top of page
  • Writer's pictureLuis A Mazariegos

March 2024 – Big Data: Multimodal Analysis of Biodiversity and Environmental Services @ RNMP

Nearly everyone recognizes the moral value of conserving and restoring ecosystems. Over the past decade, methodologies have been developed to also assign an economic value to restoration processes, primarily in terms of the amount of carbon dioxide that can be removed from the environment through photosynthesis and tree growth. The result is the estimation, calculation, verification, and sale of so-called carbon credits. This process and the development of the legal framework surrounding it have become a significant financial catalyst for organizations dedicated to the restoration of pasture lands or highly disturbed environments. Ironically, similar mechanisms do not exist to assign an economic value to the conservation of mature forests, whose value is practically nil in the carbon credit market.

Cloud forests @ MPNR hold an extraordinary biodiversity.


The concept of biodiversity credits and the regulatory framework surrounding them can fill this gap. The first challenge is to implement sampling strategies that allow assigning a value to any area in terms of the biodiversity it contains. This is challenging because species richness in tropical ecosystems is overwhelming, and our knowledge of it is just beginning. It is also challenging because the species present there have their own daily, monthly, and yearly cycles of activity, making many essentially invisible to professionals trying to count them in person.


In an attempt to address these two problems, camera traps are used to conduct counts that do not require human presence and to focus on a moderately species-rich group with relatively well-known taxonomy: mammals. Similarly, the deployment of recorders that automatically capture the soundscapes of an ecosystem has been invoked to estimate the diversity of acoustically active organisms, mainly birds. The combination of camera traps and automatic recorders thus allows estimating biodiversity solely from the two groups of animals we are most familiar with: mammals and birds.

Over 40 camera traps are deployed @MPNR providing valuable information on the presence of large cats which are an important indicator of a healthy ecosysyem.

Since 2020, BIOCONSERVANCY has taken on the challenge of estimating global diversity values that do not depend entirely on species recognition and that also allow estimating the value of a conserved area in terms of the environmental services it provides. To begin with, we needed to quantify the advances of cloud forest restoration in different areas, a task that used to demand long walk hours. Now we use a multispectral drone that takes simultaneous landscape images with cameras that emphasize blue,  green, red, near-red, and near-infrared light bands. Calculations are made with every pixel as presented by each camera allowing us to estimate forest cover, as well as cover of low-density forest, shrubs and grasses. 

Imagery and vegetation estimates of an area can be determined by obtaining multispectral data.

In the environmental gradient generated by the quantified degrees of forest restoration, the extraction, amplification, and sequencing of over 1 million soil DNA sequences confirmed the presence of more than 1,800 genera of bacteria. With this information, we can not only estimate diversity indices, but also classify them by the functional role in tree growth and soil maturation.

The Oxford Nanopore Minion system provides us with real-time, direct sequencing of DNA.


In parallel, the deployment of automatic sensors for light, temperature, and humidity allowed observing how the presence of mature cloud forests reduces both maximum temperatures and temperature fluctuation by a value between 1 to 3 degrees Celsius (1.8 to 5.4 degrees Fahrenheit), facilitating the survival of rare species. With the same type of data, we can quantify the value of cloud forests in maintaining higher values of relative humidity and estimate the percentage of time the forest captures water from horizontal precipitation.

Summary of 87.810 observations from three dataloggers, each providing 29.270 observations on temperature. There are 34 additional dataloggers deployed.


The deployment of automatic recorders with sensitivity up to 96 kHz allowed estimating diversity indices of acoustically active organisms, including bats, crickets, birds, and amphibians. This approach allows evaluating contained biodiversity without requiring the individual identification of the species involved. Finally, the use of camera traps allowed comparing the methods we integrated in a novel way with the most commonly used method for valuing biodiversity. Two advantages of our approach are already clear: the amount of available data exceeds by several orders of magnitude that obtained with traditional methods, and we are now able to quantify some of the environmental services provided by both pristine ecosystems and those under restoration.

Comparison between sound recordings in a pastureland ecosystem (spectrum above) and a pristine forest (spectrum below).

33 views0 comments


bottom of page