- Email: [email protected]
Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS Dendrochronologia xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Dendrochronologia journal homepage: www.elsevier.com/locate/dendro
ORIGINAL ARTICLE
Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland Stephen Galvin a,∗ , Aaron Potito a , Kieran Hickey b a b
Palaeoenvironmental Research Unit, School of Geography and Archaeology, National University of Ireland Galway, Galway, Ireland School of Geography and Archaeology, National University of Ireland Galway, Galway, Ireland
a r t i c l e
Article history: Received 28 May 2013 Accepted 29 March 2014 Keywords: Dendrochronology Dendroclimatology Taxus baccata Ireland
a b s t r a c t
i n f o
Tree-ring research in Ireland has typically been dominated by Quercus species, particularly Quercus petraea and Quercus robur. Recent years have seen a greater focus on multi-species reconstructions in Ireland but, due to difficulties with the hardness of the wood, missing/pinched rings and fused stems, Taxus baccata has not been included in these investigations. Despite these difficulties, a 31-tree, 204-year T. baccata chronology was successfully constructed from Killarney National Park, southwest Ireland. The chronology exhibits promising dendroclimatological potential, with climatic responsiveness equivalent to that of the other major Irish tree taxa, including Quercus. The chronology shows the strongest relationship with May–June precipitation from Muckross House synoptic station (1970–2007; r = 0.521, p < 0.01) and Valentia Observatory (1941–2007; r = 0.545, p < 0.01). November–April temperatures also exhibited a strong relationship with the chronology post-1970 (r = 0.605, p < 0.01 for Muckross House, r = 0.567, p < 0.01 for Valentia Observatory), but this relationship is not time stable and breaks down for the pre-1970 Valentia Observatory record. The long-lived nature of T. baccata, the exceptional preservation of wood and rings in this hard softwood species, as well as its prominence in Irish archaeology, all point to the potential to expand this chronology both spatially and temporally, and demonstrate T. baccata’s potential in multi-site and multi-species tree-ring studies in the region. © 2014 Elsevier GmbH. All rights reserved.
Introduction Until recently, tree-ring research in Ireland has been dominated by Quercus species, particularly Quercus petraea and Quercus robur, as these taxa were common over large areas of Irish lowlands for the last 8000 years (Gardiner, 1974). Pilcher and Baillie (1980) used six Irish Quercus chronologies to examine the climate responses of the species. Their study showed that trees in Ireland and Britain are less sensitive to changes in climate when compared to those closer to their latitudinal or altitudinal limits, which had implications for the perceived dendroclimatological potential of the region. More recently, Irish tree-ring studies have been expanded to incorporate other species. García-Suárez et al. (2009) used both Q. robur and Q. petraea, as well as Fagus sylvatica, Fraxinus excelsior and Pinus sylvestris, to reconstruct temperature, rainfall, sunshine hours and Palmer Drought Severity Index values in Northern Ireland. Their approach highlights the potential benefits of using combinations
∗ Corresponding author. Tel.: +353 091 494103; fax: +353 091 495505. E-mail address: [email protected] (S. Galvin).
of species to reconstruct environmental variables, as each species will respond slightly differently to annual or seasonal changes in temperature and precipitation. This is particularly important in regions where no single climate variable leads to notably wide or narrow tree-rings, as is the case in Ireland (Pilcher, 1994). This study focuses on exploring the dendrochronological and dendroclimatological potential of Taxus baccata as an alternative source of annual tree-ring data. T. baccata has traditionally been overlooked in the Irish record due to difficulties with the hardness of the wood, missing/pinched rings, unsymmetrical trunks and fused stems (Tabbush and White, 1996; Thomas and Polwart, 2003). Therefore, this is the first study to explore the dendrochronological or dendroclimatological potential of T. baccata in Ireland, and one of only a small number of such studies of T. baccata on a global scale. T. baccata has a wide geographic range throughout Europe (Jalas and Suominen, 1973), with prominent stands in Ireland and Britain (Ratcliffe, 1977), extending northwards to Norway and Sweden ´ 1991), eastwards to Estonia, Poland and (c. 63◦ N) (Vidakovic, Turkey, and southwards to Greece (Voliotis, 1986), northern Spain ˜ ´ 1991). In the 1994), Portugal and into Algeria (Vidakovic, (Penalba, south of its European range T. baccata is largely a montane tree,
http://dx.doi.org/10.1016/j.dendro.2014.03.004 1125-7865/© 2014 Elsevier GmbH. All rights reserved.
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
2
whereas it grows from sea level to 425 m above sea level (a.s.l.) in England and Wales and up to 470 m a.s.l. on Purple Mountain in County Kerry, southwest Ireland (Moir, 1999). The species is well suited to a mild oceanic climate, avoids areas susceptible to severe winter frosts (Godwin, 1975), and is one of the longest-living plant species native to Ireland. Historically, T. baccata has played an important role in Irish society. Its density and durability saw it employed in construction of hardwearing items such as handles, wheel dowels, and weapon shafts. The fine grain and deep colouring of the species resulted in it also being used for finer items and tableware (Perrin, 2002). While modern T. baccata is demonstrated to be suitable for use in dendrochronology (Biondi, 1992; Tabbush and White, 1996), the need to measure a number of different radii from samples indicates that analysis from incrementally cored samples may prove challenging (Moir, 1999). T. baccata cores can often contain missing or very narrow rings (Thomas and Polwart, 2003), while the frequent wedging of rings often results in difficulties in cross-dating (Yadav and Singh, 2002). Other pitfalls in estimating age include old senescent trees producing new stems from the remains of the former large trunk and appearing as young trees, and the apparent production of large trees from the fusing of several trunks. Lowe (1897) argued that the fusing of multiple stems/roots would make an individual appear older than it actually is. However, overall growth rate of the new trunk when young is unlikely to be much greater than a single tree of the same size (Thomas and Polwart, 2003). Tabbush and White (1996) argue that the coalescing of seedlings would actually make little difference to age estimates. The potential difficulties involved in using T. baccata can be overcome by selecting appropriate trees to sample, i.e. those with definite single trunks or obvious separations at the base of the trunk. In addition, the extraction of two cores per tree greatly increases the likelihood of identifying missing rings. The long-lived nature of T. baccata makes it suitable for ecological and climate investigations while the hardness of the wood, although adding to the difficulties in extracting cores, increases the preservation potential of trees and rings. Complications in obtaining data from T. baccata have undoubtedly influenced the fact that only two chronology series are available from the International Tree-ring Data Bank. The chronology compiled by Bernabei and Gjerdrum (2006) in the Alps was formed from 17 cores, the oldest dating to 1896. The second T. baccata chronology, created by Kuniholm et al. (2006), was formed from 19 cores dating to 1526 in Georgia, Eurasia, and has since been utilised to cross-date other tree-ring records in the area for archaeological purposes. Publications examining T. baccata in terms of dendroclimatology are equally as rare. Moir (1999) sampled 14 T. baccata trees in Hampton Court Palace in London, noting that February–July precipitation had a positive effect on growth. January–February and October temperatures also influenced ringwidth, as did soil-moisture. Yadav and Singh (2002) sampled 18 T. baccata trees in the western Himalaya, where the 345-yearlong chronology achieved its strongest correlations with mean March–June temperatures (r = −0.310) for the period 1898–1998. Both Yadav and Singh’s (2002) and Moir’s (1999) studies showed significant inter-species correlations, indicating that T. baccata could become an important constituent in multi-site and multispecies studies.
Study site The geology and geomorphology of the Killarney area of County Kerry is diverse, something that gives rise to a wide range of forest vegetation (Kelly, 1981). Reenadinna Wood (Fig. 1a) (50◦ 01 N; 9◦ 31 W), located on a carboniferous limestone outcrop on
Muckross Peninsula, lies approximately 4.5 km south–southwest of Killarney town and is dominated by T. baccata. Mean winter and summer temperatures are 9.6 ◦ C and 18.4 ◦ C respectively, with mean total precipitation of 598 mm in winter and 264 mm in summer (Fig. 1b). The site, part of Killarney National Park, a UNESCO Biosphere Reserve, is 20 m a.s.l. It is bounded to the north by Lough Leane and by Muckross Lake to the south. The western limit of the wood lies along the geological boundary with Devonian Old Red Sandstone while, to the east, the wood is bounded by parkland where the limestone ceases to outcrop. Through pollen analysis, Mitchell (1990) showed that T. baccata has existed in the area for up to 5000 years. Reenadinna Wood (approx. 25 ha in extent) is the most extensive T. baccata woodland remaining in Ireland (Perrin, 2002). The shorter T. baccata trees on the more broken rocky terrain of the Killarney woodlands are intermingled with taller Quercus (mainly Q. petraea) in the intervening soil-filled hollows. Little grows under the T. baccata canopy, although Corylus avellana or Ilex aquifolium become frequent locally, with the former sometimes replacing T. baccata as the dominant species. The field layer is sparse, with the most common species being Brachypodium sylvaticum, Fragaria vesca, Oxalis acetosella, Potentilla sterilis and Sanicula europaea. There is a very dense moss cover, primarily Thamnobryum alopecurum and Thuidium tamariscinum (Kelly, 1981). A more detailed description of the distribution and density of various species in and around Reenadinna Wood can be found in Perrin (2002). T. baccata trees will generally grow to between 20 and 28 m in height (Thomas and Polwart, 2003). However, the stand in the Killarney woodlands has a lower canopy of between 6 and 14 m. Thomas and Polwart (2003) suggest that this is as a result of a high-energy investment in defensive mechanisms, which increase the resistance of wood against fungi and insect attacks. Despite its poisonous properties T. baccata is susceptible to browsing and bark stripping by rabbits, hares, deer and domestic animals such as sheep and cattle (Watt, 1926; Kelly, 1981; Haeggström, 1990; Mitchell, 1998). Kelly (1975) elaborates upon this, describing T. baccata as one of the most grazing-sensitive species in the Killarney woodlands. The species is able to continue growth under severe browsing pressure (Tittensor, 1980). However, it can be killed through over-scouring by deer antlers, as was observed by Kelly (1975) in the Killarney woodlands. In response to this, part of Reenadinna Wood was enclosed in 1969 in an effort to protect the species from deer browsing (Perrin, 2002).
Materials and methods 33 healthy, well-established mature trees with no obvious injury or disease were selected from the T. baccata stand in Reenadinna Wood. In order to ensure the sample was representative, trees were selected from both within and outside the deer-proof fencing that was erected in 1969. A systematic comparison of ring widths from trees inside and outside of the deer-proof fence was conducted to determine the influence of the fence on growth trends. The cores were extracted between June and October 2007. Two cores were extracted from each tree to obtain a long-term chronology for the species. The cores were collected using a two-thread Haglöf increment borer. The extremely rigid nature of the T. baccata species resulted in two increment borers breaking just below the threads. As a result, it was necessary to coat the borers in a layer of a combination of beeswax and a wax-based wood polish, ensuring a smoother extraction of core samples. On a number of occasions a Bosch cordless drill with an adapted chuck was used to insert the increment borer a number of millimetres into the tree as
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
3
Fig. 1. (a) Location of T. baccata stands in Ireland and detailed map of study site (adapted from: Office of Public Works, 1990); (b) climatic diagram for Killarney area. Mean monthly temperature and precipitation values were calculated for the available period (1969–2007) from Muckross House synoptic station, approximately 1 km from the study site.
the rigid nature of the species made it difficult to breach the surface. Once a grip was achieved, the remainder of the core extraction process was undertaken manually. Standard procedures were used in preparing tree-ring cores for analysis (Stokes and Smiley, 1968; Fritts, 1976). Verification of cross-dating of trees was performed using COFECHA (Holmes, 1986). Cross-dating for long-term chronologies was completed
using a 32-year cubic smoothed spline with a 50% wavelength cut-off for filtering. Series were examined in 50-year segments with a 25-year lag, and inter-correlation was determined using a 99% significance level Pearson correlation coefficient. No missing rings were identified. Tree-ring widths normally contain considerable amounts of non-climatic signals that may include a biological growth trend, tree-disturbance signals, or both (Fritts, 1976), but
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model
ARTICLE IN PRESS
DENDRO-25278; No. of Pages 9
S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
4
Table 1 Correlations between seasonal temperature and precipitation values from Muckross House synoptic station and Valentia Observatory for the common time period (1969–2007). All values are significant at the 0.01 level. Season
Temperature
Precipitation
Winter Spring Summer Autumn
0.970 0.917 0.923 0.951
0.773 0.607 0.653 0.605
for this study only four poorly preserved cores from two trees were discarded. Detrending of growth variations associated with age was accomplished using ARSTAN software (Holmes, 1992). Each series was detrended using a negative exponential curve, typical of the biological growth trend in trees (Fritts, 1976; Cook et al., 1990). If this line did not provide a good fit, a linear regression line of negative slope or a horizontal line through the mean was used to detrend the ring-width series. The biological persistence was removed from the standard chronology with autoregressive modelling to produce a residual chronology (Grissino-Mayer et al., 1992; Gervais, 2006). The standard, residual and ARSTAN chronologies were correlated with monthly values of mean temperature and total precipitation from Muckross House synoptic station, located approximately 1 km from the study site at 30 m a.s.l. (Fig. 1a). This Met Éireann (the Irish national meteorological service) synoptic station is the only one in the area and provided monthly data from January 1969 onwards. In order to extend the timeframe of the climate-tree growth comparison, temperature and precipitation data from Valentia Observatory (∼50 km from the study site at 25 m a.s.l.), for the years 1873–2007 and 1941–2007 respectively, were also examined. Correlations between seasonal temperature and precipitation values for both locations can be seen in Table 1. Simple correlation coefficients were calculated for the interval of overlap between each tree-ring chronology and the monthly instrumental data for the 21-month interval beginning in April of the previous year and ending with December of the current year (Blasing et al., 1981). Here, the standard and residual indices proved more responsive than the ARSTAN index. The resulting correlations were useful in identifying particular months where precipitation and temperature were highly correlated with radial growth of T. baccata in Killarney National Park. With the recording of weather data in Muckross House synoptic station, Killarney, only commencing in 1969, it was decided to split the longer-term Valencia observatory data into pre- and post-1969 segments in order to more accurately determine if climate-growth relationships remained stable through time. The correlations achieved enabled the formation of equations that included standard and residual chronology values for the current year in addition to those lagged to one year. From these, temperature and precipitation values for the Killarney area were reconstructed and, in turn, compared to recorded values. In an effort to investigate species interrelationships, correlations were run between the T. baccata chronology and Pilcher’s (1978) Q. petraea series from Killarney National Park, the only other locally available chronology. Both series were correlated with Hurrell et al.’s (2003) principal component winter (December–March) North Atlantic Oscillation (NAO) index, which runs from 1899 onwards. The NAO is regarded as one of the primary influences upon climatological variability in Western and Northwestern Europe, and the index presents the optimal representation of spatial patterns associated with the phenomenon (Hurrell, 1995).
Table 2 Reenadinna Wood T. baccata standard and residual chronology statistics. Chronology statistic
Standard chronology
Residual chronology
Total # of trees Total # of series Oldest sampled tree (yrs.) Full chronology interval Full chronology interval (SSS ≥0.75) # of trees/radii to reach SSS threshold Mean sensitivity Standard deviation First order autocorrelation Correlation among all radii Correlation between trees Correlation within trees
31 58 244 1763–2007 1805–2007 6 0.161 0.214 0.545 0.226 0.220 0.455
31 58 243 1764–2007 1803–2007 5 0.170 0.149 −0.012 0.296 0.293 0.408
Results The construction of the deer-proof fence in 1969 appears to have had little to no impact upon growth trends of the sampled trees. The series intercorrelation of the cores extracted from outside the fence was 0.266 for the entire length of the series, and 0.299 from 1969 to 2007. The cores from within the area protected from deer browsing returned an overall series intercorrelation of 0.338, with 0.276 for the years 1969–2007. The standard and residual ring-width indices from inside and outside of the fence also follow closely (Fig. 2), with a correlation of 0.730 and 0.807, respectively, for the post1969 period. The absence of any notable growth change is likely due to the maturity of the sampled trees when the deer enclosure was established, an idea supported by Perrin et al.’s (2006) finding that the impacts of deer browsing in Killarney are restricted to saplings. This analysis showed that the mature T. baccata in Reenadinna Wood did not respond significantly to the construction of the deer-proof fence and, consequently, all tree-ring width series were pooled to produce the final chronologies. The oldest core in the standard chronology dated to the year 1763 (Table 2). However, the Subsample Signal Strength (SSS) – the increased uncertainty of a chronology when the number of its constituent core series drops in early periods (Wigley et al., 1984) – of the standard chronology drops below 0.75 in 1804. Therefore, the final chronology runs from 1805 to 2007, and is plotted in Fig. 3 along with the residual series. The mean sensitivity of the standard chronology, a measure of the potential impact of high frequency climate variability on the growth response (Fritts, 1976), is 0.161, suggesting that it can potentially be employed in climate reconstructions (Speer, 2010). The standard deviation is 0.214, with the first order autocorrelation of the standard T. baccata index in Killarney National Park at 0.545. The SSS of the residual chronology falls below 0.75 in 1802, with the chronology running from 1803 to 2007. The correlations between the standard ring-width and the mean monthly temperatures for the entire length of the Muckross House synoptic station records (1970–2007) and the same time period from the Valentia Observatory records can be seen in Tables 3a and 3b, respectively. For the Muckross House data, the months with the most significant influence on standard T. baccata ring-width are November of the previous year to April of the growth year. Combining the data for all six months gave an r-value of 0.605 (p < 0.01) and a Durbin–Watson value of 1.546. Although not particularly strong, this Durbin–Watson result does indicate some serial correlation. For the Valentia Observatory data, the months November to May (1970–2007) display the strongest influence upon T. baccata growth, producing an r-value of 0.584 (p < 0.01) and a Durbin–Watson result of 1.565. However, for the sake of consistency, it was decided to utilise the data from November to April for
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
5
Fig. 2. (a) Standard T. baccata ring-width chronologies from cores sourced from within and outside of the deer-proof fence; (b) residual T. baccata ring-width chronologies from cores sourced from within and outside of the deer-proof fence.
comparative analysis. Correlation between the standard T. baccata index and the mean temperatures for these six months at Valentia Observatory for the time period 1970–2007 produced an r-value of 0.567 (p < 0.01) and a Durbin–Watson value of 1.456. The influence of November–April temperatures on T. baccata ring widths is likely a measure of length of growing season, with
a warmer winter–spring triggering an earlier start of the growing season and resulting in a larger ring for that year. The tree-ring records may be more responsive to post-1970 temperature fluctuations due to a general warming in seasonal temperatures during this time. The Muckross House and Valentia Observatory climate records show an approximately 1 ◦ C increase in winter and spring
Fig. 3. (a) Standard T. baccata ring-width chronology for Reenadinna Wood, including 11-year running mean, number of cores and the 0.75 Subsample Signal Strength (SSS) threshold; (b) residual T. baccata chronology.
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model
ARTICLE IN PRESS
DENDRO-25278; No. of Pages 9
S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
6
Table 3a Correlations between mean monthly temperature values from Muckross House (1970–2007) and Valentia Observatory (1872–2007) and standard T. baccata treering index values, as well as total monthly precipitation values for Muckross House (1970–2007) and Valentia Observatory (1941–2007) and residual T. baccata treering index values (the numbers in bold signify the months selected for the purposes of reconstructions). Month
Apr TY-1 May TY-1 Jun TY-1 Jul TY-1 Aug TY-1 Sep TY-1 Oct TY-1 Nov TY-1 Dec TY-1 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov
Muckross House
Table 3b Correlations between standard T. baccata tree-ring index values and mean monthly temperature values in Valentia Observatory for the time periods 1872–1969 and 1970–2007, as well as residual T. baccata tree-ring index values and total monthly precipitation values for 1941–1969 and 1970–2007 (the numbers in bold signify the months selected for the purposes of reconstructions). Month
Valentia Observatory
Temperature
Precipitation
Temperature
Precipitation
0.122 0.530** 0.121 0.157 0.026 0.277 0.107 0.436** 0.056 0.390* 0.398* 0.551** 0.365* 0.308 0.068 −0.162 0.076 0.176 −0.024 0.165
0.151 −0.163 −0.040 0.021 0.045 −0.011 0.047 0.020 −0.209 0.016 0.219 0.068 −0.114 0.240** 0.277** 0.229 0.252 −0.225 −0.256 0.042
0.035 0.071 0.029 0.160 0.136 0.000 0.071 0.051 0.141 0.281** 0.218* 0.135 0.127 0.047 −0.044 0.004 0.071 0.095 −0.002 0.149
0.196 −0.175 −0.274* −0.038 0.061 0.002 −0.038 −0.099 −0.073 −0.005 0.126 0.019 −0.238 0.331** 0.455** 0.134 0.197 0.014 −0.125 0.208
Regression equation employed in temperature reconstruction (Muckross data): • (Nov-Apr Temp) = 7.608 + ((2.476 * TB-std) − (0.179 * TB-std − 1)), where Nov-Apr Temp = mean temperature values for November to April and TB-std = standardised T. baccata chronology. Regression equation employed in precipitation reconstruction (Muckross data): • (May-Jun Precip) = −269.363 + ((253.503 * TB-res) + (200.776 * TB-res − 1)), where May-Jun Precip = total precipitation values for May to June and TB-res = residual T. baccata chronology. Regression equation employed in precipitation reconstruction (Valentia data): • (May-Jun Precip) = −122.748 + ((56.714 * TB-res) + (237.316 * TB-res − 1)), where May-Jun Precip = total precipitation values for May to June and TB-res = residual T. baccata chronology. * Correlation is significant at the 0.05 level. ** Correlation is significant at the 0.01 level.
temperatures since 1980, a trend which is seen throughout Ireland (McKeown et al., 2012). The last 30 years thus account for the warmest winter–spring temperatures on record (McKeown et al., 2012). The warmer seasonal weather may be increasing its potential influence on tree growth due to a lengthened growing season, with winters prior to the 1980s largely being too cold for a small change in temperature to significantly influence tree growth. The warming trend experienced since the 1980s coincides with an increase in ring widths over the same time period (Fig. 3). The correlations between the Valentia Observatory temperature data and the standard ring-width chronology in Table 3b emphasise this point further. Concentrating on the pre-1970 time period highlights the instability of the T. baccata response to temperature. Correlation with the standard tree-ring index and mean November–April temperatures for the years 1872–1969 returned an r-value of 0.157, and a Durbin–Watson value of 1.936. Meanwhile, extending this analysis to the entire length of the Valentia Observatory dataset (1872–2007) produced an r-value of 0.262, and a Durbin–Watson value of 1.063. This shows that, although significant correlations have been achieved for the more recent decades, the temperature signal in T. baccata in Reenadinna Wood is not time stable. The significant monthly correlations facilitated the reconstruction of mean November to April temperatures for Muckross House synoptic weather station and Valentia Observatory for the years 1970–2007, the equations for which are also in Tables 3a and 3b, respectively. The observed and reconstructed temperature data
Apr TY-1 May TY-1 Jun TY-1 Jul TY-1 Aug TY-1 Sep TY-1 Oct TY-1 Nov TY-1 Dec TY-1 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov
Temperature
Precipitation
1872–1969
1970–2007
1941–1969
1970–2007
−0.024 −0.150 0.075 0.100 0.131 −0.149 −0.042 −0.065 0.098 0.245* 0.135 −0.021 −0.010 −0.132 −0.083 −0.060 0.026 0.058 −0.074 0.091
0.117 0.515** −0.011 0.118 0.035 0.144 0.115 0.386* 0.057 0.361* 0.401* 0.497** 0.353* 0.374* 0.068 −0.086 0.023 0.019 −0.047 0.167
0.117 0.515** −0.011 0.118 0.035 0.144 0.115 0.386* 0.057 0.361* 0.401* 0.497** 0.353* 0.374* 0.068 −0.086 0.023 0.019 −0.047 0.167
0.099 −0.041 −.436* −0.110 −0.056 −0.073 −0.053 −0.406* −0.044 0.091 −0.041 −0.007 −0.395* 0.383* 0.487** 0.102 0.346 0.179 −0.189 −0.042
Regression equation employed in temperature reconstruction (1872–1969): (Nov-Apr Temp) = 7.027 + ((−0.121 * TB-std) − (0.814 * TB-std − 1)), where • Nov-Apr Temp = mean temperature values for November to April and TBstd = standardised T. baccata chronology. Regression equation employed in temperature reconstruction (1970–2007): • (Nov-Apr Temp) = 5.725 + ((1.895 * TB-std) − (0.190 * TB-std − 1)), where Nov-Apr Temp = mean temperature values for November to April and TB-std = standardised T. baccata chronology. * Correlation is significant at the 0.05 level. ** Correlation is significant at the 0.01 level.
for both Muckross and Valentia follow similar peaks and troughs, with a general increase over time (Fig. 4). However, the models do tend to underestimate the fluctuations, a common characteristic of tree-ring models as the trees’ growth responses cannot cross certain thresholds (Esper and Frank, 2009). The reconstructed temperatures created utilising the Valentia Observatory data for the years 1872–1969, the equation for which can be seen in Table 3b, are, as can be expected, grossly underestimating the year-to-year fluctuations (Fig. 4). The low r-value of 0.157 ensures that the reconstruction’s ability to reflect change is weak. As with the temperature data, precipitation records from Valentia Observatory were also split into two time periods in order to allow ease of comparison with the Muckross data (Tables 3a and 3b). Table 3a shows that, for the Muckross House data (1970–2007), significant correlation values were achieved in May and June of the current year, with r-values of 0.240 (p < 0.01) and 0.277 (p < 0.01), respectively. Analysing the same time period for the Valentia Observatory data produced fewer significant correlations, with an r-value of 0.487 (p < 0.01) returned for June of the current year. However, extending the time period of examination to the entire length of the Valentia dataset (1941–2007) did return more significant correlations (Table 3a), with r-values of 0.331 (p < 0.01) and 0.455 (p < 0.01) for May and June, respectively. The pre-1970 data (Table 3b) also exhibited significant relationships to precipitation, but over a broader range of months. Because precipitation can vary so greatly from one year to the next, it was necessary to use the residual chronology as opposed to the standard chronology for these correlations as the former factors out autocorrelation in its formulation process (Meko et al., 1993). The correlation between total May–June rainfall
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
7
Fig. 4. Reconstructed and observed mean November–April temperatures in Killarney using Muckross House synoptic station data (1970–2007), and Valentia Observatory data (1872–1969 and 1970–2007).
Fig. 5. Reconstructed and observed total May–June precipitation values for Killarney, using Muckross House synoptic station data (1970–2007) and Valentia Observatory data (1941–2007).
(1970–2007) and residual T. baccata ring-widths produced an rvalue of 0.521 (p < 0.01) and Durbin–Watson value of 2.062 for the Muckross House synoptic station data, while a correlation of 0.542 (p < 0.01) and Durbin–Watson result of 2.086 was returned for Valentia Observatory (1941–2007) (Table 3a). The Durbin–Watson results, where the autocorrelation of the residuals from the regression is measured, indicate that there is little serial correlation; this is a consequence of the fact that precipitation values for one year have no influence on those of the following year. The positive relationship with summer precipitation (in a region where moisture is usually not a limiting growth factor) is due to the fact that the study site, similar to all T. baccata stands in Ireland, consists of thin, well-drained soils overlying limestone (Perrin, 2002). The equations used to reconstruct total May–June precipitation from Muckross House data (1970–2007) and Valentia Observatory data (1941–2007) can be seen in Table 3a. The reconstructed and observed values are shown in Fig. 5. Both trace similar peaks and troughs but, as with the reconstructed temperatures, the model has a propensity to undervalue the intensity of the changes. The correlation coefficient between Pilcher’s (1978) Q. petraea series from Killarney National Park and the T. baccata series is 0.207(p < 0.01) for the years the chronologies overlapped (1809–1978). The low correlation value highlights the fact that the two species react differently to similar climatic conditions, a
point emphasised further still when both series are correlated with Hurrell et al.’s (2003) winter North Atlantic Oscillation index. In a positive NAO phase, southwest Ireland should exhibit warmer and wetter conditions. While the T. baccata series shows a positive relationship with NAO (0.303, p < 0.01), the Q. petraea series exhibits a negative relationship (−0.330, p < 0.01). All three indices are presented in Fig. 6. This can be explained by species site preference. Whereas both taxa tend to thrive in warmer conditions (Gardiner, 1974; Perrin, 2002), they differ in their responses to precipitation. The thin, well-drained soils where T. baccata stands flourish are likely to promote a positive response to wetter periods associated with positive phases of NAO. Conversely, Q. petraea prefers deep, acidic soil (Lévy et al., 1992) and, as a result, growth is negatively impacted upon by water logging (Parelle et al., 2007) during positive phases of NAO. The different responses to NAO highlight the usefulness of multi-species approaches to climate reconstruction.
Discussion and conclusions For decades, tree-ring studies in Ireland had been dominated by the use of Quercus species, with García-Suárez et al. (2009) being the first to undertake a formal examination of the dendroclimatological potential of other tree species on the island. Their results
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
8
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
Fig. 6. Standard chronology indices for T. baccata (1899–2007) and Pilcher’s (1978) Q. petraea (1899–1978) with Hurrell et al.’s (2003) principal component winter NAO index. Correlation coefficient between T. baccata and NAO: 0.303 (p < 0.01); Q. petraea and NAO: −0.330 (p < 0.01).
showed that F. sylvatica, F. excelsior and P. sylvestris can be used in addition to Quercus species as part of a multi-species dendroclimatological analysis. This study shows that an additional species, T. baccata, is a reliable source of data that can be utilised in the investigation of past environmental conditions. The results garnered as part of this study show a similar strength of response to climate when compared to other Irish species (Pilcher and Baillie, 1980; García-Suárez et al., 2009) and a stronger climate response than other studies of T. baccata internationally (Moir, 1999; Yadav and Singh, 2002). T. baccata is a long-lived species, with potential to survive more than 1000 years (Thomas and Polwart, 2003). O’Connell (1987) pointed out that the natural expansion of peatlands in various parts of Ireland throughout history led to the preservation of many Quercus and T. baccata trees. However, dendrochronological investigations in Ireland are restricted by the spatial and temporal limitations of the dominant Quercus chronologies. This is despite the fact that numerous examples of well-preserved and long-lived T. baccata trees are found throughout the country. The potential for expansion of the T. baccata chronology, as well as the potential for inter-species correlation with other species (Moir, 1999; Yadav and Singh, 2002), indicate that T. baccata can become an important constituent in multi-site and multi-species studies of past environments. The facts that only two T. baccata chronologies are available from the International Tree-ring Data Bank, and only a small number of studies have examined the species’ dendroclimatological applicability (Moir, 1999; Yadav and Singh, 2002), suggest that the potential benefits of T. baccata are not necessarily recognised. This may be due to real and perceived difficulties in attaining useable cores from the species, with missing rings also being a potential drawback (Thomas and Polwart, 2003). However, such issues did not arise during the course of this study. Here it has been shown that the coring and cross-dating of T. baccata can yield fruitful results even when faced with issues such as animal browsing. In fact, it is safe to conclude that the impact of deer on older, well-established, T. baccata in Reenadinna Wood is minimal. Thus, any assumptions that browsers could have compromised the temperature and precipitation models created are refutable. The long-lived nature of the T. baccata species presents the potential to expand the chronology further than the 204 years produced in the course of this research. The chronology could also be expanded spatially. Although Reenadinna Wood in Killarney National Park is the most extensive T. baccata stand in Ireland, other stands were identified by Perrin (2002) in limestone-dominated areas in The Burren, Co. Clare; Garryland Wood, Co. Galway; St. John’s Wood in Co. Roscommon; and The Rocks of Clorhane, Co. Offaly (Fig. 1a). Individual examples can also be found in estates
and graveyards throughout the country (Perrin, 2002). A multisite approach to extending the T. baccata chronology could lead to a longer series, in turn facilitating an extension and expansion of dendrochronological and dendroclimatological studies in Ireland. Such an extension of the T. baccata chronology would also strengthen multi-species approaches to the examination of past environmental change. Significant inter-species correlations indicate the potential for T. baccata to assume an important role in the study of regionally specific environmental change and the driving factors involved, a key feature in Frank and Esper’s (2005) call for progression in tree-ring research. The multi-species study undertaken by García-Suárez et al. (2009) highlighted the benefits of this approach in Ireland. This study’s opposing, but equally significant, responses of T. baccata and Q. petraea to the NAO strengthens the argument that differential signals are key to dendroclimatological reconstructions in regions where species are not at their ecological limits. The incorporation of T. baccata would thus add to the strength of any mixed-species climate signals in the region. This study has shown the strong potential for T. baccata in creating a long-term tree-ring series, and highlights the possibility of growing this chronology through space and time.
Acknowledgements The authors wish to thank Prof. Chris Caseldine and Prof. Micheál Ó’ Cinnéide for their helpful comments on content and form, and the anonymous reviewers who provided very constructive feedback. We also wish to acknowledge funding received from the Coimbra Group during the course of this research, as well as the permission granted by the National Parks and Wildlife Service to undertake the study in Reenadinna Wood.
References Bernabei, M., Gjerdrum, P., 2006. Tree-ring data for yew (Taxus baccata) from the Alps. NOOA Satellite and Information Service, http://www.ncdc.noaa.gov/paleo/ metadata/noaa-tree-2733.html (accessed 20.01.13). Biondi, F., 1992. Development of a tree-ring network for the Italian Peninsula. TreeRing Bull. 52, 5–29. Blasing, T.J., Duvick, D.N., West, D.C., 1981. Dendroclimatic calibration and verification using regionally averaged and single station precipitation data. Tree-Ring Bull. 41, 37–43. Cook, E.R., Briffa, K.R., Shiyatov, S., Mazepa, V., 1990. Tree-ring standardisation and growth-trend estimation. In: Cook, E., Kairiukstis, A. (Eds.), Methods of Dendrochronology: Applications in the Environmental Sciences. Kluwer Academic Publ., Dordrecht, pp. 104–123. Frank, D.C., Esper, J., 2005. Temperature reconstructions and comparisons with instrumental data from a tree-ring network for the European Alps. Int. J. Climatol. 25, 1437–1454.
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
Esper, J., Frank, D.C., 2009. Divergence pitfalls in tree-ring research. Clim. Change 94, 261–266. Fritts, H.C., 1976. Tree Rings and Climate. The Blackburn Press, New Jersey, pp. 567. García-Suárez, A.M., Butler, C.J., Baillie, M.G.L., 2009. Climate signal in tree-ring chronologies in a temperate climate: a multi-species approach. Dendrochronologia 27, 183–198. Gardiner, A.S., 1974. A history of the taxonomy and the distribution of the native oak species. In: Morris, M.G., Perring, F.H. (Eds.), The British Oak. The Botanical Society of the British Isles, Cambridge, pp. 3–26. Gervais, B.R., 2006. A three-century record of precipitation and blue oak recruitment from the Tehachapi Mountains, Southern California, USA. Dendrochronologia 24, 29–37. Godwin, H., 1975. The History of the British Flora. Cambridge University Press, Cambridge, pp. 541–549. Grissino-Mayer, H., Holmes, R.L., Fritts, H.C., 1992. User’s Manual for the International Tree-ring Data Bank Program Library. Laboratory of Tree-Ring Research, University of Arizona, Tucson, 106 pp. Haeggström, C.A., 1990. The influence of sheep and cattle grazing on wooded meadows in Åland, SW Finland. Acta Bot. Fenn. 141, 1–28. Holmes, R.L., 1986. Quality control of cross-dating and measuring: a users manual for program COFECHA. In: Holmes, R.L., Adams, R.K., Fritts, H.C. (Eds.), TreeRing Chronologies of Western North America: California, Eastern Oregon and Northern Great Basin. University of Arizona Press, Tucson, pp. 41–49. Holmes, R.L., 1992. Dendrochronology Program Library, Version 1992-1. Laboratory of Tree-ring Research, University of Arizona Press, Tucson. Hurrell, J.W., 1995. Decadal trends in the north Atlantic oscillation: regional temperatures and precipitation. Science 269, 676–679. Hurrell, J.W., 2003. In: National Center for Atmospheric Research Staff (Ed.), The Climate Data Guide: Hurrell North Atlantic Oscillation (NAO) Index (PC-based). , http://climatedata guide.ucar.edu/guidance/hurrell-north-atlantic-oscillationnao-index-pc-based (accessed 12.01.13). Jalas, J., Suominen, J. (Eds.), 1973. Atlas Florae Europea. Distribution of vascular plants in Europe. 2. Gymnospermae (Pinaceae to Ephedraceae). The Committee for Mapping the Flora of Europe and Societas Biologica Fennica Vanamo, Helsinki, 40 pp. Kelly, D.L., 1981. The native forest vegetation of Killarney, south-west Ireland – an ecological account. J. Ecol. 69, 437–472. Kelly, D.L., (Dissertation) 1975. Native woodland in western Ireland with especial reference to the region of Killarney. University of Dublin, pp. 337. Kuniholm, P.I., Steele, L., Tovar, I., 2006. Tree-ring data for yew (T. baccata) from Georgia, http://hurricane.ncdc.noaa.gov/pls/paleox/f?p=519:1: 2633893827028787::::P1 STUDY ID:3799 (accessed 25.01.13). Lévy, G., Becker, M., Duhamel, D., 1992. A comparison of the ecology of pedunculate and sessile oaks: radial growth in the centre and northwest of France. For. Ecol. Manag. 55, 51–63. Lowe, J., 1897. The Yew Trees of Great Britain and Ireland. MacMillan, London, pp. 342. McKeown, M., Potito, A., Hickey, K., 2012. The long-term temperature record from Markree Observatory, County Sligo, from 1842 to 2011. Irish Geogr. 45, 257–282. Meko, D., Cook, E.R., Stahle, D.W., Stockton, C.W., Hughes, M.K., 1993. Spatial patterns of tree-growth anomalies in the United States and southeastern Canada. J. Climatol. 6, 1773–1786.
9
Mitchell, A.K., 1998. Acclimation of Pacific yew (Taxus brevifolia) foliage to sun and shade. Tree Physiol. 18, 479–757. Mitchell, F.J., 1990. The history and vegetation dynamics of a yew wood (Taxus baccata) in S.W. Ireland. New Phytologist 115, 573–577. Moir, A.K., 1999. The dendrochronological potential of modern yew (Taxus baccata) with special reference to yew from Hampton Court Palace, UK. New Phytologist 144, 479–488. O’Connell, C. (Ed.), 1987. The IPCC Guide to Irish Peatlands. Irish Peatland Conservation Council, Dublin, p. 102. Office of Public Works, 1990. Killarney National Park Management Plan. The Stationery Office, Dublin, pp. 72. Parelle, J., Brendel, O., Jolivet, Y., Dreyer, E., 2007. Intra- and interspecific diversity in the response to waterlogging of two co-occurring white oak species (Quercus robur and Q. petraea). Tree Physiol. 27, 1027–1034. ˜ Penalba, M.C., 1994. The history of the Holocene vegetation in northern Spain from pollen analysis. J. Ecol. 82, 815–832. Perrin, P., (Dissertation) 2002. The Ecology of Yew (Taxus baccata) in Ireland. Trinity College Dublin, pp. 160. Perrin, P., Kelly, D.L., Mitchell, F., 2006. Long-term deer exclusion in yew-wood and oak-wood habitats in southwest Ireland: natural regeneration and stand dynamics. For. Ecol. Manag. 236, 356–367. Pilcher, J.R., 1978. Tree-ring data for Q. petraea from Killarney, Available at http://www.ncdc.noaa.gov/paleo/metadata/noaa-tree-4220.html (accessed 15.02.13). Pilcher, J.R., 1994. Biological evidence of past environments and climatic changes. Eur. Rev. 2, 271–278. Pilcher, J.R., Baillie, M.G.L., 1980. Six modern oak chronologies from Ireland. TreeRing Bull. 40, 23–34. Ratcliffe, D.A., 1977. A Nature Conservation Review. Cambridge University Press, Cambridge, pp. 401. Speer, J.H., 2010. Fundamentals of Tree-ring Research. University of Arizona Press, Arizona, pp. 333. Stokes, M.A., Smiley, T.L., 1968. An Introduction to Tree-ring Dating. University of Chicago Press, Chicago, pp. 73. Tabbush, P., White, J., 1996. Estimation of tree age in ancient yew woodland at Kingley Vale. Q. J. Forest. 90, 197–206. Thomas, P.A., Polwart, A., 2003. Taxus baccata L. J. Ecol. 91, 489–524. Tittensor, R.M., 1980. Ecological history of yew Taxus baccata L. in southern England. Biol. Conserv. 17, 243–265. ˇ ´ M., 1991. Conifers, Morphology and Variation (M. Soljan, Vidakovic, Trans.). Gradfiˇcki Zavod Hrvatske, Zagreb, 754 pp. Voliotis, D., 1986. Historical and environmental significance of the yew (Taxus baccata L.). Israel Journal of Botany. Basic Appl. Plant Sci. 35, 47–52. Watt, A.S., 1926. Yew communities of the South Downs. J. Ecol. 14, 282–316. Wigley, T.M.L., Briffa, K.R., Jones, P.D., 1984. On the average value of correlated time series, with applications in dendroclimatology and hydrometeorology. J. Climate Appl. Meteorol. 23, 201–213. Yadav, R.R., Singh, J., 2002. Tree-ring analysis of Taxus baccata from the western Himalaya, India, and its dendroclimatic potential. Tree-ring Res. 58, 23–29.
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
ARTICLE IN PRESS Dendrochronologia xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Dendrochronologia journal homepage: www.elsevier.com/locate/dendro
ORIGINAL ARTICLE
Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland Stephen Galvin a,∗ , Aaron Potito a , Kieran Hickey b a b
Palaeoenvironmental Research Unit, School of Geography and Archaeology, National University of Ireland Galway, Galway, Ireland School of Geography and Archaeology, National University of Ireland Galway, Galway, Ireland
a r t i c l e
Article history: Received 28 May 2013 Accepted 29 March 2014 Keywords: Dendrochronology Dendroclimatology Taxus baccata Ireland
a b s t r a c t
i n f o
Tree-ring research in Ireland has typically been dominated by Quercus species, particularly Quercus petraea and Quercus robur. Recent years have seen a greater focus on multi-species reconstructions in Ireland but, due to difficulties with the hardness of the wood, missing/pinched rings and fused stems, Taxus baccata has not been included in these investigations. Despite these difficulties, a 31-tree, 204-year T. baccata chronology was successfully constructed from Killarney National Park, southwest Ireland. The chronology exhibits promising dendroclimatological potential, with climatic responsiveness equivalent to that of the other major Irish tree taxa, including Quercus. The chronology shows the strongest relationship with May–June precipitation from Muckross House synoptic station (1970–2007; r = 0.521, p < 0.01) and Valentia Observatory (1941–2007; r = 0.545, p < 0.01). November–April temperatures also exhibited a strong relationship with the chronology post-1970 (r = 0.605, p < 0.01 for Muckross House, r = 0.567, p < 0.01 for Valentia Observatory), but this relationship is not time stable and breaks down for the pre-1970 Valentia Observatory record. The long-lived nature of T. baccata, the exceptional preservation of wood and rings in this hard softwood species, as well as its prominence in Irish archaeology, all point to the potential to expand this chronology both spatially and temporally, and demonstrate T. baccata’s potential in multi-site and multi-species tree-ring studies in the region. © 2014 Elsevier GmbH. All rights reserved.
Introduction Until recently, tree-ring research in Ireland has been dominated by Quercus species, particularly Quercus petraea and Quercus robur, as these taxa were common over large areas of Irish lowlands for the last 8000 years (Gardiner, 1974). Pilcher and Baillie (1980) used six Irish Quercus chronologies to examine the climate responses of the species. Their study showed that trees in Ireland and Britain are less sensitive to changes in climate when compared to those closer to their latitudinal or altitudinal limits, which had implications for the perceived dendroclimatological potential of the region. More recently, Irish tree-ring studies have been expanded to incorporate other species. García-Suárez et al. (2009) used both Q. robur and Q. petraea, as well as Fagus sylvatica, Fraxinus excelsior and Pinus sylvestris, to reconstruct temperature, rainfall, sunshine hours and Palmer Drought Severity Index values in Northern Ireland. Their approach highlights the potential benefits of using combinations
∗ Corresponding author. Tel.: +353 091 494103; fax: +353 091 495505. E-mail address: [email protected] (S. Galvin).
of species to reconstruct environmental variables, as each species will respond slightly differently to annual or seasonal changes in temperature and precipitation. This is particularly important in regions where no single climate variable leads to notably wide or narrow tree-rings, as is the case in Ireland (Pilcher, 1994). This study focuses on exploring the dendrochronological and dendroclimatological potential of Taxus baccata as an alternative source of annual tree-ring data. T. baccata has traditionally been overlooked in the Irish record due to difficulties with the hardness of the wood, missing/pinched rings, unsymmetrical trunks and fused stems (Tabbush and White, 1996; Thomas and Polwart, 2003). Therefore, this is the first study to explore the dendrochronological or dendroclimatological potential of T. baccata in Ireland, and one of only a small number of such studies of T. baccata on a global scale. T. baccata has a wide geographic range throughout Europe (Jalas and Suominen, 1973), with prominent stands in Ireland and Britain (Ratcliffe, 1977), extending northwards to Norway and Sweden ´ 1991), eastwards to Estonia, Poland and (c. 63◦ N) (Vidakovic, Turkey, and southwards to Greece (Voliotis, 1986), northern Spain ˜ ´ 1991). In the 1994), Portugal and into Algeria (Vidakovic, (Penalba, south of its European range T. baccata is largely a montane tree,
http://dx.doi.org/10.1016/j.dendro.2014.03.004 1125-7865/© 2014 Elsevier GmbH. All rights reserved.
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
2
whereas it grows from sea level to 425 m above sea level (a.s.l.) in England and Wales and up to 470 m a.s.l. on Purple Mountain in County Kerry, southwest Ireland (Moir, 1999). The species is well suited to a mild oceanic climate, avoids areas susceptible to severe winter frosts (Godwin, 1975), and is one of the longest-living plant species native to Ireland. Historically, T. baccata has played an important role in Irish society. Its density and durability saw it employed in construction of hardwearing items such as handles, wheel dowels, and weapon shafts. The fine grain and deep colouring of the species resulted in it also being used for finer items and tableware (Perrin, 2002). While modern T. baccata is demonstrated to be suitable for use in dendrochronology (Biondi, 1992; Tabbush and White, 1996), the need to measure a number of different radii from samples indicates that analysis from incrementally cored samples may prove challenging (Moir, 1999). T. baccata cores can often contain missing or very narrow rings (Thomas and Polwart, 2003), while the frequent wedging of rings often results in difficulties in cross-dating (Yadav and Singh, 2002). Other pitfalls in estimating age include old senescent trees producing new stems from the remains of the former large trunk and appearing as young trees, and the apparent production of large trees from the fusing of several trunks. Lowe (1897) argued that the fusing of multiple stems/roots would make an individual appear older than it actually is. However, overall growth rate of the new trunk when young is unlikely to be much greater than a single tree of the same size (Thomas and Polwart, 2003). Tabbush and White (1996) argue that the coalescing of seedlings would actually make little difference to age estimates. The potential difficulties involved in using T. baccata can be overcome by selecting appropriate trees to sample, i.e. those with definite single trunks or obvious separations at the base of the trunk. In addition, the extraction of two cores per tree greatly increases the likelihood of identifying missing rings. The long-lived nature of T. baccata makes it suitable for ecological and climate investigations while the hardness of the wood, although adding to the difficulties in extracting cores, increases the preservation potential of trees and rings. Complications in obtaining data from T. baccata have undoubtedly influenced the fact that only two chronology series are available from the International Tree-ring Data Bank. The chronology compiled by Bernabei and Gjerdrum (2006) in the Alps was formed from 17 cores, the oldest dating to 1896. The second T. baccata chronology, created by Kuniholm et al. (2006), was formed from 19 cores dating to 1526 in Georgia, Eurasia, and has since been utilised to cross-date other tree-ring records in the area for archaeological purposes. Publications examining T. baccata in terms of dendroclimatology are equally as rare. Moir (1999) sampled 14 T. baccata trees in Hampton Court Palace in London, noting that February–July precipitation had a positive effect on growth. January–February and October temperatures also influenced ringwidth, as did soil-moisture. Yadav and Singh (2002) sampled 18 T. baccata trees in the western Himalaya, where the 345-yearlong chronology achieved its strongest correlations with mean March–June temperatures (r = −0.310) for the period 1898–1998. Both Yadav and Singh’s (2002) and Moir’s (1999) studies showed significant inter-species correlations, indicating that T. baccata could become an important constituent in multi-site and multispecies studies.
Study site The geology and geomorphology of the Killarney area of County Kerry is diverse, something that gives rise to a wide range of forest vegetation (Kelly, 1981). Reenadinna Wood (Fig. 1a) (50◦ 01 N; 9◦ 31 W), located on a carboniferous limestone outcrop on
Muckross Peninsula, lies approximately 4.5 km south–southwest of Killarney town and is dominated by T. baccata. Mean winter and summer temperatures are 9.6 ◦ C and 18.4 ◦ C respectively, with mean total precipitation of 598 mm in winter and 264 mm in summer (Fig. 1b). The site, part of Killarney National Park, a UNESCO Biosphere Reserve, is 20 m a.s.l. It is bounded to the north by Lough Leane and by Muckross Lake to the south. The western limit of the wood lies along the geological boundary with Devonian Old Red Sandstone while, to the east, the wood is bounded by parkland where the limestone ceases to outcrop. Through pollen analysis, Mitchell (1990) showed that T. baccata has existed in the area for up to 5000 years. Reenadinna Wood (approx. 25 ha in extent) is the most extensive T. baccata woodland remaining in Ireland (Perrin, 2002). The shorter T. baccata trees on the more broken rocky terrain of the Killarney woodlands are intermingled with taller Quercus (mainly Q. petraea) in the intervening soil-filled hollows. Little grows under the T. baccata canopy, although Corylus avellana or Ilex aquifolium become frequent locally, with the former sometimes replacing T. baccata as the dominant species. The field layer is sparse, with the most common species being Brachypodium sylvaticum, Fragaria vesca, Oxalis acetosella, Potentilla sterilis and Sanicula europaea. There is a very dense moss cover, primarily Thamnobryum alopecurum and Thuidium tamariscinum (Kelly, 1981). A more detailed description of the distribution and density of various species in and around Reenadinna Wood can be found in Perrin (2002). T. baccata trees will generally grow to between 20 and 28 m in height (Thomas and Polwart, 2003). However, the stand in the Killarney woodlands has a lower canopy of between 6 and 14 m. Thomas and Polwart (2003) suggest that this is as a result of a high-energy investment in defensive mechanisms, which increase the resistance of wood against fungi and insect attacks. Despite its poisonous properties T. baccata is susceptible to browsing and bark stripping by rabbits, hares, deer and domestic animals such as sheep and cattle (Watt, 1926; Kelly, 1981; Haeggström, 1990; Mitchell, 1998). Kelly (1975) elaborates upon this, describing T. baccata as one of the most grazing-sensitive species in the Killarney woodlands. The species is able to continue growth under severe browsing pressure (Tittensor, 1980). However, it can be killed through over-scouring by deer antlers, as was observed by Kelly (1975) in the Killarney woodlands. In response to this, part of Reenadinna Wood was enclosed in 1969 in an effort to protect the species from deer browsing (Perrin, 2002).
Materials and methods 33 healthy, well-established mature trees with no obvious injury or disease were selected from the T. baccata stand in Reenadinna Wood. In order to ensure the sample was representative, trees were selected from both within and outside the deer-proof fencing that was erected in 1969. A systematic comparison of ring widths from trees inside and outside of the deer-proof fence was conducted to determine the influence of the fence on growth trends. The cores were extracted between June and October 2007. Two cores were extracted from each tree to obtain a long-term chronology for the species. The cores were collected using a two-thread Haglöf increment borer. The extremely rigid nature of the T. baccata species resulted in two increment borers breaking just below the threads. As a result, it was necessary to coat the borers in a layer of a combination of beeswax and a wax-based wood polish, ensuring a smoother extraction of core samples. On a number of occasions a Bosch cordless drill with an adapted chuck was used to insert the increment borer a number of millimetres into the tree as
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
3
Fig. 1. (a) Location of T. baccata stands in Ireland and detailed map of study site (adapted from: Office of Public Works, 1990); (b) climatic diagram for Killarney area. Mean monthly temperature and precipitation values were calculated for the available period (1969–2007) from Muckross House synoptic station, approximately 1 km from the study site.
the rigid nature of the species made it difficult to breach the surface. Once a grip was achieved, the remainder of the core extraction process was undertaken manually. Standard procedures were used in preparing tree-ring cores for analysis (Stokes and Smiley, 1968; Fritts, 1976). Verification of cross-dating of trees was performed using COFECHA (Holmes, 1986). Cross-dating for long-term chronologies was completed
using a 32-year cubic smoothed spline with a 50% wavelength cut-off for filtering. Series were examined in 50-year segments with a 25-year lag, and inter-correlation was determined using a 99% significance level Pearson correlation coefficient. No missing rings were identified. Tree-ring widths normally contain considerable amounts of non-climatic signals that may include a biological growth trend, tree-disturbance signals, or both (Fritts, 1976), but
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model
ARTICLE IN PRESS
DENDRO-25278; No. of Pages 9
S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
4
Table 1 Correlations between seasonal temperature and precipitation values from Muckross House synoptic station and Valentia Observatory for the common time period (1969–2007). All values are significant at the 0.01 level. Season
Temperature
Precipitation
Winter Spring Summer Autumn
0.970 0.917 0.923 0.951
0.773 0.607 0.653 0.605
for this study only four poorly preserved cores from two trees were discarded. Detrending of growth variations associated with age was accomplished using ARSTAN software (Holmes, 1992). Each series was detrended using a negative exponential curve, typical of the biological growth trend in trees (Fritts, 1976; Cook et al., 1990). If this line did not provide a good fit, a linear regression line of negative slope or a horizontal line through the mean was used to detrend the ring-width series. The biological persistence was removed from the standard chronology with autoregressive modelling to produce a residual chronology (Grissino-Mayer et al., 1992; Gervais, 2006). The standard, residual and ARSTAN chronologies were correlated with monthly values of mean temperature and total precipitation from Muckross House synoptic station, located approximately 1 km from the study site at 30 m a.s.l. (Fig. 1a). This Met Éireann (the Irish national meteorological service) synoptic station is the only one in the area and provided monthly data from January 1969 onwards. In order to extend the timeframe of the climate-tree growth comparison, temperature and precipitation data from Valentia Observatory (∼50 km from the study site at 25 m a.s.l.), for the years 1873–2007 and 1941–2007 respectively, were also examined. Correlations between seasonal temperature and precipitation values for both locations can be seen in Table 1. Simple correlation coefficients were calculated for the interval of overlap between each tree-ring chronology and the monthly instrumental data for the 21-month interval beginning in April of the previous year and ending with December of the current year (Blasing et al., 1981). Here, the standard and residual indices proved more responsive than the ARSTAN index. The resulting correlations were useful in identifying particular months where precipitation and temperature were highly correlated with radial growth of T. baccata in Killarney National Park. With the recording of weather data in Muckross House synoptic station, Killarney, only commencing in 1969, it was decided to split the longer-term Valencia observatory data into pre- and post-1969 segments in order to more accurately determine if climate-growth relationships remained stable through time. The correlations achieved enabled the formation of equations that included standard and residual chronology values for the current year in addition to those lagged to one year. From these, temperature and precipitation values for the Killarney area were reconstructed and, in turn, compared to recorded values. In an effort to investigate species interrelationships, correlations were run between the T. baccata chronology and Pilcher’s (1978) Q. petraea series from Killarney National Park, the only other locally available chronology. Both series were correlated with Hurrell et al.’s (2003) principal component winter (December–March) North Atlantic Oscillation (NAO) index, which runs from 1899 onwards. The NAO is regarded as one of the primary influences upon climatological variability in Western and Northwestern Europe, and the index presents the optimal representation of spatial patterns associated with the phenomenon (Hurrell, 1995).
Table 2 Reenadinna Wood T. baccata standard and residual chronology statistics. Chronology statistic
Standard chronology
Residual chronology
Total # of trees Total # of series Oldest sampled tree (yrs.) Full chronology interval Full chronology interval (SSS ≥0.75) # of trees/radii to reach SSS threshold Mean sensitivity Standard deviation First order autocorrelation Correlation among all radii Correlation between trees Correlation within trees
31 58 244 1763–2007 1805–2007 6 0.161 0.214 0.545 0.226 0.220 0.455
31 58 243 1764–2007 1803–2007 5 0.170 0.149 −0.012 0.296 0.293 0.408
Results The construction of the deer-proof fence in 1969 appears to have had little to no impact upon growth trends of the sampled trees. The series intercorrelation of the cores extracted from outside the fence was 0.266 for the entire length of the series, and 0.299 from 1969 to 2007. The cores from within the area protected from deer browsing returned an overall series intercorrelation of 0.338, with 0.276 for the years 1969–2007. The standard and residual ring-width indices from inside and outside of the fence also follow closely (Fig. 2), with a correlation of 0.730 and 0.807, respectively, for the post1969 period. The absence of any notable growth change is likely due to the maturity of the sampled trees when the deer enclosure was established, an idea supported by Perrin et al.’s (2006) finding that the impacts of deer browsing in Killarney are restricted to saplings. This analysis showed that the mature T. baccata in Reenadinna Wood did not respond significantly to the construction of the deer-proof fence and, consequently, all tree-ring width series were pooled to produce the final chronologies. The oldest core in the standard chronology dated to the year 1763 (Table 2). However, the Subsample Signal Strength (SSS) – the increased uncertainty of a chronology when the number of its constituent core series drops in early periods (Wigley et al., 1984) – of the standard chronology drops below 0.75 in 1804. Therefore, the final chronology runs from 1805 to 2007, and is plotted in Fig. 3 along with the residual series. The mean sensitivity of the standard chronology, a measure of the potential impact of high frequency climate variability on the growth response (Fritts, 1976), is 0.161, suggesting that it can potentially be employed in climate reconstructions (Speer, 2010). The standard deviation is 0.214, with the first order autocorrelation of the standard T. baccata index in Killarney National Park at 0.545. The SSS of the residual chronology falls below 0.75 in 1802, with the chronology running from 1803 to 2007. The correlations between the standard ring-width and the mean monthly temperatures for the entire length of the Muckross House synoptic station records (1970–2007) and the same time period from the Valentia Observatory records can be seen in Tables 3a and 3b, respectively. For the Muckross House data, the months with the most significant influence on standard T. baccata ring-width are November of the previous year to April of the growth year. Combining the data for all six months gave an r-value of 0.605 (p < 0.01) and a Durbin–Watson value of 1.546. Although not particularly strong, this Durbin–Watson result does indicate some serial correlation. For the Valentia Observatory data, the months November to May (1970–2007) display the strongest influence upon T. baccata growth, producing an r-value of 0.584 (p < 0.01) and a Durbin–Watson result of 1.565. However, for the sake of consistency, it was decided to utilise the data from November to April for
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
5
Fig. 2. (a) Standard T. baccata ring-width chronologies from cores sourced from within and outside of the deer-proof fence; (b) residual T. baccata ring-width chronologies from cores sourced from within and outside of the deer-proof fence.
comparative analysis. Correlation between the standard T. baccata index and the mean temperatures for these six months at Valentia Observatory for the time period 1970–2007 produced an r-value of 0.567 (p < 0.01) and a Durbin–Watson value of 1.456. The influence of November–April temperatures on T. baccata ring widths is likely a measure of length of growing season, with
a warmer winter–spring triggering an earlier start of the growing season and resulting in a larger ring for that year. The tree-ring records may be more responsive to post-1970 temperature fluctuations due to a general warming in seasonal temperatures during this time. The Muckross House and Valentia Observatory climate records show an approximately 1 ◦ C increase in winter and spring
Fig. 3. (a) Standard T. baccata ring-width chronology for Reenadinna Wood, including 11-year running mean, number of cores and the 0.75 Subsample Signal Strength (SSS) threshold; (b) residual T. baccata chronology.
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model
ARTICLE IN PRESS
DENDRO-25278; No. of Pages 9
S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
6
Table 3a Correlations between mean monthly temperature values from Muckross House (1970–2007) and Valentia Observatory (1872–2007) and standard T. baccata treering index values, as well as total monthly precipitation values for Muckross House (1970–2007) and Valentia Observatory (1941–2007) and residual T. baccata treering index values (the numbers in bold signify the months selected for the purposes of reconstructions). Month
Apr TY-1 May TY-1 Jun TY-1 Jul TY-1 Aug TY-1 Sep TY-1 Oct TY-1 Nov TY-1 Dec TY-1 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov
Muckross House
Table 3b Correlations between standard T. baccata tree-ring index values and mean monthly temperature values in Valentia Observatory for the time periods 1872–1969 and 1970–2007, as well as residual T. baccata tree-ring index values and total monthly precipitation values for 1941–1969 and 1970–2007 (the numbers in bold signify the months selected for the purposes of reconstructions). Month
Valentia Observatory
Temperature
Precipitation
Temperature
Precipitation
0.122 0.530** 0.121 0.157 0.026 0.277 0.107 0.436** 0.056 0.390* 0.398* 0.551** 0.365* 0.308 0.068 −0.162 0.076 0.176 −0.024 0.165
0.151 −0.163 −0.040 0.021 0.045 −0.011 0.047 0.020 −0.209 0.016 0.219 0.068 −0.114 0.240** 0.277** 0.229 0.252 −0.225 −0.256 0.042
0.035 0.071 0.029 0.160 0.136 0.000 0.071 0.051 0.141 0.281** 0.218* 0.135 0.127 0.047 −0.044 0.004 0.071 0.095 −0.002 0.149
0.196 −0.175 −0.274* −0.038 0.061 0.002 −0.038 −0.099 −0.073 −0.005 0.126 0.019 −0.238 0.331** 0.455** 0.134 0.197 0.014 −0.125 0.208
Regression equation employed in temperature reconstruction (Muckross data): • (Nov-Apr Temp) = 7.608 + ((2.476 * TB-std) − (0.179 * TB-std − 1)), where Nov-Apr Temp = mean temperature values for November to April and TB-std = standardised T. baccata chronology. Regression equation employed in precipitation reconstruction (Muckross data): • (May-Jun Precip) = −269.363 + ((253.503 * TB-res) + (200.776 * TB-res − 1)), where May-Jun Precip = total precipitation values for May to June and TB-res = residual T. baccata chronology. Regression equation employed in precipitation reconstruction (Valentia data): • (May-Jun Precip) = −122.748 + ((56.714 * TB-res) + (237.316 * TB-res − 1)), where May-Jun Precip = total precipitation values for May to June and TB-res = residual T. baccata chronology. * Correlation is significant at the 0.05 level. ** Correlation is significant at the 0.01 level.
temperatures since 1980, a trend which is seen throughout Ireland (McKeown et al., 2012). The last 30 years thus account for the warmest winter–spring temperatures on record (McKeown et al., 2012). The warmer seasonal weather may be increasing its potential influence on tree growth due to a lengthened growing season, with winters prior to the 1980s largely being too cold for a small change in temperature to significantly influence tree growth. The warming trend experienced since the 1980s coincides with an increase in ring widths over the same time period (Fig. 3). The correlations between the Valentia Observatory temperature data and the standard ring-width chronology in Table 3b emphasise this point further. Concentrating on the pre-1970 time period highlights the instability of the T. baccata response to temperature. Correlation with the standard tree-ring index and mean November–April temperatures for the years 1872–1969 returned an r-value of 0.157, and a Durbin–Watson value of 1.936. Meanwhile, extending this analysis to the entire length of the Valentia Observatory dataset (1872–2007) produced an r-value of 0.262, and a Durbin–Watson value of 1.063. This shows that, although significant correlations have been achieved for the more recent decades, the temperature signal in T. baccata in Reenadinna Wood is not time stable. The significant monthly correlations facilitated the reconstruction of mean November to April temperatures for Muckross House synoptic weather station and Valentia Observatory for the years 1970–2007, the equations for which are also in Tables 3a and 3b, respectively. The observed and reconstructed temperature data
Apr TY-1 May TY-1 Jun TY-1 Jul TY-1 Aug TY-1 Sep TY-1 Oct TY-1 Nov TY-1 Dec TY-1 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov
Temperature
Precipitation
1872–1969
1970–2007
1941–1969
1970–2007
−0.024 −0.150 0.075 0.100 0.131 −0.149 −0.042 −0.065 0.098 0.245* 0.135 −0.021 −0.010 −0.132 −0.083 −0.060 0.026 0.058 −0.074 0.091
0.117 0.515** −0.011 0.118 0.035 0.144 0.115 0.386* 0.057 0.361* 0.401* 0.497** 0.353* 0.374* 0.068 −0.086 0.023 0.019 −0.047 0.167
0.117 0.515** −0.011 0.118 0.035 0.144 0.115 0.386* 0.057 0.361* 0.401* 0.497** 0.353* 0.374* 0.068 −0.086 0.023 0.019 −0.047 0.167
0.099 −0.041 −.436* −0.110 −0.056 −0.073 −0.053 −0.406* −0.044 0.091 −0.041 −0.007 −0.395* 0.383* 0.487** 0.102 0.346 0.179 −0.189 −0.042
Regression equation employed in temperature reconstruction (1872–1969): (Nov-Apr Temp) = 7.027 + ((−0.121 * TB-std) − (0.814 * TB-std − 1)), where • Nov-Apr Temp = mean temperature values for November to April and TBstd = standardised T. baccata chronology. Regression equation employed in temperature reconstruction (1970–2007): • (Nov-Apr Temp) = 5.725 + ((1.895 * TB-std) − (0.190 * TB-std − 1)), where Nov-Apr Temp = mean temperature values for November to April and TB-std = standardised T. baccata chronology. * Correlation is significant at the 0.05 level. ** Correlation is significant at the 0.01 level.
for both Muckross and Valentia follow similar peaks and troughs, with a general increase over time (Fig. 4). However, the models do tend to underestimate the fluctuations, a common characteristic of tree-ring models as the trees’ growth responses cannot cross certain thresholds (Esper and Frank, 2009). The reconstructed temperatures created utilising the Valentia Observatory data for the years 1872–1969, the equation for which can be seen in Table 3b, are, as can be expected, grossly underestimating the year-to-year fluctuations (Fig. 4). The low r-value of 0.157 ensures that the reconstruction’s ability to reflect change is weak. As with the temperature data, precipitation records from Valentia Observatory were also split into two time periods in order to allow ease of comparison with the Muckross data (Tables 3a and 3b). Table 3a shows that, for the Muckross House data (1970–2007), significant correlation values were achieved in May and June of the current year, with r-values of 0.240 (p < 0.01) and 0.277 (p < 0.01), respectively. Analysing the same time period for the Valentia Observatory data produced fewer significant correlations, with an r-value of 0.487 (p < 0.01) returned for June of the current year. However, extending the time period of examination to the entire length of the Valentia dataset (1941–2007) did return more significant correlations (Table 3a), with r-values of 0.331 (p < 0.01) and 0.455 (p < 0.01) for May and June, respectively. The pre-1970 data (Table 3b) also exhibited significant relationships to precipitation, but over a broader range of months. Because precipitation can vary so greatly from one year to the next, it was necessary to use the residual chronology as opposed to the standard chronology for these correlations as the former factors out autocorrelation in its formulation process (Meko et al., 1993). The correlation between total May–June rainfall
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
7
Fig. 4. Reconstructed and observed mean November–April temperatures in Killarney using Muckross House synoptic station data (1970–2007), and Valentia Observatory data (1872–1969 and 1970–2007).
Fig. 5. Reconstructed and observed total May–June precipitation values for Killarney, using Muckross House synoptic station data (1970–2007) and Valentia Observatory data (1941–2007).
(1970–2007) and residual T. baccata ring-widths produced an rvalue of 0.521 (p < 0.01) and Durbin–Watson value of 2.062 for the Muckross House synoptic station data, while a correlation of 0.542 (p < 0.01) and Durbin–Watson result of 2.086 was returned for Valentia Observatory (1941–2007) (Table 3a). The Durbin–Watson results, where the autocorrelation of the residuals from the regression is measured, indicate that there is little serial correlation; this is a consequence of the fact that precipitation values for one year have no influence on those of the following year. The positive relationship with summer precipitation (in a region where moisture is usually not a limiting growth factor) is due to the fact that the study site, similar to all T. baccata stands in Ireland, consists of thin, well-drained soils overlying limestone (Perrin, 2002). The equations used to reconstruct total May–June precipitation from Muckross House data (1970–2007) and Valentia Observatory data (1941–2007) can be seen in Table 3a. The reconstructed and observed values are shown in Fig. 5. Both trace similar peaks and troughs but, as with the reconstructed temperatures, the model has a propensity to undervalue the intensity of the changes. The correlation coefficient between Pilcher’s (1978) Q. petraea series from Killarney National Park and the T. baccata series is 0.207(p < 0.01) for the years the chronologies overlapped (1809–1978). The low correlation value highlights the fact that the two species react differently to similar climatic conditions, a
point emphasised further still when both series are correlated with Hurrell et al.’s (2003) winter North Atlantic Oscillation index. In a positive NAO phase, southwest Ireland should exhibit warmer and wetter conditions. While the T. baccata series shows a positive relationship with NAO (0.303, p < 0.01), the Q. petraea series exhibits a negative relationship (−0.330, p < 0.01). All three indices are presented in Fig. 6. This can be explained by species site preference. Whereas both taxa tend to thrive in warmer conditions (Gardiner, 1974; Perrin, 2002), they differ in their responses to precipitation. The thin, well-drained soils where T. baccata stands flourish are likely to promote a positive response to wetter periods associated with positive phases of NAO. Conversely, Q. petraea prefers deep, acidic soil (Lévy et al., 1992) and, as a result, growth is negatively impacted upon by water logging (Parelle et al., 2007) during positive phases of NAO. The different responses to NAO highlight the usefulness of multi-species approaches to climate reconstruction.
Discussion and conclusions For decades, tree-ring studies in Ireland had been dominated by the use of Quercus species, with García-Suárez et al. (2009) being the first to undertake a formal examination of the dendroclimatological potential of other tree species on the island. Their results
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
8
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
Fig. 6. Standard chronology indices for T. baccata (1899–2007) and Pilcher’s (1978) Q. petraea (1899–1978) with Hurrell et al.’s (2003) principal component winter NAO index. Correlation coefficient between T. baccata and NAO: 0.303 (p < 0.01); Q. petraea and NAO: −0.330 (p < 0.01).
showed that F. sylvatica, F. excelsior and P. sylvestris can be used in addition to Quercus species as part of a multi-species dendroclimatological analysis. This study shows that an additional species, T. baccata, is a reliable source of data that can be utilised in the investigation of past environmental conditions. The results garnered as part of this study show a similar strength of response to climate when compared to other Irish species (Pilcher and Baillie, 1980; García-Suárez et al., 2009) and a stronger climate response than other studies of T. baccata internationally (Moir, 1999; Yadav and Singh, 2002). T. baccata is a long-lived species, with potential to survive more than 1000 years (Thomas and Polwart, 2003). O’Connell (1987) pointed out that the natural expansion of peatlands in various parts of Ireland throughout history led to the preservation of many Quercus and T. baccata trees. However, dendrochronological investigations in Ireland are restricted by the spatial and temporal limitations of the dominant Quercus chronologies. This is despite the fact that numerous examples of well-preserved and long-lived T. baccata trees are found throughout the country. The potential for expansion of the T. baccata chronology, as well as the potential for inter-species correlation with other species (Moir, 1999; Yadav and Singh, 2002), indicate that T. baccata can become an important constituent in multi-site and multi-species studies of past environments. The facts that only two T. baccata chronologies are available from the International Tree-ring Data Bank, and only a small number of studies have examined the species’ dendroclimatological applicability (Moir, 1999; Yadav and Singh, 2002), suggest that the potential benefits of T. baccata are not necessarily recognised. This may be due to real and perceived difficulties in attaining useable cores from the species, with missing rings also being a potential drawback (Thomas and Polwart, 2003). However, such issues did not arise during the course of this study. Here it has been shown that the coring and cross-dating of T. baccata can yield fruitful results even when faced with issues such as animal browsing. In fact, it is safe to conclude that the impact of deer on older, well-established, T. baccata in Reenadinna Wood is minimal. Thus, any assumptions that browsers could have compromised the temperature and precipitation models created are refutable. The long-lived nature of the T. baccata species presents the potential to expand the chronology further than the 204 years produced in the course of this research. The chronology could also be expanded spatially. Although Reenadinna Wood in Killarney National Park is the most extensive T. baccata stand in Ireland, other stands were identified by Perrin (2002) in limestone-dominated areas in The Burren, Co. Clare; Garryland Wood, Co. Galway; St. John’s Wood in Co. Roscommon; and The Rocks of Clorhane, Co. Offaly (Fig. 1a). Individual examples can also be found in estates
and graveyards throughout the country (Perrin, 2002). A multisite approach to extending the T. baccata chronology could lead to a longer series, in turn facilitating an extension and expansion of dendrochronological and dendroclimatological studies in Ireland. Such an extension of the T. baccata chronology would also strengthen multi-species approaches to the examination of past environmental change. Significant inter-species correlations indicate the potential for T. baccata to assume an important role in the study of regionally specific environmental change and the driving factors involved, a key feature in Frank and Esper’s (2005) call for progression in tree-ring research. The multi-species study undertaken by García-Suárez et al. (2009) highlighted the benefits of this approach in Ireland. This study’s opposing, but equally significant, responses of T. baccata and Q. petraea to the NAO strengthens the argument that differential signals are key to dendroclimatological reconstructions in regions where species are not at their ecological limits. The incorporation of T. baccata would thus add to the strength of any mixed-species climate signals in the region. This study has shown the strong potential for T. baccata in creating a long-term tree-ring series, and highlights the possibility of growing this chronology through space and time.
Acknowledgements The authors wish to thank Prof. Chris Caseldine and Prof. Micheál Ó’ Cinnéide for their helpful comments on content and form, and the anonymous reviewers who provided very constructive feedback. We also wish to acknowledge funding received from the Coimbra Group during the course of this research, as well as the permission granted by the National Parks and Wildlife Service to undertake the study in Reenadinna Wood.
References Bernabei, M., Gjerdrum, P., 2006. Tree-ring data for yew (Taxus baccata) from the Alps. NOOA Satellite and Information Service, http://www.ncdc.noaa.gov/paleo/ metadata/noaa-tree-2733.html (accessed 20.01.13). Biondi, F., 1992. Development of a tree-ring network for the Italian Peninsula. TreeRing Bull. 52, 5–29. Blasing, T.J., Duvick, D.N., West, D.C., 1981. Dendroclimatic calibration and verification using regionally averaged and single station precipitation data. Tree-Ring Bull. 41, 37–43. Cook, E.R., Briffa, K.R., Shiyatov, S., Mazepa, V., 1990. Tree-ring standardisation and growth-trend estimation. In: Cook, E., Kairiukstis, A. (Eds.), Methods of Dendrochronology: Applications in the Environmental Sciences. Kluwer Academic Publ., Dordrecht, pp. 104–123. Frank, D.C., Esper, J., 2005. Temperature reconstructions and comparisons with instrumental data from a tree-ring network for the European Alps. Int. J. Climatol. 25, 1437–1454.
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004
G Model DENDRO-25278; No. of Pages 9
ARTICLE IN PRESS S. Galvin et al. / Dendrochronologia xxx (2014) xxx–xxx
Esper, J., Frank, D.C., 2009. Divergence pitfalls in tree-ring research. Clim. Change 94, 261–266. Fritts, H.C., 1976. Tree Rings and Climate. The Blackburn Press, New Jersey, pp. 567. García-Suárez, A.M., Butler, C.J., Baillie, M.G.L., 2009. Climate signal in tree-ring chronologies in a temperate climate: a multi-species approach. Dendrochronologia 27, 183–198. Gardiner, A.S., 1974. A history of the taxonomy and the distribution of the native oak species. In: Morris, M.G., Perring, F.H. (Eds.), The British Oak. The Botanical Society of the British Isles, Cambridge, pp. 3–26. Gervais, B.R., 2006. A three-century record of precipitation and blue oak recruitment from the Tehachapi Mountains, Southern California, USA. Dendrochronologia 24, 29–37. Godwin, H., 1975. The History of the British Flora. Cambridge University Press, Cambridge, pp. 541–549. Grissino-Mayer, H., Holmes, R.L., Fritts, H.C., 1992. User’s Manual for the International Tree-ring Data Bank Program Library. Laboratory of Tree-Ring Research, University of Arizona, Tucson, 106 pp. Haeggström, C.A., 1990. The influence of sheep and cattle grazing on wooded meadows in Åland, SW Finland. Acta Bot. Fenn. 141, 1–28. Holmes, R.L., 1986. Quality control of cross-dating and measuring: a users manual for program COFECHA. In: Holmes, R.L., Adams, R.K., Fritts, H.C. (Eds.), TreeRing Chronologies of Western North America: California, Eastern Oregon and Northern Great Basin. University of Arizona Press, Tucson, pp. 41–49. Holmes, R.L., 1992. Dendrochronology Program Library, Version 1992-1. Laboratory of Tree-ring Research, University of Arizona Press, Tucson. Hurrell, J.W., 1995. Decadal trends in the north Atlantic oscillation: regional temperatures and precipitation. Science 269, 676–679. Hurrell, J.W., 2003. In: National Center for Atmospheric Research Staff (Ed.), The Climate Data Guide: Hurrell North Atlantic Oscillation (NAO) Index (PC-based). , http://climatedata guide.ucar.edu/guidance/hurrell-north-atlantic-oscillationnao-index-pc-based (accessed 12.01.13). Jalas, J., Suominen, J. (Eds.), 1973. Atlas Florae Europea. Distribution of vascular plants in Europe. 2. Gymnospermae (Pinaceae to Ephedraceae). The Committee for Mapping the Flora of Europe and Societas Biologica Fennica Vanamo, Helsinki, 40 pp. Kelly, D.L., 1981. The native forest vegetation of Killarney, south-west Ireland – an ecological account. J. Ecol. 69, 437–472. Kelly, D.L., (Dissertation) 1975. Native woodland in western Ireland with especial reference to the region of Killarney. University of Dublin, pp. 337. Kuniholm, P.I., Steele, L., Tovar, I., 2006. Tree-ring data for yew (T. baccata) from Georgia, http://hurricane.ncdc.noaa.gov/pls/paleox/f?p=519:1: 2633893827028787::::P1 STUDY ID:3799 (accessed 25.01.13). Lévy, G., Becker, M., Duhamel, D., 1992. A comparison of the ecology of pedunculate and sessile oaks: radial growth in the centre and northwest of France. For. Ecol. Manag. 55, 51–63. Lowe, J., 1897. The Yew Trees of Great Britain and Ireland. MacMillan, London, pp. 342. McKeown, M., Potito, A., Hickey, K., 2012. The long-term temperature record from Markree Observatory, County Sligo, from 1842 to 2011. Irish Geogr. 45, 257–282. Meko, D., Cook, E.R., Stahle, D.W., Stockton, C.W., Hughes, M.K., 1993. Spatial patterns of tree-growth anomalies in the United States and southeastern Canada. J. Climatol. 6, 1773–1786.
9
Mitchell, A.K., 1998. Acclimation of Pacific yew (Taxus brevifolia) foliage to sun and shade. Tree Physiol. 18, 479–757. Mitchell, F.J., 1990. The history and vegetation dynamics of a yew wood (Taxus baccata) in S.W. Ireland. New Phytologist 115, 573–577. Moir, A.K., 1999. The dendrochronological potential of modern yew (Taxus baccata) with special reference to yew from Hampton Court Palace, UK. New Phytologist 144, 479–488. O’Connell, C. (Ed.), 1987. The IPCC Guide to Irish Peatlands. Irish Peatland Conservation Council, Dublin, p. 102. Office of Public Works, 1990. Killarney National Park Management Plan. The Stationery Office, Dublin, pp. 72. Parelle, J., Brendel, O., Jolivet, Y., Dreyer, E., 2007. Intra- and interspecific diversity in the response to waterlogging of two co-occurring white oak species (Quercus robur and Q. petraea). Tree Physiol. 27, 1027–1034. ˜ Penalba, M.C., 1994. The history of the Holocene vegetation in northern Spain from pollen analysis. J. Ecol. 82, 815–832. Perrin, P., (Dissertation) 2002. The Ecology of Yew (Taxus baccata) in Ireland. Trinity College Dublin, pp. 160. Perrin, P., Kelly, D.L., Mitchell, F., 2006. Long-term deer exclusion in yew-wood and oak-wood habitats in southwest Ireland: natural regeneration and stand dynamics. For. Ecol. Manag. 236, 356–367. Pilcher, J.R., 1978. Tree-ring data for Q. petraea from Killarney, Available at http://www.ncdc.noaa.gov/paleo/metadata/noaa-tree-4220.html (accessed 15.02.13). Pilcher, J.R., 1994. Biological evidence of past environments and climatic changes. Eur. Rev. 2, 271–278. Pilcher, J.R., Baillie, M.G.L., 1980. Six modern oak chronologies from Ireland. TreeRing Bull. 40, 23–34. Ratcliffe, D.A., 1977. A Nature Conservation Review. Cambridge University Press, Cambridge, pp. 401. Speer, J.H., 2010. Fundamentals of Tree-ring Research. University of Arizona Press, Arizona, pp. 333. Stokes, M.A., Smiley, T.L., 1968. An Introduction to Tree-ring Dating. University of Chicago Press, Chicago, pp. 73. Tabbush, P., White, J., 1996. Estimation of tree age in ancient yew woodland at Kingley Vale. Q. J. Forest. 90, 197–206. Thomas, P.A., Polwart, A., 2003. Taxus baccata L. J. Ecol. 91, 489–524. Tittensor, R.M., 1980. Ecological history of yew Taxus baccata L. in southern England. Biol. Conserv. 17, 243–265. ˇ ´ M., 1991. Conifers, Morphology and Variation (M. Soljan, Vidakovic, Trans.). Gradfiˇcki Zavod Hrvatske, Zagreb, 754 pp. Voliotis, D., 1986. Historical and environmental significance of the yew (Taxus baccata L.). Israel Journal of Botany. Basic Appl. Plant Sci. 35, 47–52. Watt, A.S., 1926. Yew communities of the South Downs. J. Ecol. 14, 282–316. Wigley, T.M.L., Briffa, K.R., Jones, P.D., 1984. On the average value of correlated time series, with applications in dendroclimatology and hydrometeorology. J. Climate Appl. Meteorol. 23, 201–213. Yadav, R.R., Singh, J., 2002. Tree-ring analysis of Taxus baccata from the western Himalaya, India, and its dendroclimatic potential. Tree-ring Res. 58, 23–29.
Please cite this article in press as: Galvin, S., et al., Evaluating the dendroclimatological potential of Taxus baccata (yew) in southwest Ireland. Dendrochronologia (2014), http://dx.doi.org/10.1016/j.dendro.2014.03.004