We found evidence for three different patterns of algae that occur in the hair of five sloth species: 1) The green alga in the fur of the brown-throated sloth, Bradypus variegatus, and the pygmy three-toed sloth, B. pygmaeus, is a unique species and no other green algal species were found in their fur. Microscopy of the alga on the hair revealed characteristic features to the description of Trichophilus (Weber-van Bosse 1887), which has indeed been described from the hair of Bradypus. 2) The maned three-toed sloth Bradypus torquatus was shown in our study to host a variety of algae belonging to genera known to be terrestrial, e.g. Trentepholia and Myrmecia. 3) The Hoffmann's two-toed sloth Choloepus hoffmanni and the pale-throated sloth Bradypus tridactylus, showed both patterns; they hosted terrestrial green algae from their surroundings, as well as the unique genus Trichophilus.
The green alga Trichophilus that was found in B. variegatus, B. pygmaeus and B. tridactylus was separated in the phylogenetic trees from the Trichophilus occurring in C. hoffmanni. The Bradypus-clade was supported by high bootstrap values and as Bradypus variegatus and Choloepus hoffmanni inhabit the same geographic area and habitat , the separation of Trichophilus inhabiting Bradypus suggests co-evolution between Trichophilus and Bradypus species. This is further supported by the cell sizes of Trichophilus in Bradypus and Trichophilus in Choloepus being different.
Trichophilus welckeri  has been described from sloth hair and generally has been thought to be the only green alga infesting sloth hair [6, 9–11]. However, in former phylogenetic studies [10, 11]T. welckeri (strain SAG 84.81) was shown to group with Prasiolopsis ramosa, which is contrary to our study. We suspect earlier studies to be erroneous species identification, made because of its habitat and the morphological investigation of the strain supports this. The strain SAG 84.81 resembles Prasiolopsis by forming irregular cushion-like tufts of pseudoparanchymatous filaments and cells with single, stellate, pyrenoid-containing chloroplast typical of the Prasiolales  rather than numerous small, discoid chloroplasts without pyrenoids described for Trichophilus . Therefore our understanding is that Trichophilus welckeri has been found previously in nature [6, 8, 9, 14] but has not yet been isolated into pure culture and sequenced before. It is likely not to occur in any other environment besides sloth hair given that tropical terrestrial green algal species like T. welckeri have not been found before. Its restricted occurrence may be explained by it being host-specific and possibly thermophilic.
Among all 71 sloth individuals studied, 73% hosted algae in their fur. It was either observed by the greenish color of the fur or seen under the microscope in closer observation. Seven of the 19 individuals in which we found no evidence of algae were babies still at the age of clinging to their mothers suggesting that at least Trichophilus is gained in childhood, most likely from the mother. This observation is supported by an earlier study noting that sloths gain the algae and other parasites by the time they are a few weeks old . Besides Trichophilus, other green algae found in sloth hair were terrestrial Trentepohlia, Prinzina, Myrmecia and Trebouxia -species as well as Chlorella-like algae. Among these, only Trentepohlia has been reported previously from sloth hair . It is possible that sloths gain these algae from the environment and host them in their fur. However, they may also represent sloth-inhabiting specialists at a finer evolutionary level, e.g. species level, than Trichophilus does. That there were 25 hair samples containing visible green algae but for which we were unable to get PCR-product, suggest that there were either too few algae in the sample or there were PCR inhibitors present in the DNA extracts. This problem could be overcome by culturing the microbial community from fresh hair samples. The diversity of other eukaryotic organisms in the hair was high, as expected, based on previous studies, which have observed a wide range of animals, e.g. moths, beetles, cockroaches, and roundworms . However, a surprising number and diversity of ciliates and fungi were also found in the hair of all sloth species, excluding C. didactylus, for which only two sequences were analyzed. All geographical areas studied, except Costa Rica, had similar levels of eukaryote diversity. The similar diversity is likely to be explained by the similar habitat of sloths, regardless of the geographical area they inhabit. However, regardless of the similar diversity, the species composition of eukaryotic organisms varied as was seen in the PCO and CAP-analyses (Fig. 3). The eukaryotes in B. variegatus, B. torquatus and C. hoffmanni were different (Fig. 3). Partly this separation can be explained by the differing green algae inhabiting the hair as was seen in the phylogenetic tree in which Trichophilus in Bradypus was separated from Trichophilus in Choloepus (Fig. 4). Similarly, ciliates were different on the three compared species. This can be due to differing hair structure and possibly chemistry as well as differing ecology of the species. It may also reflect the divergence of the two sloth genera about 20 million years ago , which may have led to the separation of the cohabiting eukaryotes as well.
The species composition of Naos Island C. hoffmanni was clearly different from other Choloepus samples from Panama. This may be due to the environment where the sloths are living and the fact that samples were collected on dry season. Naos Island is a small and dry island on the Pacific coast of Panama, receiving on average 1800 mm rainfall per year . Barro Colorado Island and Isla Colón, are located closer to the Atlantic coast of Panama, and therefore are moister, receiving rainfall of 2600-3300 mm per year .
The difference in the species composition in Barro Colorado Island Choloepus and Bradypus was striking (Fig. 3a) as the samples were collected from the same area, although the majority of Bradypus sequences were obtained from sloths on dry season. In contrary, Choloepus and Bradypus samples from Isla Colón were collected from the same area both on wet season being close to each other in the PCO-analyses (Fig. 3a). This suggests that the alteration of dry and wet season may have an effect on the eukaryotic species composition in the hair of sloths.
Despite the differing species compositions on sloth hair, the results (Fig. 2a) show that besides primary producers (27% algae), there are also heterotrophic organisms (52% ciliates, apicomplexans and dinoflagellates) as well as decomposers (8% fungi) in the fur, suggesting that sloth fur supports a versatile small-scale ecosystem. This was not seen in the environmental samples (Fig. 2b) in which 87% of the organisms are primary producers (algae or plants) or decomposers (fungi).