Interspecific and intraspecific variation in leaf toughness of Arctic plants in relation to habitat and nutrient supply

Leaf toughness is an important functional trait that confers resistance to herbivory and mechanical damage. We sought to determine how species composition, climate, seasonality, and nutrient availability influence leaf toughness in two types of tundra in northern Alaska. We measured leaf toughness as force to punch for 11 species of Arctic plants in tussock tundra and dry heath tundra at 17 sites distributed along a latitudinal gradient. Rubus chamaemorus and the graminoids occupied opposite ends of the leaf toughness spectrum, with R. chamaemorus requiring the least force to punch, while one of the graminoids, Eriophorum vaginatum, required the most. Leaf toughness increased with mean summer temperature for E. vaginatum and Betula nana , while it declined with warmer temperatures for the other species. Toughness of mature leaves of E. vaginatum did not vary through the growing season but declined significantly after senescence. Application of N and P fertilizer in an experimental site decreased leaf toughness in three species but had no effect on four others. Leaf toughness of four out of five species in dry heath was greater than for the same species in tussock tundra, but there was no difference in community-weighted mean toughness between tussock tundra and dry heath.


Introduction
found that wind did not affect leaf toughness as much as mechanical stimulation.
However, the more exposed conditions found in many arctic habitats may play a greater role in selecting for leaf toughness than in the lawns and pastures favored by Plantago major. At the ecosystem level, leaf toughness can affect rates of decomposition (Cornelissen et al. 1999, Pérez-Harguindeguy et al. 2000. In an extensive review, Onoda et al. (2011) examined interspecific trends in leaf toughness in response to latitudinal, temperature, and precipitation gradients. They did not find a significant trend with respect to latitude or temperature, but they did find that leaf toughness increased with decreasing mean annual precipitation. However, their survey did not include species from the Arctic. On the other hand, studies of functional traits in the Arctic have not included measures of leaf toughness (Bjorkman et al. 2018a, b).
Nevertheless, leaf toughness may be a significant trait for Arctic plants. Below-freezing temperatures and limited light mean that tundra vegetation is inactive throughout the winter, although growth proceeds rapidly once snow disappears (Khorsand Rosa et al. 2015). During the winter, leaves of evergreen species are exposed to high winds and blowing snow. Species that grow in exposed heath habitats are particularly stressed during winter due to low snow cover.
Growth is limited due to low nutrients associated with mainly organic soils (Chapin and Shaver 1985). To some extent, plants can ameliorate these harsh conditions by growing low and close together to resist the effects of wind and blowing snow. As plant height increases with warming temperatures (Bjorkman et al. 2018a), two groups of plants would be expected to show increased leaf toughness. Graminoid leaves would require greater stiffness to support themselves, while leaves of evergreen shrubs would be exposed to higher wind speeds and more blowing snow.
Rapid change in arctic ecosystems and resulting changes to plant communities are likely to change leaf toughness across the tundra biome. The expected increase in Arctic temperature by 6 2-3°C in the winter and 1°C in the summer (Chapman and Walsh 2007) could have regional and global implications leading to increased decomposition rates of organic matter within frozen soil and a deeper active layer. If more nutrients are made available as a consequence (Mack et al. 2004;Hewitt et al. 2020), leaf toughness may be altered, as faster growth may be favored over the production of structural compounds. On the other hand, the ecosystem feedbacks from climate change are complex (Wookey et al. 2009), and each may influence leaf toughness in different ways. Therefore, it is important that we systematically assess leaf toughness in relation to a variety of different environmental factors in order to project its change into the future.
We address the following questions: 1) How does leaf toughness vary between common tundra plants? 2) Is there latitudinal, seasonal, or yearly variation in leaf toughness? 3) Is there a difference in leaf toughness between tussock tundra and dry heath? 4) Is there a difference in community-weighted mean toughness between vegetation types? 5) What is the effect of added nutrients on leaf toughness? 6) How much does intraspecific variation contribute to variation in leaf toughness within communities?

Study Design
Leaves were collected from 16 different locations in northern Alaska in two arctic vegetation types: tussock tundra and dry heath (Table 1). Tussock tundra consists of 15-25 cm diameter tussocks of Eriophorum vaginatum growing in peaty soils at a density of 4-7 tussocks/m 2 with evergreen and deciduous shrubs, mosses, and lichens growing in and between the tussocks (Wein 1973, Fetcher andShaver 1982). Dry heath consists mostly of very small evergreen and deciduous shrubs and is found on exposed ridges and fell fields with rocky soils 7 with a very thin surface layer of organic soil (Shaver and Chapin 1991). Samples of fully developed leaves were collected on three occasions: in June 2015, from June 2016 until mid-September 2016, and in July 2017. Fifteen sites were sampled in tussock tundra and dry heath along a latitudinal gradient from 66ºN to 70ºN. In addition to sites in undisturbed tussock tundra, we sampled a site in the Anaktuvuk Burn, which was a large fire that occurred in (Jones et al. 2009).
We examined the effect of species and climate on leaf toughness in tussock tundra (Questions 1, 2) measured at twelve sites in late June and early July for five widely distributed species, Betula nana, Carex bigelowii, Eriophorum vaginatum, Rhododendron tomentosum, and Vaccinium vitis-idaea. (Table 1). This analysis from twelve sites had only one deciduous species (B. nana). To include more deciduous species for purposes of comparison, we sampled three additional common species, Rubus chamaemorus, Salix pulchra, and Vaccinium uliginosum, at three sites along with the five species listed above (Table 1).
To determine seasonal and year-to-year variation (Question 2), we focused on E. vaginatum because it produces leaves at different times of the season. We sampled seasonal variation from early mid-June through mid-September at three sites (Table 1). At the end of the season (24 August 2016 -15 September 2016), we compared leaf toughness of green and senesced leaves at Coldfoot, Toolik Lake, and Sagwon. To examine variation between 2015 and 2016, we used data collected from mid-June through mid-July at five sites (Table 1).
For comparing leaf toughness between vegetation types (Question 3), we used three sites that had both tussock tundra and dry heath vegetation (Table 1). A fourth dry heath site at Atigun Pass was paired with the nearby tussock tundra site at Atigun Camp for purposes of analysis.

8
To calculate community-weighted mean leaf toughness (Question 4), we used biomass values for tussock tundra and dry heath from a harvest that was carried out in 2006 by the Arctic Long-Term Ecological Research Project (Gough 2017). The tussock tundra site was 1.5 km from the site that we used to collect leaf toughness samples, while the dry heath site was 2.5 km. Both sites had the same suite of species in similar proportions as the sites sampled for leaf toughness.
All the aboveground tissue was clipped from ten 20 cm x 20 cm quadrats in both tussock tundra and dry heath and sorted by species and tissue type. Tissues were dried at 65C and weighed to give biomass for each species in the quadrat. Leaf toughness values for species from tussock tundra and dry heath were multiplied by the biomass of the same species in a quadrat to calculate community-weighted mean leaf toughness for ten quadrats in each vegetation type. We did not have leaf toughness values for all of the species in the biomass samples. For tussock tundra the biomass of the missing species was a small percentage (5.9%) of the total, while for dry heath, the biomass of missing species was a larger fraction (25%) that consisted mostly of Empetrum nigrum and Loiseleuria procumbens, the leaves of which were too small to be tested in our apparatus.
To determine the effects of added nutrients on leaf toughness (Question 5), we sampled tissue from an experimental site at Toolik Lake that was fertilized for 27 years with 10 g m -2 yr -1 nitrogen and 5 g m -2 yr -1 phosphorous and compared it with a control site that was part of the same experiment. The experiment used a randomized block design with four blocks.
To determine abundance-weighted interspecific and intraspecific trait variance variation (Question 6) in leaf toughness (de Bello et al. 2011, Siefert et al. 2015, we used data from tussock tundra and dry heath at Toolik Lake, since that was the only site with abundance data. To determine the partitioning of variance among site, species, and population (Messier et al. 2010), 9 for tussock tundra, we used data from 13 sites and eight species, while for dry heath we used data from four sites and eight species. (Table 1).

Sample processing
We sampled one leaf from all the species from the list in Table 2 that could be found in an area with a 1.5 m radius. For E. vaginatum, three tillers per tussock were collected. Once the first set of samples was gathered, we moved 5 m to the next sampling area to avoid gathering replicate genotypes. To sample the population of each species at each site, we obtained ten individual leaf samples, which were immediately stored in a plastic zip-lock bag with a moist towel to prevent desiccation.
Samples were processed no longer than 2 hours after collection. A penetrometer with a strain gauge (Imada, Inc., Model DS2-11) and a punch and die of 2 mm was used to measure the force to puncture the leaf avoiding the midrib, except for leaves that were too small that the midrib could not be avoided. The maximum force to punch (F p ) for each sample was recorded.
Leaf toughness can be expressed by the maximum force required to punch the leaf divided by the circumference of the punch (L) (Onoda et al. 2011). The leaves of E. vaginatum were less than 2mm wide (0.4-1.3 mm), so an optical comparator was used to gauge the width (W). To determine the length (L) of leaf tissue that was punched, we calculated the length of the two chords subtended by the width as given by the formula 4 × arc sin(W/2). Raw data are available in the TRY database (https://www.try-db.org) under the name Leaf Toughness of Alaskan Arctic Species in Natural Sites.

Statistical analyses
We used the JMP package (SAS Institute, 2019) to perform analysis of covariance with mean temperature from 1977 through 2017 for June and July downloaded from the SNAP data 10 archive (https://uaf-snap.org/get-data) as a measure of the environment that combined differences in latitude with differences in elevation (Questions 1-2). The model for both the five species and eight species analyses was where F p /L ij is leaf toughness of species i at site j, S i is species i, and Tj is temperature at site j. F p /L was log-transformed to stabilize the variance over a broad range of leaf toughness.
Linear regression with day of the year as an independent variable was used to evaluate changes in leaf toughness of E. vaginatum at Coldfoot, Toolik, and Sagwon during the growing season of 2016 (Question 2). The analysis was performed on each site separately. To determine the effect of leaf senescence on leaf toughness, we used two-way analysis of variance with site and green vs. senesced as factors. We also used two-way analysis of variance with site and year as factors to evaluate differences in leaf toughness between 2015 and 2016. Tukey's HSD test was used to determine differences at each site between green and senesced leaves and between 2015 and 2016. Analysis of variance tables are available as Supplementary File 1.
To determine the effect of vegetation type (Question 3), we used two-way analysis of variance with site and vegetation type as factors to analyze the response of leaf toughness for the graminoid C. bigelowii, the evergreens R. tomentosum and V. vitis-idaea, and the deciduous B.
nana and S. pulchra, which were the species that were found in both types. To compare community-weighted mean toughness for tussock tundra and dry heath at Toolik Lake (Question 4), we used a t-test assuming unequal variances.
To analyze the effect of fertilization on leaf toughness of seven species of tussock tundra (Question 5), we used a randomized block ANOVA with fertilizer as a fixed effect and block as a random effect.

11
The community-weighted means at Toolik Lake were also used to estimate intraspecific vs.
interspecific variation within communities (wITV), according to the variance partitioning approach of de Bello et al. (2011) that takes into account the relative abundance of species.
To estimate the partitioning of variation in leaf toughness in tussock tundra between site, species, and population within species and site (Question 6), we used variance component analysis with the varcomp procedure from the ape package in R (R Development Core Team 2016) to analyze the log-transformed data (Messier et al. 2010, Umaña andSwenson 2019). We used a bootstrap procedure to estimate 95% confidence intervals by resampling our dataset of the 1210 observations 700 times.

Variation between species and sites
Most of the variation was explained by differences in species, with the graminoid E.
vaginatum having the greatest values for leaf toughness, followed by the evergreen V. vitisidaea, the graminoid C. bigelowii, the evergreen R. tomentosum, and the deciduous B. nana. In addition, there was a significant (p< .001) interaction between June-July temperature and leaf toughness of different species (Fig. 1). Leaf toughness of E. vaginatum and B. nana increased significantly (p< .0001) with increasing temperature, whereas that of the other species did not change significantly.
We sampled additional deciduous species at three sites, No Name Creek, Toolik Lake, and Sagwon. Eriophorum vaginatum had the toughest leaves, followed by V. vitis-idaea (Table 3).
Rhododendron tomentosum and C. bigelowii had similar values for leaf toughness, while the four deciduous species had the lowest values (Table 4). There was considerable variation in the deciduous group, with V. uliginosum having almost twice the mean toughness as R. chamaemorus (Table 3).
In the dry heath, two evergreen species, V. vitis-idaea and R. tomentosum, had the toughest leaves along with C. bigelowii (Table 4). The leaves of the other evergreen species, Dryas octopetala, were not as tough as V. vitis-idaea and R. tomentosum.

Variation through season
We found no significant change over the growing season in E. vaginatum until the leaves started to senesce in August. Green leaves at Coldfoot were significantly (Tukey's HSD, p < 0.05) tougher than senesced leaves (5.63 kN m -1 vs. 4.73 kN m -1 ). At Sagwon and Toolik Lake, there was no difference.

Variation between years
Leaf toughness of E. vaginatum at the Anaktuvuk Burn increased from 3.66 kN m -1 in 2015 to 5.00 kN m -1 in 2016 (Tukey's HSD, p < 0.05), whereas it did not change significantly at the other sites, Sagwon, Coldfoot, Galbraith, and Toolik Lake.

Variation between vegetation types
Leaf toughness was lower in tussock tundra than in dry heath for all of the species except for R. tomentosum (Fig. 2). For C. bigelowii and S. pulchra, there was a significant interaction (p < 0.05) between site and vegetation type, but in all cases, the tussock tundra type had a lower value than the corresponding dry heath value.
Although some species appear to have greater toughness in dry heath than in tussock tundra, this does not address the question of whether community-weighted mean leaf toughness is different between the two vegetation types. There was no significant difference between mean toughness in tussock tundra and dry heath (Fig. 3), but variance was much higher in tussock tundra due to the presence of E. vaginatum in some quadrats but not others.

Effect of nutrient addition
Fertilization with nitrogen and phosphorous affected the species differently. There was no significant difference between fertilized and unfertilized plots for B. nana, E. vaginatum, R. tomentosum, and R. chamaemorus (Table 5). On the other hand, leaf toughness of C. bigelowii, V. vitis-idaea, and S. pulchra was significantly lower in the fertilized plot (Table 5).

Interspecific versus intraspecific variation in leaf toughness
Intraspecific variation in leaf toughness (wITV) in tussock tundra was 8.4% of the total variation, whereas in dry heath it was 9.2%, and the mean over both communities was 8.8%. The analysis of variance components for tussock tundra found that the variance within populations was less than in dry heath (Table 6). For tussock tundra, the proportion of the variance that was due to populations within species was almost the same as the proportion due to species. Results for dry heath were similar, although the proportions were smaller, since there was more variation within populations. For both vegetation types, the proportion due to site was 0%

Discussion
We found that leaf toughness in common plants of the Low Arctic varied across several axes. Most of the variation was between species, with E. vaginatum requiring more than five times the force per unit length than R. chamaemorus. Leaves of evergreen species were tougher than deciduous species. In this respect, the general pattern of leaf toughness follows the leaf economics spectrum (Wright et al. 2004), although the fit is imperfect. For example, the graminoids, E. vaginatum and C. bigelowii, had some of the toughest leaves, even though the 14 leaves senesce and die back in the fall. Furthermore, there was considerable variation within the evergreen group; leaves of V. vitis-idea were considerably tougher than those of R. tomentosum.
As the Arctic warms, deciduous shrubs are expected to become a larger component of the vegetation (Tape et al. 2006;Myers-Smith et al. 2011;Myers-Smith and Hik 2018). Thus, we can expect a community-level trend towards less tough leaves as deciduous shrubs increase, especially if they replace graminoids. Changes in community leaf toughness could have effects at higher trophic levels. Eriophorum vaginatum and C. bigelowii are among the toughest of the measured plants and contain higher amounts of biogenic silica than most other tundra species (Carey et al. 2017), which deters herbivory (Massey et al. 2007(Massey et al. , 2009). Further, Betula nana is highly favored by insect herbivores (Belsing 2015;Metcalfe 2019), while Salix pulchra is favored by caribou (White and Trudell 1980). Therefore, changes in plant community composition could have knock-on effects on herbivory through changes in community leaf toughness.
Environmental factors, including site temperature, vegetation type, and site fertilization, also contributed to variation. For most of the species that we studied, leaf toughness declined with increasing site temperature, which is consistent with the findings of Onoda et al. (2011) andKandlikar et al. (2018). But leaf toughness of E. vaginatum and R. chamaemorus increased with site temperature. Leaves of E. vaginatum south of timberline are longer than those farther north (Shaver et al. 1986, Fetcher andShaver 1990), so the additional toughness may be a consequence of having to maintain the leaves erect.
It is not surprising that leaf toughness did not vary through the growing season since once a leaf is developed, it is not likely to change characteristics until senescence. Year-to-year variation was not significant, except for the Anaktuvuk Burn site. The beginning of the growing 15 season on the North Slope in 2015 was much warmer than in 2016 . Because the Anaktuvuk Burn site was completely burned in (Jones et al. 2009), E. vaginatum is growing more vigorously there even though the mean temperature in June and July is similar to that of Toolik Lake (Table 1). Colder temperatures in June may have reduced the rate of leaf expansion at the beginning of the growing season, thereby producing tougher leaves.
As the landscape becomes more dominated by shrubs and leaf toughness declines, decomposition rates may increase. Leaves of the species with one of the lowest values for toughness in our study, B. nana, had the highest rates of decomposition in a study carried out in tussock tundra at Toolik Lake (McLaren et al. 2017). It was followed by R. tomentosum, V. vitisidaea, and E. vaginatum in that order, which is the inverse of their ranking in leaf toughness ( Fig. 1, Table 3). This pattern seems to hold for other ecosystems as well. Leaf tensile strength was negatively correlated with loss of litter mass for the plants of various life forms, including herbaceous dicots, woody dicots, and graminoid monocots from both Argentina and Great Britain (Cornelissen et al. 1999). This pattern was also found for leaf toughness as measured by a penetrometer and the rate of decomposition of leaves of tropical trees in the Malaysian rain forest (Kurokawa and Nakashizuka 2008). On the other hand, both evergreen and deciduous shrubs contain substantial amounts of woody tissue, which decomposes slowly. Thus, increases in shrub biomass may lead to an increase in the proportion of woody tissue and a reduction in community-weighted mean rates of decomposition (Hobbie 1990).
Individual species had tougher leaves in the dry heath than in the tussock tundra. Plants in the heath are more exposed to wind and abrasion by blowing snow and thus may benefit from having tougher leaves. The community-weighted mean was not different, however, because of the presence of E. vaginatum in the tussock tundra. Not only did it increase the variance, but it 16 raised the overall mean for the community because it has the highest values for leaf toughness.
Given the relationship between leaf toughness and decomposition, it seems reasonable to propose that mean decomposition rates might be similar in dry heath and tussock tundra.
In a warming Arctic, nutrients are generally considered to become more available (Mack et al. 2004;Schmidt et al. 2002, but see DeMarco et al. 2014). This may affect leaf toughness by increasing specific leaf area (Knops and Reinhart 2000). Fertilization experiments have resulted in increased productivity and biomass accumulation in tundra plants (Shaver and Chapin 1980). In our study, leaf toughness decreased after fertilization for three tundra species, while in four others, it was unaffected. In spinach, leaf toughness, as measured by a punch test, was negatively correlated with the amount of nitrogen provided, perhaps because the unfertilized leaves had smaller cells than the fertilized leaves (Gutiérrez-Rodríguez et al. 2013). In rainforest in Costa Rica, fertilized tree seedlings had lower leaf toughness than unfertilized plants as measured by a penetrometer in 20% shade but not in 2% shade (Nichols-Orians 1991). On the other hand, fertilization did not affect leaf toughness of dogwood and tulip poplar seedlings growing in an old field and secondary forest in Georgia (Dudt and Shure 1994). In contrast, it increased toughness for alder (Alnus crispa) and poplar (Populus balsamifera) from interior Alaska while it remained unchanged for willow (Salix alexensis) and birch (Betula papyrifera) (Irons et al. 1988). Although results vary from study to study, the general trend is for fertilization to decrease leaf toughness or leave it unaffected. Because the tundra is strongly nutrient limited (Shaver and Chapin 1980), increased nutrient availability is likely to reduce leaf toughness, at least for some species.
In the global meta-analysis by Siefert et al. (2015), the overall mean for intraspecific trait variation within communities (wITV) was 25%. In our study, wITV for leaf toughness was less than 10%, which is similar to the range reported for leaf length by Siefert et al. (2015) and much less than the value of 18% reported for the single study of leaf toughness in their survey. Thus, most of the variation in leaf toughness within a tundra plant community is due to differences between species. Such low values of wITV may result from a lack of plasticity in tundra plants as a consequence of adaptation to a stressful environment (Umaña and Swenson 2019).
Many arctic plant species have extensive distributions, which often leads to trait variation between populations and local adaption that results in ecotypes (Fetcher and Shaver 1990;Bennington et al. 2012). The analysis of variance components showed that the species and population levels accounted for similar amounts of the total variance., while there was little variation due to the latitudinal gradient. This result was similar to that obtained for four of the seven traits studied by Umaña and Swenson (2019). In their study, leaf carbon content, leaf nitrogen content, 13 N, and leaf area showed little variation due to elevation and similar amounts due to species and population. The greater amount of variation due to population may be the result of plastic or genetic adjustment to local conditions (Umaña and Swenson 2019). The amount of the variance within populations was lower for tussock tundra than for dry heath, possibly reflecting lesser heterogeneity within the tussock tundra vegetation type as well as the large influence of E. vaginatum on leaf toughness.
Leaf toughness is measured less frequently than many other functional traits such as specific leaf area. Nevertheless, it integrates several structural and cellular traits with significant implications for understanding the ecosystem processes of herbivory and decomposition.