All research mentioned in this paper was conducted in six rural and mountainous areas of the northwest of the Iberian Peninsula. In Spain: Campoo in the south of the Cantabria Autonomous Region ; Picos de Europa, a geographical region that straddles the autonomous administrative regions of Asturias, Cantabria and León province in the Autonomous Region of Castilla y León [, authors personal observations]; Piloña in central-eastern Asturias ; Caurel, in the south-east of Lugo province (Autonomous Region of Galicia) ; and Sanabria in the north-west of Zamora province (Autonomous Region of Castilla y León) . In north-eastern Portugal: Montesinho  adjoining Zamora province (Figure 1).
All six survey sites are culturally and biologically rich and most lie in protected areas, e.g. Picos de Europa National Park, Sanabria Lake Natural Park and Montesinho Natural Park. Bordering both Mediterranean and Eurosiberian floristic regions, the six sites have climates that vary from oceanic (wet and relatively mild) in Picos de Europa, Piloña, Caurel and the north of Campoo to wet-Mediterranean (drier in summer) in Sanabria, Montesinho and most of Campoo.
Landscapes include a mosaic of meadows, forests, rivers and high mountain vegetation growing on varied geological materials and soils. The predominant vegetation consists of beech (Fagus sylvatica L.) forest, several oak species, e.g. Quercus robur L., Quercus petraea (Matt.) Liebl., Quercus pyrenaica Willd.,Quercus faginea Lam. and Quercus ilex L., chestnut (Castanea sativa Mill.), broom scrubland consisting of Cytisus scoparius (L.) Link, Cytisus multiflorus (L'Hér.) Sweet, Genista florida L., and heath comprising Erica cinerea L., Erica vagans L., Erica australis L., Erica umbellata Loefl. ex L., Calluna vulgaris (L.) Hull. Fagus sylvatica, Quercus robur and Quercus petraea are more common in the wetter northern areas, while Quercus pyrenaica, Quercus faginea and Quercus ilex grow in drier areas.
Until a few decades ago the survey sites' economies were based on agriculture, cattle rearing and several less important activities. Most of the population engaged in traditional stock farming involving few animals. Short-distance vertical transhumance and long-distance southwards transhumance of cattle and sheep were particularly important in the Cantabrian Mountains. In regions such as Campoo, low salaries meant that even people working in the steel, cement and glass industries combined wage labour with livestock farming. The largely subsistence-based household economy was boosted with income from the sale of animals, eggs, butter and handicrafts. Other important economic activities were smuggling and forestry in Montesinho, and door-to-door hawking in Sanabria, chiefly using mules.
Many fields once used to grow cereals (for bread), pulses, turnip and potatoes now provide grazing for cattle. Agriculture plays only a minor role and new economic activities, such as rural tourism, are increasingly important.
Ethnobotanical data collection and analysis
Ethnobotanical information was obtained through informed consent semi-structured interviews with key informants over the last twenty years (1989–2004) (Table 1). Informants with a sound traditional knowledge of useful wild plants, mostly elderly long-time residents, were interviewed. Open questions about wild food consumption sought to ascertain knowledge about past and present-use, mode of consumption and preparation, collection time and collection sites for each species .
For this study, data were grouped into the following categories of edible plants based on folk perceptions: "vegetables", plants whose leaves, stems or even unripe fruits or seeds were consumed; "wild fruits", plants whose fruits or seeds were consumed when ripe; home-made "liqueurs" or other alcoholic drinks; "herbal teas", used in general as a digestif; plants used for "seasoning"; and finally, "flowers" and "underground organs", eaten for their sweetness.
Every plant species mentioned by an informant within one use-category was counted as one use-report (UR) [see ]. For instance, the raw fruits of Prunus spinosa in Picos de Europa were reported as consumed by 17 informants and used in liqueurs by 21, totalling 38 UR. However, a total number of 27 informants cited the species as useful since some informants reported use both for liqueurs and for raw consumption of fruits. We have rejected species with only one UR because such data are less reliable and sometimes dubious for drawing comparisons.
Voucher specimens were deposited at the herbaria of the Royal Botanical Garden of Madrid (MA, Real Jardín Botánico), the University of Oviedo (FCO, Universidad de Oviedo) and the School of Agricultural Engineering at Bragança (BRESA, Escola Superior Agrária).
To estimate the cultural significance of each species, we used the Cultural Importance Index (CI), whose definition and use are discussed in another paper [see ], with the following formula:
The index, which is based on previous indices [47,48] was obtained by adding the UR in every use-category (i, varying from only one use to the total number of uses, NU) mentioned for a species, divided by the number of informants in the survey (N).
The CI was calculated for each region. For example, Foeniculum vulgare in Montesinho was reported as used in liqueurs by 10 informants, for seasoning by 32 and for herbal teas by 23. The total number of survey participants was 90.
|CIFoeniculumvulgare = 10/90+32/90+23/90 = 0.11+0.36+0.26 = 0.722
This additive index takes into account the spread of use (number of informants) for each species and versatility, i.e. diversity of edible uses. The theoretical maximum value of the index is the total number of different edible use categories.
A mean Cultural Importance Index (mCI) of the species was used to assess wild food plant use in the Peninsular northwest as a whole. It is also useful in evaluating CI differences among the various sites. Since a null value may be due to either the species not growing in the area or growing but not being consumed, the mean value preferably needs to be calculated by considering only regions where the species grows and is available. For example, if the null values of the areas where it does not grow (Sanabria and Montesinho) are rejected, the mean value for Fagus sylvatica is 0.055; however, the figure decreases to 0.037 if all six areas are considered. Thus, the mean value takes into account species selection or rejection and availability.
To measure the cultural importance of families (CIf), we added the CI of the species from each family, following Galeano . We preferred using the sum instead of the mean as proposed by Phillips and Gentry  so as to highlight more diverse families which would otherwise be underestimated.
When comparing the edible floras of different regions, it is crucial to differentiate between plants growing in the area but not consumed and those which cannot be consumed because they are absent. To quantify this factor, a regional selection index (RSI) was created to assess differences in edible species selection or rejection among regions. It was obtained by dividing the number of species consumed at a site by the number of species growing there. For instance, the RSI for Sanabria is 0.37 (29/78), since 29 out of 78 available species are used. A regional index for each edible category can be further calculated to assess regional differences in selection among categories. For instance, the RSI for vegetables in Sanabria is 0.19 (5 out of 27 available) versus 0.4 (10/25) for fruits.