Forest Resources Assessment and Tools to Provide Information for Forest Ecosystem Management

Imprint As a federally owned enterprise, GIZ supports the German Government in achieving its objectives in the field of international cooperation for sustainable development.

Published by: Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH

Registered offices Bonn and Eschborn

4th Floor Forest Management Bureau Annex Building Department of Environment and Natural Resources Compound Visayas Avenue, Diliman, Quezon City 1101, Philippines T +63 2 697 3127

Programmes: National REDD+ System Philippines Project Forest and Climate Protection in Panay – Phase II

Authors: Ralph Lennertz, Jürgen Schade and Vincent Barrois, DFS Deutsche Forstservice GmbH

Photo credits/sources: Ralph Lennertz

URL links: This publication contains links to external websites. Responsibility for the content of the listed external sites always lies with their respective publishers. When the links to these sites were first posted, GIZ checked the third- party content to establish whether it could give rise to civil or criminal liability. However, the constant review of the links to external sitescannot reasonably be expected without concrete indication of a violation of rights. If GIZ itself becomes aware or is notified by a third party that an external site it has provided a link to gives rise to civil or criminal liability, it will remove the link to this site immediately. GIZ expressly dissociates itself from such content.

Maps: The maps printed here are intended only for information purposes and in no way constitute recognition under international law of boundaries and territories. GIZ accepts no responsibility for these maps being entirely up to date, correct or complete. All liability for any damage, direct or indirect, resulting from their use is excluded.

On behalf of the German Federal Ministry for the Environment, Nature Conservation, Building and Nuclear Safety (BMUB)

GIZ is responsible for the content of this publication.

Printing and distribution: GIZ, Quezon City

Quezon City, Philippines | 2017

Table of Contents List of Tables ...... 1 List of Figures ...... 2 List of Boxes ...... 3 Acronyms ...... 4 1. Introduction and background ...... 6 2. Objectives and definitions ...... 9 2.1 Objectives ...... 9 2.2 Areal sampling frame ...... 11 2.3 Scope and content ...... 12 2.4 Variables of interest ...... 12 2.5 Targeted precision ...... 13 2.6 Definition of terms and concepts ...... 14 2.6.1 Forest ...... 14 2.6.2 Species abundance and diversity ...... 14 2.6.3 Carbon pools ...... 14 2.6.4 IPCC Key Categories and Tiers ...... 16 2.6.5 Forest sampling-related terms ...... 16 3. Inventory design ...... 18 3.1 Sources of information ...... 18 3.1.1 Geographical-political subdivisions ...... 18 3.1.2 Forest strata and areas ...... 18 3.1.3 Soil classes ...... 20 3.1.4 Wood specific gravity ...... 22 3.2 Inventory Method ...... 22 3.3 Sampling unit design ...... 22 3.3.1 Observations and measurements at the sample points ...... 24 3.3.2 Observations and measurements in the nested plots ...... 24 3.4 Sample size and margin of error ...... 26 3.5 Sampling type and distribution ...... 27 3.6 Human and material resources ...... 29 3.7 Organization of field work ...... 29 3.8 Estimation design ...... 30 3.8.1 Merchantable volume of trees ...... 31 3.8.2 Above-ground biomass of trees ...... 31 3.8.3 Above-ground biomass of bamboos ...... 32 3.8.4 Above-ground biomass of palms ...... 32 3.8.5 Above-ground biomass of rattan and tree ferns ...... 32 3.8.6 Below-ground biomass of trees, bamboos and palms ...... 33 3.8.7 Above-ground biomass of standing dead wood ...... 33 3.8.8 Biomass of lying (downed) dead wood ...... 33 3.8.9 Biomass of litter ...... 33 3.8.10 Conversion of biomass to carbon ...... 34 3.8.11 Diversity indices ...... 34 3.8.12 Statistical inference ...... 34 4. Field data collection ...... 37 4.1 Getting to and marking of sample points ...... 37 4.1.1 Getting to the sample points ...... 37 4.1.2 Location of sample points and plot centers...... 38 4.1.3 Permanent marking of sample points and plot centers ...... 38 4.1.4 Inaccessible sample points and plot centers ...... 39 4.2 Assessment or measurement of variables ...... 40 4.2.1 Administrative location ...... 40 4.2.2 Actual coordinates ...... 40 4.2.3 Elevation...... 40 4.2.4 Slope ...... 40 4.2.5 Slope orientation ...... 40 4.2.6 Terrain ...... 41 4.2.7 Land classification ...... 42 4.2.8 Land cover ...... 42 4.2.9 Forest type ...... 42 4.2.10 Tree crown cover ...... 43 4.2.11 Plant diversity ...... 43 4.2.12 Ground coverage classes by vegetation layers ...... 43 4.2.13 Ground coverage and average depth of litter ...... 44 4.2.14 Mid-diameter and length of lying dead wood sections ...... 44 4.2.15 Observations and measurements on trees and standing dead wood ...... 45 4.3 Quality assurance and quality control ...... 52 4.4. Time and cost of the data collection ...... 53 5. Data processing ...... 55 5.1 Software ...... 55 5.2 Database architecture ...... 55 5.3 Database system application ...... 58 5.4 Quality assurance and quality control ...... 59 6. Data analysis and results ...... 61 6.1 Species diversity...... 61 6.1.1 Species diversity of closed forests (Eastern Samar FRA) ...... 62 6.1.2 Species diversity of open forests (Eastern Samar FRA) ...... 64 6.2 Stand composition ...... 66 6.2.1 Stand composition of closed forests (Davao Oriental FRA) ...... 67 6.2.2 Stand composition of open forests (Davao Oriental FRA) ...... 70 6.3 Stand structure ...... 72 6.3.1 Stand structure of closed forests (Davao Oriental FRA) ...... 73 6.3.2 Stand structure of open forests (Davao Oriental FRA) ...... 79 6.4 Timber stocks ...... 85 6.4.1 Timber stocks in closed forests (Davao Oriental FRA) ...... 85 6.4.2 Timber stocks in open forests (Davao Oriental FRA) ...... 88 6.5 Forest carbon stocks ...... 90 6.5.1 Forest carbon stocks of closed forests (Panay Mountain Range FRA) ...... 91 6.5.2 Forest carbon stocks of open forests (Panay Mountain Range FRA) ...... 92 6.5.3 Forest carbon stocks of mangroves (Panay Mountain Range FRA) ...... 93 6.6 Uncertainty of the estimates ...... 94 6.6.1 Statistical sampling error ...... 95 6.6.2 Representativeness of the sampling network ...... 96 6.6.3 Measurement errors ...... 96 6.6.4 Data entry errors ...... 97 6.6.5 Estimation design uncertainties ...... 97 6.6.6 Overall error budget ...... 97 7. Considerations for up-scaling ...... 99 7.1 Comprehensive specification of the objectives ...... 99 7.2 Efficient inventory design ...... 108 7.3 Cautious field data collection ...... 111 7.4 Tailored data processing ...... 113 7.5 Comprehensive data analysis ...... 114 8. References ...... 115 Appendix 1: Field data forms ...... 118 Annex 1: List of recorded species ...... 121

List of Tables Table 1: 2010 Land and Forest Cover Areas of the Sub-National FRAs ...... 11 Table 2: IPCC Tier 1 Soil Organic Matter Stocks ...... 20 Table 3: Overview of sub-plot sizes and assessments or measurements made on trees and dead wood ...... 26 Table 4: Sample sizes and margins of error of the sub-national FRAs ...... 27 Table 5: Deviations of initial measurements from control measurements ...... 53 Table 6: Time and cost of data collection ...... 54 Table 7: Deviations of stored data from field forms ...... 60 Table 8: Species diversity indices ...... 61 Table 9: Threatened species (Eastern Samar FRA) ...... 62 Table 10: Relative frequency, density and dominance, importance and rank of the 20 most "important" species in closed forests (Eastern Samar FRA) ...... 63 Table 11: Relative frequency, density and dominance, importance and rank of the 20 most "important" species in open forests (Eastern Samar FRA) ...... 65 Table 12: Stand composition...... 67 Table 13: Stand composition of closed forests (Davao Oriental FRA) ...... 68 Table 14: Stand composition of open forests (Davao Oriental FRA) ...... 70 Table 15: Stand structure ...... 72 Table 16: Stand structure in terms of N/ha of closed forests (Davao Oriental FRA) ...... 73 Table 17: Stand structure in terms of G/ha of closed forests (Davao Oriental FRA) ...... 75 Table 18: Stand structure in terms of AGB/ha of closed forests (Davao Oriental FRA) ...... 77 Table 19: Stand structure in terms of N/ha of open forests (Davao Oriental FRA) ...... 79 Table 20: Stand structure in terms of G/ha of open forests (Davao Oriental FRA) ...... 81 Table 21: Stand structure in terms of AGB/ha of open forests (Davao Oriental FRA) ...... 83 Table 22: Timber stocks ...... 85 Table 23: Merchantable volume in closed forests (Davao Oriental FRA) ...... 86 Table 24: Merchantable volume in open forests (Davao Oriental FRA) ...... 88 Table 25: Forest carbon stocks ...... 90 Table 26: Carbon stocks of closed forests (Panay Mountain Range FRA) ...... 92 Table 27: Carbon stocks of open forests (Panay Mountain Range FRA) ...... 93 Table 28: Carbon stocks of mangroves (Panay Mountain Range FRA) ...... 94 Table 29: Means and coefficients of variation of N/ha, G/ha, V/ha and AGB/ha ...... 95 Table 30: Statistical sampling errors of the main variables of interest ...... 96 Table 31: Overall error budget estimating V/ha ...... 98 Table 32: Overall error budget estimating AGB/ha ...... 98 Table 33: Framework for the definition of variables of interest (example) ...... 100 Table 34: Framework for the definition of ancillary variables (example) ...... 103

1

List of Figures Figure 1: Major Phases and Outputs of the FRAs ...... 7 Figure 2: Forest carbon pools (DiRocco et al. 2014) ...... 15 Figure 3: 2010 NAMRIA Land Cover Map of the Philippines ...... 19 Figure 4: 2013 BSWM FAO Soil Map of the Philippines ...... 21 Figure 5: Sampling unit design ...... 23 Figure 6: Distribution of the Sampling Units Effectively Measured (Eastern Samar FRA) ...... 27 Figure 7: Inventory camp. A) Batch of sample points (red) assigned to one inventory camp in Eastern Samar; B) Inventory camp in the Panay Mountain Range ...... 30 Figure 8: Open cycle map with "Outdoors" base layer Cangaranan River, Valderrama, Antique ...... 37 Figure 9: Google Maps v. Apple Maps Barangays Pinanag-an (Borongan City) and Patag (Maydolong) on the Suribao River, Eastern Samar ...... 38 Figure 10: Location and marking of sample points and plot centers. A) On the way to a sample point in Malinao, Aklan; B) Permanent marking of a sample point using an iron rod topped with a PVC pipe ...... 38 Figure 11: Re-location of inaccessible plots ...... 39 Figure 12: Measurements on lying dead wood sections ...... 44 Figure 13: DBH and DAB measurements (Zöhrer 1980) ...... 48 Figure 14: Diameter estimates for inaccessible measurement points ...... 49 Figure 15: Measuring and recording data ...... 52 Figure 16: Diagram of the FRA database ...... 57 Figure 17: N/ha, G/ha, V/ha and AGB/ha by number of species in closed forests (Eastern Samar FRA) ...... 64 Figure 18: N/ha, G/ha, V/ha and AGB/ha by number of species in open forests (Eastern Samar FRA) ...... 66 Figure 19: Stand composition of closed forests (Davao Oriental FRA) ...... 69 Figure 20: Stand composition of open forests (Davao Oriental FRA)...... 71 Figure 21: Stand structure in terms of N/ha of closed forests (Davao Oriental FRA) ...... 74 Figure 22: Stand structure in terms of G/ha of closed forests (Davao Oriental FRA) ...... 76 Figure 23: AGB/ha of closed forests by DBH threshold (Davao Oriental FRA) ...... 77 Figure 24: Stand structure in terms of AGB/ha of closed forests (Davao Oriental FRA) ...... 78 Figure 25: Stand structure in terms of N/ha of open forests (Davao Oriental FRA) ...... 80 Figure 26: Stand structure in terms of G/ha of open forests (Davao Oriental FRA) ...... 82 Figure 27: AGB/ha of open forests by DBH threshold (Davao Oriental FRA) ...... 83 Figure 28: Stand structure in terms of AGB/ha of open forests (Davao Oriental FRA) ...... 84 Figure 29: Merchantable volume in closed forests (Davao Oriental FRA) ...... 87 Figure 30: Merchantable volume in open forests (Davao Oriental FRA) ...... 89 Figure 31: Carbon stocks of closed forests (Panay Mountain Range FRA) ...... 92 Figure 32: Carbon stocks of open forests (Panay Mountain Range FRA) ...... 93 Figure 33: Carbon stocks of mangroves (Panay Mountain Range FRA) ...... 94

2

List of Textboxes Textbox 1: International Climate Initiative (IKI) ...... 8 Textbox 2: National REDD+ System Philippines Project ...... 8 Textbox 3: Forest and Climate Protection Project Panay - Phase II ...... 8 Textbox 4: IPCC Guidelines for National GHG Inventories ...... 10 Textbox 5: Stock Difference Method ...... 10 Textbox 6: Population and Sampling Frame in Forest Inventories ...... 11 Textbox 7: Accuracy and Precision ...... 13 Textbox 8: Optimal Sampling Unit Design ...... 24 Textbox 9: Principal Sampling Types ...... 28 Textbox 10: GlobAllomeTree ...... 31 Textbox 11: Margin of Error and Confidence Level...... 36 Textbox 12: Averaging Coordinate Measurements with GARMIN GPS Receivers ...... 40 Textbox 13: SUUNTO PM-5/360 Clinometer and Dendrometer ...... 41 Textbox 14: LTI TruePulse Laser 200 Rangefinder ...... 46 Textbox 15: Horizontal Distance Measurements with the LTI TruePulse Laser 200 ...... 47 Textbox 16: Height Measurements with the LTI TruePulse Laser 200 ...... 50 Textbox 17: Height Measurements with the SUUNTO PM-5/360 ...... 51 Textbox 18: Relational Database ...... 56 Textbox 19: Open Foris ...... 59 Textbox 20: AGB Estimation of Trees ...... 91 Textbox 21: Map Accuracy ...... 105 Textbox 22: FAO's Variable Set for National Forest Monitoring and Assessment ...... 108 Textbox 23: FAO's Sampling Unit Design for National Forest Monitoring and Assessment ...... 110

3

Acronyms AD Activity data AFOLU Agriculture, Forest and Other Land Use AFP Armed Forces of the Philippines AGB Above-ground biomass ALOS Advanced Land Observing Satellite a.s.l. Above sea level ASS Aligned Systematic Sampling AVNIR Advanced Visible and Near Infrared Radiometer BCEF Biomass Conversion and Expansion Factor BGB Below-ground biomass BMUB Bundesministerium für Umwelt, Naturschutz, Bau und Reaktorsicherheit (Federal Ministry for the Environment, Nature Conservation, Building and Nuclear Safety) BSWM Bureau of Soils and Water Management BUR Biennial Update Report C Carbon CENRO Community Environment and Natural Resources Office(r) CIRAD Centre de Coopération Internationale en Recherche Agronomique pour le Développement COP Conference of the Parties DAB Diameter above buttress DBH Diameter at breast height DEM Digital Elevation Model DENR Department of Environment and Natural Resources DFS Deutsche Forstservice GmbH Dg Quadratic mean diameter DOM Dead organic matter Dref Reference diameter ed. editor eds. editors EF Emission Factor FAO Food and Agriculture Organization FMB Forest Management Bureau FRA Forest Resources Assessment FRL Forest Reference Level GHG Greenhouse gas GIGO Garbage In - Garbage Out GIS Geographic Information System GIZ Deutsche Gesellschaft für Internationale Zusammenarbeit GmbH GPS Global positioning system GUI Graphical user interface HAC High activity clay HWSD Harmonized World Soil Database IGES Institute for Global Environmental Strategies ILUA Integrated Land Use Assessment IP Internet protocol IPCC Intergovernmental Panel on Climate Change JDK Java Development Kit JRE Java Runtime Environment LAC Low activity clay LB Living biomass LDW Lying dead wood

4

LGU Local government unit LI Litter LULUCF Land Use, Land-Use Change and Forestry MAD Mean absolute deviation MAP Mean annual precipitation MAT Mean annual temperature MNR Ministry of Natural Resources MRV Measurement, Reporting and Verification NAMRIA National Mapping and Resource Information Authority NetCDF Network Common Data Format NFMA National Forest Monitoring and Assessment NGO Non Governmental Organization NSCB National Statistical Coordination Board NTFP Non-timber forest product ODBC Open Database Connectivity p. page PNRPS Philippine National REDD+ Strategy POI Point of interest PSGC Philippine Standard Geographic Code QA Quality assurance QC Quality control RDBMS Relational Database Management System REDD+ Reducing Emissions from Deforestation and forest Degradation, and conservation, sustainable management of forests and enhancement of carbon stocks RFID Radio-frequency identification RMSD Root mean square deviation SDW Standing dead wood SE Standard edition SFM Sustainable Forest Management SINP Samar Island Natural Park SLC Scan line corrector SOM Soil organic matter SOP Standard operating procedure SQL Structured Query Language SRS Simple Random Sampling SRTM Shuttle Radar Topography Mission SU Sampling unit TCP Transmission Control Protocol TOF Trees outside forests UHF Ultra-high frequency UNCBD United Nation Convention on Biological Diversity UNEP United Nations Environmental Program UNFCCC United Nations Framework Convention on Climate Change USS Unaligned Systematic Sampling UTM Universal Transverse Mercator WGS World Geodetic System WMO World Meteorological Organization WRB World Reference Base for soil resources

5

1. Introduction and background A nation's forests cannot be properly managed without reliable information on the current conditions and trends of the resources. Moreover, to participate in the REDD+ mechanism under the United Nations Framework Convention on Climate Change (UNFCCC) to avail of results-based payments, a country must periodically measure and report man-made emission reductions and removals compared to a forest reference level (FRL). For these purposes, maps, complemented with data, some of which can only be acquired in the field, are needed.

Acknowledging the need for baseline data to start with, collected in such a way that future change could be detected to measure the impact of the policies and measures promoted towards sustainable forest management and biodiversity conservation, the National REDD+ System Philippines Project (see Textbox 2) and the Forest and Climate Protection Project Panay - Phase II (see Textbox 3), jointly implemented by the Philippine Department of Environment and Natural Resources (DENR) and the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH with funding from the International Climate Initiative (IKI) (see Textbox 1) of the German Federal Ministry for the Environment, Nature Conservation, Building and Nuclear Safety (BMUB), have implemented forest resources assessments (FRA) in their project sites, using state-of-the-art methodologies. The inventories have been conducted in the following jurisdictions: • In the Province of Eastern Samar, covering the two partner local government units (LGU) of Borongan City and Maydolong, from December 2014 to July 2015; • In the Province of Davao Oriental, covering the three partner LGUs of Caraga, Manay and Tarragona, from August 2015 to March 2016; • In the Panay Mountain Range, covering the 38 partner LGUs of Buruanga, Ibajay, Libacao, Madalag, Makato, Malay, Malinao and Nabas in Aklan, Anini-Y, Barbaza, Belison, Bugasong, Culasi, Hamtic, Laua-An, Libertad, Pandan, Patnongon, San Jose, San Remigio, Sebaste, Sibalom, Tibiao , Tobias Fornier (formerly known as Dao) and Valderrama in Antique, Jamindan and Tapaz in Capiz, and Alimodian, Calinog, Igbaras, Janiuay, Lambunao, Leon, Maasin, Miagao, San Joaquin and Tubungan in Iloilo, from February 2015 to December 2015.

The purpose of this document is to outline the FRA methodology used, to describe the software tool developed for the management and processing of the data, to illustrate the wealth of information that the acquired data yield for analyses under various perspectives, to share the lessons learned and to elaborate on whether and how the FRA methodology could and/or should be adjusted and/or enhanced to be used at wider scales and for broader scopes.

The structure of the document follows the major phases of FRA preparation and implementation depicted in Figure 1. Chapter 2 relates the various elements including important definitions collectively defining the objectives of the FRA. The inventory and estimation designs are comprehensively described in Chapter 3, where appropriate with justifications of strategic choices. The details about the implementation of the three sub-national FRAs are provided in Chapter 4 concerning to the field data collection and in Chapter 5 concerning the data processing. Data analysis from various perspectives, notably (plant) species diversity, stand composition and structure, timber and carbon stocks, as well as an approach to report uncertainties are illustrated in Chapter 6. Chapter 7 develops considerations for up-scaling, taking into consideration the lessons learned in the course of the three sub-national implementations, and the potential objectives at a wider scale.

Throughout the document, textboxes provide supplemental information on selected aspects related to the topics exposed in the main text.

6

Figure 1: Major Phases and Outputs of the FRAs

7

Textbox 1: International Climate Initiative (IKI)

Since 2008, the International Climate Initiative (IKI) of the German Federal Ministry for the Environment, Nature Conservation, Building and Nuclear Safety (BMUB) cooperates with developing and newly industrialized partner countries in the practical implementation of climate change mitigation and biodiversity conservation measures. To date, it has launched 500 climate and biodiversity projects with a total project volume of 1.7 billion euro. The resources for international climate and biodiversity activities have grown steadily since the program was launched. Whereas some 1.70 million euro was available for on-going projects in 2008, by 2014 the figure had risen to 318 million euro. In recent years, the German Government has steadily increased its commitments to climate change mitigation measures.

The projects are carried out by a broad range of entities: the German Government's major implementing organizations Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH and KfW Entwicklungsbank, multilateral organizations like United Nations organizations and multilateral development banks, NGOs, research institutes, foundations and private companies.

IKI focuses on four funding areas: (i) mitigating greenhouse gas emissions, (ii) adapting to the impacts of climate change, (iii) conserving natural carbon sinks with a focus on reducing emissions from deforestation and forest degradation (REDD+), and (iv) conserving biological diversity.

See https://www.international-climate-initiative.com/en/ for further information.

Textbox 2: National REDD+ System Philippines Project

The National REDD+ System Philippines Project aims at creating a national framework for reducing greenhouse gas emissions from deforestation and forest degradation, based on recognized ecological and social safeguards. Measures include (i) the establishment of a national implementation and coordination system for REDD+ (registry, governance, coordination and monitoring structures), (ii) the development of financing and benefit-sharing mechanisms for REDD+, (iii) the integration of ecological, social and governance standards (safeguards) in the implementation of REDD+. (iv) forest land use planning and REDD+ implementation in selected areas in Albay, Davao Oriental and Eastern Samar to prevent deforestation and forest degradation while providing co-benefits of livelihood improvements and biodiversity conservation, and (v) awareness building as well as information and knowledge management. Implemented by the Philippine Department of Environment and Natural Resources (DENR), the Project supports the implementation of the Philippine National REDD+ Strategy (PNRPS) by assisting the process towards REDD+ readiness.

Textbox 3: Forest and Climate Protection Project Panay - Phase II The Forest and Climate Protection Project Panay - Phase II aims at the protection of the Panay Mountain Range with its globally significant biodiversity and at the sustainable and climate friendly management and use of the natural resources in the adjacent areas through the upscaling and mainstreaming of the innovative and successful approaches to (i) forest land use planning (FLUP), (ii) participatory planning and implementation of forest conservation and management, (iii) forest rehabilitation, as well as (iv) agroforestry and income generation developed during the Project's first phase from August 2010 to May 2014. Implemented by the Philippine Department of Environment and Natural Resources (DENR), the Project contributes to (i) the implementation of the Strategic Plan 2011 - 2020 of the United Nations Convention on Biological Diversity (UNCBD) with focus on Aichi target 11, (ii) the achievement of global targets for Sustainable Forest Management (SFM) and (iii) the Reduction of Emissions from Deforestation and Forest Degradation (REDD+).

8

2. Objectives and definitions This chapter provides the objectives of the sub-national FRAs (Chapter 2.1). The subsequent chapters refine the objectives in terms of the areal sampling frames (Chapter 2.2), the elements sampled (Chapter 2.3), the variables of interest to be estimated (Chapter 2.4) and the precision to be achieved (Chapter 2.5). Important terms and concepts used are defined in Chapter 2.6. 2.1 Objectives The sub-national FRAs aimed to provide information about the forest condition and carbon stocks for "key" forest strata. This information comprises: • Stand and stock data estimates, such as: o Species abundance and diversity, o Stand density (N), o Basal area (G), o Merchantable volume (V), o Stand composition (proportions of species or species groups in terms of N, G and V), o Stand structure (distribution of N, G and V by diameter classes),

• Forest carbon stock estimates for key carbon pools: o Living biomass (LB), composed of: - Above-ground biomass (AGB), - Below-ground tree biomass (BGB), o Dead organic matter (DOM), composed of: - Dead wood (DW), - Litter (LI), o Soil organic matter (SOM).

The data should satisfy the requirements of the 2006 Intergovernmental Panel on Climate Change (IPCC) guidelines for national greenhouse gas (GHG) inventories (see Textbox 4) in the Agriculture, Forestry and Other Land Use (AFOLU) sector (IPCC 2006b) to: • Determine emission factors (EF), • Estimate the change of carbon stocks using the "Stock Difference Method" (see Textbox 5) by providing initial forest carbon stocks (at T0, prior to the implementation of REDD+ eligible activities).

The attribute "key" attached to forest strata and carbon pools is used in the sense of the key category analysis of the IPCC guidelines (see Chapter 2.6.4).

Living biomass and dead organic matter of key forest strata and key carbon pools are to be estimated using tier 3 methods. Soil organic matter in general and forest carbon stocks of non-key forest strata may be estimated using tier 1 methods (see Chapter 2.6.4 for the definition of the tiers).

9

Textbox 4: IPCC Guidelines for National GHG Inventories

Recognizing the problem of potential global climate change, the World Meteorological Organization (WMO) and the United Nations Environment Program (UNEP) co-established in 1988 the Intergovernmental Panel on Climate Change (IPCC). One of the IPCC's activities is to support the United Nations Framework Convention on Climate Change (UNFCCC) through its work on methodologies for national greenhouse gas (GHG) inventories.

Non-Annex I parties (developing countries) of the UNFCCC are required to submit to the secretariat national communications every four years and biennial update reports (BUR) every two years, including estimates of anthropogenic GHG emissions by sources and removals by sinks. The requirements, scattered over multiple Conference of the Parties (COP) decisions, have been compiled by the United Nations Climate Change Secretariat (2014) into a handbook on measurement, reporting and verification (MRV).

Either the 2006 IPCC or at the minimum the revised 1996 IPCC guidelines, complemented by the 2003 IPCC Good Practice Guidance for Land Use, Land-Use Change and Forestry (LULUCF), must be used. These guidelines provide detailed methods for the estimation of GHG emissions. The 2006 IPCC guidelines update and synthesize all previous guidelines. While there are some structural changes in the 2006 guidelines, including the combination of the previously separate Agriculture and LULUCF sectors into the single Agriculture, Forestry and Other Land Use (AFOLU) sector, for the most part, the inventory methods in the 2006 guidelines are updates of the previous editions (e.g. additional sources, new default emission factors).

Textbox 5: Stock Difference Method

The 2006 IPCC guidelines for national GHG inventories propose two methods of calculating carbon stock changes in a given carbon pool for a given land-use category in the Agriculture, Forestry and Other Land Use (AFOLU) sector: • The "Gain - Loss Method" also referred to as "Mass Balance", estimating the difference between increases (transfer from another carbon pool or increase of biomass [removal]) and decreases (transfer to another carbon pool or emissions) of the amount of carbon; • The "Stock Difference Method", estimating the change of carbon stocks through measurements at two (or more) points in time (which reflects the emissions and removals).

The stock difference method is robust, transparent and most common to monitor the carbon stock changes from the five activities eligible under REDD+, namely (i) reducing emissions from deforestation, (ii) reducing emissions from forest degradation, (iii) conservation of forest carbon stocks, (iv) sustainable management of forests, and (v) enhancement of forest carbon stocks.

The stock difference method requires two estimations: • "Activity Data" (AD), defined as "data on the magnitude of human activity resulting in emissions or removals taking place during a given period of time". In the forest sub-sector, AD correspond to forest land converted to other land uses to the area deforested, and for forest land remaining forest land to the area changes between the different forest strata. • "Emission Factor" (EF), defined as "a coefficient that relates the activity data to the amount of chemical compound which is the source of later emissions". In the forest sub-sector, EF correspond to the CO2 equivalent of the carbon stock per unit area of a forest stratum.

Emissions per stratum are estimated though the multiplication of the activity data by the emission factor:

Source: FAO (2013)

10

2.2 Areal sampling frame The areal sampling frames (see Textbox 6) consisted of key forest strata, namely closed forests and open forests. Mangroves have been excluded from the FRAs, because they represent less than 5% of the total forest area of the project sites. The FRAs covered the following jurisdictional units: • In the province of Davao Oriental the municipalities of Caraga, Manay and Tarragona. • In the province of Eastern Samar the city of Borongan and the municipality of Maydolong (practically covering the entire Suribao River watershed). • In the Panay Mountain Range: o In the province of Aklan the municipalities of Buruanga, Ibajay, Libacao, Madalag, Makato, Malay, Malinao and Nabas; o In the province of Antique the municipalities of Anini Y, Barbaza, Belison, Bugasong, Culasi, Hamtic, Laua An, Libertad, Pandan, Patnongon, San Jose, San Remigio, Sebaste, Sibalom, Tibiao, Tobias Fornier (formerly known as Dao) and Valderrama; o In the province of Capiz the municipalities of Jamindan and Tapaz; o In the province of Iloilo the municipalities of Alimodian, Calinog, Igbaras, Janiuay, Lambunao, Leon, Maasin, Miagao, San Joaquin and Tubungan.

The forest cover and strata boundaries were taken from the 2010 NAMRIA national forest cover map (see Chapter 3.1.2). Table 1 below summarizes the area statistics.

Table 1: 2010 Land and Forest Cover Areas of the Sub-National FRAs

FRA Closed Open Mangroves Others Total forests forests land area [ha] [ha] [ha] [ha] [ha] Davao Oriental 21,750 17,465 0 95,412 134,627 (3 municipalities) Eastern Samar 5,815 36,264 505 88,993 131,577 (2 municipalities) Panay Mountain Range 47,882 69,742 389 526,019 644,032 (37 municipalities) Total 75,447 123,471 894 710,424 910,236

Textbox 6: Population and Sampling Frame in Forest Inventories

From the statistical point of view, the population from which a forest inventory takes a sample is not the biological population of trees.

Sampling in forestry is based on the selection of sample points, not of trees. For each sample point, one or several observations and/or measurements of the values of variables are taken on population units (e.g. trees, dead wood, litter, etc.) selected according to the design of the sampling unit (e.g. fixed area plots, transects, etc.). The population actually consists of the sample points with their associated observation(s). Since sample points are dimensionless, the population is infinite ("infinite population approach") even in a limited area of interest.

The sampling frame is a list of all elements that can be selected during sampling. Since the population is infinite, the sampling frame cannot be defined by such a list, but rather by the area (areal sampling frame) to be covered by the inventory.

11

2.3 Scope and content The elements sampled in the sub-national FRAs consist of the following: • Trees (all species, including bamboos, palms, rattan and tree ferns; in the remainder of this document, the term "tree(s)" refers to all these woody vascular plants, unless specific taxa are mentioned) with a diameter at breast height (DBH) or diameter above buttress (DAB) ≥ 5 cm. • Dead wood, both standing and lying, down to a small end diameter of 5 cm (the smaller fractions are part of the litter). • Litter. • Soil type.

Trees with a DBH or DAB < 5 cm have been excluded from the FRAs and the forest carbon stock estimates because more than 96% of the AGB of tropical forests is found in trees with DBH or DAB ≥ 10 cm (Gillespie et al. 1992). This is supported by Lasco et al. (2006), who reported that 98% of the AGB in Surigao del Sur is found in trees with DBH or DAB ≥ 19.5 cm.

The inventory threshold for trees and dead wood thus consistently amounts to 5 cm (in diameter). 2.4 Variables of interest The FRAs aim to provide estimates of the following variables (also called attributes) of interest (see Chapter 2.6 for the definitions of the terms), disaggregated as much as possible and applicable by forest strata, jurisdictions, species and diameter classes: • For trees: o Species richness, o Berger-Parker, Margalef, Shannon (H' and E) and Simpson diversity indices, o Relative frequency, o Relative density, o Relative dominance, o Importance, o Quadratic mean diameter (Dg), o Density (N), o Basal area (G), o Merchantable volume (V), o Above-ground biomass (AGB), o Below-ground biomass (BGB),

• For dead wood: o Density (N) of standing dead wood, o Basal area (G) of standing dead wood, o Volume (V) of standing and of lying dead wood, o Above-ground biomass (AGB) of standing and of lying dead wood,

• Biomass of litter (LI), • Soil organic matter (SOM), • Forest carbon stock (C).

The estimates of N, G, V, AGB, BGB, LI, SOM and C are to be calculated including all following statistical parameters: • Sample size (n), • Mean (y̅), • Variance (s²), • Standard variation (s),

12

• Coefficient of variation (s%), • Standard error of the mean (S), • Margin of error (E) at 90%, 95% and 99% confidence level.

2.5 Targeted precision An inventory using probabilistic (statistical) sampling should set a targeted precision for the margin of error (E%) at a specific confidence level (see Chapter 3.8.12) that should not be exceeded for one or several variables of interest. The difference between precision and accuracy is explained in Textbox 7. The knowledge (from former inventories, or through an exploratory inventory) of the coefficient of variation (s%) allows then to determine the sample size (n):

푡2 × 푠%2 푛 = (0) 퐸%2 with • n sample size (total number of sampling units); • t two-tailed Student t-value with 푛 degrees of freedom; • s% coefficient of variation; • E% targeted margin of error at the desired confidence level (typically 90%, 95% or 99%).

In the real world, the sample size represents typically a compromise, considering the available time and budget. The three sub-national FRAs strived each to estimate the total forest carbon stock using some 200 sampling units with a margin of error at 90% confidence level hopefully not exceeding 10%, time and budget permitting. The sample sizes actually achieved and the resulting margins of error are reported in Chapter 3.4.

Textbox 7: Accuracy and Precision

Accuracy and precision are two terms that are often used equivalently, although they do not have the same meaning as illustrated below.

Source: Kleinn (2013)

The accuracy of estimates is the degree of closeness to the actual (true) value. A simple definition of accuracy is therefore the freedom from mistake or error: exactness.

The precision of estimates is the degree to which repeated estimates under unchanged conditions show the same results. The precision is estimated through the statistical standard error, or through the confidence interval or the margin of error at a given confidence level.

13

2.6 Definition of terms and concepts The sub-national FRAs used a number of terms and concepts, whose definitions (and as much as applicable sources) are detailed hereafter.

2.6.1 Forest Forest is defined according to DENR Memorandum Circular 2005-005 of 26 May 2005 as "Land with an area of more than 0.5 ha and tree crown (or equivalent stocking level) of more than 10%. The trees should be able to reach a minimum height of 5 m at maturity in situ. It consists either of closed forest formations where trees of various storeys and undergrowth cover a high portion of the ground or open forest formations with a continuous vegetation cover in which tree crown cover exceeds 10%. Young natural stands and all plantations established for forestry purposes, which have yet to reach a crown density of more than 10% or tree height of 5 m are included under forest. These are normally forming part of the forest area which are temporarily unstocked as a result of human intervention or natural causes but which are expected to revert to forest. It includes forest nurseries and seed orchards that constitute an integral part of the forest; forest roads, cleared tracts, firebreaks and other small open areas; forest within protected areas; windbreaks and shelter belts of trees with an area of more than 0.5 ha and width of more than 20 m; plantation primarily used for forestry purposes, including rubber wood plantations. It also includes bamboo, palm and fern formations (except coconut and oil palm)."

Coconuts and oil palms occurring in forests are part of the forest carbon stock.

2.6.2 Species abundance and diversity The following terms and/or indices are commonly used in ecological studies to measure species abundance and diversity: • The relative frequency of a particular species is defined as the proportion in percent (%) of the sampling units where that species has been sampled. • The relative density of a particular species is defined as its proportion in percent (%) of the total density (N), all species combined. • The relative dominance of a particular species is defined as its proportion in percent (%) of the total basal area (G), all species combined. • The importance of a particular species, typically used to determine the rank of species, is defined as the sum of its relative frequency, density and dominance. • Species richness refers to the number of (different) species. • The Margalef index measuring species richness. • The Shannon H' index measuring species abundance. • The Shannon E index measuring evenness. • The Berger-Parker and Simpson indices measuring species dominance.

2.6.3 Carbon pools Following the 2006 IPCC guidelines for national GHG inventories in the AFOLU sector (IPCC 2006b, p. 1.9), the forest carbon pools are defined as follows (see also Figure 2 for an illustration): • Living biomass (LB), composed of: o Above-ground biomass (AGB), defined as "All biomass of living vegetation, both woody and herbaceous, above the soil including stems, stumps, branches, bark, seeds, and foliage. In cases where forest understory is a relatively small component of the above- ground biomass carbon pool, it is acceptable for the methodologies and associated data used in some tiers to exclude it, provided the tiers are used in a consistent manner throughout the inventory time series." The sub-national FRAs set the inventory threshold (minimum diameter) for the living vegetation to 5.0 cm. They excluded the herbaceous vegetation, which normally does not contribute much to the forest carbon stock.

14

o Below-ground tree biomass (BGB), defined as "All biomass of live roots. Fine roots of less than (suggested) 2 mm diameter are often excluded because these often cannot be distinguished empirically from soil organic matter or litter."

• Dead organic matter (DOM), composed of: o Dead wood (DW), defined as follows: "Includes all non-living woody biomass not contained in the litter, either standing, lying on the ground, or in the soil. Dead wood includes wood lying on the surface, dead roots, and stumps, larger than or equal to 10 cm in diameter (or the diameter specified by the country)." The sub-national FRAs set the inventory threshold (minimum diameter) for dead wood to 5.0 cm. o Litter (LI), defined as follows: "Includes all non-living biomass with a size greater than the limit for soil organic matter (suggested 2 mm) and less than the minimum diameter chosen for dead wood (e.g. 10 cm), lying dead, in various states of decomposition above or within the mineral or organic soil. This includes the litter layer as usually defined in soil typologies. Live fine roots above the mineral or organic soil (of less than the minimum diameter limit chosen for below-ground biomass) are included in litter where they cannot be distinguished from it empirically." The sub-national FRAs set the inventory threshold (minimum diameter) for dead wood to 5 cm. Hence, litter comprises all non-living biomass with a size greater than 2 mm and less than 5 cm.

• Soil organic matter (SOM) "includes organic carbon in mineral soils to a specified depth chosen by the country and applied consistently through the time series. Live and dead fine roots and DOM within the soil, that are less than the minimum diameter limit (suggested 2 mm) for roots and DOM, are included with soil organic matter where they cannot be distinguished from it empirically. The default for soil depth is 30 cm."

Figure 2: Forest carbon pools (DiRocco et al. 2014)

15

2.6.4 IPCC Key Categories and Tiers According to the 2006 IPCC guidelines for national GHG inventories, Volume 1 (IPCC 2006a), a key category is "one that is prioritized within the national inventory system because its estimate has a significant influence on a country's total inventory of greenhouse gases in terms of the absolute level, the trend, or the uncertainty in emissions and removals." (IPCC 2006a, p. 4.5).

Two approaches for performing the key category analysis should be used concomitantly: • According to Approach 1, key categories are those that, when summed together in descending order of magnitude, add up to 95% of the total level (IPCC 2006a, p. 4.12). • According to Approach 2, key categories are those that, when summed together in descending order of magnitude, add up to 90% of the total uncertainty (IPCC 2006a, p. 4.19).

In other words, forest strata or carbon pools that hold less than 5% of the total forest carbon stock are not key. It is good practice to focus the available resources for the inventory onto categories identified as key. Non-key categories may be estimated using lower tier methods.

A tier represents a level of methodological complexity. Usually three tiers are provided: Tier 1 is the basic method, tier 2 intermediate and tier 3 most demanding in terms of complexity and data requirements.

Tier 1 methods are designed to be the simplest to use, for which equations and default parameter values (e.g. emission factors) are provided by IPCC. Activity data, however, must always be country- specific. For tier 1 there are often globally available sources of activity data estimates (e.g. deforestation rates), though these data are usually spatially coarse.

Tier 2 can use the same methodological approach as tier 1, but applies emission factors that are based on country- or region-specific data. Higher temporal and spatial resolution and more disaggregated activity data are typically used in tier 2.

At tier 3, higher order methods are used, including models and inventory measurement systems tailored to address national circumstances, repeated over time, and driven by high-resolution activity data and disaggregated at sub-national level. These higher order methods provide estimates of greater certainty than lower tiers. Such systems may include comprehensive field sampling repeated at regular time intervals and/or GIS-based systems of land-use and management activity data, integrating several types of monitoring. Pieces of land where a land-use change occurs can usually be tracked over time, at least statistically. In most cases these systems have a climate dependency, and thus provide source estimates with interannual variability. Models should undergo quality checks, audits, and validations and be thoroughly documented.

2.6.5 Forest sampling-related terms The forest sampling related terms are defined hereafter, in line with the terminology found in FAO (2012), Johnson (2000), Kleinn et al. (2013), Mandallaz (2008), Schreuder et al. (2004), etc. • Areal sampling frame: The sampling frame or continuum (area) from which dimensionless sample points are selected based on a statistical sampling design. • Sampling design: The statistical framework or design that describes how sample points are selected (e.g. simple random sampling or systematic sampling). • Sample point: A point selected from the areal sampling frame to which a sampling unit is associated, where data are to be collected through observations and/or measurements. • Sampling unit: Basic observation unit in forest sampling associated to a sample point, designed according to a decision rule defining which population units are to be included in the sampling unit at each sample point. Common decision rules are fixed area plots, angle-count (plot-less) samples and transects. When more than one decision rule is involved, the sampling unit is also known as cluster. • Plot: A common decision rule used to configure a sampling unit, defining an area of some geometric shape where population units are to be observed and/or measured.

16

• Variable: A characteristic of the objects of interest that can take on different values and follows a distribution, e.g. the the elevation, slope and slope orientation of a sample point, the species, diameter and height of a tree, etc. • Parameter: A characteristic of the population, e.g. the total, mean (average), range of values or distribution of the density, basal area, volume, biomass, etc.

17

3. Inventory design The inventory design used is a refinement of the methodology of the forest carbon baseline study (described by Schade et al. 2013) carried out from mid-2011 to end-2012 in Leyte in the framework of the BMUB-supported Climate Relevant Modernization of Forest Policy and Piloting of Reducing Emissions from Deforestation and Forest Degradation (REDD) Project. It takes into account evolving international standards and good practices with regards to forest carbon stock assessment in compliance with the latest (2006) IPCC guidelines for national GHG inventories.

This chapter starts with an account of the sources of information that have been used (Chapter 3.1). The following sections detail the essential inventory design elements, such as the inventory method (see Chapter 3.2), the sampling unit design (see Chapter 3.3), the sample sizes planned and actually achieved (see Chapter 3.4), the sampling type and distribution (see Chapter 3.5), the human and material resources needed (see Chapter 3.6) and the organization of the field work (see Chapter 3.7). The last section (Chapter 3.8) details the estimation design used to calculate the variables of interest and the statistical parameters. 3.1 Sources of information Apart from the tree, stand or site characteristics counted, estimated or measured in the field, the FRAs made use of the following available information, whose sources are detailed hereafter: • Geographical-political subdivisions, see Chapter 3.1.1, • Forest strata and areas, see Chapter 3.1.2, • Soil classes for the estimation of the soil organic matter, see Chapter 3.1.3, • Wood specific gravity for the estimation of the above-ground biomass of trees, see Chapter 3.1.4.

3.1.1 Geographical-political subdivisions The names of the geographical-political subdivisions were taken from the Philippine Standard Geographic Code (PSGC) developed and regularly updated by the National Statistical Coordination Board (NSCB), which can be downloaded from http://nap.psa.gov.ph/activestats/psgc/. As of December 2016, the PSGC comprises 18 regions, 81 provinces, 145 cities, 1,489 municipalities and 42,036 barangays.

In the absence of publicly available authoritative boundaries of the geographical-political subdivisions, the boundaries downloadable from the GADM database of Global Administrative Areas (http://gadm.org/) have been used.

3.1.2 Forest strata and areas The forest areas and their stratification were taken from the 2010 NAMRIA land cover map (see Figure 3), released in 2013, that was elaborated through visual interpretation of medium- to high- resolution multi spectral satellite data (116 ALOS AVNIR 2, 40 SPOT 5 and 29 LANDSAT 7 gap-filled SLC off scenes covering the national territory, acquired mainly 2010), adopting a minimum mapping area of 0.5 ha in accordance with the 2005 DENR forest definition (see Chapter 2.6.1), distinguishing the following three forest strata: • Closed forests: tree crown cover > 40%; • Open forests: 10% < tree crown cover ≤ 40%; • Mangroves.

Tree plantations have not been mapped as a separate class, since the satellite data did not warrant their comprehensive and systematic identification. The documentation of the classification and its accuracy (confusion matrix) has not been published yet. A new wall-to-wall mapping of the 2015 land cover is under way.

18

Figure 3: 2010 NAMRIA Land Cover Map of the Philippines

19

3.1.3 Soil classes The World Reference Base (WRB) soil classes have been looked up from the 2013 FAO soil map of the Philippines prepared by the Bureau of Soils and Water Management (BSWM 2013), see Figure 4. Combined with the IPCC climate region (IPCC 2006b, p. 3.39), depending on the mean annual temperature (MAT), the elevation and the mean annual precipitation (MAP), the soil organic matter carbon stocks can be estimated using the IPCC tier 1 data (IPCC 2006b, p. 2.31) summarized in Table 2.

Table 2: IPCC Tier 1 Soil Organic Matter Stocks

Climate region Soil FAO soil class Soil organic matter [t C/ha] Cambisols 65 Kastanozems Luvisols Tropical, moist High activity clay Phaeozems Regosols

Acrisols 47 Low activity clay MAT > 18°C Nitosols Elevation ≤ 1,000 m Sandy soils Arenosols 39 MAP ≤ 2,000 mm Volcanic soils Andosols 70 Wetland soils Gleysols 86 Cambisols 44 Kastanozems High activity clay Luvisols Tropical, wet Phaeozems Regosols MAT > 18°C Acrisols 60 Low activity clay Elevation ≤ 1,000 m Nitosols MAP > 2,000 mm Sandy soils Arenosols 66 Volcanic soils Andosols 130 Wetland soils Gleysols 86 Cambisols 88 Kastanozems High activity clay Luvisols Phaeozems Tropical, montane Regosols

Acrisols 63 MAT > 18°C Low activity clay Nitosols Elevation > 1,000 m Sandy soils Arenosols 34 Volcanic soils Andosols 80 Wetland soils Gleysols 86

Source: IPCC (2006b).

Fluvisols are not mentioned. Depending on their texture, they may be treated like sandy soils or wetland soils.

20

Figure 4: 2013 BSWM FAO Soil Map of the Philippines

21

3.1.4 Wood specific gravity Wood specific gravity (p, expressed in grams per cubic centimeter [g/cm³] or in tonnes per cubic meter [t/m³]) is one of the dependent variables needed when using certain allometric equations for the estimation of biomass, such as the equation developed by Chave et al. (see Chapter 3.8.2.2). The values have been looked up (and averaged whenever several gravities are available) by species or species group growing in South-East Asia from the following sources: • Global wood density database prepared by Zanne et al. (2009); • Publication on wood densities of tropical tree species by Reyes et al. (1992).

For species not found in any of the above cited sources, the average wood specific gravity for tropical tree species in Asia of 0.57 g/cm³ published by Brown (1997) has been used. 3.2 Inventory Method The sub-national FRAs adopted probabilistic (statistical) sampling. The sample consists of a certain number of sampling units, where tree, stand or site characteristics are counted, assessed or measured in circular plots spatially arranged around the sample points. If the sample point associated to the sampling unit falls into the areal sampling frame (see Chapter 2.2), the sampling unit is to be measured. 3.3 Sampling unit design Textbox 8 provides basic considerations for the optimal design of sampling units.

Each sampling unit consists of a cluster centered on the sample point, composed of the following circular plots (see Figure 5): • Four nested plots with their centers at 40 m horizontal distance from the sample point in the four cardinal directions (north, east, south and west), each consisting of two concentric circular sub-plots: o 5 m radius sub-plot (corresponding to an area of 0.0079 ha) for: - the sampling of small-sized trees with 5 cm ≤ DBH or DAB < 20 cm for the estimation of their contribution to the AGB and BGB; - the sampling of standing dead wood with DBH or DAB ≥ 5 cm for the estimation of their contribution to the DOM; - the sampling of lying dead wood down to a diameter of 5 cm for the estimation of their contribution to the DOM; - the sampling of litter for the estimation of its contribution to the DOM. o 10 m radius sub-plot (corresponding to an area of 0.0314 ha) for: - the sampling of big-sized trees with DBH or DAB ≥ 20 cm for the estimation of their contribution to the AGB and BGB; • One plot with 25 m radius centered on the sample point, for the ocular assessment of the land cover, to serve as remote sensing training and validation data.

The entire cluster is inscribed in an area of 100 m x 100 m (1 ha). One inventory team should be capable of locating a sample point and performing all related observations and measurements in one day.

Circular plots are used because among all geometric shapes, they feature the shortest perimeter for a given plot size, hence reducing the number of borderline trees. Moreover, in stands without excessive undergrowth, the plot boundaries respectively borderline trees can be conveniently checked with the aid of a rangefinder.

The sizes of the sub-plots for the sampling of small-sized and big-sized trees represent a compromise, striving to achieve a reasonable balance between the "unproductive" time invested in

22 accessing the sample points and the "productive" time invested for the measurement and/or observation of the plots.

Clusters are used to increase the representativity of the sampling units. Indeed, compared to a solid (non-clustered) sampling unit of the same area, a cluster covers a wider area. Statistically, one cluster constitutes one observational unit. For the computation of the results per ha, the following blow-up factors are applicable: • Variables counted, estimated or measured in the 10 m radius sub-plots: 10,000 / (4 × 휋 × 102) = 7.9577; • Variables counted, estimated or measured in the 5 m radius sub-plots: 10,000 / (4 × 휋 × 52) = 31.8310.

The sample points and nested plot centers were marked permanently to be prepared for their re- measurement.

N 40 m

sample point W E 40 m

nested plot composed of 2 sub-plots

S

Figure 5: Sampling unit design

23

Textbox 8: Optimal Sampling Unit Design

Two sampling principles are commonly used in forest inventories: • Selection of sample trees with a probability proportional to their frequency, using fixed area plots; • Selection of sample trees with a probability proportional to their size, more precisely to their basal area, using angle-count (plot-less) samples.

Fixed area plot sampling is particularly efficient to estimate the variables related to the frequency of the trees: density. Angle-count (plot-less) sampling is particularly efficient to estimate the variables related to the size of the trees: basal area, volume, biomass.

Though efficient, angle-count (plot-less) sampling needs well trained and enlightened field team members, because the impact of the erroneous inclusion or (more frequently) exclusion of a sample tree has considerable impact on the results.

Nowadays, an approximation of the efficiency of angle-count (plot-less) sampling through nested fixed area plot sampling (typically using 2 to 3 concentric circular or nested sub-plots) has become popular, particularly for sampling in uneven-aged stands, to achieve a balanced sampling of trees in all diameter classes, through assigning a higher probability of selection to the larger trees of which there are usually less in a stand.

The size of the sampling unit represents a compromise. Theoretically, for a given sampling intensity (proportion in percent of the areal sampling frame sampled), it is statistically more efficient to measure many small sampling units rather than few large ones. However, with many small sampling units, the ratio of the "unproductive" time invested in accessing the sample points to the "productive" time invested for the measurement and/or observation of the plots becomes unfavorable. Reasonable compromises are plots selecting on average 12 to 20 trees in uneven-aged forests, 6 to 12 trees in even-aged forests.

3.3.1 Observations and measurements at the sample points The following variables were assessed or measured at the sample points: • Administrative location: province, city or municipality and barangay. • Actual coordinates. • Elevation. • Slope. • Slope orientation. • Terrain: 11 classes (plateau; summit or crest; upper slope; middle slope; lower slope; bench or terrace; valley; plain; narrow depression; water course; dunes). • Land classification: legal status (forest land or alienable and disposable).

The following variables were assessed within a radius of 25 m horizontal distance around the sample points: • Land cover: 12 classes (forest; marshland or swamp; fallow; shrubs; wooded grassland; grassland; annual crop; perennial crop; open or barren land; built-up area; fishpond; inland water). • Forest type: 10 types (dipterocarp old growth forest; dipterocarp residual forest; mossy forest; submarginal forest; closed pine forest; open pine forest; mangrove old growth forest; mangrove reproduction forest; native tree plantation forest; other plantation forest). • Tree crown cover: 3 classes (tree crown cover ≤ 10%; 10% < tree crown cover ≤ 40%); tree crown cover > 40%).

3.3.2 Observations and measurements in the nested plots The following variables were assessed or measured at the four nested plot centers (similar to the observations or measurements at the sample points): • Administrative location: province, city or municipality and barangay. 24

• Actual coordinates. • Elevation. • Slope. • Slope orientation. • Terrain: 11 classes (plateau; summit or crest; upper slope; middle slope; lower slope; bench or terrace; valley; plain; narrow depression; water course; dunes). • Land classification: Legal status (forest land or alienable and disposable).

The following variables were assessed or measured in the 5 m sub-plots of the four nested plots: • Plant diversity. • Ground coverage classes for six (6) vegetation layers according to height (< 50 cm; 50 cm ≤ height < 130 cm; 130 cm ≤ height < 200 cm; 2 m ≤ height < 4 m; 4 m ≤ height < 10 m; height > 10 m): 4 classes (none; coverage ≤ 10%; 10% < coverage ≤ 50%; coverage > 50%). • For each of the sampled small-sized trees with 5 cm ≤ DBH or DAB < 20 cm: species, azimuth and horizontal distance (from the plot center), and DBH or DAB. • For each of the sampled standing dead wood (including stumps) with DBH or DAB ≥ 5 cm: species, azimuth and horizontal distance (from the plot center), DBH or DAB and merchantable height. • For each of the sampled lying dead wood sections (those portions that are within the 5 m horizontal distance radius circle) down to a diameter of 5 cm: mid-diameter and length. • Litter: ground coverage percentage plus average depth.

The following variables were assessed or measured in the 10 m sub-plots of the four nested plots: • Land cover: 12 classes (forest; marshland or swamp; fallow; shrubs; wooded grassland; grassland; annual crop; perennial crop; open or barren land; built-up area; fishpond; inland water). • Forest type: 10 types (dipterocarp old growth forest; dipterocarp residual forest; mossy forest; submarginal forest; closed pine forest; open pine forest; mangrove old growth forest; mangrove reproduction forest; native tree plantation forest; other plantation forest). • Tree crown cover: three classes (tree crown cover ≤ 10%; 10% < tree crown cover ≤ 40%); tree crown cover > 40%). • For each of the sampled big-sized trees with DBH or DAB ≥ 20 cm: species, azimuth and horizontal distance (from the plot center), DBH or DAB and merchantable height.

Table 3 below summarizes the sub-plot sizes and the assessments or measurements made on trees and dead wood.

25

Table 3: Overview of sub-plot sizes and assessments or measurements made on trees and dead wood

Trees Dead wood Small-sized Big-sized Standing Lying 5 cm ≤ Dref* < 20 cm Dref* ≥ 20 cm Dref* ≥ 5 cm Dref* ≥ 5 cm Sub-plot radius 5 m 10 m 5 m 5 m Species species species species - Azimuth azimuth azimuth azimuth - Horizontal distance hor. distance hor. distance hor. distance - Diameter DBH or DAB DBH or DAB DBH or DAB mid-diameter Height or length - merch. height merch. height section length

* Dref of trees and standing dead wood refers to DBH or DAB, Dref for lying dead wood refers to the small end diameter

3.4 Sample size and margin of error It was expected that one inventory team could measure an average of 16 sampling units per month, based on the experience previously gained from the forest carbon baseline study in Leyte. At this rate, the available budget for the implementation of the three FRAs should have been sufficient to measure 200 sampling units per project site. These should permit to estimate the above-ground biomass per FRA with a margin of error not exceeding ± 10% at 90% confidence level, assuming coefficients of variation of AGB/ha of ± 80% in closed forests and of ± 120% in open forests.

In all three FRAs, however, these targets could not be achieved. Table 4 below summarizes the sample sizes (n) and margins of error (E%) actually achieved.

The factors that have contributed to the lower than expected outputs are the following: • Remoteness and very difficult accessibility of the sample points, due to the predominantly steep, heavily dissected and rocky terrain, often without trails, considerably slowing down the access to the sampling units. • Information of and coordination with local officials, tribal chieftains (in Davao Oriental), police and army (for security reasons) and community members prior to the hiring of local helpers or guides and the conduct of the inventory activities, preventing the teams to swiftly proceed to the inventory camps or sample points. • Unfavorable weather conditions in the forest area (frequent rains), hampering or stalling the measurement and data recording operations. • At some times and in some places critical security situation. • In Eastern Samar landslides and fallen trees caused by typhoon Hagupit (Ruby), which struck the area in December 2014, right before the start of the field work.

Fortunately, the coefficients of variation turned out to be lower than expected (see Chapter 6.6.1). Hence, the targeted precision could be achieved in Eastern Samar, and the margins of error in Davao Oriental (± 11.5%) and in the Panay Mountain Range (± 13.3%) are not too much off target.

26

Table 4: Sample sizes and margins of error of the sub-national FRAs

FRA Stratum Area n Above-ground biomass y̅ s% E%* [ha] [t d.m./ha] [%] [%] Davao Oriental Closed forests 21,750 37 398.00 ± 51.01 ± 14.16 (3 municipalities) Open forests 17,465 44 248.32 ± 84.31 ± 21.37 Total 39,215 81 331.34 ± 62.13 ± 11.49 Eastern Samar Closed forests 5,815 18 436.46 ± 39.77 ± 16.31 (2 municipalities) Open forests 36,264 102 342.37 ± 61.41 ± 10.09 Total 42,079 120 355.37 ± 57.85 ± 8.74 Panay Mountain RanCge Closed forests 47,882 33 239.40 ± 57.31 ± 16.90 (37 municipalities) Open forests 69,742 53 176.22 ± 89.85 ± 20.67 Total 117,624 86 201.94 ± 74.33 ± 13.30

* 90% confidence level

3.5 Sampling type and distribution The sub-national FRAs used a probabilistic (statistical) sampling, where the initially planned 200 sample points per site were drawn at random without replacement from the nodes of a quadratic grid with a side length of 1 km located within the areal sampling frame. In all sites, the number of nodes materializing potential sample points was about 2 times (in Davao Oriental and Eastern Samar) to 6 times (in the Panay Mountain Range) larger than the targeted 200.

Figure 6: Distribution of the Sampling Units Effectively Measured (Eastern Samar FRA)

27

Figure 6 illustrates the grid used and the location of the sampling units effectively measured in Eastern Samar.

Textbox 9 recalls the principal sampling types used in forest.

Textbox 9: Principal Sampling Types The principal sampling types used in forest inventory are the following: • Simple Random Sampling (SRS): Sample points are selected at random and independently within the areal sampling frame. • Aligned Systematic Sampling (ASS): Sample points are selected following a rigid scheme, typically at the intersections (nodes) of a regularly spaced grid, which is placed at random on the areal sampling frame. • Unaligned Systematic Sampling (USS): Sample points are selected at random within regularly shaped polygons placed at random on the areal sampling frame. SRS ASS USS

Source: Czaplewski et al. (2004)

Though statistically sound, SRS is seldom used, because of the irregular spatial distribution of the sample points.

ASS, where the nodes of a geometrical grid materialize the locations of the sample points, is widely used, particularly in regional and national forest inventories. Intuitively, ASS seems to ensure an evenly spread, balanced distribution of sample points that is presumably more representative of the population than a random sample of the same size, where clustering of some sample points could occur. Theoretically, it yields unbiased estimates of means and totals if the origin of the grid is chosen randomly. However, it is impossible to calculate a mathematically correct estimate of the variance, and therefore of the statistical precision. Most practitioners treat points lying on systematic grids as if they were random. To a large extent this is acceptable for point estimates, but less so for variance estimates, which are usually overestimated (theoretically the converse is also possible).

Maintaining a positive inclusion probability and the ability to calculate unbiased estimates of variances and statistical precision are concerns lately often voiced by scientists (e.g. Maniatis et al, 2010), notably in the context of IPCC conform forest carbon stock inventories. USS offers an alternative that avoids the above- mentioned drawback of ASS. Contrary to the aligned variant, the sample points are randomly drawn (one and only one) within each cell of a grid, ensuring a spatially balanced distribution, yet maintaining a positive inclusion probability of all sample points in all other grids cells. The sampling units are independent of each other.

Instead of hexagonal grid cells shown in the illustration above, quadratic grid cells, which are easier to materialize with a GIS software, may be used.

28

3.6 Human and material resources The three sub-national FRAs relied on two inventory teams each, composed of a team leader (B.Sc. Forestry), an assistant (B.Sc. Forestry) and two to four helpers, recruited locally, familiar with the area and if possible knowledgeable about tree species and forest products.

The team leaders were responsible for the security of the team, for the equipment entrusted to them, and for the work of their members. They directed the members, validated the data observed or measured by their assistants, and completed the field data forms.

The assistants handled the equipment and carried out the observations and measurements.

The helpers advised on getting to the sample points, carried the equipment, opened or brushed trails, access and sighting lines, marked the sample points and plot centers, helped the assistants in carrying out the measurements, and marked the trees.

To carry out the field work, each team was equipped with the following: • One handheld IPX7 waterproof GPS receiver (GARMIN GPSMap 78 series) with proven sensitivity / ability to operate under difficult signal reception conditions (under tree cover), to retrieve the sample points and measure coordinates. • One handheld IPX7 waterproof precision compass (SUUNTO KB 14/360) graduated in degrees for the measurement of bearings or azimuth. • One handheld IP54 laser hypsometer (LASER TECHNOLOGY Inc. [LTI] TruPulse Laser 200 rangefinder or NIKON Forestry Pro rangefinder) for the measurement of tree heights using the trigonometric principle, hence capable of measuring distances and inclination angles. Regrettably, the LTI TruPulse Laser 200 hypsometer is not waterproof, and the NIKON Forestry Pro cannot measure distances of less than 10 m. A better choice is the IP55 waterproof LTI TruPulse Laser 200X. • One handheld IPX7 waterproof precision clinometer (SUUNTO PM 5/360) as alternative to and backup for the laser hypsometer (a strategy that paid off when the LTI TruPulse laser hypsometers failed to work after heavy rain). • One fiberglass distance tape, 30 m, to measure distances. • One steel diameter tape, 5 m, to measure tree diameters. Upon request of the teams, the steel tapes were replaced with fiberglass tapes to lessen the risk of injuries from the sharp cutting edges of the steel tapes. • Per sampling unit five iron rods (of at least 1 cm diameter and 50 cm length) to permanently mark the sample points and the four plot centers, forced at least 4/5 of its length into the ground, topped each with a 50 cm bright-colored 1/2 " PVC pipe to facilitate the retrieval for quality control purposes (see Chapter 4.3). • One hatchet to force the iron rods used to permanently mark the sample points and the pour plot centers into the ground. • One first aid kit. • One backpack to carry the equipment. • Personal field work gear for the team leaders and assistants (boots, rain coats, head lamps, sleeping bags, etc.). • Camping equipment (tents, mobile stoves, etc.).

3.7 Organization of field work The inventory teams recorded the data collected in the field on specifically designed paper field forms (see Appendix 1). They were guided by a manual (Lennertz et al. 2014) explaining the use and care of the equipment, the configuration of the sampling units and the orderly sequencing of the field operations to locate and permanently mark the sample points and the four plot centers, and to perform the different assessments and measurements. These standard operation procedures (SOP) should help to maximize the homogeneity of the data acquisition and to minimize operating errors.

29

At the onset of the field work on each project site, the team leaders and assistants participated in a 1-week training, to become familiar with the instruments, the sampling procedure, the observations and measurements to be performed and the data recording. The training included a refresher course in taxonomy, dendrology and the identification of common tree, bamboo, palm, rattan and tree fern species.

The sample points were grouped into batches assigned to inventory camps (see Figure 7) on the basis of their location and accessibility. A reasonable compromise had to be found between (i) the number of sample points assigned to a specific camp (ideally not less than the number of sampling units that can be observed and measured in one field mission of one to two weeks) and (ii) the distance from the camp to the furthest sample point (in principle so that a team could locate the sample point and perform all observations and measurements in one day). For safety concerns the two teams generally operated from the same camp. Due to the remoteness and poor accessibility of most forest sites, the equipment and supplies for the entire duration of a field mission (typically one to two weeks) had to be hauled on foot to the camps.

A B

Figure 7: Inventory camp. A) Batch of sample points (red) assigned to one inventory camp in Eastern Samar; B) Inventory camp in the Panay Mountain Range

3.8 Estimation design The following sections provide information on the estimation design, i.o.w. how (i) variables of interest that cannot be observed or measured directly (such as merchantable volume, biomass, diversity indices, etc.), and (ii) statistical parameters (such as mean, variance, standard error, margin of error) are calculated.

In the absence of allometric equations specifically developed for the trees, bamboos, palms, rattan and tree ferns found in the tropical rainforests of the Philippines, the biomass is calculated using the equations found in the literature and databases (notably on the web platform GlobAllomeTree, see Textbox 10). Whenever several equations are available, preference is given to expressions that do not use height, since the latter is difficult to measure in tropical rainforests, hence constituting an important source of uncertainty.

30

Textbox 10: GlobAllomeTree

GlobAllomeTree (see http://www.globallometree.org/) is the first international web platform to share and provide access to tree allometric equations, created in 2013. It builds on the convening role and technical expertise of the Food and Agriculture Organization (FAO) of the United Nations (UN), of the French Centre de Coopération Internationale en Recherche Agronomique pour le Développement (CIRAD) and of the University of Tuscia in Italy. Through collaboration with renowned research centers, allometric equations were collected in more than 78 countries in all the continents and made available through this platform. Data is provided free of charge after agreeing to the licensing requirements.

Meantime, GlobAllomeTree has extended the data available to wood density data and biomass expansion factors. The platform also offers links to download manuals (notably a manual for building tree volume and biomass allometric equations, and tutorials on wood density and tree allometric data) and a free software tool (Fantallometrik) to (i) compare existing tree allometric equations for a given geographical location (i.e. specific site, ecological zone, country or continent), (ii) calculate volume, biomass, carbon stocks and descriptive statistics and compare data with default values from the scientific literature, and (iii) to add new tree allometric equations to the user's profile, or to submit them to GlobAllomeTree for review and potential insertion in the online database.

3.8.1 Merchantable volume of trees The merchantable volume (V, expressed in cubic meter [m³] inside bark) of trees and standing dead wood is calculated based on the diameter at breast height or above buttress (Dref) and the merchantable height (H) using the Philippine regional volume equations for Dipterocarps and Non- Dipterocarps (DENR 2014). 푉 = 0.00005203 × 퐷푟푒푓2 × 퐻 Dipterocarps, Cordillera Admin. Region & Regions 1, 2 & 3 (1) 푉 = 0.00005109 × 퐷푟푒푓2 × 퐻 Non-Dipterocarps, Cordillera Admin. Region & Regions 1, 2 & 3 (2) 푉 = 0.00005171 × 퐷푟푒푓2 × 퐻 Dipterocarps, Regions 4 & 5 except Palawan (3) 푉 = 0.00005204 × 퐷푟푒푓2 × 퐻 Non-Dipterocarps, Regions 4 & 5 except Palawan (4) 푉 = 0.00004649 × 퐷푟푒푓2 × 퐻 Dipterocarps, Regions 6 & 7 & Palawan except Bohol (5) 푉 = 0.00004874 × 퐷푟푒푓2 × 퐻 Non-Dipterocarps, Regions 6 & 7 & Palawan except Bohol (6) 푉 = 0.00005231 × 퐷푟푒푓2 × 퐻 Dipterocarps, Region 8 & Bohol (7) 푉 = 0.00005109 × 퐷푟푒푓2 × 퐻 Non-Dipterocarps, Region 8 & Bohol (8) 푉 = 0.00005087 × 퐷푟푒푓2 × 퐻 Dipterocarps, Eastern Mindanao (9) 푉 = 0.00004961 × 퐷푟푒푓2 × 퐻 Non-Dipterocarps, Eastern Mindanao (10) 푉 = 0.00005019 × 퐷푟푒푓2 × 퐻 Dipterocarps, Central Mindanao (11) 푉 = 0.00005039 × 퐷푟푒푓2 × 퐻 Non-Dipterocarps, Central Mindanao (12) 푉 = 0.00004668 × 퐷푟푒푓2 × 퐻 Dipterocarps, Western Mindanao (13) 푉 = 0.00004840 × 퐷푟푒푓2 × 퐻 Non-Dipterocarps, Western Mindanao (14)

3.8.2 Above-ground biomass of trees Two different allometric equations, developed by Brown (1997) and more recently by Chave et al. (2014), respectively, are used alternatively to estimate the above-ground biomass (AGB, expressed in kilogram of dry matter [kg d.m.]) of trees.

3.8.2.1 Allometric Equation by Brown (1997) Brown's allometric equation for the estimation of the AGB of trees in moist climatic zones was developed based on the destructive measurement of 170 trees, with 5 cm ≤ Dref ≤ 148 cm: 퐴퐺퐵 = exp (−2.134 + 2.530 × ln(퐷푟푒푓)) (R² = 0.97) (15)

31

with • AGB oven-dry above-ground biomass of trees, in kg d.m. • Dref diameter at breast height (1.30 m) or above buttress (30 cm), in cm

3.8.2.2 Allometric Equation by Chave et al. (2014) Chave's et al. allometric equation for the estimation of the AGB of tropical trees was developed based on the destructive measurement of 4,004 trees, with 5.0 cm ≤ Dref ≤ 180.0 cm: 퐴퐺퐵 = exp (−1.803 − 0.976 × 퐸 + 0.976 × ln(푝) + 2.673 × ln(퐷푟푒푓) − 0.0299 × (ln(퐷푟푒푓))2) (16) with • AGB oven-dry above-ground biomass of trees, in kg d.m. • p wood specific gravity, in g / cm³ (by species or species groups, see Chapter 3.1.4) • Dref diameter at breast height (1.30 m) or above buttress (30 cm), in cm • E environmental variable measuring stress, defined as: 퐸 = (0.178 × 푇푆 − 0.938 × 퐶푊퐷 − 6.61 × 푃푆) × 10−3 (17) with • TS temperature seasonality, the standard deviation of the monthly mean temperature over a year, expressed in degrees Celsius multiplied by 100 • CWD climatological water deficit in mm per year, computed by summing the difference between monthly rainfall and monthly evapotranspiration, only when this difference is negative • PS precipitation seasonality, the coefficient of variation in monthly rainfall values, expressed in percent of the mean value

A global grid layer of E at 2.5 arc-minute resolution is available at http://chave.ups-tlse.fr/pantropical_ allometry.htm#E and has been integrated into the FRA database system application used to store, manage and analyze the inventory data (see Chapter 5.3); the values of E are extrapolated from the gridded layer based on the geographic coordinates of the sample points (see Chapter 4.2.2).

3.8.3 Above-ground biomass of bamboos The AGB of bamboos is calculated using the allometric equation developed by Priyadarsini (1998, cited in Zemek 2009, p. 94) based on the destructive measurement of Dendrocalamus asper in Indonesia, with 3 cm ≤ Dref ≤ 7 cm: 퐴퐺퐵 = 0.1312 × 퐷푟푒푓2.2784 (R² = 0.95) (18) with • AGB oven-dry above-ground biomass of bamboos, in kg d.m. • Dref diameter at breast height (1.30 m), in cm

3.8.4 Above-ground biomass of palms The AGB of palms is calculated using the allometric equation developed by Goodman et al. (2013), based on the destructive measurement of 97 palms in Western Amazonia, with 6 cm ≤ Dref < 40 cm: 퐴퐺퐵 = exp(−3.3488 + 2.7483 × ln(퐷푟푒푓)) (R² = 0.80) (19) with • AGB oven-dry above-ground biomass of palms, in kg d.m. • Dref diameter at breast height (1.30 m), in cm

3.8.5 Above-ground biomass of rattan and tree ferns No allometric equation for the estimation of the AGB of rattan and tree ferns could be found.

32

3.8.6 Below-ground biomass of trees, bamboos and palms The below-ground biomass (BGB, expressed in kilogram of dry matter [kg d.m.]) of trees, bamboos, palms and tree ferns is calculated based on the AGB using the AGB to BGB ratio (R) of the 2006 IPCC guidelines (IPCC 2006, p. 4.49): 퐵퐺퐵 = 푅 × 퐴퐺퐵 (20) with • BGB Oven-dry below-ground biomass of trees, bamboos and palms, in kg d.m. • R BGB to AGB ratio: 0.37 • AGB Oven-dry above-ground biomass, in kg d.m.

3.8.7 Above-ground biomass of standing dead wood The AGB of standing dead wood (SDW, expressed in tonnes of dry matter [t d.m.]) is calculated based on (i) the merchantable volume (V) estimated using the Philippine regional volume equations for Dipterocarps and Non-Dipterocarps (see Chapter 3.8.1) and (ii) the biomass conversion and expansion factors (BCEFs) of the 2006 IPCC guidelines (IPCC 2006, p. 4.52), (iii) divided by 2 to account for decay (Thiele et al. 2010, p. 74): 푆퐷푊 = 푉 × 퐵퐶퐸퐹푠 / 2 (21) with • SDW biomass of standing dead wood, in t d.m. • V merchantable volume inside bark of standing dead wood, in m³ • BCEFs Biomass conversion and expansion factor of merchantable growing stock volume to AGB for humid tropical natural forests, in t / m³, depending on the growing stock level: o 9.00 t d.m. / m³ for V < 10 m³ / ha o 4.00 t d.m. / m³ for 10 m³ / ha < V ≤ 20 m³ / ha o 2.80 t d.m. / m³ for 20 m³ / ha < V ≤ 40 m³ / ha o 2.05 t d.m. / m³ for 40 m³ / ha < V ≤ 60 m³ / ha o 1.70 t d.m. / m³ for 60 m³ / ha < V ≤ 80 m³ / ha o 1.50 t d.m. / m³ for 80 m³ / ha < V ≤ 120 m³ / ha o 1.30 t d.m. / m³ for 120 m³ / ha < V ≤ 200 m³ / ha o 0.95 t d.m. / m³ for V > 200 m³ / ha

3.8.8 Biomass of lying (downed) dead wood The biomass of lying (downed) dead wood (LDW, expressed in tonnes of dry matter [t d.m.]) is calculated by section based on (i) the section volume (V) estimated using the mid-diameter (Dref) and length (L) of the section assimilated to a cylinder and (ii) the average wood density for Asia (Reyes et al. 1992, p. 2), (iii) divided by 2 to account for decay (Thiele et al. 2010, p. 74):

퐷푟푒푓2 퐿퐷푊 = π × × 퐿 × 푝/ 2 40,000 (22) with • LDW biomass of lying dead wood, in t d.m. • Dref mid-diameter of lying dead wood section, in cm • L length of lying dead wood section within the sub-plot, in m • p average wood density for Asia: 0.57 t d.m. / m³

3.8.9 Biomass of litter The biomass of litter (LI, expressed in tonnes of dry matter [t d.m.]) is calculated based on (i) the volume of litter estimated using the ground coverage percentage (C) and the average depth (DPT) of litter and (ii) the average density of litter (Chojnacky et al. 2009 in South. J. appl. For. 33(1) 2009, p. 32):

33

푉 = C × 퐷푃푇 × 10,000 × 퐷 (23) with • LI biomass of litter, in kg d.m. / ha • C ground coverage percentage of litter, in% • DPT average depth of litter, in m • D average density of litter: 40 kg d.m. / m³

3.8.10 Conversion of biomass to carbon The carbon (C) equivalent of the biomass is calculated based on the AGB and BGB using the following carbon fractions (CF, expressed in tonnes of carbon per tonne of dry matter [t C / t d.m.]) of the 2006 IPCC guidelines (IPCC 2006, p 4.48): 0.47 t C / t d.m. Carbon fraction of dry matter for living biomass (24)

0.37 t C / t d.m. Carbon fraction of dry matter for dead organic matter (25)

3.8.11 Diversity indices The species diversity indices are computed as follows: • Berger-Parker index: 푁 푑 = 푚푎푥 푁 (26) • Margalef index: (푆−1) 퐷 = 푀푔 ln (푁) (27) • Shannon H' index: 푛 푛 퐻′ = − ∑푆 (( 푖) × 푙푛 ( 푖)) 푖=1 푁 푁 (28) • Shannon E index: 퐻′ 퐸 = ln (푆) (29) • Simpson index: 푛 ×(푛 −1) 퐷 = ∑푆 푖 푖 푖=1 푁×(푁−1) (30) with o 푁 total number of individuals

o 푛푖 number of individuals of species i o 푁푚푎푥 number of individuals of the most abundant species o 푆 number of species (species richness) o 푙푛 natural logarithm

3.8.12 Statistical inference Assuming for simplicity's sake a random distribution of the sample points, the estimated stratum and total means, variances, standard errors and margins of error are computed using the following formulas (see Zöhrer 1980): • Stratum means: 푛 ∑ 푗 푖=1 푦푖푗 푦̅푗 = (31) 푛푗 • Stratum variances: 2 푛푗 푛푗 ∑ 푦2 − (∑ 푦 ) / 푛 2 푖=1 푖푗 푖=1 푖푗 푗 푠푗 = (32) 푛푗−1

34

• Stratum standard errors: 푠푗 푆푗 = (33) √푛푗 • Stratum margins of error: 푠푗 퐸푗 = × 푡푗 (34) √푛푗 • Total mean: 푛 푦̅ = ∑푀 푗 × 푦̅ 푗=1 푛 푗 (35) • Total variance: 푛 푠2 = ∑푀 푗 × 푠2 푗=1 푛 푗 (36) • Total standard error:

1 2 푆 = √ × (∑푀 푃 × 푠 ) (37) 푛 푗=1 푗 푗 • Total margin of error:

1 2 퐸 = √ × (∑푀 푃 × 퐸 ) (38) 푛 푗=1 푗 푗 with

o 푦푖푗 variable (such as number of trees per ha, basal area per ha, volume per ha, biomass per ha, etc.) of sampling unit i in stratum j

o 푦̅푗 arithmetic mean of variable 푦 in stratum j o 푦̅ total arithmetic mean of variable 푦 2 o 푠푗 variance of variable 푦 in stratum j o 푠2 total variance of variable 푦

o 푆푗 standard error of the mean of variable 푦 in stratum j o 푆 total standard error of the mean of variable 푦

o 퐸푗 margin of error of the mean of variable 푦 in stratum j o 퐸 total margin of error of variable 푦 o 푀 number of strata

o 푛푗 number of sampling units in stratum j

o 푡푗 two-tailed Student t-value with 푛푗 degrees of freedom in stratum j at a given confidence level C, commonly 90%, 95% or 99% o 푛 total number of sampling units

o 푃푗 weight of stratum j

The correct interpretation of a margin of error at a given confidence interval is recalled in Textbox 11.

35

Textbox 11: Margin of Error and Confidence Level The uncertainty of the estimated arithmetic mean of a population based on a sample is estimated through the margin of error at a specific confidence level. The confidence level corresponds to the percentages of the area of the normal density curve:

Source: Yale University (1997)

For a confidence level of 95%, it is correct to say that there is a 95% chance that the confidence interval 푦̅ ± 퐸 contains the true (unknown) population mean. It is not quite correct to say that there is a 95% chance that the population mean lies within the interval.

36

4. Field data collection This chapter starts with a brief account on how the sample points and nested plot centers were located in the field and marked on the ground (Chapter 4.1). The second and main section expands on how the variables were assessed or measured (Chapter 4.2). Chapter 4.3 describes the quality assurance and control measures adopted. Chapter 4.4 concludes with a summary of the time and cost of the data collection in the three project sites. 4.1 Getting to and marking of sample points

4.1.1 Getting to the sample points The sample points were located in the field on the basis of their geographic coordinates using handheld GPS stand-alone receivers. The coordinates were uploaded from a computer as points of interest (POIs) rather than as waypoints, using the "GARMIN POI loader" software (freeware downloadable from http://www8.garmin.com/support/ mappingsw.jsp). POIs offer the advantage that unlike waypoints, they cannot be edited nor erased from the GPS receivers (unless connected to a computer and with the use of the aforementioned software).

Good sources of information to study the approach of sample points are the following: • Topographic maps in Open Cycle Map (http://www.opencyclemap.org) showing the "Outdoors" base layer, which is particularly useful for the appreciation of the relief (see Figure 8); • Satellite images in Google Maps (http://www.google.com/maps), Bing Maps (http://www.bing.com/maps) and Apple Maps (only available on Apple Mac OS and iPhone or iPad iOS operating systems), particularly where high resolution satellite data are available (see Figure 9). It is worthwhile to compare the different sources for best results, since the images are regularly updated.

As much as possible, the approach of a targeted sample point was studied together with local helpers or guides, who are well versed with the terrain, existing trails, unsurmountable barriers and/or obstacles such as steep hills or waterlogged areas to be avoided.

Figure 8: Open cycle map with "Outdoors" base layer Cangaranan River, Valderrama, Antique

37

Figure 9: Google Maps v. Apple Maps Barangays Pinanag-an (Borongan City) and Patag (Maydolong) on the Suribao River, Eastern Samar

4.1.2 Location of sample points and plot centers Considering the limited positional precision of stand-alone GPS measurements (in practice ± 10 m, as evidenced by the virtual movement of an immobilized GPS receiver, which is a remarkable precision to come close to any point on the globe from whatever origin over considerable distances, but insufficient to measure distances of less than 100 m to 200 m, since the relative precision deteriorates to 10% - 5%), the location of sample points was determined covering the last 10 m to 15 m by compass and horizontal distance measurement (referring to the azimuth / bearing and distance to the sample point displayed by the GPS receiver once the distance to the destination was less than 15 m) using a distance tape or a ranging laser, in order to prevent bias (preference for easily accessible areas) when closing in on the sample point. The same applied to the location of the four nested plot centers, situated at 40 m in the four cardinal directions (north = 0°; east = 90°, south = 180°; west = 270°) from the sample point.

The azimuth was measured with the help of a handheld precision compass.

4.1.3 Permanent marking of sample points and plot centers The sample points and the four nested plot centers of each sampling unit were permanently marked with an iron rod (of at least 1 cm diameter and 50 cm length), forced at least 4/5 of its length into the ground, topped with a 50 cm bright-colored 1/2 " PVC pipe to facilitate the retrieval for quality control purposes (see Chapter 4.3 and Figure 10).

A B

Figure 10: Location and marking of sample points and plot centers. A) On the way to a sample point in Malinao, Aklan; B) Permanent marking of a sample point using an iron rod topped with a PVC pipe

38

4.1.4 Inaccessible sample points and plot centers In the rare event that one of the four nested plot centers of a sampling unit turned out to be inaccessible because of unsurmountable obstacles, it was re-located at 80 m horizontal distance from the sample point in the next cardinal direction, turning clockwise (see Figure 11: if the western plot center is inaccessible, it may be re-located at 80 m horizontal distance to the west + 90 ° = north from the sample point).

In the equally rare event that a sample point turned out to be inaccessible, the sampling unit was abandoned. A replacement sample point was drawn at random from those nodes of the quadratic grid with a side length of 1 km (see Chapter 3.5) located (i) in the same forest stratum and (ii) at a similar elevation as the inaccessible sample point.

If the sample point or one of the nested plot centers of the sampling unit fell on an area whose land cover assessed in the field was other than forest, it was not re-located, but observed and measured as is.

re-located W 80 m

N 40 m

inaccessible

sample point W E 40 m

nested plot composed of 2 sub-plots

S

Figure 11: Re-location of inaccessible plots

39

4.2 Assessment or measurement of variables The variables to be assessed or measured were recorded on specifically designed paper field forms (see Appendix 1), following the standards and specifications describe in the following sections.

4.2.1 Administrative location The administrative location, comprising at least the region, province and city or municipality, and as much as possible the barangay, was assessed at and recorded for the sample points and the four plot centers. Hence, five such observations were recorded per sampling unit (in some cases, a sampling unit may be crossed by an administrative boundary).

4.2.2 Actual coordinates The actual Universal Transverse Mercator (UTM) coordinates, comprising the zone (in the Philippines 50 in Palawan, 52 in the eastern-most portions of Mindanao, 51 elsewhere), the northing in m and the easting in m, were measured at and recorded for the sample points and the four nested plot centers. Hence, five coordinate measurements were performed per sampling unit. The coordinates were read from the GPS stand-alone receiver, immobilized at the sample point or plot center, using averaging (see Textbox 12).

Textbox 12: Averaging Coordinate Measurements with GARMIN GPS Receivers With GARMIN GPSmap 62, 64, 76 and 78 receivers, press the MENU button twice to open the < Main Menu >, select < Waypoint Avg. >, followed by < Create Waypoint >, and wait until the sample confidence has reached 100%. This may take up to 5 minutes, under poor satellite signal reception conditions more. There is no harm letting the GPS receiver continue the averaging after the sampling confidence has reached 100%. Hence, the GPS receiver can be left at the measurement point, while performing other observations and/or measurements.

4.2.3 Elevation The elevation in m above sea level was measured at and recorded for the sample points and the four nested plot centers. Hence, five elevation measurements were performed per sampling unit. The elevation was read from the GPS stand-alone receiver.

4.2.4 Slope The slope was measured at and recorded for the sample points and the four nested plot centers. Hence, five slope measurements were performed per sampling unit. The slope corresponds to the average inclination in % measured with a handheld precision clinometer (see Textbox 13) in two opposite directions along 10 m segments (oblique distance) of an imaginary straight line passing through the sample point or plot center, respectively, and following the steepest slope gradient (where water would run off).

4.2.5 Slope orientation The slope orientation was measured at and recorded for the sample points and the four nested plot centers. Hence, five slope orientation measurements were performed per sampling unit. The slope orientation corresponds to the azimuth in ° of the downhill direction of the imaginary straight line used for the measurement of the slope gradient, read from a handheld precision compass.

40

Textbox 13: SUUNTO PM-5/360 Clinometer and Dendrometer

To adjust focus, close one eye and look through the optics. Turn the cranted black optics knob until numbers are clear.

To measure and read vertical angle, hold the clinometer vertically, so that the scale can move freely. Keep both eyes open, and aim through the optics to the target. An optical illusion makes the sighting line and the scale appear over the target. Read the value from the sighting line once the scale is stable.

4.2.6 Terrain The terrain (topography) class was assessed at and recorded for the sample points and the four nested plot centers. Hence, five terrain classes assessments were performed per sampling unit. The assessment through ocular inspection distinguished the 11 classes defined by FAO (2012): • Plateau: Relatively flat (slope ≤ 5%); terrain of great extent and high elevation, above adjacent lowlands limited by an abrupt descent scarp on at least one side; may be dissected by deep valleys and deeply incised rivers. • Summit, crest: Crest of any kind or hilltop; can be sharp or rounded. • Upper slope: Upper slope of hillside (located on the upper 1/3 of the slope) (shoulder). • Middle slope: Middle slope of hillside (slope > 5%) (back slope). • Lower slope: Lower slope of hillside (foot slope). • Bench, terrace: Horizontal zone of average width over 30 m interposed in the valley side (slope < 15%) or a terrace over 6 m width.

41

• Valley: Very wide, gently sloping depression with predominant extent in one direction commonly situated between two mountains or ranges of hills; the profile may be U- or V- shaped; includes river valley (formed by flowing water) or glacier valleys. • Plain: A large flat to very gently undulating area at a low elevation with reference to surroundings • Narrow depression: Enclosed depression or small, narrow valley or distinct crater (including ravine, gorges, gullies, canyons, etc.). • Water course: Permanent or temporary water course (river, etc.). • Dunes: Sandy hills developed through sand deposits from wind erosion or storms, often unstable and moving.

4.2.7 Land classification The land classification (legal status) was assessed at and recorded for the sample points and the four nested plot centers. Hence, five land classification assessments were performed per sampling unit. The assessment through consultation of the latest available land classification map from DENR distinguished 2 classes: • Forest land. • Alienable and disposable.

4.2.8 Land cover The land cover was assessed in and recorded for the 25 m radius plot and the four 10 m radius sub- plots. Hence, five land cover assessments were performed per sampling unit. The assessment through ocular inspection distinguished forests (further classified according to their type, see Chapter 4.2.9 below) and the 11 non-forest land cover classes used in the 2010 NAMRIA national forest cover map: • Forest: Land with an area of more than 0.5 ha and trees able to reach a minimum height of 5 m in situ with a crown cover of more than 10% (see Chapter 2.6.1 for the detailed definition). • Marshland or swamp. • Fallow. • Shrubs. • Wooded grassland. • Grassland. • Annual crop. • Perennial crop. • Open or barren land. • Built-up area. • Fishpond. • Inland water.

4.2.9 Forest type The forest type was assessed in and recorded for the 25 m radiusplot and the four 10 m radius sub- plots. Hence, five forest type assessments were performed per sampling unit. The assessment through ocular inspection distinguished the 8 natural forest types used in the conduct of the second National Forest Resources Inventory (1979 - 1988), plus 2 additional types for planted (man-made) forests: • Dipterocarp old growth forest: Tropical rain forest dominated by Dipterocarpaceae with traces of commercial logging. • Dipterocarp residual forest: Tropical rainforest dominated by Dipterocarpaceae after commercial logging.

42

• Mossy forest: Tropical rainforests of the high elevations dominated by Podocarpaceae, Myrtaceae and Fagaceae with trees of medium height and short boled, covered with epiphytes. • Submarginal forest: Tropical rainforest dominated by Leguminosae and less utilized species, mainly restricted to shallow and excessively drained lime stone soils. • Closed pine forest: Pure stands of Benguet (Pinus kesiya var. langbianensis) or Mindoro pine (P. merkusii) with crown cover > 30%. • Open pine forest: Pure stands of Benguet or Minodoro pine with 10% < crown cover ≤ 30%. • Mangrove old growth forest: Tidal forests dominated by Rhizophoraceae, Avicenniaceae or Sonneratiaceae, without traces of exploitation. • Mangrove reproduction forest: Tidal forests dominated by Rhizophoraceae Avicenniaceae or Sonneratiaceae, where utilization had been intensive and big trees had been removed. • Native tree plantation forest: Planted forest dominated by native rainforest species. • Other plantation forest: Planted forest dominated by non-native, often fast growing tree species.

4.2.10 Tree crown cover The tree crown cover was assessed in and recorded for the 25 m radius plot and the four 10 m radius sub-plots. Hence, five tree crown cover assessments were performed per sampling unit. The assessment through ocular inspection distinguished the 3 classes currently used by NAMRIA for forest cover mapping: • Non-forest: tree crown cover ≤ 10%). • Open forest: 10% < tree crown cover ≤ 40%. • Closed forest: tree crown cover > 40%.

4.2.11 Plant diversity The plant diversity was counted in and recorded for the four 5 m radius sub-plots. Hence, four plant diversity counts were performed per sampling unit. The inventory consisted of the counting of distinct higher plant species observed, even if not known by their local, official common or scientific names. To avoid repeated counting of the same species, the count was done by only one person, systematically collecting specimen of leaves from plants that can be reached from the ground.

4.2.12 Ground coverage classes by vegetation layers Ground coverage classes for six vegetation layers were assessed in and recorded four the four 5 m radius sub-plots. Hence, four times six ground coverage classes assessments were performed per sampling unit. For each of the following six vegetation layers: • Grass, herbs and mosses. • Tree regeneration, shrubs and plants with 50 cm ≤ height < 130 cm. • Tree regeneration, bushes and plants with 130 cm ≤ height < 200 cm. • Undergrowth of any kind with 2 m ≤ height < 4 m. • Lower trees and other plants with 4 m ≤ height < 10 m. • High trees with height > 10 m.

The following four ground coverage classes were assessed through ocular inspection: • None. • Coverage ≤ 10%. • 10% < coverage ≤ 50%. • Coverage > 50%.

43

4.2.13 Ground coverage and average depth of litter Litter, defined as all non-living biomass with a size > 2 mm and < 5 cm (i.e. the minimum diameter for dead wood), lying dead, in various states of decomposition above or within the mineral or organic soil (see also Chapter 2.6.3), was inventoried in and recorded for the four 5 m radius sub-plots through ocular estimates of • Ground coverage in%, and • Average depth in cm.

4.2.14 Mid-diameter and length of lying dead wood sections Lying dead wood, defined as all non-living woody biomass lying on the ground with a diameter ≥ 5 cm (i.e. the inventory threshold for dead wood and trees) not contained in the litter, was inventoried in and recorded for the four 5 m radius sub-plots. For each lying dead wood section within a sub-plot (without considering those portions extending beyond the sub-plot, see Figure 12), the following measurements were performed: • Mid-diameter: mid-diameter outside bark in cm, rounded to 0.1 cm, of the dead wood section within the 5 m radius sub-plot, without considering those portions (i) extending beyond the sub- plot, or (ii) with a diameter < 5 cm. The mid-diameter was measured using a caliper or a diameter tape. • Length: length in m, rounded to 0.1 m, of the dead wood section within the 5 m radius sub-plot, without considering those portions (i) extending beyond the sub-plot, or (ii) with a diameter < 5 cm. The length was measured using a distance tape.

If a lying dead wood section featured branches, these were measured separately.

m 5

# 03 # 02 Mid-Diameter Mid-Diameter Length Length sub-plot center

# 01 Mid-Diameter Length

Figure 12: Measurements on lying dead wood sections

44

4.2.15 Observations and measurements on trees and standing dead wood Trees and standing dead wood with DBH or DAB ≥ 5 cm were inventoried in and recorded for the four nested plots • in the 5 m radius sub-plot sampling o small-sized trees (all species) with 5 cm ≤ DBH or DAB < 20 cm; o standing dead wood with DBH or DAB ≥ 5 cm; • in the 10 m radius sub-plot sampling big-sized trees (all species) with DBH or DAB ≥ 20 cm.

For each of the sampled trees and standing dead wood, (i) the species, (ii) azimuth and (iii) horizontal distance from the plot center, (iv) DBH or DAB and (v) for standing dead wood with DBH or DAB ≥ 5 cm as well trees with DBH or DAB ≥ 20 cm the merchantable height was assessed or measured and recorded as described hereafter.

In total, the following number of trees and dead wood have been sampled in the course of the FRAs: • 3,286 trees and 41 standing dead wood in Davao Oriental, • 8,031 trees and 207 standing dead wood in Eastern Samar, • 4,794 trees and 195 standing dead wood in the Panay Mountain Range.

4.2.15.1 Species The species of each sampled tree and, as much as possible, of each standing dead wood was recorded as identified by the team mates or the local helpers, referring to the official common name or the scientific name. Local names are not suited to unequivocally identify a species, because they vary from dialect to dialect, and even from place to place. In cases where a tree could only be identified through its local name, the latter was recorded, as much as possible together with other information (such as digital pictures or specimens) that could facilitate the later identification of the species by its scientific name with the help of a dendrology expert from the local academe.

4.2.15.2 Azimuth The azimuth in ° of each sampled tree and standing dead wood was recorded as measured from the plot center using a handheld precision compass.

4.2.15.3 Horizontal distance The horizontal distance in m, rounded to 0.1 m, of each sampled tree and standing dead wood was recorded as measured from the plot center using a distance tape or a laser rangefinder (see Textbox 14 for the configuration and tilt sensor adjustment of the used LTI TruePulse Laser 200 Rangefinder, and Textbox 15 for the measurement of horizontal distances with this device).

Together with the azimuth, the horizontal distance provides the polar coordinates of the sampled trees and standing dead wood, which will be needed at the time of re-measurements to identify them one by one.

45

Textbox 14: LTI TruePulse Laser 200 Rangefinder

Press the FIRE button on top of the unit to turn power on.

Press and hold simultaneously the ▲ or ▼ buttons on the side of the unit for 4 seconds to turn power off.

The eyepiece can be adjusted by turning it.

The diopter of the viewfinder can be adjusted by turning the cranted ring at the basis of the eyepiece.

To change units, press and hold the ▼ button for 4 seconds until < UnitS > is displayed in the viewfinder. Press the FIRE button to confirm the < UnitS > option, then press the ▲ or ▼ buttons to select the unit (feet, meters or yards), and press the FIRE button to select the unit and return to the measurement mode.

The tilt sensor is aligned during assembly. Should the unit suffer a severe shock, the tilt sensor may have to be re-aligned as follows: (1) Press and hold the ▼ button for 4 seconds until < UnitS > is displayed. (2) Press the ▼ button to display the < inc > option. (3) Press the FIRE button to confirm the < inc > option, then press the ▲ or ▼ buttons to select the < CAL_Y > option, and press the FIRE button to start the calibration of the tilt sensor. < CAL_1 > is displayed. (4) Position the unit on a flat level surface. Use one finger to hold the front of the unit flat on the surface, and keep the finger in place until step (7) is completed. (5) press the FIRE button; < CAL_2 > is displayed. (6) Rotate the unit 180 °. (7) Press the FIRE button; < donE > is displayed. (8) Press the FIRE button to return to the measurement mode.

46

Textbox 15: Horizontal Distance Measurements with the LTI TruePulse Laser 200

To measure horizontal distance, select the < HD > mode by pressing the ▲ or ▼ button until < HD > is displayed at the bottom of the viewfinder.

Look through the eyepiece and use the crosshair to aim to the target along a clear line of sight, then press and hold the FIRE button.

Release the FIRE button once the horizontal distance is displayed.

4.2.15.4 Diameter at breast height or above buttress The diameter at breast height or the diameter above buttress outside bark in cm, rounded to 0.1 cm, of each sampled tree and standing dead wood was recorded as measured using a diameter tape at the following measurement points (see also Figure 13): • in general, at breast height, i.e. 1.3 m above ground (DBH) as measured from the uphill side of the stem; • for trees with prominent buttresses or basal flanges at breast height, the diameter is measured 30 cm above the end of the buttresses or flanges (DAB); • for trees with bulges, swellings, depressions, branches or other abnormalities at breast height, the diameter is measured just below and above the abnormality at a point where it ceases to affect normal stem form, and computed as the average of the two measurements; • for stumps with a total height < 1.3 m at the section.

If a tree or standing dead wood forks immediately above breast height, the diameter was measured below the swell resulting from the fork. If a tree or standing dead wood forks below breast height, the stems were considered as separate trees or standing dead woods, respectively. For leaning trees or standing dead woods, the breast height was determined along the axis of the stem.

47

Whenever it proved impossible to measure the DBH or DAB with a diameter tape as described above (e.g. when the measurement point is inaccessible), it was approximated by comparison with a metric tape held horizontally at the base of the tree (see Figure 14).

Figure 13: DBH and DAB measurements (Zöhrer 1980)

48

Figure 14: Diameter estimates for inaccessible measurement points

4.2.15.5 Merchantable height The merchantable height in m, rounded to 0.1 m, of each sampled tree with DBH or DAB ≥ 20 cm and of each sampled standing dead wood with DBH or DAB ≥ 5 cm including stumps was recorded as measured using either a laser hypsometer (see Textbox 16) or a handheld precision clinometer (see Textbox 17).

Merchantable height of trees with DBH or DAB ≥ 35 cm is defined as the linear distance along the axis of the stem from the stump height to the top merchantability limit which is restricted by forks, large limbs, sweep, crook or decay, which make segments of the stem un-merchantable for saw logs. For trees with 15 cm ≤ DBH or DAB < 35 cm, the volume section is limited by a minimum top diameter inside bark which is fixed at 60% of DBH or DAB. By this definition, the measurement to the base of the tree has to be a measurement to the place where the felling cut would be applied, usually about 50 cm above ground, or above the buttresses. Limits for merchantability are the following: • Size of limbs and knots: The sum of diameters in any ¼ m segment ½ the diameter of the log at that point. Where limb and knot diameters exceed this limit, the merchantable height cannot extend through that point, unless there is a merchantable section of 3 m or more in length above that point. • Sweep: Sweep is a curvature in a tree section. Sweep is measured in centimeters of departure of the center line of the section from a straight line joining the centers of each end of the section. The departure is measured at the midpoint of the section containing the sweep. A simple rule for maximum sweep is that departure minus allowance for long taper cannot exceed ½ the small end diameter of the section. Merchantable length is terminated below a section with excessive sweep unless there is a merchantable section of 3 m or more in length above that section. • Crook: Crook is a more or less abrupt bending or angle in a tree section. Crook is measured in cm of maximum departure of the section center line from an extension of the center line of the straight portion of the log. The maximum departure cannot exceed ½ the small end diameter of the log. Excessive crook should terminate the merchantable length unless there is a merchantable section of 3 m or more in length above that section.

Figure 15 shows some impressions of the various measurements and data recording in the field.

49

Textbox 16: Height Measurements with the LTI TruePulse Laser 200

To measure height, select the < HT > mode by pressing the ▲ or ▼ button until < HT > is displayed at the bottom of the viewfinder.

The < HD > indicator flashes, prompting to measure the horizontal distance to the tree. Look through the eyepiece and use the crosshair to aim to the tree along a clear line of sight, then press and hold the FIRE button. Release the FIRE button once the measured horizontal distance appears briefly.

The < Ang_1 > and the < Inc > indicators flash, prompting to measure the inclination to the upper point (merchantable or total height of the tree). Look through the eyepiece and use the crosshair to aim to the upper point, then press and hold the FIRE button. The measured inclination appears and is updated as long as the FIRE button is held. The measured inclination is locked once the FIRE button is released.

The < Ang_2 > and the < Inc > indicators flash, prompting to measure the inclination to lower point (base of the tree). Look through the eyepiece and use the crosshair to aim to the lower point, then press and hold the FIRE button. The measured inclination appears and is updated as long as the FIRE button is held. The measured inclination is locked once the FIRE button is released.

The measured height is displayed after the three measurements (horizontal distance, upper point, lower point).

For best results, the horizontal distance should be approximately equal to the height to be measured.

50

Textbox 17: Height Measurements with the SUUNTO PM-5/360

Height (H) can be estimated from any known horizontal distance through two readings of the% scale, aiming to the upper point (merchantable or total height of the tree) respectively to the lower point (base of the tree). For best results, the horizontal distance used for the sightings should be approximately equal to the height to be measured. Taking into consideration the sign of the lower reading (+ for readings > 0%, - for readings < 0%), the height is computed using the following formula: (푈 − 퐿) × 퐷 퐻 = 100 with • H height, in m • U reading, aiming to the upper point, in% • L reading, aiming to the lower point, in% • D horizontal distance

Example: For readings aiming to the upper point of + 62% respectively to the lower point of - 23% from a horizontal distance of 18.0 m, the formula yields: ((+62) − (−23)) × 18.0 퐻 = = 15.3 m 100

For horizontal distances of 15.0 m and 20.0 m, scales graduated in m are provided to do direct readings using the following formula:

퐻 = (푈 − 퐿) with • H height, in m • U reading, aiming to the upper point, in m • L reading, aiming to the lower point, in m

Example: For readings aiming to the upper point of + 7 m respectively to the lower point of - 5 m from a horizontal distance of 20.0 m as illustrated to the right, the formula yields:

퐻 = ((+7) − (−5)) = 12.0 m

Source: FAO (2012)

Example: For readings aiming to the upper point of + 15.5 m respectively to the lower point of + 3.5 m from a horizontal distance of 20.0 m as illustrated to the right, the formula yields:

퐻 = ((+15.5) − (+3.5)) = 12.0 m

Source: FAO (2012)

51

Figure 15: Measuring and recording data

4.3 Quality assurance and quality control Apart from selecting qualified and experienced inventory team mates (see Chapter 3.6) and their training (see Chapter 3.7), the key elements of the quality assurance were the following: • Use of a specific and detailed FRA manual (Lennertz et al. 2014), with instructions to be complied with to ensure that the field work followed SOPs, maximizing the homogeneity of the data acquisition and minimizing operating errors. • Regular supervision of the inventory teams, to check whether the inventory procedures, observations and measurements are carried out correctly.

To control the quality of the field work and to appreciate the measurement errors, 10% of the sampling units chosen at random and without prior knowledge of the inventory teams were planned to be re-measured independently. This target has been achieved in Eastern Samar. In Davao Oriental and in the Panay Mountain Range, however, time and budgetary constraints limited the control measurements to barely 5% of the sampling units.

The differences between the initial measurements and the (presumably correct) re-measurements (serving as reference) can be assessed through the mean absolute deviation (MAD) and the root mean square deviation (RMSD). Such deviations must be interpreted cautiously as long as the number of re-measured sampling units remains low (say less than about 16 sampling units).

52

In all sub-national FRAs, the following differences have been observed: • Diverging merchantable height measurements, because of the reduced visibility in the stands. Under such conditions, height measurements tend to be made from positions too close to the trees, leading to steep sighting angles, resulting in inaccurate estimates; this source of error was anticipated, hence the preference for allometric equations relying on DBH or DAB measurements only. • Diverging DBH or DAB measurements, due to non-standard measurement points above ground, to diameter tapes either not tightened or not held horizontally, or to the non-removal of vines during the initial measurements. • Diverging assessments of borderline trees (at the fringe of the 5 m and 10 m radii circles), falsely considered either to be part or not to be part of the sample; hence the importance of a through checking of such trees. • Omission of trees, especially in the small diameter range. • Omission of dead wood. • Diverging species identifications. • Trees reported dead during the initial measurements, but found alive during the re- measurements (particularly in Eastern Samar, due to the defoliating effect of typhoon Hagupit [Ruby], which affected the area shortly before the beginning of the field work).

Table 5 provides an estimate of the impact of the deviations of the initial measurements from the control measurements on the main variables of interest.

Where sufficient sampling units are available for the appreciation of the deviations (in Eastern Samar), the deviations turn out to be reasonable, as expected higher when height measurements are involved (i.e. for the estimation of the merchantable volume and of standing dead wood).

Table 5: Deviations of initial measurements from control measurements Variable of interest Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities) (2 municipalities) (37 municipalities) MAD RMSD MAD RMSD MAD RMSD [%]* [%]* [%]* [%]* [%]* [%]* Density 19.2 31.7 2.8 5.6 13.1 14.5 Basal area 23.8 29.1 3.6 5.5 7.9 9.7 Merchantable volume 27.2 33.0 14.7 27.4 30.2 42.9 Above-ground biomass 19.9 21.6 5.2 10.2 12.2 13.2 Standing dead wood 146.5 293.0 7.8 15.9 89.3 148.9 Lying dead wood 100.0 137.2 19.2 49.0 75.4 105.1 Litter 39.5 41.2 5.0 14.5 24.4 25.8 Number of plant species 20.4 25.9 5.3 9.6 18.9 22.8 based on 4 SUs based on 12 SUs based on 4 SUs

* with reference to the re-measurements

4.4. Time and cost of the data collection As reported in Chapter 3.4, the expected average output rate of 16 sampling units per month and per inventory team could not be reached. Table 6 summarizes the time and incremental cost (without the equipment) of the data collection in the three project sites.

On average, one inventory team observed and measured 9.4, 6.2 and 5.4 sampling units in the FRAs in Eastern Samar, Davao Oriental and the Panay Mountain Range, respectively. The highest productivity was observed in Eastern Samar, thanks to gentler terrain conditions and the relatively

53 small areal sampling frame (42,079 ha). The lowest productivity was observed in the Panay Mountain Range, due to the challenging topography and the comparatively large areal sampling frame, resulting in longer hiking distances to the inventory camps and sample points.

Table 6: Time and cost of data collection

Item Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities, (2 municipalities (37 municipalities, 81 SUs) 122 SUs) 86 SUs) Quantity Cost Quantity Cost Quantity Cost [PM]* [PHP]** [PM]* [PHP]** [PM]* [PHP]** Team Leaders 13.00 416,000 13.00 416,000 16.00 744,000 Assistants 13.00 364,000 13.00 364,000 16.00 672,000 Helpers 20.67 155,000 26.40 198,000 23.33 196,000 Consumables 68,000 86,000 68,000 Transportation 16,000 148,000 n.a. Total 1,019,000 1,212,000 1,722,000

* PM = person month ** 1 EUR = 48.6 - 56.0 EUR, on average 51.1 EUR

54

5. Data processing The FRA data have been processed using free open-source database and software development tools. Chapter 5.1 summarizes the characteristics of the database engine used. Chapter 5.2 briefly describes the database architecture, which is fully documented by Barrois (2017a). Chapter 5.3 presents the features of the database system application that has been developed to facilitate the entry, management and analysis of the inventory data. Chapter 5.4 concludes with a brief account of the data entry and the corresponding quality assurance and quality control measures. 5.1 Software The most popular open-source relational database management system (RDBMS) ORACLE MySQL (see https://www.mysql.com) has been used to process the FRA data. Apart from being free of license charges, the software offers the following advantages: • MySQL is available for different operating systems, notably for MICROSOFT Windows, Apple OS, Linux and Unix. • MySQL implements a client - server architecture, where tasks are partitioned between providers of a resource or service, called servers, and service requesters, called clients. Clients and servers typically communicate over a computer network on separate hardware, though both client and server may also reside in the same standalone system. • MySQL uses the most common standardized Structured Query Language (SQL) to access and manage databases. • MySQL supports very large databases, containing more than 200,000 tables and more than 50 million records. • Clients can connect to MySQL server using a variety of protocols, notably the internet (TCP/IP) and the Open Database Connectivity (ODBC). • Databases created with MySQL can be viewed and managed using a variety of free Graphical User Interfaces (GUI), e.g. MyDB Studio, phpMyAdmin, Sequel Pro, Workbench, etc.).

5.2 Database architecture The FRA data are stored in a relational database (see Textbox 18). To separate static (referential) and dynamic (transactional) data, the database is divided into two distinct sections: the common database and the inventory databases.

The common database contains all the tables and static entries that are used for the configuration, classification and referencing of the inventories and inventory data, more specifically: • The inventory properties, notably a unique identifier of each inventory, the name of the associated database, and the blow-up factors for the computation of the per ha results according to the radii of the nested sub-plots (see Chapter 3.3; the three FRAs used radii of 5 m and 10 m respectively, but the database offers the user the possibility to specify other radii). • The names of the geographical-political subdivisions (regions, provinces, cities, municipalities and barangays) according to the PSGC (see Chapter 3.1.1). • The species data (scientific and official common names, families, plant types), including the wood specific gravity of trees (see Chapter 3.1.4). • All used classification systems (terrain, land classification, land cover according to the 2010 NAMRIA national forest cover map, forest type and tree crown cover classes), see Chapter 4.2.

The inventory databases (one per inventory) contain all the data observed and measured by the inventories, more specifically: • The sample point data. • The list of inventory teammates. • The trees and standing dead wood data. • The lying dead wood data. 55

Figure 16 depicts the diagram of the database, whose architecture is fully documented by Barrois (2017a).

Textbox 18: Relational Database A relational database is a digital database whose organization is based on a relational model of data. The various software systems used to maintain relational databases are known as a relational database management systems (RDBMS).

The relational model organizes data into a set of tables composed each of columns and rows with a unique key (called primary key) identifying each row. A row (also called record) holds a data set representing a single item, e.g. a sampled tree with all the unique data pertaining to it. The columns (also called fields or attributes) structure the data of all rows of one table. A table holding the data of sampled trees could for instance feature the following columns: • Tree number (the primary key); • Sampling unit (SU) number, corresponding to the primary key of another table holding data pertaining to the SUs, such as the SU number, the coordinates, the forest type, the elevation, etc.; • Species code, corresponding to the primary key of yet another table holding data pertaining to the species, such as the species code, the botanical family, the wood specific gravity, etc.; • DBH; • etc.

Sampled trees Sampled trees Tree N° SU N° Species code DBH etc. Tree N° ...... SU N° 172 47 57 21 ... Species code 173 47 101 64 ... DBH 174 48 57 18 ... etc......

Sample points Sample points SU N° SU N° Zone Northing Easting Forest type Elevation etc. Zone ...... Northing 46 51 753,000 1,290,000 2 628 ... Easting 47 51 754,000 1,290,000 2 775 ... Forest type 48 51 755,000 1,290,000 1 609 ... Elevation ...... etc.

Species Species Species code Genus Species Family Gravity etc. Species code ...... Genus 100 Cedrela odorata Meliaceae 0.38 ... Species 101 Celtis luzonica Cannabaceae 0.55 ... Family 102 Celtis philippensis Cannabaceae 0.69 ... Gravity ...... etc.

Secondary keys, such as the SU number and the species code of the sampled trees table in the example cited above, that match the primary key of another table help to avoid storing all data items redundantly for each record. Instead, relations from the sampled tree table can be initiated using the secondary keys to lookup the data common to all trees of one SU and the data common to all trees of the same species in the two related tables. Such relations can work in both directions: from several records in one table to one record in another table (e.g. from all trees of a SU to the SU data), or from one record in one table to all records in another table (e.g. from one SU to all trees of that SU.

56

Figure 16: Diagram of the FRA database

57

5.3 Database system application To facilitate the entry, editing and analysis of the FRA data, a user-friendly database system application has been developed. The application was programmed using the ORACLE MySQL database engine (see Chapter 5.1) in combination with the widely used free application development tool ORACLE Java Standard Edition (SE) Development Kit (JDK) (see http://www.oracle.com/ technetwork/java/javase/overview/index.html). Through functions and stored procedures as well as a local Network Common Data Format (NetCDF) copy of the global gridded layer of E required for the calculation of the above-ground biomass using the 2014 Chave et al. allometric equation, the application implements the complete estimation design of the FRAs (see Chapter 3.8).

The software required for the installation and use of the FRA database system application Version 4.1, namely ORACLE MySQL Community Server 5.7.17.0 and ORACLE Java Standard Edition (SE) Runtime Environment [JRE] 8 Update 121, are all free of charge.

The FRA database system application offers the following features: • All essential data management operations: add, delete, edit, print to PDF, backup, restore data. • A series of data integrity checks, attracting the data typist's attention with the help of "traffic lights" (green = integrity check passed; orange = warning; red = integrity check failed) to missing, out-of-range, incompatible and unusual values. • Printing to PDF of a schematic representation of the location of the live trees and standing dead wood in the four nested plots of a sampling unit, facilitation their retrieval for re-mearsurement. • A comprehensive and versatile data analysis framework, performed on any selection of sampling units: o Tabulation of the frequency distribution (in %) of the sample points and plot centers according to site variable classes (elevation-, slope-, slope orientation-, terrain-, land classification-, land cover-, forest type-, tree crown cover- and vegetation layer ground coverage- classes). o Cross-tabulation of the frequency distribution (in %) of the sample points and plot centers for any combination of two site variables. o Display of the species occurrence by geographic coordinates and/or in Google Earth. o Computation of species richness, the Berger-Parker (d), Margalef (DMg), Shannon (H' and E) and Simpson (D) diversity indices. o Computation of relative frequency, relative density, relative dominance, importance, density (N/ha), basal area (G/ha), merchantable volume (V/ha), above-ground biomass (AGB/ha), below-ground biomass (BGB/ha), living biomass (LB/ha) by species and by diameter class. o Computation of standing dead wood (SDW/ha), lying dead wood (LDW/ha), litter (LI/ha), dead organic matter (DOM/ha), soil organic matter (SOM/ha) and forest carbon stock (C/ha). o Provision of tabular results including statistical precision estimates that can be exported to MS Excel, printed to PDF or depicted as pie or bar charts.

The application as well as its installation guide (Barrois 2017b) and user guide (Barrois et al. 2017) can be downloaded from http://forestry.denr.gov.ph/redd-plus-philippines/.

As an alternative, Open Foris could have been used (see Textbox 19). However, this software tools collection was only launched in October 2014, at a time when the preparations for the sub-national FRAs and the processing of their data had already started. The Open Foris "Collect" and "Calc" tools can be configured to offer similar functionalities to the FRA database system application, perhaps with a few limitations. The fully customized FRA database system application was programmed by a software developer with an input of 2 person-months. The configuration of the Open Foris tolls would certainly not have taken less time.

58

Textbox 19: Open Foris Open Foris (see http://www.openforis.org/) is a collaborative effort hosted by FAO offering free open-source software tools facilitating data collection, analysis and reporting for a wide range of monitoring purposes such as forest inventories, biodiversity assessments, deforestation monitoring, land use, land-use change and forestry (LULUCF) monitoring, detecting desertification and trees outside forest (TOF), and climate change reporting.

Presently, Open Foris features five tools: • "Collect" to configure a database, and to store and manage data collected in field-based inventories. • "Collect Mobile" to store and validate field-based inventory data using a mobile Android device. • "Calc" to perform calculations and to analyze data stored with the "Collect" tool. • "Collect Earth" to acquire and analyze GOOGLE Earth and BING Maps satellite imagery. • "Geospatial Toolkit" to process satellite imagery.

Open Foris makes use of ORACLE JRE. "Collect" and "Calc" run in a web server environment provided by the APACHE Tomcat server (see http://tomcat.apache.org/). By default, "Collect" uses the SQLite (see https://www.sqlite.org/) database engine, but can also be configured to use the PostgreSQL (see https://www.postgresql.org/) database engine, both using structured query language (SQL) to manage and query the database. "Calc" implements the inventory's estimation design and computes results using scripts written in the statistical computing language R (see https://www.r-project.org/). Aggregated results can be visualized and analyzed using the business analytics software Saiku (see http://meteorite.bi/products/saiku).

5.4 Quality assurance and quality control The FRA data were entered by trained staff with a technical background (B.Sc. Forestry). Apart from the qualification of the data typists, the key elements of the quality assurance of the data processing were the following: • Data entry closely following the data acquisition in the field, so that eventual gaps and errors observed could be ironed out with minimal effort, and the inventory teams be cautioned on typical and critical issues. • Thorough verification of the stored inventory data, assisted by completeness, value range and plausibility checks implemented in the database system application.

For quality control purposes, the stored data of 10% of the sampling units chosen at random were printed and subject to an independent comparison with the original field data forms. As for the quality control of the field data (see Chapter 4.3), the differences between the stored data and the original field data (serving as reference) can be appreciated through the MAD and the RMSD, as shown in Table 7.

The following differences have been observed: • Typing errors. • Omission of data. • Trees erroneously recorded twice.

Most data entry errors had very little impact on the variables of interest.

59

Table 7: Deviations of stored data from field forms

Variable of interest Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities) (2 municipalities) (37 municipalities) MAD RMSD MAD RMSD MAD RMSD [%]* [%]* [%]* [%]* [%]* [%]* Density 0.0 0.0 0.2 0.6 0.2 0.7 Basal area 2.0 5.4 0.1 0.1 1.2 3.3 Merchantable volume 2.5 7.2 0.0 0.1 3.0 8.3 Above-ground biomass 2.5 6.8 0.0 0.1 1.5 4.2 Standing dead wood 0.0 0.0 0.0 0.0 1.1 3.2 Lying dead wood 0.0 0.0 0.4 1.3 2.6 7.0 Litter 0.0 0.0 0.0 0.0 10.2 20.7 Number of plant species .0.0 0.0 0.0 0.0 2.7 7.7 based on 8 SUs based on 12 SUs based on 8 SUs

* with reference to the field forms

60

6. Data analysis and results The FRA data hold a wealth of information that can be analyzed from different perspectives. The most relevant types of general analyses are illustrated hereafter, starting with species diversity (Chapter 6.1). Stand composition and stand structure are dealt with in Chapters 6.2 and 6.3, respectively. Chapter 6.4 examines the timber stocks. The forest carbon stocks are scrutinized in Chapter 6.5. Chapter 6.6 proposes an approach for the reporting of the uncertainties of the estimates. The purpose of these Chapters is to show how the collected data can be translated into information, not a comprehensive presentation of the results of the three FRA, which has been done in separate reports (see Lennertz 2016a and 2016b, and Lennertz et al. 2017). 6.1 Species diversity Species diversity consists of two components: species richness and species abundance.

Species richness (i.e. the number of species observed or sampled) is the simplest measure of diversity. It is dependent on the sample size and the sampling unit size. The Margalef index also measures species richness. Despite the attempt to correct for the sample size, it remains strongly influenced by the sampling effort. Moreover, the results are very different if densities are used instead of absolute numbers (as required by definition, see Chapter 3.8.11).

Species abundance may be appreciated through the relative representation of a species, in terms of relative frequency, relative density and/or relative dominance (proportion of the basal area). The sum of the three figures corresponds per definition to the importance, which is meaningful for ranking purposes only, not for the comparison of absolute values). Shannon's H' index also measures species abundance, simultaneously taking the evenness of the species distribution into account. It increases as both the richness and the evenness increases. Shannon's E (equitability) index measures the evenness, ranging from 0 and 1. Lower values indicate more diversity, while higher values indicate less diversity (more evenness). The opposite of evenness, dominance, is measured by the Berger- Parker and the Simpson indices, ranging from 0 to 1. Again, lower values indicate more diversity, while higher values indicate less diversity (more dominance).

Table 8 provides a summary overview of the values of the species diversity indices observed in the three sub-national FRAs. The figures must be interpreted cautiously, taking into consideration both the sample size (number of sampling units) and the size of the sampling units.

The approach of the data analysis from a biodiversity perspective is illustrated hereafter based on the FRA conducted in Eastern Samar, where the forests are part of the Samar Island Natural Park (SINP).

Table 8: Species diversity indices

Variable Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities) (2 municipalities) (37 municipalities) Closed Open Closed Open Closed Open forests forests forests forests forests forests (37 SUs) (44 SUs) (18 SUs) (102 SUs) (33 SUs) (53 SUs) Species richness* 158 163 119 235 178 193 Margalef index* 21.4621 23.0822 16.4466 26.9638 23.4157 24.4401 Shannon H' index* 4.3713 4.5243 3.9892 4.5849 3.7033 4.3915 Shannon E index* 0.8635 0.8882 0.8347 0.8398 0.7147 0.8345 Berger-Parker index 0.0657 0.0899 0.1102 0.0539 0.0286 0.0249 Simpson index* 0.0187 0.0167 0.0346 0.0173 0.1278 0.0735

* referring to species identified through their scientific names; species only identified by local names are accounted as undetermined trees, bamboos, palms, rattan or tree ferns (as appropriate)

61

A total of 236 species have been identified through their scientific names, belonging to 148 genera and 62 families. The family with the largest number of species observed is the family of the Dipterocarpaceae (23 species, 5 genera, mainly Shorea), followed by the Moraceae (20 species, mainly Ficus), the Leguminosae and Myrtaceaea (12 species each), the Euphorbiaceae (11 species) and the Meliaceae (10 species).

Table 9 lists the threatened species (according to the International Union for Conservation of Nature and Natural Resources [IUCN] red list of threatened species, see http://www.iucnredlist.org/) sampled in the Eastern Samar FRA. Many Dipterocarps are considered "critically endangered" by IUCN. All species in Table 9 listed as endangered and critically endangered belong to the family Dipterocarpaceae.

Table 9: Threatened species (Eastern Samar FRA)

Vulnerable (VU) species Antipolo (Artocarpus blancoi) Malak-malak (Palaquium philippense) Balobo (Diplodiscus paniculatus) Malasantol (Sandoricum vidalii) Dalingdingan (Hopea foxworthyi) Molave (Vitex parviflora) Dalinsi (Terminalia pellucida) Nato (Palaquium luzoniense) Hamindang (Macaranga bicolor) Pahutan (Mangifera altissima) Ipil (Intsia bijuga) Pili (Canarium ovatum) Is-is (Ficus ulmifolia) Piling-liitan (Canarium luzonicum) Kalingag (Cinnamomum mercadoi) Puso-puso (Neolitsea vidalii) Katmon (Dillenia philippinensis) Takip-asin (Macaranga grandifolia) Laneteng gubat (Kibatalia gitingensis) Tanglin (Adenanthera intermedia) Lanutan (Mitrephora lanotan) Tindalo (Afzelia rhomboidea) Malakape (Psydrax dicoccos) Endangered (EN) species Mahogani (Swietenia mahagoni) Narig (Vatica mangachapoi) Mankono (Xanthostemon verdugonianus) Yakal-Mabolo (Shorea ciliata) Critically Endangered (CR) species Almon (Shorea almon) Manggachapui (Hopea acuminata) Apitong (Dipterocarpus grandiflorus) Mayapis (Shorea palosapis) Bagtikan (Shorea malaanonan) Panau (Dipterocarpus gracilis) Gisok-gisok (Hopea philippinensis) Red Lauan (Shorea negrosensis) Guijo (Shorea guiso) Tangile (Shorea polysperma) Hagakhak (Dipterocarpus validus) White Lauan (Shorea contorta) Highland Panau (Dipterocarpus hasseltii) Yakal (Shorea astylosa) Malapanau (Dipterocarpus kerrii) Yakal-Kaliot (Hopea malibato)

6.1.1 Species diversity of closed forests (Eastern Samar FRA) A total of 121 different species have been found and identified in the 18 sampling units in the closed forests, including three species whose local names could not be translated into common or scientific names. This is relatively few, compared to the 236 different species found and identified in the 120 sampling units across closed and open forests. With more sampling units in closed forests, the number of species would certainly still increase.

From 22 to 56, on average 33 different higher plant species have been observed per sampling unit.

Table 10 lists the 20 most "important" species (in the sense of the definition given in Chapter 2.6.2), led by Yakal, closely followed by Ulayan. As expected, seven Dipterocarps (Yakal, Mayapis, Guijo, Almon, Yakal-kaliot, Red Lauan and Tangile) are among the most important species, but also Ulayan (Oak), Bitanghol, two Sapotaceae (Malak-malak and Nato) and a palm (Sagisi).

62

Table 10: Relative frequency, density and dominance, importance and rank of the 20 most "important" species in closed forests (Eastern Samar FRA)

Species Relative Relative Relative Importance frequency density dominance [%] Rank [%] Rank [%] Rank [*] Rank Yakal 72.22 2 11.02 1 9.79 2 93.03 1 Shorea astylosa Ulayan (Oak) 77.78 1 6.31 3 4.05 7 88.13 2 Lithocarpus caudatifolius Bitanghol 66.67 3 8.58 2 3.21 10 78.46 3 Calophyllum blancoi Mayapis 61.11 4 2.82 6 12.35 1 76.28 4 Shorea palosapis Guijo 66.67 3 1.65 13 5.01 6 73.33 5 Shorea guiso Malak-malak 61.11 4 5.22 4 5.14 5 71.48 6 Palaquium philippense Kalipapa 61.11 4 2.44 8 2.28 12 65.82 7 Vitex quinata Ebony 55.56 5 1.60 14 1.07 20 58.22 8 Diospyros vera Nato 55.56 5 0.87 24 1.48 14 57.91 9 Palaquium luzoniense Sagisi 50.00 6 2.27 9 1.12 17 53.39 10 Heterospathe elata Almon 44.44 7 0.81 25 7.36 4 52.62 11 Shorea almon Badling 50.00 6 1.35 15 0.86 21 52.22 12 Astronia cumingiana Tiga 44.44 7 2.95 5 3.49 8 50.88 13 Tristaniopsis micrantha Yakal-Kaliot 44.44 7 2.95 5 2.39 11 49.79 14 Hopea malibato Yabnob 44.44 7 1.33 16 0.68 24 46.45 15 Horsfieldia costulata Red Lauan 33.33 9 1.25 18 8.92 3 43.50 16 Shorea negrosensis Tangile 38.89 8 0.41 37 1.43 15 40.72 17 Shorea polysperma Lanete 38.89 8 0.60 32 0.27 51 39.75 18 Wrightia pubescens ssp. laniti Duguan 33.33 9 1.92 11 1.96 13 37.21 19 Myristica philippinensis Mankono 33.33 9 2.76 7 1.10 18 37.20 20 Xanthostemon verdugonianus

* sum of relative frequency, density and dominance

Figure 17 shows that relatively few species, ranked in decreasing order of their contribution to N/ha, G/ha, V/ha and AGB/ha, constitute 50% of the totals: • Four species, namely Mayapis, Red Lauan, Almon and Yakal together represent nearly 56% of the merchantable volume; • Six species, namely Red Lauan, Yakal, Mayapis, Guijo, Almon and Tiga represent together almost 55% of the AGB, estimated using for trees the 2014 Chave et al. allometric equation factoring in wood specific gravity, see Textbox 20);

63

• Seven species, namely Mayapis, Yakal, Red Lauan, Almon, Malak-malak, Guijo and Ulayan (Oak) represent together more than 52% of the basal area; • Twelve species represent together some 51% of the density.

However, it takes 79, 66, 32 and 53 species to "explain" 95% of the total N/ha, G/ha, V/ha and AGB/ha, respectively.

Figure 17: N/ha, G/ha, V/ha and AGB/ha by number of species in closed forests (Eastern Samar FRA)

6.1.2 Species diversity of open forests (Eastern Samar FRA) A total of 246 different species have been found and identified in the 102 sampling units in the open forests, including fourteen species whose local names could not be translated into common or scientific names.

From 14 to 54, on average 30 different plant species have been observed per sampling unit.

Table 11 lists the 20 most important species, led by Ulayan (Oak), closely followed by Yakal. Seven Dipterocarps (Yakal, Mayapis, Guijo, Red Lauan, Almon, Yakal-kaliot and Palosapis) are among the most important species, but also Bitanghol, two Sapotaceae (Malak-malak and Nato) and a palm (Sagisi).

Significantly higher ranks in terms of the relative frequency compared to the relative density reveal a possibly clustered distribution of the following species: Anibong (Oncosperma tigillarium), Apanang, Bansalangin, Southern Bangkal or Hambabalud (Neonauclea formicaria) and Tibig (Ficus nota).

64

Table 11: Relative frequency, density and dominance, importance and rank of the 20 most "important" species in open forests (Eastern Samar FRA)

Species Relative Relative Relative Importance frequency density dominance [%] Rank [%] Rank [%] Rank [*] Rank Ulayan (Oak) 71.57 1 4.64 2 3.98 5 80.19 1 Lithocarpus caudatifolius Yakal 63.73 2 5.39 1 8.69 3 77.80 2 Shorea astylosa Mayapis 56.86 4 3.10 6 11.69 1 71.65 3 Shorea palosapis Guijo 61.76 3 2.87 7 4.81 4 69.44 4 Shorea guiso Red Lauan 55.88 5 2.81 8 10.28 2 68.97 5 Shorea negrosensis Bitanghol 56.86 4 4.59 3 2.51 8 63.97 6 Calophyllum blancoi Sagisi 50.98 6 3.56 5 1.19 17 55.72 7 Heterospathe elata Malak-malak 44.12 8 2.08 10 2.89 6 49.09 8 Palaquium philippense Duguan 45.10 7 1.76 11 1.67 13 48.53 9 Myristica philippinensis Kalipapa 42.16 9 1.65 14 1.05 19 44.86 10 Vitex quinata Apanang 38.24 10 4.10 4 1.78 11 44.11 11 Mallotus cumingii Almon 36.27 11 0.70 34 2.49 9 39.47 12 Shorea almon Mankono 34.31 12 1.24 17 1.92 10 37.47 13 Xanthostemon verdugonianus Dalunot 30.39 13 1.70 13 0.43 45 32.53 14 Pipturus arborescens Nato 30.39 13 0.73 32 1.19 16 32.31 15 Palaquium luzoniense Yakal-Kaliot 28.43 15 1.74 12 1.03 20 31.20 16 Hopea malibato Malatambis 29.41 14 0.93 22 0.51 35 30.86 17 Syzygium hutchinsonii Badling 27.45 16 1.15 20 0.66 27 29.26 18 Astronia cumingiana Piling-liitan 25.49 17 0.78 29 0.45 40 26.72 19 Canarium luzonicum Palosapis 24.51 18 1.00 21 1.22 15 26.72 20 Anisoptera thurifera

* sum of relative frequency, density and dominance

Figure 18 shows that relatively few species, ranked in decreasing order of their contribution to N/ha, G/ha, V/ha and AGB/ha, constitute 50% of the totals: • Four species, namely Mayapis, Red Lauan, Yakal and Guijo together represent some 53% of the merchantable volume; • Seven species, namely Red Lauan, Yakal, Mayapis, Guijo, Bansalangin (Mimusops elengi), Ulayan (Oak) and Malak-malak represent together almost 52% of the AGB;

65

• Ten species, namely Mayapis, Red Lauan, Yakal, Guijo, Ulayan, Malak-malak, Bansalangin, Bitanghol, Almon and Mankono represent together more almost 52% of the basal area; • Twenty-one species represent together some 51% of the density.

However, it takes 138, 115, 62 and 97 species to "explain" 95% of the total N/ha, G/ha, V/ha and AGB/ha, respectively.

Figure 18: N/ha, G/ha, V/ha and AGB/ha by number of species in open forests (Eastern Samar FRA)

Given the considerably different sample size in closed forests (18 sampling units) and open forests (102 sampling units), a comparison of the species diversity of the two strata, be it on the basis of the relative frequencies, densities and dominances (see Table 10 and Table 11), or on the basis of the diversity indices (see Table 8), is not feasible. 6.2 Stand composition Table 12 provides a summary overview of the stand composition in terms of density (N/ha), basal area (G/ha), merchantable volume (V/ha) and above-ground biomass (AGB/ha, estimated using for trees the 2014 Chave et al. allometric equation) observed in the tree sub-national FRAs.

Both the total stocking levels and the proportions of Dipterocarps observed in Eastern Samar are considerably higher than in the two other sites. The lowest figures are observed in the Panay Mountain Range.

The approach of the data analysis studying the stand composition is illustrated hereafter based on the FRA conducted in Davao Oriental.

66

Table 12: Stand composition

Variable Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities) (2 municipalities) (37 municipalities) Species Closed Open Closed Open Closed Open group forests forests forests forests forests forests (37 SUs) (44 SUs) (18 SUs) (102 SUs) (33 SUs) (53 SUs) N/ha Dipt. 141.5 67.1 421.3 254.4 30.6 19.6 [-/ha] Non-Dipt. 574.9 432.3 1,157.0 858.9 1,276.9 1,099.8 Others 26.3 21.7 54.8 108.4 86.2 51.9 Total 742.7 521.1 1,633.1 1,221.7 1,393.6 1,171.3 G/ha Dipt. 12.04 6.85 21.86 16.05 1.32 0.91 [m²/ha] Non-Dipt. 22.80 15.17 20.39 17.59 24.95 19.30 Others 0.40 0.23 0.68 1.58 1.38 1.07 Total 35.24 22.26 42.92 35.22 27.64 21.28 V/ha Dipt. 104.80 62.98 163.44 139.82 7.63 2.95 [m³/ha] Non-Dipt. 115.97 80.21 66.52 68.41 65.70 50.99 Total 220.77 143.19 229.97 208.23 73.33 53.87 AGB/ha Dipt. 147.92 92.02 251.91 180.93 13.54 8.98 [t d.m./ha] Non-Dipt. 249.15 155.66 182.46 158.03 224.43 165.39 Others 0.92 0.64 2.09 3.42 1.07 1.85 Total 398.00 248.32 436.46 342.37 239.40 176.15

6.2.1 Stand composition of closed forests (Davao Oriental FRA) Table 13 summarizes and Figure 19 illustrates the stand composition of the closed forests in terms of N/ha, G/ha, V/ha and AGB/ha, estimated on the basis of 37 sampling units.

In terms of basal area and AGB, Dipterocarps account for slightly more than 1/3 of the total stock. Their share is even higher (almost 50%) in terms of merchantable volume, but less (1/5) in terms of density. This stems from the fact that the average size of Dipterocarps, as revealed through the quadratic mean diameter (Dg), is considerably larger (32.9 cm) than the Dg of Non-Dipterocarps (22.5 cm).

The five most dominant Dipterocarps in terms of basal area are Tangile, Guijo, Narig, Almon and Yakal. Together, they represent 76% of the total Dipterocarp merchantable volume, and around 36% of the total merchantable volume, all species combined.

The ten most dominant Non-Dipterocarps in terms of basal area, led by Ulayan and Ulaian, closely followed by Balete (being stranglers, their size is arguable) and Nato, represent together a substantial share of G/ha (45.1%), and the lion share of V/ha (53.6%) and AGB/ha (52.6%) of their group.

The palms encountered are essentially Sagisi, and to a lesser extent Pugahan and Ulango.

67

Table 13: Stand composition of closed forests (Davao Oriental FRA)

Species group N/ha G/ha V/ha AGB/ha Species [/ha] [%] [m²/ha] [%] [m³/ha] [%] [t d.m./ha] [%] Dipterocarps Tangile 30.1 4.1 3.66 10.4 33.56 15.2 44.79 11.3 Guijo 19.8 2.7 1.78 5.1 14.39 6.5 26.14 6.6 Narig 17.4 2.3 1.35 3.8 10.53 4.8 19.62 4.9 Almon 12.7 1.7 1.20 3.4 11.35 5.1 10.15 2.6 Yakal 12.0 1.6 1.14 3.2 9.61 4.4 18.04 4.5 Other Dipt. 49.5 6.7 2.91 8.3 25.36 11.5 29.19 7.3 Sub-Total Dipt. 141.5 19.1 12.04 34.2 104.80 47.5 147.92 37.2 Non-Dipterocarps Ulayan (Oak) 48.8 6.6 1.75 5.0 7.23 3.3 16.48 4.1 Ulaian 38.7 5.2 1.68 4.8 9.01 4.1 20.15 5.1 Balete 3.0 0.4 1.44 4.1 10.11 4.6 26.01 6.5 Nato 12.3 1.7 1.26 3.6 9.04 4.1 13.37 3.4 Dacrydium beccarii 1.3 0.2 0.80 2.3 5.74 2.6 15.40 3.9 Kalingag 26.2 3.5 0.76 2.2 2.94 1.3 5.07 1.3 Lanipga 1.9 0.3 0.70 2.0 5.45 2.5 10.32 2.6 Hindang 8.6 1.2 0.70 2.0 4.38 2.0 9.20 2.3 Saguimsim 9.7 1.3 0.65 1.8 3.65 1.7 7.38 1.9 Talisay-gubat 3.9 0.5 0.55 1.6 4.62 2.1 7.70 1.9 Other Non-Dipt. 420.5 56.6 12.51 35.5 53.80 24.4 118.07 29.7 Sub-Total Non-Dipt. 574.9 77.4 22.80 64.7 115.97 52.5 249.15 62.6 Palms 25.4 3.4 0.39 1.1 - - 0.90 0.2 Bamboos 0.9 0.1 0.01 0.0 - - 0.02 0.0 Total 742.7 100.0 35.24 100.0 220.77 100.0 398.00 100.0

68

N/ha [/ha]

G/ha [m²/ha]

V/ha [m³/ha]

AGB/ha [t d.m./ha]

Non- Palms Other Dipterocarps Bamboos Tangile Guijo Narig Almon Yakal Dipterocarps Dipterocarps

Figure 19: Stand composition of closed forests (Davao Oriental FRA)

69

6.2.2 Stand composition of open forests (Davao Oriental FRA) Table 14 summarizes and Figure 20 illustrates the stand composition of the open forests in terms of N/ha, G/ha, V/ha and AGB/ha, estimated on the basis of 44 sampling units.

Like in the closed forests, Dipterocarps account for about 1/3 of the basal area and AGB of the open forests. Their share is even higher (44 %) in terms of merchantable volume, but less (13 %) in terms of density. This is thanks to their average size, in terms of Dg, which is again considerably larger (36.1 cm) than the Dg of Non-Dipterocarps (21.1 cm).

The five most dominant Dipterocarps in terms of basal area are Tangile, Narig, Guijo, Bagtikan and Yakal (the same species as in the closed forests, except Almon, which is superseded by Bagtikan, and following a slightly different ranking). Together, they represent 79% of the total Dipterocarp merchantable volume, and around 35% of the total merchantable volume, all species combined.

The ten most dominant Non-Dipterocarps in terms of basal area, led by Ulayan followed by Nato, Balete (being stranglers, their size is arguable) and Balukanag, together represent 44.6% of G/ha, 57.3% of V/ha and 55.3% of AGB/ha of their group.

The palms encountered are essentially Sagisi, some Pugahan, and very few coconuts.

Table 14: Stand composition of open forests (Davao Oriental FRA)

Species group N/ha G/ha V/ha AGB/ha Species [/ha] [%] [m²/ha] [%] [m³/ha] [%] [t d.m./ha] [%] Dipterocarps Tangile 13.9 2.7 2.63 11.8 29.24 20.4 33.52 13.5 Narig 7.2 1.4 0.79 3.5 6.69 4.7 14.10 5.7 Guijo 8.5 1.6 0.66 3.0 5.08 3.5 9.83 4.0 Bagtikan 12.1 2.3 0.60 2.7 5.42 3.8 6.68 2.7 Yakal 3.6 0.7 0.49 2.2 3.41 2.4 8.19 3.3 Other Dipt. 21.7 4.2 1.67 7.5 13.14 9.2 19.70 7.9 Sub-Total Dipt. 67.1 12.9 6.85 30.8 62.98 44.0 92.02 37.1 Non-Dipterocarps Ulayan (Oak) 46.8 9.0 1.66 7.5 8.02 5.6 16.21 6.5 Nato 8.0 1.5 1.13 5.1 10.26 7.2 14.37 5.8 Balete 2.2 0.4 0.90 4.0 5.10 3.6 18.02 7.3 Balukanag 5.4 1.0 0.62 2.8 6.35 4.4 8.74 3.5 Balobo 6.7 1.3 0.45 2.0 1.72 1.2 5.53 2.2 Moluccan sau 2.2 0.4 0.42 1.9 2.85 2.0 3.80 1.5 Mankono 3.6 0.7 0.42 1.9 2.91 2.0 6.26 2.5 Saguimsim 7.1 1.4 0.40 1.8 1.75 1.2 4.58 1.8 Talisay-gubat 2.0 0.4 0.39 1.8 3.61 2.5 5.25 2.1 Binuang 0.5 0.1 0.38 1.7 3.38 2.4 3.31 1.3 Other Non-Dipt. 347.8 66.7 8.40 37.7 34.24 23.9 69.58 28.0 Sub-Total Non-Dipt. 432.3 83.0 15.17 68.1 80.21 56.0 155.66 62.7 Palms 14.5 2.8 0.18 0.8 - - 0.48 0.2 Bamboos 7.2 1.4 0.05 0.2 - - 0.16 0.1 Total 521.1 100.0 22.26 100.0 143.19 100.0 248.32 100.0

70

N/ha [/ha]

G/ha [m²/ha]

V/ha [m³/ha]

AGB/ha [t d.m./ha]

Non- Palms Other Dipterocarps Bamboos Tangile Guijo Narig Bagtikan Yakal Dipterocarps Dipterocarps

Figure 20: Stand composition of open forests (Davao Oriental FRA)

71

6.3 Stand structure Table 15 provides a summary appreciation of the stand structure in terms of the distribution of the density (N/ha), basal area (G/ha) and above-ground biomass (AGB/ha, estimated using for trees the 2014 Chave et al. allometric equation) by diameter classes observed in the three sub-national FRAs.

The closed forests inventoried in Davao Oriental feature a very balanced distribution of the basal area over the diameter classes. In the Panay Mountain Range, the distribution by diameter classes is skewed towards the small-sized trees.

The trees in the closed and open forests in Davao Oriental are on average strikingly bigger (as revealed by their Dg), followed by the trees in Eastern Samar and in the Panay Mountain Range.

The approach of the data analysis concerned with the stand structure is illustrated hereafter based on the FRA conducted in Davao Oriental.

Table 15: Stand structure

Diameter class Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities) (2 municipalities) (37 municipalities) Closed Open Closed Open Closed Open forests forests forests forests forests forests (37 SUs) (44 SUs) (18 SUs) (102 SUs) (33 SUs) (53 SUs) N/ha [-/ha] [5 cm - 20 cm] 559.2 425.4 1,407.6 1,018.0 1,236.6 1,043.8 [20 cm - 40 cm] 119.6 57.2 169.7 146.5 121.8 104.9 [40 cm - 60 cm] 39.4 21.4 37.1 42.8 25.5 16.5 [60 cm - 80 cm] 15.2 8.5 11.5 10.0 6.0 4.0 [80 cm - 9.3 8.7 7.1 4.5 3.6 2.2 Total 742.7 521.1 1,633.1 1,221.7 1,393.6 1,171.3 G/ha [m²/ha] [5 cm - 20 cm] 7.15 4.96 14.46 10.71 10.57 8.43 [20 cm - 40 cm] 7.81 3.78 10.79 9.60 7.38 6.45 [40 cm - 60 cm] 7.31 3.84 6.67 7.72 4.56 2.94 [60 cm - 80 cm] 5.33 2.98 4.21 3.53 2.14 1.38 [80 cm - 7.61 6.70 6.80 3.66 2.99 2.09 Total 35.24 22.26 42.92 35.22 27.64 21.28 AGB/ha t d.m./ha] [5 cm - 20 cm] 44.71 29.07 93.25 63.30 57.77 45.70 [20 cm - 40 cm] 71.87 33.50 103.41 86.66 62.38 49.45 [40 cm - 60 cm] 86.46 43.38 76.17 90.81 50.49 31.23 [60 cm - 80 cm] 68.13 38.37 55.38 44.90 22.48 17.36 [80 cm - 126.83 103.99 108.26 56.70 46.29 32.48 Total 398.00 248.32 436.46 342.37 239.40 176.22 Dg [cm] Total 24.6 23.3 18.3 19.2 15.9 15.2

72

6.3.1 Stand structure of closed forests (Davao Oriental FRA) The stand structure of the closed forests is summarized hereafter in terms of the following: • Density (N/ha) by diameter class, summarized in Table 16 and illustrated in Figure 21; • Basal area (G/ha) by diameter class, summarized in Table 17 and illustrated in Figure 22; and • Above-ground biomass (AGB/ha) by diameter class, summarized in Table 18 and illustrated in Figure 24.

On average, the closed forests count per hectare 142 Dipterocarp trees, 575 Non-Dipterocarp trees, 25 palms, 1 bamboo and 18 standing dead wood.

As expected, N/ha by diameter class shows a typical inverse "J"-shaped distribution, except for trees with a DBH < 10 cm, which appear to be lacking in numbers.

The rise of N/ha for trees with DBH ≥ 90 cm is due to a few quite large Dipterocarps (essentially Tangile), Balete (being stranglers, their size is arguable), Lanigpa, Talisay-gubat and Hindang.

The distribution of the shares of the predominant Dipterocarps and Non-Dipterocarps by diameter class does not reveal any peculiarity.

Table 16: Stand structure in terms of N/ha of closed forests (Davao Oriental FRA)

Species group Density by diameter class Species [5 - 20 cm] [20 - 40 cm] [40 - 60 cm] [60 - 80 cm] [80 cm - Total [/ha] [/ha] [/ha] [/ha] [/ha] [/ha] Dipterocarps Almon 6.9 3.0 1.5 1.1 0.2 12.7 Guijo 10.3 5.2 2.6 1.3 0.4 19.8 Narig 8.6 5.2 3.4 - 0.2 17.4 Tangile 18.0 4.4 3.4 2.1 2.2 30.1 Yakal 7.8 1.5 1.5 0.8 0.4 12.0 Other Dipt. 35.2 7.6 2.9 3.1 0.7 49.5 Total Dipt. 86.8 26.9 15.3 8.4 4.1 141.5 Non-Dipterocarps Bitanghol 17.2 1.8 0.2 - - 19.1 Hindang 6.0 1.0 1.0 0.2 0.2 8.6 Kalingag 20.6 4.8 0.8 - - 26.2 Nato 2.6 5.4 3.5 0.9 - 12.3 Salingkugi 12.9 1.9 0.2 0.2 - 15.3 Ulayan (Oak) 37.0 10.3 1.0 0.2 0.2 48.8 Other Non-Dipt. 351.9 65.6 17.1 5.4 4.8 444.6 Total Non-Dipt. 448.2 90.8 23.8 6.9 5.2 574.9 Palms 23.3 1.9 0.2 - - 25.4 Bamboos 0.9 - - - - 0.9 Total 559.2 119.6 39.4 15.2 9.3 742.7 Standing dead wood 7.7 6.0 3.5 0.9 - 18.1

73

Figure 21: Stand structure in terms of N/ha of closed forests (Davao Oriental FRA)

74

On average, the basal area of the closed forests amounts to 35.2 m²/ha, which is more than the 30.5 m²/ha (for trees with DBH ≥ 15 cm) observed 1979 to 1983 by the FAO-assisted Northeastern Mindanao Pilot Project in 92 sampling units in "Old Growth Forests" of Region XI (MNR 1986).

The distribution of G/ha by diameter class does not reveal any particularity.

Table 17: Stand structure in terms of G/ha of closed forests (Davao Oriental FRA)

Species group Basal area by diameter class Species [5 - 20 cm] [20 - 40 cm] [40 - 60 cm] [60 - 80 cm] [80 cm - Total [m²/ha] [m²/ha] [m²/ha] [m²/ha] [m²/ha] [m²/ha] Dipterocarps Almon 0.11 0.18 0.29 0.44 0.19 1.20 Guijo 0.16 0.33 0.49 0.47 0.34 1.78 Narig 0.15 0.36 0.66 - 0.18 1.35 Tangile 0.20 0.30 0.72 0.74 1.70 3.66 Yakal 0.09 0.12 0.32 0.35 0.28 1.14 Other Dipt. 0.42 0.52 0.57 1.00 0.36 2.91 Total Dipt. 1.13 1.81 3.05 3.00 3.05 12.04 Non-Dipterocarps Bitanghol 0.26 0.11 0.03 - - 0.41 Hindang 0.06 0.07 0.20 0.07 0.29 0.70 Kalingag 0.29 0.33 0.14 - - 0.76 Nato 0.03 0.36 0.61 0.26 - 1.26 Salingkugi 0.16 0.10 0.04 0.08 - 0.37 Ulayan (Oak) 0.63 0.67 0.21 0.08 0.16 1.75 Other Non-Dipt. 4.35 4.25 3.01 1.84 4.11 17.55 Total Non-Dipt. 5.78 5.89 4.24 2.33 4.56 22.80 Palms 0.25 0.11 0.03 - - 0.39 Bamboos 0.01 - - - - 0.01 Total 7.15 7.81 7.31 5.33 7.61 35.24 Standing dead wood 0.11 0.46 0.54 0.24 0.00 1.35

75

Figure 22: Stand structure in terms of G/ha of closed forests (Davao Oriental FRA)

76

On average, the above-ground biomass of the closed forests amounts to 398 t d.m./ha, which corresponds to the median of the range from 280 t d.m./ha to 520 t d.m./ha referred to by IPCC as tier 1 estimate for tropical rainforest of insular Asia.

Figure 23 shows that 99% of AGB/ha is composed of trees with DBH ≥ 10 cm.

The distribution of AGB/ha by diameter class does not reveal any particularity.

Table 18: Stand structure in terms of AGB/ha of closed forests (Davao Oriental FRA)

Species group Above-ground biomass by diameter class Species [5 - 20 cm] [20 - 40 cm] [40 - 60 cm] [60 - 80 cm] [80 cm - Total [t d.m./ha] [t d.m./ha] [t d.m./ha] [t d.m./ha] [t d.m./ha] [t d.m./ha] Dipterocarps Almon 0.49 1.13 2.39 4.06 2.09 10.15 Guijo 1.24 3.85 7.07 7.65 6.33 26.14 Narig 1.32 4.47 10.14 - 3.68 19.62 Tangile 1.05 2.52 7.70 8.88 24.64 44.79 Yakal 0.74 1.42 4.73 6.00 5.16 18.04 Other Dipt. 2.56 4.15 6.01 11.45 5.01 29.18 Total Dipt. 7.40 17.54 38.04 38.04 46.91 147.92 Non-Dipterocarps Bitanghol 1.43 0.83 0.30 - - 2.56 Hindang 0.41 0.65 2.35 0.92 4.86 9.20 Kalingag 1.44 2.37 1.25 - - 5.07 Nato 0.14 3.29 6.67 3.27 - 13.37 Salingkugi 1.00 0.87 0.44 1.03 - 3.34 Ulayan (Oak) 4.33 6.19 2.34 1.14 2.49 16.48 Other Non-Dipt. 27.81 39.97 35.05 23.72 72.58 199.13 Total Non-Dipt. 36.56 54.17 48.40 30.08 79.93 249.15 Palms 0.72 0.16 0.02 - - 0.90 Bamboos 0.02 - - - - 0.02 Total 44.71 71.87 86.46 68.13 126.83 398.00

Figure 23: AGB/ha of closed forests by DBH threshold (Davao Oriental FRA)

77

Figure 24: Stand structure in terms of AGB/ha of closed forests (Davao Oriental FRA)

78

6.3.2 Stand structure of open forests (Davao Oriental FRA) As for the closed forests, the stand structure of the open forests is summarized hereafter in terms of the following: • Density (N/ha) by diameter class, summarized in Table 19 and illustrated in Figure 25; • Basal area (G/ha) by diameter class, summarized in Table 20 and illustrated in Figure 26; and • Above-ground biomass (AGB/ha) by diameter class, summarized in Table 21 and illustrated in Figure 28.

On average, the open forests count per hectare 67 Dipterocarp trees, 432 Non-Dipterocarp trees, 15 palms, 7 bamboos and 7 standing dead wood. A t-test confirms that at a confidence level of 99%, the density of trees (521/ha) is significantly lower than in the closed forests (743/ha).

The distribution of N/ha by diameter class follows a similar pattern as in the closed forests, though at a lower level. The rise of N/ha for trees with larger diameters is observed again, here for trees with DBH ≥ 80 cm, due to a few large Dipterocarps (essentially Tangile), Nato, Balukanag and Talisay- gubat. The relative density of Dipterocarps by diameter class reveals that Yakal has a reduced share in the lower diameter classes (hinting that it is not regeneration well). Among the Non-Dipterocarps, a similar, though less pronounced trend can be observed for Nato.

Table 19: Stand structure in terms of N/ha of open forests (Davao Oriental FRA)

Species group Density by diameter class Species [5 - 20 cm] [20 - 40 cm] [40 - 60 cm] [60 - 80 cm] [80 cm - Total [/ha] [/ha] [/ha] [/ha] [/ha] [/ha] Dipterocarps Bagtikan 10.0 0.9 0.4 0.4 0.4 12.1 Guijo 5.0 1.5 1.5 0.4 0.1 8.5 Narig 5.0 1.0 0.2 0.2 0.8 7.2 Tangile 6.5 2.4 2.0 1.0 2.0 13.9 Yakal 2.1 - 0.7 0.4 0.4 3.6 Other Dipt. 14.8 3.4 1.7 1.1 0.8 21.8 Total Dipt. 43.4 9.2 6.5 3.5 4.5 67.1 Non-Dipterocarps Badling 8.6 0.7 - - - 9.4 Bitanghol 5.0 1.6 0.2 - - 6.9 Lipang-kalabaw 10.8 0.5 - - - 11.4 Nato 4.3 1.0 0.6 1.3 0.8 8.0 Tibig 7.2 0.7 - - - 8.0 Ulayan (Oak) 36.2 6.9 3.5 0.4 - 46.8 Other Non-Dipt. 288.9 35.8 10.6 3.4 3.5 341.8 Total Non-Dipt. 361.0 47.2 14.9 5.1 4.3 432.3 Palms 13.8 0.8 - - - 14.5 Bamboos 7.3 - - - - 7.2 Total 425.4 57.2 21.4 8.5 8.7 521.1 Standing dead wood 2.9 2.8 0.7 - - 6.5

79

Figure 25: Stand structure in terms of N/ha of open forests (Davao Oriental FRA)

80

On average, the basal area of the open forests amounts to 22.3 m²/ha. This is significantly less than G/ha of the closed forests (35.2 m²/ha), as confirmed by a t-test at a confidence level of 99%. The order of magnitude is much higher than the 14.0 m²/ha (for trees with DBH ≥ 15 cm) observed 1979 to 1983 by the FAO-assisted Northeastern Mindanao Pilot Project in 169 sampling units in "Residual Forests" of Region XI (MNR 1986).

The distribution of G/ha by diameter class does not reveal any particularity, but confirms the observation derived from the distribution of N/ha that Yakal and Nato are underrepresented in the lower diameter classes.

Table 20: Stand structure in terms of G/ha of open forests (Davao Oriental FRA)

Species group Basal area by diameter class Species [5 - 20 cm] [20 - 40 cm] [40 - 60 cm] [60 - 80 cm] [80 cm - Total [m²/ha] [m²/ha] [m²/ha] [m²/ha] [m²/ha] [m²/ha] Dipterocarps Bagtikan 0.11 0.06 0.06 0.15 0.22 0.60 Guijo 0.07 0.07 0.25 0.13 0.14 0.66 Narig 0.08 0.07 0.04 0.08 0.52 0.79 Tangile 0.08 0.17 0.37 0.42 1.59 2.63 Yakal 0.03 - 0.13 0.13 0.20 0.49 Other Dipt. 0.17 0.24 0.34 0.36 0.57 1.68 Total Dipt. 0.54 0.61 1.19 1.27 3.24 6.85 Non-Dipterocarps Badling 0.12 0.07 - - - 0.19 Bitanghol 0.05 0.11 0.03 - - 0.19 Lipang-kalabaw 0.15 0.02 - - - 0.17 Nato 0.07 0.09 0.10 0.42 0.44 1.13 Tibig 0.04 0.04 - - - 0.08 Ulayan (Oak) 0.42 0.50 0.59 0.14 - 1.66 Other Non-Dipt. 3.39 2.29 1.92 1.15 3.01 11.75 Total Non-Dipt. 4.24 3.12 2.64 1.71 3.45 15.17 Palms 0.14 0.05 - - - 0.18 Bamboos 0.05 - - - - 0.05 Total 4.96 3.78 3.84 2.98 6.70 22.26 Standing dead wood 0.03 0.20 0.09 - - 0.32

81

Figure 26: Stand structure in terms of G/ha of open forests (Davao Oriental FRA)

82

On average, the above-ground biomass of the open forests amounts to 248 t d.m./ha, which is significantly less than AGB/ha of the closed forests (398 t d.m./ha), as confirmed by a t-test at a confidence level of 99%.

Figure 27 shows that like in the closed forests, 99% of AGB/ha in the open forests is composed of trees with DBH ≥ 10 cm.

Table 21: Stand structure in terms of AGB/ha of open forests (Davao Oriental FRA)

Species group Above-ground biomass by diameter class Species [5 - 20 cm] [20 - 40 cm] [40 - 60 cm] [60 - 80 cm] [80 cm - Total [t d.m./ha] [t d.m./ha] [t d.m./ha] [t d.m./ha] [t d.m./ha] [t d.m./ha] Dipterocarps Bagtikan 0.63 0.48 0.67 2.01 2.88 6.68 Guijo 0.57 0.81 3.47 2.18 2.79 9.83 Narig 0.64 0.88 0.67 1.49 10.41 14.10 Tangile 0.48 1.49 3.86 5.09 22.60 33.52 Yakal 0.27 - 1.85 2.23 3.83 8.19 Other Dipt. 0.95 2.05 3.54 3.78 9.42 19.70 Total Dipt. 3.54 5.71 14.06 16.78 51.93 92.02 Non-Dipterocarps Badling 0.79 0.63 - - - 1.42 Bitanghol 0.29 0.86 0.22 - - 1.36 Lipang-kalabaw 0.97 0.16 - - - 1.13 Nato 0.51 0.85 1.17 5.52 6.34 14.37 Tibig 0.22 0.39 - - - 0.61 Ulayan (Oak) 2.63 4.84 6.86 1.89 - 16.21 Other Non-Dipt. 19.55 19.99 21.07 14.19 45.73 120.56 Total Non-Dipt. 24.96 27.72 29.32 21.60 52.07 155.66 Palms 0.42 0.07 - - - 0.48 Bamboos 0.16 - - - - 0.16 Total 29.07 33.50 43.38 38.37 103.99 248.32

Figure 27: AGB/ha of open forests by DBH threshold (Davao Oriental FRA)

83

Figure 28: Stand structure in terms of AGB/ha of open forests (Davao Oriental FRA)

84

6.4 Timber stocks Table 22 provides a summary overview of the merchantable volume and its distribution by diameter classes and species groups (Dipterocarps and Non-Dipterocarps) observed in the three sub-national FRAs.

Compared to the forests inventoried in Eastern Samar and Davao Oriental, the timber stocks observed in the Panay Mountain Range are quite low.

The approach of the data analysis looking into the timber stock is illustrated hereafter based on the FRA conducted in Davao Oriental.

Table 22: Timber stocks

Diameter Davao Oriental Eastern Samar Panay Mountain Range class (3 municipalities) (2 municipalities) (37 municipalities) Species Closed Open Closed Open Closed Open group forests forests forests forests forests forests (37 SUs) (44 SUs) (18 SUs) (102 SUs) (33 SUs) (53 SUs) V/ha [m³/ha] [5 cm - Dipt. 15.09 5.19 32.56 31.39 0.92 1.08 40 cm] Non-Dipt. 35.39 18.59 41.26 33.98 26.06 19.93 Total 50.47 23.78 73.83 65.36 26.98 21.00 [40 cm - Dipt. 27.70 10.65 41.47 46.37 1.99 0.99 60 cm] Non-Dipt. 30.89 18.07 14.22 22.27 18.53 10.80 Total 58.59 28.72 55.69 68.64 20.52 11.79 [60 cm - Dipt. 30.30 12.36 32.81 28.17 3.77 - 80 cm] Non-Dipt. 33.37 13.24 5.07 6.22 6.26 6.89 Total 65.09 25.59 37.88 34.39 10.03 6.89 [80 cm - Dipt. 31.71 34.79 56.61 33.90 0.95 0.88 Non-Dipt. 33.37 30.32 5.97 5.94 14.85 13.39 Total 65.09 65.10 62.58 39.85 15.80 14.27 Total Dipt. 104.80 62.98 163.44 139.82 7.63 2.95 Non-Dipt. 115.97 80.21 66.52 68.41 65.70 50.99 Total 220.77 143.19 229.97 208.23 73.33 53.94

6.4.1 Timber stocks in closed forests (Davao Oriental FRA) Table 23 summarizes and Figure 29 illustrates the distribution of the merchantable volume in the closed forests by diameter class and main species.

On average, the merchantable volume in the closed forests amounts to 221 m³/ha.

This comes close to the order of magnitude of 233 m³/ha (for trees with DBH ≥ 15 cm) observed 1979 to 1983 by the FAO-assisted Northeastern Mindanao Pilot Project in 92 sampling units in "Old Growth Forests" of Region XI (MNR 1986). Then and now, around 1/3 of V/ha is concentrated on trees with DBH ≥ 75 cm.

However, the proportion of Dipterocarps has dramatically reduced over time, from then 65.6% to now 47.5%. Dipterocarp species whose share has strongly diminished are Apitong, Bagtikan, Red Lauan and Mayapis.

85

Table 23: Merchantable volume in closed forests (Davao Oriental FRA)

Species group Merchantable volume by diameter class Species [5 - 40 cm] [40 - 60 cm] [60 - 80 cm] [80 cm - Total [m³/ha] [m³/ha] [m³/ha] [m³/ha] [m³/ha] Dipterocarps Almon 1.54 2.92 4.56 2.32 11.35 Guijo 3.07 4.53 4.00 2.79 14.39 Narig 0.80 4.56 2.80 2.36 10.53 Tangile 2.14 6.48 7.97 16.98 33.56 Yakal 0.82 2.36 3.38 3.05 9.61 Other Dipt. 6.72 6.85 7.59 4.21 25.36 Total Dipt. 15.09 27.70 30.30 31.71 104.80 Non-Dipterocarps Bitanghol 0.76 0.34 - - 1.10 Hindang 0.43 1.28 0.33 2.34 4.38 Kalingag 2.04 0.90 - - 2.94 Nato 2.37 4.62 2.05 - 9.04 Salingkugi 0.47 0.13 0.45 - 1.05 Ulayan (Oak) 4.01 1.13 0.45 1.63 7.23 Other Non-Dipt. 25.31 22.49 13.04 29.40 90.23 Total Non-Dipt. 35.39 30.89 16.32 33.37 115.97 Total 50.47 58.59 46.62 65.09 220.77

86

Figure 29: Merchantable volume in closed forests (Davao Oriental FRA)

87

6.4.2 Timber stocks in open forests (Davao Oriental FRA) Table 24 summarizes and Figure 30 illustrates the distribution of the merchantable volume in the open forests by diameter class and main species.

On average, the merchantable volume in the open forests amounts to 143 m³/ha. Based on t-tests, this is significantly less than the merchantable volume in the closed forests (221 t d.m./ha) at a confidence level of 95%, but not at a confidence level of 99%.

The merchantable volume in the open forests is actually considerably higher than the 91 m³/ha (for trees with DBH ≥ 15 cm) observed 1979 to 1983 by the FAO-assisted Northeastern Mindanao Pilot Project in 169 SUs in "Residual Forests" of Region XI (MNR 1986). At that time, only about 10% of V/ha was concentrated on trees with DBH ≥ 75 cm, compared to 48% now. Obviously, the "Residual Forests" inventoried from 1979 to 1983 were very recently logged-over at that time, hence the very low (i) merchantable volume and (ii) proportion of trees with DBH ≥ 75 cm, while the open forests inventoried from 2015 to 2016 have to some extent recovered.

The proportion of Dipterocarps is quite comparable, 48% by then compared to 44% now. Dipterocarp species whose share has strongly diminished are Apitong and Mayapis.

Table 24: Merchantable volume in open forests (Davao Oriental FRA)

Species group Merchantable volume by diameter class Species [5 - 40 cm] [40 - 60 cm] [60 - 80 cm] [80 cm - Total [m³/ha] [m³/ha] [m³/ha] [m³/ha] [m³/ha] Dipterocarps Bagtikan 0.43 0.73 1.87 2.39 5.42 Guijo 0.54 2.11 1.39 1.03 5.08 Narig 0.59 0.23 0.94 4.93 6.69 Tangile 1.70 3.80 3.76 19.99 29.24 Yakal - 0.87 1.14 1.40 3.41 Other Dipt. 1.93 2.91 3.26 5.05 13.14 Total Dipt. 5.19 10.65 12.36 34.79 62.98 Non-Dipterocarps Badling 0.33 - - - 0.33 Bitanghol 0.66 0.15 - - 0.81 Lipang-kalabaw 0.08 - - - 0.08 Nato 0.72 0.68 3.88 4.98 10.26 Tibig 0.17 - - - 0.17 Ulayan (Oak) 3.14 4.00 0.88 - 8.02 Other Non-Dipt. 13.49 13.24 8.48 25.34 60.54 Total Non-Dipt. 18.59 18.07 13.24 30.32 80.21 Total 23.78 28.72 25.59 65.10 143.19

88

Figure 30: Merchantable volume in open forests (Davao Oriental FRA)

89

6.5 Forest carbon stocks Table 25 provides a summary overview of the forest carbon stocks by carbon pool observed in the three sub-national FRAs.

For the analysis of the total carbon stocks, the biomass estimates, usually expressed in tonnes of dry matter (t d.m.), have to be converted to carbon (C) equivalent, since soil organic matter is only expressed in the latter unit. The conversion factor, called carbon fraction of dry matter, is different for living biomass (0.47 t C/t d.m) and dead organic matter (0.37 t C/t d.m).

The above ground biomass of trees has been estimated using the 2014 Chave et al. allometric equation, factoring in wood specific gravity. Textbox 20 compares the estimates with those obtained using the allometric equation developed by Brown (1997).

The closed forests inventoried in Eastern Samar and Davao Oriental feature the highest carbon stocks (338 t C/ha and 338 t C/ha), closely followed by the open forests in Eastern Samar (280 t C/ha). The lowest carbon stocks were observed in the Panay Mountain Range (220 t C/ha in closed forests, 175 t C/ha in open forests). Due to the nature of the soils and the low elevation, soil organic matter contributes least to the total carbon stock in Eastern Samar (13% to 16%, compared to 21% to 30% in the two other sites).

The approach of the data analysis studying the forest carbon stock is illustrated hereafter based on the FRA conducted in the Panay Mountain Range, including a tier 1 estimate of the carbon stocks of the non-key mangroves forest stratum which have not been inventoried on the ground.

Table 25: Forest carbon stocks

Carbon pool Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities) (2 municipalities) (37 municipalities) Closed Open Closed Open Closed Open forests forests forests forests forests forests (37 SUs) (44 SUs) (18 SUs) (102 SUs) (33 SUs) (53 SUs) [t C/ha] [t C/ha] [t C/ha] [t C/ha] [t C/ha] [t C/ha] Living AGB 187.06 116.71 205.14 160.91 112.52 82.83 biomass BGB 69.21 43.18 75.90 59.54 41.63 30.65 Total 256.27 159.89 281.04 220.45 154.15 113.47 Dead SDW 2.93 1.03 7.18 7.63 4.62 5.85 organic LDW 1.08 1.24 3.41 3.77 2.12 1.18 matter Litter 1.98 1.92 2.43 2.39 2.45 2.39 Total 5.99 4.19 13.02 13.79 9.20 9.42 Soil organic matter 71.62 58.61 44.00 45.88 56.27 52.25 Grand Total 333.89 222.70 338.06 280.13 219.62 175.14

* AGB of trees estimated using the 2014 Chave et al. allometric equation

90

Textbox 20: AGB Estimation of Trees As stated in Chapter 3.8.2, two different allometric equations, developed by Brown (1997) and more recently by Chave et al. (2014), respectively, may be used alternatively the estimate the above-ground biomass (AGB) of trees. The table below shows that the two equations yield quite similar estimates of the total AGB.

AGB Davao Oriental Eastern Samar Panay Mountain Range of trees (3 municipalities) (2 municipalities) (37 municipalities) estimated Closed Open Closed Open Closed Open with the forests forests forests forests forests forests equation of ... (37 SUs) (44 SUs) (18 SUs) (102 SUs) (33 SUs) (53 SUs) Species [t d.m./ha] [t d.m./ha] t d.m./ha] [t d.m./ha] [t d.m./ha] [t d.m./ha] Group Brown Dipt. 156.05 96.45 280.52 194.18 13.90 5.57 (1997) Non- 256.88 173.49 165.93 153.77 239.32 189.38 Dipt. Total 412.93 269.94 446.45 347.95 253.22 194.95 Chave Dipt. 147.92 92.02 251.91 180.93 12.42 4.47 et al. Non- 249.15 155.66 182.46 158.03 225.91 169.91 (2014) Dipt. Total 397.07 247.68 434.37 338.96 238.33 174.38

With reference to the AGB estimates calculated using the equation of Chave et al., the equation of Brown yields from 2.8% to 11.8% higher estimates of the total stocks. For all species, the equation of Brown universally uses the average wood gravity of tropical timber in Asia of 0.57 gr/cm³, while the equation of Chave et al. uses specific gravities. The wood specific gravity of Dipterocarps, for instance, ranges from 0.39 gr/cm³ (Almon) to 0.89 gr/cm³ (Yakal-Mabolo). Many common Shorea species (notably Bagtikan, Mayapis, Red Lauan, White Lauan, and Tangile) feature wood specific gravities of not more than 0.51 gr/cm³. For any estimation other than total AGB stock, the equation of Chave et al. should be preferred over the equation of Brown.

6.5.1 Forest carbon stocks of closed forests (Panay Mountain Range FRA) Table 26 summarizes the carbon stocks of the open forests, illustrated in Figure 31.

On average, the closed forests feature a living biomass of 328 t d.m./ha, dead organic matter of 24.8 t d.m./ha, composed of (i) 18.2 t d.m./ha of dead wood and (ii) 6.6 t d.m./ha of litter, plus 56.3 t C/ha of soil organic matter. The bulk of the carbon stock is in the above-ground biomass (51.2%), which is essentially composed of Non-Dipterocarps (94.3%).

91

Table 26: Carbon stocks of closed forests (Panay Mountain Range FRA)

Carbon pool Biomass/ha by diameter class Carbon/ha [5 - 40 cm] [40 cm - Total Total [t d.m./ha] [t d.m./ha] [t d.m./ha] [t C/ha] [%] Living biomass Above-ground biomass 120.15 119.25 239.40 112.52 51.2 Below-ground biomass 44.46 44.12 88.58 41.63 19.0 Total living biomass 164.61 163.37 327.98 154.15 70.2 Dead organic matter Standing dead wood 12.50 4.62 2.1 Lying dead wood 5.74 2.12 1.0 Litter 6.62 2.45 1.1 Total dead organic matter 24.85 9.20 4.2 Soil organic matter 56.27 25.6 Total 219.62 100.0

C/ha [t C/ha] C/ha [t C/ha]

Figure 31: Carbon stocks of closed forests (Panay Mountain Range FRA)

6.5.2 Forest carbon stocks of open forests (Panay Mountain Range FRA) Table 27 summarizes the carbon stocks of the open forests, illustrated in Figure 32.

On average, the open forests feature a living biomass of 241 t d.m./ha, dead organic matter of 25.5 t d.m./ha, composed of (i) 19.0 t d.m./ha of dead wood and (ii) 6.5 t d.m./ha of litter, plus 52.3 t C/ha of soil organic matter. Almost half of the carbon stock is in the above-ground biomass (47.3%), which is predominantly composed of Non-Dipterocarps (96.4%).

92

Table 27: Carbon stocks of open forests (Panay Mountain Range FRA)

Carbon pool Biomass/ha by diameter class Carbon/ha [5 - 40 cm] [40 cm - Total Total [t d.m./ha] [t d.m./ha] [t d.m./ha] [t C/ha] [%] Living biomass Above-ground biomass 95.15 81.07 176.22 82.83 47.3 Below-ground biomass 35.21 30.00 65.20 30.65 17.5 Total living biomass 130.36 111.07 241.43 113.47 64.8 Dead organic matter Standing dead wood 15.81 5.85 3.3 Lying dead wood 3.20 1.18 0.7 Litter 6.45 2.39 1.4 Total dead organic matter 25.46 9.42 5.4 Soil organic matter 52.25 29.8 Total 175.14 100.0

C/ha [t C/ha] C/ha [t C/ha]

Figure 32: Carbon stocks of open forests (Panay Mountain Range FRA)

6.5.3 Forest carbon stocks of mangroves (Panay Mountain Range FRA) Mangroves have not been inventoried on the ground. Since they represent only 0.33% of the total forest area in the project site on Panay Island, they are not a key stratum. Table 28 provides nevertheless a tier 1 estimate of their carbon stocks, illustrated in Figure 33, estimated using IPCC default data.

According to these tier 1 estimates, the mangroves feature a living biomass of 286 t d.m./ha, dead organic matter of 11.4 t C/ha, composed of (i) 10.7 t C/ha of dead wood and (ii) 0.7 t C/ha of litter, plus considerable 386.0 t C of soil organic matter. The bulk of the carbon stock is in the soil organic matter (73.3%).

93

Table 28: Carbon stocks of mangroves (Panay Mountain Range FRA)

Carbon pool Biomass/ha Carbon/ha Total Total [t d.m./ha] [t C/ha] [%] Living biomass Above-ground biomass 192.00 86.59 16.5 Below-ground biomass 94.08 42.43 8.0 Total living biomass 286.08 129.02 24.5 Dead organic matter Dead wood 10.70 2.0 Litter 0.70 0.1 Total dead organic matter 11.40 2.2 Soil organic matter 386.00 73.3 Total 526.42 100.0

C/ha [t C/ha]

Figure 33: Carbon stocks of mangroves (Panay Mountain Range FRA)

6.6 Uncertainty of the estimates The estimates of all variables of interest, such as N/ha, G/ha, V/ha, AGB/ha, DOM/ha, SOM/ha and C/ha, to cite the most important ones that are summarily presented in Chapter 6, are affected with uncertainties. An approach for the reporting of these uncertainties is presented hereafter, analyzing successively the following five main sources: • Statistical sampling error (see Chapter 6.6.1). • Representativeness of the sampling network (see Chapter 6.6.2). • Measurements errors (see Chapter 6.6.3). • Data entry errors (see Chapter 6.6.4). • Estimation design uncertainties (see Chapter 6.6.5).

94

Chapter 6.6.6 combines the different sources of uncertainty for the estimates of V/ha and AGB/ha to estimate an overall error budget.

6.6.1 Statistical sampling error Table 29 shows the coefficients of variation (s%) observed in the closed and open forests in the three project sites for N/ha, G/ha, V/ha and AGB/ha. As expected, they are systematically higher in open forests than in closed forests.

Table 29: Means and coefficients of variation of N/ha, G/ha, V/ha and AGB/ha

Variable Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities) (2 municipalities) (37 municipalities) Para- Closed Open Closed Open Closed Open meter forests forests forests forests forests forests (37 SUs) (44 SUs) (18 SUs) (102 SUs) (33 SUs) (53 SUs) N y̅ 742.65 521.05 1,633.1 1,221.8 1,393.57 1,171.29 [-/ha] s% 44.15% 58.13% 46.25% 34.93% 58.01% 58.46% G y̅ 35.24 22.26 42.92 35.22 27.64 21.28 [m²/ha] s% 39.47% 62.09% 33.72% 43.62% 50.50% 59.19% V y̅ 220.77 143.19 229.97 208.23 73.33 53.94 [m³/ha] s% 58.80% 86.85% 47.28% 74.96% 88.18% 132.21% AGB y̅ 398.00 248.32 436.46 342.37 239.40 176.22 [t d.m./ha] s% 51.01% 84.31% 39.77% 61.41% 57.31% 89.85%

Table 30 shows the statistical sampling error in terms of the margin of error (E%) at a confidence level of 90% for the main variables of interest observed in the tree sub-national FRAs.

The margin of error of the SOM/ha estimates appears to be very low. This is due to the fact that following the tier 1 estimate (see Chapter 3.1.3), only few soil types and climate regions were found in the areal sampling frames of the three FRAs: • In Davao Oriental low activity clays (LAC) and high activity clays (HAC) in tropical wet and in tropical montane climate, respectively, with four corresponding SOM stocks: 60 t C/ha (tropical wet LAC), 44 t C/ha (tropical wet HAC), 63 t C/ha (tropical montane LAC) and 88 t C/ha (tropical montane HAC. • In Eastern Samar LAC and HAC in tropical wet climate. • In the Panay Mountain Range Lac and HAC in tropical wet and LAC in tropical montane climate.

Hence, there is limited variation.

The margins of error are by definition depending on the coefficients of variation and the sample size. This explains the comparatively small margins of error achieved in the closed forests of Eastern Samar, with 102 sampling units. The estimates of the merchantable volumes are affected by the highest margins of error, due to the fact that merchantable heights are highly variable. The margins of error of the total forest carbon estimates range from ± 8.05% for the open forests in Eastern Samar to ± 15.39% for the open forests in Davao Oriental.

Based on t- tests, the means of N/ha, G/ha, V/ha and AGB/ha in closed and open forests are significantly different in each of the three project sites, except for V/ha in Eastern Samar and for N/ha and V/ha in the Panay Mountain Range.

95

Table 30: Statistical sampling errors of the main variables of interest

Variable Davao Oriental Eastern Samar Panay Mountain Range (3 municipalities) (2 municipalities) (37 municipalities) Para- Closed Open Closed Open Closed Open meter forests forests forests forests forests forests (37 SUs) (44 SUs) (18 SUs) (102 SUs) (33 SUs) (53 SUs) N y̅ 742.65 521.05 1,633.1 1,221.8 1,393.57 1,171.29 [-/ha] E%* ± 12.25% ± 14.73% ± 18.96% ± 5.74% ± 17.11% ± 13.45% G y̅ 35.24 22.26 42.92 35.22 27.64 21.28 [m²/ha] E%* ± 10.95% ± 15.74% ± 13.92% ± 7.17% ± 14.89% ± 13.62% V y̅ 220.77 143.19 229.97 208.23 73.33 53.94 [m³/ha] E%* ± 16.32% ± 22.01% ± 19.39% ± 12.32% ± 26.00% ± 30.41% AGB y̅ 398.00 248.32 436.46 342.37 239.40 176.22 [t d.m./ha] E%* ± 14.16% ± 21.37% ± 16.31% ± 10.09% ± 16.90% ± 20.67% BGB y̅ 147.26 91.88 161.49 126.68 88.58 65.20 [t d.m./ha] E%* ± 14.16% ± 21.37% ± 16.31% ± 10.09% ±16.90% ± 20.67% LB y̅ 256.27 159.89 281.04 220.45 154.15 113.47 [t C/ha] E%* ± 14.16% ± 21.37% ± 16.31% ± 10.09% ± 16.90% ± 20.67% SDW y̅ 2.93 1.03 7.18 7.63 4.62 5.85 [t C/ha] E%* ± 86.61% ± 97.23% ± 53.88% ± 41.92% ± 42.51% ± 56.16% LDW y̅ 1.08 1.24 3.41 3.77 2.12 1.18 [t C/ha] E%* ± 74.68% ± 65.38% ± 75.77% ± 42.62% ± 40.51% ± 60.21% Litter y̅ 1.98 1.92 2.43 2.39 2.45 2.39 [t C/ha] E%* ± 13.87% ± 14.17% ± 22.23% ± 7.93% ± 16.56% ± 11.11% DOM y̅ 5.99 4.19 13.02 13.79 9.20 9.42 [t C/ha] E%* ± 44.68% ± 31.51% ± 35.97% ± 26.00% ± 23.94% ± 35.80% SOM y̅ 71.62 58.61 44.00 45.88 56.27 52.25 [t C/ha] E%* ± 5.17% ± 4.10% ± 0.00% ± 1.86% ± 4.05% ± 3.76% Total C y̅ 333.89 222.70 338.06 280.13 219.62 175.14 [t C/ha] E%* ± 10.95% ± 15.39% ± 13.63% ± 8.05% ± 11.95% ± 13.58% * 90% confidence level

6.6.2 Representativeness of the sampling network The design of the sampling network (see Chapter 3.5) has been made in accordance with the statistical theory that the sample points shall well present the overall population. In each project site, the 200 sample points initially targeted were distributed over the whole areal sampling frames. Because of operational difficulties (remoteness and limited accessibility of the sample points, steep and high mountains, unfavorable weather conditions, peace and order situation, etc.), it was not possible to measure all the allocated sampling units within the available time and budget. It cannot be excluded that the failure to measure all allocated sampling units affects the representativeness of the sampling network. An uncertainty of an order of magnitude of 5% in Davao Oriental, of 2.5% in Eastern Samar and of 10% in the Panay Mountain Range may conservatively be assumed.

6.6.3 Measurement errors The impact of the measurement errors has been evaluated through the re-measurement of 5% to 10% of the sampling units (see Chapter 4.3). Where sufficient sampling units have been re-measured for the appreciation of the deviations between initial and control measurements (in Eastern Samar), the uncertainties related to measurement errors (in terms of the RMSD) turn out to be of a similar or lesser magnitude than the statistical sampling errors.

96

6.6.4 Data entry errors The effect of the data entry errors has been evaluated through the comparison of the original field data and the stored data of 10% of the sampling units (see Chapter 5.4). The estimates of the variables of interest are affected by limited uncertainties most often not exceeding 3% (in terms of the RMSD) related to data entry errors.

6.6.5 Estimation design uncertainties Except for N/ha and G/ha, where no allometric models, volume equations, wood specific gravities nor conversion and/or extrapolation factors are used, the estimates of all other variables of interest are affected by uncertainties due to the lack of fit of the estimation design (see Chapter 3.8) used.

The uncertainty arising from the use of the regional volume equations for Dipterocarps and Non- Dipterocarps (see Chapter 3.8.1) for the estimation of V/ha is not documented. It may conservatively be estimated to be of an order of magnitude of 15%.

According to the authors, the uncertainty arising from the use of the allometric equation developed by Chave et al. (2014) (see Chapter 3.8.2.2) for the estimation of AGB/ha is of the order of magnitude of 10%.

The uncertainties of the other metrics used to estimate BGB/ha (the root to shoot ratio, see Chapter 3.8.6), SDW/ha (the biomass conversion and expansion factor [BCEFs], see Chapter 3.8.7), LDW/ha (see Chapter 3.8.8), LI/ha (see Chapter 3.8.9), and to convert the biomass to carbon equivalent (carbon fraction [CF] of dry matter, see Chapter 3.8.10) are difficult to evaluate.

6.6.6 Overall error budget Table 31 and Table 32 show the overall error budget of the estimates of V/ha and AGB/ha, respectively.

Unless height measurements are involved (for the estimation of V/ha), the largest uncertainties pertain to the statistical sampling error.

Measurement errors can be reduced through quality assurance measures, such as the regular supervision of the field data collection by a supervisor, and the continuous updating and enhancement of the field inventory manual based on issues and ambiguities reported from the field. Rotating the assistants associated to a team leader may also help to standardize the measurements and observations. The re-measurement of 10% of the sampling units for quality control purposes should be performed as early as possible after the initial measurements, so that feedback could improve the work of the inventory teams.

The statistical sampling error can be reduced by increasing the sample size. However, one has to consider that to halve the statistical sampling error, four times more sampling units must be measured, since the sampling error is inversely proportional to the square root of the number of sampling units.

97

Table 31: Overall error budget estimating V/ha

Source of uncertainty Stratum Davao Oriental Eastern Samar Panay (3 municipalities) (2 municipalities) Mountain Range (37 municipalities) Statistical Closed forests ± 16.3% ± 19.4% ± 26.0% sampling error* Open forests ± 22.0% ± 12.3% ± 30.4% Representativeness of Closed & ± 5.0% ± 2.5% ± 10.0% the sampling network open forests Measurement Closed & ± 33.0% ± 27.4% ± 42.9% errors open forests Data entry Closed & ± 7.2% ± 0.1% ± 8.3% errors open forests Estimation design Closed & ± 15.0% ± 15.0% ± 15.0% uncertainties open forests Figures in italic are based on an insufficient number of observations and are not reliable * 90% confidence level

Table 32: Overall error budget estimating AGB/ha

Source of uncertainty Stratum Davao Oriental Eastern Samar Panay (3 municipalities) (2 municipalities) Mountain Range (37 municipalities) Statistical Closed forests ± 14.2% ± 16.3% ± 16.9% sampling error* Open forests ± 21.4% ± 10.1% ± 20.7% Representativeness of Closed & ± 5.0% ± 2.5% ± 10.0% the sampling network open forests Measurement Closed & ± 21.6% ± 10.2% ± 13.2% errors open forests Data entry Closed & ± 6.8% ± 0.1% ± 4.2% errors open forests Estimation design Closed & ± 10.0% ± 10.0% ± 10.0% uncertainties open forests

* 90% confidence level

98

7. Considerations for up-scaling The FRA methodology and the associated software tool have been developed and progressively enhanced in close cooperation with the central, regional and local partners from DENR and the involved LGUs. They primarily respond to the specific objectives and scope of the Project. Nevertheless, they bear the potential for upscaling, with adaptations and enhancements according to the potentially broader objectives and wider scope. Considerations and recommendations for such an up-scaling are presented in the following chapters, starting with the objectives (Chapter 7.1), followed by the inventory design (Chapter 7.2), the field data collection (Chapter 7.3), the data processing (Chapter 7.4) and the data analysis (Chapter 7.5).

7.1 Comprehensive specification of the objectives FRAs acquire data of variables of interest capturing the characteristics of the forest ecosystems and of factors and/or circumstances bearing a potential to explain their current condition or to impact on their development. To be of use for decision making, the data, consisting typically of maps and descriptive statistics, must be translated through analysis, condensation and interpretation into information.

A proper starting point for designing a FRA (or to upscale the field-tested FRA methodology described so far, to account for specific circumstances, needs and/or scales) is a comprehensive specification of the objectives, comprising the following elements: • General objectives, typically one or several of the following: o Assessment of the current status. o Monitoring of change. o Hypothesis testing of a particular cause - effect relationship. • Areal sampling frame, i.e. the geographical extent to be covered by the inventory (a particular jurisdiction, watershed or forest type, etc., or the entire country). • Scope and content, i.e. the elements to be assessed (trees, bamboos, palms, rattan, tree ferns, other woody biomass, dead wood, litter, etc., inside forests and/or also outside forests). • Variables of interest to be estimated (species abundance and diversity, density, basal area, volume, above-ground biomass, forest carbon stock, etc.), including ancillary variables needed for the following purposes: o Stratification (e.g. by forest types). o Disaggregation of the results (e.g. by jurisdictions, species groups, etc.). o Estimation of the variables of interest, if these cannot be estimated directly (e.g. the AGB of trees, requiring other variables depending on the allometric equation used). • Targeted accuracy (for maps) in terms of producer's, user's and/or overall accuracy (see Textbox 21). • Targeted precision (for point estimates from sampling) in terms of the margin of error (E%) at a specific confidence level (typically 90%, 95% or 99%) (see Textbox 11). • Definition of terms and concepts used (forest, tree, volume, forest carbon stock, etc.).

The two following tables propose and illustrate a framework for the specification of the variables of interest (Table 33) together with ancillary variables (Table 34), including their scope and content as well as their targeted accuracy and/or precision.

99

Table 33: Framework for the definition of variables of interest (example)

# Variable of interest Scope & content Form(s)* Targeted Source, accuracy** or precision*** estimation design A01 Land cover Philippines map producer's and user's satellite data forest types and continuous accuracy of forests ≥ 90% other land cover or attribute uses A02 Density (N) continuous: field sampling • total • avg. per ha • s% • E% A03 Species richness count field sampling A04 Berger-Parker index continuous field sampling trees, calculated based on N bamboos, A05 Margalef index palms, continuous field sampling A06 Shannon H' index rattan continuous field sampling tree ferns calculated based on species richness and N with Dref ≥ 5 cm A07 Shannon E index continuous field sampling inside forests calculated based on species richness and N A08 Simpson index continuous field sampling based on N A09 Basal area (G) [m²] continuous: • E% ≤ ± 5% (p < 0.05) field sampling • total nationwide calculated based on Dref • avg. per ha • E% ≤ ± 10% (p < 0.05) • s% per Region • E% A10 Merch. volume (V) [m³] trees continuous: field sampling with Dref ≥ 5 cm • total estimated based on Dref and merch. height inside forests • avg. per ha using the Philippine regional volume equations • s% • E%

100

# Variable of interest Scope & content Form(s)* Targeted Source, accuracy** or precision*** estimation design A11 Above-ground biomass (AGB) trees continuous: • E% ≤ ± 5% (p < 0.05) field sampling [t d.m.] with Dref ≥ 5 cm • total nationwide AGB estimated based on Dref, SU coordinates and inside forests • avg. per ha • E% ≤ ± 10% (p < 0.05) wood specific gravity using the 2014 Chave et al. allometric equation • s% per Region bamboos • E% field sampling with Dref ≥ 5 cm AGB estimated based on Dref inside forests using the 1998 Priyadarsini allometric equation palms field sampling with Dref ≥ 5 cm AGB estimated based on Dref inside forests using the 2013 Goodman et al. allometric equation A12 Below-ground biomass (BGB) trees continuous: field sampling [t d.m.] bamboos • total estimated based on AGB and AGB to BGB ratio palms • avg. per ha with Dref ≥ 5 cm • s% inside forests • E% A13 Standing dead wood biomass standing dead wood continuous: field sampling (SDW) [t d.m.] with Dref ≥ 5 cm • total estimated based on Dref, merch. height and V/ha inside forests • avg. per ha using the Philippine regional volume equations and the IPCC 2006 biomass conversion and • s% expansion factors • E% A14 Lying dead wood biomass (LDW] lying dead wood continuous: field sampling [t d.m.] down to 5 cm • total estimated based on mid-diameter, length and the diameter • avg. per ha average wood density for Asia according to Reyes et al. (1992) inside forests • s% • E% A15 Litter biomass [t. d.m.] litter continuous: field sampling inside forests • total estimated based on average ground coverage and • avg. per ha average depth of litter using the average density of litter according to Chojnacky et al. (2009) • s% • E% A16 Soil organic matter (SOM) [t C] SOM continuous: BSWM soil map inside forests • total

101

# Variable of interest Scope & content Form(s)* Targeted Source, accuracy** or precision*** estimation design • avg. per ha estimated based on FAO soil class and climate region • s% using the IPCC 2006 tier 1 data • E% A17 Forest carbon stock [t C] Forest carbon stock continuous: sum of AGB + BGB + SDW + LDW + litter + SOM inside forests • total • avg. per ha • s% • E% etc.

* variables may have different forms: map (geodata), continuous (if the values are obtained through measurement or calculation, e.g. DBH, volume, etc.), count (if the values are obtained through counting, e.g. number of species, number of regeneration, etc.) or attribute (if the values are obtained using a classification, e.g. species, live or dead, terrain class, etc.)

** accuracy applies to maps

*** precision applies to point estimates from sampling

102

Table 34: Framework for the definition of ancillary variables (example)

# Ancillary variable Scope & Form(s) Purpose(s) Source content Site variables B01 Geographic sample points continuous estimation of AGB of field sampling coordinates trees disaggregation of results B02 Regions, Provinces, Philippines map disaggregation of http://gadm.org/ Municipalities, Cities continuous results and Barangays attribute B03 Land classification Philippines map disaggregation of DENR continuous results attribute B04 Land tenure Philippine map disaggregation of DENR forests continuous results attribute B05 Forest types Philippine map stratification NAMRIA forests continuous attribute B06 Elevation sample points continuous disaggregation of field sampling results B07 Slope sample points continuous disaggregation of field sampling results B08 Slope orientation sample points continuous disaggregation of field sampling results B09 Terrain sample points attribute disaggregation of field sampling results B10 Land cover sample points attribute disaggregation of field sampling results B11 Forest type sample points attribute disaggregation of field sampling results B12 Tree crown cover sample points attribute disaggregation of field sampling results B13 Average ground litter inside continuous estimation of litter field sampling coverage of litter forests biomass B14 Average depth of litter litter inside continuous estimation of litter field sampling forests biomass B15 Soil type SOM inside map estimation of SOM BSWM forests continuous disaggregation of attribute results etc. Tree, bamboo, palm, rattan, tree fern, standing dead wood and lying dead wood variables C01 Species attribute calculation of species field sampling richness trees, calculation of diversity bamboos, indices palms, estimation of V rattan disaggregation of tree ferns results C02 Official common attribute disaggregation of FMB name results

103

# Ancillary variable Scope & Form(s) Purpose(s) Source content C03 Scientific species, FMB genus and family validated referring to names http://www.theplantli st.org/ C04 IUCN red list status IUCN C05 Non-timber uses FMB C06 Tree species groups trees attribute disaggregation of FMB results C07 Wood specific gravity continuous estimation of AGB Zanne et al. (2009) Reyes et al. (1992) C08 Dref [cm] trees continuous calculation of G field sampling bamboos estimation of V, AGB, palms BGB, SDW rattan disaggregation of tree ferns results standing dead wood with Dref ≥ 5 cm inside forests C09 Merch. height (H) [m] trees continuous estimation of V of trees field sampling with estimation of V and Dref ≥ 20 cm AGB of standing dead standing dead wood wood with Dref ≥ 5 cm inside forests C10 Dref [cm] lying dead wood continuous estimation of LDW field sampling C11 Length [m] etc.

104

Textbox 21: Map Accuracy The accuracy of a map is commonly assessed with the help of a confusion matrix, which cross-tabulates for a sample the classes found in the map against the classes determined from a more reliable source, such as ground truth or higher resolution imagery. As shown in the example below, the columns hold the classification results from the map, while the rows hold the classification results from the reference.

Map Forest Bush Crop Urban Bare Water Un- UA [%] classified Forest 440 40 0 0 30 10 10 83

Bush 20 220 0 0 40 10 20 71 Crop 10 10 210 10 50 10 60 58 Urban 20 0 20 240 100 10 40 56 or reference

Ground Ground truth Bare 0 0 10 10 230 0 10 88 Water 0 20 0 0 0 240 10 89 PA [%] 90 76 88 92 51 86

Source: Hengl (2008)

The user's accuracy (UA) corresponds for a particular class (with reference to the map) to the proportion correctly classified (e.g. 83% for forest). It accounts for the error of commission. The producer's accuracy (PA) corresponds for a particular class (with reference to the truth) to the proportion correctly classified (e.g. 90% for forest). It accounts for the error of omission. The overall accuracy corresponds for all classes combined to the proportion correctly classified (in the example 73%).

FAO's 2012 "Manual for integrated field data collection" may serve as a source of inspiration for the selection and specification of variables of interest and ancillary variables, with caution (see Textbox 22). Apart from variables addressing administrative / organizational aspects (team composition, date of measurements, etc.), the manual proposes a breathtaking list of tree-, stand- and particularly site- related variables (the variables proposed in the framework of an "Integrated Land Use Assessment (ILUA)" are shown in italics; the variables for assessments outside forests and other wooded lands are not listed): • Trees: o Species (preferably scientific or official common name, exceptionally local name including the local dialect). o Diameter at breast height or above buttress. o Height of diameter measurement (if different from 1.30 m). o Total and bole heights. o For trees stem quality (3 classes), crown condition (5 classes), overall tree condition (5 classes), causative agents (diseases, insects, animals, etc.) (8 classes, multiple choice possible). o For stumps year(s) since last cut (5 classes). o For standing dead wood decomposition status (5 classes). • Small trees: Stem count per species. • Shrubs and bushes: Stem count, average stem diameter at 0.5 m, and average height per species. • Lying dead wood: Diameter and decomposition status (2 classes). • Litter: Depth and composition (2 classes). • Sample point location and accessibility: o Coordinates and coordinate system. o Administrative location.

105

o Global ecological zone (20 classes) and national or regional ecological zone. o Altitude. o Accessibility (6 classes). • Land use and/or vegetation cover: o Land use or cover. o For each land user or cover: Rate of conversion (4 classes) and land use or cover after conversion. o Tree canopy cover (6 classes and tree density expected in 5 years (3 classes). o Trees outside forests (TOF) distribution (5 classes) and trees density expected within 5 years (3 classes). o Shrub cover (6 classes) and average shrub height. o Herbaceous cover (6 classes). o Plant and crop residues cover (6 classes). • Protection status (10 classes). • Tenure: Land ownership (9 classes) and management agreement (7 classes). • Biophysical site: o Slope and slope orientation. o Relief (12 classes). o Soil: Soil type, soil surface condition (3 classes), organic layer thickness, topsoil and subsoil depths, texture (12 classes) and color (10 classes), coarse fragments (4 classes), pH, presence of a hardpan (3 classes), soil structural condition (3 classes), soil porosity (3 classes), topsoil color difference (3 classes), soil drainage (5 classes). o Drainage: Waterlogging (6 classes) and wetland filtering capacity (4 classes). • Socio-economic site: o Sedentary population: Number of households, average household size, number of people, adult literacy rate, main ethnic group, years since settlement (7 classes), population dynamics (5 classes), main and secondary activities (15 classes). o Nomadic and transhumant population: Number of households, average household size, main ethnic group, period of stay. o Major historical events that have affected the local people and land use (18 classes). o Proximity to infrastructures and services: Distances to closest all-weather road, seasonal road, settlement, health center, veterinary services, school, food market, input market. • Forest and other wooded land management: Stand origin (4 classes), vertical stand structure (5 classes), forest ownership (9 classes), management plan (3 classes), human disturbances impact (4 classes) and types (7 classes), timber harvesting system (7 classes), stumps removal (yes or no), branches and tops removal (yes or no), silvicultural practices (12 classes, multiple choice possible), logging technology (7 classes). • Environmental problems: Environmental problems (23 classes, multiple choice possible), problem severity (3 classes) and its trend (4 classes), soil erosion type (14 classes), fire evidence (3 classes, area, type (4 classes) and causes (11 classes), wildlife disturbances (4 classes), grazing activity (presence or absence), grazing overall quality (4 classes) and quality trend (5 classes). • Products and services: o Products: Category (29 classes, multiple choice possible) and importance per category (3 classes). o For each species used: Species name, importance (3 classes), part(s) used (12 classes), commercial end-use (7 classes), conflicts (3 classes), demand and supply trends (5 classes), harvest period, harvest frequency (7 classes) harvest trend (5 classes), harvest change reasons (9 classes), market price and unit. o For each product category and user group (4 classes): Ranking (3 classes), user rights (6 classes), main destination of product (5 classes), organization level of harvest (3 classes), gender balance (5 classes), child participation (5 classes), legislation

106

awareness and compliance (3 classes), incentives awareness and application (3 classes). o Services: Category (16 classes, multiple choice possible). o For each service category importance (3 classes), legislation awareness and compliance (3 classes), incentives awareness and application (3 classes). • Biodiversity indicators: o Insect pests, diseases and invasive species: Category (6 classes), species, category affected (7 classes), severity (3 classes). o Threatened or extinct species and varieties: Category (4 classes), species, status (2 classes). o Wildlife abundance (3 classes).

Quite evidently, the costs of a FRA increase with the complexity of its objectives. Typically, budget and time constraints lead to an iterative revision of the objectives and the ensuing inventory design (see Figure 1).

Recommendations The following recommendations concerning the definition of FRA objectives can be drawn from experience: • The objectives ("what") should come first, before caring for the inventory design ("how"). • The variables of interest should be defined together with the targeted precision for those of the highest interest or relevance. The density and the basal area are good candidates, since both are not affected by uncertainties from allometric equations. The basal area is moreover closely correlated to volume and above-ground biomass. • The targeted precision should not only be defined for the entire areal sampling frame. For subsets that are of interest, e.g. mangroves, plantations, pine forests, etc., specific targeted precisions should be defined. This may have an impact on the distribution of the sample points, or even lead to specific inventory designs, whose results will be combined at the global level. • Priority should be given to variables that can be observed (presence or absence), counted or measured objectively and reproducibly. • The observations, enumerations and measurements in the field should focus on essential variables. Numerous, complex, lengthy and barely feasible observations or measurements cause stress and contribute to increase the errors of all measurements. • Field sampling should be spared from the collection of data that can be obtained with reasonable precision from other sources (e.g. altitude, slope and slope orientation from digital elevation models (DEM); land classification, tenure and management agreements from official records; land cover or use and tree canopy cover from high-resolution imagery; etc.). • Data that are not intimately tied to the sampling unit, such as socio-economic data, should be merged with the inventory results at a higher level (stratum).

107

Textbox 22: FAO's Variable Set for National Forest Monitoring and Assessment Since 2000, FAO's National Forest Resources Monitoring and Assessment (NFMA) program has supported several countries, including Guatemala, Honduras, Kenya, Kyrgyzstan, Lebanon, Nicaragua, the Philippines, Zambia and Uruguay in the implementation of national forest inventories, using a standard methodology covering not only forest resources on forest lands, but also trees outside forests (TOF). In 2005, the methodology was broadened to cover other land uses and natural resources in the assessment, such as crops, livestock, soils, water and biodiversity features. Integrating the assessment and monitoring across forest, agriculture and other related sectors, strives to achieve a better understanding of ecosystem services and functions, and to create possibilities for analyzing land management as a whole. The new approach, termed "Integrated Land Use Assessment (ILUA)", was implemented in Kenya and Zambia.

Tomppo and Andersson (2008) undertook on behalf of FAO a scientific examination of the methodological aspects of its NFMA program. Some of their main recommendations concerning the data to be collected in the field were to "make better use of existing data for both biophysical and socio-economic variables", and to "reinforce existing quality control systems for data collection for all variables, but especially for socioeconomic and institutional data since these rely largely on indirect measurement techniques". They noted that "it is difficult to determine the spatial location of the forest use described by users in interviews", and were concerned about the measurement errors particularly of the data gained through interviews: "This type of error refers to participants' limitations related to 'memory, understanding, and willingness to respond truthfully to questions, and as a consequence distort the quality of results' (Niemi, 1993). Measurement errors in the NFMA interviews refer to difficulties to obtain valid and truthful information about people's relationships to trees and forest resources. Consider the following example: If local users do not enjoy undisputed and officially recognized property rights they are likely to be reluctant to reveal the full array of products and services that they derive from these resources. But even if users are perfectly legitimate and legal users of the resources, they may also be reluctant to provide accurate information to strangers. After all, even legal users don't have anything to gain - only lose - from revealing information about their use". "Non-responses in surveys represent another type of measurement problem that is very difficult to deal with. This may be the most difficult source of measurement error to deal with since it is hard to learn much about those individuals who did not participate in interviews (Fowler, 1993). In the NFMA interviews, field crews are under a great deal of pressure to finish their fieldwork on time. If one of the pre-selected interviewees is not available or refuses to be interviewed, field personnel will select an alternate interviewee without necessarily recording that the person who was preselected could not participate. Consequently, the NFMA interview sample is likely to be systematically biased towards those individuals who have more time to talk, and who have less to lose from divulging information about their forest use". "Another potential source of inaccuracy is the respondent's hidden agenda with regards to local forest use. Interviewees may deliberately conceal or distort information even if they do not have an apparent reason to do so". "All these measurement problems affect the accuracy of the sample estimates. The crux of it is that unlike sampling error, it is not possible to calculate the precise size of the measurement error".

Perhaps, they were too timid to challenge the measurement error of variables to be assessed by the trained inventory teams through ocular inspection. A number of those can hardly be collected in a reproducible way, such as the tree canopy, shrub and herbaceous covers according to 6 classes (no trees, < 5%, 5 - 10%, 10 - 40%, 40 - 70%, > 70%), the trend in tree density expected within 5 years according to 3 classes (decreasing, stable, increasing), etc.

7.2 Efficient inventory design As notably recommended by Tomppo et al. (2008) at the occasion of the scientific examination of the methodological aspects of FAOs NFMA program (see Textbox 23), the inventory design should take advantage of existing information. Particularly remote sensing data (satellite images) or maps can contribute to the efficiency of field sampling through stratification, provided that they are up-to-date, and that the classification is accurate. But also, research on the availability of other data, notably of ancillary variables, helps to reduce the data collection effort in the field. Such information may comprise the following: • Land and forest cover, notably from NAMRIA. • Digital elevation models, notably from the shuttle radar topography mission (SRTM, see https://lta.cr.usgs.gov/SRTM1Arc). • Soil. • Precipitation. • Geographical-political subdivisions.

108

• Land classification. • Land titles. • Forest land tenures and management agreements. • Mining tenements. • Critical watersheds. • Protected areas. • Hazard. • Accepted scientific names of plant species, notably exploring The Plant List (http://www.theplantlist.org/). • Protection status of plant species, notably from IUCN (see http://www.iucnredlist.org/) and DENR. • Allometric equations, notably exploring GlobAllomeTree (see http://www.globallometree.org/ and Textbox 10). • Wood specific density, notably exploring GlobAllomeTree (see http://www.globallometree.org/ and Textbox 10), Zanne AE et al. 2009 (see http://dx.doi.org/10.5061/dryad.234/1). • Timber and non-timber uses of species. • etc.

The fundamental inventory method will always be some kind of probabilistic (statistical) sampling, since a complete enumeration is not feasible.

If specific precision targets are set for different strata (such as closed broad-leafed forests, open broad-leafed forests, pine forests, mangroves, forest plantations, and trees outside forests), each stratum will likely be inventoried using a specific sampling unit design and distribution.

The sampling unit design is intimately related to the objectives and the scope. For trees, bamboos, palms, rattan, tree ferns and standing dead wood, fixed area plots are best suitable. Transects can be used for land area estimates, though the results will be affected by considerable sampling errors, so that the resulting area statistics cannot compete with area statistics derived from a wall-to-wall mapping. Clusters augment the representativity of sampling units.

To achieve the targeted precision(s) efficiently, relatively small sampling units should be adopted. From a purely statistical point of view, for one given sampling intensity (ratio of the aggregated sample plot area to the area of the areal sampling frame), the largest number of the smallest sampling units would be most efficient. In practice, a reasonable balance between the "unproductive" time invested in accessing the sample points and the "productive" time invested for the measurement and/or observation of the sampling units must be found.

109

Textbox 23: FAO's Sampling Unit Design for National Forest Monitoring and Assessment FAO's National Forest Resources Monitoring and Assessment (NFMA) methodology was notably used from November 2002 to July 2004 to conduct the last forest resources assessment (FRA) in the Philippines. 395 sampling units (called tracts) measuring 1 km x 1 km were distributed systematically throughout the country using a 15 ' longitude x 15 ' latitude grid. Each tract consisted of a cluster of four rectangular plots measuring 20 m x 250 m to sample trees with Dref ≥ 20 cm in the plot sections ocularly assessed as forest, and with Dref ≥ 10 cm in the plot sections ocularly appraised as other wooded land, other lands or inland water. Each rectangular plot included up to three nested sub-plots, one rectangular measuring 10 m x 20 m to sample trees with 10 cm ≤ Dref < 20 cm, bamboo, palms and rattan, and one circular with a radius of 3.99 m to sample tree regeneration, provided that the sub-plots were located on plot sections ocularly assessed as forest.

Source: Branthomme (2002)

The sampling unit is very large, likely in order to yield reasonably reliable area estimates of the ocularly assessed land use classes. If land use areas are not to be estimated through field sampling, but can be measured based on a wall-to-wall mapping, it would be more efficient to use smaller sampling units, to increase the sample size, which reduces the statistical sampling error of the estimates of most variables of interest (density, basal area, volume, etc.).

Tomppo and Andersson (2008) undertook on behalf of FAO a scientific examination of the methodological aspects of its NFMA program. One of their main recommendations was to "explore and experiment with alternative sampling designs and plot layouts". According to the authors, "in large area inventories, an efficient plot is usually rather small". "The basic alternatives are a set of concentric plots". "One of our important recommendations is that the program should seriously reconsider promoting a blue-print sampling design because what is cost-effective data collection in one country will not be the same in all countries". "We find that the current plot layout is time consuming and similar trees are measured again and again. Our preliminary analysis suggests that one would be able to generate more or less the same level of precision for estimates of biomass volumes even if one measured one third as many trees as is currently done in the NFMA".

110

Recommendations The following recommendations concerning the design of a FRA can be drawn from experience: • The design should focus on efficiency. • The inventory should adopt stratified sampling. • For the sampling of trees and standing dead wood, nested circular plots should be used, considering their following advantages: o Circular plots have the least number of borderline trees (since they feature the shortest perimeter of all geometric shapes); o As long as the radius is small (≤ 15 m), the boundary of the plots, more particularly the borderline trees can easily be checked with a distance tape or a rangefinder. o Two to three concentric (nested) sub-plots should be used to achieve a balanced sampling of trees or standing dead wood in all diameter classes. o The polar coordinates of the sampled trees and standing dead wood (azimuth and distance to the plot center) can easily be determined. They allow to plot the location of the sampled trees and standing dead wood in a sketch map, and to track their identity over the measurement cycles, which is essential if the sampling units are conceived to be permanent. • In natural forests with typically larger contiguous areas such as natural forests, clusters should be used to increase the representativity of the sampling units. • The cluster size should be chosen so that the observations and/or measurements can be achieved by one field team or, for very remote sampling units needing more time to access, two field teams splitting within the sampling unit. Clusters of a size requiring the field team(s) to return once more to complete the observations and measurements are not efficient. • For the inventory of trees outside forests, the efficiency of a two-phase sampling (first phase sample based on high-resolution remote sensing data to quantify woodlots, line plantations and single trees; second phase field sub-sample to collect average characteristics of the woodlots, line-plantations and trees observed or measured during the first phase) should be studied. The stocking levels of trees outside forests are highly variable, and it would take a huge sampling effort to achieve precisions comparable to those in forests. • For monitoring purposes, permanent sampling units should be used. Permanent sampling units are only efficient provided that the identity of the trees and standing dead wood sampled can be tracked over time. Without a one-to-one identification of the sampled trees and standing dead wood at the time of subsequent re-measurements, the permanent sampling units lose their efficiency to detect changes. Indeed, only means at different times could be compared, each affected by a statistical sampling error, so that it takes considerable changes to become statistically significant (based on a t-test). • To monitor trees inside and outside forests at a regional or national scale with stratified sampling, the permanent sampling units should be distributed systematically. • With specific precision targets per stratum, the size of the grid for the systematic distribution of the sampling units will differ from stratum to stratum. • The inventory methodology should be comprehensively documented from a conceptor's point of view. Apart from this documentation, a field inventory manual focusing on the operational aspects of the implementation of the inventory methodology must be made available to the field teams.

7.3 Cautious field data collection Field data collection is the most expensive phase in FRA. It is also the most critical phase, actually the only that generates new information otherwise not available. Shortcomings or errors committed in this phase are hardly recoverable.

Data would likely be collected by field teams coming from the Regional Offices. FMB would have to provide the training, the coordination and support, and to lead the quality control.

111

Coordination with local authorities and communities and with the Armed Forces of the Philippines (AFP) in preparation of the field work could be done by Community Environment and Natural Resources Offices (CENROs).

Planning for the location of inventory camps and the best access to the sample points remains a challenge. Persons familiar with the local situation should be consulted.

Recommendations The following recommendations concerning the field data collection can be drawn from experience: • Field data collection should focus on data quality. • The field team members should be capable to identify the majority of the species sampled. This requires training in different regions according to the prevailing species composition. • The field team members must be trained, and follow SOPs described in a field inventory manual to standardize the observations and measurements, and keep measurement errors low. • The field inventory manual should be considered as an evolving document, to be enhanced and/or amended based on the feedback from the field teams for situations initially not covered or procedures not clearly described. • The last 15 m to 10 m to the sample points should be covered by compass and horizontal distance measurement (referring to the azimuth / bearing and distance to the sample point displayed by the GPS receiver once the distance to the destination is less than 15 m) using a distance tape or a rangefinder, in order to prevent bias (preference for easily accessible areas) when closing in on the sample point). With clusters, the plot centers should be located from the sample point by compass and horizontal distance measurement. • Geotagged photographs should be taken, showing the field team members at the sample point. • The sample point and the centers of the plots should be permanently marked using iron rods (of at least 1 cm diameter and 50 cm length, preferably magnetized), forced at least 4/5 of its length into the ground. • Breast height should systematically be materialized using a reference pole, so that the DBH measurements are reasonably replicable. • The field teams should be equipped with ruggedized laser rangefinders, backed-up by opto- mechanical clinometers if the rangefinders fail. • Visibly marked trees in permanent sampling units bear the risk that the area receives a special treatment (conservation or sabotage). • The use of passive radio-frequency identification (RFID) tags (see Textbox 24) for the discrete marking of at least three sampled trees per plot should be studied. This could facilitate the retrieval of the plots. • Quality control through the independent re-measurement of 10 % of the sampling units chosen at random should be done successively and shortly after the initial measurements, to provide a feedback to the field teams. • The field team members need to be motivated, so that they collect the data with utmost care. It is recommended to associate them as well as the quality control teams in the data analysis, to keep the awareness of the "garbage in - garbage out (GIGO)" principle constantly high.

112

Textbox 24: Radio Frequency Identification Radio frequency identification (RFID) is a generic term for technologies using radio waves to transfer data from a tag to a reader for automatic identification and/or tracking. The tag is usually attached to an object to be identified. Radio transmissions are send by the reader to query the tag and by the tag to return an answer, generally containing identifying information. RFID tags can be active or passive, depending on whether or not they carry a battery. Active tags can send stronger signals and thereby achieve longer reading ranges. However, the battery incorporated in active tags limits their lifetime. In contrast, passive tags are only activated once they receive a signal from the reader, which triggers the reflection of the signal and the transmission of data. Since they lack an energy source, their signal is weaker, resulting in shorter ranges. Passive tags require no maintenance, and their lifetime is only limited by the degradation of the tag, currently ranging from 10 to 20 years.

RFID nails Mobile RFID reader

Ultra-high frequency (UHF) passive radio tags have been tested for their potential use in marking and monitoring trees in forest inventories. Farve (2014) reports read ranges of 4.5 m for the best combinations of tags and readers.

7.4 Tailored data processing The data processing must factor in both the inventory design and the objectives pursued. Compared to the collection of field data, database configuration and computer application development costs are marginal.

Recommendations The following recommendations concerning the data processing can be drawn from experience: • A reputable, free, open source and cross-platform RDBMS using SQL and implementing a client - server architecture should be used as database engine. • Careful thoughts should be given as to whether the RDBMS should be capable of handling geodata, since this currently narrows down the choice to PostgreSQL (see https://www.postgresql.org/) with its PostGIS extension (see http://postgis.net/). • The database structure should be comprehensively documented. • Regular backup copies of the database should be made and kept in a secure place. • The FRA database system application developed by the National REDD+ System Project has the potential to be up-scaled and widened. If a new development is preferred, the application provides illustrations of the features and capabilities which a user-friendly system ought to offer. • Capacities for software application maintenance and ultimately development should be developed, so that the entire inventory cycle can be mastered locally.

113

• Apart from the data collected in the field, data analysis requires reference data, notably on plant species and their attributes. Since tables can easily be imported into RDBMS systems, work could and should started to build a comprehensive table on plant species, with their official common names, accepted scientific names, species groups, protection status, potential non- timber uses, etc.

7.5 Comprehensive data analysis Data analysis is practically only bound by imagination. It should be avoided that the data end in a graveyard.

Recommendations The following recommendations concerning the data analysis can be drawn from experience: • At least one general inventory results report should be elaborated. This report should include a comprehensive analysis of the uncertainties, notably in terms of measurement error, data entry error and statistical sampling error. • Additional specific thematic analysis reports merit to be produced, to make better use of the data. • All reports must consider and state the uncertainties, at least in terms of the statistical sampling error. • The data should be shared with the academe. • The field teams and the quality control teams should be associated in the data analysis, notably to capitalize their field exposure and observations that have not and oftentimes cannot be systematically recorded. • Capacities should be enhanced for country-led data analysis and results interpretation.

114

8. References Barrois V (2017a): Database architecture for the management and analysis of forest resources assessment data. National REDD+ System Philippines Project, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Metro Manila, Philippines http://forestry.denr.gov.ph/redd-plus-philippines/publications- pdf/unpublished/FRA_Database_Architecture.pdf. Barrois V (2017b): Forest Resources Assessment Database System Application Version 4.1 installation guide. National REDD+ System Philippines Project, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Metro Manila, Philippines http://forestry.denr.gov.ph/redd-plus-philippines/publications- pdf/unpublished/FRA_Database_System_Application_V._4.1_Installation_Guide.pdf. Barrois V, Lennertz R (2017): Forest Resources Assessment Database System Application Version 4.1 user guide. National REDD+ System Philippines Project, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Metro Manila, Philippines http://forestry.denr.gov.ph/redd-plus-philippines/publications- pdf/unpublished/FRA_Database_System_Application_V._4.1_User_Guide.pdf. Branthomme A. (2002): National Forest Inventory Philippines - Field manual. FAO, Rome, Italy http://www.fao.org/forestry/download/15607-070f28f166a8e79c0fb151cacb9500731.pdf. Accessed 20 Feb 2017 Brown S (1997): Estimating biomass and biomass change of tropical forests: A primer. FAO Forestry Paper 134, FAO, Rome, Italy http://www.fao.org/docrep/w4095e/w4095e00.htm. Accessed 20 Feb 2017 BSWM (2013): Updating the Harmonized World Soil Database (HWSD): Correlation of Philippine soils into FAO's World Reference Base (WRB) for soil resources. http://www.bswm.da.gov.ph/ladaphilippines/single3.php. Accessed 20 Feb 2017 Chave J, Rejou-Mechain M, Burquez A, Chidumayo E, Colgan MS, Delitti WBC, Duque A, Eid T, Fearnside PM, Goodman RC, Henry M, Martinez-Yrizar -A, Mugasha WA, Muller-Landau HC, Mencuccini M, Nelson BW, Ngomanda A, Nogueira EM, Ortiz-Malavassi E, Pelissier R, Ploton P, Ryan CM, Saldarriaga JG, Vieilledent G (2014): Improved allometric models to estimate the aboveground biomass of tropical trees. Global Change Biology (2014) 20, 3177–3190, doi: 10.1111/gcb.12629 http://chave.ups-tlse.fr/chave/chave-gcb14.pdf. Accessed 20 Feb 2017 Chojnacky D, Amacher M, Gavazzi M (2009): Separating duff and litter for improved mass and carbon estimates. Southern Journal of Applied Forestry. 33(1): 29-34. http://www.fs.fed.us/rm/pubs_other/rmrs_2009_chojnacky_d001.pdf. Accessed 20 Feb 2017 Czaplewski R, McRoberts R, Tomppo E (2004): Knowledge reference for national forest assessments - sample design. FAO, Rome, Italy http://www.fao.org/forestry/7367/en/. Accessed 20 Feb 2017 DENR (2014): Measurement standards and procedures in the conduct of inventory for standing trees (timber). FMB Technical Bulletin No. 3. Quezon City, Philippines https://drive.google.com/file/d/0B1G5mTNoDPOFbEQzMjZWMmQ2b28/view?usp=sharing. Accessed 20 Feb 2017 DiRocco TL, Ramage BS, Evans SG, Potts MD (2014): Accountable accounting: Carbon-based management on marginal lands. Forests 2014, 5 (4), 847 - 861 http://dx.doi.org/10.3390/f5040847. Accessed 20 Feb 2017 FAO (2012): National Forest Monitoring and Assessment - Manual for integrated field data collection. Version 3.0. National Forest Monitoring and Assessment Working Paper NFMA 37/E. Rome, Italy www.fao.org/docrep/016/ap152e/ap152e.pdf. Accessed 20 Feb 2017 FAO (2013): National Forest Monitoring Systems: Monitoring and Measurement, Reporting and Verification (M & MRV) in the context of REDD+ activities. UN-REDD Programme, Geneva, Switzerland http://www.unredd.net/index.php?option=com_docman&view=download&alias=10305-national-forest- monitoring-systems-monitoring-and-measurement-reporting-and-verification-m-mrv-in-the-context-of-redd- activities-10305&category_slug=mrv-and-monitoring-296&Itemid=134. Accessed 20 Feb 2017 Farve R (2014): Using radio frequency identification (RFID) for monitoring trees in the forest: State-of-the- technology investigation. National Technology & Development Program, US Forest Service, US Department of Agriculture. https://www.fs.fed.us/t-d/pubs/pdfpubs/pdf14191805/pdf14191805dpi100.pdf. Accessed 20 Feb 2017 Gillespie AJR, Brown S, Lugo AE (1992): Tropical forest biomass estimation from truncated stand tables. Forest Ecology and Management, 48 (1992) 69 - 87. Elsevier, Amsterdam, Netherlands http://dx.doi.org/10.1016/0378-1127(92)90122-P. Accessed 20 Feb 2017

115

Goodman RC, Phillips OL, del Castillo Torres D, Freitas L, Tapia Cortese S, Monteagudo A, Baker TR (2013): Amazon plant biomass and allometry. Forest Ecology and Management, 310 (2013) 994 - 1004. Elsevier, Amsterdam, Netherlands http://www.rainfor.org/upload/publication- store/2013/Goodman/Goodman_et_al_Amazon_palm_biomass_allometry_FEM_2013.pdf. Accessed 20 Feb 2017 Hengl T. (2008): Spatial-analyst.net http://spatial-analyst.net/wiki/index.php/Main_Page. Accessed 20 Feb 2017 IPCC (1996a): Revised 1996 IPCC guidelines for National Greenhouse Gas Inventories, Volume 1 - Reporting instructions. IGES, Hayama, Kanagawa, Japan http://www.ipcc-nggip.iges.or.jp/public/gl/invs4.html. Accessed 20 Feb 2017 IPCC (1996b): Revised 1996 IPCC guidelines for National Greenhouse Gas Inventories, Volume 3 - Reference manual. IGES, Hayama, Kanagawa, Japan http://www.ipcc-nggip.iges.or.jp/public/gl/invs6.html. Accessed 20 Feb 2017 IPCC (2003): Good practice guidance for Land Use, Land-Use Change and Forestry. Penman J, Gytarsky M, Hiraishi T, Krug T, Kruger D, Pipatti R, Buendia L, Miwa K, Ngara T, Tanabe K, Wagner F (eds.). IGES, Hayama, Kanagawa, Japan http://www.ipcc-nggip.iges.or.jp/public/gpglulucf/gpglulucf.html. Accessed 20 Feb 2017 IPCC (2006a): 2006 IPCC guidelines for National Greenhouse Gas Inventories, Volume 1 - General guidance and reporting. Prepared by the National Greenhouse Gas Inventories Programme, Eggleston HS, Buendia L, Miwa K, Ngara T, Tanabe K (eds.). IGES, Hayama, Kanagawa, Japan http://www.ipcc-nggip.iges.or.jp/public/2006gl/vol1.html. Accessed 20 Feb 2017 IPCC (2006b): 2006 IPCC guidelines for National Greenhouse Gas Inventories, Volume 4 - Agriculture, forestry and other land use. Prepared by the National Greenhouse Gas Inventories Programme, Eggleston HS, Buendia L, Miwa K, Ngara T, Tanabe K (eds.). IGES, Hayama, Kanagawa, Japan http://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4.html. Accessed 20 Feb 2017 Johnson EW (2000): Forest sampling desk reference. CRC Press LLC, Boca Raton, Florida, USA Kleinn C, Beckschäfer P, Bhandari N, Fehrmann L, Lam TY, Schnell S, Seidel D, Yang H (2013): WAF-WIKI forest inventory lecture notes. University of Göttingen, Göttingen, Germany http://wiki.awf.forst.uni-goettingen.de/wiki/index.php/Category:Forest_inventory. Accessed 20 Feb 2017 Lasco RD, MacDicken KG, Pulhin FB, Guillermo IQ, Sales RF, Cruz RVO (2006): Carbon stocks assessment of a selectively logged Dipterocarp forest and wood processing mill in the Philippines. Journal of Tropical Forest Science 18 (4): 201 - 221 https://www.jstor.org/stable/43594677?seq=1#page_scan_tab_contents. Accessed 20 Feb 2017 Lennertz R (2016a): Methodology and Results of the 2015 - 2016 Forest Resources Assessment in the selected Project sites in Davao Oriental. National REDD+ System Philippines Project, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Metro Manila, Philippines http://forestry.denr.gov.ph/redd-plus-philippines/publications- pdf/unpublished/Davao_Oriental_FRA_Results.pdf. Lennertz R (2016b): Methodology and Results of the 2014 - 2015 Forest Resources Assessment in the selected Project sites in Eastern Samar. National REDD+ System Philippines Project, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Metro Manila, Philippines http://forestry.denr.gov.ph/redd-plus-philippines/publications- pdf/unpublished/Eastern_Samar_FRA_Results.pdf. Lennertz R, Fiel R, Megraso CP (2014): Manual for the forest resources assessments. National REDD+ System Philippines Project, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Metro Manila, Philippines http://forestry.denr.gov.ph/redd-plus-philippines/publications-pdf/unpublished/FRA_Field_Manual.pdf. Lennertz R, Schade J (2017): Methodology and Results of the 2015 Forest Resources Assessment in the Project Sites of Aklan, Antique, Capiz and Iloilo Adjacent to the Panay Mountain Range. Forest and Climate Protection Project Panay - Phase II. Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Metro Manila, Philippines http://forestry.denr.gov.ph/redd-plus-philippines/publications-pdf/unpublished/Panay_FRA_Results.pdf. Mandallaz D (2008): Sampling techniques for forest inventories. Chapman & Hall/CRC, Boca Raton, Florida, USA Maniatis D, Mollicone D (2010) Options for sampling and stratification for national forest inventories to implement REDD+ under the UNFCCC. Carbon Balance and Management, 5:9 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3018366/pdf/1750-0680-5-9.pdf. Accessed 20 Feb 2017 MNR (1986): Forest resources of Regions 10 and 11. Results of the BFD Northeastern Mindanao Project assisted by FAO TCP/PHI/8802, processed and compiled by RP German FRI Project. Forest Resources Inventory Project, Deutsche Gesellschaft für Technische Zusammenarbeit (GTZ) GmbH, Metro Manila, Philippines

116

Reyes G, Brown S, Chapman J, Lugo AE (1992): Wood densities of tropical tree species. United States Department of Agriculture, Forest Service, Southern Forest Experiment Station, New Orleans, Louisiana, USA http://www.fs.fed.us/global/iitf/pubs/gtr_so088_1992.pdf. Accessed 20 Feb 2017 Schade J, Ludwig R (2013): Forest carbon baseline study in Leyte. Climate-relevant Modernization of Forest Policy and Piloting of REDD Project, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH, Metro Manila, Philippines http://faspselib.denr.gov.ph/sites/default/files//Publication%20Files/Forest%20Carbon%20Baseline%20Stu dy%20in%20Leyte.pdf. Accessed 20 Feb 2017 Schreuder HT, Richard E, Ramirez-Maldonadi H (2004): Statistical techniques for sampling and monitoring natural resources. United States Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Colins, Colorado, USA https://www.fs.fed.us/rm/pubs/rmrs_gtr126.pdf. Accessed 20 Feb 2017 Thiele T, Mussong M, Mateboto J (2010): Monitoring, assessment and reporting for sustainable management in Pacific Island Countries manual. http://theredddesk.org/sites/default/files/resources/pdf/MAR- SFM%20in%20Pacific%20Island%20Countries%20-%20Manual_1.pdf. Accessed 20 Feb 2017 Tomppo E, Andersson K (2008): Technical review of FAO's approach and methods for National Forest Monitoring and Assessment /NFMA). NFMA Working Paper No. 38, Rome, Italy. http://www.fao.org/docrep/016/ap153e/ap153e.pdf. Accessed 20 Feb 2017 United Nations Climate Change Secretariat (2014): Handbook on measurement, reporting and verification for developing country parties. Bonn, Germany http://unfccc.int/files/national_reports/annex_i_natcom_/application/pdf/non-annex_i_mrv_handbook.pdf. Accessed 20 Feb 2017 Yale University (1997): Yale statistics courses. Yale, USA http://www.stat.yale.edu/Courses/1997-98/101/confint.htm. Accessed 20 Feb 2017 Zanne AE, Lopez-Gonzalez G, Coomes DA, Ilic J, Jansen S, Lewis SL, Miller RB, Swenson NG, Wiemann MC, Chave J (2009): Global wood density database. Dryad, Durham,North Carolina, USA http://datadryad.org/handle/10255/dryad.235. Accessed 20 Feb 2017 Zemek OJ (2009): Biomass and carbon stocks inventory of perennial vegetation in the Chieng Khoi watershed, northwest Viet Nam. M.Sc. thesis, University of Hohenheim, Stuttgart, Germany. https://www.uni-hohenheim.de/sfb564/public/c4_files/zemek_msc.pdf. Accessed 20 Feb 2017 Zöhrer F (1980): Forstinventur: Ein Leitfaden für Studium und Praxis. Parey, Hamburg, Berlin, Germany

117

Appendix 1: Field data forms

Forest Resources Assessment

Sample Point No. Date 2 0

Team Leader

Region Assistant

Province Helper 1

City / Municipality Helper 2

Barangay Helper 3

Measurements / observations at the Sample Point

Target coordinates Zone North m E m N

Actual coordinates Zone North m E m N

Elevation m Slope % Slope orientation °

Terrain Land classification

01 = Plateau 02 = Summit / crest 03 = Upper slope 04 = Middle slope 05 = Lower slope 06 = Bench / terrace 1 = Forest land 07 = Valley 08 = Plain 09 = Narrow depression 10 = Water course 11 = Dunes 2 = Alienable & Disposable land

Observations within a radius of 25 m horizontal distance around the Sample Point

Land cover Forest type Tree crown cover

01 = Forest 02 = Marshland / swamp 03 = Fallow 01 = Dipterocarp old growth forest 02 = Dipterocarp residual forest 1 = non-forest (tree crown cover <= 10 %) 04 = Shrubs 05 = Wooded grassland 03 = Mossy forest 04 = Submarginal forest 2 = open forest (10 % < tree crown cover <= 40 %) 06 = Grassland 07 = Annual crop 05 = Closed pine forest 06 = Open pine forest 3 = closed forest (tree crown cover > 40%) 08 = Perennial crop 09 = Open / barren land 07 = Mangrove old growth forest 08 = Mangrove reproduction forest 10 = Built-up area 11 = Fishpond 12 = Inland water 09 = Native tree plantation 10 = Other plantation forest

Sketch map of the approach to the Sample Point / Remarks

118

Forest Resources Assessment

Sample Point No. Date 2 0

Plot re-located Page Team Leader

Region Assistant

Province Helper 1

City / Municipality Helper 2

Barangay Helper 3

Measurement / observations at the Plot Center

Actual coordinates Zone North m E m N

Elevation m Slope % Slope orientation °

Terrain Land classification

01 = Plateau 02 = Summit / crest 03 = Upper slope 04 = Middle slope 05 = Lower slope 06 = Bench / terrace 1 = Forest land 07 = Valley 08 = Plain 09 = Narrow depression 10 = Water course 11 = Dunes 2 = Alienable & Disposable land

Observations within a radius of 10 m horizontal distance around the Plot Center

Land cover Forest type Tree crown cover

01 = Forest 02 = Marshland / swamp 03 = Fallow 01 = Dipterocarp old growth forest 02 = Dipterocarp residual forest 1 = non-forest (tree crown cover <= 10 %) 04 = Shrubs 05 = Wooded grassland 03 = Mossy forest 04 = Submarginal forest 2 = open forest (10 % < tree crown cover <= 40 %) 06 = Grassland 07 = Annual crop 05 = Closed pine forest 06 = Open pine forest 3 = closed forest (tree crown cover > 40%) 08 = Perennial crop 09 = Open / barren land 07 = Mangrove old growth forest 08 = Mangrove reproduction forest 10 = Built-up area 11 = Fishpond 12 = Inland water 09 = Native tree plantation 10 = Other plantation forest

Observations within a radius of 5 m horizontal distance around the Plot Center

Plant diversity (number of distinct plant species)

Grass, herbs, mosses < 50 cm Plants 50 cm - 130 cm Plants 130 cm - 200 cm

Plants 2.0 - 4.0 m Plants 4.0 m - 10.0 m Plants > 10.0 m Ground coverage classes: 0 = none 1 = coverage <= 10 % 2 = 10 % < Coverage <= 50 % 3 = coverage > 50 %

Assessment of litter within a radius of 5 m horizontal distance around the Plot Center

Ground coverage % Average depth cm

Inventory of lying dead wood with D >= 5.0 cm* # Mid-Diameter Length # Mid-Diameter Length # Mid-Diameter Length [cm] [m] [cm] [m] [cm] [m] ...... Continuation on page * within a radius of 5 m horizontal distance around the Plot Center

119

Forest Resources Assessment

Sample Point No. Date 2 0

Plot re-located Page Team Leader

Assistant

Inventory of live trees, bamboos, palms and standing dead wood with Dref >= 5.0 cm* # Species Azimuth Hor. Distance DBH / DAB M. Height**

Name Code [°] [m] [cm] [m] Live Dead . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ . . . □ □ Continuation on page

* live trees, bamboos, palms, rattan and tree ferns with 5.0 cm <= Dref < 20.0 cm within a radius of 5 m horizontal distance around the Plot Center * live trees, bamboos, palms, rattan and tree ferns with Dref >= 20.0 cm within a radius of 10 m horizontal distance around the Plot Center * standing dead wood within a radius of 5 m horizontal distance around the Plot Center ** merchantable height for all standing dead wood with Dref >= 5.0 cm and for live trees with Dref >= 20.0 cm 120

Annex 1: List of recorded species

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Adina Pertusadina multifolia (Havil.) Rubiaceae ✓ Ridsdale Aglaia tomentosa Aglaia tomentosa Teijsm. & Binn. Meliaceae 0.68 ✓ Agoho Casuarina equisetifolia L. Casuarinaceae 0.80 ✓ ✓ ✓ Agoho del Monte Gymnostoma rumphianum (Miq.) Casuarinaceae 0.86 ✓ ✓ L.A.S. Johnson Agus-us Paratrophis philippinensis Fern. - Moraceae 0.54 ✓ Vill. Akleng-parang Albizia procera (Roxb.) Benth. Leguminosae 0.51 ✓ Alagasi Leucosyke capitellata Wedd. Urticaceae ✓ ✓ ✓ Alagau Premna odorata Blanco Lamiaceae ✓ ✓ ✓ Alahan Guioa koelreuteria (Blanco) Merr. Sapindaceae ✓ Alangas Ficus heteropoda Miq. Moraceae 0.39 ✓ Alas-as Pandanus luzonensis Merr. Pandanaceae ✓ ✓ Alilaua Oreocnide trinervis (Wedd.) Miq. Urticaceae ✓ Alim Melanolepis multiglandulosa Euphorbiaceae 0.34 ✓ ✓ ✓ (Reinw. ex Blume) Rchb. & Zoll. Almaciga Agathis philippinensis Warb. Araucariaceae 0.45 ✓ ✓ Almon Shorea almon Foxw. Dipterocarpaceae 0.39 ✓ ✓ ✓ Alupag Dimocarpus longan subsp. Sapindaceae 0.70 ✓ ✓ malesianus Leenh. Alupag-amo Litchi chinensis Sonn. Sapindaceae 0.80 ✓ Amamali Leea aculeata Blume ex Spreng Vitaceae ✓ ✓ ✓ Amugis Koordersiodendron pinnatum Merr. Anacardiaceae 0.61 ✓ ✓ Amuyon Goniothalamus amuyon (Blanco) Annonaceae ✓ Merr. Anabiong Trema orientalis (L.) Blume Cannabaceae 0.33 ✓ ✓ ✓ Anagap Archidendron scutiferum (Blanco) Leguminosae ✓ I.C. Nielsen Anahaw Saribus rotundifolius (Lam.) Blume Arecaceae ✓ ✓ Anang Diospyros pyrrhocarpa Miq. Ebenaceae 0.64 ✓ Aniatam-mali Cleistanthus decurrens Hook. f. Phyllanthaceae ✓ Anibong Oncosperma tigillarium (Jack) Ridl. Arecaceae ✓ Anii Erythrina fusca Lour Leguminosae 0.25 ✓ ✓ Anilao Colona serratifolia Cav. Malvaceae 0.38 ✓ ✓ ✓ Anislag Flueggea flexuosa Muell. Arg. Euphorbiaceae 0.69 ✓ ✓ ✓ Anolang Haplostichanthus lanceolata (S. Annonaceae ✓ Vidal) Heusden Anonang Cordia dichotoma G. Forst. Boraginaceae 0.38 ✓ ✓ Antipolo Artocarpus blancoi (Elmer) Merr. Moraceae 0.43 ✓ ✓ ✓ Anubing Artocarpus ovatus Blanco Moraceae 0.61 ✓ ✓ ✓ Anuling Pisonia umbellifera (J.R. Forst. & Nyctaginaceae 0.24 ✓ G. Forst.) Seem. Apanang Mallotus cumingii Muell. Arg. Euphorbiaceae 0.49 ✓ ✓ Apitong Dipterocarpus grandiflorus (Blanco) Dipterocarpaceae 0.67 ✓ ✓ ✓ Blanco Aplas Ficus ampelas Burm.f. Moraceae 0.38 ✓ ✓ ✓ Ata-ata Diospyros mindanaensis Merr. Ebenaceae 0.65 ✓ Aunasin Ardisia paniculata Roxb. Primulaceae ✓ ✓ Auri Acacia auriculiformis Benth. Leguminosae ✓ Avocado Persea americana Mill. Lauraceae ✓ Badlan Radermachera gigantea (Blume) Bignoniaceae 0.48 ✓ Miq. Badling Astronia cumingiana S. Vidal Melastomataceae ✓ ✓ ✓

121

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Bagalunga Melia azedarach L. Meliaceae 0.46 ✓ ✓ Bagarilao Cryptocarya ampla Merr. Lauraceae ✓ Bago Gnetum gnemon L. Gnetaceae 0.61 ✓ ✓ Bagtikan Shorea malaanonan Blume Dipterocarpaceae 0.51 ✓ ✓ ✓ Bahai Ormosia calavensis Blanco Leguminosae 0.43 ✓ ✓ ✓ Bakan Litsea philippinensis Merr. Lauraceae ✓ ✓ ✓ Bakauan-gubat Carallia brachiata (Lour.) Merr. Rizophoraceae 0.66 ✓ ✓ Bakayau Cleistanthus oblongifolius (Roxb.) Phyllanthaceae 0.53 ✓ Muell. Arg. Balanti Homalanthus populneus (Geiseler) Euphorbiaceae 0.29 ✓ ✓ ✓ Pax Balat-buaya Fagraea racemosa Jack Gentianaceae 0.64 ✓ ✓ Balete Ficus balete Merr. Moraceae 0.65 ✓ ✓ ✓ Balik Hydnocarpus heterophylla Blume Achariaceae ✓ Balinghasai Buchanania arborescens (Blume) Anacardiaceae 0.45 ✓ Blume Balitbitan Cynometra ramiflora L. Leguminosae 0.79 ✓ Balobo Diplodiscus paniculatus Turcz. Malvaceae 0.63 ✓ ✓ ✓ Balukanag Chisocheton cumingianus (C.DC.) Meliaceae 0.55 ✓ ✓ ✓ Harms Banaba Lagerstroemia speciosa (L.) Pers. Lythraceae 0.55 ✓ ✓ Banai-banai Radermachera pinnata (Blanco) Bignoniaceae 0.46 ✓ ✓ Seem. Banato Mallotus philippensis (Lam.) Muell. Euphorbiaceae 0.60 ✓ Arg. Bangkal Nauclea orientalis (L.) L. Rubiaceae 0.47 ✓ ✓ ✓ Bangkal, Kaatoan Breonia chinensis (Lam.) Capuron Rubiaceae 0.34 ✓ ✓ Bangkal, Neonauclea formicaria (Elmer) Rubiaceae ✓ ✓ ✓ Southern / Merr. Hambabalud Banitlong Cleistanthus pilosus C.B. Rob. Phyllanthaceae ✓ ✓ Bansalangin Mimusops elengi L. Sapotaceae 0.82 ✓ ✓ Banuyo Wallaceodendron celebicum Koord. Leguminosae 0.56 ✓ ✓ Basikong Ficus botryocarpa Miq. Moraceae 0.43 ✓ ✓ Batete Kingiodendron alternifolium (Elmer) Leguminosae 0.49 ✓ ✓ Merr. & Rolfe Batino Alstonia macrophylla Wall. ex Apocynaceae 0.64 ✓ ✓ ✓ G.Don Batitinan Lagerstroemia piriformis Koehne Lythraceae 0.50 ✓ Bayag-usa Voacanga globosa (Blanco) Merr. Apocynaceae ✓ ✓ Bayanti Aglaia rimosa (Blanco) Merr. Meliaceae 0.69 ✓ ✓ ✓ Bayog Dendrocalamus merrillianus Poaceae ✓ ✓ (Elmer) Elmer Bayok Pterospermum diversifolium Blume Sterculiaceae 0.57 ✓ ✓ ✓ Bayuko Artocarpus fretessii Teijsm. & Binn. Moraceae 0.51 ✓ ex Hassk. Benguet Pine Pinus kesiya Royle ex. Gordon Pinaceae 0.48 ✓ Bignai Antidesma bunius (L.) Spreng. Phyllanthaceae 0.51 ✓ ✓ Bignai-pogo Antidesma montanum Blume Phyllanthaceae 0.59 ✓ Binaton Falcatifolium falciforme (Parl.) de Podocarpaceae 0.57 ✓ Laub. Binggas Terminalia citrina Roxb. ex Fleming Combretaceae 0.71 ✓ ✓ Bingliu Polyscias cenabrei (Merr.) Lowry & Araliaceae ✓ ✓ G.M. Plunkett Binoloan Syzygium acuminatissimum Myrtaceae 0.63 ✓ (Blume) DC. Binuang Octomeles sumatrana Miq. Datiscaceae 0.30 ✓ ✓

122

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Binucao Garcinia binucao (Blanco) Choisy Clusiaceae 0.75 ✓ ✓ ✓ Binunga Macaranga tanarius (L.) Muell. Arg. Euphorbiaceae 0.43 ✓ ✓ ✓ Bitanghol Calophyllum blancoi Planch. & Clusiaceae 0.46 ✓ ✓ ✓ Triana Bitanghol-sibat Calophyllum lancifolium Elmer Clusiaceae 0.53 ✓ ✓ Bitaog-Palomaria Calophyllum inophyllum L. Clusiaceae 0.60 ✓ Boga Alseodaphne philippinensis (Elmer) Lauraceae ✓ Kosterm. Bogaiat Garcinia rhizophoroides Elmer Clusiaceae 0.75 ✓ Bogo Garuga floribunda Decne. Burseraceae 0.51 ✓ Bolon Platymitra arborea (Blanco) P.J.A. Annonaceae 0.74 ✓ ✓ Kessler Bolong-eta Diospyros pilosanthera Blanco Ebenaceae 0.65 ✓ ✓ Bonot-bonot Glochidion camiguinense Merr. Phyllanthaceae ✓ Botinag Homalanthus fastuosus (Linden) Euphorbiaceae 0.57 ✓ Fern.-Vill. Botree Ficus religiosa L. Moraceae 0.44 ✓ Bridelia stipularis Bridelia stipularis (L.) Blume Phyllanthaceae ✓ Bugawak Melicope confusa (Merr.) P.S. Liu Rutaceae 0.38 ✓ ✓ Bulala (Wild Dimocarpus fumatus (Blume) Sapindaceae ✓ ✓ ✓ Rambutan) Leenh. Bulalog Parishia maingayi Hook.f. Anacardiaceae 0.51 ✓ Bunga Areca catechu L. Arecaceae ✓ ✓ Bunguas Homalium gitingense Elmer Salicaceae 0.76 ✓ Buntan Engelhardtia rigida Blume Juglandaceae 0.42 ✓ ✓ Bunud Knema mindanaensis Merr. Myristicaceae 0.53 ✓ Buri Corypha utan Lam. Arecaceae ✓ ✓ Butlig-babui Canthium gynochthodes Baill. Rubiaceae ✓ Butlo Aquilaria cumingiana (Decne.) Ridl. Thymelaeaceae ✓ Caimito Chrysophyllum cainito L. Sapotaceae ✓ ✓ Coconut Cocos nucifera L. Arecaceae ✓ ✓ ✓ Dacrycarpus Dacrycarpus cumingii (Parl.) de Podocarpaceae ✓ cumingii Laub. Dacrydium Dacrydium beccarii Parl. Podocarpaceae 0.61 ✓ ✓ beccarii Dalingdingan Hopea foxworthyi Elmer Dipterocarpaceae 0.51 ✓ ✓ Dalinsi Terminalia pellucida C. Presl Combretaceae ✓ ✓ ✓ Dalunot Pipturus arborescens (Link) C.B. Urticaceae ✓ ✓ ✓ Rob. Dalutan Lithocarpus coopertus (Blanco) Fagaceae 0.70 ✓ Rehd. Dangkalam Calophyllum obliquinervium Merr. Clusiaceae 0.64 ✓ Danglin Grewia multiflora Juss. Malvaceae 0.48 ✓ ✓ Dangula (Sasalit) Teijsmanniodendron ahernianum Lamiaceae 1.03 ✓ (Merr.) Bakh. Dao Dracontomelon dao (Blanco) Merr. Anacardiaceae 0.40 ✓ ✓ & Rolfe Dins Ochrosia glomerata (Blume) Apocynaceae 0.57 ✓ F.Muell. Dita Alstonia scholaris (L.).R. Br. var. Apocynaceae 0.39 ✓ ✓ ✓ scholaris Duguan Myristica philippinensis Gand. Myristicaceae 0.36 ✓ ✓ ✓ Duklitan Planchonella duclitan (Blanco) Sapotaceae 0.51 ✓ ✓ Bakh.f. Dulit Canarium hirsutum Willd. Burseraceae 0.49 ✓ Dungau-bundok Astronia lagunensis Merr. Melastomataceae ✓

123

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Dungo Ficus nervosa subsp. pubinervis Moraceae 0.28 ✓ (Blume) C.C. Berg Dungon Heritiera sylvatica S.Vidal Malvaceae 0.70 ✓ Dungon-late Heritiera littoralis Aiton Malvaceae 0.87 ✓ Ebony Diospyros vera (Lour.) A.Chev. Ebenaceae 0.85 ✓ ✓ Eurya nitida Eurya nitida Korth. Pentaphylacacea 0.53 ✓ e Firetree Delonix regia (Hook.) Raf. Leguminosae ✓ Galo Anacolosa frutescens (Blume) Olacaceae ✓ Blume Ganophyllum Ganophyllum falcatum Blume Sapindaceae ✓ falcatum Gatasan Garcinia venulosa (Blanco) Choisy Clusiaceae ✓ Gisok-Gisok Hopea philippinensis Dyer Dipterocarpaceae 0.67 ✓ ✓ ✓ Gubas Endospermum peltatum Merr. Euphorbiaceae 0.30 ✓ ✓ Guijo Shorea guiso Blume Dipterocarpaceae 0.71 ✓ ✓ ✓ Gulob Leea aequata L. Vitaceae ✓ Gumunan Diospyros buxifolia (Blume) Hiern Ebenaceae 0.78 ✓ Hagakhak Dipterocarpus validus Blume Dipterocarpaceae 0.54 ✓ Hagimit Ficus minahassae (Teijsm. & Moraceae 0.32 ✓ ✓ ✓ Vriese) Miq. Hamindang Macaranga bicolor Muell. Arg. Euphorbiaceae 0.30 ✓ ✓ ✓ Hantatamsi Cyrtandra villosissima Merr. Gesneriaceae ✓ Haras / Ituman Garcinia ituman Merr. Clusiaceae ✓ ✓ Hawili Ficus septica Burm.f. Moraceae 0.42 ✓ ✓ Highland Panau Dipterocarpus hasseltii Blume Dipterocarpaceae 0.56 ✓ Himbabao Broussonetia luzonica (Blanco) Moraceae 0.50 ✓ ✓ ✓ Bureau Hindang Myrica javanica Blume Myricaceae ✓ ✓ ✓ Hindang-Laparan Myrica javanica Blume Myricaceae ✓ Hinlaumo Mallotus mollissimus (Geiseler) Airy Euphorbiaceae 0.35 ✓ Shaw Igang Syzygium garciae (Merr.) Merr. Myrtaceae 0.73 ✓ Igem Dacrycarpus imbricatus (Blume) de Podocarpaceae 0.41 ✓ Laub. Igyo Dysoxylum gaudichaudianum (A. Meliaceae 0.45 ✓ ✓ ✓ Juss.) Miq. Ilang-ilang Cananga odorata (Lam.) Hook.f. & Annonaceae 0.29 ✓ ✓ Thomson Ilo-ilo Aglaia iloilo (Blanco) Merr. Meliaceae 0.53 ✓ Inyam Antidesma tomentosum Blume Phyllanthaceae ✓ Ipil Intsia bijuga (Colebr.) Kuntze Leguminosae 0.72 ✓ Ipil-ipil Leucaena leucocephala (Lam.) de Leguminosae 0.64 ✓ Wit Is-is Ficus ulmifolia Lam. Moraceae 0.38 ✓ ✓ ✓ Itangan Weinmannia luzoniensis S. Vidal Cunoniaceae 0.49 ✓ Kahoi dalaga Mussaenda philippica A. Rich. Rubiaceae ✓ ✓ Kakaag Commersonia bartramia (L.) Merr. Malvaceae 0.34 ✓ Kalambug Gordonia luzonica S. Vidal Theaceae ✓ Kalantas Toona calantas Merr. & Rolfe Meliaceae 0.29 ✓ ✓ ✓ Kaliantan Leea philippinensis Merr. Vitaceae ✓ Kalingag / Cinnamomum mercadoi S. Vidal Lauraceae 0.43 ✓ ✓ ✓ Cinamomon Kalipapa Vitex quinata (Lour.) F.N.Williams Lamiaceae 0.65 ✓ Kalokoi Ficus callosa Willd. Moraceae 0.29 ✓ ✓ Kalomala Elaeocarpus calomala (Blanco) Elaeocarpaceae ✓ ✓ Merr.

124

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Kalubkub Syzygium calubcob (C.B.Rob.) Myrtaceae 0.73 ✓ ✓ Merr. Kalulot Artocarpus rubrovenius Warb. Moraceae 0.58 ✓ Kalumpang Sterculia foetida L. Sterculiaceae 0.45 ✓ Kalumpit Terminalia microcarpa Decne. Combretaceae 0.53 ✓ ✓ ✓ Kamagong Diospyros discolor Willd. Ebenaceae 0.88 ✓ ✓ ✓ Kamandiis Garcinia rubra Merr. Clusiaceae ✓ ✓ Kamatog Sympetalandra densiflora (Elmer) Leguminosae 0.76 ✓ Steenis Kamuling Microcos stylocarpa Burret Malvaceae 0.41 ✓ Kanapai Ficus magnoliifolia Blume Moraceae 0.28 ✓ ✓ ✓ Kansulud Aglaia argentea Blume Meliaceae 0.63 ✓ Kape Coffea arabica L. Rubiaceae ✓ ✓ Karaksan Chionanthus ramiflorus Roxb. Oleaceae 0.67 ✓ Katagpo Psychotria luzoniensis (Cham. & Rubiaceae ✓ Schltdl.) Fern.-Vill. Katmon Dillenia philippinensis Rolfe Dilleniaceae 0.63 ✓ ✓ ✓ Katong-matsin Chisocheton pentandrus (Blanco) Meliaceae 0.51 ✓ Merr. Kubi Artocarpus nitidus Trécul Moraceae 0.48 ✓ ✓ ✓ Kuela Bhesa paniculata Arn. Centroplacaceae 0.66 ✓ Kulatingan Pterospermum obliquum Blanco Sterculiaceae ✓ ✓ ✓ Kupang Parkia timoriana (DC.) Merr. Leguminosae 0.34 ✓ ✓ Kurutan Olea borneensis Boerl. Oleaceae ✓ Labayanan Radermachera coriacea Merr. Bignoniaceae ✓ Lago Prunus grisea (Blume ex Muell Rosaceae 0.55 ✓ ✓ .Berol.) Kalkman Laloi Turpinia sphaerocarpa Hassk. Staphyleaceae ✓ ✓ Lamio Dracontomelon dao (Blanco) Merr. Anacardiaceae 0.40 ✓ ✓ & Rolfe Lamog Planchonia spectabilis Merr. Lecythidaceae 0.58 ✓ ✓ ✓ Lanete Wrightia pubescens subsp. laniti Apocynaceae ✓ ✓ ✓ (Blanco) Ngan Laneteng gubat Kibatalia gitingensis (Elmer) Apocynaceae ✓ ✓ Woodson Langil Albizia lebbeck (L.) Benth. Leguminosae 0.59 ✓ Langosig Trichospermum involucratum Elmer Malvaceae ✓ Lanipau Terminalia copelandi Elmer Combretaceae 0.46 ✓ ✓ Lanipga Toona philippinensis Elmer Meliaceae ✓ ✓ ✓ Lanutan Mitrephora lanotan (Blanco) Merr. Annonaceae ✓ ✓ ✓ Lanutan-dilau Polyalthia flava Merr. Annonaceae 0.51 ✓ ✓ Lanzones Lansium parasiticum (Osbeck) K.C. Meliaceae 0.71 ✓ Sahni & Bennet Lapnisan Polyalthia oblongifolia Burck Annonaceae ✓ ✓ Lapo-lapo Gyrocarpus americanus Jacq. Hernandiaceae ✓ ✓ Libas Spondias pinnata (L. f.) Kurz Anacardiaceae 0.34 ✓ Ligas Semecarpus cuneiformis Blanco Anacardiaceae ✓ ✓ Lingaton Dendrocnide stimulans (L.f.) Chew Urticaceae ✓ ✓ Lingatong Laportea brunnea Merr. Urticaceae ✓ ✓ Lingo-lingo Vitex turczaninowii Merr. Lamiaceae 0.49 ✓ ✓ ✓ Lipang-kalabaw Dendrocnide meyeniana (Walp.) Urticaceae ✓ ✓ ✓ Chew Lisak Neonauclea bartlingii (DC.) Merr. Rubiaceae ✓ ✓ ✓ Litsea cordata Litsea cordata (Jack) Hook. f. Lauraceae 0.36 ✓ Loktob Duabanga moluccana Blume Lythraceae 0.34 ✓ ✓

125

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Lubeg Syzygium lineatum (DC.) Merr. & Myrtaceae 0.73 ✓ L.M. Perry Lumbayao Heritiera javanica (Blume) Kosterm. Malvaceae 0.62 ✓ Lunas Lunasia amara Blanco Rutaceae ✓ Mabunot Gomphandra luzoniensis (Merr.) Stemonuraceae ✓ Merr. Macaranga Macaranga dipterocarpifolia Merr. Euphorbiaceae ✓ dipterocarpifolia Magabuyo Celtis luzonica Warb. Cannabaceae 0.55 ✓ ✓ Maguilik Premna cumingiana Schauer Lamiaceae ✓ ✓ Mahogany Swietenia mahagoni (L.) Jacq. Meliaceae 0.51 ✓ ✓ Makaasim Syzygium nitidum Benth. Myrtaceae 0.74 ✓ Malaanonan Shorea polita S. Vidal Dipterocarpaceae 0.51 ✓ Malabagang Phyllanthus albus (Blanco) Muell. Phyllanthaceae ✓ ✓ Arg. Malabatino Alyxia concatenata (Blanco) Merr. Apocynaceae ✓ Malabayabas Tristaniopsis decorticata (Merr.) Myrtaceae 0.91 ✓ ✓ Peter G. Wilson & J.T. Waterh. Malabignai Aporosa symplocifolia Merr. Phyllanthaceae ✓ Malabitaog Calophyllum pentapetalum var. Clusiaceae ✓ cumingii (Planch. & Triana) P.F. Stevens Malabuho Sterculia oblongata R. Br. Sterculiaceae 0.22 ✓ Malabunga Alseodaphne malabonga (Blanco) Lauraceae ✓ ✓ Kosterm. Malaikmo Celtis philippensis Blanco Cannabaceae 0.69 ✓ ✓ ✓ Malak-malak Palaquium philippense (Perr.) C.B. Sapotaceae 0.46 ✓ ✓ ✓ Rob. Malakadios Dehaasia cairocan (Vidal) C.K. Lauraceae ✓ ✓ Allen Malakalumpit Terminalia calamansanay Rolfe Combretaceae 0.50 ✓ ✓ Malakape Psydrax dicoccos Gaertn. Rubiaceae ✓ ✓ ✓ Malakatmon Dillenia luzoniensis (Vidal) Merr. Dilleniaceae 0.69 ✓ Malakauayan Podocarpus rumphii Blume Podocarpaceae 0.46 ✓ Malanangka Parartocarpus venenosa Becc. Moraceae 0.35 ✓ Malapanau Dipterocarpus kerrii King Dipterocarpaceae 0.61 ✓ ✓ Malapapaya Polyscias nodosa (Blume) Seem. Araliaceae 0.32 ✓ ✓ ✓ Malaputat Terminalia darlingii Merr. Combretaceae ✓ ✓ Malaruhat / Syzygium claviflorum (Roxb.) Wall. Myrtaceae 0.64 ✓ ✓ Panglomboyen ex A.M. Cowan & Cowan Malaruhat-puti Syzygium bordenii (Merr.) Merr. Myrtaceae 0.73 ✓ Malasangki Euonymus indicus B. Heyne ex Celastraceae 0.55 ✓ Wall. Malasantol Sandoricum vidalii Merr. Meliaceae 0.45 ✓ ✓ ✓ Malasapsap Ailanthus integrifolia Lam. Simaroubaceae 0.31 ✓ ✓ Malatambis Syzygium hutchinsonii (C.B. Myrtaceae 0.73 ✓ ✓ Robinson) Merr. Malatapai Alangium longiflorum Merr. Cornaceae 0.68 ✓ ✓ Malatibig Ficus congesta Roxb. Moraceae ✓ ✓ Malubago Hibiscus tilliaceus L. Malvaceae 0.45 ✓ Malugai Allophylus cobbe (L.) Raeusch. Sapindaceae 0.58 ✓ ✓ ✓ Manggachapui Hopea acuminata Merr. Dipterocarpaceae 0.54 ✓ ✓ Manggasinoro Shorea assamica var. Dipterocarpaceae 0.46 ✓ ✓ philippinensis (Brandis ex Koord.) Y.K. Yang & J.K. Wu Mangium Acacia mangium Willd. Leguminosae ✓ Mangkas Planchonella obovata (R.Br.) Pierre Sapotaceae 0.81 ✓

126

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Mankono Xanthostemon verdugonianus Myrtaceae ✓ ✓ Náves ex Fern. - Vill. Mapilig Xanthostemon bracteatus Merr. Myrtaceae ✓ Marang Litsea perrottetii (Blume) Fern.-Vill. Lauraceae 0.45 ✓ ✓ Marang- Artocarpus odoratissimus Blanco Moraceae 0.55 ✓ banguhan Matang-araw Melicope triphylla (Lam.) Merr. Rutaceae 0.39 ✓ ✓ ✓ Matang-hipon Breynia vitis-idaea (Burm.f.) C.E.C. Euphorbiaceae ✓ ✓ ✓ Fisch. Mayapis Shorea palosapis Merr. Dipterocarpaceae 0.42 ✓ ✓ Milipili Canarium hirsutum Willd. Burseraceae 0.49 ✓ ✓ ✓ Molave Vitex parviflora A. Juss. Lamiaceae 0.70 ✓ Moluccan sau Falcataria moluccana (Miq.) Leguminosae 0.37 ✓ ✓ Barneby & J.W.Grimes Nangka Artocarpus heterophyllus Lam. Moraceae 0.49 ✓ ✓ ✓ Narig Vatica mangachapoi Blanco Dipterocarpaceae 0.75 ✓ ✓ Narra Pterocarpus indicus Willd. Leguminosae 0.53 ✓ Nato Palaquium luzoniense (Fern.-Vill.) Sapotaceae 0.55 ✓ ✓ ✓ Vidal Natong-linis Palaquium glabrifolium Merr. Sapotaceae 0.55 ✓ Niog-niyogan Ficus pseudopalma Blanco Moraceae ✓ Pagpago Platea excelsa var. borneensis Icacinaceae 0.36 ✓ (Heine) Sleumer Pagsahingin- Canarium asperum Benth. Burseraceae 0.47 ✓ ✓ ✓ bulog Paguringon Cratoxylum sumatranum (Jack) Hypericaceae 0.59 ✓ ✓ ✓ Blume Paho Mangifera philippinensis Mukherji Anacardiaceae 0.52 ✓ Pahutan Mangifera altissima Blanco Anacardiaceae 0.59 ✓ ✓ Pakiling Ficus odorata (Blanco) Merr. Moraceae 0.32 ✓ ✓ ✓ Pakong buwaya Cyathea contaminans (Wall. ex Cyatheaceae ✓ ✓ Hook.) Copel. Palo-santo Triplaris cumingiana Fisch. & C.A. Polygonaceae ✓ Mey. Palosapis Anisoptera thurifera (Blanco) Blume Dipterocarpaceae 0.59 ✓ ✓ Pamintaogon Calophyllum soulattri Burm. f. Clusiaceae 0.43 ✓ Pamitaogen Calophyllum whitfordii Merr. Clusiaceae ✓ Panau Dipterocarpus gracilis Blume Dipterocarpaceae 0.60 ✓ ✓ Pandakaking- Tabernaemontana pandacaqui Apocynaceae ✓ ✓ gubat Lam. Pangi Pangium edule Reinw. Achariaceae 0.50 ✓ ✓ Panglongboien Syzygium simile (Merr.) Merr. Myrtaceae 0.56 ✓ Pangnan Lithocarpus sulitii Soepadmo Fagaceae 0.86 ✓ ✓ Patsaragon Syzygium crassibracteatum (Merr.) Myrtaceae 0.73 ✓ Merr. Philippine Ash Fraxinus griffithii C.B. Clarke Oleaceae 0.60 ✓ Pili Canarium ovatum Engl. Burseraceae ✓ ✓ ✓ Piling-liitan Canarium luzonicum (Blume) A. Burseraceae 0.31 ✓ ✓ Gray Pugahan Caryota cumingii Lodd. ex Mart. Arecaceae ✓ ✓ ✓ Pulahan Lansium parasiticum (Osbeck) K.C. Meliaceae 0.71 ✓ Sahni & Bennet Puso-puso Neolitsea vidalii Merr. Lauraceae ✓ ✓ ✓ Putat Barringtonia racemosa (L.) Spreng. Lecythidaceae 0.36 ✓ Putian Alangium javanicum (Blume) Wang. Cornaceae 0.73 ✓ ✓ ✓ var. jaheri Bloem. Rain Tree Albizia saman (Jacq.) Merr. Leguminosae 0.49 ✓ (Acacia)

127

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Red Lauan Shorea negrosensis Foxw. Dipterocarpaceae 0.51 ✓ ✓ ✓ Sablot Litsea glutinosa (Lour.) C.B.Rob. Lauraceae 0.50 ✓ Sagisi Heterospathe elata Scheff. Arecaceae ✓ ✓ Saguimsim Syzygium brevistylum (C.B. Rob.) Myrtaceae ✓ ✓ ✓ Merr Sakat Terminalia nitens C. Presl Combretaceae 0.58 ✓ Salaguisog Angiopteris palmiformis (Cav.) C. Marattiaceae ✓ Chr. Salinggogon Cratoxylum formosum (Jacq.) Hypericaceae 0.72 ✓ ✓ ✓ Benth. & Hook.f. ex Dyer Salingkugi Albizia saponaria (Lour.) Miq. Leguminosae 0.57 ✓ ✓ ✓ Sangilo Pistacia chinensis Bunge Anacardiaceae ✓ Sarawag Pinanga insignis Becc. Arecaceae ✓ ✓ Sinaligan Sterculia rubiginosa Vent. Sterculiaceae ✓ Spike pepper Piper aduncum L. Piperaceae ✓ ✓ ✓ Subiang Bridelia insulana Hance. Phyllanthaceae ✓ ✓ ✓ Sudiang Ctenolophon parvifolius Oliv. Ctenolophonacea 0.74 ✓ e Symplocos Symplocos lancifolia Siebold & Symplocaceae ✓ lancifolia Zucc. Taba Tristaniopsis littoralis (Merr.) Peter Myrtaceae ✓ G. Wilson & J.T. Waterh. Tabau Lumnitzera littorea (Jack) Voigt Combretaceae 0.69 ✓ ✓ Tabian Elaeocarpus monocera Cav. Elaeocarpaceae ✓ ✓ ✓ Tabon-tabon Atuna racemosa Raf. Chrysobalanacea 0.67 ✓ e Tagatoi Palaquium foxworthyi Merr. Sapotaceae ✓ ✓ Tagpo Ardisia elliptica Thunb. Primulaceae ✓ Taipo Polyosma apoensis Elmer Escalloniaceae ✓ Takip-asin Macaranga grandifolia (Blanco) Euphorbiaceae ✓ ✓ Merr. Taklang-anak Garcinia dulcis (Roxb.) Kurz Clusiaceae 0.72 ✓ Talisay Terminalia catappa L. Combretaceae 0.46 ✓ Talisay-gubat Terminalia foetidissima Griff. Combretaceae 0.60 ✓ ✓ ✓ Taluto Pterocymbium tinctorium Merr. Sterculiaceae 0.25 ✓ ✓ ✓ Tamayuan Strombosia philippinensis S. Vidal Olacaceae 0.70 ✓ ✓ ✓ Tambalau Myristica glomerata (Blanco) Kudô Myristicaceae 0.52 ✓ & Masam. Tambis Syzygium aqueum (Burm. f.) Alston Myrtaceae ✓ ✓ Tan-ag Kleinhovia hospita L. Malvaceae 0.39 ✓ ✓ Tangile Shorea polysperma Merr. Dipterocarpaceae 0.51 ✓ ✓ Tangisang- Ficus variegata Blume Moraceae 0.31 ✓ ✓ bayawak Tangisang- Ficus aurita Blume Moraceae 0.31 ✓ ✓ layugan Tanglin Adenanthera intermedia Merr. Leguminosae 0.78 ✓ ✓ Tara-tara Dysoxylum cumingianum C.DC. Meliaceae 0.72 ✓ ✓ ✓ Tarangisi Aglaia cumingiana Turcz. Meliaceae ✓ Tiagkot Archidendron clypearia subsp. Leguminosae 0.32 ✓ clypearia (Jack) I.C. Nielsen Tiaong Shorea ovata Dyer ex Brandis Dipterocarpaceae 0.64 ✓ Tibig Ficus nota (Blanco) Merr. Moraceae ✓ ✓ ✓ Tiga Tristaniopsis micrantha (Merr.) Myrtaceae 0.89 ✓ ✓ Peter G. Wilson & J.T. Waterh. Tikas-pula Canna indica L. Cannaceae ✓ Tindalo Afzelia rhomboidea (Blanco) Leguminosae 0.59 ✓ ✓ S.Vidal

128

Common name Scientific name Family Gravity* Recorded in ... [gr /cm³] Davao E. Panay Or. Samar Tipurus Palaquium polyandrum C.B. Rob. Sapotaceae 0.55 ✓ Toog Petersianthus quadrialatus (Merr.) Lecythidaceae 0.54 ✓ ✓ ✓ Merr. Tuai Bischofia javanica Blume Euphorbiaceae 0.61 ✓ ✓ ✓ Tubal Syzygium trianthum (Merr.) Merr. Myrtaceae 0.73 ✓ Tukang-kalau Aglaia pachyphylla Miq. Meliaceae 0.69 ✓ Tulo Alphitonia philippinensis Braid Rhamnaceae 0.40 ✓ Tumalim Calamus mindorensis Becc. Arecaceae ✓ Tungkao Glebionis coronaria (L.) Cass. ex Compositae ✓ Spach Ulaian Lithocarpus celebicus (Miq.) Fagaceae 0.70 ✓ Rehder Ulango Pandanus acladus Merr Pandanaceae ✓ ✓ Ulayan (Oak) Lithocarpus caudatifolius (Merr.) Fagaceae ✓ ✓ ✓ Rehder Upling buntotan Ficus heteropleura Blume Moraceae 0.32 ✓ ✓ Upling gubat Ficus ampelas Burm.f. Moraceae 0.38 ✓ ✓ Uyok Saurauia elegans Fern.-Vill. Actinidiaceae ✓ Wenzel anang Diospyros lanceifolia Roxb. Ebenaceae 0.66 ✓ White Lauan Shorea contorta S. Vidal Dipterocarpaceae 0.43 ✓ ✓ ✓ White Nato Pouteria macrantha (Merr.) Baehni Sapotaceae 0.52 ✓ ✓ Yabnob Horsfieldia costulata Warb. Myristicaceae ✓ ✓ Yakal Shorea astylosa Foxw. Dipterocarpaceae 0.73 ✓ ✓ Yakal-Gisok Shorea gisok Foxw. Dipterocarpaceae 0.76 ✓ Yakal-Kaliot Hopea malibato Foxw. Dipterocarpaceae 0.89 ✓ ✓ Yakal-Mabolo Shorea ciliata King Dipterocarpaceae 0.89 ✓ Yemane Gmelina arborea Roxb. Lamiaceae 0.43 ✓ ✓ ✓

* for tree species without specific wood gravity, the average wood specific gravity for tropical tree species in Asia of 0.57 gr/cm³ published by Brown (1997) has been used.

129

Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH

Registered offices Bonn und Eschborn

4th Floor Forest Management Bureau Annex Building Department of Environment and Natural Resources Visayas Avenue, Diliman, Quezon City, Philippines 1101 T +63 2 697 3127

E [email protected] I www.giz.de