Discussion
Contrastingly to the common paradigm that the role of ecosystem engineers is most pronounced in harsh abiotic conditions (Crain and Bertness 2006), our results generally show alimited impact of long-tailed marmots on N and P content in plant biomass and no impact on vegetation cover in an extremely arid, cold ecosystem.
A decrease in vegetation cover due to burrowing animal activity is mostly caused by the presence of mounds close to the burrow entrance, which have not been colonised by plants. (Dotter, 2009; Louw et al., 2019). Our results showed no possibility of plant regrowth on mounds in the present conditions in the studied area. Therefore, we excluded mounds from the analysis of vegetation cover and tried to answer the question, of whether marmots modify plant recruitment on actually inhabitable soil. We found high variability in the parameters we measured using aerial images, which is the reason why we decided to perform the analysis on a small scale and numerous repetitions, analysing each 20 m buffer around a burrow separately. Based on our results, the response of vegetation cover to the proximity of burrows could be positive, negative, binomial (with higher cover at highest and lowest distances from the burrow entrance) or non-existing (Fig. 2). This could suggest that those types occur depending on the balance between herbivory and nutrient supplementing through excretion. However, the relationship is probably best analysed by visual inspection of the distribution of burrows relative to vegetation cover (Fig. 1): patches of high cover occur along the slope of the river terrace, and probably reflect past paths of water flow. Marmot burrows were located both on and off these patches, and there are large areas with high vegetation cover and no burrows. The type of relationship between the distance from the nearest burrow and vegetation cover we found using the linear model depends probably on the location of the burrow relative to a vegetation patch. If it is located outside the patch, but the patch falls within the 20 buffer, then the trend is negative. When the burrow is in the patch, but the buffer reaches an area of bare soil, the trend is positive (see Fig. 1D). The edges of the vegetation patches are sharp and seem unaltered by burrow presence. The distribution of burrows is clearly not random across the searched part of the valley (Fig 1A) and can be an effect of selectivity towards physical soil properties, which we cannot account for. Marmots seem to avoid alluvial fans, especially active areas, a large part of the second river terrace and the middle part of the third terrace. Based on this, we assumed that there is no clear relationship between burrow presence and vegetation cover.
Given the limited effect of the additional nutrients on plants (see below), the slow growth of plants and the fact that marmots graze almost exclusively in the burrow area (Blumstein, 1998), it is puzzling that vegetation cover does not seem impacted by burrow proximity. Some burrow-dwelling mammals are known to create large patches of bare soil due to herbivory and trampling (Davidson et al., 2012; Harris et al., 2015; J. Zhao et al., 2021). We cannot explain this result, as we do not know the primary production of plants, the population size of marmots and their feeding requirements, as well as the time spent in different burrows. The only study on the impact of long-tailed marmot (or any other burrow-dwelling species) activity on plants conducted in a comparable environment in Eastern Pamir, found a small impact of the presence of marmot burrows on plant species composition, and the results varied among study sites. Generally, dwarf shrubs occurred more often in disturbed areas than in undisturbed areas, as opposed to herbs (Dotter, 2009), which suggest that browsing does impact the plant community, as shrubs are more resistant to herbivory.
Our results on the relationship between the distance from the burrow and N and P content in plant biomass are ambiguous. Only one out of six species, E.altaicum , showed an increase in biomass N and P content with increasing proximity of the burrow entrance. However, the effect size in this species is enormous, with N content increasing by 80% and P content by 250% with decreasing distance from the burrow (Fig. 4). This implies that there is a nutrient input caused by marmots, and so do the result of analyses of soil sampled within the high cluster (Fig. 1A). It had almost 100 times higher P (0.31 mg/kg) 10 times higher and N (0.169%) content than from other parts of the valley (0.47 – 0.125 mg/kg and 0.01 – 0.06%, respectively), which the authors of this study attribute to marmot activity (Kabala et al., 2021), as the sample was collected within the high cluster, close to several burrows (Fig. 1D). This allows us to treat the mound area as a natural fertilization experiment and interpret it as such, with necessary caution. The limited impact of nutrient addition on vegetation cover and biomass N and P content suggests, that plants are not limited by nutrient availability, but probably by abiotic factors, mainly water scarcity, as they simply lack the necessary moisture to utilize the additional nutrients (He & Dijkstra, 2014), and the effect of nitrogen addition was proven to be correlated with precipitation (Yahdjian et al., 2011). Burrowing animals are known to increase water availability for plants, as their constructions promote water infiltration and prevent water runoff and evaporation (Laundre, 1993; Whitford & Kay, 1999). However, this probably does not occur in our study area. Most of the precipitation is presumably instantly lost to evaporation, irrespective of the presence or absence of marmots. In turn, plants are adapted to these conditions and have deep roots, up to 50 cm (Kabala et al., 2021), to utilize any soil moisture. Even if there was an increase in water infiltration in the surface layer of the soil, it could probably not be used by plants. The response of E.altaicum to the marmot-derived nutrient input implies that this species, apart from being capable to utilize the additional N and P, is more efficient at water uptake and/or preventing water loss than other species. However, we lack information about the functional traits of plant species from our study, so we are not able to confirm this hypothesis.
The results of N:P stoichiometry in the distance gradient from the burrow entrance show a shift from P limitation at higher distances to N limitation close to the burrow. The use of the N:P ratio to identify the limiting nutrient, where N:P < 14 indicates N limitation, N:P < 16 P limitation and values in between indicate co-limitation has been prosed for wetlands (Koerselman & Meuleman, 1996) and proven useful in grasslands (Craine & Jackson, 2010; G. Zhao et al., 2017). We are using this tool with caution, because the high N concertation in plants inhabiting alpine habitats, and the resulting high N:P ratios (such as those we found in most species) are mainly caused by slow growth and adaptations to low temperatures, not by P limitation (Körner, 2003). Those adaptations are species-specific and most probably site-specific, therefore we are only looking at values within E. altaicum , and they indicate, that faeces are a more important source of P than N. Faeces have similar N content as plant biomass, whereas the P content is almost three times higher, indicating that they are a better source of P than plant litter. Soil P content can be also increased by the presence of marmot bones, which we have observed on several mounds. Up to 30% of marmots fail to survive the hibernation period (Blumstein & Arnold, 1998), their remains are probably ejected on the surface after hibernation or during burrow maintenance. Plants growing on initial soils are usually not P-limited, as it is abundant from weathering minerals, as opposed to N (Walker & Syers, 1976). However, in high mountain continental areas, the low temperature and precipitation can hamper weathering to the point that plans are limited by P shortage (Darcy et al., 2018).
Increased N and P content in plant biomass close to burrows are mostly caused by the input of organic matter by the burrowing animal, but it can be also an indirect effect of animal activity: (1) Disturbed soil has higher temperatures and is more susceptible to N mineralization, increasing the amount of plant-available soil N (Whicker & Detling, 1988) (2) Defoliation (including grazing) may increase N uptake by plants (Jaramillo & Detling, 1988; Yan & Lu, 2020). Therefore, we decided to use stable isotope composition as a marker of animal-derived N use by plants. Faeces of rodents are enriched in 15N compared to diet by 1.4 – 2.5‰ (Hwang et al., 2007; Sare et al., 2005) due to its discrimination in metabolic pathways. Given that long-tailed marmots mainly stay close to the burrow while foraging (Blumstein, 1998), we hypothesised, that the elevated δ15N values in faeces and the enhanced N cycling in burrow proximity would result in higher δ15N in plants growing closer to the burrow entrance. The lack of difference between δ15N values of plants and faeces in most cases (Fig. 5) implies little isotopic enrichment of the latter. Results on isotopic enrichment (trophic discrimination) in faeces are based on feeding experiments, where food is usually provided ad libitum. Fasting and nutritional stress, which is frequent in the natural environment, is known to affect δ15N values of animal tissues (Doi et al., 2017; Hertz et al., 2015) and could also alter the metabolic pathways leading to faeces enrichment in 15N. Previous studies found no clear pattern of stable isotope ratios in diet and faeces of burrowing animals. Plains vizcacha (Lagostomus maximus ) faeces had lower δ15N values by approx. 4‰ than soils and shrubs at burrow systems, with no interpretation by authors (Villarreal et al., 2008). Grasses close to arctic fox (Vulpes lagopus ) dens had similar δ15N values to grasses from control sites, despite the significant impact of fox burrows on the available N pool in soils (Gharajehdaghipour et al., 2016). Authors attributed the loss of animal-derived 15N to discrimination during mineralization and plant uptake (where high N availability could further increase discrimination). Other studies showed an increase in δ15N in plants at burrows, proving the fertilizing effect of animal faeces (Ben-David et al., 1998; García et al., 2002). However, those studies were done on burrows of predators, which have higher δ15N values in tissues and faeces due to longer trophic chains (Fry, 2006). In such a case, when groups involved in the relationship are separated by several trophic steps, each with its specific isotopic enrichment (Vander Zanden & Rasmussen, 2001), the isotopic difference and the resulting method resolution is high enough.
A mechanism typical for the interaction between burrow-dwelling animals and plants which does not occur under the circumstances we studied is the role of burrow mounds. They consist of subsurface soil and are initially bare, creating a new microhabitat for plants. They differ in physicochemical properties from the surrounding soil, promoting pioneer (Semenov et al., 2001; Swihart, 1991; Van Staalduinen & Werger, 2007) and nitrophile plants (Ballová et al., 2019; La et al., 2003). Mounds on our study sites consisted mainly of stones larger than 2 cm, which make up most of the subsurface soil layers (Kabala et al., 2021). Fine soil particles, as well as soil moisture ejected by marmots from deeper layers are probably removed by wind, like it has been found in desert environments (Sharma & Birla, 1975) long before it can be utilized by plants. Also, there is no evidence for the presence of deep organic matter in our study area (Kabala et al., 2021). Those are probably the reasons for the lack of plant succession on mounds we found, similarly to another study in Eastern Pamir (Dotter, 2009). Furthermore, marmot activity causes a loss of inhabitable sites for plants, probably for decades.
Researchers studying ecosystem engineering by burrowing animals have identified the crucial role of water availability in the interaction between those animals and plants. In arid regions, burrows and their inhabitants have a larger impact on plant species richness (Romero et al., 2015), soil nutrients (Decker, Eldridge, et al., 2019; Mallen-Cooper et al., 2019), decomposition rate (Decker, Leonard, et al., 2019) as compared to more arid environments. However, these comparisons may be biased, as there is simply more burrowing animals and more studies on ecosystem engineers in arid regions (Coggan et al., 2018). Other authors refused to identify any factor which could drive the effect size of burrow-dwelling ecosystem engineers, as it is solely site- and species-specific (Louw et al., 2019; Root-Bernstein & Ebensperger, 2013), and our results seem to fit this description. In humid, productive ecosystems where plants are mainly limited by biotic factors, such as herbivory and competition, the effect of ecosystem engineering is small (Crain & Bertness, 2006). In arid ecosystems, burrowing animals enhance nutrient cycling, which is hampered by water shortage, thus locally increasing productivity (Gharajehdaghipour et al., 2016; Yu et al., 2017). However, below a certain level of water availability, as drought directly limits plant growth, this effect decreases as it probably occurred in this study. Our results do not undermine the paradigm on the relationship between abiotic factors and the effect size of ecosystem engineers, but they lengthen the abiotic gradient by extreme habitats which are strongly understudied.
Eastern Pamir, similarly to other mountain regions, is particularly at risk due to climate change (Kohler et al., 2010). As temperatures are increasing, melting glaciers will continue to reveal patches of bare soil (Mętrak et al., 2015). But as precipitation in Eastern Pamir is expected to decrease even further (Lioubimtseva & Henebry, 2009; Normatov & Normatov, 2020), the succession of these areas will probably remain extremely slow and the impact of marmots on plants minimal. Rising temperatures may increase the rate of mineral weathering (Gislason et al., 2009), further diminishing the role of animal-derived nutrients for plants.
Despite the relatively small impact on plant nutritional status and vegetation cover, long-tailed marmots inhabiting extremely dry mountain habitats, most probably have a crucial impact on the ecosystem. Given the burrow density, marmots are probably the most abundant animal species in terms of biomass. This can make them an important food source for predators inhabiting our study area: wolves (Canis lupus ) red foxes (Vulpes vulpes ), and snow leopards (Uncia uncia ), which are known to prey on marmots, especially the two latter species (Blumstein & Robertson, 1995; Jumabay-Uulu et al., 2014; Khatoon et al., 2017). Marmot burrows can provide unique refuges of stable temperature and high humidity, crucial for the development of insects, especially in arid environments (Pike & Mitchell, 2013; Whitford & Kay, 1999). Marmot burrows can be used by other animals as the only shelter from strong winds and predators. Due to the unique character of the studied environment, the mechanisms described in this paragraph, similarly to our results on animal-plant interactions, can happen differently than in grasslands, steppes and mountain meadows. We think, that a lot of questions in this area are open to further research.