The influence of environmental change on genetic diversity across spatial and
taxonomic scales

---

Connor French, PhD Candidate

Pre-defense seminar, 2024-06-04

The Graduate Center

City University of New York

The Earth is constantly changing

The Earth is constantly changing, over short and long time scales. Here I’m showing a gif of temperature change since the last glacial maximum, 21,000 years ago. The line plot shows the mean temperature over time, with beginning on the right at 21,000 years and moving towards the present. New York has gotten a lot hotter since the last ice age! Glaciers covered New York until 10,000 years ago. Thinking about the squirrels and pigeons, they couldn’t survive on ice. So they moved out of the neighborhood, then came back in better climates. Climate fluctuations like this happen all over the world.

Populations shrink or expand…

Fragment and reconnect…

Adapt or go extinct

They shrink or expand, fluctuating in size from the past to the present. Here, I’m representing populations as amorphous blobs that correspond with their range boundaries. Populations can fragment and reconnect, leading to isolation and/or reconnection between populations. Populations can adapt to new environments or go extinct. The background environment in this figure changes from green to orange, and the portion of the population that can’t adapt goes extinct!

Genetic diversity contains signatures of that change

Genetic diversity is the variation in DNA sequences within a population. It records the history of a population’s past. In this visualization, we see each individual, represented as circles, in a population with their genealogy traced back in time. The length of those lineages represent how much genetic diversity has accumulated over time. When I play the video, you’ll see the population change over time, with individuals being born and dying. The colors represent different genetic lineages. This is a simplified example, but it illustrates how genetic diversity can change over time.

In general, the higher the genetic diversity the better, as it can provide a population with the raw material to adapt to changing environments.

Genetic diversity can tell stories about

Species

and Communities

¹

The story of a species can be told through the genetic diversity of its populations. The plot on the left shows the “effective population size” of a species of frog over time, which is a measure of genetic diversity. The x-axis is time before present, and although the pattern is noisy, we can see that the population experienced a recent expansion in the last 20,000 or so years. The story of the community can be told as well. Each of the red lines in this plot represents a lizard species in the northeastern United States. We can see that some common event led to the expansion of this community of lizards around 100,000 years ago!

1. Burbrink et al. 2016. Eco. Letters 10.1111/ele.12695

Understanding these stories is especially urgent

Climate change is happening rapidly, and it’s important to understand how these changes are impacting the genetic diversity of species and communities that are most vulnerable to shifts in their environment, at both the regional and global scales. Certain classes of animals will be particularly sensitive to changes in their environment.

Understanding these stories is especially urgent

Ectotherms are linked to their environments

Ectotherms, which are animals that cannot regulate their own body temperature, may be sensitive to fluctuations in their environment over time. This collared lizard on the left needs regular sun to keep warm and the dragonfly on the right needs a stable water source to lay its eggs. With the climate shifting rapidly, it’s important to understand how these changes are impacting the genetic diversity of species and communities that are most vulnerable to shifts in their environment.

I investigate global and regional patterns of genetic diversity in two groups of ectotherms, insects and lizards, to understand the relationship between environmental change and genetic diversity, from populations to assemblages.

My questions

1. What is the relationship between genetic diversity and environmental change in insects?

2. What can genetic diversity convey about the processes underlying species responses to environmental change?

3. How can I make integrative modeling of species responses to environmental change more accessible?

Some of my friends and family might notice that there aren’t frogs up here. Welp, a PhD is a journey that can take you in unexpected directions, and while the pandemic created circumstances that made it difficult to collect data on frogs, it led to three awesome projects that I wouldn’t have dreamed up otherwise and I’m super excited to share them with you today. There is a little frog treat at the end too.

Global determinants of insect mitochondrial genetic diversity

Ch. 1. What is the relationship between genetic diversity and environmental change in insects?

A global perspective is necessary

A global perspective is necessary for understanding broad biodiversity patterns and at-risk regions. What I mean by biodiversity is the variety of life on Earth, including the diversity of species, genes, and ecosystems. This figure shows the distribution of mammal species richness across the globe. The darker the color, the more species are present in that region. We can see that the tropics are the most diverse regions on Earth for mammals, and contain the most threatened species.

Insects comprise over 93% of the planet’s described animal diversity

Estimated number of insect species is even higher: 5.5 million! Stork 2018. Ann. Rev. Entomol. 10.1146/annurev-ento-020117-043348

Global biodiversity patterns remain undocumented for most insects

• Scientists have relied on species richness or species diversity
• Describing and cataloging the world’s insect diversity is a monumental task
• What do we do?

Genetic diversity is a promising biodiversity metric

• It does not rely on species identity for calculation
• It conveys historical information
• macrogenetics is an emerging field

Some predictions follow those from vertebrates

Latitudinal diversity gradient ¹

Human disturbance ²

Climate stability ³

1. Hillebrand 2004. Am. Nat. 10.1086/381004

2. Chapin et al. 2000. Nature. 10.1038/35012241

3. Hewitt 2000. Nature. 10.1038/35016000

I compiled the largest animal macrogenetic dataset to date

• 2,415,415 mtDNA sequences ¹
• 98,417 operational taxonomic units (OTUs) ²

For the purposes of this talk, we can think of OTUs as a proxy for species, although there are important differences that matter depending on the context of the study.

1. Barcode of Life Data System (BOLD). https://boldsystems.org/

2. OTUs are a proxy for species

I compiled the largest animal macrogenetic dataset to date

Grid cells had a 193 km by 193 km resolution and we summarized the genetic diversity for all OTUs (that is, species) in each sampled grid cell. I only retained grid cells that had at least 100 individuals for the final analysis, although I explored multiple resolutions and thresholds and ensured that results were robust across these choices.

Summarizing genetic diversity

For each OTU, I calculated the mean number of pairwise differences between sequences

Genetic diversity mean (GDM) = the average genetic diversity among all OTUs

Genetic diversity evenness (GDE) = the evenness of genetic diversity across OTUs

High GDE -> more OTUs with a similar level of genetic diversity
Low GDE -> OTUs have very different levels of genetic diversity

• Many pairwise differences equals higher genetic diversity for that OTU. But then we summarize them across all OTUs in a grid cell to get a measure of genetic diversity for that grid cell’s insect assemblage.
• While GDE can mean a lot of things depending on the context, in this case a high GDE means there aren’t one or two OTUs that dominate the assemblage, leading to higher overall diversity. GDM and GDE can be interpreted in tandem to give a holistic picture of the genetic diversity of an assemblage.

Responses

• GDE
• GDM

Predictors

• Latitude
• Climate
• Climate stability
• Human influence
• Habitat heterogeneity
• Topography

While I’m not going to go into the details of the model, I used a spatially-explicit linear model to estimate the relationship between genetic diversity (show the two metrics) and these predictors (show the predictors), which were chosen to correspond with my previously stated predictions and include other possible environmental variation. For the sake of time going forward I am focusing on the results for GDE, which was correlated with GDM, but seemed more robust to sampling variation and was more interpretable.

Global maps of insect genetic diversity evenness

GDE peaks in the subtropics

Prediction

GDE peaks in the subtropics

Observation

The subtropics are regions that are just outside the tropics, between 23 and 35 degrees latitude, and are characterized by warm temperatures and seasonal climate variation. This pattern is distinct from vertebrate patterns, which show a peak in the tropics.

GDE peaks in the subtropics

• Rapoport’s rule: species ranges are larger at higher latitudes ¹
• Genetic diversity tends to increase with range size ²
• Pattern breaks down in previously glaciated regions

1. Stevens 1989. Am. Nat. https://doi.org/10.1086/284913

2. Exposito-Alonso et al. 2022. Science. 10.1126/science.abn5642

GDE is correlated with current climate and climate stability

Mention the lack of human influence and habitat heterogeneity

GDE is correlated with current climate and climate stability

• Climate stability hypothesis: genetic diversity increases with climate stability ¹
• Evolutionary speed hypothesis: genetic diversity increases with temperature ²
• Insect diapause provides an adaptive tolerance to seasonal temperature variation ³

Evolutionary speed hypothesis- similar patterns in fish, animals, and plants

1. Hewitt 2000. Nature. 10.1038/35016000

2. Oppold et al. 2016. Proc. Biol. Sci. 10.1098/rspb.2015.2413

3. Addo-Bediako et al. 2000. Proc. Biol. Sci. 10.1098/rspb.2000.1065

To sum up

• GDE peaks in the subtropics and is distinct from vertebrate patterns
• Environmental stability leads to higher levels of genetic diversity
• Seasonally warm temperatures suggest the evolutionary speed hypothesis may be at play

What is the relationship between genetic diversity and environmental change in insects?

Demographic responses to past climate change in Brazilian Enyalius lizards using spatially-explicit coalescent modeling

Ch. 2. What can genetic diversity convey about the processes underlying species responses to environmental change?

Macro-scale patterns are interesting…

but what about processes?

The processes change over time and space, and uncovering them can help establish a causal link between environmental change and genetic diversity.

Species are expected to respond differently to environmental change

¹

Emerging evidence indicating that ectotherms from higher elevations could be more resilient due to exhibiting wider environmental tolerances than their lowland counterparts. Explain elevational midpoint, operative warming tolerance, and thermal breadth.

1. von May et al. 2017. Eco. Evo. 10.1002/ece3.2929

Species distribution models (SDMs) are useful tools

Species distribution models (SDMs) are useful tools to investigate the mechanisms underlying these expectations - quick illustration of SDMs and what they predict - State that the probability is often translated into other ecological variables, like abundance

The relationship between SDM suitabilities and abundance

• While a probability may reliably predict presence, it may not linearly predict abundance
• Wider environmental tolerances may lead to higher abundances than predicted by a linear relationship in marginal habitats

SDM suitabilities may have a non-linear relationship with abundance in high-elevation species

I investigate this relationship in two lizard species

Integrative distributional, demographic, and coalescent (iDDC) modeling

iDDC models establish a link from SDM suitability to abundance to genetic diversity

A deme is like a local population. I don’t have abundance data, so I’m using iDDC modeling to link abundance with genetic diversity

Integrative distributional, demographic, and coalescent (iDDC) modeling

Distributional

• SDMs predict the distribution of suitable habitat

Demographic

• SDMs are transformed into demographic models
  • local deme size and dispersal

Coalescent

• Coalescent modeling simulates the genealogy of a sample of DNA sequences backwards in time

A deme is like a local population and its size is analogous to its local abundance.
Coalescent modeling is a type of population genetics model that simulates the genealogy of a sample of DNA sequences backwards in time to infer the demographic history of a population.

This SDM is from the high-elevation species E. iheringii, in the southern Atlantic Forest.

Predictions

Predictions

I used simulations and machine learning to test these hypotheses

Local deme size?
Migration rate?

A threshold transformation is more likely for the high-elevation species

Predictions

A threshold transformation is more likely for the high-elevation species

Observations

Migration is estimated to be higher for the high-elevation species

Predictions

Migration is estimated to be higher for the high-elevation species

Observations

Janzen’s mountain pass hypothesis

Maps of genetic diversity correspond with environmental stability in E. iheringii

The dots on these maps are sampled localities used for species distribution modeling and genomic sequencing. The low-elevation species, E. catenatus, may be sensitive to seasonal or other environmental fluctuations not captured by our models, maintaining higher genetic diversity at their range core.

To sum it up

• Considering the relationship between predicted SDM suitability and local deme size/abundance is important in iDDC models
• High-elevation species are more strongly associated with a threshold transformation compared to low-elevation species and they have higher migration
• This supports the prediction that high-elevation species have wider environmental tolerances than low-elevation species

What can genetic diversity convey about the processes underlying species responses to environmental change?

a Python package to facilitate spatially-explicit coalescent models in msprime

Ch. 3. How can I make integrative modeling of species responses to environmental change more accessible?

Simulating these complex scenarios is difficult

Completing these simulations for my last chapter involved a TON of code and many hours of debugging code that wasn’t doing quite what I expected. In conversations with other scientists who are interested in similar types of analyses and trying out other software that got close to what I wanted, but not quite, I decided to put together my own Python software package, called spaceprime. Python is a popular, general-purpose coding language used in science and many other domains. But why’s my package called spaceprime?

space

Take spatial models of habitat suitability (i.e. SDMs) projected through time

prime

Simulate the genealogy of a sample of DNA sequences backwards in time with

msprime is a Python package that is extremely efficient at performing coalescent simulations. The user interface for msprime is very flexible, but it can be daunting for users to set up the extremely large simulations that are required for iDDC modeling. spaceprime is a package that wraps around msprime and provides a more user-friendly interface for spatially-explicit coalescent simulations.

spaceprime limits coding errors and streamlines analysis

spaceprime: 174,446 events

projections = rasterio.open("data/projections_agg.tif")
locs = gpd.read_file("data/distincta_localities_ex.geojson")
demes = sp.raster_to_demes(projections, transformation="linear", max_local_size=1000)
d = sp.spDemography()
d.stepping_stone_2d(demes, rate=0.001, scale=True, timesteps=1000)
d.add_ancestral_populations(anc_sizes = [100000], merge_time = 23000)
sample_dict, _, _ = sp.coords_to_sample_dict(projections, locs)
ts = msprime.sim_ancestry(demography=d, samples=sample_dict, sequence_length=1e6, recombination_rate=1e-9, random_seed=43, record_provenance=False)
ts = msprime.sim_mutations(ts, rate=1e-8, random_seed=99)

msprime: 4 events

demography = msprime.Demography()
demography.add_population(name="A", initial_size=10_000)
demography.add_population(name="B", initial_size=5_000)
demography.add_population(name="C", initial_size=1_000)
demography.add_population_split(time=1000, derived=["A", "B"], ancestral="C")
demography.set_symmetric_migration_rate(populations=["A", "B"], rate=0.1)
ts = msprime.sim_ancestry(samples={"A": 2, "B": 2}, demography=demography, random_seed=12)
ts = msprime.sim_mutations(ts, rate=3e-4, random_seed = 42)

Simulate thousands of populations in the same amount of code that it takes to simulate three.

With spaceprime you can

• Set up simulations
• Visualize your models
• Analyze your simulation and empirical data
• Run simulations in parallel on a cluster
• Execute these simulations fast
  • < 1 hour instead of days

The spaceprime website (will) contain

The spaceprime website (will) contain

• Documentation of all functions

• A quickstart guide and worked example(s)
  

• A guide for R users

• FAQ and other resources

More examples and tutorials are in the works, and I’m always looking for feedback on how to make the package more user-friendly and useful for the community.

https://connor-french.github.io/spaceprime/

In conclusion

• spaceprime is convenient
• spaceprime is fast
• spaceprime is accessible
• Help me create a logo and sticker!

How can I make integrative modeling of species responses to environmental change more accessible?

The influence of the environment on genetic diversity varies across scales

1. What is the relationship between genetic diversity and environmental change in insects?

• GDE peaks in the subtropics and is distinct from vertebrate patterns

2. What can genetic diversity convey about the processes underlying species responses to environmental change?

• High-elevation lizard species may have broader environmental tolerances than low-elevation species

3. How can I make integrative modeling of species responses to environmental change more accessible?

• With spaceprime

Thank you! Questions?