Ninety-four relevés were collected in two field campaigns between April and June 2017, in six areas of traditional agriculture (Tafza, Bellota, Tanakoub, Boujediane, Bou Ahmed, and Jnan Annich) (Figure 2; Table 1; Suppl. material 1). The fields with a mean size of 0.24 ± 0.04 ha were sampled by traversing in different directions until the number of plant species found no longer increases (“Tour de Champs” method; Le Bourgeois 1993). For each species, the abundance was estimated using the scale suggested by Braun-Blanquet (1952). Absolute frequency corresponds to the number of relevés in which the species was recorded, and the relative frequency corresponds to the proportion of relevés in which the species was recorded.

The plant species were identified using the relevant catalogues and Floras covering the study region (Sauvage and Vindt 1952, 1954; Fennane et al. 1999, 2007, 2014; Valdes et al. 2002a, 2002b; Fennane and Ibn Tattou 2005, 2008). Endemism and rarity assessment follow the catalogue by Fennane and Ibn Tattou (1998) which recognizes five categories for endemism (Morocco (E), Morocco and Algeria (A), Morocco and the Canary Islands (C), Morocco and the Iberian Peninsula (I), Morocco and Mauritania (M)) and three for rarity (very rare (RR), rare (R) and suspected rare (R?)). In the absence of an official red list based on the IUCN methodology and criteria, this coarse classification is widely used in Morocco.

Life forms based on the position of the perennating buds in relation to the soil surface during the unfavourable season (Raunkiær et al. 1934) were determined as phanerophytes (Phan), chamaephytes (Ch), hemicryptophytes (Hem), geophytes (G), and therophytes (Th) on the basis of our field observations and the descriptions in the consulted Floras. Taxa with two life forms were considered in both.

The chorology of the species relies on regional Floras and catalogues (Valdes et al. 2002a, 2002b; Fennane and Ibn Tattou 2005, 2008; Blanca et al. 2011). Thus, we recognized 36 different chorotypes (Suppl. material 2), which we grouped into five distributional categories: Regional (R) containing Maghrebian and Ibero-Maghrebian species; Extended (E), including the western Mediterranean and North Africa; Mediterranean broad-ranging (M), comprising the entire Mediterranean; Palearctic (P); and Cosmopolitan (C).

For the floristic status we adopted the reference by Fennane et al. (1999, 2007, 2014) to regard four groups of habitat preferences of species: i) Agricultural (AgH): cultivated or recently abandoned fields and fallow land; ii) Ruderal (R): otherwise disturbed or trampled habitats and roadsides; iii) Natural open (NHO): natural or semi-natural habitats with sparse vegetation such as rocks, scree, gravels, coastal sands, clay slopes, sandy places and cliffs, and iv) Natural closed (NHC): habitats with dense vegetation cover such as forests, scrubland, wetlands, forest clearings and grasslands.

Depending on the type of habitat, we recognized three floristic status categories: 1) agrestal species (Ag) occurring mainly in AgH, 2) apophyte species (Ap) typical for NHO and NHC and including species of former forest ecosystems, and 3) ruderal species (Rd) occurring chiefly in R and also in AgH.

Three indices were used to evaluate the diversity of segetal communities. Species richness (S), is the average number of species found in a study region. Shannon Index (1948) (H' = −∑^S_i=1 p_i log_e p_i, where S is the total number of species, and p_i is the proportional abundance of species i), expresses the diversity by considering the number of species and the abundance of individuals. The evenness index (J') of Pielou (1966) (J' = H'/log_e(S), where H' is the value of the Shannon Index, and S is the total number of species) measures the distribution of individuals within the sample, independently from species richness (Scherrer 1984). It is particularly important for comparing surveys with different species richness. The index values range from 0 to 1, and they are highest when the species have the same abundance, and lower when one or a few dominate the sample. The higher J', the more balanced is the distribution of species within the sample. The calculations were done with R software (R Core Team 2020) using the vegan package (Oksanen et al. 2020). The Jaccard index (J), calculated using PAST software (Hammer et al. 2001), quantified the variation in species composition between the areas (β-diversity). This index varies between 0 and 1, considering only positive associations. If the index increases, a large number of species occurred in both areas, suggesting that species turnover is low. If the index decreases, only a small number of species are present in both areas indicating a high turnover of species (De Bello et al. 2007). The correlation between geographical distances and the Jaccard index (J) was verified by the Mantel test (1967).

The floristic abundance data matrix consisting of 94 relevés and 209 species was classified by modified TWINSPAN (Roleček et al. 2009) using JUICE software, version 7.1.24 (Tichý 2002). Dissimilarity was calculated using the default settings (five pseudospecies cut levels, 0%, 2%, 5%, 10%, 20%; minimum group size = 5) and total inertia. Constant species were identified using a relative frequency threshold of 30%. For the names and ranks of syntaxa, we followed the EuroVegChecklist (Mucina et al. 2016).

In the sample of 94 vegetation relevés (Table 1, Figure 1), we found 209 vascular plant species belonging to 43 families and 150 genera (Suppl. material 3). Five families were represented by more than ten species and together account for 55% of the diversity: Asteraceae (21.1%), Poaceae (12.9%), Fabaceae (9.1%), Caryophyllaceae (7.2%) and Brassicaceae (4.8%). The remaining 15 families account for 45% of the diversity with two to nine species each (Suppl. material 4). The most diverse genus is Rumex with six species.

Species richness (S) showed high variability within and between the areas (Table 2). The mean number of species per relevé was 22 species with a coefficient of variation of 30.7%. The number of species per survey varied between a minimum of eleven and a maximum of 43 species. With a mean of 30 species, the Bou Ahmed area had a much higher species richness than the other five areas. In contrast, Bellota had the lowest richness, with only 18 species. The Shannon diversity index (H’) values varied similarly (Table 2; Suppl. material 5). Bou Ahmed had the highest value of H’ (3.39). For the other areas, the values of H’ ranged from 3 to 2.8. This index had a lower variability than the specific richness with a coefficient of variation ranging from 4.3% to 11.6%. Values greater than or equal to three and less than four indicate a good level of diversity (Molvær et al. 1997), as is the case for the areas of Bou Ahmed, Tanakoub and Tafza. Values below three and equal or higher than two indicate a moderate level of diversity, as is the case for Jnan Annich, Boujediane and Bellota. The values of Pielou’s evenness index (J’) were relatively low and varied between 0.09 and 0.15, showing strong differences in species abundance within the relevés with a preponderance of rare species and some dominant species (Table 2). This trend is relatively more pronounced in Bou Ahmed with E = 0.09, while the coefficients of variation vary from 12.4% for Tafza to 30.6% for Jnan Annich.

Jaccard’s similarity index (J) shows a relative heterogeneity in the floristic composition of the regions (Figure 3). The similarity values vary from a minimum of 0.509 between Bou Ahmed and Jnan Annich to a maximum of 0.935 for the Bellota - Bou Ahmed pair. Overall, we found that the similarity is rather high on mean, except for the region of Jnan Annich which is showing the lowest similarity values with other regions. No positive correlation was found between similarity index and geographical distance (r = 0.3099, p-value = 0.1462).

Therophytes are the dominant life form with 157 species representing three-fourth (75.1%) of the encountered species (Figure 4).

Figure 1. 

Location of study areas in the Tingitane Peninsula of the Rif Mountains (Morocco). BA: Bou Ahmed; BE: Bellota; BJ: Boujediane; JN: Jnan Annich; TA: Tanakoub; TZ: Tafza. Map basis: land cover from Copernicus Global Land Service (Buchhorn et al. 2020).

Figure 2. 

Traditional agroecosystem landscapes of the Rif Mountains (Morocco): (A) traditional cereal crops with almond trees inside the fields at Bou Ahmed; (B) cereal crops mixed with traditional olive orchads at Bellota; (C) a mixture of soft wheat and traditional olive groves in hillside fields at Boujediane; (D) mosaic of cereal crops and natural ecosystems (Cork oak) on erosion-prone slopes at Jnan Annich; (E) traditional agroecosystems corresponding to a bocage landscape at Tafza; (F) traditional cereal fields in agroforestry systems of mountains at Tanakoub.

Figure 3. 

Diagram of Jaccard similarity index values between the studied areas in traditional agroecosystems in the Rif Mountains (Morocco).

Figure 4. 

Life form spectrum according to Fennane et al. (1999, 2007, 2014) of the inventoried segetal plants of traditional agroecosystems in the Rif Mountains (Morocco).

Species with a Mediterranean broad-ranging distribution account for 44.1% of the total, more than half of which (54.8%) being strictly Mediterranean (Figure 5; Suppl. material 2). Cosmopolitan species ranked second in terms of occurrence, accounting for 20.2% of the total. Species of regional distribution (17) accounted for 8.1% of the total. The remaining species have either a Palearctic (19.6%) or an extended (7.7%) distribution in the Mediterranean.

Several endemic and/or rare species were found in our relevés. Overall, we recorded 33 regional to extended endemics representing 15.8% of the encountered flora (Figure 5; Suppl. material 2). For the Moroccan native flora, we noted two species considered very rare according to Fennane and Ibn Tatou (1998) (Brassica barrelieri and Echinops spinosissimus), two rare species (Filago ramosissima and Herniaria hirsuta) and two species suspected rare (Alyssum simplex and Calendula stellata).

Agrestal species comprised 46% of the encountered flora, followed by 31% apophytes, and 23% ruderal species (Figure 6). Habitat specialists occurring chiefly in a specific group of habitats accounted for 40.6%, of which 8.6% are in AgH, 8.1% R, 3.8% NHC and 20.1% occurring in NHO. On the other hand, 59.4% of the species normally inhabited two or three different habitat type groups. For example, 45.5% of the arable plants (facultative species) may colonise other habitat types beside of agricultural habitats (Suppl. material 6).

Figure 5. 

Species chorotype distribution according to Blanca et al. (2011), Fennane and Ibn Tattou (2005, 2009) and Valdes et al. (2002) of traditional agroecosystems in the Rif Mountains (Morocco).

Figure 6. 

Number and percentage of species counted per segetal flora type of traditional agroecosystems in the Rif Mountains (Morocco).

The TWINSPAN classification of the 94 relevés separated four clusters indicating area-dependent variation in species composition (Figure 7). Clusters 1–3 included relevés of the Jnan Annich, Bou Ahmed, and Tanakoub areas, respectively. Cluster 4 comprised the remaining areas of Tafza, Bellota and Boujediane. The areas differ in landscape, climate, soil and land-use characteristics (Table 1), and turned out to differ accordingly in species composition of the segetal vegetation (Table 3, full table in Suppl. material 7).

High constancy and dominance of wild grasses were found throughout all areas, with Anisantha madritensis and Lolium rigidum in clusters 1 and 2, and Avena sterilis in clusters 2, 3 and 4. The local dominance of Lolium temulentum in cluster 4, together with several other rare arable plants, indicated traditional practices in the hillside fields of the three areas represented here, such as using farm-owned seeds, cereal-legume cropping, and absence of weeding. The proportion of biennial and perennial herbs was considerable, especially in clusters 1 and 4, indicating crop-fallow rotations and manual shallow tilling by picks and hoes rather than deep ploughing. Species of slightly acidic soils were most common in clusters 2 and 3 (e.g., Brassica barrelieri, Raphanus raphanistrum, Rumex bucephalophorus, Silene gallica), indicating a component of sand to the generally heavy clayey soil texture. In contrast, in the fields of cluster 1 species of (more) neutral soils prevailed. It should be emphasized that non-native plants are rare in most of the examined fields, except in cluster 3 which contains a neophyte (Verbesina encelioides). It occurred chiefly in cannabis cropland developed by clearance of semi-natural woodland.

Figure 7. 

Dendrogram of the modified TWINSPAN classification. The numerical value on each division level represents the total inertia as a measure of cluster heterogeneity.

The total number of species found in our study area (209 vascular plant species in 94 fields) is similar or higher when compared to other studies conducted in the Mediterranean region on a comparable number of cereal fields and under approximately similar conditions: northeastern Spain, 175 species in 138 fields (Cirujeda et al. 2011); southern Spain, 80 species in 64 fields (Hidalgo et al. 1990); Greece, 103 species in 86 fields (Damanakis 1983); Turkey, 71 in 83 fields (Arslan 2018), indicating that the traditional agroecosystems of the Rif Mountains host a high level of phytodiversity. In contrast, surveys in traditional cereal fields of Crete, Portugal, Algeria and northwestern Morocco established higher total richness than the aforementioned papers: Bergmeier (2006) recorded 344 species in 74 plots in the South Aegean island of Crete; Ramôa et al. (2015) revealed 264 species in 105 crop fields of Portugal; Kazi Tani et al. (2010) reported 274 species in 100 Algerian cereal fields; Lastic (1989) found 375 species in 407 plots in the Rharb plain, and Deil (1997) 226 species in 46 wheat fields in the wider region of Tanger-Tétouan. The high local-scale (field) diversity in traditional agriculture is due to the moderate land-use intensity, the absence of herbicides and the crop rotation practice that involves fallowing (Deil 1997).

The floristic spectrum of the traditional agroecosystems (Suppl. material 4) shows the dominance of five families (Asteraceae, Poaceae, Fabaceae, Caryophyllaceae and Brassicaceae), typically reflecting the overall Moroccan flora (Fennane and Ibn Tattou 2012) and the segetal flora in western and northern Morocco (Zidane et al. 2010). These families also had higher expression in studies conducted in the northern Mediterranean region (Guillerm and Maillet 1982; Saavedra et al. 1989; Hidalgo et al. 1990; Dessaint et al. 2001; Ramôa et al. 2015; Fanfarillo et al. 2020; Munoz et al. 2020). The dominance of the Mediterranean chorotype in the Rif Mountains is in line with the figures for the entire Moroccan flora, two thirds of which are Mediterranean species, as well as with those found by Guillerm and Maillet (1982) for cereal fields in the western Mediterranean, and by Fanfarillo et al. (2019a) for traditional agroecosystems in Italy. This coincidence may be interpreted as an indication of the conservative character of the flora of traditional agroecosystems in which neophytes are rather rare (Fanfarillo et al. 2019b).

The significant percentage of apophytes in the arable fields can be linked to shallow tillage in traditional agroecosystems, as well as to the fallow period. In fact, woody apophytes such as Quercus coccifera, Chamaerops humilis, Pistacia lentiscus and Cistus salviifolius, possibly relicts of former woodlands and shrublands, continued to occur in the fields unless cleared by intensified tillage. The heterogeneous floristic status observed in the traditional agroecosystems of the Rif Mountains reflects the small size of the fields as well as the diversely structured agricultural landscapes which comprise different open natural habitats.

The largely autochthonous flora can be explained by the interlocking, transitions and permeability of the fields and the surrounding semi-natural habitats, as well as by the use of farm-owned seeds from previous harvests.

The life-form spectrum is dominated by therophytes, which account for three-quarters (75.1%) of the segetal species (Figure 4). Most annual segetal species depend on regular replenishment of the seed bank (Benarab 2008). However, the therophytes rate is lower than in the segetal flora of western Morocco which may be explained by the low-intensity agriculture in combination with regular fallow periods in our study areas, as opposed to the high-intensity management in western Morocco (Taleb and Maillet 1994; Taleb et al. 1998; Zidane et al. 2010). Hemicryptophytes are well represented in our study, where they comprise 24.4% of the arable flora. They amount to 15% more than found by Taleb et al. (1998) in autumn cereals for the entire country, approximately three times higher than that found by Hidalgo et al. (1990) in dryland wheat crops of the Cordoba region (Spain) and twice as many as found by Kazi Tani et al. (2010) on the arable land of Northwest Algeria.

Perennial plants and especially hemicryptophytes become more important in arable fields as a result of shallow tillage combined with regular fallow periods (Deil 1997; Zanin et al. 1997; Légère and Samson 1999; Streit et al. 2002; Young and Thorne 2004; Murphy et al. 2006; Trichard et al. 2013; Bergmeier and Strid 2014). Chamaephytes and phanerophytes are also remarkably well represented (6.6%). This percentage is higher than that reported from cereal crops in other parts of the Maghreb (Guillerm and Maillet 1982; Taleb and Maillet 1994; Taleb et al. 1998; Kazi Tani et al. 2010; Zidane et al. 2010). This life-form spectrum can be explained by the specificity of the traditional agroecosystems of the western Rif Mountains, which form a mosaic in which cultivated (crops, young and old fallows, abandoned fields), natural (forests, matorral), and semi-natural habitats (hedges, paths, drains and ditches) coexist (Ater and Hmimsa 2008).

The vegetation of the arable fields in the study area consists mainly of plant species characteristic of two classes – Chenopodietea and Papaveretea rhoeadis, the former with strong ruderal ties, the latter of segetal character (Table 3). Our results confirm that cereal fields in the South Mediterranean accommodate a higher proportion of species of the Chenopodietea and a lower of the Papaveretea rhoeadis than further north (Nezadal 1989; Bergmeier 2005, 2006). The classification chiefly along territorial affiliation indicates that local-scale conditions, both abiotic and land-use related, are decisive factors in our study area when it comes to vegetation differentiation. The plant communities found by us should not be understood, therefore, as syntaxa of wider than area-related validity. Species indicating the inclusion of fallows in the rotational system of cereal or cereal-legume cropping (such as Helminthotheca echioides, Dittrichia viscosa) are prominent in many of the examined fields, quite like in other fields in North and Northwest Morocco (El Antri 1985; Lastic 1989; Deil 1997). Certain structural characteristics observed by Deil (1997) about three decades ago, which contributed much to the local-scale species richness, such as patches of the palm Chamaerops humilis spared from tilling in the fields, have been removed in most places nowadays.

The first classified cluster, sampled in fields chiefly of Jnan Annich, is somewhat heterogenous, probably due to the wide altitudinal range of almost 700 m. The nutrient-rich clayey soil and the more or less pronounced thermomediterranean conditions, together with the non-intensive tillage and rotation including fallow years supports species of the order Brometalia rubenti-tectorum and, though rather indistinctly, of its alliance Cerintho majoris-Fedion cornucopiae (represented by Fedia cornucopiae). The latter is an alliance known chiefly from the Cádiz province in southern Andalusia opposite the Strait of Gibraltar (Nezadal 1989).

The second and third clusters, of Bou Ahmed and Tanakoub, respectively, differ from the first one in several species indicating less nutrient-rich and moderately acidic sandy soil. Typical are Anacyclus radiatus, Chamaemelum fuscatum, Rumex bucephalophorus and Vicia benghalensis. Noteworthy is also the fairly high constancy of rather late-ripening species such as Ammi majus, Anacyclus radiatus and Malva parviflora (in Cluster 2), Andryala integrifolia and Chamaemelum fuscatum (in Cluster 3), and Emex spinosa and Glebionis segetum (in both), together with early-flowering plants of fields with sandy soil such as Rumex bucephalophorus and Galium verrucosum. Altogether, with its strong winter-annual grass component (Avena sterilis, Anisantha madritensis, Hordeum murinum) the vegetation of the Clusters 2 and 3 may well be assignable to the Brometalia rubenti-tectorum, but the species mentioned above, which vary in abundance between fields, make it reminiscent of Rumicion bucephalophori, Diplotaxion erucoidis and Chenopodion muralis as well, vegetation alliances with, in that order, decreasing segetal and increasing ruderal affinity (see Nezadal 1989; Rivas-Martinez et al. 2002).

In the fourth cluster, from fields of three areas under subhumid climate and on nutrient-rich clay-loam soil, species of the Papaveretea rhoeadis, which largely indicate mesomediterranean climatic conditions, are not very pronounced and the frequencies of species such as Convolvulus althaeoides, Emex spinosa, Glebionis coronaria and the annual wild oats and brome-grasses suggest thermomediterranean segetal Brometalia rubenti-tectorum vegetation, but as with the other clusters, alliance affiliation remains unclear and needs further comparative studies.

Traditional farming in the rural landscapes in the Rif Mountains – characterised by a combination of low-intensity land-use, the presence of semi-natural vegetation at the field- and landscape-scale, and variation in land cover and land-use – favours the preservation of agrobiodiversity. Cropping practices within traditional agroecosystems correspond to the concept of High Nature Value agriculture (HNV), as described in European low-intensity farming systems (Beaufoy 2008; Oppermann et al. 2012; Reicosky 2015) with low inputs beneficial for biodiversity enhancement (Bignal and McCracken 2000; Pienkowski 2011; Egan and Mortensen 2012). The conservation of HNV was included in the Pan-European Biological and Landscape Diversity Strategy (PEBLDS) initiated at the Kiev Conference in 2003 (UNECE 2003). The traditional agroecosystems of the Rif Mountains are similar to HNV in the low-intensity soil tillage, low agro-chemical inputs (particularly the absence of herbicide application), low machinery, and landscape-scale land cover diversity with the coexistence of crops, fallow land, semi-natural and natural vegetation. Therefore, in a next step, it is suggested to develop HNV arable species lists and HNV evaluation systems on Moroccan arable land.

However, northwestern Morocco is experiencing profound changes in its landscapes and land cover (Taiqui 1997; Benabid 2002; Chebli et al. 2018, 2021). Such changes in the mountain production systems have caused negative effects (forest and matorral clearing, excessive pumping of groundwater, erosion, the increasing use of synthetic fertilizers, etc.) detrimental to the ecological and socio-economic sustainability (Kradi 2012). Due to the small size of the fields and the poor and eroded soils, traditional cereal growing, arboriculture and animal husbandry are insufficient to ensure the subsistence of the smallholders, leading to exodus, land abandonment or the extension of cannabis cultivation (Lara et al. 2015). Thus, traditional agroecosystems based on polyculture and biodiversity-friendly agricultural practices are facing socio-economic challenges that threaten their sustainability. Hence, conservation measures to protect traditional agriculture and agroecosystems must at the same time promote the local farmers’ socio-economic and cultural status.