Both D. nigrospiracula and D. mettleri are primarily associated with necroses of cardon cactus on the Baja California peninsula and both cardon (coastal Sonora) and saguaro on the mainland (Sonora and Arizona), although D. mettleri is also associated with senita necroses. The two species, however, are not in direct competition for resources as D. nigrospiracula develops directly in the necrotic cactus tissue whereas D. mettleri develops in the soil soaked with rot exudates at the base of the cactus. Drosophila nigrospiracula is a strong disperser [8,9] and allozyme and mitochondrial DNA (cytochrome c oxidase subunit I; COI) markers have shown little or no population structure for this species (as well as for D. mettleri) within mainland populations, and between mainland and peninsular populations [10–13]. In D. nigrospiracula, however, allozyme studies suggest some local differentiation in peninsular populations [11]. In D. mettleri, nuclear markers suggest some population structure within each geographic region, although no differentiation was found between several mainland and peninsular populations [14] in agreement with the COI data. A general lack of population structure across the Gulf seen in D. nigrospiracula and D. mettleri (Table 1), together with the fact that cardon cacti which are widely distributed on the Midriff Islands [5] support resident populations of both species [15], are observations consistent with the “stepping-stone” hypothesis of dispersal and gene flow. Migrating flies of both species, however, also have the capability to use a variety of cactus hosts [7,11,16]. Although we cannot rule out long distance dispersal, possibly wind-assisted, directly over water or via the Gran Desierto de Altar at the head of the Gulf (Figure 1), these possibilities seem unlikely given the distances involved and the harsh environmental conditions found. Columnar cacti hosts for both species are absent from the Gran Desierto de Altar (Figure 2), and potential alternate hosts for D. mettleri, such as prickly-pear Opuntia spp. [12], are also largely absent from this extremely arid region.

Genetically differentiated populations of D. mojavensis are found in four geographically isolated regions of southwestern USA and northwestern Mexico: (1) Mojave Desert (including the extreme northwestern Sonoran Desert) of southern California and northwestern Arizona; (2) mainland Sonora and northern Sinaloa; (3) Baja California peninsula and (4) Santa Catalina Island. Based on both morphological [17] and molecular characters [18–22] the four populations have been assigned to the subspecies D. m. mojavensis Patterson and Crow, D. m. sonorensis Castrezana, D. m. baja Mettler and D. m. wrigleyi Castrezana, respectively [17,23]. Molecular markers, including both nuclear and mitochondrial DNA, have shown that each subspecies is essentially panmictic throughout its range, but that significant structure exists between subspecies. The significant population structure found between D. m. sonorensis in Sonora and the peninsular D. m. baja (Table 1) is in contrast to the lack of structure seen for corresponding populations of D. nigrospiracula and D. mettleri separated by the Gulf, suggesting that the Midriff Islands represent a less effective dispersal route for D. mojavensis.

As with cardón, the principal host cacti of D. m. baja (pitahaya agria) and D. m. sonorensis (pitahaya dulce) are found on the Midriff Islands [5]. Pitahaya agria, which is present in Sonora only in a restricted region adjacent to Isla Tiburon, may itself have used these islands as “stepping stones” to disperse from the peninsula, where it is common, to mainland Sonora [4]. Because D. m. mojavensis and D. nigrospiracula show similar dispersal capability [9], with both species being essentially panmictic throughout Sonora, other explanations for genetic differentiation between peninsular and mainland flies are likely but not easily identified [21]. The observation that both mean uncorrected genetic distance (p-distance) [17] and F_st values [21] calculated from mitochondrial COI sequence data are lower between the peninsular and mainland populations, compared with pairwise comparisons from other two regions, suggests less genetic diversification and some gene flow across the Gulf, although nuclear DNA data suggest no gene flow between the two populations [20].

Populations of D. pachea are distributed throughout the mainland and peninsular regions of the Sonoran Desert, largely mirroring the distribution of its principal host cactus Lophocereus schottii (Figure 2) [24,25]. Drosophila pachea also utilizes the closely-related cactus L. gatesii, found only in a small coastal area of Baja California Sur west of La Paz [26]. Drosophila pachea has a lower dispersal ability compared to D. nigrospiracula, D. mettleri and D. mojavensis [9], suggesting a greater potential for local differentiation and population structure. A general lack of population structure, however, was found among most mainland populations of D. pachea, as well as among most peninsular populations, based on allozyme, mitochondrial DNA and microsatellite markers [10,12,25,27], in agreement with results obtained for D. nigrospiracula, D. mettleri and D. mojavensis. Although dispersal capability is lower in D. pachea, the panmixia observed within each major geographic region may be related to resource availability in that senita rots are generally found closer together compared with those of the other columnar cacti [7]. Evidence for significant structure, however, was found between mainland and peninsular populations of D. pachea using mitochondrial DNA [12] (Table 1). In pairwise comparisons among all mainland and peninsular populations, microsatellite data showed evidence for both population structure and panmixia, but a comparison between grouped populations from each of the two geographic areas showed low but significant structure in agreement with the mitochondrial DNA data [25]. Thus in both D. pachea and D. mojavensis, a common pattern of reduced gene flow between the mainland and peninsula compared with D. nigrospiracula and D. mettleri is found, although host plants for all four species of Drosophila are present on the Midriff Islands.

Compared with the other species of cactophilic Drosophila, D. arizonae shows less host plant specialization, utilizing a variety of cacti for feeding and development, principally cina (Stenocereus alamosensis) in Sonora and Opuntia spp. in Baja California Sur; decaying fruits and vegetables are also utilized in urban areas [28]. Interestingly, different molecular markers have revealed contrasting patterns of population structure in D. arizonae in the Sonoran Desert. Analyses based on COI sequences revealed essentially no structure among two nearby populations of D. arizonae from the Cape Region of Baja California Sur, and among two mainland populations from southern Sonora; a lack of structure was also found between the mainland and peninsular populations [21]. Significant structure, however, was found among populations of D. arizonae from Tucson, Arizona and from Riverside, California (just west of the northwestern limit of the Sonoran Desert), and these two northern populations showed significant structure compared to the populations from southern Sonora and Baja California Sur [21]. In contrast to the lack of structure among mainland and peninsular populations of D. arizonae found with the mitochondrial COI data set, analysis of nuclear sequences showed significant structure among populations from these two regions [20]. The nuclear data set, however, confirmed the structure noted among the other populations of D. arizonae sampled in the COI study.

The neriid cactus fly Odontoloxozus longicornis is considered here together with the chernetid pseudoscorpion (Order Pseudoscorpiones) Dinocheirus arizonensis (Banks) with which it shares an unusual symbiotic relationship. Both species are commonly found in necrotic tissue of a wide variety of cacti throughout the Sonoran Desert, with the pseudoscorpion preying upon various cactophilic insects. Although D. arizonensis has quite limited vagility which would tend to compromise its ability to colonize new rots, its dispersal is facilitated by attaching itself to the legs of O. longicornis, or other large cactophilic insects such as coleopterans in the tribe Hololeptini (see below), hitchhiking (phoretic dispersal) when the host transporter disperses to new cactus rots [29]. As with the cactophilic Drosophila (with the possible exception of D. arizonae), COI sequence analysis suggests that mainland populations of O. longicornis are panmictic over a wide geographic area [30]. Although only a single peninsular individual of O. longicornis has been sequenced [30], the haplotype of this individual was the same as the most common COI haplotype on the mainland (haplotype GY2; GenBank FJ532246), suggesting high dispersal capability between the peninsula and mainland.

Consistent with an efficient phoretic dispersal behavior, Dinocheirus arizonensis shows little population structure throughout the mainland Sonoran Desert [30]. Significant structure, however, was noted between peninsular populations and populations from the mainland (Table 1). Phylogenetic analysis showed that the populations of D. arizonensis from each of the two geographic regions resolved as distinct clades separated by a mean Kimura [31] 2-parameter (K2P) genetic distance of 2.6% [30]. Within the peninsular clade, high gene flow also was found among geographically distant populations. If phoresy was not an important dispersal mechanism for D. arizonensis, we would expect to find higher levels of population structure within both peninsular and mainland populations. The relatively high level of differentiation noted between peninsular and mainland populations suggests that the two lineages evolved in allopatry and possibly should be considered distinct taxa at the subspecies or species level. The sympatric occurrence of both lineages in southeastern Arizona [30], which probably occurred by secondary contact after differentiation, is consistent with this interpretation. In this scenario, phoretic dispersal, although promoting relatively high levels of gene flow on the peninsula and mainland, would be much less efficient across the Gulf, even considering the potential of the Midriff Islands to act as “stepping stones”. The effect of an attached pseudoscorpion on flight ability of O. longicornis has not been studied, but in all likelihood it would have a negative effect, possibly leading to the pattern of differentiation observed in D. arizonensis.

Mitochondrial DNA sequence analyses (16S rRNA and COI), combined with comparisons of morphological characters, have been conducted on seven species, or putative species, of Hololeptini associated with cactus necroses in the Sonoran Desert, three in the genus Iliotona [I. beyeri (Schaeffer), I. dorcoides (Lewis) and I. cacti (LeConte)] and four in the genus of Hololepta (H. populnea LeConte, H. vicina LeConte, and Hololepta sp. 1 and sp. 2) [33]. Based on analyses of these combined characters, I. beyeri was shown to be a valid species distinct from I. dorcoides with which it had been placed previously as a synonym [34]. In addition, phylogenetic analyses revealed sister group relationships among two pairs of species separated by the Gulf of California, Hololepta sp. 1 (peninsular) and Hololepta sp. 2 (mainland) and I. beyeri (peninsular) and I. dorcoides (mainland). Each of these species appears to be restricted to its respective geographic region, with an estimated age of divergence among each sister species pair, based on molecular clock calculations (Table 1), consistent with geological estimates for the formation of the Gulf during the late Miocene and early Pliocene [2,3,35,36]. The results suggest very limited, or a lack of dispersal of these species across the Gulf, consistent with the view that divergence and speciation in these taxa resulted from vicariance. Within the peninsular region, however, our data suggest that I. beyeri is panmictic [37], implying that it is a good disperser in its terrestrial habitat. Our sample of I. beyeri from the peninsula was collected on senita, except for a single individual found on pitahaya agria. Both cacti are present on the Midriff Islands (Figure 2).

At least ten species of very small beetles (∼2–3 mm in length) in the genus Carcinops are known to be associated with cactus necroses in the Sonoran Desert. Of these, four have been formally described: C. consors (LeConte), C. gilensis (LeConte), C. corticalis (LeConte) and C. opuntiae (LeConte); six additional species identified using both morphological and molecular characters [32] await formal description. In our sampling of cactus necroses throughout the peninsular and mainland Sonoran Desert a large number of Carcinops were found. We present a brief summary of our preliminary findings on the population genetics of this group, based on analysis of COI sequences [37], which complement the findings on the other species of cactophilic arthropods described above.

Utilizing both morphological and molecular (COI) characters given in Swanson [32] we determined that most of the peninsular Carcinops in our sample belonged to the taxon C. gilensis. Carcinops gilensis is a widely distributed histerid beetle, known from southern Arizona and southern California in the USA, and from Sonora and the Baja California peninsula in Mexico. Carcinops gilensis is typically associated with the necrotic stems of saguaro and cardon, but California specimens were taken from necrotic barrel cactus (Ferocactus sp.) [32] and peninsular specimens were collected together with the Hololeptini [33], mainly from senita. Our results reveal little population structure in C. gilensis throughout the peninsular region and, in addition, show no significant structure between samples from the Guaymas area on the mainland and populations from the mid-peninsular region. These results suggest that C. gilensis is able to disperse across the Gulf, probably via the Midriff Islands. Thus, C. gilensis and I. beyeri which share the same host on the Baja California peninsula appear to show similar dispersal capabilities throughout the peninsula, but I. beyeri which is much larger (>10 mm in length) apparently is restricted to the peninsula.

Phylogenetic analyses also revealed a sister group relationship among two undescribed species of Carcinops which are geographically separated, Carcinops sp. 1 (peninsular) and Carcinops sp. 2 (mainland; also identified by Swanson [32] but not yet formally described). Mean K2P genetic distance among the two populations was 3.9%, slightly lower than the mean value of 6.0% estimated for Hololepta sp. 1 and sp. 2 (Table 1). As with Hololepta sp. 1 and sp. 2, molecular clock calculations suggest that Carcinops sp. 1 and sp. 2 diverged during the Pliocene (Table 1). In contrast to the results seen with C. gilensis, our preliminary analyses suggest a lack of gene flow across the Gulf for Carcinops sp. 1 and sp. 2. For comparison, Table 1 also shows that genetic distance and estimated divergence time between peninsular and mainland populations (subspecies) of the hematophagous kissing bug Triatoma rubida (Uhler) (Hemiptera: Reduviidae) are similar to those obtained for Hololepta sp.1 and sp. 2 and Carcinops sp.1 and sp. 2 [38].