Introduction

Intertidal oribatid mites represent a small group of tiny arachnids that have adapted to the littoral environment. There, they live in the zone between low and high tide using diverse algae as substrate and food source (e.g. Pfingstl, 2017). For a long time, Caribbean coasts seemed to be devoid of these organisms, but a recent study (Pfingstl, 2021) revealed wide distributions of members of two families, the Fortuyniidae and the Selenoribatidae, and demonstrated that these mites are common parts of the littoral Caribbean fauna. The Caribbean region consists of continental fragments, namely the Greater Antilles, which broke off the mainland ca. 40 million years ago, and of recent volcanic islands, the Lesser Antilles which have emerged in the last 10 million years (Iturralde-Vinent & MacPhee, 1999). The geological history of this area has created unique conditions for colonization and diversification and thus features a tremendous biodiversity and high levels of endemism (e.g. Ricklefs & Bermingham, 2008).

Mites are very small arthropods with limited dispersal abilities and therefore widespread species, especially in the Caribbean with its numerous islands separated by oceanic barriers, would be rather exceptional. Indeed, several intertidal oribatid species, with supposedly wide distributions across the Caribbean, recently turned out to be subject to high cryptic diversity or to represent cryptic species complexes (Pfingstl et al., 2019a, b, 2021, 2022). Cryptic species are defined as species that are classified as a single nominal species because they are at least superficially anatomically identical (Bickford et al., 2007). This means that the morphology of theses taxa has remained stable and is highly conserved, whereas they have strongly diversified on a genetic level due to a lack of gene flow. For example, different members of the cryptically diverse Carinozetes mangrovi Pfingstl et al., 2014 a selenoribatid Caribbean intertidal mite, cannot be distinguished from each other based on morphology, but actually consist of three distinct genetic lineages, a Northern, an Antillean and a Pacific lineage (Pfingstl et al., 2019a, b). Similarly, Caribbean members of the fortuyniid intertidal Fortuynia atlantica Krisper & Schuster, 2008 show an identical morphology throughout their distribution range, but in fact include a genetically distinct species occurring on the Lesser Antilles (Pfingstl et al., 2022). An even more intriguing case is that of the Caribbean Thalassozetes barbara Pfingstl, 2013, another selenoribatid mite, that could be demonstrated to comprise seven phenotypically almost identical but genetically distinct species, with nearly every species being an island or a short-range endemic (Pfingstl et al., 2021). Although the extent of diversity and the cause for it may differ between the cases, the found morphological stasis of all is supposed to be a result of stabilizing selection caused by the extreme conditions of the intertidal environment (Pfingstl et al., 2019a, b, 2021, 2022).

Another known Caribbean species is the fortuyniid Alismobates inexpectatus Pfingstl & Schuster, 2012, which was first discovered on the small Archipelago of Bermuda in the Western Atlantic (Pfingstl & Schuster, 2012). At the time, it was also the first and thus unexpected record of this genus from the Atlantic area, therefore the name inexpectatus was given to this species. Presently, A. inexpectatus is known to occur on the shores of Florida, the Bahamas, Hispanola, Central America and several islands of the Lesser Antilles (Pfingstl, 2021), and thus shows apparently a trans-Caribbean distribution. However, in view of the above-mentioned circumstances, a wide-spread single species Alismobates inexpectatus seems to represent an unlikely hypothesis and finding cryptic diversity within this taxon would not be unexpected at all.

To screen members of A. inexpectatus for cryptic diversity, we investigated more than 500 specimens from nine different regions/islands of the Caribbean and the Western Atlantic using morphometric and molecular genetic analyses. Our specific aims were (I) to reveal cryptic taxa if present, (II) to assess morphological and genetic variation among different A. inexpectatus populations, and (III) to interpret found patterns from a bio- and paleogeographic point of view.

Material and methods

Sample collection and locations

In the years 2016–2018, coastal mites were collected during three fieldtrips to Bermuda and several regions in the Caribbean. Samples of intertidal algae were scraped off rocks or mangrove roots with a knife during low tide. Algae were then put in Berlese-Tullgren funnels for about 24 h to extract living mites. Specimens were preserved and stored in absolute ethanol for subsequent morphological and molecular genetic investigation. All samples were taken by T. Pfingstl and A. Lienhard, except for the sample from Costa Rica, which was collected by G. Kunz. Details are given in Table 1.

Table 1 Details about samples taken in the years 2016–2018. The abbreviation B.o.r means Bostrychia (intertidal alga) on rock. If two labels are given for one location, then two separate samples were taken in more or less the same spot

Molecular genetics

Specimens from Costa Rica were not used for this study, because we had no permit to perform molecular genetic analyses with this collected material.

For genetic analyses, whole genomic DNA was extracted from all 238 specimens included in this study using Chelex resin (Pfingstl et al., 2022). Region 2 of the mitochondrial cytochrome c oxidase subunit 1 gene (COI) was amplified with primers by Otto and Wilson (2001), these are: Mite COI – 2F TTY GAY CCI DYI GGR GGA GGA GAT CC and Mite COI – 2R GGR TAR TCW GAR TAW CGN CGW GGT AT. The nuclear 18S rRNA gene (18S) was amplified using the primers designed by Dabert et al. (2010). PCR amplification with subsequent DNA purification and cycle sequencing were conducted as described in Pfingstl et al. (2022), with the exception of the second purification step using the BigDye XTerminator Purification Kit according to the protocol by Applied Biosystems. Automatic capillary sequencing and sequence visualization then was conducted on an ABI3500XL (Applied Biosystems) device.

All generated sequences are available from GenBank under the accession numbers OR358561 to OR358798 for COI and OR360272 to OR360328 for 18S rRNA (see Appendix).

We sequenced 622 bp of the COI gene in all 238 specimens and 1,806 bp of the 18S rRNA gene in 57 individuals, which comprised representatives of all potential mitochondrial lineages. Additionally, sequences of nine species belonging to the superfamily Ameronothroidea, a group of mainly marine associated oribatid mites, were taken from GenBank and added to the COI and 18S rRNA alignments (see Appendix), whereof two Fortuynia species were chosen as outgroup taxa. In a first trial, we analyzed both genes separately and in a next step, fragments of both genes were combined for 63 specimens (only those with all available sequences were used) in a concatenated dataset with a final length of 2445 bp.

PhyloSuite v.1.2.2 (Zhang et al., 2020) was used to perform all phylogenetic analyses. In COI and the concatenated dataset, PartionFinder2 (Lanfear et al., 2017) was applied to find the best partitioning scheme and evolutionary model under greedy algorithm. Sequences of 18S rRNA gene were aligned with MAFFT (Katoh & Standley, 2013) using parameters '–auto' strategy and normal alignment mode. Gap sites were removed with trimAl (Capella‐Gutiérrez et al., 2009) using automated1 command. Maximum Likelihood (ML) phylogenies were inferred using IQ-tree (Nguyen et al., 2015) and Bayesian inference (BI) phylogenies using MrBayes 3.2.6 (Ronquist et al., 2012), both programs provided on the platform PhyloSuite. ML was performed under edge-linked partition model in COI and HKY + I + F model in 18S rRNA for 5000 ultrafast bootstraps (Minh et al., 2013), as well as the Shimodaira–Hasegawa–like approximate likelihood-ratio test (Guindon et al., 2010). BI phylogenies were inferred under the same models as ML, performing two parallel runs with four chains, each for 20 million generations sampling every 2,500 generation and discarding the first 25% as burn-in.

Dataset for COI haplotype networks contained only Alismobates specimens. The TCS networks (Clement et al., 2002) were reconstructed using PopArt (Leigh & Bryant, 2015).

For species delimitation analyses (SDA) based on COI data, we employed two methods: the “Assemble Species by Automatic Partitioning” (ASAP) method (Puillandre et al., 2021) and the ML partition “Bayesian Poison Tree Process “ (bPTP-ML) model (Zhang et al., 2013). The applied settings and programs/packages for the SDA analyses followed Schäffer and Koblmüller (2020), except for ASAP applying default parameters and uncorrected p-distances. To calculate Rodrigo´s P(Randomly Distinct) (P(RD)) measures, we used the species delimitation plugin (Masters et al., 2011) in Geneious Prime 2023.1.2 (https://www.geneious.com).

To assess correlation between geographical and genetic distances, a Mantel test with 10,000 permutations was performed using the program Alleles In Space (AIS) (Miller, 2005). In addition, we analyzed the genetic landscape shape, which allows the visualization of genetic distance patterns across the sampling area (Miller, 2005). The three-dimensional plots produced by AIS were illustrated from different angles.

Nucleotide diagnosis for the herein investigated Alismobates species was applied using the package Spider (Brown et al., 2012) in R (R Core Team, 2022). We used same alignments of the respective fragment as in the analyses above, with the exception that the Fortuynia species (outgroup) were excluded.

Morphometric study

For the morphometric study, specimens were placed in lactic acid (temporary slides) and measurements were performed using a compound light microscope (Olympus BH-2) and ocular micrometer. Individuals used for morphometric investigations were not the same as used for molecular genetic studies but belonged to the exact same populations (patch of algae ca. 10 cm2).

A set of 16 continuous variables (Pfingstl & Baumann, 2017; Fig. 1a, b, p.118) was measured in 271 specimens of Alismobates from nine different Caribbean regions; Bermuda: 36 specimens (BD_07, BD_21); Bahamas: 76 specimens (BH_03, BH_08, BH_16, BH_19, BH_21, BH_25); Barbados: 19 specimens (BA_15, BA_22, BA_30); Costa Rica: 20 specimens (CR_01); Curaçao: 14 specimens (CU_15); Dominican Republic: 18 specimens (DR_03, DR_10); Florida: 44 specimens (FL_12, FL_16, FL_19); Guadeloupe: 5 specimens (GU_09); Panama: 39 specimens (PA_39, PA_41). [Bermuda is part of the Western Atlantic and not the Caribbean, but for reasons of simplicity it is listed here and in the following parts under Caribbean or Northern Caribbean.]

Fig. 1
figure 1

Maximum likelihood tree (IQ-tree) inferred from a concatenated dataset of COI and 18S rRNA gene fragments. Numbers at nodes represent Bayesian posterior probability values (shown > 0.85) and bootstrap values (shown > 80) for ML. Sequences marked with an * were taken from GenBank

For univariate comparisons, mean, standard deviation, minimum and maximum of each variable were calculated. To compare the means of the variables between species a Kruskal–Wallis test was performed and for comparison between populations within each species a Kruskal–Wallis and a Mann–Whitney-U test were conducted.

Multivariate analyses investigating differences between putative species and geographic groups respectively, included a Principal Component Analysis (PCA), Non-metric Multidimensional Scaling (NMDS; based on Euclidian distances, two-dimensional) and Discriminant Analysis (LDA); all analyses were performed on log10-transformed raw and size-corrected data. No rotation was applied to the multivariate data. Size correction was done by dividing each variable through the geometric mean of the respective specimen. For species discrimination all populations from a species were pooled, and populations of A. inexpectatus were pooled according to specific geographic areas for assessing intraspecific variation (e.g. populations from Panama and Costa Rica were pooled under ‘Central America’). All analyses were performed with PAST version 3.11 (Hammer et al., 2001).

Drawings and photographs

Specimens were embedded in Berlese mountant for microscopic investigation in transmitted light and these preparations were studied and depicted using an Olympus BH-2 Microscope. Drawings were first scanned, then processed and digitized with the free and open-source vector graphics editor Inkscape (https://inkscape.org).

For photographic documentation, specimens were air-dried and photographed using a Keyence VHX-5000 digital microscope with automated image stacking.

Results

Molecular genetic results

Molecular genetic analyses of mitochondrial COI and nuclear 18S rRNA gene sequences clearly highlighted the presence of two different species. Uncorrected p-distances between these two species were 14.8% in the COI gene and 0.2% in the 18S rRNA gene. Intraspecific mean distances ranged from 2.7 to 5.3% in COI (Table 2) and were 0% in the 18S rRNA. Thus, a barcoding gap was evident in both markers. Furthermore, both species clustered in highly supported, distinct clades in the ML and BI phylogenies of the single gene analyses, as well as the concatenated dataset (Fig. 1 and Suppl. Figures). Single-locus delimitation based on COI revealed 12 to 15 putative species (excluding outgroups): ASAP shows the best support for a total of 12 putative Alismobates species with an ASAP-score of 2.00. The partition with only two Caribbean Alismobates species results in a score of 11.50 and thus significantly shows weaker support. A similar picture is given by the bPTP model with 15 well to highly supported, putative Caribbean Alismobates species (Suppl. Fig. S5).

Table 2 Mean uncorrected pairwise distances (1) between and (2) within Alismobates (A.) and Fortuynia (F.) species. Values for COI gene are given in the left lower corner and for 18S rRNA in the right upper corner. Standard deviation values shown in brackets. n/c = not possible to estimate evolutionary distances

According to the presented results, the newly discovered lineage was given species rank and will be referred to as Alismobates piratus sp. n. The calculated Randomly Distinct PRD value for Alismobates piratus sp.n. was 0.05 and is thus inconclusive; the value for A. inexpectatus was 0.09 and indicates that its structure follows the coalescent model and does not contain any further cryptic species.

The COI haplotype diversity in the new Alismobates piratus sp. n. was relatively low with only four different haplotypes which, however, were distinctly separated from each other (Fig. 2).

Fig. 2
figure 2

Map showing records of the two Alismobates species in the Caribbean and TCS haplotype network based on their COI sequences. Light red arrows on map indicate ocean currents and their directions. Each circle in the network corresponds to one haplotype and its size is proportional to its frequency, the number of mutations is indicated as hatch marks. Small black circles represent intermediate haplotypes not present in the dataset. Colors refer to different sample locations grey shades represent the different species

Haplotype diversity was higher in Alismobates inexpectatus with a total of 53 haplotypes. Specimens from Central America and the Antilles showed six and five haplotypes, respectively, with distinct structuring, whereas the Northern Caribbean group was with 42 haplotypes the most diverse one (Fig. 2). In the case of individuals from Bahamas, respectively, Bermuda, the haplotype networks revealed no geographic sub-structuring (Fig. 3).

Fig. 3
figure 3

Haplotype networks based on COI sequences of A. inexpectatus populations from the Bahamas and Bermuda. Each circle corresponds to one haplotype and its size is proportional to its frequency. The number of mutations is indicated as hatch marks. Small black circles represent intermediate haplotypes not present in the dataset. Colors refer to different locations as indicated on the respective map. a) New Providence Island, Bahamas. b) Bermuda

Genetic landscape interpolation analysis at individual level of both species qualitatively supported the results obtained from the haplotype networks. It showed basically lower genetic distances in the Northern Caribbean and Central America suggesting more connectivity between these two large areas (Fig. 4). High genetic differentiation among individuals became evident in the Western Atlantic, the Caribbean and Colombian Basin as well as in Hispaniola and Barbados.

Fig. 4
figure 4

Multidimensional graph produced by the genetic landscape shape interpolation analysis, shown from different angles. Dark blue peaks represent areas with high genetic discontinuities and yellow valleys indicate low genetic distances of individuals across the distribution of Alismobates species in the Caribbean. Most important peaks and valleys are indicated on the map below

Results of the Mantel test calculated in AIS revealed a low but statistically significant correlation between geographical and genetic distances (r = 0.53, P < 0.001) indicating that genetic distances can be explained in part by geographic distances (Fig. 5).

Fig. 5
figure 5

Mantel test for COI data of Caribbean Alismobates species with correlation coefficient (r), showing the correlation of genetic and geographic distances

Description of new species

Family Fortuyniidae Hammen, 1963.

Genus Alismobates Luxton, 1992.

Alismobates piratus sp. nov. urn:lsid:zoobank.org:act: 35A22D71-5078-4D1A-AA61-A6DCD5A8E7E5.

Type material: Holotype male (size 356 µm x 240 µm), Curaçao: Boca Ascención (CU_15) Bostrychia on rocks in the upper eulittoral area, 5 Feb. 2016; paratype (1 female 375 µm x 249 µm) same data as for holotype, both types deposited in the Senckenberg Museum für Naturkunde Görlitz (SMNG).

Etymology: The species occurs in the Caribbean area which is well known for its long historical era of piracy. The specific epithet ‘piratus’ refers to the Latin word for pirate (‘pirata’) and is given as noun in apposition.

Species diagnosis: Dark brown sclerotized mites. Average length 362 μm, mean width 233 μm. Notogaster oval in shape. No conspicuous sexual dimorphism; there is only the common sexual dimorphism in overall body size, with females being generally slightly larger. Van der Hammen’s Organ well developed, typical for the genus. Sensillum clavate, spinose. One pair of large cuticular ridges in position of prodorsal lamellae. Interlamellar setae minute. Lenticulus (light spot) large, variable in shape and with irregular borders. Areas flanking lenticulus conspicuously granular. Fourteen pairs of short and simple, notogastral setae, associated with small porose areas. Epimeral setation 3–1-2–2. Four pairs of genital setae. One pair of aggenital setae. Three pairs of adanal setae. Two pairs of anal setae. Legs monodactylous with large claw. Porose areas on trochanters III and IV and all femora. Leg setation (chaetome, solenidia): Leg I 0–4-2–3-18, 1–2-2; leg II 0–4-2–3-15, 1–1-1; leg III 1–3-1–3-15, 1–1-0; leg IV 1–2-2–3-12, 0–1-0.

Diagnostic nucleotides (only unique diagnostic characters are given): In COI, position 7 is occupied by base G, position 17 by base T, position 19 by base A, position 73 by base T, position 82 by base C, position 133 by base T, position 199 by base T, position 340 by base A, position 346 by base A, position 391 by base G, position 409 by base C, position 538 by base C, position 551 by base G, position 574 by base C, position 608 by base C and position 613 by base T.

In 18S rRNA, position 728 is occupied by base G, position 768 by base A and position 1350 by base A.

Distribution: Presently, Alismobates piratus sp. n. is known to show a distribution range from Hispanola to the Lesser Antilles. There are reports from the Lesser Antillean islands of Guadeloupe, Barbados and Curaçao, and there is a record from the northeastern coast of the Dominican Republic, namely from shores of the Samaná Peninsula (see Fig. 2). Distributions on Barbados and the Dominican Republic coincide with occurrences of A. inexpectatus whereas specimens of both species were found syntopically (in a single sample of ca. 10cm2 algae) on the coast of Samaná.

Remarks: The new species A. piratus sp. nov. is morphologically identical to A. inexpectatus (Fig. 6) and thus cannot be distinguished from it based on microscopic investigation only.

Fig. 6
figure 6

Dorsal (left) and ventral (right) depictions (legs omitted or only partially drawn) of the two genetically distinct Alismobates species, illustrating the phenotypic similarity. a) A. piratus sp. n. (specimen from Barbados BA_22); b) A. inexpectatus (specimen from the Dominican Republic DR_03)

The lengths and shapes of certain characters, i.e. sensillum and body setae, may look slightly different in some of our depictions, but these are results of slightly different observational perspectives or of different orientations of these characters. These structures are bendable and they are inserted on a convex surface, therefore a slightly different viewing angle may result in a divergent appearance. For these reasons, it is also extremely difficult to make standardized measurements of these traits. However, we observed more than 270 specimens under the microscope and could not find any reliable diagnostic character separating the two species (to the best of our knowledge), therefore, we herein provide the species diagnosis only, which is also valid for A. inexpectatus; a detailed description applying to both species is given in Pfingstl and Schuster (2012). Due to the identical phenotypes, species determination should always include the analysis of molecular genetic markers.

Morphometric results

Interspecific variation

Comparing the two species with univariate statistics, size ranges show overlaps in each variable but mean values are generally higher in Alismobates inexpectatus indicating a trend towards a larger body size in this species. Kruskal–Wallis test found significant differences in all measured morphological variables except for the anterior notogastral width nwc1 (Table 3).

Table 3 Univariate statistics and comparison of the two Caribbean Alismobates species. Minimum–maximum (mean ± standard deviation) of each measured variable given in µm. Results of Kruskal–Wallis Test (KW) are given; * = p < 0.05, ** = p < 0.01, *** = p < 0.001

Multivariate statistics – PCA based on raw data results in a clustering of both species with a small overlap, with the first three components accounting for 79.59% of total variation (PC1 53.52, PC2 19.97, PC3 6.1). Highest loadings on PC1 are shown by characters of the genital opening, namely gl and gw. PCA on size corrected data resulted in a clustering of both species with a considerable overlap and the first three components accounting for 66.48% of total variation (PC1 32.69, PC2 21.71, PC3 12.08). Highest loadings are shown by the variables ll, nwc1 and nwdm.

NMDS performed on raw and size corrected data resulted in two clusters showing a partial overlap (Fig. 7). Both clusters could be distinguished with a stress of 0.3759 and 0.4273 respectively. Linear discriminant analysis (LDA) shows a clearer separation between both species and could correctly classify 96.3% of the specimens using raw and size corrected data.

Fig. 7
figure 7

Graph showing results of Non-metric Multidimensional Scaling (NMDS) performed on raw and size-corrected morphometric data of the two Alismobates species

Intraspecific variation

Univariate comparison of populations of A. inexpectatus from the Northern Caribbean, Central America and the Antilles revealed highly significant differences in 10 out of 16 variables at least between two of the populations (Table 4). Only two characters showed no significant differences at all among all populations and these were the anterior notogastral width nwc1, and the distance between camerostome dcg. The Northern Caribbean population diverges the most and is by trend the largest, showing the highest mean values for nearly each variable.

Table 4 Univariate statistics for three geographic groups of Alismobates inexpectatus (Northern = Bermuda, Florida and Bahamas; Central America = Costa Rica and Panama; Antilles = Barbados and Dominican Republic). Range (mean ± standard deviation) of each measured variable given in µm. KW – Kruskal–Wallis Test, *0.01 < p < 0.05, **0.001 < p < 0.01, ***p < 0.001; MWU – Mann–Whitney U test,—= no significant difference, letter indicates significant difference, a = Northern Caribbean vs. Central America, b = Northern Caribbean vs. Antilles, c = Central America vs. Antilles

Variation within Alismobates piratus sp. n. is basically lower, there are only two variables, the body length bl and the anal length al, showing highly significant differences among the populations from the Dominican Republic, Curaçao, Guadeloupe and Barbados (Table 5). There are four other variables showing significantly different values but nine out of 16 variables show no significant deviations at all. Mann–Whitney-U test shows that the population from Curaçao is responsible for most of the found significant differences.

Table 5 Univariate statistics for four populations of Alismobates piratus sp. n. Range (mean ± standard deviation) of each measured variable given in µm. KW – Kruskal–Wallis Test, *0.01 < p < 0.05, **0.001 < p < 0.01, ***p < 0.001; MWU – Mann–Whitney U test,—= no significant difference, letter indicates significant difference, a = Dominican Rep. vs. Curaçao, b = Dominican Rep. vs. Guadeloupe, c = Dominican Rep. vs. Barbados, d = Curaçao vs. Guadeloupe, e = Curaçao vs. Barbados, f = Guadeloupe vs. Barbados

Multivariate analyses of A. inexpectatus populations resulted generally in large overlaps but with a distinct trend for diversification. NMDS on size corrected data resulted in three overlapping clusters (Fig. 8) that could be distinguished with a stress of 0.37. LDA showed better results with fewer overlaps and with 83.19% correctly classified specimens using size corrected data. The Northern Caribbean population shows the largest cluster with most variation in both analyses.

Fig. 8
figure 8

Graph highlighting results of Non-metric Multidimensional Scaling (NMDS) and Linear Discriminant Analysis (LDA) performed on size corrected data of seven Caribbean Alismobates inexpectatus populations. Slightly coloured and framed areas refer to geographic regions: Northern Caribbean = Bermuda, Florida, Bahamas; Antilles = Barbados, Dominican Rep.; Central America = Costa Rica, Panama

NMDS performed on size corrected data of A. piratus sp. n. resulted in four clusters, with the population from Curaçao showing only minor overlaps, whereas the other populations from the Dominican Republic, Guadeloupe and Barbados strongly overlap (Fig. 9). Populations could be separated with a stress of 1.9. When these populations are subjected to LDA, populations from Guadeloupe and Curaçao form separate clusters and the Dominican Republic and Barbados still show slight overlaps. 93.2% of the specimens were correctly classified using size corrected data.

Fig. 9
figure 9

Graph showing results of Non-metric Multidimensional Scaling (NMDS) and Linear Discriminant Analysis (LDA) performed on size corrected data of four Caribbean Alismobates piratus sp. n. populations

Discussion

Cryptic diversity

All collected Alismobates specimens from across the Caribbean show conformity in their morphology and though morphometric data indicates certain differentiation in size and shape, all measured variables show overlaps in their ranges leaving no distinct trait for separation of any group. Molecular genetic data, on the other hand, show distinct structures and strong diversification. Species delimitation analyses support the existence of more than 10 different Caribbean Alismobates species, but tree based methods are known to often oversplit species (e.g. Dellicour & Flot, 2018). The present dataset is highly structured due to the geographic setting with many isolated islands and species delimitation approaches apparently identify somewhat divergent singletons as distinct species. It is reasonable not to assume so many species. Divergence on the COI and 18S rRNA gene fragments clearly exceeds intraspecific variation for at least two distinct lineages and thus reveal the existence of just a single genetically distinct species, namely Alismobates piratus sp. n. The contrast between morphology and genetic sequences renders A. piratus sp. n. a so called ‘cryptic species’, which is defined as a species that is difficult to distinguish with traditional morphology-based taxonomic methods (Knowlton, 1993). Despite the identical phenotype, the present results clearly show that A. inexpectatus and A. piratus sp. n. are separately evolving metapopulation lineages, which makes them species in the sense of the unified species concept (De Queiroz, 2007). Moreover, individuals of both supposed species were found syntopically at the exact same location and did not show any signs of hybridization, consequently they show reproductive isolation and thus are also species according to the biological species concept (Mayr, 1940) (Table 6).

So why do both species still show the same morphology? Extreme and homogeneous environments may cause stabilizing selection that reduces morphological change usually accompanying speciation (Bickford et al., 2007). The intertidal zone represents such an extreme environment and thus may be responsible for the observed morphological stasis. Cryptic diversity was also reported in several other Caribbean intertidal oribatid mite taxa, i.e. Carinozetes, Fortuynia and Thalassozetes, (Pfingstl et al., 2019a, b, 2021, 2022) which confirms a correlation between the intertidal environment and morphological stasis.

However, the reasons for cryptic speciation in these Caribbean intertidal oribatid mite species may differ. In the cryptically diverse Carinozetes, neither the found biogeographic pattern nor the ecological needs could explain the genetic diversification of the lineages (Pfingstl et al., 2019a, b), and thus the cause for speciation remains unknown in this case. In the cryptic Thalassozetes, on the other hand, vicariance is most likely the reason, because nearly all species are clear island endemics showing very poor dispersal ability (Pfingstl et al., 2021). Vicariance is also supposed to be responsible for the diversification of the northern Caribbean Fortuynia atlantica and the Lesser Antillean F. antillea Pfingstl et al., 2022 (Pfingstl et al., 2022). The present Alismobates species show a somewhat similar pattern with A. inexpectatus being more distributed in the northern and western parts of the Caribbean, and A. piratus sp. n. being restricted to Hispaniola and the Lesser Antilles, whereas distributional ranges are overlapping in the latter areas. At some point in geological history, there may have been a biogeographic barrier between more northern and Antillean ancestral populations of Caribbean Fortuynia and Alismobates, which caused allopatric speciation. Sometime later the barrier might have vanished and species expanded distribution ranges resulting in the observed sympatric patterns, at least in Alismobates. However, based on the present results, we are not able to allocate this suggested geographic barrier to any geological event or formation and further studies are needed to identify the reason for the speciation of the Caribbean Alismobates species.

Phylo-and biogeographic patterns

Haplotype network analysis of mitochondrial COI sequences show that Alismobates piratus sp. n. populations from different Antillean islands exhibit a distinct pattern with strong differentiation among them. Accordingly, there was no recent gene flow between populations and they have been separated for a long time. Most of the Lesser Antilles are relatively recent volcanic islands with an age of less than 10 million years (Iturralde-Vinent & MacPhee, 1999), the oldest rock of Guadeloupe, for example, is dated 4.7 million years (Maury et al., 1990). Presently, there is no reliable substitution rate inferred for the COI gene of mites, but using the general arthropod substitution rate of 1–1.15%/my (DeSalle et al., 1987) would result in a separation of approx. 2.7–3 million years among the A. piratus sp. n. populations. This date would coincide with the emergence of at least some of the Lesser Antilles, but as said before, the substitution rate is not reliable and thus we cannot assign the diversification of the populations to a certain time period. However, it is assumable that the A. piratus sp. n. populations from the Lesser Antilles are derived from stocks in Hispaniola because this Greater Antillean Island is much older with an approximate age of 40 million years (Iturralde-Vinent & MacPhee, 1999) and there is a recent population present on this landmass. Certain arachnids show strong diversification within the islands of Jamaica and Hispaniola and basically a higher diversity in the Greater Antilles relative to the Lesser Antilles (Crews & Gillespie, 2010), indicating that these older landmasses are centers of species diversity within the Caribbean. This could also apply to Alismobates with the ancestral species having diversified on the Greater Antilles and descendants having subsequently colonized newly emerged islands. Nevertheless, further comprehensive sampling and studies, including more islands of the Greater and Lesser Antilles, are necessary to verify this assumption.

The A. inexpectatus populations, on the other hand, show varying haplotype patterns in different Caribbean regions. First, the populations from the Dominican Republic/Hispaniola and Barbados exhibit large divergence, which is well in accordance with the pattern found in A. piratus sp. n. populations from the same areas. Second, the Central American populations show deep splits and distinct structures although all individuals were collected on the Bocas del Toro archipelago in a range of less than 30 km. The Central American Landmass has been subject to tremendous changes during the last 40 million years and the Panamanian Isthmus was finally completed about 2.5–2.3 million years ago (e.g. Iturralde-Vinent, 2006). The latter event led to enormous environmental changes, which may have caused the separation of A. inexpectatus populations in this area. Moreover, the closure of the Central American land bridge caused a deflection of the circumtropical current and gave birth to the Caribbean current, flowing along the South and Central American coastline (e.g. Iturralde-Vinent & MacPhee, 1999). Intertidal mites are supposed to be dispersed over long distances by drifting on ocean currents (Pfingstl, 2013; Schatz, 1991), consequently populations from more eastern South American coastal areas may have occasionally colonized the Bocas del Toro archipelago via the Caribbean current. Thus, the distinct heterogenous genetic structure found in Panamanian A. inexpectatus populations may be a product of the dynamic geological and environmental history of the Central American Isthmus during the Pliocene. Third, northern Caribbean populations show less distinct structuring in the COI gene sequences and haplotype network analyses indicate a common origin and subsequent expansion events on the Bahamas and the Bermudas. The Bahamas platform has been a stable carbonate block since the Cretaceous but emerging islands have only been created by accumulation and sedimentation during the Pleisto- and Holocene (Carew & Mylroie, 1997). It seems that New Providence Island has been colonized by a few individuals sometime during this period and then populations have expanded and slowly diversified over the island. A single haplotype is shared by northeastern and southeastern populations indicating that initial colonization may have happened in the eastern parts of this landmass. A similar haplotype pattern is found in Bermudian populations. Bermuda is a small remote oceanic landmass that emerged from the sea ca. 900,000 years ago (Thomas, 2004). The present data indicates that it was colonized early in this period and then populations slowly diversified over the island, whereas a single haplotype is still found in four out of the five sampled locations. Another interesting pattern is found in the specimens from the Florida Keys; there are two groups of haplotypes that are only distantly related, which indicates that one of the groups most likely originates from a different area in the Caribbean.

The genetic landscape analysis of all Caribbean Alismobates shows high divergence in the Western Atlantic and parts of the Caribbean Sea, like the Colombia and the Caribbean Basin, and less genetic divergence in the Western and northern Caribbean areas (see Fig. 4). The former areas represent vast stretches of open ocean and it makes absolute sense that these areas are rather effective barriers for the dispersal of tiny mites. The latter areas, on the other hand, mostly represent continuous continental coastlines which may allow gene flow along the coast, at least over short distances. However, there is another important factor that seems to strongly contribute to the observed pattern, namely the Gulf Stream. This strong current flows along the Central American coast, passes Florida and the Bahamas and finally runs past Bermuda (see Fig. 2). Considering long distance transport of mites drifting on the ocean’s surface, the Gulf Stream could explain the strong connectivity between Central America and the Northern Caribbean, as well as between the Northern Caribbean and the far remote landmass of Bermuda. A strong link and gene flow between Central America and North America was also shown in the intertidal oribatid mite Thalassozetes balboa Pfingstl et al., 2019a, b, because populations from Panama and Florida share several haplotypes (Pfingstl et al., 2021). Other studies (Pfingstl, 2013; Schatz & Schuster, 2012) already hypothesized that the Gulf Stream may be responsible for the colonization of Bermuda by several oribatid mites from the Northern Caribbean, and the recent study on the Caribbean intertidal mite F. atlantica (Pfingstl et al., 2022) supported this theory for the first time with molecular genetic data. The present results further corroborate a link between the Northern Caribbean and Bermuda, and highlight the important role of the Gulf Stream for the dispersal of intertidal oribatid mites.