Research Methods
1. Study Area and Sample Collection
Surface sediment samples were provided by Dr. Anne de Vernal (Université du Québec à Montréal, GEOTOP). Surface sediments were collected between August 27th and September 23rd 2008 in Labrador Sea, Davis Strait, Baffin Bay, North Water Polynya, Jones Sound, and Lancaster Sound (Figure 5) using a box core.
Fig 5. Map of study area showing station numbers for
each box core location.
each box core location.
2. Sample Preparation
Surface sediments were cleaned to obtain a siliceous slurry from which diatoms can be counted. The technique used HCl to remove carbonates from the slurry and H2O2 to remove organic matter, leaving only siliceous material in the slurry. This is the general technique used to clean diatoms from sediment. Siliceous slurry were then dried on coverslips (strewn mounts) and mounted using Naphrax, a high resolution mounting medium.
3. Diatom Identification and Counting
Three hundred to four hundred diatoms were counted and identified from each sample (excluding Chaetoceros sp. spores) under 1000X on a Leica light microscope (following Williams, 1990). Each frustrule was counted as one diatom. To be counted, at least half of the diatom frustrule was present. Scanning electron microscopy was used to supplement the identification of diatom species. Various resources were used to identify the diatoms including Cleve (1965), Priddle and Fryxell (1985), Medlin and Priddle (1990), Hasle and Syversten (1997), as well as relevant taxonomy journals.
Figure 6. Scanning electron micrographs showing two diatom species found in the arctic surface sediment samples (left: centric diatom; right: pennate diatom).
4. Environmental Data
Monthly circum-arctic sea ice concentrations from 1979-2008 were provided by Dr. Christian Haas (University of Alberta). Using ArcMap, the fifteen stations (from which surface sediments were collected) were sampled for monthly sea ice concentration data from 1979-2008.
Figure 7 (right). Circum-arctic sea ice concentration map. The shades between black-white represent sea ice concentrations based on a 25X25 Km grid.
5. Multivariate Statistical Analysis
Three different multivariate statistical techniques were used to analyze both the species and environmental data. The ultimate goal is to determine the relationship between the diatom assemblages and the monthly sea ice concentrations. Another variable is the latitude, which is based on the station ID number. All analyses were performed using the statistical software package 'R'.
Hierarchical Cluster Analysis
A hierarchical cluster analysis was chosen to group stations (where surface sediments were collected) by diatom species. This method classifies the stations based on the "distance" of one station to another. The "distance" is calculated based on the Bray-Curtis distance matrix. In the context of this study, the distance between one station and another is based on the similarity of diatom species found in the surface sediments. If there is a high similarity of diatom species, then the distance between the stations is lower, and therefore they will tend to be grouped closer together.
Multiple Regression Tree (MRT)
A multiple regression tree is generally used to determine the relationship between predictor variables and response variables. MRT analysis groups variables in a way that results in the minimum variability within the group. In this study the predictor variables are the mean monthly sea ice concentrations, while the response variables are the diatom species presence/abundance.
Non-linear Multidimensional Scaling (NMDS)
A direct gradient analysis using nonmetric multidimensional scaling was used to determine the effects of monthly sea ice concentrations on diatom species found in surface sediments. NMDS is an ordination technique. NMDS was chosen for this study because it can work with non-normalized data, as is the case with the data in this study. Multidimensional scaling requires the use of a distance matrix. For this study a Bray-Curtis distance matrix was used.
Hierarchical Cluster Analysis
A hierarchical cluster analysis was chosen to group stations (where surface sediments were collected) by diatom species. This method classifies the stations based on the "distance" of one station to another. The "distance" is calculated based on the Bray-Curtis distance matrix. In the context of this study, the distance between one station and another is based on the similarity of diatom species found in the surface sediments. If there is a high similarity of diatom species, then the distance between the stations is lower, and therefore they will tend to be grouped closer together.
Multiple Regression Tree (MRT)
A multiple regression tree is generally used to determine the relationship between predictor variables and response variables. MRT analysis groups variables in a way that results in the minimum variability within the group. In this study the predictor variables are the mean monthly sea ice concentrations, while the response variables are the diatom species presence/abundance.
Non-linear Multidimensional Scaling (NMDS)
A direct gradient analysis using nonmetric multidimensional scaling was used to determine the effects of monthly sea ice concentrations on diatom species found in surface sediments. NMDS is an ordination technique. NMDS was chosen for this study because it can work with non-normalized data, as is the case with the data in this study. Multidimensional scaling requires the use of a distance matrix. For this study a Bray-Curtis distance matrix was used.
