Verification of soil carbon sequestration : uncertainties of assessment methods

suédois. Populärvetenskaplig sammanfattning: Kunskap om mängden kol som växter tar upp i från atmosfären är i dag en viktig frågeställning inom naturvetenskaplig forskning. Resultaten av sådan forskning kan bl.a. användas vid internationella konventioner som t.ex. Kyoto avtalet. I denna studie har jord prover från semi-arida agro-ekosystem i Sudan statistiskt analyserats för att se om skillnaden i organiskt kol över tid och rum kan detekteras (“minimum detectable difference” (MDD)). Proverna kom från tre olika fält. ”MDD” är den minsta detekterade skillnaden mellan medelvärden när variansen, signifikansnivån, statistiska styrkan och antalet prover är kända. Inom de tre fälten varierade variansen mellan 51 000 och 212 000. Den höga variabiliteten gör en verifiering av mängden kol som jorden kan ta upp svår. Över 1000 jord prov skulle behövas för att observera en 5 % förändring i organiskt kol (motsvarande 25 g m-2). En fördubbling av variansen innebär också ett fördubblat behov av antalet jordprover för att få samma “MDD”. En s. k. ”jackknife routine” användes för att simulera delar av det ursprungliga datasetet (de observerade jordproverna). Delarna representerade antalet jordprover och varierade mellan 2 och 99 samt simulerades 500 gånger var. Från resultaten beräknades medelvärden, variansen och deras 95 % konfidensintervall. Användning av det översta konfidensintervallet av variansen i ”MDD” beräkningar kan leda till fördubblat behov av antalet jordprover. En sensitivitets analys samt framtida scenarier utfördes med ekosystem modellen CENTURY. Målet av denna analys/scenarier var att observera effekten på organiskt kol vid varierande klimat (nederbörd, temperatur), markanvändning, jordens kornstorleksfördelning, andel efterlämnad biomassa efter skörd och kväve (N) fixering. Modellen visade sig vara mest känslig för mängden N fixering av träd. Effekten av klimatförändringar var mest framträdande då mängden N fixering var hög. Möjliga framtida klimatförändringar visade ingen ökning i organiskt kol. Olika typer av odlings metoder och grödor leder till signifikanta skillnader i mängden organiskt kol i marken. Hundra år kontinuerlig odling av hirs (Pennisetum typhoideum) leder till en förlust av 0,3 g C m-2 år-1 medan odling:träda ratio 5:10 och endast betning leder till en ökning av 0.4 g C m-2 år-1 respektive 1.5 g C m-2 år-1 . De presenterade resultaten är simulerade helt utan N fixering. Vid närvaro av N fixerande vegetation ökar samtliga presenterade resultat.

anglais. Knowing the amounts of atmospheric carbon soils can potentially sequester is gaining great interest among the scientific community. International conventions, such as the Kyoto Protocol, help promote such research. In this study, soils samples from a semiarid agro-ecosystem of the Sudan were statistically analyzed to evaluate if changes in soil organic carbon (SOC) over time or space were cost-effectively detectable or not, given a sample size (with minimum detectable difference (MDD) method). One hundred (100) samples were taken from each of three fields. The MDD is the smallest detectable difference between treatment means once the variation, significance level, statistical power and sample size are specified. The variances of the SOC contents varied between 51 000 and 212 000. This makes short-term verification of carbon sequestration difficult as over 1000 soil samples would be required to observe a 5% change (25 g m-2) in SOC over time or space. A doubling of the variance also doubles the required amount of soil samples to achieve the same MDD. A “jackknife routine” was setup to simulate alternative data sets from the available data (observed soil samples). These datasets represented sample sizes ranging from 2 to 99 and were extracted 500 times for each sample size. For the results, the mean, the variance and their 95% confidence interval were calculated. Using the upper 95% confidence interval for the variance estimates in the MDD calculations can lead to as much as a twofold increase in the sample size requirement compared with the average variance of all samples. A sensitivity analysis and future scenarios were done with the CENTURY ecosystem model to evaluate the effects of varying climates (precipitation and temperature), land management and harvest practices, soil texture and nitrogen (N) fixation by trees on SOC. The model showed to be most sensitive to N fixation by trees, and the effects of variations in climate were amplified when N fixation was allowed in the system. Probable future climate scenarios did not indicate a significant increase in SOC, whereas different land management practices did. Long fallow periods (crop:fallow ratio 5:10) and grazing showed the potential of this semi-arid area to sequester up to 0.4 g C m-2 yr-1 and 1.5 g C m-2 yr-1 respectively, while soils under continuous cultivation of millet (Pennisetum typhoideum) acted as a C source, losing 0.3 g C m-2 yr-1 over the next century. These numbers result in simulations without N fixation allowed from trees. The presence of N fixing legumes will increase the sequestration rates.

anglais. Summary: It is well documented that atmospheric carbon dioxide (CO2) is increasing globally. Many scientific studies also show that increased CO2 in the atmosphere may raise the mean global temperature and disturb climates in unforeseen ways. The Kyoto Protocol is a framework aiming for an international collaboration for the reduction of atmospheric greenhouse gases (GHG), of which CO2 is one of the most important. Within the Kyoto Protocol, there are some “mechanisms” designed to help countries reach their reduction targets as cost-effectively and efficiently as possible. One of those mechanisms is called the Clean Development Mechanism (CDM), which allows developed countries to invest in sustainable development projects in developing countries. Some studies show that improved land management practices can result in an increase of the rate at which carbon is sequestered from the atmosphere into soils, and this is one of the basis on which the CDM concept is designed. If significant investments are made to promote projects that could sequester carbon, it is important to be able to quantify as precisely as possible the changes in soil organic carbon (SOC) over time or space in order to relate the changes to the Kyoto Protocol and to justify adequate monetary sums for projects. A reason why soils should be considered as carbon “sinks” is that carbon in soils can have very long residence times (up to thousands of years) if it can reach a stable state in which is becomes protected from physical and chemical destructive processes. SOC can be estimated though direct soil sampling. The number of soil samples over an area will determine the accuracy of the estimates. The detection of changes in SOC over time or space will also be relevant to the number of soil samples taken each time or place. Three hundred soil samples (100 samples for each of 3 fields) from a semi-arid agroecosystem of the Sudan, Africa were statistically analyzed to see if the detection of such changes would be possible for short-term verification. The variance is a statistical parameter explaining the variability of the samples in regards to the mean value. A number of statistical tests are designed to evaluate if two or more groups of samples are significantly different or not. If a statistical test results in fields not being significantly different, then changes in SOC cannot be detected. A method known as the “minimum detectable difference” can provide the number of soil samples required to statistically differ sampling groups, given the variability (variance) of each group. This is helpful for the design of efficient sampling schemes for the verification and monitoring of SOC. Given the variability of the samples analyzed, more than 1000 samples would be required over at least two time periods or places in order to detect a 5% increase of the SOC. Nevertheless, as the variability decreases by half, so does the number of soil samples required to detect the same change in SOC. Estimating the variance for soil properties can be difficult because many factors can influence important changes over very small ground areas. For this reason, a “jackknife routine” was used. Such a routine is used to simulate alternative data sets from the available data (observed soil samples). These datasets represented sample sizes ranging from 2 to 99 and were extracted 500 times for each sample size. For the results, the mean, the variance and their 95% confidence interval were calculated. It is therefore possible to evaluate how representative are the overall mean and variance for each sampling group. Computer models can be used to simulate SOC from different scenarios. One of those models, called CENTURY, was tested to evaluate its sensitivity to variations in some input parameters concerning climate, land management and harvest practices, soil texture and plant physiology regarding the modeling of SOC. A number of future scenarios were also designed to have an idea of the faith of SOC in the study area for different land management practices and climate change scenarios. Future climate scenarios did not indicate important changes in SOC, but improved land management, such as the stop of continuous cultivation and the return of long fallow periods (crop:fallow ratio 5:10 years) did show important increases in SOC. More SOC has the benefit of increasing soil fertility, crop productivity (food security), as well as helping combat other serious problems such as desertification and the decline in biodiversity. In a world where adequate food supplies and climate can cause rigorous problems, such considerations should be reflected upon. It is nevertheless important to remember that scientific proof can only be valid if socio-economic and political issued are also considered.