Seed banks and post‐fire recovery of invasive alien Metrosideros excelsa in South Africa: Implications for control.

The New Zealand bottlebrush, Metrosideros excelsa (Myrtaceae), was introduced into South Africa in the 1940s as an ornamental plant and subsequently naturalized in coastal parts of the Cape Floristic Region. Knowledge of M. excelsa's propagule pressure and response to fire in South Africa can inform evaluation of the species' invasion potential and management. We assessed M. excelsa's canopy and soil seed banks in burnt and unburnt stands; and post‐fire regeneration via reseeding and resprouting in relation to fire severity and tree size. Soil seed banks were assessed using an emergence technique, but no seedlings emerged from soil collected under burnt or unburnt plants suggesting that the species does not maintain a viable soil‐stored seed bank or germination triggers were not met (though unlikely). The annual seed crop in the canopies of unburnt trees was approximately 0.5–23 million seeds per tree. Viability, assessed through tetrazolium stain testing, of canopy‐borne seeds on unburnt trees was 18%, whereas a canopy fire, ranging in severity from low to extreme, killed all canopy‐borne seeds. Fifteen months post‐fire, the seedling to pre‐fire (live) tree ratio was 0.01, whereas 43% of burnt trees survived via basal resprouting. Fire severity had a non‐linear effect on tree survival (survival was highest after medium and high fire severity and lowest after low and extreme fire severity), while larger trees were more likely to survive fire. These results suggest that even low‐severity (safe) burning may be a useful control measure as it kills canopy‐borne seeds and causes substantial mortality, particularly of smaller trees. An opportunistic evaluation of an uncontrolled (no comparison with untreated individuals) foliar herbicide application to post‐fire resprouting individuals also showed considerable (91%) mortality. However, rigorous herbicide trials (also with herbicide in conjunction with burning) are required to inform the management of M. excelsa. [ABSTRACT FROM AUTHOR]