Molecular Ecology (1992)1, 55-63

MINI-REVIE W

Applications of random amplified polymorphic DNA

(RAPD) in molecular ecology
H. HADRYS, M. BALICK* and B. SCHIERWATERt
Yale University, Department of Biology, 165 Prospect Street, New Hnven, CT 06511 and ‘New York Botanical Garden, Bronx,
Ny 10458, USA

Abstract

Molecular genetic markers have been developed into powerful tools to analyse genetic
relationships and genetic diversity. As an extension to the variety of existing techniques
using polymorphic DNA markers, the Random Amplified Polymorphic DNA (RAPD)
technique may be used in molecular ecology to determine taxonomic identity, assess
kinship relationships, analyse mixed genome samples, and create specific probes. Main
advantages of the RAPD technology include (i) suitability for work on anonymous
genomes, (ii) applicability to problems where only limited quantities of DNA are
available, (iii) efficiency and low expense.
Keywords: DNA fingerprinting, DNA probes, kinship analysis, paternity determination, RAPD,
taxonomic identifications
Received 3 February 1992

Introduction probes (Burke et nl. 1991b) and PCR amplified simple

sequence microsatellite loci (Tautz 1989). Potential
Within the past few years molecular genetic approaches
applications are frequently thwarted by the requirement
have become of increasing importance to studies in be-
for significant quantities of DNA in the case of RFLP
havioural ecology and population biology. For instance,
analysis or by lack of relevant DNA sequence information
DNA fingerprinting technologies have revolutionized
in the case of conventional PCR-based techniques. Recent
approaches to our understanding of animal social
criticisms are that DNA fingerprinting requires special
systems by permitting analyses of kinship reIationships
molecular training, is labour-intensive, and is relatively
(e.g. Burke & Bruford 1987; Burke et 01. 1991a; Jones,
expensive
- (Weatherhead & Montgomerie 1991).
Lessels & Krebs 1991; Gyllensten, Jakobsson & Temrin
It has been argued that DNA fingerprinting is so
1991; Packer et nl. 1991; Pemberton, Banaoft & Amos
essential to behavioural ecology and population biology
1991; Schlotterer, Amos & Tautz 1991; Smith ef al. 1991).
that it would be highly unfortunate to have its application
Despite constant progress in methodology, application
limited to a few specialized laboratories rather than to the
of DNA markers to many problems in behavioural and
broad community working in these fields (Weatherhead
population ecology has been limited by technical con-
& Montgomerie 1991). Major progress in technology
siderations (Lewin 1989; Kirby 1990; Burke et al. 1991a;
development is expected in two directions: (I) increase in
Pemberton et al. 1991). The most frequently used DNA
analytical power per unit effort, and (2) simplification in
markers include RFLPs (restriction fragment length
technology, and ultimately reduction in expense. Use of
polymorphisms) visualized by Southern blot hybridi-
random amplified polymorphic DNA (RAPD) markers,
zation to different types of single-locus or multilocus
detected by PCR amplification of small inverted repeats
Correspondence author: Heike Hadrys, Zoologisches Institut der
scattered throughout the genome, adds a new tech-
Technischen Univenitat, Pockelsstr. 10a. D-3300 Braunschweig, nology of DNA fingerprinting to the molecular analysis of
Germany. relatedness between genotypes. The introduction of
*Present address: Zoologisches Institut der Universitiit, Siesmayerstr. RAPD fingerprinting is a substantial contribution toward
70, 06000 Frankfurt a.M., Germany. the second direction.
56 H . HADRL’S. M. BALICKand B. SCHIERWATER
RAPD technology is only some 2 years old, and con-
sequently published applications are limited. However,
the rapidly growing interest in using RAPD technology
justifies an early review on the current literature and the
potentials of the method. In this paper we shall review
the principle and original applications of the RAPD
technology, discuss its applications in molecular ecology,
and point out the particular advantages as well as limita-
tions of RAPD markers.
Principle of RAPD analyses
The PCR-based RAPD technique (Williams rt nl. 1990) is
an attractive complement to conventional DNA finger-
printing in ecology. RAPD analysis is conceptually
simple. Nanogram amounts of total genomic DNA are
subjected to PCR using short synthetic oligo-
nucleotides of random sequence. The amplification pro-
tocol differs from the standard PCR conditions (Erlich
1989) in that only a single random oligonucleotide primer
is employed and no prior knowledge of the genome
subjected to analysis is required. When the primer is
short (e.g. 10-mer), there is a high probability that the
genome contains several priming sites close to one
another that are in an inverted orientation. The technique
essentially scans a genome for these small inverted
repeats and amplifies intervening DNA segments of
variable length. The profile of amplification products
depends on the templateprimer combination and is
reproducible for any given combination (see below). The
amplification products are resolved on agarose gels and
polymorphisms serve as dominant genetic markers,
which are inherited in a Mendelian fashion (Williams ct
al. 1990; Carlson et al. 1991; Martin, Williams & Tanksley
1991; Welsh, Peterson & McClelland 1991). Amplification
of non-nuclear RAPD markers is negligible because of the
relatively small non-nuclear genome sizes.
Two modifications of detecting RAPD markers have
been described as DNA Amplification Fingerprinting
(DAF)and Arbitrarily Primed Polymerase Chain Reaction
(AP-PCR). DAF uses short random primers of 5-8 bp and
visualizes the relatively greater number of amplification
products by polyacrylamide gel electrophoresis and
silver staining (Caetano-Anolles, Bassam & Gresshof
1991). AP-PCR uses slightly longer primers (such as
universal M13) and amplification products are radio-
actively IabelIed and also resolved by polyacrylamide gel
electrophoresis (Welsh h McClelland 1990;Welsh et al.
1991b).
Standard RAPD analysis is performed according to the
original methods (Williams et al. 1990) using short
oligonucleotide primers of random sequence which are
commercially available (Operon Technologies, Inc.,
Alameda, Calif.). Only high-molecular-weight, i.e. non-
degraded, DNA should be subjected to RAPD analyses.
Amplification products can be resolved by gel electro-
phoresis on 1.4% agarose gels.
Applications
Here we illustrate several potential applications of RAPD
fingerprinting in molecular ecology, including deter-
mination of taxonomic identities, detection of inter-
specific gene flow, assessment of kinship relationships,
analysis of mixed genome samples, and production of
specific probes.
Determinntiorz of tnxomnic identity
By employing different oligonucleotide primers,
molecular characters can be generated that are diagnostic
at different taxonomic levels. For any given primer,
RAPD amplification products can be broadly classified
into two groups: variable (polymorphic) or constant (non-
polymorphic). These definitions are relative for a given
operational taxonomic unit (OW). For instance, consider
a RAPD analysis of several individuals within a species,
and several species within a given genus (Fig. 1).
Constant fragments diagnostic for a genus may be iden-
tified, as well as fragments which are polymorphic
between species within the genus. Both types of product
can be exploited for establishing systematic relationships.
In this example, constant fragments operationally
identify members of a certain genus exclusively if the
fragment is a unique polymorphism in a comparison of
genera (genus-specific character in Fig. 1). Note, the
determination of taxonomic relatedness is only valid
between taxa for which the diagnostic RAPD fingerprint
patterns have been established. Similarly, fragments
polymorphic at the species level will operationally iden-
tify members of a given species if the fragment is constant
among all members of that species (species-specific
character in Fig. 1). An example of such a marker is
provided in RAPD fingerprints from two dragonfly
species in Fig. 2. Fragments polymorphic among indi-
viduals may also be utilized to determine clonal identity,
as is frequently required for asexually reproducing
organisms. Clone-specific markers have been identified
in hydroids (Schienvater, unpubl.), clonal “individual”-
specific markers in fungal mycelia (Smith, Bruhn &
Andcrson 1992), cultivar-specific markers in broccoli and
cauliflower (Hu & Quiros 1991), strain-specific markers in
mice (Welsh et al. 1991a), species-specific markers in irises
(Arnold, Buckner & Robinson 1991) and tomato (Klein-
Lankhorst et al. 1991) and genus- and family-specific
markers in palms (M. Balick & S. Dellaporta, in prep-
aration). Thus RAPD products can be generated that
 RAPD F I N G E R P R I N T I N G 57

Fig. 1 Use of polymorphic and n m -

polymorphicRAPD fragments for genome
scanning. Schematic of RAPD fingerprints
of genera, species, and individuals illus-
trating the potential use of polymorphic
A: genera B: specks c: individuals and non-polymorphic RAPD fragments as
diagnostic markers. In gel A, a genus 3
specific band is identified by its presence
1 2 3r4 5 1 2 3 4 5 in all species of that genus (and con-
oaonn sequently all individuals of all species of
genus 3 tested); this genus-specific frag-
ment can serve as a diagnostic DNA
marker in RAPD analyses or a genus-
specific probe in KFLP analyses. Similarly,
a species-specific fragment (gel B) is iden-
tified by its presence in all individuals of
species 3, that can serve as a species
specific RAPD marker or probe (see Fig.
2). Finally, gel C shows five polymorphic
RAPD fragments between different indi-
viduals of species 3; these may be used for
intraspecific analyses, such as paternity
determinations (see Fig. 3), by means of
RAP0 fingerprinting.

serve as diagnostic molecular characters at different

taxonomic levels.
As illustrated in Fig. 1, the specificity of any single
RAPD marker may range from the level of the individual
to higher taxonomic levels. However, taxon identification
by diagnostic RAPD markers can only be done by com-
A
.. B parison within a set of genotypes of known R4PD ampli-
fication profile for a given primer. The specificity of
RAPD markers has to be determined empirically for each
genome within a set of genomes under investigation by
screening several primer-templa te combinations
(analogous to the screening of restriction enzyme-probe
combinations in Southem-blot-based fingerprinting).
Homology of diagnostic markers, especially between
members of higher Oms, may be established by
Southem analysis to exclude the possibility of co-
migration of fragments of the same size from non-
homologous loci (see below).

Analyses of interspecific gene flow and hybrid speciation

A straightforward conclusion of the outlinedpotential of
h i 2 3 4 5 6 7 8 ~ the RAF'D method to identify diagnostic markers for
different OTUs is that RAF'D can be applied to analyse
Fig. 2 Example of determination of taxonomic identities. The fusion of genotypes at different taxonomic levels. At the
primer (5'CCACAGCAm amplifies a constant RAPD level of the individual, RAPD markers may be applied to
marker (arrow) shared by the dagonfly prthm'JPe parentage analysis (see below); at the population or
(lanes 1-4, four unrelated individuals) and Orthetrum cmr~escms
(lanes 5-8, four unrelated individuals). In addition, note
species level RAPD may be used to detect hybrid pop-
the occurrence of a constant marker (arrow) diagnostic for 0. ulations Or species. The detection Of genome hybrids
c ~ u l f f c e n s . Each marker appeared in all individuals tested relies upon the identification of diagnostic RAPD markers
with this primer (n = 38). for the parental genotypes under investigation.
58 H . HADRYS, M. BALICKand B. SCHIERWATER

4-

1 2 3 4 5 6 7 8 9 1011121314
Fig. 3 Determination of paternity, RAPD fingerprint of a guarded Amx partfrenopcfemale, the guarding mate, the offspring of this pair,
and several unrelated males revealed by primer B 07 (S’GCTGACGCAC). Lanes 1.2 = unrelated control males; 3 = guarding male; 4 =
mother; 7-14 = individual offspring and entire clutches. Note that a characteristic band (arrow) from the guarding male (lane 3) is
detected in all offspring samples. Further note that this same band is detected in synthetic offspring using the guarding male (lane 5),but
is absent from the synthetic offspring of unrelated males (lane 6). To achieve statistically significant analyses we pooled three RAPD
fingerprints with three different primers and calculated band-sharing coefficients for known first-degree relatives (mother and
offspring), putative first-degree relatives (guarding male and offspring) and controls (known non-firstdegree relatives) for seven
families and several offspring clutches per family. In all cases, the controls shared statistically significant fewer bands with any of the
offspring clutches than did the putative parents.

Arnold et al. (1991) have demonstrated interspecific

gene flow between two Louisiana iris species, Iris fulva
and I. hexagonu, by analyses of species-diagnostic RAPD
markers. Using two 10-bp and one 16-bp primers they
reported four species-specific markers from different
populations of I. fulua. These markers were missing in I.
hexugona, but were present at intermediate frequencies in
experimental F1 hybrids (I. fulm x I . hexngonn) and at
variable frequencies in a natural contemporary hybrid
population. Conclusions from RAPD analyses on the
occurrence of interspecific gene flow between two iris
populations were consistent with results from RFLP
analyses of the chloroplast genome. Furthermore,
variable frequencies of species-diagnostic markers were
also found in the putative hybrid species 1. nelsonii. Here,
RAPD markers may be useful in investigating the role of
hybridization in the origin of I. nelsonii.
Using the AP-PCR modification of RAPD, Welsh et al.
(1991a) identified F1 hybrids from different inbred maize
lines. Other groups have begun to use RAPD for analyses
of hybridization events where allozyme studies have not
proven to be sensitive enough for hybrid genotypes, for
example, in hybridization along vertical zonations in
natural populations of Dnphnia (B. Streit, personal
communication) or in plant breeding programmes
(Crowhurst et al. 1991; Hu & Quiros 1991; Martin et al.
1991; Quiros et ul. 1991).
Defermimfionof paternity and kinship relationships
By employing fragments that are polymorphic among
individuals, RAPD analysis may be used to assess pater-
nity and kinship relationships in large offspring samples.
A common problem in behavioural ecology is to deter-
mine the actual father from a number of potential fathers.
The earliest application of RAPD fingerprinting to pater-
nity analyses resolved the question of paternity in an
unknown mating system of the dragonfly A m parthenope
(Hadrys 1991; H. Hadrys et nl., in preparation).
A. pnrflrenope males guard ovipositing females over
extended periods of time and thereby regularly take the
chance of suffering severe injury or even death by attacks
hom conspecific males trying to split the tandem pairs.
The presence of spermatodesmids, i.e. sperm bundles
encased in a slowly degrading matrix, has suggested that
the male may guard a female in order to assure a sub-
sequent mating success rather than immediate fertili-
zation success. R4PD analyses of tandem males, tandem
females and the offspring clutches identified the tandem
male as the father of the immediately oviposited offspring
clutches. Figure 3 shows RAPD fingerprints of the
guarding male, the guarded female, the offspring, and
several unrelated males. The RAPD markers poly-
morphic among potential fathers show the guarding male
to be the actual father. For statistical analyses the number

of polymorphic markers can be increased by increasing
the number of diagnostic primers, and conventional
band-sharing coefficients can be applied (Lynch 1990,
1991; Burke et al. 1991a;Keane et al. 1991; H. Hadrys et al.,
in preparation). Band-sharing coefficients between
unrelated individuals are highly dependent on the
primer-template combination used (see Fig. 2). Although
we found background band-sharing coefficients of RAPD
markers generally higher than those known from multi-
locus probe finge printing, this does not represent a
serious problem. Because linkage between different
arbitrary priming sequences is extremely unlikely, the
number of independent polymorphic markers analysed
can be rapidly increased by pooling markers revealed by
several primers. The amplification of monomorphic
RAPD markers may be kept to a minimum by choosing
the ~ g h t primer-template combination, and any
monomorphic markers may be removed from the
analysis to decrease background band sharing.
In principle RAPD markers can formally be treated as
Mendelian alleles, and for parentage analysis analytical
approaches may be developed which are based on
knowledge of allelic frequencies, e.g. as used in statistical
analyses of single-locus fingerprint profiles. In practice,
however, allelic frequencies of scorable dominant RAPD
markers in diploid organisms might be difficult to
estimate and markers revealed by the same primer may
be linked (cf. Williams ef RZ. 1990; Carlson ef el. 1991).
The segregation and linkage of RAPD markers has
been demonstrated in several genetic studies. The ex-
pected nuclear transmission of RAPD markers to F1
offspring has been reported for hybrids of maize inbred
lines (Welsh et al. 1991a) as also has the segregation of
RAPD polymorphisms in experimental crosses of
conifers (Carlson et al. 1991) and the linkage of RAPD
markers to resistance genes in lettuce (Michelmore, Paran
& Kesseli 1991; Paran, Kesseli & Michelmore). RAPD
markers have also been used in demonstrating the intro-
gression of two parental genomes in an iris hybrid species
(Arnold et al. 1991).
Conventional RFLP fingerprinting techniques are ill-
suited for the analysis of paternity and estimation of
reproductive success in species with large offspring
clutches, because of the need to determine paternity for
each individual offspring. RAPD fingerprinting provides
a ready alternative for such cases. Synthetic offspring
may be produced by mixing equal amounts of the DNA of
the mother and the potential father (Fig. 3). The ampli-
fication products from the synthetic offspring should
ideally contain the full complement of bands (H. Hadrys
et al., in preparation; cf. Carlson efal. 1991; see also below)
that appear in any single offspring of these parents.
However, certain combinations of alleles may lead to
amplification artefacts (e.g. heteroduplex formation, see
RAPD FINGERPRINTING 59
below), and certain markers may only get amplified from
an offspring but not from either of its parents. The
Occurrence of non-parental bands in offspring from
known primate pedigrees has raised concerns in paren-
tage determinations when single individuals are
analysed (Riedy, Hamilton & Aquadro 1992). The
Occurrence of non-parental bands in offspring from
known primate pedigrees has raised concerns in
parentage determinations when single individuals are
analysed (Riedy, Hamilton & Aquadro 1992). In contrast,
the synthetic offspring is a complete representation of
both parental genomes and will match the profile of a
large sampIe of offspring, since in both the ‘synthetic’ and
the real offspring dutch, allele combinations that may
cause amplification artefacts are represented in equal
amounts. The degree of mismatch between synthetic
offspring and actual offspring clutches is indicative of
mixed paternity, which can be measured by quantitative
determination of mixed genome samples.
A na lysing mixed genome samples
The RAPD technique may be used to generate quanti-
tative estimates of the relative proportions of different
genomes in mixed DNA samples. In many polygamous
mating systems, especially in insects, sperm competition
and mixed paternity may occur. Here, confirmation
of the presence of more than one paternal genotype
would be highly desirable. Using DNA from a dragonfly,
Orthetrurn coerulescms, we reconstituted mixed-genome
samples experimentally by varying the relative propor-
tion of DNA from two male individuals over two orders of
magnitude. This DNA was amplified and the relative
amounts of individual male-diagnostic amplification
products quantified by densitometry. The relative inten-
sity of diagnostic bands from two different individuals
varied predictably with the relative DNA concentrations
of each genome in the reaction, and the DNA concen-
tration of a given genome could be estimated from band
intensities as long as the relative amount of this genome
was at least 20% (Fig. 4; see also Michehore ef al. 1991).
Note that this approach is based on well amplified poly-
morphic bands, requires precise knowledge of the
diagnostic markers for each genome being scanned, and
requires prior calibration experiments.
Generating novel specific probes
Any diagnostic RAF’D marker can be eluted from the gel,
reamplified, radiolabelled with 32P and serve as an
inexhaustible supply of probe in Southern analyses (Fig.
5). Such probes may be used to exclude the possibility of
60 H . H A D R Y S , M . B A L I C K a n d B . S C H I E R W A T E R

(C)

1
1

Fig. 4 Analysis of mixed genome samples. Quantitative estimates of the presence of a genome in mixed DNA samples by densitometry.
(A) RAPD fingerprints of two males (a, b) of Orthetrrrrn cwrrilesceirs with primer B l l (5'GTACACCCGT) showing a diagnostic marker for
each (see arrow). (B) Mixed reactions with varving proportions of DNA from the two males. The total amount of DNA per reaction was 25
ng, with the proportions a:b (in '2) as follows; lane 1.50:jO; lane 2,40:60; lane3,30:70; lane 4,20:80. The intensity of bands is reproducible
between reactions (cf. Caetano-Anolles el nl. 1991; hlichelmore et nl. 1991) and co-varies with total amount of DNA template in a non-
linear fashion. (C) Densitometric analyses (DESAGAdensitometcr) of lanes 1-4 of the fingerprint shown in (B). x-axis is the distance in
mm; y-axis is the relative extinction at 540 nm. Peaks of the diagnostic band from male b are numbered refemng to lanes in (B).
Calculations of the area of single peaks of the curve (corresponding to single bands on the gel) are relative estimates of the amount of
DNA amplified. Using the BMDP statistical software (BMDPAR)the best fit regressions between percentage of known male DNA (Y)in
the reaction to percentage of maximal band intensity of the characteristic polymorphic band (Imax, i.e. when using 100% known
template DNA) follow sigmoidal models of the form: Y = qre"'-* / (9 - r + re"'"'). Parameters 9, r , s must be determined by calibration
for each such analysis. In this example, parameters and asymptotic standard deviations were 9 = 108 f 10.1. r=30 f 8.4, s=0.035 f
0.012. For each useful marker, I,, has to be experimentally determined to establish the correct amount of total mixed sample DNA in
the reaction because band intensity may decrease with DNA concentrations exceeding Ima,.. This method produced reproducible
estimates of known genome presence in mixed samples over the range of 2040%.For more-sensitive estimates Southern analyses are
preferable.

co-migration of fragments of different sequence b u t Difficulties and limitations of RAPD

similar sue. O t h e r applications include generation of
probes for taxonomic analysis o r t h e quantitative esti-
mation of t h e presence of a certain g e n o m e in a mixed
sample by Southern analysis (cf. Michelmore rt nl. 1991;
Jeffreys et al. 1991). The specificity of t h e probes can be
further improved by eluting diagnostic RAPD b a n d s from
the gel, reamplifying, subcloning a n d sequencing t h e
fragments, and eventually selecting a partial consensus
sequence a s a probe. Williams et al. (1990)report that six
of 11 RAPD markers tested as probes in RFLP analyses
were useful hybridization probes, because all of them
hybridized to single-copy DNA. The other five, however,
were not useful, because they hybridized to middle- or
highly repetitive DNA i n the soybean genome. RAPD
probes have also been u s e d t o detect RFLPs in tomato
species (Martin et al. 1991; Klein-Lankhorst et al. 1991).
fingerprinting
The following technical considerations a n d potential
difficulties merit attention:
77ie size of tlir primer
Primer size will determine t h e degree of specificity in
genome scanning. It m a y be expected that primers of
short length will amplify an unreasonably large n u m b e r
of sequences and that larger primers will amplify too few
sequences t o be routinely informative. Beyond a certain
primer size (c. 15-mer) increasing primer length m a y also
increase non-specific primer annealing, consequently
increasing t h e probability of random non-reproducible
amplification patterns. All studies using standard RAPD

Fig. 5 Use of amplification products as probes. (a) RAPD
fingerprint gel (primer 814 5’TCCGCTCTGG) of a family of
Orthetrum cwrulmens (lanes 1-9). Lane 5 is the synthetic oif-
spring of the parents of the family, and lane 7are larvae from the
last oviposited egg clutch that lack a diagnostic marker (arrow).
(b) A sample (5 pl) of the diagnostic marker was eluted from the
gel and reamplified in a 100-p.1reaction volume containing 50 WCi
of 32dATPunder standard RAPD PCR conditions (Williamset RI.
1990). The radioactive amplification product was used as a
hybridization probe to a Southern blot of the original RAPD
fingerprint.The autoradiogram shows the diagnostic marker to
be homologous in different lanes and also identifies a smaller
RAPD fragment (arrow)containing homologous sequence to the
probe. Note that the probe hybridizes substantially more
strongly to the synthetic offspring (lane 5). which contains the
pooled RAPD markers from both parents. This result was re-
producible with synthetic offspring from different parents; its
basis is not yet understood and it suggests that care should be
taken in using probes with synthetic offspring samples. Also
note, the probe used here did not hybridize to any RAPD
markers (revealed with the same primer) from the unrelated
dragonfly Anax parthenope.
RAPD FINGERPRINTING 61
conditions (fragment separation on agarose gels) have
found 10-bp primers to be an efficacious size. A G+C
content of the primer similar to the G+C content of the
analysed genome will maximize the frequency of binding
sites and hence amplification products.
Sensitivity to renction conditions
Being PCR-based, the principal limitations of RAPD
fingerprinting arise from its sensitivity to reaction con-
ditions, and slight changes in the conditions may affect
the reproducibility of amplification products (Williams et
al. 1990; Arnold et al. 1991; Carlson et 01. 1991; Klein-
Lankhorst et al. 1991). The technique is sensitive to (a)
shape of the temperature profile , (b) type of polymerase
used and (c) M$+ concentration. The amplification
profile is sensitive to Taq or DNA concentration. The
shape of the temperature profile is a property of the
thermal cycler and must be standardized. Only strictly
standardized reaction conditions will guarantee repro-
ducible amplification products. Furthermore, we found
that the optimal concentration of template DNA per
reaction may vary substantially from typical conditions
(25 ng per reaction) depending on the primer-template
combination used (cf. Carlson et al. 1991).
The possibility of co-migration
An assumption of the use of the RAPD technique is that
amplified fragme.nts are unique, i.e. that the procedure
does not amplify two distinct fragments which co-
migrate on gels because of similar size. Co-migration in
the RAPD technique is easily detected by eluting indi-
vidual PCR products from gels and reprobing the
products via Southern analysis (Fig. 5). Alternatively,
polyacrylamide gel electrophoresis may be used to
increase the resolution of band separation.
Non-reproducible amplificntion products
As with other genetic markers, some RAPD fragments
may be ambiguous and not easy to score (Williams et al.
1990). These unclear and non-reproducible fragments,
which may derive from non-specificic primi,ng or from
heteroduplex formation between related amplification
products (or other secondary structure artefacts, which
can prevent normal amplification patterns) are not useful
as genetic markers. However our own work, as well as
the work of several others (e.g. Williams et nl. 1990;
Arnold et al. 1991; Hu & Quiros 1991; Klein-Lankhorst et
al. 1991) have all shown that if the RAPD amplification is
repeated two or more times, the majority of markers is
clearly reproducible and scorable. As in many cases of
using PCR, sometimes amplification products are found
62 H . HADRYS, M . BALICKand B . SCHIERWATER
even in the absence of template DNA in the reaction
(Innis et nl. 1990; Klein-Lankhorst et al. 1991). However, in
ail reported cases those 'ghost' bands disappear if the tem-
plate DNA under investigation is added to the reaction.
Other considerations
The most common DNA fingerprint technologies differ
substantially in (i) complexity of technological pro-
cedures, (ii) required amount of DNA, (iii) sequence
information needed for a genome being scanned, (iv)
analytical power of assigning genotype relatedness, (v)
expense in tenhs of labour and money, (vi) broadness of
applications. In this context, RAPD fingerprinting seems
IiJcely to have wide potential for applications in molecular
ecology, and requires the least in technology, labour and
expenses. The cost of producing one individual DNA
fingerprint by Southern hybridization can be very sub-
stantial (Weatherhead & Montgomerie 1991);in contrast,
the average expense for one RAPD fingerprint can be as
low a s US$2.00. O n the other hand, RAPD markers are
the least informative of all known DNA markers and they
are dominant (heterozygosity is normally not detectable).
Consequently, the analytical power of RAPD markers is
not competitive with analyses using sequence infor-
mation or single locus probe fingerprint technologies.
However, RAPDs are detected more easily than RFLPs
and can be competitive with RFLPs even in analyses of
genomes with high levels of heterozygosity (Williams et
al. 1990; Carlson et nl. 1991; Hu & Quiros 1991).
Potential Future Applications
We note here several additional applications currently
under development.
1. Sex determinution. In many ecological (as well as agri-
cultural and legal) applications it would be convenient to
have available markers that were sex-specific. We expect
that little difficulty will be encountered in developing
RAPD markers with this characteristic.
2. Gentrution of specific PCR primers for anonymous genoms.
A major limitation in the application of PCR to ecological
problems is the absence of sequence information for the
vast majority of organisms. We suggest that this difficulty
may be overcome for many applications by using a
RAPD-based strategy for developing 'designed PCR
primers. Specifically, RAPD primers with an embedded
restriction site may be used to detect fragments showing
the desired properties (e.g. detecting a particular taxon).
These fragments may then be cloned and sequences used
to develop specific 'designed' PCR primers for diagnostic
markers.
3. Quantitative analysis of mixed biosamples. Analogous to
the analysis of mixed paternity samples in dragonflies,
analysis of field samples of different species or other
Oms (e.g. plankton sampling) may be performed.
4. Phylogeny. RAPD markers may prove to be useful
characters for cladistic analysis.
Acknowledgments
Many substantial contributions to the manuscript derive
from the ideas of Dr Stephen L. Dellaporta and Dr Leo W.
Buss. We thank Matt Dick, Robert Domtt and Neil
Blackstone for critical comments. Our work is supported
by D U D , DFG, NATO, NIH, NSF and ONR.
References
Arnold ML, Buckner CM, Robinson JJ (1991) Pollen-
mediated introgression and hybrid speciation in
Louisiana irises. Proceediiigs of the National Academy of
Sciences of the USA, 88, 1398-1402.
Burke T, Bruford MW (1987) DNA fingerprinting in
birds. Nuture, 327, 149-152.
Burke T, Hanotte 0, Bruford MW, Cairns E (1991a)
Multilocus and single locus minisatellite analysis in
population biological studies. In: Burke T, Dolf G,
Jeffreys AJ, Wolff R (eds) D N A Fingerprinting:
Apprmches niid Applicntions, pp. 154-168. Birkhauser,
Basel.
Burke T, Dolf G,Jeffreys AJ, Wolff R (eds)(1991b) D N A
Fingerprintiiig: Apyrmches mid Applications. Birkhauser,
Basel.
Caetano-Anolles G, Bassam GJ, Gresshof PM (1991)
High resolution DNA amplification fingerprinting
using very short arbitrary oligonucleotide primers.
Biotechnology, 9, 553-556.
Carlson JE, Tulsieram LK, Glaubitz JC, Luk VWK,
Kauffeldt C, Rutledge R (1991)Segregation of random
amplified DNA markers in F1 progeny of conifers.
Theoretical and Applied Genetics, 83, 194-200.
Crowhurst RN, Hawthorne BT, Rikkerink EHA,
Templeton M D (1991)Differentiation of Fiisarium solnni
f . sy. cucurbitae race-1 and race-2 by random ampli-
fication of polymorphic DNA. Current Genetics, 20,
391-396.
Erlich HA (1989) PCR Technology. Stockton Press, New
York.
Gyllensten UB, Jakobsson S, Temrin H (1991) No
evidence for illegitimate young in monogamous and
polygynous warblers. Nature, 343, 168-170.
Hadrys H (1991) Postkopulatorische Mannchen-
konkurrenz und RAPD Fingerprinting in Odonaten.
Verhndlungen der Deufschen ZuoIogischen Grsellschuft,
84,306-307.

Hillis D, Moritz C (1990)Molecrilar Systematics. Sinauer,
Sunderland, Mass.
Hu J, Quiros CF (1991) Identification of broccoli and
cauliflower cultivars with RAPD markers. Plmt Cell
Reports, 10, 505-511.
Innis MA, Gelfand DH, Sninsky JJ, White TJ (1990) PCR
Protocols. Academic Press, San Diego.
Jeffreys AJ, MacLeod A, Tamaki K,Neil DL, Monckton
DG (1991) Minisatellite repeat coding a s a digital
approach to DNA typing. Nature, 354,204-209.
Jones CS, Lessels CM, Krebs JR (1991) Helpers-at-the-
nest in European Bee-eaters (Meriops apiaster). In: Burke
T, Dolf G, Jeffreys AJ, Wolff R (eds) D N A Fingerprirrting:
Approaches and Applications, pp. 169-192. Birkhauser,
Basel.
Keane B, Waser PM, Danzl-Tauer L, Minchella DJ (1991)
DNA fingerprinting: estimating background band
sharing in banner-tailed kangaroo rats. Animal
Behaviour, 42, 141-143.
Kirby LT (1990) D N A Fingerprinting: An Introduction.
Stockton Press, New York.
Klein-Lankhorst RM,Vermunt A, Weide R, Liharska T,
Zabel P (1991) Isolation of molecular markers for
tomato (L.esculentum) using random amplified poly-
morphic DNA (RAPD). Theoretical and Applied Genetics,
83, 108-114.
Lewin R (1989) Limits to DNA Fingerprinting. Science,
243, 1549-1551.
Lynch M (1990) The similarity index and DNA finger-
printing. Molecular Biology nnd Evolution, 7, 478-484.
Lynch M (1991) Analysis of population genetic structure
by DNA fingerprinting. In: Burke T, Dolf G, Jeffreys
AJ, Wolff R (eds) D N A Fingerprinting: Apprmchrs n ~ t d
Applications, pp. 113-126. Birkhauser, Basel.
Martin GB, Williams JGK, Tanksley SD (1991) Rapid
identification of markers linked to a Psezdonzonns
resistence gene in tomato by using random primers
and near-isogenic lines. Proceedings of the Nntiorzal
Academy of Sciences of the USA, 88, 2336-2340.
Michelmore RW, Paran I, Kesseli RV (1991) Identi-
fica tion of markers linked to disease-resistance genes
by bulked segregant analysis. A rapid method to detect
markers in specific genomic regions bv using segre-
gating populations. Proceedings of the National Academy
of Scienres of the USA, 88,9228-9832.
Paran I, Kesseli R, Michelmore R (1991) Identification of
restriction fragment lengthy polymorphism and
random amplified polymorphic DNA markers linked
to downy mildew resistance genes in lettuce, using
near-isogenic lines. Genome,34, 1021-1027.
Packer C, Gilbert DA, Pusey AE, OBrien SJ (1991)A
molecular genetic analysis of kinship and cooperation
in African lions. Nature, 351, 562-565.
Pemberton J, Bancroft D, Amos B (1991) Behavioural
RAPD FINGERPRINTING 63
Ecology and DNA fingerprinting: the lab rats’ riposte.
Trends in Ecology and Evolution, 6 , 299-300.
Quiros CF, Hu J, This P, Chevre AM, Delseny M (1991)
Development and chromosomal localization of
genome specific markers by polymerase chain reaction
in Brassica. Theoretical and Applied Genetics, 82, 627-632.
Riedy MF, Hamilton WJ,Aquadro CF (1992) Excess of
non-parental bands in offspring from known pedigrees
assayed using RAPD PCR. Nucleic Acids Research, 20,
918.
Schlotterer C, Amos B, Tautz D (1991) Conservation of
polymorphic simple sequence loci in cetacean species.
Nature, 354, 63-65.
Smith HG, Montgomerie R,Poldmaa T, White BN, Boag
PT (1991)DNA fingerprinting reveals relation between
tail ornaments and cuckoldry in barn swallows,
Hirundo rustica. Behavioural Ecology, 2, 90-98.
Smith ML, Bruhn JN, Anderson JB (1992) The fungus
Armillaria bulbosa is among the largest and oldest living
organisms. Nature, 356, 428431.
Tautz D (1989) Hypervariability of simple sequences as a
general source for polymorphic DNA markers. Nucleic
Acids Research, 17, 6463-6471.
Weatherhead PJ, Montgomerie RD (1991) Good news
and bad news about DNA fingerprinting. Trerrds iiz
Ecology and Evolution, 6, 173-174.
Welsh J, McClelland M (1990) Fingerprinting genomes
using PCR with arbitrary primers. Nricleic Acids
Research, 18, 7213-7218.
Welsh J, Petersen C, McClelland M (1991a) Poly-
morphisms generated by arbitrarily primed PCR in the
mouse: application to strain identification and genetic
mapping. Nucleic Acids Research, 20, 303-306.
Welsh J, Honeycutt RJ, McClelland, Sobral BWS (1991b)
Parentage determination in maize hybrids using the
arbitrarily primed polymerase chain reaction
(AP-PCR). Theoretical and Applied Genetics, 82, 47-76.
Williams JGK, Kubelik AR, Livak KJ, Rafalski JA,
Tingey SV (1990) DNA polymorphisms amplified by
arbitrary primers are useful as genetic markers. Nucleic
A d s Rpsearch, 18, 6531-6535.
This paper is a result of collaborative work between the groups of
Leo Buss and Stephen Dellaporta (Yale, Biology) to develop and
apply new molecular genetic tools to developmental genetics
and population ecology. Heike Hadrys is studying the evolution
of mating systems in odonates by combining field work with
molecular genetic amlyses at the Zoological Institute
Braunschweig and at Yale University. Michael Ballick cotlabor-
ates with Stephen Dellaporta on conservation biology and
systematics of palms. Bernd Schienvater has performed 3 years
of postdoctoral research with Leo Buss on molecular develop
mental genetics and evolutionary ecology of hydroids.