BiologicalJoumal ofthe Linnean Sociep (1998), 63: 81-98.

With 3 figures

The significance of relatedness and gene flow

on population genetic structure in the
subsocial spider Eresus cinnaberinus
(Araneae: Eresidae)

JES JOHANNESEN', THOMAS BAUMANN', ALFRED SEITZ' AND

MICHAEL VEITH'
'Institut @r <oologie, Universitat Main<, Saarstrge 21, 0-55099 Main<, Germany;
21nsitut@r<oologie, Martin Luther-Universitat Halle- Wttenberg Kriillwit<m Stx 44,
0-06099 Halle, Germany

Received 2 9 May 1997; acceptedfor publication 4 September 1997

Interdemic selection, inbreeding and highly structured populations have been invoked to
explain the evolution of cooperative social behaviour in the otherwise solitary and cannibalistic
spiders. The family Eresidae consists of species ranging from solitary and intermediate
subsocial to species exhibiting fully cooperative social behaviour. In this study we, in a
hierarchical analysis, investigated relatedness of putative family clusters, inbreeding and
population genetic structure of the subsocial spider Exsus cinnabffinus. Five hierarchical levels
of investigation ranging from large scale genetic structure (distances of 250 and 50 km level
1 and 2) over microgeographic structure (20 h'and 4 km', level 3 and 4) to a single hill
transect of 200 m (level 5 ) were performed. The purpose of level 5 was two-fold (1) to
investigate the relatedness of putative family groups, and (2) to evaluate the influence of both
family living and sampling design on higher level estimates. Relatedness estimates of putative
family groups showed an average relatedness of R=0.26. There was no indication of
inbreeding. In contrast to social spiders, genetic variation was abundant, He=0.10. The
population genetic structure was intermediate between social and asocial spiders. Genetic
variance increased continually across hierarchical levels. Family structured neighbourhoods
biased differentiation estimates among level 5 samples (FsT= 0.04) and level 3 and 4 samples
(0.07<FsT<0.18),and apparent inbreeding among level 3 and 4 samples, &>O, was caused
by disjunct sampling from separate neighbourhoods. Larger scale samples were highly
differentiated 0.12<FsT<0.26, depending on level and sampling design. Due to a distance
effect family living did not influence estimates of the higher level 1. Although the dispersing
sex among social spiders and the subsocial E. cinnebarinus differ, females versus males, female
behaviour of both sociality classes lead to high genetic variance.
0 1998 The Linnean Society of London

ADDITIONAL KEY WORDS:-Arachnida - dispersal - isolation - evolution of sociality

- allozymes.

Correspondence to: J. Johannesen. E-mail jesjo@hydra.biologie,uni-mainz.de.

81
0024-4066/98/010081+ 18 $25.00/0/bj970186 0 1998 The Linnean Society of London
82 J. JOHANNESEN ETAL.

CONTENTS

Inroduction . . . . . . . . . . . . . . . . . . . . . . . 82
Material and methods . . . . . . . . . . . . . . . . . . . 84
Sampling design . . . . . . . . . . . . . . . . . . . . 84
Electrophoresis . . . . . . . . . . . . . . . . . . . . 87
Genetic data analysis . . . . . . . . . . . . . . . . . . 88
Results . . . . . . . . . . . . . . . . . . . . . . . . 89
Genetic differentiation . . . . . . . . . . . . . . . . . . 89
Relatedness . . . . . . . . . . . . . . . . . . . . . 91
Discussion . . . . . . . . . . . . . . . . . . . . . . . 93
Relatedness and genetic structure . . . . . . . . . . . . . . 93
Implications for evolution of sociality in spiders . . . . . . . . . . 95
Acknowledgements . . . . . . . . . . . . . . . . . . . . 96
References . . . . . . . . . . . . . . . . . . . . . . . 96

INTRODUCTION

Three main hypotheses have been put forward to explain altruistic behaviour:
kin selection, group selection and reciprocity (Michod, 1993). Family structured kin
selection models assume group-living of related individuals among which interactions
take place (e.g. Wade & Breden, 1981; Michod, 1982; Uyenoyama, 1984). Such
models have been a powerful tool to explain the evolution of social behaviour in
the haploid-diploid genetic system of Hymenoptera (Michod, 1993). Individual
selection, which within groups opposes the evolution of altruism, may be overcome
by genetically homogenizing group members through inbreeding or interdemic
selection and thereby increasing genetic variance among groups. Among diploid
organisms group selection and/or inbreeding are thought to play an important role
in the evolution of social behaviour leading to highly structured populations in, for
example, termites (Redly, 1987; Kaib et al., 1996), naked mole-rats (Reeve et al.,
1990), shrimps (D@, 1996), and spiders (Lubin & Crozier, 1985; Smith, 1986;
Roeloffs & Riechert, 1988; Smith & Engel, 1994; Smith & Hagen, 1996). In the
study of evolution of social behaviour subsocialityprovides an opportunity to compare
the genetic status of species which are intermediate between asocial and fully social.
In contrast to ants and termites, which all are social, spiders show varying degrees
of social behaviour.
The majority of spiders are solitary, territorial and cannibalistic. In at least seven
families a few species exhibit social cooperative behaviour (AvilCs, 1997), although
within each family most species retain the original solitary and cannibalistic way of
life. It therefore seems likely that sociality in spiders has arisen independently several
times (Kullman, 1972;Wickler & Seibt, 1993).Social spiders show no morphologically
different or sterile castes and most individuals within colonies reproduce (Riechert
& Roeloffs, 1993). Social spiders therefore are not eusocial as are ants and termites.
However, social spiders do experience conditions hypothesized as being prerequisite
for eusociality among diploid organisms, namely gradual metamorphosis, extensive
parental care (subsociality), and long lasting niches (Alexander, Noonan & Crespi,
1991). Kullmann (1972) defined three criteria with which to recognize a spider as
social: tolerance, interaction and cooperation. The threshold step seems to be
tolerance among reproductives (Kullmann, 1972).
Evolution of social and altruistic behaviour in spiders has been explained in a
variety of contexts. Individual selection and mutualistic advantage of snare-building
 GENETICS OF A SUBSOCIAL SPIDER 83

and predation have been advocated for evolution of altruistic behaviour (Brach,
1977). However, cooperative hunting which is common to all social spiders is also
thought just to be a consequence of an aggregated lifestyle, and not the driving
force (Packer & Ruttan, 1988; Seibt & Wickler, 1988). Within the family Eresidae
pedogenesis, i.e. retention of juvenile characters, has been put forward to explain
tolerance among adults (Wickler & Seibt, 1993).This hypothesis, which is based on
fewer moults by social than by subsocial Eresids to reach sexual maturity, may, on
the other hand, be a consequence of reduced growth and not a precondition for
sociality. Under good food conditions subsocial Stegodyphus lineatus will tolerate each
other beyond the juvenile stage (Schneider, 1995). Interdemic selection, which
involves differential colony extinction and proliferation relative to the individual,
and inbreeding have been favoured in recent years as explanations of social evolution
(Avilts, 1986; Roeloffs & Riechert, 1988; Smith & Hagen, 1996). A phenomenon
observed in all social spiders is a female biased sex ratio (e.g. Vollrath, 1982; Elgar
& Godfray, 1987; Avilts, 1986, 1993; Lubin, 1991). All social spider colonies
propagate by budding or swarming of nest related females (Vollrath, 1982; Lubin
& Robinson, 1982) and inbreeding is thought to be pervasive. High levels of
inbreeding and relatedness among females would, in the light of local mate com-
petition, bias the sex ratio towards the dispersing sex (Bulmer, 1986). Riechert and
Roeloffs (1993) mention little or no courtship in social spiders which is expected
without mate competition, and would be suggestive of inbreeding as there is no
selective need for males to compete as all are genetically identical. However, mate
competition is observed in several social spiders (Henschel, Lubin & Schneider,
1995).Genetic studies have found very little genetic variation within colonies. Indeed,
neighbouring colonies may by fixed for alternative alleles (Lubin & Crozier, 1985;
Smith, 1986; Roeloffs & Riechert, 1988; Smith & Engel, 1994; Smith & Hagen,
1996).
The gap between cannibalistic and social spiders seems difficult to bridge. A few
species of spiders are considered subsocial: they experience extended parental care
and may, after dispersing from the natal nest, live communally for a period of their
life. For these species no genetic research has been done and several questions arise
in a study of intermediate sociality comparative to social species. Do intermediate
social species outbreed or is there indication of inbreeding, are individuals distributed
non-randomly, e.g. in family groups, or do they disperse widely?
Eresus cinnabmhus ( O h . , 1789) (formerly E. nger Pet.) is a periodically social, or
subsocial, spider. It belongs to the cribellate spider family Eresidae in which
permanent social behaviour has been described from five species (Riechert &
Roeloffs, 1993).Eresus cinnaberinus is found in a variety of dry biotypes, such as grass-
and heathland and stony steppe habitat (Baumann, 1996 and references therein).
The subsocial phase of E. cinnaberinus is characterized by the first weeks the newly
emerged spiderlings spend in the maternal nest. The female catches prey and feeds
the young by regurgitation. When the female dies she is consumed by the spiderlings.
Only after this extended tolerance phase do spiderlings disperse, most probably only
within the close vicinity of the maternal nest (Ratschker, 1995). From now on they
live solitary underground in silk-lined burrows. Cooperative predation and occasional
cohabitation among juveniles has been reported (Holl & Reinbach, 1991). The tube
webs are often found in a clumped distribution. It seems that once a burrow has
been established an individual does not disperse any further. Upon reaching sexual
maturity males disperse whereas females probably stay in their burrows. Baumann
84 J. JOHANNESEN ETAL.

(in press), in a three year study of dispersal, collected only six females and juveniles
in pitfall traps. In a mark-recapture study of dispersing males a median dispersal
distance of 13 metres, with a maximum distance of 59.43 metres, was observed,
N= 540 (Baumann, in press). Ballooning has never been observed (Ratschker, 1995).
In the study area E. cinnaberlnus females live for about 4 years and males 2-3 years.
In this study we investigated relatedness and gene flow within and among groups
of E. cinnaberinus. The aim of this study was to estimate the amount and distribution
of genetic variation, for comparison with that of asocial and social spiders in an
attempt to infer possible genetic predispositions for a transition to more permanent
social behaviour. More specifically we were interested in how to assess where a
population, or rather deme, begins and ends, whether inbreeding take place, whether
web clusters consist of related individuals and the extent of local mating.
We applied two approaches. First we estimated relatedness of individuals within
putative family groups and differentiation among these in a single habitat patch.
Secondly, we compared more geographically widespread samples in order to
determine larger scale differentiation. By employing these two approaches we wished
to gain insight into how family structure and/or sampling artefacts may influence
the results of the larger scale analyses, and thereby infer at which hierarchical level
a population and/or deme can be defined. A family structure of living may bias the
estimated differentiation among populations towards an apparently stronger degree
of isolation than is actually present. If this is the case it would be appropriate to
investigate the relatedness of individuals within groups and not just the differentiation
among groups (Pamilo, 1984).

MATERIAL AND METHODS

Sampling desZgn

Samples were collected in three German states: Rheinland-Pfalz, Sachsen-Anhalt,

and Thuringen (Fig. 1 and Table' 1). The sampling design consisted of five
geographical levels of analysis (Fig. 2). Samples of levels 1-4, which explore larger
scale differentiation, are designated with capital letters (RP and SA-A to SA-H)
(Table l), whereas samples of level 5, which investigates genetic structure along a
200 m hill transect, are designated with lower case letters (a-c). Samples were divided
into sub-samples. Abbreviations of sub-samples of levels 1-4. were given an area
coding RP or SA and a number, e.g. RP 1.1 or SA 4.1, where RP 1.1 signifies
Rheinland-Pfalz sub-sample no 1 (*.I), which is only included in level 1 analyses
(1 .*), and SA 4.1 signifies sub-sample no 1, which is included in all analyses to level
4 (table 1). Level 5 sub-samples were numbered 5.1 to 5.15 (Fig. 3). Individuals of
level 5 were only included in level 5 analyses.
The first level analysis included all sampling locations. In Rheinland-Halz col-
lections were made from six locations. These consisted of three central localities
(RP-1 to 3) with a distance of c. 4 km and three outlying localities c. 50 km away.
These six locations were considered sub-samples of just one Rheinland-Pfalz sample
since Rheinland-Halz in southwest Germany was used only as an outgroup com-
parison for large scale differentiation. The second level included the sub-samples
 GENETICS OF A SUBSOCIAL SPIDER 85

CIMRITZ (level 4)

BRANDBERGE

Figure 1. A schematic presentation of geographic locations, with a map of level 3 (Halle) and, in the
upper left hand corner, level 4 (Gimritz) sampling locations. Localities were situated in Rheinland-
Pfalz (RP) and Sachsen-Anhalt/Thilhgen (SA and TH). Ellipsoids (SA-D to SA-H) indicate samples
with 2 4 sub-samples, circles (SA-C) with one sub-sample.

from Kyflhauser in Thiiringen, and Gliicksburger Heide and Halle in Sachsen-

Anhalt. The greatest distance between level 2 samples was similar to the Rheinland-
Pfalz samples, about 50km. The third and fourth level of analysis was performed
at Halle (Fig. 1). This area consists of porphyry and calcareous hills disjunctly
distributed in the landscape. Hills are separated by farm land or urban development.
The third level constisted of samples from Franzigmark (SA-D), the sample SA-C,
and Gimritz (SA-E to SA-H). The Gimritz Nature Reserve (level 4) consisted of
four samples collected along hill slopes. Samples from Gimritz consisted of 2-4 sub-
samples of 8-10 individuals. A total of 14 sub-samples were obtained. A sub-sample
was collected within an area of approximately 10 x 10 metres. Caution was taken
not to collect from adjacent webs. The distance between sub-samples within a
sample was 100-150m. The habitat between sub-samples was not in all cases
continuos, i.e. apparent sub-optimal habitat could divide these. The pattern of sub-
sampling at Gimritz was repeated for the slope at Franzigmark (three sub-samples),
4 km distance. Additionally, two sub-sampleswere obtained from the isolated patches
SA-3.1 and SA-3.2 about 1 and 4 km from Franzigmark, respectively. In the data
analysis these two sub-samples were treated as one sample, SA-C (Table 1) to
enhance sample size.
86 J. JOHANNESEN ETAL.

TABLE1. Sample locations and sample sizes. Abbreviations include an area coding RP or SA and a
number, e.g. Rp I. 1 or SA 4.1, where the first digit indicates the sampling level, and where the second
digit is the sub-sample of a given level. See also Figure 1 for map locations

Sample location Abbreviation Sample Sample Sub-samplc

designation size size
~~ ~

RHEINLAND-PFW (RP)
NSG Martinstein RP-1.1 RP 60 14
Kunoweg, SchoObdckelheim RP-1.2 7
Idar-Obentein RP-1.3 11
NSG Rotenfels, Bad Munster am Stein RP- 1.4 11
NSG Pommern RP-1.5 11
Battenberg, Pfalz RP-1.6 6
SACHSEN-ANHALT/TH~~INGEN (SA)
Kyffhauser SA-2.1 SA-A 10 10
Glucksburger Heide SA-2.2 SA-B 24 24
Halle
Lunzberg SA-3.1 SA-C 18 8
Brandberg SA-3.2 10
Franzigmark
Franzigmark 1 (Stahl) SA-3.3 SA-D 30 10
Franzigmark 2 (Draht) SA-3.4 10
Franzigmark 3 (Beton) SA-3.5 10
Gimritz
Teichgrund Cdunetum SA-4.1 SA-E 37 8
Teichgrund oben SA-4.2 9
Teichgrund DFL SA-4.3 10
Teichgrund Zufahrt SA-4.4 9
Saalehang SA-4.5 SA-F 40 10
Zornberg SA-4.6 10
Lauchengrund sudl. Eisenbahn SA-4.7 10
Lauchengrund SA-4.8 10
Kuppc I36 SA-4.9 SA-G 38 10
Sudspitze Kuppe I3 SA-4. I0 10
Wddchen SA-4. I 1 10/ 159'
Hang sudl. Wittsack SA-4.12 8
HoU-Hang SA-4.13 SA-H 20 10
Hollunder-Hang SA-4.14 10
~ ~~ ~

' Individuals for level 5 analysis.

Finally, analysis at the fifth and lowest level took place at Waldchen in the Gimritz
Nature Reserve along a 200 m hill transect. Here the Gimritz sampling design was
repeated, with a similar number of sub-samples (1 5 ) and individuals (159) (Fig. 3).
Each sub-sample (5.1-5.15) corresponded to a putative family group. Three samples
were defined. A main sample, sample by of nine sub-samples was situated in the
centre of the transect. These sub-samples were collected within an area of ap-
proximately 5 x 40 metres. Samples a and c were situated 25 m and 80 m from
sample b, respectively. Samples a and b were divided by apparently sub-optimal
habitat. The area of sample a was 2 x 4 meters; of sample c 5 x 10 meters. Sub-
samples consisted of 5-22 individuals. For all individuals located coordinates and
age were registered. In contrast to the situation at Gimritz, individuals from sub-
samples were often collected in the immediate vicinity of each other. The individuals
of the Mth (Waldchen) level were not included in the higher (1-4) level analyses in
order to avoid sample size biased results.
 GENETICS OF A SUBSOCIAL SPIDER 87

Level 1. Total

Thiiringen(1)

L I
I-.
Level 2. SACHSEN-ANHALT/THLJRI"GEN

Giicksburger- 30-50 b Halle (19)

Heide (1)
KyfFshauser (1)

Franzigmark (3)
Brandberg (1)
Lunzberg (1)

Level 4. GIMRITZ (4 b2)

Gimritz, 4 samples-14 samples

Distance between samples 0.5-2 km

Level 5. W (200 m transect)

~ C H E N

Waldchen (SA-4.11)3 samples-15 sub-samples*

*1 sub-sample not included in 3 sample FsT estimation

Figure 2. Hierarchical sampling design for Erem cinnaberinus consisting of five levels. The number of
sub-samples within levels 1-3 are given in brackets. Location names are given in Table 1.

ElectmphoresiJ

Spiders were stored in liquid nitrogen. Abdominal tissue was homogenized in

80ml Pgm-buffer (Harris & Hopkinson, 1978) by ultrasound and centrifuged at
14 000 rpm for 2 minutes. Analysis was done by cellulose acetate electrophoresis
(Hebert & Beaton, 1993). A total of 18 enzyme systems representing 24 loci could
be scored: Aat-2, -2 (EC 2.6.1. l), Acon (EC 4.2.1.3), Adh (EC 1.1.1. l), Ak (EC 2.7.4.3),
Apk-2, -2 (EC 2.7.3.3), Fum (EC 4.2.1.2), G-3-Pdh (EC 1.2.1.12), G-6-Pdh (EC
1.1.1.49), Gpd (EC 1.1.1.8), Hbdh (EC 1.1.1.30), Hk (EC 2.7.1.1), Zdh-2, -2 (EC
1.1.1.42),Ldh (EC 1.1.1.27-28), Mdh-2, -2 (EC 1.1.1.37), Mpi (EC 5.3.1.8), Pep-B -
2,-2,-3 (leucine-glycine-glycine) (EC 3.4.1 1 or 13), PS; (EC 5.3.1.9),Pgm (EC 5.4.2.2).
Three buffer systems were used: Tris-Maleic acid pH=7.0 (adjusted from T M
88 J. J 0 H A " E S E N E l AL.

-40 0 40 80 120 160

Metres

Figure 3. Level 5 transect sample for relatedness estimates at Wddchen. The letters a, b, and c refer
to samples for F-stahrics. Numbers 1-1 5 refer to sub-samples for F-stalktics and relatedness estimates.
The four demes for relatedness estimates (Table 4) consisted of three samples a, b, and c and the
isolated sub-sample 13.

pH = 7.8, Colgan 1986) for Acon, Ak, Apk, G-6-Pdh, Gpd, Hbdh, Ldh, P&; Pgm; Tris-
Citrate pH = 8.2 (Richardson, Baverstock & Adams, 1986) for Fum, G-3-pdh, Zdh,
Mdh; Tris-Glycine pH=8.5 (Hebert & Beaton, 1993) for Aat, Adh, Mpi,Hk,Pep-B.
All enzymes were run for 30 minutes. Adh, Ak, Gpd, Hbdh, and Pep-B were run at
200V; all others at 250V.

Genetic data anabsysis

Putative loci were analysed only if they displayed patterns consistant with known
quaternary structure (Richardson et al., 1986). If a locus only showed one band it
was interpreted as monomorphic. The furthest migrating allele was designated 1,
the second furthest 2, and so forth. Loci were tested for departure from expected
Hardy-Weinberg proportions with the Louis-Dempster (1987) exact test. Exact
probability genotypic linkage disequilibria based on the algorithim of Weir (199 1)
were tested for all subsets of samples. Allelic distribution patterns were analysed
with the exact probability test of Raymond & Rousset (1995a). This test was only
applied in the third and fourth level analyses for the comparison and interpretation
of allele distribution patterns within and among hills. The above tests were all tested
with the program GENEPOP (Raymond & Rousset, 1995b).Genetic differentiation
among samples was analysed at different levels of regional subdivision with the
estimates of &, Frr and FSTby employing the F-statistics of Weir & Cockerham
(1984).A hierarchical distribution of genetic variation was quantified with Wright's
(1978) hierarchical F-statistics. At level 5 (Waldchen),F-statistics were obtained both
for the three samples and for the 15 sub-samples. Standard deviations were obtained
using the jack-knife procedure. Finally, isolation by distance (Slatkin, 1993) was
investigated among sub-samples of level 5 to evaluate sub-sample relationships.
 GENETICS OF A SUBSOCIAL SPIDER 89

Under the equilibrium assumption of Nem= (1 /FsT-l)/4, the number of migrants

per generation (Nem)between sample pairs can be calculated. A negative linear
correlation between log (Ncm)and the log of the geographical distance (i) indicates
isolation by distance. Since pair-wise 4 . m estimates are not independent, the
association between log (Nem)and log (i) was tested with a Mantel test (subroutine
Mantel from the program package GENEPOP; Raymond & Rousset, 1995b).
Relatedness estimates were obtained using the program Relatedness 4.2b (Queller
& Goodknight, 1989). Estimates were tested for significance by a one-tailed t-test
using the number of included sub-samples as degrees of freedom. Between age-class
estimates were tested with two-tailed t-tests. All estimates were weighted by total
individuals. All relatedness estimates were tested with and without respect to deme
division. At level 5 (Waldchen) sub-samples were divided in to four demes. The four
demes correspond to samples a-c of the FqT estimates, and the isolated cluster 5.13
(Fig. 3). Average population, age-class, and nearest group relatedness were analysed.
Relatedness estimates from level 3 and 4 (Halle and Gimritz) included samples and
sub-samples from samples SA-D to SA-H.

RESULTS

Genetic dajiientiation

Levels I and 2
Across all samples, 18 of the 24 loci investigated were polymorphic (allelefrequency
tables provided on request). For all sub-samples no significant deviations from
Hardy-Weinbergproportions were observed. However, due to small sub-sample size
and the conservative nature of Hardy-Weinberg tests and this finding might not be
unexpected. The average sub-sample heterozygosity was significantly lower in
Rheinland-Pfalz, H, = 0.060 (range: 0.007-0.106), than in Sachsen-Anhalt/Thur-
ingen, He=0.094 (range: 0.024-0.144), P<0.05. There was no indication of in-
breeding, & % 0, when comparing sub-samples. However, among samples & was
significantly greater than zero (Table 2).
Genetic differentiation among sub-samples of the two German regions (level 1)
was high, FST = 0.26 (Table 2). This was primarily caused by fixation of the G$d 2-
allele, an increased P p 1-allele frequency, and no Fum variation in Rheinland-Pfalz
(Table 2). Repeating the analyses for the samples (we treated Rheinland-Pfalz as
one sample) we found almost the same FsT=0.22, indicating that level 1 analysis
was truly influenced by an among region variance and not the division of samples
into sub-samples. The hierarchical analysis of genetic variance (Wright, 1978),based
on sub-samples, indicated about 60% of the variance estimate was caused by regional
separation, and 40% within regions (Table 3). Differentiation among sub-samples
within Rheinland-Pfalz, FST = 0.17 0.04, was comparable to that of sub-samples
within Sachsen-Anhalt, FST = 0.18 +_ 0.03.

Levels 3 and 4
Decreasing the geographical range from level 3 (2Okm') to level 4 (4km2)also
decreased the differentiation estimate, e.g. sample FST = 0.12 to 0.07 (Table 2). The
inbreeding estimate, &, was not significantly different from zero when estimated
90 J. J 0 H A " E S E N ETAL.

TABLE 2. F-statistics for different sampling schemes and Hardy-Weinberg (HW) proportions for levels
3 to 5. HW estimates are based on all individuals samples within a level. N= number of sub-samples
or samples

N 4.1 4s Fsr HW

Level 1
All sub-samples 27 0.30k0.07 0.05 k0.03 0.26 k0.07
All samples 9 0.32f0.08 0.12 k0.03 0.22 f0.07
Level 2
Rheinland-Pfalz 6 0.18f0.04 0.02 k0.06 0.17k0.04
Sachsen-Anhalt/ 21 0.22k0.03 0.05 k0.02 0.18 k0.03
Thiiringen
Level 3 NO
H d k , sub-samples 19 0.21k0.04 0.0 1 f 0.04 0.20+0.04 i'=149, df32, P<O.OOl
HaNe, samples 6 0.21 k0.04 0.11 k0.03 0.12f0.03
Level 4 NO
Gimritz, sub-samples 14 0.16k0.05 0.03 k0.04 0.13+_0.02 i'=60, df28, RO.001
Gimritz, samples 4 0.17+0.06 0.10+0.04 0.07 f 0.02
Level 5 YES
Wddchen, sub-samples 15 0.02+0.02 -0.14k0.03 0.13k0.03 i2=
13, df22, P=0.92
Waldchen, samples 3 0.02k0.03 - 0.02 f0.02 0.04k0.02

TABLE 3. Hierarchical FsT(Wright, 1978) for all sub-samples and for

level 3 (HaUe) samples (SA-C to SA-H). FSRsignifies the differentiation
among sub-samples or samples relative to a regional level. FRris the
differentiation within the regional level relative to the total level. The
regions of all sub-samples were Rheinland-Halz, Glilcksburger Heide,
Kyffshauser, and Halle. At Halle three regions based on sample locations
were defined, (1) SA-C, (2) SA-D, (3) SA-E - SA-H

FXY vc
All sub-samples N= 27
F5n 0.180 0.48 1
Fur 0.111 0.333
F,I 0.271 0.813
Halle, samples N= 6
4r 0.053 0.151
FRT 0.050 0.299
F, I 0.100 0.148

VC variance component.

among sub-samples, but was when estimated among samples. Hierarchical FST
statistics for level 3 (Halle) showed 50% variance within and between areas (Table
3). In these tests we treated the two outlying sub-samples, SA 3.1 and SA 3.2, as
one sample (SA-C). Hereby we gain a conservative differentiation estimate, because
by increasing the genetic variation within a sample, the among sample differentiation
estimate decreases. This sample was not in Hardy-Weinberg proportions, R0.05.
This is not unexpected as the two sub-samples are separated by a distance of 4 km.
None of the remaining five samples (SA-C to SA-H) at Halle deviated significantly
from expected Hardy-Weinberg proportions. In contrast, the allelic distribution
 GENETICS OF A SUBSOCIAL SPIDER 91

among sub-samples within samples were significantly different from another. De-
viations from expected Hardy-Weinberg proportions were found both for the total
Halle sample and for the Gimritz sample (Table 2).

Lvel 5
The Waldchen co-ordinate sample (159 individuals, 79 age-class 1, 57 age-class
2, 23 age-class 3 +) did not differ significantly form Hardy-Weinberg proportions,
P= 0.92 (Table 2). Hardy-Weinberg proportions were also observed for both age-
class 1, P=O.99, and age-class 2 P=0.97. Of a total of 66 possible painvise
locus genetic linkage disequilibria comparisons thirteen were significant. The allele
distribution was significantly different between age-class 1 and 2.
Differentiation estimates among samples a-c (Fig. 3) were calculated in two ways:
first including all individuals, second by only employing the first 30 individuals
collected from sample 2, with the purpose of keeping the among sample size nearly
equal. FST estimates were equal indicating slight structuring, F5T= 0.04. Neither F,,
nor &T were significantly different from zero. The allelic distributions among samples,
for both calculations, were significantly different from one another.
For the 15 sub-samples a high degree of structure, FsT=0.13, and a significant
excess of heterozygotes, &= -0.14 were observed (Table 2). The allelic distribution
was highly significantly different among sub-samples. However, fiT= 0 indicated
that defining level 5 sub-samples as populations was erroneous and that individuals
were part ofjust one hill population. This finding corroborated the Hardy-Weinberg
proportions.

Relatedness

Level 5
The observed differentiation among sub-samples observed of level 5 could in
large part be explained by the sampling of related individuals within sub-samples.
The average sub-sample relatedness estimate was R =0.287, which, in the absence
of inbreeding (positive assortative mating), corresponds to individuals that on average
were related as half-sibs (Table 4). There were no indications of inbreeding, F= 0,
thus the estimates were not caused by substructure or isolation by distance. Low
sub-sample estimates of R were influenced by grouping different age-classes. The
tentative division of level 5 into four demes was not justified. The four deme
relatedness estimate did not differ from the one deme estimate (Table 4). This
contrasts with the results from the level 3 relatedness investigation (see below).
+
Due to too few age-class 3 individuals we only analysed relatedness within this
age-class and not relative to age-classes 1 and 2 and two-sub-sample comparisons.
Two year olds, R = 0.320, were significantly more related to one another than were
one year olds, R= 0.206, and three year olds, R= 0.180, and the population estimate
was therefore influenced quite strongly by the two year olds. A high variance was
found within and among all age-classes; and especially so for age-class 1 (Table 5).
In half of the sub-samples, RxO; for the other half, Rx0.30-0.60, which is in the
range of half to full-sibs. Relatedness between age-classes 1 and 2, R = 0.1 14, (Table
4) was significantly less than among two year olds, P<O.OOl, and nearly significant,
P= 0.06 than among one year olds. Again, this nearly significant result was influenced
92 J.JOHANNESEN E T A

TABLE 4. Relatedness estimates, Rfstandard error taken over sub-

samples, N, included in the analysis, and inbreeding estimate F. All
estimates are weighted by total number of group individuals. Estimates
were gained for all sub-sample members (all), 1, 2 and 3 year olds,
relatedness between 1 and 2 year olds (1,2 year) , and relatedness of sub-
sample individuals including individuals of the nearest sub-sample (2sub).
The Halle estimate omits sample SA-C. Maximum number of sub-
samples possible were for Halle 17, Gimritz 14, and Wiildchen 15
Category

Level 3 (Halle)
5 demes 17 0.163f0.044 0.095 f0.034
1 deme 17 0.243 f 0.033 0.145f0.022
Level 4 (Gimritz)
4 demes 14 0.176f0.055 0.126f0.037
1 deme 14 0.228*0.041 0.158 f0.027
Level 5 (Wadchen)
4 demes all 14 0.279 f 0.070 0.015 f0.053
4 demes all 2sub 14 0.128 f0.08 1 0.015 f0.053
I deme all 15 0.287 f0.055 0.0 13f 0.039
I deme all 2sub 15 0.145 k0.065 0.0 13f 0.039
1 deme 1 year II 0.206+0.103 0.054 f 0.062
1 deme 1 year Zsub 13 0.096 f0.094 0.054f0.062
1 deme 2 year 8 0.320*0.065 0.005 f0.095
1 deme 2 year 2sub 9 0.297 f0.062 0.005 f0.095
1 deme 3 year 5 0.180f0.111 -0.098fO.110
1 deme 1,2 year 7 0. I14f0.078 0.025 fO.040
I deme 1,2 year 2sub 11 0.009 f 0.069 0.006 f0.040

TABLE 5. Sub-sample relatedness estimates for different age-classes. N= total number of sub-sample
individuals. N (*) and R (*) are the number of age-class individuals (1, 2, or 3) and their relatedness,
respectively. (-) signifies a group with one or no individuals of that age-class and analysis not possible

Sub-sample N NU) R (1) (2) R (2) N (3) R (3)

5.1 22 6 0.13 9 0.41 6 0.03
5.2 8 2 -0.29 - - 5 0.02
5.3 10 2 0.43 7 -0.01 - -
5.4 5 5 0.06 - - - -
5.5 11 10 0.57 - - - -
5.6 13 13 0.43 - - - -
5.7 10 2 -0.44 8 0.30 - -
5.8 5 - - 4 0.08 - -
5.9 15 15 0.30 - - - -
5.10 7 - - 6 0.68 - -
5.11 13 - - 8 0.46 5 0.44
5.12 10 9 0.57 - - - -
5.13 10 10 0.01 - - - -
5.14 10 4 0.03 4 0.05 2 0.06
5.15 10 - - 6 0.34 3 0.38

by the high age-class 1 variance. Including the nearest sub-sample (Table 4: 2sub)
caused the relatedness estimate within age classes to decline significantly for age-
classes 1 and 1 x 2, RO.01, but not for age-class 2 (Table 4).
Isolation by distance tests mirrored relationships within sub-samples. AU tests
 GENETICS OF A SUBSOCIAL SPIDER 93

TABLE
6. Isolation by distance regressions among level 5 (Wddchen)
samples. A negative linear correlation between log (N&) and the log of
the geographical distance (i) indicates isolation by distance. Note: P
values are based on Mantel tests, not regression analyses

Sub-samples included Regression P

All logNrm = 0.478-0.147 log(i) 0.300
Sample b logNrm=0.912-0.692log(i) 0.120
Samples a and b logNcm=0.6004.384log(i) 0.0 12
Samples b and c logN-m =0.673-0.202 log(i) 0.300

showed a negative correlation between gene flow and geographic distance. A

significant correlation was observed only for sub-samples of samples a and b, P=
0.012. Within sample 1 a negative, but non-significant correlation was observed,
P=O.12 Mantel test (Table 6). Across all sub-samples no isolation by distance was
found. This finding was most likely caused by sub-samples of sample c, which were
situated furthest away and showed low within sub-sample relatedness.

hvels 3 and 4
The samples of Halle (omitting sample SA-C) and Gimritz were analysed with
and without respect to deme (correspondingto a sample) structure. Without division
into demes relatedness among sub-sample individuals seemed high, R = 0.243 and
R = 0.228 (Table 4). However, the inbreeding estimate, F, was significantly greater
than zero revealing that the relatedness coefficient was caused by population division
and not by family members within sub-samples. Grouping sub-samples into demes
(samples) caused the relatedness of sub-sample individuals to decrease, R =0.163
and R= 0.176. However, F remained greater than zero, which indicated, as did the
F$T estimates of level 3 and 4, that sub-samples within samples were collected from
separate neighbourhoods.

DISCUSSION

Relatedness and genetic structure

Family clustering coupled to a secondary effect of subdivided habitat proved the

differentiation of E. cinnabmhus samples intermediate between social, FSTz 0.50-0.80
(Lubin & Crozier, 1985; Smith, 1986; Roeloffs & Riechert, 1988; Smith & Engel,
1994; Smith & Hagen, 1996) and asocial spiders FsTx 0-0.15 (Galindo-Ramirez &
Beckwitt, 1983; Riechert, 1993; Ramirez & Fandino, 1996; Ramirez & Froehlig,
1997,Johannesen and Veith, unpublished [most allozyme studies in asocial spiders
deal with phylogenetic relationships and intraspecific differentiation are difficult or
not possible to assess, e.g. Pennington, 1979; Cesaroni et al., 1981; Terranova &
Roach, 1987; Rowell, 1990; Steiner & Greenstone, 1991; Colgan & Gray, 19921).
As in most outbreeding species there was no indication of inbreeding and levels of
genetic variation were high compared to social spiders.
The initial definition of a cluster of webs as a group of related individuals
was only speculative. However, the age similarity indicated, and genetic analyses
94 J.JOHANNESEN E T A

confirmed, that many web clusters consisted of related individuals. Level 5 analyses
indicated family or sib-groups in three ways: (1) a significant sub-sample excess of
heterozygotes &<O, compared to Fm = 0; (2) genetic linkage disequilibrium at 20%
of the pairwise locus comparisons caused by sampling sib-groups; and (3) the
population estimate of R= 0.26 showed a degree of relatedness. Different relatedness
estimates within age-classes, two year olds R=0.30, one year olds R=0.21, and
three year olds R = 0.18, liiely occurred because we defined the putative sub-samples
as family goups, but can also be explained by differential dispersal of sibs between
years. Spiderlings from neighbouring nests may disperse and settle in the vicinity of
each other. This probably led to both a high within age-class variance of R and the
differences among age-classes. The observed isolation by distance among sub-samples
of level 5 indicated that females are sedentary and thus suggests a vertical female
genealogy. Relatedness between age-classes, within samples and among neighbour
groups increased both inter- and intrademic variance causing a microgeographic
genetic structure.
High differentiation among social spider clusters originate due to closed colonies
(overview by Kechert & Roeloffs, 1993), whereas the E. cinnabainus system is open.
Within samples of levels 3 and 4 some of the sub-samples were rather discretely
distributed and &>O showed a Wahlund effect on a small geographic scale. This
finding was likely due to disjunct sampling and contrasted the results of level 5
(Waldchen) where no indication of inbreeding was found, &=O among samples
a-c. Disjunct sampling of levels 3 and 4 sub-samples was confirmed in the level 5
analysis which showed that along a transect the individual heterozygosity relative
to total heterozygosity was zero, F,=O. & seen in the relatedness estimates, and
an isolation by distance among sub-samples, the slight structuring of level 5, FsT>O,
was caused by a sampling artefact of family living (and probably limited female
dispersal). As mentioned above, a Wahlund effect at Halle and Gimritz indicated
inbreeding within samples SA-D to SA-H, but within sub-samples no inbreeding
was observed. Thus, family clustering also biased the FsT estimates of levels 3
and 4 where individuals within samples probably were collected from different
neighbourhoods. Instead of family colonies as in social spiders one could speak
of neighbourhoods of E. cinnaberinus. On the other hand, the Hardy-Weinberg
disequilibrium of the total level 4 population (Gimritz)contrasted the results of the
level 5 (Waldchen) population indicating at level 4 (4km2)a true subdivision of
samples. This subdivision likely originated as a consequence of limited dispersal and
was enhanced by a patchy habitat. The subdivision was corroborated by Wright's
hierarchical analysis (Table 3). For an dispersing outbreeding organism within a
small area as Gimritz (4km2)and Halle (20 km') differentiation was very high, and
indicated little gene flow within a few kilometres.
What then defines a population and/or a deme of E. cinnaberinus? This question
can not be answered in a straightforward manner because of both a within patch
and an among patch variance component. By increasing the sample size (relative
to the sub-sample)and the geographical range the differentiation as expected declined
(e.g. Gimritz sample versus Gimritz sub-sample)because more variation was included
within each sample and because we may have put adjacent more related neigh-
bourhoods in different samples. Thus, the among patch differentiation estimate is
biased from the sedentary life mode of E. cinnaberinus. The differentiation at Gimritz
must lie between the sample and sub-sample FST estimate, 0.08-0.18. As sub-samples
likely often belonged to separate neighbourhoods the estimate probably lies in the
 GENETICS OF A SUBSOCIAL SPIDER 95

upper half. Our tentative answer would be, despite the Wahlund effect, to characterise
Gimritz (level 4) as a single population divided into patch demes, where a deme is
defined as the level of random mating (&=O and no ‘deme’ effect on R within the
patch Waldchen). Each patch deme is neighbourhood structured, which is the reason
for the high level of differentiation. This genetic structure could be extrapolated in
a hierarchical fashion to all higher levels and differentiation expands as concentric
rings from level 5 to level 1. It contrasts the findings for social spiders, where
variation is related to colonies but not to geographical areas (Lubin & Crozier,
1985; Smith, 1986; Roeloffs & Riechert, 1988; Smith & Engel, 1994; Smith &
Hagen, 1996). Only between levels 3 and 2 was no additional increase in variance
observed. This was probably due to collecting only one sample within each of two
areas. One reason could be that the great structuring within level 3 (Halle) will not
cause much additional variance at only 30-50 km distance. The similar differentiation
estimates within Rheinland-Pfalz and Sachsen-Anhalt/Thuringen could support this
explanation. However, because E. cinnaberinus is neighbourhood structured means
that collecting only one sample per area may give a false estimate of an area’s
genetic variation, and thus the variance among areas. In either case, due to the
neighbourhood structure of E. cinnabmhus, the sampling of only two outlying samples
(from different areas) for inclusion in level 2 analysis proved inadequate to evaluate
differentiation at this level. Spiders collected from the Western and Eastern German
regions (level 1) suggested little, if any, gene flow between these.

Implicationsfor evolution ofsociali& in spiders

All social spiders experience female dispersal or founding of new colonies by

propagules of parent colonies (Vollrath, 1982). In contrast, in asocial spiders it is
the males that disperse and seek mates (Foelix, 1981). Males of the likewise subsocial
spider Stegodyphus lineatus also disperse (3’. Lubin, pers. comm.). The mode of dispersal
thus seems essential for social evolution in spiders. Shifting dispersal from males to
females may increase the potential relatedness of offspring and inbreeding might
cause cooperative behaviours to evolve. This implies inbreeding, high intergroup
variance, and maybe evolution of social behaviour by interdemic selection (Roeloffs
& Riechert, 1988; Smith & Hagen, 1996). Estimates of relatedness within colonies
of Agelena consociata have been estimated to 0.556 (Riechert & Roeloffs, 1993) and
an inbreeding coefficient of 0.69 in Skgodyphus dumicola has been reported (Wickler
& Seibt, 1993).
Evidence from E. cinnaberinus showed it to be an outbreeder, with high levels of
genetic variation, and high intergroup variance (on the scale of an outbreeding
organism). Within a deme relatedness estimates showed family clustering and
restricted dispersal. We may be witnessing two structuring processes. First, increasing
relatedness among neighbour females creating family neighbourhoods, and second
outbreeding caused by dispersing males. It is conceivable that by extending the
neotene period clusters of related spiderlings will sustain tolerance, but male dispersal
persists and opposes intrademic relatedness over generations. For kin selection or
inbreeding to propagate sociality the mode of dispersal must change too. In this
sense a male-female conflict may, in terms of establishing an increased genetic
variance among lineages, halt the evolution of social behaviour of E. cinnabm‘nus.
For interdemic selection to mediate permanent social conditions by increasing
96 J. J 0 H A " E S E N ETAL.

relatedness through inbred groups turnover must be high, although a patchy habitat
can compensate by increasing genetic variance (Slatkin, 1981).
Interdemic selection for cooperative behaviour in spiders has been invoked
primarily for social Agelenidae and Theridiidae (Lubin & Crozier, 1985; Smith,
1986; Riechert & Roeloffs, 1993; Smith & Hagen, 1996),whereas sociality in Eresids
may be due to individual selection by pedogenesis (Kraus & Kraus, 1990; Wickler
& Seibt, 1993),and may be driven by a different selection pressure. In an interdemic
selection model for social behaviour, as has been suggested for A g e h a consociata,
aggression between males is scarce, which would be expected if males are highly
related because no selective advantage among identical males exists (Riechert &
Roeloffs, 1993). On the other hand, in social Eresids female choice or male-male
competition is often observed (Henschel et al., 1995). Competition behaviour is
observed in E. cinnaberinus too, and dispersing males have been observed to engage
in fights (T. Baumann, unpublished). However, it is possible that one selection
regime initiates an evolutionary process and another takes over once a certain limit
is reached. For example, once tolerance has been achieved by individual selection
in outbreeding populations, selection of family groups may take over.
Intermediate sociality may be a distinct social condition and not an ancestral
stage towards sociality. We do not know whether E. cinnaberinus is still evolving
towards permanent sociality or has reached its social limits. Communal breeding in
bees and wasps probably should not be viewed as a route to eusocial conditions
(Packer, 1993). Lack of inbreeding or kin selection has been showed for several
communal bees (Kukuk & Sage, 1994; Danforth, Neff & Barretto-KO, 1996).
Although the dispersal modes among social spiders and the intermediate social E.
cinnaberinus differ, females versus males, it is likely female behaviour of both that
cause high levels of differentiation among populations.

ACKNOWLEDGEMENTS

We thank Lorenz Becker for helping collecting spiders, Jes SOe Pedersen for
helping with the Relatedness 4.2b estimates, and Yael Lubin and Jutta Schneider
for discussions on the manuscript. Martin G. Ramirez and an anonymous reviewer
made valuable comments on the manuscript. Permissions for collections were
provided by Regierungsbezirk Koblenz and Regierungsprasidium Halle. The study
was supported by the German Federal Ministry of Science and Technology (grant
no. 3395 19A).

REFERENCES

Alexander RD,Noonan KM, Crespi BJ. 1991. The evolution of eusociality. In: Sherman PW,
JarvisJUM, Alexander RD,eds. l7k biology ofthe naked mole-rat. Princeton: University Press Princeton,
344.
AvilCs L. 1986. Sex-ratio bias and possible group selection in the social spider Anelosimus eximz'us.
American Naturalist 128: 1-12.
AvilCs L. 1993. Interdemic selection and sex ratio: a social spider perspective. A d a n Naturalist 142:
320-345.
AvilCs L. 1997. Causes and consequences of cooperation and permanent-sociality in spiders. In: Choe
 GENETICS OF A SUBSOCIAL SPIDER 97

JG, Crespi BJ, eds. n2e evolution of social behviour in insects and arachnids. Cambridge, Cambridge
University Press, 476-498.
Baumann T. 1996. Relations of invertebrate species to habitat quality in the fragmented porphyry
landscape near Halle (Sachsen-Anhalt, Germany). In: Settele J, Margules C, Podschlod P, Henle
K, eds. Species survival in Fragmented landscapes. Dordrecht: Kluwer Academic Publishers.
Baumann T. in press. Habitat selection and distribution power of the spider Eresus cinnaberinus (Oliv.,
1789) in the porphyry landscape near Halle (Saale). Aoceedings ofthe 16th. European coll. Armhnology.
Brach V. 1977.Anelosimus studiosus (Araneae:Theridiidae) and the evolution of quasisocialityin theridiid
spiders. Evolution 31: 154-161.
Bulmer MG. 1986. Sex ratio theory in geographically structured populations. Huedip 56: 69-73.
Cesaroni D, Allegrucci G, Caccone A, Sbordoni MC, de Matthaeis E, di Rao My Sbordoni
V. 1981.Genetic variability and divergence between populations and species ofNesticus cave spiders.
Genetica 56: 81-92.
Colgan DJ. 1986. Developmental changes in the isozymes controlling glycolysis in the acridine
grasshopper, Caledia captiva. ROWSArchives of Developmental Biology 195: 197-201.
Colgan DJ, Gray MR. 1992. The genetic structure of the Badumna candida group of species (Araneae,
Amaurobiodea). Bulletin ofthe British Arachnological Sociep 9: 86-94.
Danforth BN, Neff JL, Barretto-KO P. 1996. Nestmate relatedness in a communal bee, Perdita
texana (Hymenoptera: Andrenidae), based on DNA fingerprinting. Evolution 5 0 276-284.
D u e JE. 1996. Eusociality in a coral-reef shrimp. Nature 381: 512-514.
Foelix RF. 1981. Biology ofspiders. Cambridge, Mass: Harvard University Press.
Elgar MA, Godfray HCJ. 1987. Sociality and sex ratios in spiders. Eendc in Ecology and Evolution 2:
6-7.
G a l i n d o - R 6 z MyBeckwitt R. 1983.Gene diversityamong isolated populations of the Calinornia
trapdoor spider (Mygalomorphae,Ctenizidae).Aoceediqs ofthexnth International Congress ofArachnology,
Panama. 95-99.
Harris H, Hopkinson DA. 1978.Handbook o f e n i p ~electrophmsk in human g e n e h . Amsterdam: North-
Holland Publishing Company.
Hebert PDN, Beaton MJ. 1993. Methodologiesfor alloryme anahsk using cellulose acetate electmphoresk.
Beaumont, Tx: Helena Laboratories.
Henschel JR, Lubin YD, Schneider J. 1995. Sexual competition in an inbreeding social spider,
Slpgodyphus dumicola (Araneae: Eresidae). Insectes Sociam 42: 419-426.
Holl A, Reinbach M. 1991. Soziobiologie der Rohrenspinne Eresus n&er (Araneida, Eresidae).
Vkhandlungen Deuhche /Zbologische Gesellschaft 84: 3 1 1.
Kaib My Hussender C, Epplen C, EpplenJT, Brand R. 1996. Kin-based foraging in a termite.
Aoceedings ofthe Royal Sociep oflandon, series B 263: 1527-1532.
Kraus 0, Kraus M. 1990.The genus Stegodyphus:systematics,biogeography,and sociality(Araneida,
Eresidae). Acta /zoologia Fennica 190 223-228.
Kukuk PF,Sage GK. 1994.Reproductivity and relatedness in a communal halicine bee, Lusioglossum
(Chilalictus) hernichalceum. Insecles so cia^ 41: 443-455.
Kullmann EJ. 1972. Evolution and social behavior in spiders (Araneae, Eresidae and Theridiidae).
Amm'can <oologist 12: 419426.
Louis EJ, Dempster ER. 1987. An exact test for Hardy-Weinberg and multiple alleles. Biometries
43: 805-8 1 I .
Lubin YD. 1991. Patterns of variation in female-biased colony sex ratios in a social spider. Biological
Journal ofthe Linnean Sociep 43: 297-3 1 1.
Lubin YD, Crozier RH. 1985. Electrophoretic evidence for population differentiation in a social
spider Achaearanea wau (Theridiidae). Insectes Sociaux 32: 297-304.
Lubin YD, Robinson MH. 1982. Dispersal by swarming in a social spider. SGience 216: 319-321.
Michod RE. 1982. The theory of kin selection. Annual Reveiw ofEcokogy and $ s h a & s 13: 23-55.
Michod RE. 1993. Inbreeding and evolution of social behavior. In: Thornhill NW, ed. "h natural
histoly of inbreeding and outbreeding: theoretical and emperiial perspectives. Chicago: University of Chicago
Press, 74-96.
Packer C, Ruttan L. 1988. The evolution of cooperative hunting. Amoican Naturalist 132: 159-198.
Packer L. 1993. Multiple-foundress associations in sweet bees. In: Keller L, ed. Queen number and
socialip in insects. Oxford: Oxford University Press, 2 15-233.
Pamilo P. 1984. Genotypic correlation and regression in social groups: multiple alleles, multiple loci
and subdivided populations. Genetics 107: 307-320.
98 J. JOHANNESEN ETAL.

Pennington BJ. 1979. Enzyme genetics in taxonomy: Diagnostic enzyme loci in the spider genus
Meta. Bulletin ofthe Brihih Arachnological So&& 4: 377-392.
Queller DC, Goodnight KE 1989. Estimating relatedness using genetic markers. Evolution 43:
258-275.
Ramirez MG, Fandino LB. 1996. Genetic variability and gene flow in Metepeira ventura (Araneae,
Araneidae).Journal ofArachnology 24 1-8.
Ramirez MG, Froehlig JL. 1997. Minimal genetic variation in a coastal dune arthropod: the
trapdoor spider Aptostichus simus (Cyrtaucheniidae).Conservation Biology 11: 256-259.
Ratschker UM. 1995. Bemerkenswerte Spinnenfunde in den St. Pauler Bergen in Karnten (Araneae,
Atypidae-Eresidae-Theridiidae).Carinthia ZZ 185.1105: 723-728.
Raymond M, Rousset F. 1995a. An exact test for population differentiation.Evolution 49: 1280-1 283.
Raymond M, Rousset F. 1995b. GENEPOP (V.1.2): population genetics software for exact tests
and ecumenicism. Journal ofHmdi& 86: 248-249.
Reeve HK, Westneat DF, Noon WA, Sherman PW, Aquadro CF. 1990. DNA ‘fingerprinting’
reveals high levels of inbreeding in colonies of the eusocial naked mole-rat. Aoceedings ofthe National
Acadmy ofS&ce USA 87: 249&2500.
Reilly LM. 1987. Measurements of inbreeding and average relatedness in a termite population.
American Naturalist 130: 339-349.
Richardson BJ, Baverstock PR, Adams M. 1986. A l l o g m electrophoresis. A handbookfor animal
systematics and population studies. San Diego: Academic Press, Inc.
Riechert SE. 1993. Investigation of potential gene flow limitation of behavioral adaptation in an
aridlands spider. Behauioural Ecology and Sociobiology 32: 355-363.
Riechert SE, Roeloffs RM. 1993. Evidence for and consequences of inbreeding in the cooperative
spiders. In: Thornhill NW, ed. i’h natural histoy of inbreeding and outbreeding: theoretical and mperical
perspectives. Chicago: Chicago University Press, 283-303.
Roeloffs R, Riechert SE. 1988. Dispersal and population genetic structure of the cooperative spider,
A g e h a consociata, in west African rainforest. Evolution 42: 173-183.
Rowell DM. 1990. Ftted fusion heterozygosity in D e h a cancerides Walck. (Araneae: Sparassidae): an
alternative to speciation by monobrachial fusion. Genetica 8 0 139-157.
Schneider J. 1995. Survival and growth in groups of a subsocial spider (Stegodyphtrs lineatus). Insects
S O ~ 42: W 237-248.
Seibt U, Wickler W. 1988. Why do ‘family spiders’, Stegodyphus (Eresidae),live in colonies?Journal of
Arachnology 16: 193-198.
Slatkin M. 1981. Population heritability. Evolution 35: 859-57 1.
Slatkin M. 1993. Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47:
264-279.
Smith DR. 1986. Population genetics of Anelosimus e x i m k (Araneae, Theridiidae).Journal ofArachnology
1 4 201-217.
Smith DRYEngel MS. 1994. Population structure in an indian cooperative spider, Stegodyphw
surmnomm Karsch (Eresidae). Journal OJArachnologr 22: 108-1 13.
Smith DR, Hagen RH. 1996. Population structure and interdemic selection in the cooperative spider
Anelosimus eximius. Journal ofEvolutionay Biology 9 589-608.
Steiner WWM, Greenstone MH. 1991. Segregation studies of isozyme variation in Metuphidippus
galathea (Araneae, Salticidae).Journal ofArachnolo~1 9 157-160.
Terranova AC, Roach SH. 1987. Genetic differentiation in the genus Phid$pus (Araneae, Salticidae).
Journal ofArachnology 14: 385-391.
Uyenoyama MK. 1984. Inbreeding and the evolution of altruism under kin selection: effects on
relatedness and group structure. Evolution 38: 778-795.
Vollrath F. 1982. Colony formation in a social spider. <eitschriftfiir Twpsychologie 60: 3 13-324.
Wade MJ, Breden F. 1981. Effect of inbreeding on the evolution of altruistic behavior by kin
selection. Evolution 35: 844-858.
Weir BS. 1991. Genetic data analysis. Sunderland, m.Sinauer Publ.
Weir BS, Cockerham CC. 1984. Estimating F-statistics for the analysis of population structure.
Evolution 38: 1358-1370.
Wickler W, Seibt U. 1993. Pedogenetic sociogenesis via the ‘sibling-route’ and some consequences
for Stegodyphus spiders. Ethology 9 5 1-18.
Wright S. 1978. Evolution and the genetics offiopulations, vol4. Variabili& within and among natural Populations.
Chicago: University of Chicago Press.