This sourcebook is intended to be a ready reference for practitioners (including World Bank stakeholders, clients in borrowing countries, and World Bank project leaders) seeking state-of-the-art information about good land management approaches, innovations for investments, and close monitoring for potential scaling up. This sourcebook is divided into three parts: the first part identifies the need and scope for sustainable land management (SLM) and food production in relation to cross-sector issues such as freshwater and forest resources, regional climate and air quality, and interactions with existing and emerging infectious diseases. It introduces the concept of production landscapes and analysis of trade-offs and establishes a framework for linking indicators that provide a measure of the outcomes of SLM. It then categorizes the diversity of land management (that is, farming) systems globally and the strategies for improving household livelihoods in each type of system. For the farming system types, a set of SLM principles and common but important issues for future investments are identified. The second part of the sourcebook focuses on three major farming system types and presents a range of investment notes and innovative activity profiles. The third part provides users of the source book with easy-to access, web-based resources relevant for land and natural resource managers. The resources are available in the public domain, and readers can access the web sites of various international and national agencies.

Soil respiration is the second largest flux of carbon between terrestrial ecosystems and the atmosphere and is affecting climate sensitivity and vulnerability of the terrestrial carbon stock. Monitoring soil carbon dioxide efflux is a complex task, due to the high spatial and temporal variability of the fluxes. For this reason, more than 30 sampling points are required to attain reliable estimates of ecosystem soil respiration. However, the number of sampled points is often limited by labour, time and budget constraints. Stratified sampling is an alternative to random sampling as a method to reduce the number of sampling points when an effective proxy variable is available for the definition of the strata. In order to evaluate different sampling strategies we tested, with a Monte Carlo simulation, the effectiveness of random and stratified samplings, using experimental data collected in three alpine ecosystems (two forests and one grassland). We evaluated an innovative method for defining the strata to be sampled. The method is based on an initial sampling of soil respiration from a large number of candidate points in order to account for the spatial variability. The minimum number of sampling points required to adequately represent soil respiration for the entire area were then selected using stratified statistical sampling. We show that this method is unbiased and that it reduces considerably the uncertainty in the sampling process compared to random sampling. The method was highly effective in the two forest ecosystems, characterized by a high spatial variability in soil respiration and by a high temporal correlation of the fluxes. On the contrary the method was not so effective in the grassland site, where fluxes have lower spatial variability and temporal correlation. However, the stratified sampling offered a consistent reduction of the error (%) of the estimated annual soil CO₂ efflux in all the ecosystems. At the grassland ecosystem the average reduction of the error (%) of the annual CO₂ efflux was about 12%, while at the forest ecosystems the average reductions were 55% and 57%, respectively.