Hookworm (Necator americanus) transmission in inland areas of sandy soils in KwaZulu-Natal, South Africa

Authors


Authors
M. L. H. Mabaso, National Malaria Research Programme, South African Medical Research Council, P.O. Box 70380, Overport 4067, South Africa. Fax: +27 31 203 4700; E-mail: mabasom@mrc.ac.za (corresponding author).
C. C. Appleton, School of Life & Environmental Sciences, University of Natal, Durban 4041, South Africa. E-mail: appleton@biology.und.ac.za
J. C. Hughes, School of Applied Environmental Sciences, University of Natal, Private Bag X01, Scottsville 3209, South Africa. E-mail: hughesj@nu.ac.za
E. Gouws, 16 Rue de la Canonniere, Geneva, Switzerland. E-mail: gouwse@who.int

Summary

This study extended the association between hookworm transmission in KwaZulu-Natal and the sandy coastal plain by investigating the parasite's occurrence in isolated areas of sandy soils further inland. A school-based prevalence survey was carried out in selected inland sandy areas and in surrounding areas dominated by clay soils within a narrow altitudinal range of between 500 and 700 m to reduce the effect of altitude on climate-related factors (rainfall and temperature). Sandy areas situated on the coastal plain were included in the analysis for comparative purposes. Soil samples (0–50 mm depth) were collected from each locality to assess their nematode loadings and to analyse selected physical and chemical properties. Significant differences were found between the moderate prevalence of hookworm infection among children living in inland areas with sandy soils (17.3%) and the low prevalence in surrounding non-sandy areas (5.3%, P < 0.001), and between infection among children living in all inland areas (9.3%) and the high prevalence on the coastal plain (62.5%, P < 0.001). Amounts of fine and medium sand were highest in both the coastal plain soils and in inland sandy areas and these fractions showed a significant positive correlation with hookworm prevalence and nematode loadings. Clay, coarse sand and organic matter contents were highest in surrounding non-sandy soils and showed a significant negative correlation with the nematode variables. No statistically significant correlations were found with soil pH at study localities. We conclude that properties of inland sandy soils, particularly particle size distribution, correlate well with hookworm prevalence and nematode loadings and therefore provide a more suitable habitat for nematodes than surrounding non-sandy areas. These results suggest that particle size distribution of sand fractions, organic matter and clay content in the soil influence the survival of hookworm larvae and hence the parasite's transmission.

Introduction

Human parasitic nematodes exhibit varying degrees of complexity in their modes of transmission. For soil-transmitted intestinal nematodes, development in the soil is of fundamental epidemiological significance since after a focus of transmission becomes established in the soil, it will be maintained only as long as the prevailing conditions in the interstitial environment permit, primarily temperature and moisture (WHO 1964). This is particularly true for hookworms (Ancylcostoma duodenale and Necator americanus), which have thin-shelled eggs and free-living larvae that are very sensitive to desiccation and cold temperatures. Other common soil-transmitted intestinal nematodes such as Ascaris lumbricoides (roundworm) and Trichuris trichiura (whipworm) have no free-living larval stages but rather thick-shelled eggs that allow for prolonged survival under harsh environmental conditions (WHO 1964). As such hookworms require particular soil types and tolerate only a narrow range of physical conditions within the soil.

Several studies have proposed an association between sandy soil types and hookworm infection: Augustine and Smile (1926) and Beaver (1953) on the coastal plain of the southern USA; Hsieh et al. (1971) on the narrow coastal belt of Liberia; Michaelsen (1985) in sandy areas of northern Botswana, and Evans and Joubert (1989) and Evans et al. (1990) in the Kalahari sands of north-eastern Namibia. In KwaZulu-Natal, South Africa, studies by Appleton and Gouws (1996) and Appleton et al. (1999) have shown that N. americanus, the dominant hookworm species, is confined to the low-lying, sandy coastal plain where it occurs at a prevalence that ranges from 95% in the north to 42% in the south, i.e. prevalence decreases as the coastal plain narrows with increasing south latitude.

In a more recent study, Mabaso et al. (2003) showed that this pattern of lowland transmission was determined by altitude-dependant climatic variables, especially temperature and soil-related factors, particularly clay content. When these variables were analysed together in a multivariate analysis the influence of soil type on hookworm distribution was masked by the fact that the effect of climate on the soil depends largely on the nature of the soil itself (Foth 1984). Understanding the environmental limits of these parasites can improve our knowledge of their biology and epidemiology and in turn aid the targeting of national control strategies (Brooker et al. 2002). This study therefore examines further the relationship between the physical and chemical properties of sandy and non-sandy soils and hookworm transmission in areas where the climate, particularly temperature, may be limiting.

There are several areas in KwaZulu-Natal where sandy soils similar to those on the coastal plain occur inland and these offer a unique opportunity to further probe the role of soil type in the complex interaction between hookworm transmission and climatic factors. The aim of this study is to determine hookworm prevalence in selected communities located on inland sandy soils to see whether transmission occurs away from the coastal plain and if so, how it is affected by the change in location and climate (most importantly temperature). We expect that if hookworm is present in the areas of interest its prevalence should at least be higher than in the surrounding areas characterized by clay soils. We also analyse characteristics of the different soil types for possible correlations with nematode loadings and prevalence data in order to assess their suitability for the free-living and infective larval stages.

Materials and methods

Selection of study localities

To assess the presence of hookworm infection in areas of sandy soils situated inland of the coastal plain in KwaZulu-Natal, the Durban Land Type Map (Land Type Survey Staff 1986) was used to select seven rural communities within the Maphumulu district (Figure 1). One school was randomly selected from each community. To control for the effect of altitude on climate-related factors (rainfall and temperature), the inland study localities were chosen to lie within a narrow altitude range of 500–700 m. Three schools (Nyamazane, Mushane and Tshelabantu) were thus selected from communities situated on inland sandy soils (red dune cordon sands) and four from outside this area of interest, on high clay-content soils (dark coloured and red clayey soils). Two of the latter were in the north (Empungeni and Thimuni) and the other two in the south (Mary Gray and Nsuze-Gcwensa). Two schools (Sakhesethu and Sizani) situated on the red dune cordon sands of the coastal plain were included in the analysis for comparative purposes.

Figure 1.

Map of the study area (Maphumulo district, KwaZulu-Natal) showing the distribution of the selected schools situated on (i) the red dune cordon sands and (ii) the dark coloured and red clayey soils.

Field and laboratory methods

Survey data were collected from primary school pupils in grades one to five (age 5–20 years). Faecal samples were collected from 12 boys and 12 girls selected randomly from each grade at each school. Identification of hookworm species and assessment of prevalence (% infected) were done using the modified Harada–Mori test tube cultivation technique (Goldsmid 1967) and the modified formal-ether stool examination concentration method (Allen & Ridley 1970), respectively, as described by Mabaso et al. (2003).

Recovery of soil nematodes

Counts of total nematode loadings (i.e. free-living and parasitic forms) in soil samples from the nine study localities were used to provide an indirect measure of the suitability of these soils to support the free-living and infective stages of N. americanus. Ten 200-g soil samples were collected from each locality, using a spade to scoop soil from the surface horizon (0–50 mm depth). Each 200-g soil sample was transferred to a separate jar.

Soil samples were collected (i) at least 100 m apart in schoolyards and play grounds, (ii) about 300 m from the school premises and (iii) from the immediate surroundings of local households. These were then taken to the laboratory where nematodes were recovered using a series of modified Baermann apparatus (Cheesebrough 1981).

Soil analysis

To investigate the characteristics of the different soil types for possible correlation with epidemiological data, 200-g portions of the soil samples from all nine study localities were analysed for the following: pH in a 1:5 soil:1M KCl solution; organic carbon by the Walkey–Black procedure (Walkey & Black 1934); particle size by the pipette method for clay (<0.002 mm), silt (0.002–0.05 mm), fine sand (0.05–0.25 mm), medium sand (0.25–0.50 mm) and coarse sand (0.50–2.00 mm) (Gee & Bauder 1986).

Statistical analysis

A total of 1093 pupils were examined from the seven schools located up to 53 km inland and from two schools situated on the coastal plain. To compare prevalence of hookworm infection among individuals, study localities situated on similar soil types were grouped. Four major groups were created: the coastal plain (CP) with two localities (n = 336), inland sandy area (SA) with three localities (n = 379) and inland non-sandy areas north (IN) with two localities (n = 189) and south (IS) of SA with two localities (n = 189). Differences in prevalence between groups of study localities were analysed using a chi-square test. Analysis of variance with Tukey pairwise multiple test comparison was carried out to test for differences in soil variables for combinations of study localities. Spearman correlation coefficients were used to assess relationships between prevalence of hookworm infection, nematode loadings and soil variables in the nine study localities (P < 0.05 and r > 0.5 were considered significant).

Results

Prevalence of hookworm infection

There were significant differences (χ2 = 27.3; P < 0.001) in the prevalence of hookworm infection among children living between the inland study localities situated in the SA with a moderate prevalence of 17.3% and surrounding localities situated north (IN) and south (IS) of SA with low prevalence of 5.0% and 5.5%, respectively. The chi-square test also showed a significant difference (χ2 = 321.1; P < 0.001) between hookworm prevalence among children living in all the inland study localities (9.3%) and those in localities on the CP with a prevalence of 62.5%.

Soil variables

Table 1 gives the results for particle size distribution, pH and organic matter for soils in the CP, SA, IN and IS localities. Statistical pairwise combinations for these data are given in Table 2. No significant differences were found between the combination of the coastal plain and inland sandy areas, and between non-sandy areas with respect to organic matter content. Further, no significant differences were found either between sandy areas with respect to proportion of fine sand or between non-sandy areas with respect to clay content. There was also no significant difference between combinations of all study localities with respect to pH. Although statistical significance was not tested, study localities differed with respect to mean nematode loadings with 348 worms per 200 g soil from CP followed by 126 worms per 200 g soil from SA, and 114 and 98 worms per 200 g soil from IN and IS, respectively.

Table 1.  Particle size distribution (expressed as percentages) and mean values (standard deviation) for pH and organic matter used in the analysis (n = 10 replicates in each locality)
Group of study localitiesCoarse sand, 0.50–2.00 (%)Medium sand, 0.25–0.50 (%)Fine sand, 0.05–0.25 (%)Silt, 0.002–0.05 (%)Clay, <0.002 (%)pH (KCl)Organic matter (%)
  1. CP refers to the coastal plain, SA the inland sandy area, IN and IS the surrounding non-sandy areas north and south of SA, respectively.

CP (n = 2)2.7 (1.09)42.5 (5.3)43.5 (5.6)5.5 (1.5)5.6 (1.03)5.8 (0.9)1.6 (1.11)
SA (n = 3)15.0 (5.9)18.0 (2.6)40 (10.7)9.3 (2.9)17.4 (5.7)5.6 (1.0)2.5 (1.6)
IN (n = 2)20.2 (2.3)16.1 (1.5)23.9 (2.8)6.4 (2.1)32.7 (2.1)6.4 (1.4)4.3 (3.4)
IS (n = 2)16.1 (3.9)16.6 (2.2)24.3 (2.5)8.4 (5.6)34.4 (2.2)5.4 (0.9)3.4 (1.2)
Table 2.  Comparison of soil variables between pairwise combinations of study localities showing mean difference (standard error) in particle size distribution
Pairs of study localitiesSoil variables
Coarse sand, 0.50–2.00 (%)Medium sand, 0.25–0.50 (%)Fine sand, 0.05–0.25 (%)Silt, 0.002–0.05 (%)Clay, <0.002 (%)pH (KCl)Organic matter (%)
  1. * The mean difference is significant at P < 0.05.

  2. CP refers to the coastal plain, SA the inland sandy area, IN and IS the surrounding non-sandy areas north and south of SA, respectively.

CPSA12.35* (1.17)24.49* (0.92) 3.39 (0.29)3.78* (0.87)11.85* (1.06)0.16 (0.31)0.94 (0.58)
CPIS13.43* (1.28)25.89* (1.01)19.10* (2.10)2.91* (0.95)28.85* (1.16)0.36 (0.34)1.82* (0.64)
CPIN17.21* (1.28)26.40* (1.01)19.49* (2.10)0.87 (0.95)27.16* (1.16)0.58 (0.34)2.69* (0.64)
SAIS 1.08 (1.17) 1.40 (0.92)15.70* (1.92)0.87* (0.87)17.00* (1.06)0.21 (0.31)1.88* (0.57)
SAIN5.21* (1.17) 1.91 (0.92)16.08* (1.92)2.90* (0.87)15.32* (1.16)0.74 (0.31)1.86* (0.64)
ISIN4.13* (1.28) 0.51 (0.95) 0.39 (2.10)2.04 (0.95) 1.67 (1.16)0.95 (0.34)0.86 (0.64)

Spearman correlation analysis

Only the medium and fine sand fractions showed significant positive correlations with hookworm prevalence and nematode loadings (Table 3). Clay content showed a significant negative correlation with hookworm prevalence and nematode loadings. The coarse sand fraction and organic matter showed moderate negative correlations, although not statistically significant, with both prevalence and nematode loadings. No statistically significant correlations were found with pH.

Table 3.  Correlations between mean hookworm prevalence, nematode loadings and soil variables in selected study localities (n = 9) (P < 0.05 and r > 0.5 are considered significant)
Soil variablesHookworm prevalenceNematode loadings
r valueP valuer valueP value
Coarse sand (%)−0.660.05−0.600.09
Medium sand (%)0.920.000.780.01
Fine sand (%)0.920.000.820.01
Silt (%)−0.170.660.001.00
Clay (%)−0.950.00−0.950.00
pH (KCl)0.250.510.200.61
Organic matter (%)−0.580.10−0.620.08

Discussion

The moderate hookworm prevalence among children living in the inland sandy area (17.3%) contrasted with the low prevalence (5.3%) among children living in the surrounding areas dominated by clay soils confirm the importance of sandy soil types in the transmission of hookworm infection. The difference in hookworm prevalence between the inland sandy area and the coastal plain (62.5%) can be attributed partly to the suppressing effect of lower temperatures inland, i.e. mean annual temperature of 17 °C (Schulze 1982) and partly to differences in their soil properties (Table 2). However, elsewhere in Africa local differences in hookworm infection have been associated with the interaction between abiotic factors such as rainfall and soil type and biotic factors such as human population density and hygiene (Tedla 1986). We therefore acknowledge the limitation imposed in our analysis by exclusion of demographic and socioeconomic variables.

In the present study, the soil variables used in the analysis showed similar correlations to both hookworm prevalence and nematode loadings (Table 3). Only the medium and fine sand fractions showed significant positive correlations and both these fractions are found in highest proportions in study localities situated on the coastal plain, followed by localities in the inland sandy area. Although proportions of medium sand are relatively high at all inland localities, clay content in non-sandy areas showed a strong negative correlation with both hookworm prevalence and nematode loading followed by the coarse sand fraction and organic matter, both of which are found in highest proportions in these areas. It is also important to note that although inland sandy soils (red dune cordon sands) have a much lower clay content (17.4%) than the surrounding non-sandy soils (dark coloured and red clayey soils) with 32.7–34.4% clay, their clay content is still considerably higher than the red dune cordon sands on the coastal plain (5.6%). However, these sandy areas are well drained and therefore provide a more suitable habitat for nematodes than surrounding clayey soils.

The disadvantage of clay soils is that they usually retain more water than sandy soils. Clay becomes sticky and plastic when wet because it is cohesive (Singer & Munns 1987) and, as a result, clay soils tend to be less well aerated than sandy soils and this probably restricts the movement of nematodes. Hookworm larvae are obligatory aerobes and migration through the soil allows them to move towards the surface or deeper into the soil if conditions (particularly cold temperatures and desiccation) become inimical to their survival (WHO 1964; Smith 1990). Sand particles are comparatively large and hence have a much smaller surface area to volume ratio than equal masses of clay or silt. Also, sand particles normally increase the size of interstitial spaces, facilitating the movement of air and drainage (Foth 1984). Such conditions provide sufficient oxygen and a suitably moist medium for larval hookworm survival and migration.

However, only the medium and fine sand fractions showed a positive correlation with hookworm prevalence and nematode loadings, while the coarse sand fraction showed significant negative correlations. Coarse sand occupies more volume but does not contribute much space for air or water in the soil (Singer & Munns 1987). In addition, coarse sands have high permeability and drain more quickly than fine sands, and thus tend to produce drier conditions. This probably explains the high hookworm prevalence and nematode densities in areas with low amounts of coarse sand in the soil (Table 1).

The observed inverse relationship between nematode variables and organic matter can be explained by the fact that light sandy soils are frequently well aerated, which restricts the accumulation of organic matter (Foth 1984). Although heavy clay soils may contain adequate organic material, there are few if any nematodes because the air and water spaces are greatly reduced (Crofton 1966). Since organic matter is a source of food and energy for the majority of soil organisms, for hookworms this could mean that the faecal micro-fauna or the organic material at the point of defecation are probably sufficient for the development of the feeding free-living larval stages (L1 and L2). While the pH of the soil influences the activity and relative abundance of different groups of soil organisms, the present narrow pH range of 5.4–6.4 does not allow for any generalizations to be made on its influence on hookworm transmission.

We conclude that properties of inland sandy soils, particularly particle size distribution, correlate well with hookworm prevalence and nematode loadings and therefore provide a more suitable habitat for nematodes than surrounding non-sandy areas. These results suggest that particle size distribution of sand fractions, organic matter and clay content of the soil influence the survival of hookworm larvae and hence the parasite's transmission.

Acknowledgements

We acknowledge the help and cooperation of the following people during this study: Mr Ndlovu (Senior Environmental Health Officer, Stanger), Mr V. Roberts (Department of Agriculture and Environmental Affairs, c/o Cedara College of Agriculture, Pietermaritzburg), Colleen Archer and Frank Sokolic (School of Life & Environmental Sciences, University of Natal, Durban). We are grateful to Carrin Martin (GIS Unit National Malaria Research Programme, South African Medical Research Council). The South African Medical Research Council provided financial support.

Ancillary