Root hemiparasitic plants are associated with high diversity in temperate grasslands

Authors


Abstract

Questions

Is the incidence of root hemiparasitic plants in non-forest vegetation associated with high diversity? Are root hemiparasites more associated with species-rich vegetation than other species?

Location

Czech Republic.

Methods

Plot size-corrected species richness, Shannon diversity and Pielou's evenness were computed for a representative set of 18 101 vegetation plots representing all main types of terrestrial open (non-forest) habitats of the country. Null models of species richness assuming occurrence of a random species with given occurrence frequency, reflecting higher incidence probability in species-rich plots, were constructed for 16 common root hemiparasitic species. The null model distribution of species richness was subsequently compared with the actual mean species richness of plots containing the respective root hemiparasites. Median values of plot Shannon diversity and evenness were computed for each species in the database. Values obtained for plots containing individual root hemiparasites were compared with distribution of values for other species in the database.

Results

The occurrence of 11 of 16 root hemiparasites studied was associated with high species richness significantly more than under random expectations; three species were negatively associated. Three root hemiparasites were among the top 5% of all species associated with high species richness and Shannon diversity, and eight were among the top 25%. Almost 50% of the top 1% most species-rich plots contained at least one root hemiparasitic species.

Conclusions

We demonstrated a positive association between the incidence of most root hemiparasites and diversity of non-forest terrestrial plant communities at a broad landscape scale. This finding scales up the results of experimental studies that showed some root hemiparasites act as ecosystem engineers, increasing vegetation diversity. Root hemiparasites should be regarded as important biodiversity indicators and potential drivers of biodiversity. As such, conserving their wild populations or promoting their establishment should become a goal of nature conservation and ecological restoration.

Nomenclature
Danihelka et al. (2012)

 

Introduction

Biological diversity has become a central topic in ecology. Vegetation ecologists are particularly interested in factors influencing differences in the number of species between sites and ecosystems (Palmer & White 1994). At a fine scale, temperate grasslands are the most species-rich plant communities globally (Wilson et al. 2012; Chytrý et al. 2015), but mechanisms allowing for the co-existence of so many species within a small area, and their relative importance, are still the subject of lively debate (Wilson 2011). Among these, niche differentiation is probably the most often cited. To support their growth and survival, various plant species use the same essential resources, which limits the chance of niche differentiation. However, there are specialized functional groups of plants with different mechanisms of resource acquisition. These include, for example, the ability to fix atmospheric N through symbiotic bacteria in plants of the Fabaceae, or uptake of resources from other species in parasitic plants. Existence of these functional groups and the interactions of their members create opportunities for niche partitioning in the plant community.

The effects of specialized enemies have long been considered as a method to support species co-existence (Janzen–Connell effects; Janzen 1970), although mainly in the tropics. Recently, it was demonstrated that these effects are common and sufficiently strong to also support species co-existence in temperate grasslands (Petermann et al. 2008). They are mostly considered in connection with fungal infections and insect herbivores (e.g. Bagchi et al. 2014), but in principle can be associated with any consumer group showing at least some degree of specialization, including root hemiparasitic plants.

Root hemiparasites form a specialized functional group of green plants that attach to roots of other plant species in order to take up resources from their xylem (Sivicek & Taft 2011; Spasojevic & Suding 2011). Some root hemiparasitic species are even considered to be ecosystem engineers (Bardgett et al. 2006; Decleer et al. 2013). This is based on their ability to harm their host, alter competitive hierarchies in the community, affect mineral nutrient cycling and create regeneration gaps in the sward (Gibson & Watkinson 1989; Phoenix & Press 2005; Mudrák & Lepš 2010; Demey et al. 2015; Lepš & Těšitel 2015; Těšitel et al. 2015a). Individual host plant species may differ in their sensitivity to hemiparasite infection, and in their ability to exploit the nutrients mobilized and regeneration gaps created by hemiparasites. These differences can increase niche partitioning and promote species co-existence and community diversity. Although hemiparasites are not highly specialized, they might also support species co-existence by preferential use of productive and dominant hosts, similar to Janzen–Connell effects. It has also been demonstrated that the “luxury” use of resources by hemiparasites (Seel & Press 1994) can lead to a significant decrease in total community biomass in the presence of hemiparasites in comparison with analogous communities without them. Thanks to the well-known negative dependence of species richness on productivity in fertile temperate grasslands (particularly at biomass values >500 g m−2 dry mass; Al-Mufti et al. 1977; Grace 1999; Crawley et al. 2005), hemiparasites might support species diversity by limiting the total biomass in grassland communities.

Results of manipulative experiments testing the effects of hemiparasitic species on species diversity have been mixed, with positive (Mizianty 1975; Pywell et al. 2004; Westbury et al. 2006), negative (Gibson & Watkinson 1992) and neutral (e.g. Mudrák & Lepš 2010) effects being reported. Non-target effects, namely facilitation of weedy species (Joshi et al. 2000; Wagner et al. 2011) or support for resistant dominant species (Mudrák & Lepš 2010), are observed occasionally. In addition, data on long-term effects of hemiparasites on community diversity are lacking. Despite this uncertainty and ambiguity, hemiparasites have been suggested and used in nature conservation as a tool for increasing grassland species richness (e.g. Smith et al. 2000; Pywell et al. 2004; Westbury et al. 2006), although there is a risk of establishment of species-poor vegetation dominated by hemiparasites and hosts tolerating the infection.

Here we aim to complement previous experimental studies through analysis of the association between the incidence and diversity of hemiparasitic species of open (non-forest) terrestrial plant communities at the country scale. Specifically, we test the following hypotheses: (1) root hemiparasitic species occur more frequently in species-rich vegetation than would be expected by chance; and (2) root hemiparasites are more strongly associated with high vegetation diversity than other plant species.

Methods

Vegetation data

Data on the association between the occurrence of root hemiparasitic plant species and plant species diversity were obtained from the Czech National Phytosociological Database (Chytrý & Rafajová 2003), which contains records of vegetation plots (relevés) from the Czech Republic. We used a stratified subsample of the database following the resampling criteria used by Chytrý et al. (2005) in order to reduce differences in sampling intensity among areas and vegetation types. This resulted in a set of 31 512 plots with 2006 species, covering all the main types of vegetation in the country. From this data set, we selected 18 101 plots representing non-forest and non-aquatic vegetation (hereafter ‘all plots’ or ‘database’). For each plot in the database, percentage cover–abundance of all vascular plants present (derived from original records on the Braun-Blanquet or Domin scale) and assignment to vegetation type (phytosociological classification into vegetation classes; Chytrý 2007–2013) were available.

We computed three diversity indices for each plot in the database: species richness (i.e. number of vascular plant species, α-diversity, S), Shannon index of diversity (H′ = −Σpi ln pi, where pi is the proportional cover-abundance of species i) and Pielou's evenness (H′/ln S) for all plots in the database (natural logarithm was applied in all computations). The number of species was standardized by the plot size using the species–area curve to account for variable plot size in the database. This was done by fitting a species–area model:

display math(1)

where c and z are parameters from non-linear least-squares fitting and A is plot size in m2. The number of species in each plot was then standardized to the same plot size (16 m2 as a common size of non-forest plots):

display math(2)

The corrected number of species is used in all analyses. Shannon diversity and evenness are not size-corrected because their dependence on area is very weak in the range of plot sizes used (1.04–100.00 m2).

Association of root hemiparasites with diversity

First, we described species richness patterns of plots containing at least one of the 16 common root hemiparasites (those with at least 20 occurrences in the database) by fitting the plot richness values using negative binomial distribution. Subsequently, we tested statistical significance of the association between the incidence of individual hemiparasites and species richness of the plots. Such analysis must consider the fact that most species are more frequent in plots with a higher number of species than is the average for the whole data set. Therefore, we used null models including frequency of individual hemiparasites (n) and species richness of the plots to account for this effect. For each hemiparasite, we randomly chose n plots from the whole database, with probability weighted by their species richness, and computed the mean species richness of this sample. This step was repeated 999 times to generate the null distribution of species richness. Actual mean species richness of the plots containing individual hemiparasites was then compared with the null distributions, and P-values were determined from the null distribution quantiles. Further, we compared associations with diversity between the 16 root hemiparasites and other species in the database. All root hemiparasites studied were members of the Orobanchaceae, with the exception of Thesium linophyllon (Santalales). Species with at least 20 occurrences (n = 1039) in the database were considered in the analysis. For each species, we computed median values of Shannon diversity and evenness of the plots where it was present. Thus, we obtained measures of associations of each species with community diversity, which are independent of species frequency (|r| of all relations between abundance and diversity indices were <0.0008). Finally, we compared the values of median diversity indices for root hemiparasites with the distribution of median diversity indices of all the other species.

Statistical analysis used R v 3.2.3 (R Foundation for Statistical Computing, Vienna, AT) and the R package ‘vegan’, v 2.3–5 for computation of diversity indices.

Results

The distribution of species richness across plots followed a negative binomial distribution (Fig. 1a). Plots containing individual root hemiparasites were notably shifted towards higher species richness in comparison with the richness distributions of all plots (Fig. 1a, Appendix S1). The minimum species richness recorded in plots containing a root hemiparasite was mostly between 15 and 20, which is close to the median of the overall species richness (Fig. 1). Major parts of these species richness distributions were located within the upper quartile of the overall species richness. Several root hemiparasites (Rhinanthus major, Rhinanthus minor, Melampyrum cristatum, Thesium linophyllon) occurred in the most species-rich plots in the database. Moreover, 88 of 182 positive outliers of species richness in all plots included a hemiparasitic species (Fig. 1b). Null model comparisons identified significant positive associations between species richness and incidence of 11 of 16 hemiparasitic species (Fig. 2). Melampyrum nemorosum and Pedicularis palustris did not show any significant associations, whereas Bartsia alpina, Melampyrum pratense and Melampyrum sylvaticum were significantly associated with species-poor vegetation (Fig. 2).

Figure 1.

Distributions of species richness in all plots and plots with individual hemiparasitic species in the database. Lines correspond to fitted negative binomial distribution for all plots and fitted negative binomial probability density scaled by hemiparasite frequency. The histogram is shown for richness in all plots, whereas histograms for individual root-hemiparasitic species are presented in Appendix S1. Yellow and brown dots close to outliers for all plots correspond to plots containing and not containing hemiparasitic species, respectively. n corresponds to the number of all plots or plots containing a hemiparasitic species.

Figure 2.

Comparison between the mean observed species richness across plots containing individual root hemiparasitic species and species richness distributions in the null models. Lines represent the 2.5%–97.5% quantile ranges of the mean species richness under the random expectation. Circles correspond to the mean observed species richness of plots with hemiparasites. Vertical line denotes mean species richness in the database. Asterisks denote significance of associations with species richness: *P < 0.05, **< 0.001.

Most root hemiparasites had median values of species richness higher than the median of other species in the database (Fig. 3a, Appendix S2). Melampyrum cristatum, Rhinanthus major and Thesium linophyllon were among the top 5% of species associated with high species richness. The proportion of hemiparasites in the top 5% was significantly higher than that of non-hemiparasites (2 × 2 contingency table, χ2 = 4.264, = 0.039). In addition, Rhinanthus minor, Euphrasia officinalis and Odontites luteus were close to the top 5% limit. In contrast, the association with high species richness was negative for Bartsia alpina, Melampyrum pratense and M. sylvaticum. A very similar pattern was observed for Shannon diversity (Fig. 3b). Associations with high evenness values were, in general, lower than those with species richness and Shannon diversity. Nevertheless, 50% of root hemiparasites were in top 25% of the other species (Fig. 3c).

Figure 3.

Comparisons between medians of diversity indices, including (a) species richness, (b) Shannon diversity and (c) evenness, of plots containing the 16 root hemiparasites (colour symbols) and corresponding median distributions of plots containing all the other species in the database (n = 1039; gray bars). 50%, 75% and 95% quantiles of the distributions are depicted by different gray levels. The scores of the symbols on the right-hand (green) y-axis indicate frequencies of the hemiparasitic species in the database.

Discussion

We have demonstrated significant positive associations between incidence of many root hemiparasites and high α-diversity of local plant communities using a data set extending over a broad landscape and geographic scale. The most notable diversity patterns include (1) absence of most hemiparasitic species from species-poor vegetation, (2) high frequency of hemiparasites in extremely species-rich plots, and (3) stronger association with species-rich vegetation for many hemiparasitic species than for the rest of the flora. Three of the hemiparasitic species, Bartsia alpina, Melampyrum pratense and M. sylvaticum, were exceptions to these patterns, being associated with species-poor vegetation. An explanation of these exceptions is not straightforward. Restricted regional species pool size of some of the habitats where these species occur (e.g. bogs, springs, acidic heathlands; Sádlo et al. 2007) may account for these exceptions, at least in part.

We suggest that the positive association between incidence of most hemiparasites and community diversity identified in our study could correspond to three mechanisms. First, hemiparasites might grow in habitat types with a large species pool (species pool effects). Second, species-rich vegetation may be suitable for hemiparasites, especially due to the low intensity of above-ground competition which limits persistence of hemiparasite populations (habitat suitability effects; Těšitel et al. 2013). And third, hemiparasites may facilitate species co-existence and thus promote biodiversity (ecosystem engineering effects).

It is likely that all three mechanisms influence the association of hemiparasites with diversity. Rhinanthus major, Thesium linophyllon and Melampyrum cristatum, three species associated with the highest species richness and Shannon diversity, grow mostly in dry grasslands (Těšitel et al. 2015a), a habitat having a large regional species pool (Sádlo et al. 2007). High frequency of hemiparasites in extremely species-rich plots may thus be attributable, to a large extent, to the species pool effects. Absence of most root hemiparasites from species-poor vegetation, however, cannot be explained by the species pool effect. Low species richness occurs either in stressed low-productivity habitats or in high-productivity habitats where competitive exclusion prevents co-existence of competitively subordinate species (Grime 1979; Keddy 2005). Hemiparasites perform poorly under either of these extremes (Těšitel et al. 2015b) and consequently rarely occur in such conditions (Těšitel et al. 2015a). This habitat suitability effect can be quite strong and might explain both the absence of hemiparasites from species-poor vegetation and their general preference for diverse vegetation. The ecosystem engineering effects, i.e. increase in community diversity driven by hemiparasites, would be the most interesting of all these mechanisms, but more experimental evidence is needed to confirm this at a landscape scale. Published evidence suggests that ecosystem engineering effects may also positively affect habitat suitability for hemiparasites. Most grassland hemiparasites occur mainly at sites of intermediate productivity and nutrient availability (Fibich et al. 2010; Těšitel et al. 2015a). Under such conditions, they benefit most from the parasitic uptake of resources and their rapid transformation into growth and fitness attributes (Těšitel et al. 2015b). High nutrient availability and productivity are associated with high intensity of competition for light, which strongly reduces both diversity (Hautier et al. 2010) and population density (Těšitel et al. 2011, 2013) of hemiparasites. Hemiparasites are known to decrease community biomass production, especially in productive vegetation (Ameloot et al. 2005), which may increase both diversity and habitat suitability for hemiparasites. However, many hemiparasitic species require at least a moderate abundance of mineral nutrients (Těšitel et al. 2015b). Their nutrient-rich litter may enhance nutrient cycling and possibly increase productivity in nutrient-poor, low-productive habitats (Quested 2008; Spasojevic & Suding 2011; Demey et al. 2013). At such sites, an increase in productivity might be associated with an increase in diversity, following the humped-back productivity–diversity relationship (Keddy 2005; Fraser et al. 2015).

The hemiparasitic species for which we documented a positive association with diversity can be regarded as biodiversity indicators, regardless of the mechanism underlying their association with high diversity. Such an indicator role was suggested for hemiparasites of the genus Castilleja in North American grasslands (Sivicek & Taft 2011). Here, we expand this to a whole series of hemiparasites occurring in Central European grasslands and other types of open terrestrial vegetation. Given available experimental evidence (Mizianty 1975; Smith et al. 2000; Pywell et al. 2004; Westbury et al. 2006), it is likely that at least some of the hemiparasites influence community properties in a way that affects community diversity. Such species, in particular Rhinanthus spp., could be considered as biodiversity drivers, although their ecosystem engineering effects on diversity might differ depending on habitat type and might be related to community diversity itself. Given the general positive association with diversity and lack of species-poor plots containing these hemiparasites, their use as facilitators of grassland diversification can be considered as a safe method for restoration ecology, with minimum risk of non-target effects that might threaten biodiversity.

Conclusion

Our study demonstrates an association of the functional group of root hemiparasitic plant species with high local diversity of plant communities across a large region and many non-forest terrestrial vegetation types. However, we show that individual root hemiparasites display contrasting biodiversity associations. This highlights the significance of biological differences even among species belonging to the same functional group, a pattern repeatedly shown to affect parasitic interactions between plants (Rowntree et al. 2014; Demey et al. 2015).

Acknowledgements

We thank Dana Holubová for managing the Czech National Phytosociological Database used in this study and Conor Redmond for language improvement. PF, JL and MC were supported by the Czech Science Foundation project 14-36079G (Centre of Excellence PLADIAS) and JT by the Czech Science Foundation project 14-26779P.

Ancillary