Our study was conducted on Port-Cros Island (6.40 km²), protected by National Park status since 1963 and located in the north-western Mediterranean Sea (43°00′N, 6°21′E). This hilly island has a maximum elevation of 196 m above sea level and is 15 km from the mainland. Climate is sub-humid, temperate Mediterranean with an average annual rainfall of 582.4 mm and an average annual temperature of 16.5 °C (Levant Island Meteorological Office, 1997–2007). The siliceous island is 80% covered by mixed forests of the sclerophyllous oaks Quercus ilex and Quercus suber and the Aleppo pine Pinus halepensis. Port-Cros Island is home to introduced vertebrates, especially cats and rats, and a Mediterranean endemic seabird of particular conservation interest, the Yelkouan shearwater (Puffinus yelkouan).

To test the relationship between the measurement of rat incisors and rat weight, we needed rats of known weight. 47 free-living ship rats were trapped at Port-Cros island during 4 consecutive nights on 4 lines of 30 “INRA” traps (BTT mécanique, France) placed at 10 m intervals [18]. All trapped individuals were killed, sexed, weighed, frozen and returned to the laboratory for incisor extraction. Each rat's incisors (2 upper and 2 lower incisors) were removed, washed and measured with a stereo microscope containing a micrometer slide with a measurement accuracy of 0.01 mm. We measured the antero-posterior diameter (AP) of upper and lower incisors at the tooth curve tangent (Fig. 1), the part of the tooth least likely to vary in diameter, avoiding incisor extremities which can be thinner or broken. We obtained five variables: APur (AP for upper incisor right), APul (AP for upper incisor left), APlr (AP for lower incisor right), APll (AP for lower incisor left), and Rat Weight (Rweight). All the variables were tested and did not have a standard distribution. Spearman rank tests were therefore performed to test for a possible correlation between Rat Weight and incisor measurements [19].

These correlations between rat weight and incisor measurements were then applied to rat incisors found during a recent feral cat diet study based on scat analysis [13]. Feral cat scats were collected on paths on Port-Cros Island every two months and analyzed in the laboratory by washing over a 0.5-mm sieve under a stream of hot water and separating all items [13], particularly incisors, to perform measurements. During the same year as the cat diet study, rats were trapped on the study site and weighed. A Mann–Whitney test was performed to test differences between mean weight of trapped rats and eaten rats. A Chi square-test was performed to test homogeneity between the six weight classes determined (1–50 g, 51–100 g, 101–150 g, 150–200 g, 201–250 g, 251–300 g) for trapped rats and eaten rats [19]. The two first and the two last weight classes were pooled, due to the small sample size.

Eyeballs were removed from the sample set of 47 killed rats used to determine the relationship between incisor size and rat weight. Both eyeballs of each rat were removed just after rat death and placed, in pairs, in a 10 percent buffered formalin solution [20,21]. After fixing in formalin for 2 months (previously determined in a sample test as the most suitable tinning time) lenses were removed, dried in an oven at 80 °C and regularly weighed to the nearest 0.01 mg [20], until weight variation ceased (after 28 hours, as determined previously by a sample test). Each lens of the pair was weighed. If each lens in a pair weighed nearly the same, the joint weight of the pair was used [17,21]. If the weights of the two lenses differed, the higher weight was used and doubled to obtain the weight of the pair (the lower weight being generally due to damage during extraction of the lens). The same Spearman rank tests as above were performed to test for possible correlation between Lens Weights (Lweight) and incisor measurements (APur,APul,APlr,APll) [19].

47 wild ship rats were trapped in the field: 28 females and 19 males, with weights ranging from 19 to 270 g. The correlations between rat weight and incisor measurements were all statistically significant (p<0.001) whatever the type and the position of incisor we considered (Table 1). We obtained the same level of statistical significance when correlations were tested separately for males and females (Table 1). The results of this significant correlation between rat weight and incisor measurements were used to determine the weight of rats eaten by cats, based on the measurements of rat incisors found in cat scat.

The equation used was:

The correlations between lens weight (Lweight) and incisor measurements (APur,APul,APlr,APll) were all statistically significant (p<0.001) whatever the type and the position of incisor (Table 1). We obtained the same level of statistical significance, no matter whether correlations were tested separately for males and for females or both together (Table 1).

This study demonstrates a strong correlation between rat weight and measurements of the antero-posterior diameter of upper and lower rat incisors. This simple measurement of rat incisors, one of the most frequently found and best-preserved rodent items in cat scats, can thus be used as an easy surrogate to obtain an accurate evaluation of biomass of each rat eaten by cats. Moreover, there can be no bias, since rodents are entirely ingested [22]. This approach offers a key to solving the problem posed by the wide range of biomass of rats [13], one of the most frequent preys for feral cats, especially on islands [23,24]. Obtaining the precise biomass for this primary and introduced prey opens the way to more precise calculation of the proportion represented by this prey in the overall biomass intake by the predator.

From the incisors found in cat scats, we can identify the predation patterns of feral cats as a large range of rats of different weights, from the smallest, just out of their shelters, to the largest adults. The fact that cat scats revealed a higher proportion of the smallest rats than was found among those trapped in the field may indicate that the predation pattern is linked more to “catchability” (the smallest, i.e. the youngest, being more naïve and easier to catch) than to availability [25]. However, a review of 66 published feral cat diet studies [26] indicates that only 10 express the feral cat diet in biomass of preys, and even then without taking into consideration the precise weight of preyed rats. Moreover, using the mean weight of rodents sampled by trap-lines for the biomass calculation is not only too rough, but may also induce a significant bias, since cats appear preferentially to eat smaller rats. Consequently, an accurate evaluation of prey biomass is essential and our study should contribute to this by providing a simple and easy-to-use tool for future diet studies.

The strong correlation found between rat incisor measurements and rat lens weights, a crude estimator of mammal age, showed that the age of each rat could potentially be derived from simply measuring of rat incisors. Although the correlation between age and eye lens weight has been demonstrated for the Rattus genus [20], no abacus of lens weight in relation to rat age is, unfortunately, yet available for ship rats. To determine the age of rats eaten by feral cats, ship rats should be bred from rats originating from the study site. With this protocol, the pattern of predation by cats on each age-class of rats (juveniles, sub-adults, adults) can be determined easily, enabling the impact of cats on the population dynamics of this introduced mesopredator to be better calculated. In particular, some surprise effects, like mesopredator release after cat eradication, could be evaluated with more accuracy [27,28]. This should improve understanding of the complex longstanding interactions between the dynamics of alien and native species [29,30].