olate dispersal behavior to the landscape level (17, 21,27,28). Explicit tests of such models are needed (18). The tight fit between observed and predicted patterns of seed rain in our habitat patches provides strong support for the key assumption that small-scale behavioral responses can drive landscape-scale distributional patterns. From a conservation perspective, impacts of corridors can be predicted on the basis of behaviors that are relatively simple to measure (29).

SummaryWe have recently reported the presence of covalently linked pilus-like structures in the human pathogen, Group B Streptococcus (GBS). The pilus operon codes for three proteins which contain the conserved amino acid motif, LPXTG, associated with cell wall-anchored proteins together with two genes coding for sortase enzymes. Analysis of the eight sequenced genomes of GBS has led to the identification of a second, related genomic island of which there are two variants, each containing genes coding for proteins with LPXTG motifs and sortases. Here we show that both variant islands also code for pilus-like structures. Furthermore, we provide a thorough description and characterization of the genomic organization of the islands and the role of each protein in the assembly of the pili. For each pilus, polymerization of one of the three component proteins is essential for incorporation of the other two proteins into the pilus structure. In addition, two sortases are required for complete pilus assembly, each with specificity for one of the pilus components. A component protein of one of the newly identified pili is also a previously identified protective antigen and a second component of this pilus is shown to confer protection against GBS challenge. We propose that pilus-like structures are important virulence factors and potential vaccine candidates.