Trees, soil, debris, wood chips, water, and air samples were collected according to previously described techniques, with limits of detection as previously described (Kidd et al., unpub. data). Sampling was conducted in the environments surrounding the residences of those with reported infections, including homes, habitats, and nearby parks and wooded areas.

A 35-km traffic corridor connecting 2 highly visited provincial parks was sampled at ≈500-m intervals (publicly accessible areas only). This sampling was performed on 6 nonconsecutive days in October 2004 and included 92 sites, 64 located at the roadside (designated “road sites”) and 28 located ≈100 m from the road (designated “forest sites”). For this series of samples, 169 trees of 8 different species were swabbed, and soil samples were collected from the rhizospheres of 77 of these trees. Global positioning system coordinates and tags were used to identify trees. Sites were designated positive by the presence of at least 1 tree or soil sample positive for C. gattii. Positive sites were resampled in June and December of 2005, including the original positive tree(s) and adjacent trees.

Many sites were sampled multiple times within 2–3 years to investigate the longitudinal colonization patterns of C. gattii in the environment. Except in these analyses, all data presented consider only the first swab, soil, and air samples collected at each sampling point to minimize statistical artifacts.

To investigate the effect of forestry activities on C. gattii aerosolization, air samples were collected during 2 independent tree removal efforts in a C. gattii–endemic area of Vancouver Island. A red alder (Alnus rubra) and a Douglas fir (Pseudotsuga menziesii var. menziesii) tree were removed by arborists on the same day in August 2002. The trees were tested for C. gattii colonization by swab and air sampling immediately before felling; in addition, air samples were collected at different tree heights during felling.

In collaboration with a municipal garden waste removal service, swabs of garden waste and nearby trees were collected from properties on 17 residential streets in October 2003. Air samples were collected around the garden waste and at the outlet of the wood chipper. Samples of wood chips were also collected from the wood chipper.

The wheel wells of vehicles used for Vancouver Island sampling were routinely swabbed; 63 swabs have been taken since July 2003. In addition, wheel wells of privately owned mainland- and Vancouver Island–based vehicles were swabbed to evaluate C. gattii carriage on vehicles not involved in the sampling effort. Two hundred vehicles in a mainland university campus parking lot and 200 vehicles in a Vancouver Island university campus parking lot were randomly selected for testing.

Eighty swabs of footwear worn by personnel participating in C. gattii sampling were taken at various Vancouver Island sampling sites during July 2003–June 2004. In addition, 9 swabs were taken from surfaces that came in contact with C. gattii–positive footwear (i.e., carpet, steel flooring, and a plastic bag). To investigate the potential for survival of C. gattii on footwear, shoes worn in a C. gattii–endemic area of Vancouver Island were stored for 333 days, with swabs taken periodically. The shoes were worn in nonendemic areas for short periods (≈4 h) on days 144, 153, and 154.

Cryptococci were initially identified by using Staib agar (12). Isolates were subcultured on malt extract agar and confirmed as C. gattii by using canavanine-glycine-bromothymol blue media (13), serotyping, or both (Iatron Laboratories, Tokyo, Japan). Sample positivity was scored both binarily and by the detected C. gattii concentration. Swab concentration was estimated by the presence of colonies on progressive streak lines, on a scale of 0 to 4+. The molecular types of selected isolates were identified by using previously described restriction fragment length polymorphism (RFLP) methods (6,14).

Environmental sampling data was compiled by using Microsoft (Redmond, WA, USA) Access 2002, and statistical analyses were performed with SPSS 14 (SPSS Inc., Chicago, IL, USA). Geographic data were assembled, and sampling points were mapped on 1:50,000 scale National Topographic System of Canada (NTS) grids (Kidd et al., unpub. data).

We initially investigated longitudinal patterns of C. gattii tree and soil colonization. Figure 1 illustrates these colonization patterns. Consistently positive swab and soil results were observed for some trees and their rhizospheres over 2–3 years, and these were designated “permanently colonized.” For other trees, an initial positive swab result was followed by a series of negative samples, which indicated a transient presence of C. gattii in these locations. Intermittently positive swab and soil results were also observed, in which the series of samples effectively oscillated between positive and negative for C. gattii over time with no discernable seasonal pattern. This intermittent positivity was probably due to fluctuations in the cryptococcal population over time, above and below limits of detection. This situation might arise following the introduction of C. gattii to a new location or substrate.

We hypothesize that permanent colonization is established once the cryptococcal population reaches a critical mass. This concept forms the basis of a model for C. gattii introduction to new areas of the Pacific Northwest, in which the fungus must adapt to new microclimates and compete with local microbiota. Areas of recent C. gattii dispersal may go through a period of intermittent positivity before either failing to become colonized (transience) or becoming permanently colonized. Determinants of colonization resulting in high C. gattii concentrations include low soil moisture and organic carbon content (Kidd et al., unpub. data).

The emergence of C. gattii infection among humans and animals with no travel history to endemic areas raised the possibility of dispersal within the Pacific Northwest (6) and the observed colonization of C. gattii on wood products such as wood chips and mulch, in bodies of fresh water and seawater, in air, and in soil, suggests there could be several mechanisms for this dispersals (Kidd et al., unpub. data). The unique opportunity for investigation of C. gattii as an emerging pathogen in BC has facilitated insight into the ecology and distribution of this pathogen. We believe the dispersal mechanisms of C. gattii could be applied as a model for other organisms.

The mechanisms of C. gattii dispersal discussed here are similar to those described in a number of reports for other mammalian pathogens. For example, Coccidioides immitis is similar to C. gattii in that it primarily colonizes soil, and disease is acquired by inhaling aerosolized spores. Coccidioidomycosis outbreaks have occurred as a result of soil disturbances as well as windborne dispersal. Such outbreaks were documented after a California earthquake (18), after separate point-source exposures among archaeology students in northern California and Utah (19,20), and after a windstorm in Kern County, California, that led to many coccidioidomycosis cases in non–disease-endemic areas of the San Joaquin Valley in California (21).

Blastomyces dermatitidis is the cause of blastomycosis outbreaks in humans and animals in Wisconsin (22,23), and dispersal occurs by way of rivers (24–26). This dispersal is similar to the isolation of C. gattii from bodies of water in BC. B. dermatitidis infection has also been reported in 2 captive California sea lions in the adjacent US states of Wisconsin and Illinois (27), although no related environmental sampling was discussed in these cases.

We have compiled evidence that suggests human-mediated dispersal of C. gattii may be important, including the detection of multiple C. gattii strains in the wheel wells of vehicles and on footwear. Footwear and farm vehicles were found to be involved in mechanical transmission of porcine reproductive and respiratory syndrome virus in Minnesota (28). Similarly, contaminated footwear and vehicles have been implicated in the spread of foot and mouth disease, resulting in the establishment of strict disinfection protocols in certain parts of the world (29,30). In addition, dispersal of the oomycete plant pathogen Phytophthora ramorum, causing sudden oak death in North America and Europe, is mediated by human activity as well as natural factors (31–33).

A specific route of C. gattii introduction to the Pacific Northwest has not been established, although 1 hypothesis implicates importation of contaminated trees (9,10). The data presented here indicate that C. gattii could also have been introduced by mechanical vectors such as vehicles or footwear, or by wooden pallets or crates (34) that are not routinely inspected for microbial contamination upon entry into Canada. Bird and animal migration may be involved in C. gattii dispersal through passive transport as well. Certainly, a large number of migratory birds pass through the disease-endemic area on Vancouver Island.

Given numerous possibilities for dispersal of C. gattii, until more study is done on conditions favoring or inhibiting de novo colonization, attempts to restrict further dispersal of the organism would be impractical and unlikely to succeed. However, public health, medical, and veterinary personnel, as well as the public in the Pacific Northwest and elsewhere, must be made aware of C. gattii and symptoms of infection so the disease can be identified correctly and managed effectively.