Fire and grazing in a mesic tallgrass prairie impacts on plant species and functional traits.

Spasojevic et al 1Running head: Plant functional traits and disturbance Fire and grazing in a mesic tallgrass prairie: impacts on plant species and functional traits Marko J Spasojevic1, Rebecca J Aicher1, Gregory R Koch2, Emily S Marquardt3, Nicholas Mirotchnick4, Tiffany G Troxler2 and Scott L Collins5 81Department of Ecology and Evolutionary Biology, 321 Steinhaus Hall, 9University of California-Irvine, Irvine, CA 92697-2525 USA 102Department of Biological Sciences, Florida International University, Miami, Fl 1133199 USA 123Department of Biology and Biochemistry, University of Houston, Houston, TX 1377204 USA 144Department of Ecology, Evolution and Environmental Biology, Columbia 15University, New York, NY 10027 USA 165Department of Biology, MSC03-2020, University of New Mexico, 17Albuquerque, NM 87131 USA 18 19 20 21 22 23 24 Spasojevic et al 25 26 27 28 29Abstract 30Fire is a globally distributed disturbance that impacts terrestrial ecosystems 31and has been proposed to be a global "herbivore” Fire, like herbivory, is a 32top-down driver that converts organic materials into inorganic products, 33alters community structure, and acts as an evolutionary agent Though 34grazing and fire may have some comparable effects on ecosystem processes 35in grasslands, they not have similar impacts on species composition and 36community structure However, the concept of fire as a global herbivore 37implies that fire and herbivory may have similar effects on plant functional 38traits Using long-term data from a mesic, native tallgrass prairie with a long 39evolutionary history of fire and grazing, we tested if trait composition 40between grazed and burned grassland communities would converge, and if 41the degree of convergence depended on fire frequency Additionally, we 42hypothesized that eliminating fire from frequently burned grassland would 43result in a state similar to unburned grassland, and, conversely, that adding 44fire into a previously unburned grassland would cause composition to 45become more similar to that of frequently burned grassland We found that 46grazing and burning once every four years showed the most convergence in 47trait, suggesting that these communities operate under similar deterministic Spasojevic et al 48assembly rules and that fire and herbivory are similar disturbances to 49grasslands at the trait-group level of organization We found that fire reversal 50had different effects depending on treatment The formerly unburned 51community that was then burned annually became more similar to the 52annually burned community in trait composition suggesting that function 53may be rapidly restored if fire is reintroduced After fire was removed from 54the annually burned community, however, trait composition developed along 55a unique trajectory indicating that a time lag is necessary for structure and 56function to return to the community 57 58Keywords: Disturbance, Fire, Grazing, Grassland, Plant Functional Traits 59 60Introduction 61Fire and herbivory are two of the most common globally distributed 62disturbances that have profound impacts on the structure and function of 63terrestrial ecosystems (Milchunas et al 1988, Bond and Van Wilgen 641996) Recently, Bond and Keeley (2005) proposed that fire is, in essence, a 65global "herbivore." Fire, like herbivory, is a top-down driver that converts 66organic materials into inorganic products, alters community structure, and 67acts as an evolutionary agent (Bond and Keeley 2005) Unlike herbivores, 68however, “combustive consumption” by fires is based on physical properties 69of available fuels, and can have rapid and severe impacts over large areas In 70addition, while fires often have relatively uniform impacts over large areas, Spasojevic et al 71the effects of grazing are much more spatially and temporally variable 72(Fuhlendorf and Engle 2004, Collins and Smith 2006) Although grazing and 73fire have some similar effects on ecosystem processes, their impacts on 74species composition differ (Collins 1987, Keeley et al 2003, Trager et al 752004, Uys et al 2004) For instance, frequent fires which are non-selective, 76decrease diversity, whereas selective grazing increases diversity in mesic 77tallgrass prairie (Collins and Smith 2006) 78 While patterns of species composition have been shown to differ in 79response to fire and grazing, the Bond and Keeley (2005) hypothesis 80suggests that these disturbances may have similar impacts on plant 81functional trait composition Functional traits integrate the evolutionary and 82ecological history of a species (Ackerly and Reich 1999, Cavender-Bares et al 832004) and both fire and grazing act as evolutionary agents (Bond and Keeley 842005) While these two disturbances can both select for specialized traits, 85such as serotiny in the case of fire (Bond and Van Wilgen 1996) or thorns in 86the case of grazing (Belovsky et al 1991, Young et al 2003), fire and grazing 87may also both select for the same suite of more general traits For example, 88both fire and grazing decrease the height of plants (Diaz et al 2001, Noy89Meir and Kaplan 2002), and increase the abundance of C4 plants (Collins et 90al 1998), legumes (Towne and Knapp 1996, Coppedge and Shaw 1998, Noy91Meir and Kaplan 2002), and non-native plants (Dantonio and Vitousek 1992, 92Seabloom et al 2003) While species composition differs in burned and 93grazed grasslands, these disturbances may be selecting for similar traits that Spasojevic et al 94are distributed among different species However, the effects of grazing and 95fire on the abundance of plant functional traits may differ depending on both 96the frequency and the intensity of burns (Engle and Bidwell 2001, Briggs et 97al 2002, Heisler et al 2003) 98 Changes in the fire regime can have strong effects on the species and 99trait composition of plant communities (Pausas 1999, Franklin et al 2001, 100Franklin et al 2005, Kahmen and Poschlod 2008) Frequent fire has been 101shown to increase the abundance of C4 grasses, legumes and annual species 102(Towne and Knapp 1996, Collins et al 1998, Kahmen and Poschlod 2008), 103while unburned communities tend to be composed of perennials, shrubs, C3 104species, broad-leaved evergreens and deciduous species (Collins et al 1998, 105Heisler et al 2003, Grund et al 2005, Kahmen and Poschlod 2008) Thus, 106removal of fire from a regularly burned system or the addition of fire to a 107previously unburned community should shift the community from dominance 108of one group to the other Additionally, the removal of fire from an ecosystem 109will have impacts on aboveground biomass (fuel load) which can, in turn, 110impact species diversity and recovery from a future fire (Thaxton and Platt 1112006) The addition of fire to a previously unburned system can impact 112ecosystem function as well as community composition For example, fires 113tend to increase soil NH4+ and NO3- in some systems (Wan et al 2001), and 114cause large gaseous losses of nitrogen (Goode et al 2000) and carbon 115(Johnson et al 2007) in others 116 Using long-term data from a mesic grassland, we tested the following Spasojevic et al 117hypotheses to examine the impacts of fire, grazing and fire reversal on 118species and trait composition: (1) trait composition in grasslands that are 119grazed or burned will converge because these disturbances exert similar top120down pressures on vegetation and community physiognomy, (2) the degree 121of trait convergence of grazed and burned grassland will depend on the 122frequency of burning, (3) removal of fire from a frequently burned grassland 123will result in a shift in species and trait composition to a state similar to an 124unburned grassland and, conversely, (4) the introduction of fire into an 125unburned grassland will cause trait composition to become more similar to 126that of a frequently burned grassland 127 128Methods 129Study Site 130This study was based on data collected at the Konza Prairie Biological Station 131(KPBS), a 3487 native tallgrass prairie located in the Flint Hills of Kansas, 132USA ranging from 320 to 444m above sea level All data used in this study 133were from upland sites which are characterized by shallow, rocky, cherty, 134silty clay loams Replicate watersheds at KPBS have been burned 135experimentally at 1-, 4-, and 20-year intervals since 1972 (Knapp et al 1361998) In 1987 bison were introduced to a 1000 area of KPBS where they 137have free access to watersheds subjected to the above fire treatments In 1382001, fire treatments were reversed on four ungrazed watersheds (fire 139reversal treatments); burning was stopped on two watersheds that had been Spasojevic et al 140burned annually since 1972 and 1978, and annual burning treatments were 141started on two watersheds that had been burned infrequently since 1973 and 1421980 143 The vegetation at KPBS is predominantly native unplowed tallgrass 144prairie, with some woody vegetation in gallery forests along drainages 145(Knight et al 1994) as well as in infrequently burned watersheds (Briggs et 146al 2002) The grassland is dominated by a matrix of C4 perennial grasses, 147including Andropogon gerardii, A scoparius, Sorghastrum nutans, and 148Panicum virgatum Although grass biomass dominates, interstitial forb 149species comprise greater than 75% of the species richness (Towne 1502002) Common perennial forbs include Aster spp., Kuhnia eupatoroides, 151Salvia azurea, and Solidago spp Woody species include Symphoricarpos 152orbiculatus, Cornus drummondii, Prunus americana, Rhus glabra and 153Juniperus virginiana (Briggs et al 2005) 154 155Vegetation Sampling 156We used long-term data from grazed and ungrazed watersheds subjected to 1571-yr, 4-yr and 20-yr fire frequencies, and fire reversal treatments Vegetation 158in each watershed was sampled in five permanently located 10 m2 circular 159quadrats equally spaced along each of four 50 m transects for a total of 20 160permanent quadrats per site in May and late August of each year Percent 161cover of species in each quadrat was estimated using a modified 162Daubenmire scale (1 = < 1%, = 2-5%, = 6-25%, = 26-50%, = 51- Spasojevic et al 16375%, = 76-95%, = 95-100%) Abundance of each species was 164determined by converting the Daubenmire scale to the midpoint of the cover 165range and then averaging species across the 20 quadrats at a site 166 167Trait Groups 168We constructed trait groups using a published trait database that included 169species from KPBS (Cleland et al 2008) The traits included life history 170strategy (annual, biennial, or perennial), detailed life form (C grass, C4 grass, 171leguminous forb, non-leguminous forb, evergreen shrub, or deciduous shrub), 172height within the canopy, clonality, and nativity Although certain plant traits 173exhibit plasticity, all of the traits we used not vary substantially with 174environmental conditions We performed hierarchical clustering on the trait 175data using Perl, which used a single linkage algorithm to merge clusters 176based on the minimum Euclidean distance to generate 15 clusters as our 177trait groups (Table 1) Complete trait data was present for all species in the 178database for the trait group clustering The cut-off for the number of clusters 179was determined by the limited species pool (no trait groups could have only 180one species), and after 15 clusters, the successive trait groups were not 181biologically meaningful We used these trait groups as an approximation of 182response patterns to fire and grazing in this mesic grassland ecosystem 183 184Data Analysis 185To describe differences in diversity between watersheds we calculated Spasojevic et al 186richness and Simpson's diversity index for each year of each watershed 187using PRIMER 5.2 (Clarke and Gorley 2001) We compared richness and 188diversity across years among watersheds using repeated measures ANOVA in 189JMP 5.1 We used a non-metric multidimensional scaling (NMDS) routine in 190PRIMER to describe differences among the unburned, annually burned, 191quadrennially burned and grazed watersheds for species composition and 192trait composition using Bray-Curtis dissimilarity for each year for each 193watershed Bray-Curtis dissimilarity is a semi-metric measure of dissimilarity 194used for continuous numerical data and does not group samples by shared 195zeros in the dataset (Beals 1984) 196 197Results 198Fire Grazing 199Species Composition 200We found that grazing and fire not have similar impacts on species 201composition in grasslands (Fig 1A) Overall, richness and diversity were both 202higher in grazed compared to burned communities (Richness: F1,3= 25.8380, 203P< 0.001, Simpson’s Diversity: F1,3=15.4026, P