Mackenziella psocoides was found in the shallow mattress of mosses (dominant Brachythecium albicans and intermixed Ceratodon purpureus) growing on a path of concrete slabs in a peri-urban context (Fig. 2A–C). The path is enclosed between a road and a concrete sidewalk, with only a thin border of ground covered with herbs. The mosses were sampled with its substrate, being a 1–2 cm deep layer of organic matter and soil particles mixed with coarse sand and small gravels. At the time of collection, the sample was humid, but unfrozen (sampling occurred 24 hours after the last frost event). The second collection occurred after one week of cold temperatures (two nights near 0°C, and up to 10°C), without precipitation. The day of collection itself was rainy and the new samples were well-moistened.

Mackenziella psocoides was very abundant in sample A (> 100 specimens), and in low abundances in sample B (4 specimens) and C (~ 30 specimens) (B and C each being roughly twice the size as sample A in terms of moss sampled). Even in sample A, M. psocoides was evidently the smallest contribution to the overall collembolan biomass.

Hypogastrura vernalis (Carl, 1901) was dominant in terms of relative abundance and biomass in all three samples. Hemisotoma thermophila (Axelson 1900) was rather common in all samples. Folsomides angularis (Axelson, 1905) was common in sample B but absent from A and C. Sminthurididae sp. (likely Sminthurides sp., only juveniles found, whitish, each with the tibiotarsal III organ) was rather common in sample A, and almost absent from sample B and C. Sminthurinus cf. elegans was found in low numbers in all samples. Sphaeridia pumilis (Krausbauer, 1898) was found only in sample C, in roughly the same numbers as M. psocoides. Lepidocyrtus lanuginosus (Gmelin 1790), Orchesella cincta (Linnaeus, 1758) and Agrenia sp. were found in moderate abundance (but high biomass) in sample A and C, but not B.

The three sequenced libraries for M. psocoides resulted in three identical sequences. Thus, a single sequence was used to represent the species in the phylogenetic tree. The recovered tree (Fig. 8) was compatible with the monophyly of all orders. Mackenziella psocoides was found to be the sister group of Sphaeridia pumilis, this node being the sister of another clade formed by Sminthurides aquaticus and Stenacidia violacea, all supported with high bootstrap support (>=99%) in the ML tree and with maximum posterior probability (1) in the Bayesian analyses. In the rest of the text, we refer to Sminthurididae as a monophyletic group, including M. psocoides. Within Symphypleona, the relations were as such: (Katiannidae, (Sminthuridae, Sminthurididae)) with a moderate support to the basal node (bootstrap = 84.4%).

Figure 8. 

Maximum clade credibility tree based on the trimmed alignment of the combined 18S rDNA and 28S rDNA. Node support values: posterior probability/nonparametric bootstrap (100 replicates)/aLRT (1000 replicates); blue bars indicate 95% HPD intervals of the age estimates, age estimate in red above the branches. Nodes with less than 75% bootstrap or 0.7 posterior probability support are collapsed. Habitus of Sphaeridia pumilis and M. psocoides represented next to the corresponding labels (M. psocoides drawing modified after Fjellberg 1989).

The mean crown age for Mackenziella + Sphaeridia was estimated at ~79 Ma (42–113 Ma). The crown age of Sminthurides + Stenacidia was estimated at 68.79 Ma (31–106), the crown age of Sminthurididae at ~126 Ma (113–158). The origin of the four Collembola orders seems to be rooted in the Paleozoic (or possibly Mesozoic considering the lower part of the range), with mean crown age of ~261 (169–363) Ma, ~285 (194–398) Ma, ~159 (100–309) Ma and ~336 (211–427) Ma for Entomobryomorpha, Symphypleona, Poduromorpha and Neelipleona respectively. These results have to be taken with caution, considering the restricted taxon sampling used in our analyses and, more importantly, the known collembolan fossil record being very scarce. The ML and Bayesian trees are independently shown in Fig. S1 and Fig. S2. The complete analysis folder is deposited on Zenodo (https://doi.org/10.5281/zenodo.8171774).

No evident morphological synapomorphies of Sphaeridia + Mackenziella could be found. Direct optimization of character states at the (Sphaeridia, Mackenziella) node results in: globular body shape, orthognathous head, simple MACO, mouth parts present, slender antennae, large clypeal area, complete mouth parts, eyes with eight ocelli, prothorax without prominent bulge, absence of vesicles on metathorax (male), long furca at least reaching the prothorax segment when folded under the body, long mucro with a pair of posterior lamellae, retinaculum with presence of chaetae (adult) and basal tubercles, five pairs of abdominal trichobothria, absence of Tibiotarsus III organ. Those ancestral character states are all unchanged in the genus Sphaeridia, and also apply to the Sminthurididae ancestor (but state of MACO arguably ambiguous). The ancestral body size estimation indicates a reduction of the size in branch leading to Sphaeridia and Mackenziella (with an ancestor estimated around 650 µm, against 910 µm for the ancestor of Sminthurididae). The tree annotated with all estimated ancestral body sizes is provided in Fig. S3.

Mackenziella psocoides is reported from North America (Hammer 1953), Scandinavia (Fjellberg 1988, 2007), Western and Central Europe (Hüther 1961, Pomorski 2000, Berg 2018, this paper), North Africa (Fjellberg 1989) and South Africa (Liu et al. 2012). Known localities are indicated in Fig. 9 and Table S3. Despite this wide distribution range, the species is uncommonly found. Of the two dozen reported findings, only Fjellberg (1989) found it in very large abundance (about 150 specimen), in the Canary Islands (Tenerife, La Palma and La Gomera). Another population of Mackenziella is regularly found in the region of Cape Town, South Africa (Liu et al. 2012 and C. Janion-Sheepers pers. comm.), where it is assumed to be invasive (Liu et al. 2012). Our specimens could not be distinguished from the description by Fjellberg (1989) based on specimens from the Canary Islands. Given that Fjellberg (1988, 2007) also recognized the same species in Scandinavia, we assume that the European and North African populations belong to the same morphological species. Hammer (1953) original description based on two specimens from Canada, indicated 4+4 teeth on the retinaculum (instead of 3+3 as in our specimens). Fjellberg (1989) examined the type specimens but could not distinguish the actual state of their retinaculum. Novel collection in the Nearctic and in the Southern Hemisphere will be necessary to confirm the morphological homogeneity of M. psocoides populations worldwide. Since some collembolan species are known to hide a large molecular diversity behind homogeneous morphology (e.g., Schneider and D'Haese 2013, von Saltzwedel et al. 2017), further genetic investigations are also desirable.

Figure 9. 

Reported findings of M. psocoides. Marker: blue – type locality, yellow – this work new locality, green – previous records.

Mackenziella psocoides is related to poor habitats exposed to drought: moss and vegetation on sand and rocks, sandy meadow (Berg 2018). In the Canary Islands, it was found on wind swept open ridges in the forested zone with sparse moss cover on the ground (A. Fjellberg, pers. comm.). It was found in proximity to a coal mining site near Spremberg in Germany (D. Russel, pers. comm.): we sampled this locality in November 2022 but could not find it there. The population sampled by us for this paper was apparently restricted to the concrete slabs path; similar mosses sampled around, but growing directly on the ground, did not yield it. In summer, the shallow substrate on the slabs is likely to dry out, without any options for the individuals to escape in the depth. Fjellberg (1989, 2007) suggested M. psocoides to be adapted to dry habitat through drought resistant eggs.Berg (2018) reached the same conclusions by reviewing all the findings of M. psocoides so far, indicating a preference for nutrient-poor, exposed habitats with shallow vegetations. This trait could also have allowed M. psocoides to reach a cosmopolitan distribution through airborne dispersal of the eggs (Fjellberg 1989). Our sampling shows the ability of M. psocoides to reach and colonize a small island of suitable habitat.

M. psocoides seems also to have a winter affinity. It was active shortly after the defrosting of its habitat. The defrosting may have triggered a rapid bloom of the population from diapause eggs, which was then already in decline one week later. However, fine observations would be needed to ascertain this. It is unclear if the individuals themself can withstand drought or frost through mechanisms of anhydrobiosis or cryoprotective dehydration known in several springtail species (Holmsrtrup 2018 and references therein).

Among the springtails found in the same habitat, we find notable the presence of F. angularis and Sphaeridia pumilis, both widespread and common species. Folsomides angularis is an indicator of dry habitats, well known for coping with drought through anhydrobiosis (Belgnaoui and Barra 1989). In Europe, it is commonly found in low-growing vegetation on exposed rocks (Fjellberg 2007), and it can settle in extreme environments such as the Namib desert gravel plains (Collins et al. 2019). Sphaeridia are the closest relatives of M. psocoides. Sphaeridia pumilis and M. psocoides were found together in one of the samples. This proximity was also observed by Fjelberg (1989) in the Canary Islands, who extracted a few individuals of M. psocoides from Euphorbia balsamifera litter and sand samples (collected dry, wetted before Berlese extractions), among several hundreds of Sphaeridia pumilis and Folsomides species (dominant F. terrus Fjellberg, 1992) (Fjellberg, comm. pers.). While we did not find any further mention of specific drought resistance for Sphaeridia pumilis, drought resistant eggs have been reported in Sphaeridia (Greenslade 1981). This trait could be ancestral to Mackenziella + Sphaeridia lineage. However, Sphaeridia does not only contain xerophilous species: the species from the tropical cloud forest of Ecuador (Bretfeld and Trinklein 2000) are likely never exposed to drought. The reduction of the mouthparts in Mackenziella (compared to the well-developed chewing-type mandibula in Sphaeridia) and its rarity suggest some specialization in food resources and micro-environmental conditions, while the co-occurring Sphaeridia pumilis may be more generalist.

We found M. psocoides, F. angularis and Sphaeridia pumilis in the same habitat, but with low overlap in the three distinct samples, hinting at a possible space or time exclusion of the species at fine scale. Further sampling is required to understand the dynamic of the springtail community in exposed habitats.

Our phylogenetic inference based on the nuclear rDNA, not only confirms Fjelberg (1989) views that Mackenziella is related to Sminthurididae but reveals that M. psocoides is a member of Sminthurididae, actually a close relative to Sphaeridia. The precise phylogenetic positioning of Mackenziella allows to draw its singular evolutionary history, as a member of Sminthurididae that reverted to an elongated body shape. The present phylogeny also indicates the paraphyly of Appendiciphora, with two possible evolutionary scenarios:

(1) the independent acquisition of the female anal appendages: once in Katiannidae (and presumably Arrhopalitidae not sampled here), and once in Sminthuridae (and presumably Bourletiellidae and Dicyrtomidae not sampled here)

(2) loss of those appendages in Sminthurididae.

The bootstrap support remained weak and we do not aim to further address this question here.

Eye. In most Sminthurididae, the eye is composed of eight ocelli (labeled A to H after Guthrie 1906). The ocelli C and D are usually reduced, e.g., in Sminthurides aquaticus and Sphaeridia pumilis (Fig. 10A, B). In species reported with reduced eyes such as Sminthurides sexoculatus Betsch and Massoud, 1970, C and D are missing. Palacios-Vargas et al. (2018) mislabelled the ocelli in the eye of Denisiella betschi Palacios-Vargas et al., 2018 and Denisiella rhizophorae Palacios-Vargas et al., 2018, and the small ocelli C and D have been overlooked (reduced, but still visible on their figs 12A and 25, in a similar configuration than in Sminthurides and Sphaeridia).

Figure 10. 

Left eye SEM microphotographs in frontal view. A Sminthurides aquaticus, B Sphaeridia pumilis. Arrow points in the anterior direction.

The reduction of the eye in M. psocoides can be described as follows: enlargement of C, loss of D, loss of either A or G (Fig. 5B). In Mackenziella a single chaeta is found between H and G, compared to two chaetae in the other genera of Sminthurididae (Figs 5B, 10A) (state unknown in Sminthuridia Massoud and Betsch, 1972).

Male antennal clasping organ (MACO). The four represented genera of Sminthurididae in our dataset form two clades: one including species with larger body size and complex MACO (Sminthurides + Stenacidia) and one with smaller species with simple male antennal clasping organ (Sphaeridia + Mackenziella). Following a parsimony criterion, one would assume the simple MACO to be the ancestral trait of Sminthurididae. Indeed, all Sminthurididae possess the modified chaetae of the simple clasping organ (Massoud and Betsch 1972, refer also to fig. 1 in da Silva Medeiros 2022). Those are, on Ant. II: a large chaeta (b1) on a tubercle, at least one additional smaller chaetal element (two elements only in Sminthuridia Massoud and Betsch, 1972) being small spines in the simple organ version. On Ant III, the simplest form seems limited to the large chaeta c3, in Sminthuridia. In Sphaeridia and Sminthuridia, the modified macrochaetae, possibly homologous to b2 and b3 (Ant. II), were initially not named in the antennal nomenclature of the Sminthurididae (Bestch and Massoud 1970, Massoud and Betsch 1972). Massoud and Betsch (1972) noted the variability of their presence in Sphaeridia (from 0 to 2), based on the original species description of the time, and represented a MACO devoid of additional chaetal element as the simplest form found in Sphaeridia (their fig. 9). da Silva Medeiros et al. (2022) followed this view, but associated the microchaetae of Sminthuridia to b2 and b3. At least several species of Sphaeridia possess two microchaetae in the same region.

The presence of b2 and b3 is possibly a plesiomorphy in Sminthurididae, with subsequent loss in M. psocoides and some species of Sphaeridia. However, a detailed re-analysis of Sphaeridia and Sminthuridia morphology and their internal phylogeny are necessary to gain a fine understanding of the homologies and evolution of the MACO. On the other hand, the complex MACO involves a large number of shared sub-organs following a similar organization. Those organs are on Ant. II: at least a trichobothria (Tra1) and from one to three additional modified chaetal elements (b4, b5, b6) generally mounted on a tubercle. On Ant. III from one to two additional modified chaetal elements (c1 and c2), and additional unnamed small spines and processes (variable among the genera).

Suprageneric subdivisions of the Sminthurididae based on the MACO were once suggested by Richard (1968) (in its restricted extent at the time and under the name Sminthuridinae): the Sminthuridini Börner, 1906 grouped the genera with a complex MACO and the Sphaeridiini Richard, 1968 accommodated Sphaeridia, the only genus with simple MACO at the time (also known now is Sminthuridia). However, those subdivisions were rejected by Betsch 1980 and not followed by any subsequent authors.

Claws. We note that the structure of the claws of M. psocoides is not perfectly matching the general description provided by Betsch (1980) for Symphypleona, stating “the external surface of the claw is convex and always without tegumentary grain” (translated from French). In M. psocoides, this is true in the apical part of the claw. On the basal part, the external smooth surface covers only the anterior and posterior edges of the claw, where it joins the weakly developed pseudonychia (Fig. 6C). We observed a similar condition in Sphaeridia pumilis, and Sminthurides aquaticus (for the last species, it is ascertained in claw II and III, but unclear in the fine and elongated claw I). The trait is difficult to assess with light microscopy. Claw II of Denisiella rhizophorae, also fine and long, is almost fully smooth (Palacios-vargas et al. 2018, their Fig. 12F). Further analyses are needed to understand the evolution of the claws in Symphypleona.

Tibiotarsus I. The modification of the ventro-apical chaeta ja we reported in M. psocoides is also apparent in Sphaeridia pumilis. After verification, we recognized that this chaeta was generally modified in Symphypleona (seen in representatives of Sminthurididae, Katiannidae, Sminthuridae, Arrhopalitidae, Dicyrtomidae and Bourletiellidae), a fact that is overlooked in the major syntheses on this order (Richard 1968, Betsch 1980, Bretfeld 1999).

Ventral abdominal chaetotaxy. The reduction or total loss of the ventral, anterior group of chaetae on Abd. IV may be the result of a common neotenic evolution of Mackenziella and Sphaeridia.

From a morphological point of view, the grouping of Mackenziella and Sphaeridia lacks clear synapomorphies. The simple MACO does not have specific innovations that would be missing in the lineages with the complex MACO and may be interpreted as the plesiomorphic state within Sminthurididae. However, the nuclear rDNA operon brings a strong support to the clade.

We expect the Sminthuridini (Börner, 1906) to be a natural group, including the 10 genera of Sminthurididae with a complex MACO (da Silva Medeiros et al. 2022). The Sphaeridiini Richard, 1968 may de facto be used to regroup Sphaeridia and Mackenziella. However, Frans Janssens drew our attention to the homonymy of Sphaeridiini Richard, 1968 with Sphaeridiini Latreille, 1802 (Coleoptera: Hydrophilidae). We propose the replacement name Sphaeridiaini (type-genus: Sphaeridia Linnaniemi, 1912) formed on the entire generic name as the stem, as per recommendation 29A of the Zoological Code of Nomenclature. To adjust to the current taxonomic levels in use in Collembola higher taxa (Bellini et al. 2022), we suggest to raise those taxa to the subfamily level (Sminthuridinae and Sphaeridiainae). Mackenziellidae is a junior synonym of Sminthurididae.

We consider the position of Sminthuridia to remain unclear. Sminthuridia possesses the tibiotarsal III organ, an apomorphic character also present in most of the (above defined) Sminthuridinae, but absent in Sphaeridia and Mackenziella. Betsch (1980) used the presence-absence of the tibiotarsal III organ for the first split of its determination key of the Sminthurididae, and also represented this split in his “phyletic scheme” of the Symphypleona, while stating that the character was not informative for the family evolution. The presence of the tibiotarsal III organ but absence of a complex MACO suggests a sister relationship between Sminthuridia and the Sminthuridinae.

Our phylogenetic reconstruction (Fig. 8) indicates that the common ancestor of Sphaeridia and Mackenziella was a roundish Sminthurididae, bearing traits similar to modern Sphaeridia species. The ancestral body size estimation indicates that the ancestor of Sphaeridia and Mackenziella underwent some size reduction. Of course, our sampling is limited and do not allow for a precise inference of ancestral body size. Sphaeridia pumilis (0.5 mm) is among the largest species of the genus (minimal 0.3 mm observed in Sphaeridia furcata Dunger & Bretfeld, 1989). Most species from the other genera of Sminthurididae have a body size ≥ 0.5 mm (0.7 mm for Stenacidia violacea and 1 mm for Sminthurides aquaticus), smaller species being exceptional (e.g., 0.35 mm for Sminthurides monnnioti Massoud & Betsch, 1966). Since our sampled Sminthurididae species are in the upper size range of their respective genus, increased sampling would likely result in inferior or equal body size estimation for the ancestor of Sphaeridia and Mackenziella.

The transformation process from a Sphaeridia-like ancestor toward M. psocoides can be described as follows:

(1) Shortening and bulkening of the antennae

(2) Reduction of the clypeal area

(3) Reduction of the maxilla and loss of the oral fold, the maxillary outer lobe and of the mandibula.

(4) Straightening of the head from an orthognathic to a more prognathic position.

(5) Reduction of the eye.

(6) Stretching of the trunk from a globular shape toward an elongated shape (including increased distance between the retinaculum and the ventral tube). The trunk deflation emphasizes the dorsal bulges aligned with the chaetal pattern, well-marked until Abd. II. The bulges are more or less marked in other Sminthurididae (Fig. 1A–C).

(7) Formation of a granulated bulge (b) that could be interpreted as the reformation of a pronotum (or analogous structure). Alternatively, it could originate from the head occiput, pushed back dorsally during the head returns to horizontal position.

(8) Reduction of the furca from the long one reaching the first thoracic segment, toward a short one barely the ventral tube (when folded under the body).

(9) Modification of the long mucro with a pair of posterior lamellae, typical for Symphypleona and Neelipleona (Bretfeld 1986, 1999, Schneider 2017), into a short mucro with an anterior hardened ridge.

(10) Loss of chaetae and basal tubercles on the retinaculum.

(11) Loss of three of the five pairs of abdominal trichobothria (one of those remaining pairs being further reduced to small s-chaetae in the female). As suggested by Fjellberg (1989), the strongly reduced chaetotaxy of Th. II tergite is inherited from its globular ancestor.

The shortening of the antennae and the reduction of the mass of the abdomen may be the result of a neotenic process, since those traits are also observed in juveniles of Symphypleona (Fig. 1K). Assuming a “Sphaeridia-like” ancestor, we can wonder what conditions allowed M. psocoides lineage to undergo this specific evolutionary trajectory. Fjellberg (1989) suggested that M. psocoides was adapted to reach food in the inter-leaf spaces of the mosses. In general, it is clear that M. psocoides gained a much slender profile compared to Sphaeridia, from body elongation and size reduction. The straightening of the head and the reduction of anterior trichobothria may indeed allow M. psocoides to reach as far as possible into reducing spaces. The persistence of one anterior trichobothria in the male complicates this straightforward scenario: perhaps the male focuses more on finding females than feeding (the male is also even smaller than the female).