Holocene vegetation dynamics and agricultural development at ancient Bubastis (Egypt): insights from phytolith analysis

1 Introduction

The emergence of agriculture during the Holocene was an important factor influencing the development of many early societies. However, this process was neither uniform nor linear. Instead, societies followed diverse and region-specific developmental trajectories in which agricultural practices interacted with - and did not simply determine - changes in settlement patterns, social complexity, and technological innovation. In some regions, the domestication of plants and animals, irrigation systems, and the expansion of trade networks developed alongside varying degrees of mobility and different sociopolitical organizations; in others, these innovations acted as key drivers of broader societal transformations. Such variability is well documented in early Holocene societies worldwide, including in Mesopotamia (Weiss and Zohary, 2011; Zeder, 2011), the Indus Valley (Fuller, 2006; Petrie and Bates, 2017), China (Liu and Chen, 2012; Zhao, 2011), and Mesoamerica (Piperno and Pearsall, 1998; Staller et al., 2006).

Egypt offers another striking example of this transformation. Changing climatic conditions and evolving landscapes in the Nile Delta during the Holocene—particularly the shift from a crevassing system to meandering Nile branches—led to a transition from mixed subsistence strategies to a predominantly agrarian economy based on cereal cultivation and pasture farming (Garbe et al., 2023; Pennington et al., 2017, 2020; Zhao et al., 2020, 2022). This process coincided with a delta-wide increase in settlements around 4000 B.C.E., likely driven by increased agricultural productivity and population (Garbe et al., 2023; Pennington et al., 2017; Stanley and Warne, 1993a). Bubastis, situated in the southeastern Nile Delta, evolved during the same period into a regional center (Lange-Athinodorou, 2024). While geoscientific studies have addressed various geoarchaeological topics (Garbe et al., 2024a, 2024b, 2025; Meister et al., 2021), insights into the paleobotanical context remain limited. A detailed reconstruction of vegetation history–such as through phytolith analyses—is essential for understanding the agricultural practices, adaptation strategies, and environmental management (Piperno, 2006) employed by the inhabitants of ancient Bubastis.

This study reconstructs the paleoecology of ancient Bubastis and its surrounding landscape, with particular emphasis on vegetation history and human-induced environmental change. Building on earlier lithostratigraphic and paleotopographic work by the authors (Garbe et al., 2024a, 2024b, 2025), it expands the perspective to include paleovegetation. Fieldwork conducted between 2022 and 2024 combined sediment data with phytolith-based paleobotanical analyses, a well-established approach in Egypt (Attia, 2018; Madella et al., 2014; Marinova et al., 2013) and globally (Leceta et al., 2024; Portillo et al., 2009; Rosen, 1992; Strömberg, 2004; Zuo et al., 2021). Phytolith assemblages from seventeen sediment cores are examined within the established stratigraphic framework to address three questions: (1) how natural vegetation evolved from the Late Pleistocene into the Holocene, (2) when and where cereal cultivation began near Bubastis, and (3) how settlement growth reshaped vegetation and land-use patterns. The results provide the first phytolith-based reconstruction of vegetation history at Bubastis and offer new insights into long-term human–environment dynamics in the Nile Delta.

2 Study site

2.1 Nile Delta: Holocene landscape dynamics and settlement activities

The Nile Delta is an alluvial plain that has evolved through a long series of delta-forming processes, influenced by natural factors such as tectonics, climate, and sea-level fluctuations (Butzer, 1976, 2002; Pennington et al., 2017; Said, 2012). Since the Quaternary period, the alluvial deposits of the Mit Ghamr Formation (Pre-Nile sediments) and the Bilqas Formation (Neo-Nile sediments) have defined the surface morphology of the Nile Delta. At the beginning of the Holocene, the Nile Delta was largely covered with Pleistocene deposits from the Mit Ghamr Formation, consisting predominantly of sands, transported by ancient Nile branches up to around 6000 cal. B.C.E. These fluvial sands contribute to local mounds known as the Geziracover Formation, commonly referred to as Geziras or “turtlebacks” (Pennington, 2017; Pennington et al., 2017).

During the mid-Holocene African Humid Period, from 6000 to 4000 cal. B.C.E., the Nile Delta experienced increased discharges and enhanced sediment rates. Known as the ‘Large-Scale Crevassing’ delta, the area was characterized by a complex network of small streams and floodplains including extensive swamps and marshes, leading to the silty or clayey and organic-rich Bilqas-2 Formation. From 4000 to 3500 cal. B.C.E., aridification and slowing sea-level rise led to reduced sediment input, promoting the development of meandering rivers and well-drained floodplains, resulting in deposition of less organic-rich attributed to the Bilqas-1 sediments (Goodfriend and Stanley, 1999; Pennington et al., 2017, 2020; Stanley and Warne, 1993b, 1998). Throughout the Holocene, shifting river branches reshaped the delta. Up to seven major branches existed historically, including the Canopic, Saitic, Tanitic, Pelusiac, Sebennitic, and Mendesian (Arbouille and Stanley, 1991; Bietak, 1975; Butzer, 2002). Today, only the Rosetta and Damietta branches remain (Figure 1A).

The transformation from a crevassing to a “Meandering” delta strongly influenced settlement patterns. The emergence of well-drained floodplains supported pastoral activity and agriculture, enabling the growth of early urban centers that relied on waterways for both local and long-distance transport and communication (Garbe et al., 2023; Pennington et al., 2017, 2020). From ∼2800 cal. B.C.E., urban centers emerged along major branches, notably at bifurcation points such as Mendes and Bubastis in the eastern delta (Bietak, 1975; El-Mahmoudi et al., 1998).

However, the shifting fluvial landscape of the Nile Delta frequently disrupted the settlement history of many cities, often resulting in their abandonment as nearby river channels changed course (Bietak, 1975; Butzer, 2002). A notable exception is the ancient city of Bubastis. Situated in the southeastern Nile Delta, Bubastis sustained continuous human habitation for over three millennia (Lange-Athinodorou, 2024; Lange-Athinodorou et al., 2019).

2.2 The ancient city of Bubastis (Tell Basta) and its paleoenvironmental history

Bubastis, located on the southeastern edge of Zagazig in the southeastern Nile Delta, was historically linked to the Pelusiac and possibly the Tanitic Nile branches (Figure 1A), although the exact courses of these waterways remain uncertain (Bietak, 1975; El Mahmoudi and Gabr, 2009; Ullmann et al., 2020). The city's strategic location established it as a vital hub for transregional trade, contributing to its foundation around 3250 B.C.E., and its continued significance until its decline under Roman rule around 200 C.E. (Lange-Athinodorou, 2024.; Meister et al., 2021).

The city gained prominence as the main cult center for the local goddess Bastet. The temple dedicated to Bastet became particularly important during the Sixth Dynasty around 2300 B.C.E. (Lange, 2006). In this period, Bubastis also served as a key administrative center with a residence of high-ranking provincial officials, as evidenced by the discovery of a provincial palace from the middle of the 4th Dynasty to the end of the 5th Dynasty (c. 2550–2350 B.C.E.), located on a mound west of the Temple of Bastet (Bakr and Lange, 2017; Lange-Athinodorou, 2024; Lange-Athinodorou and Es-Senussi, 2022; Figure 1B). This residence was then overbuilt by temples dedicated to the veneration of kings Teti and Pepi I of the Late Old Kingdom (beginning of the Sixth Dynasty, around 2300 B.C.E., Lange et al., 2006; Lange-Athinodorou and Es-Senussi, 2022). The city's role as a major administrative hub during the Middle Kingdom is further supported by the excavation of a large governor's palace from the 12th Dynasty and an associated cemetery dating to 1830 B.C.E., situated on a mound in the northern part of the ancient city (Bietak and Lange, 2014; Lange et al., 2016; Lange-Athinodorou, 2024; Figure 1B). Bubastis reached its peak during the 22nd Dynasty (c. 900–800 B.C.E.), accompanied by major architectural projects, especially the expansion and renovation of the Temple of Bastet (Kitchen, 1996; Lange, 2008; Lange-Athinodorou, 2019a, 2019b, 2024; Naville, 1891). This temple, located on the central elevated mound of the city, was surrounded by two sacred canals, as evidenced by recent geoarchaeological findings (Lange-Athinodorou et al., 2019; Meister et al., 2021).

For paleolandscape reconstructions, numerous geoscientific investigations have been conducted in the study area in recent years, comprising over one hundred boreholes. These revealed five key stratigraphic units distinguished by their depositional milieu (Garbe et al., 2024a, 2024b, 2025). The uppermost layer in all boreholes consists of an anthropogenic layer with mixed grain sizes, rich in ancient pottery (unit I, buff color in figures logs). In almost all cores, this is underlain by a fluvial layer (unit II, greenish color), mostly devoid of ceramics. These nutrient-rich loamy-clayey floodplain deposits are thought to have been used for agricultural purposes due to their favorable soil properties. Basal deposits of unit III (greyish color) consist of fluvial sandy sediments associated with the Mit Ghamr Formation or the Geziracover Formation. In a few boreholes, a further fluvial loamy-clayey layer occurs beneath unit I, interpreted as artificial canals (unit IV, bluish color). In some cores, the stratigraphy is interrupted by anthropogenically displaced deposits of unit VI (reddish color) with atypical characteristics, classified as sandy colluvial sediments (Garbe et al., 2025).

The investigations also revealed several Pleistocene “Gezira” mounds underlying key areas of the city (Garbe et al., 2024a, 2024b, 2025). These mounds, or Koms (Arabic for “settlement mound”), later served as foundations for major building structures, including the Governor's palace of the Middle Kingdom in the north, the Temple of Bastet on the central elevation, and earlier palaces together with the Temple of Pepi I on the western Kom (Garbe et al., 2024a, 2024b, 2025). Settlement in the eastern part of Bubastis probably began later, during the Late Period (c. 664–332 B.C.E.), and gradually expanded eastward, reaching its peak in the Ptolemaic Period (c. 332–30 B.C.E.). Preliminary assumptions suggest that area S1 was initially used for agriculture and displaced as settlement expanded. In contrast, area S2 remained sparsely inhabited and appears to have been used primarily for agriculture (Garbe et al., 2025). These findings represent the first comprehensive geoarchaeological results in the eastern part of the city, which for a long time was only referred on scattered surface finds and historical texts as the main residential area of the city (areas S1 and S2) and was also home to the Temple of Hermes and the Dromos—the main street of the city (Figure 1B; Lange et al., 2016; Lange-Athinodorou, 2019b; Wilkinson, 1843). Building on earlier geoscientific work on paleoenvironmental and settlement development, this phytolith-based paleobotanical analysis seeks to clarify the paleovegetation history of the study area, with particular emphasis on the agricultural evolution of the region.

3 Materials and methods

3.1 Coring and sediment analyses

The boreholes were examined following the methodology outlined by Garbe et al., 2024a, 2024b, 2025; detailed descriptions and photographs are provided in Supplementary Material SM1. In total, 17 boreholes were investigated within the study area. Boreholes C1—C5 are situated around the Temple of Pepi I area in the western part of the city. Boreholes C6—C14 are distributed throughout area S1 in eastern Bubastis, while C15—C16 are located around area S2 to the northeast and C17 further northeast, outside of the ancient settlement area (Figure 2). As part of earlier research on landscape reconstruction at the Temple of Pepi I (Garbe et al., 2024a) and paleotopographic studies in eastern Bubastis (Garbe et al., 2025), cores C1—C16 were previously published in a slightly modified configuration. For organizational clarity and to streamline the presentation of results, these boreholes have been renamed in the current study (see Supplementary Material SM1).

Due to transport restrictions on sample material from archaeological sites, sediment samples were analyzed macroscopically on-site, following the methodology outlined by Garbe et al., 2024a, 2024b, 2025. No additional laboratory analyses were conducted. Chronostratigraphic assessment was based on 42 well-preserved pottery fragments recovered from the cores, using approaches established in previous studies (Garbe et al., 2024a, 2024b, 2025; Lange-Athinodorou et al., 2019; Meister et al., 2021). Dating relied on typological comparison with other Nile Delta reference collections (e.g., Ballet et al., 2018; Lange-Athinodorou et al., 2025; Schiestl and Seiler, 2012), covering eight periods from the Old Kingdom (c. 2850–2180 B.C.E.) to Roman Period (c. 30 B.C.E.—200 C.E.).

3.2 Phytolith analyses

A total of 130 phytolith samples were recovered from 17 boreholes. The quantity and depth of samples in each borehole varied based on factors such as location and initial on-site interpretation results. In certain boreholes (e.g., cores C3, C9, and C11), samples were primarily collected from depths associated with the potential onset of human occupation to identify original vegetation patterns and human influence. Due to restrictive export regulations, the samples were analyzed on site using a rapid extraction method according to Katz et al., 2010 with simplified field equipment. Microscopic analysis was performed using a portable microscope from ioLight (Model 1) at 400× magnification. To establish a suitable basis for statistical evaluation, a minimum of approximately 200 identifiable phytoliths per sample were counted whenever possible (Albert et al., 1999; Meister et al., 2017; Piperno, 2006).

The identification of the morphological types of the phytoliths relied on subject-specific literature (Albert and Weiner, 2001; Ball et al., 1999; Leceta et al., 2024; Portillo et al., 2014, 2021; Rosen, 1992; Tsartsidou et al., 2007) and modern plant reference collections from the Mediterranean region (Attia, 2018; Bremond et al., 2004; Dal Corso et al., 2017; Delhon et al., 2003; Portillo and Albert, 2011), with adherence to the “International Code for Phytolith Nomenclature” where applicable [International Committee for Phytolith Taxonomy (ICPT), 2019]. Subsequently, the morphotypes were assigned to plant groups ‘Grasses’, ‘Cereals’, ’Shrubs and trees’, and ‘Palms’. Phytoliths that could be classified to established morphotypes but not assigned to a plant group based on the available reference collections, were recorded as ‘Unattributed’, acknowledging that future expansions of reference datasets or methodological advances may allow their identification. In contrast, phytoliths whose diagnostic features were obscured due to dissolution processes were categorized as ‘Weathered Morphotypes’.

Spheroid echinate morphotypes are assigned to palms in this study but can be confused with sponge-derived particles in fluvial settings. While freshwater sponge spicules are indeed common in lacustrine, wetland, and river sediments, the specimens identified here show none of the morphological irregularities characteristic of sponge spicules (Rasbold et al., 2023). Moreover, none of the sediments contain additional indicators of sponge remains.

Silica skeletons represent multicellular structures composed of several individual phytolith morphotypes and, as such, are treated separately from single cell phytoliths in some studies due to methodological considerations in quantitative analyses, as their multicellular nature may require adapted counting strategies (D’Agostini et al., 2022). Although this distinction is widely acknowledged in the literature, silica skeletons were quantified as single entities at the time of analysis and, due to the application of the Katz protocol and the resulting loss of both slides and processed material, they unfortunately cannot be re-counted. Consequently, silica skeletons remain included in the main count; however, it should be noted that this may introduce a degree of bias relative to strictly single-cell-based quantification, while still ensuring internal consistency within the dataset.

Additionally, phytolith concentrations were quantified per gram of sediment. A normalization to the commonly used Acid Insoluble Fraction (AIF) was not applied (Albert and Weiner, 2001), as the extraction followed a simplified protocol based on Katz et al. 2010, which did not allow for intermediate weighing of the residue after acid treatment. Consequently, the exact mass of the acid-insoluble fraction is unknown and cannot be reliably determined. Any attempt to estimate AIF values would therefore introduce substantial uncertainty. Therefore, the resulting dataset is only partially comparable to broader regional studies, as the methodological constraints necessarily limit its alignment with standardized extraction-based datasets. The quantification per gram of sediment used in this study, which, although potentially influenced by variations in sediment composition, provides only a consistent basis for internal comparison within the dataset (Meister et al., 2017).

Multivariate statistical approaches are widely applied in palaeobotanical research to structure complex assemblages and enhance the interpretative strength of the results (Hermans et al., 2025; Strömberg, 2007). In this study, a Correspondence Analysis (CA) was performed on the phytolith-based frequency data, as CA appropriately captures differences in the relative composition of samples and is frequently used in phytolith research (Jarl and Bruch, 2023; Olatoyan et al., 2023; Poliakova et al., 2025; Witteveen et al., 2024). The four plant groups, together with the categories unattributed and weathered morphotype, were included to assess (dis)similarity and distinguishability among phytolith assemblages across the sedimentological units. All calculations were carried out using the freely available software PAST version 5.3, which provides a robust implementation of CA including biplot visualization (Hammer et al., 2001).

4 Results of phytolith analyses

A total of 47 different morphotypes were identified from the 130 analyzed phytolith samples. Several significant or commonly observed morphotypes are highlighted in Figure 3. For a simplified overview, some morphotypes with similar characteristics and belonging to the same higher-rank group are combined into aggregated categories (e.g., Elongates includes all decorated Elongates except Elongate dentritic and Elongate Entires). An overview of all higher-rank groupings is provided in Table 2.1 in Supplementary Material SM2, while the full distribution of individual morphotypes across all samples is presented in Supplementary Material SM3. Based on reference collections, the phytolith morphotypes and the above-mentioned higher-rank groups were assigned to the plant groups ‘Grasses’, ‘Cereals’, ’Shrubs and trees’, and ‘Palms’ (Table 1) or categorized either as ‘Unattributed’ or as ‘Weathered morphotype’.

Table 1. Classification of the individual morphotypes and morphotype categories after ICPN 2.0 and ICPN 1.0 into the relevant plant groups and the related reference literature.

Due to the quantity of results, the following chapter presents the findings of the two most significant boreholes in detail, while the results of the remaining boreholes will be discussed later (see Supplementary Material SM2 and SM3 for illustrations and data of all cores). To facilitate understanding and correlation of the sediments with the phytolith results, the presented cores are preceded by a brief summary of the corresponding sediment data.

Coring C10 is located at the northeastern edge of area S1 and reaches a total length of 8 m (Figures 2, 4). At the top there is a ∼340 cm thick layer of anthropogenic debris (unit I) followed by ∼350 cm thick fluvial sediments of unit II. The bottom layer comprises the Gezira sediments of unit III, which are ∼110 cm thick. A total of five datable ceramic fragments, distributed throughout unit I and unit II at depths ranging from ∼60 cm to ∼560 cm, were recovered from core C10. The upper ceramic can be dated to the time from the Ptolemaic Period (c. 332–30 B.C.E.) to the Roman Period (c. 30 B.C.E.—200 C.E.), while the remaining ceramics can be attributed to the Ptolemaic Period (Garbe et al., 2025).

A total of 16 phytolith samples were analyzed in borehole C10 (Figure 4). Samples from 0.40–2.90 m belong to unit I (6 samples), 3.40–6.40 m to unit II (8 samples), and 6.90–7.20 m to unit III (2 samples). Phytolith concentrations vary significantly, from 2,307,035 (unit II) to 2,900 (unit III) per gram of sediment. Concentrations are moderately stable from 0.40–3.40 m, peak sharply at 3.70–3.90 m and drop abruptly. Counting 200 phytoliths was not possible at 6.40, 6.90, and 7.20 m.

The most frequently occurring plant group are grasses (Ø = 49.4%, σ = 14.5, n = 16), with Elongate entires (Ø = 30.9%, σ = 10.5, n = 16), Elongates (Ø = 10.8%, σ = 7.8, n = 15) and Rondel (Ø = 5.3%, σ = 2.5, n = 16) as the most common morphotypes for this group. Their distribution is highly variable, ranging from c. 32% to 57% in unit I, c. 29%–78% in unit II, and c. 21% to 38% in unit III. Peak values occur at depths of 3.70 m and 3.90 m in unit II, whereas the lowest value is recorded at 6.90 m in unit III. Shrubs and trees are also frequent (Ø = 24.6%, σ = 11.0, n = 16), especially the morphotypes Blockys (Ø = 14.3%, σ = 6.9, n = 16), Tabulars (Ø = 5.8%, σ = 6.0, n = 15) and Ovates (Ø = 4.4%, σ = 2.7, n = 16). The occurrence is relatively consistent in unit III (c. 38%–40%), but shows considerable variability in unit I (c. 17%–35%) and unit II (c. 6%–46%). The highest diversity of morphotypes is recorded at a depth of 6.40 m, whereas the lowest occurs at 3.90 m. Another although rather marginal plant group, is cereals (Ø = 6.2%, σ = 4.4, n = 14). Within this group, the morphotypes identified are Silica skeleton (Ø = 4.5%, σ = 3.8, n = 14) and Elongate dendritic (Ø = 1.7%, σ = 1.8, n = 12). The cereals occur almost completely above 5.40 m and are very consistently distributed in these samples (c. 5%–8%), except for samples 3.40 m–3.90 m (c. 12%–15%). Palms (Ø = 0.1%, σ = 0.2, n = 3) with the morphotype Spheroid echinate is a very low occurring plant group. These occur exclusively at depths of 3.70 m, 5.40 m and 6.40 m with values of <1%. The remaining phytoliths are divided among the unattributed group (Ø = 0.9%, σ = 0.9, n = 11), primarily featuring the Oblongs morphotype (Ø = 0.8%, σ = 0.8, n = 11) and the weathered group (Ø = 18.9%, σ = 9.3, n = 16). The Weathered morphotype exhibits large variability, ranging from c. 1% at a depth of 3.90 m (unit II) to c. 39% at a depth of 6.90 m (unit III).

Another core, revealing significant results is C3, situated at the center of the Temple of Pepi I area in the western part of Bubastis reaching a total depth of 7 m (Figures 2, 5). The uppermost ∼390 cm consists of anthropogenic debris (unit I), underlain by ∼60 cm of colluvial sediments (unit V). Underneath there is a ∼120 cm thick fluvial deposit (unit II), followed by ∼130 cm of Gezira sediments forming the basal layer (unit III). Four datable ceramic fragments were recovered from depths between ∼470 cm and ∼635 cm, distributed across units II and III. The upper ceramic within unit II can be assigned to the Early Old Kingdom (c. 2850–2570 B.C.E.), while the lower ceramic dates to the Old Kingdom (c. 2850–2180 B.C.E.). The upper ceramic within unit III can be dated to the time span from Early Old Kingdom to Old Kingdom, the lower ceramic can be assigned to the Old Kingdom (Garbe et al., 2024a).

Three phytolith samples were analyzed in borehole C3 (Figure 5). The top two samples at depths of 3.50 m and 3.70 m belong to unit I, the lowest sample at a depth of 4.70 m to unit II. Phytolith concentrations show clear differences between 924,203 (3.50 m) and 156,407 (4.70 m) phytoliths per 1 g sediment. More than 200 phytoliths could be counted in all samples.

The distribution of plant groups closely resembles that observed in the previously described core. The most dominant group is formed by grasses (Ø = 45.6%, σ = 2.0, n = 3), followed by shrubs and trees (Ø = 22.8%, σ = 3.3, n = 3). Cereals represent a smaller proportion (Ø = 4.5%, σ = 2.3, n = 3), identified through the morphotypes Silica skeleton (Ø = 3.4%, σ = 1.4, n = 3) and Elongate dendritic (Ø = 1.1%, σ = 1.2, n = 2). Particularly noteworthy are the cereal remains within unit II, which—together with the earliest dated ceramic fragments from the Early Old Kingdom—provide compelling evidence for the earliest agricultural activity in the study area. A substantial proportion of Weathered morphotypes was also recorded (Ø = 23.1%, σ = 3.7, n = 3), while unattributed types (Ø = 4.0%, σ = 4.2, n = 3) and palms, which were entirely absent, played only a marginal role.

The remaining phytolith results are summarized below. Figure 6 presents the mean values of the respective phytolith groups for all samples within each unit across all boreholes. Because not every borehole contains samples from all units, the number of data points per unit varies accordingly. Borehole C7 contains samples from unit IV, which has been omitted here due to insufficient statistical significance. The distributions shown in Figure 6 broadly mirror the patterns described above. With only a few outliers, the units display relatively consistent internal variability. In units I and II, grasses constitute the dominant group, mostly ranging between 40%–50%, followed by shrubs and trees, which mostly occur between 15%–30%. In unit III, these two groups are observed at nearly equal frequencies, generally between 20%–40%. Cereal phytoliths show a clear increase from the lower to the upper units. They are almost absent in unit III, occur slightly more frequently in unit II (though remaining below 10%), and reach their highest values in unit I (mostly 5%–10%). Palms are rare overall and appear only slightly more often in units II and III (up to ∼5%) than in unit I (∼1%). The unattributed group remains consistently low across all units (c. 0%–10%). Weathered morphotypes exhibit larger variability within units and reach their highest proportions in unit III (mostly 35%–55%), compared to units I and II, where they generally range between 25% and 40%.

To ensure the robustness of the results, a Correspondence Analysis (CA) was performed (see Figure 7 for results and table sheet ‘Statistics (CA)’ in SM3 for raw data output). The four plant groups, together with the categories unattributed and weathered morphotype, were included to evaluate the (dis)similarity and distinguishability of phytolith assemblages across the sedimentological units, thereby capturing the major compositional gradients within the dataset. The majority of samples plot within the 95 % confidence ellipse, indicating a largely homogeneous structure of the phytolith assemblages across most units. Only a small number of samples fall clearly outside this range, representing distinct outliers that deviate markedly from the main compositional pattern (e.g., the two data points at the bottom right from core 17, at depths of 8.70 m and 13.70 m). Most samples cluster near the center of the ordination space and in close proximity to the vectors of grasses, shrubs and trees, and weathered morphotypes, indicating broadly similar assemblage compositions and the absence of strong dominance by any single phytolith category. This central clustering reflects the fact that these groups vary in a comparable manner across the sedimentological units and therefore exert a similar influence on the main composition. This pattern does not apply to groups unattributed, cereals, and palms, as no samples plot in the immediate vicinity of these vectors. Unit I plots predominantly towards the grasses and cereals vectors, indicating stronger variation towards these groups, although its samples remain largely dominated by a combination of grasses and shrubs and trees. Samples from unit II occupies a more balanced position between these groups, reflecting a more even contribution and only a few samples showing a tendency towards cereals. Unit III is located centrally within the ordination space, with a slight shift towards the weathered morphotypes vector. Assemblages from unit IV, represented by only a few samples, display a similarly balanced composition. Overall, these patterns correspond well and support the trends observed in the univariate results.

5 Discussion

The following discussion addresses differences in phytolith concentrations, as well as the distribution of the individual plant groups within the sedimentological units, before relating these findings to the landscape reconstruction models proposed by Garbe et al. 2024a, 2025. In addition to the phytolith results from boreholes C3 and C10 presented above, the analysis also considers data from all boreholes (see Supplementary Material SM2 and SM3 for phytolith results of all cores).

5.1 Phytolith concentrations

Interpretations of phytolith assemblages in the literature commonly highlight the importance of considering both formation pathways and post-depositional processes. While many phytoliths are deposited in situ, they may also be relocated by fluvial or aeolian processes or by animal activity, particularly in arid environments (Fredlund and Tieszen, 1994; Neumann et al., 2009; Osterrieth et al., 2009; Piperno, 2006; Qader et al., 2023). In the Nile Delta, fluvial transport plays a significant role and could potentially influence fluvial sediments (unit II and unit III). However, studies from arid environments suggest that such transport has only a limited effect on locally preserved phytolith assemblages (Garnier, 2013; Garnier et al., 2013) and phytolith assemblages generally reflect the local vegetation (Boyd et al., 2022, Neumann et al., 2009; Ramsey and Rosen, 2016). Therefore, no taphonomic distinction is made between locally deposited and transported phytoliths in this study. Phytolith preservation is generally poor in most samples, as evidenced by the high abundance of weathered morphotypes and the low frequency of multicellular phytoliths. However, the reduced occurrence of multicellular phytoliths may also reflect taphonomic processes such as fragmentation and disarticulation of silica skeletons during transport, deposition, bioturbation and post-depositional alteration, rather than solely indicating poor preservation or dissolution (Alexandre et al., 1997; Cabanes et al., 2012; Cabanes and Shahack-Gross, 2015; Fraysse et al., 2009; Jenkins, 2009; Madella and Lancelotti, 2012; Shillito et al., 2011, Strömberg et al., 2018). Although high concentrations are typical near archaeological sites (Cabanes et al., 2012), values in the study area range widely—from 423 (C16, 7.70 m) to 2,307,035 (C10, 3.70 m) phytoliths per gram of sediment—but remain consistent with other phytolith studies (Albert and Henry 2004; Marinova et al. 2013).

Unlike natural soils, anthropogenically influenced profiles show more balanced phytolith concentrations due to mixing processes (Qader et al., 2023; Strömberg et al., 2018), expected in cores near the Temple of Pepi I and in areas S1 and S2. This pattern is evident in many boreholes, where the highest phytolith concentration is not reached in the uppermost sample (e.g., C5, C7 or C10). In contrast, cores C2–C4, C6, C13, and C16 exhibit more natural vertical trends.

Influence of current vegetation can be largely disregarded in C1–C14 due to sparse vegetation, but intense agricultural activity near C15–C17 may increase phytolith concentration in the upper layers, as seen in core C16. Although mixing current vegetation in the remaining boreholes probably represents a minor factor, an influence on the phytolith concentration, especially in the uppermost soil layers, cannot be completely excluded.

5.2 Morphotype assemblages

Among the 47 different morphotypes identified in the phytolith analysis, which were classified into four different plant groups and the categories of unattributed and weathered morphotypes, there are some considerable differences between the boreholes and the individual sedimentological units. In general, grasses are the dominant plant group across nearly all samples in the study area, followed by shrubs and trees. Whilst the two groups still occur in roughly equal proportions in Pleistocene unit III, the balance shifts over time, with grasses becoming increasingly dominant. Despite these fluctuations in their concentration, both groups have consistently shaped regional vegetation since the Late Pleistocene—Early Holocene transition (c. 9000 B.C.E.) as evidenced by the samples from Pleistocene unit III sediments. It should be noted that the Elongate entires morphotype is assigned inconsistently in the literature, having been attributed either to woody taxa (Tsartsidou et al., 2007) or to grasses (Albert and Weiner, 2001; Strömberg, 2004). This reflects a broader challenge in phytolith interpretation, as occurrences in woody taxa are considered more likely to result from grass contamination (Albert and Weiner, 2001). In this study, the morphotype is therefore classified as representing grasses. This might significantly alter the distribution between the two groups but does not change the overall conclusion that both were consistently dominant in the region. As noted above, these two groups occur most frequently across all units and thus dominate the vegetation of the study area over time. Accordingly, the following section about varying occurrence of phytoliths in the distinct units will concentrate primarily on the plant groups cereals and palms.

Cereal morphotypes occur sporadically in unit III. Their signature morphotypes, notably Elongate dendritic and partly Silica skeleton, can provide valuable information on the agricultural use and human impact of an area. While silica skeletons are not inherently diagnostic of cereals, since they represent preserved multicellular plant tissues that may originate from different taxa, their occurrence within suitable assemblages can support the identification of cereal remains, especially when characteristic features such as dendritic structures are observed (Ball et al., 1999; Bremond et al., 2004; Piperno, 2006; Portillo et al., 2014, 2021; Rosen, 1992). However, the presence of cereal-type morphotypes (up to c. 1.5%) in boreholes C4, C6, C10, and C13 is most likely the result of anthropogenic relocation processes. Given that neither wild nor domesticated cereals (i.e., wheat or barley) nor any evidence of human activity are documented for the region during the Late Pleistocene and Early Holocene, an agricultural origin of these morphotypes can be effectively ruled out. The occurrence and distribution of palms also vary—they are absent near the Temple of Pepi I, sporadically present in boreholes C6 (area S1) and C16 (area S2), and occur more frequently in borehole C17 (c. 13%) outside of Bubastis. Accordingly, during the transition from the Late Pleistocene to the Early Holocene, the vegetation was characterized only by scattered palms.

From the Early Holocene onwards, the region was periodically inundated by Nile floods from the nearby Pelusiac and/or Tanitic branches (Bietak, 1975; Ullmann et al., 2020). These floods submerged most of the terrain, gradually depositing unit II floodplain sediments. Rich in organic matter, these nutrient-rich sediments are well-suited for cultivation and could contain cereal-related phytoliths if used agriculturally after human occupation around 3250 B.C.E. (Garbe et al., 2025.; Pennington et al., 2017; Said, 2017). Even though the plant parts in which most diagnostic cereal phytoliths are produced—e.g., glumes, lemmas and paleas (Piperno, 2006; Rosen, 1992)—are typically removed during harvest and often transported away from the fields for further processing, the long-term cultivation of these fertile fluvial deposits over centuries would nonetheless facilitate the incorporation of plant-derived microremains into the soil matrix. The gradual accumulation of plant residues, root activity and traces of processing during growth can lead to an overall accumulation of plant remains (Tsartsidou et al., 2007; Weisskopf et al., 2014), which could contribute to the presence of cereal-related phytoliths. However, their overall abundance is expected to remain relatively low.

In the area around the Temple of Pepi I, cereal morphotypes were identified within unit II of cores C3–C5. Their very low abundance (<3%) suggests only limited, short-term agriculture, but also an external input of the cereals through relocation processes is possible. Given the interpretative challenges associated with cereal-related morphotypes in nutrient-rich fluvial unit II sediments mentioned above, as inflorescence parts that produce most diagnostic cereal phytoliths are typically removed during harvest and processed elsewhere, the assemblage may also reflect cereal-processing activities rather than cultivation.

In the boreholes of area S1, cereal-type morphotypes were only observed in boreholes C9, C10 and C14 within unit II. The remaining boreholes contain either no unit II or no cereal morphotypes, indicating that no agriculture was practiced at these sites. Likewise, a local agriculture and the removal of phytolith-bearing plant parts during harvest and their subsequent processing elsewhere cannot be fully excluded here. Low proportions in C9 (c. 1%) and C14 (<4%) indicate only limited or short-term cultivation. In contrast, borehole C10 shows significantly higher occurrence (up to c. 15%) across several samples, pointing to more extensive and sustained agricultural activity or cereal processing at this location. In area S2 only in core C15 cereal morphotypes were found. Cereals were identified in all three samples of unit II, but only in small amounts (<4%) suggesting only limited or short-term agricultural use or further processing. No cereals were detected in boreholes C16 or C17 (outside Bubastis), indicating no cultivation in those areas or complete removal of all plant parts in this area.

Palm morphotypes in unit II show varied distribution. They are absent near the Temple of Pepi I, occur in small amounts (up to c. 2%) in areas S1 and S2 (C6, C9, C10 and C14), and appear more frequently (up to c. 23%) in borehole C17 outside Bubastis. Accordingly, palm trees represented a minor to very minor part of the vegetation cover around Bubastis during deposition of unit II but were more common in the landscape further outside.

With the beginning of the occupation of Bubastis (around 3250 B.C.E.), human activities—especially the construction, use, and alteration of settlement structures—led to the gradual accumulation of anthropogenic sediments (unit I), forming substantial deposits over time (Garbe et al., 2025). As these sediments predominantly reflect settlement-related debris rather than agricultural soils, only few cereal morphotypes of local agrarian origin would be expected within unit I. However, inputs resulting from cereal processing or storage activities are plausible.

Around the Temple of Pepi I, cereal morphotypes were identified within unit I in cores C2–C5 (up to c. 7%). The increased occurrence in anthropogenic unit I compared to the other units indicates local agriculture in the immediate vicinity, cereal processing activities or storage facilities in this area (Dietrich et al. 2019). Due to the proximity of the Temple of Pepi I, a ritual utilization of cereals is also possible.

All boreholes in area S1 contain cereal morphotypes within unit I, though their frequencies vary considerably. High proportions in cores C7 (up to c. 34%), C11 (up to c. 23%), C13 (up to c. 10%), and C14 (up to c. 12%) point to intensive and long-term cereal processing or storage activities in this part of the settlement. Given the consistently elevated values even in deeper samples, and the current absence of agricultural use of the Tell, a recent introduction can be excluded. Lower proportions in other boreholes of area S1 still indicate cereal-related activity, but likely on a smaller scale or over shorter intervals. Outside the present extent of ancient Bubastis, cereal morphotypes in unit I were identified only in core C16. Notably, a redeposited layer of unit III between units I and II contains up to 9% cereal morphotypes, suggesting a phase of cereal-related activity that may relate to agricultural input documented in unit II of cores C10 and C15. However, in the uppermost samples, recent agricultural activity in the surrounding fields cannot be entirely ruled out as a contributing factor. No cereal morphotypes were found in unit I of cores C15 and C17, indicating an absence of cereal use or processing in these areas.

Palm morphotypes in unit I appear very irregularly. While no palms could be observed in the area around the Temple of Pepi I, they are observed in very small amounts (up to c. 2%) in the boreholes (C6, C7, C9, C11, C13, C14, C15 and C16) across areas S1 and S2 and entirely absent in C17 outside of Bubastis.

Borehole C7 includes a fluvial layer (unit IV), possibly linked to the sacred canals of the Temple of Bastet (Garbe et al., 2025; Meister et al., 2021). Due to slow-flowing waterbody, cereal morphotypes are unexpected; their presence (up to c. 5%) likely results from surrounding input or subsequent anthropogenic relocation processes.

5.3 Vegetation evolution from transition Late Pleistocene/Early Holocene to present time

Based on the archaeological record, sedimentological results and datable pottery from the boreholes, providing primarily a topographical and paleoenvironmental landscape reconstruction of the study area (Garbe et al., 2024a, 2024b, 2025; Lange-Athinodorou et al., 2019; Meister et al., 2021), these findings, combined with the phytolith results, allow a more comprehensive vegetation reconstruction in the region. Although some boreholes contain only a limited number of pottery assemblages—affecting their informational value and reliability—they nonetheless serve as useful chronological markers for the distinct sediment layers. When interpreting these finds, the potential for displacement must be taken into account, whether through natural processes such as fluvial transport or—particularly in ancient settlement contexts—through anthropogenic activities including the relocation or reworking of sediments. However, the absence of abrasion or erosion on the pottery fragments suggests that they were not transported by water. This supports the assumption that the ceramics likely originate from contexts in proximity to the respective boreholes (Meister et al., 2021; Toonen et al., 2022). The following reconstruction is, though in a more schematic and simplified form, particularly well-defined for the western and eastern part of Bubastis, located mainly between the western and eastern elevation including area S1 (Figure 8, left), as well as for the more northern part of Bubastis (Figure 8, right), which extends from the northern elevation eastward including area S2.

During the transition from the Late Pleistocene to the Early Holocene (c. 9000 B.C.E.), the Nile Delta was shaped by a pre-Nile regime, resulting in the Pleistocene Gezira deposits of the Mit Ghamr Formation (unit III). The landscape in the study area was characterized by a hilly or undulating terrain with sharp elevation changes over short distances and influenced by an extensive river network (bluish color in Figure 8A). Several large mounds, formed by active fluvial processes, contributed to localized sedimentation and erosion, creating a highly diverse topography. Surveys have identified major elevated features in the western, northern, central and eastern parts of the area, with additional prominent formations scattered throughout the central region in area S1, alongside numerous smaller mounds and gentle rises, which contribute to the area’s complex morphology. The area also includes several depressions, notably one between the western and central elevations and another situated to the south of the northern elevation. Additional deeper zones are mainly found in area S2 and east of the archaeological site (Garbe et al., 2024a, 2024b, 2025; Meister et al., 2021).

As outlined in Chapter 5.2, phytolith analysis indicates that the area was predominantly covered by grasses, as well as shrubs and trees during this period, while palms played only a minor role and were entirely absent around the western elevation (Figure 8A). These findings align with other studies suggesting that the Nile Delta at this time featured distinct grasslands and scrublands with scattered trees and palms, set within a relatively wet, marshy environment (Bernhardt et al., 2012; Pennington et al., 2017; Said, 1993).

From 6000 B.C.E., as the region transitioned first to a Large-Scale Crevassing Delta and later to a Meandering Delta, the investigated area experienced periodic flooding from the nearby Pelusiac and/or Tanitic Nile branch (for its possible location see Bietak, 1975; El Mahmoudi and Gabr, 2009; Ullmann et al., 2020). This flooding submerged almost all minor Gezira elevations and deeper elevated zones, leading to the gradual accumulation of clayey/loamy Bilqas sediments of unit II (Figure 8B).

Human occupation began around 3250 B.C.E., with initial settlements concentrated in the western and northwestern parts of ancient Bubastis, as evidenced by numerous archaeological findings. During the earliest phase, however, only a cemetery dating to the First and Second Dynasties was established on the western Kom—the western elevated mound that serves from this time for settlement activities (Figure 8B, Lange-Athinodorou, 2024). Additionally, pottery recovered from unit II and unit III around this area, dated to the Early Old Kingdom (c. 2850–2570 B.C.E.) suggests that the region was already at least partially affected by human activity during this period.

The discovery of cereal remains in unit II of core C3, alongside Early Old Kingdom pottery, marks the onset of agrarian activity in this region during that period (Figure 8B)—approximately 2,000 years later than previously documented for other sites in the Nile Delta (Rowland and Bertini, 2016; Wilson, 2007; Wilson et al., 2014; Zhao et al., 2022, 2025), and around 400 years after the initial occupation of Bubastis. While earlier and more extensive agricultural use of the area is conceivable, it remains unverified due to the lack of evidence, because geoarchaeological investigations in the northwestern part of Bubastis, where the earliest occupation started, are currently not feasible, limiting our understanding of the region's earliest agrarian history.

Aside from the notable emergence of cereal morphotypes in the area of the western Kom, the phytolith analyses do not indicate any significant shifts in regional vegetation. The remaining parts of the study area—particularly the eastern sector—appear to have remained largely unaffected, with the landscape continuing to be dominated by grasses, as well as shrubs and trees, interspersed with occasional palms (Figure 8B).

In the course of the Late Old Kingdom (c. 2570–2180 B.C.E.), building activity intensified in the area surrounding the western and central elevations. This led to the accumulation of anthropogenic sediments (unit I), primarily due to the construction and later removal of the Governor's Palace, as well as the erection of the Temple of Pepi I on the western Kom and the Temple of Bastet on the central elevation—referred to as the central Kom (Figure 8C). The widespread presence of Old Kingdom and especially Late Old Kingdom pottery fragments and at various depths in unit I at the western Kom provides clear evidence of intensive human activity and settlement expansion (Garbe et al., 2024a). This expansion contributed to a decline in natural vegetation, although the overall distribution of the remaining vegetation of grasses, shrubs and trees remained largely unchanged. Findings of cereal morphotypes in the central and northeastern part near the Temple of Pepi I (cores C3—C5) suggest that small-scale cereal cultivation, processing, or utilization continued in the immediate proximity during this period (Figure 8C).

Archaeological excavations in the northern part of the city have revealed several cemeteries (Bakr and Lange, 2017). The prominent elevation of the northern mound—referred to as the northern Kom—was selected for cemetery construction during the Late Old Kingdom (c. 2570–2180 B.C.E.), likely to protect the burial grounds from flooding, as the cemetery on the western Kom before. Consequently, settlement activity and the accumulation of anthropogenic sediments also extended from the northern Kom (Figure 8C). The general scarcity of anthropogenic deposits from the Old Kingdom in the eastern part of Bubastis suggests that these zones were not settled during this time. Instead, especially in area S2, the landscape was potentially used for agriculture, benefiting from frequent flooding and the accumulation of nutrient-rich fluvial sediments like proposed by Garbe et al., 2025. However, the phytolith results do not indicate any evidence of agricultural activities and no significant changes in vegetation patterns during this period in the eastern part of Bubastis (Figure 8C). While small-scale agricultural activity or post-processing cannot be entirely excluded, especially given the removal of phytolith-bearing inflorescence parts during harvesting, there is no evidence to suggest intensive cultivation.

Until the Late Period (c. 664–332 B.C.E.), the settlement structure of Bubastis underwent significant changes. While the areas surrounding the central Kom continued to be primarily influenced by the Temple of Bastet, and the northern Kom remained associated with cemetery use, the Temple of Pepi I at the western Kom was removed (Figure 8D). Sedimentological data provide no clear indication of subsequent substantial human activity in the western Kom. In the eastern part of area S1, the Temple of Hermes was likely constructed on the eastern elevated mound—referred to as the eastern Kom (Garbe et al. 2024b, 2025). Moreover, settlement activities expanded beyond the Koms, especially to the east into area S1, however, certain zones within the settlement area remained uninhabited (Garbe et al., 2025).

This development was accompanied by a further reduction in natural vegetation within zones of intensified settlement activity. Cereal assemblages from borehole C10 indicate widespread agricultural practices at the northeastern edge of area S1 or intensive cereal post-processing and, regarding the different sedimentological units containing these morphotypes, probably also cereal processing activities in the surrounding zones especially further west near borehole C7 (Figure 8D). In regions less impacted by human activity—such as the areas east of the city and area S2—grasses, shrubs and trees continued to dominate the landscape. The absence of evidence for subsequent use of the area surrounding the Temple of Pepi I suggest that the temple ruins at the western Kom likely remain undisturbed, gradually being reclaimed by natural vegetation including grasses, shrubs and trees.

Further findings from the study area point to a gradual eastward expansion and increasing settlement intensity, particularly within area S1, continuing up to the Ptolemaic Period (c. 332–30 B.C.E.). This expansion was accompanied by a further decline in natural vegetation in these parts of the settlement. Although the precise extent of settlement activity cannot be determined, it is approximately outlined in Figure 8E (Garbe et al., 2025). The distribution of cereal phytoliths—especially in cores C10 and C14—indicates sustained intense agricultural use along the northeastern edge of area S1 and probably in neighboring parts in area S2, with buildings showing evidence of cereal processing activities present throughout the area S1, as evidenced by cores C7, C8, and C13. In contrast, regions that remained relatively undisturbed by human activity, such as the areas east and northeast of the Tell, retained their natural vegetation, dominated by grasses, shrubs and trees.

Until the Roman Period (c. 30 B.C.E.–200 C.E.), the settlement areas continued to expand, likely reaching their maximum extent by covering the entirety of area S1 and most of area S2 (Figure 8F; Garbe et al., 2025). This expansion appears to have almost entirely displaced natural vegetation within the inhabited zones. However, phytolith analyses indicate that the small patches of remaining vegetation did not undergo notable changes compared to earlier periods. As the settlement areas became densely built-up, agricultural activities were pushed to the outskirts of Bubastis. Evidence of cultivation or cereal post-processing activities on fields during the Roman Period is limited to sediments from borehole C15 in area S2 (Figure 8F). Given the need to supply the large urban population with food and supported by samples indicating intensive cereal use (cores C7 and C10), it is highly likely that extensive agricultural zones existed in the surrounding region. Nevertheless, these zones did not extend to borehole C17 further northeast of Bubastis, where no traces of agricultural activity were detected.

Following the abandonment of the city around 200 C.E., most of the area investigated became silted up. In the modern landscape, only the stone remains of the Temples of Pepi I and Temple of Bastet, along with collapsed mudbrick walls in the northern part of the settlement, are still visible on the surface. With the exception of surface ceramic artefacts, the remaining areas exhibit no identifiable evidence of earlier occupation or development (Figure 8G). Striking contrasts characterize the present-day appearance of the study area: area S1, situated within the protected archaeological zone of Tell Basta, consists of entirely uncultivated, gently undulating terrain, whereas area S2, located to the north of S1, is markedly flat and intensively used for modern construction and agriculture (Garbe et al., 2025). The vegetation cover has been altered by both historical and contemporary anthropogenic influences. Area S1 is sparsely vegetated, with the western section almost completely barren, while the eastern section supports slightly denser plant growth, including grasses and small shrubs such as halfa grass (Desmostachya bipinnata L. Stapf) and camelthorn (Alhagi maurorum Medik.), as well as scattered trees like the Jerusalem thorn (Parkinsonia aculeata L.). In contrast, area S2 and the region further east of the Tell are characterized by dense vegetation resulting from intense current cultivation, including rice (Oryza sativa L.), various cereal crops (e.g., Triticum aestivum L.), berseem clover (Trifolium alexandrinum L.), and onions (Allium cepa L.), typical of the Nile Delta region (Wunderlich, 1989).

6 Conclusions

This study provides the first integrated paleovegetation reconstruction for the ancient city of Bubastis, combining phytolith analyses from 17 boreholes with previously established stratigraphic and paleoenvironmental frameworks. The results reveal long-term landscape and vegetation dynamics shaped by the interplay of natural Nile Delta processes and human activity from the Late Pleistocene through the Roman Period.

During the transition from the Late Pleistocene to the Early Holocene, the study area was characterized by a mosaic of dominant grasses, as well as shrubs and trees within a geomorphologically diverse setting dominated by sandy Gezira mounds. Phytolith assemblages indicate that palms were only sparsely present, and no evidence exists for wild or domesticated cereal cultivation during this early phase. The subsequent deposition of Bilqas floodplain sediments from the Early Holocene onwards created fertile soils in the region; however, the earliest clear indicators of cereal agriculture or cereal post-processing activities appear in the Early Old Kingdom, particularly in the surroundings of the Temple of Pepi I. This marks the onset of localized cereal cultivation around Bubastis, occurring around 2000 years after earlier agricultural developments elsewhere in the Nile Delta.

From the Old Kingdom onward, human occupation increasingly reshaped the vegetation structure. Settlement growth around the western, central, and northern elevations led to widespread deposition of anthropogenic sediments (unit I), accompanied by a reduction in natural vegetation. Cereal morphotypes become more frequent from this period onward, reflecting both agricultural production in the settlement's periphery and cereal processing or storage within residential zones. Especially in area S1, high cereal proportions in unit I point to intensive and sustained cereal-related activities, while more moderate values in other areas suggest smaller-scale or episodic use. Agricultural activity is particularly evident along the northeastern margin of area S1 and parts of area S2 during the Late Period and into the Ptolemaic and Roman periods, indicating that cultivation zones shifted outward as the settlement grew denser.

Overall, the phytolith evidence demonstrates that agricultural fields were consistently located outside the main settlement area, with cereals entering the urban area primarily through processing, consumption, and storage. By the Roman Period, natural vegetation had largely disappeared from the inhabited areas, while agricultural production continued in the surrounding floodplain landscapes. However, the results also indicate that vegetation change in the study area is driven mainly by anthropogenic influences, while the effects of changing climatic conditions are comparatively minor. The region's dominant vegetation—grasses, shrubs and trees—remains, despite marginal fluctuations relatively stable and is only locally replaced by cultivated and processed cereals in settlement areas.

This study highlights the value of combining phytolith data with geomorphological and archaeological information to reconstruct the environmental history of ancient urban centers. For Bubastis, the results clarify not only the development of local vegetation but also the spatial organization of agricultural production and food-processing activities over several millennia.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fearc.2026.1778152/full#supplementary-material

Overview of the sedimentological results. Supplementary material at the source

Overview of the phytolith results. Supplementary material at the source

Overview of the original phytolith data. Supplementary material at the source

Footnotes

Edited by: Marco Madella, Pompeu Fabra University, Spain

Reviewed by: Charlotte Diffey, University of Oxford, United Kingdom

Marc Testé, Université Paris 8, France