Taking the pulse of Earth's tropical forests using networks of highly distributed plots

TL;DR: In this paper, the authors show how a global community is responding to the challenges of tropical ecosystem research with diverse teams measuring forests tree-by-tree in thousands of long-term plots.

About: This article is published in Biological Conservation.The article was published on 2021-05-25 and is currently open access. It has received 66 citations till now.

Summary

1. Introduction

• As the most diverse and productive ecosystems on Earth, tropical forests play essential roles in the carbon and water cycles and maintenance of global biodiversity.
• They also affect their health in multiple ways, providing rich pharmacopoeias to traditional and modern societies, and capable of changing the course of history when pandemic zoonotic pathogens emerge as forests and wildlife are exploited.
• Tropical forests are also critical to determining the degree and impact of anthropogenic climate change.
• While vital to their past and future, efforts to measure and monitor them have until recently been localised and largely disconnected.
• Providing tools to ensure tropical scientists can manage, share and analyse their data themselves, ForestPlots.

2. Network development

• Tropical research plots that tag, measure, identify and follow forests tree-by-tree have existed for decades.
• The very first permanent sample plots the authors are aware of in the tropics were installed in 1857 by the German forester Brandis, who worked for the British in Burma (now Myanmar) and later in other parts of India (Dawkins and Philip, 1998).
• This was expanded to draw together inventory data from >100 sites in Amazonia and then African forest plots to include some of the longest running monitoring sites worldwide (Peacock et al., 2007).
• Now, ForestPlots.net supplies ecological informatics to colleagues in scientist-led networks from 54 countries working across 44 tropical nations (Fig. 1).
• Developing this functionality has supported a surge in multi-site and multi-national analyses that are increasingly initiated by scientists from the tropics, gradually supplanting the traditional model where researchers from the Global North lead.

3. Environmental representation

• Within each continent coverage has been focused on the moist tropical lowlands with sampling extending into montane and drier forest systems most effectively in South America (Fig. 4c).
• Plots also cover the complex edaphic variation present in Amazonia (Quesada et al., 2012) where they encompass landscape-level variability within old-growth forests (Anderson et al., 2009, 2010).
• Fuller environmental coverage can help networks address challenges such as monitoring of protected area effectiveness (Baker et al., 2020) and providing calibration-validation of Earth Observation space-borne sensors (Chave et al., 2019).
• Beyond the lowland humid tropics, special effort is also needed for long-term, ground-based monitoring in particular environments.

4. Discovery: forest ecology across the tropical continents

• RAINFOR, AfriTRON and T-FORCES plots have generated ecological and biogeographical insights that have only been achievable via largescale collaboration.
• Thus the role of soils and species composition in affecting biomass carbon is a key reason why ground data are essential for mapping forests (Chave et al., 2019).
• The authors cannot simply focus on carbon and achieve biodiversity conservation, and vice versa.
• African forests have many fewer stems than their Asian and South American counterparts, but South American forests have considerably less biomass.
• The implication is that other factors related to the evolutionary and historical happenstance of each continent matter.

5. Discovery: tropical forest change

• The single most significant scientific impact of these multiple permanent plot networks has been to transform their understanding of how tropical forests function in the Earth system.
• Large-scale imbalances observed in the global carbon balance have cast doubt on this assumption (e.g. Taylor and Lloyd, 1992).
• A quarter of a century of research since then has rejected the notion that ‘intact’ tropical forests are unaffected by atmospheric changes and reinforced the central concept that all tropical forests are being influenced by a suite of large-scale contemporary anthropogenic drivers.
• (2) Biomass dynamics have also accelerated in Amazonia.
• The long-term increase in carbon stocks of African forests was recently updated and confirmed, with three times as many plots showing continued sink strength (Hubau et al., 2020).

6. Challenges and the future of tropical forest monitoring

• Large-scale plot networks have not only made a series of crucial scientific discoveries and advances, but even more profoundly the Social Research Network model pioneered by RAINFOR since 2000 has influenced how the science itself is being done.
• A partial developmental solution to this involves providing network contributors the opportunity to lead analyses with the expectation that these new leaders then support others with their analyses.
• So here the authors address this question in terms of the human effort made thus far and the financial investment needed to monitor across continents.
• The average effort in the field, herbarium, and lab to install a typically remote and diverse 1-ha tropical forest plot and analyse its species and soil sums to 98 person-days, with an additional effort of 38 person-days to support and sustain these teams and data management.

7. Achievements, impact and potential

• Tropical forest science has come a very long way.
• As new technologies for probing forests become available, the highly distributed standardized long-term plots and networks of skilled tropical researchers represent critical infrastructure to enhance and calibrate new insights as they arise.
• Yet twenty years of hard-won scientific results show that reliable and highly distributed monitoring is irreplaceable.
• The article is attributed collectively as ForestPlots.
• In conclusion, the ongoing cost of monitoring Earth’s remaining tropical forests on the ground is extraordinarily small compared to the great scientific and practical benefits it provides.

Acknowledgments

• This paper is a product of the RAINFOR, AfriTRON and T-FORCES networks and the many other partner networks in ForestPlots.
• The authors are particularly indebted to more than one thousand four hundred field assistants for their essential help in establishing and maintaining the plots, as well as highly distributed rural communities and institutions.
• For additional assistance the authors thank Michel Baisie, Wemo Betian, Vincent Bezard, Mireille Breuer-Ndoundou Hockemba, Ezequiel Chavez, Douglas Daly, Armandu Daniels, Eduardo Hase, Muhammad Idhamsyah, Phillipe Jeanmart, Cisquet Keibou Opepa, Jeanette Kemp, Antonio Lima, Jon Lloyd, Mpanya Lukasu, Sam Moore, Klaus Scipal and Rodrigo Sierra.
• The development of ForestPlots.net and curation of data has been funded by several grants including NE/B503384/1, NE/N012542/1 ‘BIO-RED’, ERC Advanced Grant 291585 ‘T-FORCES’, NE/F005806/1 ‘AMAZONICA’, NERC New Investigators Awards, NE/N004655/1, ‘TREMOR’, the Gordon and Betty Moore Foundation (‘RAINFOR’, ‘MonANPeru’), ERC Starter Grant 758873 ‘TreeMort’, EU Framework 6, a Royal Society University Research Fellowship, and a Leverhulme Trust Research Fellowship.
• Finally the authors thank their late, great colleagues whose unique contributions helped make possible all that the networks and ForestPlots.

Figures

Fig. 2. Growth of pan-tropical forest monitoring since the mid-twentieth-century. Top: Plot-censuses curated at ForestPlots.net by date of census. Bottom: Cumulative number of contributors to ForestPlots.net by date of first recorded fieldwork. Growth was slow following the first census in 1939, only reaching 100 censuses by 1969. For early censuses, records of field team personnel and leaders are often sparse or absent. Note that ‘contributors’ are defined inclusively to reflect members of indigenous communities, protected area guards, parataxonomists, students, and technicians, as well as principal investigators, botanists, and other specialists.

Fig. 1. Current extent of ForestPlots.net. Top: Pantropical plot sampling density per 2.5 degree square with the 4062 multiple- and single-inventory plots hosted at ForestPlots.net. These plots contribute to 24 networks including RAINFOR, AfriTRON, T-FORCES, ATDN, BIOTA, COL-TREE, FATE, GEM, Nordeste, PELD, PPBio, RAS, RBA and SEOSAW. Forest cover based on the Global Land Cover 2000 database (JRC, 2003) with tree cover categories: broad-leaved evergreen; mixed leaf type; and regularly flooded. Our plots also extend into neotropical and African savannas; Bottom: The same plot sampling but displayed at higher-resolution (1-degree grid cells) for each focal continent, South America, Africa, and Southeast Asia and Australia.

Fig. 5. Pantropical forest carbon storage is independent of species richness. There are no clear within-continent or pantropical relationships between carbon stocks and tree species richness per hectare in structurally intact oldgrowth tropical forests. Figure adapted from Sullivan et al. (2017).

Table 1 Networks contributing to ForestPlots.net. We report the 24 international, national, and regional plot networks contributing to and supported by ForestPlots.net in 2020, in order of date of affiliation. Note that some plots contribute to more than one network, in some cases the plots managed at ForestPlots.net are fewer than the total number of plots of the network, while others are not ‘networked’ but managed by individual researchers. Hence, cross-network totals do not correspond precisely to the number of plots managed. We include 20 tropical networks with multi-census plots plus four large-scale floristic-focussed scientific networks (ATDN, CAO, sANDES, RedGentry) that work exclusively with single-census data. All numbers compiled September 2020. As an open collaborative project ForestPlots.net welcomes all contributors with carefullymanaged plots.

Fig. 6. Tropical continental macroecology. Remarkable continental differences in species richness, stem density and carbon stocks emerge among lowland tropical moist forests when densely sampled plot networks are combined. Graphics depict probability densities such that the whole area for each continent sums to 1. Note that the y-axis scale for each variable thus varies depending on the range of the x-axis: for continents with larger variation in x, the probability density at any point along the y axis is correspondingly smaller. Analysis adapted from Sullivan et al. (2017, 2020).

Fig. 4. Network coverage of geographical and climate space. Analyses include >1500 permanent plots managed at ForestPlots.net. (a) Top panels: (1) Geographic distance between multicensus plots across the humid tropical forest biome; and (2) Minimum climate dissimilarity (Euclidean distance on variables scaled by their standard deviation, accounting for mean annual temperature, temperature seasonality, mean annual precipitation and precipitation seasonality), where for each cell environmental distance represents how dissimilar a location is to the most climatically similar plot in the network. Note that some poorly sampled areas are mostly deforested, such as Central America, Madagascar, and much of tropical South and Southeast Asia. The baseline map depicts WWF terrestrial ecoregions (Olson et al., 2001). (b) Middle panel: Tropical plots displayed in global biome space (Whittaker diagram), showing the main concentration of plots from lowland wet through to moist forests and savanna, with some samples in cooler montane climates. (c) Lower panels: Plots displayed within tropical humid and sub-humid climate space, with plots displayed colour-coded by continent (see Fig. 2) and symbol size corresponding to total census effort. Note the important differences in baseline climatic conditions between continents.

Fig. 3. Growth of ForestPlots.net and its contributing networks since 2000. Top: Cumulative upload of unique plot censuses to ForestPlots.net by date of upload (pre-2009 uploads to pre-internet versions allocated evenly back to network beginnings); Bottom: Cumulative peer-reviewed scientific articles based on network plots, excluding research based on single-plot studies.

Frequently asked questions

Q1. What have the authors contributed in "Taking the pulse of earth’s tropical forests using networks of highly distributed plots" ?

Here the authors show how a global community is responding to the challenges of tropical ecosystem research with diverse teams measuring forests tree-by-tree in thousands of long-term plots. The authors review the major scientific discoveries of this work and show how this process is changing tropical forest science. Their core approach involves linking long-term grassroots initiatives with standardized protocols and data ForestPlots. 

Q2. What are the future works mentioned in the paper "Taking the pulse of earth’s tropical forests using networks of highly distributed plots" ?

Net experience demonstrates that collaborative, multi-polar structures help ensure breadth and resilience while supporting and encouraging the leaders of the future. Meanwhile, maintaining permanent plots is as much an expression of hope in the future as a stake in an immediate scientific outcome, as rewards may accrue to others distant in time and space. Ultimately, sustained funding in and by tropical countries themselves will ensure they not only have strong training programmes to develop the core field and analytical skills that scientists need, but equal opportunities for career development. Collaboration is gratifying, but letting go of their egos can be challenging, while in larger groups there is greater risk that individuals feel their contributions go unnoticed.