I really like this pdf. The Nesse references also discusses the evolutionary advantages for anxiety.
Abstract: This article reviews etiological and comparative evidence supporting the hypothesis that some genes associated with the autism spectrum were naturally selected and represent the adaptive benefits of being cognitively suited for solitary foraging. People on the autism spectrum are conceptualized here as ecologically competent individuals that could have been adept at learning and implementing hunting and gathering skills in the ancestral environment. Upon independence from their mothers, young individuals on the autism spectrum may have been psychologically predisposed toward a different life-history strategy, common among mammals and even some primates, to hunt and gather primarily on their own. Many of the behavioral and cognitive tendencies that autistic individuals exhibit are viewed here as adaptations that would have complemented a solitary lifestyle. For example, the obsessive, repetitive and systemizing tendencies in autism, which can be mistakenly applied toward activities such as block stacking today, may have been focused by hunger and thirst toward successful food procurement in the ancestral past. Both solitary mammals and autistic individuals are low on measures of gregariousness, socialization, direct gazing, eye contact, facial expression, facial recognition, emotional engagement, affiliative need and other social behaviors.
The evolution of the neurological tendencies in solitary species that predispose them toward being introverted and reclusive may hold important clues for the evolution of the autism spectrum and the natural selection of autism genes. Solitary animals are thought to eschew unnecessary social contact as part of a foraging strategy often due to scarcity and wide dispersal of food in their native environments. It is thought that the human ancestral environment was often nutritionally sparse as well, and this may have driven human parties to periodically disband. Inconsistencies in group size must have led to inconsistencies in the manner in which natural selection fashioned the social minds of humans, which in turn may well be responsible for the large variation in social abilities seen in human populations. This article emphasizes that individuals on the autism spectrum may have only been partially solitary, that natural selection may have only favored subclinical autistic traits and that the most severe cases of autism may be due to assortative mating.