SUCCESS OF POST-FIRE BROADCAST SEEDING AS A TOOL FOR RESTORATION OF A HAWAIʻI DRY FOREST

Abstract

Dry forest ecosystems in Hawai‘i once supported a rich diversity of native and endemic species. However, the diversity, composition, distribution, ecosystem structure, and function of dry forests have been significantly altered due to human impacts, invasive species, and wildfires. As such, there is a great need for landscape-level restoration of dry forests, and the development and testing of methods for reintroducing native species in altered habitats. Wildfire, in particular, is a significant threat to vulnerable Hawaiian dry forests, with thousands of acres burned annually. These burned areas have high restoration potential through the use of methods such as native seed broadcasting, yet few studies have investigated the effectiveness of direct seeding for ecological restoration following wildfire in Hawaiian lowland dry forests. A broadcast seeding experiment was established at the Waikōloa Dry Forest Preserve in 2018 after a 18,000 acre wildfire burned through a portion of the preserve (22 acres). To restore native plants to the area after the fire, the seeds of 10 native plants were broadcasted: ʻaʻaliʻi (Dodonaea viscosa), ‘āweoweo (Chenopodium oahuense), ʻilima (Sida fallax), wiliwili (Erythrina sandwicensis), ‘āwikiwiki (Canavalia hawaiiensis), kolomona (Senna gaudichaudii), ko‘oko‘olau (Bidens menziesii), māmane (Sophora chrysophylla), pua kala (Argemone glauca), and ʻūlei (Oseomeles anthyllidifolia). Seeds were broadcasted under both a short (three rounds of herbicide, between October 2018- July 2019) and long-term (bi-annual, October 2018-October 2021) weed management treatment and in a fenced and unfenced site. A vegetation survey was completed three years post-fire to determine the density, richness, and diversity of native plant species between treatments. Compared to the short-term treatment, which observed only 3 species, the long-term weed management three years post-fire contained a higher diversity of native species with four of the eleven seed scattered species present. The long-term removal of non-native plant species allowed the native ʻaʻaliʻi (Dodonaea viscosa), ‘āweoweo (Chenopodium oahuense), ʻilima (Sida fallax), and wiliwili (Erythrina sandwicensis) to germinate and young plants to establish. There were significantly more ‘āweoweo and ʻaʻaliʻi in the long-term weed management treatment compared to the short-term weed management treatment, and the density, richness, and diversity of native seedlings established were also higher in the long-term weed management treatment. The density (mean stem counts) of ‘āweoweo was 6× greater (1483% increase) in the long-term compared to short-term treatment after 3 years. ʻA‘ali‘i density was 4× greater (989% increase) and ʻilima density was 8× greater (1077% increase) in the long-term compared to short-term treatment after three years. A Multiple Response Permutation Procedure (MRPP) showed that species composition was significantly different (p<0.05) between the long-term weed management treatment and short-term weed management treatment. Finally, no scattered seed species germinated in the presence of ungulates within the unfenced site. Controlling invasive plants (e.g., fountain grass (Cenchrus setaceous)) with herbicide long-term (three years) allowed the native seeds that were broadcasted to germinate and establish before the invasive grass reinvaded the area. Broadcast seeding can be an effective restoration tool in post-fire Hawaiian lowland dry forests, but only within an ungulate-free fenced unit. Moreover, long-term weed management is critically needed to increase the establishment of native plant species.