8The evolution of host specialization has been studied intensively, yet it is still often difficult to determine 9 why parasites do not evolve broader niches -in particular when the available hosts are closely related 10 and ecologically similar. Here, we used an experimental evolution approach to study the evolution of 11 host specialization, and its underlying traits, in two sympatric parasites: Anostracospora rigaudi and 12Enterocytospora artemiae, microsporidians infecting the brine shrimp Artemia franciscana and Artemia 13 parthenogenetica. In the field, both parasites regularly infect both hosts, yet experimental work has 14 revealed that they are each partially specialized. We serially passaged the parasites on one, the other, or 15 an alternation of the two hosts; after ten passages, we assayed the infectivity, virulence, and spore 16 production rate of the evolved lines. In accordance with previous studies, A. rigaudi maintained a higher 17 fitness on A. parthenogenetica, and E. artemiae on A. franciscana, in all treatments. The origin of this 18 specialization was not infectivity, which readily evolved and traded off weakly between the host species 19 for both parasites. Instead, the overall specialization was caused by spore production, which did not 20 evolve in any treatment. This suggests the existence of a strong trade-off between spore production in 21A. franciscana and spore production in A. parthenogenetica, making this trait a barrier to the evolution 22 of generalism in this system. This study highlights that the shape of between-host trade-offs can be very 23 heterogeneous across parasite traits, so that only some traits are pivotal to specialization. 24 KEYWORDS 25 Ecological specialization, niche, host specificity, experimental evolution, parasite life history, multi-host 26 parasites, Artemia, microsporidians. 27