The authors have declared that no competing interests exist.
Conceived and designed the experiments: SM. Performed the experiments: SM. Analyzed the data: SM. Wrote the paper: SM. Helped to interpret results and contributed to the writing: SJ YS MTA TBH.
Despite increasing control measures, numerous parasitic and infectious diseases are emerging, re-emerging or causing recurrent outbreaks particularly in Asia and the Pacific region, a hot spot of both infectious disease emergence and biodiversity at risk. We investigate how biodiversity affects the distribution of infectious diseases and their outbreaks in this region, taking into account socio-economics (population size, GDP, public health expenditure), geography (latitude and nation size), climate (precipitation, temperature) and biodiversity (bird and mammal species richness, forest cover, mammal and bird species at threat). We show, among countries, that the overall richness of infectious diseases is positively correlated with the richness of birds and mammals, but the number of zoonotic disease outbreaks is positively correlated with the number of threatened mammal and bird species and the number of vector-borne disease outbreaks is negatively correlated with forest cover. These results suggest that, among countries, biodiversity is a source of pathogens, but also that the loss of biodiversity or its regulation, as measured by forest cover or threatened species, seems to be associated with an increase in zoonotic and vector-borne disease outbreaks.
The Asia-Pacific region, and particularly Southeast Asia, is recognized as a hotspot for biodiversity
Infectious disease (ID) incidence has clearly increased during recent decades
The increased interactions between humans and wildlife resulting from habitat fragmentation are also affected by changes in wildlife species richness and community composition. At the local level, a reduction in biodiversity may lead to an increase in the prevalence and transmission rates of certain vector-borne diseases as reviewed by Keesing
The original level of biological diversity is also important to explain the overall richness of infectious diseases
Here, we explored the effects of biodiversity on patterns of infectious disease richness and outbreaks in Asia-Pacific. We considered two predictions related to the diversity of infectious diseases and the number of infectious disease outbreaks. First, the richness of ID should increase with increasing biodiversity (after controlling for nation area, population size, climate and surveillance effort), supporting regionally the pattern observed globally by Dunn et al.
Infectious disease data were obtained from GIDEON (Global Infectious Diseases and Epidemiology Network,
Distributions of variables were normalized using log-transformation or asin-square root transformation. We performed a principal component analysis (PCA) using the package ade4 (version 1.5–2) implemented in the R freeware programming environment (R Development Core Team, 2012). Correlation analyses and PCA allowed the identification of highly correlated variables such as latitude and temperature, or evapo-transpiration and mean annual precipitation (
A. Correlation among geographic (latitude, evapotranspiration, nation area size), climate (mean temperature, mean precipitation), and socio-economic variables (population size, GPD
We used general linear models (GLMs) using package lme4 implemented R 2.10 to explain disease richness and number of outbreaks as a function of our explanatory variables. In order to explain to explain the richness in infectious diseases, an initial model included the following explanatory variables (i.e. with the exclusion of latitude and evapo-transpiration, which are strongly correlated respectively with temperature and with precipitation): mean annual temperature, mean annual precipitation, bird and mammal species richness (in log), population size (in log), nation size (in log), number of surveys, GDP and health expenditure. These explanatory variables included the ones selected in the study of Dunn et al.
A second initial model, in order to explain the total number of outbreaks, included the following explanatory variables: mean annual temperature, mean annual precipitation, bird and mammal species richness (in log), population size (in log), nation size (in log), number of survey, GDP, health expenditure and richness of infectious diseases. We added also two other variables that represent the importance of impacts on biodiversity, the number of threatened bird and mammal species and the percent forest cover of the country. Additionally, we separated the disease outbreaks into three types: total, zoonotic, and vector-borne, and modelled each category. Selection of the best models was done using a step forward procedure based on AICc criterion. Initial and intermediate models are given in
In order to illustrate graphically the effects of patterns of biodiversity and biodiversity loss on the richness and total number of outbreaks, we used the residuals of the selected regression models.
The best model showed that the richness of infectious diseases is positivey linked to temperature (i.e. climate), population size and richness in bird and mammal species (see
Dependent variable | Explanatory variables | Effect | F (P) | VIF | R2 (P) |
Richness of infectious diseases | Temperature | +0.001 (0.97) | 14.3 (<0.001) | 1.37 | |
Population size | +0.02 (<0.0001) | 32.4 (<0.0001) | 1.99 | ||
Richness of birds and mammals | +0.12 (0.0036) | 66.4 (<0.0001) | 1.54 | ||
0.82 (<0.0001) |
Initial variables were: nation area, population size, richness in bird and mammal species, mean temperature, mean precipitation, surveys, GDP and health expenditure. Selection of the model was based on AIC criterion. The multicollinearity among independent variables is assessed by the variance inflated factor (VIF).
Using the GIDEON data base, and over the period from 1950 to 2008, 124 different diseases with epidemics were identified in Asia-Pacific countries. An increase in both the number of outbreaks and the number of different diseases causing outbreaks through time was observed (
Three models were created to explain the number of total infectious diseases, the total number of zoonotic infectious diseases and the total number of vector-borne diseases with outbreaks (for complete model selection process see
The number of total infectious diseases with outbreaks was found to be positively associated with temperature, forest cover, health expenditure, surveys, number of threatened bird and mammal species and population size (
A. Total number of infectious diseases with outbreaks and number of threatened vertebrate species. B. Total zoonotic diseases with outbreaks and number of threatened vertebrate species. C. Total number of vector-borne diseases with outbreaks and forest cover.
Dependent variable | Explanatory variables | Effect (P) | VIF | F (P) | R2 (P) |
a. number of infectious diseaseswith outbreaks | Mean temperature | −0.02 (0.20) | 2.27 | 0.2 (0.69) | |
Forest cover | −0.84 (0.037) | 2.08 | 0.36 (0.55) | ||
Health expenditure | +0.00005 (0.003) | 1.62 | 11.3 (0.003) | ||
Surveys | +0.02 (0.002) | 3.93 | 11.9 (0.003) | ||
Number of threatened species | +0.72 (0.022) | 2.24 | 98.8 (<0.0001) | ||
Population size | +0.21 (0.016) | 3.69 | 104.2 (<0.0001) | 0.92 (<0.0001) | |
b. number of zoonotic diseaseswith outbreaks | Mean temperature | −2.47 (0.12) | 2.11 | 0.02 (0.89) | |
Forest cover | −0.90 (0.071) | 1.95 | 0.03 (0.86) | ||
Health expenditure | +0.0004 (0.013) | 1.63 | 7.4 (0.01) | ||
Surveys | +0.02 (0.009) | 3.77 | 10.3 (0.004) | ||
Number of threatened species | +0.59 (0.08) | 2.27 | 64.5 (<0.0001) | ||
Population size | +0.24 (0.01) | 3.60 | 86.7 (<0.0001) | 0.89 (<0.0001) | |
c. number of vector-borne diseaseswith outbreaks | Health expenditure | +0.0007 (0.001) | 1.40 | 21.8 (<0.0001) | |
Richness of Infectious Diseases | +1.2 (0.0003) | 6.03 | 6.8 (0.02) | ||
Forest cover | −2.1 (<0.0001) | 1.71 | 23.2 (<0.0001) | ||
Richness of birds and mammals | +2.1 (0.003) | 2.97 | 81.4 (<0.0001) | ||
Population size | −0.11 (0.21) | 5.16 | 53.9 (<0.0001) | 0.90 (<0.0001) |
Initial variables were: nation area size, population size, richness of bird and mammal species, number of threatened vertebrate species, proportion of forest, mean temperature, mean precipitation, surveys, GDP and health expenditure. Selection of the best models was based on AIC criterion (see all models in
The total number of zoonotic diseases with epidemics was found associated with mean temperature, forest cover, health expenditure, surveys, number of threatened bird and mammal species and population size (
The total number of vector-borne diseases with epidemics was correlated positively with health expenditure, richness of infectious diseases, richness of bird and mammal species, and population size and negatively with forest cover and population size (
Although the number of epidemics has increased worldwide
The results of our analyses seem to support that effectively biodiversity if a source of infectious diseases in this region, but that biodiversity loss may have increased the number of outbreaks of infectious diseases over the last decades.
Our study is a step forward in the analysis of the relationship between biodiversity and infectious diseases. Several studies have emphasized the need to preserve vertebrate biodiversity and community composition in order to significantly reduce the risk of emergence
Our dataset showed a dramatic increase in the number of outbreaks, both total number and number of diseases presenting outbreaks, reported in Asia-Pacific over the last sixty years. This is in accordance with observations made in Europe
Our statistical models have identified several potential explanatory factors for the increase in outbreaks. The inclusion of mean temperature underscores the importance of climate in the occurrence of epidemics.
For each of the three investigated types of disease, outbreaks are correlated with either biodiversity at threat and/or proportion forest cover, a proxy measurement for biodiversity. However, it should be noted that the use of the number of threatened species the forest cover may describe quite poorly the dynamics of biodiversity loss and more accurate and precise indicators of biodiversity loss should be used. New accurate and precise indicators that integrate changes in ecosystem functions and are comparable among nations and scientifically sound are necessary (see
Our results seem to support the hypothesis of the potential role of biodiversity as a buffer of pathogen spread. As shown at the local scale, biodiversity could reduce outbreaks of vector-transmitted diseases through a dilution effect
Our results obtained at a regional scale seem to agree with many of these previous observations made at the local level, which showed an increase in the incidence of infectious diseases with a reduction in biodiversity
The ecosystem approach has been endorsed by the Millennium Ecosystem Assessment
The results presented here have one major implication for surveillance. The effect of biodiversity on infectious diseases suggests that higher resolution surveillance, both geographic and temporal, is needed. With better surveillance one could, for example, study the statistical relationships between the changing uses of land (fragmentation of landscape, increasing or reducing forest areas) and the occurrence of epidemics in order to investigate different scenarios for the effects of changes in biodiversity on infectious risks
(DOC)
(DOC)
We thank Katharine Owers for her contribution at gathering some data. We thank Prof. R. Coker for his suggestions and encouragements. We also thank four anonymous reviewers for their helpful comments.