To our knowledge, we are the first to combine both litter traits and spectral data of six common alpine plant species; three of them even over the course of 24 months of decomposition. Generally, more litter decomposition studies – ideally multiple-method studies like this one – are needed to better understand decomposition patterns of multi-species litter, not only in alpine ecosystems. The investigated Alpine decomposition patterns were in line with our expectations based on previous studies (e.g. Rief et al. 2012) and the literature as in this study recalcitrant SHRUB_Vv litter showed the slowest and both high-quality GRASS_Dg and FORB_Gs litter a significantly faster decomposition (Montané et al. 2010; Duan et al. 2013; Berg 2014). Further, we found both synergistic and antagonistic non-additive effects on LML in litter mixtures, similar to the few-species mixtures in the study of Duan et al. (2013). Analyses of the NIR spectra of litter material successfully separated litter types, functional groups as well as the decomposition stages. However, reliable predictions were possible only for the total C content due to methodological constraints.
Near-infrared reflectance spectra and partial least squares predictions
The successful prediction of soil properties by using NIRS such as total C and N content, soil moisture, and others, has been shown by Chang et al. (2001), resulting in correlations of measured and predicted values that exceeded R2 > 0.80. In this study, we successfully validated our PLS model with the initial litter material. Further, we were able to validate PLS predictions for total C and total N content, and thus also for the C:N ratio of the decomposed SHRUB_Vv, GRASS_Dg, and FORB_Gs litters (Table A2). These parameters were measured for all the litter material along the entire decomposition experiment. However, when attempting to predict the other litter traits of the decomposed from the freshly fallen litter material, we encountered difficulties in terms of extremely high deviations of the predicted values.
Generally, litter traits considerably change during the decomposition process (Cornelissen and Thompson 1997; Berg 2014). Therefore, the PLS model which was built on the initial litter material (i.e. 0 months) was not able to predict the litter traits of the decomposed litter (6 to 24 months). In other words, PLS predictions work properly only for the same or very similar kind of “product” (CAMO Software 2017). We assume that, as the decomposed litterbag material differed significantly in their spectral composition after 6, 12, and 24 months in the field (Fig. A3 and A4), also their litter traits differed too much to be classified as similar “product”. However, we believe that measuring material at each decomposition stage and use that data for building corresponding PLS models would result in highly accurate predictions, as we could already prove for the total C and N contents.
Thus, NIRS offers a powerful tool in soil ecology and in decomposition studies (Albrecht et al. 2008). According to our experiences with NIRS and the PLS predictions of alpine litter traits along the decomposition process of 24 months, we advise everyone to collect and estimate generously the needed initial freshly fallen litter material and to be aware that both the litterbag approach and the NIRS measurements need a sufficient amount of litter material left at the end of the decomposition experiments (at least 20 g air-dried mass).
Leaf litter traits
For freshly fallen litter material, we found four out of the six alpine plant species to be significantly different in their litter traits and therefore well separable (Fig. 2). Our results are in line with the study of Rief et al. (2012), which was conducted at the same study site and compared the palatability of eight alpine plant species to invertebrate decomposers; five of them were also used in our study. Regarding decomposability, the plant components can be divided into three groups: rapidly degrading carbohydrates and proteins (the latter expressed here as N), slower degrading structural components such as cellulose and hemicellulose (ADF), and recalcitrant lignin and polyphenols (ADL) (Berg and Laskowski 2005).
The three litterbag species, each belonging to a different functional groups, differed significantly in their initial chemical traits (Fig. 2). The evergreen dwarf shrub V. vitis-idaea (i.e. SHRUB_Vv) can be found mainly in abandoned pastures at the study site (Steinwandter et al. 2017) along with other Vaccinium species; it is wide-spread in natural alpine heathland of Eurasia (Meusel et al. 1978) and North America (Flora of North America Editorial Committee 2009). The freshly fallen SHRUB_Vv litter was characterised by the highest lignin content (i.e. ADL), which was also reflected by a high total C content that exceeded 50% of their total litter biomass; on average, lignin is composed of 65% of carbon (Montané et al. 2010). Additionally, freshly fallen litter of Vaccinium spp. and other dwarf shrubs was reported to contain higher amounts of secondary metabolites such as phenols and tannins compared with grass and forb species (Montané et al. 2010; Berg and McClaugherty 2014), thus being responsible for low palatability to large decomposers such as earthworms (Rief et al. 2012). These recalcitrant compounds impair the establishment and growth of microorganisms by forming resistant complexes with proteins and nitrogen (Hättenschwiler and Vitousek 2000), further inhibiting and regulating decomposition processes. However, Rief et al. (2012) reported that after 12 months in the field, the total phenolic concentration in aged dwarf shrub litter decreased considerably (e.g. approximately 90%), most likely making aged SHRUB_Vv litter again more attractive to the decomposer community.
We found a relatively low total N content in freshly fallen litter types, with SHRUB_Vv exceeding that of FORB_Gs but not GRASS_Dg. Generally, a high N concentration was found to boost the decomposition of hemicellulose (i.e. ADF) in the first phase of decomposition, but to hamper that of lignin (i.e. ADL) in the long term, thus leading to lower decomposition rates (Berg 2014). This might be true for the SHRUB_Vv litter material where we found high initial lignin:N, lignin, and N values (Table 1). Further, as the total C content of SHRUB_Vv treatments was still high after 12 and 24 months in the field, it might indicate that these litter still contained large concentrations of lignin (Berg 2014). Personal observations confirm this, as tough and recalcitrant Vaccinium spp. leaves of different decomposition stages and ages could be found in the litter layer at our study site (see also Seeber and Seeber 2005). The lignin:N ratio was found to be a good predictor of decomposability in tropical and Mediterranean climates, but not in the temperate region (Aerts 1997). Even though we did not test it, our data may indicate that lignin:N could be a useful and applicable predictor for the decomposability of alpine litter.
Regarding the C:N ratios that can be used as index for palatability and litter quality (Amelung et al. 2018), we obtained relatively high values for all freshly fallen litter types ranging from 30 to almost 50. This would suggest the initial litter material to be generally of low-quality mainly due a relatively low total N versus a high total C content (Amelung et al. 2018). However, the SHRUB_Vv C:N ratios decreased only slowly along decomposition, therefore becoming only slowly more favourable to the decomposer community. On the other hand, the C:N ratios of perennial GRASS_Dg and FORB_Gs significantly dropped after six month, therefore becoming more favourable to decomposers already in the early stages.
The speed and rates of decomposition for SHRUB_Vv, GRASS_Dg, and FORB_Gs were in line with our expectations and the literature (Montané et al. 2010; Duan et al. 2013; Berg 2014). All the litter material degraded rapidly in the beginning followed by gradually decelerating decomposition. SHRUB_Vv as an evergreen plant showed the lowest decomposition rate, while the perennials GRASS_Dg and FORB_Gs degraded significantly faster. Compared with studies from the lowland, we observed lower decomposition rates most probably due to the harsher environmental conditions at our Alpine study site (Gavazov 2010). Generally, decomposition rates highly depend on climate (i.e. temperature and (soil) moisture), the composition of the litter material (i.e. nutritional quality and physical toughness), site-specific local factors, and the decomposer community (Cornelissen and Thompson 1997; Berg and McClaugherty 2014). A fast decomposition can be observed when both temperature and moisture are high, as found in the tropics (Aerts 1997; Bradford et al. 2016). However, cold climates – for example high-elevation and high-latitude environments – only partly fulfil these criteria, as they are characterised by high amounts of precipitation but low temperatures. Additionally, snow can persist for a long period and can return in summer months (Körner 2003). In these cold biomes, temperature, when not limited by soil moisture (Aerts 2006), was found to be the main driver of decomposition rates (Gavazov 2010; Bradford et al. 2016) along with other factors such as the decomposer community composition (Gessner et al. 2010; Kitz et al. 2015). As larger decomposers were excluded in this study, the observed decomposition rates can be accounted solely to the leaf-litter associated microbiome (Knapp et al. 2011).
Interestingly, litter decomposition was found to be quite high and active under the snow cover (Sjögersten and Wookey 2004; Baptist et al. 2010), due to the insulation of the snow pack that keep temperatures close to the freezing point. We recorded decomposition rates up to 42% LML after the first six months (Table 2) when the litterbags were covered by a 30 cm thick snow cover for most of the time, which is in line with the literature (Saccone et al. 2013).
Non-additive effects are defined as effects that are greater or less than the sum of two or more net (i.e. additive) effects (Hättenschwiler et al. 2005), the former being called synergistic and the latter antagonistic. We observed both non-additive effects, depending on the functional groups present, with an inconsistent pattern at the beginning of the decomposition processes and a synergistic trend after six months. This is consistent with Duan et al. (2013) who studied litter mixtures of up to 25 species in alpine meadows on the Tibetan Plateau and found antagonistic as well as synergistic effects in few-species mixtures, especially when shrubs were included. However, with an increasing number of litter types mixed with shrubs, mostly synergistic effects were recorded. In our study, GRASS_Dg showed the highest synergistic effects of all treatments followed by the SHRUB_Vv, hinting that adding at least one other functional litter type already increases the overall decomposition rate considerably (Fig. 2). The effect was even more pronounced for SHRUB_Vv and GRASS_Dg in the 3-species mixtures with a highly significant effect for GRASS_Dg + SF. Cuchietti et al. (2014) reported that high-quality litter can stimulate the decomposition of low-quality litter and vice-versa, hence benefit the decomposability of all interacting litter types. Our results for SHRUB_Vv and GRASS_Dg in mixtures fit into this scheme as both litter types significantly accelerated each other’s decomposition (i.e. SHRUB_Vv + G, GRASS_Dg + S, and GRASS_Dg + SF). The reason might be a nutrient transfer from the high to the low-quality litter, leading to an elevated attraction of the species-specific microbiome (Hättenschwiler et al. 2005). Rief et al. (2012) found well separable communities of microorganisms inhabiting litter of eight alpine plant species from the same study site, which also included five species investigated from our study. A different picture was found for FORB_Gs, for which only adding SHRUB_Vv slightly increased decomposition (Table 2 and Fig. S6). In fact, we assume that in the first phase of decomposition GRASS_Dg and FORB_Gs negatively influence each other, the latter hampering decomposition in mixtures more severely. However, FORB_Gs had the highest mean decomposition rate of all tested litter types, therefore adding one or two litter types might slow down the decomposition processes due to a deficit of N and/or P (Zheng et al. 2017).
The finding that alpine low-quality dwarf shrub litter decomposition is accelerated when present in mixtures with high-quality litter types is of actual interest regarding ongoing land-use changes in Alpine pasturelands (Schirpke et al. 2013). These were shaped by extensive low-input agriculture for centuries, but are increasingly taken out of management due to socio-economic reasons (MacDonald et al. 2000), although being of high conservation status in the Habitats Directive (European Commission 1992). In sub-alpine pastureland, abandonment is followed by dwarf shrub encroachment, aggravated by global warming (Brigham et al. 2018), leading to high amounts of litter accumulation (Seeber and Seeber 2005; Steinwandter et al. 2017) and slow rates of decomposition (Montané et al. 2010), with mostly negative effects on various ecosystem services. However, these effects could be partly mitigated by the observed synergistic effects when low-quality litter occurs in mixtures with high-quality litter and by generally higher net-decomposition rates induced by global warming (Cornelissen et al. 2007; Luo et al. 2010; Duan et al. 2013).