Effects of Canada Goose (Branta canadensis) and Snow Goose (Chen caerulescens) herbivory on tidal marsh recession at the Westham Island marsh

In the Fraser River Estuary of British Columbia, tidal marshes have been receding and converting into unvegetated mudflats since the 1980s. While there are many hypotheses for this recession, the effect of avian herbivory is poorly understood. This study assessed how Canada Goose (Branta canadensis) and Snow Goose (Chen caerulescens) herbivory affected cover of tidal marsh vegetation that was comprised mainly of three-square bulrush (Schoenoplectus pungens) in the Westham Island tidal marsh. I conducted two field-based exclosure experiments, marsh edge and mudflat, that used exclosure plots to reduce specific goose herbivory in a randomized block design. Each experiment consisted of four blocks each of which was comprised of four treatments: open to goose herbivory, excluded all goose herbivory, primarily excluded Canada Goose herbivory, or primarily excluded Snow Goose herbivory. The marsh edge experiment used exclosures centered on the vegetated edge of the marsh, while the mudflat experiment was conducted in the unvegetated mudflat and were transplanted with S. pungens. Based on results from July to October of 2020, percent cover of tidal marsh vegetation was about 20% lower in plots open to Canada Goose herbivory versus those that excluded geese. Snow Goose herbivory could not be accurately assessed as they arrived when S. pungens were dormant. Thus, deterring goose herbivory may be an important consideration for land managers in restoring tidal marshes. Additionally, I compared percent cover from drone-derived remote sensing to traditional ground-based visual estimates of percent cover of S. pungens in the tidal marsh. One per month, from July to October of 2020, I used a drone to take photos of the exclosures from the previous experiments, and used pixel counts to calculate the percent cover of S. pungens. I then used a t-test to compare the drone-derived percent cover to the ground-based estimates and found no significant difference (t = 0.58, p = 0.56). I then plotted a linear regression model and found a strong correspondence between both methods (R² = 0.99, p = 1.3e-139). So, remote sensing using drones appears to be an effective alternative to visual estimates of percent cover of tidal-marsh vegetation in the Westham island tidal marsh.