Previous Page Table of Contents Next Page


Conclusions and recommendations


A great amount of methodological knowledge and case study information was gained from the multiple activities reported in this volume. In particular, this knowledge and information concerned the following:

The methods and procedures developed represent the results of intensive research, refinement, development and testing in three case studies in Latin America and the Caribbean region.

The case studies represent the range of conditions in terms of ecological variability, in terms of the impact of human activities, and in terms of the decision-making structure related to LUCs.

The methodology makes use of a range of knowledge and tools. This makes it imperative that a multidisciplinary team undertake its application in assessment, inventory and monitoring exercises in the field. At least one member of the team should be a botanist. Ideally, the team would comprise soil scientists, ecologists, botanists, foresters and experts in land evaluation or resource planning, modelling and GIS/remote sensing. The level of expertise of such a team may range from a competent technician with sufficient experience in the area of study to graduate or postgraduate professionals.

As discussed in the body of this report, there are several complex technical issues that require further analysis, refinement and testing in order to become part of a robust methodology for the assessment, inventory and monitoring of carbon stocks and sequestration. Most of the technical issues relevant to the objectives of this project concern the accuracy of estimates of carbon content in the carbon pools in aboveground and belowground biomass and in SOM.

Remote-sensing techniques and products proved useful for providing a landscape to the study, for estimation of biomass, and for assessment of biodiversity and land degradation.

These tools also provided a geographical framework for partitioning the spatial variability of ecological parameters relevant in estimating the different pools of C in biomass and in the soil.

The use of nested sampling quadrats and of a supervised classification of a satellite image for stratification was found adequate for estimating biomass.

Nested quadrat dimensions provided a good compromise between accuracy of estimates derived from measurements and the amount of work required for field measurements of volume, species counting and degradation assessment.

The regression equation method for estimating aboveground biomass proved to be better than the use of volume measurements at quadrats and their spatial interpolation. The latter would have required a full geostatistical study in order to generate reliable estimates over an entire area, whereas the regression equation method simplifies computations.

The use of multispectral satellite images and band ratio indices related to biomass estimates from quadrats on the ground proved very promising for the upscaling of biomass estimates where forests are the dominant land cover types.

There are no universal regression equations to use for biomass estimation from specific site measurements of volume in all ecological conditions. In each area, these equations will have to be developed afresh or adapted to the conditions in which they will be used with prior rigorouscalibration.

It proved impractical to obtain reliable tree crown measurements in areas with dense tropical forest cover, such as in Bacalar, because the density of crown cover obscured the limits of tree crowns. In these circumstances, it was found more useful to measure the volume of tree trunks and crowns for a subset of sampling quadrats, and relate them to band ratio indices of reflectance in a satellite image through a regression equation. The image was then used for the upscaling of ground estimates.

The methodology does not account for the so-called leakages of biomass resulting from the multipurpose use of the forest by subsistence farmers, which is common in developing countries.

Biomass estimates for crops can be obtained reliably from crop growth models, the AEZ methodology or from the local records of experimental stations. Crop yield estimates can also be derived from these sources.

The search for promising LUTs with potential for carbon sequestration required a full land suitability assessment exercise, for which the FAO framework for land evaluation proved to be most useful. Criteria concerning the efficiency of CO2 assimilation or other forms of biomass accumulation should be part of the suitability assessment procedure.

The models for suitability assessment based on decision trees and including criteria for carbon sequestration proved to be very effective in the selection of PLUTs. However, they may not necessarily be practically feasible as their development requires time, information on land qualities and the availability of decision-tree software.

Soils represent a larger pool of terrestrial C than biomass. Therefore, carbon sequestration in soils is of great importance in mitigating greenhouse gas concentrations. Land management practices have the strongest effect on the fate of SOM and on carbon sequestration, or on its release to the atmosphere as CO2. This was demonstrated by the results obtained through SOM turnover modelling in all three case studies.

The turnover of SOM must be modelled over a definite period of time in order to determine whether carbon sequestration is possible for soils under a given LUT.

Model parameterization is fundamental to accurate and reliable projections of quantities of SOM over a period of time. Data requirements for parameterization and the degree of complexity of the parameterization process vary with the complexity of the model and the detail and specificity of model outputs.

In using any organic matter turnover simulation model, model calibration is a necessary step in order to determine model behaviour and the bounds of accuracy of model predictions in the area of concern.

It proved fairly straightforward to parameterize and run simulations with the RothC-26.3 model. However, this model does not include all the possible partitions of SOM; nor does it consider the numerous interactions of C with other elements, particularly N and S.

The CENTURY model is powerful and very complete. It simulates to a high level of detail the environmental compartments and SOM fractions in which the organic matter is decomposed and degraded, accounting for the interactions with other elements and compounds in the soil, such as N and S. However, this model proved too large and complex for the ordinary professional and mid-level technician. Its parameterization was very laborious and difficult, particularly in terms of data requirements and the specification of input and output variables. As a trade-off, the model offers a wide range of output variables to examine, in detail, carbon fractions and their interactions with other nutrients.

The development of software for a GUI for the CENTURY model (named Soil-C) represents an important improvement. This interface software makes transparent to the user the intricacies of the CENTURY model parameterization, specification of the management events, and all the circumstances of the scenario being modelled. Input and output is made much simpler, making the model more accessible.

The link between the Soil-C interface and a GIS still requires further software development and testing in order to include interfaces to commonly-used GIS software.

The RothC-26.3 model should be used where only limited detail is required about the fractions and pools of SOM, or where few data are available for model parameterization.

The CENTURY model with the Soil-C interface can be used: (i) where greater accuracy is required or greater detail in terms of the number of partitions or pools of SOM, or in explaining the fate of additions of organic materials under detailed management of SOM; (ii) where land and crop management are compounded by various factors that need to be included in the model, such as interactions of C in organic matter with other important elements and compounds; and (iii) where greater accuracy of estimates is required.

The results obtained from all three case studies showed that land management is crucial in determining carbon sequestration in soil, particularly in agricultural land. Land management turned out to have even more weight in enhancing carbon sequestration than the selection of carbon-efficient crops through land evaluation, although the latter is also an important factor.

Staple crops in associations or in rotations, particularly with N-fixing legumes (e.g. alfalfa and beans) appear to hold the greatest promise for carbon sequestration in soils. This can be further enhanced by including fruit trees in the association.

Irrigation, or moisture availability in the soil, and the addition of sufficient organic inputs from crop residues or other sources, such as FYM, trigger the process of SOM accrual in the soil, particularly of the resistant fraction of SOM. The threshold value of organic inputs to the soil to enable carbon sequestration varies with soil type, climate conditions and prior land management history, particularly the management of organic matter.

It was demonstrated that SABA can be converted into stable, continuous cropping on the same field where sufficient amounts of organic inputs are brought into the field from a variety of sources in the family unit production system.

The modelling results showed that grasslands were not as efficient as expected in carbon sequestration in the soil, except for some instances in the dry tropics of Cuba.

The parametric semi-quantitative approach used for rapid assessment of land degradation based on indicators observed in the field and standard climate data sets yielded satisfactory results. Thus, it was possible to determine the causative factors and intensity of a given type of land degradation and its relative importance with respect to other types of degradation.

In Texcoco, physical, biological and chemical land degradation are acute. In particular, physical degradation caused by soil erosion and compaction is readily evident. Biological degradation, consisting of SOM depletion, appears to be the main determining factor, inhibiting carbon sequestration. Chemical degradation is quite evident in the lower part of the watershed.

The assessment of biodiversity using the indices selected in the methodology provided an initial picture of the state of plant diversity in the area. These biodiversity indices need to be combined with rapid assessments of soil biodiversity and faunal assessments, in order to provide a more complete picture of biodiversity. However, the measurement or counts of faunal species (both macro and micro) on the ground and in the soil need to be balanced against the time and effort required. It is clear that the methodology will need to incorporate standard techniques for rapid assessment of fauna and soil biodiversity.

Species identification and cataloguing proved to be the bottleneck in the field procedures for biodiversity assessment in the case studies. A botanist and a zoologist are essential members of a field assessment team. Indigenous knowledge of plant species should also be used for plant identification.

The problem of upscaling and “spatialization” of point estimates is a generic and important one. The accuracy of estimates (biomass, SOM, indicators of degradation, biodiversity indices, etc.) will depend on the accuracy of the spatial interpolation techniques used for estimation at sites not measured or observed.

The experiences with the application of the methods and procedures developed in the case studies indicate that the methodology can be applied to routine assessments of carbon stock and sequestration potential elsewhere in the Latin American and Caribbean region with a well-organized, multidisciplinary team of middle-level professionals, with relatively little training.

The time taken to carry out all the phases of this study, from research on methods to data processing and reporting, including the three case studies, was about two years. However, once the methodology is in place and the tools have been learned and customized, the time involved in an assessment and generation of scenarios is estimated at about four months for each case study, for areas comparable in size and provided that all the elements necessary for field and laboratory work are present.


The experiences generated in this project provided answers to many technical and practical questions. However, they also left some partially answered. While the methods appear to be complete in some aspects, in others there are obvious improvements to be made. There are seven specific recommendations for follow up:

Previous Page Top of Page Next Page