rest of

the

araneines

(Al,

D2, and

J2)

necessitates

an

independent

evolution of

attack

wrap-

ping

in the

groups

"other

Araneinae" and

Tetragnathinae-Metinae.

Postulation of

such

a

convergence

is justified

by

three

considerations: 1)

convergence has

already

been

documented

for

post-attack

wrap-

ping

behavior

in

at

least

four

groups

of

spiders

(Eberhard,

1967;

Rovner

and

Knost, 1974;

Robinson and

Lubin,

1979);

2)

convergence has

also

occurred

in

im-

mobilization

wrapping in

such

diverse

groups as

Pholcidae,

Hersiliidae,

Oecobi-

idae, and

Theridiidae (pers.

observ.); and

3) the details of

araneine attack

wrapping

seem to be

consistently

different

(13)

from

those of

other araneids

(12)

(see

Appendix

3).

The same

general

argument

favoring

independent

evolution

of

attack

wrapping

could also

be applied

to

wrapping

attacks

by

Theridiosomatidae

(still

poorly docu-

mented),

and this

could permit

a

system

of

relationships (Fig.

lb)

more

in

accord

with

other schemes

of

relationships

that

have

been

proposed

on

the basis

of

adult

morphology

and

which is

thus

probably

more

likely

to be

correct.

It

is

interesting

to note that

the

char-

acter

A3

which is

characteristic of Ne-

philinae can be

performed

only

in

webs

in

which

the

separation

between radii is

small

compared to the

span

of

the

spider's

legs,

and

is also

energetically

the least

costly

since the

spider

moves

directly

from

one

attachment to the next

(see

Peakall and

Witt,

1976,

for a

preliminary

discussion

of

the costs of

movements

during

web con-

struction).

This

suggests

that

the

evolution

of

A3

behavior

may

have

occurred

at

the

same

time as or

after the

development

of

tightly

meshed

webs.

All

of the few

known

nephiline

webs-Nephila

madagascaren-

sis

(Wiehle,

1931),

N.

clavipes,

N.

macu-

lata

(Robinson

and

Robinson,

1973),

Her-

ennia

ornatissima

(Robinson

and

Lubin,

1979a),

and

Nephilengys

malabarensis

(pers.

observ.-are

indeed

tightly

meshed,

and

the

Micrathena

species

(gracilis

and

#2200)

which

have

apparently

conver-

gently

evolved A3

behavior

also

make rel-

atively

tightly

meshed

webs and

only

use

this

behavior

in

the

central

portion

where

the

mesh

is

smaller.

Nephiline

radius con-

struction

behavior

(F2),

with

its

strong

emphasis on radii

originating on

the tem-

porary spiral

rather

than at

the hub

may

also

be

adapted

to

produce

tightly meshed

webs

(viz.

similar

radius

patterns

in

the

very

tightly

meshed

webs of

Cyrtophora

and

Mecynogea).

At

least

until

more

ne-

philine

webs

are

discovered

and

de-

scribed, one

can

tentatively

suppose

that

some of

the

unique behavioral

characters

of

this

group

represent

adaptations

spinning

tightly

meshed orbs.

This

interpretation

is

contrary

to

the

schemes of web

evolution

proposed

by

Kaston

(1964)

and

Kullmann

(1972)

be-

cause it

has the

distinctive

nephiline webs

secondarily

derived from

more

typical

orbs

(autapomorphic)

rather than

ancestral to

the first

typical orbs. The

interpretations

of

Kaston and

Kullmann

were

based on

far

fewer

data than

those

available now.

In

addition

both

authors

took

as

primitive

those

araneid webs

that

most

closely

re-

semble the

webs of certain

species

of

the-

ridiids

and

linyphiids.

There is a

great

variety

of web

forms

however

in

these

families,

especially

in

Theridiidae

(see

for

example

Bristowe,

1958;

also

Marples,

1955a, 1955b;

Kullmann,

1970;

Mascord,

1970;

Forster

and

Forster,

1973;

Eber-

hard,

1977b,

1979;

Carico,

1978;

Clyne,

1979;

pers.

observ. of

Chrysso

spp.

and

Chrosiothes

sp.).

These authors

give no

reason

for

assuming

links

to the

particular

species

they

chose

(other

than that

the

species are common

temperate

forms and

thus better

known).

Their

decision to con-

sider

certain

araneid web

characters

prim-

itive

thus seems

arbitrary.

In

addition the

web character on which

both

place

emphasis-the

presence

of a

mesh

or "barrier"

web on one or

both

sides

SPIDER

BEHAVIOR AND

TAXONOMY

1073

of

the orb-would

seem to have

a

number

of

possible

functions (e.g., Hingston,

1922c; Robinson

and Robinson, 1973; Lu-

bin, 1975) and to

be

a

good example

of

the

non-conservative

characters mentioned by

Levi (1978a). The

presence of barrier webs

near the orbs of

such diverse genera as

Metepeira

(Araneinae, Araneae-e.g., Si-

mon, 1892), Spilasma

(Araneinae, Ara-

neae-pers.

observ.), Arachnura (Ara-

neinae,

Arachnureae-Main, 1976),

Argiope

(Araneinae, Argiopeae-e.g., Lu-

bin, 1975), Gasteracantha

(Araneinae,

Gasteracantheae-pers. observ.),

Leu-

cauge

(Metinae-e.g., Comstock, 1948),

and the uloborid genera

Philoponella (Pe-

ters, 1953;

Eberhard, 1969;

Lahmann

and

Eberhard, 1979;

Opell, 1979) and Ulobo-

rus (Lubin et al.,

unpubl.) as well as in

the nephiline genera

Nephila (Nephileae)

and Phonognatha

(Phonognatheae-

Main, 1976)

indicates

that the

presence

of

a

mesh

cannot be

taken

as a reliable

in-

dication

that

other

aspects

of that

species'

web

are

primitive.

A

further difference

between the scheme

proposed here

and

those

of Kaston and

Kullmann is my

placement of Cyrtophora

and

Mecynogea

(=Allepeira)

as

a

deriva-

tive group within

the Araneinae, far

from

the ancestral stem

of

the orb weavers. The

objections to

their

choice

of

primitive

character states

apply

here also.

In

addi-

tion,

while the

building

behavior data are

limited (these genera do not

spin sticky

spirals

and have

distinctive

radius

con-

struction

behavior),

the existence

of both

F1

and F3 radius construction behavior

in

Argiope

anasuja,

a

member

of

a

genus

closely related

to

Cyrtophora

in

its adult

morphology (Levi,

1978),

attack behavior

(Robinson, 1975),

and

egg

sac form

(Si-

mon, 1892)

is

in

accord with

a

derived

position

for

Cyrtophora

and

Mecynogea.

Robinson and Robinson

(1978) suggested

placing Argiope near the base

of

the ara-

neid line on the

basis of mating behavior

(mating site).

Their

argument

is not con-

vincing

since this character seems to be

non-conservative

(see Appendix 3)

and

since,

as

they

themselves

point out,

it

is

not

possible

to establish whether the more

simple type

of

courtship

is

primitive

or

derived. They imply that Argiope may be

close to Nephilinae, but it is clear that the

data

here do not support such a grouping.

Attack behavior (Robinson, 1975; Appen-

dix

3)

and

the occurrence of tarsal rubbing

in male courtship as well as other details

of

mating

behavior

(Robinson and Rob-

inson, 1980) also argue for separation of

the two groups and a relatively more de-

rived position for Argiope.

Some

authors (e.g., Kaston, 1948; For-

ster, 1967; Lehtinen, 1967, 1974)

have

placed tetragnathines in a separate family,

but

the data

here could only support

this

classification

if

metines

were

also placed

in

this

separate family and

if

the hitherto

unquestioned

araneid

group Nephilinae

were

put

in

its

own

family. Lehtinen (1975)

proposed, on the basis of the structure

of

genital organs,

trichobothrial

patterns

on

legs, and color patterns (no precise data

were given however), that Metinae and

Leucauginae be split from Tetragnathidae

and combined with

Nesticidae and

Mi-

metidae

in

a family separate from both

Araneidae and Tetragnathidae. The data

here

do

not

support

this

split since

Leu-

cauge

and

other metines seem identical

to

the

tetragnathines

in

all the

characters

ex-

amined.

In

addition

they

share the con-

struction

of

orb webs with

Tetragnathinae

but

not

Nesticidae

and

Mimetidae.

One

would have

to

postulate

a

secondary

loss

of orb webs

in

nesticids and

mimetids

to

justify grouping

them with

the

orb weav-

ers.

SUMMARY

Some

details of orb web

construction

and attack behavior

are

evolutionarily

conservative

and

appear

to be useful

in

defining

subfamilies and families and

de-

termining relationships.

Their

patterns

of

distribution

among

the

at least 148

species

in

at least 55

genera surveyed

here

agree

in

general

with

classical taxonomic

schemes based

on

adult

morphology.

The

data

suggest

that

convergent

evolution

of

orb

webs

may

have

occurred

in

two lines

(uloborids

and

araneoids). They

also

in-

dicate

that several

previous proposals

re-

1074

WILLIAM

G.

EBERHARD

garding the evolution of orb weavers and

their webs may be incorrect. Certain be-

haviors appear to constitute autapomor-

phies for Uloboridae, Nephilinae, and

Araneinae, while others may be synapo-

morphies for Theridiosomatidae-Anapi-

dae.

ACKNOWLEDGMENTS

Dr.

Herbert W. Levi spent many hours

identifying long series of araneid spiders

for

me. Without his dedicated and invalu-

able aid this

project would

never have

been begun.

I

hope that the results pre-

sented here help make his

effort

seem

worthwhile.

I

am also grateful to Drs. Norman Plat-

nick and Brent Opell for identifications of

anapids

and

uloborids respectively,

to

Jonathan Coddington

for tentative names

for

theridiosomatids,

and

Dr. P. N.

Witt

for

sending specimens

of Mallos

gregalis.

I

benefitted greatly from timely discus-

sions with

Dr.

Platnick and

J. Codding-

ton. They, along

with H.

W. Levi,

C.

E.

Valerio,

B. P.

Opell

and

M.

J. West

Eber-

hard read preliminary versions of the

manuscript.

The observations were made over a

span

of about

10

years, during

which time

I

received aid from many people.

I

am

especially grateful

to Central

Hidroelec-

trica

Anchicaya', Carlos Rodriguez,

the

Dixon Stroud family, and Drs. Luis

Ar-

ango,

Carlos

Valerio,

Madhav

Gadgil,

and

A.

J.

T.

Johnsingh for help

and hos-

pitality.

The work was financed by the Comite

de

Investigaciones of

the Universidad del

Valle, Cali, Colombia,

and the Vicerec-

toria de

Investigacion

of the Universidad

de Costa

Rica.

Note

added

in

proof:

Fukumoto

(1973,

1981), working

with

Nephila, Tetragnatha,

and

five

species

in

four

araneinine

genera

(probably

all

species

were different

from

those

in

this

study) recognized

some

of the

similarities and differences

in

sticky spiral

construction

(characters A, B,

and

C)

which are documented

here.

I

thank

M.

Stowe for

kindly translating

these articles

from

Japanese.

LITERATURE

CITED

BRIGNOLI,

P.

M. 1979.

Contribution

i

la

connais-

sance

des

Uloboridae

palearctiques

(Araneae).

Rev.

Arachnol.

2(6):275-282.

BRISTOWE,

W.

S. 1958.

The

World

of

Spiders.

Col-

lins,

London.

304

p.

CARICO,

J. E. 1978. Predatory

behavior

in

Euryopis

funebris

(Hentz)

(Araneae:

Theridiidae)

and

the

evolutionary

significance

of

web

reduction,

p.

51-58.

In

P.

Merrett

(ed.),

Arachnology,

Symp.

Zool.

Soc.

London

42.

CLYNE,

D.

1973.

Notes

on the

web

of

Poecilopachys

australasia

(Griffith

and

Pidgeon,

1833)

(Aranei-

da:

Argiopidae).

Aust.

Ent.

Mag. 1(3):23-29.

.

1975.

Come

into

my parlour,

said

the spi-

der;

techniques

of spider

predation.

Mantis

Wild-

life

Films,

Turramurra

2074,

N.

S.

W.,

Australia.

1979.

The

Garden

Jungle.

Collins,

London.

184 p.

COMSTOCK,

J.

H. 1940.

The Spider

Book.

Revised

and

edited by

W.

J.

Gertsch.

Comstock

Pub.

Assoc.,

Ithaca.

729 p.

CRANE,

J.

1975.

Fiddler

Crabs

of

the

World

(Ocy-

podidae:

Genus

Uca).

Princeton

Univ.

Press,

Princeton.

736 p.

CROME,

W.

1954.

Beschreibung,

Morphologie

und

Lebensweise

der

Eucta

kaestneri sp.

n.

(Araneae,

Tetragnathidae).

Zool.

Jb.

(Syst.)

82:425-452.

DARWIN,

C.

1859.

On

the

Origin

of

Species

by

Means

of

Natural

Selection,

1st ed. John

Murray,

London.

503

p.

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.

EBERHARD,

W.

G.

1967.

Attack

behavior

of

digue-

tid

spiders

and

the

origin

of

prey

wrapping

in

spiders.

Psyche

74:173-181.

.

1969.

The

spider

Uloborus

diversus Marx

and its

web.

Ph.D. thesis.

Harvard

Univ.,

Cam-

bridge.

.

1972.

The

web

of

Uloborus

diversus

(Ara-

neae:

Uloboridae).

J.

Zool.,

London

166:417-

465.

.

1976.

Physical

properties

of sticky

spirals

and

their

connections: sliding

connections

in

orb

webs.

J.

Nat.

Hist.

10:481-488.

.

1977a.

Aggressive

chemical

mimicry

by

a

bolas

spider.

Science

198:1173-1175.

.

197

7b.

'Rectangular

orb'

webs

of

Synotaxus

(Araneae:

Theridiidae).

J.

Nat.

Hist.

11:501-507.

.

1979.

Argyrodes

attenuatus (Theridiidae):

a web

that

is

not a snare.

Psyche

86:407-413.

.

198 1a.

The

single

line web

of Phoron-

cidia

studo

Levi: a

prey

attractant

(Araneae,

Theridiidae)? J.

Arachnol.

9:229-232.

1981b.

Notes

on

the natural history

of

Taczanowskia

sp.

(Araneae:

Araneidae).

Bull.

Br.

Arachnol.

Soc.

5(4):175-176.

.

1981c.

Building

behavior

and web

ten-

sions

in orb

weavers.

Bull.

Br. Arachnol.

Soc.

5(5):

189-204.

EICKWORT,

G.

C.,

AND

S.

F.

SAKAGAMI.

1979.

A

classification

of nest

architecture

of

bees

in

the

tribe

Augochlorini

(Hymenoptera:

Halictidae;

SPIDER

BEHAVIOR

AND

TAXONOMY

1075

Halictinae), with

description of a

Brazilian nest

of Rhinocorynura

inflaticeps.

Biotropica

11:28-

37.

EVANS, H. E.

1966. Comparative Ethology

and

Evolution of the

Sand Wasps. Harvard

Univ.

Press, Cambridge.

526 p.

FORSTER, R. R.

1967. The Spiders of

New Zealand.

Part I. Otago

Mus. Bull. 1:7-124.

.

1970.

The Spiders of New Zealand.

Part

III. Otago Mus. Bull. 3:11-184.

FORSTER, R. R.,

AND L. M. FORSTER.

1973. New

Zealand Spiders.

Collins, London.

254 p.

FORSTER,

R.

R.,

AND

N. I.

PLATNICK. 1977. A

re-

view of

the

spider

family Symphytognathidae

(Arachnida,

Araneae). Amer. Mus.

Novit.

2619:1-

29.

GORNER, P.

1966. Uber die Koppelung

der

optisch-

en

und kinasthetischen

Orientierung

bei

den

Trichterspinnen

Agelena labyrinthica (Clerck)

und

Agelena gracilens

C.

L.

Koch.

Z.

vergl. Physiol.

53:253-276.

GREENE, A. 1979.

Behavioral characters

as indi-

cators of yellow

jacket phylogeny (Hymenoptera:

Vespidae).

Ann.

Entomol.

Soc.

Amer.

72:614-

619.

HINGSTON,

R. W.

G. 1922a. The snare

of the giant

wood

spider (Nephila

maculata).

Part I.

J.

Bom-

bay Nat. Hist.

Soc. 28(3):642-649.

. 192

2b.

The snare of the giant

wood spider

(Nephila

maculata).

Part II.

J.

Bombay Nat.

Hist. Soc. 28(4):911-917.

.

1922c. The snare

of

the giant

wood

spider

(Nephila

maculata).

Part III.

J.

Bombay Nat.

Hist. Soc. 28(4):917-923.

HOLM, A.

1939.

Beitriige zur Biologie

der Theridi-

iden. Festr. Strand

5:56-67.

JACOBI-KLEEMANN,

M. 1953. Ueber die

Locomo-

tion

der Kreuzspinne

Aranea

diademata

beim

Netzbau (nach

Filmanalysen).

Z.

vergl. Physiol.

34:606-654.

KASTON, B. J. 1948.

Spiders of Connecticut.

Conn.

St.

Geol.

Nat. Hist.

Survey

Bull.

70:5-874.

.

1964.

The evolution of spider

webs. Amer.

Zool. 4:191-207.

KULLMANN,

E. 1971. Bemerkenswerte

Konvergen-

zen im Verhalten

cribellater und ecribellater

Spinnen.

Freunde

Kolner

Zoo

13(4):123-150.

. 1972.

The

convergent development

of orb-

webs

in cribellate and ecribellate

spiders.

Amer.

Zool. 12:395-405.

KULLMANN, E.,

AND

W.

ZIMMERMANN.

1976.

Ein

neuer

Beitrag zum Cribellaten-Ecribellaten

Problem: Beischreibung

von

Uroecobius

ecribel-

latus n. gen.

n. sp. und Diskussion

seiner phy-

logenetischen

Stellung.

Ent. Germ.

3(1/2):29-40.

LAHMANN, E.,

AND

W. EBERHARD.

1979.

Factores

selectivos

que

afectan

la tendencia

a

agruparse

en la

arafia

colonial

Philoponella

semiplumosa

(Araneae;

Uloboridae).

Rev.

Biol.

Trop.

2

7(2):23

1-240.

LEHTINEN,

P.

1967.

Classification

of the cribellate

spiders

and

some allied

families,

with notes on

the

evolution

of the suborder

Araneomorpha.

Ann.

Zool.

Fenn. 4:199-468.

.

1975. Notes on the

phylogenetic classifica-

tion

of

Araneae.

Proc. 7th Int.

Arachnol.

Congr.

:26-2

9.

LEVI, H. W. 1968. The

spider genera Gea and Ar-

giope

in

America

(Araneae: Araneidae). Bull.

Mus. Comp. Zool.

136:319-352.

.

1978a. Orb-webs

and phylogeny of orb-

weavers, p.

1-15.

In

P.

Merrett (ed.), Arachnol-

ogy, Symp. Zool.

Soc.

London 42.

.

1978b. Orb-weaving

spiders and their webs.

Amer. Sci. 66:734-742.

.

1980a. The

orb-weaver genus Mecynogea,

the subfamily Metinae, and

the genera Pachy-

gnatha, Glenognatha,

and Azilia of the subfamily

Tetragnathinae north

of Mexico. Bull. Mus.

Comp. Zool.

149:1-75.

.

1980b. Orb-webs:

primitive or specialized.

Proc' 8th Int. Arachnol.

Congr. (Vienna):367-

370.

LONGMAN, H. 1922. The

magnificent spider:

Di-

chrostichus magnificus

Rainbow.

Proc.

Roy. Soc.

Queensland

33:91-98.

LUBIN,

Y.

D.

1973.

Web

structure and function: the

non-adhesive

orb-web of

Cyrtophora

muluccensis

(Doleschall)

(Araneae:Araneidae). Forma

et

Functio 6:337-358.

.

1975. Stabilimenta

and barrier webs in the

orb

webs

of

Argiope argentata

(Araneae,

Aranei-

dae)

on

Daphne

and

Santa Cruz

Islands,

Gala-

pagos. J.

Arachnol.

2:119-126.

LUBIN,

Y.

D.,

W. G.

EBERHARD,

AND

G. G.

MONT-

GOMERY.

1978. Webs of

Miagrammopes (Ara-

neae: Uloboridae)

in

the

neotropics. Psyche

85:1-

23.

MAIN,

B.

1976. Spiders.

Collins, London. 296 p.

MARPLES, B. J. 1955a.

A

new type of web spun by

spiders

of the

genus

Ulesanis

with the

description

of

two

new

species.

Proc. Zool. Soc. London

125:75

1-760.

1955b. Spiders

from Western Samoa. J.

Linn. Soc.

Zool. 42:453-504.

.

1962. Notes on

spiders of the family Ulo-

boridae.

Ann. Zool.

(Agra.) 4(1):1-11.

MARPLES, M. J.,

AND

B. J. MARPLES. 1937. Notes

on

the

spiders Hyptiotes

paradoxus

and

Cyclosa

conica. Proc. Zool. Soc.

London Series

A

1937:2

13-22 1.

McKEOWN, K. C. 1952.

Australian Spiders. Angus

&

Robertson,

London.

287

p.

MASCORD,

R.

1970. Australian

Spiders.

Ch. E.

Tut-

tle

& Co., Rutland,

Vt.

MICHENER,

C.

D.,

M.

L.

WINSTON,

AND

R. JAN-

DER. 1978. Pollen

manipulation and related ac-

tivities

and

structures

in

bees of

the

family Api-

dae.

Univ. Kansas

Sci.

Bull.

51:575-601.

MUMA,

M.

1971.

Biological

and

behavioral notes

on

Gasteracantha

cancriformis

(Arachnida:

Aranei-

dae).

Fla.

Entomol.

54:345-351.

OPELL, B. P. 1979. Revision of

the genera and trop-

ical

American

species

of the

spider family

Ulo-

boridae.

Bull. Mus.

Comp. Zool. 148:443-549.

PEAKALL,

D.

B.,

AND

P.

N. WITT.

1976.

The en-

ergy budget

of an

orb

web-building spider. Comp.

Biochem. Physiol.

54A:187-190.

1076

WILLIAM

G.

EBERHARD

PETERS,

H. 1931.

Die Fanghandlung

der

Kreuz-

spinne

(Epeira

diademata

L.).

Experimentelle

Analysen

des

Verhaltens.

Z.

vergl.

Physiol.

15 :693-748.

.

1953.

Beitriige

zur

vergleichenden

Etholo-

gie

und Oekologie

tropischen

Webespinnen.

Z.

Morph.

Okol.

Tiere

42:278-306.

.

1954.

Estudios

adicionales

sobre

la

estruc-

tura

de la

red concentrica

de

las arafias.

Comun-

ic. Inst. trop.

Invest.

cien.

Univ.

Salvador 3(1):

1-

18.

PETRUNKEVITCH,

A. 1926.

The value

of instinct

as

a taxonomic

character

in

spiders.

Biol.

Bull.

50:427-432.

RISCH,

P. 1977.

Quantitative

analyses

of

orb-web

patterns

in four species

of spiders.

Behav.

Genet.

7:

199-238.

ROBINSON,

M.

H.

1969.

Predatory

behavior

of

Ar-

giope

argentata

(Fabricius).

Amer. Zool.

9:161-

173.

.

1975.

The

evolution

of

predatory

behavior

in

araneid

spiders,

p.

292-312.

In

G. Baerends,

C.

Beer,

and

A. Manning

(eds.),

Function

and

Evolution

in

Behaviour.

Clarendon

Press,

Ox-

ford.

ROBINSON,

M.

H.,

AND Y.

D.

LUBIN. 1979a. Spe-

cialists and

generalists:

the

ecology

and behavior

of some

web-building

spiders

from

Papua

New

Guinea.

I. Herennia

ornatissima,

Argiope

ocy-

aloides,

and

Arachnura

melanura (Araneae:

Ar-

aneidae).

Pac. Ins. 21:97-132.

.

1979b.

Specialists

and

generalists:

the ecol-

ogy

and

behavior of

some web-building spiders

from

Papua

New

Guinea.

II.

Psechrus

argenta-

tus

and Fecenia

sp. (Araneae:

Psechridae).

Pac.

Ins.

21:133-164.

ROBINSON,

M. H.,

AND

J. OLAZARRI.

1971.

Units

of behavior

and

complex

sequences

in the

pred-

atory

behavior

of

Argiope

argentata

(Fabricius):

(Araneae:

Araneidae).

Smithsonian

Contr.

Zool.

65:1-36.

ROBINSON,

M.

H.,

AND

B. ROBINSON.

1973. Ecol-

ogy

and

behavior of the

giant

wood

spider

Ne-

phila

maculata

(Fabricius)

in New Guinea.

Smithsonian

Contr.

Zool. 149:1-76.

.

1975.

Evolution

beyond

the orb web:

the

web of the

araneid

spider

Pasilobus

sp.,

its struc-

ture, operation

and construction.

Zool.

J.

Linn.

Soc.

56(4):301-314.

.

1978.

The evolution

of courtship

systems

in

tropical

araneid

spiders,

p.

17-29.

In

P. Merrett

(ed.),

Arachnology,

Symp.

Zool.

Soc. London

42.

.

1980. Comparative

studies

of the courtship

and

mating

behavior of

tropical

araneid

spiders.

Pac. Ins.

Monogr.

36:1-218.

ROBINSON,

M. H., B. ROBINSON,

AND

W. GRANEY.

1974. The

predatory

behavior

of

the nocturnal

orb web

spider

Eriophorafuliginea

(C.

L.

Koch)

(Araneae:

Araneidae).

Rev.

Per. Entomol.

14(2):304-3

15.

ROVNER, J.,

AND

S.

J.

KNOST. 1974. Post-immo-

bilization

wrapping

of

prey

by lycosid

spiders

of

the herbaceous

stratum.

Psyche

81:398-415.

SEYFARTH,

E.,

AND

F.

BARTH.

1972.

Compound

slit sense organs

on the

spider

leg:

mechanore-

ceptors involved

in

kinesthetic

orientation.

J.

Comp. Physiol. 78:176-191.

SIMON,

E. 1892. Histoire Naturelle des

Araignees.

I.

2nd ed. Lib. Encycloped. Roret,

Paris. 1084 p.

STOWE, M. 1978. Observations of two

nocturnal orb

weavers

that

build

specialized

webs:

Scoloderus

cordatus and Wixia ectypa (Araneae:

Araneidae).

J. Arachnol. 6:141-146.

WIEHLE, H. 1928.

Beitriige

zur Biologie

der Ara-

neen, inbesondere zur

Kenntnis des Radnetz-

baues. Z. Morph.

Okol. Tiere 15:262-306.

.

1931. Neue Beitriige zur

Kenntnis des

Fanggewebes der Spinnen aus den

Familien

Ar-

giopidae, Uloboridae, und Theridiidae.

Z.

Morph.

Okol. Tiere

22:348-400.

WITT, P. N. 1965. Do we live in

the best of all

possible

worlds?

Spider

webs

suggest

an answer.

Perspect. Biol. Med. 8:475-487.

WITT, P. N., C. REED,

AND

D. B.

PEAKALL.

1968.

A

Spider's Web. Springer,

N.Y.

107

p.

Corresponding

Editor:

D.

J.

Futuyma

APPENDIX 1.

Behavior

descriptions

and probable

directions of change

Spiders

were observed

in the wild and in a

large

screen

cage.

Those building

in

the

dark were illu-

minated at least

periodically

with a headlamp.

Super

8

movies were

made of several

species

as

they

spun

sticky spiral

(see Appendix

2) and

were analyzed

frame by frame.

The positions

of web threads

were

sufficiently

clear

in

the movies to

give good

confi-

dence that the

drawings made

from them (Figs.

2-4)

are

precise,

and enough repetitions

observed

to

as-

certain that

the behavior described

is

typical

and

highly stereotyped.

In

no

case

did

I

see

significant

variation among

individuals of a single species

in the

behavior patterns

described

here except

for the

dif-

ferences

(noted

in

Appendix

2)

between

young

and

old

individuals

of

some

Nephilinae

and two

species

of

Micrathena in character

A

and some

araneines

in

character

D. Most observations

were

of mature fe-

males.

The characters

are denoted

by letters,

and the

character states

by

numbers (e.g.,

character

B

has

two states,

B1

and B2).

The descriptions

are

orga-

nized with respect

to the stage

of web construction

in

which the

behavior occurs.

The probable

func-

tions of some

of the behaviors are

discussed else-

where

(Eberhard,

1981c).

I. Sticky Spiral

Construction

All orb weavers

whose behavior is known

place

the

sticky

thread

on

their webs after

building

a "scaf-

fold" of

radii,

frame

lines,

hub,

and

(with

some ex-

ceptions)

temporary

non-sticky spiral

lines.

They

start the sticky spiral

near

the

edge

of

the web and

gradually

work

inward.

Figure

1

illustrates

this

pro-

cess,

and

gives

the

names

used

for the various lines

SPIDER

BEHAVIOR AND

TAXONOMY

1077

and

legs

used

in

the behavioral

descriptions

which

follow.

A. Determination

of sticky spiral attachment

point

A1:

Tap forward

with

leg

iI

(Fig. 2).

The

spider

faced inward as she attached to

rN, proceeded

toward

the

hub,

and

then

turned

to

face out the

(rN+1)

and moved

forward rather than

sideways out

to

encounter the

inner loop, finally turning

1800

to

attach

again. This

abrupt turning to face

inward

and

then outward was

especially

dramatic as the

spider

laid sticky spiral

near the hub. As the spider

moved

out

rN+l,

iI

tapped

or

pushed forward to contact the

inner

loop while ol was held in

nearly

the same

po-

sition but

was involved

in

gripping

the radius.

A2: Tap

sideways with leg oI (Fig. 3).

The spider

moved sideways

away

from

the hub along

rN+l

wav-

ing

or

tapping

repeatedly

with

the outer of her front

legs which was directed

laterally to

the side toward

which she

was

moving (Fig. 2).

Her

body

was

thus

oriented

perpendicular

to

the

radius,

and her outer

leg

I

was more or

less parallel to it.

A3: Extend leg

oIV backward (Fig. 4). Instead of

moving inward

toward the hub and then

back out-

ward,

the

spider

moved

almost

directly

from

one

attachment to the

next, sidling

across the web as she

faced inward toward the hub. The front

legs

were

not brought near the site

where an attachment

was

about

to

be

made, although

oII sometimes

tapped

toward the inner

loop

several radii

in

advance

of

the

current

attachment

site. The

spider used

oIV

to

probe

for

the point

where

rN+1

intersected

the

inner

loop.

A4: No contact.

The spider's body was small com-

pared

to

the

spaces

between radii and between

loops

of

sticky spiral.

After

attaching

to

rN

and

moving

inward

to

reach

rN+l,

the

spider

faced outward

as

she moved

away

from the hub

along

rN+l,

but

stopped

when still

several

body lengths

short

of the inner

loop

and

turned

to attach.

B.

Legs holding

rN+1

near

attachment

point

when

attachment made

B

1:

oIII

just

inside the attachment

point,

oIV

just

outside

it.

As the

spider

neared

the attachment

point,

leg

oIII

seized

rN+1

just

inside

the

point

where

the

attachment was to

be made.

Then as the

spider

po-

sitioned

her

abdomen to make the

attachment,

oIV

grabbed the radius

just

outside the attachment

point

(Figs. 2, 3, 4).

B2: oIV just

outside

the

attachment point

and

iIV

just

inside

it. The

spider

ceased

combing

sticky

cri-

bellum

silk

with her

legs

IV

just

before

making

the

attachment,

and

gripped rN+1

on either side

of

the

attachment

point

with these

legs

as she

faced

inward.

C. Push

current

segment

with iIV

just

before

at-

taching

C1: Push the

line. Leg

iIV

was extended

(Fig. 4)

and/or bent

ventrally (Fig. 2) just

before the attach-

ment was

made;

in

each case the

tip

of the

leg

moved

away

from

the

spinnerets,

so

additional

line

was

probably

drawn out

(see Eberhard,

1981

c).

In

TEMPORARY SPIRAL

-

- jV

\

INNER

LOOP

CURRENT SEGMENT

F,,

RN

RN+1

FIG. 1.

Path

taken by a

typical

araneid as it

moves

between one

attachment

of the

sticky spiral

and the next

(sticky

spiral lines

shown

thicker

than

others).

The

line

is

attached to

radius rN at

point x,

and

then the

spider

moves (dotted

line) to

radius

rN+1

via the

outermost

loop

of

temporary spiral. It then

moves

out

rN+j

and

attaches the

current

segment of

sticky spiral

(the

line which it

has produced as it

moves) at

point y.

The size of

the spider

compared

to the

distances

between radii and

between the

sticky

spiral

loops

is

relatively

constant

within

species but

varies

widely between them. The

legs designated as

inner

(i)

are

closer to the

hub,

and

the outer

(o)

ones

closer to the

web's

edge.

When

the spider doubles

back

(i.e.,

moves

clockwise

instead of

counterclock-

wise

in

the

figure),

i

legs

become

o

legs and vice

versa.

Nephila

clavipes,

Micrathena

gracilis and

M. sex-

spinosa

I

was

able to

ascertain

that the

exact point

where the

sticky

line

contacted

the

leg

was

variable,

and

was

usually

on

the side rather

than the

tip

of the

tarsus.

It

appeared

that

the line

snagged

on

the stiff

hairs

covering

the

sides

of the

leg.

C2:

Sticky line not

pushed with

legs. The

current

segment

was

not touched

by

any leg

as it

was at-

tached to

rN+l.

D.

Contact with

temporary spiral

as lay

sticky spiral

Dl:

Lose

contact. The

spider moved

beyond the

outermost

loop

of

temporary spiral

and was com-

pletely out

of

contact with it as she

attached

at least

some

of

the

outermost

loops

of

sticky spiral.

D2: Maintain

contact. The spider

maintained con-

tact with the

temporary spiral

at all

times while

spin-

ning

sticky spiral.

The

outline

of

the area covered

by

sticky

silk

thus more or less

reflected the

outline

of

the

outer

loop

of

temporary spiral.

D3: No

temporary

spiral. There

was no

temporary

spiral

in

the web.

1078

WILLIAM

G. EBERHARD

3

6

2

20

tsp

olV

t

4X

inner

oi

Rn

Rn.1l

Rn

Rn+o

16

17

20

21

tsp

tst

oil

ii

'~~~current

o

ol

segment

o lR

inner

lo

oil

l

oop-

i

Rn

Rnw+s Rn

Rn+1

22

24

25

tp

30

tsp

II ~~~~~~~~~~IIV

olil

IV

ollI-

~~~~~~~~~~~~~current

inner

oil

current)

oIl

sd6ment-en

loop-,0

o

emn

oIl oV

0lV

I

'Rn

Rni1

Rn

Rn+1

FIG.

2.

Al

determination

of

sticky

spiral

attachment

point

behavior,

illustrated from

films of

Leucauge

sp.

(#1556) taken

looking

down

at the

spider

as

it

moved

counterclockwise

on a horizontal web.

Conventions

are

as described for

Figure

3.

Body:

faced inward

as

attachment

was

made

to

RN (2)

and as

the

spider moved

inward

along

RN

(3,

6);

then turned

to

face

more

or less

along

the

temporary

spiral

(11),

and

then

away

from

the hub

as the

spider

moved out

RN+1

(16-2

1);

finally

it

turned

1800

briskly (2 2) just

after

leg

ii

touched

the

inner

loop

and

stayed

in

this

position

as

the

attachment was

made.

Leg

pi:

held

RN

as the

attachment

was

made;

after

variable behavior while the

spider

moved in

and across the

temporary

spiral,

it

was

extended

forward and

was bent

ventrally

at the mt-tibia

joint

as 'the

spider

moved

out

RN+1

(16,

17)

until

it

made

contact

with

the

inner

loop (20),

then

swung

back

to

seize

RN+1

(24)

as the

spider

turned

to

make

the

attachment.

Legs

ol and

oil: held

nothing

while the

attachment

was

made

(2);

waved

and

contacted

RN+1

as the

spider

moved

inward toward

the

hub and

across

the

temporar-y spiral bridge,

then

moved

along

RN+s

in

a

"hand-over-hand"

pattern (e.g.,

17, 20);

as soon

as

leg

il

touched

the

inner

loop, leg

oI

swung

to

the

side

as the

spider began

its 1800 turn

(21,

22),

but

leg

oil

retained its hold on

RN+1

until

oII

and

oIV

had

seized

this

thread

(24),

and then it

also

swung

to the

side

(25).

Leg

oIV:

held

RN just

beyond

the

attachment

point

as the

attachment was

made

(2);

retalned its hold

on

RN

as

the

spider

moved inward

(3,

6),

then seized

the

current

loop

of

sticky

spiral (16),

pulled

it

once

(17),

then held

it

as the

spider

moved

out

RN?l,

finally

releasing

its hold

(22)

and

moving

to seize

RN+S

just

outside the

point

where

oil

was

holding

it

(24);

it

maintalned

this

hold until the

attachment

was made.

Leg

iIV: after

releasing

the

sticky spiral

line

(generally

"plucking"

it with

a

snap--3),

moved

irregularly

until

oIV released

its hold on the

current

segment (24)

and,

just

before the

attachment was

made,

pulled

out

more

silk

by

extending

with a

straightening

at

the

femur-

patella-tibia

joints

(30).

SPIDER

BEHAVIOR

AND

TAXONOMY

1079

E. Attach sticky

spiral

to each radius

crossed

El: Attach

to each one.

The

spider

attached the

sticky line to

each radius

she crossed

as she moved

around the

web. Apparent

exceptions

occurred

only

on webs

in

which all radii

were not in the same

plane

and the spider

did not

"encounter" all radii

which

crossed her

path.

E2: Skip

some radii. The

spider crossed some

radii

as she laid sticky spiral

without

attaching

to them.

This behavior,

which can be

deduced from

finished

webs, was

especially common

near the

hub.

II. Radius

Construction

F.

Order of thread placement

F1:

One trip from hub-one

radius laid (Fig. 5).

Starting

from the hub,

the

spider

moved to the

edge

of the web along a pre-existing

radius, paying

out

a

new line as

she went. She

attached this

line to the

frame, but

then immediately

broke

it

and

rolled it

up

as

she returned to the

hub, laying

a

replacement

line behind

her as she

went.

Usually

radius

con-

struction was uninterrupted

by

hub

construction,

and

the hub

loops

were

laid

only

late

in

the radius

construction stage

or after

it

was finished.

F2:

One

trip from hub-two

radii attached

at two

points

on the frame

(Fig.

6). Starting

either at the

hub as she

laid hub loops

or at some point

away

from the

hub as she spun

the temporary spiral,

the

spider

moved to the

edge

of the

web, paying

out

a

new line as

she went.

She attached this

line to the

frame, then

moved farther

along the frame

and at-

tached it

again;

finally

she used

the line laid on the

way

out for

support

as

she returned to the

hub, lay-

ing a second

new radius

as she went. On arriving

at

the point of

origin she

resumed laying hub

or tem-

porary spiral

line. Thus

in

contrast to the

preceding

case,

hub

loop

construction

was an

integral

part

of

radius construction.

F3:

One

trip

from the

hub-two radii attached at

a

single point

on the frame

(Fig.

7). Cyrtophora

webs

have no

sticky spiral,

but

they

do have

very tightly

spaced

radii. These were

laid two at a time

along

with hub

and

"temporary

spiral" lines,

but differed

from

F2

behavior

in

being

attached

to the frame only

once instead of

twice,

and

to the hub or

temporary

spiral

at

two

points

rather

than one. This

type

of

behavior could be deduced

from

inspection

of fin-

ished webs since

it

results

in

characteristic "V" shaped

intersections of radii with

frame threads.

F4:

One

trip

from

the hub-one double radius

(Fig. 8).

Interrupting

hub construction,

the

spider

walked from the hub along

a

pre-existing

radius and

attached the line she laid

behind her to the

frame,

then returned to the

hub

along

this

line, laying

another behind which was

attached to this one when

the

spider

resumed hub

construction and

which ef-

fectively

doubled the

new radius.

Ill. Hub

Destruction and

Replacement

Gi: Hub left intact.

The

spider

left the hub

of the

web intact after

finishing

the

sticky spiral.

G2:

Hub center removed. The

spider

moved to the

hub

and apparently

ingested

the

threads

in

the

very

center,

and

left this hole

in

the

finished web.

G3:

Hub center removed

and

replaced.

After

mak-

ing a

hole at the center of the hub the

spider

laid

additional

lines which more or less

"sewed

up"

the

hole

she

had created.

G4:

Entire hub removed.

The

spider

destroyed

the

entire

hub rather than

just

the inner

part,

and either

reattached

the radii

directly

to

each other or

made

new

hub loops which were attached to the radii

out-

side the

points

where

they

had been

joined previ-

ously.

G5:

No

hub.

There

was

no

localized area where

radial

lines

converged,

and no lines were

broken

af-

ter

completion

of

the

sticky spiral.

IV.

Temporary

Spiral Form

Hi:

Spiral.

As

the name

indicates, the non-adhe-

sive

line

spun just

after radius

construction and

which

joined

the radii

together away

from the hub

approximated a

spiral, at least near

the hub.

H2:

Circle. The

temporary spiral

consisted of a

single

circle, or,

in

the case of

Epeirotypus (?) sp.

(#2170), two circles,

one inside the

other.

H3:

No

temporary

spiral.

No

temporary spiral

was

spun.

V. Attack Behavior

I1:

Attack all prey

by biting.

Spiders attacked prey

by

biting,

and

only

wrapped them,

if

at all, after a

bite had been

administered.

I2:

Attack wrap

with rotation in

the web. Spiders

attacked at least

some

prey by

wrapping before

bit-

ing.

The wrapping

behavior often

included rotating

the

prey

while

it

was still

attached to

the web (usu-

ally

to a

radius)

so that it

spun

"rotisserie-fashion"

(or

"bobbin-fashion"

in

Robinson,

1969) and

the

turning

movement of

the

prey

itself

appeared to pull

wrapping

silk from the

spider's spinnerets.

When

the

prey

was very large the

spider

ran

around and over

it

rather than

spinning

it.

I3:

Attack wrap

without rotation in

the web. Spi-

ders attacked at least some

prey by

pulling silk

from

their

spinnerets with

legs

IV

and

laying it onto the

prey before

administering

a

bite. The

prey was not

spun

rotisserie-fashion while it

was still

attached

to

the

web,

but

was sometimes

rotated

slowly while

being

wrapped

after

it

had been cut

free from the

web.

I4: Attack

wrap with

rotation

unspecified. Most

accounts

of

attack

wrapping

in

the

literature do not

specify

whether

the

prey

was rotated

while

in

the

web or

not.

Species

for which

only

this

type

of

in-

formation

is available

are

assigned

to

this

category.

This

is

thus not a

distinct

character

state, but rather

is

equivalent

to

"either

2

or 3."

It

should

be

noted that attack

behavior

varies

with

the

identity

of the

prey (e.g.,

Robinson, 1969)

and

is

thus

more difficult to

characterize

than web-build-

ing

behavior.

Failure to

observe a

given

type

of at-

tack

may signify

that the

spider

is

incapable

of

that

1080

WILLIAM G. EBERHARD

0

12

20

28

Rn

olV

oXl

ol

l n

;

inner

loop

Rn+

tsp

Rn

32

36

44

52

Current

segment~

l

Rnfps

of oa

tsp oIl

oil

Rn

0

88

92

Rnio

Rn+i

olI li lVI

Rn+1

olli

IlV

inner

oI0V1

loop

tsp

o

ol

tsp

FIG. 3.

Determination of

sticky spiral

attachment

point

behavior of

type

A2,

illustrated

from

films

(18

fps)

of

Alpaida

rhodomelas

(Tacz.) (#1757)

taken

looking

at

the ventral side

of the

spider

as

it

moved

downward

near the

edge

of

a

nearly vertical web.

Numbers indicate

the

number

of

frames since the last

attachment of

the

sticky spiral

(represented by

thicker

lines)

occurred. The

positions

of

some

of the

legs

(especially legs

III)

were often unclear when

they

crossed

the

body,

and

are not drawn in

some figures;

in

some cases notes

taken in the field

were

used to arrive at

the

descriptions

of both

leg

and

thread positions.

When

legs

moved

rapidly they

were blurred

in the

film, and this is

indicated

in

the

figures

by thickening

of

their

outlines (e.g.,

leg

iII

in frame

20).

The movements of different parts

of

the body are

most conveniently

described

separately;

only

movements of the

legs

which seem to move

in

consistent patterns and were

important

in

the extraction and

placement

of the sticky thread

are

described

(other legs

seemed to be mainly

involved

in

supporting the spider

in

the

web).

Body: faced inward as

the

attachment

was made to rN (0),

and

as

the

spider

moved

inward

along rN (12, 20);

then it turned to

face

more or less along the temporary

spiral (28, 32)

and maintained this

position

as

rN+1

was reeled

in

(44-88);

finally

it

turned 900 (just

starting

in

100)

to face inward

again

while the attachment

was made to

RN+,.

Leg oI: was held

in

the air near

RN

while

the

attachment was

made

(0-12),

then

moved variably as the

spider

walked inward along RN and

along

the

temporary spiral (20-28);

just

after oII contacted

RN+,

oI was extended laterally

parallel to

RN+,

(32) and tapped

several times until

it touched the inner

loop of sticky

spiral (88); then it

was brought back

and held

in

front

of

the

spider

and remained relatively

quiet and out of

contact with threads

as the attachment

was made.

Leg

oII: was held near

RN

as the attachment

was

made and as the spider started

inward (12, 20);

after

irregular

movements it contacted

RN+,

and

moved

out this line and

then reeled it in

using a "hand-over-

hand" motion

in

conjunction with

oIII, each leg

releasing the line just

after the other grasped

it, and then

extending

laterally

to

grab

it

again

and

flex

ventrally so

the

leg tip

was brought close to

the spider's body

just

as the other

leg

released its hold and reached

out

laterally

for the

stages illustrated

in

32-52). Just

after leg

oI

touched

the inner loop (88),

oII

ceased the

hand-over-hand movement

and swung

anteriorly

while the

spider

turned

to make the attachment.

Leg

oIII: held RN just inside

the attachment point

as the attachment was

made

(0),

then moved

irregularly

until

it

began

moving

"hand-over-hand" with oII

along

RN+1;

when

oI

contacted

the

inner loop (88)

it

maintained its grip

on

RN+,

as the spider

turned to make

the

attachment.

Leg

oIV: held RN just beyond the

attachment point

while the attachment

was made (0),

then moved

along RN (12, 20)

before beginning to pull

out sticky spiral

line (e.g., 32, 36) as

the spider moved

toward the next

attachment point.

At

first oIV pulled

silk by itself,

later

in

alternation

with iIV (e.g., 80,

SPIDER

BEHAVIOR

AND

TAXONOMY

1081

kind of behavior,

but

it

may also mean only

that

the

appropriate stimuli

were not present

to

elicit

that

behavior. Only

extensive observations

can distin-

guish

between these

possibilities,

and the data

for

species

of some

groups (particularly

theridiosomatids

and anapids) are as

yet extremely fragmentary.

VI.

Male

Courtship Behavior

Robinson and

Robinson (1980) noted three clear

groupings of

character states associated with mating

site and courtship

mode (direct contact

vs.

thread

vibration).

Unfortunately these patterns seem to have

little to do with

the taxonomic relationships of the

species

involved since

species which are usually placed

in

one small,

distinct group (Argiopeae-the genera

Argiope and

Gea-see Levi, 1968, and also Robinson

and

Robinson,

1980,

for evidence that

these genera

are indeed

closely

related)

fall

in

all

three

of

their

major categories.

One characteristic, however, may

be useful and is

included here. Nearly all the obser-

vations of this character were made

by

Robinson and

Robinson, and

species

are classified

according

to

their criteria.

J. Tarsal Rubbing

by Males

Jl: Without

tarsal rubbing. Males did not perform

tarsal rubbing

(TR) movements while courting fe-

males.

J2: With tarsal

rubbing. Courting males rubbed

their legs together

with

a

motion similar

to that made

when

cleaning

their

legs by rubbing

them

together.

Functional

Independence of

Characters

The characters

B2, C2, E2, and F4 consistently

occurred

only together

(in Uloboridae),

and

it

is

rea-

sonable to ask

if

these are

independent

of

each

other

and of the fact

that all

uloborids

spin

cribellate

sticky

silk.

I

have

argued

elsewhere

(Eberhard, 1976)

that

the

relatively

non-extensible nature of cribellate

sticky

silk

may

indeed be

functionally

related to

E2,

and

believe

it

is

not

unreasonable to

suppose

that

it

is

also related to C2

(see

functional

interpretation

of Cl

in

Eberhard,

1981c).

There is however

no ob-

vious relation between

any

of these

characters

and

B2

and

F4.

The only other

completely consistent combination

was

B1, C1,

and

E1

(in Araneoidea). Again

E1 and

Cl

may

be

functionally

associated with the

very

ex-

tensible

nature

of araneoid orb

weavers'

sticky

silk

and their

ability

to make

"pulley"

connections

to

ra-

dii

(nevertheless, some

groups

do not seem

able to

make such connections-see

Eberhard, 1976).

There

are no

other

obviously necessary relationships.

Character

A3 and

perhaps

also

A4

are

necessarily

associated

with the

relationship

between

web

mesh

size

and

spider body

size:

A3

would

be

impossible

in

relatively widely

meshed webs

(see text) and A4

would

seem less

likely though

not

impossible

in webs

with very small

meshes. These web characteristics

are however

independent

of the other character states.

Probable Directions of Change in

Character States

A. The

probable primitive nature of

Al

with re-

spect

to

A2

is

suggested by

the

exceptional

behavior

of

Tetragnatha sp. (#2043) and Chrysometa species

(#'s 1824, 0-6)

which all

perform

Al

behavior

near

the edges of their webs, but switch to

A2

near the

hubs (Appendix 2). One could consider that

A2 in

these species is either

a

remnant of an ancestral be-

havior Which

has

been lost

in

all

other

observed

Te-

tragnathinae-Metinae (i.e., a symplesiomorphy) or a

new, derived behavior which evolved from

Al

in

both groups (i.e.,

either

a synapomorphy

or a con-

vergence). The morphological

differences between

Tetragnatha and Chrysometa are so great that they

have been placed

in

different subfamilies, so syn-

apomorphy appears

to be ruled

out,

and

the

second

hypothesis thus requires convergence. Nevertheless

it seems more likely than symplesiomorphy when one

examines

the

details of the

spiders'

behavior. This

is because A2 seems to be just a modification of

Al

in

which the last part

of

the behavioral sequence is

omitted, and

it is

"derived from"

Al in

the course

of

the construction of each web as is illustrated in Fig.

9.

A

selective advantage for A2 over Al, which could

explain why the proposed convergence occurred, is

easy

to

imagine since

the

turning back and

forth

which is characteristic

of Al

must be wasteful

of

energy, especially

near the hub where the

spider

can

easily

use

the

closely spaced

radii to move more

di-

rectly from one

attachment to the next. It

is inter-

esting

that

Chrysometa

is considered

by

some

(Levi,

1978a) to be near to the ancestral stock of araneids

on

morphological grounds.

The

exceptional

A2 be-

havior of

spider

#2173

(undescribed genus) (all

other

metines and tetragnathines performed Al) may rep-

resent the culmination

of

this process.

A

modification of

A2

similar

to but not the same

as

A3

is

present

in

some

Micrathena

(Appendix 2).

The non-identity suggests a convergence with

A3

behavior

in

Nephilinae

and thus leaves the deri-

vation of

A3

in

the

latter

group

in doubt.

Again

en-

ergetic efficiency

could

explain

the

convergence.

A4

seems

closest

to Al

since both involve the same

orientation of the

spider's body along

the

radius.

A4,

88). Finally, as the spider began to turn to attach, oIV seized RN+1 close

to the

point

oII

had

been

holding

(100) and maintained this

hold until

the attachment

was made.

Leg iIV: stretched

the

sticky spiral

line

just as

it

was attached

to

RN (0) and

then

helped support the spider

as it moved inward along RN and along

the temporary spiral (20-32); then it began pulling out sticky spiral line (e.g., 52-80),

at

first alternating

with oIV and later with consecutive pulls by itself

until the

last pull

ended in the

stretching

of the line

(100)

as it was attached to

RN+1l

1082

WILLIAM G.

EBERHARD

1

v

~~~~~2

oilli

iIV

inner

01V

Oil

lo

Rn+1

4V

7X9

10

FIG. 4. A3 determination

of sticky

spiral attachment point

behavior, illustrated

from films

of

Nephila

clavipes

taken looking at the

ventral side of

a

mature

female spider

as she moved clockwise near the

bottom

edge of a more or less vertical orb. Conventions are as in Figure 3. Body: faced inward the entire

time,

moving only slightly sideways from one attachment to the next.

Leg

oIV: held RN just outside the attachment

point (1); then it followed

olII

to grip

RN+1,

moving laterally

until the lateral surface

of the tarsus contacted

the radius; it was then extended so that the tarsus slid along the radius until its dorsal surface contacted

the

inner loop

of sticky spiral (10)

(or the outer loop of temporary spiral-apparently no distinction was

made

between the two); it then grabbed the radius near the junction and pulled it toward the spinnerets

(15)

as

the attachment was made just

inside this

point. There was

some variation in the amount

the

tarsus

oIV

actually slid along

RN+1

before contacting a spiral. Leg

olII:

held RN just inside the attachment point

(1),

SPIDER

BEHAVIOR AND

TAXONOMY

1083

As ,'4

B

s

A

l

o

-

---c

-

FIG. 5. Fl

radius

construction

behavior;

new

lines

are

solid. The

spider

moves to

a frame line

along

a

preexisting

radius,

laying

a

new line behind

it

(a).

This

line is attached to the frame

(b),

and

as

the

spider

moves back toward the hub

it

breaks this

new

line

and rolls

it

up

with

the

pedipalps

(c),

laying

another new line behind

it.

This line is

fastened

at

the

hub

(d),

and

the

spider

usually

proceeds

to

lay

another radius without

making

any

hub

loop

attach-

ments.

which

presumably

involves a

memory

of distances

moved inward and outward

along

radii,

might

seem

to be

a

very

specialized

character,

but

in

fact a sim-

ilar

kinesthetic sense has been

noted

in

two

groups

of

spiders

unrelated

to orb weavers

(Gorner,

1966;

Seyfarth

and

Barth,

1972).

In

addition,

modifica-

tions of

the

experiments

of

Hingston

(1927)

show that

some

orb

weavers which

ordinarily

perform

Al

or

A2

also

appear

to have the

ability

to

sense

and re-

member these distances

(Eberhard,

unpubl.);

and A4

seems to have evolved

convergently

in

the

araneinine

r-4

FIG. 6. F2 radius

construction;

new lines are sol-

id. The

spider interrupts hub loop construction (or

in

other

cases, temporary spiral construction) to

move to the frame along a preexisting

radius,

laying

a new line behind it which it

holds with one leg IV

(a).

This line is

attached to

the

frame

(b),

and then

the spider moves farther along the frame and at-

taches it

again, maintaining

its

hold on the first new

line

with

leg IV (c). Finally

it

returns to the hub,

laying

a second

line behind

it.

As it moves it slides

legs

III

along the first new line, and when they en-

counter

the

junction

between this

line and the hub

or

temporary spiral loop,

the

spider

attaches the

newer radius

to the other and resumes hub construc-

tion

(d).

This

description

is the

same as that of

Hingston (1922) of

N. maculata

behavior in all de-

tails

except

his

claim that when

the first new

line

is

attached to the frame one of the

legs

IV

holds the

radius which the spider walked along from hub to

frame.

then

moved to grab

RN+,

(7) and held

this radius

and pulled it

toward the

spinnerets as the attachment

was

made.

Leg iIV:

stretched the current

segment as it

was attached to

RN (1), then

was inactive

until stretching

the next

current

segment (15, 20)

before its

attachment to

RN+,.

Leg oII:

moved

infrequently and was

apparently used

mainly

for

support

by mature

females.

Immature individuals,

however,

directed

it

retro-

laterally and tapped

toward the inner

loop of sticky

spiral before

most but not all

attachments.

Contact with

the

inner loop was made

several

radii

in

advance of

the one

where the attachment

was about

to be made.

The

leg usually

tapped until it hit the

inner loop;

seemingly in

contrast with leg

oIV, it seemed to

distinguish

sticky

spiral

from

non-sticky lines such

as

radii

and

temporary spiral.

The movements

typical

of

immature

N.

clavipes were also

executed by

mature females

of N.

maculata and Herennia

ornatissima,

and an im-

mature

Nephilengys

malabarensis

(mature

individuals of this last

species were

not observed).

1084

WILLIAM G.

EBERHARD

FIG. 7. F3 radius

construction;

new lines are sol-

id. The spider interrupts hub

loop

(or temporary spi-

ral) construction to move to the frame along a preex-

isting radius, laying a new line behind it (a). It

attaches this line to the frame (b) and then moves

back

to the hub along the new

line,

laying a second

line as it goes

(c).

This line is then attached to the

other radius

bounding

the

sector where the new lines

are

laid,

and

the spider resumes hub loop (or tem-

porary spiral) construction (d). Some

radii

laid this

way were attached to each other part way along their

length (e) by Mecynogea sp. (#1040).

genus Cyrtarachne

which

spins

reduced

orbs. Thus

the relationship of

A4

to

the other character states

is not clear.

B, C, D,

E.

There is no a priori way

to

distinguish

primitive from derived states

in

these characters.

G, H, J.

On

the basis of

simplicity, G1, H2,

and

Jl

might seem primitive with respect to alternative

states.

In

each

case

however these

simpler

states

are

only

omissions of

given behaviors,

and

would

thus

be

easy to

derive from other states.

Thus there

is

no

a

priori certainty

of direction of evolution

in

these

characters.

F. F

1

might seem derived with respect to the other

states since

it

involves the

complex

behavior

of

breaking

and

rolling up

one

line while

laying

another

behind rather than

simply walking

under an

unbro-

ken

line. Essentially

identical behavior

is

known,

however

in

non-orb-weaving species

such as

Pho-

roncidia pukeiwa (=Ulesanis pukeiwa) (Marples,

1955a), P. studo (Eberhard, 1981a), Argyrodes

attenuatus

(Eberhard, 1979), Synotaxus

turbinatus

and

two

Chrysso species (pers. observ.) (all

in the

araneoid

family Theridiidae).

Indeed

it

is not

far

g

~~~~~~-

---;----

----

b

C~

FIG. 8. F4 radius

construction;

new lines

are

sol-

id. The

spider interrupts

hub

loop

construction to

move to a frame

line

along

a

preexisting

radius,

lay-

ing

a new line behind it.

This

line

is then

attached

to the frame

(b),

and the

spider

lays

a second

line as

it returns to

the

hub.

This

line

is attached to the

first,

thus

doubling

it,

and the

spider

resumes hub

loop

construction

(from

Eberhard,

1972).

The first

radii

laid

(during

frame thread

construction)

were

exceptional:

they

entailed little or no hub

thread con-

struction,

and

after the

frames

were

complete

or

nearly

so,

the

spider

removed the accumulation of

threads

in

the center of the web

and

proceeded

to

begin

"normal"

radius construction

as illustrated

here.

from

the

behavior used

by

spiders

of

many

families

to ascend their

safety

lines after

dropping

from

some

support

above. Thus

there

is no

clear a

priori

way

to

distinguish

primitive

from derived.

Arguments

supporting

the

probable

derivation of

F3

from

Fl1

are

given

in

the text.

I. It

seems

likely

that

Ii

is

primitive

with

respect

to

12

and 13. This is because attack

wrapping

is a

complex

behavior

generally

associated with web

building,

and

lack

of

this

behavior

is

widespread

and

is

typical

of

groups

which are

undoubtedly

primitive

with

respect

to the

groups

discussed here

in

many

characters

including

web

building

(Robinson,

1975;

also

Eberhard,

1967;

Rovner and

Knost, 1974;

Rob-

inson

and

Lubin,

1979).

Due

to

the demonstrated

advantages

of

attack

wrapping

in

subduing

large,

powerful

prey

(data

summarized

in

Robinson,

1975)

it

is also

unlikely

that once

acquired

it

will be lost.

The

only

possible

exceptions

would

be

in

cases

in

SPIDER

BEHAVIOR

AND TAXONOMY

1085

RN

RN+1

RN

RN+1

FIG. 9. A

diagrammatic

representation

(horizontal

web

seen

from

above)

of the transitional

Al-A2

determination

of

sticky spiral

attachment

point

behavior of

Tetragnatha

sp. (#2043).

Only legs

I

are

shown.

Leg

oI

was extended

along

RN+1

prior

to

grabbing

it

(solid

drawings).

If,

as

in

A,

it

contacted the inner

loop

before

grabbing

the

radius,

the

spider

immediately

swung

to face inward and attached the current

segment

of

sticky spiral

(dotted lines),

performing

an

A2

sequence

of behavior.

If

it

grabbed

the radius before

contacting

the inner

loop (B),

it turned instead

to face out

RN+,

and used

iI

to

locate the inner

loop (dotted

lines

in

B),

performing

an

Al

sequence.

Occasionally

the

spider began

to

swing

its

body

to face outward

before it had finished

extending iI,

and arrested

this

movement when that

leg

contacted

the inner

loop (A2

"tinged"

with

Al).

Thus

A2

appeared

to be

derived from

Al in

that it consisted of a

part

of normal

Al

behavior,

and

vestigial

Al

movements were sometimes

included

in

an

A2

sequence.

Al

was

performed

consistently

near the

edge

of the

web,

and

was

gradually

replaced by

A2

as the

spider

moved inward.

which most

or

all

of

the spider's prey

were relatively

harmless

(e.g., non-stinging

or non-biting)

and/or

either extremely

efficient

or extremely

inefficient

in

escaping

from the web.

A

possible example

of

the

former are

the bolas spiders

and their

kin which

ap-

pear to specialize

on moths (McKeown,

1952; Clyne,

1973; Robinson

and Robinson,

1975; Eberhard,

1977);

none of these

spiders attack

wraps (but

it is not clear

if this represents

a secondary

loss-see

caption to

Fig. 1 of the

text). An example

of the latter

type are

theridiosomatids

which only

occasionally

attack wrap

and which seem to capture

mostly small

nematocer-

ous flies (J.

Coddington,

pers. comm.; pers.

observ.).

It should

be noted that

acceptance of

the ideas that

attack wrapping

in

araneids

is derived and that once

acquired

it will not be

lost except

in the cases just

described

requires that one

consider

the evolutionary

origin of

attack wrapping

in

the

non-orb-weaving

araneoid family Theridiidae to be independent of

that

in

Araneidae (unless, as

noted in

the text, one

accepts the improbable loss of orb webs in theridiids).

This is not unreasonable since theridiid attack wrap-

ping at least often

involves the use

of

a

type

of

silk

(from the aggregate gland) which is never used in

this way by araneids (this gland's product constitutes

the sticky spiral),

and

because prey wrapping

seems

to have evolved

independently

in

a

number of

spi-

ders (see text). It is interesting to note

in

this con-

nection that the different combs on legs IV used in

wrapping

in Theridiidae and

Uloboridae argue for

a

separate origin

of

wrapping

behavior

in

these

fam-

ilies (the combs are on different segments and formed

by modifications

of

different structures-see Opell,

1979; pers. comm.) (see Eberhard, 1979, for evidence

that such combs

do indeed

function

in

wrapping be-

havior).

1086

WILLIAM

G. EBERHARD

c

'i-4

.

=

(1) E E

-5

:;;

0

- =

9

0

S.-

;.4

--d

m u

z

(u

-

Cd

u

(U

cd

Cd

W

$..4"

4

cd

>

(U

>

U

'o

(U 'w

0

'IC

.-

=

--I

::I

cd

c

'n

u u

o ::I

7

o w

(U

-Z

-

W

Cd

w

+-J m

in.

,

14

d

cn

:

Cd

m

(U

;zq

cn

i--j

w

bo

o

4.5

00

bo

M

-Cs

-z

0

Cd bo

0 CZ

0

(n W

4-4

V

- o g.-

(u

o o

-C$

w

>

0

4--j

U)

Cd

bo

$:

>,

un

Cd

cd

$.4

Cd

ts

7s

-C$

0

bo

4-j

0 4-i 4-i

ce

w

g M

4.5

4-j

+-j j..j Cd

Cd

cd

Cd

0

Cd

CZ CZ CZ

CZ

Cd

W

$

w

-Cs

bo

9'.

bo

w

bo

$4

0

1.4

(u

-4-J

+-j

(n

>

cn

0

<U

(U

w > E

Z

s. -

>, =

<U

4--j

Cd

u

1:1.

-4

S

-,C

$.4

= E

u

o

cn

:

cn

Cd

00

oo

M

CIA

cn

cn t-

00

C', C',

C'S

cd

cri

-C$

u bo

bo

o

cn g

9

CZ

(U ;

5

M -C$

0

.-

E

"..,

0.

u

-C$ ::I

E

(U

0

u

U)

<U

U)

4--4

(u

o

(u

cd

cn

bo

-;::I

cd

(u

crj

>,

0

4-J

0

Cd

0

Cd w

tzx

w

>

s.

+-j

0

4-j

+-j

(n

u U w

w

cn

a

cd

0

cd

0

> Cd

9

cd

u

4-4

E u

0

cd

(u

0

-C$

0 =

$.4

r.

+-J

4-j

cd

-j

Cd

t4..,

aj

4-)

Cd

0

0 u g

'r

0

cn u

.

0

'Ci

E ,

>,

" -C$

bo -c$

-,

bo

.td-

cd

0

g

cd

Cd

M

cd

Cd

x

Cd

x

s-4

-C$

r.

-Cs

i--j

.1 -

bo

cd

0 0 0

4-j

<U

tn

cn

U

t:

4-j

,M

0

bo

'r.

(n

i..J .-

W

U

(n

tz

Cd

u

cd

g

cd

-,t

Cd

t: ,

cd

>1

cd

g-.

-.4

0

cd

'r

>

cn

'd

>

Cd

tZ

0

cd

-zi

bo

cd

c

o

bt

0

to

r4

Cd

bo

9'.

;

bo

oz

Cd

bo

4, u

>

> u

Cd

bo

W

0

Cd

u Q)

'C$

o

Cd

4

-C$

Cd

Z

cd

g-. g-. (u

::I

$4

Cd

X

M

o

4-j

(u

u o

0

g u

0

cn

C)

u cd

cd

$4

IC$ 0

0

w $4

4-4

i

bo Cd

9

Cd

SPIDER

BEHAVIOR AND

TAXONOMY

1087

C9

C-CI)

CC4

_-

C"9

N

C

N

CN

N

C Q _ __ _ ___ _ ________ _~~~~~~~~~CI

-

00 00

00

0

0~~~~~~~~

cn

~ ~ ~ ~ ~ ~ ~ ~ ~~~~0

_0 __

_ _.

_ _

_

_ _

_

_ _ _

_ _

_

<^

Z 0

_ __

__

_

__

_

__

___

__

t

.E N~~~~~~~~~~~~~

1088

WILLIAM

G.

EBERHARD

U-)

10

C',

en

i-

-)

Cj)

Cen

C-

-ai

00_

00000

_

-~C-c

_4

_ _

00

S Q

0)

m

_

_ _

_

_ _

_

c9

_

cL

C;~~~~~~~~~

t-t

X c c O O e e t e _ = c e cro X

O~~~~~~~~~ajce

~0) C 0~-0~ o

0)0) Ch 0h I 1 c9 c9 Ch ~ Ch Ch ?O > C.

c

z

SPIDER

BEHAVIOR

AND

TAXONOMY

1089

In

tO tO

C

r-,

-r,

r-

--

in

\C

r- ON (N

06

,O

c

m ~ ~ ~ ~

\

n

\C

O' C ON

0

N cni

C

0

c9

C9l

\C

_- I'_

r-O,l-0

00

00 00 00

C.9

,,I-c9c9

c9

c

c9

in

-c9

r9

in

t9

c c9

-

',I

in 00

0909c9

X

Q

o ~ ~ ~~t-

tt

t9

o

.0

0

C

'00

0

oo

s

t:t

( t- t

't-

4z

(U

t:t 0 DeO

C~~~~~~~~~~~~~~~~~~~~)

~ ~ ~ ~ ~ ~ ~ ~ '

X

_ c _

e c O

_ c O

% N e

m c %

e _ N

e N c

~~~~~~~~~~~~~~~~t:Ot

C

%

0~~~~~~~~~

?~ ~~

a

C/)

Y

00X

t:t

C~

~~~s

S

-~~~tt

's;

=~~~~~~~

ol)

~ UC~

00

C

.

V

C'S

1090

WILLIAM

G.

EBERHARD

(

eeN (N (N (N (N (N

c-

N

CN

C"

-4

C"

eN

C'

c

C" C"

Cl

C)

C"l

C"

C"l

CN CN

CN

cl

r

l

C)

N

CN CN

C

N

C)

"

C"l C"l

C"

CN

CN Cl

0~~~~~~~~~~~~~

.on

0

N

(zN(N (N(N(NC N

(N(N(csN

(9N(N

(N (N ~N

(N

00

-

~ ~ ~ ~ 00(N C~~~~~t'.r-00

in

i

' '

-0

0

,-

,- o

0

i

0N

00

_l

U

00

CZ

-0

w

C

>

N~~~~~~~~~~~~~~~(

u

=

~~~~~c

v

~

~~~~~~~~~~~~~~~~~

C

oR

E~~~~~~~~~~~~~~~~~~~~~~~E

,~~~~~c

Q

1>,

2

r aS o

SPIDER

BEHAVIOR AND TAXONOMY 1091

C"

(

eeN (N (N

(N

(N (N (N

cl-

c

g" g"

C"

g"

g

Q

C"l

g" g"

C" C)l C)l (N gN(

N(

0~~~~~~~~~~~~~0

(

m

~~~~~N (N (N (N(N N(N(N(N(N(N N

CN (N

az

~ ~ ~ ~ ~ ~ ~ ~ ~~~r

\C -.-I in

0

N(NN

(N (N (NN

(N(NN

C(N(N

(N

~tt

(N

00~~~~~~o

t~~~~~~

t~-CC W

D OOi ;O00 tk

00~~~

Z

~ ~ 0 N-

N

)

<N-

\

U)

00

t

U)

eX

C ~~~~~~~C

(u

( L

;

=N

SX

O

-

-Z!

0

U~~~~~~~~(

1092

WILLIAM

G.

EBERHARD

C"

cn

n cn

n

cn

c-

C"

(N (N

n cn

1- 1- --_-

v

~ ~ ~ ~ C l

CZ

Ca Ca Ca

Ca CZ

CZ

C

0n0

t ~t

0

~ " o

a-

_ln

00

C' -4

00

in

r-

-4r

4-

0~~~~~~~~~~~~0

*>~~~~~~~~~~~~~0

0

-6

d

m

0

Cl)

0 ~~~~~~~~~~~~~Cd

l

C

_Id t:t em NNNP

X .s O c ? .M 0

t~~~~~

=~

H i j 9 W - w i 2 X ' E E . t

. cX-Z

EH

?

>

SPIDER

BEHAVIOR AND TAXONOMY

1093

=4

g

4)

44, cd

'o

>

C's

4.)

>

C,3

C's

a) , C's

C's

bi)

r-

Z

4) biD

Cd

0

-a

=-

-Zj

biD

r.

C's

0

ti

u

C13

6

C13

NO

Z

0

E.

0

"o

0

C13

Z

'A

CU

r.

0

>

0

>

C13

C's

0 u

a7l

r.

a

v

0

-

C

4.)

0

pq

1-4

cl-I

iz

ln

=

0

4.)

>,

bl)

C's

0

-1

C13,

X4

4.)

C's

u

M

(D -t

.-

-

:

4.) "

r.

C's

0

r.

CN

C's -.C

o

06

m

Z 0

C13

o o

cd

C's

>

C's

-

=

I..

r.

z

>

0

Z

C's

o

0

Z

a

>

CIS

0

Cd

0

biD

z

-o

q,-

0

bi)

bO

C's

r.

0

o

-d

r.

C's

co

C's

0

Z

CN

-bl)

0

>

0

U

-

7--

C,3

0

C'S

Z

-I..

1

-m

:3o

Z

In

'o-

C13

0

Z

-.4

m

0

'3

4.)

U

> > >

U)

-

4

Z3

U!

0

In

4) -

= 'n

iz

0

I-Cd

r

7

:6

4)

-

_

:

'(5

0

o

0

cd

C13

C's

bt

C's r-

o

U 4.)

0

a.

o

>,

-

;..

=

4

cn

r. Cd

Z

0

U

0

Z:,

'C.-

cd

0

".

a

In

Cd

r.

r-

-Zx

-

In

bO

Cd

Wu

u-m

cl

m

Z

w

CN

r

Wc

p

0

Cd

v

10'

M -

z

=

-

r.

In

o

C)

o

19

-

.2

E

r,

P4

0

P4

a

r

Cd

0

C,3

0

Cd

cd

4.)

b

0

4.)

U - 4.)

C,3

-i5

.-

U

,

C's

M

'5

: 'o

=

4.)

:6

-

C,3

r

im.

C)

0

C,3

U)

E

0 r

M

r.

g

r

4.)

-

0

K

C'3

c

-0

0 C%

E;a

:

-a

.

0

7

0

M.

0

In

0

C)

cd

bo

Cd

0

Cd

cd

CIS

E

'o

C's

'o

A:

Cd

4.) 0

M 0

cd

C,3

4.)

M 0

r

>

0

I..

>

bO

4.)

r.

.0

0

C13

r.

z

-,;5

U

.-.z

:

0

c)

=

0

r.

.0

C)

bO

0

im.

tn

E

c)

v

z

0

>

C13

=s

C,3

U

0

O 0

r.

c

C13

0

0 0

M W

w

C-1 3

C,3

r-1. C,3

14 0

CZ

0

M

biD

>

bl)

W

0

>

o

4)

-

>,

r.

W)

0

>

0

'n

4.)

-

.-

4.)

zls

r. r-

C,

>

o

C13

>

0

>

M

0

v 0 v

-

0

1.

n

'o

0)

C13

r-

>

C)

CI.

o

Cd

0

O

0 0-

0

cd

a :3

r-

C13

M

04

n

$M.

v

4.)

.

-

(:, :.-Z

0.-

0 -

r

>

0

4.)

.0 >,

a

C'S

bO

,

n

-

'o

Z7

C13

0

-w

r

0

Z

c)

-.4

0

>,,=

U

7

C,3

>,

aq-,

C,3

c)

o o

C,3

'n

7

-

.Z

: -

a

0

Cd

0

zu

v

w

a

>,

=

,,

-0

r.

't

cd

0

o a IL)

4

b,

Z3

0 -

w Z 'n

w

=

0

Cd

o

0 ,

.

-

$M.

'S

0

'0'

-

'n

"

cd

>

0

'o

r.

bj)'-

Cl 0

'n

U

cd

0 CIS

'o

>

C13

;>

a

_A

C's

$M.

14

0 o o o

o" o

C13

v

C)

r.

U) a P4

-

=

,

r.

$..

-

C13

o

En

>,

4)

r

M 0

bO

Cd

r.

',;

4.).-

r

b

0

O

O'

M

7

>

bll

4.)

=s

r r

0

bO

r

ql)

M

0 0

0

c,3

o

C'3

bO a

Z

C13

C13 o C13

r,

C, 34i l) r,

C)

-0

bb,,

(:)

-

4

-

.- C-

:

-

Z

.-

U

0

4j

4.)

T.

U

C,3

0

cd

:Y-

r

:3

:3,0

biD

4.)

0

v

U

v

0

I...

5.

(::)

4.)

o

C,3

0

To

2

P.

0

ar

r:

=

-1

-

0

1094 WILLIAM G. EBERHARD

APPENDIX

3.

Attack behavior of at least 81 species in at least 38 genera.

Observations with

no source indicated are

original,

and numbers refer to

descriptions

in

Appendix

1.

It

should be

kept

in

mind

that

"4"

is not a

separate category,

but rather

indicates ignorance

of

whether

the

character state is

2

or 3.

"( )"

indicates that less than 6 attacks were

observed, and applies only to original

observations.

Spider

Attack

behavior Source

Araneidae

Tetragnathinae

Dolichognatha spp. #533 2

#0-2

2

Metinae

Leucauge sp. near venusta 2

L. spp. #527

2

#526

2

#0-21-5

2

Chrysometa spp. #532 2

#0-6

(2)

Nephilinae

Nephila clavipes

1

Robinson,

1975

N.

maculata

1

Robinson,

1975

N.

constricta

1

Robinson,

1975

N. turneri

1

Robinson,

1975

Nephilengys

cruentata

1

Robinson,

1975

Herennia ornatissima

1

Robinson, 1975

Araneinae

Argiopeae

Argiope argentata 3

Robinson and Olazarri, 1971;

pers.

observ.

A. savignyi 4

Robinson, 1975

A

florida

4

Robinson,

1975

A. aurantia 3

Robinson, 1975; pers. observ.

A. aemula 4

Robinson,

1975

A. picta 4

Robinson,

1975

A. ocyaloides

3

Robinson and Lubin, 1979a

A. sp. 4

Robinson, 1975

Cycloseae

Cyclosa caroli

3

C.

conica

3

Marples

and

Marples,

1937

C.

triquetra

3

C.

furcata or bifurcata

3

Salassina

sp. #2226

3

Mangoreae

Mangora melanocephala

3

M.

sp. #1569,

1641 3

Acacesia hemata

3

Cyrtophora cylindroides

4

Lubin,

1973

C.

monulfi 4

Lubin,

1973

Eustalafuscovittata

3

Spilasma artifer

3

Araneae

Araneus diadematus

4

Peters,

1931

A.

marmoreus

4

Robinson,

1975

A.

rufipalpis

4

Robinson,

1975

Alpaida leucogramma

3

Eriophorafuliginea

3

Robinson

et

al.,

1974

E.

nephiloides

4

Robinson,

1975

E.

edax

3

Wagneriana sp.

#574

(2)

SPIDER

BEHAVIOR

AND TAXONOMY

1095

APPENDIX 3. Continued.

Spider

Attack

behavior Source

Metazygia gregalis

3

M. sp. #0-21-2 3

Metepeira labyrinthica

3

M. sp. or spp. #254 (2)

#1313

3

Wixia

ectypa

2a

Stowe,

1978

Arachnureae

Arachnura melanura 3 Robinson and Lubin, 1979a

Hypognatheae

Hypophthalma sp.

3

Gasteracantheae

Gasteracantha

cancriformis

lb

Muma, 1971; pers.

observ.

Micratheneae

Micrathena clypeata

1

Robinson,

1975

M. schreibersi

1

Robinson,

1975

M. sexspinosa

1

Robinson,

1975

Mastophoreae

Mastophora dizzydeani

1

Dichrostichus

magnificus

1

Longman,

1922

Cyrtarachneae

Poecilopachys australasia

1

Clyne,

1973

Pasilobus sp.

1

Robinson and Robinson,

1975

Celaenieae

Celaenia excavata

1

McKeown, 1952

Taczanowskia sp. (1) Eberhard, 1981b

Theridiosomatidae

Olgulnius spp. #1292 (1)

#EG2-22III76

(2)

Epeirotypus spp. #1093 (1)

#1054

(1)

#1603 (1)

#802 (1)

Uloboridae

Uloborus walckenaerius

2

Marples,

1962

U. congregabilis

2

Marples,

1962

U. diversus

2

Eberhard,

1967

U.

trilineatus

2

Philoponella semiplumosa

2

P. republicana (2)

P.

tingena (2)

P.

vittata

(2)

P.

oweni

2

P. arizonica

2

P. para (2)

P. sp. nearfasciata #Moz-14

2

Zosis geniculatus

2

Marples, 1962, pers. observ.

Hyptiotes paradoxus

2

Marples,

1962

Miagrammopes simus

2

Lubin et al.,

1978

M.

intempus

2

Lubin et

al.,

1978

M. sp. near unipus (2)

Lubin

et al.,

1978

a

Prey

is

wrapped

on branch

rather

than

in

web, spider's

web

is so

unusual that

comparisons

with

orb weavers are

probably

not

justified.

b

Robinson (1975) attributed

attack

wrapping

to

this

species

Muma's

account,

on

which

Robinson

based

his

decision,

is

unclear

and

is

open

to

other

interpretations (Muma, 1971) My

own

observations

(20

attacks

on a

variety

of

potentially dangerous [wasps,

bees]

and

innocuous

prey

suggest

that this

species

does

not

attack wrap.

Robinson later

(Robinson and Lubin, 1979a)

included

Gasteracantha

in

a

list

of

genera lacking

attack

wrapping.