18

FROM: Ubick, D., P. Paquin, P.E. Cushing, and V. Roth (eds). 2005. Spiders of North America: an
identification manual. American Arachnological Society. 377 pages.

Chapter 2

PHYLOGENY AND CLASSIFICATION

OF SPIDERS
Jonathan A. Coddington
ARACHNIDA
Spiders are one of the eleven orders of the class Arachnida, which also includes groups such as harvestmen (Opiliones), ticks and mites (Acari), scorpions (Scorpiones), false
scorpions (Pseudoscorpiones), windscorpions (Solifugae),
and vinegaroons (Uropygi). All arachnid orders occur in
North America. Arachnida today comprises approximately
640 families, 9000 genera, and 93,000 described species, but
the current estimate is that untold hundreds of thousands
of new mites, substantially fewer spiders, and several thousand species in the remaining orders, are still undescribed
(Adis & Harvey 2000, reviewed in Coddington & Colwell
2001, Coddington et al. 2004). Acari (ticks and mites) are
by far the most diverse, Araneae (spiders) second, and the
remaining taxa orders of magnitude less diverse. Discounting secondarily freshwater and marine mites, and a few
semi-aquatic or intertidal forms, all extant arachnid taxa
are terrestrial. Arachnida evidently originated in a marine
habitat (Dunlop & Selden 1998, Dunlop & Webster 1999),
invaded land independently of other terrestrial arthropod
groups such as myriapods, crustaceans, and hexapods
(Labandeira 1999), and solved the problems of terrestrialization (desiccation, respiration, nitrogenous waste
removal without loss of excess water, and reproduction)
in different ways. Although the phylogeny of Arachnida is
still controversial (Coddington et al. 2004), specialists agree
that the closest relative of Araneae is a group of orders collectively known as Pedipalpi: Amblypygi, Schizomida, and
Uropygi (Shultz 1990).
PHYLOGENETIC THEORY AND METHOD
Systematics is the study and classification of the different kinds of organisms and the relationships among them.
Good classifications are predictive: knowing one feature
predicts many others. If one knows that an animal has
spinnerets on the end of the abdomen, it will also have
fangs and poison glands (lost in a few spiders), eight legs,
two body regions, male palpi modified for sperm transfer,
and it will spin silk: in short, it is a spider. All spiders share
these features because they inherited them from a common
ancestor, but todays spiders have evolved to differ among
themselves. For example, the earliest spiders had fangs that
worked in parallel (orthognath, like tarantulas and their
allies), but later in spider evolution one lineage developed
fangs that worked in opposition (labidognath, like the
majority of spiders in North America). Much later within
the labidognath lineage, some evolved the ability to coat silk
lines with a viscid, semi-liquid glue, useful for entrapping
and subduing prey. This nested pattern of branching lineages (phylogeny), results from evolutionary descent with
modification (Fig. 2.1). The vast majority of similarities
and differences among species are due to phylogeny. Jumping spiders (Salticidae) all have huge anterior median eyes
because they are relatively closely related, and wolf spider
(Lycosidae) eyes exhibit their characteristic eye pattern for
the same reason. Phylogeny explains more biological pattern than any other scientific theory (e.g., ecology, physiology, ethology, etc.), and therefore classifications based on
phylogeny will be maximally predictive. Besides huge front

eyes, jumping spiders also share many other anatomical,

behavioral, ecological, and physiological features. Most
important for the field arachnologist they all jump, a useful
bit of knowledge if you are trying to catch one. Taxonomic
prediction works in reverse as well: that spider bouncing
about erratically in the bushes is almost surely a salticid.
Another reason that scientists choose to base classification on phylogeny is that evolutionary history (like all history) is unique: strictly speaking, it only happened once.
That means there is only one true reconstruction of evolutionary history and one true phylogeny: the existing classification is either correct, or it is not. In practice it can be
complicated to reconstruct the true phylogeny of spiders
and to know whether any given reconstruction (or classification) is true. Indeed, scientists generally regard truth
in this absolute sense as beyond their reach. Instead they
strive to make their hypotheses as simple as possible, and as
explanatory as possible. Simpler and more general hypotheses win. They win through comparison of predictions
made by the hypothesis to factual observation. Scientific
hypotheses (e.g., explanations, classifications, taxonomies,
phylogenies) are constantly tested by discovery of new traits
and new species. To the extent that the hypothesis is good,
it accommodates and comfortably explains new data. If the
new data do not fit the theoretical expectations, sooner or
later a new hypothesis or a revised version of the old one
takes its place. In biological classification, and phylogeny
reconstruction in particular, scientists have developed a
number of technical terms to describe the various ways that
classifications or phylogenies do, or do not, correspond to
fact (Fig. 2.1). Any group in a classification is said to be a
taxon (plural taxa) or clade, and in theory corresponds to
one common ancestral species and all of its descendants.
Such clades are said to be monophyletic (mono = single,
and phylum = race).
In the preceding examples, spiders (Araneae), labidognath spiders (now called Araneomorphae), sticky-silk
spinners (Araneoidea), jumping spiders (Salticidae) and
wolf spiders (Lycosidae) are all thought to be monophyletic
groups, clades, and taxa. Each of these groups is distinguished by one or more uniquely evolved features or innovations. Such characters are said to be derived, because
they are transformations of a more primitive trait. Orthognath chelicerae is the original, primitive (plesiomorphic)
condition for spiders, and labidognath chelicerae is the
later, derived (apomorphic) condition. The only acceptable
evidence for monophyletic groups are shared, derived characters, or synapomorphies (syn = shared, apomorphy =
derived morphology) such as the evolution of viscid silk in
Araneoidea (Fig. 2.1).
Sometimes systematists (scientists who infer phylogeny
and use the results to classify organisms) make mistakes
and group taxa based on primitive characters or plesiomorphies. Such groups, containing a common ancestor
and some but not all of its descendants, are then termed
paraphyletic. In Figure 2.1, the grouping Orthognatha is
paraphyletic because it is based on a primitive character,
orthognath or paraxial chelicerae, and because it includes
the common ancestor of all spiders but excludes some

19

Spiders of North America

descendants, i.e. the Labidognatha. Even worse, sometimes

groups dont even include any common ancestor at all and
are then termed polyphyletic (Big Spiders, Lycosidae +
Mesothelae in Fig. 2.1 would be polyphyletic). Polyphyletic
groups are usually based on convergent features and paraphyletic groups on primitive features.
Classifications (and phylogenies) need not be strictly
binary or dichotomous: in Figure 2.1 the three-way fork
uniting Araneoidea, Lycosidae, and Salticidae intentionally
doesnt indicate which is most closely related to which. If
nodes are dichotomous, the two daughter lineages are often
called sister taxa, or, informally, sisters.
In practice systematists infer phylogeny by compiling
large tables or matrices of taxa and their traits or features. Traits may be anything presumed to be genetically
determined and heritable, such as morphology, physiology, behavior, or, increasingly, DNA sequences. The ideal
approach would encapsulate all comparative knowledge
about the group in question. Based on evidence external

to the analysis (or even an a priori assumption) one taxon

in the analysis is specified to join at the root of the tree,
and powerful computer algorithms are used to find the
most plausible tree (or branching diagram, also termed a
cladogram) that unites all taxa and best explains the data.
Systematists adopt the initial null hypothesis that all similarities are due to phylogeny. The fit between the tree and
the data decreases to the extent that one must suppose the
same trait arose two or more times independently (convergent evolution) or was lost secondarily. An example of
the former might be big. Not all big spiders are each
others closest relatives (but some are). An example of the
latter is the absence of true abdominal segmentation in all
spiders. Spiders are arthropods and arthropods typically
have segmented abdomens; spider relatives also have segmented abdomens. Rather than suppose that all arthropods with segmented abdomens gained the condition
independently, and thus that spiders reflect the ancestral
unsegmented arthropod, it becomes very much simpler

ea

an
eo

id

ae
Ar

Ly
co

id
ltic
Sa

sid

ae

m
alo
M
yg

es

ot

he
la

or

ph
ae

(Monophyletic)
Entelegynae

Viscid silk

(Polyphyletic)
Big Spiders

Female fertilization ducts

(Paraphyletic)
Orthognatha

Orthognath

Synapomorphy

primitive
plesiomorphic

Cheliceral
Orientation

derived
apomorphic

Labidognath

Opisthothelae
Clades
Spinnerets

Time

Araneae

Fig. 2.1 Taxa are in regular, characters in italic font. Only synapomorphies (shared, derived characters) are valid evidence of monophyletic groups
(clades). Paraphyletic groups are usually based on plesiomorphies, polyphyletic groups on convergences.

20

Mesothelae

Liphistiidae
Atypoidea

Atypidae
Antrodiaetidae
Mecicobothriidae

Avicularoidea

"Dipluridoids" + Hexathelidae
Crassitarsae

"Nemesiioids" + Microstigmatidae
Barychelidae
Theraphosodina
Theraphosidae (Ischnocolinae)
Theraphosoidea
Paratropididae
Opisthothelae
Theraphosidae (Theraphosinae)
Rastelloidina
"Cyrtaucheniioids"
Paleocribellatae
Hypochilidae
Actinopodidae
Migoidea
Austrochilidae
Migidae
Austrochilioidea
Domiothelina
Gradungulidae
Idiopidae
Araneomorphae
Filistatidae
Ctenizidae
Caponiidae
Tetrablemmidae
Haplogynae
Segestriidae
Dysderidae
Neocribellatae
Orsolobidae
Oonopidae
Pholcidae
Diguetidae
Plectreuridae
Ochyroceratidae
Leptonetidae
Telemidae
Araneoclada
Sicariidae
Drymusidae
Scytodidae
Eresidae
Eresoidea
Periegopidae
Oecobiidae
Hersiliidae
Entelegynae
Mimetidae + Malkaridae?
Huttoniidae
Palpimanoidea
Palpimanidae
Stenochilidae
Micropholcommatidae
Holarchaeidae
Pararchaeidae
Archaeidae
Mecysmaucheniidae
Canoe Tapetum Clade
Deinopidae
Deinopoidea
Uloboridae
Orbiculariae
Araneidae
Tetragnathidae
Theridiosomatidae
Araneoidea
Symphytognathoids
Mysmenidae
Anapidae
Symphytognathidae
Derived Araneoids
Pimoidae
Linyphioids
Linyphiidae
Reduced Piriform Clade
Cyatholipidae
Cyatholipoids
Araneoid Sheet Web Weavers
Synotaxidae
Nesticidae
Nicodamidae
Theridioids
Phyxelididae Spineless Femur Clade
Theridiidae
Titanoecoids
Titanoecidae
Dictynidae
Zodariidae
Zodarioids
Cryptothelidae
Sparassidae
Divided Cribellum Clade
Anyphaenidae
Clubionidae
Corinnidae
Zoridae
Liocranidae
Dionycha
Philodromidae
RTA Clade
Salticidae
Selenopidae
Cithaeronidae
Thomisidae
Ammoxenidae
Trochanteriidae
Gnaphosoidea
Gallieniellidae
Lamponidae
Gnaphosidae
Prodidomidae
Neolanidae
Stiphidioids
Stiphidiidae
Fused Paracribellar Clade
Agelenidae
Agelenoids
Desidae
Amaurobiidae
Amaurobioids
Amphinectidae
Tengellidae
Metaltellinae
Unplaced families
Zorocratidae
Miturgidae
Chummidae
Ctenoid
complex
Lycosoids
Ctenidae
Cybaeidae
Ctenidae (Acanthoctenus)
Cycloctenidae
Zoropsidae
Psechridae
Hahniidae
Senoculidae
Halidae
Oxyopidae
Pisauridae
Homalonychidae
Miturgidae
(Mituliodon)
Higher Lycosoids
Synaphridae
Trechaleidae
Lycosidae
Mygalomorphae

Fig. 2.2 Phylogeny of Araneae

Spiders of North America

to suppose that the original arthropod was segmented

and that it was spiders that changed to lack abdominal
segmentation. The tree that requires the fewest hypotheses
of convergent evolution and/or secondary loss is preferred
as the current working hypothesis. Of course new taxa and
characters can be added, so that in practice the preferred
tree can -and usually does -change at least a little bit with
each new analysis. The best classifications are derived from
phylogenetic analyses, but to date rather few groups of spiders have been analyzed phylogenetically.
SPIDER PHYLOGENY
Spiders currently comprise 110 families, about 3,600
genera, and nearly 39,000 species (Platnick 2005). Paleontologists to date have described roughly 600 fossil
species (Selden 1996, Dunlop & Selden 1998), but these
have primarily been significant in dating lineages; thus
far fossils have not seriously challenged or refuted inferences based on the recent fauna. Strong evidence supports
spider monophyly: cheliceral venom glands, male pedipalpi
modified for sperm transfer, abdominal spinnerets and silk
glands, and lack of the trochanter-femur depressor muscle
(Coddington & Levi 1991). Roughly 67 quantitative phylogenetic analyses of spiders at the generic level or above have
been published to date, covering about 905 genera (about
25% of the known total), on the basis of approximately
3,200 morphological characters (summarized in Coddington & Colwell 2001 and Coddington et al. 2004). On
the one hand, overlap and agreement among these studies
is just sufficient to permit stitching the results together
manually (Fig. 2.2); on the other, they are so sparse that
many relationships in Fig. 2.2 are certain to change as more
information accumulates and more taxa are studied. Figure
2.2 is not itself the result of a quantitative analysis but is,
essentially, an amalgamation of individual cladograms
published for particular lineages. Although there are several spider phylogenies above the species level based on
DNA (Huber et al. 1993, Garb 1999, Gillespie et al. 1994,
Piel & Nutt 1997, Hedin & Maddison 2001, Vink et al. 2002,
Maddison & Hedin 2003a, b, Arnedo et al. 2004), this field
is still in its infancy.
The basics of spider comparative morphology have been
known for over a century, but the first explicitly phylogenetic treatment of spider classification was not published
until the mid-1970s (Platnick & Gertsch 1976). This
analysis resolved a long-standing debate by clearly showing a fundamental division between two suborders: the
plesiomorphic mesotheles (the southeast Asian Liphistiidae with two genera and about 85 species) and the derived
opisthotheles (everything else). Whereas mesotheles show
substantial traces of segmentation, for example in the
abdomen and nervous system, the opisthothele abdomen
shows no segmentation (although color patterns in many
spider species still reflect ancient segmentation patterns)
and the ventral ganglia, or nerve centers, are fused. Opisthotheles include two major lineages: the baboon spiders
(tarantulas) and their allies (Mygalomorphae, 15 families
worldwide with about 300 genera and 2,500 species) and
the so-called true spiders (Araneomorphae, 94 families
worldwide with about 3,200 genera and 36,000 species)
(Platnick 2005).
Mygalomorphs look much more like mesotheles than
araneomorphs. They tend to be rather large, often hirsute
animals with large, powerful chelicerae. Nearly all lead
quite sedentary lives, usually in burrows, which they rarely
leave, and they rely little on silk for prey capture. Some do
fashion trip-lines from silk or debris, which effectively
increase their sensory radius beyond the immediate area

21

of the burrow entrance, and some diplurid species do spin

elaborate sheet webs, but they lack one key innovation present in araneomorphs: the piriform silk essential to cement
silk to silk, or silk to substrate. Whereas any araneomorph
can dab a tiny dot of piriform cement to anchor its dragline and almost instantly trust its life to the bond, mygalomorphs must spin structures several centimeters across to
anchor silk to substrate, and that is a long and laborious
process. Without piriform silk, substantial innovations in
web architecture are essentially impossible.
Mygalomorphs rarely balloon, and therefore their powers
of dispersal are limited to walking. Usually, the juveniles do
not walk far, and so mygalomorph populations are highly
clumped: when you find one, its siblings and cousins are
generally not far away. Mygalomorph known species diversity is barely 7% of araneomorph diversity, so the diversification rates of these two sister taxa clearly differ, but
the reasons remain mysterious. The contrast in dispersal
mechanisms may be a partial explanation, but it also may
be that mygalomorph species are simply much more difficult to discriminate morphologically. Araneomorphs lead
much more vagile lives (including dispersal by ballooning),
and the group is much more diverse.
Within mygalomorphs, the atypoid tarantulas (Atypidae,
Antrodiaetidae) seem to be the sister group of the remaining lineages (Raven 1985a, Goloboff 1993), although some
evidence suggests including Mecicobothriidae in the atypoids. The sister group to the atypoids is the Avicularioidea, of which the basal Dipluridae may be a paraphyletic
assemblage (Goloboff 1993). One of the larger problems
in mygalomorph taxonomy worldwide concerns the paraphyletic Nemesiidae, currently 38 genera and 325 species
(Goloboff 1995). The remaining mygalomorph families
(represented in North America only by Ctenizidae, Cyrtaucheniidae, and Theraphosidae, Fig. 2.2) divide into two
distinct groups: the theraphosodines baboon spiders or
true tarantulas and their allies (Prez-Miles et al. 1996)
and the typically trap-door dwelling rastelloidines (Bond
& Opell 2002). Less work has gone into mygalomorph phylogeny, and because the diverse Dipluridae, Nemesiidae,
and Cyrtaucheniidae seem to be paraphyletic, the number
of mygalomorph families may increase substantially. For
all their size and antiquity, mygalomorphs have remarkably
uniform morphology. Because fewer reliable, distinctive
features can be compared, mygalomorph phylogeny has
been a difficult and frustrating subject. Perhaps molecular
data will advance the subject.
Araneomorphs include over 90% of known spider species: they are derived in numerous ways and appear quite
different from mesotheles or mygalomorphs. Mesotheles
are the only spiders with an anterior median pair of distinct
spinnerets and mygalomorphs have lost them completely.
A complex, important synapomorphy of araneomorphs is
the fusion and reduction of the anterior median spinnerets
to a cribellum, a flat sclerotized plate that bears hundreds
to thousands of silk spigots that produces very fine, dry,
yet extremely adhesive, silk (cribellate silk). Spider dragline
silk is justly famous because it is tougher than Kevlar (Craig
2003, Gosline et al. 1986, 1999, Scheibel 2004), but in many
ways, cribellate silk is even more amazing. Its stickiness
seems to be based on electron-electron interactions (van
der Waals forces) between the silk and the surface to which
it sticks (Hawthorn & Opell 2002). Insofar as other natural
and man-made glues operate on gross chemical principles,
this atomic-level universal glue mechanism also seems
worthy of biotechnological attention.
Other animals use silk throughout their lives, but no
other group of animals even comes close to the diversity,

22

intricacy, and elegance of silk use by spiders (Eberhard

1990, Craig 2003). Systematists first became aware of this
richness through comparative studies of behavior (Eberhard 1982, 1987b, 1990, Coddington 1986b, c), and only
later by paying attention to the morphology that produced
all that diversity. Although cytologists had been studying spider silk glands since the early 20th century (Kovoor
1987), it was not until the advent of the scanning electron
microscope, and the cladistic reinterpretation of cytological data in the light of spigot diversity (Coddington 1989)
that systematists began to plumb the immense variation in
spinneret spigots and silks for phylogenetic research. Now
spigots, silks, and silk use are some of the richest sources
of comparative data in spiders (e.g., Platnick 1990a, Platnick et al. 1991, Eberhard & Pereira 1993, Griswold et al.
1999a).
Although many araneomorph lineages independently
abandoned the sedentary web-spinning lifestyle to become
vagabond hunters, the plesiomorphic foraging mode seems
to be a web equipped with cribellate silk. Paleocribellatae
contains just one family, Hypochilidae (two genera, 11
species). It is a famous North American taxon because it is
sister to all remaining araneomorphs (Neocribellatae), and
therefore retains a fair number of primitive traits (Platnick et al. 1991, Catley 1994). Within Neocribellatae, the
monophyly of Haplogynae is weakly based on cheliceral
(chelicerae fused with a lamina instead of teeth), palpal
(tegulum and subtegulum fused rather than free), and
spinneret characters (vestiges of spigots in former molts
lacking) (Platnick et al. 1991, Ramrez 2000). The cribellate
Filistatidae (three North American genera) is is sister to the
remaining haplogynes, and a quantitative phylogeny of the
family has been published (Ramrez & Grismado 1997).
All haplogynes except Filistatidae evidently lost the cribellum, but Pholcidae (Huber 2000), Diguetidae, Ochyroceratidae, and, debatably, Scytodidae and Segestriidae still
build prey-catching webs. The cellar spiders (Pholcidae)
are exceptional for their relatively elaborate, large webs.
Some pholcid genera have independently invented viscid
silk (Eberhard 1992). Some of the most common and ubiquitous synanthropic spider species are pholcids, so one
should not assume that a taxon that branched off relatively
early in evolution is necessarily primitive, poorly adapted,
or non-competitive. The remaining haplogyne families live
either in tubes (Dysderidae) or are vagabonds that tend to
occur in leaf litter or other soil habitats and are not commonly encountered by the casual collector. Because they
are still poorly known, many of the new spider species discovered each year tend to come from haplogyne families.
The araneomorph Entelegynae is supported by several
important, yet poorly understood synapomorphies (Griswold et al. 1999a) in reproductive and spinning systems.
Entelegynes have more complex reproductive systems
in which the female genitalia (epigynum) has external
copulatory openings. In all other spiders copulation takes
place through the gonopore. This secondary set of openings provides a flow-through sperm management system:
the male deposits sperm in the epigynum that connects to
spermathecae that connect to the uterus. This has major
implications for reproductive behavior and the relation
between the sexes (Eberhard 2004). For one thing, the
flow-through system means the first male to mate with a
female usually sires most of the spiderlings (Austad 1984).
The copulatory ducts that lead from the epigynum to the
spermathecae are often extremely contorted in entelegyne
females. This has led to hypotheses that such complexity
actually make it more difficult for males to inseminate
females (Eberhard 1985, 1996). For whatever reason, fer-

tilization ducts in entelegyne spiders have been secondarily

lost only five times and in mostly small groups: twice in
distal palpimanoid families (this may be primary absence
rather than secondary loss, Huber 2004), a small sub-clade
of uloborids, some anapids, and a rather uniform, if speciose, sub-clade of tetragnathids (Hormiga et al. 1995). A
second consistent but enigmatic entelegyne synapomorphy
is cylindrical gland silk. These glands and spigots appear
only in adult females, and it is thought that the silk is used
only in egg sacs, but the specific contribution of cylindrical
gland silk to egg sac function remains unknown.
Male entelegyne genitalia are also greatly modified. Plesiomorphic male spider genitalia are usually simple, tapering, pyriform bulbs. Pyriform bulbs lack apophyses or, if
apophyses are present, they are small and quite simple
(e.g., Fig. 2.3). Other parts of the male palp, such as the
cymbium, patella, and tibia, are likewise unadorned. In
contrast, entelegyne male genitalia can be bewilderingly
complex. The bulb has two or three divisions (always
subtegulum and tegulum, sometimes with an elaborate
embolic division. The tegulum usually has two apophyses (conductor and median apophysis) in addition to the
embolus. Any or all of these in entelegynes can be wonderfully complex, with knobs, levers, grooves, hooks, serrations, sinuous filaments, and spiraling parts (e.g., Figs.
2.4-2.5). Unraveling the homology of entelegyne male
genitalia is a major problem (Coddington 1990). Entelegyne bulbs also work differently. The plesiomorphic bulb
ejaculates via muscles that force the sperm out. Entelegyne
bulbs lack those muscles and work hydraulically instead.
The male pumps blood into the bulb to raise its internal
pressure, which serves both to expand and uncoil its various parts. Glands empty their contents into the sperm duct
and force the sperm out (Huber 2004). For sperm transfer
to occur, this complicated structure must interact precisely
with the correspondingly complex female genitalia (Huber
1994, 1995). In addition, the various parts of an entelegyne
bulb are usually connected only by thin, flexible membranes that inflate like balloons during copulation. As a
whole the bulb is so flexible that at least some of the males
complexity doubtless serves only to stabilize and orient his
own genitalia during copulation.

2.4

2.3

2.5

Spiders of North America

Female epigyna have corresponding ledges, pockets,

ridges and protuberances externally, and often labyrinthine
ductwork internally. One hypothesis is that the female
complexity is essentially defensive, the result of antagonistic co-evolution (Chapman et al. 2003): females as a whole
invest so much more in their offspring than do males
(even in spiders), that they should choose mates carefully
(Alexander et al. 1997). Female genitalic complexity may
be a challenge to males such that only high-quality males
succeed, or succeed much better than low-quality males.
Males, in turn, have evolved complex and highly flexible
genitalia, the better to win over choosy females. This explanation assumes, of course, that overall quality of males is
tightly correlated with male copulatory prowess. In any
event, the difficulty of attaching to and navigating female
genitalia may give the female more time to assess the male
and break off mating if she chooses.
Three small families, known as eresoids, have thus far
always clustered near the base of Entelegynae in phylogenetic analyses: Oecobiidae, Hersiliidae, and Eresidae (Coddington 1990, Griswold et al. 1999a). Only the first two
occur in North America. Their phylogenetic relationships
are controversial. Oecobiids (Glatz 1967) and hersiliids (Li
et al. 2003) share a unique attack behavior: they are the only
spiders known to run swiftly around the stationary prey
encircling it with silk as they go. The behavior could also be
convergently evolved, although in accordance with the null
hypothesis of phylogenetics, until proven otherwise, we
presume the similarity is explained by descent. Certainly
no obvious features tie any of these families closely to any
other entelegynes. Because they are entelegyne yet share no
more derived apomorphies with other entelegyne clades,
they seem to be the basal entelegyne group.
The Palpimanoidea is a controversial entelegyne group
(10 families, 54 genera), of which only the pirate spiders
(Mimetidae) occur in North America. For years mimetids
were thought to be araneoids based on setal morphology,
general appearance, and the overall complexity of their
palps, but then were transferred to palpimanoids (Forster
& Platnick 1984). Recent research, however, suggests that
mimetids are araneoids after all, in which case Palpimanoidea is polyphyletic (Schtt 2000, 2003, Huber 2004).
Males can also have knobs or apophyses elsewhere on
their palpi. About half of entelegyne species have one in
particular, the retrolateral tibial apophysis, that defines a
clade of 39 entelegyne families (the RTA clade: Sierwald
1990, Coddington & Levi 1991, Griswold 1993, Fig. 2.2).
Huber (1994, 1995) found that the RTA usually, but not
always, serves to anchor and orient the male bulb to the
female genitalia prior to expansion of the hematodochae.
Orbiculariae is one of the largest entelegyne lineages.
It consists of two superfamilies, Araneoidea (13 families,
1,000 genera), and Deinopoidea (2 families, 22 genera).
The monophyly of Orbiculariae is controversial because
the strongest apomorphies are all behavioral: both groups
spin orbwebs (Coddington 1986c and references therein).
Prior to strictly phylogenetic classification in spiders the
two groups were thought to be only distantly related. The
cribellate Deinopidae and Uloboridae were included in
the Cribellatae, and authors often commented on the
detailed similarity of these orbs to those of the classical,
ecribellate Araneidae (reviewed in Scharff & Coddington
1997). However, as noted above, the cribellum is primitive
for Araneomorphae. On the one hand, araneoids or their
ancestors must therefore have lost it and, on the other,
groups based on plesiomorphies are false. When the old,
polyphyletic Cribellatae collapsed (Lehtinen 1967, Forster
1967, 1970b), deinopids and uloborids had nowhere else to

23

go, as it were, and so the form of the web (and the striking
similarities in behavioral details) constituted a strong block
of synapomorphies. But if orbweavers were monophyletic,
the six araneoid families that spin sheet or cobwebs must
have lost the orbweb. Against this view is the hypothesis
that the orbweb is an unusually efficient and profitable
design to catch prey. In general the more adaptive a feature
is, the more likely it is to evolve independently; perhaps
the araneoid and deinopoid web forms are convergent.
This view argues that the orbweb is so superior a predation strategy that any spider lineage capable of it would
have evolved it independently (and never lost it). Little evidence thus far suggests that orbwebs are drastically better
than other web architectures (although they are widely
regarded as better-looking!). Indeed, ecological evidence
points the other way (Blackledge et al. 2003). Another difficulty for the monophyly hypothesis is that the deinopoid
orb is cribellate (dry adhesive silk), and the araneoid orb
uses viscid silk. The missing link, it is argued, would
have had neither. The obvious rejoinder is that perhaps
they had both at one point, but one of the good effects of
modern quantitative analysis is that people spend less time
arguing about irresolvable issues, and more time seeking
new evidence. The orb web diphyly argument particularly
needs evidence that deinopoids share strong synapomorphies with some non-orb weaving group. Evidence against
orbweaver monophyly is starting to appear from molecular
evidence (Hausdorf 1999, Wu et al. 2002), but these studies
are small, omit many important taxa, and do not confirm
each others results.
Araneoidea (ca. 11,000 species) is much larger than
Deinopoidea (ca. 300 species). Only one deinopid species occurs in North America (in Florida and, possibly,
Alabama). Araneoids are ecologically dominant species
throughout the world but especially in north temperate
areas such as North America, where Linyphiidae swamps
any other spider family in both species diversity and sheer
abundance. Current phylogenetic results (Hormiga 1994b,
2000, Griswold et al. 1998) indicate that Linyphiidae and
five other families form the monophyletic araneoid sheet
weaver clade, which thus implies that within Araneoidea,
the orb was lost only once (or transformed into a sheet
web). Linyphiidae spin sheets as do Pimoidae. The classic cobwebs of Theridiidae and Nesticidae would then
be derivations from a basic sheet, which, considering the
web of black widows, Steatoda, and other apparently basal
theridiid genera (Benjamin & Zschokke 2002, Agnarsson
2004, Arnedo et al. 2004, ), seems plausible. Araneoid sheet
web weavers account for the bulk of araneoid species diversity (713 genera, 7,600 species worldwide). Perhaps sheet
or cobwebs are not so bad after all (Griswold et al. 1998,
Blackledge et al. 2003).
Although the most recent analysis suggests that the
sister taxon of Orbiculariae is approximately all remaining entelegyne families (possibly including eresoids and
palpimanoids), the evidence for this is quite weak for several reasons (Griswold et al. 1999a). First, non-orbicularian entelegyne families have received little phylogenetic
research, so such overarching conclusions are premature.
Second, the problem is intrinsically difficult. Resolving the
entelegyne node requires an analysis that includes several
representatives from all major entelegyne clades, including
relevant enigmas such as Nicodamidae (Harvey 1995) and
Zodariidae (Jocqu 1991a). That means a very large matrix
and an even larger scope of characters. Such a matrix is not
easily constructed, and will probably require collaboration
of numerous specialists.

24

Although araneologists refer to large groups like the

amaurobioids (Davies 1998a, 1999, Davies & Lambkin
2000, Wang 2002), lycosoids (wolf spiders, Griswold
1993), and Dionycha (two-clawed hunters, Platnick
1990a, 2000a, 2002), their monophyly is also tenuous at
best. Amaurobioids (which currently contains lycosoids as
a subgroup) are defined by a few small changes in spinneret
spigot morphology only visible with the scanning electron
microscope. Basal amaurobioid families present in North
America are Desidae, Amaurobiidae, and Agelenidae (a
mixture, by the way, of cribellate and ecribellate groups).
The group is quite heterogeneous, including everything
from hunters to elaborate web builders. Dictynidae may fall
close to these families as well (e.g., Bond & Opell 1997).
Lycosoids were formerly thought to be defined by quite
an unusual and convincing synapomorphy, the grateshaped tapetum. The tapetum is a reflective layer within
the eye that probably serves to increase sensitivity. In most
lycosoids, the tapetal architecture is like a barbecue or
street-drain grate -an arrangement of parallel bars and
holes, whereas in other spiders the tapetum shows no particular pattern, or is in the form of a simple canoe (Canoe
Tapetum Clade, Fig. 2.2). The reflection of the grate-shaped
tapetum of the posterior median or lateral eyes can be visible at great distances in the field. At night it is common
with a headlamp to see the green eye shine of a lycosoid
5, or even 15 meters away. In the latest analysis, however,
the grate-shaped tapetum evolves twice (in stiphidiids and
lycosoids sensu stricto). In North America, the lycosoid
families are Ctenidae, Lycosidae, Miturgidae, Oxyopidae,
Pisauridae, Trechaleidae, and Zoropsidae (introduced).
Among these families a few genera still spin webs, but the
majority have given up webs for a vagabond lifestyle.
The monophyly of Dionycha is equally tenuous.
Dionycha is an old hypothesis -at one time many classifications divided entelegyne families into two-clawed
(Dionycha) and three-clawed spiders (Trionycha). But
three claws is primitive for a group even larger than all
spiders, so a group defined by it would be paraphyletic.
Two claws, however, is arguably a derived condition. This
argument is weak (some spider groups placed elsewhere are
two-clawed), and if strong evidence connected a dionychan
family elsewhere, it would be preferred. As it happens,
however, phylogenetically rigorous arguments link any
dionychan group outside Dionycha, so the group stands
solely on the simple fact of the two-clawed condition. One
group within Dionycha, the Gnaphosoidea (7 families,
Gnaphosidae and Prodidomidae in North America), does
share an apomorphy with interesting functional implications (Platnick 1990a, 2000a, 2002): all have obliquely
angled posterior median eyes with flat, rather than rounded
lenses. As a flat lens cannot bend light, and there seems little
reason otherwise to have eyes, the flat lens has always been
a mystery. It turns out that at least the gnaphosid Drassodes
uses these modified eyes to orient to polarized light, which
in turn allows them to move about in the habitat and to
return to the same spot (Dacke et al. 2001a, b).
The phylogenetic relationships of these non-orbicularian entelegynes, therefore, is poorly known. Because they

were originally defined by plesiomorphies, many of the

classical entelegyne families (most seriously Agelenidae,
Amaurobiidae, Clubionidae, Ctenidae, and Pisauridae)
were, and still probably are, paraphyletic. Dismembering these assemblages into monophyletic units has been
difficult because the monophyly of their components or
related families is also often doubtful (e.g., Amphinectidae,
Corinnidae, Desidae, Liocranidae, Miturgidae, Tengellidae,
Stiphidiidae, Titanoecidae). Corinnids, liocranids, zoropsids, and ctenids have been recently studied (Bosselaers &
Jocqu 2002, Silva Dvila 2003), and neither analysis recovered a monophyletic Liocranidae, Corinnidae, or Ctenidae
(indeed, rather the opposite). Bosselaers (2002) analyzed
Zoropsidae, and it does seem to be monophyletic. The
basic phylogenetic structure of Anyphaenidae was studied
by Ramrez (1995a, 2003).
Therefore neither the RTA clade, nor the two-clawed
hunting spider families (Dionycha) may be strictly monophyletic, although each presumably contains within it a
large cluster of closely related lineages. Dionychan relationships are quite unknown, although some headway has
been made in the vicinity of Gnaphosidae (Platnick 1990a,
2000a, 2002). In contrast, Lycosoidea was supposedly based
on a clear apomorphy in eye structure, but recent results
suggest that this feature evolved more than once, or, less
likely, has been repeatedly lost (Griswold et al. 1998). The
nominal families Liocranidae and Corinnidae are massively polyphyletic. The nodes surrounding Entelegynae
will certainly change in the future.
In summary, most of the major clades in 20th century
spider classifications were fundamentally flawed, which
means books and overviews published prior to the last
two decades have been superceded. Major lineages such
as Trionycha, Cribellatae, Tetrapneumonae, Orthognatha,
and Haplogynae (older definition) were all based on plesiomorphies. By the mid-1970s the higher-level classification
of spiders had collapsed to the extent that catalogs of that
era began to ignore higher classification (and still do). From
that rubble has emerged the hypothesis presented in Figure
2.2, but events on the horizon suggest that its clarity may
be shortlived. For one thing, a small cadre of workers from
the mid 1980s on tried to cover as much ground as possible on the family-level, via studies that barely overlapped.
Thus, the various studies underlying Figure 2.2, although
quantitative, have not been tested by other workers, denser
taxonomic sampling, or new sources of data. Figure 2.2
remains a first draft of spider phylogeny. As the latter processes proceed, our understanding of spider phylogeny will
doubtless improve. The underlying observations are solid,
and as more data accumulates, we can expect more stability
in the results. In sum, phylogenetic understanding of spiders has advanced remarkably since the early 1980s. We are
approaching a truly quantitative estimate of spider phylogeny at the family level, but phylogenies below that are going
to require much denser taxonomic sampling and rigorous
comparative study.