Two sides of the coin: Feedback-driven landscape formation results in trade-off between establishment and resilience of marram grass

Habitat-modifying plants engineer biogeomorphic landscapes through self-reinforcing interactions with their physical environment, or so-called 'biogeomorphic feedbacks'. Nevertheless, benefits can vary across a biogeomorphic landscape gradient and between plant-life stages. For instance, European marram grass forms dunes by trapping sediments which triggers plant growth, in turn promoting sediment trapping. Yet, by increasing dune height and vegetation cover, marram grass mitigates sediment dynamics, inhibiting sediment-growth feedbacks, which ultimately leads to its demise. However, little is known about how dune formation affects the growth and survival of marram grass at different life stages. Therefore, we performed a two-level field experiment testing the effect of position on marram grass across the biogeomorphological dune gradient (beach, foredune, backdune) on (i) the establishment success of juvenile transplants and (ii) the resilience of mature plants to disturbance by above-ground biomass removal, over one growing season. Although juvenile transplants grew similarly well across the dune gradient, significantly fewer beach transplants (67%) survived compared to the foredune- and backdune transplants. Conversely, survival of mature disturbed marram grass (100%) was unaffected, yet recovery was highest at the beach and significantly decreased across the dune gradient. We could link these opposing responses to habitat modification. In heavily modified dune habitats sediment stabilization aided juvenile establishment, whereas the high sediment dynamics of unmodified beaches facilitated adult resilience indicating dune formation invokes a trade-off between establishment and resilience. Our findings highlight the importance of assessing life stage-dependent differences in environmental requirements of habitat-modifying plants to understand population dynamics and landscape-forming processes.