Facilitation is an interaction where one species (the benefactor) positively impacts another (the beneficiary). However, the reciprocal effects of beneficiaries on their benefactors are typically only documented using short-term datasets. We use Azorella selago, a cushion plant species and benefactor, and a co-occurring grass species, Agrostis magellanica, on sub-Antarctic Marion Island, comparing cushion plants and the grasses growing on them over a 13-year period using a correlative approach. We additionally compare the feedback effect of A. magellanica on A. selago identified using our long-term dataset with data collected from a single time period. We hypothesized that A. selago size and vitality would be negatively affected by A. magellanica cover and that the effect of A. magellanica on A. selago would become more negative with increasing beneficiary cover and abiotic-severity, due to, e.g., more intense competition for resources. We additionally hypothesized that A. magellanica cover would increase more on cushion plants with greater dead stem cover, since dead stems do not inhibit grass colonization or growth. The relationship between A. magellanica cover and A. selago size and vitality was not significant in the long-term dataset, and the feedback effect of A. magellanica on A. selago did not vary significantly with altitude or aspect; however, data from a single time period did not consistently identify this same lack of correlation. Moreover, A. selago dead stem cover was not significantly related to an increase in A. magellanica cover over the long term; however, we observed contrasting results from short-term datasets. Long-term datasets may, therefore, be more robust (and practical) for assessing beneficiary feedback effects than conventional approaches, particularly when benefactors are slow-growing. For the first time using a long-term dataset, we show a lack of physical cost to a benefactor species in a facilitative interaction, in contrast to the majority of short-term studies.

Following Raath-Krüger et al., the outcome of the interaction between Azorella selago and Agrostis magellanica was inferred from changes in Azorella size and dead stem cover and Agrostis cover through time. Twelve long-term monitoring plots (“Plot” in the dataset), which were established in 2003 at three altitudes (“Altitude” in the dataset) on the island’s eastern and western aspects (“Aspect” in the dataset), were resurveyed in 2016. Fifty Azorella individuals were surveyed from each of the plots. Each Azorella individual was photographed in the summer of 2002/2003 from directly above at a height of 1.5 m, with a scale bar included within each photograph. Each Azorella individual was photographed again in the summer of 2016 using the same methods.

To assess how Azorella size and stem mortality changed in relation to Agrostis cover, the photographs for each year were analysed in Fiji ImageJ and Adobe Photoshop. Azorella size (“AzoArea03” and “AzoArea16”)  and dead stem area (“DS03” and “DS16”) were measured using the polygon area selection tool, the wand tracing tool and by adjusting colour thresholds in Image J. Azorella size was defined as the total horizontal surface area of the cushion plants (including live stems, dead stems enclosed by live stems, dead stems contiguous with live stems, and parts of the cushion plants that were covered by other plants) as observed from directly above in the photographs. The extent of dead stems was used as a measure of plant vitality, where higher cover of dead stems on a plant was assumed to represent lower vitality. The area of Agrostis “AgrArea03” and “AgrArea16”), other vascular plants (“V03” and “V16”) and non-vascular plants (“NV03” and “NV16”) were measured using the same methods. Where other vascular and non-vascular plant species were growing on the edge of cushion plants and the edges of the cushion plants were not directly visible, we interpolated the plant edge based on the shape of the cushion plants. Similarly, we interpolated the cover of other vascular plant species and mosses when their cover was obscured by Agrostis cover; however, this was rare. In almost all cases, the vascular and non-vascular plants measured on the Azorella individuals were alive. Generally, very few dead individuals of other species are observed growing on Azorella, likely due to strong wind on Marion Island rapidly removing dead individuals. Dead stems were defined as any portions of Azorella individuals that were black or grey (and/or had very low stem densities) and/or if Azorella stem tips consisted entirely of brown leaves.

The Agrostis area, area of other vascular plants and mosses, and dead stem area variables that  were measured in ImageJ were then converted into cover values (%). The cover values were used for the analyses. The cover values are labelled as follows in the dataset:

·         PerAgr03: Cover (%) of A. magellanica on A. selago calculated as: AgrArea03/AzoArea03) x 100.

·         PerCO03: Combined cover (%) of other vascular plants and mosses on A. selago in 2003, calculated as: (CoverOther03/AzoArea03) x 100.          

·         PerDS03: Dead stem cover (%) on each A. selago individual in 2003 calculated as: (DS03/AzoArea03) x 100.     

·         PerDS16: Dead stem cover (%) on each A. selago individual in 2016 calculated as: (DS16/AzoArea16) x 100.                 

·         PerAgr16: Cover (%) of A. magellanica on A. selago calculated as: (AgrArea16/AzoArea16) x 100.

·         PerCO016: Combined cover (%) of other vascular plants and mosses on A. selago in 2016, calculated as: (CoverOther16/AzoArea16) x 100.