70 H
. De Wolf et al. J. Exp. Mar. Biol. Ecol. 246 2000 69 –83
ability e.g. Crisp, 1978; Palumbi, 1994, 1996. In marine gastropods, for example, species with planktonic dispersing larvae display higher levels of gene flow and less
population genetic differentiation e.g., Mitton et al., 1989; Benzie and Williams, 1992; Brown and Murray, 1995, than nonplanktonic developing species with poor dispersal
capacity e.g. Johannesson et al., 1993; Johnson and Black, 1995; Rolan-Alvarez et al., 1995; Trussell, 1996. Nevertheless, even in species with high dispersal abilities there
are several factors that may limit actual dispersal and or gene flow, thus creating opportunities for genetic differentiation as well Palumbi, 1994, 1996. These limitations
include: invisible gene flow barriers, isolation by distance, behavioral limits to dispersal, selection, and the recent history of a species Palumbi, 1994, 1996.
For instance, in the Macaronesian i.e. Azores [AZ], Madeira [MA], Canary Islands [CA] and Cape Verde Islands [CV] planktonic developing periwinkle, Littorina striata,
King and Broderip, 1832, no genetic population differentiation is detected at mi- crogeographical scales i.e. 5–100 m De Wolf et al. 1998a, whereas preliminary
allozyme data tentatively suggest a tendency towards an isolation by distance IBD relationship among more distant populations i.e. up to 500 km within an archipelago
AZ Backeljau et al., 1995. In addition, esterase and random amplified polymorphic DNA RAPD patterns reveal a higher degree of genetic variability in the CV than in the
other archipelagos De Wolf et al. 1998b,c.
In this paper we explore the population genetics of L . striata at macrogeographic
scales by surveying allozyme data from populations covering the entire known geographic distribution of the species i.e. up to 2000 km. We particularly aim at
assessing whether the allozyme variation reveals a macrogeographical patterning and if so, whether this patterning is indeed related to IBD.
2. Materials and methods
2.1. Sampling sites Forty-two populations of L
. striata n 5 1640 were collected at comparable wave-
exposed sites in 13 Macaronesian Islands, including the archipelagos of the AZ, MA, ˜
CA and CV. Sampled islands included: in AZ, Sao Miguel AZ1–12, Pico: AZ13–14, ˜
Santa Maria AZ15, Faial AZ16–18, and Sao Jorge AZ19–20; in MA, Madeira MA1–2, Porto Santo MA3 and Deserta Grande MA4; in CA, Tenerife CA1–4
˜ and Gran Canaria CA5–6; and finally in CV, Sao Nicolau CV1–2, Sal CV3–7, and
˜ Sao Vicente CV8–12 Table 1.
2.2. Allozyme electrophoresis After field collecting, periwinkles were starved for 4 days and subsequently frozen
and stored at 2 808C. Sample preparation and protocols for vertical polyacrylamide gel
electrophoresis were as described by Backeljau and Warmoes 1992. Five polymorphic enzyme loci were analyzed. Glucose phosphate isomerase GPI, E.C. 5.3.1.9, 6-
phosphogluconate dehydrogenase PGD, E.C. 1.1.1.44, and malate dehydrogenase
H . De Wolf et al. J. Exp. Mar. Biol. Ecol. 246 2000 69 –83
71 Table 1
Populations that were analysed with their corresponding abbreviations Archipelago
Island Sampling site
Abbreviation ˜
Azores AZ Sao Miguel SM
Mosteiros AZ1–AZ2
Capellas AZ3
Santana AZ4
Villa de Nordeste AZ5
Faial da Terra AZ6
˜ Povoc¸ao
AZ7 Caloura
AZ8 Lagoa
AZ9 Santa Clara
AZ10 Feteiras
AZ11 Ferereia
AZ12 Santa Maria SAM
Anjos AZ13
˜ Sao Jorge SJ
F. da St. Cristo AZ14
Urzelina AZ15
Pico PI Lajes
AZ16 Calhau
AZ17 Faial FA
P. da Almoxariffe AZ18
Capelinhos AZ19
P. da Espalama AZ20
Madeira MA Madeira MA
Canic¸io de Baixo MA1
P. de St. Caterina MA2
Porto Santo POS Porto Santo
MA3 Deserta Grande DEG
Doca MA4
Canary Is. CA Gran Canaria GRC
Agoustin CA1–CA4
Tenerife TEN Playa da America
CA5–CA6 ˜
Cape Verde Is. CV Sao Nicolau SAN
Harbour CV1–CV2
Sal SAL Santa Maria
CV3–CV4 Jose Fonseca
CV5–CV6 Pedro de Lumen
CV7 ˜
Sao Vicente SAV Baia das Gatas
CV8–CV9 Mindelo
CV10–CV11 Calhau
CV12
MDH, E.C. 1.1.1.37 were resolved in a continuous gel and tray buffer are identical Tris–citric acid buffer at pH 8.0, while hydroxybutyric acid dehydrogenase HBDH,
E.C. 1.1.1.30 was resolved in a continuous Tris–boric acid EDTA buffer at pH 5 8.9.
Mannose phosphate isomerase MPI, E.C. 5.3.1.8 was analyzed using a discontinuous buffer system with a Tris–glycine tray buffer and a Tris–HCl gel buffer, both at pH 9.0.
Enzyme stainings were adapted from Harris and Hopkinson 1976.
2.3. Statistical analysis Allele frequencies, mean numbers of alleles per locus MNA and heterozygosity
levels H 5 observed, H
5 expected were estimated with the BIOSYS program
obs exp
Swofford and Selander, 1989. Genotype frequency departures from Hardy–Weinberg
72 H
. De Wolf et al. J. Exp. Mar. Biol. Ecol. 246 2000 69 –83
HW equilibrium conditions were tested with exact probability tests, implemented by the GENEPOP software v. 1.1 Raymond and Rousset, 1995, which applies the Markov
chain method proposed by Guo and Thompson 1992. Interarchipelago differences in MNA or H
were investigated by means of two
obs
analyses of variance, using the software package STATISTICA v. 5.0 Statsoft, 1995. Genetic population differentiation can be expressed by means of hierarchical F-
statistics. When a hierarchical arrangement of populations is assumed, in this case populations P being placed within islands I, archipelagos A and total distribution
area T, the variance of the observed genetic differentiation among the populations var
can be split up into its variance components. A series of F-statistics can be
ST
obtained e.g., F , F , F and F
, of which the terms in the following equation
PI IA
PT AT
1 2 F 5 1 2 F ? 1 2 F ? 1 2 F
PT PI
IA AT
represent respectively: total differentiation, differentiation among populations within islands, differentiation among islands within archipelagos, and differentiation among
archipelagos. These F-values are not additive. Hence, F reflects only the additional
IA
variance among islands beyond that which exists among populations, and F reflects
AT
only additional variance among archipelagos beyond that which exists among islands Wolf and Campbell, 1995. F-values and corresponding variance components were
calculated with the WRIGHT78 option in BIOSYS Swofford and Selander, 1989. Allele frequency heterogeneities among the four archipelagos, were evaluated with
Fisher exact tests applied to R 3C contingency tables as implemented by the GENEPOP
v. 1.1 software Raymond and Rousset, 1995. The differentiation among islands was further analyzed by means of correspondence
analysis, executed with the NTSYS v. 1.80 software Rohlf, 1993. Gene flow Nm among archipelagos was estimated using private allele frequencies
Slatkin, 1985; Slatkin and Barton, 1989. Pairwise Nm values were plotted against pairwise geographical distances, enabling us to calculate the correlation coefficient and
corresponding regression equation. A significance level of 5 was used throughout. The sequential Bonferroni technique
was employed to correct for false assignments of significance by chance alone multiple test problems Rice, 1989.
3. Results