Differentiation under isolation and genetic structure of Sardinian hares as revealed by craniometric analysis,mitochondrial DNA and microsatellites

Hares (Lepus capensis Linnaeus 1758) were probably introduced into Sardinia in historical times. Previous studies indicated North Africa as the most likely source area but did not exclude the occurrence of hybridization events with continental brown hares (L. europaeus Pallas 1778) perhaps introduced for hunting purposes. We implemented both morphometric and genetic approaches to verify the genetic isolation of the Sardinian population. Specifically, we conducted a multivariate analysis of craniometric data and analysed 461 bp of the mitochondrial control region and 12 autosomal microsatellites in Sardinian hares, using North African cape hares and European brown hares as reference populations. Sardinian hares displayed a peculiar skull shape. In agreement, both nuclear and mitochondrial markers remarked the distinctiveness of this population. Observed and expected heterozygosity were 0.52 and 0.61, while haplotype and nucleotide diversity were 0.822 and 0.0129. Self‐assignment based on Bayesian cluster analysis was high (average membership 0.98), and no evident signs of introgression from continental brown hares were found. Our results support the hypothesis that the Sardinian hares have been introduced from North Africa, remained genetically isolated since the founding event and evolved independently from the source population. This long‐lasting isolation and the consequent genetic drift resulted in a differentiation, perhaps accompanied by an adaptation to local environmental conditions.     

Phylogenetic analysis of mtCR-1 sequences of Tunisian and Egyptian hares (Lepus sp. or spp., Lagomorpha) with different coat colours

North African hares are currently considered belonging to cape hares (Lepus capensis), except for an isolated occurrence of L. victoriae in NW Algeria. However, the few existing molecular data are not unequivocal. Here, we study sequence variation (415 bp) in the hypervariable domain-1 of the mitochondrial (mt) control region, of hares with different coat colour from north-central Tunisia and NW Egypt, to test Petter's [(1959): Eléments d’une révision des Lièvres africains du sous-genre Lepus. Mammalia 23, 41–67] hypothesis that North African hares belong to L. capensis. Seven Tunisian and one Egyptian haplotypes were revealed from 28 hares and compared phylogenetically to 245 haplotypes of various Lepus species downloaded from GenBank. Neighbour joining (NJ) and principal coordinate (PCO) analyses based on a Tamura-Nei 93 distance matrix, as well as maximum parsimony (MP) analysis concordantly grouped all currently obtained haplotypes together into one monophyletic clade, and revealed relatively close relationships to the clades of African scrub hares (L. saxatilis) and brown hares (L. europaeus). The three distinguished coat colour types of Tunisian hares were paralleled only to a small extent by sequence differentiation. Haplotypes of L. capensis from the nominal Cape province of South Africa, North Africa, and China clustered into different major clades, respectively, with Chinese L. capensis haplotypes forming only a subclade within a major clade that encompassed predominantly “mountain/arctic hare-type sequences” in addition to sequences of several other palaearctic and nearctic species. One further Chinese L. capensis haplotype clustered into the L. comus clade. These results indicated occurrence of introgression and/or shared ancestral polymorphism. Such an evolutionary scenario implies using nucelar markers in addition to mtDNA for phylogenetic inferences in the genus Lepus; nevertheless, mtDNA is still useful for inferring phylogenetic history and biogeography of hares.     

Molecular Approaches Revealing Prehistoric, Historic, or Recent Translocations and Introductions of Hares (Genus Lepus) by Humans

Although only of medium size, and thus of little nutritional value compared to big game such as mammoths and ungulates, hares (Lepus spp.) probably have always been a food source for humans, as documented in archaeological finds. Nowadays, hares, particularly such species as the brown hare (L. europaeus), are among the most important game species in many European countries. For hunting, perhaps religious reasons, and in connection with certain myths, hares have been and are still being intentionally translocated. Ancient translocations by humans can be inferred from the presence of hares on islands that had no mainland connections, at least during the Pleistocene, the major evolutionary period of the genus Lepus. We review some of the literature on anthropogenic translocations of hares. We focus on three examples [the brown hare (L. europaeus), the Corsican hare (L. corsicanus), and the Sardinian hare (L. capensis)], where some molecular data could be used to trace the translocation routes and possible origins of introduced hare populations. Certain molecular marker systems, such as sequences of the hypervariable part I (HV-1) of the mitochondrial control region, show high variability in hare species and are thus promising for tracing both recent and ancient origins of translocated hares. Some other molecular marker systems as well as caveats connected with the use of such marker systems in the genus Lepus are also discussed.     

Genetic isolation of insular populations of the Maghrebian bat,Myotis punicus,in the Mediterranean Basin

Aim We investigate the population genetic structure of the Maghrebian bat, Myotis punicus, between the mainland and islands to assess the island colonization pattern and current gene flow between nearby islands and within the mainland. Location North Africa and the Mediterranean islands of Corsica and Sardinia. Methods We sequenced part of the control region (HVII) of 79 bats across 11 colonies. The phylogeographical pattern was assessed by analysing molecular diversity indices, examining differentiation among populations and estimating divergence time. In addition, we genotyped 182 bats across 10 colonies at seven microsatellite loci. We used analysis of molecular variance and a Bayesian approach to infer nuclear population structure. Finally, we estimated sex‐specific dispersal between Corsica and Sardinia. Results Mitochondrial analyses indicated that colonies between Corsica, Sardinia and North Africa are highly differentiated. Within islands there was no difference between colonies, while at the continental level Moroccan and Tunisian populations were highly differentiated. Analyses with seven microsatellite loci showed a similar pattern. The sole difference was the lack of nuclear differentiation between populations in North Africa, suggesting a male‐biased dispersal over the continental area. The divergence time of Sardinian and Corsican populations was estimated to date back to the early and mid‐Pleistocene. Main conclusions Island colonization by the Maghrebian bats seems to have occurred in a stepping‐stone manner and certainly pre‐dated human colonization. Currently, open water seems to prevent exchange of bats between the two islands, despite their ability to fly and the narrowness of the strait of Bonifacio. Corsican and Sardinian populations are thus currently isolated from any continental gene pool and must therefore be considered as different evolutionarily significant units (ESU).     

First record of planktonic crustaceans in Sardinian reservoirs

Sardinian man-made lakes are reservoirs of species richness, hosting zooplankton taxa from the Mediterranean region and North Africa. To provide a first record of the taxa composition and diversity of zooplankton communities, we sampled 15 reservoirs during 2008–2009, from the north of the island to the south, representative of a range of size, depth, renewal time, and trophy. The survey was complemented by seasonal sampling in one of the largest lakes studied. Water samples collected from surface to bottom provided data on hydrochemistry and trophy. Crustacean dormant stages were inspected from sediments of the richest, and most diverse, Lake Sos Canales. RDA suggested that productivity, water depth, renewal time and altitude were the main variables related to taxa composition. The ubiquitous Copidodiaptomus numidicus, and its persistence in the water column, resulted from the production of subitaneous eggs throughout the year, an adaptive strategy in perennial water bodies. Genetic analyses of DNA sequences of the diagnostic gene ND5 placed the Sardinian Daphnia pulex in the North American group. Moreover, the ND5 sequence found in Sardinia was identical with that of an asexual hybrid clone between the American D. pulex and American D. pulicaria that replaced native D. pulex throughout Africa. The presence of this ND5 haplotype in Sardinia shows that this invasive clone also poses an invasive threat to native populations in Europe.     

On shortcomings of using mtDNA sequence divergence for the systematics of hares (genus Lepus): An example from cape hares

Despite the well-known fact that evolutionary patterns of single genes or sequences are not necessarily paralleled by organismic evolution, using mitochondrial sequence divergence for inferring phylogenetic relationships among hare species is not uncommon. Earlier studies reported interspecific introgression in the genus Lepus that may slur systematic conclusions drawn exclusively from mtDNA data. We examined pairwise divergence in partial mtHV-1 sequences and nuclear gene pools based on microsatellite and allozyme loci separately in a South and North African population of cape hares, Lepus capensis, and did not find significant correspondence on the individual level within either population. Also, the former population had a significantly higher level of mtHV-1 sequence divergence compared to the latter, but levels of individual nuclear gene pool divergence did not differ significantly between the two populations. Hence, the absence of correspondence between mtHV-1 sequence divergence and nuclear gene pool divergence among individuals within a population might be independent of the population-specific level of differentiation among individuals. Our results complement earlier findings in hares, and strongly recommend to include nuclear gene pool evidence for systematic inferences within this genus, even under the absence of mitochondrial introgression.     

Phylogeographic inferences from the mtDNA variation of the three-toed skink, Chalcides chalcides (Reptilia: Scincidae)

Genetic diversity was analyzed in Chalcides chalcides populations from peninsular Italy, Sardinia, Sicily and Tunisia by sequencing 400 bp at the 5' end of the mitochondrial gene encoding cytochrome b (cyt b) and by restriction fragment length polymorphism (RFLP) analysis of two mitochondrial DNA segments (ND-1/2 and ND-3/4). The results of the phylogenetic analysis highlighted the presence of three main clades corresponding with three of the four main geographical areas (Tunisia, Sicily and the Italian peninsula), while Sardinia proved to be closely related to Tunisian haplotypes suggesting a colonization of this island from North Africa by human agency in historical times. On the contrary, the splitting times estimated on the basis of cyt b sequence data seem to indicate a more ancient colonization of Sicily and the Italian Peninsula, as a consequence of tectonic and climatic events that affected the Mediterranean Basin during the Pleistocene. Finally, the analysis of the genetic variability of C. chalcides populations showed a remarkable genetic homogeneity in Italian populations when compared to the Tunisian ones. This condition could be explained by a rapid post-glacial expansion from refugial populations that implied serial bottlenecking with progressive loss of haplotypes, resulting in a low genetic diversity in the populations inhabiting the more recently colonized areas.     

Satyrinae butterflies from Sardinia and Corsica show a kaleidoscopic intraspecific biogeography (Lepidoptera,Nymphlidae)

The Mediterranean islands of Sardinia and Corsica are known for their multitude of endemics. Butterflies in particular have received much attention. However, no comprehensive studies aiming to compare populations of butterflies from Sardinia and Corsica with those from the neighbouring mainland and Sicily have been carried out. In the present study, the eleven Satyrinae species inhabiting Sardinia and Corsica islands were examined and compared with continental and Sicilian populations by means of geometric morphometrics of male genitalia. Relative warp computation, discriminant analyses, hierarchical clustering, and cross‐validation tests were used to identify coherent distributional patterns including both islands and mainland populations. The eleven species showed multifaceted distributional patterns, although three main conclusions can be drawn: (1) populations from North Africa and Spain are generally different from those belonging to the Italian Peninsula; (2) populations from Sardinia and Sicily often resemble the North Africa/Spain ones; Corsica shows transitional populations similar to those from France; and (3) sea barriers represent filters to dispersal, although their efficacy appears to be unrelated to their extension. Indeed, the short sea straits between Sardinia and Corsica and between Sicily and the Italian Peninsula revealed a strong effectiveness with respect to preventing faunal exchanges; populations giving onto sea channels between Corsica and Northern Italy and between Sicily and Tunisia showed a higher similarity. A comparison of island and mainland distributions of the eleven taxa have helped to unravel the complex co‐occurrence of historical factors, refugial dynamics, and recent (post‐glacial) dispersal with respect to shaping the populations of Mediterranean island butterflies. © 2010 The Linnean Society of London, Biological Journal of the Linnean Society, 2010, 100 , 195–212.     

Molecular evidence of conspecificity of South African hares conventionally considered Lepus capensis L., 1758

Conventionally, Lepus capensis is considered to range across large parts of Africa, the Middle East, Central and Far East Asia. However, a recent morphological study restricts cape hares tentatively to a small range in the Western Cape Region of South Africa and groups all other L. capensis-type hares from South Africa into a new species: L. centralis. Here, we studied molecular relationships among L. capensis-type hares from South Africa. Phenotypically and morphologically the individuals matched either the newly described L. capensis or L. centralis. We examined 66 hares for allelic variation at 13 microsatellite loci and for sequence variation of the hypervariable domain 1 of the mitochondrial control region. All tree-generating analyses of the currently obtained sequences and all South African cape hare sequences downloaded from GenBank revealed monophyly when compared to sequences of various other Lepus species. A network analysis indicated close evolutionary relationships between hares of the “L. capensis-phenotype” and the “L. centralis-phenotype” (according to Palacios et al. 2008) from the southwest of the Western Cape, relative to their pronounced evolutionary divergence from all other more central, northern, and north-eastern L. capensis-type hares. F-statistics, a Bayesian admixture STRUCTURE model, as well as a principal coordinate analysis of microsatellite data indicated close genetic relationships among all South African L. capensis-type hares studied presently. A coalescence model-based migration analysis for microsatellite alleles indicated gene flow between most of the considered subspecies of cape hare, including L. capensis capensis and L. capensis centralis, theoretically sufficient to balance stochastic drift effects. Concordantly, AMOVA models revealed only little effects of partitioning microsatellite variation into the two suggested morpho-species “L. capensis” and “L. centralis”. Under an “Interbreeding Species Concept” (e.g. a strict or relaxed Biological Species Concept), the current molecular data demonstrate conspecificity of the two proposed morpho-species “L. capensis” and “L. centralis”. Based on the present molecular data the differentiation of subspecies of cape hares from southern Africa is discussed.     

Isozyme variation inFaidherbia albida (Leguminosae, Mimosoideae)

Genetic variation has been assessed in 30 populations (931 families) ofFaidherbia albida (Leguminosae, Mimosoideae) from across its entire African range, using six isozyme loci identified by five enzyme systems. Among the populations studied a null allele was proposed to explain the absence ofLap-1 activity in populations from southern and eastern Africa. The mean percentage of polymorphic loci per population, the mean number of alleles per locus and the mean genetic diversity within populations were 31.7%, 1.6 and 0.127 respectively. Genetic diversity was greatest in populations from West Africa and lowest in populations from eastern/southern Africa, with Ethiopian/Sudanese populations intermediate. The overall degree of genetic differentiation between populations (G_ST) indicated that approximately 56% of the enzyme variation resided within populations. Clustering of Nei's unbaised genetic distances calculated between all populations produced a dendrogram that generally followed the geographic distribution of the populations. Two major groups were identified that may be considered the eastern/southern African and the Ethiopian/West African clusters. Within the Ethiopian/West African cluster two subclusters could be recognised, one broadly corresponding to those populations from Ethiopia/Sudan and the other to those populations from West Africa. The implications of these results for theories regarding the origin ofF. albida in Africa are discussed.     

Where do Sardinian orchids come from: a putative African origin for the insular population of Platanthera bifolia var. kuenkelei?

Because of the complex geoclimatic dynamics of the Mediterranean basin, resulting from a combination of Tertiary geological processes and Quaternary climatic oscillations, the current species composition of the main continental islands is a mosaic resulting from relative contribution of vicariance, geodispersal and long‐distance dispersal by seed. To help distinguish among the contributions of these different dispersal modes to the orchid species richness on the island of Sardinia, we investigated the genetic structure and phylogeographic relationship of the only Platanthera bifolia var. kuenkelei population on the island. By using plastid DNA sequence variation, we show that this recently discovered population is more closely related to the Tunisian P. bifolia var. kuenkelei than to the geographically closer P. bifolia populations from peninsular Italy. Results from haplotype network construction support the hypothesis that the current distribution of P. bifolia var. kuenkelei (Sardinia and Tunisia) is most likely explained by over‐sea dispersal. However, haplotype diversity and mismatch analysis of this unique Sardinian population suggest that, if a long‐distance dispersal event did occur, it was not recent. More likely, temporarily favourable climatic conditions generated a suitable habitat on the island (also in terms of suitable mycorrhizas and pollinators) and thus a transient opportunity for seedling growth and population establishment of P. bifolia var. kuenkelei from North Africa. © 2011 The Linnean Society of London, Botanical Journal of the Linnean Society, 2011, 167 , 466–475.     

Phylogeography of two cryptic species of African desert jerboas (Dipodidae: Jaculus)

The lesser Egyptian jerboa Jaculus jaculus is a desert dwelling rodent that inhabits a broad Arabian–Saharan arid zone. Recently, two distant sympatric lineages were described in North‐West Africa, based on morphometric and molecular data, which may correspond to two cryptic species. In the current study, phylogenetic relationships and phylogeographical structure among those lineages and geographical populations from North Africa and the Middle East were investigated. The phylogeographical patterns and genetic diversity of the cytochrome b gene (1110 bp) were addressed on 111 jerboas from 41 localities. We found that the variation in Africa is partitioned into two divergent mitochondrial clades (10.5% divergence relating to 1.65–4.92 Mya) that corresponds to the two cryptic species: J. jaculus and J. deserti. Diversifications within those cryptic species/clades were dated to 0.23–1.13 Mya, suggesting that the Middle Pleistocene climatic change and its environmental consequences affected the evolutionary history of African jerboas. The third distant clade detected, found in the Middle East region, most likely represents a distinct evolutionary unit, independent of the two African lineages. © 2012 The Linnean Society of London, Biological Journal of the Linnean Society, 2012, ??, ??–??.     

Genetics,geography, and culture: The population of S. Pietro Island (Sardinia,Italy)

An interesting aspect of the island of Sardinia (Italy) is the wide range of genetic variability within the island itself. The variability is widened by the presence of some populations of different ethnic origin who speak a language other than Sardinian. This work deals with the study of the genetic structure of the Carloforte population which inhabits the tiny island of S. Pietro 4 km off the southwest coast of Sardinia. S. Pietro was first populated in 1738 by emigrants coming from the island of Tabarka (Tunisia) who spoke an archaic form of the Ligurian dialect. Data on genetic polymorphisms in the Carloforte population are presented and discussed in relation to some Sardinian and Italian populations. Data on demographic and matrimonial structure are also presented. The genetic analyses show the Carloforte population as being clearly separated from both Sardinians and continental Italians. The isolation of Carloforte, highlighted by language diversity, endogamy, and consanguinity levels and marriage area, supports the idea of genetic diversity linked to cultural peculiarity. © 1996 Wiley-Liss, Inc.     

Paternal lineages in Libya inferred from Y‐chromosome haplogroups

Many studies based on genetic diversity of North African populations have contributed to elucidate the modelling of the genetic landscape in this region. North Africa is considered as a distinct spatial‐temporal entity on geographic, archaeological, and historical grounds, which has undergone the influence of different human migrations along its shaping. For instance, Libya, a North African country, was first inhabited by Berbers and then colonized by a variety of ethnic groups like Phoenicians, Greeks, Romans, Arabs and, in recent times, Italians. In this study, we contribute to clarify the genetic variation of Libya and consequently, of North African modern populations, by the study of Libyan male lineages. A total of 22 Y‐chromosome‐specific SNPs were genotyped in a sample of 175 Libyan males, allowing the characterization of 18 Y‐chromosomal haplogroups. The obtained data revealed a predominant Northwest African component represented by haplogroup E‐M81 (33.7%) followed by J(xJ1a,J2)‐M304 (27.4%), which is postulated to have a Middle Eastern origin. The comparative study with other populations (～5,400 individuals from North Africa, Middle East, Sub‐Saharan Africa, and Europe) revealed a general genetic homogeneity among North African populations (F_ST = 5.3 %; P‐value < 0.0001). Overall, the Y‐haplogroup diversity in Libya and in North Africa is characterized by two genetic components. The first signature is typical of Berber‐speaking people (E‐M81), the autochthonous inhabitants, whereas the second is (J(xJ1a,J2)‐M304), originating from Arabic populations. This is in agreement with the hypothesis of an Arabic expansion from the Middle East, shaping the North African genetic landscape. Am J Phys Anthropol 157:242–251, 2015. © 2015 Wiley Periodicals, Inc.     

Biochemical Genetic Relationships Among Tunisian Hares (Lepus sp.), South African Cape Hares (L. capensis), and European Brown Hares (L. europaeus)

Tunisian hares (n = 45), currently assigned to Lepus capensis, were assayed for allelic variation at 40 allozyme loci, and allele frequencies at 32 loci were directly compared with earlier data of South African cape hares (L. capensis, n = 9) and European brown hares (L. europaeus, n = 244) to reveal genetic relationships among them. European mountain hares (L. timidus, n = 200) were used for outgroup comparison. In the Tunisian hares 27.5% of the loci were polymorphic with 2–4 alleles. Among all alleles at polymorphic loci, 15.1% occurred exclusively in Tunisian hares, 5.7% exclusively in cape hares, and 7.5% exclusively in brown hares at low frequencies. Not a single locus showed alternately fixed alleles between the samples of the L. capensis/L. europaeus complex. Levels of absolute and relative genetic differentiation among the samples of the L. capensis/ L. europaeus complex were low, relative to pairwise comparisons involving mountain hares. Diverse cluster analyses and multidimensional scaling of various pairwise genetic distance matrices concordantly grouped Tunisian hares with brown hares, and South African cape hares clustered only slightly farther apart, whereas mountain hares were distinctly separate. These results suggest regionally distinct phylogenetic units within an overall cohesive gene pool in the L. capensis/ L. europaeus complex, supporting Petter's view that all North African hares belong to L. capensis except for one local population of savanna hares, and that cape hares and brown hares are conspecific.     

Y chromosome binary markers to study the high prevalence of males in Sardinian centenarians and the genetic structure of the Sardinian population

We have analyzed a sample of 40 centenarians and 116 young controls from Sardinia, with a set of new Y chromosome binary markers, to evaluate if Y chromosome genes are involved in the high prevalence of males among centenarian Sardinians (1/2 vs. 1/4 in other populations studied). The results indicate that none of the seven lineages that account for >97% of the Y chromosome diversity in Sardinia provide an advantage with respect to the extreme longevity. However, our results, although based on the male-specific Y chromosome polymorphisms, give a clear profile of the pattern of genetic variability in Sardinia. Indeed they indicate that the Sardinian population had two main founder populations that have evolved in isolation for at least the last 5,000 years. These findings set the stage for future studies on longevity and other complex traits in Sardinia.     

Isolation and characterization of microsatellite loci in the striped mullet,Mugil cephalus

Forty‐three microsatellites were isolated from grey mullet (Mugil cephalus) of Mediterranean origin using the standard enrichment methods followed by enhanced chemiluminescence for the rapid identification of clones containing microsatellites. Eleven microsatellites were characterized to assess genetic diversity of different populations surrounding the island of Sardinia in Italy. The versatility of these markers was also assessed using an Australian population. All 11 microsatellites were highly polymorphic with a mean heterozygosity of 0.834. The number of alleles ranged from 14 to 38 in the Sardinian populations and from 11 to 25 in the Australian samples.     

Where are you from,stranger? The enigmatic biogeography of North African pond turtles (Emys orbicularis)

The European pond turtle (Emys orbicularis) is a Nearctic element in the African fauna and thought to have invaded North Africa from the Iberian Peninsula. All North African populations are currently identified with the subspecies E. o. occidentalis. However, a nearly range-wide sampling in North Africa used for analyses of mitochondrial and microsatellite DNA provides evidence that only Moroccan populations belong to this taxon, while eastern Algerian and Tunisian pond turtles represent an undescribed distinct subspecies. These two taxa are most closely related to E. o. galloitalica with a native distribution along the Mediterranean coast of northern Spain through southern France to western and southern Italy. This group is sister to a clade comprising several mitochondrial lineages and subspecies of E. orbicularis from Central and Eastern Europe plus Asia, and the successive sisters are E. o. hellenica and E. trinacris. Our results suggest that E. orbicularis has been present in North Africa longer than on the Iberian Peninsula and that after an initial invasion of North Africa by pond turtles from an unknown European source region, there was a phase of diversification in North Africa, followed by a later re-invasion of Europe by one of the African lineages. The differentiation of pond turtles in North Africa parallels a general phylogeographic paradigm in amphibians and reptiles, with deeply divergent lineages in the western and eastern Maghreb. Acknowledging their genetic similarity, we propose to synonymize the previously recognized Iberian subspecies E. o. fritzjuergenobsti with E. o. occidentalis sensu stricto. The seriously imperiled Moroccan populations of E. o. occidentalis represent two Management Units different in mitochondrial haplotypes and microsatellite markers. The conservation status of eastern Algerian pond turtles is unclear, while Tunisian populations are endangered. Considering that Algerian and Tunisian pond turtles represent an endemic taxon, their situation throughout the historical range should be surveyed to establish a basis for conservation measures.     

Space-time diversification of Androcymbium Willd. (Colchicaceae) in western South Africa

 We examined patterns of cpDNA RFLP variability using 21 restriction endonucleases in 21 populations of Androcymbium that represent 12 endemic species distributed in the winter rainfall areas of South Africa to explore the diversification of the genus in its area of maximum species diversity. Our results are supportive of a diversification landscape characterized by continued opportunistic short-range invasion, naturalization, and rapid speciation, in which the selective action of the different environments where Androcymbium species occur determined their colonization success and subsequent short-range geographic expansion. The historical presence of fire, the constraint imposed by the low concentration of nutrients throughout southwestern South Africa and the different reproductive capabilities of Androcymbium species have also likely stimulated species' diversification. Our divergence time estimates bolster the view that speciation of South African Androcymbium initiated in the late Eocene, intensified in the Oligocene and proceeded more sporadically during the Miocene. These chronological estimates also substantiate the previous hypothesis that most lineages of Androcymbium in South Africa are much more ancient than their North African relatives, whose diversification began in the late Miocene-early Pliocene. Received July 25, 2001 Accepted December 6, 2001