Fig 11. A cluster dendogram for ponds in the mire. Ponds were grouped based on percent composition of different vegetation communities in and around the ponds. Colored boxes indicate the four specific groups. The red line indicates the cutoff line for each group identified. The blue star indicates an outlier pond which belongs in a different group.
Pond Types The 15 ponds in the mire location were split into classes based on both within and surrounding vegetation communities using a clustering technique (Fig 11). Four distinct pond types were identified including open-water, moss dominated, sedge dominated, and lake fed ponds. While open-water and lake-fed ponds had similar within pond vegetation, they differed in outer vegetation composition, which is in part, related to the ponds' proximity to a nearby lake. At times, the lake will overflow into the ponds, changing the pond chemistry substantially. Pond19 was grouped in the lake-fed class, but based on field observations this pond belongs in the open-water class. Which groups each pond falls into is important because pond types are used in later analyses when comparing chemical and physical characteristics.
Fig 12. Principle Components Analysis for physical and chemical pond variables at each location. See text for the specific vector label names.
Relationships between variables across the sites Principle component analyses for all three locations showed varying relationships (Fig 12). While water depth (WDEPTH) is inversely related to methane concentrations (CH4) in the tundra and larch systems, there is no observable relationship in the mire (Fig 12 a,c). Similarly, dissolved oxygen (DO) shows an inverse relationship with carbon dioxide (CO2) in the mire and larch locations, but not in the tundra (Fig 12 b,c). Furthermore, soil organic carbon content (represented as CLOST) has a positive relationship with CH4 in the mire and larch systems(Fig 12 b,c), but was not measured in the tundra. This suggests that while standard physical and chemical like water depth and dissolved oxygen can influence concentrations of CH4 and CO2, they may not always be the most important factors to consider.
The effect of fire and vegetation While there does not appear to be a consistent relationship between a common physical or chemical variable and dissolved CH4 or CO2, what is apparent is the distinct separation of groups in both the mire and the larch study sites. For example, in the larch ponds, there is a clear separation between the unburned ponds and the burned ponds (Fig 12c). Statistical comparisons between the groups confirmed that there is a difference between at least one of the groups (perManova; p=0.03). Upon post-hoc analysis, we confirmed that there is a significant difference between the unburned ponds and the low/high severity ponds, but not between the low and high severity ponds (TukeyHSD; p=0.001 and p=0.07, respectively). Similarly, there is also a separation of ponds in the mire according to pond type. Ponds defined as moss dominated and sedge dominated appear to be different from both each other and open-water ponds (Fig 12b). The distinct separation of ponds types in both the larch and mire sites suggests that while physical and chemical variables govern some influence on CH4 and CO2 concentrations, the secondary influences of vegetation composition and fire disturbance may be a more important indicator.
Focusing on methane: An univariate approach Our results suggest that pond types (ie. disturbances) not only govern the chemical characteristics of ponds, but more specifically, may control concentrations of CH4 and CO2. For example, unburned ponds had substantially higher methane concentrations than pond in both burn sites (Fig 13; one-way ANOVA, p=0.001). This is most likely driven by higher quantities of organic matter in the surrounding soils and sediments in the unburned sites. In the burn sites, surface vegetation and organic matter was lost in the fire by combustion , leaving behind more recalcitrant carbon which is less available for microbial use, leading to less CH4 production.
Fig 13. A boxplot representing the dissolved methane concentrations in the ponds in the larch location. Concentrations in the unburned pond are significantly higher than concentrations in both burn sites (ANOVA, p =0.001)
In the mire, open-water ponds had the highest concentrations of CH4 (represented as "Flux" in Fig. 14), while moss dominated ponds had little to no CH4 (Repeated measures ANOVA; Table 1). Open- water ponds had little to no vegetation at the surface or on the bottom and were surrounded by eroding banks, characteristics associated with high CH4 fluxes from ponds in northern Canada (Laurion et al. 2010). The relatively high fluxes from open water ponds could be a result of differences in gas exchange velocities, relatively low photosynthetic CO2 fixation, and high terrestrial input of C and organic matter from adjacent thawing and eroding soils as compared to moss and sedge dominated ponds. Of note, while only CH4 is represented below, CO2 shows similar patterns and relationships.
Fig 14. Methane fluxes for each pond type in the mire over the course of one field season. Fluxes were calculated using a wind based equation that is closely tied to methane concentration. Fluxes from each pond types were compared using repeated measures ANOVA (Table to the left).