Genetic Aspects of the Worldwide

Colonization Process of Ceratitis capitata

A. R. Malacrida, F. Marinoni, C. Torti, L. M. Gomulski, F. Sebastiani, C.
Bonvicini, G. Gasperi, and C. R. Guglielmino

Multilocus enzyme electrophoresis data from 26 polymorphic loci (124 alleles) were
used to analyze the genetic aspects of the worldwide colonization of Ceratitis cap-
itata (medfly). Eighty-two samples of 17 populations were collected from six
regions throughout the species range: Africa, extra-Mediterranean islands (Madeira
and Gran Canaria), Mediterranean region, Latin America (Guatemala), Pacific (Ha-
waii), and Australia. The variability parameters (H, P, A) reveal that the geographical
dispersal of medfly from its ancestral source area (East Africa) is associated with
a great reduction in variability. The pattern of decreasing variability occurs at two
regional levels: in the African–Mediterranean region where the differentiation is
gradual, and in the Latin American–Pacific region where some ancestral variability
is still present as a consequence of recent colonization. The UPGMA phylogenetic
tree, derived from Nei’s genetic distances, shows the presence of intraspecific dif-
ferentiative processes affecting mainly the two island populations, Réunion and
Hawaii. The population genetic changes observed in the species range are consis-
tent with both the chronology and the historical circuitous course of the medfly
colonization process.

The Mediterranean fruit fly (medfly) (Cer- al. 1991; Malacrida et al. 1992) and DNA
atitis capitata) is a fast-colonizing species variation ( Baruffi et al. 1995; Gasparich et
which in the last 100 years has spread al. 1995; Gomulski et al. 1996; Haymer and
from its supposed origin in tropical Africa McInnis 1994; McPheron et al. 1994). From
to a number of countries including the these previous studies it has been possi-
Mediterranean basin, parts of South and ble to suggest an African origin for this
From the Dipartimento di Biologia Animale, Università
di Pavia, Piazza Botta 9, I-27100, Pavia, Italy (Malacrida,
Central America, and Australia ( Fletcher species in the sub-Saharan East region
Marinoni, Torti, Gomulski, Sebastiani, Bonvicini, and 1989b). In these regions it exists in a wide ( Kenya) and to evidence the presence of
Gasperi), Dipartimento di Genetica e Microbiologia, variety of climates. History of infestation intraspecific differentiative processes of
Università di Pavia, Pavia, Italy (Guglielmino), and Is-
tituto di Genetica Biochimica ed Evoluzionistica del and geographical spread of this species ancestral versus peripheral populations.
CNR, Pavia, Italy (Guglielmino). We are very grateful to are well documented (for a comprehen- Very high levels of genetic variability have
Jeffrey R. Powell for critical reading of the manuscript sive review see Robinson and Hooper been detected in the African native popu-
and for his valuable suggestions. We appreciate the
help of two anonymous reviewers for their highly valu- 1989). The oldest derived populations are lations: this high level of genetic variabil-
able comments. We are indebted to the following peo- found in the Mediterranean area. More re- ity may reflect the genetic plasticity of this
ple for having provided the wild-collected specimens
of medfly: T. Mukiama ( University of Nairobi, Kenya);
cent colonization events characterize the polyphagous species which, in a very
E. Osir ( ICPE, Nairobi, Kenya); S. Quilici ( IFRA, Saint- New World populations. The question of short time, has reached a nearly cosmo-
Pierre, La Réunion); A. Mazih ( Institute de Agrono- why this species has become a major pest politan geographical distribution (Malacri-
mique et Vétérinaire-Hassan II, Agadir, Morocco); P.
Ross (CIT-INIA, Madrid, Spain); A. Barbosa (Madeira); has been approached by studies on zoo- da et al. 1996). With respect to the colo-
A. Economopoulos ( IMBB, Heraklion, Greece); F. De geography (Maddison and Bartlett 1989), nization pattern, we suggested that medfly
Lima ( Department of Agriculture, Perth, Western Aus- on population biology (Papadopoulos et populations, like those of Drosophila mel-
tralia); B. Katsoyannos ( University of Thessaloniki,
Thessaloniki, Greece); D. O. McInnis ( USDA-ARS, Ho- al. 1996), and on the analysis of the life- anogaster ( David and Capy 1988), can be
nolulu, Hawaii); P. Rendon ( USDA-APHIS, San Miguel, history strategies that this species has divided into three main categories: ances-
Petapa, Guatemala); Y. Roessler (Citrus Marketing
Board, Israel); L. Suess ( University of Milan, Milan, It-
evolved ( Fletcher 1989a). tral, ancient, and new populations from
aly); G. Zervas ( NCSR, Democritos, Athens, Greece). A description of medfly population sub-Saharan Africa, the Mediterranean ba-
This research was supported by grants from the Eu- structure is important both for a better sin, and the New World, respectively (Ma-
ropean Communities Commission ( ECC) (Project FAIR
PL96-1972), the National Ministry of the University and understanding of the history and future lacrida et al. 1992). Because historical pro-
Scientific Research and Technology (M.U.R.S.T. ‘‘Funds evolutionary potential of this pest species cesses seem to be important determinants
40%’’), and the International Atomic Energy Agency ( Vi- and its populations as well as for risk as- of medfly geographic variation, we have
enna, Austria). Address correspondence to Dr. Mala-
crida at the address above or e-mail: mala- sessment. The genetic aspects of the col- undertaken more extensive medfly sam-
[email protected]. onization process have been studied pling from throughout the species range
q 1998 The American Genetic Association 89:501–507 through the analysis of protein (Gasperi et to study the genetic consequences of the

501
worldwide colonization process. In this ar- two samples from Fodele collected in neutral mutation and that the genetic vari-
ticle we examine the results obtained with 1995. The Chios Island population consist- ability is at equilibrium between mutation
a multilocus enzyme electrophoresis ap- ed of three samples collected from August and genetic drift. Nei’s genetic distance is
proach on the distribution of genetic vari- 1992 to September 1993. The Greek flies expected, for a sample of many loci, to
ability in ancestral and derived popula- derive from two samples: one from Salon- rise linearly with time. The trees were con-
tions from the Mediterranean basin, Aus- icco and the other from Attica, collected structed using the unweighted pair-group
tralia, Latin America, and Pacific areas. in March and July 1994, respectively. arithmetic average ( UPGMA; Sneath and
The results are used to infer (1) a histori- Latin American region. This region is Sokal 1973) and neighbor-joining methods
cal interpretation of the population genet- represented by the population from Gua- with the PHYLIP 3.5c package ( Felsenstein
ic changes found in different geographic temala. The Guatemalan flies derive from 1993). The bootstrap method ( Efron 1982)
populations and (2) to assess the exis- three samples collected near Antigua, was applied to test the robustness of tree
tence of population substructuring in the from coffee berries, from December 1989 topology.
species range. to March 1994. F statistics were calculated in the med-
Pacific region. The Hawaiian flies derive fly sampling area, grouping the 17 above
from two samples collected in Mauna Loa mentioned populations into nine geo-
Materials and Methods
and Kauai in April 1992. graphical groups. Groups and populations
Populations of C. capitata Australian region. The Australian flies constitute the basis of a hierarchical de-
Eighty-two samples of 17 populations of C. derive from nine samples collected near sign for studying the degree of the genetic
capitata were collected from five regions Perth, from different hosts, from 1992 to isolation within and between groups. The
throughout the species range: Africa, ex- 1994. standardized genetic variance (FST; Wright
tra-Mediterranean islands, Mediterranean 1965) within and between each of the geo-
basin, New World, and Australia. Electrophoretic Studies graphical groups was estimated. For each
African region. In this region the follow- Preparation of samples and electrophoret- FST value obtained for each locus, we test-
ing three populations were sampled: Ken- ic procedures are described in Gasperi et ed the significance of deviation from zero
ya, Réunion, and Morocco. The Kenyan al. (1991). For each population sample at with the heterogeneity chi-square test
population was represented by 12 samples least 25 individuals were assayed at the (Workman and Niswander 1970), x2 5
of pupae, collected on coffee berries, over following 26 enzyme loci: Mpi, Mdh-1, Mdh- 2NFST(k 2 1) with (k 2 1)(s) degrees of
10 years (August 1984–August 1994) from 2, Hk-1, Hk-2, Est-1, Est-2, Pgi, Zw, Pgd, Fh, freedom, where N is the total sample size,
two farms near Nairobi. The population on Had, Idh, Pgm, Got-1, Got-2, Ak-1, Ak-2, Adh- k is the number of alleles for the locus,
Réunion was sampled 14 times (October 2, Gpt, Pgk, Me, a-Gpdh, Aox, Acon- 1, Acon- and s is the number of subpopulations.
1987–December 1994): 10 times near St. 2. Staining for enzyme activities, after elec- The total sample size (N) for each couple
Denis and 4 times in other places on the trophoresis, was based on the recipes of of groups is calculated on the basis of the
island. Pupae were collected from several Meera Khan (1971) and Harris and Hopkin- average sample size per population.
different host fruits. The Moroccan popu- son (1976). For each population we cal- To estimate the relationships between
lation was sampled nine times ( February culated the genotypic absolute frequen- FST values and geographical distances, log
1992–May 1993) from the Argania spinosa cies and the weighted mean allele frequen- geographic distances were calculated be-
forest on the western coast of that coun- cy of each locus across the relative sam- tween all pairs of groups and were re-
try (Mazih and Debouzie 1996). ples. gressed against FST values. To test the sig-
Extra-Mediterranean islands.One sample nificance of the relationships, we used the
was collected in Gran Canaria (October Data Analysis Mantel test (Mantel 1967) with 10,000 ran-
1994) from Prunus persica and one in Ma- Every population was tested for confor- dom permutations to correct for lack of
deira (October 1995) from Pirus communis. mance to the Hardy–Weinberg equilibrium independency between data points (GE-
Mediterranean region. The following nine at each polymorphic locus using the exact NEPOP, version 3, described by Raymond
populations were sampled: Spain, Sardin- probability test provided by the BIOSYS-1 and Rousset 1995).
ia, Procida Island, north Italy, Libya, Israel, program (Swofford and Selander 1981). As Gene flow estimates (Nm) were derived
Crete, Chios Island, and Greece. The Span- recommended by Rice (1989), the sequen- from the private allele method of Slatkin
ish population was sampled twice, near tial Bonferroni test was used on all the 156 (1985) and calculated between the nine
Valencia (September 1994) and near Tar- tests to determine if any cases showed sig- geographical groups. The mean frequency
ragona (October 1994). The population nificant departure from Hardy–Weinberg of private alleles [p(1)], indicates the av-
from Sardinia was sampled six times (Oc- expectations, with a 5 0.05. The single erage level of gene flow among subpopu-
tober 1985–October 1990) in the western populations were tested for the degree of lations according to the formula
and eastern parts of central Sardinia. The genetic variability. The average number of e {2[ln[p(1)]12.44]/0.505}
Procida Island population was sampled alleles per locus (A), percentage of poly- Nm 5 ,
N/25
seven times (June– August 1985). Three morphic loci (P), and mean heterozygosity
north Italy samples were collected around (H) were estimated for each population. where N is the average number of individ-
Milan: in October 1990, September 1992, We also estimated the expected number of uals sampled per population.
and November 1995. The Libyan popula- alleles per locus: ne 5 1/(1 2 H).
tion was sampled once in August 1992. The relationships between the popula- Results
The Israel population was sampled twice tions were represented through a dendro- Estimates of Genetic Variability in the
near Tel Aviv in May 1991 and July 1993. gram obtained from Nei’s (1978) unbiased 17 Geographic Populations
The Crete population consisted of one genetic distances. In Nei’s method it is as- We scored a total of 124 alleles for the 26
sample from Sisses, collected in 1993, and sumed that all loci have the same rate of polymorphic loci which can be considered

502 The Journal of Heredity 1998:89(6)

 Table 1. Parameters of genetic variability in 17 populations of C. capitata from five geographic regions to be a random, although small, sample of
N a
A H b
HEc
the genome. These loci are widely distrib-
O
Population (6SE) (6SE) ne P (6SE) (6SE) uted over the genetic map of C. capitata
(Malacrida et al. 1988). Most of the poly-
African area
morphic loci in each population were in
Kenya 309.2 3.7 1.16 50.0 0.137 0.161
(15.2) (0.4) (0.036) (0.042) good agreement with Hardy–Weinberg ex-
Réunion 255.8 2.4 1.10 26.9 0.091 0.103 pectations. Only 5 (Est-1 and Got-2 in Kenya,
(7.5) (0.3) (0.031) (0.037) Hk-2 and Mpi in Réunion, Zw in Israel) out
Morocco 224.2 2.9 1.10 23.1 0.090 0.097
(6.2) (0.2) (0.031) (0.034) of 156 exact probability tests were found
Extra-Mediterranean islands to show significant statistical deviations
Canary 28.9 1.4 1.09 23.1 0.079 0.075 after Bonferroni correction.
(0.6) (0.1) (0.029) (0.028) The amount of allozyme variability with-
Madeira 28.0 1.4 1.09 26.9 0.083 0.080
(0.8) (0.2) (0.030) (0.028)
in each of the 17 populations, collected in
the six geographic regions, are estimated
Mediterranean area
Spain 55.6 1.5 1.07 23.1 0.068 0.074
by the parameters A, ne, P, and H ( Table
(0.9) (0.1) (0.026) (0.030) 1). The African ancestral populations
Sardinia 103.6 1.5 1.06 23.1 0.057 0.059 ( Kenya, Réunion, Morocco) are more poly-
(5.2) (0.2) (0.021) (0.022)
Procida 99.8 1.5 1.06 19.2 0.057 0.059
morphic with respect to the derived pop-
(6.1) (0.1) (0.023) (0.023) ulations, that is, those from the Mediter-
North Italy 44.2 1.3 1.04 19.2 0.035 0.037 ranean basin, Guatemala, Hawaii, and Aus-
(1.6) (0.1) (0.017) (0.018)
Libya 27.5 1.2 1.03 15.4 0.032 0.044 tralia. Among African populations, Kenya
(1.5) (0.1) (0.016) (0.024) has the highest average number of alleles
Israel 65.9 1.3 1.04 11.5 0.035 0.040 per locus and it is polymorphic for 50% of
(2.0) (0.1) (0.019) (0.022)
Crete 73.7 1.2 1.03 7.7 0.027 0.032 its loci. Furthermore, the observed num-
(1.9) (0.1) (0.022) (0.026) ber of alleles (A) is three times higher than
Chios 94.5 1.3 1.01 7.7 0.015 0.017 the expected number (ne) on the basis of
(3.9) (0.1) (0.008) (0.009)
Greece 58.6 1.4 1.04 11.5 0.040 0.044 the observed heterozygosity; this high A
(0.6) (0.1) (0.020) (0.023) estimate is due to the presence of several
Pacific low frequency endemic alleles. These re-
Hawaii 51.6 1.3 1.5 26.9 0.052 0.068 sults, which indicate Kenya as the most
(3.6) (0.1) (0.021) (0.026)
polymorphic population, confirm the gen-
Latin American eral trend of decreasing genetic variation
Guatemala 83.7 1.3 1.06 11.5 0.054 0.060
(3.2) (0.2) (0.033) (0.035)
from this putative African source area to-
ward the periphery of the species range
Australia 208.8 1.3 1.03 11.5 0.032 0.034
(4.8) (0.1) (0.018) (0.018) ( Baruffi et al. 1995; Gasperi et al. 1991). We
tried to interpret these variability data, es-
a
Average number of individuals per locus. timated in the species’ range, on the basis
b
Observed heterozygosity based on direct count of heterozygotes.
of the history of medfly worldwide colo-
c
Unbiased heterozygosity ( Nei 1978).
nization, the stages of which are summa-
rized in Table 2. The relationship between
the date of the earliest record of each con-
Table 2. Chronological records of the worldwide colonization of C. capitata
sidered geographical population (as pre-
Diffusion area sented in Table 2) and its degree of vari-
Geographic region Country Date of the earliest record ability, in terms of heterozygosity, are
shown in Figure 1. The correlation coeffi-
Africa South East Africa a
putative source area Fletcher 1989b cient between the time of colonization and
South Africa 1889 Back and Pemberton 1918
Extra-Mediterranean islands Canarya early 1800 Fimiani 1989 the heterozygosity is statistically signifi-
Madeiraa 1829 0 cant (r 5 20.69, P , .05). This result sug-
Mediterranean area Spaina 1842 Fimiani 1989
Algeria 1850 0
gests that the steps in the decrease of het-
Tunisia 1855 0 erozygosity are related to the timing of the
South Italya 1863 0 colonization, especially in the Mediterra-
France 1885 0
Portugal 1898 0 nean basin. The colonization of this area
Israela end 1800 0 is an old event ( Fimiani 1989), and we can
Turkey 1904 0 suppose that the populations are derived
Greecea 1915 0
ex-Yugoslavia 1947 0 from each other through subsequent bot-
Latin American Brazil 1905 Enkerlin et al. 1989 tlenecks. On the other hand, the medfly
Costa Rica 1955 0 invasion into Latin America and the Pacif-
Nicaragua 1960 0
Panama 1963 0 ic area has been caused by more recent
Guatemalaa 1975 0 and unrelated events ( Harris 1989).
Mexico 1977 0
Pacific Hawaiia 1910 Harris 1989 Genetic Distances and Cluster Analysis
Australia Australiaa 1897 Hooper and Drew 1989
The Nei unbiased genetic distance (D) es-
a
Countries for which population variability data were obtained in this study. timates reveal the presence of differentia-

Malacrida et al • Genetic Variation and Medfly Colonization History 503

 bor-joining tree obtained with the same versity with respect to each of the two
dataset shows the same differentiative Mediterranean groups. The within-group
processes among populations. estimates suggest that the eastern Medi-
terranean group is the most heteroge-
Hierarchical F Statistics Analysis neous.
To determine the effect of spatial scale on To test for isolation by distance in the
gene frequencies we conducted a hierar- species range, we regressed the FST esti-
chical analysis of population differentia- mates with log of the geographic distance
tion in the species range. For this purpose between the pairwise combinations of all
we partitioned the 17 populations into the nine geographic groups of popula-
nine geographical groups ( Table 3) repre- tions. This regression analysis showed
senting (1) the ancestral Kenya popula- that there was no significant relation be-
tions, (2) the African-related population tween the FST estimates and the geograph-
Figure 1. Correlation between the date of the earliest from Réunion, (3–6) four groups of old ad- ical separation ( FST 5 2288 1 0.108 log
record of each considered geographic population (as ventive populations from the Mediterra- distance, r2 5 0.051, Mantel P 5 0.149).
presented in Table 2) and the average heterozygosity However, when we removed from this
(H) at 26 biochemical loci (r 5 20.69, P , .05).
nean area, (7) the Australian group, and
(8 and 9) the two recently derived popu- analysis the two recently derived popula-
lations from Latin America and the Pacif- tions from Latin America and the Pacific
tion processes among the 17 geographical ic—Guatemala and Hawaii. FST estimates, (Guatemala and Hawaii) and the Austra-
medfly populations. The largest D values calculated as the weighted average over lian group, there was a clear pattern of iso-
occur with all the comparisons involving all loci within and between the nine geo- lation by distance ( Figure 3) in the pair-
the populations of Réunion (D 5 0.110– graphical groups, are shown in Table 3. Al- wise comparisons involving the African
0.145) and Hawaii (D 5 0.072–0.145). most all FST values differ significantly from populations and the four groups of the old
Therefore Réunion and Hawaii are the zero according to the heterogeneity chi- adventive populations from the Mediter-
most differentiated populations. Low D square test, suggesting a high degree of ranean area (FST 5 21.198 1 0.364 log dis-
values are observed between Kenya and heterogeneity in the species range. tance, r2 50.38, Mantel P 5 0.033). We can
all the remaining populations, confirming The important outcome of Table 3 is conclude that in the African–Mediterra-
Kenya’s ancestral status. All the compari- that very low FST values are observed be- nean area, the genetic differentiation be-
sons involving the other populations show tween the source area (south East Africa) tween populations increases with geo-
very low D values, indicating a general and all the other derived groups, exclud- graphic distance.
high genetic similarity among the other ing Réunion. The genetic heterogeneity of
derived populations. Among the lowest D the two island populations, Hawaii and Ré- Gene Flow
values are those between the Australian union, with respect to the other geograph- From the above analysis it seems clear
population and the populations from the ical groups is evident; however, Hawaii that the history and the time of coloniza-
Mediterranean area (D 5 0.010–0.028). doesn’t appear differentiated from the an- tion must have influenced the contempo-
A UPGMA tree obtained with Nei’s ge- cestral population of Kenya (FST 5 0.066). rary genetic structure of medfly popula-
netic distances representing the 17 popu- Australia shows low levels of genetic di- tions. Taking into account the important
lations from the six regions is shown in distinction between contemporary and
Figure 2. The corresponding consensus historical gene flow (Slatkin 1987), we
tree obtained from 100 bootstrap resam- used allozyme data to derive indirect es-
plings of the original dataset shows a sim- timates of migration using Slatkin’s private
ilar pattern. In Figure 2 we report, under allele approach. Table 4 reports the parti-
the corresponding node, only those boot- tion of the number of private alleles and
strap values greater than 50%. Réunion their average frequencies detected in each
and Hawaii are the most divergent popu- pairwise comparison between the nine
lations; subsequently, Kenya and Guate- geographic groups of populations. In each
mala separate from the populations of the comparison, low values of the total aver-
Mediterranean area together with Austra- age frequency of private alleles corre-
lia and the extra-Mediterranean islands. It spond to high Nm estimate ( Table 5). High
is clear from Figure 2 that the differentia- Nm values are observed in all the compar-
tion among the Mediterranean popula- isons involving the ancestral Kenya pop-
tions is minimal and it is characterized by ulation: a maximal Nm estimate of 56.9 has
multifurcations; this means that the Med- been ascertained in the pairwise compar-
iterranean clusters are composed of pop- ison with the Iberian–African group, which
ulations whose genetic distances are not in turn shows high levels of gene flow with
statistically significant. The affinity of the the western and eastern Mediterranean ar-
Australian and the Mediterranean popula- eas (Nm 5 132.0 and 139.0, respectively)
tions is evident: Australia is included with- and Australia (Nm 5 42.8).
in the Mediterranean cluster in 51% of the Figure 2. Tree derived from Nei’s unbiased genetic Genetic isolation, in terms of gene flow,
bootstrapped replications. A high consen- distances representing the 17 populations of C. capita- has been ascertained for the two island
ta. Under the corresponding node are reported only
sus value (75%) supports the association the bootstrap values greater than 50%, obtained from populations—Réunion and Hawaii—which
between Morocco and Spain. The neigh- 100 bootstrap resamplings of the original dataset. maintain gene flow levels greater than 1

504 The Journal of Heredity 1998:89(6)

 Table 3. FST estimates and x 2 analysis within a and between b the nine geographical groups of C. capitata populations

Groups 1 2 3 4 5 6 7 8 9

1 South East Africa 0.261 0.014 0.023 0.070 0.063 0.021 0.066 0.067
( Kenya) — 5917.8 500.1 1861.1 7619.8 1986.6 812.5 1411.5 3305.9
(168)* (146)* (178)* (150)* (152)* (148)* (146)* (150)*
2 Réunion 0.182 0.341 0.429 0.431 0.248 0.182 0.442
— 3033.2 7418.1 7215.7 9845.0 2341.9 3220.3 6809.2
(84)* (138)* (88)* (90)* (78)* (86)* (82)*
3 Extra-Mediterranean islands 0.026 0.006 0.102 0.242 0.007
(Canary, Madeira) 36.6 135.4 20.008 20.014 349.0 889.9 189.7
(24) (104) — — (34)* (34)* (32)*
4 Iberian–African group 0.023 0.017 0.028 0.133 0.032
(Morocco, Spain) 20.004 564.32 486.0 394.0 1847.2 586.3
— (112)* (110)* (108)* (108)* (106)*
5 West Mediterranean group 0.009 0.056 0.209 0.006
(Sardinia, Procida, North Italy) 71.3 20.010 518.1 2067.5 108.4
(57) — (40)* (46)* (42)*
6 East Mediterranean group 0.119 0.055 0.303
( Libya, Israel, Crete, Chios, Greece) 184.4 334.4 1525.5 20.023
(90)* (40)* (44)* —
7 Guatemala 0.302 0.095
821.5 479.1
— (26)* (26)*
8 Hawaii 0.316
1886.3
— (28)*
9 Australia —

a
Only if the group is composed of at least two populations. For groups 3 and 4, composed of only two populations, single samples of each population are at the low level of
the hierarchy.
b
When each of the compared groups is composed of only one population, single samples are considered at the low level of the hierarchy.
Chi-square estimates have not been computed for negative FST values, which result from negative variance components in the hierarchical analysis.
The degrees of freedom are in parentheses. The asterisk indicates tablewide Bonferroni statistical significance at a 5 0.05 (Rice 1989).

with the ancestral Kenyan population. Ré- terranean islands (H 5 0.081), 70% within of isolation by distance. The second re-
union has a high number of private alleles, the Mediterranean area (H 5 0.041), and gional level involves the recent expansion
some of which are fixed ( Baruffi et al. 77% in Australia (H 5 0.032). A loss of vari- into Latin America and the Pacific, where
1995), while Hawaii contains few private ability of 61% is observed within the pop- some ancestral variability is still present
alleles, but at very high frequencies. ulations from Hawaii and Guatemala (H 5 in the derived populations, since probably
0.053). there has not been sufficient time for the
We observe that the loss of variability two populations to reach a proper equilib-
Discussion
in the derived populations occurs at two rium variability.
Genetic Consequences of the regional levels. The first is at the African– The phylogenetic tree portrays well all
Colonization Process Mediterranean level: the decrease is grad- the above described aspects of the medfly
The analysis of variability parameters, ual from Kenya to Greece and it is tightly colonization process. In addition it reflects
such as H, P, ne, and A, clearly reveals the correlated with the timing of the medfly the presence of a great amount of differ-
dynamic aspects of the population genetic expansion. In this range the genetic differ- entiation affecting the two island popula-
changes associated with the worldwide entiation is also positively related with the tions, Réunion and Hawaii. The old popu-
spread of the medfly. The East African geographic distance, with a clear pattern lation from Réunion and the new one from
area, represented here by the Kenyan pop- Hawaii may represent two examples of in-
ulation, is confirmed as the ancestral cipient geographic differentiation due to
source area of the species ( Baruffi et al. island isolation. All the other derived pop-
1995; Gasperi et al. 1991). This native pop- ulations still maintain genetic affinity with
ulation is characterized by a high number the ancestral Kenyan population; among
of alleles per locus, most of which are at them, the Guatemalan population is the
low frequency and found only in this pop- closest. The Mediterranean populations
ulation: we consider these alleles as ‘‘an- form a compact cluster with a very short
cestral’’ ones. The geographical dispersal range of genetic distances. However, incip-
of medfly from the source area is associ- ient genetic substructuring within this
ated with a gradual and great reduction of area is supported by the hierarchical anal-
variability, which is indicative of a coloni- ysis data showing the eastern Mediterra-
zation process probably through subse- nean group as the most heterogeneous.
Figure 3. Plot of FST against log geographic distance
quent bottlenecks. The loss in variability for all pairs of nine population groups. White circles Agreement of the Genetic Data With
is quantifiable: from the ancestral level of indicate comparisons including New World and Austra- the Proposed Worldwide Colonization
H 5 0.137 in Kenya, 34% is lost within the lian populations; black circles indicate comparisons Pattern
not including New World and Australian populations.
African area (H 5 0.090 in Morocco and The regression line is shown only for the latter com- From its supposed origin in southern East
0.091 in Réunion), 41% in the extra-Medi- parisons (FST 5 21.198 1 0.364 log distance; r2 5 0.38). Africa ( Hagen et al. 1981), the medfly

Malacrida et al • Genetic Variation and Medfly Colonization History 505

 Table 4. Partition of the number of private alleles (npa) and their average frequencies, p(1), in 1989). In the Americas it was first reported
parentheses, in each pair group comparison
in central America in Costa Rica in 1955
Pair of groups npa (p(1)) (Gallo et al. 1970), spreading to other
a b Group a Group b Total countries, being detected in 1976 in Gua-
temala, and moving to Mexico along the
SE Africa Réunion 47 (0.049) 10 (0.204) 57 (0.076) coffee belt ( Harris 1989).
SE Africa Extra-Medit. islands 60 (0.021) — 60 (0.021)
SE Africa Iberian-African group 37 (0.012) 16 (0.008) 53 (0.011) The relatively recent historical associa-
SE Africa West Medit. group 57 (0.019) 2 (0.018) 59 (0.019) tion of the medfly populations explains
SE Africa East Medit. group 58 (0.019) 3 (0.008) 61 (0.019)
SE Africa Guatemala 68 (0.031) 2 (0.100) 70 (0.033)
the observed genetic similarities between
SE Africa Hawaii 66 (0.027) 2 (0.199) 68 (0.032) the African ancestral and the derived pop-
SE Africa Australia 65 (0.023) 1 (0.026) 66 (0.023) ulations. It is not surprising that the Med-
Réunion Extra-Medit. islands 28 (0.133) 5 (0.402) 33 (0.173) iterranean populations exhibit a modest
Réunion Iberian-African group 17 (0.162) 33 (0.063) 50 (0.097)
Réunion West Medit. group 25 (0.137) 7 (0.311) 32 (0.175) genetic differentiation with respect to the
Réunion East Medit. group 28 (0.103) 10 (0.201) 38 (0.129) others, in view of the presumed longer his-
Réunion Guatemala 32 (0.119) 3 (0.731) 35 (0.171)
Réunion Hawaii 33 (0.109) 6 (0.400) 39 (0.154)
tory of occupation of the Mediterranean
Réunion Australia 31 (0.105) 4 (0.500) 35 (0.150) area contrasted with the more recent col-
Extra-Mediterranean islands — — 5a (0.025)a onization of the Pacific and Latin Ameri-
Extra-Medit. isl. Iberian-African group — 39 (0.009) 39 (0.009) can regions. In the plot of Figure 2, genetic
Extra-Medit. isl. West Medit. group 5 (0.039) 10 (0.026) 15 (0.030)
Extra-Medit. isl. East Medit. group 5 (0.039) 10 (0.004) 15 (0.015) distances between African, European, Lat-
Extra-Medit. isl. Guatemala 10 (0.160) 4 (0.154) 14 (0.158) in American, and Pacific populations are
Extra-Medit. isl. Hawaii 8 (0.224) 4 (0.157) 12 (0.202)
Extra-Medit. isl. Australia 7 (0.046) 3 (0.012) 10 (0.036)
the results of a sequence of historical
Iberian-African group — — 43a (0.009)a events during the colonization.
Iber.-Afr. group West Medit. group 39 (0.004) 5 (0.041) 44 (0.008) Also the Slatkin’s Nm estimates reflect
Iber.-Afr. group East Medit. group 37 (0.007) 3 (0.002) 40 (0.007) the common ancestry of populations in
Iber.-Afr. group Guatemala 47 (0.025) 2 (0.121) 49 (0.028)
Iber.-Afr. group Hawaii 45 (0.036) 2 (0.222) 47 (0.044) the recent past, since the majority of the
Iber.-Afr. group Australia 44 (0.014) 1 (0.002) 45 (0.014) considered private alleles are ‘‘ancestral’’
West Mediterranean group — — 8a (0.055)a African alleles. On this basis, the high Nm
West Medit. group East Medit. group 8 (0.030) 8 (0.002) 16 (0.016)
West Medit. group Guatemala 12 (0.102) 1 (0.192) 13 (0.109) estimates detected at a macrogeographi-
West Medit. group Hawaii 13 (0.146) 4 (0.121) 17 (0.140) cal level between Africa and the Mediter-
West Medit. group Australia 11 (0.039) 2 (0.005) 13 (0.034) ranean basin are consistent with both the
East Mediterranean group — — 12a (0.023)a
East Medit. group Guatemala 13 (0.057) 2 (0.121) 15 (0.065)
chronology and with the above mentioned
East Medit. group Hawaii 13 (0.131) 4 (0.139) 17 (0.133) historical circuitous course of medfly
East Medit. group Australia 10 (0.013) 1 (0.002) 11 (0.012) spreading ( Figure 4). The Nm estimates
Guatemala Hawaii 4 (0.331) 6 (0.468) 10 (0.413) between south East Africa to the Iberian–
Guatemala Australia 4 (0.200) 6 (0.141) 10 (0.165)
Hawaii Australia 5 (0.134) 5 (0.331) 10 (0.233)
African area and from this region to the
western and eastern parts of the Mediter-
a
npa and p(1) estimates within each geographical group. ranean area are consistent with the pro-
posed route of colonization from the
source area to northwest Africa and from
seems to have traveled with man to the 1842) and then to other coastal, northern
there to Spain. The Iberian–African area
Mediterranean coast (Maddison and Bart- and eastern Mediterranean locations ( Fi-
seems to have played the major role in the
lett 1989). The fly may have established miani 1989). It was introduced into Austra-
entry of medfly into the Mediterranean ba-
itself progressively from Africa to the Med- lia from Europe around 1897 as a second-
sin. Within this area our data are consis-
iterranean regions of Spain ( De Breme ary colonization event ( Hooper and Drew
tent with a colonization pattern from the
west to the east. Secondary colonization
Table 5. Gene flow estimates within and between the nine geographical groups of C. capitata events such as those from the Mediterra-
populations according to Slatkin’s method nean basin to Australia ( Fimiani 1989) ex-
1 2 3 4 5 6 7 8 9 plain our high Nm estimates between
these groups. However, Slatkin gene flow
1 South East Africa — 1.5 15.2 56.9 24.6 19.4 6.2 7.2 15.7
( Kenya)
estimates depend on relatively few rare al-
2 Réunion — 0.3 0.9 0.4 0.5 0.3 0.4 0.5 leles and can be strongly influenced by on-
going gene flow deriving from trading ac-
3 Extra-Mediterranean islands 10.2 87.0 9.3 26.9 0.3 0.2 6.0
(Canary, Madeira) tivities. This seems to be the case for gene
4 Iberian–African group 94.4 132.0 139.0 8.6 3.8 42.8 flow estimates between Guatemala and
(Morocco, Spain) the Iberian–African region: the Nm value
5 West Mediterranean group 4.1 31.8 0.7 0.5 9.3
(Sardinia, Procida, North Italy) (8.6) appears slightly greater compared to
6 East Mediterranean group 11.6 1.5 0.4 54.6 the Nm estimate between Guatemala and
( Libya, Israel, Crete, Chios, Greece)
7 Guatemala — 0.04 0.3
Kenya (6.2), in spite of the greater genetic
affinity between these two last popula-
8 Hawaii — 0.2 tions. In the same way, the high Nm esti-
9 Australia — mate (87.0) between the Iberian–African
group and the extra-Mediterranean islands

506 The Journal of Heredity 1998:89(6)

 electrophoresis in human genetics. Amsterdam: North-
Holland.
Haymer DS and McInnis DO, 1994. Resolution of popula-
tions of the Mediterranean fruit fly at the DNA level using
random primers for the polymerase chain reaction. Ge-
nome 37:244–248.
Hooper GHS and Drew RAI, 1989. Australia and South Pa-
cific islands. In: Fruit flies: their biology, natural enemies
and control (Robinson AS and Hooper G, eds). Vol. 3A.
Amsterdam: Elsevier Science; 67–72.
Maddison PA and Bartlett BJ, 1989. Contribution towards
the zoogeography of the Tephritidae. In: Fruit flies: their
biology, natural enemies and control (Robinson AS and
Hooper G, eds). Vol. 3A. Amsterdam: Elsevier Science; 27–
35.
Malacrida AR, Gasperi G, Baruffi L, Biscaldi GF and Milani
R, 1988. Updating of the genetics of Ceratitis capitata
(Wied.). In: Modern insect control: nuclear techniques and
biotechnology. Vienna, Austria: International Atomic En-
ergy Agency; 221–227.
Malacrida AR, Guglielmino CR, D’Adamo P, Torti C, Mari-
noni F, and Gasperi G, 1996. Allozyme divergence and phy-
logenetic relationships among species of tephritid flies.
Heredity 76:592- -602.
Malacrida AR, Guglielmino CR, Gasperi G, Baruffi L, and
Milani R, 1992. Spatial and temporal differentiation in col-
onizing populations of Ceratitis capitata. Heredity 69:101–
111.
Figure 4. Slatkin’s Nm estimates along the probable route of C. capitata colonization process from the south East Mantel N, 1967. The detection of disease clustering and a
Africa source area to the Mediterranean basin, to Australia, and to Latin America and Pacific areas. The dashes generalized regression approach. Cancer Res 27:209–220.
arrows indicate the probable gene flow influenced by ongoing trading activities. Mazih A and Debouzie D, 1996. Infestation rate of argan
fruit (Argania spinosa) by the Mediterranean fruit fly (Cer-
atitis capitata) in relation to phenology and maturation of
could be influenced by trading activities Enkerlin D, Garcia LR, Fidel Lopez M, 1989. Mexico, the fruit. Entomol Exp Appl 81:31–38.
Central and South America. In: Fruit flies: their biology,
subsequent to the establishment of the natural enemies and control (Robinson AS and Hooper McPheron BA, Gasparich GE, Han HY, Steck GJ, and Shep-
medfly in these islands. G, eds). Vol. 3A. Amsterdam: Elsevier Science; 83–90. pard WS, 1994. Mitochondrial DNA restriction map for the
Mediterranean fruit fly, Ceratitis capitata. Biochem Genet
In conclusion, from these genetic data it Felsenstein J, 1993. PHYLIP (phylogeny inference pack- 32:25–33.
appears that medfly could be considered age) version 3.5c. Seattle: University of Washington.
Meera Khan P, 1971. Enzyme electrophoresis on cellulose
as a case in which single ancestral popu- Fimiani P, 1989. Mediterranean region. In: Fruit flies: acetate gel: zymogram patterns in man- mouse and man-
their biology, natural enemies and control (Robinson Chinese hamster somatic cell hybrids. Arch Biochem Bio-
lations in eastern and southern Africa give AS and Hooper G, eds). Vol. 3A. Amsterdam: Elsevier phys 145:470–483.
rise to several populations in the recent Science; 39–50.
Nei M, 1978. Estimation of average heterozygosity and ge-
past. The isolation by distance evidenced Fletcher BS, 1989a. Life history strategies of tephritid netic distance from a small number of individuals. Genet-
in the African–Mediterranean area is com- fruit flies. In: Fruit flies: their biology, natural enemies ics 89:583–590.
and control (Robinson AS and Hooper G, eds). Vol. 3B.
patible with a hierarchical migration struc- Amsterdam: Elsevier Science; 195– 208. Papadopoulos NT, Carey JR, Katsoyannos BI, and Kou-
ture. On the other hand, the structure of loussis NA, 1996. Overwintering of the Mediterranean fruit
Fletcher BS, 1989b. Movements of tephritid fruit flies. fly (Diptera: Tephritidae) in northern Greece. Ann Entomol
Latin American and Pacific populations In: Fruit flies: their biology, natural enemies and control Soc Am 89:526–534.
can be the product of few geographically (Robinson AS and Hooper G, eds). Vol. 3B. Amsterdam:
Elsevier Science; 209–219. Raymond M and Ruosset F, 1995. GENEPOP: population
separated colonization events followed by genetics software for exact texts and ecumenicism. J Her-
Gallo DNO, Wiendel FM, Silveira Neto S, and Ricardo ed 86:248–249.
subsequent expansions. These two colo- PLC, 1970. Manual de entomologia. Sao Paulo: Agron-
nization processes, one ancient and one omica Ceres. Rice WR, 1989. Analyzing tables of statistical tests. Evolu-
tion 43:223–225.
new, are correlated with human mobility Gasparich GE, Sheppard WS, Han HY, McPheron BA,
and Steck GJ, 1995. Analysis of mitochondrial DNA and Robinson AS and Hooper G (eds), 1989. Fruit flies: their
capacities at different times and are there- biology, natural enemies and control. Vols. 3A and 3B. Am-
development of PCR-based diagnostic molecular mark-
fore connected with the history of human er for Mediterranean fruit fly (Ceratitis capitata) popu- sterdam: Elsevier Science.
trading activities. lations. Insect Mol Biol 4:61–67. Slatkin M, 1985. Rare alleles as indicators of gene flow.
Gasperi G, Guglielmino CR, Malacrida AR, and Milani R, Evolution 39:53–65.
1991. Genetic variability and gene flow in geographical Slatkin M, 1987. Gene flow and the geographic structure
References
populations of (medfly) Ceratitis capitata (Wied). He- of natural populations. Science 236:787– 792.
Back EA and Pemberton CE, 1918. The Mediterranean redity 64:347–356.
fruit fly. USDA Bulletin 640. Washington, DC; U.S. De- Sneath PHA and Sokal RR, 1973. Numerical taxonomy. San
partment of Agriculture; 1–44. Gomulski L, Morandi PA, Brogna S, Bourtzis K, Savakis Francisco: W. H. Freeman.
C, and Gasperi G, 1996. Intron size variation traces the
Baruffi L, Damiani G, Guglielmino CR, Bandi C, Malacri- Swofford DL and Selander RB, 1981. BIOSYS-1. A computer
world-wide colonization history of the medfly, Ceratitis
da AR, and Gasperi G, 1995. Polymorphism within and program for the analysis of allelic variation in genetics.
capitata. In: Proceedings of the XX International Con-
between populations of Ceratitis capitata: comparison Urbana, Illinois: University of Illinois, Urbana.
gress of Entomology, Florence, Italy, August 25–31,
between RAPD and multilocus enzyme electrophoresis Workman PL and Niswander JD, 1970. Population studies
1996. 229.
data. Heredity 74:425–437. in southwestern Indian tribes. II. Local genetic differentia-
David JR and Capy P, 1988. Genetic variation of Dro- Hagen KS, William WW, and Tassan RL, 1981. Mediter-
tion in the Papago. Am J Hum Gen 22:24–49.
sophila melanogaster natural populations. Trend Genet ranean fruit fly: the worst may be yet to come. Calif
Agric 35:5–7. Wright S, 1965. The interpretation of population structure
4:106–111.
by F-statistics with special regard to systems of mating.
De Breme F, 1842. Note sur le genre Ceratitis de M. Mac Harris EJ, 1989. Hawaiian islands and North America.
Evolution 19:355–420.
heay ( Diptera). Ann Soc Entomol France 11:183–190. In: Fruit flies: their biology, natural enemies and control
(Robinson AS and Hooper G, eds). Vol. 3A. Amsterdam: Received June 17, 1997
Efron B, 1982. The jackknife, the bootstrap, and other
Elsevier Science; 73–81. Accepted February 13, 1998
resampling plans. Conf Board Math Sci Soc Ind Appl
Math 38:1–92. Harris H and Hopkinson DA, 1976. Handbook of enzyme Corresponding Editor: Ross MacIntyre

Malacrida et al • Genetic Variation and Medfly Colonization History 507