**4. Discussion**

We recently reported that APBs can be major components of endolithic intertidal ecosystems and could potentially be euendolithic in nature [15], for which no precedent existed. Alternatively, these APBs may constitute secondary colonizers of opened pore space that rely on metabolic interactions with cyanobacteria, as they commonly do in other benthic environments like microbial mats or microbialites [50,54,55]. We hypothesized that examining colonization using molecular techniques and photopigment analysis specifically targeting APBs could help solve this question, in that early colonizers of bare substrates can be logically assumed to be active borers, while a dependency on cyanobacteria should result on delayed colonization by APBs. The temporal dynamics of endolithic population of Chloroflexales indeed sugges<sup>t</sup> that this group of APBs are not euendoliths but instead act as secondary colonizers whose populations do not attain significance until communities of cyanobacteria are mature and the substrate has significantly eroded. The case of the proteobacterium *Erythrobacter* sp. was clearly di fferent, since significant populations of *Erythrobacter* were present early in the colonization process and were sustained through the experimental period. *Erythrobacter* are aerobic anoxygenic phototrophs that conduct photoheterotrophy, have a low BChl *a* content, and require a source of organic carbon [24,56]. Our endolithic sequences were most similar to those in Group I *Erythrobacter* genomes [57]. Under our hypothesis, these organisms could still be euendoliths, even though their populations remained low throughout the experiment. Alternatively, since these small unicellular bacteria are abundant in coastal marine waters [24,56], they could have easily washed into fresh pits made by cyanobacteria in exposed tiles. Our current data cannot fully solve these alternatives. In fact, the metabolic action of photoheterotrophs can increase pH levels around cells, leading the precipitation, not dissolution, of calcium carbonate [15], which would make a boring activity more di fficult [58]. By contrast, the lack of the more complex photosynthetic Chloroflexales and low total bacteriochlorophylls suggests that, during colonization, euendolithic cyanobacteria dominate the photosynthetic niche due to their ability to excavate habitable space and utilize the mineral carbon for autotrophy [18,59]. Only once su fficient habitable space has been created by cyanobacteria can significant populations of APBs develop.

**Figure 4.** Detailed phylogenetic relationships of sequences in the "unknown boring cluster (UBC)", with environmental (uncultured) cyanobacterial sequences from stromatolites (shaded in green) and the closest known euendolith cluster (shaded in blue). Branch lengths are substitutions per site and node labels indicate bootstrap values.

We found that the patterns of endolithic cyanobacterial succession within hard intertidal carbonates sustain three distinct phases (early, late succession, and steady-state climax). In our habitat, early colonization is predominantly conducted by a previously undescribed group of euendolithic cyanobacteria (UBC) that rapidly colonizes rock to maximal levels within 3 months. This clade

could exceed 40% of endolithic cyanobacterial populations early on. Cluster 1 (*Leptolyngbya-*like) organisms also contribute to early colonization but only reach 60% of the biomass of UBC. By 9 months of incubation, the three canonical groups of euendolithic cyanobacteria, *Leptolyngbya (*which we tentatively equate to the *Plectonema terebrans* morphotype), boring members of the Pleurocapsales, and *Mastigocoleus testarum* gain a foothold. Finally, as the community reaches a steady-state climax composition, euendolithic cyanobacteria are displaced in relative importance by other cyanobacteria and by significant populations of Chloroflexalean APBs. The initial large abundance of the UBC could be explained by the presence of fast-growing propagules in natural seawater that quickly attach and bore into the substrate. Since boring microorganisms are fixed in place in their boreholes, competition for space, which can influence patterns of distribution in benthic cyanobacterial communities [60] is likely not a relevant factor until significant proportions of the rock surface become colonized. Hence, having an early foothold on the substrate may have ensured their persistence through time, as we observed. However, UBC did not continue to increase in population size through the colonization, unlike the total cyanobacterial population, which did. The dynamics of the Cluster 1 (*Leptolyngbya*-like) members were not very di fferent from those of UBC, although they seemed to sustain net population losses in late stages of colonization. The net gains in later stages can be attributed to Cluster 3 (*Mastigocoleus*) and, even more so, Cluster 2 (Pleurocapsalean) cyanobacteria (Figure 3d). As these slow colonizers begin to excavate more carbonate, they could reach a threshold where individual pore spaces become connected and pioneer organisms are no longer fully insulated from competition for space. Chlorophyll and qPCR data sugges<sup>t</sup> that this carrying capacity is reached by 9 months of incubation. This density-dependent competition would also explain the overall decline in cyanobacterial evenness/Shannon diversity with successional progress. Finally, at maturity, as endolithic space has been colonized and the rock has become porous, nonboring endoliths can begin to colonize. One can imagine a scenario where nonboring endoliths, which need not spend energy for excavation, can outcompete borers in the outermost sections of the rock. Euendoliths would still have a competitive advantage deeper within the rock. This would be consistent with the relative decline of boring cyanobacterial ASVs in steady-state climax communities, as they are better adapted to di ffusion-limited conditions. Interestingly, we did not see a di fference in cyanobacterial pigment concentrations between the 9-month samples and steady-state climax communities, which suggests that nonboring phototrophs may colonize the upper interior of the rock, shading the deeper euendoliths and contributing to their decline.

A comparison of our results with prior colonization studies shows that there exist similarities, as well as marked di fferences, with the dynamics of porous, biogenic coral skeletons. For example, early work [4,61] demonstrated the divergence in euendolith composition between shells and inorganic calcites. However, careful consideration must be taken as both substrate composition [9,10,12,14] and water depth [9,10] influence community structure, and, as discussed above, there are substantial di fferences in methodology. Even bearing those caveats in mind, the fact that all four major euendolithic clades are present after 3 months of colonization corroborates the prior conclusions that cyanobacterial colonization happens swiftly, in as little as 4 weeks, with *Plectonema, Mastigocoleus, Solentia,* and *Hyella* species all present [3,8–10]. Interestingly, there are no reports of any *Chroococcidiopsis*-like organism that could potentially represent our UBC. We also found that though *Mastigocoleus* does colonize quickly, it does not reach large abundances until the community approaches a steady-state climax composition, in contrast to the findings from corals where it is one of the first and most abundant pioneer organisms. Our observations on Cluster 1 *Leptolyngbya*-like euendoliths agree with the patterns of *P. terebrans* described by Grange et al. [11]. We find that this cluster peaks in abundance after 6 months, which was also found for coral systems. However, when comparing 9-month Cluster 1 *Leptolyngbya-*like populations to those of steady-state climax communities, we found that Cluster 1 *Leptolyngbya-*like populations were less than 10% of the 9-month totals, whereas in corals, *P. terebrans* remains very abundant through maturity [6]. Cluster 2 Pleurocapsalean euendoliths were not very abundant (sometimes < 1%) in previous colonization experiments and surveys, which was attributed to their alleged susceptibility to grazing by fish and chitons due to their shallow mode of boring [11]. This was

clearly not the case in our system, with Cluster 2 Pleurocapsalean organisms being the most abundant euendoliths after 9 months. Perhaps grazing pressure was unusually low in our setting, even though we did see abundant, actively grazing chitons during sampling. Though the abundance of eukaryotic euendoliths are widely reported in coral systems [6,11], we did not find a significant contribution of plastid 16S rRNA genes in our samples, and those that were there were not phylogenetically related to known euendoliths.

In summary, by applying molecular approaches to euendolithic systems we were able to confirm that Chloroflexalean APBs act as secondary colonizers of marine carbonates, illustrate the complex dynamics of cyanobacterial colonization, and define a new clade of likely euendolithic cyanobacteria, highlighting the differences and similarities in succession dynamics between mineral and biogenic carbonates. Our work provides a first look at the complex colonization dynamics that drive bioerosion on these substrates.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-2607/8/2/214/s1, Figure S1: Rarefaction curves of tile communities, Figure S2: PCoA analysis of bacterial and cyanobacterial community composition, Figure S3: Phylogenetic placement of euendolith clusters and colonization dynamics, Figure S4: ASV differential abundance analysis of 3-month and 9-month colonized tiles. Table S1: BIOM tables containing sequences counts for each euendolith cluster at all 3 time points and steady-state climax communities.

**Author Contributions:** Conceptualization, D.R. and F.G.-P.; methodology, D.R.; validation, D.R., and F.G.-P.; formal analysis, D.R.; investigation, D.R.; resources, D.R. and F.G.-P.; data curation, D.R.; writing—original draft preparation, D.R.; writing—review and editing, D.R. and F.G.-P.; visualization, D.R.; supervision, F.G.-P.; project administration, D.R.; funding acquisition, F.G.-P. All authors have read and agree to the published version of the manuscript.

**Funding:** This research was partly funded by the National Science Foundation, gran<sup>t</sup> number EAR 1224939 to F.G.-P.

**Acknowledgments:** We would like to thank A. Garrástazu, S. Velasco Ayuso, and B. Guida for field support.

**Conflicts of Interest:** The authors declare no conflict of interest.
