Global Change Biology Global Change Biology (2014) 20, 2211–2220, doi: 10.1111/gcb.12548

Archaeobotanical evidence for climate as a driver of ecological community change across the anthropocene boundary
and B R I A N J . C O P P I N S C H R I S T O P H E R J . E L L I S , R E B E C C A Y A H R , R O C
IO B E L I N C H ON Royal Botanic Garden Edinburgh, 20A Inverleith Row, Edinburgh EH3 5LR, UK

Abstract The biodiversity response to climate change is a major focus in conservation research and policy. Predictive models that are used to project the impact of climate change scenarios – such as bioclimatic envelope models – are widely applied and have come under severe scrutiny. Criticisms of such models have focussed on at least two problems. First, there is an assumption that climate is the primary driver of observed species distributions (‘climatic equilibrium’), when other biogeographical controls are often reliably established. Second, a species’ sensitivity to macroclimate may become less relevant when impacts are down-scaled to a local level, incorporating a modifying effect of species interactions structuring communities. This article examines the role of different drivers (climate, pollution and landscape habitat structure) in explaining spatial community variation for a widely applied bioindicator group: lichen epiphytes. To provide an analysis free of ‘legacy effects’ (e.g. formerly higher pollution loads), the study focused on hazel stems as a relatively short-lived and recently colonized substratum. For communities during the present day, climate is shown to interact with stem size/age as the most likely explanation of community composition, thus coupling a macroclimatic and community-scale effect. The position of present-day communities was projected into ordination space for eight sites in England and compared to the position of historical epiphyte communities from the same sites, reconstructed using preserved hazel wattles dating mainly to the 16th Century. This comparison of community structure for the late- to post-Mediaeval period, with the post-Industrial period, demonstrated a consistent shift among independent sites towards warmer and drier conditions, concurrent with the end of the Little Ice Age. Long-term temporal sensitivity of epiphyte communities to climate variation thus complements spatial community patterns. If more widely applied, preserved lichen epiphytes have potential to generate new baseline conditions of environment and biodiversity for preindustrial lowland Europe. Keywords: biodiversity, environmental change, industrialization, lichens, wattles Received 22 October 2013; revised version received 17 January 2014 and accepted 28 January 2014

Introduction Lichen epiphytes are a functionally important indicator guild used to understand the interrelationships between environmental change and biodiversity. Lichen epiphytes are diverse in forests and woodland landscapes (Ellis, 2012), contributing to ecosystem function by structuring food webs (Stubbs, 1989; Petterson et al., 1995; Gunnarson et al., 2004) and supporting water and nutrient cycles (Knops et al., 1996; Antoine, 2004). They are sensitive to pollutants (Nash, 2010), providing bioindicators for environmental degradation related to fossil fuel combustion (Hawksworth & Rose, 1970; Geebelen & Hoffman, 2001), agricultural nitrogen (N) deposition (Van Herk, 1999; Van Herk et al., 2003) and consequent human Correspondence: Dr Christopher J. Ellis, Royal Botanic Garden Edinburgh, 20A Inverleith Row, Edinburgh EH3 5LR, UK, tel. +44 0131 248-2993, fax +44 0131 248-2901, e-mail: [email protected]

© 2014 John Wiley & Sons Ltd

health impacts (Cislaghi & Nimis, 1997). More recently, attention has turned to the possible effect of human-induced climate change on lichens. Studies have indicated that regional lichen distributions may be controlled by climate (McCune et al., 1997; Jovan & McCune, 2004; Giordani & Incerti, 2008) and may shift in response to climate change scenarios (Ellis et al., 2007). These results signal a potential threat to epiphytic diversity from climate change if species are unable to track suitable climate space. This threat would emerge if increased rates of population loss under a suboptimal macroclimatic regime are coupled with lowered rates of colonization weakening either a rescue effect or community turnover within heavily fragmented forest landscapes (cf. Hannah et al., 1995; Travis, 2003). Active debate surrounds lichen epiphyte sensitivity to climate variability and the relevance of climate change as a threat; this debate for lichens captures more general uncertainties when estimating the biodiversity 2211

2212 C . J . E L L I S et al. response to global change. Principally, studies across divergent ecological guilds have questioned the widely held assumption that macroclimate is a primary explanation of species distributions, bringing into doubt a wide body of bioclimatic evidence founded on this principle (Pearson & Dawson, 2003; Thuiller et al., 2005). The bioclimatic debate is summarized here at two scales (biogeographical and ecological); first in general terms and then second more specifically for lichen epiphytes. At a biogeographical scale, distribution records may be explained not simply by climate, but by a range of alternative and reliable processes (Svenning & Skov, 2004; Willner et al., 2009), weakening the core assumption of ‘climatic equilibrium’. This weak explanatory power may remain hidden in standard methodologies because species distributions which tend to be spatially aggregated can be modelled statistically in an apparently effective, though spurious way, against autocorrelated climate surfaces (Beale et al., 2008; Chapman, 2010). At an ecological scale, the relevance of coarse-grained macroclimate has been questioned (Kennedy, 1997) compared for example, to the importance of local microhabitat and autogenic community processes including species interactions. These smaller-scale effects may significantly modify and thus be critically important in understanding the community-mediated species-response to macroclimatic variability (Brooker, 2006). For lichens at a biogeographical scale, publicly available distribution records of the type used to implement bioclimatic models (cf. Sober
on & Townsend Peterson, 2004; Newbold, 2010) are accumulated over multi-decadal time periods. Patterns in these data may therefore be structured by a range of drivers apart from climate and most notably the air pollution regime (McCune et al., 1997; Geiser & Neitlich, 2007; Ellis & Coppins, 2009). During the industrial and postindustrial period in Western Europe and North America, the pollution regime has been a dynamic one, including a reduction in long-standing high SO2 levels (Baumgardner et al., 2002; Vestreng et al., 2007) alongside stable or increasing rates of N deposition (Baumgardner et al., 2002; Holland et al., 2005; Dentener et al., 2006). Accumulated lichen distribution records therefore span periods of dynamic environmental change, such that their accurate interpretation is dependent on a complex amalgam of drivers, with climate one possible component among many. For lichens at an ecological scale, sampling along altitudinal gradients has brought into question the relative importance of macroclimate, compared to the role of within- and between-site microhabitat heterogeneity in controlling community composition (Moning et al.,

2009). Apparent macroclimatic sensitivity may emerge from the process of community assembly by speciessorting into microhabitats, that is, if microhabitats themselves are spatially differentiated across a landscape, as suggested by regional trends in forest/woodland composition or stand structure (cf. Angelstam & D€ onz-Breuss, 2004; Ellis, 2012). In this article, we perform a test for lichen epiphytes which is relevant to the broader bioclimatic debate because it aims (i) to confirm or refute the role of climate in structuring species distributions spatially and using time-series data and (ii) to examine this sensitivity to climate in terms of community-scale patterns. First, we field-sampled trends in lichen epiphyte community structure along spatial gradients for individual hazel stems (Corylus avellana) which represent a relatively short-lived and widely distributed substratum for epiphytes in Britain. This sampling design has three important attributes: (1) Focusing on a single substratum type avoids the confounding effect of landscapescale spatial variation in woodland structure which could generate spurious correlations with climate; (2) The limited time span for colonization onto hazel stems avoids the obfuscating effect of past environmental regimes, which may complicate long-term distribution records accumulated over extended periods of field recording (‘legacy effects’); (3) The focus on community structure tests for a signature macroclimatic effect which would be apparent despite the established role of smaller-scale ecological processes, including species interactions during autogenic succession (Ellis & Ellis, 2013). Second, we provide a unique temporal dimension to our sampling which extends time-series data in community ecology by several centuries. We compared present-day epiphyte communities with a novel baseline sampled from the archaeobotanical record, yielding high-resolution community data for the landscape of lowland Europe prior to the Industrial Revolution (Mathias, 1983) and during the Little Ice Age (Mann, 2002), when both the climate and pollution regime were substantially different. This temporal approach has strong potential to validate the functional significance of species turnover where this is related to the presentday climate, if equivalent time-series shifts in community structure are found to be consistent across multiple independent sites with the magnitude and direction of historical climate change.

Materials and methods

Hazelwood study sites and field sampling A database of 115 widely distributed hazelwood sites in mainland Britain was compiled from literature and internet © 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220

C O M M U N I T Y R E S P O N S E T O H I S T O R I C A L C L I M A T E C H A N G E 2213 searches. These sites were matched to 5 km-gridded GIS layers for four contrasting variables, which were preselected to summarize relevant landscape-scale climate and pollution regimes. Individual climate and pollution variables can show strong regional covariation within our study area (Ellis & Coppins, 2009) and we aimed to target a subset of variables which previous ordination analysis had demonstrated to be the least correlated among a range of potential factors (Ellis et al., 2011), and thereby avoid data mining during regression analysis (see Epiphyte Community Analysis, below). The four chosen variables were: (i) mean annual temperature (°C) and (ii) annual precipitation (mm), both for a baseline period of 1961– 2006 made available by the UK’s Met Office (Perry & Hollis, 2005), and (iii) SO2 as lg kg 1 at 1013 hPa and at 25 °C based on a spatial interpolation for both 1987 and 2004–2006 averaged values and (iv) NHx as KgN ha 1 yr 1 using interpolation of 2004–2006 averaged values made available by the UK’s Centre for Ecology and Hydrology (cf. RoTAP, 2012). The variables capture contrasting aspects of the climate (rainfall, temperature) and relate to a period of declining (1987 SO2) and subsequently lower (2004–2006 SO2) fossil fuel pollution, alongside the increasing relative importance of agricultural N deposition (see Introduction). Thirteen hazelwood sample sites were selected at different positions along continua in the climate and pollution variables (Fig. 1) aiming to minimize potential covariation among environmental effects: (i) the two hazelwood sites with the

lowest values for SO2, (ii) the two sites with the lowest values for NHx, (iii) two contrasting sites with the highest SO2 and NHx values respectively and (iv) using scatter plots to identify an additional suite of seven sites whose pollution values were uncorrelated with those for the climatic setting. Differently sized buffers were used to calculate the extent of ancient woodland with continuity of tree cover >250 years (Roberts et al., 1992; Spencer & Kirby, 1992) within a spatial radius of 1, 5 and 10 km extending outwards from the selected hazelwood sample sites. Spatial analyses were conducted in ArcMap v. 10 (ESRI, 2012). Although compiled from a range of sources across different available time periods, the environmental data characterized known regional variability including strong environmental contrasts among the sample sites (Table S1). Epiphyte communities were field sampled by randomly selecting a single hazel stool located centrally within each site, and then following a random walk procedure (cf. Lisewski & Ellis, 2010) to encounter consecutive stools. For each selected stool a single stem was sampled within each of nine stratified size categories: 0–2 cm; 2–4 cm; 4–6 cm; 6–8 cm; 8–10 cm; 10– 12 cm; 12–14 cm; 14–16 cm and >16 cm dbh. A random aspect was selected between 0 and 360 degrees to mark the starting position at a stool, progressing clockwise to sample consecutive stems along the direction of travel. Up to 10 stems per size category were sampled per site, across a maximum of 60 hazel stools. On each of the sampled stems, the occurrence of epiphytic lichens was estimated using a ladder transect of 5,

Fig. 1 The location of 13 hazelwood sites from which hazel stems were sampled (present-day environment), and eight villages from which preserved epiphyte communities were sampled, relating to the pre-Industrial environment. © 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220

2214 C . J . E L L I S et al. 1 9 1 cm squares, placed at 1.3 m height at four contrasting aspects (N, S, E, W). Additional ladders were sampled at randomly selected intermediate aspects for progressively larger stems (Fig. S1).

Pinheiro et al., 2013; Pinheiro & Bates, 2000). We selected from the alternative models applied to DCA axes one and two using Akaike’s Information Criterion (AIC) and optimized the best model (lowest AIC) with backwards selection using loglikelihood tests (Zuur et al., 2009).

Epiphyte community analysis Species abundances were calculated for individual hazel stems as frequency of occurrence. Singleton species which occurred in only one sample were excluded from analysis, with the reduced community matrix used for ordination by  detrended correspondence analysis (DCA; Leps & Smilauer,  2003), implemented in CANOCO v. 4.5 (Ter Braak & Smilauer, 2002). DCA sample scores summarized community compositional turnover among sampled hazel stems and the first two orthogonal DCA axes were adopted as dependent community response variables. The sample scores for these DCA axes were compared to the contrasting landscape-scale environmental effects: mean temperature, annual precipitation, SO2 (1987, 2004–2006), NHx and ancient woodland at the 1, 5 and 10 km scale, in addition to stem diameter (dbh) which was log transformed to a normal distribution. A suite of exploratory full models included the main effect of each environmental variable separately, with dbh and their interaction term. These preliminary fits to the DCA sample scores were implemented using generalized linear mixed models (GLMM), with site identity and stool identity as hierarchically nested random effects (‘nlme’ package, in R:

Archaeobotanical sampling Preliminary studies investigating lichen epiphytes as an archaeobotanical resource (Yahr et al., 2011) had confirmed that entire communities are excellently preserved on roundwood wattles of infill panels (Yahr & Ellis, 2009; Fig. 2) because these have been protected by a coating of daub (Sunshine, 2006). The preserved communities are representative of epiphyte diversity at the time the roundwood was harvested and incorporated as a wattle into a panel. The use of wattles for constructing infill panels rapidly declined during the transition to industrialization in England (Graham, 2004; Sunshine, 2006) and combined with archaeological survey reports it was possible to target sampling towards pre-18th Century material. Wattles were sampled from the exposed internal panels of 12 private homes, within eight villages in lowland England (Fig. 1). The choice of sample sites was informed by archaeological survey information which is made publicly available for vernacular buildings (English Heritage, 2013). Wattles sampled from a single source (infill panel) within a building are referred to here as an ‘artefact’. Sampled artefacts were returned to the herbarium at the Royal Botanic Garden Edinburgh for gentle

(a)

(c)

(b)

Fig. 2 A typical timber framed building of the type from which wattles were sampled (a) and an exposed wattle panel (b); with a cleaned wattle artefact dating from the 16th Century (c) showing exceptional preservation of the preserved lichen epiphyte community. © 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220

C O M M U N I T Y R E S P O N S E T O H I S T O R I C A L C L I M A T E C H A N G E 2215 cleaning with distilled water, and lichen epiphyte communities were quantified as species frequency of occurrence within 1 cm2 contiguous quadrats positioned over the bark surface. The degree of preservation of material made it possible to identify species using standard microscopy and chemical analysis. Taxonomic nomenclature follows Smith et al. (2009).

Time-series comparison The optimized GLMMs explaining variation in present-day epiphyte community composition (see Epiphyte Community Analysis, above) were used to predict the position in ordination space (i.e. the sample score along DCA axes one and two) for the eight sites from which historical artefacts had been sampled (Fig. 1) based on their present-day environmental setting. This positioning of sites from which artefacts had been sampled formed a null hypothesis of no change in ecological community structure resulting from controlling environmental effects (Cnull). Furthermore, these same eight sites could be positioned into ordination space using their preserved epiphyte communities, by borrowing a technique from palaeoecology which applies a weighted averaging procedure as a transfer function (cf. Birks, 1995; Ellis et al., 2008) to estimate sample scores for the observed preindustrial community structure (Cobs). The difference between Cnull (the expected position in ordination space based on the present-day environment) and Cobs (the position in ordination space reconstructed from historical epiphyte communities) can be used to infer change in ecological community composition caused by a shift along the controlling environmental gradients represented in DCA axes one and two. We tested these differences statistically: the sampled scores (Cobs) for historical epiphyte communities were compared to the expected scores based on the present-day environment (Cnull) using a paired t-test (R Core Team, 2013).

Results Samples from the 13 present-day hazelwood sites included 140 lichen species distributed across 788 stems (mean number of stems per site = 60.62  20.91 1SD). The estimated mean age of sampled hazel stems was

22  14 years 1SD, based on a diameter growth rate of 0.329 cm yr 1 derived from stem dendrochronology (C.J. Ellis & S. Eaton: Royal Botanic Garden Edinburgh, unpublished data). This places the epiphyte colonization of stems within a post-1980 declining and subsequently low SO2 environment (cf. Rose & Hawksworth, 1981; Hawksworth & McManus, 1989). Removal of singletons reduced the community matrix to 109 species and 681 hazel stem samples. Analysis of the reduced data set by DCA indicated a primary gradient (axis one) with high species turnover (gradient length = 8.67 units) which explained 3.1% of variation in the species data, with axis two explaining an additional 2.4% of variation; given the large size of the matrix, axes one and two appeared to provide substantial explanatory power. With the exception of autocorrelated SO2 values for the two different time periods, the sampling of presentday hazelwood sites successfully separated any strong covariance among contrasting variables representing the same environmental feature (climate or pollution); however, SO2 did correlate significantly, though weakly, with mean annual temperature and annual precipitation (Table 1). Furthermore, ancient woodland extent (ha) covaried across scales of 1 and 5 km, and 5 and 10 km, with additional covariance between ancient woodland extent at the 10 km scale and mean temperature (Table 1). Variation in community composition (DCA axis sample scores) was explained using the best model fits (lowest AIC) and optimized GLMMs (Table 2): (i) for DCA axis one, with annual precipitation and stem dbh as significant fixed effects and including their interaction and (ii) for DCA axis two, with mean annual temperature and stem dbh as significant fixed effects and including their interaction. Models with alternative explanatory variables relating to pollution and ancient woodland extent had lower AIC values than those which incorporated precipitation and temperature.

Table 1 Product moment correlation coefficients showing covariance among the environmental variables used to explain epiphyte community composition [1] [1] SO2 (‘87) [2] SO2 (‘04-’06) [3] NHx [4] Mean temp. [5] Ann. precip. [6] AWI_1 km [7] AWI_5 km [8] AWI_10 km

[2]

[3]

[4]

[5]

[6]

[7]

[8]

– 0.604* 0.428

– 0.778**

–

– 0.818*** 0.239 0.6* 0.619* 0.128 0.399 0.376

– 0.094 0.724** 0.673* 0.063 0.26 0.443

– 0.131 0.11 0.139 0.313 0.381

– 0.526 0.221 0.506 0.564*

*P < 0.05, **P < 0.01, ***P < 0.001. Values in bold are significant at P < 0.05. © 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220

– 0.254 0.368 0.343

2216 C . J . E L L I S et al. Table 2 Fixed effects and regression diagnostics (GLMM) for optimum models explaining hazel stem sample scores along DCA axes one and two

Random effects variance Site Stool Fixed Effects (estimate and significance) Intercept Annual precip. Mean temp. Stem dbh (log) Annual precip. * Stem dbh (log) Mean temp. * Stem dbh (log) Residual variance

DCA axis one

DCA axis two

0.64 0.06

0.14 0.18

0.526 (t = 0.879, P = 0.379503df) 0.002 (t = 3.694, P = 0.003511df) n/a 0.709 (t = 2.541, P = 0.0114503df) 0.00072 (t = 3.133, P = 0.0018503df) n/a 0.94

6.602 (t = 6.432, P < 0.0001503df) n/a 0.25 (t = 2.248, P = 0.04611df) 2.28 (t = 3.202, P = 0.0015503df) n/a 0.328 (t = 4.221, P < 0.0001503df) 0.61

Archaeobotanical sampling and time-series comparison Sampling of wattles across the eight sites for midland and southern England (Fig. 1) yielded 63 discrete artefacts, i.e. single wattles or small bundles of wattles contained within the same panel, which could be positively identified as hazel (Corylus avellana) based on wood anatomy. Archaeological reports relevant to each building from which artefacts were sampled, provided by historical building specialists, indicated that material spanned a period from the mid-15th Century to the mid-18th Century, with a majority of artefacts dating to the 1500s. Twenty-four lichen epiphytes (Table S2) were recorded from the hazel artefacts, along with the nonlichenized bark fungus Hypoxylon howeanum. Of the 24 species, two of these were not recorded from hazelwood stems in the present-day landscape (Bacidia arcuetina and Melaspilea sp. C) with a further two species occurring in one stem sample only (Pyrenula chlorospila, Pyrenula corylii) and which had therefore been excluded from the community analysis by DCA. All preserved epiphyte communities were recorded from small roundwood, with a mean dbh of 2.18  0.97 cm 1SD. We projected DCA scores for the sites from which historical artefacts had been sampled, using their present-day environmental data and artefact dbh, based on the GLMM interpretation of axis scores (Table 2). Projected values represented a null hypothesis (Cnull) of no difference when compared to the observed historical epiphyte communities sampled from artefacts (Cobs), which were positioned onto the DCA axes using community weighted averaging (Fig. 3a and b). We thus rejected our null hypotheses of no change in environmentally controlled epiphyte community structure. Instead differences between Cobs and Cnull demonstrated change between the late-Mediaeval and post-Industrial periods corresponding to: (i) a shift

towards a drier climatic regime, indicated by higher values for Cobs than Cnull on DCA axis one and (ii) a shift towards a warmer climatic regime, with higher values for Cobs than Cnull on DCA axis two. Tested using a paired t-test (Fig. 3a and b), the difference in DCA scores (Cnull v Cobs) was statistically significant for both axes (P < 0.05).

Discussion We formulated and performed a robust test for the role of alternate larger-scale environmental drivers (climate, pollution, landscape-scale ancient woodland habitat) in structuring the composition of epiphyte communities. To do this, we selected sample sites aiming to avoid strong covariation among a summary suite of climate and pollution variables and focussed on hazel stems because they present a relatively short-lived substratum with epiphyte communities free of legacy effects. On this basis, we demonstrate that climate variability may play a key role in structuring lichen epiphyte community composition. This is significant for three reasons. First, strong observational evidence for the sensitivity of lichens to industrial pollution has been well established since the 19th Century (Hawksworth, 1971). Since the 1960s, SO2 in particular has been the primary explanation of regional lichen distributions across industrialized landscapes, including those of Britain (Gilbert, 1965, 1970; Hill, 1971). Following pollution controls, lichen epiphytes have begun to colonize into previously affected areas and a recovery of the lichen flora has been observed (Seaward, 1998). However, this recovery of lichen epiphytes is modified firstly by the effect of increased atmospheric N (Van Herk et al., 2003; Mitchell et al., 2005). Secondly, former pollution impacts are subject to a lag period, possibly related to (i) species dispersal and low recolonization rates and © 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220

C O M M U N I T Y R E S P O N S E T O H I S T O R I C A L C L I M A T E C H A N G E 2217 (a)

(b)

Fig. 3 The mean ordination scores (1SD) for preserved epiphyte communities sampled from historical artefacts (Cobs), positioned by weighted averaging, compared to sample scores (Cnull) for the same sites though estimated based on the present-day environmental setting (cf. Table 2) for (a) DCA axis one, and (b) DCA axis two. For the identification of sample site codes (A1–A8), see Fig. 1.

(ii) the delayed recovery of substratum chemistry, such that lichen species or communities sampled from longer-lived substrata may continue to reflect past pollution regimes (Bates et al., 1990, 2001; Gilbert, 1992). Given the extent to which pollution has been a key driver explaining lichen distributions and community structure in our study region, it is significant that for relatively short-lived hazel stems there is now evidence that climate may exert a stronger control in the assembly of community composition than pollutants. We posit that this climatic control of community composition is likely to become an increasingly significant factor in explaining lichen biogeography as pollution levels decline across industrial Europe (Amann et al., 2011) and as dispersal-limited species associated with longer-lived substrata trend towards their equilibrium distributions in a lowered pollution environment. Second, the climatic signature is evident despite sampling at a community scale. It is reasonable to expect that a clear macroclimatic signature may be weak when investigated for community patterns because of the modifying effect of species interactions (Ara
ujo & Luoto, 2007), including autogenic feedbacks (Camill & Clark, 1998; Ellis, 2008) which may increase community resilience to macroclimatic variability (Suttle et al., 2007). There is clear evidence that epiphyte communities are subject to autogenic successional processes (Ellis & Ellis, 2013) including species interactions (e.g. facilitation and competition) which may modify the © 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220

community response to larger-scale environmental change (Brooker, 2006). Indeed, the importance of epiphyte succession along a stem chronosequence was confirmed for hazel, because stem dbh was both a significant main effect in explaining sample position along ordination axes and was included as an interaction term with the climatic setting. Previous studies have likewise indicated that successional processes show predictable variability dependent on the climatic regime (Peterson & McCune, 2001; Ellis & Coppins, 2006). Third, climatic sensitivity evidenced through a spatial correlation of epiphyte community structure with temperature and precipitation during the present day, was matched by a temporal shift in community composition that is consistent with environmental change at the end of the Little Ice Age. This result is based on the use of a novel archaeological resource: lichen epiphytes preserved on preindustrial wooden building structures (Fig. 2). These preserved epiphytes provide high-resolution community-scale data for a well-established bioindicator guild (see Introduction), accessible for a period since at least c. AD 1400 and therefore extending ecological data sets by several centuries, i.e. comparable data sets commencing in the 19th Century (well within the period of industrial impact) are considered long-term (Magurran et al., 2010). For the purposes of this study, the data provided a new time-series test of climate sensitivity, by offering a retrospective analysis

2218 C . J . E L L I S et al. which can help validate the inferred response to spatial climatic variation (Hill et al., 1999; Ara
ujo et al., 2005; Green et al., 2008). A directional shift in community structure was characterized by species associated with a cooler and wetter climate sampled from historical preindustrial artefacts and which contrast with warmer and drier environments occurring at the same sites in the present day. Our interpretation assumes the relatively local supply of hazel stems for construction (e.g. within c. 10 km) and is defensible given our sampling from lowstatus homes, for which the proximate sourcing of materials stands in contrast to the Mediaeval transport of large and high quality timber to support prestige building (Rackham, 1986, 2003). Furthermore, the historical hazel stems would likely have been sampled from an actively managed short-rotation coppice system (Rackham, 1986, 2003), while the present-day samples were from neglected coppice woods and seminatural hazelwoods in north-western Scotland. We mitigated this discrepancy by sampling across a broad range of stem sizes for present-day contrasting environmental settings. Epiphyte communities demonstrate successional trends with stem size/age (cf. Degelius, 1964; Stone, 1989; Hilmo, 1994) and it was possible to account for the small average stem size of actively coppiced hazel (i.e. early successional communities associated with historical samples) in calculating the present-day expected community composition (Cnull). Overall, our evidence indicates a direct sensitivity to climate, with lichens responsive to Little Ice Age cooling (Mann, 2002) in terms of the preserved epiphyte communities, and contrasting with early-21st Century warmer and drier conditions. In summary, we believe our results quantify for the first time the effect on lichen communities of long-term historical climatic change. The corollary is that lichen epiphytes are also likely to be directly sensitive to future human-induced climate change, adding significant complexity in understanding the recolonization of species into formerly polluted areas during recovery of the British lichen diversity (Seaward, 1998). To quantify this temporal change we utilized a novel archaeological resource to generate high-resolution community data for the late- to post-Mediaeval period; we therefore highlight the existence of a bioindicator guild (lichen epiphytes) preserved within preindustrial timber framed buildings (Yahr & Ellis, 2009; Yahr et al., 2011). Where structural elements are shown to be original and of known provenance, there is potential to apply this same technique widely across lowland Europe to develop a new understanding of the biodiversity response to climate change and industrialization across the anthropocene boundary.

Acknowledgements This research was funded by a grant from The Leverhulme Trust. We thank the private home owners who provided access to their buildings from which to sample wattles. We thank two anonymous reviewers who commented on and improved an earlier draft manuscript.

References Amann M, Bertok I, Borken-Kleefeld J et al. (2011) Cost-effective control of air quality and greenhouse gases in Europe: modeling and policy applications. Environmental Modelling and Software, 26, 1489–1501. Angelstam P, D€ onz-Breuss M (2004) Measuring forest biodiversity at the stand-scale an evaluation of indicators in European forest history gradients. Ecological Bulletins, 51, 305–332. Antoine ME (2004) An ecophysiological approach to quantifying nitrogen fixation by Lobaria oregana. The Bryologist, 107, 82–87. Ara
ujo MB, Luoto M (2007) The importance of biotic interactions for modelling species distributions under climate change. Global Ecology and Biogeography, 16, 743–753. Ara
ujo MB, Whittaker RJ, Ladle RJ, Erhard M (2005) Reducing uncertainty in projections of extinction risk from climate change. Global Ecology and Biogeography, 14, 529–538. Bates JW, Bell JNB, Farmer AM (1990) Epiphyte recolonization of oaks along a gradient of air pollution in south-east England, 1979–1990. Environmental Pollution, 68, 81–99. Bates JW, Bell JNB, Massara AC (2001) Loss of Lecanora conizaeoides and other fluctuations of epiphytes on oak in S.E. England over 21 years with declining SO2 pollution. Atmospheric Environment, 35, 2557–2568. Baumgardner RE, Lavery TF, Rogers CM, Isil SS (2002) Estimates of the atmospheric deposition of sulfur and nitrogen species: clean air status and trends network, 1990–2000. Environmental Science and Technology, 36, 2614–2629. Beale CM, Lennon JJ, Gimona A (2008) Opening the climate envelope reveals no macroscale associations with climate in European birds. Proceedings of the National Academy of Sciences, 105, 14908–14912. Birks HJB (1995) Quantitative palaeoenvironmental reconstructions. In: Statistical Modelling of Quaternary Science Data (eds Maddy D, Brew JS), pp. 161–254. Quaternary Research Association, Cambridge. Brooker RW (2006) Plant-plant interactions and environmental change. New Phytologist, 171, 271–284. Camill P, Clark JS (1998) Climate change disequilibrium of Boreal permafrost peatlands caused by local processes. The American Naturalist, 151, 207–222. Chapman DS (2010) Weak climatic associations among British plant distributions. Global Ecology and Biogeography, 19, 831–841. Cislaghi C, Nimis PL (1997) Lichens, air pollution and lung cancer. Nature, 387, 463– 464. Degelius G (1964) Biological studies on the epiphytic vegetation on twigs of Fraxinus excelsior. Acta Horti Gotoburgensis, 27, 11–55. Dentener F, Drevet J, Lamarque JF et al. (2006) Nitrogen and sulfur deposition on regional and global scales: a multimodel evaluation. Global Biogeochemical Cycles, 20, GB4003. Ellis CJ (2008) Interactions between hydrology, burning and contrasting plant groups during the millennial-scale development of sub-montane wet heath. Journal of Vegetation Science, 19, 693–704. Ellis CJ (2012) Lichen epiphyte diversity: a species, community and trait-based review. Perspectives in Plant Ecology, Evolution and Systematics, 14, 131–152. Ellis CJ, Coppins BJ (2006) Contrasting functional traits maintain lichen epiphyte diversity in response to climate and autogenic succession. Journal of Biogeography, 33, 1643–1656. Ellis CJ, Coppins BJ (2009) Quantifying the role of multiple landscape-scale drivers controlling epiphyte composition and richness in a conservation priority habitat (juniper scrub). Biological Conservation, 142, 1291–1301. Ellis CJ, Ellis SC (2013) Signatures of autogenic succession for an aspen chronosequence. Journal of Vegetation Science, 24, 688–701. Ellis CJ, Coppins BJ, Dawson TP, Seaward MRD (2007) Response of British lichens to climate change scenarios: trends and uncertainties in the projected impact for contrasting biogeographic groups. Biological Conservation, 140, 217–235. Ellis CJ, Rochefort L, Gauthier G, Pienitz R (2008) Paleoecological evidence for transitions between contrasting landforms in a polygon-patterned High Arctic wetland. Arctic, Antarctic, and Alpine Research, 40, 624–637.

© 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220

C O M M U N I T Y R E S P O N S E T O H I S T O R I C A L C L I M A T E C H A N G E 2219 Ellis CJ, Yahr R, Coppins BJ (2011) Archaeobotanical evidence for a massive loss of epiphyte species richness during industrialization in southern England. Proceed-

Mitchell RJ, Truscot AM, Leith ID et al. (2005) A study of epiphytic communities of Atlantic oak woods along an atmospheric nitrogen deposition gradient. Journal of

ings of the Royal Society B, 278, 3482–3489. English Heritage (2013) Images of England. Available at: http://www.imagesofengland.org.uk/ (accessed 4 march 2014). ESRI (2012) ArcMap v. 10.1. ESRI, Redlands, California. Geebelen W, Hoffman M (2001) Evaluation of bio-indication methods using epiphytes by correlating with SO2-pollution parameters. The Lichenologist, 33, 249– 260.

Ecology, 93, 482–492. Moning C, Werth S, Dziock F, B€assler C, Bradtka J, Hothorn T, M€ uller J (2009) Lichen diversity in temperate montane forests is influenced by forest structure more than climate. Forest Ecology and Management, 258, 745–751. Nash TH (2010) Lichen sensitivity to air pollution. In: Lichen Biology (ed Nash TH), pp. 301–316. Cambridge University Press, Cambridge. Newbold T (2010) Applications and limitations of museum data for conservation and

Geiser LH, Neitlich PN (2007) Air pollution and climate gradients in western Oregon and Washington indicated by epiphytic macrolichens. Environmental Pollution, 145, 203–218. Gilbert OL (1965) Lichens as indicators of air pollution in the Tyne Valley. In: Ecology and the Industrial Society (eds Goodman GT, Edwards RW, Lambert JM), pp. 35–47. Blackwell Scientific Publications, Oxford.

ecology, with particular attention to species distribution models. Progress in Physical Geography, 34, 3–22. Pearson RG, Dawson TP (2003) Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecology and Biogeography, 12, 361–371. Perry M, Hollis D (2005) The generation of monthly gridded datasets for a range of

Gilbert OL (1970) Further studies on the effect of sulphur dioxide on lichens and bryophytes. New Phytologist, 69, 605–627. Gilbert OL (1992) Lichen reinvasion with declining air pollution. In: Bryophytes and Lichens in a Changing Environment (eds Bates JW, Farmer AM), pp. 159–177. Clarendon Press, Oxford. Giordani P, Incerti G (2008) The influence of climate on the distrubution of lichens: a case study in a borderline area (Liguria, NW Italy). Plant Ecology, 195, 257–272.

climatic variables over the UK. International Journal of Climatology, 25, 1041–1054. Peterson EB, McCune B (2001) Diversity and succession of epiphytic macrolichen communities in low-elevation managed conifer forests in western Oregon. Journal of Vegetation Science, 12, 511–524. Petterson RB, Ball JP, Renhorn K-E, Esseen P-A, Sj€ oberg K (1995) Invertebrate communities in boreal forest canopies as influenced by forestry and lichens with implications for passerine birds. Biological Conservation, 74, 57–63.

Graham T (2004) Wattle and Daub: craft, conservation and a Wiltshire case study. Unpublished MSc, University of Bath, Bath. Green RE, Collingham YC, Willis SG, Gregory RD, Smith KW, Huntley B (2008) Performance of climate envelope models in retrodicting recent changes in bird population size from observed climatic change. Biology Letters, 4, 599–602. Gunnarson B, Hake M, Hultengren S (2004) A functional relationship between spe-

Pinheiro JC, Bates DM (2000) Mixed Effects Models in S and S-PLUS. Springer-Verlag, New York. Pinheiro J, Bates D, Debroy S, Sarkar D, Team RC (2013) nlme: linear and nonlinear mixed effects models. R Core Team. Available at: http://cran.r-project.org/web/ packages/nlme/index.html (accessed 4 march 2014). R Core Team (2013) R: A Language and Environment for Statistical Computing. R Foun-

cies richness of spiders and lichens in spruce. Biodiversity and Conservation, 13, 685–693. Hannah L, Carr JL, Lankerani A (1995) Human disturbance and natural habitat: a biome level analysis of a global data set. Biodiversity and Conservation, 4, 128–155. Hawksworth DL (1971) Lichens as litmus for air pollution: a historical review. International Journal of Environmental Studies, 1, 281–296. Hawksworth DL, McManus, PM (1989) Lichen recolonization in London under con-

dation for Statistical Computing, Vienna. Available at: http://www.R-project.org/ (accessed 4 march 2014). Rackham O (1986) The History of the Countryside. Phoenix Press, London. Rackham O (2003) Ancient Woodland. Castlepoint Press, Kirkcudbrightshire. Roberts AJ, Russel C, Walker GJ, Kirby KJ (1992) Regional variation in the origin, extent and composition of Scottish woodland. Botanical Journal of Scotland, 46, 167–189. Rose CI, Hawksworth DL (1981) Lichen recolonization in London’s cleaner air. Nat-

ditions of rapidly falling sulphur dioxide levels, and the concept of zone skipping. Botanical Journal of the Linnean Society, 100, 99–109. Hawksworth DL, Rose F (1970) Qualitative scale for estimating sulphur dioxide air pollution in England and Wales using epiphytic lichens. Nature, 227, 145–148. Hill DJ (1971) Experimental study of the effect of sulphite in lichens with reference to atmospheric pollution. New Phytologist, 70, 831–836.

ure, 289, 289–292. RoTAP (2012) Review of Transboundary Air Pollution: Acidification, Eutrophication, Ground Level Ozone and Heavy Metals in the UK. Centre for Ecology and Hydrology, Penicuik. Seaward MRD (1998) Time-space analysis of the British lichen flora, with particular reference to air quality surveys. Folia Cryptogamica Estonica, 32, 85–96.

Hill JK, Thomas CD, Huntley B (1999) Climate and habitat availability determine 20th century changes in a butterfly’s range margin. Proceedings of the Royal Society B, 266, 1197–1206. Hilmo O (1994) Distribution and succession of epiphytic lichens on Picea abies branches in a Boreal forest, central Norway. The Lichenologist, 26, 149–169. Holland EA, Braswell BH, Sulzman J, Lamarque J-F (2005) Nitrogen deposition onto the United States and western Europe: synthesis of observation and models. Eco-

Smith CW, Aptroot A, Coppins BJ, Fletcher A, Gilbert OL, James PW, Wolseley PA (2009) The Lichens of Britain and Ireland. British Lichen Society, London. Sober
on J, Townsend Peterson A (2004) Biodiversity informatics: managing and applying primary biodiversity data. Philosophical Transactions of the Royal Society B, 359, 689–698. Spencer JW, Kirby KJ (1992) An inventory of ancient woodland for England and Wales. Biological Conservation, 62, 77–93.

logical Applications, 15, 38–57. Jovan S, McCune B (2004) Regional variation in epiphytic macrolichen communities in northern and central California forests. The Bryologist, 107, 328–339. Kennedy AD (1997) Bridging the gap between general circulation model (GCM) output and biological microenvironments. International Journal of Biometeorology, 40, 119–122.

Stone DF (1989) Epiphyte succession on Quercus garryana branches in the Willamette Vally of western Oregon. The Bryologist, 92, 81–94. Stubbs CS (1989) Patterns of distribution and abundance of corticolous lichens and their invertebrate associates on Quercus rubra in Maine. The Bryologist, 92, 453–460. Sunshine P (2006) Wattle and Daub. Shire Publications, Risborough. Suttle KB, Thomsen MA, Power ME (2007) Species interactions reverse grassland

Knops JMH, Nash TH, Schlesinger WH (1996) The influence of epiphytic lichens on the nutrient cycling of an oak woodland. Ecological Monographs, 66, 159–179.  Leps J, Smilauer P (2003) Multivariate Analysis of Ecological Data using CANOCO. Cambridge University Press, Cambridge. Lisewski V, Ellis CJ (2010) Epiphyte sensitivity to a cross-scale interaction between habitat quality and macroclimate: an opportunity for range-edge conservation. Biodiversity and Conservation, 19, 3935–3949.

responses to changing climate. Science, 315, 640–642. Svenning J-C, Skov F (2004) Limited filling of the potential range in European tree species. Ecology Letters, 7, 565–573.  Ter Braak CJF, Smilauer P (2002) CANOCO v. 4.5: Software for Canonical Community Ordination. Microcomputer Power, New York. Thuiller W, Lavorel S, Ara
ujo MB, Sykes MT, Prentice IC (2005) Climate change threatens plant diversity in Europe. Proceedings of the National Academy of Sciences,

Magurran AE, Baillie SR, Buckland ST et al. (2010) Long-term datasets in biodiversity research and monitoring: assessing change in ecological communities through time. Trends in Ecology and Evolution, 25, 574–582. Mann ME (2002) Little Ice Age. In: The Earth System: Physical and Chemical Dimensions of Global Environmental Change (eds Maccracken MC, Perry JS), pp. 504–509. John Wiley & Sons, Chichester.

102, 8245–8250. Travis JMJ (2003) Climate change and habitat destruction: a deadly anthropogenic cocktail. Proceedings of the Royal Society B, 270, 467–473. Van Herk CM (1999) Mapping of ammonia pollution with epiphytic lichens in The Netherlands. The Lichenologist, 31, 9–20. Van Herk CM, Mathijssen-Spiekman EaM, De Zwart D (2003) Long distance nitrogen

Mathias P (1983) The First Industrial Nation: The Economic History of Britain 1700–1914. Routledge, London. McCune B, Daey J, Peck JE, Heiman K, Will-Wolf S (1997) Regional gradients in lichen communities of the southeast United States. The Bryologist, 100, 145–158.

air pollution effects on lichens in Europe. The Lichenologist, 35, 347–359. Vestreng V, Myhre G, Fagerli H, Reis S, Tarras
on L (2007) Twenty-five years of continuous sulphur dioxide emission reduction in Europe. Atmospheric Chemistry and Physics, 7, 3663–3681.

© 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220

2220 C . J . E L L I S et al. Willner W, Di Pietro R, Bergmeier E (2009) Phytogeographical evidence for post-glacial dispersal limitation of European beech forest species. Ecography, 32, 1011–

Yahr R, Coppins BJ, Ellis CJ (2011) Preserved epiphytes as an archaeological resource in pre-industrial vernacular buildings. Journal of Archaeological Science, 38, 1191–

1018. Yahr R, Ellis CJ (2009) Lichens in the attic. The Building Conservation Directory, 2009, 13–14.

1198. Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM (2009) Mixed Effects Models and Extensions in Ecology with R. Springer, New York.

Supporting Information Additional Supporting Information may be found in the online version of this article: Figure S1. The association between stem size and sampling effort (as an adjusted number of ladder transects) for individual hazel stems: Spearman’s rank r = 0.92, P < 0.0001 with 786 df. Table S1. Summary environmental data for 13 hazelwood sites from which hazel stems were sampled in the present-day landscape (cf. Fig. 1). Table S2. Lichen species recorded from historical artefacts (hazelwood wattles), with their DCA axis one and two scores used in weighted averaging to position preserved communities into ordination space (cf. Fig. 3a and b); ‘n/a’ indicates that a species was not sampled from the present-day hazel stems. Nomenclature follows Smith (2009).

© 2014 John Wiley & Sons Ltd, Global Change Biology, 20, 2211–2220