No change without a cause – why climate change remains the most plausible reason for shrub growth dynamics in Scandinavia

Büntgen & Schweingruber (2010) challenge the methods and the main conclusions of our recently published research article ‘Establishing a missing link: warm summers and winter snow cover promote shrub expansion into alpine tundra in Scandinavia’ (Hallinger et al., 2010). Therein, we present three main lines of evidence for an elevational shrubline expansion in northern Sweden: a pronounced increase of radial and vertical shrub growth; an age structure that indicates an up-slope range expansion of shrubs; and significant positive climate–growth relationships with summer temperatures and winter snow cover. Over the last few years, mounting evidence from around the arctic has suggested that shrub growth in northern tundra biomes has increased over the last decades (Sturm et al., 2001; Kullman, 2002; Jia et al., 2004; Jagerbrand, 2005; Tape et al., 2006; Bunn et al., 2007) and is probably linked to summer warmth or increasing mean air temperature (Shiyatov et al., 2002; Rayback & Henry, 2006; Bär et al., 2008; Forbes et al., 2009; Rozema et al., 2009). Our results contribute to this growing amount of indication for a pan-Arctic shrub expansion, as postulated by Tape et al. (2006), and include analyses of both radial and vertical growth changes in shrubs. To address the critique of Büntgen & Schweingruber (2010), we have re-analyzed our data and present additional morphological findings on vertical growth. However, none of our re-analyses changes or questions our main conclusions. The seemingly relevant pitfalls of our work concerned: climate change in Scandinavia, ring-width increase and ecology (issues not enumerated separately but addressed repeatedly in Büntgen & Schweingruber, 2010); averaging different stem heights and tree-ring standardization and the danger of artificial index inflation. In the following, we will first address each issue raised by Büntgen & Schweingruber (2010) and then present additional evidence for a shrub expansion and its possible causes in our research area. Büntgen & Schweingruber (2010) present a well-replicated tree-ring data set from northern Scandinavia spanning the last 1500 yr to show that there is no long-term trend from generally colder Little Ice Age conditions to recent warmer summer temperatures in Scandinavia (i.e. during the last 200 yr). According to Büntgen & Schweingruber (2010), the nonexistent long-term trend of summer warming (supported by climate data; cf. Callaghan et al., 2010) questions the link between warm summer temperatures and shrub growth. While we agree that there has been no long-term trend in summer temperatures over the last 200 yr (and never stated this in Hallinger et al., 2010), the conclusion drawn by Büntgen & Schweingruber (2010) is questionable: shrubs do not need long-term (200-yr-long) trends of increasing temperatures to increase growth but often react rapidly to changes in the environment occurring on yearly to decadal resolution (Bret-Harte et al., 2002; Epstein et al., 2004). In periods with above-average temperatures, such as 1920 to 1930 and 1970 to present, shrubs are able to increase growth, while they react with decreasing growth in colder periods (such as 1940 to 1960, Hallinger et al., 2010; Fig. 3). The warming events in the first half of the 20th century and in the present century are sufficient to explain elevated radial and vertical shrub growth. Moreover, while no trend in summer temperatures has been observed, significant increases in autumn and spring temperatures (2.5 and 2.9°C) imply an increase in growing degree days over the last century (Callaghan et al., 2010). As we have the impression that our results regarding ring-width increase have been profoundly misunderstood, we present some of them here in a different way (Fig. 1a,c), including information that was only shown in part in Hallinger et al. (2010) as a result of space considerations (Fig. 1b). It should be noted that we did not employ the standardized ring-width curves to extrapolate vertical growth, as incorrectly stated by Büntgen & Schweingruber (2010) but clearly described by Hallinger et al. (2010). In our study, we report a substantial vertical growth increase (shoot elongation) at all elevational levels starting in 1970. After assuming growth rates to remain constant between consecutive shrub disks (respectively between the onset of growth (pith date) between those disks) and subsequent averaging of all individuals, it is apparent that there was an earlier peak of vertical growth c. 1930 and a growth reduction before the recent increase (Fig. 1a). This growth pattern strongly resembles the radial growth pattern (Fig. 1c); also described in Hallinger et al. (2010, Fig. 3). Vertical and radial growth of Juniperus nana shrubs in northern Sweden. (a) Vertical growth (cm yr−1) at (i) 770 m, (ii) 800 m, (iii) 900 m, (iv) 1000 m and (v) 1100 m above sea level (asl), and (vi) arithmetic mean. (b) Raw measurements before power-transformation (PT). (c) Chronologies: (I) index calculation: ratios; (II) PT, individual stem-height levels standardized using conservative methods and index calculation: residuals; (III) PT, individual stem-height levels standardized using the regional curve standardization (RCS) method, index calculation: residuals; (d) sample depths: black, number of individuals; grey, number of radii; (e) difference between chronology I and chronology II; (f) difference between chronology I and chronology III. RW, ring width. Thus, we acknowledge the influence of the 1930 warming peak on shrub growth and support the thought of Büntgen & Schweingruber (2010) that our average shrub ages of 84 yr correspond to germination conditions during this period that were characterized by both warm summers and winters. As current climatic conditions are similar, the numerous shrub saplings that we observed at higher elevations might be an indication for yet another (climate-driven) germination event in our study area. The long-term survival of saplings will, of course, depend on mortality rates that might be rather high in the northern mountain tundra (Körner, 2003). Büntgen & Schweingruber (2010) speculate sapling establishment of shrubs to start with a decade-long period of suppressed growth and a subsequent growth release, similar to conifers in North Russia (Esper & Schweingruber, 2004). We are, however, sceptical of whether this also applies to the shrub species we sampled. We record 13 individuals (of 34) that show a growth pattern similar to the one described for the Russian conifer saplings, but all except one of these shrubs germinated in the 1950s to 1960s, when shrub growth was generally low. In shrubs that had germinated before 1950, shrub growth was regularly not suppressed and we therefore assume that the observed suppression is caused by external forcing (low temperatures). This also justifies the inclusion of the first 25 yr of shrub growth in our calculations of radial growth (Hallinger et al., 2010). Even if we – to be on the safe side – divide the mean ring-width of the period 1955 to 1980 (and not the mean of the first 25 yr of shrub ring-width) by the mean of the last 25 yr of radial growth (1981 to 2006), we still receive an average radial growth increase of 51%, with the highest shrubs almost doubling (91%) their average ring width after 1980 (compared with 61% and 73%, respectively, in Hallinger et al., 2010). Büntgen & Schweingruber (2010) just state that our thought model of a positive relationship between plant age and elevation (an age–class structure with increasing plant age with increasing elevation) lacks scientific grounds but fail in specifically stating why they come to this opinion and do not provide any evidence for it. This comes as a surprise as it is written in Schweingruber & Poschlod (2005) that (first described for alpine and arctic environments) environmental stress reduces plant growth and therefore prolongs individual plant life. As environmental stress usually increases with increasing elevation (Körner, 2003), we deem it logical to assume increasing average plant ages with increasing elevation. In the case of an up-slope shrub increase, however, we would expect average shrub ages at shrubline to be younger than further down-slope as a sign for increased recruitment at the margins of the distributional range (Hallinger et al., 2010). Generally, younger individuals can also denote regular recruitment without the actual movement of the upper limit: this would result in a thickening of shrub patches and is, by definition (Tape et al., 2006; Hallinger et al., 2010), one of the criteria for shrub expansion. It is also true that older individuals can be relicts of an earlier favourable period for recruitment, but there has to be a reason why they are not distributed evenly along the elevational gradient. Similar reasoning can be found for a possible elevational willow expansion in northwest Canada (Danby & Hik, 2007) and for an elevational treeline advance in northeast Canada (Gamache & Payette, 2005). Ward (1982) reported that slow-growing juniper shrubs tend to live for longer, and Clifton et al. (1996) found that shrub longevity increased with latitude and attributed it to increasingly extreme climatic conditions. The prostrate growth form of many shrubs, including the dwarf junipers (Juniperus nana Willd.) studied in Hallinger et al. (2010), causes wedging and missing rings, and eccentricity of the pith. The comparison and final averaging of several radii from different stem heights (serial-sectioning) is therefore beneficial for the accurate cross-dating and the development of shrub growth curves (Kolishchuk, 1990; Bär et al., 2006; Hallinger, 2007; Hallinger et al., 2010). Büntgen & Schweingruber (2010) now argue that the continuous inclusion of successively younger wood with potentially higher mean ring-width (at higher stem levels) will result in an average growth curve that is systematically elevated and thus biased towards its end. While this reasoning is understandable, it generally does not apply to the shrub species we studied. Fig. 2(a,b) shows the ring width–stem height profiles of 34 dwarf juniper shrubs. Thus, the average shrub usually shows wider rings at the stem base than at all height levels higher up (Fig. 2b) and the averaging of successive height levels will therefore systematically lessen the magnitude of the (average) outermost rings. Only the typical ring width–stem height profile of a tree (Fig. 2c; adapted from Krause & Eckstein, 1992) would result in the schematic diagram of biased mean growth curves presented by Büntgen & Schweingruber (2010). Ring width (RW) profiles of 34 Juniperus nana shrubs. (a) Individual shrub stem-height profiles (average RW of up to 12 height levels scaled to the maximum level that occurred at 400 cm). (b) Arithmetic mean and smoothed (10 successive stem heights) average RW profile indicating that the largest rings occur at the stem base. (c) Typical RW profile of a tree with increasing ring width towards the top of the stem (spruce, generalized after Krause & Eckstein, 1992). Before studying the influence of external factors, such as climate, on shrub- or tree-rings, it is essential to remove biological age trends (Cook & Kairiukstis, 1990; Esper et al., 2003). Thereby, artificial index inflation (end-effect) is possible when ring indices are calculated by division of actual growth by a standardization curve and if this curve decreases towards zero and/or underestimates the measurements (Cook & Peters, 1997). As a remedy, Büntgen & Schweingruber (2010) recommend the use of an adaptive power transformation (PT) to make the measurements homoscedastic (and thus stabilize the variance), permitting differencing instead of dividing for index calculation. Additionally, they advise the standardization of individual radii from different stem heights before averaging them into growth curves or chronologies to circumvent the potentially biasing effects (see below) of averaging individual height levels of different length and mean growth. Our standardization involved only five shrubs (15%) with sloping standardization curves (slope different from 0) and in those cases, the fitted growth curve did not undershoot the actual measurements (Hallinger et al., 2010; the resulting chronology is displayed in Fig. 1c, I). Still, following the protocol suggested by Büntgen & Schweingruber (2010), we power-transformed and detrended individual height levels (horizontal line, but for the same 15% of samples negative linear regression or negative exponential where appropriate), used differencing for index calculation and stabilized the variance of the resulting chronology (Osborn et al., 1997; Fig. 1c, II). In addition, we repeated the protocol but applied the method of regional curve standardization (based on a regional curve showing declining radial growth trend with age, Fig. 1c, III). The original as well as the recalculated chronologies, and the resulting differences, are displayed in Fig. 1(c–f). The differences, especially concerning the suspected end-effects, are negligible in the main period of interest (pre-1920 differences probably relate to decreasing sample size), which is not surprising as a result of our conservative detrending applied (Hallinger et al., 2010) and the slight, but systematic, underestimation of average mean growth curves in the original chronology (Fig. 2b). Optimized correlation between all different standard shrub chronologies and the June to July temperature improved (from an average r = 0.4, 1913 to 2004; Hallinger et al., 2010) during the second half of the 20th century (r = 0.6–0.7 for the year of growth and r = 0.4 for the year before growth; P < 0.05; 1970 to 2004). Recent research has shown an increase in the smoothed mean annual air temperatures in North Sweden by 2.5°C between 1913 and 2005, mainly through an increase in spring and autumn temperatures causing an increase in the number of growing degree days (Callaghan et al., 2010). Additionally, most recent warming (since 2005) has exceeded that of the 1930s (Callaghan et al., 2010), with summers being exceptionally warm since 1996 (Van Bogaert et al., 2009). Shrubs have been found to strongly react on artificially elevated temperature (Bret-Harte et al., 2002; van Wijk et al., 2004; Jonsdottir et al., 2005; Rozema et al., 2009) and they increased during the last 10 yr within the direct vicinity (Olofsson et al., 2009), and during the last 50 yr within the larger vicinity (Kullman, 2007), of our research area. Therefore, an increase of the juniper population studied would be no surprise, as nonclimatic factors presumably do not significantly apply, as discussed later. Recent findings from our research area show that woody species have increased their growth and (elevational) range: shrub biomass and shrub height of Betula nana L. increased during the recent abnormally warm decade (Olofsson et al., 2009). Aspen (Populus tremula L.), a thermophilic tree species, increased its range in the subalpine birch forest as well as at treeline, where it was found to be up to 180 m higher than at the beginning of the 20th century (Van Bogaert et al., 2010). The birch (Betula pubescens Ehrh. ssp. czerepanovii (Orlova) Hämet-Ahti) treeline increased locally up to 145 m up-slope (Van Bogaert et al., 2011). In the Scandes of West-Central Sweden, subalpine and alpine plant species shifted 200 m on average up-slope. There, juniper was found at a location 105 m higher than in the early 1950s (Kullman, 2007). These findings support our conclusions (Hallinger et al., 2010) and rebut the assumptions of Büntgen & Schweingruber (2010) that recent decades compared with earlier decades are not marked by an elevated degree of ecological changes. In detail, the causes attributed to the observed vegetation dynamics vary according to life form and ecology of the respective species: for birches at treeline, warm summer temperatures seem to be of high importance for yearly growth and snow depth is positively correlated with elevational shifts of the treeline (Van Bogaert et al., 2011). Herding disturbances and browsing by reindeers, as well as defoliation caused by moth outbreaks, are the main causes preventing a uniform treeline rise (or even causing retreat of the treeline) in the area (Van Bogaert et al., 2011). For aspen, a positive relationship of germination with summer temperature of the year before, summer precipitation and decreased competition by birch has been found (Van Bogaert et al., 2010). The shrub expansion first mentioned in the area (Sandberg, 1963) was a strong growth increase and range expansion of willows (Salix sp.) in the two rather cool decades following the 1930 warming event and a strong reduction of the reindeer population in the 1930s (Van Bogaert et al., 2011). Both factors probably influenced willow germination and establishment rates positively. Whereas herbivory has been found to play a role for deciduous shrubs and trees in the study area, juniper in our study area is probably, if at all, only indirectly influenced by herbivores: juniper shrubs are not browsed by reindeers, and rodents and insects have not been reported to feed on it with significant impact as far north in its distributional range (Ward, 1977; Manseau et al., 1996; Moen & Danell, et al., 2007; et al., 2009; et al., 2009). In three in of the juniper distributional range we have never to juniper shrubs. Whereas the browsing of deciduous shrubs in the vicinity of juniper might potentially competition and thus further juniper growth, there was competition by shrub species 900 m above sea level (asl), where growth increases were It be that by the highest number of on record in the 1930s might have some of the vegetation and for while the subsequent strong in reindeer caused by a Moen & Danell, would have establishment of the Similar decreases in have been reported to of establishment in juniper & However, our data do not support this as most of the germination in our research area occurred in the 1950s when reindeer were and in the early when reindeer were still increasing three five It be however, that the of used to denote recruitment are rather in this study and should be further increased to a of germination While juniper is generally & et al., 2007), it can be to as a result of snow cover (Kullman, 2007) or of and these factors rather apply to the treeline where winter temperatures are higher and snow cover therefore is et al., 2009). While we have far found no of or in our research area, with the current trend of snow cover (Callaghan et al., 2010), events might potentially an to shrub increases in the that influence shrub growth is by that has increased in during the 20th century the to et al., 2009). shrubs in a positive or a negative way to on the in on the amount of on the and on the individual of the and of the & 1996; & et al., Whereas significant increases in vertical and radial growth be reported from a on Juniperus growing on 2003), nana has after yr of with and & a increase in vegetation cover from to Generally, species that are found on strongly to than species that are adapted to & The Juniperus has a of on a of to et al., for the subalpine area at which is to than the part of Sweden & has decreased since and has been over the last two decades at a of is c. & 2004). As a result of the and the recent and rates at that do not show any with the recent increase in ring we deem it rather that is linked to the recent increase in shrub growth, we that elevated of the of nana to higher of have shrub growth earlier we that an expansion of our to within the area and around the arctic would be to our results and potentially increase the evidence for a shrub expansion. increased sample number would be beneficial for a of the of the germination We are that the factors plant growth are and and that of factors will therefore remain a further into the different factors at work would be After of all possible present in our study area, of the data to and after the of our methods according to the by Büntgen & Schweingruber (2010), the results of our study have been As juniper was not found to be significantly to herbivory in our research area, as shrubs and another thermophilic species have increased within the vicinity of the research area and according to our results (Hallinger et al., 2010), we in with et al. and that climate is the most for the observed and increases in radial and vertical shrub growth, as well as for the up-slope establishment of shrubs at high elevations in northern Sweden. While none of the environmental factors can be by evidence was found to support their influence in this In with Büntgen & Schweingruber (2010), we agree that conclusions can the long-term of but would to stress that this not only applies to of research but also to the scientific and on was by the of the Hallinger, an and the for and We Callaghan and the for their

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.