Introduction

Digestive plasticity is a particular case of phenotypic plasticity that refers to changes in digestive features due to changes in both internal (e.g., energetic demands) and external environmental conditions (e.g., food quality and availability). Digestive plasticity has been demonstrated in several vertebrates, including fish (Olsson et al. 2007), amphibians (Naya et al. 2010), reptiles (Starck and Beese 2001), birds (Starck and Rahmaan 2003), and both small (Sabat and Bozinovic 2000; Naya 2008) and large (Jiang et al. 2002) mammals. A change in landscape structure often entails modifications in food quality, and this provides an excellent arena to explore digestive plasticity of resident species. Over the last quarter of the 20th century, the agricultural intensification and woodland fragmentation produced a major change of landscape in Europe (Stoate et al. 2010). Their consequences varied across wildlife species, producing for example, a long-term decline in the hare (Lepus europeaus) population (Edwards et al. 2000) and an increase in the amount of crop damage caused by the demographic and geographic expansion of deer populations (Putman and Moore 1998).

In this context, the European roe deer (Capreolus capreolus), originally considered a forest dweller, has been favored by the land-use changes and invaded a multitude of habitats including agro-ecosystems (Andersen et al. 1998). Although this ecological flexibility has been mostly ascribed to behavioral plasticity (Hewison et al. 2001), little is known about the physiological responses of roe deer (e.g., changes in digestive characteristics) that could potentially underlie the success of this mammal in agricultural landscapes.

The roe deer has been classified as a typical ‘browsing’ species in terms of the botanical composition of its diet (Tixier et al. 1997), its digestive morphology (König et al. 1976) and physiology (Clauss et al. 2001; Rowell-Shäfer et al. 2001; Behrend et al. 2004). While seasonal influences on the roe deer digestive tract have been demonstrated (Holand 1994), few differences between “forest” and “field” roe deer have been identified, apart from ruminal papillation patterns (Hoffmann et al. 1988). However, data on general digestive characteristics, like the weight of the digestive tract contents, have not been available so far. Such measurements are important for the understanding of seasonal and geographic differences in feeding strategy within the species (Holand 1994). Although little empirical information exists, roe deer of the agricultural landscapes may attain higher body mass (Fruzinski et al. 1982), presumably due to a higher diet quality (Hewison et al. 2009). Since these responses should be particularly apparent in the most plastic parts of the body (e.g., digestive system), the study of how the digestive tract of roe deer varies across a local landscape gradient should provide insights into the morphophysiological plasticity that is at the core of the species ecological flexibility.

In this work, we analyzed the influence of local landscape structure and diet quality on the mass of both the reticulorumen and distal fermentation chamber among juvenile and adult roe deer inhabiting an agro-ecosystem characterized by a gradient of land use, i.e., from a purely forested area to a predominantly agricultural plain. Since the quality of the diet is greater in more diverse cultivated landscapes (Abbas et al. 2011), we predicted that roe deer inhabiting the mainly cultivated areas of the study site would have relatively smaller proximal and distal fermentation chambers (i.e., chamber mass for a given body mass) due to a higher diet quality compared to those inhabiting the more forested areas of this same landscape.

Materials and methods

Study area

The study area (10,000 ha) is a mixed landscape of open fields and remnant woodland belonging to southwest European lowland colline downy oak forests (Bohn and Neuhäusl 2003), located in the Aurignac canton, southwest France (N43°13′, E0°52′). Today, about 33% of the total area is cultivated, mostly with wheat and barley (51%), sunflower (15%), maize (10%), soya (5%), sorghum (8%), and rape (4%). Meadows cover about 34%, hedges 7%, woodland patches 14%, and a central forest (630 ha) 7% of the study area. The human population is present in small villages and farms connected by an extensive road network (for a more detailed description, see Hewison et al. 2001).

Roe deer data

Roe deer samples (n = 47) were taken from legal hunts carried out by local hunting teams during autumn–winter (2005–2007), with most of the deer (76%, n = 36) being harvested from November 1 to January 31. Immediately after each shooting, the hunters recorded the precise locality where the deer were harvested on a map. Within the next few hours, our team was called to the locality to collect the deer. After converting the precise place of shooting to UTM coordinates, we initiated the sampling procedure, recording sex and mass (using a spring scale to the nearest 0.1 kg) and collecting the entire gastrointestinal tract and head in separate plastic bags. At the laboratory, after removal of mesenteries and adhering fat, the reticulorumen (RR) and the distal fermentation chamber (DFC, cecum and the ansa proximalis of the colon ascendens) were dissected, rinsed, dried with paper towels, and then weighed. The same two investigators (B.C and E.S) carried out all dissections in order to minimize inter-observer bias. Once the skulls were boiled in a 1% potassium hydroxide (KOH) solution, tooth wear was used to assign deer to one of the following age classes: 1.5 years of age (yearling), 2.5–3.5 years of age, 4.5–6.5 years of age, and older than 6.5 years of age. However, because errors are common using this technique (Hewison et al. 1999), we retained only two adult age classes for the analysis in this paper (yearlings and adults of more than 2 years of age).

Landscape description

In our study area, the annual home range size of roe deer ranges between 17 and 200 ha depending on landscape openness (Cargnelutti et al. 2002). To describe the local landscape structure plausibly used by a given deer, we used a Forest index (FI hereafter, see Hewison et al. 2001) that takes into account both the distance between patches of woodland and their surface area and varies from 0 (no wood within an 800-m radius) to 200 (entirely wooded). Within our study site, variation in this Forest index translates a gradient of increasing openness across the landscape, incorporating not only changes in woodland fragmentation but also the proportion of meadows and the degree of cultivation of the local habitat (Hewison et al. 2001, 2009).

Fermentation test

The RR contents of ten deer hunted in the most open local landscape and ten animals from the most forested one were submitted to in vitro fermentation to measure the rate of gas production at 4, 8, 12, and 24 h. The gas produced during fermentation represents a measure of food degradation and consists of nearly equal parts of the CO2 produced from a bicarbonate buffer reaction with the volatile fatty acids developing during fermentation and the waste gases of fermentation (Blümmel et al. 1999). Diet gas production is considered an excellent indicator of diet quality in ruminants (Hummel et al. 2006).

Statistical analysis

To investigate whether the mass of either RR or DFC were influenced by age, diet quality, and landscape structure, we evaluated a set of general linear models (GLM) in which the dressed mass of each digestive component (as response variable) was explained by the fixed factor age (two modalities: yearling or adult), the FI (as a proxy of landscape structure) and dressed body mass. Dressed body mass was included as a co-variate to control for the effects of the known allometric relationships between body parts (Christians 1999). Because our sample size was not large, and in order to avoid model over-parameterization, sex was not included as a factor. Indeed, the roe deer is only slightly sexually dimorphic (Andersen et al. 1998), and by including body mass as a co-variable, we controlled for correlated between-sex differences in the mass of digestive components. Additionally, because each year the precise location of hunting within the landscape varies and because we did not consider additional environmental factors which may explain any inter-annual variation in digestive morphology (e.g., density, precipitation, or temperature), in a initial approach the year of collection was included as a random factor in a general mixed linear modeling procedure to prevent pseudoreplication (Zuur et al. 2007). However, year of collection was subsequently excluded from the analysis since it did not improve the selected model.

To achieve a linear relationship between the response and explanatory variables and to minimize the residual variance, the Forest index had to be log-transformed prior to its inclusion in the set of candidate models. Once the best model was selected, we confirmed the general assumptions of linear models by exploring the residual pattern. The model selection procedure was based on a theoretic information approach based on the Akaike Information Criterion corrected for small sample sizes (AICc, see Burnham and Anderson 2002). In brief, for each of the candidate models we estimated the AICc and selected the model with the lowest AICc value. We then ranked the remaining competing models according to their AICc value and subsequently estimated their Akaike differences (Δi) with respect to the best model (lowest AICc) and the Akaike weight (w i) of each model (Burnham and Anderson 2002).

Finally, since gas production of each RR content was measured at different time intervals, differences in fermentation rate between the predominantly wooded or cultivated landscapes were evaluated by a linear mixed model in which landscape type was a two modality fixed factor and the repeated measurement was a random factor. All tests were performed using R version 2.10.1 (R Development Core Team 2009).

Results

The relative mass of RR (for a given body mass) of deer was mainly influenced by the age of animals and not by landscape structure (Table 1). Woodland cover, however, had a clear positive effect on relative DFC mass (for a given body mass, Table 2B; Fig. 2).

Table 1 Descriptive statistics for raw values of both reticulorumen (RR) and distal fermentation chamber mass (DFC, cecum, and the ansa proximalis of the colon ascendens) of yearlings (n = 20) and adults (n = 27) in roe deer from a mixed landscape in southern France
Full size table
Table 2 Model selection for explaining the observed variability in relative mass of the reticulorumen (A) and the distal fermentation chamber (cecum and the ansa proximalis of the colon ascendens, B) in yearling (n = 20) and adult (n = 27) roe deer harvested in a mixed landscape in southern France
Full size table

Our model selection showed that the best model for explaining the relative mass of RR included only the age term (wiAge = 0.55, Table 2A). Yearlings had lower RR mass for a given body mass than adults (by 1.4 times, Fig. 1), with 3.6% of the observed variability in relative RR mass explained by age. The second competing model that included the effect of woodland fragmentation had little support for being the best model (Δi > 2 units, see Burnham and Anderson 2002); however, after correcting for both body mass and age, we observed a slight positive influence (β = 9.36, SE = 25.71) of forest cover on relative RR mass.

Fig. 1
figure 1

Reticulorumen mass corrected for body mass as function of age (two classes: yearling if 1 ≤ age < 2, and adult if age ≥ 2 years) and reticulorumen mass corrected for body mass

Full size image

Independently of the age of animals, deer harvested in the more forested areas had relatively heavier DFC for a given body mass (β = 28.69, SE = 10.01) than their counterparts from the more open landscapes (Fig. 2). In fact, DFC weight for animals between 20 and 23 kg from the most forested locations (Log(FI) > 1.6) was on average 1.7 times heavier (mean = 66.6 g, SE = 4.84, min = 51, max = 80, n = 5) than their counterparts sampled in the open areas (Log (FI) < 1.6, mean 39.06 g, SE = 2.25, min = 30.4, max = 44, n = 5). Our model selection showed that the model that included the simple effect of landscape structure (wiForest index = 0.63, Table 2B) explained 15.4% of the observed variability in relative mass (once corrected for body mass) of the distal fermentation chamber mass.

Fig. 2
figure 2

Relationship between landscape structure (Forest index, log-transformed) and relative mass of the distal fermentation chamber (DFC, corrected for body mass) in roe deer from a mixed landscape in southern France

Full size image

Gas production in RR contents was on average 1.4 times higher (AICc of the model including the landscape term = 91.45 versus AICc of the constant model = 106.34, ΔAICc = 14.89) in deer harvested in more open landscapes (average ± SE = 23.23 ± 4.14 ml/200 mg, versus 16.49 ± 1.89 ml/200 mg for forest deer). However, three RR samples from the most open landscapes (Log (FI) < 1.4) contained not only green forage material but also a high proportion of maize. When the same statistical comparison was performed without these samples, a difference in gas production between open (17.12 ± 2.39 ml/200 mg) and forest landscapes still existed (AICc of the model including the landscape term = 21.78, versus AICc of the constant model = 27.46, ΔAICc = 5.68).

Discussion

As we originally predicted, deer from the more forested habitats of our study site had relatively more digestive tissue mass (mainly in the distal fermentation chamber) than their counterparts from the more open areas of this landscape. Indeed, the mass of the DFC was more variable than that of RR in our roe deer, with the lightest relative DFC mass observed in animals harvested in areas with a low proportion of woods, despite the fact that absolute body mass is higher in these same landscapes (Hewison et al. 2009). On the other hand, the fact that RR mainly varied between age classes, with yearlings having a lower relative RR mass for a given body mass than adults, suggests that digestive capacity increases with age. These results suggest that deer living in the agricultural matrix require less digestive capacity.

In our study site, the observed digestive plasticity could be explained by the access to a high-quality diet of the deer foraging outside the forest (Hewison et al. 2009). This finding is supported by the fact that we also found a higher gas production in deer from the more open landscapes. Deer that forage outside the forest certainly have more opportunities for incorporating highly nutritious food in their diet such as cultivated crops and herbaceous meadow species (for example, only rumens of deer harvested in the agricultural landscapes contained maize). In the same study area, Abbas et al. (2011) have shown that diet quality increases with landscape openness, with more cell content and less lignin and hemicellulose in diets from the more open landscapes, linked with the higher consumption of cultivated plants in winter in these areas.

These results suggest that roe deer inhabiting agro-ecosystems might benefit from a lower overall daily metabolic cost associated with rumination (i.e., reduction in gut tissue mass). However, when comparing the magnitude of the differences between the roe deer living in the contrasting landscape contexts within this study area to the between season differences measured with similar techniques in other roe deer populations (König et al. 1976; Holand 1992), it is clear that seasonal mass fluctuations in the digestive tract are much more pronounced than the landscape-related ones we observed here. For example, in highly seasonal environments, the roe deer gut mass (RR and DFC) may increase by 50% when the landscape is covered with the snow and food quality is reduced (Holand 1992). In other species (e.g., the Mongolian gazelle Procapra gutturosa), when high-quality forage becomes scarce, the weight of both the reticulorumen and large intestine increases by more than 27% (Jiang et al. 2002). Therefore, we suggest that digestive plasticity in species such as roe deer, which evolved as an adaptation to highly seasonal environments, also facilitates corresponding adaptation to varying landscape structure.

Landscape changes due to agricultural intensification generally reduce the roe deer’s native woodland habitat, but it has also increased the availability of highly energetic resources, notably in the form of cultivated crops. As a consequence, this easy-to-digest food allows roe deer to reduce the mass of their digestive tissue and ingesta load. For deer inhabiting predominantly agricultural habitats, this could lead to a reduction in metabolic costs associated with digestion, and consequently a reduction in daily energy expenditure allocated for rumination. The better body condition (Hewison et al. 2009) and immunocompetence (Navarro-Gonzalez et al. 2011) of roe deer living in the more open landscapes of our study area could therefore be partly explained by this energetic advantage. Whether the landscape-related differences in digestive tissue mass observed in this study could suffice to result in significant metabolic savings needs to be explored in future studies by measuring and modeling energy expenditure and intake.