Land health surveillance and response: A framework for evidence-informed land management

Land health surveillance and response A framework forevidence-informed land managementKeith D Shepherd a Gemma Shepherd b Markus G Walsh c

a World Agroforestry Centre (ICRAF) PO Box 30677-00100 Nairobi Kenyab United Nations Environment Programme PO Box 30552-00100 Nairobi Kenyac The Earth Institute at Columbia University co Selian Agricultural Research Institute POB 2704 Arusha Tanzania

A R T I C L E I N F O

Article historyReceived 28 February 2014Received in revised form 24 August 2014Accepted 2 September 2014Available online

KeywordsAssessmentDiagnostic screeningIntervention evaluationLand degradationPreventive measuresRisk factors

A B S T R A C T

Degradation of land health ndash the capacity of land relative to its potential to sustain delivery of ecosys-tem services ndash is recognized as a major global problem in general terms but remains poorly quantifiedresulting in a lack of specific evidence to focus action Land health surveillance and response is designedto overcome limitations of current assessment approaches It is modelled on science principles and ap-proaches used in surveillance in the public health sector which has a long history of evidence-informed policy and practice

Key elements of the science framework are (i) repeated measurement of land health and associ-ated risk factors using probability based sampling of well defined populations of sample units (ii)standardized protocols for data collection to enable statistical analysis of patterns trends and associa-tions (iii) case definitions based on specific diagnostic criteria (iv) rapid low cost screening tests to permitdetection of cases and non-cases in large numbers of samples (v) cost-effectiveness evaluation of in-terventions based on projected reduction in risks and problem incidence (vi) design of statisticallyanalysable studies to evaluate interventions in the real-world (vii) meta-analysis of these data to guidedesign of public policy and intervention programmes and (viii) integrating surveillance and the com-munication and use of results into operational systems as part of regular policy and practice

The scientific rigour of land health surveillance has potential to provide a sound basis for directingand assessing action to combat land degradation Specialized national surveillance units should be es-tablished to harness and realign existing resources to provide integrated national land health systemsAn international unit is needed to provide science and technology support to governments and developstandards whereas an international agency should coordinate land health surveillance globally Appli-cation of the surveillance framework could result in a shift away from a focus on rehabilitation of severelydegraded land towards a preventive approach that focuses more on reducing distal risks at national andregional levels

copy 2014 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-NDlicense (httpcreativecommonsorglicensesby-nc-nd30)

1 Introduction

Land degradation (or terrestrial ecosystem degradation) is gen-erally recognized as a major global environmental and developmentproblem but despite much investment in research and numerousassessments the degree extent and impacts of land degradationremain controversial especially in developing countries (Young1998 Koohafkan 2000 Eswaran et al 2001 Gisladottir andStocking 2005 UNEP 2007) The lack of specific evidence and in-formation at all scales makes it difficult for international andgovernmental policy makers to prioritize and direct interventions

to improving and protecting land health Consequently politicalbacking is often given to human causal relationships that lackempirical support (Lambin et al 2001 Thomas and Middleton1994)

While local participatory and expert-opinion based assess-ments are important for understanding stakeholder perceptions andthe design of local solutions their lack of standardization pre-vents aggregation of results and provision of reliable informationfor decision making and support at higher levels of scale and overtime (MA 2005a Vogt et al 2011) Thus a primary problem forlarge area planning and action is a lack of coherent and rigoroussampling and assessment frameworks that enable comparison ofdata on land condition across a wide range of environmental con-ditions and scales and the integration of such frameworks intodecision-making processes at different levels of scale This

Corresponding author Tel +254 20 7224173 fax +254 20 7224001E-mail address kshepherdcgiarorg (KD Shepherd)

httpdxdoiorg101016jagsy2014090020308-521Xcopy 2014 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license (httpcreativecommonsorglicensesby-nc-nd30)

Agricultural Systems (2014) ndash

ARTICLE IN PRESS

Please cite this article in press as Keith D Shepherd Gemma Shepherd Markus G Walsh Land health surveillance and response A framework for evidence-informed land man-agement Agricultural Systems (2014) doi 101016jagsy201409002

Contents lists available at ScienceDirect

Agricultural Systems

journal homepage wwwelseviercom locate agsy

situation contrasts with the situation in the public health sectorwhere surveillance is the main source of information guidingdecisions and actions in health policy and practice (Teutsch andChurchill 2000) Therefore surveillance frameworks used in publichealth management may have much to offer for guiding land healthmanagement

Public health surveillance which has been active since the 1940shas been defined in simple terms as ldquothe routine ongoing collec-tion analysis and dissemination of data to those in public healthwho need to knowrdquo (Brookmeyer and Stroup 2004) It provides thescientific and factual foundation and database for informed deci-sion making and appropriate public health action We takesurveillance to include the full cycle of communication responseand evaluation and not only data collection and dissemination (egRemmington and Nelson 2010) An example of the contribution ofsurveillance in the public health sector is the major achievementof the Global Burden of Disease Study which provided data on in-cidence and prevalence by age sex and region of over 100 diseasesand injuries and a comparative assessment of 20 major risk factors(Lopez 2005) The data were used to analyse the impact of the dif-ferent risks on the avoidable future burden of disease evaluate thecost-effectiveness of alternative interventions and establish clear in-tervention strategies (WHO 2002) Nothing comparable exists forassessing the state of the land resource base and for definingintervention strategies at any scale

In fact land management problems share many of the complexfeatures of public health problems especially those of non-communicable diseases and thus similar scientific approaches maybe applicable These similarities include

1 A rapidly increasing burden of health problems in develop-ing countries partly linked to demographic factors and theproblems become chronic if not addressed

2 Problems often occur together as syndromes with a commonset of symptoms rather than in isolation

3 Health problems exist as a continuum and there is difficultyin defining the normal case and diagnosing poor versus goodhealth

4 Problems are associated with a range of physical biologicalsocial and economic determinants (or risk factors) both atindividual and population levels

5 Risks are often inter-related and act together to cause a healthproblem They range from proximal risks acting directly tocause the problem to distal risks that are further back in thecausal chain

6 Risk factors are often separated from outcomes in time some-times by many decades making it difficult to establishcausality

7 Risks are generally greater for the disadvantaged in our so-cieties and poverty is a major risk factor

8 Evaluating the cost-effectiveness of alternative preventive andrehabilitation interventions is complex and must considermany factors Interventions may cause unintended side effects

9 Different stakeholder groups (scientists policy makers public)perceive risks differently requiring two-way communica-tion processes

10 Few resources are allocated to surveillance in developing coun-tries presenting challenges such as how best to communicatesurveillance and research findings in this setting which stillneeds much local research

Given these similarities the overall goal of this paper is to applyscientific principles used in public health surveillance to the designof a methodological framework for evidence-informed land man-agement and policy and hopefully stimulate its wide applicationThe specific objectives are to (i) elucidate the scientific principlesand system components of public health surveillance (ii) de-scribe their applicability to land health surveillance and (iii) suggest

opportunities for deployment of land health surveillance systemsand identify areas for their further research

2 Principles of land health surveillance

21 Definition

The definition of land health surveillance is closely modelled onthe definition of public health surveillance given by the Centres forDisease Control and Prevention (Thacker and Berkelman 2008)

ldquoLand health surveillance is the ongoing systematic collectionanalysis and interpretation of data essential to the planning im-plementation and evaluation of land management policy and practiceand application of these data to promote protect and restore landand ecosystem healthrdquo (UNEP 2007) A surveillance system in-cludes a functional capacity for data collection analysis anddissemination linked to land health programmes Surveillance aimsto provide information for action but excludes research on landhealth and full-scale implementation of delivery programmes

Land health is defined here as the capacity of land relative toits potential to sustain delivery of essential ecosystem services (thebenefits people obtain from ecosystems) which are well de-scribed in MA (2003) and Kibblewhite et al (2008) Land health isnot directly equated with human health but rather the scientific con-ceptual and statistical approaches in land health surveillance aremodelled on those employed in public health The parallels in prin-ciples are closer for surveillance of chronic non-communicablediseases than for communicable diseases

22 Functions

In broad terms the ultimate goal of land health surveillance isto tell us (i) where land problems exist (ii) whom and what theyaffect (iii) where programmatic and prevention activities shouldbe directed and (iv) how well they are working Land health sur-veillance systems are designed to provide timely information foraction to improve land productivity and maintain essential ecosys-tem functions In more specific terms land health surveillance hasa number of functions (Table 1)

In public health different types of surveillance systems are usedfor different purposes (Stroup et al 2004) and examples of poten-tial parallel applications in land health can be identified (Table 2)These generally follow the progression of surveillance steps of (i)establishing risks to key health problems as a basis for interven-tion targeting (ii) evaluating interventions and (iii) establishing finaloutcomes Integrated surveillance systems are likely to be of most

Table 1Principal functions of land health surveillance

1 Identify land health problems11 Assess and monitor land health status12 Quantify risk factors of land degradation and sustainable land

management13 Provide early warning of land degradation

2 Establish quantitative objectives for land health promotion21 Specify objectives for prevention of land degradation22 Specify objectives for land rehabilitation

3 Provide information for the design and planning of land managementintervention programmes and resource allocation priorities31 Prioritize intervention areas based on degree and extent of risks to land

health32 Prioritize interventions based on cost-effectiveness analysis33 Spatially target interventions

4 Determine the impact of specific interventions41 Empirically test effectiveness of interventions42 Establish outcomes in terms of reduction in risks and health

improvement5 Identify research service and training needs for different stakeholder groups

ARTICLE IN PRESS


2 KD Shepherd et alAgricultural Systems (2014) ndash

value in the land sector as existing surveillance systems are poorlydeveloped

23 Science principles and system components

Land health surveillance systems deploy the scientific prin-ciples (Table 3) and systems components (Fig 1) of public healthsurveillance The foundation of surveillance is a population based(epidemiological) approach to the study of health problems Proba-bilistic sampling schemes are used in conjunction with standardizedprotocols to establish the prevalence of health problems in popu-lations and to statistically establish associated determinants orrisk factors

Risk is defined (WHO 2002) as a probability of an adverseoutcome (eg probability of soil erosion occurring) and a risk factoras a factor that raises that probability (eg steep slopes) Much effortis put on quantifying and forecasting the trends in the burden ofhealth problems in populations and their risk factors This infor-mation forms the main basis for targeting interventions (Fig 1)

Analysis of the cost-effectiveness of alternative interventions isa key tool for selection of interventions for testing in the real worldThis is done using statistically sound study designs with the use ofcontrols The use of standardized approaches permits combinedanalysis of results from a number of studies (meta-analysis) Theresults of meta-analysis are published and used to design overallstrategies for improving health policy and practice

Surveillance systems engage in communicating results and en-suring their use in the design and evaluation of public programmes(Fig 1) This also involves engaging in the overall design of inter-vention strategies and programmes and evaluation of theeffectiveness of the overall surveillance and response system Thussurveillance forms an integral part of a health system As withany systems approach this is not a strictly linear process andthere are iterations and feedbacks between the components andsteps

3 Applying surveillance principles

31 Sampling populations

Land health surveillance is concerned with the health of popu-lations of land sample units that are known to be representativeof those populations measured by health status indicators The goalis to make inferences from the sample to the population (Cochran1977) The primary sampling unit is a fixed area of land (eg sat-ellite pixel field sampling plot or soil auger hole of fixed area) Wecan also consider fixed volumes of land where soil depth and heightabove the land surface are considered The sample unit size is pref-erably fixed so that prevalence estimates can be expressed on a perunit area basis and scaled and also to avoid comparing units oflargely varying size Where the objective is to assess land healthchanges or interventions on human welfare then individual landusers households or communities may constitute the sampling unit

Spatial hierarchical nesting of sampling is important in ecologybecause different process operate at different scales and there isnormally interest in aggregating results to higher levels of scale(Gallant et al 2008) For example fixed area sample units can benested within climate or agro-ecological zones river basins wa-tersheds districts or countries Stevens and Olsen (2004) discussfurther considerations for sampling natural resources over large areasPopulations are also often stratified on important conditioningfactors before drawing a sample to spread samples to ensurerepresentativity increase precision and compare groups

Table 2Types of surveillance systems in public health and examples of parallels in land health

Surveillance system Purpose in public health Example for land health

Population or Prevalence Provide basic data on prevalence and incidence ofdifferent health problems in populations and their keydeterminants (risk factors)

Establish prevalence of different land degradationproblems (such as deforestation rangeland degradationsoil salinity) in a target area and their associated risks

Determinant or risk factor Obtain data on population distributions of key healthdeterminants (risk factors) and preventive behavioursand monitor changes in risks

Survey and track key risk factors (eg soil conservationpractices land tenure) for major land health problems ina target area

Intervention Gather data on intervention impacts at individual andpopulation levels

Evaluate impacts of individual soil conservationpractices and of sustainable land managementprogrammesprojects on land health

Outcome Collect data on final outcomes in terms of frequency anddistribution of health in a population

Collect data on frequency and distribution of specificland degradation problems that have been targeted forintervention in a target area

Syndromic Use health-related data that precede diagnosis to signala sufficient probability of a case or an outbreak towarrant further public health response

Use drought forecasts and data on areas vulnerable toovergrazing to signal impending rangeland degradationand plan preventive response by the government

Integrated Combine several surveillance objectives into oneintegrated system to help better describe the webs ofcausation that result in health problems

Implement national land health surveillance systemsthat integrate data collection and use on land health riskfactors and intervention impacts

Table 3Surveillance science principles (synthesized from Teutch and Churchill 2000 WHO2002 McQueen and Puska 2003 Brookmeyer and Stroup 2004 Lawson andKleinman 2005)

1 A focus on the health of populations rather than individuals2 Sampling designs are used to make inferences from the sample to the

population3 The frequency of health problems in populations such as prevalence and

incidence is measured4 Standardized protocols are used for data collection5 Case definitions are used to specifically and consistently diagnose health

problems6 Screening tests are used for assigning individuals to cases7 The association between health problems and risk factors is measured and

assessed using statistical (risk quantification) models including significancetesting and confidence interval estimation

8 Spatial and syndromic surveillance are commonly employed to detect spatialpatterns in health problems and in symptoms that commonly co-occur

9 Future needs are projected using mathematical forecasting models to detectpatterns in data collected over time

10 Cost-effectiveness analysis is used to screen alternative interventions11 Intervention impacts are rigorously evaluated using experimental designs

with controls12 Meta-analysis is performed and used as the primary source of information for

design of public policy and health programmes13 Operational surveillance systems are built into everyday health policy and

practice for both design and evaluation

ARTICLE IN PRESS


3KD Shepherd et alAgricultural Systems (2014) ndash

(eg different land use systems) but it is important that the strataare objectively defined

Surveillance commonly uses different types of sampling designProbability (random) sampling designs are used to avoid selectionbias Probability sampling also allows standard errors and confi-dence intervals to be calculated and hypotheses to be tested withoutintroducing additional assumptions A small probability sample canprovide more accurate information than a much larger sample thatsuffers from selection bias (eg Hubbard 2014) although under somecircumstances a small bias may be worth having if it gives a largereduction in variance For example convenience sampling of landor farms close to roads is a common selection bias in land studiesNon-probability designs may be used to take advantage of data thathas been collected for other purposes (eg data from testinglaboratories)

Population-based sampling has rarely been applied in land re-source surveys (McKenzie and Grundy 2008 Sonneveld and Dent2009 Nachtergaele and Petri 2010) The target population the sam-pling units and sampling strata are often not rigorously defined withthe result that there is an inability to generalize results to a widerpopulation or area The lack of sampling frames appears to be ageneral deficiency in many monitoring initiatives in agriculture en-vironment and livelihoods (Shepherd et al 2013) However UNEP(2007) demonstrated population based surveillance within senti-nel sites which was then extended to continental level hierarchicalprobabilistic sampling design in the Africa Soil Information Service(AfSIS 2014) (Fig 2) The fixed-area based sampling frame allowsprevalence estimates to be expressed on a per unit area basis (eg of area affected by a constraint) and at different scales

32 Prevalence and risk factors

A public health risk framework (WHO 2002) can readily beapplied to land health (Fig 3) A key goal of surveillance is to quan-titatively establish association between the occurrence ordevelopment of a health problem and a set of risk factors Preva-lence (or cross-sectional) studies examine the relationship betweenhealth problems and other variables as they exist in a defined pop-ulation at one particular time Prevalence is the number of instances

of a particular health problem in a population at a given time In-cidence is the number of new instances of a particular healthproblem during a given period in a specified population

Risk factors can be biophysical or socioeconomic factors and caninclude behavioural as well as inherent characteristics Protectiveas well as hazardous risk factors are also considered Modifiable riskfactors which are factors that can be controlled or influenced areof most interest as they identify areas for action to overcome a healthproblem They include health behaviours (eg type of land man-agement) health status (eg state of degradation) and policies (egland tenure policy) Risk factors do not necessarily imply a causallink with a health problem For example non-modifiable riskfactors are often important conditioning variables (eg climatic zoneslope)

Distal determinants or risk factors are remote or far in posi-tion time or resemblance to the health problems whereas proximaland process determinants are nearest in time andor distance to theoccurrence of a health problem and more likely to have well definedcausal linkage to the problem (Fig 3) Preventive measures focuson the causes while treatment (rehabilitation) focuses on theoutcomes

Although risk factors can be postulated from prevalence studiesincidence data are required to confirm a causal link This requires long-term monitoring using consistent and standardized sampling framesand measurement protocols Because area-based probabilistic sam-pling is rarely used in land assessment few unbiased prevalence andstill less incidence data exist (eg on the extent of degraded range-land or of strong soil acidity in a country and their trends) and errorson area-based estimates are rarely known

While avoiding major risks is the key to prevention reversingrisks to land health is also often the prime target of rehabilitationtreatments (eg reducing overgrazing to rehabilitate degraded pastureland replacing soil nutrients extracted in harvests to replenish soilfertility reversing woody cover depletion in landscapes to restorelandscape ecological functions) However current evidence on riskfactors for land degradation is based largely on case studies andexpert opinion derived from a limited number of observations (egGeist and Lambin 2004) and there is a need for a more quantita-tive approach

Fig 1 Key components of a health surveillance and response system Connecting lines are omitted between components to convey the iterative nature of the surveillanceprocess This is the authorsrsquo interpretation based on a wide range of public health surveillance texts

ARTICLE IN PRESS



Once key determinants of population health have been quanti-tatively established efforts are focused on surveillance of the keyrisk factors usually through low cost surveys (Bonita et al 2003)Behavioural risk factor surveillance systems using low costmeasurements need to be established for land health (eg moni-toring of area under good agricultural practice degree of perennialland cover)

As with human health problems land health problems oftenoccur in association with one another linked to a common set ofcauses and can then be defined as syndromes (WBGU (GermanAdvisory Council on Global Change) 1995 Petschel-Held et al 1999Geist and Lambin 2004 Reynolds et al 2007) (Table 4) In medi-cine and psychology the term syndrome refers to the associationof several clinically recognizable features signs or characteristicsthat often occur together so that the presence of one feature pro-vides early warning to the presence of the others For examplechange in perennialannual grass composition could provide earlywarning of loss of permanent ground cover leading to soilphysical degradation processes in rangelands

33 Dosendasheffect relationships

The relationship between the probability of a health problem oc-curring (risk) and the level of a risk factor (the dosendasheffectrelationship) is critical for planning of control policy (Rose 2008)

Exposure may have little adverse effect until it reaches a certain highlevel beyond which risk increases rapidly (Fig 4a) This shape ofrelationship produces a tipping point at moderate to high levels ofexposure and this behaviour is hypothesized for many ecosystemdegradation problems (MA 2005b) If much of the burden of a healthproblem is concentrated within an identifiable group with high ex-posure then this presents a clear target for action The pattern alsoimplies that risk is negligible at low levels of exposure and can beignored On the other hand a linear dosendasheffect relationship overthe whole range of exposure (Fig 4b) implies that any level of therisk factor should be regarded as hazardous (Rose 2008) In thiscase risk should be reduced as far as economic and social costsallow

A curvilinear relationship (Fig 4c) with shallow slope at low ex-posure is a common pattern for many human health problems (egheart disease in relation to blood pressure levels) and lends itselfto screening policies that are hinged on a critical limit at the pointwhere the slope begins to increase sharply (Rose 2008) Here a dualstrategy is applicable giving some priority to treating high-risksample units while also reducing exposure for units at low riskas a preventive measure The pattern of Fig 4d requires that therisk factor is kept at moderate levels to minimize risk Thissituation would apply for example to managing soil pH at moder-ate levels to prevent nutrient imbalances for plant growth that occurat either high or low pH Establishing dosendasheffect relationships for

Fig 2 Probability-based sampling scheme used in the Africa Soil Information Service (httpwwwafricasoilsnet) Left Koumlppen-Geiger climate zones in Africa clipped tothe AfSIS project area Yellow circles (dots) show the location of the 60 AfSIS sentinel sites Top right AfSIS sentinel site (blue dots are sampling plots) of standard 10 times 10 kmsize Bottom right Sampling plot consisting of three 100 m2 subplot within a 1000 m2 plot Source Varinggen et al 2010

ARTICLE IN PRESS



major land health problems should be a priority for designingintervention strategies and is a key objective for land healthsurveillance

34 Case definitions and diagnostic screening

Clear case definitions are important for ensuring validity of com-parisons across times and sites in surveillance studies screening ofhigh-risk sub-populations and guiding rehabilitative actions (Table 5)A case definition is a set of diagnostic criteria that must be ful-filled in order to identify a sample unit as a case of a particularproblem (Last 2001) Ideally it should include quantifiable andexplicit criteria where possible (Thacker and Birkhead 2002) and

be easily applied (Gregg 2002) Case definitions may be based onobservations laboratory criteria a combination of both or a scoringsystem with points for each criterion that matches the features ofthe health problem (Last 2001) Case definitions must include theunit of surveillance and may be further qualified by various otherfactors such as a sub-population time and place A compendiumof rigorous case definitions (eg WHO 1999) does not currently existfor different land health problems However diagnostic criteria orattributes are available in several fields that could serve as a start-ing point For example diagnostic criteria are used in assessmentof rangeland health (eg Herrick et al 2010) forest health (Schreuderet al 2004) land quality (eg FAO 1976) and soil constraints(Sanchez et al 2003)

Fig 3 Framework for assessment of risk factors associated with land healthAdapted form WHO (2002) and Murray et al (2003)

Table 4Examples of land health syndromes and associated risk factors

Syndrome Associated individualproblems or signs

Proximal risk factors Distal risk factors

Rangeland degradation Displacement of perennial grasses by annualgrassesShrub encroachmentSoil structural degradationIncreased surface run-off and soil erosionReduced soil water holding capacityReduced soil carbon content

OvergrazingWood extractionSensitive soils

Drought climate changeRestricted access to traditionalreserve grazing areasLack of de-stocking policy

Soil fertility decline in crop land Poor crop yieldsVisual nutrient deficiency symptoms in cropsSoil fertility tests below reference valuesIndicator plant species

Low input useContinuous croppingPoor soil conservation

PovertyMarket accessInput supply chainsOff-farm incomeCultural attitudesEducation

Deforestation Decline in indigenous forest coverConversion of forest to agricultural landSelective logging

RoadsCommercial agricultureProximity to agricultural land

Forest protection policiesDemand for agricultural productsPopulation growth

Soil heavy metal pollution Stunted plant growthElevated soil and plant test valuesHuman health problems

Proximity to industrial ormining enterprisesIndustrial waste emissions controlIndustrial waste disposal practices

Government and industry pollutionemission and waste disposal policiesPublic awarenessRecycling incentives

ARTICLE IN PRESS



Health problems generally exist as a continuum of severitybut for practical reasons the diagnostic continuum is oftendichotomized into lsquocasesrsquo and lsquonon-casesrsquo or lsquoaffectedrsquo and lsquonon-affectedrsquo to aid decisions on interventions Where there is uncertaintydefinitions can classify cases by levels of probability (eg uncon-firmed case possibleprobable case or confirmed case) For examplefield signs may indicate a probable case which is confirmed by alaboratory test

In clinical investigations test results are compared against a ref-erence population and may be made more specific using conditioningfactors (Jones and Payne 1997) Clinical trials for example dosendasheffect trials may also be used to establish reference values Insurveillance studies however the lsquonormalrsquo case is rarely known and

most attention is directed to detect passing limits based on ob-served or expected patterns for example relative to populationaverage values of indicators or expected prevalence (Lawson andKleinman 2005) There has generally been a lack of consistent mea-surement methods in international land degradation studies (Linigeret al 2008 FAO-LADA 2009a 2009b Sonneveld and Dent 2009)and improving case definitions and reference levels for land healthwill require systematic measurement and monitoring to establishexpected patterns and responses to conditioning factors (dosendasheffect relationships) For example soil organic matter is almostuniversally considered to be one of the most useful indicators ofsoil health but what constitutes an adequate level of soil organicmatter to maintain essential ecosystem services conditioned on keyfactors is still difficult to define (Sanchez et al 2003)

Case definitions must be accompanied by screening tests thatwill be used to assign samples as affected (case) or non-affected (non-case) or alternatively provide a probability of being a case Frontlinescreening tests need to be rapid low cost but repeatable methodsthat can be applied to the large sample numbers typically re-quired for surveillance Error associated with the screening test mayproduce false positives (or false negatives) and samples screenedas affected may be subjected to further tests or bioassays to confirma diagnosis (confirmed case)

For aboveground (vegetation) attributes of land health remotesensing is an example of a potential low cost screening tool (egRoumlder and Hill 2009) and well suited to the multiscale fixed areasampling methods with the spatial resolution of the pixel sizes ofdifferent sensors ranging from hundreds of metres to less than ametre However the lack of systematic ground monitoring cur-rently poses serious limits on the interpretation and validation ofremote sensing information We expect far wider variationin land variables than in indicators of human health and the ref-erenced population will have to be more narrowly definedand based on conditioning factors such as climate zone and soilmineralogy

Infrared spectroscopy a proximal tool closely related to remotesensing has been proposed as a screening tool for assessing soil andplant health within a surveillance framework (Shepherd and Walsh2002 2007) In particular the method predicts key soil functionalattributes that respond only slowly to management changes suchas soil organic carbon and cation exchange capacity and is gener-ally more repeatable than conventional soil and plant analyticaltechniques (Shepherd et al 2005)

35 Forecasting

Forecasting is a method used to predict future events using math-ematical models to detect patterns in data collected over timeProjecting the future burden of health problems based on popula-tion distributions of problems and risk factors and populationdynamics is a critical tool for intervention targeting in public health(WHO 2002) Recent advances in spatio-temporal modelling (Lawsonand Kleinman 2005) will be critical for forecasting and scenario anal-ysis in land health problems especially for projecting thedevelopment of keys risk factors Use of stochastic modelling whichexplicitly represents uncertainties and permits value of informa-tion analysis is a promising approach for dealing with problems withsparse data (eg Briggs et al 2006 Hubbard 2014)

4 Intervention targeting

An intervention is taken to mean any promotive preventive cu-rative or rehabilitative action where the primary intent is tosafeguard or improve health (WHO 2002) In our case health is theinherent capacity of land to sustain delivery of essential ecosys-tem services and our primary intention is to improve land health

Fig 4 Different relationships between level of risk of a health problem and levelof exposure The shape of the relationship has important implications for controlpolicyModified from Rose (2008)

Table 5Example case definition for soil salinity

Etiology Accumulation of soluble salts in soils to a level thatadversely affects the growth of most crop plants Saltaccumulation is a result of natural or human actionHuman causes include sub-optimal irrigation waterquality and application practices and clearance ofwoody vegetation in dryland areas with shallow watertables

Clinical signs White crust visible on the soil surface or on newlyexposed surfaces of soil excavations Inhibited seedgermination patchy and stunted plant growth withsmall leaves of deep blue-green colour

Laboratory criteriafor diagnosis

Electrical conductivity of saturated soil extract at25 degC gt 04 S m-1 within 1 m of soil surface (Sanchezet al 2003)

Case classification Probable a clinically compatible condition withetiological evidence and a history of occurrence (egsalt crust in an irrigation scheme that has beenpreviously diagnosed with soil salinity)Confirmed a clinically compatible condition that islaboratory confirmed

Comment Clinical signs may be absent Salinity may bediagnosed without laboratory confirmation if theclinical and etiological evidence is overwhelming (egthick salt crusts on soil surface in an irrigationscheme) Soil salinity commonly co-occurs with soilsodicity in a saline-sodic soil syndrome

ARTICLE IN PRESS



ultimately for improved human well-being by any of the aboveactions Interventions can be categorized into those that arepopulation-based which seek to reduce risks in the population asa whole (reduce erosion risk on all land units) and individual-based which target individuals within the population (egthose identified as severely degraded or at high risk) Inter-vention targeting and priority setting includes the combineduse of two key approaches risk assessment and cost-effectivenessanalysis

Application of a health risk assessment framework (WHO 2002)to land would provide estimates for1 The attributable burden of land health problems at the present

time ie the proportion of land health problems that results frompast exposure

2 The avoidable burden over the next several decades for a stan-dardized range of reductions in risk factors ie the proportionof future land health problems that is avoidable if current andfuture exposure levels are reduced to those specified by somealternative or counterfactual distributionThe type of surveillance data required to estimate the burden

attributable to specific risk factors are risk factor levels dosendasheffect relationships and current and projected land health problemprevalence and risk factors In addition there is need to express dif-ferent land health outcomes (eg reduction in specific ecosystemservices) on a common scale equivalent to the disability-adjustedlife year (Murray 1996) Risk factors for which a given level of re-duction provides the greatest reduction in burden are then obviouscandidates for intervention Where several risk factors jointly con-tribute to health problems strategies that simultaneously tackle theset of risk factors may be accorded high priority An outcomemeasure for land could express the gap between the actual healthstatus of a population of land units and some reference status interms of degree of functional impairment As an intermediate stepthe effect of interventions could be assessed in terms of relative re-duction in key behavioural risk factors (eg soil cover) which areusually quicker and cheaper to monitor than health status This isclearly an area for further research

Once risks are well established the next step is to analyse inter-ventions in terms of what interventions are available their effectivenessin reducing risks either singly or in combination and what resourcesare required to implement them Cost-effectiveness is considered to bethe clearest simple way to promote value for money in health (Musgroveand Fox-Rushby 2006) and it is this information that is precisely lackingin order to make the case for land health interventions to ministriesof finance Lessons may be gleaned from the CHOICE (CHOosing In-

terventions that are Cost-Effective) (Murray et al 2000 Tan-Torres Edejeret al 2003) For prioritizing interventions in highly uncertain data poorsituations typical of developing countries Bayesian theory combinedwith value of information analysis offers a promising approach (egBriggs et al 2006 Hubbard 2014)

41 Intervention evaluation

411 Study designsSurveillance is an essential tool for measuring the real world

impact of interventions and public policy (Bonita et al 2003)Because health surveillance is by definition oriented towards actionthe surveillance system itself is also evaluated for example in termsof whether surveillance information has been communicated to thosewho need to know and whether the information has had abeneficial impact on the health problem (Teutsch and Churchill2000)

Evaluation of intervention efficacy or effectiveness inthe public health sector invariably employs study designsto allow scientifically rigorous evaluation (Katz 2010) Designs varyin the strength of evidence they provide for causation and to caterfor different circumstances such as ease of accessing sample unitsand ethical considerations (Table 6) There are also manyoptions with respect to randomized vs non-randomized studiestypes of randomization (eg stratified clustered unequal alloca-tion) and nesting as well as a range of analytical tools to dealwith threats to validity (Katz 2010 Lipsey and Cordray 2000)With intelligent design and analysis some evidence can be gainedon intervention effectiveness in most situations althoughdifficulties increase for interventions operating on more distalfactors

Measuring intervention impacts on problem incidence in a realworld setting often necessitates long-term studies (ten or moreyears) well beyond the life of a typical donor-funded project Suchstudies may require for example multiple measurements of theoutcome prior to and after the intervention However early resultscan be achieved by identifying proximal markers that are highly pre-dictive of the final outcome variables of interest providing moreoutcomes in shorter follow-up times (Katz 2010) The focus shouldalso be on intermediate outcome variables that have the largestuncertainty (Hubbard 2014)

There is often opportunity for superimposing intervention eval-uation studies on population surveillance studies such as conductingcase control studies to evaluate interventions after the fact Forexample the effect of a soil conservation intervention on land deg-

Table 6Experimental designs for intervention evaluation (synthesized from Schlesselman 1982 and Katz 2010)

Design type Features Principal use

Pre-intervention vs post-intervention withone groupLongitudinal cohorta Repeatedly assess the same sample units over time To assess intervention impact on selected individuals

over time when random selection within cohorts isunethical or impractical

Serial cross-sectional Assess different sample units from the same populationover time

To assess intervention impact on the population as awhole assess changes in practice patterns

Post-intervention vs pre-intervention withcomparison group

As above but follows a group exposed to theintervention over time and a group without theintervention

As above but provides stronger causal evidence ofintervention effect

Post-intervention only (retrospective studies) Looks backwards to see if an association exists betweena past exposure and a health outcome

When no pre-intervention measurement was made egwhere an intervention is a law or policy

Cohort Compares problem frequency in groups that are similarin all respects except exposure to the intervention

Allows study of multiple potential effects of anintervention both risks and benefits

Case-control study Compares groups with and without a specific healthproblem and assesses frequency of interventionexposure in each group

Especially useful for problems that have very lowprevalence

a A cohort is a group of sample units that share a common characteristic or experience within a defined period

ARTICLE IN PRESS



radation could be evaluated by a case control study whereby sampleunits with and without signs of soil erosion are identified withina surveillance study and then the frequency of implementation ofthe intervention in each group is assessed Natural resource man-agement projects should be required to build in evaluation plansthat statistically evaluate the interventions being applied Reliablelearning would be accelerated if governments and donors fundedfewer but better designed projects

412 Quality standards and meta-analysisIn the public health sector studies involving human subjects

require a review of the whole protocol by an institutional reviewboard which checks the quality and soundness of the design andmeasurement protocol in addition to ensuring the rights of humansubjects are protected Testing of land management interventionsin almost all cases involve human subjects as well Some prin-ciples for ethics in public health surveillance (Heilig and Sweeney2010) may be applicable to land health surveillance For exampledata should be stored and managed in a physically and technolog-ically secure environment and the access to surveillance data onidentifiable individuals controlled However the general prin-ciples of open access data and publication of all results derived formthe data should be upheld

Public health policy and practice is strongly guided by pub-lished meta-analyses of intervention evaluation trials Meta-analysis refers to the statistical analysis of data from separate butsimilar studies leading to a quantitative summary of the pooledresults (Stangl and Berry 2000 Last 2001) Although there aremethods for combining data from heterogeneously reported studiesmeta-analysis is only feasible when scientifically rigorous proto-cols and reasonably consistent measurement and reportingapproaches have been used in different studies Sadly this is not oftenthe case in land-related studies (eg Rudel 2008) For example inthe public health sector recommended standards are available forreporting results from randomized (Moher et al 2001) and non-randomized (Des Jarlais et al 2004) trials to facilitate meta-analysis and guard against errors and biases The CochraneCollaboration (Higgins and Green 2011) provides a good model thatcould be applied to land health and the principles are already beingapplied to evidence-based conservation (Centre for Evidence-BasedConservation 2010)

5 Using and communicating surveillance information

51 Users and uses of surveillance information

Effective surveillance systems do not simply make the assumptionthat data will be used but purposively plan use of surveillance find-ings Groups with an interest in surveillance data and information canbe categorized into stakeholders a subset of direct users and those whohave principal responsibility for defining and establishing the surveil-lance system (Ottoson and Wilson 2003) Stakeholders and users mayspan a number of organizations (Table 7) including policy research ac-ademic practitioners community groups private sector and mediaSuccessful surveillance systems find ways to manage the collabora-tion among these groups and ensure enduring partnerships Althoughthere has to be some centralized coordinating body or design team in-volvement of target audiences in the design of surveillance systems iscritical for effective utilization of results (eg Mitchell et al 2006) Thereare challenges due to fact that decisions that affect land health man-agement are spread among different ministries (eg forestry agriculturelivestock water environment lands) and that agricultural forestry andenvironmental extension infrastructures are mostly very weak indeveloping countries

52 Communicating results

Public health surveillance has a long history of communicatingsurveillance results (Goodman et al 2000 Churchill 2000 WHO2002 Remmington and Nelson 2010) and the lessons are equallyapplicable to communicating results of land health surveillance(WHO 2002 2004) The communication process which basicallytakes facts and packages them to convey meaning involves severalbasic steps (i) interpret data into information (ii) translate infor-mation into actionable messages (iii) identify sender target audienceand communication channel (iv) communicate the message and(v) evaluate the impact (Goodman et al 2000 Remmington andNelson 2010)

Involvement of target audiences in the design of the dissemi-nation process can provide valuable early feedback on relevance andeffectiveness making the cycle more of an iterative than a linearprocess (eg Mitchell et al 2006) There are similar experiences inagricultural extension (Roling and Wagemakers 1998) and envi-ronmental management (Lee 1993) In many of these contexts anldquoactionable messagerdquo forms the starting point of the discussionHowever how best to communicate surveillance and research find-ings in developing countries still needs much local research

The increasing use of digital means for gathering and storing in-formation coupled with the use of Internet and mobile phonetechnology is transforming access to information in remote ruralareas and educational establishments in developing countries Thispresents a significant opportunity to catalyze progress through re-use and re-purposing of data and is changing how individuals interactand collaborate Information technology will thus play a pivotal rolein land health surveillance systems in developing countries

53 Intervention strategies

531 Prevention vs treatment strategiesIn public health much scientific effort and most health re-

sources have been directed towards treating health problems ratherthan preventing them (WHO 2002) but recent efforts are focusingon reducing risks to health as the key to prevention This ap-proach is based on one of the most fundamental axioms in preventivemedicine ndash that a larger number of people exposed to a small riskmay generate many more cases than a small number exposed to ahigh risk (Rose 2008) Risk typically increases across the range ofa risk factor so that population-based strategies that seek to shiftthe whole distribution of risk factors have the potential to controlpopulation incidence (Fig 5) Just as in human health a large partof the burden of land degradation in many developing countriescould result from a small number (5 to 10) of risk factors so thatmajor impacts in reducing land degradation would come from imple-menting strategies to reduce those risks For example strategies thatseek to maintain moderate levels of soil cover over the whole of acountry may be more effective in reducing national soil erosion thanstrategies that seek to increase soil cover over only small areas athigh risk or rehabilitate already degraded areas

532 Population-wide vs high-risk preventive strategiesThe choice of preventive strategy between population-wide in-

tervention or targeting only high-risk individuals will depend onthe population distribution of risk factors for key land health syn-dromes and the shape of the dosendasheffect relationships Combiningthese two pieces of information provides information on the the-oretical impact of a specified reduction in a given risk factor on thetotal burden of a health problem and allows assessment of how thebenefit is shared among the different levels of risk exposure (Rose2008) Cost effectiveness will depend on (i) the prevalence of high-risk cases in the population and (ii) the costs of identifying themrelative to the cost of the population intervention

ARTICLE IN PRESS



Population-wide strategies are generally more effective thantargeting the most affected in situations where the population meanpredicts the prevalence of cases and risk is widely diffused throughthe population In these situations even a small shift in distribu-tions may have a large effect on the number of individuals fallingin the high vulnerable tail of the distribution (Rose 2008)Population-wide strategies present surveillance challenges in thatsmall changes in risk need to be detected over large populations(eg of land units) and it may be difficult to demonstrate interven-tion impacts in a convincing way However a small increase in landproductivity or reduction in erosion over large land areas can ag-gregate up to be very significant at river basin or national levelsPopulation-based interventions may be less attractive to minis-tries and the public at first sight because although they maybring maximum benefit to the population as a whole theyprovide only very small benefits to each individual in the short-term a phenomenon that Rose (2008) terms the lsquopreventionparadoxrsquo

Rose (2008) refers to the high-risk preventive strategy as ldquoa tar-geted rescue operation for vulnerable individualsrdquo If the problemis confined to an easily identifiable minority of the population andcan be successfully controlled in isolation of the wider popula-tion then the approach is adequate For example this is more likely

to apply to prevention of point source pollution problems whereasnon-point source and diffuse problems require a populationapproach Currently many land management intervention strate-gies are in fact targeted at rehabilitation of highly degraded landsuch as erosion hot spots and saline soils in irrigated area despitethe fact that efforts to rehabilitate ecosystems are usually onlypartially successful and incur very high cost Preventive actions areoften high-risk individual interventions such as soil conservationprogrammes in steeply sloping areas or biodiversity conservationstrategies confined to biodiversity hot spots and protected areasCost-effectiveness analysis of alternative intervention strategies isneeded to design optimal strategies or mixes of strategies

Risk prevention strategies require a change in ownershipof responsibility for tackling health risks away from individuals atthe extremes towards governments and ministries tacklingpopulation-wide risk levels (WHO 2002) This is because individ-ual behaviour depends substantially on the economic and socialstructures for which governments are responsible (Rose 2008) Hencecontrolling distal risks to health may be more effective than con-trolling proximal risks (WHO 2002 Rose 2008) and the same islikely to apply to many land health problems For example tack-ling land tenure and pricing policies may be more effective inreducing over-grazing at the overall population level than imple-

Table 7Examples of types of land health surveillance data or findings and examples of their uses

Audience Information products Uses

Farmers andpastoralistscommunity-basedorganizations

bull Prevalence data and maps of land healthconstraints in a locality

bull Proximal behavioural risk factors forland degradation

bull Cost-effectiveness analysis of specificland management interventions

bull Enhance individual and community knowledge of predominant land healthconstraints and hazards in the locality to help mobilize action

bull Guide screening of appropriate management interventions for testing byindividual land users or communities

bull Guide good preventive practice by individuals and communitiesbull Enhance individual and community knowledge on trade-offs and risks

associated with different management interventionsLocal

governmentplannersdevelopmentassistanceorganizations

bull Prevalence and incidence data and mapsof land health constraints in a district

bull Information on proximal behaviouralrisk factors for land degradation

bull Maps targeting intervention strategiesand priorities in relation to constraints

bull Early warning of land degradationoutbreaks

bull Evidence-based evaluation of performance andrisks for specific land management interventions

bull Standardized operational norms or case definitionsand screening tests for assessing goodpoor land health

bull Plan public information and awareness campaigns on prevalent land healthproblems best preventive practice and rehabilitation interventions

bull Knowledge of land health status in a district assess needs of differentgroups and areas

bull Plan land health intervention programmes target priority areas define andmonitor measurable objectives and targets

bull Take early action in relation to new land degradation outbreaksbull Prepare funding proposals to central government and donors based on

evidence of problemsbull Adjust surveillance programmes in light of user feedback evaluation of

interventions and new emerging threats

Planning andfinanceministries

bull Identification of priority risk factors for preventionof land degradation at national level

bull Information on time trends in land healthand associated risk factors

bull Reliable and comparable estimates of theburden of land degradation in relation tofactors such as poverty region

bull Cost-effectiveness analysis to identify highmedium and low priority interventions toprevent or reduce land health risks

bull Evaluation of targeting strategies population-wideversus high-risk individuals distal versus proximalrisks primary versus secondary preventionprevention vs rehabilitation

bull Formulate risk prevention and rehabilitation policies for land healthmanagement and set priorities and targets

bull Formulate concrete and specific action plans and monitor impactsbull Evidence-based reporting of progress on land health management in

fulfilment of commitments to UN and other conventions and internationalagreements (eg UNCCD UNCBD MDGs)

bull Improve public awareness and understanding of risks to land healthbull Identify opportunities for combining risk reduction strategies including

with other sectors (eg human health)bull Identify priorities for investments in land health surveillance systems to

strengthen the scientific evidence base

UN bodiesconventions

bull Scientifically sound multiscale data and informationon land health status and risks over time

bull Priority research and training needs in landhealth care and surveillance

bull Information on priority intervention strategiesand their cost-effectiveness

bull Scientifically credible and systematic assessments of land health status andrisks at different scales including early warning

bull Evidence-based information on land health intervention priorities as a basisfor policy development advocacy science coordination and capacity building

bull Scientifically credible monitoring and impact assessment with respect toachieving goals of international conventions related to land management

Internationaldonors

bull Reliable information on intervention priorities andstrategies for land health management includingin relation to other sectors (eg poverty reductionstrategies food security human development)

bull Information on research and training needsin land health management and surveillance

bull Formulation of development assistance plans and priorities related to landmanagement based on scientifically sound data and information

bull Formulation of well-targeted capacity building assistance programmes forland health care

ARTICLE IN PRESS



menting grazing management interventions through individualinterventions aimed at high-risk populations

533 Other considerations for strategyThere are a number of other considerations for intervention

strategy summarized below1 Consider the interactions among several risk factors associated

with the same health problem One risk factor (eg acceleratedrun-off as a risk to soil erosion) may have less importance at lowlevels than at high of another risk factor (eg soils with low orhigh erodibility)

2 Take advantage of the many synergies between land health andhuman health interventions For example both share a numberof common distal risk factors such as education levels andpoverty therefore joint interventions aimed at distal risks maybe mutually reinforcing and cumulative evidence from bothsectors may provide more pressure to implement policy changethan either sector operating alone

3 Because human health and land health are exposed to somecommon highly uncertain risks (such as climate change genet-ically modified organisms toxic chemicals and outbreaks of rarediseases) resource allocation decisions need to be made betweenmanaging these highly uncertain risks using the precaution-ary principle as a guide and tackling well known risks (WHO2002)

4 There is a need to consider risk transitions in developing coun-tries For example land health problems associated with lowagricultural input use are transitioning to pollution problems fromover-use of inputs and breakdown of biological cycles (UNEP2007) and surveillance systems will need to monitor both setsof risk factors

5 Frameworks for planning and evaluation of evidence-based in-terventions and policy interventions could help increase rigourin land health programmes (eg RE-AIM King et al 2010) asmany of the factors and considerations for planning and pro-moting land health intervention strategies are similar to thoseencountered in public health

54 Implementing national surveillance systems

For a surveillance system to be effective the surveillance prin-ciples and components must be integrated and put into operationas part of regular health policy and practice (Teutsch and Churchill2000) The World Health Organization has a number of guidelineson making health surveillance work (WHO 2001) National landhealth surveillance system could draw heavily on existing re-sources and have a relatively small core staff (Fig 6) We estimatethat a land hand health surveillance system could operate in manyAfrican countries including field survey and laboratory costs at lessthan 005 of the public expenditure The surveillance unit couldbe housed in one specialized division (eg agricultural research) ofa ministry which would elicit the cooperation of others For exampleYoung (1991 1998) suggested a new role for soil and resource surveydepartments in soil monitoring whose traditional role has de-clined since basic reconnaissance surveys have been completed

The surveillance unit must have the expertise to fulfil several crit-ical functions (Fig 6) There will be a need for specialized expertiseto harness advances in earth observation using remote sensing fromspace in the field and on the laboratory bench Strong biostatis-tics support is needed to harness new statistical approaches forhandling hierarchical spatial problems and making probabilistic in-ferences There is also a role to uphold scientific rigour and theconsistent application of standardized measurement protocols acrossstudies and regions so that data can be coherently analysed atdifferent levels of spatial scale and over time For example a cen-tralized soil laboratory would also be required to provide consistentreference analysis to support networks of new low cost spectro-scopic techniques for soil analysis (Shepherd and Walsh 2007) Thereis need for revival of dedicated field teams capable of collectingconsistent data in remote rural areas Specialized communica-tions expertise will be required to ensure surveillance data andinformation is transformed into actionable messages and reachestarget audiences Finally overall coordination and evaluation of thesurveillance programme will be required These sets of skills haveimportant implications for the development of university curricu-la and capacity strengthening in developing countries (Swift andShepherd 2007)

Fig 5 Reducing the average level of a risk factor in a population (a) shown for meanvalues of 50 45 40 and 35 units dramatically reduces the proportion of the pop-ulation that is at high risk (gt60 units) (b) Based on Rose (2008)

ARTICLE IN PRESS



There should be links with surveillance approaches in othersectors (Fig 6) whereby land health provides a set of risk factorsfor problems such as freshwater and marine ecosystem healthatmospheric pollution and human wellbeinghealth This is the onlyrigorous way to assess impacts of land degradation on other sectorsand human wellbeing Equally other sectors could provide risk factordata for land health surveillance and much efficiency could be gainedthrough coordinated surveillance systems For example there isalready considerable interest work on assessing environmental risksto human health through environmental health surveillance (WHO2002)

Justifications for funding are similar to those for public healthsurveillance (Rein 2010) land health surveillance information isa non-rival public good with multiple primary secondary and ter-tiary uses and users Everyone can use the information freely withoutdiminishing its value to any individual user Hence making thedata more broadly accessible increases the value of surveillancesystems

An international scientific and technical support unit could gen-erate large efficiencies by providing advisory and training servicesto backstop national programmes This unit would also play an im-portant role for further research on surveillance problems and thedevelopment and dissemination of standardized protocols as wellas fulfil the role of ethical and quality reviews of surveillance pro-tocols An international research body such as the CGIAR consortiumwould best house this unit Finally there is a role for an interna-tional body such as the United Nations to foster and coordinatesurveillance activities across countries and regions and provide ad-vocacy and large-scale capacity building similar to the role that WHOplays in global public health surveillance

These are positive developments in uptake of land health sur-veillance principles Building on earlier developments (ICRAF 2002)and testing (UNEP 2007) the principles are being applied at sub-Saharan Africa scale in the Africa Soil Information Service (AfSIS2014) and at national level by the Ethiopia Soil Information System

(EthioSIS 2014) as well as being deployed by the CGIAR insustainable land management projects and sentinel landscapes

6 Conclusions

The results of this review can be synthesized into a set of pos-tulates to help guide further research and action in this area

1 Application of surveillance and response principles could ac-celerate progress in tackling major global problems relatedto land health and management especially in developingcountries

2 Standardized methods provide data that are more useful forplanning successful interventions than locally adaptedmethods

3 Risk factors are important over large enough areas and broadenough ranges of contexts to make broad scale surveillanceand intervention planning useful

4 A large part of the burden of land degradation in many de-veloping countries results from a small number (5 to 10) ofrisk factors

5 Interventions based on systematic surveillance will be morecost-effective than alternatives

6 For diffuse land degradation problems (such as loss of bio-diversity soil fertility depletion and soil erosion) populationlevel preventive interventions will be more cost-effective thanthose focusing on high risk sub-groups

7 Wider use of study designs in sustainable land manage-ment projects will reduce the time required to identify cost-effective interventions

8 Better evidence on cost-effectiveness and impact of inter-ventions will increase government and donor support forpromising interventions

9 Guidelines and standards for the design and consistent re-porting of results of intervention studies will facilitate meta-analysis for policy guidance

Fig 6 Suggested institutional arrangement for a national land health surveillance system showing examples of linkages to government departments other surveillancesystems and an international support unit

ARTICLE IN PRESS



10 National land health surveillance systems could be estab-lished at relatively low cost by re-aligning existing resourcesand capitalizing on new technology

11 An international scientific and technical support unit couldefficiently build capacity in surveillance systems improve dataand information quality and facilitate international meta-analysis

12 An international advocacy body such as the United Nationscould effectively accelerate adoption of national surveil-lance systems and promote global standards

Priorities for further development of research on land healthsurveillance and response include1 Systematic review of land health problems case definitions risk

factors and dosendasheffect relationships2 Protocols for measuring socioeconomic determinants of land

health with emphasis on behavioural risk factors3 A generalized framework for risk assessment and cost-

effectiveness analysis of land health interventions4 Guidelines and example applications for use of statistically

rigorous study designs in land management projects and stan-dardized reporting of results for use in meta-analysis

Acknowledgements

We thank Richard Coe for helpful comments on the manu-script The following agencies funded various aspects of this workover a number of years the Rockefeller Foundation the Bill ampMelinda Gates Foundation the Alliance for a Green Revolution inAfrica the World Bank the Government of Ireland Government ofUK (DFID) Government of Canada Government of Norway Gov-ernment of Finland Government of Switzerland and Governmentof Germany and the CGIAR Research Program on Water Land andEcosystems These funding agencies had no involvement in thewriting of this paper or in the decision to submit the article forpublication

References

AfSIS 2014 The Africa soil information service lthttpwwwafricasoilsnetgt(accessed 20141001)

Bonita R Winkelmann R Douglas KA de Courten M 2003 The WHO stepwiseapproach to surveillance (steps) on non-communicable disease risk factors InMcQueen DV Puska P (Eds) Global Behavioural Risk Factor SurveillanceKluwer AcademicPlenum Publishers New York pp 9ndash22

Briggs A Claxton K Sculpher M 2006 Decision Modelling for Health EconomicEvaluation Oxford University Press Oxford UK

Brookmeyer R Stroup DF (Eds) 2004 Monitoring the Health of PopulationsStatistical Principles and Methods for Public Health Surveillance Oxford UniversityPress Oxford UK

Centre for Evidence-Based Conservation 2010 Guidelines for systematic review inenvironmental management Version 40 Environmental evidence lthttpwwwenvironmentalevidenceorgAuthorshtmgt (accessed 20141001)

Churchill RE 2000 Using surveillance information in communications marketingand advocacy In Teutsch SM Churchill RE (Eds) Principles and Practice ofPublic Health Surveillance Oxford University Press New York pp 364ndash389

Cochran WG 1977 Sampling Techniques 3rd ed John Wiley and Sons New YorkDes Jarlais DC Lyles C Crepaz N the Trend Group 2004 Improving the reporting

quality of nonrandomized evaluations of behavioral and public healthinterventions the TREND statement Am J Public Health 94 361ndash366

Eswaran H Lal R Reich PF 2001 Land degradation an overview In Bridges EMHannam ID Oldeman LR Pening de Vries FWT Scherr SJ SompatpanitS (Eds) Responses to Land Degradation Proc 2nd International Conference onLand Degradation and Desertification Khon Kaen Thailand Oxford Press NewDelhi India pp 20ndash35

EthioSIS 2014 Ethiopia Soil Information System Ethiopian AgriculturalTransformation Agency lthttpwwwatagovetprojectsethiopian-soil-information-system-ethiosisgt (accessed 20140623)

FAO 1976 A Framework for Land Evaluation Soils Bulletin 32 FAO RomeFAO-LADA 2009a Field Manual for Local Level Land Degradation Assessment in

Drylands LADA-L Part 1 Methodological Approach Planning and Analysis FAORome

FAO-LADA 2009b Field Manual for Local Level Land Degradation Assessment inDrylands LADA-L Part 2 Local Assessment Tools and Methods for FieldworkFAO Rome

Gallant JC McKenzie NJ McBratney AB 2008 Scale In McKenzie NJGrundy MJ Webster R Ringrose-Voase AJ (Eds) Guidelines for SurveyingSoil and Land Resources 2nd ed CSIRO Publishing Collingwood Australia pp27ndash43

Geist HJ Lambin EF 2004 Dynamic causal patterns of desertification BioSience54 817ndash829

Gisladottir G Stocking M 2005 Land degradation control and its globalenvironmental benefits Land Degrad Dev 16 99ndash112

Goodman RA Remington PL Howard RJ 2000 Communicating information foraction within the public health system In Teutsch SM Churchill RE (Eds)Principles and Practice of Public Health Surveillance Oxford University PressNew York pp 168ndash175

Gregg MB 2002 Conducting a field investigation In Gregg MB (Ed) FieldEpidemiology 2nd ed Oxford University Press Oxford UK pp 62ndash77

Heilig CM Sweeney P 2010 Ethics in public health surveillance InLee LM Teutsch SM Thacker SB St Loius ME (Eds) Principles and Practiceof Public Health Surveillance 3rd ed Oxford University Press Oxford UK pp198ndash216

Herrick JEV Lessard C Spaeth KE Shaver PL Dayton RS Pyke DA et al 2010National ecosystem assessments supported by local and scientific knowledgeFront Ecol Environ 8 403ndash408

Higgins JPT Green S (Eds) 2011 Cochrane handbook for systematic reviews ofinterventions version 510 [updated March 2011] The Cochrane Collaborationltwwwcochrane-handbookorggt (accessed 20141001)

Hubbard DW 2014 How to measure anything finding the value of ldquointangiblesrdquoIn Business 3rd ed John Wiley and Sons Inc Hoboken NJ

ICRAF 2002 Improved land management in the Lake Victoria basin annual technicalreport July 2001 to June 2002 Natural Resource Problems Priorities andPolices Programme Working Paper 2002-2 World Agroforestry Centre NairobiKenya

Jones RG Payne BR 1997 Clinical Investigation and Statistics in LaboratoryMedicine The Association for Clinical Biochemistry (ACB) Venture PublicationsLondon

Katz MH 2010 Evaluating Clinical and Public Health Interventions A PracticalGuide to Study Design and Statistics Cambridge University Press CambridgeUK

Kibblewhite MG Ritz K Swift MJ 2008 Soil health in agricultural systems PhilTrans R Soc B 363 685ndash701

King DK Glasgow RE Leeman-Castillo B 2010 Reaiming RE-AIM usingthe model to plan implement and evaluate the effects of environmentalchange approaches to enhancing population health Am J Public Health 1002076ndash2084

Koohafkan AP 2000 Land resources potential and sustainable land managementan overview Nat Resour Forum 24 69ndash81

Lambin EF Turner BL Geist HJ Agbola SD Angelsen A Bruce JW et al 2001The causes of land-use and land-cover change moving beyond the myths GlobEnviron Change 11 261ndash269

Last JM (Ed) 2001 A Dictionary of Epidemiology Oxford University Press NewYork

Lawson AB Kleinman K (Eds) 2005 Spatial and Syndromic Surveillance for PublicHealth John Wiley and Sons Ltd Chichester UK

Lee KN 1993 Compass and Gyroscope Integrating Science and Politics for theEnvironment Island Press Washington DC

Liniger N van Lynden G Nachtergaele F Schwilch G 2008 A Questionnaire forMapping Land Degradation and Sustainable Land Management CDEWOCATFAOLADA ISRIC FAO Rome

Lipsey MW Cordray DS 2000 Evaluation methods for social intervention AnnuRev Psychol 51 345ndash375

Lopez AD 2005 The evolution of the Global Burden of Disease framework fordisease injury and risk factor quantification developing the evidence base fornational regional and global public health action Global Health 1 5 doi1011861744-8603-1-5 Published online 2005 April 22

MA 2003 Millennium Ecosystem Assessment Ecosystems and Human Well-BeingA Framework for Assessment World Resources Institute Island Press WashingtonDC

MA 2005a Ecosystems and Human Well-Being Volume 3 Policy ResponsesMillennium Ecosystem Assessment World Resources Institute Island PressWashington DC

MA 2005b Ecosystems and Human Well-Being Volume 1 Current State and TrendsMillennium Ecosystem Assessment World Resources Institute Island PressWashington DC

McKenzie NJ Grundy MJ 2008 Approaches to land resource survey In McKenzieNJ Grundy MJ Webster R Ringrose-Voase AJ (Eds) Guidelines for SurveyingSoil and Land Resources 2nd ed CSIRO Publishing Collingwood Australia pp15ndash25

McQueen DV Puska P (Eds) 2003 Global Behavioural Risk Factor SurveillanceKluwer AcademicPlenum Publishers New York

Mitchell RB Clark WC Cash DW Dickson NM (Eds) 2006 Global EnvironmentalAssessments Information and Influence MIT Press Cambridge

Moher D Schulz KF Altman DG 2001 The CONSORT statementrevised recommendations for improving the quality of reports ofparallel group randomized trials BMC Med Res Methodol 1 2 lthttpwwwbiomedcentralcom1471-228812gt (accessed 20141001)

Murray CJ Evans DB Acharya A Baltussen RM 2000 Developmentof WHO guidelines on generalized cost-effectiveness analysis Health Econ 9235ndash251

ARTICLE IN PRESS



Murray CJ Ezatti M Lopez AD Rodgers A Vander Hoorn S 2003 Comparativequantification of health risks conceptual framework and methodological issuesPopul Health Metr 1 1ndash20

Murray CJL 1996 Rethinking DALYs In Murray CJL Lopez AD Cambridge MA(Eds) The Global Burden of Disease Harvard University Press on behalf of theWorld Health Organization and the World Bank pp 1ndash89

Nachtergaele F Petri M 2010 Mapping Land Use Systems at Global and RegionalScales for Land Degradation Assessment Analysis Version 11 FAO Rome

Ottoson JM Wilson DH 2003 Did they use it Beyond the collection of surveillanceinformation In McQueen D Puska P (Eds) Global Behavioral Risk FactorSurveillance KluwerPlenum New York

Petschel-Held G Block A Cassel-Gintz M Kropp J Luumldeke MKB MoldenhauerO et al 1999 Syndromes of global change a qualitative modelling approachto assist global environmental management Environ Model Assess 4 295ndash314

Rein DB 2010 Economic and policy justification for public health surveillance InLee LM Teutsch SM Thacker SB St Loius ME (Eds) Principles and Practiceof Public Health Surveillance 3rd ed Oxford University Press Oxford UK pp32ndash43

Remmington PL Nelson DE 2010 Communicating public health surveillanceinformation for action In Lee LM Teutsch SM Thacker SB St Loius ME(Eds) Principles and Practice of Public Health Surveillance 3rd ed OxfordUniversity Press Oxford UK pp 146ndash165

Reynolds JF Smith DMS Lambin EF Turner II BL Mortimore M BatterburySPJ et al 2007 Global desertification building a science for drylanddevelopment Science 316 847ndash851

Roling NG Wagemakers MAE (Eds) 1998 Facilitating Sustainable AgricultureParticipatory Learning and Adaptive Management in Times of EnvironmentalUncertainty Cambridge University Press Cambridge UK

Rose G 2008 The Strategy of Preventive Medicine New edition (Original editionwas published in 1992) Oxford University Press New York

Roumlder A Hill J 2009 Recent advances in remote sensing and geoinformaticprocessing for land degradation assessment Taylor and Francis Group London

Rudel TK 2008 Meta-analyses of case studies a method for studying regional andglobal environmental change Glob Environ Change 18 18ndash25

Sanchez PA Palm CA Buol SW 2003 Fertility capability soil classification a toolto help assess soil quality in the tropics Geoderma 114 157ndash185

Schlesselman JJ 1982 Case-Control Studies Design Conduct and Analysis OxfordUniversity Press New York

Schreuder HT Ernst R Ramirez-Maldonado H 2004 Statistical Techniques forSampling and Monitoring Natural Resources Gen Tech Rep RMRS-GTR-126 USDepartment of Agriculture Forest Service Rocky Mountain Research Station FortCollins CO

Shepherd KD Walsh MG 2002 Development of reflectance spectral libraries forcharacterization of soil properties Soil Sci Soc Am J 66 988ndash998

Shepherd KD Walsh MG 2007 Infrared spectroscopy ndash enabling an evidence-baseddiagnostic surveillance approach to agricultural and environmental managementin developing countries J Near Infr Spectrosc 15 1ndash19

Shepherd KD Vanlauwe B Gachengo CN Palm CA 2005 Decomposition andmineralization of organic residues predicted using near infrared spectroscopyPlant Soil 277 315ndash333

Shepherd KD Farrow A Ringler C Gassner A Jarvis A 2013 Review of theEvidence on Indicators Metrics and Monitoring Systems Commissioned by the

UK Department for International Development (DFID) World Agroforestry CentreNairobi lthttpr4ddfidgovukoutput192446defaultaspxgt (accessed20141001)

Sonneveld BGJS Dent DL 2009 How good is GLASOD J Environ Manage 90274ndash283

Stangl DK Berry DA (Eds) 2000 Meta-Analysis in Medicine and Health PolicyMarcel-Dekker New York

Stevens DL Jr Olsen LR 2004 Spatially Balanced Sampling of Natural ResourcesJ Am Stat Assoc 99 262ndash278

Stroup DF Brookmeyer R Kalsbeek D 2004 Public health surveillance in actiona framework In Brookmeyer R Stroup DF (Eds) Monitoring the Health ofPopulations Statistical Principles and Methods for Public Health SurveillanceOxford University Press Oxford UK pp 1ndash35

Swift MJ Shepherd KD 2007 Saving Africarsquos Soils Science and Technology forImproved Soil Management in Africa Joint NEPAD ICRAF TSBF-CIAT PublicationWorld Agroforestry Centre Nairobi

Tan-Torres Edejer T Baltussen R Adam T (Eds) 2003 Making Choices in HealthWHO Guide to Cost-Effectiveness Analysis World Health Organization Geneva

Teutsch SM Churchill RE (Eds) 2000 Principles and Practice of Public HealthSurveillance Oxford University Press New York

Thacker SB Berkelman RL 2008 Public health surveillance in the United StatesEpidemiol Rev 10 164ndash190

Thacker SB Birkhead GS 2002 Surveillance In Gregg MB (Ed)Field Epidemiology 2nd ed Oxford University Press Oxford UK pp26ndash50

Thomas DSG Middleton NJ 1994 Desertification Exploding the Myth WileyChichester UK

UNEP 2007 Global Environmental Outlook (GEO-4) Environment for Develop-ment United Nations Environment Programme Progress Press Ltd ValletaMalta

Varinggen T-G Shepherd KD Walsh MG Winowiecki L Desta LT Tondoh JE 2010AfSIS Technical Specifications Soil Health Surveillance Version 10 Africa SoilInformation Service Nairobi

Vogt JV Safriel U Von Maltitz G Sokona Y Zougmore R Bastin G et al 2011Monitoring and assessment of land degradation and desertification towardsnew conceptual and integrated approaches Land Degrad Dev doi101002ldr1075

WBGU (German Advisory Council on Global Change) 1995 World in Transition TheThreat to Soils 1994 Annual Report Economica Verlag GmbH Bonn

WHO 1999 WHO Recommended Surveillance Standards 2nd ed World HealthOrganization Geneva

WHO 2001 Making surveillance work A series of four modules produced by theWorld Health Organization Geneva lthttpwhqlibdocwhointhq2001WHO_VampB_0108pdfua=1gt (accessed 20141001)

WHO 2002 The World Health Report 2002 Reducing Risks Promoting Healthy LivesWorld Health Organization Geneva

WHO 2004 World Report on Knowledge for Better HealthndashStrengthening HealthSystems World Health Organization Geneva

Young A 1991 Soil monitoring a basic task for soil survey organizations Soil UseManag 7 126ndash130

Young A 1998 Land Resources Now and for the Future Cambridge University PressCambridge UK

ARTICLE IN PRESS



Land health surveillance and response A framework for evidence-informed land management

Introduction

Principles of land health surveillance

Definition

Functions

Science principles and system components

Applying surveillance principles

Sampling populations

Prevalence and risk factors

Dosendasheffect relationships

Case definitions and diagnostic screening

Forecasting

Intervention targeting

Intervention evaluation

Study designs

Quality standards and meta-analysis

Using and communicating surveillance information

Users and uses of surveillance information

Communicating results

Intervention strategies

Prevention vs treatment strategies

Population-wide vs high-risk preventive strategies

Other considerations for strategy

Implementing national surveillance systems

Conclusions

Acknowledgements

References

situation contrasts with the situation in the public health sectorwhere surveillance is the main source of information guidingdecisions and actions in health policy and practice (Teutsch andChurchill 2000) Therefore surveillance frameworks used in publichealth management may have much to offer for guiding land healthmanagement

Public health surveillance which has been active since the 1940shas been defined in simple terms as ldquothe routine ongoing collec-tion analysis and dissemination of data to those in public healthwho need to knowrdquo (Brookmeyer and Stroup 2004) It provides thescientific and factual foundation and database for informed deci-sion making and appropriate public health action We takesurveillance to include the full cycle of communication responseand evaluation and not only data collection and dissemination (egRemmington and Nelson 2010) An example of the contribution ofsurveillance in the public health sector is the major achievementof the Global Burden of Disease Study which provided data on in-cidence and prevalence by age sex and region of over 100 diseasesand injuries and a comparative assessment of 20 major risk factors(Lopez 2005) The data were used to analyse the impact of the dif-ferent risks on the avoidable future burden of disease evaluate thecost-effectiveness of alternative interventions and establish clear in-tervention strategies (WHO 2002) Nothing comparable exists forassessing the state of the land resource base and for definingintervention strategies at any scale

In fact land management problems share many of the complexfeatures of public health problems especially those of non-communicable diseases and thus similar scientific approaches maybe applicable These similarities include

1 A rapidly increasing burden of health problems in develop-ing countries partly linked to demographic factors and theproblems become chronic if not addressed

2 Problems often occur together as syndromes with a commonset of symptoms rather than in isolation

3 Health problems exist as a continuum and there is difficultyin defining the normal case and diagnosing poor versus goodhealth

4 Problems are associated with a range of physical biologicalsocial and economic determinants (or risk factors) both atindividual and population levels

5 Risks are often inter-related and act together to cause a healthproblem They range from proximal risks acting directly tocause the problem to distal risks that are further back in thecausal chain

6 Risk factors are often separated from outcomes in time some-times by many decades making it difficult to establishcausality

7 Risks are generally greater for the disadvantaged in our so-cieties and poverty is a major risk factor

8 Evaluating the cost-effectiveness of alternative preventive andrehabilitation interventions is complex and must considermany factors Interventions may cause unintended side effects

9 Different stakeholder groups (scientists policy makers public)perceive risks differently requiring two-way communica-tion processes

10 Few resources are allocated to surveillance in developing coun-tries presenting challenges such as how best to communicatesurveillance and research findings in this setting which stillneeds much local research

Given these similarities the overall goal of this paper is to applyscientific principles used in public health surveillance to the designof a methodological framework for evidence-informed land man-agement and policy and hopefully stimulate its wide applicationThe specific objectives are to (i) elucidate the scientific principlesand system components of public health surveillance (ii) de-scribe their applicability to land health surveillance and (iii) suggest

opportunities for deployment of land health surveillance systemsand identify areas for their further research

2 Principles of land health surveillance

21 Definition

The definition of land health surveillance is closely modelled onthe definition of public health surveillance given by the Centres forDisease Control and Prevention (Thacker and Berkelman 2008)

ldquoLand health surveillance is the ongoing systematic collectionanalysis and interpretation of data essential to the planning im-plementation and evaluation of land management policy and practiceand application of these data to promote protect and restore landand ecosystem healthrdquo (UNEP 2007) A surveillance system in-cludes a functional capacity for data collection analysis anddissemination linked to land health programmes Surveillance aimsto provide information for action but excludes research on landhealth and full-scale implementation of delivery programmes

Land health is defined here as the capacity of land relative toits potential to sustain delivery of essential ecosystem services (thebenefits people obtain from ecosystems) which are well de-scribed in MA (2003) and Kibblewhite et al (2008) Land health isnot directly equated with human health but rather the scientific con-ceptual and statistical approaches in land health surveillance aremodelled on those employed in public health The parallels in prin-ciples are closer for surveillance of chronic non-communicablediseases than for communicable diseases

22 Functions

In broad terms the ultimate goal of land health surveillance isto tell us (i) where land problems exist (ii) whom and what theyaffect (iii) where programmatic and prevention activities shouldbe directed and (iv) how well they are working Land health sur-veillance systems are designed to provide timely information foraction to improve land productivity and maintain essential ecosys-tem functions In more specific terms land health surveillance hasa number of functions (Table 1)

In public health different types of surveillance systems are usedfor different purposes (Stroup et al 2004) and examples of poten-tial parallel applications in land health can be identified (Table 2)These generally follow the progression of surveillance steps of (i)establishing risks to key health problems as a basis for interven-tion targeting (ii) evaluating interventions and (iii) establishing finaloutcomes Integrated surveillance systems are likely to be of most

Table 1Principal functions of land health surveillance

1 Identify land health problems11 Assess and monitor land health status12 Quantify risk factors of land degradation and sustainable land

management13 Provide early warning of land degradation

2 Establish quantitative objectives for land health promotion21 Specify objectives for prevention of land degradation22 Specify objectives for land rehabilitation

3 Provide information for the design and planning of land managementintervention programmes and resource allocation priorities31 Prioritize intervention areas based on degree and extent of risks to land

health32 Prioritize interventions based on cost-effectiveness analysis33 Spatially target interventions

4 Determine the impact of specific interventions41 Empirically test effectiveness of interventions42 Establish outcomes in terms of reduction in risks and health

improvement5 Identify research service and training needs for different stakeholder groups

ARTICLE IN PRESS







































ARTICLE IN PRESS














ARTICLE IN PRESS










ARTICLE IN PRESS























ARTICLE IN PRESS









35 Forecasting












ARTICLE IN PRESS
































ARTICLE IN PRESS



















ARTICLE IN PRESS

















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References





































ARTICLE IN PRESS














ARTICLE IN PRESS










ARTICLE IN PRESS























ARTICLE IN PRESS









35 Forecasting












ARTICLE IN PRESS
































ARTICLE IN PRESS



















ARTICLE IN PRESS

















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References












ARTICLE IN PRESS










ARTICLE IN PRESS























ARTICLE IN PRESS









35 Forecasting












ARTICLE IN PRESS
































ARTICLE IN PRESS



















ARTICLE IN PRESS

















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References








ARTICLE IN PRESS























ARTICLE IN PRESS









35 Forecasting












ARTICLE IN PRESS
































ARTICLE IN PRESS



















ARTICLE IN PRESS

















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References





















ARTICLE IN PRESS









35 Forecasting












ARTICLE IN PRESS
































ARTICLE IN PRESS



















ARTICLE IN PRESS

















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References







35 Forecasting












ARTICLE IN PRESS
































ARTICLE IN PRESS



















ARTICLE IN PRESS

















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References






























ARTICLE IN PRESS



















ARTICLE IN PRESS

















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References

















ARTICLE IN PRESS

















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References















































Internationaldonors





ARTICLE IN PRESS















ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References













ARTICLE IN PRESS








6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References






6 Conclusions












ARTICLE IN PRESS











Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References









Acknowledgements


References












































ARTICLE IN PRESS











































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References









































ARTICLE IN PRESS




Introduction


Definition

Functions







Forecasting



Study designs










Conclusions

Acknowledgements

References

Land health surveillance and response: A framework for evidence-informed land management

Documents

Transcript of Land health surveillance and response: A framework for evidence-informed land management