Macroecological patterns of rodent population dynamics shaped by bioclimatic gradients

Abstract

Long-term studies of cyclic rodent populations have contributed fundamentally to the development of population ecology. Previous research has shown macroecological patterns of population dynamics in relation to latitude, but without disentangling the role of underlying ecological and climate drivers. We collected 26 rodent time-series from the tundra biome and assessed how population dynamics characteristics of the most prevalent species varied with latitude and environmental variables. While we could not find a relationship between latitude and population cycle peak interval, other characteristics of population dynamics had latitudinal patterns. The environmental predictor variables provided insight into causes of these patterns, as i) increased proportion of optimal habitat in the landscape led to higher population cycle amplitudes in all species and ii) mid-winter climate variability had negative impacts on cycle amplitude in Norwegian lemmings and grey-sided voles. These results indicate that biome-scale climate and landscape change can be expected to have profound impacts on rodent population cycles and that the macro-ecology of such functionally important tundra ecosystem characteristics is likely to be subjected to transient dynamics.

Keywords

Bioclimatic zones, climate, field vole, grey-sided vole, latitude, macroecology, Norwegian lemming, population dynamics, tundra ecosystem, tundra vole

Introduction

Studies of the cyclic population dynamics of small rodents have contributed greatly to our understanding of population dynamics (Berryman 2002, Stenseth 1999, Turchin 2003). In particular, long-term series have provided an opportunity for macroecological studies (Cornulier et al. 2013, Ehrich et al. 2020, Hansson and Henttonen 1985). Such studies can reveal general patterns across large scales, enabling comparisons of climatic and ecological drivers of population dynamics and ecosystem functioning (Blackburn 2004, Kerr et al. 2007). Notably, many of the previous large-scale analyses of Fennoscandian rodents have shown strong latitudinal clines, with a northward increase in rodent cycle peak interval and amplitude (Bjørnstad et al. 1995, Hanski et al. 2001, Hansson and Henttonen 1988, Korpela et al. 2013, but see Angerbjörn et al. 2001). This gradient and its suggested connection to a species richness gradient within the predator guild has found its way into major ecology text books (Begon et al. 2006). Some studies based on exceptionally long-term data have, however, indicated that such macroecological pattern may be transient (Henden et al. 2009, Steen et al. 1990).
In his book summarizing a century of research on rodent population dynamics, Charles Krebs (2013) proposed to compile quantitative time series to test macroecological hypotheses as a research agenda for the next century. Examples of such hypotheses are the roles of food-web interactions (Hanski et al. 2001, Oksanen et al. 2008), diseases (Burthe et al. 2006), species-specific traits (Andreassen et al. 2013), landscape configuration (Magnusson et al. 2015), and climate (Kausrud et al. 2008, Tkadlec and Stenseth 2001) in shaping rodent dynamics. Although surrogate variables such as latitude and altitude may provide valuable clues about the underlying processes, quantification of more mechanistic variables is needed to reach beyond pattern description (Krebs 2013). Typically, several environmental variables change along latitudinal and altitudinal gradients, making their respective effects difficult to disentangle. For instance, previous studies of the Fennoscandian latitudinal gradient have merged data from different biomes (e.g. boreal forest and tundra) and the different rodent communities specific to these biomes (Bjørnstad et al. 1995, Hanski et al. 1991).
We propose that focusing on macroecological patterns of rodent population dynamics within a single biome allows for stronger inferences owing to less confounding between ecological and climatic variables. We first assessed whether biome-specific analyses of Fennoscandian rodents detected similar latitudinal patterns of population cycle characteristics as previous studies. We focused on the tundra biome, as rodent cycles have a particularly strong impact on tundra food-web dynamics (Ims and Fuglei 2005). We then evaluated whether variables describing winter climate variability and landscape composition would give more insight to the observed patterns. We hypothesized that the Norwegian lemming (Lemmus lemmus ), the only rodent species endemic to the Fennoscandian tundra, is more sensitive to more variable winter climate than other tundra-dwelling rodent species (Ims et al. 2011, Kausrud et al. 2008). Accordingly, we predicted that in regions with more variable winter climate the lemming makes up a smaller proportion of the rodent community and has lower amplitude cycles. We further hypothesized that landscape structure has implications for both rodent community structure and species-specific cycle amplitudes, as maximum population growth rates and consequently highest amplitudes should be related to high proportions of optimal habitat in the landscape (Bondrup-Nielsen and Ims 1988, Dalkvist et al. 2011, Lidicker 2000). Accordingly, we predicted that different rodent species would both dominate the rodent community and reach the highest cycle amplitudes in the parts of the tundra biome where their primary habitats occur.

Material and Methods

Study system

The tundra biome covers the arctic and oroarctic parts of Fennoscandia. Despite substantial variation in climate, the relatively simple food web has essentially a similar structure across the region. The low alpine bioclimatic zone is predominantly dwarf-shrub tundra and the middle alpine zone graminoid tundra (definitions according to Moen 1998, see Table A3). These tundra types have similar vegetation composition throughout the region, although the dwarf-shrub community has more arctic features in the north (Virtanen et al. 2016).
We focused on the four most abundant and widespread rodent species in the Fennoscandian tundra; besides the Norwegian lemming, the grey-sided vole Myodes rufocanus and two ecologically closeMicrotus -voles (M. agrestis and M. oeconomus , considered here as one functional group), and refer to them as genera. Based on food preferences (Soininen et al. 2013a), the low alpine zone contains optimum habitats for grey-sided voles. Lemmings, in turn, reach their highest numbers in the middle alpine zone (Ekerholm et al. 2001, Ims et al. 2011), which is dominated by their preferred food plants, i.e. graminoids and mosses (Soininen et al. 2013b). TheMicrotus -voles dwell in lush, moist, grass-rich habitats mainly found as patches in the low alpine zone (Hansson 1969, Henden et al. 2011).

Time-series and spatiotemporal replication

We collected time-series from 26 different locations in the Fennoscandian tundra where lemmings occurred and where snap-trapping data on all captured species were available for ≥ 10 years (Table 1, Figures 1 and A1, Appendix 2). Some of the locations also included trapping in adjacent ecotone forests (Appendix 2). The time-series at the different locations have various degrees of spatial replication (sampling units such as quadrats or trap lines ranging from 1 to 74 per location; Table 1). To link rodent population dynamics with environmental variables, we focused on analyses at the sampling unit level as the area extent of locations ranged from <1 km2 to >1000 km2(Appendix 2).
For analyses at the sampling unit level, we excluded sampling units if i) trapping had been conducted less than 10 years, unless a new unit was established in the vicinity to replace the previous unit, ii) the time-series were disconnected by gap years resulting in shorter than 10 years of continuous data or iii) no rodents were ever trapped. We also excluded locations from these analyses if data were not available per sampling unit. In total, 22 locations (n = 385 sampling units) fulfilled all criteria.
We used data from fall trapping season when available, as this is the season included in most data series (Table 1). We assume that fall data are more comparable between series than spring data, given that the varying match between the timing of spring trapping and phenology likely causes much noise in the data. For locations without fall trapping data, we used data from spring trapping season (n=3 locations) or pooled data across variable trapping dates (n=4 locations). To account for the effect of sampling seasons we i) included sampling season as an additive factor in all regression models, and ii) tested whether excluding locations without fall data affected the results of the best models.
To make the time-series comparable, we used the number of rodents captured per 100 trap nights per sampling unit as an abundance index in all analyses.

Characteristics of rodent population dynamics

We focused on characteristics of rodent population dynamics that have consequences for ecosystem functioning, namely cycle amplitude ,peak interval, peak sharpness , mean densities andcommunity contribution (cf. Hanski et al. 1991, Henden et al. 2008, Krebs 2013, Turchin et al. 2000, Table A1). Variables and their calculation are described in Table 2.
We assessed the presence of temporal trends in community contribution (Text A4, Figure A12, Table A12). As no such trends were evident, we proceeded with the approach of using rodent population cycle characteristics aggregated over time as spatial replicates.

Environmental predictor variables

We derived environmental predictor variables from raster data. It was not a priori clear how large an area around a sampling unit best predicts the local rodent numbers. We therefore extracted environmental predictor variables at three spatial extents: 1 km2, 9 km2, and 25 km2 around each sampling unit. Because the results differed only little, we present only the largest extent (25 km2). We chose this extent because it had the highest number of locations where any sampling unit had any middle alpine zone within their buffers (n=3, 8, and 9, at 1 km2, 9 km2 and 25 km2, respectively). Results at other extents are given in Text A2, Tables A8, A9, Figures A6, A7.
To assess winter climate impact on rodent population dynamics, we extracted the long-term mean number of days in January-March when the daily mean temperature was above zero. An increase in the number of days would represent a more variable winter climate as the baseline is 0 or very low number of days with temperature above 0oC (i.e., stable “winter climate”). This metric was available for the entire region and is linked to rodent winter demography (Aars and Ims 2002). We first created annual raster maps, depicting the number of days in January-March with above-zero temperature. We chose this period because we expect snow-covered conditions throughout the study area. The annual maps were based on gridded daily mean temperature raster maps of Fennoscandia, available from the Norwegian Meteorological institute, Climatology Division (senorge.no). The daily maps are estimated by a residual interpolation approach, applying terrain and other predictor variables to define a trend that is removed from the observed temperatures before they are interpolated into a 1 x 1 km gridded field. The trend is then added to the interpolated field to obtain a spatially continuous gridded temperature map (Tveito et al. 2005). Based on the annual maps, we calculated a mean per sampling unit across a buffer zone (5 km x 5 km) and the years when trapping was conducted at that unit.
To assess landscape composition, we used two approaches. First, we used a map of tundra bioclimatic zones in Norway (Table A3, Moen 1998), published by Blumentrath and Hanssen (2010). The map is based on modeling the tree-line altitude and thereafter estimating the elevation limits of the bioclimatic zones (Blumentrath and Hanssen 2010). The map has pixel size 25m x 25m. For each sampling unit in Norway (n=221), we extracted landscape composition by centering the sampling unit in the middle of a 25 km2 (5 km x 5 km) square and calculating the proportions of bioclimatic zones within the square. Second, we used July mean temperature (°C) as a proxy of bioclimatic zones, allowing inclusion of all locations (n=367 sampling units). We used a temperature raster map of the July mean temperature for the normal period of 1981-2010, available from the Norwegian Meteorological institute, Climatology Division (Hanssen-Bauer et al. 2015). The map is based on a residual interpolation approach as described for the winter climate data. Within the bioclimatic zones, July mean temperature data was distributed as follows (mean °C ± sd): low alpine (10.6 ± 1.3), middle alpine (8.7 ± 1.5), and high alpine zone (6.8 ± 1.4) (Figure A4). To extract the July temperature variable for each sampling unit, we proceeded similarly as described for the winter climate data. As July mean temperature was less than 50% correlated with the variable describing winter climate variability (rho = 0.41), we proceeded to use both variables in common models.

Statistical analyses of macroecological patterns in rodent population dynamics

We first assessed latitudinal patterns in the rodent population dynamics characteristics. At the level of sampling unit, we constructed a linear mixed effect model for each rodent genus and each characteristic, with latitude and trapping season as fixed variables and location as a random variable. As location-level data has previously been used to assess latitudinal patterns (Bjørnstad et al. 1995, Hansson and Henttonen 1985), we also ran linear models of latitude impact on community contribution and peak interval using location-level data.
We then assessed the effect of environmental variables on the population dynamics characteristics, focusing on community contribution and amplitude. We constructed two model sets: i) model set for all data (n=385 sampling units from 22 locations) using July temperature and winter climate variability as predictor variables, and ii)model set for Norwegian data (n=239 sampling units from 17 locations) using the proportion of optimal bioclimatic zone (low alpine for voles and middle alpine for lemmings) as predictor variable instead of July temperature. For each rodent genus and both model sets, we included all additive combinations of relevant predictor variables, together with trapping season as a fixed variable and location as a random variable. Visual inspection of the data indicated a non-linear effect of summer temperature for the two vole genera (i.e., temperature optimum, Figure 3), and we therefore included a quadratic term of temperature in these models. In all models for community contribution, we log-transformed the response variable to achieve close to normal distribution.
We assessed if, despite the large-scale synchrony in the occurrence of rodent population peaks, there was spatial autocorrelation in the population dynamics characteristics beyond the extent of location. To do this, we assessed the evidence for a spatial autocorrelation of the predicted random effects for location (Text A3, Figures A10, A11, Table A10). When there was evidence for such autocorrelation, it could be removed by including latitude as an additional covariate (Table A10), and we checked if results were robust to the inclusion of latitude as a covariate (Table A11). We selected the best models in each candidate model set based on AICcc (Burnham and Anderson 2002). Model selection was run with and without latitude as a covariate when there was evidence for spatial autocorrelation.
All data analyses were done in the software R, using packages lme4 (linear mixed effect models, Bates et al. (2008)), AICcmodavg (AICc based model selection, Mazerolle (2012)), and raster (extracting climate data, Hijmans and Etten (2012)). We used 95% confidence intervals to measure uncertainty for effects, and inspected model fit to assumptions using diagnostic plots.

Results

Characteristics of rodent population dynamics

At the sampling unit level, the community contributions of all three rodents ranged from 0-100% (Table A2). However, lemmings andMicrotus were abundant in only few locations. The median of community contribution across sampling units was > 50% in two locations for lemmings and in three locations for Microtus , while the same was true for ten locations for grey-sided voles. At the location level, community contribution of grey-sided voles and lemmings ranged from almost absence (1-3%) to complete dominance (80-88%), while Microtus reached at most 57% community contribution (Figure 1b, Table A2).
Peak interval ranged from 2 to 13 years at the sampling unit level. The very long maximum intervals arose from sampling units where a peak was absent despite being present at other sampling units within the same location. Consequently, peak intervals at sampling unit scale which were longer than twice the mean across all units (i.e., > 8 years) were removed from the analyses. This resulted in a peak interval range from 2 to 6.8 (mean 3.8 years, Table A2). Peak interval was less variable at the location level than at the sampling unit level (ranging from 3.2 to 4.7 with a mean of 3.9, Table A2).
At the sampling unit level, mean density was the population dynamics characteristic with clearest differences between the rodent genera (Figure 2, see Table A2 for all values in this paragraph and the associated measures of uncertainty). Grey-sided voles mean densities were on average higher than those of lemmings and Microtus (mean across all sampling units; 1.3, 5.3, and 2.6 for lemmings, grey-sided voles and Microtus , respectively). Grey-sided voles also had the highest sampling unit specific mean densities, respectively two and five times higher than for lemmings and Microtus . Amplitudes varied less, although the mean across lemming amplitudes was slightly lower than those of voles (2.3, 2.8, and 2.5 for lemmings, grey-sided voles, and Microtus , respectively). In contrast, the mean across lemming skewness was higher than those of voles (1.9, 0.5, and 1.3 for lemmings, grey-sided voles and Microtus , respectively). This indicates that lemming peaks were on average lower and sharper than vole peaks.
The characteristics of population dynamics were connected in all species in a similar manner (Figure A3). High community contribution, high mean density, high amplitude and low (below-zero or zero) skewness tended to occur together, as did low community contribution, low mean density, low amplitude and high (above-zero) skewness (Figure A3). This indicates that independent of species identity, the dominant species in the rodent community had high and round population peaks, whereas lower and sharper peaks characterized less abundant species. However, lemming skewness always remained above-zero (Figures 2, A3), indicating that sharp peaks were a consistent characteristic of this species.

Latitudinal patterns of population dynamics

The relationship to latitude differed between species (Figure 2, Table 3). Based on sampling unit specific analyses, the lemming community contribution decreased northwards, but the other lemming characteristics showed no latitudinal patterns. Grey-sided voles’ community contribution increased northwards, as did their mean density and amplitude, whereas their peak skewness decreased (i.e. peaks were less sharp). AlsoMicrotus’ mean density and amplitude increased northwards, but less strongly than those of grey-sided voles (Figure 2, Table 3). The mean density of the rodent community (i.e. all species combined) increased northwards, but we found no latitudinal patterns in peak interval. Location level patterns of community contribution were similar to patterns at sampling unit level (Table A5, Figure A5). Location level peak interval had no clear latitudinal trends, either. We explored visually patterns between peak interval and other variables (location, environmental variables, Figures A2, A9), but found no patterns.

Effects of climate and landscape on population dynamics

The mean number of days with above-zero temperatures during January-March ranged from 0.9 to 13.9 days per sampling unit, while July mean temperature ranged from 7.6 to 12.7 °C (for all values in this paragraph, see Table A4). Among the Norwegian locations where we had data for alpine bioclimatic zones, low alpine tundra dominated independent of spatial scale. Within a 25 km2neighborhood, low alpine zone made up an average of 81% (range 8-99%), while mid alpine zone made up just 3% (range 0-43%). Furthermore, only 9 out of 17 Norwegian locations had sampling units with any middle alpine zone within their buffers, while low alpine zone was present at all locations. All variables were correlated with latitude; the correlation was positive for July temperature and low alpine tundra, and negative for the other variables (Figure A8).
For lemmings, high community contribution and high amplitudes were related to the colder parts of the landscape (see Table 4 and Figure 3 for this and subsequent paragraphs). The model set with all data indicated a negative effect of July temperature on both aspects of the species population dynamics. The model set with only Norwegian data supported this by indicating a positive effect of middle alpine zone on community contribution. Winter climate variability was not included in the best models for lemming community contribution, but it had a negative effect on lemming amplitude.
For grey-sided voles, the different model sets indicated different effects. The model set for all data related community contribution positively to winter climate variability and amplitude positively to July temperature. In contrast, the model set for only Norwegian data related community contribution negatively to the optimal bioclimatic zone and amplitude negatively to winter climate variability.
The Microtus community contribution was related to the surrounding landscape. The model set for all data indicated a negative effect of July temperature, with an increasing impact at higher temperatures. The model set with only Norwegian data indicated a positive effect of the proportion of low alpine zone. The results forMicrotus amplitude indicated a negative effect of high July temperatures and a positive effect of the proportion of the low alpine zone. Winter climate was not included in any of the best models forMicrotus . However, it was included in the second-best models, and in the best models for amplitude at the most local scale (Tables A6, A7).

Discussion

Our study is the first biome-specific macroecological analysis of a rodent community at the scale of a biogeographic region (i.e., Fennoscandia). Earlier studies based on rodent trapping series from Fennoscandia have been instrumental for demonstrating macroecological latitudinal patterns of vertebrate community dynamics (Angelstam et al. 1984, 1985, Bjørnstad et al. 1995, Hanski et al. 1991, Hansson and Henttonen 1985, Steen et al. 1990) and as baselines for generating hypotheses of the underlying drivers of these dynamics (e.g. Hanski et al. 1993, Hanski et al. 2001, Hansson and Henttonen 1988, Korpela et al. 2013, Korpela et al. 2014). Our approach enabled us to verify that the previously found latitudinal gradient of the cycle amplitude was also present within the tundra biome, albeit only found in voles and not in lemmings. In contrast, we found no evidence for the previously found northwards increasing peak interval (Bjørnstad et al. 1995, Hanski et al. 2001). Further, the rodent community characteristics were related to landscape composition. Thus, bioclimatic zonation appears to be a strong predictor of structure and functioning of tundra rodent communities and as such more informative than latitudinal gradients. Furthermore, increasing winter climate variability decreased cycle amplitudes of both lemmings and grey-sided voles, implying that impacts of a warming winter climate may not necessarily be divergent between lemmings and voles as we hypothesized. Some of the relationships we identified were species-specific, demonstrating that lumping functionally different species in analyses of population dynamics characteristics should be done with great care, as it may mask relevant species-specific patterns. Even the lumping of two ecologically similar Microtus species in the present analyses may have affected our results. Taken together, environmental variables provided new understanding beyond latitudinal patterns.
Voles and lemmings have been described to have distinct shapes of the cycles, with vole peaks being of lower amplitude and rounder than those of lemmings (Turchin et al. 2000). Although we found lemming peaks to be on average sharper than vole peaks, their cycle amplitudes were within the same range as those of voles. Our results of cycle topology are thus only partly in line with the hypothesis by Turchin et al. (2000), i.e. different trophic interactions creating the different shapes of rodent population cycles. Moreover, our results on cycle topologies may relate to lemming peaks being observed less frequently than vole peaks, often with no individuals during the low phase, which can result from the scarcity of time-series sampling lemmings in their optimal bioclimatic zone (middle alpine), or different trappability. The community contribution of lemmings increased together with the proportion of middle alpine bioclimatic zone of the landscape, conforming well to smaller scale studies (Ekerholm et al. 2001, Ims et al. 2011, Kleiven et al. 2018). We thus expect that time-series collected in the middle alpine bioclimatic zone could have higher lemming cycle amplitudes than those observed in the existing data. Comparing lemming data from the middle alpine zone against vole data from the low alpine zone would relate each species to their optimal parts of the landscape, and thus provide a better case for comparing cycle topologies.
Our results matched only partly the earlier macroecological descriptions of Fennoscandian rodent population dynamics (Angerbjörn et al. 2001, Bjørnstad et al. 1995, Hanski et al. 1991, Hansson and Henttonen 1988, Korpela et al. 2013). In particular, we found no support for the latitudinal gradient in rodent population peak interval, unlike Bjørnstad et al. (1995), and Hanski et al. (1991). A lack of patterns within the tundra biome could indicate that this pattern arises from comparisons between biomes (e.g. less variable peak intervals in the tundra than in the boreal biome). Furthermore, it is unlikely that peak interval remains fixed over several decades, given the variation observed across Norway during the 20th century (Henden et al. 2009, Steen et al. 1990). Our findings thus support the conclusion of Henden et al. (2009); that the latitudinal gradient of small rodent population dynamics characteristics in Fennoscandia is not a temporally persistent phenomena and may rather be a case of transient dynamics (Hastings et al. 2018).
The species-specific properties of population dynamics did, however, show some of the same latitudinal patterns as described earlier (e.g. Hanski et al. 1991, Hansson and Henttonen 1985). The cycle amplitude of both vole species increased northwards, and the grey-sided vole displayed a prominent northward increase of community contribution, mean density, and cycle amplitude, concurrently with increasingly round peaks. However, both the abundance of low alpine bioclimatic zone (for Norway) and July temperature (for all locations) were positively correlated with latitude. Thus, the latitudinal patterns of voles appear to be related to an increasing abundance of the optimal bioclimatic zone in the landscape (cf. Bondrup-Nielsen and Ims 1988, Lidicker 2000). An increasing quality of the low alpine zone vegetation towards north, in terms of increasing palatability (Virtanen et al. 2016), could also contribute to explain this pattern.
Population dynamics of lemmings showed little latitudinal patterns, except for a decreasing community contribution towards the north. On the other hand, landscape composition appears to be a stronger predictor of the species dynamics than latitude, as both cycle amplitude and community contribution of lemmings had a positive relationship with the colder parts of the landscape. Our findings thus support the idea that landscape structure is an important determinant of both rodent community structure (Ecke et al. 2017) and species-specific population dynamics (Bondrup-Nielsen and Ims 1988, Le Vaillant et al. 2018, Lidicker 2000, Magnusson et al. 2015, Pyke et al. 1977).
We found little evidence for winter climate impacts on lemming community contribution. Yet, variable winter climate and community composition appear related, as the frequency of above-zero winter temperatures was positively associated with the community contribution of grey-sided voles. Furthermore, population cycle amplitude of lemmings and grey-sided voles decreased with increasing winter climate variability. Andreassen et al. (2020) found a similar pattern in boreal forests, relating the highest vole cycle amplitudes to the highest altitudes with presumably the coldest winters. Similarly, Ruffino et al. (2016) found a higher lemming peak in a continental study area with presumably colder winter climate than in a coastal study area, but an opposite pattern for grey-sided voles. Our results are thus only partly in line with these more local studies, but it appears evident that winter climate variability can affect both rodent community composition and the cycle amplitudes of both voles and lemmings.
Lemmings have been suggested to be more sensitive to warm winter climate than voles, potentially because their low-growing food plants can easily be encapsulate by ice after melt-freeze events (Ims et al. 2008, Ims et al. 2011). Our analyses did not reveal any dichotomy between lemmings and grey-sided voles in terms of winter climate effects but provided less evidence for the Microtus -voles sensitivity for winter climate than for the other tundra rodents. These seemingly contrasting results among the genera must, however, be interpreted with caution. First, lemmings and Microtus voles were scarce in most locations (n=5 and n=2 locations with more than 50% lemmings and Microtus , respectively), which potentially affected our ability to detect strong winter climate impacts in these genera. Moreover, although within-year spring and fall abundances are usually well correlated (Cornulier et al. 2013, Kausrud et al. 2008), winter climate is expected to have the most direct impact on spring abundances. Indeed, previous local-scale studies demonstrating impact of winter climate in Microtus and lemmings were based on population growth rates between fall and spring (Aars and Ims 2002). Thus, winter climate impacts are more likely to be revealed by analyses of e.g., snap-trapping data from the spring or by camera trapping data collected throughout the year. Furthermore, our variable for winter climate was calculated from gridded meteorological data across fixed mid-winter dates. Locally measured data on snow structure would provide a more mechanistic variable (Domine et al. 2018, Kausrud et al. 2008), while winter length can be decisive for cyclicity (Bierman et al. 2006). Given the climate-change driven changes of snow conditions (Pall et al. 2019) and the key role of rodents in tundra food webs (Ims and Fuglei 2005), we encourage future studies to probe into the mechanisms of snow condition impacts on rodent population dynamics.

Future perspectives

We propose that the Fennoscandian tundra and its rodent community are well suited for further biome-specific macroecological studies. The tundra biome extends across more than 10 latitudinal degrees in Fennoscandia and covers distinct climate gradients (Moen 1998, Virtanen et al. 2016), and small rodent dynamics in the tundra appear to have more pervasive food web impacts than in other biomes (Ims and Fuglei 2005, Krebs 2011, Olofsson et al. 2012). We here show that in this region, rodent population dynamics characteristics vary greatly within the biome and between the rodent genera. More focused assessments of causes of such variation have been called for (see Krebs 2013, Myers 2018), as most previous studies have been restricted to a few locations and local context dependencies are therefore almost unknown (cf. Soininen et al. 2018).
We see considerable scope for improvements for future macroecological studies based on small rodent population time series from the Fennoscandian tundra. Better insight may be achieved by i) extension of small rodent monitoring to achieve greater representation of higher alpine zones, ii) harmonization of practices and protocols, iii) development of environmental predictor data layers across country borders, iv) collection of data on tundra rodent dynamics throughout different seasons (e.g. Mölle et al. 2021), and v) development of predictor variables targeting winter climate impacts on rodents.
Long-term data in ecology is important in the face of anthropogenic driven changes of land-use, climate, and contaminant loads (Berteaux et al. 2017, Ecke et al. 2020, Ims and Yoccoz 2017). The scientific community has recognized its importance (Haase et al. 2016, Lindenmayer et al. 2012), but continued funding remains a challenge (Callaway et al. 2012). Yet, continued funding and increased coordination are prerequisites to achieve an efficient macroecological study design. The tundra biome is also the terrestrial biome on earth most affected by climate change (Box et al. 2019, CAFF 2013, Post et al. 2009) and the existing spatial configuration of population cycle characteristics is likely to change accordingly. Monitoring of the tundra biomes’ key players according to a macroecological protocol is a valuable approach to detect the impacts of climate change on tundra ecosystem functioning.

Supplementary material

Appendix 1: additional figures, tables, and methodological details
Appendix 2: additional information on rodent-time-series

References

Andreassen, H. P., Glorvigen, P., Rémy, A. and Ims, R. A. 2013. New views on how population‐intrinsic and community‐extrinsic processes interact during the vole population cycles. - Oikos 122: 507-515.
Andreassen, H. P., Johnsen, K., Joncour, B., Neby, M. and Odden, M. 2020. Seasonality shapes the amplitude of vole population dynamics rather than generalist predators. - Oikos 129: 117-123.
Angelstam, P., Lindström, E. and Widén, P. 1984. Role of predation in short-term population fluctuations of some birds and mammals in Fennoscandia. - Oecologia 62: 199-208.
Angelstam, P., Lindström, E. and Widén, P. 1985. Synchronous short‐term population fluctuations of some birds and mammals in Fennoscandia‐occurrence and distribution. - Ecography 8: 285-298.
Angerbjörn, A., Tannerfeldt, M. and Lundberg, H. 2001. Geographical and temporal patterns of lemming population dynamics in Fennoscandia. - Ecography 24: 298-308.
Bates, D., Maechler, M. and Dai, B. 2008. Lme4: Linear mixed-effects models using s4 classes (version 0.999375-37) [computer software].
Begon, M., Townsend, C. R. and Harper, J. L. 2006. Ecology, From Individuals to Ecosystems. - Blackwell Publishing.
Berryman, A. A. 2002. Population cycles: the case of trophic interactions. - Oxford University Press, Oxford.
Berteaux, D., Thierry, A. M., Alisauskas, R., Angerbjörn, A., Buchel, E., Doronina, L., Ehrich, D., Eide, N. E., Erlandsson, R., Flagstad, O., Fuglei, E., Gilg, O., Goltsman, M., Henttonen, H., Ims, R. A., Killengreen, S. T., Kondratyev, A., Kruchenkova, E., Kruckenberg, H., Kulikova, O., Landa, A., Lang, J., Menyushina, I., Mikhnevich, J., Niemimaa, J., Noren, K., Ollila, T., Ovsyanikov, N., Pokrovskaya, L., Pokrovsky, I., Rodnikova, A., Roth, J. D., Sabard, B., Samelius, G., Schmidt, N. M., Sittler, B., Sokolov, A. A., Sokolova, N. A., Stickney, A., Unnsteinsdottir, E. R. and White, P. A. 2017. Harmonizing circumpolar monitoring of Arctic fox: benefits, opportunities, challenges and recommendations. - Polar Research 36 (sup1):
Bierman, S. M., Fairbairn, J. P., Petty, S. J., Elston, D. A., Tidhar, D. and Lambin, X. 2006. Changes over Time in the Spatiotemporal Dynamics of Cyclic Populations of Field Voles (Microtus agrestis L.). - The American Naturalist 167: 583-590.
Bjørnstad, O. N., Falck, W. and Stenseth, N. C. 1995. A geographic gradient in small rodent density fluctuations: a statistical modelling approach. - Proceedings of the Royal Society B: Biological Sciences 262: 127-133.
Blackburn, T. M. 2004. Method in macroecology. - Basic Appl Ecol 5: 401-412.
Blumentrath, S. and Hanssen, F. 2010. Beregning av areal -In: Nybø (ed) Datagrunnlaget for ”Naturindeks i Norge 2010”, DN-utredning 4-2010 (in Norwegian).
Bondrup-Nielsen, S. and Ims, R. A. 1988. Predicting stable and cyclic populations of Clethrionomys. - Oikos 178-185.
Box, J. E., Colgan, W. T., Christensen, T. R., Schmidt, N. M., Lund, M., Parmentier, F.-J. W., Brown, R., Bhatt, U. S., Euskirchen, E. S. and Romanovsky, V. E. 2019. Key indicators of Arctic climate change: 1971–2017. - Environmental Research Letters 14: 045010.
Burnham, K. P. and Anderson, D. R. 2002. Model selection and multimodel inference : a practical information-theoretic approach. - Springer.
Burthe, S., Telfer, S., Lambin, X., Bennett, M., Carslake, D., Smith, A. and Begon, M. 2006. Cowpox virus infection in natural field vole Microtus agrestis populations: delayed density dependence and individual risk. - Journal of Animal Ecology 1416-1425.
CAFF 2013. Arctic Biodiversity Assesment. Status and trends in Arctic biodiversity. - Conservation of Arctic Flora and Fauna.
Callaway, R., Holt, R. D., Källersjö, M., MacLean, D. A., Magurran, A., Moore, A., Moss, B., Sæther, B.-E., Wedell, N. and Willis, K. 2012. International Evaluation of Ecology and Evolutionary Biology in Finland 2006–2010. Evaluation Report. - In: Publications of the Academy of Finland 3/12.
Cornulier, T., Yoccoz, N. G., Bretagnolle, V., Brommer, J. E., Butet, A., Ecke, F., Elston, D. A., Framstad, E., Henttonen, H., Hörnfeldt, B., Huitu, O., Imholt, C., Ims, R. A., Jacob, J., Jedrzejewska, B., Millon, A., Petty, S. J., Pietiäinen, H., Tkadlec, E., Zub, K. and Lambin, X. 2013. Europe-Wide Dampening of Population Cycles in Keystone Herbivores. - Science 340: 63-66.
Dalkvist, T., Sibly, R. M. and Topping, C. J. 2011. How predation and landscape fragmentation affect vole population dynamics. - PLoS ONE 6:
Domine, F., Gauthier, G., Vionnet, V., Fauteux, D., Dumont, M. and Barrere, M. 2018. Snow physical properties may be a significant determinant of lemming population dynamics in the high Arctic. - Arctic Science 4: 813-826.
Ecke, F., Angeler, D. G., Magnusson, M., Khalil, H. and Hörnfeldt, B. 2017. Dampening of population cycles in voles affects small mammal community structure, decreases diversity, and increases prevalence of a zoonotic disease. - Ecology and Evolution 7: 5331-5342.
Ecke, F., Benskin, J. P., Berglund, Å. M., de Wit, C. A., Engström, E., Plassmann, M. M., Rodushkin, I., Sörlin, D. and Hörnfeldt, B. 2020. Spatio-temporal variation of metals and organic contaminants in bank voles (Myodes glareolus). - Science of the Total Environment 713: 136353.
Ehrich, D., Schmidt, N. M., Gauthier, G., Alisauskas, R., Angerbjörn, A., Clark, K., Ecke, F., Eide, N. E., Framstad, E., Frandsen, J., Franke, A., Gilg, O., Giroux, M.-A., Henttonen, H., Hörnfeldt, B., Ims, R. A., Kataev, G. D., Kharitonov, S. P., Killengreen, S. T., Krebs, C. J., Lanctot, R. B., Lecomte, N., Menyushina, I. E., Morris, D. W., Morrisson, G., Oksanen, L., Oksanen, T., Olofsson, J., Pokrovsky, I. G., Popov, I. Y., Reid, D., Roth, J. D., Saalfeld, S. T., Samelius, G., Sittler, B., Sleptsov, S. M., Smith, P. A., Sokolov, A. A., Sokolova, N. A., Soloviev, M. Y. and Solovyeva, D. V. 2020. Documenting lemming population change in the Arctic: Can we detect trends? - Ambio 49: 786-800.
Ekerholm, P., Oksanen, L. and Oksanen, T. 2001. Long-term dynamics of voles and lemmings at the timberline and above the willow limit as a test of hypotheses on trophic interactions. - Ecography 24: 555-568.
Hanski, I., Hansson, L. and Henttonen, H. 1991. Specialist predators, generalist predators, and the microtine rodent cycle. - Journal of Animal Ecology 60: 353-367.
Hanski, I., Turchin, P., Korpimäki, E. and Henttonen, H. 1993. Population oscillations of boreal rodents: regulation by mustelid predators leads to chaos. - Nature 364: 232-235.
Hanski, I., Henttonen, H., Korpimäki, E., Oksanen, L. and Turchin, P. 2001. Small-Rodent Dynamics and Predation. - Ecology 82: 1505 - 1520.
Hanssen-Bauer, I., Førland, E. J., Haddeland, I., Hisdal, H., Mayer, S., Nesje, A., NIlsen, J. E. Ø., Sandven, S., Sandø, A. B., Sorteberg, A. and Ådlandsvik, B. 2015. Klima i Norge 2100 - kunnskapsgrunnlag for klimatilpasning. - In: KSS rapport. p. 203.
Hansson, L. 1969. Spring Populations of Small Mammals in Central Swedish Lapland in 1964-68. - Oikos 20: 431-450.
Hansson, L. and Henttonen, H. 1985. Gradients in density variations of small rodents - the importance of latitude and snow cover. - Oecologia 67: 394-402.
Hansson, L. and Henttonen, H. 1988. Rodent Dynamics as Community Processes. - Trends in Ecology & Evolution 3: 195-200.
Hastings, A., Abbott, K. C., Cuddington, K., Francis, T., Gellner, G., Lai, Y.-C., Morozov, A., Petrovskii, S., Scranton, K. and Zeeman, M. L. 2018. Transient phenomena in ecology. - Science 361:
Henden, J.-A., Ims, R., Yoccoz, N., Sørensen, R. and Killengreen, S. 2011. Population dynamics of tundra voles in relation to configuration of willow thickets in southern arctic tundra. - Polar Biology 34: 533-540.
Henden, J. A., Bårdsen, B. J., Yoccoz, N. G. and Ims, R. A. 2008. Impacts of differential prey dynamics on the potential recovery of endangered arctic fox populations. - Journal of Applied Ecology 45: 1086-1093.
Henden, J. A., Ims, R. A. and Yoccoz, N. G. 2009. Nonstationary spatio-temporal small rodent dynamics: evidence from long-term Norwegian fox bounty data. - Journal of Animal Ecology 78: 636-45.
Hijmans, R. J. and Etten, J. v. 2012. raster: Geographic analysis and modeling with raster data. R package version 2.0-12.
Haase, P., Frenzel, M., Klotz, S., Musche, M. and Toll, S. 2016. The long-term ecological research (LTER) network: Relevance, current status, future perspective and examples from marine, freshwater and terrestrial long-term observation. - Ecological indicators 65: 1-3.
Ims, R. A. and Fuglei, E. 2005. Trophic interaction cycles in tundra ecosystems and the impact of climate change. - BioScience 55: 311-322.
Ims, R. A., Henden, J. A. and Killengreen, S. T. 2008. Collapsing population cycles. - Trends in Ecology & Evolution 23: 79-86.
Ims, R. A., Yoccoz, N. G. and Killengreen, S. T. 2011. Determinants of lemming outbreaks. - PNAS 108: 1970-1974.
Ims, R. A. and Yoccoz, N. G. 2017. Ecosystem-based monitoring in the age of rapid climate change and new technologies. - Current Opinion in Environmental Sustainability 29: 170-176.
Kausrud, K. L., Mysterud, A., Steen, H., Vik, J. O., Ostbye, E., Cazelles, B., Framstad, E., Eikeset, A. M., Mysterud, I., Solhoy, T. and Stenseth, N. C. 2008. Linking climate change to lemming cycles. - Nature 456: 93-97.
Kerr, J. T., Kharouba, H. M. and Currie, D. J. 2007. The macroecological contribution to global change solutions. - Science 316: 1581-1584.
Kleiven, E. F., Henden, J.-A., Ims, R. A. and Yoccoz, N. G. 2018. Seasonal difference in temporal transferability of an ecological model: near-term predictions of lemming outbreak abundances. - Scientific Reports 8: 1-6.
Korpela, K., Delgado, M., Henttonen, H., Korpimäki, E., Koskela, E., Ovaskainen, O., Pietiäinen, H., Sundell, J., Yoccoz, N. G. and Huitu, O. 2013. Nonlinear effects of climate on boreal rodent dynamics: mild winters do not negate high-amplitude cycles. - Global Change Biology 19: 697-710.
Korpela, K., Helle, P., Henttonen, H., Korpimäki, E., Koskela, E., Ovaskainen, O., Pietiäinen, H., Sundell, J., Valkama, J. and Huitu, O. 2014. Predator–vole interactions in northern Europe: the role of small mustelids revised. - Proceedings of the Royal Society B: Biological Sciences 281: 20142119.
Krebs, C. J. 2011. Of lemmings and snowshoe hares: the ecology of Northern Canada. - Proceedings of the Royal Society B: Biological Sciences 278: 481-489.
Krebs, C. J. 2013. Population fluctuations in rodents. - The University Press Of Chicago.
Le Vaillant, M., Erlandsson, R., Elmhagen, B., Hörnfeldt, B., Eide, N. E. and Angerbjörn, A. 2018. Spatial distribution in Norwegian lemming Lemmus lemmus in relation to the phase of the cycle. - Polar Biology 41: 1391-1403.
Lidicker, J., William Z 2000. A food web/landscape interaction model for microtine rodent density cycles. - Oikos 91: 435-445.
Lindenmayer, D. B., Likens, G. E., Andersen, A., Bowman, D., Bull, C. M., Burns, E., Dickman, C. R., Hoffmann, A. A., Keith, D. A., Liddell, M. J., Lowe, A. J., Metcalfe, D. J., Phinn, S. R., Russell-Smith, J., Thurgate, N. and Wardle, G. M. 2012. Value of long-term ecological studies. - Austral Ecology 37: 745-757.
Magnusson, M., Hörnfeldt, B. and Ecke, F. 2015. Evidence for different drivers behind long-term decline and depression of density in cyclic voles. - Population Ecology 57: 569-580.
Mazerolle, M. J. 2012. AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). Version 1.24.
Moen, A. 1998. Nasjonalatlas for Norge: Vegetasjon (in Norwegian). - Statens kartverk.
Myers, J. H. 2018. Population cycles: generalities, exceptions and remaining mysteries. - Proceedings of the Royal Society B: Biological Sciences 285: 20172841.
Mölle, J., Kleiven, E. F., Ims, R. and Soininen, E. 2021. Using subnivean camera traps to study arctic small mammal community dynamics during winter. - Arctic Science e-First:
Oksanen, T., Oksanen, L., Dahlgren, J. and Olofsson, J. 2008. Arctic lemmings, Lemmus spp. and Dicrostonyx spp.: integrating ecological and evolutionary perspectives. - Evolutionary Ecology Research 10: 415-434.
Olofsson, J., Tømmervik, H. and Callaghan, T. V. 2012. Vole and lemming activity observed from space. - Nature Climate Change advance online publication:
Pall, P., Tallaksen, L. M. and Stordal, F. 2019. A climatology of rain-on-snow events for Norway. - Journal of Climate 32: 6995-7016.
Post, E., Forchhammer, M. C., Bret-Harte, M. S., Callaghan, T. V., Christensen, T. R., Elberling, B., Fox, A. D., Gilg, O., Hik, D. S., Hoye, T. T., Ims, R. A., Jeppesen, E., Klein, D. R., Madsen, J., McGuire, A. D., Rysgaard, S., Schindler, D. E., Stirling, I., Tamstorf, M. P., Tyler, N. J. C., van der Wal, R., Welker, J., Wookey, P. A., Schmidt, N. M. and Aastrup, P. 2009. Ecological dynamics across the arctic associated with recent climate change. - Science 325: 1355-1358.
Pyke, G. H., Pulliam, H. R. and Charnov, E. L. 1977. Optimal foraging: A selective review of theory and tests. - The Quarterly Review of Biology 52: 137-153.
Ruffino, L., Oksanen, T., Hoset, K. S., Tuomi, M., Oksanen, L., Korpimäki, E., Bugli, A., Hobson, K. A., Johansen, B. and Mäkynen, A. 2016. Predator–rodent–plant interactions along a coast–inland gradient in Fennoscandian tundra. - Ecography 39: 871-883.
Soininen, E. M., Ravolainen, V. T., Bråthen, K. A., Yoccoz, N. G., Gielly, L. and Ims, R. A. 2013a. Arctic small rodents have diverse diets and flexible food selection -PLoS ONE 8: e68128.
Soininen, E. M., Zinger, L., Gielly, L., Bellemain, E., Bråthen, K. A., Brochmann, C., Epp, L. S., Gussarova, G., Hassel, K., Henden, J.-A., Killengreen, S. T., Rämä, T., Stenøien, H. K., Yoccoz, N. G. and Ims, R. A. 2013b. Shedding new light on the diet of Norwegian lemmings: DNA metabarcoding of stomach content. - Polar Biology 36: 1069-1076.
Soininen, E. M., Henden, J. A., Ravolainen, V. T., Yoccoz, N. G., Bråthen, K. A., Killengreen, S. T. and Ims, R. A. 2018. Transferability of biotic interactions: Temporal consistency of arctic plant–rodent relationships is poor. - Ecology and Evolution 8: 9697-9711.
Steen, H., Yoccoz, N. G. and Ims, R. A. 1990. Predators and small rodent cycles: an analysis of a 79-year time series of small rodent population fluctuations. - Oikos 115-120.
Stenseth, N. C. and Framstad, E. 1980. Reproductive Effort and Optimal Reproductive Rates in Small Rodents. - Oikos 34: 23-34.
Stenseth, N. C. 1999. Population Cycles in Voles and Lemmings: Density Dependence and Phase Dependence in a Stochastic World. - Oikos 87: 427-461.
Tkadlec, E. and Stenseth, N. C. 2001. A new geographical gradient in vole population dynamics. - Proceedings of the Royal Society B: Biological Sciences 268: 1547-1552.
Turchin, P., Oksanen, L., Ekerholm, P., Oksanen, T. and Henttonen, H. 2000. Are lemmings prey or predators? - Nature 405: 562-565.
Turchin, P. 2003. Complex Population Dynamics: A Theoretical/Empirical Synthesis. - Princeton University Press.
Tveito, O. E., Bjørdal, I., Skjelvåg, A. O. and Aune, B. 2005. A GIS-based agro-ecological decision system based on gridded climatology. - Meteorological Applications 12: 57-68.
Virtanen, R., Oksanen, L., Oksanen, T., Cohen, J., Forbes, B. C., Johansen, B., Kayhko, J., Olofsson, J., Pulliainen, J. and Tømmervik, H. 2016. Where do the treeless tundra areas of northern highlands fit in the global biome system: toward an ecologically natural subdivision of the tundra biome. - Ecology and Evolution 6: 143-158.
Aars, J. and Ims, R. A. 2002. Intrinsic and climatic determinants of population demography: The winter dynamics of tundra voles. - Ecology 83: 3449-3456.

Figures