Effect of media and heterotrophs on autotroph
Many phylogenetically diverse heterotrophic bacteria have been isolated and identified from environments dominated by cyanobacteria (e.g, cyanobacterial blooms) . In mixed cultures, the interaction between cyanobacteria and other bacteria can vary from mutualistic (mostly connected with nutrient exchanges and cycling)  to growth-inhibition or cell-lysis of one of the actors.
Cultivation conditions are a very important issue affecting the outcome of these interactions. From our preliminary experiment, SM/BG11 broth showed the least adverse influence on Nostoc growth from the various broths tested (including media with carbon sources, data not shown). However, when adding SM to cyanobacteria, their growth was negatively affected slightly for a few initial days due to the new condition. But later, as shown by the second measurement (day 7), the Nostoc cultures appeared to adapt to this cocktail and were growing even better than in pure BG11 (Figure 1a). Indeed, according the literature, Nostoc muscorum can grow heterotrophically if glucose is present ; as is the case in SM medium.
BUZ 2 had a strong significant negative effect on cyanobacterial growth under nutrient rich conditions (SM/BG11). However it had no effect under nutrient poor conditions (SM washed out with BG11). One explanation could be that BUZ 2 first fed on SM and used trace elements, iron and nitrate from BG11, reached high cell numbers, and after finishing the carbon sources, started to attack the cyanobacterial filaments. Hence the interaction could start as nutrient and space competition and later switch to predation. In washed cultures (nutrient poor), Fibrella was not able to reach a density of cells which would have allowed them to destroy cyanobacterial filaments. Indeed, the ratio between pathogen (or predator) and host (or prey) is an important factor in natural pathogenicity (or predation) and also species dependent . The ratios of BUZ 2 to Nostoc cells were very different in SM/BG11 in comparison to washed cultures in BG11 media. In SM/BG11, the cell ratios progressed from 1:1 (starting point), to 241:1 (3 d), 314:1 (6 d), 17:1 (11 d) and 1:2 (21 d). Conversely, in nutrient poor BG11 medium, the ratios favoured Nostoc with values 1:1, 1:7, 1:12 and 1:18218. Therefore, by varying the ratio in favor of BUZ 2, it could be still possible that the heterotroph may harm the cyanobacterium under nutrient poor conditions, but this remains to be tested.
Fibrisoma sp. strain BUZ 3 had no significant effect on the autotroph. Microscopic pictures (Figure 3 c and f) suggest physical contact between BUZ 3 and cyanobacterial cells, but cell numbers of BUZ 3 might have been too low to significantly harm cyanobacterial growth. Indeed, the ratios of BUZ 3 to Nostoc (maximal ratio was 24:1 after seven days) did not reach as high as in the BUZ 2 case. This might be due to their low growth rate (in comparison to BUZ 2) or to their physiological characteristics. In washed cultures in BG11, all of the ratios were tilted towards higher numbers of cyanobacteria (lowest cell ratio 1:3120 in the last measurement).
Negative effects of heterotrophic bacteria on cyanobacterial growth could be caused via different modes of action, including nutrient and space competition, production of secondary metabolites in the close vicinity of entrapped cyanobacteria (so called indirect way) and/or contact lysis and parasitism (direct way). The observed physical interaction between the filaments of the heterotrophs and the autotroph possibly imply either contact lysis or entrapment with production of extracellular compounds. This is in agreement with present literature, where the five members of the order Sphingobacteriales showed similar mechanisms when tested on filamentous or non-filamentous cyanobacteria [2–4]. Interestingly, extracts of both our tested heterotrophs had no effect on cyanobacteria. However, it should be noted that we had filtered the extract from heterotrophs that had been grown in the absence of cyanobacteria (in pure cultures). The possibility remained that production of harmful substances may depend on the presence of the prey (in our case cyanobacteria), or could be released only on the cell surface of the prey. This possibility was tested using the supernatant from mixed cultures of BUZ 2 and Nostoc cultivated together for two days. This supernatant displayed negative effects on cyanobacterial growth after three days. In comparison, Sallal  showed that Flexibacter flexilis adsorbed to Oscillatoria williamsii and then released a lysozyme lysing the cyanobacterium. Furthermore, extracellular metabolites in the cell filtrates (from pure bacterial cultures) led to growth inhibition. Rashidan and Bird  analyzed the mode of action of two Cytophaga strains on cyanobacteria. Given the fact that lysing enzymes were not secreted into the medium (extracts gained from pure and mixed cultures as well), and it appeared that no special attachment organelles exist on the surface of Cytophaga, it seems likely that surface lytic enzymes were involved in the lytic action . The lytic bacteria also inhibited photosynthesis, glycolate dehydrogenase and nitrogenase. The wide host range of tested Flexibacter (which includes Nostoc muscorum) indicates the possibility of using those bacteria in controlling cyanobacterial growth .
A novel strain of filamentous helical Saprospira did not produced cyanobactericidal compounds into medium but showed group behaviour when cultivated together with filamentous Anabaena sp . It formed three-dimensional reticular structures and lysed the host through direct contact. Furthermore, its group behavior greatly accelerated the cyanobactericidal process in comparison to individual filaments. Therefore a relevant question arises: do filamentous heterotrophs mainly act as a bundle of filaments or as single filaments? Unfortunately, only Shi et al.  explicitly mentioned that the Saprospira strain is filamentous and that it builds up such group structures. Figure 3e is an example indicating that the tested heterotrophs could aggregate and cover parts of cyanobacterial filaments. It is not known how often this happens, and whether this kind of behavior is representative for the tested heterotrophs. Further experiments could shed light on the occurrence and mechanism of this mode of action. Concerning other closely related taxa, Flexibacter flexilis, Cytophaga sp. (according Reichenbach ) and Chytinophaga sancti should be filamentous. However, this could not be seen it the pictures of strains tested by Sallal  nor by Rashidan and Bird . Therefore, additional studies are necessary. The appropriate use of electron microscopy and atomic force microscopy, as done by Malfatti and Azam  could clarify this issue.
Effects of media and phototroph on heterotrophs
Our experimental set up allowed us to check the densities and survival of heterotrophs in different situations [see Additional file 1]. This issue is often not reported in lytic studies of cyanobacteria. We detected better growth of both heterotrophs in the mixture of SM/BG11 than in SM or BG11 alone (Figure 4). This could be expected as BG11 contains nitrate, iron and trace elements such as B, Mo, Co, Cu and Mn which are very important for heterotrophic metabolism. Furthermore, our tested heterotrophs could use cyanobacterial cells and cyanobacterial supernatant as nutrient sources (Figure 4). However, in the presence of Nostoc, the outcome of competition can depend on ratios between heterotroph and autotroph. Both heterotrophs could survive well when washed in BG11 (nutrient poor) during the whole experiment (Figure 4c, d). This was not always true when Nostoc was present. It is possible that Nostoc uses some strategy to out-compete the heterotrophs. An example are some strains of Nostoc muscorum that produce toxins which kill heterotrophs . However, it is not known whether the strain we use produces toxins.
Our results suggest that the outcome of interaction between filamentous heterotroph and autotroph depends on the presence of nutrient in cultivation media, which boost or reduce the numbers of heterotrophs and hence change the ratios between the actors. To understand these interactions, additional research is needed. In particular, it is necessary to shed light unto the mode of action for lysis by heterotrophs and the possible defence mechanisms of the autotrophs.