Identifying foundation species in North American forests using long-term data on ant assemblage structure 1, 1,2 3 4 SYDNE RECORD, TEMPEST MCCABE, BENJAMIN BAISER, AND AARON M. ELLISON 1Department of Biology, Bryn Mawr College, 101 North Merion Avenue, Bryn Mawr, Pennsylvania 19010 USA 2Department of Earth and Environment, Boston University, 1 Silber Way, Boston, Massachusetts 02215 USA 3Department of Wildlife Ecology and Conservation, Institute of Food and Agricultural Sciences, University of Florida, 110 Newins-Ziegler Hall, PO Box 110430, Gainesville, Florida 32611 USA 4Harvard University, Harvard Forest, 324 North Main Street, Petersham, Massachusetts 01366 USA Citation: Record, S., T. McCabe, B. Baiser, and A. M. Ellison. 2018. Identifying foundation species in North American forests using long-term data on ant assemblage structure. Ecosphere 9(3):e02139. 10.1002/ecs2.2139 Abstract. Foundation species are locally abundant and uniquely control associated biodiversity, whereas dominant species are locally abundant but are thought to be replaceable in ecological systems. It is impor- tant to distinguish foundation from dominant species to direct conservation efforts. Long-term studies that remove abundant species while measuring community dynamics have the potential to (1) aid in the identi- fication of foundation vs. dominant species and, (2) once a foundation species is identified, determine how long its effects persist within a community after its loss. Long-term data on ant assemblages within two canopy-manipulation experiments—the Harvard Forest Hemlock Removal Experiment (HF-HeRE) and the Black Rock Future of Oak Forests Experiment (BRF-FOFE)—provide insights into how ant assemblages change and reassemble following the loss of Tsuga canadensis or Quercus spp. Previous research docu- mented foundation species effects on ants in the HF-HeRE for up to four years after T. canadensis loss. Six additional years of data at HF-HeRE presented for the first time here show that removal of T. canadensis resulted in taxonomic and some measures of functional shifts in ant assemblages that persisted for ten years, further supporting the hypothesis that T. canadensis is a foundation species at Harvard Forest. In contrast, ant assemblages at BRF-FOFE varied little regardless of whether oaks or other tree species were removed from the canopy, suggesting that Quercus species do not act as foundation species at Black Rock Forest. Deer and moose exclosures within each experiment also allowed for comparisons between effects on ants of foundation or dominant tree species relative to effects of large herbivores. At HF-HeRE, effects of T. canadensis were stronger than effects of large herbivores on taxonomic and functional diversity of ant assemblages. At BRF-FOFE, in contrast, effects of Quercus species were weaker than effects of large herbi- vores on ant taxonomic diversity and some measures of ant functional diversity. These findings illustrate the importance of distinguishing between the roles of irreplaceable foundation species and replaceable dominant ones in forested ecosystems along with other drivers of biodiversity (e.g., herbivory). Key words: Adelges tsugae; dominant species; Formicidae; foundation species; functional diversity; Quercus; Tsuga canadensis. Received 22 December 2017; accepted 9 January 2018. Corresponding Editor: Debra P. C. Peters. Copyright: © 2018 Record et al. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. E-mail: [email protected] INTRODUCTION becausemorespeciesareavailableintheformerto fill functional roles when species are lost (i.e., the Ecosystems with high biodiversity are hypothe- insurance hypothesis sensu Yachi and Loreau sized to be more resilient to changing environmen- 1999). However, not all species are functionally tal conditions than those with lower biodiversity equivalent, and the loss of keystone predators ❖ www.esajournals.org 1 March 2018 ❖ Volume 9(3) ❖ Article e02139 RECORD ET AL. (Paine 1966), dominant species (Whittaker 1965), myriad of native and non-native insects and or foundation species (Ellison et al. 2005b)may pathogens (Lovett et al. 2016). While we are have unexpectedly large or even irreversible sys- mourning these impending losses (Foster 2014) tem-wide impacts. In many terrestrial ecosystems, and working to control these insects and patho- foundation species and dominant species tend to gens, we also have an unparalleled opportunity be trees and share the following attributes: large to study how forests respond to, and reorganize size, local abundance, occupying basal positions in after, the loss of foundation or dominant species local food webs, and controlling ecosystem pro- by assessing how their loss changes the taxo- cesses and dynamics principally through non- nomic and functional biodiversity of associated trophic interactions (Baiser et al. 2013). In addition, species. Furthermore, within these studies we foundation species differ from dominant species in can investigate effects of other drivers of forest that the former are thought to be irreplaceable in dynamics (e.g., ungulate herbivory, earthworms) terms of their control on population and commu- with additional manipulations. nity dynamics and ecosystem processes, whereas Here, we used two forest canopy-manipulation the latter are considered replaceable (Ellison et al. experiments to ask how biodiversity within forests 2005b). Consequently, identifying foundation spe- changes and reorganizes after the experimental cies should be a key priority for conservation removal of T. canadensis or Quercus species. Large efforts to preserve the processes that they modu- herbivore exclosures within each experiment also late within their communities and ecosystems. enabled us to test for additive and interactive In eastern North American forests, Tsuga effects on biodiversity of both the loss of canopy canadensis (L.) Carriere (eastern hemlock) has been tree species and herbivory by large ungulates. identified as a foundation species (Ellison et al. Specifically, we examined the effects of these exper- 2005b, 2014); stands dominated by T. canadensis imental treatments on the species and trait diver- are different, both structurally and functionally, sity of ground- and soil-nesting ant assemblages. from stands dominated by other conifers or mix- Ants are a particularly good taxon to use for these tures of various deciduous species. Hemlock- studies because they are abundant and widespread dominated stands are dark, cool, and moist; have omnivores; are known to be responsive to local acidic, nutrient-poor soils with slow rates of nutri- environmental conditions such as canopy cover, ent cycling (Orwig et al. 2013); and are populated light availability, and soil temperature (Rescaso by generally species-poor assemblages of associ- et al. 2014); modulate and control some soil ated plants and animals (Rohr et al. 2009, Sackett ecosystem processes (Del Toro et al. 2012), often in et al. 2011, Orwig et al. 2013). concert with the canopy tree species (Kendrick On the other hand, many eastern North Ameri- et al. 2015); and are well understood both taxo- can forests are dominated in terms of numbers or nomically and functionally in northeastern North biomassbyoneormoreQuercus(oak)species America (Ellison et al. 2012, Del Toro et al. 2015). (Schuster et al. 2008). Unlike T. canadensis,how- Previous work by the authors reported that the ever, it is not clear whether oaks determine experimental removal of hemlock had large uniquely the structure and function of the forests impacts on ant assemblage structure for four years they dominate. Although oak masts are important (Sackett et al. 2011). There are now data available for certain organisms such as small mammals and for six additional years in experimental plots simu- ticks (Ostfeld et al. 1996, McShea et al. 2007), most lating hemlock loss. Here, we revisit the hypothesis core ecosystem processes of oak-dominated for- that hemlock is a foundation species for ant assem- ests, including leaf-litter decomposition rates and blages with these six additional years of data (i.e., associated soil nutrient dynamics (Polyakova and 10 yr total; Hypothesis 1). It is unknown whether Billor 2007), root respiration (Levy-Varon et al. oak species in eastern North American forests act 2012), and net ecosystem exchange (Papale and as foundation species. We hypothesize that if oaks Valentini 2003), are statistically indistinguishable are not foundation species, then the loss of oaks from forests dominated by other deciduous species will have no detectable impact on the structure of or from forests with no clear dominant species. ant assemblages (Hypothesis 2). Alternatively, if Many trees, including T. canadensis and Quer- oaks are foundation species, then the loss of oaks cus spp., are threatened with functional loss by a will have a significant impact on assemblages of ❖ www.esajournals.org 2 March 2018 ❖ Volume 9(3) ❖ Article e02139 RECORD ET AL. ants, analogous to what has been seen for hemlock. plot/block), all T. canadensis individuals were gir- Finally, we hypothesize that when a foundation dled (as done by Yorks et al. 2003) to simulate speciesispresent(asestablished by Hypotheses 1 physical impact of loss of hemlock caused by the and 2), the effect of the foundation species on ant non-native hemlock woolly adelgid (Adelges tsu- assemblages