Diet analyses

We used a total of 336 faecal samples of five ruminant species for estimation of the needle content in their winter diet: red deer (Cervus elaphus), roe deer (Capreolus capreolus), fallow deer (Dama dama), white tailed deer (Odocoileus virginianus), chamois (Rupicapra rupicapra) and mouflon (Ovis aries). Material originated from three areas of the Czech Republic and was collected for previous studies focused on feeding ecology of herbivores (Homolka & Heroldová 1992, Homolka 1995). Food supply on all used localities contained sufficient quantity of spruce needles.

Fresh faeces were collected seasonally on all localities and then air dried. We used three seasons in accordance with predicted changes in botanical composition of herbivore diet (autumn: October–November; winter1: December–January; winter2: February–March). The proportion of individual food components in the ungulate diet was estimated by the microscopic analysis of the undigested items in faeces. Reliable relationship between diet (rumen content) and faeces composition was confirmed by Homolka & Heroldová (1992). From the faecal samples collected, one pellet was removed and used to prepare a microscopic slide. The representation of various food components was estimated on the basis of their relative coverage in the microscopic field. In evaluating the overall character of the diet, the components were pooled to form primary forage classes: grasses, bramble, leaves and twigs of deciduous trees, blueberry, needles, forbs, seeds and others.

Quality of main diet component

For comparison of the nutritional quality of spruce needles and other main diet component (> 5% in the diets) we calculated the metabolizable energy (ME), which we used as indicator of nutritional value. We picked samples of plants that the deer usually browse (leaves of bramble, terminal parts of shoots of deciduous trees and terminal parts of spruce twigs with needles) at foot hills in the Jeseníky Mountains in October 2006. The samples were dried in a ventilated drying chamber at 60°C to constant weight and analyzed for the content of Kjeldahl nitrogen, fat, fibre, acid detergent lignin, nitrogen free extract and ash with standard method (AOAC 1980). These values were used for calculation of ME by the procedure of Kamler & Homolka (2005).

NIRS estimating of the needle content

Sixty five faecal samples were used as standards for testing possibility of NIRS to estimate needle content in diet. Measurements were carried out using a Nicolet Antaris, a near infrared reflectance spectrophotometer, in the 9860–4100 cm^-1 wavelengths range, and 50 scans from each sample was collected. For analysis, one pellet with flat area abraded on sand paper was used to simplify process of measurement. The pellet was measured lying direct on the cell of integrating sphere of the spectrometer. The spectra produced by the spectrometer represent the total chemical and physical properties of a sample. Chemical information appears at a specific point of the spectrum. Physical properties of a sample, such as particle size, are eliminated by mathematical corrections. Data were analysed using software TQ Analyst. Partial least square regression (PLS) was chosen for the calibration equation quantifying the relationship between NIR absorption and reference values (Martens & Naes 1991). PLS is recognized as a very powerful tool for developing models from spectroscopic data (Schenk & Westerhaus 1991) and is preferred in the studies analysing biological materials (Gillon et al. 1999, Xiccato et al. 1999). For validation of our equation we used standard error of cross validation (SECV). SECV is used to estimate the error of prediction for unknown samples by simulating the prediction process and leaving part of a data set out, thus developing the calibration model on the rest of data matrix, and making predictions for samples left out. This process is repeated several times so that each sample is left out once.

Fig. 1.

Variability in the needle content in diet of five deer species in three season of the year in various areas (BH — Brdy Hills, BM — Beskydy Mountains, DH — Drahanská Highland).

Needles in herbivore diet

Consumption of needles by large herbivores was common in snowy winter. At all three localities needles formed a significant part of the herbivore winter diet. At the end of the vegetation season (autumn), needles appeared in the animal diet only sporadically (volume in the diets less than 1% on average). In winter, volume of needles in the ruminant diet increased to 40%. In general, concentrate selectors (roe and white-tailed deer) consumed more conifers than other species (mouflon, red and fallow deer) in average 55.9 ± 28% and 23.0 ± 18% resp. (Fig. 1).

The volume of needles in diet depends mostly on the overall structure of the food supply. If there is a food source of higher quality present, the animals prefer this source. It can be demonstrated on two following examples. In the locality BM, red deer and roe deer consumed relatively little amount of needles (27% resp. 24% of food volume) in the mountain area in the period of deep snow cover as they had enough beech branches broken off by icing to feed on (at steep slopes with low snow cover they reached grasses, ferns and blueberry). In the locality BH, the autumn food supply included bramble which formed a significant share of the white tailed deer diet (81%) (browser) as well as fallow deer (18%) (intermediate feeder). Snow that fell in the middle of the winter worsened the access of fallow deer to grasses and at the same time the supply of bramble lowered. The share of needles in the diets of both species increased significantly to 67% in white tailed deed and 24% in fallow deer. By the end of the winter, the reserves of bramble were depleted and almost the only food source of some nutritional value for white tailed deer were twigs of conifers that formed more than 80% of its diet (Fig. 1). When the snow melted, fallow deer could reach the grasses again and they regained the dominant position in its diet (85%). White tailed deer substituted bramble by needles during the winter (r = -0.957; p < 0.001; N = 32), fallow deer substituted grasses as well as bramble (r = -0.898; p < 0.001; N = 33, resp. r = -0.795; p < 0.001; N = 33).

Quality of needles and other main diet components

According to the amount of ME, needles were the lowest quality food among the nine studied components. Content of ME in needles was by half lower than in leaves of bramble and by third lower than in sprouts of broadleaves woody plants (Table 1). Low preference of spruce needles in the herbivore diet can be therefore explained by the nutritional quality. In contrast to that, the content of crude protein in the needles was comparable or higher than in other studied components (Table 1). The correlation between ME and crude protein in our diet component was low (r² = 0.307, p = 0.0127).

Table 1.

Content of crude protein (CP % of dry matter = N × 6.25) and metabolizable energy (ME MJ/Kg % of dry matter) in the most important winter diet components.

Fig. 2.

Reference and NIR predicted values of needle content in the samples used for calibration.

Detection of needles by NIRS

A good relationship between values derived by the standard microscopic analyses and the predicted NIRS values for the content of needles was discovered. The needle content of reference samples ranged from 2 to 96% (Fig. 2). The standard error of calibration (SEC) of 6.41 and coefficient of determination (R²) of 0.98 were both indicative of good calibration statistics (Fig. 2). The calibration was validated by standard error of cross validation which was 18.8% for six used factors. This accuracy is comparable to the estimated precision of our standard microscopic technique. These results demonstrate that NIRS is a reliable technique for quantifying the content of spruce needles in faecal samples and we can processed native samples of pellets without any fitting.

Spruce needles in herbivore diets

Needles were an important component of the winter diet of the studied species and its content well corresponded with quality of their diet. Spruce needles were not consumed during vegetation period and in winter they were presented in the diet as important food source after depletion of more attractive sources only. This low feeding attractiveness of needles is due to its low quality (Cederlund et al. 1980, Prieditis 1984). Consumption of needles also depends on the feeding type of a species (Hofmann 1989). The species that are adapted to utilization of grasses (red deer, mouflon) prefer grasses also in winter and needles form substantial part of their diet especially during snow cover. On the contrary, for the species accustomed to browsing on twigs consuming a minimal amount of grasses, needles formed up to 90% of their diet depending not only on snow cover but also on availability of other food resources (seeds, forbs, sprouts of broadleaved tree and shrubs).

Therefore, the presence of needles in diet is a good indicator of a food supply in the environment, with respect to feeding type of the ruminant, date of sample collection and snow cover height. High consumption of needles during a longer winter period shows impaired relations between the herbivores and the environment, disturbed either by changes of vegetation structure, or by high number of animals. The usual content of needles from localities with varied food supply was 30–45% of volume in winter. For example, Hodgman et al. (1996) tested the reliability of several faecal indicators for monitoring of the diet quality and intake in feeding trials with simulated diets. In three winter diets, the conifer browse formed only 0–20%. Low feeding preference of needles by herbivores correlates with quality of spruce that is lower than quality of all other main components of food supply. Low quality of needles was confirmed for example by Cederlund et al. (1980).

Diet quality

The quality of diet was evaluated according to the content of metabolizable energy. The energy content is considered as the main indicator of food supply quality for wild herbivores and also is considered as the main factor affecting food choice of herbivores (Berteaux et al. 1998). Many studies used nitrogen content in the diet or faeces as a simple and effective indicator of nutritional quality of ruminant diets (Latham et al. 1999). The applicability of nitrogen content is predominantly based on correlation between energy and nitrogen content in the diet rather then on lack of protein in the diet. The relationship between energy and nitrogen varies widely and therefore energy content reflects the value of food much more accurately than the content of nitrogen (Kamler & Homolka 2005). Our results support the importance of energy content. The content of nitrogen in our needle samples was higher than in several components of higher feeding attractiveness for herbivores and the correlation between the nitrogen content and ME was weak (r² = 0.307, p = 0.0127). The ratio CP/ME was 1:0.68 in spruce, 1:0.99 in blueberry and 1:1.33 in rowan. Therefore, the content of nitrogen itself has only a limited value as an indicator of food supply quality despite the observed correlation.

NIR estimation of needles

There is no doubt that NIRS is an efficient tool for this type of analyses and further applications of this method are feasible. For testing the accuracy of our calibration we used the PLS regression method that is preferred for studying biological materials (Gillon et al. 1999, Xiccato et al. 1999). Our study verified the applicability of NIRS for the estimation of the content of spruce needles in the winter diet of free-living ruminants from their faeces and we verified this procedure for estimating quality of environment or suitability of the animal numbers. We would also emphasise other advantages such as speed and low costs of the analysis with sufficient accuracy of the estimation. For example, about 30 samples can be processed in an hour with only easy pre-preparation. This means, that the number of samples that can be processed is not limited which saves everyday laboratory cost and thus solves the usual major problem with limited possibility of analysed samples. Our procedure has a potential for being used on a large scale for monitoring winter nutritional conditions of ungulates. Furthermore, validity of the estimated parameters is not influenced by sample freshness (Pearce et al. 1993, Leite & Stuth 1994).