REPORT IV: Basemap report

 

Monitoring Illegal Killing of elephants (MIKE)

Central African Pilot Project


TECHNICAL REPORT No 4

SITE BASE MAPS, DATA BASES AND MIKE SITE REPORTS

Rene Beyers and John Hart

1 October 2001

 

TABLE OF CONTENTS

1 SUMMARY AND ACKNOWLEDGEMENTS

2 INTRODUCTION

3 OVERVIEW OF SITE-BASED DATA MANAGEMENT

4 MIKE PILOT PROJECT DATA BASES AND BASE MAPS

5 DATA BASE UPDATES AND ARCHIVES

6 REPORTING

7 CAPACITY BUILDING FOR DATA MANAGEMENT

8 SPATIAL MODELLING OF ELEPHANT DISTRIBUTION AND ABUNDANCE

9 CONCLUSIONS AND LESSONS LEARNED

10 BIBLIOGRAPHY

11 ANNEXES

ANNEX 1. Evaluation of available hardware and software (in French).

ANNEX 2. Evaluation of SE Cameroon monitoring program and database for the MIKE program.

ANNEX 3. Field data collection forms for socio-economic surveys of MIKE sites used in the Pilot Project.

ANNEX 4. Summary of Base Map development process (Example Okapi Wildlife Reserve)

ANNEX 5. Pilot Project base maps

Ituri/ Okapi Wildlife Reserve Odzala National Park Central Gabon / Lope Reserve

ANNEX 6. Site Reporting form for MIKE (Intensive survey sties).

ANNEX 7. Spatial Modelling of Elephant Distribution and Abundance

 

1. SUMMARY AND ACKNOWLEDGEMENTS

Site base maps developed from GIS-linked, geo-referenced databases are important tools for data synthesis and analysis, and in the transmission and utilization of information. Base maps have proven to be the single most important document in developing a coherent site-based monitoring program.

This report presents three site base maps, which were developed by the MIKE Pilot Project. The pilot project initiated a new base map for the Ituri/ Okapi Wildife Reserve (DR Congo). The pilot project added information to existing base maps for Lope Reserve (Central Gabon), and Odzala National Park (Northwest Congo).

Base maps were used as a basis for designing and executing elephant dung count surveys, for Law Enforcement Monitoring (LEM), and as a basis for spatial co-variate analysis of elephant distributions during the pilot project.

The most important step in creating a site database is to clearly identify the objectives for monitoring, to which it will be linked. Specific objectives for information use will define the contents of the database, data collection protocols, and data analysis. Technical support, training and data update procedures must also be planned for from the start. Site data bases and base maps must be accessible to potential users, simple enough to update and maintain, and provide products to meet a wide array of needs.

Databases are developed incrementally, and recommended data layers for first generation MIKE base maps include:

Site boundaries Human access and settlement Hydrology and major habitat types Protected area infrastructure and anti-poaching activities (LEM) MIKE elephant Survey locations

These features are likely to be included in any wildlife monitoring site database in Central Africa.

Steps in creating a model database are presented, and the experience of the pilot project is used to provide recommendation on computer and software needs, information transmission, use of satellite imagery, ground verification and electronic (cyber tracker) data collection.

Data archives are inadequately developed in the sub-region, and must be incorporated in the execution of the next phase of MIKE.

The pilot project established that data reporting and transmission of site information to a centralized national program and the CITES / MIKE subunit will require major strengthening and development. A draft Site Report Form, developed during the pilot project is presented. It is recommended that for Central African MIKE sites, this form be prepared and submitted with the elephant survey and LEM spreadsheets provided in Technical Reports 2 and 3, on a trimester (4-month) basis. A model meta-data base structure to aid in management of multi-site data is introduced.

A preliminary spatial analysis of factors affecting elephant distribution and abundance in Odzala NP and vicinity is presented.

The report concludes with an overall evaluation of lessons learned from the pilot project, and recommendations for the development and extension of site-based monitoring programs, including MIKE, in Central Africa.

We would like to acknowledge the important financial contribution of Unites States Fish and Wildlife Service (USFWS) and Wildlife Conservation Society (WCS) in support of the Central African MIKE Pilot Project. The European Union’s Central African Forest Conservation Project, (ECOFAC) gave us access to their major site databases, and provided technical support for the development of the MIKE maps. We would like to acknowledge Leonard Akie, ECOFAC data base manager, Pascal Janvier, (TREES), Nadine Laporte and Didier Devers (University of Maryland), Jean Marc Froment (Chef de Composante, ECOFAC/Odzala) and Emmauael Demerode (Projet Vision / Lope) for important contributions. Madame Elize and Emmanuel Obame Nguema assisted in digitising base maps. Finally, we would like to thank and congratulate the pilot project field teams, and the National Elephant Officers of Gabon, DR Congo and Congo for their contributions in collecting and synthesizing field data for the development of the site base maps.

2. INTRODUCTION

The overall goal of the MIKE program is to provide information on trends in elephant distribution and abundance, and in the factors affecting these, in particular illegal elephant killing. The first three MIKE Pilot Project Technical reports are concerned with information gathering. They provide a rationalization for the design of surveys and suggested methodologies and field protocols for forest elephant inventories, and anti-poaching or law enforcement monitoring (LEM).

Data collection is clearly essential for a monitoring program, but it is not sufficient. All monitoring programs, including MIKE, must plan for data management data syntheses, preliminary analyses and data archives, and for the communication of monitoring results. In essence, monitoring programs must provide for the transformation of data into information.

Turning data into information, and reporting the results proved to be one of the most difficult and problematic components of the MIKE pilot project. This was due to a number of constraints including:

Physical difficulties and high costs of communication Weak links between dispersed field operations and a national, centralized reporting process Inadequate data management Lack of suitable training and needed capacity building for data management.

The linkage between field surveys and reporting have a bearing on the structure and management of national MIKE programs. This will be dealt with in the final Technical Report, Number 6. Capacity building for information management will be further addressed in Technical Report 5 summarizing MIKE training. In this report we will focus on the inputs and structure required to manage and synthesize survey information at the level of the field site.

There may be some question as to why the MIKE monitoring program should concern itself with anything more than data collection and its transmission to CITES. The pilot project experience has shown that both the fidelity of the information acquired, and the sustainability of the monitoring program itself will depend upon the development of capacity, at the national and most significantly, the site level to manage and analyse data. MIKE’s future depends upon the application of this information, not only to the CITES mandate, but also to the management and conservation of elephants in the range states.

National elephant programs represented by the National Elephant Officers and by the MIKE sub-regional Steering Committee are the institutional base for MIKE in each of the range states (See Report Number 6). In Central Africa, at least, the MIKE program will also collaborate with international NGOs and selected private sector partners in developing an information base for elephant conservation. The MIKE program must inform national elephant management priorities, and it must also be accessible to wildlife managers and site-based partners who are responsible for elephant conservation and management.

One of the most important conclusions of the MIKE Pilot Project is that data management and synthesis should be initiated in the field by the same individuals responsible for collecting the information. This capacity should be developed at key locations, including designated MIKE sites and other protected areas where MIKE field teams will be based. Information should be reported from these site databases to national MIKE programs.

There are a number of other reasons to develop data management at the site level:

Data syntheses and preliminary analyses provides a basis for the evaluation of survey design, data quality and field protocols. Site-based data management is the principal mechanism by which MIKE monitoring data will be available to site-based elephant conservation and management programs. Site-based data archives represent a backup in case of data loss or failed communication at the national level.

Cost for site-based data management will include expenses of replicated computers and associated hardware, software and power sources, and a major investment in training. A decentralized approach to data management will require reinforcement of communication between the field and national office. These costs and constraints are not insignificant.

Given the experience of the pilot project, centralized data management is not adequately developed in any country in the region (though some countries have made a commitment to this). It will neither be developed rapidly, nor in isolation of site-based inputs. At the completion of the 24-month pilot project in July 2001,few elephant officers showed that they could manage MIKE data, and communicate results on their own. One conclusion of the pilot experience is that for the immediate future, data management remain de-centralized and site-based, and that centralized data management programs, including MIKE, be built from the bottom up, from site-based data management programs.

The objective of this Pilot Project Technical Report are to provide a basic framework for information management at the level of the survey site and field base, and to provide draft protocols for reporting forms to communicate MIKE site data from the field to the national MIKE office, and from there to CITES.

The specific products included here are:

An overview to the function and process in site databases and mapping, with special reference to the MIKE program. Pilot project products: including

Site base maps Model data management program Site reporting form

Recommend data management procedures for MIKE sites. An overview of capacity building needs for site database and GIS management.

The report also provides an introduction to spatial modelling, an analytical approach that allows an evaluation of factors explaining variation in elephant distribution and abundance, and identifying correlates of illegal elephant killing.

3. OVERVIEW OF SITEBASED DATA MANAGEMENT

All monitoring programs must plan at the outset for a means to manage data and information and, and to communicate results. The essential elements of this process include :

Data summary and synthesis Incorporation of monitoring data and information into site databases linked to Geographic Information Systems (GIS) software. Updating of site databases and base maps. Archiving data bases Communication of information to CITES and other users.

Databases and base maps are the critical component of site-based monitoring programs (Figure 3.1). Base maps inform survey design and receive feedback from the field in the form of geo-referenced survey data in a four-step updating process.

Figure 3.1. The GIS and base map in the monitoring cycle

Maps are the single most important tool for organizing, analysing and communicating monitoring information. GIS programs are an especially useful tool for managing monitoring data. With GIS programs map production is directly linked to data bases. This provides a means to update maps rapidly and permits maps to be produced in relation to specific objectives and outputs.

 

3.1 Principles in Developing Site-based Monitoring Programs

The development of all monitoring programs, including the MIKE program is guided by the following principles:

3.1.1 Identify Objectives:

The most important step in establishing a monitoring program is to clearly identify the objectives for monitoring, the needs for information, and how this information must be presented to meet user needs. The more specific the objectives, and data collection protocols, the more functional a data base.

The purpose of the MIKE monitoring is two fold: to provide information to CITES in support of the MIKE mandate, and secondly, to contribute to an information base for site-based elephant conservation and for the development and application of national elephant management policies.

The specific objectives of the site data bases for the MIKE program are to provide:

A GIS-based, geographically referenced template for entering spatially explicit data, which was collected either remotely or in the field. Base maps to guide sampling design and the placement of survey transects. Cartographic and tabular outputs for navigation in the field and locating sampling units. A data set for spatial modelling of elephant dung data.

3.1.2 Incremental development.

Data bases evolve continually. Yet it is essential as soon as possible in this process, to produce a useable data base, and a base map that is spatially comprehensive, and that contains the minimum, immediately essential information. It is also important that the initial data base and map are a platform on which additional information and coverages can be added as these become available, and that a process for data updating is foreseen from the start.

The following were considered the minimum essential inputs for the first generation base maps for MIKE survey sites:

A geo-referenced base map, preferably from a recent satellite image and digitised in Arc View (or compatible GIS program)

Protected Area boundaries and associated infrastructure and key sites (patrol posts). Human settlements (classified at least by relative population size), Roads, navigable rivers or other access Major hydrology

Pilot project inputs to this basic map included

Data from elephant inventories (See Reports 1 and 2). Law Enforcement Monitoring (LEM) (See Report 3), including

Indicators of elephant poaching: carcasses, poacher camps, encounters with poachers, etc. Anti-poaching effort and results (infrastructure, mobile patrol routes, arrests, seizures.

Additional inputs for selected maps included:

Major habitat formations. Zones and locations of economic activity. Topography Important landscape features

3.1.3 Field verification and data collection.

Assure a process for verification and collection of geo-referencing data at the outset. Rationalized survey design and data collection protocols are essential for this. After building a well-documented database information can be readily updated.

During the pilot project, available, existing data sets were evaluated, and incorporated into site databases as possible. Original data on human presence, socio-economic variables, human access and habitat were collected by field teams using survey designs and field protocols presented in the Pilot Project technical reports

3.1.4 Funding and Technical Support.

It is unlikely that all resources for a base map and base map will be available at the outset. It is essential to use available funds to produce a first useable product, and then seek additional funding to improve and expand this. The first priority is to ensure that data base managers are identified and that their basic training and support is provided for.

In the MIKE pilot project, base maps were developed in collaboration with site partners to better ensure continuity following the pilot phase.

3.1.5 Training

Training and support for data base management is a continuous process, and should include auto- didactic approaches to enhance sustainability. Project leaders should have a specified program to evaluate capacity and effectiveness of training.

The pilot project training program is presented and evaluated in Technical Report 5.

3.2 Overview of the Process of Developing Data Bases and Base Maps.

Figure 4.2 gives an example of the process that was followed for the production of a site database, base map and other GIS products. GIS products include maps and associated attribute tables. Attribute tables contain information that is linked to features on a map. For example the village layer shows the location of villages on a map and the linked record in the attribute table includes the name of the village, the number of inhabitants, main ethnic groups and other socio-economic data. Both the cartographic component of the data and the attribute tables can be updated, queried and analysed.

The basic steps used in producing site base maps are the following:

Evaluate site-based hardware soft ware and communication needs. Identify a suitable geo-referenced base image or map Integrate existing databases for the site. Evaluate gaps and provide training for site-based data managers. Collect field data.

Each of the steps is analysed below, with specific recommendations based on the pilot experience provided

 

 

Figure 4.2: overview of the technical process of the creation of a site base map and other GIS outputs.

 

3.2.1 Hardware, Software and Communication Capacity

Data management and reporting in the pilot project and subsequent phases will be done electronically. It will be necessary to have computers at each site where databases will be managed. For the pilot project, site-based computers belonging to the primary collaborating partner (EU/ECOFAC) were used at Odzala. The pilot project supplied a site-based laptop and software for Ituri, and provided a centrally located desktop computer in Libreville for the national office, for the Gabon program. In addition, national elephant officers were provided with lap top computers and basic software in Cameron, CAR, Congo and DR Congo.

Significant hardware and software problems were encountered at a number of sites, necessitating professional support. This will continue to be the case in subsequent phases of MIKE. All monitoring programs, including MIKE, should plan that hardware service and backup are provided for in any monitoring budget.

Site-based computer systems, including backups, are the essential component of the monitoring infrastructure. A centrally based data management program, without field based computing power and capacity for electronic data transmission is inefficient, and not recommended.

Electronic field data collection (cyber trackers) was tested in Odzala, and results are generally favourable. An assessment is provided in Technical Report 2. The adaptation of Cyber-tracker to the MIKE program should be pursued if all technical issues can be dealt with successfully.

Software used during the MIKE program included Window’s based MS Word, Excel, Access and PowerPoint. Site databases were managed in a Geographic Information System (GIS) with the Arcview environment from ESRI as the main component. Satellite image analysis was done in PCI, ERDAS Imagine and Arcview Image Analyst. However the latter will generally not be done at the site level. MapInfo Professional 5 was used for certain spatial operations and some elephant officers used MapInfo as their primary GIS package.

We recommend that as a minimum, each site should posses a data base manager with capacity to use word processing, Excel, and have some capacity to manage maps in ArcView (though full competence is not necessary to start with). We also recommend that Access or a comparable data management software, and Distance 3.5 (for analysis of line transect data) be included in the priority software kit for each site (See Report 2). As with hardware, back up program CDs must be available at each site, and technical support available on a scheduled basis to trouble shoot, and assist with software problems..

Essentially all national officers and field leaders used computers during the pilot project. In all cases, however, further training is needed. Some officers need remedial basic training. Untrained or inadequately trained field staff seriously corrupted data files and program files during the course of the pilot project.

Communication from field sties remains a critical problem in Central Africa. No pilot site had an adequately developed communication system (An adequate system was defined as including VH radio, satellite phone and fax, and preferably also satellite email,). Odzala had all of these components but email. Some pilot sites lacked any form of communication. Inadequate communication is a very important gap that must be filled during the next phase of MIKE. Collaboration with site-based partners may provide the best array of options to build this capacity. At sites with severely limited communication capacity, data transfer will need to be accomplished physically, with National Officers personally collecting data files during scheduled site visits.

All pilot sites had some form of power generation. Back up solar or other power sources must be foreseen. As with communication infrastructure, investments in site-based power sources could be shared with site-based partners in the next phase of MIKE.

The form used to evaluate existing site based infrastructure for data management is provided in ANNEX 1.

3.2.2 Geo-referenced Base Image

It is essential to initiate a base map with a cartographic platform to which added coverages and additional information can be readily incorporated. To date we have used satellite images, radar imagery, air photo compilations, and commercially available reference maps for base images. Of these, at the pilot sites satellite images proved to be the best base image for GIS map development.

The most important advantage of satellite images is that they show many of the major human settlement and human access features and basic cover types we are immediately concerned with. They have proven to be easiest document to geo-reference, and have presented the least problems with disjunctions (décallage) as new coverage types have been brought in (such as forestry concession limits).

Satellite images present the problem in that many features cannot be interpreted without ground truthing or field verification. However, it is important to point out that while images contain much potential information (including land use and cover types). It is not necessary to be able to interpret all of these image features to begin to use the image for mapping. Many of the essential features for first generation maps were readily extracted from images without need for field verification. Nevertheless, field geo-referencing using hand held Geo-positioning devices (GPS) is essential and must be planned for early on in the base map development.

The imagery we used in the pilot project base maps was made available without cost (in collaboration with remote sensing partners). The first generation base maps were geo-referenced in the field, and digitised with less than three months of field controls, and computer work.

3.2.3 Integration of Existing Data Bases

Much information of pertinence to the MIKE program was available at all pilot sites. This included anti-poaching patrol reports, village censuses and surveys, faunal inventories, habitat maps. In many cases the information was not accompanied by adequate geo-referencing. However, other data sources are well documented and can be brought into the site data base. An evaluation of existing data and information, including its accessibility, coverage and format, should be undertaken at the initiation of the MIKE program to determine its suitability for MIKE data bases. It is also important to specify how existing data will be updated.

3.2.4 Training Data Base Managers

Most of the sites with active mapping programs have used nationally based resources to train staff in GIS management. In every case, locally-based national staff managed the base maps. Training in GIS and data base management have been among the most important inputs in improving staff capacity. Training is dealt with in further detail in Report Number 5.

3.2.5 Data Collection

Field data collection for data bases should be based on appropriate survey design and follow specific protocols. MIKE data include elephant population inventories, and Law Enforcement Monitoring. Survey design and protocols for data collection are provided in Pilot Projects Reports 1 through 3. Additional site data include basic habitat maps, human settlement, and economic activity. Preliminary data collection forms were developed during the pilot project for human settlement surveys in Ituri Forest (see below), and habitat data was collected on elephant inventories. These inputs were incorporated into the base maps.

4. MIKE PILOT PROJECT DATA BASES AND BASE MAPS

This chapter presents the pilot project data bases and base maps produced during the pilot project. In addition, we provide specific technical recommendations for modifying and improving data bases for the next phase of the monitoring program. A presentation of overall lessons learned and general recommendations is found in chapter 7, below.

4.1 Overview of Site Databases

Site databases were developed at the three principal pilot sites: Ituri (Okapi Wildlife Reserve), Odzala National Park and vicinity, and Central Gabon (Lope Reserve). The pilot project contributed to existing data bases at Odzala, and initiated a new data base in Ituri and Lope. In addition, an existing monitoring database for Southeast Cameroon tri-national zone was evaluated during the pilot project (See ANNEX 2.

Table 4.1 gives a summary of the features that were included in the spatial site databases for all three field pilot sites.

 

Table 4.1. Initial Data base inputs for MIKE monitoring program

SITE

FEATURE/COVERAGE

SOURCE

INITIATOR/

FUNDER

LOPE

River network

Radar

MIKE

Villages

Topographic maps/Tabular census data

MIKE

Forest exploitation permits

WRI 1)

WRI / MIKE

Roads

Topographic map

MIKE

Law enforcement infrastructure

DFC 2)

MIKE

Law enforcement : patrols and results

DFC

MIKE

Park boundaries

DFC

MIKE

ODZALA

Socio-economic data for villages in the south

Field data

MIKE (with ECOFAC 3))

Villages for the region

Topographic maps & local knowledge

ECOFAC / MIKE

Roads

Topographic maps

ECOFAC

 

Rivers in and outside the park

Topographic maps

MIKE / ECOFAC

Vegetation classes for part of the park

Landsat 7 image

TREES 4) / ECOFAC

Park boundaries (new and old)

ECOFAC

Law enforcement

Cybertracker data

ECOFAC

ITURI

Hydrology: main rivers

Landsat 5

MIKE (with UMD 5))

Main road network

Landsat 5 and field

MIKE

Villages

Field and Landsat 5

MIKE

Camps for mining, fishing, etc.

Field

MIKE

Park boundaries

CARPE and Landsat 5

MIKE

Edos (forest clearings)

Field

MIKE

Law enforcement and results

Field

MIKE

Vegetation classes

Landsat 5 and field validation

UMD / MIKE

 

TABLE NOTES

1) WRI: World Resources Institute, Washington D.C., USA

2) DFC: Direction de la Faune et de la Chasse, Gabon

3) ECOFAC : Conservation et utilisation rationnelle des Ecosystèmes Forestiers d’Afrique Centrale, EU

4) TREES : Tropical Ecosystem Environment observation by Satellite, Joint Research Centre, European Commission, Ispra, Italy

5) UMD: NASA Pathfinder project, University or Maryland, USA

4.2 Field Data Collection

Data base inputs include existing data and information (see above), opportunistic information for the site and systematic surveys managed by MIKE and collaborating monitoring programs.

Field data collection forms and survey protocols for MIKE elephant and LEM surveys are presented in detail in Pilot Project Technical Reports (1 through 3). In addition, the pilot project undertook socio-economic surveys at two sites, Odzala and Ituri. Data collection forms for these exercises are presented in ANNEX 3.

We recommend that systematic data collection incorporate the following:

A rationalized survey design be developed for all data collection (this will include design unbiased and model-unbiased survey design. (See Technical Report 1). All protocols, including socio-economic and other surveys, be harmonized across MIKE sites. Data collection be made compatible with a cybertracker data collection.

4.3 Review of Base Maps

Two base maps were produced by the pilot project and contributions were made to a third one. We use the base map development for the Ituri Forest pilot site (Okapi Wildlife Reserve) to illustrate the process of database and base map development. Prior to the MIKE program, no comprehensive, current map existed for the Ituri region, and so the pilot project provided the first base map ever for this area. ANNEX 4 summarizes this base map process.

Base maps produced by the pilot project for three pilot sites are provided in ANNEX 5. Specific details for each of these maps is given in the text below.

Table 4.3 gives an overview of the sources of the features that were included in the different site base maps.

 

TABLE 4.3 Coverage features included on MIKE site base maps.

Source

Feature

Satellite data

Cartographic data

Tabular and text data

Field data (GPS)

Human access

Roads

Roads

 

Road access and condition, point features (bridges etc.)

Rivers

Rivers

 

River access, point features (waterfalls, etc.)

Human Presence

Major towns and human occupation

Villages and towns

Socio-economic data

Human settlements and socio-economic data

Human Activity

Forest exploitation

Forest and other exploitation concessions

 

Human activity sign

Conservation status

Verification of protected area boundaries

Protected area boundaries

Protected area boundaries

 
 

Protected area infrastructure

Protected area infrastructure

Protected area infrastructure

   

Law enforcement

Patrol routes, law enforcement results

   

Conservation or development project

 

Habitat

Hydrology

Hydrology

 

Major river systems

 

Vegetation

Vegetation

 

Vegetation ground verification

Countries

 

Country boundaries

Demographic, social, economic and political status of country

 

Known Elephant range

 

Elephant range

Elephant range

 

 

Table 4.3: Geographic features of base maps and sources of data

Interaction between the field and the lab is crucial to get good quality data, to interpret the image, to solve punctual problems with field data collection and to avoid collecting redundant or useless information.

In reality, during the pilot project, we had very limited means of communication and it often took many weeks to exchange information between the field and the lab. It would be extremely useful to invest in good communication tools (e.g. satellite phone with email link) whenever and wherever such a mapping exercise is undertaken. Field data files could be uploaded, entered in the GIS and checked in a matter of hours, so that field teams can be guided and timely informed of eventual problems.

Roads, still water and rivers were digitised on screen from the satellite image. Only the bigger rivers and some others not covered by canopy were reliably digitised. Condition of the roads was described in the field in terms of accessibility (by truck, by car, by bicycle). Roads themselves were digitised on screen from the image. Trailheads of important connecting footpaths were localised in the field and plotted on the map. These footpaths will be followed in the future using continuous GPS tracking and then plotted on the map. Navigability of rivers was also recorded in the field, but this information is still incomplete.

Park boundaries were taken from the CARPE (Central African Regional Program for the Environment) GIS database for Central Africa. Boundaries were adjusted to natural limits (rivers and roads) that were visible on the satellite image. When proper marking of the park will be done in the field, the GIS should be updated accordingly.

New, geo-referenced field data was gathered by field teams on the ground. Data input can also take advantage of aerial videography and other remote detection methods. The crucial feature is to ensure adequate geo-referencing. In Ituri, field teams recorded all villages around the park and geographic positions were taken with GPS. They collected socio-economic (presence of schools, clinics, economic activities) and demographic data. Data collection sheets for the Ituri area are shown in ANNEX 3. These forms also permit updating of existing data on these coverages. Whenever another settlement in the forest was found (mining camps, fishing camps) a GPS position was recorded.

Vegetation classification of the image is currently being done using unsupervised and supervised classification algorithms combined with field data of vegetation recorded on the MIKE transects.

The geographic component of the GIS can be further refined and more detailed geographic data (e.g. smaller rivers, forest clearings) can be added as information becomes available or as a more detailed analysis of images and other sources is carried out. Other "base map" information, which would be useful to add when available, includes topography and administrative boundaries.

Specific elements of the three base maps are further discussed under each heading below:

4.3.1 LOPE

A few topographic maps exist for the Lope region. However the more detailed 1:200 000 and 1:50 000 maps are not available for the Lope reserve itself. Rivers and roads were originally digitized from the 1:1 000 000 topographic maps. Serious problems with location accuracy and completeness for major rivers on these maps were discovered after the first draft and it was subsequently dropped. Using GPS, we also found that the existing topographic maps were not always well geo-referenced. Therefore we used radar imagery to digitise the hydrological network. It is very difficult to find good satellite images in the visible light spectrum for Gabon because of permanent cloud cover during most of the year. Radar imagery doesn’t suffer from this as radio microwaves travel easily through cloud cover. Unfortunately we had no time to verify the quality of the ortho rectification of the source image and location accuracy of the digitised map since this was done near the end of the field work.

4.3.2 ODZALA

Rivers and roads were digitized from topographic maps. A shift of up to 250 m was noted in the field between GPS and map locations. Location of human settlements was digitized from the topographic maps and verified with ECOFAC team members for their existence.

A more detailed socio-economic survey of the villages and temporary settlements south of the park was undertaken by a consultant who was hired by the MIKE Pilot project. Besides valuable data, this survey also provided an opportunity to test and develop field data collection protocols and forms for spatially linked socio-economic data. Forms and protocols were harmonized with ECOFAC data protocols to ensure compatibility of datasets with ECOFAC sites. (ECOFAC is one of the biggest conservation projects in Central Africa and manages 8 sites in the region).

An unsupervised classification was carried out by Pascale Janvier for the European TREES project on a landsat 7 image for a big part of the park. Field data collected by MIKE teams include data on vegetation and will be used to ground-truth the image and will lead to a more reliable classification.

4.3.3 ITURI

Refer in this section to Annex 4 for a summary of the Ituri Base Map process. No reliable or accurate maps were available for this area. The regional CARPE dataset was initially used but was found too inaccurate for working at the smaller scale of the Ituri reserve for which this database was not designed. Other sources of information, including digitized aerial photographs, were also explored but none of these were accurate enough and showed particular problems with geo-referencing. We decided that satellite images would give us the best base to build upon. A Landsat 5 image from the southern part of the reserve was geo-referenced with field GPS locations of features that were visible on the image. We collaborated on this with the NASA Pathfinder project, hosted at the University of Maryland at College park. . In the first step a cloud-fee or cloud-poor image was selected and Ground Control Points (GPC) were identified on the image that could be easily and precisely located in the field (e.g. a river confluence, a road crossing a river, etc.). GPS recordings of these points in the field permitted subsequent geo-referencing of the image in the lab. Accurate topographic maps could also be used as a backdrop and make digitising of geographic features much easier (features are clear and distinct), but for most of the region good maps were not available and corrections of these maps were often necessary.

Several expeditions were set up by the MIKE field teams to obtain information on human access (road condition, major river access), human presence (villages, camps), socio-economic data and major footpaths. GPS locations of all human settlements around the park were recorded and overlaid on the satellite image. This permitted a visual check of human occupied and human modified habitat.

Roads were digitised from the satellite image; condition of the road was described in the field and added to the attribute table in the GIS. Main rivers were also digitised from the satellite image.

Fine scale printouts were made and compared with previously digitized rivers from aerial photographs. Important rivers were traced on the printouts with input of expert knowledge. This aided in the digitizing process. GPS points taken in the field also helped identify important rivers.

 

4.4 Use of Base Maps by the MIKE Program

The site base maps constituted a well geo-referenced template for adding a wide range of spatially linked information on wildlife, law enforcement monitoring, human impact, economic and political features, etc.

Three lines of GIS output were particularly useful in the Pilot Project:

Base for developing elephant dung count survey plan Law Enforcement Monitoring Maps Covariate modelling for spatial analysis

4.4.1 Sampling plan

The GIS provided the base map for allocating sampling effort and laying out the sampling design. Although no stratification was done for the pilot sampling plan, we considered stratifying for accessibility, which would not be possible without proper base maps containing information on access. This process is detailed in Technical Report 1.

In the case of Odzala and Lope, a grid was generated, overlaid on the base map and sampling units (recce-transects) were located in the centre of systematically sampled grid cells. Coordinates of start and end points of recce-transects were stored in a table.

Map, table and GPS were taken into the field to locate sampling locations. The maps and GIS were also used to determine the most efficient route between sampling locations on a particular monitoring trip.

Sampling design and placement of transects has recently been automated in DISTANCE 4.0 software using Arcview compatible files. With small modifications for the proposed MIKE sampling design this program could be used in the future to generate sampling designs.

4.4.2 Law Enforcement Monitoring

Spatially linked law enforcement data can easily be entered in the GIS and plotted on the base maps. Spatio-temporal data on law enforcement effort (patrol routes, patrol frequency etc.) and law enforcement results (indicators of illegal killing of wildlife, poachers arrests, etc.) are among the most important correlates of elephant distribution, and illegal elephant killing. A GIS-produced map provides a strong visual and analytical aid for this assessment and can help identify priority areas, resource needs, management needs etc. Law enforcement monitoring data is essential for understanding the relationship between protection, elephant killing and elephant abundance. Protection is an important covariate in spatial modelling explained below.

Technical Report 3 presents the results of the pilot project law enforcement monitoring program.

5. DATA BASE UPDATES AND ARCHIVES

Data bases serve little use unless the information is updated on a rational basis, and properly archived. Data archives must include information on survey design, and sets must be archived. The same schedule for data updates need not apply for all data fields. The important point is that surveys and data analysis anticipate and respond to expressed information needs, and that information be captured in an adequate data storage system. Protocols for both updating and data archiving should be agreed to at the start of the monitoring program.

5.1 Database Updates

Data updates can be done on in a variety of ways, but systematic surveys should include timing for resurveys, and inputs into a data base should be done on a scheduled and controlled basis to ensure quality control. This latter recommendation would be modified in the case of a data base being linked to a cyber tracker, in which case updates would be linked to the cyber program.

Although the pilot project did not deal with this directly, the pilot experience nevertheless has provided some insight into the process and qualified recommendations for an updating program. We provide a range of recommended update periods. Factors affecting this include the rate of change of the phenomena in question, available funds for surveys, logistical constraints, and the availability of survey teams. Schedules for data updating will have to remain flexible, as highly dynamic situations will require more frequent surveys.

Elephant surveys: We suggest at forested sites systematic ground surveys using MIKE methods occur on a 24 to 36 month cycle, with surveys covering 9 month to 12 month periods, separated by 12 to 24 months. LEM Monitoring should occur in conjunction with anti-poaching programs, and should provide a continuous monitoring of anti-poaching activities at some level. Monitoring reports should be adapted and scheduled to provide pertinent inputs to guide anti-poaching programs. LEM summaries for MIKE should be reported in conjunction with site report forms (see Section 6 below). For the pilot project we recommended site reports on a 4 month basis (three times/year). However, monthly, 6-month or annual reporting schedules are also possible. Human access and settlement. Updates every two to five years. The choice of timing depends upon the rate of change in the survey region. Collaboration with national administrations, churches, private sector or other partners may be possible for these censuses. Socio-economic co-variates. Updates every two to five years, as above. Habitat. Updates every two to ten years, as above. Updates for these parameters should be based on integration of updated imagery. Availability of these images will determine the schedule of updates..

5.2 Data Archives.

During the pilot project we explored options of data archives with ECOFAC.

Of the pilot sites, only Odzala had a system of data archives, but this system was not evaluated during the pilot project. Within the Central African sub-region, data archives are rare and generally not well managed. Where archives occured they were usually limited to a single site, with negligible backup at the national or regional level. This is certainly an areas that will require significant input in the next phase of MIKE.

Decisions such as who and how data can be accessed, data ownership etc. must be well-understood and agreements reached before proceeding with the full-scale program. At present, MIKE data is a national product, and managed through the national MIKE structure.

Data archives are a major product of all long term monitoring programs. Cooke et al (2001) lay out basic principles and recommended practices for data archives that should be incorporated into the MIKE program:

Assign descriptive file names. File names should reflect file content includes location, survey period, investigator, file type. Use consistent and stable file formats. Use ASCII formats, and retain the same file structure throughout. The first row should contain identification information and companion file listings. Define parameters and all data fields. Be sure that the data field titles are spelled out, and that the units of measure are indicated. Use standard, unambiguous notation for dates and time and spatial coordinates. Specify codes for missing data, and unknown values. It is critical to be able to distinguish an observed "zero" from no observation made.. Use a consistent data organization. Standard practice is to use columns as data fields, and each observation in its own row. Perform regular quality control. Check for data entry mistakes by doing sort and extract exercises and investigate all outlier data values. Conduct statistical summaries of results (frequency distributions, means and variances). In some cases double data entry may be used to check for data entry errors.. Provide documentation. The documentation accompanying a data file should be written in a separate file. Do not use acronyms. Plan to have this document read up to 20 years into the future. Provide the following information for each file (or set of linked files):

Data set name Why data collected What data were collected What instruments used Who was responsible for data collection, and key participants/team members. How each parameter was measured. Summary statistics generated.

6. REPORTING

6.1 Pilot Project Reporting Forms

The MIKE site summary reporting form developed during the pilot project is presented in ANNEX 6. This form provides data on elephant survey and LEM effort, and results. Data on protected status, and basic socio-economic parameters and indicators are also provided. The site report form is to be submitted with attached elephant survey (recce transect) and LEM spread sheets. The two spread sheets are presented and discussed in detail in Technical Reports 2 and 3

The site report form is designed to be submitted on a regular basis to CITES. Originally it was proposed that the reports be made on a monthly basis by national elephant officers to their national steering committee representative. During the pilot project, we learned that logistical and financial constraints and communication problems would make it unlikely that national officers could visit all the MIKE survey sites in their country on a monthly basis. In addition, many sites had low levels of activity so that monthly visits would produce very little information. By the same token, if site reports are filed at too infrequent a basis, then information will likely be lost and the data quality degraded.

We propose a compromise timing of trimestrial report, with each report and accompanying spreadsheets covering a four month period.

Not all national officers provided final summary reports following the pilot project. This was due to a number of factors, including turnover of national officers at the end of the pilot project, communication problems, lack of computer capacity, and lack of familiarity with the data. The three pilot site officers (field team leaders), in contrast, all completed and submitted forms for their specific site, and in some cases undertook analyses as well. We recommend that the MIKE programme consider direct reporting of by site officers to the MIKE Sub-regional Coordinating Officer in some cases. Further details for reporting structure are provided in Technical Report 6 which evaluates national elephant programs.

The draft reporting form included in this report is designed to completed by each of the Field Team Leaders in a national MIKE program, in collaboration with the national officer. Copies should be retained by the Field Team Leader, collaborating NGO or other site partners.

The geographic area covered by a given site report is the area covered by the designated field team. Normally, these teams would be based in protected areas, and specifically in designated MIKE sites. Surveys, however would cover a much wider area, and the site report would summarize information over this wider area as well. Recommendations for developing a strategic approach to a national survey program are presented in Report 6.

6.2 Metadatabase

Site data can be managed in a metadatabase. This has potential to be developed to a national MIKE data management system. A simple metadatabase for spatial information was developed to keep track of incoming spatial information and spatial processing of GIS layers. A very simple user friendly interface was developed in Filemaker Pro 4 for Windows (figure 4.3) to handle data entry and retrieval.

We had often difficulties to track down the source and data of much of the original source information and maps. Besides satellite images, for which a good reference was always available, we depended on existing maps and other data in digital format but very rarely this information was accompanied by a description of the source and the quality and uncertainty of the information. It was therefore often difficult to gauge the quality of the source material and this became sometimes only apparent in the field. It is critical to ensure proper documentation of source data for site data bases in the next phase of MIKE.

The database was also useful to systematically record the updating and processing of GIS layers and spatial data. Non-spatial data such as field datasheets, reports etc could also be entered in the database.

We want to emphasize here that a system to keep track of incoming spatial information is essential for the production of a reliable product. Local data managers should be encouraged to use a system, which allows them to organise multiple files and sources of spatial information and to keep track of the important process of updating.

Figure 4.3: main window interface of spatial metadatabase for the Central African MIKE Pilot project

 

7. CAPACITY BUILDING FOR DATA MANAGEMENT

Essentially all sites will need capacity to maintain and utilize data bases, update data files, print and modify maps, and ensure some level of data analysis. Some of this capacity, in particular that which requires more advanced training, will be shared between sites. At least one person at each site should have computer capacity.

MIKE national officers should have the same basic computer capacity as the Site Officers (Field Team Leaders). In addtiona, they should have training in GIS and Access data bas management.

An evaluation of the pilot training program and recommendations for strengthening capacity are provided in Report 5..

8. SPATIAL MODELLING OF ELEPHANT DISTRIBUTION AND ABUNDANCE

Certain environmental and human factors are known to have an important effect on elephant distribution and abundance. These factors can be used in a model to explain variation in elephant densities across the study area. In this model elephant density is dependent on a set of independent variables, covariates with their respective constants that describe the relationship between the independent variable (e.g. distance from roads) and the dependent variable (elephant abundance).

An important covariate in the model that initiated the MIKE project is the CITES decision to down list the elephant from appendix I to appendix II for three Southern African countries. The original goal of the MIKE project was to show if there was a link between this decision and increased illegal killing of elephants and consequent decline of elephant abundance. However in our understanding it was generally accepted in the course of the Central African Pilot project that this causal link would be extremely difficult or even impossible to establish. It was agreed that the wider objective of MIKE was to inform the CITES debate by documenting trends in illegal killing and abundance of elephants and the factors that drive these trends.

Many variables that have an influence on elephants are spatially explicit.

For example in previous studies a strong correlation was found between distance from roads and elephant densities in Gabon (e.g. Barnes et al. 1991, Barnes et al. 1995). Areas close to roads are easily accessible by poachers and suffer from heavy hunting, while areas further away have higher elephant densities. Condition and accessibility of the road, distance to the nearest town etc. could also play a role.

Spatial variables that may have an impact on elephants in the Central African forests include:

Human access: distance from roads, navigable rivers and major footpaths Human presence: villages, temporary camps Human activity and exploitation of natural resources: forestry, mining, agriculture. Protection: distance from the boundary of a protected area, distance from protection bases, patrolled areas and patrol routes Distance from an international boundary Habitat, vegetation types Topography

Non-spatial variables include:

The CITES downlisting of elephants Socio-economic status of the country Political status of the country: peaceful areas versus war zones

A spatial model that includes spatial covariates can be fitted to the transect or recce-transect data (Hedley, Buckland and Borchers,1999) using generalized additive models.

The benefits of spatial modelling are:

an improvement in precision of encounter or density estimates and thus our ability to detect significant trends in elephant densities. an assessment of which variables are related to animal densities a density map for the study area showing projected densities of elephants (or elephant dung)

The model will help us understand the complex relationships between poaching and elephants. When temporal data become available a spatial model could be fitted to each year’s data to obtain a density estimate for any location for which a trend estimate is required (Buckland and Underwood, 2000). Trends in elephant densities can then be interpreted in relation to spatial factors affecting these.

An example analysis (annex 7) was carried out on the Odzala elephant data, modelling elephant dung encounter rates as a function of distance from roads and villages, and distance from protection infrastructure and patrol routes. The preliminary results of this analysis show an important improvement in precision (coefficient of variation was almost halved) and a projected density map of elephant dung in the surveyed area.

 

9. CONCLUSIONS AND LESSONS LEARNED

The following points summarize the major conclusions of the pilot project experience with data management and reporting, and provide recommendations for reporting for the next phase of MIKE.

1) Monitoring programs must be site based and developed at a landscape scale

Monitoring must inform site-based management. Monitoring has to provide information important to the site manager and collaborating NGOs, as well as providing information for CITES/MIKE.

Interpretation of MIKE survey and LEM results will depend upon information of the broader landscape context of the survey area.

This includes the larger spatial matrix of the surrounding human occupied areas. Issues pertinent to MIKE include:

Impact of forestry and mining activities on elephant distribution and abundance, and on illegal killing. Relationship of illegal killing to settlement, international frontiers, military operations. Evidence of trans boundary elephant movements.

2) Site base maps and site data bases are essential

Maps and their GIS linked databases have proven to be the most effective means to integrate and present a broad and diverse information base for each site. It is important to maintain the databases at the site itself where the information will be readily available. Data archives, if not at the site, must also be accessible.

A high quality geo-referenced satellite image has proven to be the best base image for a GIS platform. Good images are readily verified in the field, are the best platform on which to incorporate new data fields and new imagery.

3) Information management and transmission has not developed at same pace as data generation

Much biological and sociological information exists at a number of sites. More attention to sampling design and analysis is needed in many cases. Many surveys and wildlife inventories were conducted as one-time efforts with little consideration for further sampling. Repeated surveys over time, however, will be the only way to evaluate trends in the key features of a protected area. If the design of data collection programs is adequately simple these can be repeated over time despite changing staff and resources. Simple, standardized data collection constitutes the most valuable long-term program.

There was limited capacity to analyse, manage and transmit information at essentially all MIKE pilot sites. As a result, much information that is properly collected is inadequately utilized, and sometimes lost. There is a poor cost benefit ratio of information cost to information utility.

Computer capacity must be maintained at each site. During the course of the pilot year, three of the five field laptops for the national officers had significant problems that constrained information management and transmission. Monitoring program must budget for computer and software maintenance, and plan for data back up.

 

4) Develop monitoring programs incrementally:

Start with base map and simple databases that can be built upon and developed in steps. Focus on information that is needed immediately: This includes LEM, key faunal surveys, selected socio-economic information.

5) Train continuously and evaluate capacity frequently.

The focus should include data base management and capacity to produce and communicate specified output (maps, spread sheets, site reports)

6) Integrate MIKE with other monitoring programs from the start.

Monitoring programs must be harmonized if they are to succeed. Information needs of diverse monitoring programs will likely overlap, and this should be incorporated into collaborative design. The MIKE program must make data management and information transmission its first priority.

10. BIGLIOGRAPHY

Barnes, R.F.W., Barnes, K.L., Alers, M.P.T and A. Blom 1991. Man determines the distribution of elephants in the rain forests of northeastern Gabon. African Journal of Ecology, Vol. 29, pp. 54-65

Barnes-R-F-W; Blom-A; Alers-M-P-T; Barnes-K-L 1995. An estimate of the numbers of forest elephants in Gabon.. Journal of Tropical Ecology 11(1): 27-37

Buckland, S.T. and F.M. Underwood. 2000. Analysis of data and survey design for the mike central African pilot project: first interim report

Cook, R.B., R.J. Olson, P Kanciruk and L Hook. 2001. Best Practices for preparing ecological data sets to share and archive. Bulletin of the Ecological Society of America 82 (2): 138-141.

Hedley, S.L., Buckland, S.T. and Borchers, D.L. 1999. Spatial modelling from line transect data. Journal of Cetacean Research and Management1, 255-64.

Hedley, S.L., Buckland, S.T. and Borchers, D.L. in prep. Spatial models for line transect sampling.

 

11. ANNEXES

ANNEX 1: Evaluation of available hardware and software (in French).

ANNEX 2: Evaluation of SE Cameroon monitoring program and database for the MIKE program.

ANNEX 3: Field data collection forms for socio-economic surveys of MIKE sites used in the Pilot Project.

ANNEX 4: Summary of Base Map development process (Example Okapi Wildlife Reserve)

ANNEX 5: Pilot Project base maps

Ituri/ Okapi Wildlife Reserve

Odzala National Park

Central Gabon / Lope Reserve

ANNEX 6: Site reporting form for MIKE (Intensive survey sties).

ANNEX 7: Spatial Modelling of Elephant Distribution and Abundance