That’s why TMA has spent the past five years building one of Ecuador’s most advanced bioacoustic monitoring programs. Working with partners like Dr. Shawn McCracken (Texas A&M–Corpus Christi), Rainforest Connection (RFCx), and Huawei Technologies, we’ve deployed a landscape-scale network of programmable acoustic recorders (AudioMoths) and satellite-connected Guardian devices across the Capuchin Corridor. These tools allow us to listen to the forest continuously (24/7), capturing the full soundscape of birds, mammals, amphibians, insects, and even human activity.

In a landscape where more than 98% of native forest has already been lost, sound is revealing a picture of biodiversity that our eyes alone could never detect.

Sound Reveals What the Eyes Rarely See

Many species that define the Pacific Forest of Ecuador are nearly invisible to human observers. Some call only before dawn, some move too fast to be reliably seen, and others—like the Ecuadorian Capuchin monkey—are so critically endangered that visual encounters are extraordinarily rare.

Bioacoustics bypasses this challenge entirely. Wildlife species use sound to communicate, defend territory, navigate, and find mates. These vocalizations carry across hundreds of meters, cutting through dense vegetation. Acoustic recorders automatically pick up these signals, often detecting species weeks or months before a human survey might encounter them.

Beginning in 2020, TMA and our partners built an automated monitoring program using AudioMoths (programmable acoustic recorders) to understand which species live in and around protected areas like the Jama-Coaque Reserve and Cerro Pata de Pájaro, and how they respond to different land uses across the wider landscape.

Bioacoustics Unlocks Landscape-Scale Insights

Our team deployed more than 50 AudioMoths across nine transects, each covering three distinct habitat types: cattle pasture, cacao agroforestry farms, and native forest. Each device was programmed to record one minute of sound every five minutes around the clock—a design that has produced more than two million recordings to date.

This design allowed us not only to document which species were present, but also to understand how wildlife moves through a fragmented landscape.

The findings were clear and powerful: native forest and agroforestry farms hosted far more species than cattle pasture. Agroforestry, in particular, showed high conservation value—reinforcing TMA’s efforts to support regenerative cacao farming as a wildlife-friendly alternative to cattle ranching or monoculture farming.

The AudioMoths also detected numerous threatened species, including the Gray-backed Hawk, Slaty Becard, Ecuadorian Mantled Howler Monkey, and—most excitingly—the critically endangered Ecuadorian Capuchin.

These kinds of landscape-wide insights are nearly impossible to obtain from visual surveys alone and extremely difficult to achieve using only camera traps.

Why Camera Traps Aren’t Enough

Camera traps are an essential tool, and TMA uses them extensively. But they come with built-in limitations: they record only what passes directly in front of their sensors. Small wildlife may not trigger them, arboreal species may not pass by the camera’s field of view, and many species avoid trails where cameras are placed.

Bioacoustic devices, on the other hand, monitor everything audible within hundreds of meters. Instead of a single, narrow window, they capture the full acoustic footprint of the forest. And unlike a single snapshot of a passing animal, bioacoustics can reveal continuous patterns such as daily behavior cycles, seasonal shifts, and responses to habitat change (i.e. deforestation).

Where a camera trap might miss a species entirely, a single call in the distance can confirm its presence for bioacoustic analysis.

AI Makes Acoustic Monitoring Fast, Accurate, and Scalable

Collecting sounds from the forest is only half the breakthrough. The real innovation comes from the AI model developed by TMA and Rainforest Connection using the Arbimon analysis platform.

With more than 2 million audio recordings from 50+ sites across the Capuchin Corridor, we trained a Convolutional Neural Network (CNN)—essentially an AI species-detection engine—to automatically identify wildlife vocalizations. Think of it as Shazam, but for birds, monkeys, amphibians, and insects.

Every species’ call has a distinctive “sound fingerprint” when visualized on a spectrogram (see below). Once the AI model learns this pattern, it can scan thousands of hours of audio and instantly highlight moments that match. Over time, the model becomes increasingly skilled at distinguishing species—even when calls overlap or background noise is strong.

The result: near-perfect performance across all trained species classes and extremely low false-positive rates—accuracy that would be impossible through manual listening alone.

Where a team of biologists might take months or years to review so many recordings, the AI can do it in minutes. This technological advance frees our field staff to focus on what matters most: conservation action.

Visualization of pattern matching analysis in Arbimon platform. Species detected in this example (circled in green as it calls) is the Slate-coloured Grosbeak (Saltator grossus).

Acoustic Monitoring Protects Forests in Real Time

Bioacoustics isn’t just about studying biodiversity, it also helps protects it.

Rainforest Connection’s Guardian devices detect the acoustic signatures of chainsaws and gunshots and send real-time alerts to TMA’s field team via text message. During the project, one Guardian on the northern extent of the Jama-Coaque Reserve repeatedly triggered chain-saw alerts from a neighboring property, allowing our park rangers to monitor the situation closely and ensure the activity didn’t enter the Reserve.

This type of rapid detection is simply not possible with camera traps or with periodic patrols. The soundscape of the entire landscape gives us time to respond, not just time to analyze.

Satellite-connected Guardian device placed in the canopy of the rainforest in the Jama-Coaque Reserve to detect potential threats in real-time. 

A Conservation Tool for the Future

Bioacoustic monitoring doesn’t replace field biologists; it supercharges their ability to protect the threatened Pacific Forest of Ecuador. It creates an objective, continuous, scalable record of ecosystem health that complements camera traps, on-the-ground expertise, and community knowledge.

As TMA expands acoustic monitoring across new properties and restoration sites, sound is becoming one of our most powerful allies in safeguarding what remains of the Pacific Forest of Ecuador.

The post Listening to the Forest appeared first on TMA.

• Wildlife & Ecology
• Reforestation & Agroforestry
• Carbon & Climate Change

Jacqueline Tliemat, PhD candidate at Texas A&M, installing a camera trap in the canopy of an emergent tree in JCR.

Wildlife & Ecology

Canopy-based wildlife monitoring : TMA together with Dr. Shawn McCracken (Texas A&M Corpus Christi) and his students Jacqueline Tleimat and Rebecca Davis installed camera traps, acoustic recorders (AudioMoths), and climate sensors high in the canopy of trees to study local wildlife populations, with a focus on the critically-endangered Ecuadorian Capuchin Monkey, the Black Mantled Howler Monkey, and other internationally threatened species. To date, our team has analyzed over 220,000 images and documented over 100 species (see Wildlife Insights).

• Publication: Monitoring the habitat and spatial associations of two threatened primates along a conservation area in western Ecuador.

Bird banding specialist Holly Garrod and her team at work in JCR.

Reforestation & Agroforestry

Active reforestation vs passive natural forest restoration: TMA joined forces with esteemed restoration ecologists Dr. Rebecca Cole and Dr. Leland Werden, from the Crowther Lab at ETH Zurich, on a long-term reforestation study that compare the efficacy of active tree-planting versus natural restoration. The study is being conducted in nineteen sites across Colombia, Costa Rica, Ecuador, Mexico, Panama, and Peru. One of those sites is the Jama-Coaque Reserve.

Cacao-Based Agroforestry Demonstration Site: TMA has been experimenting with cacao varieties and growing methods since 2008. JCR currently contains the single largest repository of DNA-verified pure Ancient Nacional cacao in Ecuador—featured in Smithsonian Magazine. We actively manage 9 experimental cacao plots with 5 different types of Nacional cacao, and we closely track the genetics of every single plot. This work was the precursor to TMA’s Regenerative Cacao program with over 100 farmers in the region.

Biodiversity of agroforestry systems: Sophie Roberts from Yale University used remote sensing techniques to assess farm-level diversity metrics in JCR and across 30 regenerative cacao farms in the region.

• Publication: Incorporating drone imagery and field-data to access geographic and social drivers of agroforestry biodiversity in western Ecuador.

Professor Shawn McCracken with a research team member sorting through canopy camera gear in the Bamboo House.

Carbon & Climate Change

Aboveground Biomass Inventory: Researchers from the Universidad Técnica de Manabí and Macalester College conducted aboveground biomass inventories by combining ground-based tree species plot data with allometric equations, drone and satellite imagery, and machine learning algorithms.

Carbon assessment: The Landscapes and Livelihoods Group (TLLG) used satellite imagery to perform a land cover assessment and then applied the data from the aboveground biomass inventories (referenced above) to estimate the carbon value of the Jama-Coaque Reserve and the entire Capuchin Corridor.

Forest Carbon Ledger: After being treated to a front-row seat to the limitations of REDD+ carbon accounting, TMA developed an alternative forest carbon accounting methodology called the Forest Carbon Ledger (FCL). FCL avoids the biggest pitfalls of REDD+. Namely, it uses objective data rather than speculative projections and subjective counter-facturals. It calls for performance-based PES payments according to the total amount of CO2 stored in the forest, which is amortized in annual increments and paid ex-post.

Future Research Opportunities

Margay (Leopardus wiedii) captured on a camera trap in JCR.

The post Research Highlights from the Jama-Coaque Reserve appeared first on TMA.

Note: the Capuchin Corridor is located in the Pacific Forest of Ecuador, known for its remarkable diversity of forest types. This is one of the reasons why it is such an ecologically unique place. It also makes carbon estimation more difficult because each forest class needs to be assessed separately. In some cases, we went so far as to measure various sub-classes as well.

Background

To estimate the carbon value of the Capuchin Corridor, we conducted several biomass inventories in the Jama-Coaque Reserve (JCR) and other sites throughout the Capuchin Corridor. The primary study was conducted in partnership with the Universidad Técnica de Manabí, under the guidance of Dr. Carlos Salas-Macias and Dr.Ezequiel Zamora-Ledezma in 2022-2023. Xavier Haro-Carrión, from Macalester College, as well as Mike Ellis, from Tulane, also provided biomass inventories of the Capuchin Corridor.

We compared their findings to each other and to other biomass inventories in nearby areas or comparable regions. The results were consistent across all biomass inventories. From these results, we took the averages for each forest class. We include links to published studies and/or raw data is for each biomass inventory that is referenced below.

Carbon Density Summary

Below is a summary of the average carbon density per hectare of each major land class.

Notes:

• Measured in metric tons of carbon per hectare (Mg C/ha)
• Total carbon stock includes carbon from above-ground biomass, below-ground biomass, necromass, and soil organic carbon.
• “AGB” refers to carbon from above-ground biomass only.
• See below for detailed definitions of forest types and land class distinctions.

Biomass Inventories

Below are the results of biomass inventories and estimations for each land cover class and subclass. Complete notes and definitions are provided below.

Notes & Definitions:

• “AGB” means above-ground biomass.
• “Mg” is the abbreviation for megagrams, which is equivalent to one metric ton.
• “Mg C/ha” signifies metric tons of carbon per hectare.
• “Mature forest” is defined as forests with minimal anthropogenic intervention, good structural and floristic composition, and closed canopy.
• “Evergreen” forest, in this context, is defined as a broadleaf forest in which the vast majority of its trees have leaves at any given time of the year.
• The category “Evergreen Forest” includes moist forest and cloud forest.
• The “weighted average” of “Mature Evergreen Forest” is the average of cloud forest (25%) and mature moist forest (75%), in line with their relative occurrence in the Capuchin Corridor.  
• “Secondary forest” is defined as forest with a history of significant human intervention, currently has a closed canopy, typically 8-20m canopy height.
• “Tropical dry forest,” also known as “tropical deciduous forest” is a tropical forest in which the majority of tree species loose their leaves for most of the dry season.
• “Semi-deciduous forest” is defined as a forest comprised of a mix of evergreen trees and deciduous trees.
• In this case, “Dry Forest & Semi-Deciduous Forest” includes tropical dry forest (i.e., deciduous forest), secondary semi-deciduous forest, and mature semi-deciduous forest. The “weighted average” is an average of each of those carbon densities weighted by the relative occurrence of each class in the Capuchin Corridor: 10% dry forest, 10% mature semi-deciduous forest, and 80% secondary semi-deciduous forest.  
• “Early-successional forest” is defined as a young secondary forest in the early stages of regeneration–typically 2-10 years old.
• “Fallows” is land that is submitted to a slash-and-burn cycle, in which natural vegetation is allowed to grow unabated for 1-7 years (during which time it is “fallow”) and then is slashed and burned for corn, cattle, or other agricultural uses.

Sources:

• Haro-Carrión et al. (2021)
• Ellis (2022)
• Salas-Macias et al. (2017)
• Eguiguren et al. (2021)
• Smith et al. (2020)
• See below for details of Salas-Macias & Zamora-Ledezma (2023)

Salas-Macias & Zamora-Ledezma (2023)

The primary biomass inventory was conducted by Dr. Carlos Salas-Macias and Dr.Ezequiel Zamora-Ledezma with the Universidad Técnica de Manabí in 2022 and 2023. Their findings have not yet been officially published. Below is an explanation of their methods and sources.

Methods

Four sample plots of 20 x 50 m (1000 m2) were used to determine the carbon stored in living aboveground biomass in each of four types of forest: Moist evergreen forest; cloud forest; 10-year-old early successional forest; and dry/semi-deciduous forest. Within each plot, data on total height and diameter at breast height (DBH) of all individuals with DBH greater than or equal to 5 cm were recorded. For the estimation of biomass (kg) the allometric equation proposed by (Chave et al., 2005) was used, which also incorporates information on wood density obtained from the Global wood density database (Zanne et al., 2009). In cases where there was no data on wood density of the species, the genus or family value was used (Honorio & Baker, 2010). To carry out the conversion of aerial biomass to carbon, it was assumed that the carbon content is 50% of the aerial biomass of each living tree (Penman et al., 2003; Hetland et al., 2016; Tashi et al., 2016, Vijayakumar et al., 2016), The data is expressed in Mg ha-1.

Bibliographical references

• Chave, J., Andalo, C., Brown, S., Cairns, M. A., Chambers, J. Q., Eamus, D., Fölster, H., Fromard, F., Higuchi, N., & Kira, T. (2005). Tree allometry and improved estimation of carbon stocks and balance in tropical forests. Oecologia, 145(1), 87–99. https://doi.org/10.1007/s00442-005-0100-x
• Hetland, J., Yowargana, P., Leduc, S., & Kraxner, F. (2016). Carbon-negative emissions: Systemic impacts of biomass conversion: A case study on CO2 capture and storage options. International Journal of Greenhouse Gas Control, 49, 330–342. https://doi.org/10.1016/j.ijggc.2016.03.017
• Honorio, E., N., & Baker, T., R. (2010). Manual para el monitoreo del ciclo del carbono en bosques amazónicos. Instituto de Investigaciones de la Amazonia Peruana. Universidad de Leeds. Lima. Recuperado de http://www.rainfor.org/upload/ManualsSpanish/Honorio_Baker2010%20Manual%20carbono.pdf
• Penman, J., Gytarsky, M., Hiraishi, T., Krug, T., Kruger, D., Pipatti, R., Buendia, L., Miwa, K., Ngara, T., Tanabe, K., Wagner, F. (2003). Good practice guidance for land use, land-use change and forestry. Institute for Global Environmental Strategies. ISBN 4-88788-003-0. Recuperado de https://www.ipcc-nggip.iges.or.jp/public/gpglulucf/gpglulucf_files/GPG_LULUCF_FULL.pdf
• Tashi, S., Singh, B., Keitel, C., & Adams, M. (2016). Soil carbon and nitrogen stocks in forests along an altitudinal gradient in the eastern Himalayas and a meta-analysis of global data. Global Change Biology, 22(6), 2255–2268. https://doi.org/https://doi.org/10.1111/gcb.13234
• Vijayakumar, D., Raulier, F., Bernier, P., Paré, D., Gauthier, S., Bergeron, Y., & Pothier, D. (2016). Cover density recovery after fire disturbance controls landscape aboveground biomass carbon in the boreal forest of eastern Canada. Forest Ecology and Management, 360, 170–180. https://doi.org/10.1016/j.foreco.2015.10.035
• Zanne, A.E., Lopez-Gonzalez, G., Coomes, D.A., Ilic, J., Jansen, S., Lewis, S.L., Miller, R.B., Swenson, G., Wiemann, M.C., Chave, J. (2009), Data from: Towards a worldwide wood economics spectrum, Dryad, Dataset, https://doi.org/10.5061/dryad.234

Total Carbon Stock

Carbon in above-ground biomass (AGB) is the largest carbon pool in forests and it’s the most responsive to deforestation and forest degradation. It’s also the easiest (and therefore most reliable) to measure.

Nevertheless, we also calculated other carbon pools (e.g., soil organic carbon, below-ground biomass, necromass, etc.) by extrapolating their values from above-ground biomass. This process was aided by numerous academic publications on the subject, most notably Eguiguren et al. (2021) “Estimating carbon stocks across forest types in the Ecuadorian lowland forest.”

Notes & Definitions:

• Measured in metric tons of carbon per hectare (Mg C/ha).
• “AGB” means above-ground biomass.
• “BGB” means below-ground biomass.
• “SOC” means soil organic carbon.
• “Necromass” is the mass of dead timber in a forest, dead bacteria etc.
• For land class definitions and sources, see the “Notes & Definitions” in the “Biomass Inventories” section.
• BGB, Necromass, and SOC values were extrapolated from AGB values in line with the ratios for lowland forest in northwest Ecuador in Eguiguren et al. (2021)

Cloud forest of the Capuchin Corridor

The post Biomass Inventory of the Capuchin Corridor in Coastal Ecuador appeared first on TMA.

FCL is a methodology we developed in response to many of the defects we encountered with REDD+ during the measurement and estimation process. For an in-depth comparison of these two methodologies, check out Comparing the Forest Carbon Ledger to REDD+: Using the Capuchin Corridor in Ecuador as a case study. For a deeper understanding of how the FCL works, and why it was created, refer to Flipping REDD+ on Its Head: The Forest Carbon Ledger (FCL) is a new valuation method. The biomass numbers in this assessment are sourced from the Biomass Inventory of the Capuchin Corridor that we conducted in partnership with researchers from the Universidad Técnica de Manabí.

Brief Explanation of the Project Areas

The Capuchin Corridor spans a 43-kilometer mountain range that runs parallel to the Pacific Ocean. This corridor is home to one of the last major remnants of Ecuador’s Pacific Forest—an ecosystem that has already lost 98% of its native forest over the course of the last century.

The Capuchin Corridor contains a wide range of distinct tropical forest types, including cloud forest, moist evergreen forest, tropical dry forest, semi-deciduous forest, and a small remnant of primary-growth wet forest at the top of Cerro Pata de Pájaro. It also contains a patchwork of agricultural land that has been eating away at the forest for the last century.

In its entirety, the Capuchin Corridor is 40,000 hectares. The carbon estimates in this report analyze 37,000 hectares.

The Camarones River Basin occupies 2,980 hectares in the heart of the Capuchin Corridor and is the site of the Jama-Coaque Ecological Reserve—TMA’s flagship project. It’s also the place where TMA pilots its projects before expanding out into the rest of the Capuchin Corridor.

Brief Explanation of the Numbers

There are three sets of measurements involved in assessing the CO2 benefits of a given forest: land cover, biomass inventory, and the final estimate of net CO2 benefits. Below is a summary of each set of calculations, starting from the conclusion.

Annual CO2 Benefit (Summary)

Annual CO2 benefit of the Capuchin Corridor and the Camarones River Basin are calculated using the following variables:

• Different carbon accounting methodologies: Forest Carbon Ledger (FCL) vs REDD+
• Different carbon pools: all carbon pools (ie., total carbon stock) vs above-ground biomass (AGB)
• Different project scope: entire project area vs only the areas in which direct intervention will take place

Table 1A. Annual CO2 benefit using FCL vs REDD+

Notes:

• Measured in metric tons of CO2 per year (Mg CO2 yr-1)
• Total carbon stock includes carbon from above-ground biomass, below-ground biomass, necromass, and soil organic carbon.
• “AGB” refers to carbon from above-ground biomass only.
• Land cover assessment provided by The Landscapes & Livelihoods Group (TLLG)
• Takes into account the entire project areas.
• All monetary values are in USD$

Table 1B. Annual CO2 benefit according to project scope

Notes:

• Measured in metric tons of CO2 per year (Mg CO2 yr-1)
• Land cover assessment by The Landscapes & Livelihoods Group (TLLG)
• Uses FCL for carbon accounting

Table 1C. Annual CO2 benefit per hectare 

Notes:

• Land cover assessment by The Landscapes & Livelihoods Group (TLLG)
• Takes into account the entire project areas

Forest Carbon Ledger (FCL)

The FCL is an annual ledger that tracks the carbon stock of a forest year-over-year. Below are the FCL estimates that measure carbon from above-ground biomass at the present point in time. As stated above, a detailed account of FCL calculations in the Capuchin Corridor can be found in the case study. For a comprehensive explanation of how FCL works, and why it is more reliable than REDD+, refer to Flipping REDD+ on Its Head: The Forest Carbon Ledger (FCL) is a new valuation method.

Table 2A. FCL estimate of the Capuchin Corridor (above-ground carbon, year 0)

Table 2B. FCL estimate of the Camarones River Basin (above-ground carbon, year 0)

Notes to tables 2A and 2B:

• Assumes a risk buffer & leakage discount of 15%.

The summaries for FCL calculations of Total Carbon Stock and the Direct Intervention Area can be found in Table 1B. For the complete calculations, refer to our database Carbon Stock of the Capuchin Corridor.

REDD+ Estimate

The following REDD+ carbon estimation was performed by The Landscapes & Livelihoods Group (TLLG), a carbon project developer and monitoring specialist accredited by Plan Vivo, a carbon offset standard with a strong emphasis on empowering rural communities.

REDD+ is a methodology for valuing the CO2 benefit of a forest. It does this by estimating how much CO2 would be released into the atmosphere from the cutting and/or burning of trees in a hypothetical “business-as-usual” scenario. It then compares this number to the amount of CO2 that would presumably be released into the atmosphere in a “project scenario,” in which actions are taken to prevent the cutting and/or burning. The difference between these two numbers = net carbon benefit.

The measurements below take into account the entire project areas and assume a risk buffer & leakage discount of 15% and an effectiveness rate of 75%.

Table 3A. Annual CO2 benefit (REDD+)

The baseline numbers in the table above are informed by projections of annual deforestation and degradation rates in the future—in this case, over a five-year period of the future. These deforestation rates are extrapolated from historic deforestation rates in the same area as well as adjacent areas that are judged to be comparable in terms of topography, demographics, road access, economic pressures, political conditions, etc. It’s a subjective estimate.

Table 3B. Deforestation/Forest Degradation Projections

Land Cover Assessment

A land cover assessment is the first step toward quantifying the carbon value of a forest. This requires breaking the forest down into its various different land classes, which (ideally) can each be distinguished by aerial imagery paired with machine learning algorithms. We contracted two different specialists to perform land cover assessments of both areas.

The first analyst, Sake Alkema, worked at IUCN-Netherlands at the time; now he’s a remote sensing specialist for the global nature tech company Satelligence. The second remote sensing specialist was the Plan Vivo-accredited carbon project developer The Landscapes and Livelihoods Group (TLLG).

To assist in this process, we helped train the latter’s algorithm by recording ground data at 211 different points throughout the project area. Below are the results from both land cover assessments.

Table 4A. Land cover assessment by Sake Alkema / IUCN-NL (2018)

Table 4B. Land cover assessment by TLLG (2022)

Land cover map of the Capuchin Corridor prepared by The Landscapes & Livelihoods Group (TLLG).

Biomass Inventories & Carbon Density

To estimate the initial carbon stock of the forest, we conducted several biomass inventories in the Capuchin Corridor, most notably in partnership with the Universidad Técnica de Manabí. Then we compared their results to each other and to other biomass inventories in nearby areas or comparable regions. The results were consistent across all biomass inventories. From these results, we took the averages for each forest class.

Below is a summary of the carbon densities of each major land class. To explore the results in greater detail, including biomass results for each sub-class, refer to Biomass Inventory of the Capuchin Corridor.

Table 5. Average carbon density per hectare 

Notes:

• Measured in metric tons of carbon per hectare (Mg C/ha)
• Total carbon stock includes carbon from above-ground biomass, below-ground biomass, necromass, and soil organic carbon.
• “AGB” refers to carbon from above-ground biomass only.

Mature wet forest of Cerro Pata de Pájaro.

Conclusion

For reasons amply explained in the case study and FCL white paper, we believe that FCL is preferable to REDD+ as a methodology for estimating the CO2 benefit of the Capuchin Corridor and Camarones River Basin.

To be conservative, we’ve chosen to limit our carbon inventory to above-ground biomass (AGB), rather than use total carbon stock. AGB is the most straightforward carbon pool to measure and it is, by far, the most responsive to forest loss/gain. The numbers below refer to the entire project areas.

The post Carbon Assessment of the Capuchin Corridor & Camarones River Basin appeared first on TMA.