Our study addresses rather typical restoration works in Australia (i.e., returning natural/native ecosystems following agricultural development), and some of the difficulties faced when attempting to define ecosystem development toward a given restoration target. In this regard, restoration plantings in the Lurg Hills generally developed in different ways across the two landforms studied (i.e., upper and lower hills). All sites were found in a hybrid state sharing some components with the remnant (historical) ecosystems, but also some novel proportional combinations of species assembly. Despite differences between groups, trends towards the restoration target were evident for species composition of seedlings only in the lower hills, and for bud/flower and fruit production only in the upper hills. For example, after 8 years, restored sites in the lower hills had a seedling composition that was very different to both unrestored and younger restored sites. Sites also became more similar to the remnant with time, particularly 12 years after restoration.
A range of development scenarios have been identified where sites may converge, diverge or deviate from the restoration target (Suding 2011). The majority of sites in the lower hills converged towards the target and also increased in similarity to sites of the same age class with time. This supports the idea that, as a community ages, the number of potential development trajectories may decrease due to plant competitive ability (Nuttle et al. 2004). However, a range of factors may have influenced these similarities within age classes. As Hobbs and Norton (2004) point out, the sites’ starting conditions, order of species introductions and subsequent management will all interact to affect how a community will develop. Seed or microsite availability can also have a strong influence on the composition and structure of ecosystems through the failure of species to recruit (Clark et al. 2007). There were some instances where seedlings seemed to derive from adjoining remnant vegetation, as evidenced by patterns of seedling dispersal in relation to remnant trees, maturity of restored vegetation and restoration species lists. However, the majority of species recruiting were the same as those planted during restoration. While seeds were not tested for viability, it is likely much of the regeneration resulted from mature planted individuals (e.g., fruits were present on medium shrubs after only 4 years). Many studies show that recruitment of trees and shrubs is often affected to a greater extent by the availability of microsites, areas with suitable conditions for germination and growth (Clarke and Davison 2001; Clark et al. 2007; Gómez-Aparicio 2008). However, as the extent of potential recruitment is ultimately determined by the availability of viable seed (Clarke and Davison 2001), the production of adequate seed in restored ecosystems is clearly essential for achieving a target community. Introducing seed of additional species may help reduce any seed limitations (Hobbs and Norton 2004; Young et al. 2005; Clark et al. 2007) which may be caused by decreased pollination and dispersal opportunities (Bennett et al. 2009).
The lack of recruitment at two restored sites (4–6 years and 12–14 years) in the lower hills suggests divergence, which occurs when sites that have undergone similar restoration develop along different trajectories (Suding 2011). Structurally, these sites differ from other sites of the same age, which have a range of species recruiting. It would be worth investigating the barriers that are preventing recruitment, such as herbivory, microsite variations or seed availability. Some species present in reference ecosystems may not re-establish under current conditions unless regeneration niches are provided (Zedler et al. 2012), yet recruitment of seedlings 8 years after restoration indicates niches have been provided for many indigenous species—including rare species such as Goodia medicaginea, which was found recruiting in small numbers in both landforms. In addition, no woody weeds were recorded at either landform. Therefore, seedlings and the structure of the tree and shrub layers were composed solely of remnants or planted indigenous species sourced from local provenance.
Zedler et al. (2012) argue that sites that need constant redirection towards their target ecosystem might be more sustainable if retained as novel ecosystems, thereby acknowledging the barriers preventing veritable restoration. With the exception of the two sites lacking seedlings, if diverse recruitment of indigenous species continues to occur, hybrid ecosystems could be retained in preference to novel ones, and have the potential to become more similar to a natural or historical ecosystem with time. If biota cannot regenerate, development into a novel ecosystem will probably occur (Hobbs et al. 2009). The insignificant difference in species composition of seedlings in the upper hills may be due to further factors (not investigated in the present context) involving more subtle or site-specific management practices, e.g., due to different historical land-usage intensities. Alternatively, potentially large variability in abiotic factors across the steeper upper hills (such as aspect, slope, moisture availability and proximity to larger patches of remnant vegetation) may be an influence.
Restored sites had buds, flowers and fruits after 4 years in the upper hills, with sites becoming more similar to the remnant as rapidly as 8 years after restoration. The absence of a clear trend towards a restoration target for bud/flower and fruit production in the lower hills may be due to the presence of canopy trees in both the unrestored and remnant sites, and the diverse range of functional groups with buds/flowers or fruits in the restored sites, which were fewer in the remnant. This resulted in greater similarity between the remnant and unrestored sites. While the more fertile, gentler slopes of the lower hills may have influenced the production of buds/flowers and fruits on a diverse range of functional groups, the different stages of ecological development of the restored sites and the remnant may have been a factor. As only one remnant was surveyed in each landform, this was a limitation of the study. The different vegetation structure in the late stage of ecological development of the remnant is also likely to have an impact when assessing restoration success; as Suding and Hobbs (2009) point out, each reference site may reflect only one of the many stages of development, any of which would form the model for restoration.
Has restoration in the Lurg Hills been successful?
Despite much debate surrounding the use of remnant ecosystems as restoration targets (Hobbs and Harris 2001; Harris et al. 2006; Hobbs et al. 2009; Palmer et al. 2006; Choi 2007; Comín 2010), these selected reference systems are still commonly used to determine site-specific restoration success. In this regard, seedling recruitment and reproductive parameters provide limited insight into whether sites are developing towards a restoration target. The establishment of self-sustaining ecosystems is an ecological attribute the Society for Ecological Restoration International Science and Policy Working Group (2004) proposes using to evaluate restoration success. Given seedlings recruited in most sites in the Lurg Hills, most of the restoration could be deemed successful at this point in time. The presence of seedlings signifies a life cycle has successfully been completed, as transplanted individuals have flowered, set fruit, dispersed seed and recruited new seedlings (Menges 2008). In addition, microsite conditions have clearly been conducive to germination and growth. While it may be preferable from a management perspective for sites to converge together along a predictable pathway towards a restoration target, Suding (2011) argues that resilience may be increased if there is variability at a landscape scale. Resilience is vital for an ecosystem to adapt and recover from disturbances, while still maintaining ecological function and health (Folke et al. 2004; Walker and Salt 2006; Clewell and Aronson 2007). Diversity is thought to influence ecosystem resilience (Grant 2009; Allen et al. 2010); therefore, variability among sites of the same age may in fact be beneficial in the longer term, increasing resilience through differences in diversity as seedlings mature and alter vegetation structure and composition. This is particularly relevant under uncertain future conditions including climate change, as increased human demand places further pressure on ecosystems, which may not remain stable (Folke et al. 2004).