Giorgio Nicoletti

Check out our recent paper just out in Physical Review E!
journals.aps.org/pre/abstract...
with Alice Doimo, @giorgionicoletti.bsky.social, Davide Bernardi
We look into stochasticity arising from finite carrying capacity in classical metapopulation models to understand survival statistics.

Finite-size scaling of survival statistics in metapopulation models link.aps.org/doi/10.1103/...

Happy to be part of the global ecology starter pack!

go.bsky.app/F86HYqj

Landscape and environmental heterogeneity support coexistence in competitive metacommunities | PNAS Metapopulation models have been instrumental in quantifying the ecological impact of landscape structure on the survival of a focal species. Howeve...

Thanks for sticking around! If you want to know more, check out our paper in PNAS and stay tuned for even more results about spatial ecosystems!

www.pnas.org/doi/abs/10.1...

We believe that our framework will enable the systematic study of how landscape structure shapes the dynamics of interacting ecological niches - a crucial point, because a sufficient level of landscape heterogeneity is essential for sustaining biodiversity in spatially extended ecosystems

9/9

Finally, we connect landscape and habitat heterogeneity by introducing spatial correlation between nearby habitats, using models of aquatic and terrestrial landscapes. Correlations increase the total population and strengthen the buffering effect due to the spatial clustering of niches

8/9

What is going on? The crucial idea is that the spatial structure helps to buffer competition, increasing diversity and population - species that thrive in one landscape may go extinct in another. Hence, species coexistence is fundamentally entangled with the spatial structure of the ecosystem

7/9

Then, we add landscape heterogeneity by introducing complex dispersal networks! We find that the fraction of coexisting species increases with habitat heterogeneity, and at the same time landscape heterogeneity allows for more species to coexist while increasing the total population

6/9

We first study a "mean-field case" where all habitat patches are connected to focus on the effect of habitat heterogeneity alone. If it's large enough, coexistence of all species becomes possible thanks to the emergence of ecological niches in different habitats - species start to "localize"

5/9

And this is where we use the physics of disordered systems! We obtain exact results for an ecosystem where the landscape-mediated fitness is random and follows a generic probability distribution: a spatially disordered landscape where we can tune its spatial heterogeneity and see its effects

4/9

To quantify habitat heterogeneity, we measure the balance between colonization and extinction via a local “landscape-mediated fitness”: species with large colonization and low extinction rates in a patch have high fitness, and vice-versa. This is akin to the classical metapopulation capacity

3/9

Without heterogeneity, when all habitats are equivalent, we find that the only possible solution is monodominance - the fittest species overtakes the entire ecosystem, regardless of the underlying dispersal network! But what happens if not all habitats are equally good for different species?

2/9

We start by generalizing the seminal metapopulation model by Hanski and Ovaskainen. Species explore and settle on a complex landscape described by a network of habitat patches. Each habitat has a finite amount of colonizable space, generating an effective competition term between species

1/9

Take-at-home message: we study a general model of spatially extended ecosystems where species compete for spaces in different habitat patches. We find that spatial heterogeneity is the key mechanism allowing the coexistence of a large number of species, and fosters the emergence of ecological niches

Landscape and environmental heterogeneity support coexistence in competitive metacommunities | PNAS Metapopulation models have been instrumental in quantifying the ecological impact of landscape structure on the survival of a focal species. Howeve...

Better late than never: it's time to tell you about my main project at EPFL 🇨🇭! What happens in competitive metacommunities when you take into account spatial structure and habitat heterogeneity? Once more, a great collaboration with @lambdapp and the people at @liphlab!

www.pnas.org/doi/abs/10.1...

🚀 New research alert! 🚀

Led by @gbarzon.bsky.social, @giorgionicoletti.bsky.social and Daniel Maria Busiello, our latest work in Physical Review Letters (journals.aps.org/prl/abstract...) reveals how excitation-inhibition balance tunes the timescale of information coding in neuronal populations! 🧠🔬

Optimal brain processing requires balance between excitatory and inhibitory neurons, study suggests The brain's ability to process information is known to be supported by intricate connections between different neuron populations.

Balanced excitatory and inhibitory neuron activity is crucial for optimal brain information processing, highlighting the importance of neural network stability in encoding external signals effectively. doi.org/g87h5w

Flash post - my first last-author work with @gbarzon.bsky.social and @dmbusiello.bsky.social is out today in @apsphysics.bsky.social Physical Review Letters!

"Excitation-Inhibition Balance Controls Information Encoding in Neural Populations"

More about it soon!

doi.org/10.1103/Phys...

And finally a bonus for reading the whole thread: another cool video of metapop dynamics using a DEM from the Himalayas!

A cool application is elevation-based networks obtained from Digital Elevation Maps (DEMs). By imposing that exploring uphill is harder than downward one, our model captures the dendritic features of complex topographic landscapes

9/9

Overall, our approach is highly interpretable and extremely flexible, allowing for changes in both the local node dynamics and the exploration ones. Not only that: it works for directed networks as well!

8/9

We can study network fragmentation, showing that our model perfectly captures segregation in modular networks and the detrimental effects of landscape fragmentation. Further, we can prove analytically that survival is generally favored in more dense network

7/9

And of course, network topology influences everything - in particular, the survival of the population, quantified by the metapopulation capacity. We find that multiple short paths and hubs in general favor survival and strong localization of the surviving individuals

6/9

The topological contribution of the underlying networks is naturally encoded in the metapopulation kernel, depending on the species' dispersal. Our model can reproduce classical exponential distance-based kernels, but can capture much more complex network structures

5/9

But where did the topology go? Our model features an exact all-to-all kernel that quantifies the influence of one node to the other - and it does so by naturally taking into account all possible paths the explorers may have taken to move between the two nodes!

4/9

With the core assumption that network exploration is faster than the local dynamics, we obtain an effective fully-connected model for the local population - the metapopulation - which is identical in form to classical metapop models such as the Hanski and Ovaskainen's one

3/9

We attempt to answer this question with ecology in mind. We start from a microscopic model of metapopulation dynamics and make the distinction between
👉 a local settled population on nodes
👉 explorers moving around and colonizing empty sites

2/9

Phenomenological metapop models have proven to be very successful - from predicting species survival in complex landscapes to epidemic spreading in contact networks and urban mobility patterns. But what happens to metapopulation dynamics in arbitrary network topologies?

1/9

Our model can be applied to any network, including elevation-based ones obtained from DEMs, allowing us to study ecological patterns in real-world scenarios (and make cool videos). Keep reading for all the details!

Take-at-home message: we derive a general metapop model from interpretable, agent-based dynamics on any network topology. The properties of metapopulation are encoded in an effective kernel that incorporates both the microscopic dynamics and the network structure at once!