Landscape and microhabitat features determine small mammal abundance in forest patches in agricultural landscapes

Introduction

Land use change, intensification of agricultural practices, fragmentation of natural habitats and the consequent alteration of biological communities lead to a widespread decline in farmland biodiversity, measured across many different taxa (Batáry et al., 2020; Benton, Vickery & Wilson, 2003; Geiger et al., 2010; Kleijn et al., 2011). In agricultural landscapes, a principal task for the conservation of farmland biodiversity and related ecosystem services is the understanding of the relationships between habitat features, habitat pattern and species population dynamics (Fahrig, 2007; Le Roux et al., 2008; Tscharntke et al., 2005). The need to preserve and enhance agricultural landscape diversity via management of non-crop habitats (i.e., field margins, woodlots, hedgerows, and perennial grasslands) to support high levels of agricultural biodiversity is largely confirmed (Carvell et al., 2011; Dainese et al., 2015; Dainese et al., 2016; De Simone, Sigura & Boscutti, 2017).

Small forest patches may deliver several important ecosystem services to human society, but they receive little attention compared to large forest ecosystems. In particular, small forest patches represent central elements in the long-standing conflict between the need of agriculture productions and biological conservation (Biber et al., 2015; Benayas, Bullock & Newton, 2008).

In a recent review, Decocq et al. (2016) focused on the ecosystem services delivered by small forest patches of agricultural landscape highlighting their potential to support biodiversity both in situ (forested patch) and in the proximity area enhancing landscape connectivity. Here, patch structural features (pH of soil, nutrient availability, light) and habitat configurations patterns (i.e., forest patch size, patch isolation) were defined as drivers for internal species richness and community composition.

Rodents and other small mammals are rarely target of conservation studies since they are often considered as pest or menace to production (Brown et al., 2007; Butet, Paillat & Delettre, 2006; Dickman 1999; Putman, 1989), thus associated with negative impacts on human activities (Capizzi, Bertolino & Mortelliti, 2014). However, small mammals are pivotal for important ecosystem functions (Dickman 1999). They contribute to soil aeration (Laundré & Reynolds, 1993) and play an important role as consumers of weeds and insects (Bricker, Pearson & Maron, 2010; Gliwicz & Taylor, 2002). Small mammals are crucial for diaspore dispersal of plants (Bogdziewicz et al. 2020; Steele, Wauters & Larsen, 2005) and fungi (Capizzi & Santini, 2007; Sieg, 1987). In several ecosystems they sustain food chains, representing the main prey biomass for several predator groups, such as birds, reptiles, and mammals (Martin, 1994; Salamolard et al., 2000; Šálek et al., 2010) and greatly contribute to the overall complexity of foodwebs (Butet & Leroux, 2001; Korpimaki & Norrdahl, 1991). For instance, low prey densities have been proved to affect raptor nesting densities in agricultural landscapes (see e.g., Michel, Burel & Butet, 2006).

The populations of some small mammal species have experienced a decline, sometimes drastic, probably because of agriculture intensification (Macdonald et al., 2007; Palmeirim, Santos-Filho & Peres, 2020), acting at both landscape and patch levels. In particular, the intensification of agricultural landscapes has been shown to decrease the density of rare and stenoecious species in favor of generalist ones (Burel et al., 2004; Gentili, Sigura & Bonesi, 2014; Millàn de la Peña et al., 2003), provoking idiosyncratic responses. For example, Silva, Hartling & Opps (2005) and Serafini, Priotto & Gomez (2019) found that both structural complexity (e.g., coverage and shape of forest patches) and landscape heterogeneity (e.g., different land-use covers) promoted small mammal species diversity. On the other hand, Michel, Burel & Butet (2006) demonstrated that, in areas where land-uses are relatively homogeneous (less landscape heterogeneity), species evenness but not species richness was negatively affected by the increasing of agricultural land-use intensification. At the patch scale, forest-floor characteristics, presence of coarse woody debris, understory vegetation, canopy composition (Carey & Harrington, 2001), forest age, vegetation complexity (Pearce & Venier, 2005), presence of deadwood (logs and rotten trunks) and litter structure (Cox, Dickman & Cox, 2000; Kemper & Bell, 1985; Marsh & Harris, 2000; Simonetti, 1989; Szymañski et al., 2020; Yahner, 1986) can also influence small mammal occurrence. A high microhabitat complexity through high shrub diversity, high vegetation cover and low percentage of ground covered by bare-soil are known to increase overall small mammal abundance and species richness (Gelling, Macdonald & Mathews, 2007; Silva & Prince, 2008; Szymañski et al., 2020).

Most of the previous studies evidenced strong effect of environmental drivers on small mammal. Nonetheless, comprehensive assessments at multiple ecological scales are still scarce and focus on species autecology. In this light, comparing species with different behavior (i.e., generalist vs specialist) should be encouraged to get new insights into future agricultural landscape management.

In this study we investigated the effects of environmental features on small mammal abundance paying particular attention to forest specialists and habitat generalist species. We considered three different ecological scales: microhabitat, habitat (patch), and landscape. In particular, we tested the effects of landscape composition and configuration, forest structure and forest ground features on two small mammal species (i.e., Apodemus sylvaticus, and A. agrarius) in remnants of wood vegetation along an intensification gradient of agricultural land use. These species were chosen as they represent typical species of lowland woodlots, very common in the study area (Dorigo, 2018) and can easily be captured with live-capture systems.

Specifically, we ask two questions: (1) What are the relative and independent effects of heterogeneity of landscape elements, composition and configuration on small mammals abundance in woodlots within agricultural landscapes? (2) Does the effect of these mosaic properties differ between species with different ecological needs?

We expected to find a strong interplay between landscape composition and habitat features in determining small mammals species distribution and abundance. In particular, we expected to find contrasting responses between the species considered, hypothesizing that the generalist A. sylvaticus could be more affected by landscape composition and configuration rather than local forest characteristics.

Materials and methods

Results

In all sampling sessions 868 captures were recorded, of which 36.5% were recaptured individuals. The total capture effort was 3024 trap-nights, with a capture index (Ic = N ×100 catch / trap-nights) of 28.7. Among the catches, 55.3% was represented by A. agrarius, 40.1% of A. sylvaticus, 4.1% by Rattus norvegicus and the remaining 0.5% by R. rattus, Microtus arvalis and Crocidura suaveolens.

Apodemus sylvaticus was caught in all sampling sites while A. agrarius occured in 16 sites and, together, these species represented about 95% of the individuals caught. Both rodent species were caught in all sampling sessions. For all the investigated scales the null models showed a low weight (weight <0.001) and were not considerable as plausible (delta AIC >12).

Overall model

The multi-model inference applied to the selected variables from each environment structural level (overall model) showed one supported model both in 250 m and 500 m buffers (Tables 7, 8). The models confirmed the importance of the extension of permanent crops and the undecomposed litter layer, highlighting a significant interaction of the study species with the variables (p < 0.001; RVI>0.7). In particular, where the landscape had a higher percentage of permanent crops, the abundance of A. agrarius increased, whereas the abundance of A. sylvaticus decreased (Fig. 1A). Instead, an increase of undecomposed litter favors A. sylvaticus, decreasing the abundance of A. agrarius (Fig. 1B). The models including variables within the 250 and 500 m buffers explained respectively 56 and 51% of the total variation in small mammals abundance.

Table 7:

List of plausible models performed with multi-model inference in 250 m buffer.

The estimates are reported: the intercept (Int), the variables considered in each model (prm_250 = percentage of surface area occupied by permanent crops; spcs = species; und = undecomposed litter percentage), R², AIC, ΔAIC and model weight (wi). Relative Variable Importances values and significance are shown in bold.

	(Int)	prm_250	spcs	und	prm_250:spcs	spcs:und	R2	AIC	delta	weight
	20,68	0,1873	+	-0,267	+	+	0,56	214,8	0	0,748
		0,73***	0,91.	0,81***	0,7***	0,75***

DOI: 10.7717/peerj.12306/table-7

Table 8:

List of plausible models performed with multi-model inference in 500 m buffer.

The estimates are reported: the intercept (Int), the variables considered in each model (prm_500 = percentage of surface area occupied by permanent crops; spcs = species; und = undecomposed litter percentage), R², AIC, ΔAIC and model weight (wi). Relative Variable Importances values and significance are shown in bold.

	(Int)	prm_500	spcs	und	prm_500:spcs	spcs:und	R2	AIC	delta	weight
	101,5	1,267	+	-1,295	+	+	0,51	336,9	0	0,904
		0,99***	1	0,97**	0,98**	0,91*

DOI: 10.7717/peerj.12306/table-8

Discussion

Our findings shed new light on the interplay between different ecological scales in determining small mammals abundance and distribution. We found species with different ecological niches to differ significantly in landscape and habitat features, suggesting the occurrence of alternative strategies for species coexisting in fragmented forest patches in agricultural landscapes.

Considering the contribution of each variable group and their selection, we found both landscape composition and microhabitat features to significantly affect species abundance, with contrasting species responses. On the contrary, wood structure did not directly influence the abundance of the two studied species, not confirming previous studies evidencing significant relationships between abundance of small mammal forest species and forest tree density (Capizzi & Santini, 2007) and size (Capizzi, Battistini & Amori, 2003). Albeit not significant, our results also highlighted positive relationships between small mammal abundance and tree density, as well as number of tree species and number of shrub species, that could be interpreted as direct relationship with the complexity of the ecological niche. Several studies reported a strong relationship between deadwood abundance and the presence of small mammals. In fact, deadwood represents a source of potential shelters and hosts large invertebrate communities which are potential sources of food (Amori et al., 2015; Bellocq & Smith, 1997; Bowman et al., 2000; Harmon et al., 1986; Kemper & Bell, 1985; McCay, 2000; Miller & Getz, 1977; Szymañski et al., 2020; Tallmon & Mills, 1994; Yahner, 1986). Several studies also indicate that the decay stage may be important in habitat selection of small mammals (Bowman et al., 2000; McCay, 2000). Availability of understory cover is also a key condition for small mammals (Barry, Botje & Grantham, 1984; Bellocq & Smith, 1997; Carey & Harrington, 2001; Cox, Dickman & Cox, 2000; Marsh & Harris, 2000; Simonetti, 1989). Probably its importance is connected with food availability and the related possibility to reduce either the areas of predation (Bellocq & Smith, 1997; Miller & Getz, 1977; Simonetti, 1989) or the competition, by expanding the available space (Montgomery, 1980). Regarding microhabitat structure, the bare soil cover, presence of deadwood and litter structure are known to affect small mammal populations (Amori et al., 2015; Cox, Dickman & Cox, 2000; Denny et al., 2021; Kemper & Bell, 1985; Marsh & Harris, 2000; Simonetti, 1989; Yahner, 1986). Our results confirmed these findings, but highlighted the pivotal role of the litter decomposition stage. Litter decomposition rate is driven by environmental conditions, chemical composition of soil and intensity of soil organism activity (Aerts, 1997; Harmon et al., 1986).

The speed and degree of degradation of leaves depends mainly on the species of tree (He et al., 2019; Mathews & Kowalczewski, 1969; Nykvist, 1962) and other environmental variables such as temperature and humidity. In addition, the presence of large invertebrates such as earthworm, millipedes and other taxa is also relevant, enhancing the degradation of organic matter (Bocock, 1964; Mathews & Kowalczewski, 1969; Tresch et al., 2019). The dynamics of the soil degradation process is connected to invertebrate biodiversity: well developed temperate woodland soils can host up to a hundred species of animals (Anderson, 1975; Tresch et al., 2019). Moreover, leaf litter quality affects numerous trophic levels such as insects, other invertebrates (Koivula et al., 1999; Sabo, Soykan & Keller, 2005) and small mammals (Canova & Fasola, 1991; Hansson, 1978; Kaminski et al., 2007). Litter cover affects the use of habitat by black rats (Cox, Dickman & Cox, 2000), which spend most of the time foraging on the ground, probably obtaining food from the litter (Cox, Dickman & Cox, 2000).

A dense litter cover would produce suitable mesic conditions in soil and litter ensuring appropriate microenvironments for litter-dwelling arthropods used as food (Yahner, 1986). Wood mice catch and find invertebrate prey, which are an important source of food, under litter cover (Piper, Lewis & Compton, 2014). The microclimate conditions of the litter can influence movements of small mammals, whom sounds are more difficult to hear in moist than in dry litter, influencing predation (Vickery & Bider, 1981). Predation risk is considered one of the most important factors determining selection of microhabitat in small mammals (Bellocq & Smith, 1997; Fragoso, Santos-Reis & Rosalino, 2020; Simonetti, 1989).

Despite this evidence, we found a different interaction of the two considered species with the environment conditions. In fact, abundance of A. agrarius was inversely related with the percentage of undecomposed litter, while the opposite trend was observed in A. sylvaticus.

Probably A. agrarius is more dependent on litter quality because it generally avoids xeric conditions, as reported by Yahner (1986) for other species in relation to the moisture of the habitat. This species is known to prefer habitats with dense soil cover and it usually moves through the litter stratum because of its fossorial behaviour (Kuncová & Frynta, 2009), probably to escape predators (Orlandi & Paolucci, 2004).

On the contrary, A. sylvaticus occurs in both woody and in open habitats (Heroldová et al., 2007) avoiding cut crops or cut set aside (Tattersall et al., 2001) and bare soil (Tew, Todd & Macdonald, 2000), showing this species is poorly influenced by the structure of litter, in disagreement with what is reported in other studies (e.g., Balestrieri et al., 2017).

Moreover, A. sylvaticus is apparently not affected by wood characteristics probably because this ubiquitous species prefers the diversity of the complex mosaic created by agricultural landscape rather than the continuous forest habitat (Geuse, Bauchau & Boulengé, 1985; Kozakiewicz et al., 1999).

It is well known that composition of landscape can influence the population dynamics of small mammals (Fischer & Schröder, 2014; Silva, Hartling & Opps, 2005). Our results confirmed this evidence pointing out the positive role of some non-natural habitats (i.e., permanent crops) in agreement with what was found in other studies (Fitzgibbon, 1997; Macdonald et al., 2007; Tattersall et al., 1999; Vieira et al., 2009). Even though some studies showed small mammal abundances to be related with landscape structure (Fischer, Thies & Tscharntke, 2011; Kozakiewicz et al., 1999; Macdonald et al., 2007; Van Apeldoorn et al., 1992; Vieira et al., 2009), in our study, no significant relationship emerged considering landscape configuration. We observed a positive relationship between the abundance of A. agrarius and the presence of permanent crops, despite the percentage of natural woods was relatively low in the studied area (ca. 8% on average). This result suggests that permanent woody crops, such as poplars and orchards, may be important habitats for the study species, especially if placed in the surroundings of woodlots, offering canopy cover and in some cases a dense leaf litter. On the contrary, we did not observe such strong relation for A. sylvaticus, which seems less influenced by the presence of permanent crops in the landscape but with a negative relationship. It seems that A. sylvaticus in Mediterranean areas uses mostly nests positioned far away from orchards, probably to avoid predation (Rosalino et al., 2011), although other studies show that the species is quite common in poplar plantations (Balestrieri et al., 2017) and orchards (Dickman & Doncaster, 1987).

The differences found for the two species could be linked to their ecological requirements.

Our results suggest that A. agrarius, a species connected to the presence of wood mosaic, uses permanent plantations in highly modified landscapes probably as surrogate of forest habitats. The abundance of this species was negatively related with the amount of arable land so that it becomes absent in very simplified landscapes (Fischer, Thies & Tscharntke, 2011). Then, permanent crops could become an important secondary habitat for this rodent in intensive agricultural landscapes.

A. sylvaticus, instead, has a broad ecological niche and can exploit many different habitats (Halle, 1993; Michel, Burel & Butet, 2006), such as fallow lands (Heroldová et al., 2007; Macdonald et al., 2007) and crops (see for instance Tattersall et al., 2001). This species has high dispersal abilities (Marsh & Harris, 2000) and is supposed to have higher abilities to navigate in crops (Sozio, Mortelliti & Boitani, 2013).

In cultivated landscapes, A. sylvaticus moves from hedgerows to crop fields according to seasons (Ouin et al., 2000). At the time of the crop, wood mice massively overrun the fields and remain there until the time of ploughing (Heroldová et al., 2007); during winter more mice are found in field edges than in the center of fields, and when the crop are cut down they prefer hedgerows, woodland edges (Macdonald et al., 2000; Tattersall et al., 2001), woods and fallow lands (Tew & Macdonald, 1993). This allowed them, to avoid predation, which can bring high rates of mortality (Tew & Macdonald, 1993). The wood mouse is therefore well represented in intensified sites because it certainly benefits from landscape heterogeneity as it can find and exploit many resources (Michel, Burel & Butet, 2006). As observed by Bellows, Pagels & Mitchell (2001) for other rodent species, the lack of relations between the species and microhabitat characteristic is probably because of its ability to exploit a variety of sources.

We did not observe significant relationships between species abundance and percentage of settlements in the surroundings, within the considered buffers. This fact may be interpreted as the low presence of urban areas, reduced in the study sites to a few scattered agglomerates of buildings (ca 8.5% in a 500 m buffer).

Conclusions

Our findings revealed that, in the study areas, landscape composition rather than landscape configuration significantly affect small mammals species in the considered agricultural landscapes, which, along with microhabitat features and species interaction, drives species abundance.

Our results showed that composition of agricultural landscapes is not perceived by all species in the same way. Species respond to each ecological scale in relation to their own ecological requirements, spatial behaviors, mobility, and dependency on habitat (Kozakiewicz et al., 1999; Serafini, Priotto & Gomez, 2019). The maintenance of wider woodlots (or hedgerows) is, hence, necessary for several aims, as: (i) protecting wildlife species with different ecological needs, (ii) maintaining/enhancing habitat continuity, and (iii) avoiding structure simplification (Dondina et al., 2016). Functional connectivity among habitat patches is proven to be one of the most important factors affecting the presence of small mammals in urban areas (Fitzgibbon, Putland & Goldizen, 2007) but also matrix composition and management is fundamental since forest species can cross agricultural surroundings (Mortelliti et al., 2013; Paise, Vieira & Prado, 2020). Our results showed that the presence of anthropogenic habitats in the landscape with plausible landscape connectivity functions (such as permanent plantations) can also affect small mammals.

Studies have been carried on connectivity in the agricultural matrix, but the effect of forest plantations on functional connectivity has received limited attention (Mortelliti, Westgate & Lindenmayer, 2014).

Permanent tree crops, especially those aimed at producing woody biomass, have a negative effect on biodiversity when they replace native forest vegetation (Brockerhoff et al., 2008; Greene, Martin & Wigley, 2019). It is already commonly believed that their effect may have less impact than that of the agricultural matrix, since they are working as a semi-permeable barrier for organisms, representing a compromise between maintaining the economic value of an area and reducing the effect of habitat fragmentation, increasing connectivity (Norton, 1998; Brockerhoff et al., 2008; Vanbeveren & Ceulemans, 2019). Landscape connectivity as well as the habitat function, could greatly increase if management rules are properly applied (Norton, 1998; Lindenmayer, Hobbs & Salt, 2003; Zitzmann, Reich & Schaarschmidt, 2021). This process may guarantee, for example, the permanence of mature plants within plantations (Hanowski, Niemi & Christian, 1997) or increase the shrub and herbaceous layers (Moser et al., 2002), whose positive effect on small mammals is already known (Balestrieri et al., 2017). Even the understory cover and litter can promote biodiversity, if they are not completely eliminated (Christian et al., 1997; Christian et al., 1998; Moser et al., 2002; Vanbeveren & Ceulemans, 2019).

Our results showed that several properties of agricultural land mosaics had a strong influence on small mammal abundance, and all of them are within the reach of farmers management. In this study, we get new insight to support the concept of ‘wildlife-friendly farming’ (Fischer et al. 2014), which emphasizes the ecological interactions of farmed and unfarmed coexisting areas and the improvement of natural and cultivated areas as strategy for an agricultural landscape management aimed to increasing biodiversity. This would contribute to the implementation of policy for the conservation of agricultural landscape at local and regional scales, and to farm landscape design.

Supplemental Information