Jan Leps Dept of Botany University of South

Jan Leps, Dept of Botany, University of South Bohemia, České Budejovice, Czech Republic Biotic and abiotic effects on species and functional trait composition: lessons from a grassland experiment with dominant removal

OHRAZENI – a seminatural meadow Regular mowing ceased in late eighties

Molinia caerulea Nardus stricta

Species diversity and “interesting plants” (e. g. red list species) concentrated in “traditional”, i. e. mown, unfertilized Dactylorhiza majalis Senecio rivularis

14 Carex species Carex pulicaris C. hartmanii

Scorzonera humilis

Myosotis nemorosa

Factorial experiment, 3 replications • Mowing (once a year, in June) • Ferilization (65 [50] g of commercial NPK/m 2 - 12% N (nitrate and ammonium), 19% P (as P 2 O 5) and 19% K (as K 2 O)) • Dominant (i. e. Molinia caerulea) removal (in spring 1995, but re-weeding necessary time from time) • Yielding 24 plots, 2 m × 2 m each • Central 1 m x 1 m sampled, followed by detailed analysis of 50 cm × 50 cm grid of 10 cm × 10 cm – including seedling counts • Experiment started 1994, baseline data available

Ohrazení (http: //mapy. atlas. cz)

Detailed recording of vegetation in all the 15 years Sprouts of woody plants removed

Today, I want to demonstrate • General pattern of species composition changes • Scale dependence of changes in species richness • Importance of species trait for the species response to treatments • Effect of single species on various community processes/functions

Error bars =95% confidence intervals Problem for interpretation, in later years, Molinia removal would have little effect in fertilized mown (as it have very low cover also in the control plots)

Species composition

DCA - Molinia is passive species - log transformed cover. Starting points in 1994 show the random variability - the divergence of trajectories show the differentiation according to treatments between 1994 and 2007. Centroids of Year x Treatment

Principal response curves triangles - mown circles unmown full symbol - fertil. open symbol - unfert solid line - control broken l. - removal

Principal response curves • Multivariate counterpart of Repeated measurement ANOVA - the first RDA axis, which is plotted on vertical axis against the time (horizontal axis), captures the main differentiation among categories of YEAR * TREATMENT interaction • The common temporal trend is subtracted from the data - YEAR as covariable(s) • The horizontal axis corresponds to the control (in our case, unmown, unfertilized, no removal)

Principal response curves triangles - mown circles unmown full symbol - fertil. open symbol - unfert solid line - control broken l. - removal

Mown - unfertilized (=traditional)

Mown - unfertilized & Molinia removed

Mown - fertilized (=intensive)

Mown - fertilized & Molinia removed

Unmown - unfertilized (=abandoned)

Unmown - unfertilized & Molinia removed

Unmown - fertilized (=abandoned eutrofized meadow)

Unmown - fertilized & Molinia removed

Species richness

In mown plots, increase of the number of species during first six years, regardless of removal of Molinia, in unmown plots, removal has positive effect on species richness. NB. – Molinia is thread only in unmown plots! NSP of vascular plants per m 2

In unmown plot, continuous decrease. Initial positive effect of removal ceased after 10 yrs. In mown plots, initial increase (5 years) followed by decrease, no effect of removal.

Take home message • Increase in soil nutrients can lead to competitive exclusion - nevertheless, in community of established perennial plants, the exclusion can take rather long (six years in our case). • Mgmt recommendations based on short term experiments can be misleading

Species richness dynamics depends on spatial scale • Repeated measures ANOVA • 3 main plot factors (Mowing, Fertilization, Removal) • 2 Rep Mes factors – year and plot size (from 10 x 10 cm 2 to 50 x 50 cm 2) • Number of species log transformed – i. e. relative change of the richness • Very long ANOVA table, with many significant interactions

Unmo Mown Unfertilized Fertilized

During the first eight years, number of species increases on small plots, but is constant on larger plots

Positive effect of mowing is most pronounced on the small spatial scale 50 x 50 10 x 10 Data from 2008, averaged over fertilization and removal treatments

The species lost from the community under various management types • Are not a random subsample, but species with specific ecological traits (contrary to Hubbell’s neutral theory) • IMPORTANCE OF SPECIES TRAITS • (mostly based on 2004 biomass data)

Species based approach to trait response to environment • 1. Calculate the environmental response for each individual species (we have used constrained ordination framework / RDA) • 2. Predict the species response on the basis of species traits (various regressions, regression trees)

Species response to fertilization (RDA score, positive values mean that the species gains from fertilization) Cirsium palustre

Potential height is good predictor of response to fertilization and mowing • With increasing nutrients, the plants are released from competition for nutrients, but simultaneously, the importance of competition for light increases - the taller plants are in advantage. • The taller plants are harmed more by mowing • Higher asymmetry of competition for light (in comparison with competition for nutrients) explains decline in species richness.

The role of the dominant (Molinia caerulea) consequences of its removal

Litter - Molinia produces large amount of slowly decaying litter. Its removal causes decrease of litter amount (with exception of mown & fertilized conditions) As a consequence, seedling recruitment is supported by Molinia removal. Molinia competes also after its death.

Seedling number - Averaged over 1996 -2007

In the presence of Molinia, the peak of biomass is shifted from June to August (and is slightly higher) - only weak statistical support Unfertilized plots only shown The presence of a single species can considerably shift the seasonal biomass dynamics

In mown plots, the removed Molinia is replaced by other grasses, however, there is no other grass which would be able to replace Molinia in unmown plots. Take home message: The dominant species (when removed) is not always replaced with the species from the same functional group.

Thanks for the help Thanks for money Czech Science Foundation (GACR) Framework V – VISTA project Iva Spackova, Alena Vitova, Petr Macek, Francesco de Bello, Jiri Dolezal, Vojtech Lanta, Jonathan Titus, Eva Chaloupecka, Katerina Palkova, David Zeleny

Thank you for your attention

Two approaches to analyse species trait response • Species based: can we predict the species response on the basis of its traits? – Traits are predictors of species response (trait value is a fixed characteristics of individual species) • Community (plot) based: how do the community (weighted) average [or variability] respond to environmental characteristics? – Traits (averages, variability) are response. Trait plasticity can be evaluated

Community (plot) based approach • 1. Calculate the plot composite characteristics (proportions of functional groups, weighted averages of quantitative traits) • 2. Analyse the composite characteristics by univariate (ANOVA) or multivariate (RDA) methods

Changes in proportions of four “functional” groups 1. Grasses 2. Other graminoids (mostly sedges) 3. Forbs (other than legumes) 4. Legumes (not very common in our plots)

Methodological note 1 • Regression tree – highly non-parametric regression approach – Well designed to account for the non-additivity • The tall plants are always harmed by mowing • In a group of “not so tall” plants, also other factors play a role: e. g. those with a rosette respond more positively

Methodological note 2 • Species based approach – species are considered independent observations – problem of phylogenetic relatedness – Do we need phylogenetic correction? – To which extend could be the similarity of species responses explained by the similarity of their traits, and to which extend by the phylogenetic relatedness?

RDA, additive effects of treatments (no interactions) Proportion of four (functional? ) groups in various treatments. Grasses supported by both, mowing and fertilization

RDA - “species traits” (treatment specific) Additive effects

Separating the effect of species composition and plasticity • Plants in fertilized plots could be taller than in unfertilized either because – The species composition changed in favor of genuinely taller plants (Prunella vulgaris was replaced by Holcus lanatus) – The species composition does not change, but the same species grow taller in fertilized plots – Combination of the two

Vegetative height

Over all the traits • For the change in species composition (i. e. for fixed traits), mowing is more important, whereas for the plastic response (i. e. for specific values and for the interaction), the fertilization is much more important. I. e. fertilization changes traits of individual species, mowing is a selection force, discriminating according to species traits

Univariate analyses with individual traits • The community based approach (i. e. weighted averages) resulted more frequently in significant results • It is easier to predict what will be the prevailing traits in the community, than to predict, how will individual species behave

Individual RDA analyses. Note, that in CANOCO, RDA is scaled so that the total variability (before subtracting the effect of covariables) equals 1. M – mowing, F –Fertilization, S – specificity (i. e. fixed vs. specific values). * means interaction. Analysi s no. Respons e Explan atory variabl es Covari ables Variability after subtracting covariables Variability explained by the explanatory variable(s) F P 1 Fixed M, F - 1. 000 0. 423 3. 293 0. 004 2 Fixed M F 0. 833 0. 255 3. 978 0. 012 3 Fixed F M 0. 745 0. 167 2. 609 0. 022 4 Specific M, F - 1. 000 0. 547 5. 431 0. 002 5 Specific M F 0. 593 0. 139 2. 767 0. 046 6 Specific F M 0. 861 0. 408 8. 095 0. 004 7 All M*S, F*S plot ID, S 0. 163 0. 130 17. 483 0. 002 8 All M*S plot ID, S, F*S 0. 063 0. 030 7. 979 0. 002 9 All F*S plot ID, S, M*S 0. 134 0. 100 26. 987 0. 002