Abstract
In recent decades the biometric analysis of shellfish resources (mainly marine mollusc shells) has been used by many researchers to support the hypothesis of their more intensive human use and to estimate the increase in human populations. This paper studies the plates of the crustacean Pollicipes pollicipes found in three Holocene archaeological levels at the Jaizkibel 3 shell-midden (Gipuzkoa, Basque Country, Northern Spain). First, by analysing a sample from each level, the Minimum Number of Individuals (MNI) was calculated from the Number of Remains (NR). Second, by using one of the plates (right scutum), the biometry of P. pollicipes in the three levels was reconstructed. The measurements are compared with a sample of P. pollicipes, collected a few kilometres away from the site of J3. The results indicate that the barnacles were not overexploited by the human groups.
Keywords
Introduction
Shellfish metrical data are used as a source of information about the exploitation of marine resources in the past (between locations and over time at particular archaeological sites) in different regions all over the world. Mollusc shells, mainly the shells of limpets, but also of other gastropods, are used in almost all of these investigations. For some researchers the stability in shellfish size and age classes is interpreted as evidence that human pressure was not too heavy, while their decrease is considered as clear signs of overexploitation (e.g. Erlandson et al., 2008; Mannino and Thomas, 2002; Steele and Klein, 2008).
Biometrical analysis of other marine resources that are also found in the prehistoric sites is much more scarce. This is the case for crustaceans such as sessile barnacles (Moos and Erlandson, 2010) and decapoda (crabs) (e.g. Gruet, 2002, 2010; Gruet and Laporte, 1995; Jerardino and Navarro, 2002; Losey et al., 2004).
The current research refers to Cirripedia pedunculata. The objective is to reconstruct the size of the stalked barnacle Pollicipes pollicipes from the remains documented at a Holocene archaeological site in northern Spain. The research is based on the plates found by archaeological methods in the different levels of the shell midden at Jaizkibel 3 (J3) (Gipuzkoa, Basque Country) (Figure 1). On one hand, from this evidence, we will determine the number of individuals from the number of remains. On the other hand, with the assistance of modern analogues and applying statistical tests, we shall attempt to assess the evolution in size of P. pollicipes in the sequence. Through all the information being analysed and taking into account the other faunal remains documented at J3, we will assess the role of P. pollicipes in the human diet in prehistoric times.

Sites in northern Spain where goose barnacles have been recorded in contexts clearly assigned to the Mesolithic, Neolithic, Chalcolithic and Bronze Age.
Biology of Pollicipes pollicipes
The stalked barnacle (also called gooseneck barnacle) Pollicipes pollicipes (Gmelin, 1790) is a marine pedunculate cirriped with a flexible stalk, which attaches to the rock by cementing glands. The rest of the body (upper part) is formed by the capitulum, a structure protected by five large calcareous plates (two terga, two scuta and the carina) that are separated from the scaly peduncle by a whorl of smaller plates (in sets of ten) (Figure 2a). It is a strictly littoral species and forms dense aggregates over the wave-exposed rocky shore of the low/middle tidal zone.

(a) Main parts of Pollicipes pollicipes: A: right side; B: anterior side; C: left side; D: posterior side; 1: right scutum; 2: right tergum; 3: rostre; 4: left tergum; 5: left scutum; 6: carina. (b) J3 (Basque Country, Northern Spain). Level B. Pollicipes pollicipes plates recovered with a mesh with screen sizes of 8 mm (left) and 2 mm (right).
P. pollicipes is nowadays found in the eastern Atlantic from southwest England through France, Spain, Portugal and West Africa to Dakar (Senegal, ≈15°N) (Barnes, 1996; Southward, 2008; Van Syoc et al., 2010), excluding the Cape Verde Archipelago (≈ 15°N to ≈ 17°N) inhabited by Pollicipes caboverdensis (Fernandes et al., 2010). The peduncle is the edible part of the animal, to be exact, the orange cylinder inside it, which is nowadays considered a delicacy. Today this species is heavily exploited in northern Spain and Portugal because of the great commercial demand and the high prices in the market (Bernard, 1988; Jacinto et al., 2010; Molares and Freire, 2003). This poses a risk to population sustainability (Barnes, 1996).
Pollicipes pollicipes remains in the archaeological sites
Calcareous plates are the only remains that have been recovered at archaeological sites in western Europe. This evidence indicates that stalked barnacles were collected and consumed as food in the past (e.g. Álvarez-Fernández, 2010; Álvarez-Fernández et al., 2010b; Dean, 2010; Dupont et al., 2008; Gutiérrez, 2011). In northern Spain these animals have been documented in shell middens (debris of human activity consisting mainly of mollusc shells) dated from the early and middle Holocene onwards. They are present at sites belonging to the Mesolithic (e.g. La Poza l’Egua), Neolithic and Chalcolithic (e.g. Los Gitanos), and Bronze Age (e.g. La Garma B) (Álvarez-Fernández, 2011) (Figure 1).
The lack of remains of these crustaceans at archaeological sites before the Mesolithic in northern Spain could be due to the absence of this animal on the shores of the region in a cold climate (Campo et al., 2010), and its colonisation of the area during the Holocene (Álvarez-Fernández et al., 2010b). This colonisation took place very rapidly, as the radiocarbon dates from archaeological sites indicate. In the province of Asturias they are already present c. 9500 cal. BP (La Poza l’Egua, level 1) and in the Basque Country c. 8500 cal. BP (level Amck at Kobeaga II and level D at J3).
Jaizkibel 3 (J3) archaeological site
Location and stratigraphy
This site, located in Hondarribia (Basque Country) (Figure 1) is a taffoni-type rock shelter in sandstone, formed by the large-scale alveolization of the rock. The location dominates the Basque coastal corridor, precisely at the passage to the Pyrenees, in a privileged location, on the last stretch of a small valley where a perennial active watercourse provides the local inhabitants with fresh water. Despite its proximity to the current intertidal line (about 200 m) and the dominant northwestern orientation of the seaward façade, J3 faces southwest, making it exceptionally inhabitable, and in fact, it was occupied approximately between 8500 and 3000 BP.
The rock shelter was discovered by a local prospector in 1985 and was excavated throughout 2001 (Iriarte et al., 2005, 2010). These authors opened an exploratory square metre, which was later extended southwards because of the find of a human burial. Both units were respectively referenced as K12 and K10. Each square was divided into nine sectors measuring 33 cm × 33 cm (in K10, only the six sectors to the north of the section, numbers 4 to 9, were excavated). The sediment was dug by natural levels, although relatively thick layers (<5 cm) were dug by thin arbitrary spits to maintain control.
Different moments in the occupation are recognised, most of them characterized by a massive presence of malacological and anthracological remains, as well as occasional appearances of fish and crustacean remains, lithic fragments, pottery, etc. The levels have been designated in alphabetical order, from the surface (Level A) to the bedrock (Level G):
Level A is the superficial level.
Levels B and C are the first two accumulations of anthropic remains. They are attributed to a post-Mesolithic phase and yielded over 50 prehistoric fragments of pottery, including some rims and a base of small recipients.
Level D is the third shell midden level. Inside it was located the burial of an adult male. Radiocarbon determinations of charcoal and a human bone indicated a Mesolithic chronology for this level. The anthropological study of the bones indicates that the estimated age of the man was between 20 and 40 years.
Level E consists only of sandy sediment without faunal remains.
Level F is the fourth level of shell midden. Two radiocarbon dates on shell and charcoal indicate a Mesolithic chronology.
Level G, with hardly any faunal remains, is similarly dated in the Mesolithic period.
Sediment from the different sectors and layers was sieved when dry, using meshes of 8, 4, 2, 1 and 0.5 mm. Subsequently, in the laboratory, samples of archaeological and sedimentary material from each sector were floated and simultaneously cleaned. After drying and separating the final residue in the sediment, the stalked barnacle remains were separated from the rest of the material thus obtained.
Shell midden composition
A sample of the archaeological material from J3 has been analy-sed (Álvarez-Fernández et al., 2010a, 2013). All the materials come from Square K12, sector 2. They were selected randomly. A total of 45,313 remains have been documented in the different layers of the rock shelter. Of these, 45,237 are marine remains. All the different species of marine fauna are still present today on the coast next to the site, in the intertidal zone.
Molluscs are the most abundant remains in levels B, C, D and F (>95%). The genus Patella predominates throughout the sequence, with percentages up to 94% of the molluscs. Other species recovered, equally living in the intertidal zone, are gastropods (Osilinus lineatus, Stramonita haemastoma, Gibbula sp., Melarhaphe neritoides) and the mussel Mytilus sp. However, their remains are merely testimonial. In turn, P. pollicipes plates have only been documented in Levels B, C and D and only represent percentages between 3.95% (Level B) and 0.31% (Level D) of the marine faunal remains discovered at the site (Figure 2b). Other species of crustaceans are documented in the four different levels. Most of the sessile barnacle plates (Chthmalus sp. and Balanus perforatus) were found isolated in the sediment. Plates of acorn barnacles are also found in association with limpets. These small crustaceans, lacking any dietary value, probably became detached from the external surface of the molluscs taken to the site as food. Finally, echinoderm spines and fragments of carapaces and fish bones are also present, but are scarce (Table 1).
Number of remains (NR) and percentages of the marine fauna in the four archaeological levels (B, C, D and F) at J3.
Other faunal evidence (terrestrial and river snails, bones and teeth of mammals, and the bone of a bird) is also scarce. Small bones and teeth belonging to the human burial were also documented. Some lithic remains complete the assemblage. Fragments of pottery are not present in the different levels of Sector 2.
Materials and methods
Archaeozoological methodology
Large and small plates of P. pollicipes were documented at the site through the use of a triple screen with mesh sizes of 8, 4 and 2 mm (the remains found in the 1 mm mesh were too fragmented to be used in our research) (Álvarez-Fernández et al., 2010b) (Figure 2b). 530 plates, complete and fragmented were documented in Sector 2: 329 in Level B, 134 in Level C and 67 in Level D.
In order to increase the size of the samples to allow biometric studies, plates were also separated from Sectors 3, 5 and 8 in Square K12, following the same methodology as used in Sector 2. The total number of plates documented in the four sectors reaches 2144. The four sectors represented 44.4% of the excavated area where all the levels exist (in K10 only one layer with a shell accumulation was represented, Level B).
Calculation of the MNI
In a stalked barnacle, terga and scuta plates are found on both sides of the capitulum, whereas this has only one carina plate, situated on the posterior face and one rostrum plate on the anterior face. The shapes of the terga and scuta mean that they can be positioned and lateralized. In this way, based on a given number of plates found by archaeological methods it is possible to calculate the MNI from the NR. The simplest way of calculating the MNI is by using the number of carina or rostra plates, but it can also be calculated from the right and left specimens of both terga and scuta, as their shapes are different on each side of the capitulum. In most of the cases, plates are not complete, so fragments of tergum, scutum and carina plates are taken into account to improve the accuracy of the MNI (compare Álvarez-Fernández et al., 2010a; Dupont et al., 2008; Gutiérrez, 2011). The formula used is:
The most abundant plate (tergum right, tergum left, scutum right, scutum left, carina and rostrum) indicates the MNI in each archaeological level.
Biometric reconstruction
To study potential changes in size, biometric data were collected from P. pollicipes in the different levels of J3, to the nearest 0.1 mm. Maximum length and width were measured for scuta plates, since they are the most numerous and least fragmented and because the length and width of this plate is well correlated. The samples come from Sectors 2, 3, 5 and 8 in Square K12. The data obtained, both quantitative and qualitative, have been analysed by means of different statistical methods.
The missing length and width measurements were re-constructed by means of regression models. The dependent variable in each model was the length or width of the scutum depending on which parameter has not been measured.
In order to verify whether the biometric relationships are maintained over time, the relationship between the length and width of the two scuta (right and left) was studied.
Biometric comparisons with recent samples
In order to compare the size of archaeological and modern P. pollicipes, two samples were collected in Jaizkibel Mountain (denominated J6-1 and J6-2) in October–November 2011. They come from a location on the coast about 2 km to the northeast of the archaeological site. The sampling was undertaken by scraping four 30 × 30 quadrats (Borja et al., 2006a; De la Hoz and García, 1993). In the laboratory, the barnacles were sorted and counted, including both adults and juveniles (with <10 mm capitulum length, according to Cruz, 1993).
The capitulum of each stalked barnacle was allowed to dry and the plates were separated. Maximum lengths and widths of scuta plates (left and right) of each stalked barnacle were measured. The data obtained, both quantitative and qualitative, have been used to generate descriptive statistics.
The changes in the biometric measurements between the scuta plates of archaeological specimens found in the different levels and the modern individuals have been compared, using either an ANOVA or the Kruskal-Wallis Test (in the case of non-norma-lity). Then, when appropriate, the Mann-Whitney U-test was applied with the Bonferroni correction to avoid the possible increase in Type 1 risk.
Results
Determination of the remains and calculation of the MNI
In the selected Sectors of J3 (2, 3, 5 and 8), a total of 2144 complete and fragmented P. pollicipes plates were classified (Level B: 1334; Level C: 468; Level D 342). All the remains (whole and fragmented plates) were quantified. Large plates (whole and fragmented) were identified without difficulty. Whole and fragmented small plates (rostrolateral, lateral, etc.) were also documented. However, their small size, and in some cases their fragmentation, meant it was impossible to identify them at plate level. Instead, they have all been included in the same group (Other plates).
In Level B, the MNI is 170 and was calculated with the carina, taking into account COMP + PMFP + LFP. In Level C, the MNI is 62 and in Level D it is 64. In both of these, the calculation was based on right scuta plates from COMP + MPFP + PFP + LFP (Table 2).
Number of remains (NR) and Minimum number of individuals (MNI) of goose barnacles in Levels B, C and D at J3 Rock shelter.
Notes: COMP: complete plate; PFP: proximal fragment of plate; PMFP: proximo-mesial fragment of plate; MFP: mesial fragment of plate; MDFP: meso-distal fragment of plate; DFP: distal fragment of plate; LFP: longitudinal fragment of plate; IFP: indeterminate fragment of plate.
Biometry of archaeological data
A large percentage of right and left scuta from the three levels at J3 are fragmented (Levels B and C: 28%; Level D: 26.5%) because of the effect of trampling. The length and width of these plates were reconstructed using regression models.
Two models were used to calculate the length of the scutum as a function of the width, depending on which scutum, left (L) or right (R) was being determined. These models were, respectively:
A further two models were used to calculate the width of the scutum as a function of the length, depending on which scutum, left (L) or right (R) had to be determined. These models were, respectively:
All the models display very high explanatory values (R2 > 0.95) and great predictive power, which means that the measures could be predicted with great accuracy.
The descriptive statistics of the biometric measurements (length and width) in the different archaeological levels are given in Table 3.
Descriptive statistics with the variables being studied in the three archaeological levels (B, C and D) at J3.
The total archaeological sample consists of 345 measurements of right scuta and 295 measurements of left scuta (including both measurements of whole specimens and lengths and widths of fragmented specimens calculated with the above regression mo-dels). As no significant difference was seen between the lengths of right and left scuta, the following analyses only used the information about the length of the right scutum plates, as this was the largest and most significant sample.
Table 3 shows how the highest median values for right scutum lengths were reached in Level B and the lowest in Level C.
The results of the Kruskal-Wallis test show, however, that no significant differences are seen between the right scutum lengths for the barnacles found in the three archaeological levels being studied here (p = 0.13). This test was chosen because the norma-lity hypothesis does not hold in level B.
Comparison between archaeological and modern samples
The biometric measurements of archaeological and modern samples of right scutum lengths are given in Table 4. The mean and median of the archaeological individuals are larger than in the modern sample.
Descriptive statistics of the variable (right scuta) in the three archaeological levels (B, C and D), and the modern sample.
In turn, Figure 3(a) shows that the percentage of scuta with lengths of between 6 and 8 mm is higher in the modern samples. In contrast, scuta with lengths larger than 16 mm are more abundant in the archaeological sample.

(a) Histogram with the percentage of right scutum measured at different interval lengths, both modern (above) and archaeological (below). (b) Box plots with the lengths of right scuta (in mm), comparing the individuals in the modern sample and in the three archaeological levels (B, C and D) at J3.
The analysis of the modern and archaeological samples (Le-vels B, C and D), in Tables 3 and 4, shows that the scutum lengths are smaller in modern individuals than in the three archaeological levels. The box plots comparing the descriptive results are given in Figure 3(b).
At an inferential level, it should be pointed out that the modern scutum plates also exhibit a non-normal distribution. Therefore, the Kruskal-Wallis test was used to compare the lengths in the four groups. The results show that the differences found at a sample level are also maintained at a population level (p<0.000). In order to determine whether all the groups differ significantly or only some of them, pairs were compared with the Mann-Whitney U test. To avoid an increase in Type 1 risk, the Bonferroni correction was applied. The analysis shows that the only significant differences are found between the sizes of the modern scuta and those in the three archaeological levels (p<0.000). However, as explained above, there are no significant differences between the sizes in the archaeological levels.
To determine whether or not the biometric relationships are maintained over time, the dispersion diagram in Figure 4 shows the relationship between length and width of right scuta (R2 = 0.969; r = 0.984; p < 0.001). As can be seen, modern barnacles exhibit smaller and less variable scuta. However, the biometric relationship remains stable over time.

Relationship between length and width of right and left scuta (in mm) in modern and archaeological samples from J3 (Levels B, C and D).
Discussion and conclusions
J3 is one of the few prehistoric sites in northern Spain with a shell-midden consisting of different levels (B, C, D and F). In the four archaeological levels, the most abundant remains are marine molluscs. Limpets and topshells (O. lineatus) make up >95.5% of these. Owing to this large volume with a marine origin at J3 (>45,000 remains, in Sector 2 alone), we can conclude that they formed a major part of the diet of the human groups occupying the site during different phases of the Holocene (Álvarez-Fernández et al., 2013). This conclusion is further supported by a human burial in Level D, where the skeletal isotopic composition revealed high δ13C values, which indicates that the man’s diet included the consumption of a considerable amount of marine protein (Arias, 2005–2006).
As well as molluscs, the most recent levels at J3 (Levels B, C, D) have yielded P. pollicipes plates, indicating this crustacean was gathered for food on the shore, together with the molluscs.
J3 is, to date, the only prehistoric site in southwest Europe with different archaeological levels (two of them Mesolithic and one post-Mesolithic) in which sufficient plates of P. pollicipes have been found to allow the calculation of the MNI and a biometric study.
The MNI was calculated with the most numerous plates, both whole and fragmented specimens (carina in Level B, right scutum in Level C and right scutum in Level D). The quantification of the fragmented larger plates (carina, tergum and scutum) increased the final MNI. Level B, which is attributed to a post-Mesolithic phase, yielded the largest number of individuals. In contrast, a much smaller number was found in Level D.
The statistical treatment of the data obtained by using the barnacles’ right scuta succeeded in determining their sizes in Levels B, C and D. This plate was chosen as it was the most abundant and best conserved. The fragmented plates were reconstructed by means of regression models.
The medians of the right scutum lengths in the archaeological levels were used to show that no significant differences exist between the three levels. This means that there is no evidence to support the hypothesis of more intensive human use of P. pollicipes at J3 during recent Prehistory.
However, the presence of plates from juvenile individuals in the three archaeological levels indicates that bunches with juvenile and adult stalked barnacles were gathered on the rocks and transported to the site. Here the smallest P. pollicipes were discarded.
The archaeological remains were then compared with a mo-dern biological sample collected in the intertidal zone near the archaeological site, in order to determine whether or not any biometric differences existed. It was seen that the right scuta of mo-dern barnacles are smaller than the archaeological specimens. Moreover, the largest barnacles are found in greater numbers in the archaeological samples (Álvarez-Fernández, 2010; Álvarez-Fernández et al., 2011).
Our research concludes that P. pollicipes was not intensively exploited in any of the three archaeological levels of J3, but it is today, when larger individuals are collected first and the younger have less time to mature before being gathered. The coastal zone of the Jaizkibel Mountains is presently heavily exploited by poachers. Owing to the overexploitation of barnacles in northern Spain and, for example, in the northern part of Portugal in modern times (Borja et al., 2006a; Cunha and Weber, 2001; Molares and Freire, 2003), P. pollicipes are now found only in wave-beaten zones with difficult access (Borja et al., 2006b).
Before now, biometrical research on marine resources in northern Spain was based only on mollusc shells. Some resear-chers propose that in this region, after the end of the Palaeolithic, marine resources in general, and molluscs in particular, were intensively exploited. Therefore, in reference to molluscs, the larger individuals tended to be collected first and the younger had less time to mature before being gathered. Consequently, both the average age of the natural population and the average size of the shells were pushed down (Gutiérrez, 2009; Ortea, 1986). If this hypothesis based on overexploitation were true, a statistically significant difference in species size would be expected. Thus, the shell sizes in more recent levels would be statistically smaller than in older layers.
The data given here for the plates (scuta) of the barnacles in Levels B, C and D at J3 indicate no significant size differences exist. In addition, the biometric study of limpets (diameters) in the four levels (B, C, D and F) at J3 reached the same conclusion (Álvarez-Fernández et al., 2013) and therefore it cannot said that these resources were overexploited at the site.
We may conclude, in comparison with the numbers of molluscs, which the limited presence of P. pollicipes plates in Levels B, C and D at J3 indicates, that these crustaceans were not very important in the diet of prehistoric groups. Their capture was probably opportunistic. P. pollicipes were picked while the human groups gathered molluscs in the intertidal zone, mainly during the low tide of the spring tide, when these animals of the low/middle tidal zone are accessible, probably by land. This, in fact, supports the idea that P. pollicipes stocks were not overexploited in the past.
A comparative study of our biometric results for Pollicipes pollicipes, using the working methodology proposed in this investigation, with other stalked barnacle collections from archaeological sites in the same region, from synchronic and diachronic points of view, and with other kinds of data for molluscs, other crustaceans, etc., will enable a better understanding of the possible human impact on coastal ecosystems in northern Spain in prehistoric times.
Footnotes
Acknowledgements
The authors are grateful to Yves Gruet for the sessile barnacle identification and to Miriam Cubas for through-provoking discussions and suggestions. We also wish to thank R-P Carriol (Muséum National d’Histoire Naturelle, Paris) and an anonymous reviewer for their careful reviews and valuable comments.
Funding
This research was undertaken in the context of the project: The exploitation of marine resources in Atlantic Europe during the late Pleistocene and the Holocene (HAR2011-29907-C03-03/HIST), funded by the Programa Nacional de Humanidades del Plan de I + D + i, Ministerio de Ciencia e Innovación (Spain). Sampling and analyses of current P. pollicipes were funded by the Department of Food, Fisheries and Environment of the Basque government.
