Harrison's rule scales up to entire parasite assemblages but is determined by environmental factors.

Harrison's rule states that parasite body size and the body size of their hosts tend to be positively correlated. After it was proposed a century ago, a number of studies have investigated this trend, but the support level has varied greatly between parasite/host associations. Moreover, while the rule has been tested at the individual species level, we still lack knowledge on whether Harrison's rule holds at the scale of parasite and host communities. Here, we mapped flea (parasites) and rodent (hosts) body sizes across Mongolia and asked whether Harrison's rule holds for parasite/host assemblages (i.e. whether a parasite's average body size in a locality is positively correlated with its host's average body size). In addition, we attempted to disentangle complex relationships between flea size, host size and environmental factors by testing alternative hypotheses for the determinants of fleas' body size variation. We gathered occurrence data for fleas and rodents from 2,370 sites across Mongolia, constructed incidence matrices for both taxa and calculated the average body sizes of fleas and their hosts over half-degree cells. Then, we applied a path analysis, accounting for spatial autocorrelation, trying to disentangle the drivers of the correlation between parasite and host body sizes. We found a strong positive correlation between average flea and host size across assemblages. Surprisingly though, we found that environmental factors simultaneously affected the body sizes of both fleas and hosts in the same direction, leading to a most likely deceptive correlation between parasite and host size across assemblages. We suggest that environmental factors may, to a great extent, reflect the environmental conditions inside the hosts' burrows where fleas develop and attain their adult body size, thus influencing their larval growth. Similarly, rodent body size is strongly influenced by air temperature, in the direction predicted by Bergmann's rule. If our findings are valid in other host-parasite associations, this may explain the dissenting results of both support and lack thereof for Harrison's rule.