Non-native fallows hold high potential for restoration through agroforestry in a Pacific Island ecosystem

Agricultural land abandonment affects millions of hectares of cultivated lands globally. While ending cultivation can lead to spontaneous reforestation and ecological benefits, the resulting landscapes often have lower social and agricultural benefits than the native forests and agricultural systems they replace, especially when non-native species dominate successional pathways. This is the case in many Pacific Islands including HawaiModified Letter Turned Commai, where approximately 45 % of agricultural lands are unmanaged and non-native forests make up nearly 40 % of total forest cover. Agroforestry systems that integrate native and non-native culturally important plants present a potential pathway to increase social and ecological benefits of unmanaged agricultural lands; however, under-standing what the restoration potential is of different agroforestry systems remains a question. We collaborated with a Native Hawaiian-led, community-based organization to explore this potential. We asked, 1) does the composition of agroforestry species planted (i.e., treatment) affect restoration success, and if so, do other factors mediate the effect of treatment, and 2) how do ecological conditions two years after starting restoration compare to conditions pre-restoration? We set up ten 12 x 15 m restoration plots and one reference plot on former pasture land regenerating as non-native forest. Then, we used a functional trait-based approach to select two agroforestry species mixes. Both mixes had high cultural value and each had traits to address a different primary ecological goal: erosion control and early successional facilitation. We monitored the plant communities before restoration and at six months, one year, and 1.5 years post-planting. We used multivariate analysis and structural equation modeling (SEM) to analyze the differences between treatments over time. We found that measures of restoration success did not vary significantly between treatments but did change from baseline. Results of the SEM indicated that understory weed cover was a significant driver of understory cover of agroforestry species, but that vari-ability in agroforestry cover was primarily a result of management factors not tested in the model. This study provides a first step in documenting what non-native forest to agroforest transitions can look like. Our findings suggest that non-native fallows have a high potential for restoration through agroforestry in Pacific Island ecosystems.