Two levels of host-specificity in a fig-associated Caenorhabditis

Background Biotic interactions are ubiquitous and require information from ecology, evolutionary biology, and functional genetics in order to be completely understood. However, study systems that are amenable to investigations across such disparate fields are rare. Figs and fig wasps are a classic system for ecology and evolutionary biology with poor functional genetics; C. elegans is a classic system for functional genetics with poor ecology. In order to help bridge these disciplines, here we describe the natural history of a close relative of C. elegans, C. sp. 34, that is associated with the fig Ficus septica and its pollinating Ceratosolen wasps. Results To understand the natural context of fig-associated Caenorhabditis, fresh F. septica figs from four Okinawan islands were sampled, dissected, and observed under microscopy. C. sp. 34 was found in all islands where F. septica figs were found. C. sp. 34 was routinely found in the fig interior and almost never observed on the outside surface. Caenorhabditis was only found in pollinated figs, and C. sp. 34 was more likely to be observed in figs with more foundress pollinating wasps. Actively reproducing C. sp. 34 dominated younger figs, whereas older figs with emerging wasp progeny harbored C. sp. 34 dispersal larvae. Additionally, C. sp. 34 was observed dismounting from plated Ceratosolen pollinating wasps. C. sp. 34 was never found on non-pollinating, parasitic Philotrypesis wasps. Finally, C. sp. 34 was only observed in F. septica figs among five Okinawan Ficus species sampled. Conclusion These observations suggest a natural history where C. sp. 34 proliferates in young F. septica figs and disperses from old figs on Ceratosolen pollinating fig wasps. The fig and wasp host specificity of this Caenorhabditis is highly divergent from its close relatives and frames hypotheses for future investigations. This natural co-occurrence of the fig/fig wasp and Caenorhabditis study systems sets the stage for an integrated research program that can help to explain the evolution of interspecific interactions.


Background
Interactions at a broad range of scales structure the organization of biological systems. Within ecology, the biotic environment is a major determinant of the distribution and abundance of both species and communities, and so understanding the origins and maintenance of interspecific interactions is a key goal within the field. Yet, interspecific relationships taken as an aggregate are composed of millions of interactions between individual organisms [1,2], and the nature of those individuals is in turn strongly dependent upon the interactions of thousands of genetic elements comprising their overall genetic composition [3]. Thus a thorough explanation of how and why species interact with one another is ultimately dependent upon information about the genetic bases of such interactions, of which we currently know very little. A full analysis of all of these interactions, from gene to ecosystem, requires the development of study systems in which the power of modern genetic approaches can be used within the context of a compelling ecological circumstance. Here we begin to establish such a system using a newly discovered nematode species that lives in association with the classic fig-fig wasp ecological system [4].
Eukaryotic laboratory model systems have been rightly heralded for their contributions to our understanding of genetics [5][6][7]. However, only a fraction of their genes are annotated, and there are thousands of genes that as of yet have no known function [8]. Understanding the natural ecological functional context of these genes holds the potential to unlock this mysterious fraction of the genome [8]. Conversely, an understanding of the molecular biology of gene function can be used to inform ecology and evolutionary biology-those interested in the molecular basis of adaptive traits (such as the wing patterns of Heliconius butterflies [9] or coat color in crows [10]), physiological systems that structure species distributions [11][12][13], and the underpinnings of hostmicrobe interactions [1]) all need functional genetic tools to address their questions [14]. Are such tools also needed to understand the interspecies interactions that underlie most ecological theory?
Successfully traversing these broad fields requires the development of appropriate study systems-particularly systems wherein questions spanning multiple levels of biological organization can be simultaneously addressed. And although there are systems with compelling ecology and evolution (such as Heliconius [15], ants/acacias [16], and Darwin's finches [17]) and systems with well-established and powerful functional genetics (such as fruit flies [6], yeast [5], and worms [7]), systems with a good knowledge of both are rare. The development of good functional genetics in established ecological systems [8] and/or the development of good ecology in established genetic systems [14] is necessary to bridge these gaps.
A classic system for coevolutionary studies is the fig microcosm [4]. The subject of decades of research efforts [18][19][20], this system has revealed important advances regarding mate competition [21][22][23], sex ratio allocation [22,24], and the maintenance of interspecific interactions [25], among others. Furthermore, this system entails a textbook mutualism in figs and their associated wasps: figs need wasps for pollination, and wasps lay their eggs in fig ovules [4]. This system is amenable to experimental manipulation in the field, and evolutionarily-relevant measurements such as the number of seed, wasp progeny, and wasp foundresses are easily ascertained [4]. Thus, this is a powerful system for investigating a number of fundamental questions in ecological and evolution. Likewise, a classic model genetic system is the roundworm Caenorhabditis elegans. Like most genetic models, it is easy to rear in the laboratory and is amenable to sophisticated genetic manipulations. Furthermore, the background knowledge concerning its molecular, cellular, and developmental biology is simply vast-we arguably know more about this species than any other metazoan.
Recently, the nematode Caenorhabditis sp. 34, a novel sister species to C. elegans, has been discovered inside the fresh figs of Ficus septica [26]. As multiple reverse genetic techniques are applicable across the genus [27,28], this species is particularly well-positioned to connect functional genetics with natural ecology. To this end, here we describe the natural ecology

Collection sites
C. sp. 34 was originally isolated from the fresh figs of Ficus septica on the island of Ishigaki in Okinawa Prefecture, Japan by Natsumi Kanzaki (Figure 1). To further probe the natural context of this species, F. septica figs were sampled from additional Okinawan islands ( Figure 2 and Table 1). In May 2015, F. septica was sampled from Ishigaki and Iriomote islands, while in n May 2016 sampling of figs was expanded to include the islands of Ishigaki, Iriomote, Miyako, and Yonaguni (Supplemental Table 4). Sampling was also attempted on the islands of Okinawa (main island) and Tarama: F. septica was not found at all on Tarama, and although F. septica was identified on Okinawa main island, figs were not sampled because no easily-accessible figs could be picked. F. septica was typically found at the edge of vegetation on roadsides, but sampling was also performed in the public areas of Banna Park (Ishigaki) and Uenootakejoshi Park (Miyako). In May 2015 and May 2016, additional Ficus species were also sampled when accessible figs were found. Images revealing geographic position information of sampled plants were generated with Mapbox.  (Figure 1d). L1 and L2 stage animals were observed but not noted as they tended to coincide with adult animals and were more difficult to morphologically identify. The dispersing morphotype ( Figure 1d) that dominated later stage figs ( Fig. 3f) was confirmed to be Caenorhabditis in the field via pharynx morphology under compound light microscopy, COI sequencing with phylogenetic analysis (Figure 4), and their development into reproductive stage Caenorhabditis under culture conditions ( Figure 5). As stress conditions can promote both L1 arrest and dauer larva formation in Caenorhabditis [31], and the microscopic power necessary to identify key morphological features of dauer larvae [32] was not available in the field, dispersing animals could not be assigned to such specific developmental stages. This is particularly relevant given the extreme morphological divergence of C. sp. 34 [26]. Furthermore, specific Caenorhabditis species assignment is problematic as mating tests are typically necessary for species assignment in Caenorhabditis [30]. However, it is highly likely that these animals are C. sp. 34 as they share the same specific ecological niche, geographic locality, morphology (Figure 1), and nearly identical COI DNA sequences with the C. sp. 34 reference genome ( Figure 4). Regardless, reproductive or dispersal Caenorhabditis were noted as "abundant" if ≥20 individuals were observed and "rare" if <20 individuals were observed. Dissected figs were observed under a Nikon SMZ-2 dissection microscope, and pharynx morphologies in young larvae were observed with mounted live specimens under a AmScope M100C-LED compound light microscope. DNA sequencing and phylogenetic analysis Ficus, wasp, and nematode species were initially identified via morphological characteristics. Subsequently, DNA was isolated from some ethanol-preserved, F. septica-derived wasp and nematode specimens and sequenced to verify genus identity. For wasp samples, preserved animals were washed three times in PBS and subsequently crushed with a pestle in a 1.5 mL Eppendorf tube. DNA was then isolated from the suspension with a Qiagen Blood and Tissue DNeasy kit. For worm DNA samples, preserved single individuals were washed three times in PBS and digested with 5% Proteinase K in Tris-EDTA buffer for 1 hour at 58°C. This solution was immediately used for PCR after a 10 minute, 95°C incubation for enzyme deactivation. For wasp and nematode identification, the mitochondrial cytochrome oxidase I (COI) locus was amplified with primers LCO1490 (5'-GGTCAACAAATCATAAAGATATTGG-3') and HCO2198 (5'-TAAACTTCAGGGTGACCAAAAAATCA-3') [33]. PCR reactions were performed with the New England BioLabs Phusion High Fidelity PCR kit. For all reactions this thermocycler program was implemented: 98°C for 10 min. initial denaturation; 98°C for 10 sec. denaturation; 45°C for 30 sec. annealing; 72°C for 30 sec. extension (37 cycles); 72°C for 10 min. final extension. Sanger sequencing was performed by Genewiz. Sequences were then queried with BLAST to the NCBI GenBank database to identify closely related taxa.  (GenBank accession KP015022) and Schistonchus guangzhouensis (GenBank accession EU419757), and the marine rhabditid Litoditus marina (which was a high BLAST hit for an unidentified nematode species found among our preserved specimens, GenBank accession KR815450). Sequences of the phylogenetically informative taxa Pristionchus pacificus, C. japonica, and C. elegans were retrieved from WormBase [34]. The C. sp. 34 COI sequence was retrieved from the genome assembly (https://www.ncbi.nlm.nih.gov/nuccore?term=382947%5BBioProject%5D). Sequences were aligned with MUSCLE [35], and a maximum likelihood phylogenetic analysis was performed with RaxML [36] under a GTR-gamma model with 1000 bootstrap replicates. Analyses were performed with (335 bp alignment) and without (344 bp alignment) wasp taxa.

Wasp capture, nematode dispersal observations, figs temperature measurements
Parasitic and pollinating fig wasps emerging from intact F. septica figs were caught in a plastic bag ( Figure 1e). These insects were then killed and placed on Nematode Growth Medium (NGM) agar plates seeded with E. coli OP50 bacteria [37]. Plates were monitored for disembarking nematodes three hours and two days after plating. Nematodes of a given morphotype were confirmed to be C. sp. 34 via pharyngeal morphology and, in some cases, subsequent development into reproductive phase C. sp. 34 ( Figure 5).    Parasitodiplogaster is a known diplogastrid parasite of Ficus-associated wasps [38], and Schistonchus is a known tylenchid (clade IV [39]) Ficus plant parasite [40]. Although this analysis is insufficient to resolve known Caenorhabditis phylogeny [41], 17 individuals (blue squares) share near identical sequence with C. sp. 34. An unidentified nematode species (red squares) was also observed; this species is unlikely to be Schistonchus or Parasitodiplogaster as it does not cluster with representative species in this analysis and returns rhabditid nematodes in a BLAST query to GenBank (top BLAST hits include the marine rhabditid Litoditis marina [42], Phasmarhabditis sp. [43], and Acrostichus sp. [44]; all 88-89% identity). This tree was generated with RaxML under a GTR-gamma model and 1000 bootstrap replicates. Scale bar represents 0.1 substitutions/site. ** = >90% bootstrap support. *** = 100% bootstrap support.

C. sp. 34 reproduces in young figs and disperses in old figs
Caenorhabditis nematodes can undergo alternative developmental trajectories depending on environmental conditions [47]. If conditions are favorable, animals develop into adults capable of reproduction. But in crowding, starvation, or otherwise stressful conditions, animals develop into the long-lived, stress-resistant dauer larva [47]. It is this dauer stage that is used for dispersal to new food sources in the wild [48].

Caenorhabditis has only been found in F. septica figs among Okinawan Ficus
A number of Caenorhabditis species are associated with a variety of plant substrates [50,51]. However, pollinating fig wasps tend to be associated with only one or two species of Ficus [4,52], which suggests that fig wasp-associated Caenorhabditis may also be limited to specific Ficus species. To determine if this is so, figs from additional Okinawan Ficus species were sampled. Of the nine Ficus species reported to be in the sampling locales [53], four species were found with fresh figs aside from F. septica (Table 2). No figs aside from F. septica were contained C. sp. 34 nematodes (Table 2), despite some of these species being known to harbor multiple nematode groups [54,55]. Thus, this particular fig-associated C. sp. 34 is possibly a host specialist and restricted to one species of Ficus.

F. septica figs harbor interior temperatures that are comparable to C. sp. 34 labrearing temperatures
The environmental parameters defining Caenorhabditis ecological niche space are nearly entirely unknown [50]. Among these, temperature influences a multitude of life history traits in Caenorhabditis, including survival and reproductive rate [56,57], as well as the dauer entry switch [58].  [26] was utilized. Regardless, these observations provide a unique snapshot into the natural context of C. sp. 34. Future estimates of additional natural environmental parameters will be essential in informing hypotheses regarding the evolution and ecology of these organisms.

Discussion
The  [62]. Figs are also a key resource for over a thousand bird and mammal species, who in turn aid in seed dispersal [63]. As a consequence of this microcosm complexity, this remains an influential and active system for study in ecology and evolution [64][65][66][67]. However, none of the species in these communities are particularly amenable to functional genetics and laboratory studies-both of which are crucial for refining the explanatory power of evolutionary science. Conversely, as thousands of genes in multiple long-standing eukaryotic laboratory model systems have no known functions [8], it is likely that their natural ecological contexts (which have often been neglected) will be needed to thoroughly understand their genomes. As a consequence, there have been calls to integrate ecological, evolutionary, and functional genetic approaches [8,14]. Here, we have described the natural history of C. sp. 34, a close relative of the model genetic organism C. elegans. What has been observed in this Caenorhabditis study, together with the known biology of the fig microcosm, can then be used to inform hypotheses regarding the evolution of interspecific relationships in both systems.
Caenorhabditis typically proliferates on rotting plants and disperses on invertebrate carriers. And although the features defining niche specialization in this group remain uncertain, it seems clear that there is variation in its extent. Some species appear limited in their geographic range (C. sinica has only been found in east Asia [68]), whereas others are globally distributed [50]. Interspecific variation in seasonal predominance of wild populations has been observed, consistent with variation in fitness at different temperatures [51]. Furthermore, different Caenorhabditis species have been found associated with different bacterial communities [69], consistent with variation in bacterial preference [70]. There is also interspecific variation in the extent of dispersal carrier specificity. Some Caenorhabditis are promiscuous in their choice of carrier; C. elegans has been found on snails, slugs, isopods, and myriapods [51]. Other species (such as C. japonica, C. angaria, and C. drosophilae), despite intensive sampling, have only been observed dispersing on one insect species in a highly host-specific manner [50,71]. The existence of C. sp. 34 in the fresh figs of a single species of Ficus and observations of its  [26], suggesting that this change in natural history has promoted the evolution of novelties within this species  Figures 1-2), consistent with previous observations of inbreeding in pollinating wasps [22,72]. Variation in founder population size and its inbreeding effects have been shown to have consequences in local mate competition and sex ratio allocation in fig wasps [22,23]. This may then also hold for C. sp. 34, although it is possible that resource availability is different for nematodes (probably bacterial food) and wasps (fig ovules). Furthermore, pollinating fig wasps have been shown to exhibit tremendous dispersal distances [73], which is consistent with low levels of observed population structure for a number of fig wasp populations [72,74,75]. These patterns from wasps suggest C. sp. 34 may also have unique population genetic features. Male/female Caenorhabditis species tend to be incredibly diverse with enormous population sizes, and C. brenneri is among the most diverse eukaryotes known [76]. The expected inbreeding in C. sp. 34 should reduce diversity, as has been seen in C. japonica, another Caenorhabditis male/female species with high host-specificity [77]. The selfer C. elegans displays reduced diversity, low global population structure, yet high local structure [78,79]. This is consistent with a boom-and-bust natural history with high migration and largely clonal local populations initiated by single founders [48,50,80]. As C. sp. 34 is dispersed by wasps that can migrate over long distances while exhibiting small founder populations (Supplemental Figure 1), they may have more population genetic features in common with selfing lineages than expected of a typical gonochoristic Caenorhabditis species.
C. sp. 34 also displays differences in developmental timing and developmental decision-making from their close relatives [26]. Their developmental rate is very slow compared to its close relatives, and dauer larvae (an alternative developmental trajectory favored under stress and dispersal conditions) are rarely seen in laboratory populations. Here, we find that reproductive stage animals are enriched in younger figs and smaller, dispersing larvae are found in older figs (Fig. 3). It was not possible to absolutely confirm that these were morphological dauer larvae due to limitations in microscopy in a field setting. However, given that nearly all Caenorhabditis observed on invertebrate carriers are in the dauer stage [48], it is likely that animals found in older figs and fig wasps were indeed dauer larvae. Given that figs typically take weeks to develop [18], and that C.  [40] and the wasp parasite Parasitodiplogaster [38], C. sp. 34 is unlikely to inflict direct harm on figs or wasps as a parasite. This is because C. sp. 34 maintains its typical Caenorhabditis pharyngeal morphology throughout the reproductive stages observed in fresh figs (plant parasitic nematodes typically have pharyngeal stylets [81]), and proliferative animals are not associated with wasps ( Fig. 3; Fig. 5). As a particle feeder, it is possible C. sp. 34 eats Ficus pollen, thereby affecting host fitness. This seems unlikely, however, as C. elegans cannot ingest particles greater than 4 microns in diameter [82], and Ficus pollen tends to be larger than this on average [83]. C. sp. 34 may affect pollinator wasp fitness through phoresy by somehow adversely affecting pollinating wasp travel across figs. Considering the size of C. sp. 34 dauer larvae (Fig. 1), the low C. sp. 34 dauer larvae load on emerging pollinating wasps (Supplemental Fig. 2), the pervasiveness of phoresy as a dispersal strategy [84], and the contingency of worm success on wasp success in this case, a large cost to wasp dispersal ability also seems unlikely.  [4], and contemporary metagenomic tools can define microbial communities [1], the interplay between C. sp. 34 activity, microbial communities, and host fitness should be able to be interrogated in the future.
As our understanding of the Caenorhabditis-associated microbiota is rapidly increasing [69,85,86], this affords an exciting opportunity for future research.
Notably, C. sp. 34 was found dispersing on pollinating Ceratosolen wasps, and not Philotrypesis parasitoid wasps emerging from figs of the same tree ( Figure 5). In contrast to pollinating wasps, who must enter the fig to lay eggs, Philotrypesis wasps do not enter the fig and use long ovipositors to lay eggs from the fig exterior [87]. This suggests that dispersing C. sp. 34 must discriminate within the fig to find the appropriate carrier. This would likely be a novel behavior, as its close relatives are not fig-associated and tend to be promiscuous in carrier choice [51] (although some preferences in C. remanei have been noted [88]). The more distantly-related C. japonica has been shown to have behavioral preferences for its shield bug host [89], and similar findings have been shown for Pristionchus nematodes and their host beetles [90]. Nematode occupancy biases on pollinating wasps relative to parasitic wasps have been observed in the figassociated parasitic Schistonchus and Parasitodiplogaster nematodes [45,91]. This typical preference for pollinating wasps has been recapitulated in a laboratory framework with Schistonchus using traditional chemotaxis assays with wasp-derived volatiles and cuticular hydrocarbons [92]. Similar studies could be extended to the culturable C. sp. 34 to interrogate the genetic basis of novel behaviors.

Conclusion
The elegance of contemporary molecular biology resides in the explanatory power generated by conceptual continuity across multiple hierarchical levels [93] (aka vertical integration [94]). Such continuity is rarely found in evolutionary science-it remains unclear how the disparate pieces of population-level processes, environmental effects, developmental events, and historical contingencies interact to generate diversity in nature. Here, we described the natural history of a close relative of C. elegans that is associated with figs and fig wasps. The fig-fig wasp system is a legendary study system in evolution and ecology, and C. elegans is a legendary one in model systems genetics. Here then is a serendipitous convergence of research organisms that can facilitate the conceptual connection of their respective disciplines. The functional genetics of C. sp. 34 has the potential to inform the molecular basis of how ecologically-relevant phenotypes are generated, whereas the evolution and ecology of the fig system can inform how populationlevel and environmental forces sort said variation. This all begins with a simple understanding of where and how this organism lives in nature.

Declarations
Author contributions GCW collected and analyzed the data. GCW and PCP wrote the paper.