The distribution of small mammal burrows was highly skewed. Kolmogorov-Smirnov test indicated that the data did not fit a normal distribution (p<0.001 in both cases) either inside or outside fenced pastures. Normality was not obtained after Box-Cox transformations. Thus, the burrow density of small mammals was compared between open and fenced pastures by using nonparametric tests that considered landscape factor. Spearman correlation tests were used to quantify the relationship between the burrow density of small mammals on open pastures and the surface of fenced pastures in settlements in which the fenced areas were all measured, controlling for the landscape factor. A multiple logistic regression model was used to relate median burrow density of small mammals to dog infection in the settlements. The dependent variable was a presence/absence vector (0/1) (dog was negative or positive for E. multilocularis). Independent variables included dog's age and sex, droppings collected versus purged fecal samples, and median density of small mammal burrows. All these analyses used SPSS release 10 (SPSS, Chicago, IL, USA).
Landscape type influenced the abundance of small mammal burrows (p<0.001). Post hoc Tukey multiple comparison test on ranks confirmed that the densities of small mammal burrows were different among different landscape types (p<0.05), except for the comparison between flatland and piedmont. The burrow densities of small mammals on open pastures were significantly higher than those on fenced pastures in 3 of 4 landscapes ().
Comparisons of open and fenced pastures small mammal burrow densities, stratified by landscapes*
The Spearman correlations showed larger fenced areas associated with higher density of small mammal burrows in the open pastures in all landscape types (). The relationships between the area of fenced pastures and the burrow density of small mammals inside the fenced pastures in the 4 landscapes were not significant (valley [rs = –0.08, p = 0.32], flatland [rs = –0.46, p = 0.02], piedmont [statistics not applicable because of 3 observations only], and valley entrance [rs = 0.08, p = 0.736]), except for flatland.
Relationship between surface of fenced pastures and densities of small mammal burrows in open pastures*
Feces samples, of which 159 (63.1%) were purged, were obtained from 252 dogs (mean age 4.4 years; 183 males). E. multilocularis infection rate was 16.7% (42/252); the infection rate was 18.2% (29/159) for purged samples and 14.0% (13/93) for sampled droppings. Multiple logistic regressions showed that the median density of small mammal burrows in the open pastures was significantly positively related to dog infection (p = 0.003, odds ratio 1.05, 95% confidence interval 1.02–1.08). No correlation to dog age (p = 0.52), sex (p = 0.78), or sample collection method was seen (p = 0.380).
The higher median burrow density of small mammals was linked to a higher prevalence of E. multilocularis in dogs in these Tibetan pastoralist communities. Thus, partial fencing increases populations of potentially susceptible small mammal species in open pastures and consequently higher contamination pressure by dogs.
In a previous study, we showed that partial fencing around Tibetan settlements in winter pasture was significantly and independently associated with the risk for human AE in surveyed villages (8
). We assumed that the underlying reason might be overgrazing, exacerbated by reducing communal pastures near the settlements because of introduction of partial fencing around group tenure pastures acquired by Tibetan pastoralist families. Overgrazing may have promoted population outbreaks of small mammal intermediate hosts of the parasite and increased the density of the small mammal intermediate host, especially O. curzoniae
. This increase consequently favored higher dog E. multilocularis
infection and, thus, transmission to humans (8
). This study supports this hypothesis.