Existing paleontological data indicate marked eukaryote diversification in the Neoproterozoic, ca. 800 Ma, driven by predation pressure and various other biotic and abiotic factors. Although the eukaryotic record remains less diverse before that time, molecular clock estimates and earliest crown-group affiliated microfossils suggest that the diversification may have originated during the Mesoproterozoic. Within new assemblages of organic-walled microfossils from the ca. 1150 to 900 Ma lower Shaler Supergroup of Arctic Canada, numerous specimens from various taxa display circular and ovoid perforations on their walls, interpreted as probable traces of selective protist predation, 150–400 million years before their first reported incidence in the Neoproterozoic. Selective predation is a more complex behavior than phagotrophy, because it requires sensing and selection of prey followed by controlled lysis of the prey wall. The ca. 800 Ma eukaryotic diversification may have been more gradual than previously thought, beginning in the late Mesoproterozoic, as indicated by recently described microfossil assemblages, in parallel with the evolution of selective eukaryovory and the spreading of eukaryotic photosynthesis in marine environments.
A sensu lato definition of predation could be summarized as the action of an organism killing for nutritional purposes. In detail, predatory interactions represent an array of feeding habits, from microbes feeding on microbes to carnivorous metazoans. The only fundamental outcome is the death of the prey. Parasitism, browsing, or scavenging is therefore precluded from this definition, although there is overlap between all of these behaviors [1–4].
Because the prey is killed, predation plays a substantive role in evolution via natural selection. Thus, it has been regarded as a key factor in many major biological transitions, including the origin of Domain Eukarya [5–7], the appearance of sexual reproduction , of multicellularity , the evolution of protist biomineralization , and the diversification of acanthomorphic (process-bearing) protists .
A major diversification of eukaryotes, beginning ca. 800 Ma, is documented in paleontological archives by the increase in taxonomic richness of complex taxa [12,13]. Possible drivers for such increase notably include abiotic factors (geochemical, redox, climatic, and tectonic changes) [13–15], but also biotic factors, such as the development of predatory behavior by eukaryotes preying on eukaryotes (eukaryovory; [9,16]) or the appearance of sponges and green algae . A three-step model for eukaryote diversification  proposed an earlier diversification, ca. 1.1 Ga, based on the appearance of biological innovations in early eukaryotic cells, including the evolution of eukaryotic photosynthesis in marine environments indicated by the occurrence of the microfossil red alga, Bangiomorpha . A revised age for this fossil and cross-calibrated molecular clock analyses  indicate the evolution of crown-group eukaryotes at least by 1.045 Ga and the acquisition of the chloroplast by 1.25 Ga (but see ref. ), providing a minimum age for Archaeplastida and thus indirectly for the last eukaryotic common ancestor (LECA). Nevertheless, the occurrence of ornamented and process-bearing acritarchs and macroscopic carbonaceous compressions interpreted as unambiguous but taxonomically unresolved eukaryotes in 1.7–1.65 Ga successions indicates that the domain may have originated at least in the late Paleoproterozoic (review in ref. ).
Here, we report circular and ovoid perforations on microfossils from three formations of the lower Shaler Supergroup (1.15–0.9 Ga; [23,24]), indicating the presence of predatory behavior at the Mesoproterozoic–Neoproterozoic transition. The perforations are similar to those documented in younger material [16,25] and in modern protists preyed upon by vampyrellid amoebae. This older evidence of predation provides significant insights into eukaryote macroevolution and diversification.
The lower Shaler Supergroup
The lower Shaler Supergroup, outcropping in the Brock Inlier and Coppermine areas of northwestern Canada (see Supplementary Material), consists of alternating carbonate rocks and quartzarenite, with subordinate siltstone and mudstone, and was deposited in shallow water in the Amundsen basin. The basin preserves sediments deposited in an intercontinental sea that was intermittently connected to an open ocean [26,27]. Although geochronologically well constrained (see ref. ), the age of the lower Shaler Supergroup, and especially the investigated interval, mainly relies on maximum depositional ages obtained from U-Pb dating of inherited zircon . Thus, the Grassy Bay and Nelson Head formations could have been deposited anytime between 1013 ± 25 Ma and before 892 ± 13 Ma, the age of the stratigraphically overlying Boot Inlet Formation (Re-Os on black shale; see Supplementary Material) . Similarly, the Escape Rapids Formation may have deposited anytime between 1151 ± 13 and 892 ± 13 Ma. Considering this age uncertainty, the studied interval should be interpreted as, at least early Neoproterozoic, possibly late Mesoproterozoic, in age; i.e. at the Mesoproterozoic–Neoproterozoic transition.
The perforated microfossils are preserved as carbonaceous compressions in shales of the Grassy Bay, uppermost Nelson Head and upper Escape Rapids formations, but are more abundant in the ca. 1 Ga Nelson Head Formation samples. They were extracted from the shales by acid maceration, with protocols that minimize manipulation . The microfossils are preserved in a variety of marine to fluvial settings. Grassy Bay and Nelson Head formations were deposited mainly in shallow marine deltaic environments, but the top of the Nelson Formation represents a barrier island lagoon. The upper Escape Rapids Formation records shallow and sub- to intertidal marine settings .
Description of the perforations
Perforations are reported for four identified species, two unnamed forms and unidentified fragments from large vesicles. Following previously defined criteria for fossil eukaryote identification , two species and one unnamed form are interpreted as unambiguously eukaryotic. They display conspicuous ornamentation on their wall surface such as little cushion-shaped outpockets (Culciculisphaera revelata Riedman and Porter, 2016 ; Figure 1A,1B) and sets of ridges, either parallel and concentric [Valeria lophostriata (Jankauskas, 1979) Jankauskas, 1982 , Figure 1C,1D] or randomly distributed (unnamed form, Figure 1E,1F). This latter form resembles specimens of Volleyballia dehlerae Porter and Riedman, 2016 , but the ornamentation is located on the inner wall surface and the ridges are not parallel. The two other species [Leiosphaeridia crassa  Jankauskas et al. 1989 , L. jacutica  Jankauskas et al. 1989 ] and one unnamed form with wooly wall texture (Figure 1G–1N) are probably also eukaryotic, but lack any particular taxonomic feature that could help resolve their affinities.
Microfossils with and without perforations.
The perforations are circular and ovoid ranging from 0.1 to 7.1 µm in minimum diameter (Figure 2). The average diameter of the perforation shows no correlation with the vesicle diameter (Figure 3). The depth of the perforations also varies, in a few cases perforating only the external envelope of the microfossil (Figure 1A), and shows, on most microfossils, a narrow range of diameters within the same specimen.
Perforation sizes in microfossils.
Comparison between microfossil and perforation diameters.
Their occurrence in different taxa, but not on all specimens of each taxon, their irregular distribution on the vesicle wall, and the lack of co-variation with the vesicle diameter suggest that they are not a morphological character of the species, such as pores or an excystment structure. Similarly, their occurrence on well-described, widespread taxa (Leiosphaeridia and Valeria) suggests that they are not a specific variation augmented by diagenesis. They differ from irregular holes left by secondary pyrite or other mineral grain imprints and do not exhibit a raised rim as the holes left by pyrite framboids . The specimens do not show any variety of degradation; they are either perforated or they are not. Moreover, no specimens display large zones of degradation, pitting or irregular perforations indicative of postmortem decomposition, for instance by bacterial scavenging [37,38]. As pointed out by Porter , to be considered as the result of postmortem microbial activity, the perforations observed in the present microfossils should have stopped at the same time and the same stage in all specimens within an assemblage. This constitutes an improbable scenario considering the number of perforated microfossils and the time interval along which they occurred.
Origin of the perforations
The perforations on the microfossils' organic walls reported here are similar in morphology, size range, and distribution to those reported from the 780–740 Ma Chuar Group, interpreted as direct evidence of predation by protist on protist, or eukaryovory [9,16,39]. They are also similar to the 1–3 µm circular holes observed on Cerebrosphaera from the 750 Ma Svanbergfjellet Formation, Svalbard , interpreted as predator or scavenger traces (Figure 2). The perforations are unlikely to be the result of diagenetic effects and rather suggest the presence of predatory behavior during deposition of the lower Shaler Supergroup. Because predation increases morphological innovation in prey, it constitutes an important selective driver of diversity [2,9,11,16,40]. For instance, an increased diversity of acanthomorphs could indirectly, but not necessarily, illustrate such selective pressure [2,9].
In the literature, various examples of similar perforations on organic-walled microfossils exist (Figure 2), but the authors did not highlight or fully document them. Loron and Moczydlowska  illustrate a specimen of Cerebrosphaera from the upper formation of the ca. 800 Ma Visingsö Group, Sweden, with numerous minute circular holes on the surface (plate 1, Figure 2). Similarly, in the Tonian of the Lena–Anabar basin, Siberia, Nagovitsin et al.  illustrate a minutely perforated fragment of Cerebrosphaera (Figure 9e) and report spheroidal vesicles with numerous circular holes on the surface (Figure 9d). Finally, a filamentous specimen from the ca. 1.65 Ga Ruyang Group also displays circular perforations (; Figure 2), as noted by Porter . It is probable that all of these examples represent other traces of predatory behavior, but more detailed investigation is required.
Perforations on microfossils are a way to get through the wall, to access cellular content. Such examples of predation on modern eukaryotes are numerous, but a good analog for Proterozoic predators could be the ‘vampire amoebae’ Vampyrellidae ( and refs. therein). These naked filose microorganisms with processes (filipodia) are known to predate, through perforation, cells and cysts of algae, fungi, protozoa, and even small metazoans and are common in today's soil, freshwater, and marine environments [44,45]. They chemically bore numerous, irregularly distributed perforations and circular depressions on the prey's wall with the tip of their filipodia by removing or dissolving disks of cell wall material [44,46–48]. When the wall is breeched, they suck up cell content or enter the cell to digest it from inside [45,47]. As summarized by Porter , there is a specificity of perforation sizes among modern vampyrellids, ranging from 0.2 to 6 µm.
The characteristics of fossil material permits the hypothesis that there is also a specificity of prey, with nutritional specialization, as observed in modern vampyrellids , and that different sizes or types of holes originate from different predatory taxa. An example would be the half-moon holes observed on vase-shaped microfossils (VSMs) by Porter (; Figure 5a–d). Nevertheless, the ability to make simple spheroidal or ovoid perforations, as reported here, is probably convergent among perforating protists and is also known in fungi-predating nematodes [49,50]. Therefore, it is impossible to confidently identify the predators .
This way of feeding differs from phagocytosis (engulfing a whole cell or a particle) and probably requires, in early eukaryotes, a more complex behavior and elaborate cellular organization, digestive enzymatic machinery, and chemical communications to sense the prey, as in modern vampyrellids . Likewise, as in vampyrellids, active, possibly controllable, appendices (processes) or pseudopods around the vesicles would require the presence of a complex cytoskeleton, but also a sophisticated level of control over cellular development [12,29,51].
Macroevolutionary implications of earlier predatory behavior
Because predation leads to the death of the prey, its impact on selection is considerable and co-evolution of the predator–prey system is viewed as an ‘arms race’: acquisition of improved defense in prey driving more selective predation, leading to even better adaptations [1,52,53]. Eukaryovory may have appeared early in eukaryote history. By definition, LECA was capable of phagotrophy [54,55], which should have represented a boost in fitness at that time. One question arises: if predatory behavior appeared early in eukaryotic evolution, and presumably had such a strong effect on fitness, why did eukaryotes take a long time to diversify from the time of their origin to the Neoproterozoic (late Tonian) increase in diversity? Knoll and Lahr  proposed that early predation would have first targeted prokaryotic preys, and much later eukaryotic preys, triggering a rise in diversity in the mid-Neoproterozoic. Another possible hypothesis is that early predators may have been opportunistic, targeting prokaryotic but also eukaryotic cells, reducing the impact on selection, for example, to a positive feedback loop of size increase (see ref. ).
However, the timing of diversity increase needs re-evaluation with the discovery of new diverse microfossil assemblages in the late Mesoproterozoic of western and central Africa [56,57] and with the ongoing study of the lower Shaler Supergroup microfossils. The new hypothesis presented here is that it is the transition from nonselective to selective predation that triggered the arms race: eukaryovory evolved from nonspecific phagocytosis of whole (prokaryotic and eukaryotic) cells to perforations through eukaryotic cell walls of selected preys. A highly developed cellular machinery is already a prerequisite for phagotrophy [51,58], but wall perforation requires additional biochemical enzymatic mechanisms and chemical sensing for prey detection, as in modern vampyrellids . Finally, feeding through perforation of the prey may yield significant benefits : rapid access to the cell content (increased by the multiple borings), lower volume of cell walls to digest by perforating small points, and protection against opportunistic surrounding bacteria or protists by preventing the burst of the prey after its cell lysis.
As mentioned above, the ability to perforate walls requires an increase in cytological complexity, biochemistry, and control of a complex cytoskeleton and, probably, the ability to sustain active processes (appendices) or pseudopods. Processes appear at least ca. 1.6 Ga, based on the occurrence of Tappania, an organic-walled microfossil bearing a variable number of heteromorphic and asymmetrically distributed processes , that possibly performed osmotrophy (passive uptake of dissolved nutrients) . However, the emergence of selective predation, shown by perforations through microfossil walls, started at least between1.15 and 0.9 Ga, the age of the studied rocks.
When placed on a diversification timescale model (Figure 4; [18,22]), the transition to selective predation in the Mesoproterozoic is contemporaneous with major changes in eukaryote evolution, including the development of multicellularity, photosynthesis, and, more generally, the appearance of crown groups. Some of these innovations may have resulted from predation pressure (see ref. ), for instance, the emergence of multicellularity .
Protist predation in context with biotic and abiotic changes during the Proterozoic.
Molecular clock estimates suggest that LECA originated ca. 1.9–1.0 Ga , or earlier 1.9–1.7 Ga , and indicate a rapid divergence of major clades in the following 300 million years  or ca. 1.2–1.0 Ga  (see reviews in refs [60,63]). Existing paleontological data, however, show that the Domain Eukarya is at least 1.7 Ga [18,64] and diversified ca. 800 Ma, notably in response to predation and abiotic events [12,13].
The evidence for selective predation in the lower Shaler Supergroup supports the development of major eukaryote clades in the late Mesoproterozoic, as estimated using molecular clock [18,62] and diversification models . It also suggests that Neoproterozoic expansion of Domain Eukarya may have been more gradual than previously observed in paleontological models [12,13,65]. The growing ecological complexity documented throughout the Neoproterozoic (appearance of VSMs, protist biomineralization, radiation of process-bearing acritarchs, or later metazoans; see refs [9,16,40,66]) may be the escalation of an arms race that started during the Mesoproterozoic–Neoproterozoic transition, possibly 1.15 Ga, triggered by selective predation pressure. Moreover, recent microfossil studies indicate that eukaryotes were already thriving by 1 Ga [55,56], supporting the idea of a more progressive expansion of the domain, an hypothesis to be tested by the study of other contemporaneous assemblages, including the lower Shaler Supergroup microfossils.
Predation by protist on protist (eukaryovory) emerged earlier than previously documented, during the Mesoproterozoic–Neoproterozoic transition, possibly by 1.15 Ga.
Emergence of predation through perforation implies the development of a more complex cellular machinery than for phagocytosis, to sense and select preys.
The transition from unspecific predation (phagocytosis) to selective predation (by perforation) may have triggered and driven the diversification of eukaryotes, in parallel to other biological innovations.
The short-time radiation of eukaryotes documented in the Neoproterozoic may have been more progressive, rooting back in the Mesoproterozoic.
Research support from the Agouron Institute, the FRS-FNRS, and the ERC Stg ELiTE FP7/308074, and technical support by M. Giraldo are gratefully acknowledged. Fieldwork logistics for sample collection were provided by the Geological Survey of Canada's Geomapping for Energy and Minerals Program. Field assistance was provided by T. Gibson (McGill University). We thank the guest editors for their invitation to contribute to this special issue and two reviewers for their constructive comments.
The Authors declare that there are no competing interests associated with the manuscript.