Reviews and Syntheses: Composition and Characteristics of Burrowing Animals along a Climate and Ecological Gradient, Chile

Although the burrowing activity of some species (e.g. gophers) is well studied, a comprehensive inventory of burrowing animals in adjacent biomes is not yet known, despite the potential importance of burrowing activity on the physical and chemical evolution of Earth’s surface. In this study, we review the available information with a focus on: a) an inventory of burrowing vertebrates and invertebrates along the climate and 15 ecological gradient in Chile; b) the dimensions and characteristics of burrows; and c) calculation of excavation rates by local species compositions. Methods used include a literature compilation (>1000 studies) of Chilean burrowing animal species integrated with global, species-specific excavation rates. A field study augments literature findings with quantification of the zoogeomorphic effects on hillslope mass transport at the animal community level and along the arid to humid-temperate climate gradient within the Chilean Coastal Cordillera 20 (27-38° S latitude). The literature review indicates 45 vertebrate and 345 invertebrate burrowing species distributed across Chile in different biomes. Burrowing depths for Chilean mammals range between 3 m (e.g. for skunks, Conepatus) to 0.25 m (for rock rats, Aconaemys). For invertebrates, burrowing depths in Chile range between 1 m for scorpions to 0.3 m for spiders. In comparison, globally documented maximum burrow depths reach up to more than 6 m for 25 vertebrates (gopher tortoises and aardvarks) and 4 m for invertebrates (ants). Minimum excavation rates of local animal communities observed from field sites in Chile are 0.34 m3 ha-1 yr-1 for the arid site, 0.56 m3 ha-1 yr-1 for the semi-arid site, 0.93 m3 ha-1 yr-1 for the mediterranean site and 0.09 m3 ha-1 yr1 for the humid-temperate site, with the latter likely an underestimation. The calculated minimum Chilean excavation rates are within the large range of globally observed single species rates ranging between 0.01 and 30 146.20 m3 ha-1 yr-1 for vertebrates and from 0.01 to 53.33 m3 ha-1 yr-1 for invertebrates. Taken together, results highlight not only the diverse and latitudinally varying number of burrowing vertebrates and invertebrates present in different biomes, but also fosters the understanding of how burrowing activity changes over a gradient and is influenced by mean annual temperature, mean annual precipitation, slope aspect and latitudinal related incoming solar energy. 35

. Thus, the effects of burrowing animals along gradients in temperature, precipitation, and 80 vegetation are seldomly studied.
We build upon the previous global inventories and focus on burrowing vertebrates and invertebrates along the climate and ecological gradient in Chile. Our approach complements recent global compilation on ants as geomorphological agents in Viles et al. (2021) and a global review about insects as such agents (Bétard, 2020).
From existing literature, we compile the burrowing taxa of Chile, their distribution, burrow details such as tunnel 85 size, depth and geographic extent, as well as excavation rates (when available). Globally published excavation rates of animals are summarized for comparison. In addition, we present new observations from four Chilean study areas spanning diverse biogeographic zones and document the number of burrow entrances of present taxa for vertebrates and invertebrates. From measurements of burrow entrances and tunnel lengths, taxa independent excavation rates are calculated as minimum values. With this method it is possible to sample and estimate species 90 independent zoogeomorphic effects at the hillslope scale, and along a climate gradient.

Literature compilation: Burrowing animals in Chile
To identify burrowing animal species in Chile and compile burrow related information, as well as global animal excavation rates, a literature search was conducted using multiple online databases and published studies.

95
Complementary searches were done in English and in Spanish. Additionally, available information was accessed at the scientific collections of Laboratorio de Entomología Ecológica de la Universidad de La Serena (LEULS), La Serena, Chile, and the División Aracnología del Museo Argentino de Ciencias Naturales "Bernardino Rivadavia" (MACN-Ar), Buenos Aires, Argentina. The species-specific information compiled included species taxonomy, common name(s), alimentation type, special characteristics, group size, density, excavation 100 characteristics (tunnel diameter, (maximum) burrow depths, details about burrow type), preferred habitat and distribution limit by elevation. For vertebrates, body size and weight were also compiled. The data compilation focused on species that excavate in an active manner. Excluded were species that are restricted to the use of burrows made by sympatric species, as well as ground dwellers that hide under rocks or vegetation and also soildwelling animals that move through the soil by pushing material aside ("swimming") (Gabet et al., 2003). The full 105 list of excavating species identified for Chile is provided in a companion data publication (Übernickel et al., 2020).
Hereafter we use the data in this data publication for our figures and analysis. The subset of data regarding the four study sites were summarized as the burrowing species lists of vertebrates (Table S1) and invertebrates (Table   S2). In addition, published excavation rates of animals worldwide were compiled (Table 1). For the calculation of the excavation rates from volume measurements of publications, an intermediate near surface pedolith bulk density 110 of 1.2 g cm -3 was assumed (Amelung et al., 2018). compiled at a resolution of 0.5 latitudinal degrees (approx. 55 km) (Fig. 1). The topography, mean annual precipitation (MAP) and mean annual temperature (MAT) were also compiled (Fick and Hijmans, 2017). The distribution of invertebrate burrowing species in Chile are not yet available. To estimate the diversity of sympatric https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License. species along Chile we generalized the mostly single study site information of the species to the extent of the respective Chilean region (Übernickel et al., 2020) and plotted the amount of species per region (Fig. 2).  The data compilation is based on the associated data publication to this study, Übernickel et al. (2020). The map source is http://freevectormaps.com, Other* refers to 130 Isopterans, Neuropterans and Hemipterans https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License.

Site description
The study sites are part of the German-Chilean priority program EarthShape (Earth surface shaping by biota; 135 www.earthshape.net). The sites are located within three National Parks (NP) and one private reserve along the Chilean Coastal Cordillera to minimize anthropogenic disturbances. The latitudinal distance between these areas extends over 1,300 km and includes NP Pan de Azúcar (~26° S) in the arid north, the semi-arid Private Reserve Santa Gracia (~30° S), mediterranean climate NP La Campana (~33° S), and the humid-temperate NP Nahuelbuta (~38° S) in the south (Fig. 3). For previous detailed information concerning site descriptions, soil analysis, as well 140 as local vegetation see Bernhard et al. (2018), Oeser et al. (2018) and Schaller et al. (2018). In the following we summarize the climate, soil classification and topographic characteristics per location (Tabs. S1 and S2).  (Rundel et al., 1996). It has a MAP of 12 mm and a MAT of 16.8° C (Fick & Hijmans 2017). The elevation is 330 m a.s.l., the vegetation cover is < 10 % and the slopes range between 25° and 40° 150 (Oeser et al. 2018), with the plots located on several slopes. The soil is classified as a Regosol. Bulk density varies between 1.3 Mg m -3 at the north-facing slope and 1.5 Mg m -3 at the south-facing slope and the texture is sandy loam in both slopes (Bernhard et al., 2018).
The Private Reserve Santa Gracia contains plots on hillslopes near 29.759° S 71.166° W. It has a MAP of 66 mm and a MAT of 13.7° C (Fick & Hijmans 2017). The elevation is around 680 m a.s.l. and vegetation cover is 30-40 classified as Cambisol due to a differentiation between surface horizons and subsoil. Bulk densities are homogenous in the first 20 cm depth and 1.5 Mg m -3 for both slopes, but texture predominates as sandy loam in the north-facing slope, while the south-facing slope has a loamy sand texture (Bernhard et al. 2018). Known disturbances of this ecosystem are goats feeding on disturbed (reduced) vegetation (Armesto et al., 2007).

160
NP La Campana is the mediterranean site and contains plots at the sector Ocoa, around 32.9562° S 71.0637° W.
The NP was established in 1967 with the boundaries legalized in 1985 (Macdonald et al., 1988). It has a MAP of 367 mm and a MAT of 14.1° C (Fick & Hijmans 2017). The elevation is around 730 m a.s.l., with a vegetation cover of close to 100 %, the north-facing slope has an angle of 12° and the south-facing slope of 23° (Oeser et al. 2018). The soil is classified as Cambisol. Bulk densities vary between 1.5 Mg m -3 at the north-facing slope and 165 1.1 Mg m -3 at the south-facing slope and the general soil texture is sandy loam for the first 20 cm of depth (Bernhard et al., 2018). Known disturbances of the ecosystem are (illegally) grazing cows (Rundel and Weisser, 1975).
NP Nahuelbuta is the southernmost site, with plots located near 37.8087° S 73.0137° W. This NP was established in 1936 (Servicio Agrícola y Ganadero, SAG, 1970). It has a MAP of 1,469 mm and a MAT of 6.6° C (Fick & Hijmans 2017). The elevation is around 1,240 m a.s.l. and vegetation cover is ~100 %. The north-facing slope has 170 an angle of 13° and the south-facing slope of 15° (Oeser et al. 2018). The soil is classified as an Umbrisol. The bulk density is 0.9 Mg m -3 at the north-facing slope and 0.8 Mg m -3 at the south-facing slope and the texture is sandy-clay loam (Bernhard et al., 2018).

Data compilation
Data acquisition took place during three field campaigns in November 2016, May 2017 and March 2018. Each 175 site was visited during every field campaign. The plots per site were selected with the criteria to be located in eastto-west oriented valleys, two plots per opposing ~north and ~south facing slopes with one plot near both the top and one plot near the bottom of the hillslope (Fig. 3B). The size of the plots was 10 x 10 m in 2016 and 2018; in May 2017 5 m x 5 m sized plots were analyzed and numbers later scaled for comparability.
The burrow measurement procedure was to walk uphill within the plot in approximately 50 cm separated tracks, 180 measuring the burrow entrances along the path and marking them to avoid duplicate measurements. The procedure and the parameters taken were optimized during the three field campaigns. First, the diameter of burrow entrances was digitally measured from photographs using ImageJ (v1.51m9). When the entrance was oval the major and minor axes were measured. In 2017 and 2018, the time effort was reduced by counting burrow entrances and taking measurements on site. From 2017 onwards minimum tunnel length was determined as well, measuring manually 185 from the entrance to the reachable maximum end of the tunnel. Furthermore, the direction of the tunnel into the ground was labelled as either vertical, horizontal or 45°. In 2018 the ground cover of the plots was estimated (Table   S3). Note, for NP Pan de Azúcar the two plots per hillslope scheme was not completely fulfilled, as slopes at this site were too short. Plots were selected as similar to the original set up as possible.

190
From the collected data we correlated entrance diameter and minimum tunnel lengths (Fig. 4), measured the distribution of burrow entrances regarding the plot position on the top or bottom of the slope, distribution of entrances relative to the slope aspect (north-or south-facing), and distribution of entrances along the climate gradient ( Fig. 5) and calculated minimum burrowing depth (Fig. 6) and minimum excavation rates (Fig. 7). Note, the approach to evaluate the number of burrow entrances is a proxy to burrowing activity, as the actual density of 195 https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License. the individuals remains unknown. Also, by counting the burrow entrances, the number of burrows is not exactly registered, as one burrow may have several entrances. Statistical analysis on the burrow entrance data was not applicable as no replicates were collected. Finally, ground cover impeded the collection of sufficient data at the southernmost site (Nahuelbuta).  Minimum burrowed depths (Fig. 6) were approximated from the 2018 data, using the measured tunnel length, angle of the tunnel and the law of sines. The minimum excavation rate in m 3 ha -1 yr -1 (Fig. 7) was estimated based 230 https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License. on tunnel volumes measured in March 2018. March is late summer in Chile, and represents the closest estimate of burrowing activity within a year before autumn and winter rainfall reset the surface. For each entrance the minimum excavated volume was calculated, dependent on the entrance's approximate shape. For round and oval entrances, the volume was calculated using the entrance radius, or major and minor axis, respectively. Crevice shaped entrances were treated as ovals for conservative volume approximation. The volumes were converted from 235 values per plot to m 3 ha -1 yr -1 . Note that for tunnel volumes and excavation rate minimum values are reported, because additional volume behind a first curve or obstacle in the tunnel could not be measured.
In addition, the solar energy input and its effects may influence the distribution of the burrow entrances as they vary with the slope aspect (Gallardo-Cruz et al., 2009). To detect a link between of solar energy input and the quantity of burrow entrances on north-and south-facing slopes, the diurnal course of potential direct radiation was 240 calculated (Fig. 8). We assumed cloud free conditions and the potential direct radiation on an inclined surface using the terrain angle, based on the coordinates of the study sites following Bendix (2004). The calculations were conducted for the 21 st June and December respectively, the period of highest and lowest solar radiation. For simplicity, atmospheric transmission is fixed to 0.8 and potentially occurring shadows of neighboring topography and/or vegetation were neglected. Plotted values are for noon (12 p.m.).

245
Figure 8: Calculated solar radiation and sum of animal burrow entrances separated per animal group (large and small animals) for the slopes along the studied climate gradient; the solar radiation calculation follows Bendix (2004); the slopes are oriented N(orth)-facing and S(outh)-facing; the depicted data is pooled from all three field campaigns 2016, 2017 and 2018;

250
Abbreviations: PA= Pan de Azúcar, SG= Santa Gracia, LC = La Campana, NA = Nahuelbuta, note that in PA the sum of the south-facing plot entrances is of one plot, and the north-facing entrances sum of three plots.

Literature compilation on Chilean burrowing animal species
The literature survey included >1,000 references and revealed 45 vertebrate and 345 invertebrate species in Chile 255 with excavating activity. The full species list is provided in the associated data publication (Übernickel et al., 2020). Of the 45 burrowing vertebrate species identified for Chile, 40 are mammals, two are birds and three are reptiles. Most burrowing species are rodents i.e. mice, (mole-) rats, rabbits, cavies (guinea pigs), chinchillas and nutrias. Other excavating mammals present include foxes, skunks, armadillos and some insectivores. Regarding Higher taxonomic groups with burrowing species are beetles, with 157 species including 63 darkling beetles (Tenebrionidae), 13 ground beetles, (Carabidae), 11 trogids (Trogidae), 29 dung beetles and 41 scarab beetles 270 (both of Scarabaeidae); 13 bees (Colletidae); 5 wasps (Crabronidae and Sphecidae); 23 ants (Formicidae); 1 ant lion (Myrmeleontidae); 1 termite (Rhinotermitidae) and 1 cicada (Cicadidae) species (Übernickel et al., 2020). Bétard (2020) mentions additional orders and families of insects that burrow on a global scale. Of these (Blattodea, Dermaptera, Diptera, Embioptera, Ephemeroptera, Lepidoptera, Mecoptera, Megaloptera, Odonata, Orthoptera, Plecoptera, Trichoptera), however we did not find supporting literature for equivalent Chilean species. The 275 following sections provide details concerning the findings summarized separately for vertebrates and invertebrates in Chile, subdivided into the geographic distribution in Chile, composition and spatial occurrence at study sites and their burrow characteristics.

Burrowing vertebrates
The distribution of Chilean burrowing mammal species varies with latitude and altitude along Chile ( Fig. 1 D), 280 with most species having restricted distributions in distinct parts of the country. A maximum of 18 sympatric species were identified around 35° S. Most diversity is present at the mediterranean zone (central Chile) with commonly 11 to 16 species per grid cell. Trends in the species distribution are gradual with latitude and fewer species are present at higher elevations, i.e. the Andes ( Fig. 1 A). Similarly, species number decreases with decreasing mean annual temperature, i.e. Patagonia in the south (MAT, Fig. 1 B). Finally, species numbers 285 decrease with decreasing mean annual precipitation, i.e. northern arid lowlands (MAP, Fig. 1 C).
Other burrowing vertebrates in addition to mammals includes birds and reptiles. Among the burrowing birds only the burrowing owl (Athene cunicularia) and the parrot "Tricahue" (Cyanoliseus patagonus) are identified. These species have distributions at the Chilean main land; Chilean seabirds, i.e. burrowing penguins and petrels, were excluded from the list, as in general they are exclusively present on islands (Chester, 2008;Figueroa and Stucchi, 290 2008; Reyes Arriagada et al., 2013;Zavalaga and Alfaro-Shigueto, 2015). Of the three burrowing reptile species, two are Liolaemus species. Chile has 99 known species of Liolaemus sp. (Ruiz de Gamboa, 2020), but only for the two Liolaemus sp. the burrowing behavior was found documented. The third burrowing reptile is the Chilean Racerunner Callopistes maculatus (Übernickel et al., 2020) with a distribution from just north of Antofagasta to Cauquenes (Contreras et al., 2020).

295
The species composition in the four study sites is composed of 27 out of the total 45 Chilean burrowing vertebrates (Table S1). The NP Pan de Azúcar study area has 13 documented (and up to 15, including burrowing species potentially present) burrowing vertebrate species, 13 (up to 17) are present in Santa Gracia, 14 (up to 17) in NP La Campana and 11 (up to 15) in NP Nahuelbuta. Very few species are present along the entire studied gradient, https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License.
The knowledge of the geographic distribution and density of burrowing vertebrates in Chile is sparse for most species, with no further details other than the morphometric description of the animals available (Übernickel et al., 2020). Most details are available for some rodents, especially for the cururo, Spalacopus cyanus, and degu, Octodon degus. Within the known burrowing species, the density of individuals varies considerably on a local 305 level, with most species occurring in densities usually less than one individual ha -1 , up to an approximate 300 individuals ha -1 under favorable, time restricted circumstances associated with (for example) El Niño Southern Oscillation (ENSO) events, see discussion. Mammal group sizes vary from a solitary lifestyle, usual for armadillos, skunks and foxes, to groups of more than 15 to 20 individuals for most rodents. These rodent colonies generally inhabit multi-entrance burrows (Iriarte W., 2007;Kolb, 1985;Pearson, 1988;Redford and Eisenberg, 1992; 310 Schmid-Holmes et al., 2001;Shepherd and Ditgen, 2013). The burrow entrances are usually clumped in a small area (Iriarte W., 2007;Pearson, 1984Pearson, , 1988Pearson, , 1951, with larger spacing between complex burrows. There are usually less than 2 colonies ha -1 (Cofré and Marquet, 1999;Patton et al., 2015;Redford and Eisenberg, 1992).
Mammal burrow dimensions are highly variable and for most species the areal extent of burrows is not reported.
The following provides an overview of the known basic dimensions available for Chilean species (Übernickel et
The large group of burrowing small mammals, including rodents and insectivores, vary in their tunnel diameter between 2 and 11 cm. By diameter the entrances are roughly classifiable to distinct groups. The largest rodentmade entrances with diameters of 7 to 10 cm are made by tuco-tucos, Ctenomys sp., rats Rattus sp., rock rats,
valdivianus. The smallest entrance diameters of 2.5 to 4 cm are made by grass mice, Abrothrix sp. / Akodon sp., the mouse, Loxodontomys sp. and the house mouse, Mus musculus.

330
For non-mammal burrowing vertebrates information on burrow characteristics is scarce. Concerning lizards, field observations reveal that burrows of the lizard Callopistes vary considerably in size, and frequently they occupy burrows made by Spalacopus cyanus or other rodents. The observed entrance diameters is 10 to 15 cm, 0.6 to 2 m for tunnel lengths and they have also been observed to shelter under rocks (pers. com. A. Cortés Maldonado, J. P. Castillo). Regarding birds, burrowing parrots make entrances that vary largely in size, 14-49 cm horizontally and 335 8-25 cm vertically.

Burrowing invertebrates
The distribution of burrowing invertebrate species is heterogeneous along the north to south extent of Chile (Fig.   2). To summarize, the burrowing animal diversity is highest in the semi-arid "Norte Chico" (small north) of Chile,  Snelling and Hunt, 1975) are distributed over all Chilean biomes from sea level to 2500 m a.s.l. (Ipinza-Regla et al., 1983). A maximum of 47 sympatric species is present in Central Chile (antmaps.org, also for genus distributions). Of burrowing species, the large sized "hormigones" (Camponotus), is one of the most abundant and widely distributed genera in Chile (Snelling and Hunt, 1975). The fire ant genus Solenopsis includes S. gayi, the most common native ant in Chile with the widest distribution. Of the Neuropterans, antlions are present in northern 370 Miller and Stange, 2016) and central Chile.
Species compositions of burrowing invertebrates are unknown for most parts of Chile. We approximated the compositions by counting all burrowing species with published presence within the regions of the study sites.
"Likely present" species according to current knowledge were also included. We figure the following species diversity (Table S2) Known details of the spatial distribution of burrowing Chilean invertebrates are very scarce. Habitat preferences, e.g. specific substrates species are associated to, are registered for very few species (Übernickel et al., 2020). Praocis (Mesopraocis) sp. are usually found within the uppermost 30 cm of the substrate, but have also been found 410 close to 2 m depth (pers. obs. J. Pizarro-Araya). They lay eggs on surfaces or at depths between 5 and 10 cm. Both burrowing bee genera, Caupolicana and Cadeguala, burrow their nests in the soil to around 50 cm depth. The burrows of Cadeguala occidentalis reach extensions of 120 * 170 cm with over 150 entrances. The "digger wasp" Sphex sp. burrows reach 33 cm depth in regions outside of Chile (Brockmann, 1980). The burrows of the cicada Tettigades chilensis reach depths of 45 cm.

Published excavation rates of burrowing animals
Globally, excavation rates of single vertebrate and invertebrate species are published for specific sites (Table 1) ha -1 yr -1 ). A range of excavation rates from 10 to 19 m 3 ha -1 yr -1 is reached by the European mole in Russia and the arctic ground squirrel. Excavation rates up to 10 m 3 yr -1 are reported for badgers, bettongs (rat kangaroos), cururos (rodents) and porcupines. In contrast, small excavation rates are reported for rabbits (0.68 m 3 ha -1 yr -1 ). Similarly, the globally published excavation rates for invertebrates also vary significantly (Table 1)

Maximum burrow depths of vertebrates and invertebrates
Globally, the documented maximum depth of animal burrows is 6 m (gopher tortoises and aardvark dens, Platt et 435 al., 2016). Among invertebrates, the basal taxa of scorpions and spiders reach burrow depths of 70 and 60 cm, respectively (Framenau and Hudson, 2017;Talal et al., 2015). Ants' nests are documented up to depths of 4 m, and cicadas reach depths of 2.5 m (Bétard, 2020;Tschinkel, 2004). Chilean burrowing mammals are described to reach maximum burrow depths of 3 m for skunks, 1.5 m for armadillos, > 80 cm for rabbits, < 75 cm for tuco-tucos, ~ 60 cm for mole rats (Chelemys), degus and cururos, ~ 440 28 cm for cavies and < 25 cm for rock rats (Aconaemys). For invertebrates maximum burrowing depths are estimated to range from 100 cm for scorpions to 30 cm for spiders. Higher taxa burrow depths are 60 cm for beetles, 52 cm for bees and 45 cm for cicadas.

Study: Observed animal burrows in Chilean study areas
A total of 1,590 burrow entrances were found within the plots and measured throughout the three field seasons in 445 2016 to 2018 (yellow stars, Fig. 1, Fig. 3). During all field visits animal burrow entrances were identified at all sites and in all plots. In general, there are two types of animal burrows including: single and small entrances, like those of most invertebrates, or clumped multiple and larger entrances, typical of most rodent burrows. To simplify analysis, burrow entrances were separated into two groups using a diameter of 2.5 cm as threshold. The threshold is based on the relation of a burrow entrance diameter to the hosts' body width (e.g. Gabet et al., 2003;Vleck, 450 1981). The smallest entrance diameter reported for mammals is 2-3 cm for Mus musculus (house mouse, Table   S1). Here we follow the approach of Kelt et al. (2004) and define a threshold diameter of 2.5 cm to include grass mice. Given this, entrance diameters < 2.5 cm were classified as from "small animals", mostly invertebrates, and ≥ 2.5 cm for "large animals", mostly mammals (Fig. 9).
Field observations in Santa Gracia revealed the presence of the lizard Callopistes maculatus identified by 455 characteristic feces adjacent to burrow entrances, located next to a dry riverbed. As for burrowing birds, the owl   Across slopes, more entrances are found on the north-facing slopes than on the south-facing slopes in the two 470 northern sites. In the mediterranean climate of the NP La Campana, the number of registered entrances is the same across both slopes. For the southernmost NP Nahuelbuta available data is too sparse to allow any conclusion from visual inspection. Along the climate gradient, more entrances are found in arid (n=443) or semi-arid sites (n=624), compared to the mediterranean (n=362) or humid-temperate study areas (n=136).

475
The number of entrances per slope and along the climate gradient could be related to incoming solar radiation.
The number of entrances is plotted together with the calculated summer and winter radiation energy on north and south-facing slopes (Fig. 8). The maximum of the solar radiation at noon throughout a year varies with latitude, and the amount decreases from 26° S (PA) to 38° S (NA). During the summer, at latitudes >30° S (Santa Gracia) the radiation is higher on north-facing slopes than on south-facing slopes. During winter, there is higher solar 480 radiation on all north-facing slopes, compared to south-facing slopes. The amount of burrow entrances decreases from the semi-arid site (Santa Gracia) towards the south. Within each site, the majority of the small animal entrances is located in accordance with the higher radiation on north-facing slopes. The number of large animal burrow entrances is also higher on north-facing slopes at the arid and semi-arid sites. In contrast, at the mediterranean and humid-temperate sites, large animal entrance numbers are higher at the south-facing slopes.

485
Similar to the incoming solar energy, MAT is high at the arid site, peaks at the semi-arid Santa Gracia site, and then decreases towards the south (Fig. 10). The MAT in Chile varies from desert maximum values of 20° C in the north to temperatures around 0° C or below in the south of Chile. Small animal and large animal entrances follow https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License. the same pattern decreasing towards the south with decreasing MAT. MAP counteracts this pattern (Fig. 10). The MAP in Chile varies from near zero in the north to above 200 mm in the humid south.

Burrow dimensions
The burrow dimensions measured were entrance diameter, minimum tunnel length and tunnel orientation. The correlation of the diameter and the minimum tunnel lengths are positive for both groups (Fig. 4), steeper for small 500 animals (slope: 5.84) and shallower for large animal entrances (slope: 3.83). The longest straight tunnel length was 47 cm with a diameter of 3.5 cm. For invertebrates the longest detected tunnel length was 16 cm with 2 cm diameter. The minimum burrowed depth, calculated from minimum tunnel length and tunnel orientation, has a mean value of < 6 cm depth (range of mean values for small and large animals per plot: 1.6 -11.6 cm, Fig. 6).
Maximum values of calculated burrowing depth are < 26 cm for all sites and slopes. For reference, we plotted soil 505 thickness (Oeser et al. 2018) and mobile layer thicknesses from Schaller et al. (2018) from the same sites (Fig. 6).
Both the soil and mobile layer thicknesses have values around 20 cm depth in the north and increase towards the south reaching values around 60 to 90 cm depth.
Even through the number of entrances generated by small animals is always larger, the excavation volumes generated by large animals is about 1.5 magnitudes larger than small animal entrances on the same slope. Comparing across slopes in each study area the sum of minimum excavation rates (Fig. 7) are larger on the north-515 facing slope in NP Pan de Azúcar (south-facing: 0.007 m 3 ha -1 yr -1 ; north-facing 0.675 m 3 ha -1 yr -1 ) and also in Santa Gracia (south-facing: 0.135 m 3 ha -1 yr -1 ; north-facing: 0.988 m 3 ha -1 yr -1 ). In La Campana, the contrary pattern is visible, i.e. a larger excavation rate on the south-facing slope (south-facing: 1.459 m 3 ha -1 yr -1 ; north-facing 0.397 m 3 ha -1 yr -1 ). At the southernmost site, NP Nahuelbuta, the excavation rate is similar on both slopes (southfacing: 0.090 m 3 ha -1 yr -1 ; north-facing: 0.088 m 3 ha -1 yr -1 ). Focusing on the subgroup of small animal minimum 520 excavation rates, they are larger on the north-facing slopes at the three northern sites. For large animal minimum excavation rates a clear pattern is not observed with the available data.

Literature compilation on Chilean burrowing animals
In this literature review we focus on an inventory of burrowing vertebrates and invertebrates along the climate and 525 ecological gradient in Chile. Currently known burrowing species vary considerably along Chile, mammal diversity was highest for the mediterranean biome, invertebrates peaked in their diversity in the semi-arid areas.

Relationship between burrows and climate
Previous work suggested the composition of burrowing animals is related to a specific site's climate (Crawford et al., 1993;Paton et al., 1995). This finding is confirmed in our literature review. For mammals, we found a 530 maximum of 18 sympatric species in central Chile with a mediterranean climate in the lowland. Some burrowing species are adaptable to varying climate conditions such as the Chilean rodent cururo Spalacopus cyanus (Begall and Gallardo, 2000;Contreras, 1986), that adapted its body size and behavior to the local conditions, to adapt over a larger climatic gradient. According to their known distribution, foxes, Lycalopex griseus, L. culpaeus and small mammals Abrothrix olivaceus, A. longipilis, Phyllotis darwini, and Mus musculus seem also very flexible in the 535 climate in which they live. For invertebrates, the quantity of sympatric species has a maximum of 131 species in the region of Coquimbo (IV), part of the "Norte Chico" semiarid climate (Fig. 2). Some invertebrate groups are restricted to specific biomes such as darkling beetles in the arid and semi-arid regions, solifuges in semi-arid regions, trap door spiders or tarantulas in mediterranean areas, or dung beetles in humid settings. Fewer species, mostly ants, ant and camel (sun) spiders, as well as scarab beetles have a very wide distribution across several 540 biomes.

Burrowing animal excavation rates
The published global species specific excavation rates reach maximum reported values around 146 m 3 ha -1 yr -1 (Mountain pocket gopher, Ingles, 1952) (Table 1). More common are rates up to 25 m 3 ha -1 yr -1 for vertebrates.
For most burrowing invertebrates data about excavated volumes remains unknown. The species-specific values reported in literature are potentially higher than the average burrowing activity, as most study sites were selected because they had a high number of burrows of the target species present (rather than doing a plot scale inventory as in this study). Study outcomes may also vary to a certain degree by different sampling methods, as e.g. burrow 550 volumes may reach more than 20 times the volume of the mounding at the surface (Bétard, 2020). For a more https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License. general discussion readers are referred to previous work by Hausmann (2017) and Wilkinson (2009), and Smallwood and Morrison (1999a) on gopher burrows.

Observations of animal burrow entrances in four Chilean study areas
This study consolidated previous results from literature regarding burrowing animals' composition and spatial 555 occurrence variation across the four studied biomes along a climate gradient. In addition, our field study areas ( Fig. 3) documented the zoogeomorphic effects on hillslope mass transport at the animal community level and along the climate gradient. We found the distribution of burrow entrances is heterogeneous for both vertebrates and invertebrates burrowing species within a biome as well as along the climate gradient. More burrow entrances were found at north-facing slopes than on opposing slopes in two of the four biomes, and the highest excavation 560 rates were found in the semi-arid and the mediterranean biomes. Links of burrow activity with soil thickness, MAT, MAP and solar radiation are discussed in the following.

Distribution of burrow entrances
Varying soil thickness at top and bottom slope positions of hillslopes could influence the burrow entrance distribution. A thicker soil layer at bottom slope positions potentially provides more substrate and nutrients for 565 vegetation to grow bulbs and rhizomes in, that in turn may provide more favorable food conditions for burrowing herbivores. The soil thickness measurements on south-facing slopes of our study areas increased from arid to humid-temperate sites from approximately 20 cm to 90 cm and also increased from top to bottom measurements within the slopes (Oeser et al. 2018, Fig. 6). Despite these findings, we did not find a difference in the distribution of burrow entrances within top and bottom slope positions (Fig. 5). Soil thickness appears not to be a limiting 570 factor for burrow entrance distribution between top and bottom slope.
Burrow entrance distribution could also be influenced by the incoming solar energy, that is higher on north-facing slopes in comparison to south-facing slopes in the southern hemisphere and has an effect on the available energy and also on water balances (Gallardo-Cruz et al., 2009). The effect of the hillslope aspect (and hence available solar energy) on animal excavation activity has been addressed in previous work. In the northern hemisphere in 575 an alpine region in Canada the total sediment displacement of animals has been highest on east and south-facing slopes, compared to north-and west-facing ones (Hall and Lamont, 2003). Furthermore, in a Chilean animal trap study at a semi-arid north-central site, north-facing slopes have had the doubled abundance of small mammals, mostly burrowing species, compared to south-facing slopes (Jiménez et al., 1992). Our findings of a higher number of entrances on north-facing slopes than on south-facing slopes were consistent with both studies in the arid and 580 semi-arid biomes (Fig. 5).

Differences in the distribution of large and small animal burrows
We observed small animal burrows to be mostly distributed evenly over the plot surface areas, large animal burrows were mostly distributed in a clumped manner. Furthermore, large animal burrow entrances were present on all slopes but not in all plots. We adopted to this distributional difference by separation of the data into two which may overprint all volumes excavated by small animals that were up to three-quarters of locally counted 590 entrances (Fig. 7).
In addition, even single large animal dens, i.e. fox dens, may overprint all other burrowing activity. In NP Pan de Azúcar, adjacent to the plots within a 90 m radius, we observed larger holes that gave the impression to be dens or burrow entrances. 21 of these structures were measured in 2017, which were oriented in all directions. At the bottom of the slopes they were excavated farther than further uphill. Occasionally feces were found in front of the 595 entrances. The diameter of the structures was on average 15 cm (range: 7-57 cm), height was on average 14 cm (range: 4-45 cm), and the excavated depth was on average 34 cm (range: 10-76 cm).

Excavated depths
Animal groups have different burrowing depths. By burrowing, animals are able to maintain favorable conditions of temperature and humidity compared to above surface conditions (e.g. Bétard, 2020). Described burrowing 600 depths in literature for Chilean burrowing animals ranges for vertebrates from 0.25 to 3 m and for invertebrates between 0.3 and 1.0 m. The observed maximum excavated depth measured at all four study sites was < 26 cm ( Fig. 6). A previous study at the same study sites had revealed an increasing depth of the mixed layer from arid to more humid conditions, with a maximum of 85 cm in NP La Campana, followed by 70 cm in NP Nahuelbuta, 45 cm in Santa Gracia and 17.5 cm in NP Pan de Azúcar (Fig. 6) (Schaller et al. 2018). The calculated minimal 605 burrow depth reached the soil mixing depth values from literature only at the arid site. Based on the maximum burrow depths from literature for the differing taxa in Chile, we presume that real burrow depths are similar to the mobile layer depths reported in the study of Schaller et al. (2018). The observed result is likely an underestimation caused by the measurements procedure, i.e. the inability to easily determine the full burrow extent deeper than the first curvature after the entrance.

Relationship between burrows, climate, and climate variation
A correlation between the burrowing activity and regional climate was hypothesized. As indicators of gradients in climate along the extent of Chile we investigated incoming solar radiation, MAT and MAP. The burrowing activity of both, vertebrates and invertebrates, along the climatic gradient appeared to be linked to all three variables (Fig.   8, 10). The quantity of animal burrow entrances across the study sites showed the highest activity in the semi-arid 615 Santa Gracia study area (Fig. 3), followed by arid NP Pan de Azúcar, NP La Campana and lowest activity for NP Nahuelbuta. Because of the clumped distribution patterns of large animal entrances (see discussion above), the small animal entrances resemble a more stable indicator. Previous work has found a temperature gradient to have an effect on arthropod density (Tiede et al., 2017) with fewer species at lower mean temperatures. Our results support this finding.

620
As previously stated, species groups with the largest zoogeomorphic effects vary with climate, e.g., in humid areas dominant bioturbators are earthworms (Lumbricidae), whereas in arid areas the effects of mammals are highest (Paton et al., 1995;Whitford and Kay, 1999;Wilkinson et al., 2009). In our results, the minimum excavation rates reached a maximum for invertebrates in the semi-arid Santa Gracia, and for vertebrates the highest impact was found in mediterranean La Campana (Fig. 7). Nevertheless, a more detailed understanding of animal activity, 625 especially within the invertebrates, is necessary to assign the impact to specific groups.
In addition to the gradual change in MAP with latitude ( Fig. 1), climate variation due to ENSO cycles can generate variations in the local climate conditions. ENSO consists of increased precipitation events in periodic intervals of, https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License. on average, of every four years and lasting 2-10 years (Cane, 1983). The temporarily increased water availability during ENSO enhances vegetation in arid regions (Gutiérrez et al., 2000;Jaksic, 2001), including rainfall triggered 630 seed production (Gutiérrez et al., 1997). The variation in changing food availability from increased vegetation affects animal densities over several orders of magnitude (Fuentes and Campusano, 1985;Holmgren et al., 2006;Jaksic, 2001;Lima et al., 1999;Meserve et al., 1995Meserve et al., , 2011Milstead et al., 2007;Pearson, 1975;Previtali et al., 2009Previtali et al., , 2010. Rodent populations are described to react with "ratadas", very high abundances of rodents, to mast seeding production, triggered by favorable conditions for plants. This applies, for example, to the granivore rat 635 Oligoryzomys sp. and Abrothrix sp. that feed on bamboo seeds (Chusquea sp.) in Chile (Boric-Bargetto et al., 2012;Gallardo and Mercado, 1999;González et al., 2000) and also to other small mammal communities in Northern Chile (e.g. Jiménez et al., 1992;Milstead et al., 2007). Epigean arthropod assemblages of the Chilean coastal desert have also reacted to an ENSO event with an augmented number of individuals in most of the recorded taxa, with emphasis on Hymenoptera (Formicidae) and Coleoptera (Tenebrionidae) (Cepeda-Pizarro et 640 al., 2005). The years during with data acquisition were conducted in this study did not cover an ENSO cycle, so we cannot corroborate the results of previous studies with our observations.

Minimum excavation rates
The species independent excavation rates in four biomes studied in Chile (Fig. 3) were quantified. Although the field observations presented provide only initial estimates, they nevertheless highlight trends and patterns present 645 at the hillslope scale. Largest total sediment volumes were displaced in the south-facing mediterranean slope (1.46 m 3 ha -1 yr -1 , NP La Campana). Along the climate gradient the minimum excavation rates increased from arid to mediterranean, with a decrease at the humid-temperate site such that values were 0.34 m 3 ha -1 yr -1 for the arid site, 0.56 m 3 ha -1 yr -1 for the semi-arid site, 0.93 m 3 ha -1 yr -1 for the mediterranean site, and 0.09 m 3 ha -1 yr -1 for the humid-temperate site. Compared to published species-specific excavation rates (Table 1), our Chilean study area 650 estimates are in the range observed for invertebrate species. As discussed above, the direct comparison of calculated excavation rates with species specific rates from literature has to be evaluated with caution. On one hand, the presented animal community values are likely to present a more representative excavation rate, than the numbers of species-focused studies. On the other hand, the presented values (0.09 to 1.45 m 3 ha -1 yr -1, Fig. 7) are minimum excavation rates limited by the study design. The local species independent excavation rates are likely 655 to be higher, but currently unquantifiable with available observations. For the invertebrates, observations resulted in a range of excavation rates between 0.003 to 0.063 m 3 ha -1 yr -1 (Fig.   7). The values of identified small animal burrow entrances were likely more reliable from the arid and semi-arid sites, because of the bias in litter cover at the southern sites. The minimum excavation rates obtained were up to one order of magnitude larger than the reported value for alkali bees as single invertebrate species (0.005 m 3 ha -1 660 yr -1 ) (Cane, 2003). The order of magnitude for excavation rates by small animals in this study are two to three orders below the excavation rates of large animals. Regardless, the number of entrances and excavated volumes of small animals also have important impacts on other factors than downhill erosion, such as soil mixing, water runoff or infiltration, bulk density and a range of further secondary effects of bioturbation (Bétard, 2020; further discussion about this topic in Birkby, 1983;Carlson and Whitford, 1991;Jouquet et al., 2006;Lobry de Bruyn and 665 Conacher, 1990;Thorp, 1949;Whitford, 1996). https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License.

Study caveats and future research needs
In our literature review large gaps of knowledge appeared regarding excavating habits in most species, distributional ranges of species, species' densities, excavated masses or volumes, and burrowing depths. Without better knowledge of these values the quantification of downhill erosion by burrowing animals is limited. The 670 method presented in this study is offered as an approach to deduce minimum excavation rates for different animal burrow entrances, independent of animal taxa. In the following, improvements upon the methods used here are discussed.
One issue encountered is the increasing ground cover by vegetation that occurs towards the south and reduces the detection of entrances in our two southernmost study areas (Fig. 3). In NP La Campana the ground was covered 675 with litter where in the southern most NP Nahuelbuta dense understory vegetation, mainly bamboo twigs and litter, covered the ground. The amount of burrow entrances is very likely to be reduced and also the minimum excavated volume is presumably underrepresented. Removal of the ground litter would imply a massive disturbance of the plot surface, that would reduce the probability of repeated observation of undisturbed burrowing behavior.
A second issue for consideration is the seasonality of a species' activity needs to be taken into account when 680 estimating a species' abundance and effect on burrowing. Individual species' activity patterns in Chilean vertebrates and invertebrates may result from species specific reproduction cycles (e.g. Yunger et al., 2002).
Different groups have activity peaks over different times within seasons, making repeated measurements during a year time span favorable. Furthermore, repeated measurements over several year variations in climate variations (e.g., ENSO cycles) and vegetational change are recommended, and dry periods with low plant abundances, mass 685 seedlings, etc. should also be considered. Repeated measurements could also give insight for arid sites. For example, it currently remains unclear if the burrow entrance amounts in NP Pan de Azúcar are high because rain does not reset the yearly burrowing activity records, or the density of burrowing individuals is comparatively high, even if caused by fewer known species numbers.
A third consideration is the high variation around the positive correlations between diameter and tunnel lengths 690 (Fig. 4). This may be an artefact of measurements limited to the first obstacle (see methods section). In addition, the excavation rate estimate is only a minimum. With measurements including the curvatures of the tunnels, the slopes of the correlation are expected to be steeper, possibly more similar to the regression shown for small animals. Identification of the burrow builders on species level and known burrow dimensions for the respective species would also be an approach to improve the minimum estimation of excavation rates.

695
Finally, comprehensive datasets are valuable to improve the conceptual understanding of downhill sediment fluxes. To upscale the conclusions from the hillslope scale to catchment or regional scales, further parameters obtainable with methods from other disciplines would improve the results. Soil texture measurements, as it influences the amount of energy spent, and therefore the overall activity of the respective burrowing species (Price and Podolsky, 1989) could provide improved insights into burrow distributions. Geomorphological techniques 700 (see review by Viles, 2020) would provide additional information, such as the form of sediment traps beneath monitored areas that enable the estimation of total material moved downhill. Soil mixing rates and differences in mixing rates as dependent on the depth, and over time spans of decades to millennia, could be resolved by luminescence or cosmogenic nuclide techniques applied to depth profiles (Gray et al., 2020;Reimann et al., 2017;Schaller et al., 2018). In addition, photogrammetry of mound modifications over time would be useful for 705 estimating burrow related sediment fluxes at the surfaces from loose soil patches. UAV remote sensing techniques could serve to more rapidly estimate burrow density in areas without vegetation cover (Lawton et al., 2006). Lidar https://doi.org/10.5194/bg-2021-75 Preprint. Discussion started: 6 May 2021 c Author(s) 2021. CC BY 4.0 License. data measuring vegetation structure could facilitate to identify the occurrence and abundance of vertebrate (Müller et al., 2010) and invertebrate assemblages (Müller et al., 2018;Vierling et al., 2011).

710
This review provides a synthesis of existing knowledge and new observations concerning the zoogeomorphic effect of burrowing animals in Chile. This is a first step of quantification of the complete burrowing animals' community at given sites on a hillslope scale and identified patterns that are some of the principal components that drive differences of animal burrowing effects on hillslopes along a climatic gradient. The quantification of zoogeomorphic effects on hillslope scales is in its infancy and additional observations are needed. The key findings 715 of this study are: This study documents 45 burrowing vertebrate and 345 burrowing invertebrate species and species distribution summaries for both groups (Figs. 1D and 2). Most burrowing mammal species are present in the mediterranean climate of Chile. For invertebrates most known burrowing species are present in the semi-arid region of Coquimbo (IV) and adjacent regions. A second invertebrate peak of high species diversity is in the humid-temperate area 720 around the region of Biobío (VIII).
For the Chilean study areas, minimum excavation rates range between 0.34 m 3 ha -1 yr -1 for the arid site, 0.56 m 3 ha -1 yr -1 for the semi-arid site, 0.93 m 3 ha -1 yr -1 for the mediterranean site and 0.09 m 3 ha -1 yr -1 for the humidtemperate site, with the latter being vastly underestimated due to vegetation and litter cover. Relative to single species rates from literature, our calculated excavation rates are in the range of previous invertebrate rates and low 725 for rates of vertebrates.
In this study we discuss the relationships between burrowing animal activity and different metrics such as topographic slope and aspect, and latitudinal site specific variations in solar radiation and climate. We found more burrow entrances on north-facing than on south-facing slopes. On the climate gradient animal burrow entrances showed the highest activity in the semi-arid Santa Gracia study area (Fig. 3), and a gradual decrease in NP La

730
Campana and lowest activity in NP Nahuelbuta -following decreasing MAT and increasing MAP as well as decreasing incoming solar energy towards the south. Arid NP Pan de Azúcar reflected less activity than the semiarid region, potentially due to the extreme conditions that are more challenging for animals to adapt to.

Author contribution
KÜ and TAE conceived the study. KÜ designed the experiments and carried them out, with contribution of LP.

735
KÜ, SB and JPA conducted the literature survey. KÜ prepared the manuscript and figures with contributions from TAE and JPA. All co-authors reviewed and improved the final manuscript.