Path analyses of cross-sectional and longitudinal data suggest that variability in natural communities of blood-associated parasites is derived from host characteristics and not interspecific interactions

Table 1 Comparison of models from stage 1

Response variables included in the model		Dataset and target parasite
Type	Variable	Cross-sectional				Longitudinal
		M	B	S	H	M	B	S	H
Environment	^aTemperature	0	0	0.02	0	n/a	n/a	0	0
Host	^bAge	17	27	96	0.2	n/a	25	28	32
	^bBody condition	0	0	0	0	n/a	0	0	0
	^bReproductive status	35	22	0	99	n/a	29	43	43
	Sex	13	24	0	0	n/a	23	29	29
Vector	^bFlea burden	0	0	n/a	0	n/a	0	n/a	0
Vector	^bTick burden	0	0	0	n/a	n/a	0	0	n/a
Bacteria	^b Mycoplasma presence	n/a	27	n/a	n/a	n/a	23	n/a	n/a
Bacteria	^b Bartonella presence	35	n/a	n/a	n/a	n/a	n/a	n/a	n/a

The models explain occurrence/abundance or temporal reduction for each of the four parasitic species in the cross-sectional and longitudinal datasets, respectively. Values are weights (w _i) in percentages of Akaike information criterion corrected for sample size—the relative likelihood of the current model, given the data and the set of models. Weights are normalized across the set of candidate models to summate to one, and are interpreted as probabilities. M = Mycoplasma haemomuris-like bacterium, B = Bartonella sp., S = Synosternus cleopatrae fleas, H = Hyalomma impeltatum ticks, n/a = not applicable. The best models (w _i > 10) are marked in bold and were used for stage 2 (Additional file 1: Table S1, Figure S1)
^aFor the longitudinal analyses, we used the between-period temperature differences
^bFor both datasets, we used the relevant factor measured in the first period

ISSN: 1756-3305