Understanding why some individuals are more prone to carry parasites and spread diseases than others is a key question in biology. Although epidemiologists and disease ecologists increasingly recognize that individuals of the same species can vary tremendously in their relative contributions to the emergence of diseases, very few empirical studies systematically assess consistent individual differences in parasite loads within populations over time. Two species of fleas (Oropsylla montana and Hoplopsyllus anomalous) and their hosts, California ground squirrels (Otospermophilus beecheyi), form a major complex for amplifying epizootic plague in the western United States. Understanding its biology is primarily of major ecological importance and is also relevant to public health. Here, we capitalize on a long-term data set to explain flea incidence on California ground squirrels at Briones Regional Park in Contra Costa County, USA. In a 7 year study, we detected 42,358 fleas from 2,759 live trapping events involving 803 unique squirrels from two free-living populations that differed in the amount of human disturbance in those areas. In general, fleas were most abundant and prevalent on adult males, on heavy squirrels, and at the pristine site, but flea distributions varied among years, with seasonal conditions (e.g., temperature, rainfall, humidity), temporally within summers, and between flea species. Although on-host abundances of the two flea species were positively correlated, each flea species occupied a distinctive ecological niche. The common flea (O. montana) occurred primarily on adults in cool, moist conditions in early summer whereas the rare flea (H. anomalous) was mainly on juveniles in hot, dry conditions in late summer. Beyond this, we uncovered significantly repeatable and persistent effects of host individual identity on flea loads, finding consistent individual differences among hosts in all parasite measures. Taken together, we reveal multiple determinants of parasites on free-living mammals, including the underappreciated potential for host heterogeneity – within populations – to structure the emergence of zoonotic diseases such as bubonic plague.