Collaria

Castro-Huertas, Valentina, Morales, Irina & Forero, Dimitri, 2025, Is the grass greener on the other side? Phylogeny and biogeography of grass-feeding plant bugs of the genus Collaria Provancher (Hemiptera: Miridae: Mirinae), Journal of Natural History 59 (5 - 8), pp. 511-542 : 530-536

Collaria group and the monophyly of Collaria

Schwartz (2008), in his phylogenetic hypotheses for the tribe Stenodemini , proposed the Collaria group as composed of Collaria and Nabidomiris , based on the following non-homoplastic characters: eye ovoid and located in the medial portion of the head; male genitalia with dorsal and ventral ornamentation on the secondary gonopore; membrane of the vesica with a small spinose or filamentous sclerite basal to the secondary gonopore; and female genitalia with the dorsal surface of the posterior wall with a large process. In addition, two homoplastic characters also supported the Collaria group: the rounded humeral angle of the pronotum, and the proepisternum strongly rounded. In our study, we recovered a sister-group relationship between Collaria + Nabidomiris supported by two of the characters originally proposed by Schwartz (2008): the postocular region elongated and the basal region of the endosoma with a ribbon-like sclerite. The eye located on the medial portion of the head is used by Schwartz (2008) as one non-homoplastic character to support the Collaria group clade; however, the eye position corresponds to the relative anteocular and postocular lengths, and not to the eye position in the head per se. Therefore, we coded the anteocular and postocular structure as separate characters. Additionally, we recovered in our analysis that the dorsal surface of the pronotum clearly divided into lobes supports the clade Collaria + Nabidomiris .

Schwartz’s hypothesis (2008) used genera as terminals, assuming a priori their monophyly, and thus disregarding species in the analyses. Nine species of Collaria were analysed by Schwartz (2008): C. guaraniana , C. improvisa , C. meilleurii , C. obscuricornis , C. oculata , C. oleosa , C. scenica and two undescribed species, all coded as a single terminal, Collaria . Considering the diagnosis of Collaria offered by Schwartz (2008), our analysis agrees with one synapomorphy: the mandibular plates produced. Additionally, the generic diagnosis of Collaria did not consider the homoplastic characters recovered in his analyses: the setae on the first and second antennal segments long and erect (character 14–1; Schwartz 2008), and the metatibia with long erect setae (character 18–1; Schwartz 2008). Having erect setae on the first antennal segment was recovered as a synapomorphy for Collaria in our analyses.

Collaria species groupings

Missing data had an important effect on the species grouping hypothesis within Collaria . In our analysis of the complete matrix, the recently described species C. boulardi ( Brazil), C. elsae ( Madagascar), and C. minuta ( French Guiana) were recovered as sister species to the remaining species of the genus, whereas C. danae ( Ivory Coast) was recovered as part of the otherwise Neotropical species clade. The phylogenetic position of C. boulardi , C. danae , C. elsae , and C. minuta derived from the complete matrix is doubtful. We did not examine specimens of these species, and because they are known only from females, the percentage of missing data for these species in our complete matrix surpasses 60%. Additional material for these species could contribute to more robust phylogenetic and biogeographic hypotheses.

In our analyses with the reduced matrix, Collaria is composed of two main clades that correspond to the biogeographic distribution of the species: an Afrotropical species clade and a clade of Nearctic species + C. royi with a Neotropical species clade ( Figure 7 , clades C and E + F, respectively).

The Afrotropical clade (clade E) is supported by genital characters, as well as most of the species groupings. Similarly, the Neotropical species clade is supported by genitalic characters, but the groupings within the clade are weakly supported, except for the clade formed by C. scenica and C. oleosa . The genitalia in Collaria are asymmetrical and include various sclerites and processes on the male endosoma and the female posterior wall. Thus, genitalic structures were informative in our analyses, supporting several clades.

The Nearctic species + C. royi were recovered as the sister group of the Neotropical species, supported by two homoplastic characters. Collaria meilleurii and C. oculata are similar in endosomal structure in having a long sclerite and the medial left sclerite with a semicircular shape; whereas they are distinguished by the colouration patterns and the shape of the ventral right sclerite of the endosoma ( Morales et al. 2016). Colour patterns are not included as evidence in our analyses, because Collaria species are usually entirely brown or pale brown with some interspecific variation.

Biogeography of Collaria

Attempts to explain species distribution require an integrated approach, because the processes behind the distribution patterns involve diverse aspects of historical biogeography, ecology, and even biology and behaviour ( Wiens and Donoghue 2004). Despite several limitations hindering our understanding of the biogeographic processes that have resulted in the observed distributional patterns of Collaria species, the results of our biogeographic analyses provide a set of hypotheses that could be further tested with additional data. Our results on the biogeography of Collaria show a mix of geographically explicit events among the Afrotropical, Nearctic, and Neotropical areas, including vicariance, sympatry, and founder events.

The disjunct distribution between the Afrotropical and the Nearctic (except C. royi ) + Neotropical clades was inferred as a vicariant event in our analysis ( Figure 8 , node 1). This pattern needs alternative explanations besides being of Gondwanan origin, considering the age of diversification for plant bugs (159.7–167.5 MYA, Johnson et al. 2018; Oh et al. 2023), and of Stenodemini in particular, which ranges between 75 and 80 MYA ( Oh et al. 2023). Gondwana started breaking up during the late Jurassic, between 150 and 165 MYA, but South America and Africa were still connected until 95–110 MYA, when northern South America and Africa started separating ( Sanmartín and Ronquist 2004), much earlier than the origin of stenodemines. The emergence of several lineages of stenodemines, including Collaria , might thus have happened in the early Palaeogene or late Cretaceous, between 55 and 80 MYA ( Oh et al. 2023), coinciding with the diversification of grasses ( Bouchenak-Khelladi et al. 2010), and after Africa and South America had already drifted apart.

A transoceanic disjunction pattern is found in several insect groups (eg Ye et al. 2017; Bourguignon et al. 2018). Traditionally, the tectonically driven vicariance of Gondwanan origin has been the principal mechanism to explain the distribution of organisms with transoceanic disjunctions ( Amorim et al. 2009). However, the inclusion of dating analysis in phylogenetic approaches has allowed testing this scenario and providing alternative mechanisms at the same time. One of these is long-distance dispersal, which has arisen as a possible hypothesis when there are inconsistencies between the divergent times of the biological groups and the estimated timing of Gondwana break-up ( Givnish et al. 2004). Usually, a long-distance dispersal mechanism is commonly invoked in many grass-feeding insects; for example, Satyrini butterflies ( Nymphalidae : Satyrinae ) ( Peña et al. 2011), Spodoptera moths ( Lepidoptera : Noctuidae ) ( Kergoat et al. 2012), and grassland leafhoppers of the tribe Chiasmini ( Hemiptera : Cicadellidae : Deltocephalinae ) ( Zahniser and Dietrich 2015). Nonetheless, 50 MYA the distance between Africa and South America was about 1000 km on their closest path ( Oliveira et al. 2010), making a long-distance dispersal mechanism challenging.

Even though mirids are usually regarded as poor dispersers at long distances because of their small size and limited flight power ( Schuh and Stonedahl 1986), various data might suggest otherwise. Miridae are actually good flyers, better than other heteropterans studied ( Southwood 1960). For instance, Creontiades dilutus (Stål) can migrate over long distances over land, up to 1500 km, and can potentially disperse across a range of 5000 km ( Hereward et al. 2013); this species can also disperse from mainland Australia to Tasmania ( Hill 2017). It also has been noted that mirids can be found at sea far from land (about 600 km) ( Wheeler 2001) which is congruent with data on the colonisation of oceanic islands by mirids (eg Leston 1957; Asquith 1997). These dispersal abilities probably can be explained by a combination of favourable wind conditions and flight abilities ( Wheeler 2001).

An alternative to the long-distance dispersal mechanism to explain the observed vicariant pattern of species of Collaria , distributed between the Neotropical and Nearctic + Afrotropical clades, is geodispersal through a series of large dry land extensions found in the South Atlantic at 50 MYA (Brigs 2003; Oliveira et al. 2010; Ezcurra and Agnolín 2012). Between 80 and 50 MYA, there was a series of large islands that connected Africa and South America in the South Atlantic ( Oliveira et al. 2010). Evidence of a South Atlantic dispersal pattern has been shown by some groups of plants and vertebrates (eg Briggs 2003; Renner 2004). Predominant palaeocurrents and palaeowinds from Africa to South America ( Oliveira et al. 2010) lead us to favour the idea of a geodispersal of Collaria from Africa to South America through this island chain between 40 and 50 MYA. These islands were non-existent by 20 MYA ( Oliveira et al. 2010), severing the connection between the two continents, and, thus, probably creating the vicariant pattern observed between South America and Africa in Collaria .

The African clade was reconstructed as having three unambiguous vicariant events. The first one is between C. schwartzi , distributed in Central and Eastern Africa, and the remaining species found in Africa, which have a more south-eastern and Madagascan distribution ( Figure 8 , node 2). The observed pattern could have formed during the Miocene (8–23 MYA). Initially, the rainforest stretched from coast to coast, but around the mid Miocene (~17 MYA) the uplift of eastern Africa started to separate the eastern and western rainforests ( Dijkstra 2007; Wichura et al. 2010). The consequence was the expansion of the grass-dominated biome between the mid and late Miocene (8–16 MYA) ( Jacobs 2004; Utescher and Mosbrugger 2007), which could have had an impact on the diversification of grass-associated Collaria . During the Pliocene and early Pleistocene, a major uplift also created the Great Rift Valley and shaped the Congo Basin, which could have divided the continent into eastern and western regions ( Dijkstra 2007; Wichura et al. 2010). These complex expansions and contractions of grass-dominated biomes, coupled with changes in temperature ( Couvreur et al. 2021), might have affected the diversification of African Collaria species.

The second vicariant event was between C. nigra , distributed into East Africa, and the remaining African species, distributed in Madagascar, West Africa and Eastern Africa ( Figure 8 , node 4). This event could similarly be explained by the uplift of the East African plateau. The third vicariant event was between the African species distributed in the continent and species found in Madagascar, and the Réunion and Grande Comore Islands ( Figure 8 , node 5). Madagascar and Africa separated as part of the Gondwana break-up into eastern and western areas, about 155–170 MYA ( Masters et al. 2006). Réunion Island is located about 680 km east of Madagascar on the western Indian Ocean, and it is composed of three shield volcanoes ( Michon et al. 2007). The archipelagos around Madagascar vary widely in origin, age, and structure, although their formation could respond to Gondwana separating from Africa ( Ali and Aitchison 2008). Recent studies propose that faunal composition, high endemic biodiversity, and the presence of unique lineages in these islands could be the result of dispersal and in situ diversification processes ( Strijk et al. 2012). If we accept that the previous vicariant event in Africa probably occurred in the mid-Miocene, the present distributional pattern between Africa and Madagascar and related areas is likely not the product of a vicariant event, but could instead correspond to more recent dispersion events.

The event between the North American and the African species C. royi in our analysis was reconstructed as a founder event ( Figure 9 , node 14). Several Stenodemini lineages probably originated between the latest Cretaceous and the early Palaeogene, from 55 to 80 MYA ( Oh et al. 2023). A biogeographic relationship between these two areas has rarely been documented or explained. Hembree (2006), in a biogeographic analysis of amphisbaenians, showed with a molecular data set that African areas had complex relationships with both Asia and North America, rather than with South America. In his analysis, North America was closely related to both Western Asia and Northern Africa, a relationship which was not satisfactorily explained. Certain Scarabaeinae dung beetle taxa have been shown to have a North American–African relationship (Phillips 2016). Africa was connected to Eurasia during the Miocene ( Steinthorsdottir et al. 2021), and an exchange of fauna with North America was hypothesised to have occurred from the Miocene up to the Pleistocene via tectonic or sea level changes ( Davis et al. 2002). Boreotropical migration has been invoked to explain the current distribution of certain groups of plants ( Smedmark and Anderberg 2007; Wei et al. 2015; Huang et al. 2019) and animals ( Brunke et al. 2017; Ye et al. 2017), and thus could be an explanation for the pattern of Collaria species distribution between North America and Africa. This would imply that the African and North American Collaria clade diversified during and after the Miocene. Alternatively, because the phylogenetic support for the relationship between C. royi and the Nearctic clade is weak ( Figure 7 ), the biogeographic pattern might be an artefact and, thus, its interpretation remains tentative.

The Nearctic Region comprises cold temperate and arid subtropical areas, shaped by two main events: the glaciation dynamic and aridity during the Holocene ( Morrone 2019). The Nearctic species of Collaria occur in Eastern North America ( Figure 9 , node 15), which could be a response to glaciation dynamics during the Holocene, approximately 11,700 years ago. An Eastern North America distribution pattern is considered an endemic area, as shown by several biological groups ( Escalante et al. 2010; Weirauch et al. 2017). The north-eastern region in the United States was influenced and shaped by the advance and retreat of these glaciers, forming glacial lakes in low areas between or in front of glaciers, and shaped its current topography, drainage patterns, and organism distribution ( Sandford 1929). Collaria meilleuri is distributed in Canada, the north-eastern United States and Appalachian regions, and C. oculata mostly in the Southeastern Plains and Florida (Southeast Coastal Plain) ( Morales et al. 2016; Weirauch et al. 2017), with a broad overlapping area between the two, which could be a response to similar glaciation processes.

The Neotropical clade exhibits only one vicariant event between the north-western and the south-eastern species ( Figure 9 , node 10). Nonetheless, this distribution pattern is not always clear in the case of C. oleosa and C. scenica , in which both are distributed from the southern portion of the Chacoan subregion of the Atlantic Forest to the north of the Mexican transition zone, including the Antillean and Brazilian subregion ( Morrone 2015). The disjunction between the north-west and south-east areas could be explained by the origin of the Atlantic Forest with open vegetation formations since the Miocene and its complete isolation since at least the Pliocene. These events have been invoked to explain similar distribution patterns in other biological groups ( Sigrist and Carvalho 2009; Batalha-Filho et al. 2013; Silva and Noll 2014). The patterns of the clade containing C. boliviana , C. manoloi , C. oleosa , and C. scenica are mostly explained by sympatric events ( Figure 8 , nodes 5 and 6). Given these sympatric patterns, besides geographical factors, other biological aspects such as intraspecific biochemical recognition could be at play ( Byers et al. 2013; Sai et al. 2021). The founder event between C. husseyi and C. guaraniana ( Figure 8 , node 6) could be explained by the vegetation changes between the Araucaria Forest province and the Atlantic and Parana Forests regions.

Information on species distributions is scattered in natural history collections and various scientific publications, affecting biogeographical inferences through the incomplete taxonomic and distributional knowledge ( Arias et al. 2011; Kreft and Jetz 2010). Therefore, it is necessary to collect data and specimens from areas with an obvious lack of information, which may enable more precise identification of possible events and patterns they have generated ( Silva and Noll 2014). In the case of Collaria , the most obvious gaps are found in Africa ( Figures 8 and 9 ). Despite these shortcomings, we were able to propose biogeographic hypotheses within Collaria , which should be tested with additional data (eg molecular). New data, and refined phylogenetic hypotheses, could help date the nodes of the phylogeny, and thus help us test the critical biogeographical events identified here that shaped the evolution of Collaria .

Figures

Figure 7. Single most parsimonious tree obtained with the implied weighting (IW) analysis and using k = 30. Main apomorphies mapped on the tree, non-homoplasious changes as black polygons, homoplasious changes as white polygons. Support values in the branches are within square brackets (Bremer support, absolute symmetric resampling and “Group present/Contradicted” (GC) frequencies, respectively). Clades A to F. Geographic distribution: Nearctic region (blue), African region (marine blue), Neotropical region (green).

Figure 8. Optimal reconstructions found for the geographical data and the phylogenetic hypothesis, nodes 1–7. Symbols at the nodes of the phylogeny are as follows: black square, vicariance; white square, sympatry; white circle, point sympatry; and white triangle, founder event. In the maps, the ancestral range is drawn in green; the vicariance is drawn in red and blue squares to indicate the disjunction; regarding the founder event and point sympatry, the ‘founder’ and ‘point’ descendants are indicated by white squares.

Figure 9. Optimal reconstructions found for the geographical data and the phylogenetic hypothesis, nodes 8–15. Symbols at the nodes of the phylogeny are: black square, vicariance; white square, sympatry; white circle, point sympatry; and white triangle, founder event. In the maps, the ancestral range is drawn in green; the vicariance is drawn in red and blue squares to indicate the disjunction; regarding the founder event and point sympatry, the ‘founder’ and ‘point’ descendants are indicated by white squares.