In central Chile, the bur beak chervil (Anthriscus caucalis M. Bieb.; Apiaceae) is an annual naturalized herb introduced from Europe at least 120 years ago. Anthriscus is distributed in vegetation formations such as sclerophyllous shrublands (locally known as "matorral") and spiny savannas of Acacia caven (locally known as "espinal"). In matorral formations, Anthriscus grows at the edge of native woody fragments. Because these fragments are refuges where native herbs recruit, we studied the impact of Anthriscus on the diversity and survival of native forbs established in these sheltering microsites. First, we characterized the spatial distribution of Anthriscus in the matorral, sampling in different micro-habitat types. We differentiated three microhabitat types: under the canopy of a fragment, at the edge of the canopy of a fragment, and in open sites outside the fragments. A total of 40 1 x 1 m quadrates were randomly distributed in each habitat type. Inside each of them, we recorded the number of seedlings and established plants, including Anthriscus. Then we evaluated experimentally the effect of Anthriscus on diversity and evenness of the local herb assemblages. For this purpose we conducted a field trial using 34 metallic enclosures (0.5 x 0.5 m) arranged in pairs. In each pair, Anthriscus individuals were removed from one plot, the other paired plot acting as control. We periodically recorded the presence and abundance of the remanent species of herbs inside the plots, and then we characterized the species diversity and evenness over time (Shannon's index, H' and Pielou's index, J'). Finally, in a second experiment we measured experimentally the presence or absence of Anthriscus against the survival of four native herb species (Bowlesia incana, Bromus berteroanus, Pectocarya linearis, and Moscharia pinnatifida). Here we used 20 0.5 x 0.5 m plots where we randomly transplanted seedlings of native herbs and Arnthiscus. Then, for each species and plot we determined their survival (%) according to the number of seedlings initially transplanted. The samplings show strong association between the presence of Anthriscus on edge habitat in the matorral. The maximum densities were noted in these microhabitat types whereas in open areas and under-fragment sites Anthriscus shows very low or null densities. At the end of the first trials, the plots with Anthriscus showed a Shannon diversity index H' = 0.41 (+/- 0.11 SE), while in plots without Anthriscus this value was 1.19 (+/- 0.1 SE), both as averages. Pielou's evenness index (J') yielded values of 0.23 (+/- 0.06 SE) and 0.59 (+/- 0.04 SE) for treatments with and without Anthriscus, respectively. Similarly, the second trials shows that the survival of the four native herbs was drastically decreased in the presence of Anthriscus: by 64% for Bowlesia incana, 43% for Bromus berteroanus, 46% for Moscharia pinnatifida, and 76% for Pectocaria linearis. Our study shows that the effects of Anthriscus include an inhibition of the establishment of native plants and a decrease in their survival in edge habitats, therefore affecting the composition and diversity of the local herb layer. Thus, Anthriscus is invading a refuge habitat for native herbs in the Chilean matorral, decreasing the native herb diversity and survival.