The best model for Floristic Quality values over the first 50 years of succession was a negative exponential increase to an asymptote. This trajectory was consistent whether or not non-native species were included in calculations and it was robust to scales of vegetation sampling. Initial field condition had some early effects on Floristic Quality values, but trajectories quickly converged among fields and values did not vary between treatments over the long-term. The consistency of FQA value trends despite large temporal variation in species dissimilarity among fields suggests that values are dictated by deterministic successional processes over early- to mid-successional stages.
The Trajectory of Floristic Quality Values Over Time
A few studies have reported community-level Floristic Quality values over time. Matthews, Spyreas & Endress (2009) tracked 29 wetland restorations in Illinois for 5–14 years after their creation. Although Floristic Quality values were far more variable among sites and over time compared to those in our study, the majority of their sites were also best described by an asymptotic trajectory model. A similarly shaped logarithmic trajectory best described FQI values in eight Ohio wetland restorations (Gutrich & Hitzhusen 2004), which on average reached an asymptote 8 years after their creation. Finally, values from an Indiana grassland restoration generally increased over 13 years (McIndoe, Rothrock & Reber 2008), although the shape of the trajectory was too erratic to be defined.
While asymptotic trends are most often supported, there appear to be stark differences among studies and systems in the length of time until values plateau. Peaks within 5–10 years typify wetland restorations, whereas at least three decades were necessary in our study’s upland fields. Comparatively rapid peaks to Floristic Quality in wetland restorations could have several causes. First, Conservative species are planted in most of these restorations. This is compared to BSS fields, which underwent natural colonization and showed gradually increasing trends. Second, relatively low dispersal limitation and high productivity in wetlands allows for rapid establishment by highly competitive taxa whose dominance then resists new colonizations (Chen et al. 2010). Finally, emergent wetlands could have earlier peaks because their terminal state as a herbaceous community lacks the woody and shade-tolerant forest taxa accompanying the ongoing physiognomic change of BSS fields to forests.
Succession and Floristic Quality
The Floristic Quality trajectories of BSS fields were notable for their consistent shape (Figs S2–S3, Supporting information) and variation over time (Fig. 1). Additionally, there were no patterns in Floristic Quality values related to year of abandonment or spatial position at the site (data not shown). Therefore, while minor differences in slopes or asymptote values were apparent, no field FQA values took idiosyncratic or divergent paths, suggesting that they were dictated by historical contingency or spatial stochasticity (Vaughn & Young 2010). Similar successional trends to Floristic Quality values may not seem surprising for fields sharing the same species pool and abandoned under similar abiotic conditions (soils, etc.), as this would likely lead to similar species assemblages in fields. However, species dissimilarity among BSS fields was actually quite variable over time, while FQA trends remained consistent. Thus, different species in different fields were producing the same Floristic Quality trends across the site. This is particularly surprising for a metric like FQI, the components of which, species richness and composition, are frequently erratic and unpredictable during succession (Matthews 1979; Christensen & Peet 1984). Furthermore, initial field conditions (hayfield vs. bare ground) are known to have differentially affected fields in other aspects for 30 years or more after abandonment (e.g. relative representation by annuals and forage grasses, native vs. exotic richness, Meiners, Pickett & Cadenasso 2002), but Floristic Quality values between treatments followed nearly identical trend lines throughout. In total, these results suggest that Floristic Quality was dictated by deterministic processes over time and that FQA measures behave predictably in unmanipulated habitats over early- and mid-successional timeframes.
This finding is also supported by comparing patterns of richness and Floristic Quality in plots vs. fields. While Floristic Quality values had similarly increasing trajectories when calculated per plot, per field or at the site level, species richness behaved differently at different scales. Richness (total and native) per field exhibited distinctly unimodal trends, whereas species richness per individual plot has remained very consistent in BSS plots over time (Meiners, Pickett & Cadenasso 2002). Therefore, species of greater Conservatism replaced less Conservative species in plots, without a net change in species density per plot. However, the same increasing Floristic Quality trends were generated by different increasingly Conservative species in different fields.
On the other hand, species life form was clearly related to successional trends in Floristic Quality values, especially for dominant plants. For example, the first group to dominate was comprised of weedy ephemeral taxa with low C values (e.g. Ambrosia artemisiifolia L. C = 0, Erigeron annuus (L.) Pers. C = 0), whose populations collapsed within 10 years (Meiners, Rye & Klass 2008). The second group to ascend was comprised of slightly more Conservative perennial herbaceous taxa (e.g. Aster pilosus Willd. C = 1, Solidago juncea Aiton, S. canadensis L., S. gigantea Aiton, S. rugosa Mill. C = 2, Apocynum cannabinum L. C = 2). The third group was made up of the trees, shrubs and woody vines that dominated during later years of the study (e.g. Acer rubrum L. C = 3, Rubus allegheniensis Porter C = 3, Cornus florida L. C = 5, Vitis spp. C = 4). They first increased Floristic Quality values as they came to dominate communities and then maintained values at their asymptotic levels as old-field herbs declined. However, despite the seeming coupling of life form with species Conservatism levels during succession, life form and Conservatism are not synonymous. Both highly Conservative and non-Conservative species are well represented among all life history, functional group and species trait categories in regional floras. Further study of the yet untested relationship between life form and species Conservatism certainly seems warranted.
A fourth group of species influencing temporal patterns in Floristic Quality values were non-native species, which generally decreased over time in BSS fields relative to natives. Non-natives directly decrease Floristic Quality values when included in metric calculations (equations 1, 3 and 4; Fig. 2). However, because there were no differences in the shapes of trajectories for metrics that included or excluded non-natives, non-native presence or richness alone did not determine Floristic Quality value trajectories. Non-native species effects on Floristic Quality values can also occur as an indirect function of invader dominance by displacing native species with higher C values or by decreasing opportunities for them to establish. Even though several of the most invasive plants in North America (e.g. Rosa multiflora, Microstegium vimineum, Lonicera japonica, Alliaria petiolata, Lonicera maackii; Meiners, Pickett & Cadenasso 2001; Gibson, Spyreas & Benedict 2002; Spyreas et al. 2004) are common in BSS fields, decreasing overall non-native dominance may have explained the asymptotic trajectory shape in these fields, rather than the peak-and-decline trajectory sometimes observed for FQA values over time. Therefore, our study does not dispute the majority of evidence that suggests considerable depressive effects on Floristic Quality from strong invasions (e.g. Spyreas et al. 2010). As non-native species and their impacts have been suggested as being comparatively minimal in mature forests (Von Holle, Delcourt & Simberloff 2003; Meiners, Rye & Klass 2008; Martin, Canham & Marks 2009), it will be highly informative to follow continued maturation of BSS vegetation with respect to non-native invasions and their effects. Furthermore, because understoreys contain a disproportionate amount of the plant diversity in these forests, future study should consider invasion in different strata and their effects on Floristic Quality in different strata.
Even though BSS fields had become young forests by the end of the study, and despite their adjoining old-growth forest seed source, their understoreys show a glaring absence of Conservative shade-tolerant native forest herbs. Conservative forest herbs were sporadically detected in plots throughout the study period (e.g. Actea pachypoda Elliott C = 5, Athyrium felix-feminina (L.) Roth C = 7, Circaea lutetiana L. C = 6, Monotropa uniflora Small C = 8, Phryma leptostachya L. C = 8, Podophyllum peltatum L. C = 6), but these were singular occurrences that did not persist. The potential for future sustained colonization by these taxa could initiate a second period of increasing Floristic Quality values in BSS fields. However, the notoriously slow migration and establishment by such species into mature forests suggests that this will not occur for hundreds of years, even with adjacent propagule sources (Matlack 1994; Brunet & von Oheimb 1998; Singleton et al. 2001; Spyreas & Matthews 2006). Recolonization rates by Conservative species in other habitat types have not been directly studied, but long-term comparisons of site histories suggest that if passive recovery by remnant taxa occurs in non-forest habitats, it will be measured over centuries as well (Gibson & Brown 1991; Kirkman et al. 2004; Ejrnæs et al. 2008). For example, Conservative species are notably absent from grassland restorations even with propagule sources that are directly adjacent (Kindscher & Tieszen 1998; Foster et al. 2007).
Implications for the Use of Floristic Quality Assessment
It could be argued that the increases in Floristic Quality values demonstrated here provide evidence that ‘hands-off’ approaches to restoration are likely to be successful given enough time; however, we reject this interpretation. Restorations are prone to failure from non-native species invasions (Matthews, Spyreas & Endress 2009). Furthermore, the maximum values in BSS fields (Mean C = 2·25, FQI = 17) were still well below values in remnant habitats with intact floras (e.g. Mean C = 5–6, FQI = 45–55, Swink & Wilhelm 1994), as the highly Conservative species characterizing remnant habitats did not establish. Barring a few exceptional cases (e.g. in North America, Sperry 1994; Gardner 1995), even the oldest restoration projects show considerable deficiencies in their Floristic Quality. Therefore, restoration efforts would do well to focus on Conservative species. In instances where restorations have achieved FQA value parity with remnants, they have received massive planting and management efforts over dozens of years (e.g. repeated overseeding, hand planting of plugs, careful introduction of missing Conservative species, meticulous monitoring, regular prescribed fire, invasive species control).
Three conclusions can be drawn from these results with respect to assumptions underlying FQA’s use. First, by illustrating the consistency of Floristic Quality metrics during succession, we demonstrate the robustness of FQA for use across temporal gradients. Second, because these fields reached an asymptote in their FQA values even though they continue to undergo rapid successional turnover (data not shown), temporal changes in FQA values cannot be considered synonymous with succession or with the successional states of communities. Finally, while the relationship between Floristic Quality and time since anthropogenic disturbance may be consistent and predictable, it is not simple (i.e. it is nonlinear). Therefore, FQA users must carefully consider background successional trends in Floristic Quality when using FQA metrics across temporal gradients or for habitats of different ages. For example, Tulbure, Johnston & Auger (2007) concluded that an increase by an invasive species did not decrease a community’s Floristic Quality over time. However, the lack of an invasion effect may have been obscured by background increases in Floristic Quality that were likely occurring across the site, which was undergoing rapid succession after a recent disturbance. Similarly, controlling for ambient successional changes in Floristic Quality values in a study of deer browsing effects on the floras of young grassland restorations may have allowed for treatment differences to have been better discerned (Anderson, Dorick & Crispino 2007).
While the asymptotic trajectory model we have described will require further testing for its general applicability in other habitat types, successional stages, regions and landscape settings, we suggest it for use as a baseline expectation for predicting Floristic Quality values over early- to mid-successional timeframes. Deviations from this expected baseline trajectory could highlight relative successes or failures in recovery progress or management practices at sites. Comparative study of site trajectories and their deviations from the expected baseline en masse would reveal patterns in the relative importance of specific ecological constraints to the recovery of community-level Floristic Quality.