Introduction

Islands in the Atlantic have been colonised during the late Holocene by African and European populations, resulting in various histories of adaptation to, and transformation and degradation of ecosystems there (Norder et al. 2020; Castilla-Beltrán et al. 2021; Raposeiro et al. 2021). The Canary Islands saw the arrival of humans ca. 2,000 years ago and were settled by people who developed their ways of life there based on crops and domesticated animals brought from the mainland of Africa, as well as using the resources of the islands, including endemic and native plants and animals (Morales et al. 2009; de Nascimento et al. 2020). There are many unanswered questions regarding how the indigenous inhabitants and then the European colonisers shaped the landscapes of these islands. Archaeology and palaeoecology can reveal particular relationships between cultural practices and adaptation to environmental change, as well as the human role in landscape degradation and extinctions of taxa.

The Canary Islands are situated off the northwest coast of Africa, Fuerteventura being just 96 km away from there (Fig. 1). This island group comprises seven islands, four islets and many rocks that vary in climate and height, from the dryer and lower eastern islands of Lanzarote and Fuerteventura to the higher and more ecologically diverse central and western islands of Gran Canaria, Tenerife, La Gomera, La Palma and El Hierro. These islands display an outstanding habitat diversity, with an array of ecosystems that include coastal desert scrub, Mediterranean thermophilous woodlands, subtropical cloud forests, high altitude coniferous forests and scrub above the tree line (Fernández-Palacios et al. 2021). The varied environments of the Canary Islands together with their isolation for more than 20 million years has created a considerable diversity of flora and fauna, with very many taxa found nowhere else, including over 650 endemic vascular plant species (Arechavaleta et al. 2010).

Fig. 1
figure 1

Inset, location of the Canary Islands. Main figure, Tenerife, current vegetation map including the windward zonal vegetation and its schematic altitudinal distribution (del Arco 2006)

Full size image

Upon the first colonization of the Canary Islands, humans introduced a series of domestic plants and animals, elements of a transported landscape directly brought from northwest Africa. Ethnohistorical and archaeological sources have confirmed the introduction of Capra aegagrus hircus (feral goats), Ovis orientalis aries (sheep), Sus scrofa domestica (pigs) and Canis familiaris (dogs), as well as cereals such as Hordeum vulgare (barley) and Triticum durum (hard wheat), legumes such as Lens culinaris (lentil), Vicia faba (faba bean) and Pisum sativum (pea), together with fruit trees such as Ficus carica (fig) (Machado Yanes 2007; Morales et al. 2009). While the herding of livestock has been proven to be a major subsistence activity among the indigenous populations, evidence of agriculture has also been found on all the islands (Morales and Gil 2014b; Hagenblad and Morales 2020; Morales et al. 2023). Abundant resources for various purposes were also provided by native taxa. Current knowledge on the use of plant resources by indigenous people is mainly limited to seed and charcoal analyses (Morales et al. 2021; Vidal-Matutano et al. 2021b). In comparison, there has been little study of the micro-botanical remains such as pollen, phytoliths and starch grains from archaeological contexts.

Palaeoecological research on the Canary Islands has followed two main lines of investigation, 1, the effect of climate changes on the vegetation, causing ecosystems to move to a different altitude or changing the composition of taxa over time; and 2, the effects of the arrival of humans on the vegetation. This has been studied from proxy evidence such as fossil pollen and macro- and microcharcoal particles stored in natural basins such as former lakes or small volcanic calderas (de Nascimento et al. 2009, 2016; Nogué et al. 2013; Ravazzi et al. 2021). These proxies, however, come with various biases. First, many plants are insect pollinated, and their pollen production and dispersal are low (de Nascimento et al. 2015; Nogué et al. 2022). Second, the persistent northeasterly trade winds blowing in these latitudes may carry pollen to the islands from ecosystems on the mainland of the Maghreb, Sahara or the Iberian Peninsula, so that the pollen detected on the Canary Islands may include some extra-regional components (Hooghiemstra et al. 2006). Phytoliths are not as easily transported by wind as pollen grains, and are thus less likely to travel long distances (Piperno 2006, p 21). Finally, pollen preservation conditions on the Canary Islands can be poor due to a lack of permanent water, oxidation and the coarseness of the sediments. In contrast, the volcanic soils show great potential for phytolith preservation (Delmelle et al. 2015), as they are generally acidic, varying from pH 5 to 7, only vertisols being slightly alkaline at around pH 8 (Fernández-Caldas et al. 1978).

Due to the large number of endemics in the native vascular flora of the Canary Islands, up to 50% (Fernández-Palacios et al. 2021), a specific modern phytolith reference collection is needed. Also, some taxa on the islands are affected by a ‘secondary island woodiness’ process, by which herbaceous taxa which arrive from the mainland in time become woody on islands (Lens et al. 2013; Zizka et al. 2022), potentially producing phytoliths that differ from their mainland relatives. Few studies have so far attempted to provide an understanding of phytoliths in plants and soils in Macaronesia. The first ones were archaeological (Afonso Vargas 2004, 2009, 2012) and another had an ecological and palaeoecological perspective, with selected morphotypes uploaded to the PhytCore online database, such as Brachypodium distachyon, Bromus rubens, Cenchrus ciliaris and Sonchus congestus (PhytCore 2022; Andréu Diez 2017). However, the preparation of a detailed quantitative and qualitative reference collection of phytoliths is fundamental for understanding their production by the plants of the islands and is still lacking. Here we build upon previous work to expand knowledge of phytolith production in a selection of plants relevant for the study of past socio-ecological interactions in the Canary Islands. We aim to provide a systematic quantification of morphotypes in 109 taxa of all major terrestrial ecosystems of this archipelago, including coastal scrub, thermophilous woodlands, laurel forest, pine forests on moister windward slopes or thinner pine woodlands on drier leeward slopes and mountain summit scrub at over 2,000 m (Fig. 1). This study can hopefully serve as a guide for future studies in Macaronesian archaeology and palaeoecology, because the Canary Islands share an important number of species and genera with other Macaronesian archipelagos, especially with Madeira. Our main research questions are 1, which species/genera/families are the main phytolith producers and 2, is there preliminary evidence for potentially diagnostic phytoliths and/or assemblages that can help identify particular plant species/genera/families in future archaeological and palaeoecological studies?

Material and methods

Collection and processing of plant material

Plant specimens of 109 taxa were collected on various field trips to the islands of Tenerife, Gran Canaria and La Palma between 2015 and 2021 (ESM 1 Table S1). As a first step in the creation of an extensive reference collection, we studied leaves and stems/branches of a single specimen per taxon, processing them together, and aiming to achieve a broad overview of the range of phytoliths produced by endemic and native plants. Some fruits were also processed, including those of Arbutus canariensis, Ficus carica and Juniperus turbinata ssp. canariensis (ESM 1 Table S2).

Processing of the samples was carried out at the ecology laboratory of the Instituto Universitario de Enfermedades Tropicales y Salud Pública de Canarias (IUETSPC, University of La Laguna). Plant samples were processed using the dry-ashing procedure (Parr et al. 2001; Piperno 2006). Samples were first cleaned with laboratory detergent and placed in an ultrasonic bath for 30 min, then dried overnight at 90 °C. The dry plant content was weighed on a high precision scale, and then ashed in a muffle furnace at 550 °C for 3 h, after which the ashes were weighed again. They were then treated with HCl (10%) and HNO3 (10%), and washed with distilled water. After drying the samples, we added 1 ml of water to each and extracted 50 μl of phytolith material after stirring. When the water evaporated, we mounted the material on to microscope slides using Canada balsam.

Analysis of samples and recording phytoliths

We counted a maximum of 200 phytoliths at 400 × and 500 × in a maximum of 10 transects per slide. Photographs of the most characteristic morphotypes were taken. We named the phytolith morphotypes according to the principles of the International Code of Phytolith Nomenclature 2.0 (International Committee for Phytolith Taxonomy - ICPT 2019).

We calculated the phytolith concentration per g of dry plant material from the numbers of phytoliths recorded in the slide transects. We calculated percentages and concentrations of phytoliths in RStudio. We classified the plant species into three categories based on their phytolith content, 1) high producers, taxa with over 10,000 phytoliths per g of dry plant material, 2) medium producers with values less than 10,000 but above 500 phytoliths per g and 3) low producers, with values less than 500 phytoliths per g.

Contamination in the form of volcanic glass and phytoliths of other taxa, such as Poaceae short cells, was found in some samples. Some examples of contaminants are Dracaena draco, Frankenia ericifolia, Juniperus turbinata ssp. canariensis and Pinus canariensis. This can be explained by a process of strong adherence of phytolith silica structures to porous bark and stems (Tsartsidou et al. 2007). Phytoliths thought to be contamination were left out of the final sum. According to Albert et al. (2003) such contamination can represent up to 30% of the total amount of phytoliths recovered from these taxa.

Statistical analyses

To assess the various species groups based on their phytolith assemblages we carried out cluster and ordination analyses using the package RVegan in Rstudio v. 1.1.456 (Oksanen et al. 2007). Cluster analysis using Ward’s method, Euclidean distance, is useful to break species groups into clusters based on the similarity of their phytolith assemblages (Mercader et al. 2009). We analysed only the species that yielded over 10 phytoliths in the total sum of our counts, and those with fewer were excluded.

Results and discussion

The resulting dataset (ESM 2) includes 123 phytolith morphotypes from 107 different plants, including trees, shrubs, herbs and ferns, and their quantification in estimated concentrations of phytoliths per g plant material and percentages (Fig. 2, ESM 1 Table S2). These phytoliths consist of 16 broad morphotypes (Table 1, Figs. 3–6, ESM 1 Table S2). The current state of research suggests that surface 3D or 2D shape, texture and ornamentation are the most important features for identification (Table 1, 2). These are currently distributed across coastal scrub, thermophilous woodland, laurel forests, pine forests or woodlands and summit scrub ecosystems, and some are also typical of plants of abandoned and cultivated fields. In the following subsections, we will first focus on the diagnostic potential of selected phytoliths (Table 2), their likely anatomical origin and their significance for archaeological and palaeoenvironmental research in the Canary Islands. Then we discuss production and preservation biases and assess taxonomic grou**s based on statistical analyses.

Fig. 2
figure 2

Phytolith concentration estimates (phytoliths per g of dry plant material) of the main phytolith producers in the reference collection

Full size image
Table 1 The main morphotypes with diagnostic potential in the phytolith reference collection of Canary Islands plants
Full size table
Fig. 3
figure 3

a, Acicular bulbous (ACI_BUL, Sideritis soluta); b, Acicular nodulate (ACI_NOD_1, Echium webbii); c, Acicular nodulate (ACI_NOD_1, Ficus carica); d, Acicular psilate (ACI_PSI_1, Forsskaolea angustifolia); e, Acicular segmented (ACI_SEG_1, Micromeria lachnophylla); f, Acicular psilate (ACI_PSI_2; Ficus carica); g, Acicular segmented (ACI_SEG_2, Nepeta teydea); h, Acicular segmented (ACI_SEG_3, Sideritis soluta); i, Acicular striate (ACI_STR, Rubia fruticosa); j, Acicular uncinate (ACI_UNC, Forsskaolea angustifolia); k, l, Acute bulbosus (ACU_BUL, Piptatherum coerulescens); m, Acute helical (ACU_HEL, Bromus tectorum); n, Bilobate (BIL, Hyparrhenia hirta); o, Bilobate (BIL, Aristida adscensionis); p, Brachiate (BRA, Descurainia borgeauana); q, Bulbous granulate (BUL_GRA, Pteridium aquilinum); r, Bulbous nodulate (BUL_NOD, Heberdenia excelsa); s, Cross (CRO, Hyparrhenia hirta); t, Ellipsoidal rugose (ELI_RUG, Bituminaria bituminosa); u, Articulated Ellipsoidal (ELI, Juniperus cedrus); scale bars = 50 μm

Full size image
Table 2 Descriptors and codes for the morphotypes with diagnostic potential in the Canary Islands phytolith reference collection
Full size table

Recorded phytoliths and their diagnostic potential

Coastal scrub

The coastal scrub (cardonal, tabaibal) of the Canary Islands is a desert-like shrubby community that is thought to have occupied half of the area of the islands before human colonization (del Arco et al. 2010). This vegetation type is exposed to a strong water stress and is dominated by succulent species of the genus Euphorbia. The coastal scrub was an important ecosystem due to its resources and closeness to the coast, which were vital for island societies. This was probably the first ecosystem encountered and used by human settlers; for instance, some of the earliest radiocarbon dates in Tenerife come from the site of Las Estacas rock shelter (Galván Santos et al. 2000). As they are from leaves and stems, phytoliths of plants in this ecosystem could show plant use in archaeological sites, and complement pollen analysis for reconstructing changes in coastal scrub in response to human impact and climate.

We found diagnostic phytoliths in Rubia fruticosa, which has Hair base striate (HAI_BAS_STR_3) (1.8%), Acicular striate (ACI_STR) (54%) and Polygonal striate morphotypes (2%) (Figs. 3i, 4t, u). This species is distributed across all the coastal areas in the Canary Islands, as well as Madeira and the Selvagens islands off the coast of Morocco. Its use since pre-European times as a valuable fodder resource has been suggested (Pais Pais 1996). Another taxon, Neochamaelea pulverulenta had a high concentration of Tracheary pitted (TRA_PIT_1) (80%) (Fig. 6o), similar to those produced by Plocama pendula (66%) (Fig. 6q), probably silicified vascular tissues for transporting water (Piperno 2006). N. pulverulenta is particularly relevant to archaeological studies due to its use for funerary practices by indigenous societies (Atoche Peña and Ramírez Rodríguez 2008). The consumption of its ripe fruits has also been ethnographically noted in Gran Canaria (Morales and Gil 2014b). The genus Echium includes Echium auberianum, E. brevirame, E. webbii, E. wildpretii, among many others, which are part of a group of ca. 25 endemic shrubs distributed in all vegetation communities. In the coastal scrub of La Palma the endemic species Echium brevirame is very abundant and is often used as fodder for livestock and as firewood (del Arco 1993; Pais Pais 1996). This species, as also the other studied Echium species, also has many phytoliths including Acicular psilate (61%) and Acicular nodulate (ACI_NOD_1) (28%), Hair base (HAI_BAS_1) (3%) and Tabular bulbous (TAB_BUL) (6.4%) (Figs. 3b, 4n, 6j). Acicular morphotypes found in Echium and Rubia are good examples of hairs that attach to hair bases, and their origin is the outer epidermal tissue of the plant (Piperno 2006).

Fig. 4
figure 4

a, Elongate areolate (ELO_ARE, Pteridium aquilinum); b, Elongate dentate/dendritic (ELO_DEN, Avena barbata); c, d, Elongate bulbous (ELO_BUL_2, Pinus canariensis); e, Elongate bulbous Echium webbii; f, g, Elongate bulbous (ELO_BUL_1, Arbutus canariensis); h, i, Elongate clavate (ELO_CLA), Carex spp.; j, Elongate striate (ELO_STR_2, Ranunculus cortusifolius); k, Fusiform nodulate (FUS_NOD, Erysimum scoparium); l, m, Fusiform striate (FUS_STR, Erysimum scoparium); n, Hair base (HAI_BAS_1, Echium webbii); o, Hair base striate (HAI_BAS_STR_1, Cistus monspeliensis); p, Hair base granulate (HAI_BAS_GRA, Cistus monspeliensis); q, Hair base (HAI_BAS_2, Tuberaria guttata); r, s, Hair base striate (HAI_BAS_STR_2, Bituminaria bituminosa); t, u, Hair base striate (HAI_BAS_STR_3, Rubia fruticosa); scale bars = 50 μm

Full size image

Rumex lunaria (Polygonaceae) is extremely abundant in the coastal scrub, dominating recent lava flows, as well as disturbed areas on the coast and centre of the islands. It has been documented as having been used both for feeding livestock (Nogales et al. 1992) and for funerary purposes, mostly serving as a bed of litter for the deceased (Diego Cuscoy 1953; del Arco 1993). A variety of Elongate (32%) and Blocky (23%) forms, apparently without diagnostic potential, were found in this species. Their abundance in combination with other proxy evidence such as pollen could help identify their local presence. Finally, species of the Poaceae family, such as Aristida adscensionis and Hyparrhenia hirta, are commonly found in the coastal vegetation and are key resources for livestock grazing (del Arco 1993), and are likely to be some of the main phytolith producers. They produce Bilobates with convex edges (BIL_1) (20%) and Cross (CRO) (6%) (Fig. 3n, s) in H. hirta as well as Acute bulbosus (7%). Aristida adscensionis also had Acute psilate (37%) types, together with diagnostic long-shaft Bilobate (BIL_2) (24%) (Fig. 3o).

Thermophilous woodlands

Thermophilous woodlands (bosques termófilos) have some remarkable species yet they are some of the most heavily transformed ecosystems of the archipelago (Castilla-Beltrán et al. 2021). These woodlands are dominated by various trees, such as Juniperus turbinata ssp. canariensis (juniper), Olea cerasiformis (wild olive), Pistacia atlantica (African mastic), Phoenix canariensis (Canary palm) and Dracaena draco (Macaronesian dragon tree) (del Arco et al. 2006).

From these characteristic species, only Phoenix canariensis stands out as a high producer of diagnostic phytoliths. As with other palm species, its phytolith production is very great and dominated by Spheroid echinate (SPH_ECH) (85%) (Levin 2019; An and **s of elements (species and morphotypes) in the diagram, those that plot in the lower left quadrant, those in the lower right, and those that cluster in the upper centre of the diagram (Fig. 8). The ordination analysis is useful to visualize how some morphotypes can help differentiate between species groups, including Acicular psilate, Hair base, Tracheary, articulated Polygonal, Spheroids, Amoeboid and Elongate sinuate. Their explanatory power in the ordination is a product of their great abundance. While an analyst can get a first overview of which species are likely to be represented in a sample as high percentages of these common types, they are not diagnostic of any particular species. It is by the combination of these common types, together with rare diagnostic ones, that particular plants can be identified. We suggest that when trying to record rare yet diagnostic phytoliths, analysts should make quick scans of slides before and after reaching the standardized minimum count of 250 phytoliths, as is recommended for archaeological samples (Zurro et al. 2016). Overall, quantitative analyses (cluster and ordination) show that there is no clear-cut pattern in terms of how taxonomic or ecological relationships are reflected in phytolith assemblages. For instance, Apollonias barbujana and Picconia excelsa grow in a similar habitat, laurel forest, but are in different taxonomic groups (Lauraceae vs. Oleaceae). They are clustered together, but far away from the other laurel forest tree species (Laurus novocanariensis and Persea indica), which belong to the same Lauraceae family.

Fig. 8
figure 8

Principal component analysis plot showing the main associations of phytolith morphotypes (blue arrows, black lettering) and species (grey lettering, full species names in ESM 1 Table S1) in two axes (PC1, PC2). It includes species in the reference collection with a phytolith sum >10 in the total count

Full size image

While some species within the same genus often produce similar phytoliths, as do Juniperus cedrus ssp. cedrus and J. turbinata ssp. canariensis, which cluster together, as well as both Carlina salicifolia and C. xeranthemoides, others such as the various species of Euphorbia produce very different phytolith assemblages. The smallest distance analysed is between Euphorbia balsamifera and Neochamaelea pulverulenta, which belong to very distant families, but share the same habitat, indicating that their similar assemblages with articulated Polygonal and anticlinal are probably a functional response to the same environmental factors (Tsartsidou et al. 2007). Another clear example is that of Ovate facetate phytoliths in laurel forest trees, which have been found in Apollonias barbujana, Persea indica and Visnea mocanera, and are similar to Ovate facetate found in Erica arborea. Such a pattern confirms that while taxonomy is central, adaptations to the environment can also play a role in phytolith production. This has interesting implications, as phytolith assemblages could be successfully linked to vegetation zones, if they are systematically studied in soils from different ecosystems and climatic zones in the islands.

Conclusions and future directions

By assembling and studying the first reference collection of phytoliths for the Canary Islands, we have managed to categorise the various plant species depending on the types of phytoliths they produce and amounts, and also by noting types with diagnostic potential based on their general shape (either 3D or 2D), surface texture, ornamentation and wall thickness, among other features such as size and the distribution of surface ornamentation. From this we have presented here a first overview of the main phytoliths produced by taxa of the different vegetation zones in the Canary Islands. We have shown the complexity of interpreting phytoliths from there, as production and preservation biases need to be taken into account to interpret phytolith assemblages depending on their context.

Future directions in using this line of proxy evidence for archaeological and palaeoecological studies include the checking of phytoliths in various parts of plants with known cultural value, depending on archaeological research questions. A biogeographical perspective should also be taken into consideration. Many species in the Canary Islands show insular secondary woodiness, such as Echium and Sonchus; their phytoliths may be very different to those of their mainland relatives, a topic that should be explored in future studies.

The study of phytolith representation in modern and ancient soils of the islands would also help understand patterns of vegetation distribution of particular plant groups to reveal past ecological and climatic changes. Finally, particular case studies would help to understand issues of differential preservation and taphonomic processes. Combining phytolith studies with those of other macro- and microfossils, and using molecular analytical techniques, would reveal the full potential of phytolith studies in achieving a more complete understanding of past vegetation change and culture-environmental interactions in the Canary Islands and other islands in the Macaronesian region that share similar taxa.