CBGP, INRA, CIRAD, IRD, Montpellier SupAgro, Univ. Montpellier, Montpellier, France.
Centre Régional Agrhymet, USAid/Fews-Net, Niamey, Niger.
Infect Genet Evol. 2018 Sep;63:307-315. doi: 10.1016/j.meegid.2017.10.006. Epub 2017 Oct 5.
Trypanosomes are protozoan parasites found worldwide, infecting humans and animals. In the past decade, the number of reports on atypical human cases due to Trypanosoma lewisi or T. lewisi-like has increased urging to investigate the multiple factors driving the disease dynamics, particularly in cities where rodents and humans co-exist at high densities. In the present survey, we used a species distribution model, Maxent, to assess the spatial pattern of Trypanosoma-positive rodents in the city of Niamey. The explanatory variables were landscape metrics describing urban landscape composition and physiognomy computed from 8 land-cover classes. We computed the metrics around each data location using a set of circular buffers of increasing radii (20m, 40m, 60m, 80m and 100m). For each spatial resolution, we determined the optimal combination of feature class and regularization multipliers by fitting Maxent with the full dataset. Since our dataset was small (114 occurrences) we expected an important uncertainty associated to data partitioning into calibration and evaluation datasets. We thus performed 350 independent model runs with a training dataset representing a random subset of 80% of the occurrences and the optimal Maxent parameters. Each model yielded a map of habitat suitability over Niamey, which was transformed into a binary map implementing a threshold maximizing the sensitivity and the specificity. The resulting binary maps were combined to display the proportion of models that indicated a good environmental suitability for Trypanosoma-positive rodents. Maxent performed better with landscape metrics derived from buffers of 80m. Habitat suitability for Trypanosoma-positive rodents exhibited large patches linked to urban features such as patch richness and the proportion of landscape covered by concrete or tarred areas. Such inferences could be helpful in assessing areas at risk, setting of monitoring programs, public and medical staff awareness or even vaccination campaigns.
锥虫是一种世界性的原生动物寄生虫,感染人类和动物。在过去的十年中,由于 Lewisi 锥虫或类似 Lewisi 锥虫的非典型人类病例报告数量增加,促使人们研究导致疾病动态变化的多种因素,特别是在啮齿动物和人类高密度共存的城市。在本研究中,我们使用物种分布模型 Maxent 来评估尼日尔城市中锥虫阳性啮齿动物的空间模式。解释变量是描述城市景观组成和地貌的景观指标,这些指标是从 8 个土地覆盖类别中计算出来的。我们使用一组半径不断增大的圆形缓冲区(20m、40m、60m、80m 和 100m)来计算每个数据位置周围的指标。对于每个空间分辨率,我们通过使用完整数据集拟合 Maxent 来确定特征类和正则化乘数的最佳组合。由于我们的数据集较小(114 个事件),我们预计与将数据划分为校准和评估数据集相关的不确定性很大。因此,我们使用代表 80%事件随机子集的训练数据集和最佳 Maxent 参数进行了 350 次独立模型运行。每个模型都生成了一张覆盖尼日尔的栖息地适宜性图,该图被转换为一个二值图,该二值图通过最大化敏感性和特异性的阈值来实现。将生成的二进制地图组合起来,可以显示出指示锥虫阳性啮齿动物具有良好环境适宜性的模型的比例。Maxent 在使用来自 80m 缓冲区的景观指标时表现更好。锥虫阳性啮齿动物的栖息地适宜性表现为与城市特征相关的大型斑块,例如斑块丰富度和被混凝土或焦油覆盖的景观比例。这种推断可能有助于评估风险区域、制定监测计划、提高公众和医务人员的认识,甚至开展疫苗接种运动。