Differences in Beta Diversity Between Exotic and Native Grasslands Vary with Scale Along a Latitudinal Gradient Leanne M

Ecology, Evolution and Organismal Biology Ecology, Evolution and Organismal Biology Publications

4-2015 Differences in Beta Diversity Between Exotic and Native Grasslands Vary with Scale Along a Latitudinal Gradient Leanne M. Martin Iowa State University

Brian J. Wilsey Iowa State University, [email protected]

Follow this and additional works at: http://lib.dr.iastate.edu/eeob_ag_pubs Part of the Biodiversity Commons, Ecology and Evolutionary Biology Commons, and the Plant Breeding and Genetics Commons The ompc lete bibliographic information for this item can be found at http://lib.dr.iastate.edu/ eeob_ag_pubs/88. For information on how to cite this item, please visit http://lib.dr.iastate.edu/ howtocite.html.

This Article is brought to you for free and open access by the Ecology, Evolution and Organismal Biology at Iowa State University Digital Repository. It has been accepted for inclusion in Ecology, Evolution and Organismal Biology Publications by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Differences in Beta Diversity Between Exotic and Native Grasslands Vary with Scale Along a Latitudinal Gradient

Abstract Biodiversity can be partitioned into alpha, beta, and gamma components, and beta diversity is not as clearly understood. Biotic homogenization predicts that exotic species should lower beta diversity at global and continental scales, but it is still unclear how exotic species impact beta diversity at smaller scales. Exotic species could theoretically increase or decrease beta diversity relative to natives depending on many factors, including abiotic conditions, community assembly history, management, dispersal rates of species, and connectivity among patches. We sampled plant species abundances in 42 novel, exotic- and native-dominated (remnant) grasslands across a latitudinal gradient in the tallgrass prairie region, and tested whether exotic and native grasslands differed in beta diversity at three scales: across sites within the entire biome, across sites within regions, and across locations within sites. Exotic-dominated grasslands differed from native-dominated grasslands in beta diversity at all scales, but the direction of the difference changed from positive to negative as scales went from large to small. Contrary to expectations, exotic-dominated grasslands had higher beta diversity than native-dominated grasslands at the largest scale considered. This occurred because the identity of dominant exotic species varied across the latitudinal gradient, with many exotic grassland pairs exhibiting zero similarity, whereas native-dominated grasslands differed more gradually with distance. Beta diversity among sites within a region was variable, with exotic-dominated grasslands having 29% higher beta diversity than native grasslands in the south and 33% lower beta diversity in the north. Within sites, beta diversity was 26% lower in exotic-dominated than native grasslands. Our results provide evidence that different regional identities and abundances of exotics, and lack of connectivity in fragmented landscapes can alter beta diversity in unexpected ways across spatial scales.

Keywords alpha diversity, beta diversity, exotic species, gamma diversity, grasslands, homogenization, invasive species, native species, tallgrass prairie R

Disciplines Biodiversity | Ecology and Evolutionary Biology | Plant Breeding and Genetics

Comments This article is from Ecology 96 (2015): 1042-1051, doi:10.1890/14-0772.1. Posted with permission.

This article is available at Iowa State University Digital Repository: http://lib.dr.iastate.edu/eeob_ag_pubs/88 Ecology, 96(4), 2015, pp. 1042–1051 Ó 2015 by the Ecological Society of America

Differences in beta diversity between exotic and native grasslands vary with scale along a latitudinal gradient

1 LEANNE M. MARTIN AND BRIAN J. WILSEY Department of Ecology, Evolution, and Organismal Biology, 253 Bessey Hall, Iowa State University, Ames, Iowa 50011 USA

Abstract. Biodiversity can be partitioned into alpha, beta, and gamma components, and beta diversity is not as clearly understood. Biotic homogenization predicts that exotic species should lower beta diversity at global and continental scales, but it is still unclear how exotic species impact beta diversity at smaller scales. Exotic species could theoretically increase or decrease beta diversity relative to natives depending on many factors, including abiotic conditions, community assembly history, management, dispersal rates of species, and connectivity among patches. We sampled plant species abundances in 42 novel, exotic- and native-dominated (remnant) grasslands across a latitudinal gradient in the tallgrass prairie region, and tested whether exotic and native grasslands differed in beta diversity at three scales: across sites within the entire biome, across sites within regions, and across locations within sites. Exotic-dominated grasslands differed from native-dominated grasslands in beta diversity at all scales, but the direction of the difference changed from positive to negative as scales went from large to small. Contrary to expectations, exotic-dominated grasslands had higher beta diversity than native-dominated grasslands at the largest scale considered. This occurred because the identity of dominant exotic species varied across the latitudinal gradient, with many exotic grassland pairs exhibiting zero similarity, whereas native-dominated grasslands differed more gradually with distance. Beta diversity among sites within a region was variable, with exotic-dominated grasslands having 29% higher beta diversity than native grasslands in the south and 33% lower beta diversity in the north. Within sites, beta diversity was 26% lower in exotic-dominated than native grasslands. Our results provide evidence that different regional identities and abundances of exotics, and lack of connectivity in fragmented landscapes can alter beta diversity in unexpected ways across spatial scales. Key words: alpha diversity; beta diversity; exotic species; gamma diversity; grasslands; homogenization; invasive species; native species; tallgrass prairie.

INTRODUCTION composition can change due to species sorting according Species diversity can be partitioned into alpha, beta, to abiotic conditions, such as soil types or climates. and gamma components in order to understand how Second, differences in assembly history, in which timing species diversity varies at multiple scales across space and arrival order of plant species vary within an (Whittaker 1960). Alpha (local) diversity and gamma environment, can alter beta diversity (Chase 2003, (regional) diversity are commonly measured, but beta Questad and Foster 2008, Martin and Wilsey 2012). diversity is not as clearly understood. Beta diversity, Patches within an area that have the same assembly defined as turnover or patchiness in species identities history should have lower beta diversity, whereas and abundances across space, provides us with the basic patches that differ in their history should have higher understanding of how species are distributed, as well as beta diversity. Third, animal activities, such as grazing, how alpha and gamma diversity are related (e.g., urine deposition, or digging, can alter plant species Whittaker 1960, Olden and Poff 2003, Novotny et al. composition among patches and increase beta diversity 2007, Wilsey 2010, Anderson et al. 2011, Leprieur et al. (Steinauer and Collins 1995). Fourth, metacommunity 2011). theory predicts that dispersal rates and connectivity of Beta diversity of plant communities can, theoretically, habitat patches can influence beta diversity (Wilson vary depending on many factors (e.g., Loreau 2000, 1992, Leibold et al. 2004). Dispersal rates and connec- Chase 2003, Soininen et al. 2007). First, community tivity are inversely related to beta diversity (Loreau 2000, Soininen et al. 2007, Qian 2009) such that species Manuscript received 22 April 2014; revised 4 September with greater numbers of propagules with more time to 2014; accepted 17 September 2014. Corresponding Editor: disperse among connected sites should, theoretically, N. J. B. Kraft. lower beta diversity by occupying more habitat patches 1 Present address: Kansas Biological Survey, 2101 Con- stant Avenue, University of Kansas, Lawrence, Kansas 66047 (all other things being equal). All of these factors can USA. E-mail: [email protected] interact, such that abiotic conditions and assembly 1042 April 2015 EXOTIC AND NATIVE BETA DIVERSITY DIFFERS 1043 history can both alter community composition and beta which homogenization vs. differentiation occurs. Our diversity. lack of understanding of how exotics are altering One human impact with the potential to alter these patterns of beta diversity at multiple, finer scales is determinants of beta diversity is introduction of multiple unfortunate, because these are scales at which manage- exotic (non-native) plant species (Hobbs et al. 2006). ment decisions take place (Gering et al. 2003, Martin et The usual prediction is that exotic species should al. 2005, Wilsey et al. 2005). homogenize biota at broad (continental and global) Understanding how exotic species are impacting beta spatial extents. However, to understand whether exotic diversity at multiple scales is important. Beta diversity species can homogenize biota, one must first examine the declined with latitude (e.g., Hillebrand 2004, Soininen et level of beta diversity that exists for native species at a al. 2007) for both exotic and native species in North particular scale. Native species at continental and global America (McKinney 2004b, Qian and Ricklefs 2007, scales are theoretically adapted to vast variation in soil Qian 2008). Studies comparing beta diversity across types and biomes in which they have a long evolutionary broad extents between native and exotic plant species history of interaction with their environments. Thus, have typically utilized floristic surveys (e.g., Qian and native species composition changes across biome Ricklefs 2006, Qian et al. 2008), and they have assisted boundaries, resulting in maximum levels of beta our understanding of the effects of exotic species on beta diversity at global and continental extents. Exotic diversity. Floristic tests determine whether beta diversity species, however, are now establishing and generating is higher or lower for exotic vs. native species, based on persistent, novel communities (Hobbs et al. 2006, species presence or absence. However, this method Kulmatiski 2006, Wilsey et al. 2011, Martin et al. assumes that species have equal abundances across sites. 2014). Exotic species are predicted to homogenize biota, Since exotic species can become highly dominant in or lower beta diversity, compared to native species at landscapes (Wilsey et al. 2009, 2011, Martin et al. 2014), global and continental extents because they are being taking species abundances into account when testing for dispersed by humans across the Earth and continents, effects of exotics will be important (Cleland et al. 2004, and inhabit anthropogenically disturbed areas, which Wilsey et al. 2005, Seabloom et al. 2013, Martin et al. are increasing in their extent at large scales (Ellis and 2014). Ramankutty 2008). Management practices may encour- We sampled plant species abundances in 42 novel, age establishment of exotic species in some cases, exotic- and native-dominated grasslands across a especially when exotic forage species are seeded and latitudinal gradient in the tallgrass prairie region of the promoted. Floristic studies suggest that exotic species United States, which is a fragmented system of grassland can, indeed, homogenize biota, measured by decreased fields surrounded by crop fields and other intensively beta diversity at continental and global extents (McKin- managed lands. We tested whether exotic and native ney and Lockwood 1999, McKinney 2004b, Qian and grasslands differed in beta diversity at three scales: Ricklefs 2006). across sites within the biome (across the entire gradient), At scales finer than the continent, however, such as across sites within regions (southern, central, and within soil types or biomes, exotic plant species could northern), and across locations within sites. This either increase or decrease beta diversity. Native species allowed us to assess whether exotic species homogenize, compositions will tend to be more homogeneous within differentiate, or do not change biota at multiple scales than across soil types or biomes. Most exotic species, within a biome. Finally, we asked whether any especially grassland species, were intentionally intro- differences found at the within-site scale varied across duced by humans (Mack and Lonsdale 2001), and the latitudinal gradient. Previously, we found that site- humans are likely to have introduced different species to level (alpha) diversity was much lower in exotic- (species different places, resulting in variation in community richness 13.2; Simpson’s diversity 2.5) than native- assembly histories and management. If humans are dominated (species richness 30.8; Simpson’s diversity introducing exotic species to different areas, and exotics 6.8) grasslands (Martin et al. 2014), but beta diversity are dispersing within soil types or within entire biomes was not assessed. Thus, our study uniquely contributes that have highly connected habitats, then exotics could to understanding how replacement of native- by exotic- lower beta diversity relative to native species composi- dominated systems may be altering beta diversity by tions at those scales. However, if exotic species are not utilizing species abundance data, and by comparing dispersing from patch to patch, due to low levels of exotic- vs. native-dominated sites at multiple scales connectivity among suitable habitats, short time frames, across a gradient. or different management regimes, then exotic species METHODS could cause higher beta diversity relative to native species (Olden and Poff 2003, McKinney 2004a). Site selection and sampling design Consistent with this, floristic studies have found mixed Sampling of 21 exotic- and 21 native-dominated results on whether exotic species are altering beta grasslands took place in 2010 and 2011 along a diversity at finer scales (McKinney 2004a, Qian et al. latitudinal gradient in the tallgrass prairie region, from 2008). Therefore, we know little about the scales at Texas to Minnesota, USA. This is an ideal region to test 1044 LEANNE M. MARTIN AND BRIAN J. WILSEY Ecology, Vol. 96, No. 4 for differences in beta diversity because 85–99.9% of ment of species abundances for distance-decay and original tallgrass prairie has been converted to alterna- regional beta diversity analyses. Abundances from all tive land uses, including exotic-dominated rangelands, pin hits within each plot were combined to assess beta cropland, and human land-use (Samson and Knopf diversity among plots within sites. 1994). Sites were owned by a variety of public and private entities. Native-dominated grasslands were Relationships between frequency and abundance previously unplowed remnants and had a variety of To understand whether abundances of species that grazing, haying, and burning land-use histories. They were in exotic or native sites were related to their had historically high levels of connectivity across the frequency of occurrence, we used linear regression. We entire biome prior to European settlement, but are summed the total number of hits for each species (i.e., currently not connected due to fragmentation. Exotic- abundances) that was present in exotic or native sites dominated grasslands were previously plowed or un- and regressed it against the proportion of exotic or plowed, grazed, hayed, or had a combination of these native sites, respectively, in which they were found (i.e., factors, or land managers or owners did not know long- relative frequency). We compared slopes with Proc term land-use history. Exotic status in tallgrass prairie GLM using SAS with origin as a categorical variable, grasslands can be due to past planting or invasion from and with ln(abundance) as a dependent variable and outside, but which process was involved for all of our relative frequency as an independent variable (Littell et sites is not known. However, no exotic sites were al. 2004). currently purposefully being managed for high exotic abundance to our knowledge; they had not been plowed Diversity indices for at least seven years; and no site was grazed or hayed We decomposed species diversity into its spatial during the year of sampling. Sites were selected such that components of a, b, and T diversity for all levels at for every native remnant, an exotic-dominated grassland which beta diversity differences were tested. This with similar soil types was located in the same geologic allowed us to calculate and interpret the more com- region at approximately the same latitude (see Martin et monly measured a and T spatial indices of diversity al. [2014] for analysis indicating soils in exotic and native using multiplicative partitioning of beta diversity, where 0 sites did not significantly differ). For every native site, an b ¼ T/mean a diversity, using Shannon’s (eH ¼ 0 exotic site was sampled on the same day during peak exp[Rln( pi ) 3 pi]), where H is Shannon’s diversity biomass season for native sites. Native and exotic sites index, and pi is the relative abundance of each species i ranged in size from 1.4–17.6 ha (mean 6.5 ha) and from (Wilsey 2010), as well as species richness (S). For beta 1.1–55.8 ha (mean 11 ha), respectively. Point-intercept diversity along the gradient, we calculated T diversity by sampling (described in Sampling), confirmed that sites combining data from all native sites or from all exotic selected as native or exotic in the field could be assigned sites, with a diversity being calculated at the site level. to a categorical variable (see Results). We divided grassland sites into three regions according to their locations in the southern (Texas and Oklahoma, Sampling n ¼ 16 sites), central (Kansas and Missouri, n ¼ 8 sites), We used a point-intercept sampling technique to or northern (Iowa, Minnesota, and South Dakota, n ¼ measure species abundances within each site. This 18 sites) tallgrass prairie regions to assess whether beta technique consisted of placing four, 1 m long pins diversity across sites within regions differed between perpendicular to the ground at each corner of a 20 3 50 native and exotic grasslands. For beta diversity within cm frame at 25 random locations (plots) within each regions, we calculated T diversity by combining data site, and counting the number of pin hits per species. from all native sites or from all exotic sites within a Thus, there were four pins dropped at each of 25 plots, region, again with a diversity at the site level. For beta totaling 100 pin drops per site. Point intercept results are diversity within sites, we calculated T diversity by highly correlated with biomass (mean r2 across 35 combining data from all 25 quadrats within a site, and 0 species ¼ 0.89, range ¼ 0.69–0.99, in Wilsey et al. 2011 a diversity from each quadrat. Shannon’s (eH ) and experiment in tallgrass prairie). Species were designated species richness were calculated in R v.2.13.2 statistical as either native or exotic using national references when software using the diversity and specpool functions, possible (49% of species; Flora of North America respectively, in the vegan 2.0 package (Oksanen et al. 1993þ), and regional references when necessary (Great 2011, R Development Core Team 2011). Plains Flora Association 1986, Gleason and Cronquist 1991, Eilers and Roosa 1994, Diggs et al. 1999, Tests of beta diversity Yatskievych 2006). We were unable to identify 5% of Latitudinal gradient.— We assessed whether exotic- species to native or exotic status based on available plant dominated grassland sites differed in beta diversity from material, and these were removed from calculations of native-dominated sites within the biome by comparing proportion exotic (i.e., when assessing native or exotic distance-decay rates in species similarity across the status of grassland). Species hits from all plots within latitudinal gradient using a permutation procedure sites were combined to generate one site-level measure- (Anderson et al. 2011). We used an exponential decay April 2015 EXOTIC AND NATIVE BETA DIVERSITY DIFFERS 1045 function because exotic-dominated sites were better modeled with an exponential decay than a linear model (linear AIC ¼51.8, r2 ¼ 0.30; exponential decay AIC ¼ 60.5, r2 ¼ 0.33), and exponential and linear models were similar for native sites (linear AIC ¼361.3, r2 ¼ 0.37; exponential decay AIC ¼361.9, r2 ¼ 0.37). The permutation procedure was necessary because of many zero-similarity values associated with exotic sites. We used the nls procedure in R v.2.13.2 to generate a nested ðb1xþb2xgÞ model, y ¼ (a1 þ a2g)e , where y is pairwise Bray- Curtis similarity (1 Bray-Curtis dissimilarity) using species abundance data; x is pairwise Euclidean distance using latitude; and g represents a group variable, where g is 1 for native sites and 0 for exotic sites (R Development Core Team 2011). We interpreted the model coefficients as a1 as the intercept coefficient for exotic sites, a1 þ a2 as the intercept coefficient for native sites, b1 as the decay coefficient for exotic sites, and b1 þ FIG. 1. Regression of abundance of species (ln-transformed b2 as the decay coefficient for native sites. We used a total number of hits) and frequency of species occurrence permutation test to test the null hypothesis H0: b2 ¼ 0, (proportion of sites) in either exotic- (solid circles) or native- under which both groups have the same decay. The dominated (open circles) sites. permutation procedure consisted of randomly reassign- ing native and exotic categories and recalculating tions to test for homogeneity of multivariate dispersion pairwise distance and similarity values 10 000 times to in the vegan 2.0-2 package, R version 2.13.2 (Oksanen et generate a null distribution of b2 coefficients. If distance- al. 2011, R Development Core Team 2011). decay rates did not differ, then the observed b2 would be Within-site beta diversity.—We tested for differences similar to the mean generated under the null distribu- in beta diversity within sites by comparing average Bray- tion, and if they differed, then the observed b2 would be Curtis dissimilarity of plots to their median within each significantly higher or lower than the mean b2 generated grassland site as a measure of beta diversity using under the null distribution. We calculated the two-tailed vegdist and betadisper functions in the vegan package p value as the proportion of times the absolute value of 2.0-2, R version 2.13.2 (Oksanen et al. 2011, R the b2 estimates under the null distribution (with random Development Core Team 2011). We then used a general assignment) was greater than the b2 estimate with actual linear model (Proc GLM in SAS version 9.3) and Type I data, i.e., SS to test for within-site differences in beta diversity 0 1 between all native and exotic sites (Littell et al. 2004). XN Explanatory variables included area as a covariate p @ bi . bˆ A=N ¼ j 2 j j 2 j because beta diversity within a site could increase with i larger area, as well as latitude, origin (native or exotic), i and latitude by origin interactions. We included latitude where N is the number of permutations, b2 is coefficient b2 estimated under the ith iteration of the permutation, and the latitude 3 origin interaction as explanatory ˆ and b2 is the estimated b2 with actual data. variables to determine whether beta diversity within sites Regional beta diversity.—We tested for differences varied across the gradient, and whether this differed within each region using a permutation test of homo- between exotic and native grasslands. Finally, we geneity of multivariate dispersion (PERMDISP), which correlated the multiplicative diversity indices with the avoids problems of lack of independence among distance to median in multivariate space to determine pairwise site comparisons (Anderson et al. 2006). This how closely related these values were to each other. test consists of measuring average dissimilarity from RESULTS each observational unit to their group centroid or median in multivariate principal coordinates analysis Point-intercept sampling confirmed the native–exotic (PCoA) space and conducting a permutation test on status of grasslands. Native sites ranged from 0–30% residuals. Here, we used Bray-Curtis dissimilarity and exotic cover and exotic sites ranged from 68–100% the group median, which is more conservative but exotic cover (see Plate 1; see also Martin et al. 2014). conceptually similar to using the group centroid, to There were 262 species total across all sites (i.e., total measure average dissimilarity for native vs. exotic sites gamma). Species abundances (ln[number of hits]) were within each region (Anderson et al. 2006), and we tested positively related to relative frequencies within exotic for differences by permuting residuals 999 times. We and native sites (exotic y ¼ 0.81 þ 17.28[frequency], r2 ¼ used the vegdist function to calculate Bray-Curtis 0.53; native y ¼ 1.44 þ 7.80[frequency],r2 ¼ 0.55; Fig. 1). dissimilarity, and the betadisper and permutest func- However, the slope was steeper for exotic species than 1046 LEANNE M. MARTIN AND BRIAN J. WILSEY Ecology, Vol. 96, No. 4

TABLE 1. Values of spatial components for the multiplicative partitioning of beta diversity at all spatial scales for tallgrass prairie plants of exotic and native origin.

0 eH S Scale and Origin T Mean ab T Mean ab Among sites across gradient Exotic 14.4 3.4 4.2 135 12.7 10.6 Native 29.8 11.2 2.7 212 31.2 6.8 Among sites within regions Northern Exotic 4.6 3.3 1.4 54 11.2 4.8 Native 22.2 12.6 1.8 131 37.1 3.5 Central Exotic 12.7 5.6 2.3 55 18.8 2.9 Native 13.2 9.5 1.4 66 28.3 2.3 Southern Exotic 7.8 2.5 3.1 56 11.4 4.9 Native 21.2 10.5 2.0 87 26.1 3.3 Among quadrats within sites (mean 6 SE) Exotic 3.4 6 0.1 2.0 6 0.0 1.7 6 0.0 12.7 6 0.3 2.7 6 0.0 4.7 6 0.1 Native 11.2 6 0.1 4.1 6 0.0 2.7 6 0.0 31.2 6 0.4 5.9 6 0.1 5.3 6 0.0

H0 0 Note: Here b ¼ T/mean a diversity, using Shannon’s (e ¼ exp [Rln( pi ) 3 pi]), where H is Shannon’s diversity index, and pi is the relative abundance of each species i; S is species richness. for natives (t ¼ 6.76, P , 0.001). No exotic species lands had higher beta diversity than native-dominated occurred in a frequency higher than 12 sites, whereas grasslands within the biome (i.e., across the latitudinal some native species had very high frequencies, occurring gradient). Specifically, similarity in species composition in all or nearly all native sites (Fig. 1). The most frequent among exotic-dominated sites decayed at a faster rate exotic species at exotic sites were Poa pratensis L. (12 of than in native-dominated sites (Fig. 2). The observed b2 21 sites), Bromus inermis Leyss. (10 sites), Cirsium parameter was significantly greater than the mean arvense (L.) Scop. (6 sites), Melilotus alba and officinalis generated under the null distribution (observed model, (L.) Lam. (6 sites), and Sorghum halepense (L.) Pers. (6 y ¼ (0.50 þ0.06xg)e(0.22xþ0.16xg), P ¼ 0.005). Although sites). The most frequent native species at native sites native sites changed in species composition across the were Andropogon gerardii Vitman and Sorghastrum gradient, they always had at least some level of species nutans (L.)Nash(21of21sites),Schizachyrium and abundances in common. In contrast, pairs of exotic scoparium (Michx.) Nash (20 sites), Solidago canadensis sites often had no species in common, and this occurred L. (19 sites), Carex spp. (19 sites), Sporobolus compositus across nearly the entire range of pairwise distances (Fig. (Poir.) Merr. (18 sites), and Symphyotrichum ericoides 2). Distance-decay results were similar when we used (L.) G.L. Nesom (18 sites).

Diversity indices 0 Alpha and gamma diversity indices (eH )were consistently lower for exotic than native grasslands at all spatial scales (see also Martin et al. [2014] for tests showing site-level alpha richness and diversity were 56% and 63% lower, respectively, in exotic grasslands; Table 1). The beta diversity indices were qualitatively similar (i.e., higher or lower) to results obtained using dissimilarity measures (Table 1). Shannon’s beta diver- 0 sity index (eH ), but not beta diversity calculated with species richness, was highly correlated with diversity values obtained by measuring distance of species composition across quadrats (Appendix A: Fig. A1). FIG. 2. Distance decay in Bray-Curtis similarity (y-axis) Tests of beta diversity across all pairwise distances calculated using units of degrees Beta diversity differed between native and exotic latitude (x-axis) between 21 exotic-dominated grassland sites (solid line) and between 21 native-dominated grassland sites grasslands, but the direction and magnitude of the (dashed line). Sites were sampled across a latitudinal gradient in difference changed with scale. Exotic-dominated grass- the tallgrass prairie region, USA. April 2015 EXOTIC AND NATIVE BETA DIVERSITY DIFFERS 1047

Jaccard’s similarity index (data not shown). Regional scale differences in beta diversity were variable. Exotic- dominated grasslands had 29% higher beta diversity than native grasslands in the south (PERMDISP F1,14 ¼ 4.9, P ¼ 0.04), 40% higher beta diversity in the central region (PERMDISP marginally signiﬁcant F1,6 ¼ 4.7, P ¼ 0.10), and 33% lower beta diversity in the north (PERMDISP F1,16 ¼ 9.9, P ¼ 0.004; Fig. 3). The difference changed in direction at the within site scale. Beta diversity was 26% lower, on average, within exotic-dominated grasslands than within native grasslands (origin F1,37 ¼ 25.8, P , 0.001; Fig. 4), and the area of sites did not signiﬁcantly affect beta diversity (area covariate F1,37 ¼ 0.16, P , 0.69). Beta diversity did not change with latitude (GLM latitude F1,37 ¼ 0.3, P ¼ 0.62), and exotic-native differences were consistent across latitudes (GLM latitude x origin F1,37 ¼ 0.0, P ¼ 0.97).

DISCUSSION Previously, we reported on lower alpha diversities in exotic communities than in native communities that they replaced (Martin et al. 2014). Here, we found that exotic- and native-dominated grasslands differed in beta diversity at all scales. Exotic-dominated grasslands did not consistently have lower beta diversity at the scales we measured. Rather, we found that exotic grasslands had higher, lower, or similar beta diversity, depending on the scale of observation. Contrary to hypotheses suggesting that exotics should exhibit decreased beta diversity as scale increases (Olden and Poff 2003), we found the opposite trend. Exotic grasslands had higher beta diversity across sites along the entire latitudinal gradient; higher, lower, or similar beta diversity at regional scales; but lower beta diversity within sites than did native grasslands. Our results provide evidence that different regional identities and abundances of exotics, FIG. 3. Species composition in multivariate principal and a lack of connectivity in fragmented systems, can coordinates analysis (PCoA) space for axes 1 and 2, and alter beta diversity in unexpected ways across scales. permutation test of homogeneity of multivariate dispersion (PERMDISP) results for exotic- and native-dominated grass- Even though reduced beta diversity was not a land sites in the (a) northern, (b) central, and (c) southern consistent phenomenon across scales, differences found tallgrass prairie (TGP). Smaller gray points represent the here and in other studies (e.g., Olden and Poff 2003, median species composition of all native or all exotic sites for McKinney 2004a) are not necessarily contradictory. that region. Instead, our results suggest that to understand beta diversity, we should consider how it is differentially of the entire landscape composed of exotic vs. native affected by assembly history, species introductions, species, and this should be considered in the future to patch connectivity, and management practices. Exotic determine whether results of beta diversity studies are species identity changed across sites, and sites were not consistent with ours. connected, leaving patches highly dominated by a single The trend we found at the largest scale occurred exotic species that did not spread to other sites. This led because the identity of dominant exotic species varied to a patchy distribution of strongly dominated stands. across the latitudinal gradient, with many exotic Native sites, which were once one continuous, connected grassland pairs exhibiting zero similarity. For example, system, tended to have high within-site beta diversity, exotic-dominated grasslands at southern latitudes were but shared species along the latitudinal gradient, dominated primarily by C4 species such as Bothriochloa lowering beta at larger scales compared to exotic sites. ischaemum (L.) Keng (King Ranch bluestem), S. Native sites are now rare in the tallgrass prairie system, halepense (Johnson grass), Panicum coloratum L. and exotic grasslands are common (Samson and Knopf (kleingrass), or Cynodon dactylon (L.) (Bermuda grass; 1994). Our study did not take into account proportions Martin et al. 2014). However, these species were never 1048 LEANNE M. MARTIN AND BRIAN J. WILSEY Ecology, Vol. 96, No. 4

FIG. 4. Average distance to the median of species composition in quadrats within sites and GLM (general linear model) results using Bray-Curtis dissimilarity for exotic- and native-dominated grasslands in a latitudinal gradient in the tallgrass prairie region, USA. The F statistic represents the exotic–native difference. found at northern latitudes, which were dominated and a more homogeneous landscape across the tallgrass primarily by C3 species including B. inermis (smooth prairie region (Loreau 2000). Thus, levels of native beta brome) (Martin et al. 2014). Previously, we showed an diversity we observed in the tallgrass prairie biome could abrupt shift in the dominant photosynthetic mode be reminiscent of the once highly connected tallgrass across the gradient in exotic, but not in native, prairie ecosystem. grasslands (Martin et al. 2014). Combined, these data When exotic species were introduced or invaded the suggest that changes in beta diversity with exotic species tallgrass prairie biome, we hypothesize that many introductions could be associated with important different assembly history events and management changes in functional diversity. These differences may practices occurred in a highly fragmented grassland not always be permanent and altering management environment lacking the high levels of connectivity that practices in exotic systems by seeding native species, once existed. For example, different species could have mowing, or haying could increase functional diversity in been intentionally introduced and managed as forage some cases (Foster et al. 2009, Weigelt et al. 2009, species in the north vs. the south, and these species have Questad et al. 2011). In contrast, native-dominated not spread across the gradient (Martin et al. 2014). grasslands differed more gradually with distance, Lindborg and Eriksson (2004) found that historical because some native species were found consistently connectivity of grasslands was an important determi- across the gradient (e.g., A. gerardii [big bluestem] and nant of plant species diversity in grasslands in Sweden. S. nutans [Indian grass]). Common management practices, such as mowing, could Our results along the entire latitudinal gradient also have unintentionally promoted exotic species (e.g., differed from ﬁndings of reduced beta diversity from Gibson et al. 1993). Experimental evidence, however, exotics in other studies (e.g., Qian and Ricklefs 2006), indicates that exotic species can also become dominant possibly because of differences in native beta diversity at without human management (Wilsey et al. 2009, 2011). the scale studied. At global or continental scales, where Thus, past management may not be the sole factor many biome types are present, beta diversity of native promoting dominance of exotics, and understanding species is very high. When exotics are introduced or human- vs. plant-mediated dynamics at broad scales will invade into anthropogenically disturbed habitats at be important. Dickson et al. (2012) and Martin and these largest scales, they persist across many biomes Wilsey (2012) experimentally varied the arrival order of and continents and homogenize ﬂora. Within biomes, native and exotic species and found that exotic- however, native species compositions do not vary as dominated communities developed when exotic species much as across biomes, as we found in the tallgrass establish before natives. Thus, we hypothesize that, prairie biome. Prior to fragmentation, native prairies regardless of whether there were purposeful or haphaz- were highly connected across the biome for very long ard introductions, the lower connectivity and dispersal periods of time. This likely allowed for high rates of rates among sites can explain why exotics have not dispersal, leading to very high levels of local diversity spread across the biome. April 2015 EXOTIC AND NATIVE BETA DIVERSITY DIFFERS 1049

PLATE 1. (Left) A diverse, native grassland and (right) an exotic, ‘‘novel’’ grassland. Photo credit: B. J. Wilsey.

At the regional scale, we found that exotic sites had considering only total number of species, which suggest reduced beta diversity in the north, but higher beta that exotics are increasing species richness at regional diversity in the south. We hypothesize that compared to levels (e.g., Ellis et al. 2012), our results suggest that the north, the south either had more species introduced exotic introductions are resulting in diversity declines (and thus more community assembly events and in tallgrass prairie grasslands. The novel ﬁnding here is associated management), weaker competitors, or less that this reduced level of alpha diversity in exotic stressful conditions that permitted establishment of systems has interesting impacts on beta diversity. Beta more exotic species (Alpert et al. 2000, Sax 2001, Liu diversity can sometimes be high even where alpha and etal.2005).Therefore,whenexoticspecieswere gamma are low. Furthermore, lower alpha diversity in introduced in the south, they were able to cause greater exotic sites could help explain higher beta diversity divergence in species composition relative to native across the gradient. If one of only a few species is prairies they replaced. replaced by another species across exotic sites, beta Exotic grasslands exhibited lower beta diversity within diversity would be higher compared to replacing one sites consistently across the gradient, which supports the species across the much more diverse native sites. homogenization hypothesis and other work showing Finally, having multiple sites strongly dominated by a increased dominance of exotic species at local levels. single but separate species is not the same (and not Local dominance by exotic species could be due to intuitively as diverse) as having multiple species in each multiple factors, including past management, greater site, even when sites themselves are similar to one levels of competitive suppression, soil feedbacks, evolu- another. tion of increased competitive ability, or novel weapons, Conclusion among others (Blossey and Notzold 1995, Callaway and Ridenour 2004). In contrast to low connectivity among We found that beta diversity can increase, decrease, sites, connectivity among patches within sites is high. or remain unchanged with exotic species introductions Thus, exotic species have a greater opportunity to depending on scale and location in the tallgrass prairie disperse within sites. We hypothesize that the high level biome, suggesting that we are beginning to understand of local dominance and dispersal probably led to the full range of impacts of exotics on this measure of homogenization within sites. Recent experimental evi- diversity. Our work suggests that interpreting beta dence also suggests that, when exotic species are diversity patterns at multiple scales will enable us to interacting in communities, exotic species themselves develop a better understanding of how human can cause declines in both species and functional management, assembly histories, dispersal abilities, diversity (Wilsey et al. 2011, Martin et al. 2014), which and connectivity of habitats lead to patchiness could help explain lower levels of alpha diversity within in nature. exotic grasslands. ACKNOWLEDGMENTS Although beta diversity was sometimes higher This publication was developed under STAR Fellowship among exotic than native grasslands at large scales, Assistance Agreement no. 91722701-0 awarded by the U.S. alpha and gamma diversity and richness were consis- Environmental Protection Agency (EPA). It has not been tently lower in exotic grasslands at all spatial scales formally reviewed by the EPA. The views expressed in this publication are solely those of the authors, and EPA does not (Martin et al. 2014). This suggests that exotic endorse any products or commercial services mentioned in this introductions have resulted in diversity declines for publication. We thank The Nature Conservancy, Minnesota the most common measures of diversity. Unlike studies Department of Natural Resources, U.S. Fish and Wildlife 1050 LEANNE M. MARTIN AND BRIAN J. WILSEY Ecology, Vol. 96, No. 4

Service, Iowa Department of Natural Resources, Guthrie Hillebrand, H. 2004. On the generality of the latitudinal County Conservation Board (Iowa), Missouri Department of diversity gradient. American Naturalist 163:192–211. Conservation, University of Kansas Field Station, Oklahoma Hobbs, R. J., et al. 2006. Novel ecosystems: theoretical and State University, Oklahoma Department of Wildlife Conser- management aspects of the new ecological world order. vation, USDA ARS, and private landowners for providing Global Ecology and Biogeography 15:1–7. sites. The work was also partially supported by USDA-NIFA Kulmatiski, A. 2006. Exotic plants establish persistent commu- grant 2014-67003-22067. nities. Plant Ecology 187:261–275. Leibold, M. A., et al. 2004. The metacommunity concept: a LITERATURE CITED framework for multi-scale community ecology. Ecology Alpert, P., E. Bone, and C. Holzapfel. 2000. Invasiveness, Letters 7:601–613. invasibility and the role of environmental stress in the spread Leprieur, F., P. A. Tedesco, B. Hugueny, O. Beauchard, H. H. of non-native plants. Perspectives in Plant Ecology, Evolu- Durr, S. Brosse, and T. Oberdorff. 2011. Partitioning global tion and Systematics 3:52–66. patterns of freshwater fish beta diversity reveals contrasting Anderson, M. J., et al. 2011. Navigating the multiple meanings signatures of past climate changes. Ecology Letters 14:325– of b diversity: a roadmap for the practicing ecologist. 334. Ecology Letters 14:19–28. Lindborg, R., and O. Eriksson. 2004. Historical landscape Anderson, M. J., K. E. Ellingsen, and B. H. McArdle. 2006. connectivity affects present plant species diversity. Ecology Multivariate dispersion as a measure of beta diversity. 85:1840–1845. Ecology Letters 9:683–693. Littell, R. C., W. W. Stroup, and R. J. Freund. 2004. SAS for Blossey, B., and R. Notzold. 1995. Evolution of increased linear models. SAS Institute Inc., Cary, North Carolina, competitive ability in invasive nonindigenous plants: a USA. hypothesis. Journal of Ecology 83:887–889. Liu, J., S. Liang, F. Liu, R. Wang, and M. Dong. 2005. Invasive alien plant species in China: regional distribution patterns. Callaway, R. M., and W. M. Ridenour. 2004. Novel weapons: Diversity and Distributions 11:341–347. invasive success and the evolution of increased competitive Loreau, M. 2000. Are communities saturated? On the ability. Frontiers in Ecology and the Environment 2:436– relationship between a, b, and T diversity. Ecology Letters 443. 3:73–76. Chase, J. M. 2003. Community assembly: when should history Mack, R. N., and W. M. Lonsdale. 2001. Humans as global matter? Oecologia 136:489–498. plant dispersers: getting more than we bargained for. Cleland, E. E., M. D. Smith, S. J. Andelman, C. Bowles, K. M. Bioscience 51:95–102. Carney, M. C. Horner-Devine, J. M. Drake, S. M. Emery, Martin, L. M., K. A. Moloney, and B. J. Wilsey. 2005. An J. M. Gramling, and D. B. Vandermast. 2004. Invasion in assessment of grassland restoration success using species space and time: non-native species richness and relative diversity components. Journal of Applied Ecology 42:327– abundance respond to interannual variation in productivity 336. and diversity. Ecology Letters 7:947–957. Martin, L. M., H. W. Polley, P. P. Daneshgar, M. A. Harris, Dickson, T. L., J. L. Hopwood, and B. J. Wilsey. 2012. Do and B. J. Wilsey. 2014. Diversity, photosynthetic mode, and priority effects benefit invasive plants more than native ecosystem services differ between native and novel ecosys- plants? An experiment with six grassland species. Biological tems. Oecologia 175:687–697. Invasions 14:2617–2624. Martin, L. M., and B. J. Wilsey. 2012. Assembly history alters Diggs, G. M., B. L. Lipscomb, and R. J. O’Kennon. 1999. alpha and beta diversity, exotic-native proportions and Shinner’s and Mahler’s Illustrated Flora of North Central functioning of restored prairie plant communities. Journal Texas. Botanical Research Institute of Texas, Fort Worth, of Applied Ecology 49:1436–1445. Texas, USA. McKinney, M. 2004a. Do exotics homogenize or differentiate Eilers, L. J., and D. M. Roosa. 1994. The vascular plants of communities? Roles of sampling and exotic species richness. Iowa: an annotated checklist and natural history. University Biological Invasions 6:495–505. of Iowa Press, Iowa City, Iowa, USA. McKinney, M. 2004b. Measuring floristic homogenization by Ellis, E. C., E. C. Antill, and H. Kreft. 2012. All is not loss: non-native plants in North America. Global Ecology and plant biodiversity in the Anthropocene. PLoS ONE 7:e30535. Biogeography 13:47–53. Ellis, E. C., and N. Ramankutty. 2008. Putting people in the McKinney, M. J., and J. L. Lockwood. 1999. Biotic homog- map: anthropogenic biomes of the world. Frontiers in enization: a few winners replacing many losers in the next Ecology and the Environment 6:439–447. mass extinction. Trends in Ecology and Evolution 14:150– Flora of North America Editorial Committee, editors. 1993þ. 453. Flora of North America North of Mexico. 16þ vols. New Novotny, V., et al. 2007. Low beta diversity of herbivorous York, New York, USA. insects in tropical forests. Nature 448:692–695. Foster, B. L., K. Kindscher, G. R. Houseman, and C. A. Oksanen, J., F. Guillaume Blanchet, R. Kindt, P. Legendre, Murphy. 2009. Effects of hay management and native species P. R. Minchin, R. B. O’Hara, G. L. Simpson, P. Solymos, sowing on grassland community structure, biomass, and M. H. H. Stevens, and H. Wagner. 2011. vegan: community restoration. Ecological Applications 19:1884–1896. ecology package. R package version 2.0-2. http://CRAN. Gering, J. C., T. O. Crist, and J. A. Veech. 2003. Additive R-project.org/package¼vegan partitioning of species diversity across multiple spatial scales: Olden, J. D., and N. L. Poff. 2003. Toward a mechanistic implications for regional conservation of biodiversity. understanding and prediction of biotic homogenization. Conservation Biology 17:488–499. American Naturalist 162:442–460. Gibson, D. J., T. R. Seastedt, and J. M. Briggs. 1993. Qian, H. 2008. Latitudinal gradient of beta diversity for exotic Management practices in tallgrass prairie: large- and small- vascular plant species in North America. Diversity and scale experimental effects on species composition. Journal of Distributions 14:556–560. Applied Ecology 30:247–255. Qian, H. 2009. Beta diversity in relation to dispersal ability for Gleason, H. A., and A. Cronquist. 1991. Manual of vascular vascular plants in North America. Global Ecology and plants of Northeastern United States and adjacent Canada. Biogeography 18:327–332. New York Botanical Garden, New York, New York, USA. Qian, H., M. L. McKinney, and I. Ku¨hn. 2008. Effects of Great Plains Flora Association. 1986. Flora of the Great Plains. introduced species on floristic similarity: comparing two US University Press of Kansas, Lawrence, Kansas, USA. states. Basic and Applied Ecology 9:617–625. April 2015 EXOTIC AND NATIVE BETA DIVERSITY DIFFERS 1051

Qian, H., and R. E. Ricklefs. 2006. The role of exotic species in Weigelt, A., W. W. Weisser, N. Buchmann, and M. Scherer- homogenizing the North American ﬂora. Ecology Letters 9: Lorenzen. 2009. Biodiversity for multifunctional grasslands: 1293–1298. equal productivity in high-diversity low-input and low- Qian, H., and R. E. Ricklefs. 2007. A latitudinal gradient in diversity high-input systems. Biogeosciences 6:1695–1706. large-scale beta diversity for vascular plants in North Whittaker, R. H. 1960. Vegetation of the Siskiyou Mountains, America. Ecology Letters 10:737–744. Oregon and California. Ecological Monographs 30:279–338. Questad, E. J., and B. L. Foster. 2008. Coexistence through Wilsey, B. J. 2010. An empirical comparison of beta diversity spatio-temporal heterogeneity and species sorting in grass- indices in establishing prairies. Ecology 91:1984–1988. land plant communities. Ecology Letters 11:717–726. Questad, E. J., B. L. Foster, S. Jog, K. Kindscher, and H. Wilsey, B. J., P. P. Daneshgar, and H. W. Polley. 2011. Loring. 2011. Evaluating patterns of biodiversity in managed Biodiversity, phenology and temporal niche differences grasslands using spatial turnover metrics. Biological Conser- between native and novel exotic-dominated grasslands. vation 144:1050–1058. Perspectives in Plant Ecology, Evolution and Systematics R Development Core Team. 2011. A language and environment 13:256–275. for statistical computing. R Foundation for Statistical Wilsey, B. J., L. M. Martin, and H. W. Polley. 2005. Predicting Computing, Vienna, Austria. http://www.r-project.org plant extinction based on species-area curves in prairie Samson, F., and F. Knopf. 1994. Prairie conservation in North fragments with high beta richness. Conservation Biology America. Bioscience 44:418–421. 19:1835–1841. Sax, D. F. 2001. Latitudinal gradients and geographic ranges of Wilsey, B. J., T. B. Teaschner, P. P. Daneshgar, F. I. Isbell, and exotic species: implications for biogeography. Journal of H. W. Polley. 2009. Biodiversity maintenance mechanisms Biogeography 28:139–150. differ between native and novel exotic-dominated communi- Seabloom, E. W., et al. 2013. Predicting invasion in grassland ties. Ecology Letters 12:432–442. ecosystems: is exotic dominance the real embarrassment of richness? Global Change Biology 19:3677–3687. Wilson, D. S. 1992. Complex interactions in metacommunities, Soininen, J., J. J. Lennon, and H. Hillebrand. 2007. A with implications for biodiversity and higher levels of multivariate analysis of beta diversity across organisms and selection. Ecology 73:1984–2000. environments. Ecology 88:2830–2838. Yatskievych, G. 2006. Steyermark’s Flora of Missouri. Steinauer, E. M., and S. L. Collins. 1995. Effects of urine Missouri Botanical Garden Press, St. Louis, Missouri and deposition on small-scale patch structure in prairie vegeta- Missouri Department of Conservation, Jefferson City, tion. Ecology 76:1195–1205. Missouri, USA.

SUPPLEMENTAL MATERIAL

Ecological Archives Appendix A is available online: http://dx.doi.org/10.1890/14-0772.1.sm