Assessing and mapping of carbon in biomass and soil of mangrove forest and competing land uses in the Philippines

Assessing and mapping of carbon in biomass and soil of mangrove forest and competing land uses in the Philippines

Mangrove forests provide many ecosystem goods and services, and are important carbon (C) sinks in the tropics. Yet, land use conversions in mangroves still continue, especially in Southeast Asia. Carbon stocks in biomass and soil as well as the soil emissions of greenhouse gases (GHG) are important parameters to quantify, monitor and map in mangrove area, and are vital inputs for assessing the impact of mangrove conversion on C budget. This study was conducted in a section of tropical intertidal zone in Honda Bay, Philippines, with the following objectives: 1) evaluate the biomass C stocks in mangrove forests and land uses that replaced mangroves, 2) examine the potential of Sentinel satellite radar and multispectral imagery for mapping the aboveground biomass in mangrove area, 3) investigate the soil C stocks and the potential of GIS-based Ordinary Kriging for mapping the C stocks in mangrove soil, and 4) assess the soil fluxes of greenhouse gases and the potential of Ordinary Kriging for mapping the soil GHG fluxes. I used intensive field assessments, combined with laboratory analysis, remote sensing and GIS methods, to achieve the above objectives.

To address the first objective, the biomass C stocks of the study land uses were quantified. Their relationships with selected canopy variables were also evaluated. Results reveal that for mangrove forests, the mean biomass was 22.4 to 178.1 Mg ha-1, which store 10 to 80 MgC ha-1 or 47.9 MgC ha-1, on average. Leaf Area Index significantly correlated with mangrove biomass C. In contrast, the biomass C stock of the land uses that replaced mangroves was, on average, 97% less than that in mangrove forests, ranging from zero in salt pond and cleared mangrove, 0.04 Mg C ha-1 in abandoned aquaculture ponds, to 5.7 Mg C ha-1 in the coconut plantation. C losses in biomass from conversion were estimated at 46.5 Mg C ha-1, on average.

For the second objective, the potential of Sentinel imagery for the retrieval and predictive mapping of aboveground biomass in mangrove area was evaluated. I used both Sentinel SAR and multispectral imagery. Biomass prediction models were developed through linear regression and Machine Learning algorithms, each from SAR backscatter data, multispectral bands, vegetation indices, and canopy biophysical variables. The results show that the model based on biophysical variable Leaf Area Index (LAI) derived from Sentinel-2 was more accurate in predicting the overall aboveground biomass. However, the SAR-based model was more accurate in predicting the biomass in the usually deficient-to-low vegetation cover replacement land uses. These models had 0.82 to 0.83 correlation/agreement of observed and predicted value. Overall, Sentinel-1 SAR and Sentinel-2 multispectral imagery can provide satisfactory results in the retrieval and predictive mapping of aboveground biomass in mangrove area.

In the third objective, the soil C stocks of the study land uses were quantified to estimate C losses in soil owing to conversion. I also evaluated the potential of GIS-based Ordinary Kriging for predictive mapping of the soil C stock distribution in the entire study site. On average, the soil C stock of mangrove forests was 851.9 MgC ha-1 while that of their non-forest competing land uses was less than half at 365.15 MgC ha-1. Aquaculture, salt pond and cleared mangrove had comparable C stocks (453.6, 401, 413 MgC ha-1, respectively) and coconut plantation had the least (42.2 MgC ha-1). Overall, C losses in soil owing to land use conversion in mangrove ranged from 398 to 809 MgC ha-1 (mean: 486.8 MgC ha-1) or a decline of 57% in soil C stock, on average. It was possible to map the site-scale spatial distribution of soil C stock and predict their values with 85% overall certainty using Ordinary Kriging approach.

To achieve the fourth objective, the soil fluxes of CO2, CH4 and N2O in the study land uses were investigated using static chamber method. I also evaluated the potential of GIS-based Ordinary Kriging for predictive mapping of the soil GHG fluxes in the entire study site. Results show that the emissions of CO2 and CH4 were higher in mangrove forests by 2.6 and 6.6 times, respectively, while N2O emissions were lower by 34 times compared to the average of non-forest land uses. CH4 and N2O emissions accounted for 0.59% and 0.04% of the total emissions in mangrove forest as compared to 0.23% and 3.07% for non-forest land uses, respectively. Site-scale soil GHG flux distribution could be mapped with 75% to 83% accuracy using Ordinary Kriging.

This study has shown that C losses in biomass and soil arising from mangrove conversion are substantial (63%; 571 MgC ha-1). Moreover, mangrove conversion heavily altered the soil-atmosphere fluxes of GHG, increasing the N2O fluxes by 34 times. The use of Sentinel imagery for biomass mapping, as well as the application of Ordinary Kriging for soil mapping of C stocks and GHG fluxes, offer good potentials for mangrove area monitoring. This study advances current knowledge on the C stocks and soil GHG fluxes in mangrove area and the C emissions owing to mangrove conversion. The mapping techniques presented here contribute to advancing the knowledge for mapping the biomass and soil attributes in mangrove ecosystem.