An approach that aligns human action with the capacity of ecosystems for self-organization and uses artificial intelligence to support decisions in forest restoration

Large-scale forest restoration needs to go beyond the idea of simply rebuilding ecosystems by planting trees. This is what researchers Pedro H. S. Brancalion, from the Center fo Carbon Research in Tropical Agriculture at the University of São Paulo (CCARBON/USP), and Marcelo P. L. Medeiros, from the Salata Institute for Climate and Sustainability, Harvard University, argue in an article published in the journal Trends in Ecology & Evolution, one of the main international scientific publications in the areas of ecology and evolution.

Entitled “Human, natural, and artificial intelligence in forest restoration”, the article was made available online on September 7, 2026 by Cell Press. The paper discusses how the integration of human knowledge, natural processes, and artificial intelligence (AI) can contribute to addressing one of the great contemporary challenges: restoring forests on a scale sufficient to respond to the climate, biodiversity loss, and environmental degradation crises.

 Trends in Ecology & Evolution is an international reference journal in ecology and evolution. The journal has an impact factor of 17.3 and a Q1 rating, occupying a prominent position among scientific journals in its area.

A forest is not just a collection of trees

The starting point of the article is a critique of a still common conception of restoration: the idea that it would be possible to “rebuild” a forest in the same way that one builds an infrastructure, selecting components and putting them together to produce a previously planned system.

Although tree planting is an important tool in certain situations, the authors point out that a forest is much more than the sum of its trees. Its biodiversity, structure, and functioning result from complex interactions between plants, animals, microorganisms, soil, water, climate, and ecological processes that develop over time.

“Restoring forests is not simply about planting trees, but about creating the conditions for ecosystems to recover their complexity. Artificial intelligence can expand our ability to understand this complexity, but it does not replace natural processes or eliminate the limits of human action.”

Pedro H. S. Brancalion, CCARBON/USP researcher and co-author of the study

This complexity also helps to explain why natural regeneration can play a key role in the recovery of ecosystems. In several situations, the recovery of native forests occurs through ecological succession, in which communities reorganize and gradually develop their complexity without all stages being directly controlled by humans.

For this reason, the authors propose a change of perspective: instead of trying to control and rebuild nature, restoration should seek to create the conditions for ecological processes themselves to lead to the recovery of ecosystems.

Three forms of intelligence

It is in this context that the article proposes the integration of three forms of intelligence: human, natural and artificial.

 Human intelligence is responsible for setting goals, mobilizing knowledge and resources, making decisions, and creating the conditions necessary for ecological recovery to happen.

 Natural intelligence corresponds to the ability of ecosystems themselves to organize, adapt, and develop complexity through processes such as ecological succession and interactions between organisms and the environment.

Artificial intelligence, on the other hand, can expand the human capacity to understand and deal with this complexity. Machine learning technologies can, for example, integrate large volumes of data from satellites, drones, sensors, acoustic monitoring, and environmental DNA to identify patterns and support decisions about where, when, and how to carry out interventions.

Rather than replacing human decision-making, AI could contribute to making restoration more precise, adaptive, and able to account for the heterogeneity of landscapes.

From the logic of control to the logic of adaptation

The use of artificial intelligence, however, also involves risks. The authors warn that AI systems are not neutral: their results depend on the data used to train them, the algorithms employed, and the values and interests incorporated into the decision-making processes.

An inadequate application could, for example, reinforce the tendency to prioritize tree planting to the detriment of natural regeneration or favor species for which data are more available, failing to consider organisms and ecological interactions that are still poorly studied.

The adoption of these technologies can also produce social impacts. While automation can increase the efficiency of restoration activities, it can also alter job opportunities and income generation in local communities. Therefore, the authors argue that decisions about the use of AI should consider not only efficiency gains, but also human well-being, local livelihoods, and social priorities.

To restore is to allow nature to rebuild itself

The main message of the article is that not even the most advanced technologies can reproduce the complexity of a native forest.

For the authors, the challenge of large-scale restoration is not simply figuring out how to plant more tree species, more quickly and over larger areas. The challenge is to align human actions with ecosystems’ self-organizing capacities, using technologies such as artificial intelligence to better understand this complexity and guide interventions when they are needed.

This perspective proposes an important shift in the way of thinking about restoration: from an approach based on trying to control and rebuild ecosystems to an approach based on facilitating and guiding their recovery.

By integrating human intelligence to initiate and drive restoration processes, natural intelligence to generate ecological complexity, and artificial intelligence to support decisions and adaptive management, forest restoration can move towards a model that is more sensitive to the particularities of each landscape and the limits of human capacity to control nature.

The article “Human, natural, and artificial intelligence in forest restoration” is available at Trends in Ecology & Evolution.

Reference:
Brancalion, P. H. S.; Medeiros, M. P. L. Human, natural, and artificial intelligence in forest restoration. Trends in Ecology & Evolution. 2026. DOI: 10.1016/j.tree.2026.08.008