Genetic Connectivity of Northern Leopard Frogs
In the Prairie Pothole Region
A large adult northern leopard frog (Rana pipiens)
Why do we care about connectivity?
• Maintain genetic diversity through migration between populations
• Habitats can be recolonized if populations are locally extirpated
• Allows spread of new adaptations to cope with changing environments
The Prairie Pothole Region
One of the most wetland-dense regions in the world
Land use dominated by highly productive farmland in the US and Canada
Undergoing rapid land use change
>50% wetland loss in North Dakota
Even higher in other parts of the PPR
A leaping leopard frog
Study Species: The Northern Leopard Frog
Very mobile amphibian species
Can migrate up to 5 km between habitats over the course of a season
Leopard frogs have varying habitat needs over the course of their life cycle
- Breeding wetlands - usually small, temporary or semi-permanent wetland without fish
- Wet meadows - where adult frogs can forage for insects
- Overwintering sites - large, deep wetlands or streams/rivers with flowing water where frogs can escape the North Dakota frost
A leopard frog (top left) and a series of wetlands and meadows showcasing the different habitats a leopard frog needs at various points in its life cycle.
Methods
Site selection
Previous work from our lab showed leopard frogs in North Dakota were genetically clustered by river basin
Chose to focus on Lake Oahe/James River (light purple) basins. This population cluster is spread across areas with highly agricultural land use as well as areas with more grassland habitat.
Selected 28 sites meant to capture a range of nearby land uses.
Collecting Genetic Data
Toe clips from 20 frogs/site
Extract DNA from tissue samples and build Best-RAD (restriction associated digest) libraries - randomized snapshot of the genome
Found ~3000 SNP (single nucleotide polymorphism) markers we can use to genotype the frogs we sampled
Data Analysis: Genetic Diversity
Does the amount of agricultural land use in the surrounding landscape affect frogs' genetic diversity?
Calculate observed heterozygosity and inbreeding coefficient within each site
Results
No clear relationship between genetic diversity and nearby agriculture
Data Analysis: Population Structure
Are there clusters of genetically similar frogs that we can identify?
Multiple clustering methods:
- Discriminant Analysis of Principal Components
- fineRADstructure
- BAPS
Results
6 unique genetic clusters
Genetic clusters are spatially organized
Relatively weak overall structure - most frogs are pretty similar genetically
Data Analysis: Landscape Resistance
Are there elements of the landscape that are affecting gene flow?
Examined 9 landscape variables that could affect frogs' ability to migrate between habitats:
- Land use
- Proportion of years a field was planted with corn (2006-2018)
- Proportion of years a field was planted with soybeans (2006-2018)
- Proportion of years a field was planted with wheat (2006-2018)
- Distance from a permanent wetland (likely overwintering sites)
- Heat Load Index (a measure of direct solar radiation)
- Topographic Wetness Index (a measure of where water accumulates)
- 30-year average number of days with >5mm of precipitation during active season (April to September)
- Topographic Roughness (measures flatness/steepness of surface)
How landscape resistance analysis works:
Calculate pairwise genetic distances among sites (Fst, Jost's D) to quantify how genetically different frogs from different sites are.
Test landscape factors with different combinations of resistance. Higher resistance means that frogs are less likely to move through that type of landscape. We can use this to calculate a "resistance distance" telling us relatively how hard it is for frogs to initiate gene flow between sites.
The simplest resistance surface has a constant value. Here, the resistance distance is equal to the straight-line geographic distance between sites.
Straight-line geographic distance vs genetic distance among sample sites. We can see that geographic distance explains some of the genetic differences, but much of the variation is left unexplained.
For the categorical variables, we can test out different resistance values for each category and find a set of values that does a better job at explaining genetic differences than our straight-line geographic distance model. Check out how land use is translated into resistance values:
How land uses are translated to resistance values. In the land use single variable model, grassland and pasture have very low resistance (greens) while cropland and open water have high resistance (oranges and reds)
For continuous variables, we can transform the data in a number of ways to represent resistance. Often, moderate values for many environmental variables are easiest for an animal to traverse, so we have to transform our data and test out multiple transformations accordingly to see which transformation best matches up with our genetic data.
See how topographic roughness is translated to resistance below:
We see that resistance is lowest in flatter areas (low roughness) and is highest in areas with a lot of elevation change (high roughness).
Transformation for topographic roughness data. Resistance rises quickly when starting from low levels of roughness, but plateaus at high levels of roughness.
Now we can start combining the resistances of different landscape variables to see if multiple landscape variables can better explain our genetic differences than one landscape variable alone.
After we test out many combinations, we start to see a pattern emerge. Most of our best models include a combination of the land use and topographic roughness variables.
We can now take a weighted average of all of our best performing models and create our final resistance surface. This is our best guess to how the landscape might be affecting gene flow based on all of the data we've looked at.
Once the final resistance surface is made we can simulate gene flow like an electric current. The current will flow through the areas of least resistance and reveal where frogs are likely to go when they are migrating between habitats.
See below for how the current map is made from the final resistance map:
