Morbidity in the Marshes: Using Spatial Epidemiology to ... western malaria ajpa... · Morbidity in the Marshes: Using Spatial Epidemiology to Investigate Skeletal Evidence for Malaria

Morbidity in the Marshes: Using Spatial Epidemiology toInvestigate Skeletal Evidence for Malaria in Anglo-SaxonEngland (AD 410–1050)

R.L. Gowland* and A.G. Western

Department of Archaeology, Durham University, South Road, Durham DH1 3LE, UK

KEY WORDS cribra orbitalia; Plasmodium vivax; Anopheles atroparvus; enamel hypoplasia

ABSTRACT Concerns over climate change and itspotential impact on infectious disease prevalence havecontributed to a resurging interest in malaria in thepast. A wealth of historical evidence indicates thatmalaria, specifically Plasmodium vivax, was endemic inthe wetlands of England from the 16th century onwards.While it is thought that malaria was introduced to Brit-ain during the Roman occupation (AD first to fifth centu-ries), the lack of written mortality records prior to thepost-medieval period makes it difficult to evaluate eitherthe presence or impact of the disease. The analysis ofhuman skeletal remains from archaeological contexts isthe only potential means of examining P. vivax in thepast. Malaria does not result in unequivocal pathologicallesions in the human skeleton; however, it results in he-molytic anemia, which can contribute to the skeletal con-

dition cribra orbitalia. Using geographical informationsystems (GIS), we conducted a spatial analysis of theprevalence of cribra orbitalia from 46 sites (5,802 indi-viduals) in relation to geographical variables, historicallyrecorded distribution patterns of indigenous malaria andthe habitat of its mosquito vector Anopheles atroparvus.Overall, those individuals living in low-lying and Fen-land regions exhibited higher levels of cribra orbitaliathan those in nonmarshy locales. No corresponding rela-tionship existed with enamel hypoplasia. We concludethat P. vivax malaria, in conjunction with other comor-bidities, is likely to be responsible for the patternobserved. Studies of climate and infectious disease in thepast are important for modeling future health in relationto climate change predictions. Am J Phys Anthropol147:301–311, 2012. VVC 2011 Wiley Periodicals, Inc.

The history of endemic malaria in Britain has gainedincreasing attention over recent years due to escalatingconcerns surrounding climate change and the re-emer-gence of infectious diseases (Marchant et al., 1999;Rieter, 2000; Lindsay and Thomas, 2001; Reiter et al.,2004; Lindsay et al., 2010). Analyses of the impact ofmalaria on past populations have been based on textualmortality records, and have therefore been restricted torecent times (Dobson, 1994, 1997; Nicholls, 2000; Kuhnet al., 2003). The research described here uses spatialepidemiology to analyze the prevalence of skeletal indi-cators of poor health as evidence for malaria in easternEngland. This is the first time that paleopathologicalindicators have been modeled in this way.It seems likely that malaria was introduced to Britain

during the Roman occupation in the first to fifth centu-ries AD (Sallares, 2002: 34). A reference in Bald’s Leech-book, a ninth century AD medical text, and numerousothers in Chaucer and Shakespeare, may be to malaria(Bruce-Chwatt, 1976; Cameron, 1993). It has howeverbeen suggested that malaria did not become endemicuntil the post-medieval period (Dobson, 1994). The lackof substantial written records prior to the 16th centurymeans that we cannot know whether the disease waspresent and how much impact it had. Documentaryreferences increase from the 15th century (Dobson,1994); they do not specifically refer to malaria until the19th century, but often use the nonspecific term ‘‘ague’’to describe intermittent fevers widespread in the generalpopulation at this time (Shute and Maryon, 1974; Bruce-Chwatt, 1976; Nicholls, 2000). These would haveincluded typhus as well as enteric and relapsing fevers(Shute and Maryon, 1974; Hutchinson and Lindsay,

2006). Not all deaths recorded as ‘‘ague’’ should thereforebe attributed to malaria (James, 1929; Dobson, 1994:45).The only means of assessing the presence of malaria in

the past is the analysis of skeletal remains. Plasmodiumvivax is the most likely form of indigenous malaria inBritain (James, 1929), and clinical tests using bloodsmears on ‘‘ague sufferers’’ in the early 20th century posi-tively identified the presence of P. vivax (Dobson, 1994:55). While P. vivax malaria does not result in unequivocallesions in the human skeleton, it is strongly associatedwith hemolytic anemia which can contribute to the preva-lence of the condition cribra orbitalia (Buckley and Tayles,2003; Buckley, 2006; Gowland and Garnsey, 2010). Thearea of investigation for this research was easternEngland, in particular the low-lying wetland areas wheremalaria was recorded in historical times (Dobson, 1994:48). The skeletal sample dates to the Anglo-Saxon period(AD 410–1,050), for which substantial numbers of inhu-mation burials have been excavated, but no written mor-tality records exist. The Romano-British transgression

Grant sponsor: British Academy; Grant number: ref SG090361.

*Correspondence to: Dr. Rebecca Gowland, Department of Archae-ology, Durham University, Durham, DH1 3LE, UK.E-mail: [email protected]

Received 14 July 2011; accepted 29 October 2011

DOI 10.1002/ajpa.21648Published online 20 December 2011 in Wiley Online Library

(wileyonlinelibrary.com).

VVC 2011 WILEY PERIODICALS, INC.

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 147:301–311 (2012)

(300–600 AD) and the Middle-Saxon warm period(700–800 AD) were significant climatic events occurringduring this time, which resulted in flooding of the low-lands and warmer temperatures (Rippon 2000, 2010;Cracknell, 2005), potentially providing an ideal environ-ment for mosquito abundance. Using geographical infor-mation systems (GIS), we conducted a spatial analysis ofthe prevalence of cribra orbitalia and, for comparativepurposes, dental enamel hypoplasia (regarded as a non-specific indicator of poor health) in relation to geographi-cal variables (geology and topography) and historicallyrecorded distribution patterns of indigenous malaria andthe habitat of its mosquito vector Anopheles atroparvus.Spatial epidemiology using mapping applications such asGIS enable accurate mapping and overlaying of manytypes of data (Pfeiffer et al., 2010) and are particularlyuseful for archaeological data; since, samples aresmaller than in clinical contexts, thus contextualisation isrequired on a wide scale to investigate underlying etiologi-cal factors.

MALARIA AND SKELETAL REMAINS

Direct evidence: biomolecular

Direct evidence for malaria can only come from theidentification of the relevant plasmodium species DNAin the remains of past victims of the disease. The DNAof Plasmodium falciparum has been retrieved from thesoft tissues of Egyptian Mummies (e.g., Taylor et al.,1997; Nerlich et al., 2008) and skeletal remains from a5th century AD infant cemetery in Umbria, Italy (Sal-lares et al., 2004). However, current PCR systems do notappear to be able to amplify routinely the DNA ofmalaria pathogens from ancient bones (Zink et al.,2002). For example, Pinello (2008) sampled 159 skeletonsfrom areas of known malaria endemicity in later medie-val England, but failed to retrieve P. vivax DNA. Pinelloconcluded that the structure and pathogenicity ofP. vivax might prohibit its study from archaeologicalskeletal material using current biomolecular techniques.Another potential avenue for direct diagnosis from

skeletal remains is the use of immunological techniques.These seek to identify antigens produced in response tothe malaria pathogen and are frequently used as a diag-nostic tool in clinical contexts. Claims for the successfulidentification of P. falciparum using such techniqueshave been made from soft tissue samples obtained froman Egyptian Mummy (Bianucci et al., 2008) and onarchaeological bones dating to 16th century Italy (Forna-ciari et al., 2010). However, positive test results canoccur in the absence of the pathogen due to a rangeof diagenetic and contaminant factors and a number ofauthors have raised serious doubts as to the reliability ofthese techniques when applied to archaeological bone(e.g., Brandt et al., 2002; Setzer, 2010).

Indirect evidence: skeletal

Cribra orbitalia refers to a condition that manifests assmall holes (known as foramina) in the orbits of theskull. These foramina result from an expansion of thediploe (marrow hypertrophy) and a thinning of the outercortex of the bone. A similar condition observed on thecranial vault is referred to as porotic hyperostosis, aterm under which cribra orbitalia is sometimes sub-sumed. These lesions are commonly recorded in bothchild and adult skeletons from archaeological contexts.

The lesions in adult skeletons are believed to representhealed childhood episodes of the condition (Stuart-Mac-adam, 1985, 1992).For many years, cribra orbitalia has been interpreted

as an indicator of iron deficiency anemia resulting frompoor diet, parasites, and/or a high pathogen load. Mostinterpretations of cribra orbitalia in past populationsinvoke a synergistic combination of all of these factors.Other conditions that result in a similar appearance toporotic hyperostosis are vitamin D or C deficiencies andinfectious processes (Ortner, 2003; Lewis, 2007). In a his-tological study, Wapler et al. (2004: 336) noted that while54.1% of macroscopically observed lesions in their studysample were associated with anemia and hypertrophy ofbone marrow, 32.4% were likely to have been producedsolely through inflammation (osteitis). Diagnosis is alsocomplicated by the interrelated nature of many defi-ciency diseases; for example folic acid deficiency willaffect iron and vitamin C status (Fairgrieve and Molto,2000). In such cases, it would be difficult to determinethe precise etiology of cribra orbitalia lesions. Overall,therefore, there is the potential for misidentification ofcribra orbitalia and misattribution of anemia as beingthe sole agent of the lesions observed macroscopically inskeletal material.While clinical literature in the 1960s and 70s reported

radiological findings of porotic hyperostosis to be associ-ated with iron deficiency anemia (See, for example,Aksoy et al., 1966), it is now understood that only thoseanemias linked to a significant and sustained increase inred blood cell production (erythropoeisis) can directlyresult in these lesions; iron deficiency results in suppres-sion of the formation of marrow and new red blood cellsand, therefore, cannot lead to porotic hyperostosis andcribra orbitalia (Walker et al., 2009, 111). Rothschild(2002, 609) has also argued that the hair-on-end sign(indicative of porotic hyperostosis) could only reflect‘‘hyperregenerative marrow’’ and could not be attributedto chronic iron deficiency anemia, since in this case ‘‘ifthere is inadequate iron for the production of blood cells,the marrow may actually be hypo-regenerative.’’ Roths-child (2002, 609) suggested that parasitic infection andhemolytic anemia were among the causes of skeletalchanges. Walker et al. (2009) consider both hemolyticand megaloblastic anemias, the latter caused by vitaminB12 and B9 deficiencies, to have the potential to producethe levels of marrow hypertrophy necessary for macro-scopic observations of porotic hyperostosis and cribraorbitalia (although this has been disputed by Oxenhamand Cavill, 2010).Clinical studies have observed that severe anemia is

one of the leading causes of morbidity and mortality inpatients with malaria (Chang and Stevenson, 2004).Symptoms of malaria include cyclical fever and chills,headache, weakness, vomiting, and diarrhea. In vivaxmalarial patients, blood serum levels of magnesium, zincand iron are significantly depleted (Balloch et al., 2011).The infection leads to hemolysis, the process of an abnor-mal breakdown of red blood cells that are subsequentlyremoved by the spleen. When hemolysis exceeds the rateof erythropoiesis, an individual becomes anemic.P. vivax parasites infect the immature red blood cells

(reticulocytes) and generally cause long-term chronicinfections and hemolytic anemia but low mortality(Calton et al., 2008). In cases of malarial anemia, theeffects of hemolysis are further exacerbated by theaccompanying dyserythropoiesis, where the production

302 R.L. GOWLAND AND A.G. WESTERN

American Journal of Physical Anthropology

of new red blood cells is inadequate; bone marrow cellsare increased but mature red cell output is insufficient(Abdalla and Pasvol, 2004: 63). Dyserythropoiesis resultsin the prolonging of anemia and is a frequently observedsequela of chronic malaria (Wickramasinghe and Abdalla,2000; Abdalla and Wickramasinghe, 2004: 219). P. vivaxparasites have a low biomass but studies have shown thatfor every infected red blood cell destroyed by P. vivax, �32noninfected red blood cells are removed from circulation(Anstey et al., 2009; Bassat and Alonso, 2011). Therefore,P. vivax causes severe hemolytic anemia in sufferersthrough a number of different mechanisms, includingdestruction of noninfected red blood cells, destruction ofreticulocytes, and increased fragility of infected and non-infected red blood cells (Anstey et al., 2009: 221).Megaloblastic morphological abnormalities of blood

cells also occur as part of the dyserythropoietic processas a direct result of malaria infection (Abdalla and Wick-ramasinghe, 2004: 221). Of significance for studies ofcribra orbitalia, these abnormalities cannot be differenti-ated from those caused by vitamin B12 or folate deficien-cies (Abdalla and Wickramasinghe, 2004: 223–224).In cases of chronic hemolytic anemia, marrow hyper-

plasia and reconversion from yellow to red marrow hasbeen observed as a response to the hemolysis in anattempt to generate sufficient compensatory new redblood cells (Buckley, 2006). In nonadults, the conversionfrom red to yellow marrow is arrested. Radiological signsinclude a coarsened or accentuated trabecular pattern ofbone marrow with widened medullary spaces and corti-cal thinning (Vigorita, 2008). The clinical literatureregarding skeletal changes in response to malaria andassociated hemolytic anemia is extremely sparse,although bone marrow is described as ‘‘extremely hyper-cellular’’ in chronic malaria (Pasvol and Abdalla, 2004:44). There is no direct clinical evidence beyond marrowhyperplasia linking cribrotic lesions with P. vivax. Thisis because cribra orbitalia is not of significance in a clini-cal context and it is difficult to observe radiographicallyin living patients. Most discussion has been in the paleo-pathological literature. Angel (1966) first made the asso-ciation between endemic malaria and high prevalencesof porotic hyperostosis in the Mediterranean, althoughhe initially suggested that thalassemia was responsible.Porotic hyperostosis is observed in genetic hemolyticanemias such as beta-thalassemia major and sickle celldisease. However, genetic anemia is rare in comparisonto acquired anemia. Furthermore, the severity and dis-tribution of the skeletal lesions associated with geneticanemias tend to be much greater than observed inacquired anemia, thus enabling these conditions to bedifferentiated in the paleopathological record (Hershko-vitz et al., 1997; Lagia et al., 2007; Lewis, early view).In conclusion, there is a clinically established associa-

tion between P. vivax malaria, chronic anemia, and mar-row hypertrophy. Previous paleopathological studieshave noted an association between P. falciparum malaria(established using DNA) and porotic hyperostosis (Soren,2003). The association between P. vivax malaria andhemolytic anemia has also been highlighted by Buckley(2006) in relation to porotic hyperostosis and cribra orbi-talia in archaeological remains excavated from Polyne-sia. Furthermore, the clinical literature emphasizes thesynergistic relationship between malaria and other infec-tious diseases, nutritional disorders and parasites, oftentermed ‘‘comorbidities’’ (Chang and Stevenson, 2004;Anstey et al., 2009), all of which are routinely implicated

in the etiology of cribra orbitalia. It therefore seemshighly probable that infection by P. vivax, in conjunctionwith other comorbidities, will produce the sustainedincrease in red blood cell production necessary to resultin cribra orbitalia in skeletal remains. All other factorsbeing equal, the prevalence of this condition should behigher in areas where malaria is endemic.

MATERIALS AND METHODS

A spatial epidemiological study of the skeletal remainswas conducted to evaluate correlations between tradi-tionally recognized nonspecific skeletal indicators of poorhealth (cribra orbitalia and enamel hypoplasia), geogra-phy, topography, and historically recorded evidence ofoutbreaks of malaria. A survey of Anglo-Saxon inhuma-tion sites was undertaken using Historic EnvironmentalRecords (HERs) from 27 counties and district authoritiesin the east of England and these were mapped usingGIS Arcview version 9.3 (Figs. 1 and 2). Subsequently,cribra orbitalia and, for comparison, enamel hypoplasiaprevalence rates were collated from published andunpublished skeletal reports and linked to the surveydata in the GIS database. The basic methodologies forrecording the presence or absence of cribra orbitalia fol-lowed Brothwell (1981) or Stuart-Macadam (1991). Forenamel hypoplasia, defects were most frequentlyrecorded using Buikstra and Ubelaker (1994) and Broth-well (1981, 159). There will doubtless have been someinterobserver variability in recording; however, becausethis study relies on presence/absence of lesions only,rather than degrees of severity, its impact should beminimal. The total number of sites for which cribra

Fig. 1. Distribution of Anglo-Saxon cemeteries in easternEngland.

303SKELETAL EVIDENCE FOR MALARIA


orbitalia crude prevalence rates (CPRs) were availablewas 46, comprising of the skeletal remains of 5,802 indi-viduals. Sites were excluded where they contained fewerthan 15 individuals and where bone preservation wasnotably poor. The basic age composition of the popula-tions was also noted (Table 1).Anglo-Saxon environmental and topographical data

was extrapolated within a 2 km circumferential bufferzone for each site using British Geological Society super-ficial and bedrock geology digital map layers at 1:50,000scale obtained through the EDINA geology digimap serv-ice. Slope data was obtained from the ShareGeo Openwebsite (www.sharegeo.ac.uk) SRTM Slope DEM forGreat Britain and was recorded for the central point ofeach site.Historical data from a British Museum (Natural His-

tory) survey carried out in 1900 AD noting the locationsof Anopheles maculipennis (now known as Anophelesatroparvus) mosquito in England was mapped on GIS(Lang, 1918). Since this early study was not systematic,these data were cross-referenced with a second contem-poraneous survey of the distribution of ‘‘ague’’ inEngland (Nuttall et al., 1901).

Spatial distribution

Methodology. This study uses the general approachtaken in clinical spatial epidemiological studies as out-lined by Pfeiffer et al. (2010). The sample of 46 cribraorbitalia CPRs were mapped and tested for clustering orspatial autocorrelation (SA). SA is a measure of ‘‘thedegree of spatial similarity observed among neighboringvalues’’ (Pfeiffer et al., 2010: 34). Ideally, true prevalencerates (TPRs) should be used in paleopathologicalresearch due to these rates accounting for differential

bone preservation across sites. Unfortunately, the sampleof TPRs of cribra orbitalia gathered from the literaturewas too small to be used for SA investigation. SinceCPRs can be affected by differential bone preservation,CPR values were statistically assessed against theknown corresponding TPR values in the study sampleusing both Pearson’s r (r2 5 0.68, two-tailed P value is\0.0001) and Spearman’s Rank (r2 5 0.72, two-tailed Pvalue is \0.0001) correlations (Fig. 3). These tests con-firm that there is a highly significant correlationbetween the two datasets. Given that the majority of thepopulation sizes are large and that populations of poorpreservation were excluded, it is considered here thatthe CPRs should be reliable proxies for TPRs.The distribution of sites was first analyzed to detect

any spatial clustering irrespective of prevalence valuesusing the average nearest neighbor tool in GIS (DeSmith et al., 2007). The average nearest neighbor testcalculates an index based on the average distance fromeach point to its nearest neighboring point. The nearestneighbor ratio was 0.93 with a P value of 0.36 and a Zscore of 20.914010 (standard deviation), demonstratingthat the selected sites were randomly distributed acrossthe study area.Due to the small sample size, only global estimates

detecting first order effects in SA were made. A firstorder effect is defined as ‘‘one that produces a variationin point density in response to some causal variable’’ (DeSmith et al., 2007). This is exemplified by environmentalconditions acting as habitats for vectors that may causedisease. Second order effects are locally produced andare a measure of the interaction between neighbors andare used, for example, in the epidemiological analysis ofcontagious diseases (De Smith et al., 2007). Factors cor-relating with the spatial distribution of a disease, suchas habitat, could however be placed in either categorydepending on the scale upon which the environmentalvariables are analyzed and how narrow an area a factoris confined to (Pfeiffer et al., 2010: 14).In this case, global effects in cribra orbitalia CPR were

calculated using Moran’s I to assess the SA between at-tribute values in adjacent areas (Pfeiffer et al., 2010:46). This method assumes heterogeneity of the distribu-tion of the population and also requires a normal distri-bution of data. To meet these requirements prevalencerates have been used rather than raw counts so thatthere is no bias in skewed population distribution. Therates were transformed using square root values, therecommended method for transforming percentage ratesbetween 0 and 30% (Gomez and Gomez, 1984: 307) andthe resulting Gaussian distribution of the data was con-firmed by carrying out the Kolmogorov-Smirnov test fornormality (KS 5 0.12; P [ 0.10) and the Shapiro-Wilktest for normality (W 5 0.95; P 5 0.05). Moran’s I wascalculated using a zone of indifference weighting, so thatthe influence of points within a specified distance bandwere weighted more highly and that there was a sharpdrop off in influence from those points lying outside theband. The appropriate distance band value was calcu-lated by GIS by formulating Z scores for a series ofincremental band values of increasing distance fromeach point taking into account the values of its neigh-bors. This process identifies the maximum SA or maxi-mum clustering of values. The value correlating withthe last increase in those Z-scores (standard deviations)is the appropriate distance band value, in this case61.1 Km.

Fig. 2. Map of sites with cribra orbitalia (CPR) data.



Getis Ord General G and Gi* tests were also used toexplore any clustering further and to identify clusteringof high (hot spots) and/or low values (cold spots). Thesetests were run also using the zone of indifference weight-ing to ensure continuity in the conceptualization of spa-tial relationships and comparability with Moran’s I.

RESULTS

The Moran’s I global test of variation revealed a statis-tically significant level of SA of the cribra orbitalia CPRvalues (Z 5 3.39; \0.001). Having identified a trend inSA, the distribution of values was analyzed using theGetis Ord General G and Gi* tests to identify hot or coldspots in clustering. The Getis Ord General G test indi-cated a statistically significant clustering of high valuesin the sample data (Z 5 2.92; \0.01). Hot and cold-spots

identified through the Getis Ord General Gi* test areillustrated in Figure 4. One statistically significant (P \0.05, Z score [ 1.96) hot spot is identified in the Cam-bridgeshire region and one cold spot (P \ 0.05; Z score\21.96) in the Northumbrian area.

To evaluate this distribution, some associated environ-mental factors potentially relating to the contrastingrates were analyzed. Northumbria is mainly hilly with ahard-rock coastline, whereas Cambridgeshire is flat andassociated with the Fens, an open coastal lowland areagreatly affected by marine and freshwater flooding inthe past. Historically, the coastal lowlands are alsoknown to have provided an ideal habitat for mosquitoes.

Geographical factors associated with hotspots

Methodology. The second part of this study aims toexplore the relationship between geographical factors

TABLE 1. The prevalence of cribra orbitalia and enamel hypoplasia in the sample and corresponding ‘‘malarial’’ or ‘‘nonmalarial’’categorizations (early AD 410–649, mid AD 650–849, and late AD 850–1050)

SiteTotal

Population Date

CribraOrbitaliaCPR (%)

CribraOrbitaliaTPR (%)

DentalEnamel

HypoplasiaCPR (%)

Nonadults0–17

years (%)Adults181 (%)

Malarial(1 5 yes,2 5 no)

Addingham, West Yorkshire 45 Mid-Late 8.9 – – 16.9 83.1 2Ailcy Hill, Ripon, N. Yorks 27 Mid 7.4 – 14.8 18.9 81.1 2Binchester, Durham 54 Early 5.5 – – 8.9 91.1 2BishopsmillSchool, Norton, Tees 86 Mid 3.5 – 18.6 14.0 86.0 2Burwell 167 Early Mid 9 15.8 – – – 2Caister on Sea, Norfolk 139 Mid-Late 18 – – 23.0 77.0 1Castledyke, Barton on Humber 200 Early 3 3.5 30 23.1 76.9 1Church End, Cherry Hinton, Cambs 683 Early 5.8 8.5 4.4 40.7 59.3 2Cleatham, Kirton in Lindsey, Lincs 60 Early 3.3 11.8 40 18.3 81.7 2Cuxton, Kent 34 Early Mid 8.8 42.9 29.4 70.6 2DixonKeld, Masham, N. Yorks 57 Early Mid 3.5 – 1.75 17.5 82.5 2Eccles, Kent 166 Early Mid 3.6 25.9 21.3 78.7 1Edix Hill Barrington, Cambs 148 Early Mid 10.1 16 10.1 31.1 68.9 1Farmers Avenue, Castle Mall, Norwich 84 Late 34.5 – 46.1 31.0 69.0 1Finglesham, Kent 201 Early Mid 17.4 43.2 19.9 22.9 77.1 1Great Chesterford, Cambs 173 Early 9.8 25.8 – 49.7 50.3 2Great Houghton, Northampton 23 Early 8.7 – 13 22.7 77.3 2Highfield Farm, Littleport, Cambs 86 Early 29.1 59.5 45.3 29.1 70.9 1Jarrow, Northumberland 170 Mid 2.3 – 11.2 42.9 57.1 2King’s Garden Hostel, Cambridge 21 Mid 4.8 5 38.1 38.1 61.9 1Marina Drive, Beds 49 Mid 26.5 38.2 – 58.1 41.9 2Melbourn 2, Cambs 30 Mid 13.3 23 – – – 2Monkwearmouth, Northumberland 327 Mid 4.3 – 0.9 35.5 64.5 2Nazeingbury, Essex 150 Mid 10 – – 11.0 89.0 1North East Bailey, NorwichCastle 112 Late 8 – – 46.6 53.4 1NorthElmhamPark, Norwich 206 Late 2.4 – – 18.9 81.1 1Norton, Cleveland 125 Early 0.8 – 35.2 30.4 69.6 2Oakington, Cambridge 25 Early 32 42.1 13 54.2 45.8 1Ocklynge Hill, Eastbourne 23 Early 4.3 – 8.7 35.7 64.3 2RaundsFurnell, Northants 361 Late 10.8 – 11.9 46.8 52.6 1RedCastle, Thetford, Norfolk 85 Late 1.2 – – 28.2 71.8 2Riccall Landing, Yorkshire 64 Mid-Late 6.3 – 47.2 18.9 81.1 2School Street,Ipswich 95 Late 8.4 – 12.6 16.8 83.2 1South Acre, Norfolk 119 Early 6.7 – 31.1 31.8 68.2 2St Peters Tip, Broadstairs, Kent 328 Early Mid 5.2 12.8 – – – 2Staunch Meadow, Brandon, Suffolk 158 Mid 11.1 – 60.1 19.6 80.4 1Tanners Row, Pontefract 178 Early Mid 13.5 – 20.2 55.1 44.9 2Thetford 2, Norfolk 81 Late 17.3 – 14.8 47.4 52.6 2Updown, Eastry, Kent 65 Early Mid 3.1 5 – 19.0 81.0 2Village Farm, Spofforth 94 Mid 11.7 13.3 30.9 29.8 70.2 2Water Lane, Melbourn, Cambs. 59 Early 27.1 34.8 1.1 18.3 81.7 2West Field Farm, Ely, Cambs 15 Mid 13.3 25 20 28.6 71.4 1West Heslerton, Yorkshire 132 Early 5.3 – 10.6 24.6 75.4 2WickenBonhunt, Saffron Walden 222 Mid-Late 10.4 – – 36.6 63.4 1Wolverton, Milton Keynes 80 Early 17.5 26.4 23.8 45.9 54.1 2York Minster 60 Mid 10 – – – – 2



and the distribution of cribra orbitalia CPRs. In order toobjectively quantify this relationship, the variables ofslope, surface soil type and the historical presence or ab-sence of the A. atroparvus mosquito were analyzed usingspatial distribution and statistical analysis.The 481 observations of A. atroparvus mosquitoes as

collated by Lang (1918) were mapped and categorizedwhere possible according to the frequency noted (Fig. 5).An interpolated map of the distribution of A. atroparvuswas created (Fig. 6). Missing values were interpolatedaccording to the average value of at least 10 nearestneighbor scores. Neighboring points were selectedequally in all directions on the basis that spatial depend-ence was isotropic and weighted according to their prox-imity. This map ignores any constraints on mosquitoabundance placed by the local environment or elevationand represents a ‘‘smoothed’’ version of the data at anational level. Although anopheles mosquitoes have beenobserved to fly only a maximum of 12 km, and more gen-erally only 4.5 km (Kaufmann and Briegel, 2004: 140), itis likely that the extent of malaria is greater than thatof the minimum extent of the mosquito habitat becausepeople, animals and mosquitoes all migrate. In this con-text, the interpolation map is a useful indicator of thefull extent of malarial transmission.In Figure 6, observations of ‘‘ague’’ during the 19th

century (Nuttall et al., 1901) have been overlain onto theinterpolated map of A. atroparvus distribution to assessconsistency between the two datasets. The distributionof ‘‘hot’’ and ‘‘cold’’ spots for cribra orbitalia values weresubsequently overlain onto the interpolation map of A.atroparvus observations to examine any spatial correla-tions between prevalence and A. atroparvus distribution(Fig. 7).The observations of the distribution of A. atroparvus

mosquitoes collated by Lang (1918) were not the resultof a systematic study but were merely a record of posi-tive sightings, as Lang himself stressed. In order tomore accurately define the locales most likely to presenta suitable habitat for A. atroparvus mosquitoes, a 50 mresolution map was created based upon a query thatsearched for the full extent of superficial soil types con-taining the plotted points of A. atroparvus observationthat were of a slope of less than 18 (i.e., of flat topogra-phy) over a 3 3 3 m2 moving panel. Mosquitoes areweak fliers, generally reaching no higher than 5–6 m up

Fig. 4. Spatial analysis of cribra orbitalia values showing‘‘hot’’ and ‘‘cold’’ spots. High values (1 and x) 5 ‘‘hot’’ and lowvalues (diamonds) 5 ‘‘cold’’ (Z scores = 1.96 and = 21.96 are sig-nificant to the P < 0.05 level).

Fig. 3. The relationship between CPR and TPR for cribraorbitalia.

Fig. 5. Distribution of anopheline mosquito presence (afterLang 1918).



from ground level, very often keeping much closer toground level (Service, 1971). This map represents, there-fore, the minimum extent of the most likely areas ofmosquito habitats or wetland marsh (Fig. 8).To quantify the relationship between cribra orbitalia and

the geographical variables, each Anglo-Saxon cemeterysite was subsequently scored as ‘‘Malarial’’ or ‘‘NonMa-larial,’’ according to whether one or more 50m grid squarespositive for malarial habitat or an identified ‘‘malarial’’location point of historically observed ‘‘ague’’ fell within a 2km buffer zone around each site (Table 1). Statistical anal-ysis using a two tailed Fisher’s Exact Test was undertakento identify any meaningful correlation between ‘‘malarial’’and ‘‘nonmalarial’’ categorization and cribra orbitaliaCPRs. The same process was undertaken using dentalenamel hypoplasia CPRs to identify any similarity or dif-ference in environmental associations between the twoconditions. Where two sites were located within the same 2km buffer zone and the geology types present were consis-tently similar across the zone, an average was taken of theCPRs for each buffer zone to maximize the population sizeassociated with each zone, thus helping to overcome anyinconsistencies in archaeological sampling and resulting ina more representative prevalence rate. The final cribraorbitalia CPRs and dental enamel hypoplasia CPRs werecategorized as above or below average for the purposes ofthe statistical test (Dunn and Clark, 2009:141–164).

RESULTS

GIS spatial distribution mapping indicates a similardistribution pattern in the historically observed locations

of ‘‘ague’’ outbreaks and the inverse distance weightedinterpolated map of A. atroparvus distribution in Eng-land. It can be inferred from the overlap in distributionthat the two proxy measures for the likelihood ofmalaria are reasonably reliable in assessing the extentof historical malaria and malarial habitats in England.The spatial distribution of the hot- and coldspots in

cribra orbitalia CPRs identified by the Getis Ord GeneralGi* test plotted against the interpolated values ofA. atroparvus distribution identifies the cribra orbitaliahotspot as occurring in an area positive for A. atroparvusand, thereby, ‘‘ague.’’ Conversely, the cold spot for cribraorbitalia occurred in an area negative for A. atroparvusand ‘‘ague.’’ An association between high prevalencerates in cribra orbitalia and areas positive for A. atropar-vus and, thereby, ‘‘ague’’ is demonstrated.The Fisher’s exact test identified a statistically signifi-

cant association between areas identified as ‘‘malarial’’or ‘‘nonmalarial’’ (using the geographical factors of slopeand superficial soil types in association with the locationpoints of historically recorded A. atroparvus and Ague)and Anglo-Saxon cemetery sites with above or below av-erage cribra orbitalia CPR’s (two-sided P value 50.0077). In contrast, no statistical relationship betweenthe ‘‘malarial’’ or ‘‘nonmalarial’’ cemetery sites and den-tal enamel hypoplasia CPRs was observed (two-sidedP value 5 0.7098).

Fig. 6. The distribution of ‘‘ague’’ recorded during in the19th century overlaying an interpolation map of the A. atropar-vus distribution (darker areas indicating greater anophelinepresence). Fig. 7. Interpolated inverse distance weighted map of A.

atroparvus presence and distribution of hot and cold spots forcribra orbitalia CPR. Purple and blue areas of shading indicatehigh concentrations of A. atroparvus (Z scores = 1.96 and =21.96 are significant to the P < 0.05 level).



DISCUSSION

The interpretation of cribra orbitalia lesions has longbeen a matter of debate in the paleopathological litera-ture, mainly due to the lack of a clinical base from whichto interpret them. Spatial distribution analysis of cribraorbitalia using GIS provides a new perspective fromwhich to interpret these lesions. There is a similarity inspatial distribution between the historically recordedpresence of the malarial vector A. atroparvus, ‘‘ague,’’and the cribra orbitalia hotspots at a national level. Fur-ther, there is a statistically significant correlationbetween cribra orbitalia CPRs with more specific geo-graphical locales, including the factors of slope and su-perficial soil types, suggesting that wetland populationshad a lower health status than those in dryland areas.Individuals living in low-lying and Fenland regions ex-hibit overall higher levels of cribra orbitalia than thosein nonmarshy locales, in particular, Northumbrian popu-lations in areas of hard-rock coastline. The lack of anycorresponding correlation between dental enamel hypo-plasia and geographical location suggests that specificdisease processes related to local environmental factorsproduced cribra orbitalia lesions, and were either dis-tinct from or more sensitive to these factors than thoseproducing dental enamel hypoplasia.There are a few outliers of high values of cribra orbita-

lia that do not correspond to marshland environments,but this is to be expected given the vagaries of archaeo-logical sample composition and more significantly, therange of etiological factors that cause these lesions. Onesuch site (Marina Drive) is a small sample of a mid-lateSaxon date with a relatively high proportion of nona-

dults (58.1%), which may partly explain the higher levelsof cribrotic lesions observed here, since cribra orbitalia isoften more frequently observed in nonadults. The quan-tification of cribra orbitalia prevalence rates may lead toparadoxical issues; a population of high comorbidity andhealth stress is likely to have a high rate of childhoodmortality and archaeologically would be manifest in thepresence of higher numbers of nonadults. To compareprevalence of cribra orbitalia per population, age-relatedmortality is not considered here, nor is the question ofthe extent to which morbidity provides a life-sustainingadaption to an environment and the extent to which thisis represented by cribra orbitalia lesions. However, toensure that there are not biases in the date presentedhere due to different proportions of nonadults/adultsbetween the cemetery sites, an examination of the demo-graphics between the ‘‘malarial’’ and ‘‘nonmalarial’’ popu-lations was undertaken. There is no statistically signifi-cant difference in the proportion of nonadults (Unpairedt test P 5 0.80) and no age-related effects contributingto the correlation observed. A lack of higher numbers ofsubadults in the ‘‘malarial’’ populations may beexplained by the fact that P. vivax infections are associ-ated with low mortality but chronic morbidity. However,it is not possible to quantify sampling issues with regardto preservation and limited excavations, where nona-dults are frequently observed to be under-represented(Buckberry, 2000).Another factor for consideration is that this study

rests on the assumption that individuals interred withinAnglo-Saxon cemeteries, and the pathologies their skele-tal remains display, are related to the local environmentof the cemetery site. However, it is possible that anynumber of individuals interred in a particular cemeterymigrated into the area after childhood, when it is gener-ally accepted that cribra orbitalia lesions are formed.There is currently little scientific knowledge of Anglo-Saxon migration; we can only state that from the pres-ent limited isotopic studies, groups do appear to containindividuals of both local and nonlocal origins. Signifi-cantly, for cribra orbitalia prevalence, studies have indi-cated that subadults were most frequently local individu-als (Montgomery et al., 2005).The results of this study correspond to those by Gow-

land and Garnsey (2010) of cribra orbitalia and enamelhypoplasia in Roman Italy. Cribra orbitalia has longbeen recorded in particularly high frequencies in Medi-terranean populations (Keenleyside and Panayatova,2006; Walker et al., 2009) and is most often interpretedas resulting from an inadequate weaning diet and poorsanitation (e.g., Salvadei et al., 2001, Facchini et al.,2004). Gowland and Garnsey (2010) found that siteswith the highest recorded prevalence of cribra orbitaliacoincided with marshy areas where malaria is believedto have been endemic. Likewise, no corresponding pat-tern was observed with respect to enamel hypoplasia.Gowland and Garnsey (2010) concluded that cribra orbi-talia prevalence at many sites in Roman Italy were asso-ciated with acquired hemolytic anemia caused directlyby malaria.It should be recognized that the anemia of P. vivax

malaria is complex in etiology. Contemporary researchdemonstrates that areas in which malaria is endemiccontain a wide variety of coexisting diseases and arereferred to as hotspots for comorbidity (see, for example,Kazembe et al., 2007). Many of these co-occurring dis-eases cause anemia as part of their own pathological

Fig. 8. Distribution of superficial soils of low gradient withA. atroparvus presence (identified as ‘‘Malarial’’) and cribra orbi-talia prevalence.



processes, and in addition may interact with and exacer-bate malarial anemia in the individual (Anstey et al.,2009: 222). For example, the severity of anemia has beenobserved to increase with coinfection of intestinal para-sites (Echeverri et al., 2003). However, although the im-portance of comorbidities in contributing to severe andfatal vivax malaria is likely to be underestimated (DaSilva Ventura et al., 1999; Fenn et al., 2005; Fernandezet al., 2008; Anstey et al., 2009: 225), malaria itself isbelieved to be the most common cause of severe anemia(Pasvol and Abdalla, 2004: 31).During the Anglo-Saxon period, the dramatic flooding

and later rise in temperature occurring as a result ofchanging climatic conditions would have created idealhabitats for mosquito abundance but concomitantlyunstable and unsanitary living conditions for people,such as is seen in marginalized fishing communitiestoday (Verduijn, 2000, Westaway et al., 2009). That theFenlands were indeed a treacherous area of wetland dur-ing the Saxon period was recorded in the early eighthcentury in the Life of St. Guthlac;

‘‘There is in the Midland district of Britain a most dismal fen ofimmense size, which begins at the banks of the river Grantanot far from the camp which is called Gronte (Cambridge) andstretches from the south as far north as the sea. It is a verylong tract, now consisting of marshes, now of bogs, sometimeswith black waters overhung by fog, sometimes studded withwoodland islands and traversed by the windings of tortuoussteams.’’ (Felix’s Life of Guthlac, cited by Hill, 1981:11).

Adaptation to wetland environments during the Earlyand Middle Saxon period is thought to have involved atransient and seasonal pattern (Rippon, 2002: 58). Exca-vations of Anglo-Saxon settlements demonstrate thatthose dwelling on the Fen-edge, exploiting the fertile wet-land and coastal lowlands, would have lived in woodenhousing with earthen floors, in close proximity to live-stock, with animals usually housed within the same dwell-ing (Carr et al., 1988). Surrounding areas of stagnantwater and a relatively warm climate would also have pro-vided an ample habitat for the mosquito A. atroparvus,especially where artificial reclamation may have extendedmosquito breeding grounds (Serandour et al., 2007). Thismosquito has been shown to over-winter in dwellings inclose proximity to their blood meals (humans or animals),thus usually infecting several inhabitants of the samedomicile (Reiter, 2000; Hulden et al., 2005). The ability tothrive in the ever-changing marshlands is a testament tothe resilience and ingenuity of the local populations.Nonetheless, reclaiming the wetlands would have been ahazardous occupation and would have presented a num-ber of serious health risks to the resolute inhabitants.Climate is a critical factor influencing infectious dis-

ease prevalence. Although archaeological skeletal data-sets are compromised by taphonomic factors as well asinterpretational and etiological complexities, they pro-vide our only means of analyzing this relationship overlonger time-periods. By linking skeletal evidence withthe increasing quantities of data available for past ecol-ogy and climate, we can essentially ‘‘populate’’ this envi-ronmental evidence and examine the impact of climatechange fluctuations on the health and demography ofpast peoples. While we may be technologically far-removed from those living in the past, these data mayprovide an important dimension for modeling futurehealth in relation to climate change predictions.

CONCLUSIONS

This research has used the techniques of spatial epide-miology in a paleopathological context to address the issueof morbidity in the wetlands in relation to the putativepresence of malaria. The study has highlighted a numberof important and significant methodological and interpre-tive findings, which may be summarized as follows:

1. There is a statistically significant correlation betweencribra orbitalia prevalence and underlying geologyand topography in eastern England.

2. There is no such correlation with enamel hypoplasia,indicating that it is not generalized health stress thatresulted in the observed cribrotic lesions, but specificconditions that are causing acquired anemia.

3. There is a spatial correlation between elevated levelsof cribra orbitalia and putative evidence for histori-cally recorded A. atroparvus and malaria.

4. Environmental evidence indicates that temperatureand precipitation in marshland areas during the earlyto mid-Anglo-Saxon period would have provided favor-able breeding conditions for A. atroparvus.

5. We hypothesize that endemic malaria, in conjunctionwith related comorbidities, are responsible for theobserved pattern of cribra orbitalia. These data poten-tially provide the earliest indirect evidence formalaria in England.

6. Future developments in biomolecular techniques mayallow Plasmodium vivax to be detected directly inarchaeological bone, but this is not yet possible.

7. Given the nonspecific etiology of many skeletal indica-tors of health stress a consideration of specific localconditions may ultimately aid in deciphering morecontext-specific etiological causes. Generalized inter-pretations that more often abound in the paleopatho-logical literature are not always helpful in under-standing the prevalence and distribution of theselesions.

8. Spatial epidemiological studies using GIS are notcommon in palaeopathology, but can be an invaluabletool for integrating historical, geographical, and paleo-pathological variables.

ACKNOWLEDGMENTS

For data and contextual information, the authors thankAbby Antrobus, Anthea Boylston, Victoria Brown, Jo Buck-berry, Anwen Caffell, Andrew Chamberlain, Andy Chap-man, Stephen Coleman, Elizabeth Craig, Nick Crank,Natasha Dodwell, Corinne Duhig, Holly Duncan, NaomiField, Julia Habeshaw, Gail Hama, Betina Jakob, DavidKlingle, Tom Lane, Andrew Newton, Rose Nicholson, Nata-sha Powers, Charlotte Roberts, Brett Thorn, and JudithWalton. They would also like to thank all HER staff. Theauthors are extremely grateful to Tim Thompson, AndrewMillard, Graham Philip, Peter Rowley-Conwy, MikeChurch, the editors at AJPA, and two anonymousreviewers for their insightful comments on earlier drafts ofthis article. All errors remain their own. This research wasfunded by the British Academy (ref SG090361).

LITERATURE CITED

Abdalla SH, Pasvol G. 2004. Pathogenesis of the anemiaof malaria. In: Abdalla SH, Pasvol G, editors. Malaria: ahaematological perspective. London: Imperial College Press.p 61–96.



Abdalla SH, Wickramasinghe SN. 2004. The bone marrow inhuman malaria. In: Abdalla SH, Pasvol G, editors. Malaria: ahaematological perspective. London: Imperial College Press.p 213–248.

Aksoy M, Camli N, Erdem S. 1966. Roentgenographic bonechanges in chronic iron deficiency anemia. Blood 27:677–687.

Angel LEW. 1966. Porotic hyperostosis anemias malarias andmarshes in prehistoric eastern Mediterranean. Science153:760–762.

Anstey NM, Russell B, Yeo TW, Price RN. 2009. The pathophys-iology of vivax malaria. Trends Parasitol 25:220–227.

Balloch S, Memon SA, Gachal GS, Baloch M. 2011. Determina-tion of trace metals abnormalities in patients with vivaxmalaria. Iran J Parasitol 6:54–59.

Bassat Q, Alonso PL. 2011. Defying malaria: fathoming severePlasmodium vivax disease. Nat Med 17:48–49.

Bianucci R, Mattutino G, Lallo R, Charlier P, Jouin-Spriet H,Peluso A, Higham T, Torre C, Rabino Massa E. 2008. Immu-nological evidence of Plasmodium falciparum infection in anEgyptian child mummy from the early dynastic period. J ArchSci 35:1880–1885.

Brandt E, Weichmann I, Grupe G. 2002. How reliable areimmunological tools for the detection of ancient protein in fos-sil bones. Int J Osteoarcheol 12:307–316.

Brothwell DR. 1981. Digging up bones, 3rd ed. Ithaca, NY: Cor-nell University Press.

Bruce-Chwatt LJ. 1976. Ague as malaria. J Trop Med Hyg79:168–176.

Buckberry JL. 2000. Missing, presumed buried? Bone diagenesisand the under-representation of Anglo-Saxon children. Assem-blage 5:http://ads.ahds.ac.uk/catalogue/adsdata/assemblage/html/5/buckberr.html.

Buckley HR. 2006. ‘The predators within’: investigating the relation-ship between malaria and health in the prehistoric Pacific Islands.In: Oxenham M, Tayles N, editors. Bioarchaeology of SoutheastAsia. Cambridge: Cambridge University Press. p 309–332.

Buckley HR, Tayles N. 2003. Skeletal pathology in a prehistoricPacific island skeletal sample: issues in lesion recording,quantification and interpretation. Am J Phys Anthropol122:303–324.

Buikstra JE, Ubelaker DH. 1994. Standards for data collectionfrom human skeletal remains. Fayetteville, AR: ArkansasArchaeological Survey Research Series no. 44.

Calton JM, Escalante AA, Neafsey D, Volkman SK. 2008. Com-parative evolutionary genomics of human malaria parasites.Trends Parasitol 24:545–550.

Cameron ML. 1993. Anglo-Saxon medicine. Cambridge: Cam-bridge University Press.

Carr R, Tester A, Murphy P. 1988. The middle Saxon settlementat Staunch Meadow, Brandon. Antiquity 62:371–377.

Chang K-H, Stevenson MM. 2004. Malarial anemia: mecha-nisms and implications of insufficient erythropoiesis duringblood stage malaria. Int J Parasitol13–14:1501–1516,1572.

Cracknell B. 2005. Outrageous waves: global warming andcoastal change in Britain through two thousand years. Wilt-shire, UK: The Cromwell Press.

Da Silva Ventura A, Pinto A, Silva R. 1999. Malaria por Plas-modium vivax em craiancas e adolescents: aspectos epideliolo-gicos, clınicos e laboratoriasis. J Pediatr (Rio J) 75:187–194.

De Smith MJ, Goodchild MF, Longley PA. 2007 GeoSpatial anal-ysis: a comprehensive guide to principles, techniques and soft-ware tools. Leicester, UK: Matador.

Dobson MJ. 1994. Malaria in England: a geographical and his-torical perspective. Parassitologia 36:35–50.

Dobson MJ. 1997. Contours of death and disease in early mod-ern England. Cambridge: Cambridge University Press.

Dunn OJ, Clark VA. 2009. Basic statistics: a primer for the bio-medical sciences, 4th ed. Hoboken, NJ: Wiley.

Echeverri M, Tobon A, Alvarez G, Carmona J, Blair S. 2003.Clinical and laboratory findings of Plasmodium vivax malariain Colombia, 2001. Rev Inst Med Trop Sao Paulo 45:29–23.

Facchini F, Rastelli E, Brasili P. 2004. Cribra orbitalia and cri-bra cranii in Roman skeletal remains from the Ravenna areaand Rimini (I-IV AD). Int J Osteoarcheol 14:126–136.

Fairgrieve SI, Molto JE. 2000. Cribra orbitalia in two tempo-rally disjunct population samples in the Dahkleh Oasis,Egypt. Am J Phys Anthropol 111:319–331.

Fenn B, Morris SS, Black RE. 2005. Comorbidity in childhoodin northern Ghana: magnitude, associated factors, and impacton mortality. Int J Epidemiol 34:368–375.

Fernandez J, Idrovo A, Cucunuba Z, Reyes P. 2008. Validity ofstudies on the association between soil-transmitted helminthsand the incidence of malaria: should it impact on health poli-cies? Rev Bras Epidemiol 11:1–13

Fornaciari G, Guiffra V, Ferroglio E, Gino S, Bianucci R. 2010.Plasmodium falciparum immunodetection in bone remains ofa member of the Renaissance Medici family (Florence, Italy,16th Century). Trans R Soc Trop Med Hyg 104:583–587.

Gomez KW, Gomez AA. 1984. Statistical procedures for agricul-tural research. New York: Wiley.

Gowland RL, Garnsey P. 2010. Skeletal evidence for health,nutritional status and malaria in Rome and the empire. JRom Arch Suppl Series 78:131–156.

Hershkovitz I, Rothschild BM, Latimer B, Dutour O, LeonettiG, Greenwald CM, Rothschild C, Jellema LM. 1997. Recogni-tion of sickle cell anemia in skeletal remains of children. AmJ Phys Anthropol 104:213–226.

Hill D. 1981. An atlas of Anglo-Saxon England. Oxford: Black-well Publishing.

Hulden L, Hulden L, Heliovaara K. 2005. Endemic malaria: anindoor disease in northern Europe. Historical data analysed.Malar J 4 Published online http://www.malariajournal.com/content/4/1/19.

Hutchinson RA, Lindsay SW. 2006. Malaria in the Englishmarshes. Lancet 367:1947–1952.

James SP. 1929. The disappearance of malaria from England.Proc R Soc Med 23:71–87.

Kaufmann C, Briegel H. 2004. Flight performance of themalaria vectors Anopheles gambiae and Anopheles atroparvus.J Vector Ecol 29:140–153.

Kazembe LN, Muula AS, Appleton CC, Kleinschmidt I. 2007.Modelling the effect of malaria endemicity on special varia-tions in childhood fever, diarrhoea and pneumonia in Malawi.Int J Health Geogr 6: 33. Published online http://.ij-health-geographics.com/content/6/1/33.

Keenleyside A, Panaytova K. 2006. Cribra orbitalia and porotichyperostosis in a Greek colonial population (5th to 3rd centu-ries BC) from the Black Sea. Int J Osteoarchaeol 16:373–384.

Kuhn KG, Campbell-Lindrum D, Armstrong B, Davies R. 2003.Malaria in Britain: past, present and future. Proc Natl AcadSci USA 100:9997–10001.

Lagia A, Eliopoulos C, Manolis S. 2007. Thalassemia: macroscopicand radiological case study. Int J Osteoarchaeol 17:269–285.

Lang WD. 1918. A map showing the known distribution in Eng-land and Wales of the anopheline mosquito, with explanatorytext and notes. London: British Museum.

Lewis ME. 2007. The bioarchaeology of children. Cambridge:Cambridge University Press.

Lewis ME. Early View. Thalassaemia: its diagnosis and inter-pretation in past skeletal populations. Int J Osteoarchaeol.Published online, DOI: 10.1002/oa.1229.

Lindsay SW, Hole DG, Hutchinson RA, Richards SA, Willis SG.2010. Assessing the future threat from vivax malaria in theUnited Kingdom using two markedly different modellingapproaches. Malar J 9:70.

Lindsay SW, Thomas CJ. 2001. Global warming and risk ofvivax malaria in Great Britain. Global Change Hum Health2:80–84.

Marchant P, Eling W, van Gemert G-J, Leake CJ, Curtis C F.1999. Could British mosquitos transmit falciparum malaria?Parasitol Today 14:344–345.

Montgomery, JA, Evans JA, Powlesland D, Roberts CA. 2005.Continuity or colonization in Anglo-Saxon England? Isotopeevidence for mobility, subsistence practice, and status at WestHeslerton. Am J Phys Anthropol 126:123–128.

Nerlich AG, Schrout B, Dittrich S, Thomas J, Zink AR. 2008.Plasmodium falciparum in Ancient Egypt. Emerg Infect Dis14:1317–1319.



Nicholls A. 2000. Fenland ague in the nineteenth century. MedHist 44:523–550.

Nuttall GHF, Cobbett L, Strangeways-Pigg T. 1901. Studies inrelation to malaria 1. The geographical distribution of anophe-les in relation to the former distribution of ague in England.J Hyg 1:1–44.

Ortner DJ. 2003. Identification of pathological conditions inhuman skeletal remains. New York: Academic Press

Oxenham M, Carvill I. 2010. Porotic hyperostosis and cribraorbitalia: the erythropoietic response to iron deficiency ane-mia. Anthropol Sci 118:199–200.

Pasvol G. Abdalla SH. 2004 Definition, epidemiology, diagnosisand management of the anemia of malaria. In: Abdalla S,Hoffman S L, editors. Malaria, a haematological perspective.London: Imperial College Press. p 29–60.

Pfeiffer D, Robinson T, Stevenson M, Stevens K, Rogers D,Clements A. 2010. Spatial analysis in epidemiology. Oxford:Oxford University Press.

Pinello C. 2008. Attempted ancient DNA detection of Plasmo-dium vivax in Medieval and Post-Medieval Britain. Unpub-lished PhD, University of Manchester.

Reiter P. 2000. From Shakespeare to Defoe: malaria in Englandin the Little Ice Age. Emerg Infect Dis 6:1–11.

Reiter P, Thomas C J, Atkinson PM, Hay SI, Randolph SE, Rog-ers DJ, Shanks GD, Snow RW, Spielman A. 2004. Globalwarming and malaria: a call for accuracy. Lancet Infect Dis4:323–324.

Rippon S. 2000. The transformation of the coastal wetlands: ex-ploitation and management of the marshland landscapes innorthwest Europe during the Roman and Medieval Periods.Oxford: Oxford University Press.

Rippon S. 2002. Infield and outfield: the early stages of marsh-land colonisation and the evolution of the medieval field sys-tem. In: Lane T, Coles J, editors. Through wet and dry:essays in honour of David Hall. Lincolnshire: Heritage Trustof Lincolnshire/WARP Occasional Paper,17. p 54–70.

Rippon S. 2010. Landscape change during the ‘long eighth cen-tury’ in Southern England. In: Higham NJ, Ryan MJ, editors.Landscape archaeology of Anglo-Saxon England. Woodbridge,Suffolk: The Boydell Press. p 39–64.

Rothschild BM. 2002. Hair standing on end as a manifestationof iron deficiency? Radiology 224:609–610.

Sallares R. 2002. Malaria and Rome. Oxford: Oxford UniversityPress.

Sallares R, Bouwman A, Anderung C. 2004. The spread ofmalaria to Southern Europe in antiquity: new approaches toold problems. Med Hist 48:311–328.

Salvadei L, Ricci F, Manzi G. 2001. Porotic hyperostosis as amarker of health and nutritional conditions during childhood:

studies at the transition between Imperial Rome and theearly Middle Ages. Am J Hum Biol 13:709–717.

Serandour J, Girel J, Boyer S, Ravanel P, Lemperiere G, Rave-ton M. 2007. How human practices have affected vector-bornediseases in the past: a study of malaria transmission in Al-pine valleys. Malar J 6:115–125.

Service MW. 1971. The daytime distribution of mosquitoes rest-ing in vegetation. J Med Entomol 8:271–278.

Setzer TJ. 2010. Malaria in prehistoric Sardinia (Italy): an ex-amination of skeletal remains from the middle Bronze Age.Unpublished PhD, University of South Florida.

Shute PG, Maryon M. 1974. Malaria in England, past, presentand future. J R Soc Promot Health 94:23–29.

Soren HD. 2003. Can archaeologists excavate evidence ofmalaria? World Archaeol 35:193–205.

Stuart-Macadam P. 1985. Porotic hyperostosis: representative ofa childhood condition. Am J Phys Anthropol 66:391–398.

Stuart-Macadam, P. 1991. Anaemia in Roman Britain. In: BushH and Zvelebil M, editors. Health in past societies. Biocul-tural interpretations of human remains in archaeological con-texts. Oxford, Tempus Reparatum, British ArchaeologicalReports. Int Series 567:101–113.

Stuart-Macadam P. 1992. Porotic hyperostosis: a new perspec-tive. Am J Phys Anthropol 87:39–47.

Taylor GM, Rulland P, Molleson T. 1997. A sensitive polymerasechain reaction method of the detection of plasmodium speciesDNA in ancient human remains. Anc Biomol 1:193–203.

Verduijn RJC. 2000. Basic needs of 39 coastal fishing commun-ities in Kanniy Akumari district, Tamil Nadu, India: a surveyto investigate and prioritise problems regarding services andinfrastructure. India: Bay of Bengal Programme BOBPIMM/1,Chennai, India.

Vigorita VJ. 2008. Orthopaedic pathology, 2nd ed. Philadelphia:Lippincott, Williams and Wilkins.

Walker PL, Bathurst RR, Richman R, Gjerdrum T, AndrushkoVA. 2009. The causes of porotic hyperostosis and cribra orbita-lia: a reappraisal of the iron deficiency-anemia hypothesis.Am J Phys Anthropol 139:109–125.

Wapler U, Crubezy E, Schultz M. 2004. Is cribra orbitalia syn-onymous with anemia? Analysis and interpretation of cranialpathology in Sudan. Am J Phys Anthropol 123:333–339.

Westaway E, Barratt C, Seeley J. 2009. Educational attainmentand literacy in Ugandan fishing communities: access for all?MAST 8:74–97.

Wickramasinghe SN, Abdalla SH. 2000. Blood and bone marrowchanges in malaria. Best Pract Res Clin Haematol 13:277–299.

Zink AR, Reischl U, Wolf H, Nerlich AG. 2002. Molecular analysis ofancient microbial infections. FEMSMicrobiol Lett 213:141–147.



Morbidity in the Marshes: Using Spatial Epidemiology to ... western malaria ajpa... · Morbidity in the Marshes: Using Spatial Epidemiology to Investigate Skeletal Evidence for Malaria

Documents