Early fishing

The earliest record of fish at a hominin site comes from the Koobi Fora Formation in Kenya and dates to 1.95 million years ago [1]. Since these early remains were discovered, archaeologists have paid closer attention to the use of fish in anthropogenic sites, identifying fish remains or indirect evidence for fishing based on use wear traces or residues on stone tools from a number of Lower and Middle Paleolithic and Middle Stone Age sites in Africa and Eurasia [5–10]. Marean [9] argues that the earliest evidence for a coastal adaptive strategy appears with early Homo sapiens about 110 kya in South Africa. Although the presence of fish is increasingly apparent in these early sites and others, they are not common and even if they were eaten, fish do not become substantial or routine components of human diets until much later.

Fish take on increasingly important roles in the Upper Paleolithic economies in east and south Asia after this date [3,11]; and increase in ubiquity and abundance throughout the Upper Paleolithic in other parts of Europe and Asia as well [12–15]. Direct evidence for the exploitation of coastal Mediterranean species and cold-adapted Atlantic species appears in the Solutrean at Cueva de Nerja in Spain [16,17]. Relief sculpture and painting as well as body adornments, from Upper Paleolithic sites in Europe also reveal that fish became sufficiently important to hold a role in forager imaginations. The significance of fish remains increases in Epipaleolithic and Mesolithic sites, especially those situated in coastal, lake or riverine settings in Northern Europe and the Atlantic [18]. Specialized fishing activities are also well documented in the Mediterranean during this period at Vela Spila in Croatia [19], Franchthi Cave in Greece [13,20] and Grotta dell’Uzzo in Italy [21]. In southeastern and northern Europe, an increase in the role of freshwater fish in the diet of forager populations is attested from the early Mesolithic onward on the basis of fish remains and stable isotope values in bone collagen [22–29], as well as in dental calculus [27].

Fishing technology

Despite the interest in the time depth of aquatic resource exploitation and broad spectrum human diets, fishing technology is elusive throughout the Paleolithic. While fishing technology may have included a wide range of tools such as traps, lines, gorges, weights, barbed points, harpoons and hooks, such artefacts are rarely recovered from ancient prehistoric sites. Barbed points and harpoons appear early on in Africa [30] and were important cultural markers of Upper Paleolithic techno-complexes such as the Magdalenian and/or the Azilian [31,32] but their association with fishing is not secure [33,34]. Other fishing technologies such as nets or traps, are made largely of perishable materials that preserve only in unusual conditions, thus their early appearance is difficult to pinpoint in the archaeological record. Indisputable fishing technology like hooks do not appear until the end of the Upper Paleolithic and the Epipaleolithic when zooarchaeological evidence for fishing also expands in earnest as part of a larger shift toward the exploitation of smaller game species as diets diversify.

Possible straight fish hooks (gorges or hameçon droits) were in use during the Magdalenian when artistic representations of fish appear [35], but the earliest fish hooks known thus far are made of shell and come from the cave of Asitau Kuru (formerly known as Jerimalai) in East Timor [36]. A broken shell-fish hook from the site has been dated between 16 and 23 kya [3]. Another early fish hook, also made from shell, was found at Sakitari Cave in Japan (22,380−22,770 cal BP) [11]. Archaeological excavations at the site of Makpan Cave have also recovered entire hooks, roughouts and preforms, exposing the entire chaine operatoire for the production of shell hook. The earliest bone fish hooks, date to Late Epipaleolithic (Natufian) sites in the southern Levant [37] and include those from JRD presented here. After this, bone fish hooks become widespread. Slightly more recent hooks have also been found in the Late Upper Paleolithic deposits at the site of Wustermark 22 in northeast Germany (ca. 12,300 cal BP) and are frequently found in Mesolithic sites in Europe [38,39]. Later in the Mesolithic, waterlogged deposits where anaerobic conditions allowed the preservation of otherwise perishable botanical remains, are a primary source for prehistoric fishing gear, including fish-traps, nets, gorges, hooks, harpoons, weirs and more [25,28,29,40–42].

Early fishing and fishing technology in the Epipaleolithic southern Levant

In the southern Levant abundant fish remains were found at the site of Gesher Benot Ya’aqov located on the banks of the Upper Jordan River within 1.5 kilometer of JRD [7,43,44]. Fish bones make sporadic appearances in Middle and Upper Paleolithic sites in the Levant, but they are always rare [45]. Indisputable evidence for invested human fishing first appears at Early Epipaleolithic Ohalo II (ca. 23 cal kya) on the shore of the Sea of Galilee [46–49]. The fish bone assemblage is accompanied by material remains that probably served as fishing technology such as net sinkers [50,51]. Importantly, fragments of twisted fibers were preserved in the waterlogged sediments and may have been used to make nets or traps [52].

After this time, fish are much more ubiquitous in Levantine archaeological sites [53], especially in sites close to major water bodies such as the Mediterranean Sea and the freshwater system of the Jordan Valley [54]. Fish are found in most Natufian sites with good preservation and careful recovery [4,37,45,50,53,55], but are especially abundant at Natufian sites in the Upper Jordan Valley including Eynan in the Hula Valley [56,57] and Nahal Ein Gev II near the eastern shore of the Sea of Galilee [58].

The Early Natufian layer at Kebara Cave was home to the first fish hooks found in the Levant, as well as harpoons that may have been used for fishing. Two complete hooks, one plain and one grooved around the neck, and two broken hooks were recovered [37,59]. Numerous bone fish hook fragments were also found in the Final Natufian layer at Eynan [57,60] as was one broken hook at Early Natufian Hof Shahaf near the northwestern corner of the Sea of Galilee [61]. A single hook was found at Late Natufian Hayonim Terrace [62]. After the Natufian, the use of fish hooks continues into the Pre-Pottery Neolithic, although primarily at Mediterranean coastal sites. They are especially common at the water-logged site of Atlit Yam [63–65]. The switch to metal as the primary raw material for fishing hooks production took place as early as the Chalcolithic period in the Levant [66].

The site of Jordan River Dureijat (JRD)

Jordan River Dureijat (JRD) is an ephemeral, short-term encampment located on the southern shore of Paleolake Hula that was intermittently occupied over a span of about 10,000 years (ca. 20,000 to 10,000 cal BP) during the Levantine Epipaleolithic period [4]. Stratigraphic, radiocarbon and technological evidence indicate that JRD is one of very few sites to be occupied in each of the Early, Middle and Late Epipaleolithic phases.

JRD is located on the east bank of the Upper Jordan River where it flows south out of the Hula Valley in northern Israel (Fig 1). The site was discovered during an archaeological survey preceding a massive drainage operation of the Jordan River in 1999 [67,68]. JRD’s archaeological horizons stretch across 50 m of the river bank adjacent to the outlet of the small Dureijat Stream (Fig 1). A test excavation undertaken in 2002 [69] was followed by six full excavation seasons between 2014 and 2019. Excavation exposed a sequence of sediments that accumulated within the fluctuating water levels of Paleolake Hula. The archaeological horizons are found within near-shore layers that accumulated during low water stands and reflect human activity on the bank of Paleolake Hula. These archaeological horizons are separated by archaeologically sterile, fine silt layers deposited when the lake level was higher (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

a) Location of the site of JRD; b) view of JRD from the east and its location on the Jordan River adjacent to the Dureijat stream; c) map of excavation areas and test trenches.

https://doi.org/10.1371/journal.pone.0257710.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

a) Stratigraphy, chronology and cultural affiliation of Area B sequence; b) Area D stratigraphic sequence.

https://doi.org/10.1371/journal.pone.0257710.g002

The site’s stratigraphic sequence and chronology were established in Area B, the primary excavation area at the site (Fig 2D). The chronology was built from numerous radiocarbon dates (Fig 2; [4]. The lowest layer excavated at the site (Layer 5) dates to ca. 20,000 years cal BP and is affiliated with an Early Epipaleolithic lithic tradition. The lithics from Layer 4 are clearly ascribed to the Middle Epipaleolithic, Geometric Kebaran tradition and date to 17,460–15,750 cal BP. The three upper, layers 3c, 3b and 3a, are affiliated with the Late Epipaleolithic Natufian culture (15,000–11,500 cal BP) based on radiocarbon dates and the presence of typical flint artifacts such as microlithic lunates [4,70]. The archaeological sequence of the site ends with silty Layer 3–0, dated to the initial Holocene. Because the accumulation of Layer 3–0 is complex, and the anthropogenic input is not as significant as in other archaeological layers, the context is less secure than the archaeological horizons below. Nevertheless, the finds from this layer include artifacts such as El-Khiam arrowheads and limestone axes that can be clearly ascribed to the Early Neolithic.

During the 2019 season, an additional test excavation (Area D) was opened 10 meters to the north of Area B (Fig 1D). This area yielded lower artifact densities, likely because of its original location in the nearshore, shallow water of the paleolake itself. Nevertheless, in addition to faunal and lithic remains, three bone fish hooks and a single grooved pebble were excavated from this area. The stratigraphy of Area D has been correlated with the stratigraphy in Area B (Fig 2).

The archaeological evidence from JRD indicates that although some of the intermittent visits by Epipaleolithic peoples were longer and involved more intensive activities especially during the Natufian period, the excavated area was never used for habitation. JRD was a place that people visited again and again to take advantage of the confluence of diverse lake shore resources. Like other archaeological sites on the banks of the Upper Jordan River [71–73], the sediments at JRD have been waterlogged since they accumulated. The rare outstanding preservation of organic remains sets the site apart from most other Levantine Epipaleolithic sites [4,69,74].

Stratigraphic provenance of the JRD hooks

All of the hooks were recovered from in situ excavation contexts with known provenience and stratigraphic location (Table 1; Fig 4). All but two hooks (#7 and 17, Table 1), were found during the sorting of sediments from these contexts after they were wet sieved through 2 mm mesh in the Jordan River. Three broken hooks from Area B (#15, 16 and 18), and one complete hook from Area D (#12), were found at the contact between Layer 3–0 and 3a (Fig 2). Based on the radiocarbon chronology and associated Early Neolithic archaeological finds, Layer 3–0 is likely Early Neolithic in age [4]. The upper surface of 3a may have been disturbed by humans after it accumulated. The similarity of the four hooks to the Natufian assemblage suggest that they derive from the very end of the Epipaleolithic, but they could also belong to the very early Neolithic.

The rest of the JRD bone hooks were found in the three Natufian layers at the site (3a, 3b and 3c), or their stratigraphic equivalents in Area D (Fig 2; Table 1). Six hooks were excavated from Layer 3a and one from the Area D equivalent, Layer D4. Three hooks (#1, 2 and 5) were found in Layer 3b and a single hook, the largest (#13; Figs 3 and 4), was excavated from Layer D8 –which is stratigraphically equivalent to Layer 3c in Area B. A single broken hook (#10) was recovered from the west part of Area B (Square M98) where Layer 3c cuts into Layer 4. In this area [4], the stratigraphy becomes more complex and the separation of the layers is challenging and thus the stratigraphic affiliation of hook #10 is the least secure. It is likely that it belongs to Natufian Layer 3c rather than Middle Epipaleolithic Layer 4. It should be noted that Layer 3c and Layer 4 were excavated across a large areal surface (>35 meters; Fig 1D) and all sediments were sieved and sorted. Still, neither layer yielded any hooks other than the large and morphologically divergent hook #13. Hooks were also absent from Layer 5, the oldest layer of the site dating to the Early Epipaleolithic.

In comparison to other archaeological horizons at JRD, Layer 3b is thin, with a low density of archaeological finds. It probably represents a short period of non-extensive human activity in near-shore shallow water. In accordance, the volume of sediments from this layer was small and thus sediments were only sampled for sieving and sorting. Nevertheless, three of the best preserved and unique hooks were unearthed from this layer (Table 1). In contrast, Layer 3c, which encapsulates the earliest stage of the Natufian represented at JRD, is a much more clearly defined archaeological horizon that contained diverse archaeological remains (flint tools, limestone net sinkers, basalt tools, animal bones and botanical remains) compared to other layers, yet it yielded only a single bone tool that can be classified as a hook, but clearly deviates from all other hooks in size and morphology.

Preservation and raw material

Like the faunal remains from JRD, the state of preservation of the hooks is good overall (Fig 3). Observations under high magnification revealed post-depositional alteration, such as soil and salt concretions, bacteria damage, rootlet etching and sediment patination on a few specimens, but were not severe enough to affect the functional interpretation of the hooks.

All of the hooks from JRD are made of bone except for one fragmentary hook (#16) made of tooth enamel (likely wild boar tusk). Heavy working and reshaping of the bone during hook manufacture erased any diagnostic features on the bones making it challenging to determine the taxon of the bones used to produce the JRD hooks. Analytical methods that can identify taxon based on protein identification are destructive so could not be used in this study. The size and the natural curvature of the hooks, as seen in the section view of hooks #1, 3 and 9 (Fig 4), suggest that some were produced from the long bones of small and medium sized herbivores, such as gazelle or fallow deer, the two most common ungulates in the JRD assemblage.

Terminology and typology of fish hooks

The terminology used to describe the parts of the fish hook is largely standardized for modern hooks [75,76], although the terms for some parts vary by region and manufacturer (i.e., some prefer the term tip over point or bite over throat). Here, we largely follow the terminology of Thomas et al. [77] developed for contemporary hooks (Fig 5). Nevertheless, because prehistoric hooks sometimes differ from modern hooks in certain features, we also use terminology developed to describe prehistoric and historic fish hooks from the Pacific region [78,79].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Fish hook terminology and measurements used in the current study.

https://doi.org/10.1371/journal.pone.0257710.g005

The morphology and specific attributes of the fish hooks provide crucial information about their function. Determining the size of certain dimensions and the range of hook attributes is necessary to assess change in the function of the hooks over time, the type of fish targeted by the prehistoric anglers and their fishing techniques. For example, the length, width, and bite size (Fig 5) limits both the size of the bait and the size of the fish that can be captured (small hooks can capture large and small fish, whereas large hooks can capture only large fish). Likewise, the size of the gape, bite and point-angle affect how easily a fish can swallow the hook and how likely the point is to get stuck in the fish’s mouth [e.g., 80]. The width of the cross-section of the bend influences the ability of the hook to withstand torque without snapping during fishing. The measurements can thus, provide some insight into the fishing strategies and fish types targeted by the fishers from JRD.

Measurement of the fish hooks

Studies of prehistoric fish hooks are often regional in character and based on small assemblages [e.g. 39,81–85]. In addition, artifact features are often described qualitatively, introducing ambiguity in the classification and hampering the comparison among assemblages. In their study on Scandinavian fish hooks, Olson et al. [86] proposed a series of measurements of the linear dimensions and angles of the most common segments typically used to describe modern day fish hooks [87]. Given the importance of providing standardized methods of measurement to compare fish hooks from different times and places, and the challenge of taking accurate measurements of prehistoric hooks posed by their irregular shape and fragile nature, we extracted some of their measurements using high-resolution 3D scans. Digital 3D models of the hooks were obtained using Polymetric® PTM-4c structured light scanners at the Computational Archaeology Laboratory (CAL) at the Hebrew University of Jerusalem. The use of high resolution 3D models allows precise and objective measurements to be taken and accurately replicated without having to handle these highly fragile artifacts, including parameters that cannot be measured using traditional methods such as the position of the center of mass [88]. In addition, the scans can be used with the open source MeshLab software [89] to virtually refit scanned fish hook fragments. Finally, the highly accurate 3D models provide a digital archive of these hooks, preserving them for future study.

The 3D models were manually positioned so that the hook’s shank was parallel to the y-axis with the bend on the plane defined by the y- and x-axes using the Artifact3-D software [freely available upon request to CAL; 88]. Since JRD hooks are almost flat, a 2D outline was created for each complete or refitted item using the ‘Process Object’ function to simplify the measurement procedure. Outlines are defined by a series of vertexes with width and length coordinates. Based on the position of the center of mass (automatically defined by Artifact 3-D), the outlines were then segmented into three parts: the bend (below the x-axis), the shank, and the point (above the x-axis and on either side of the y-axis) (Figs 5 and 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6.

Measurements extracted based on the profile (legend a) and the cross-section (legend b) of the hooks.

https://doi.org/10.1371/journal.pone.0257710.g006

We used the 3D models to extract the maximum length and width, gape, and bite (Fig 5). measurements proposed by Olson et al. [86]. In addition, the angle between the point and the shank was calculated. Finally, we used the 3D models to measure the width and thickness of the transversal cross-section of the bend at its lowest point and its maximum axis (width and thickness respectively) with the ‘Create Cut’ function in Artifact3-D.

The following measurements were taken from each hook when possible (Figs 5 and 6):

Maximum length (L; mm): distance on the y-axis between the lowest and highest vertex of the outline (length). Measured on all complete and refitted items.

Maximum width (W; mm): distance on the x-axis from the first vertex at the left and the last vertex on the right of the profile. Measured on all complete and refitted items.

Hook size (HS; mm): the square root of the product of bend thickness and height. Extracting the square root of the product allows comparison of values with different dimensionalities (1D and 2D). Measured for all complete items.

Gape (G; mm): distance on the x-axis between the highest vertex of the point and the shank.

Bite (B; mm): distance on the y-axis between the highest vertex of the point and the lowest vertex of the profile. Measured on all complete and refitted items.

Point angle (PA; deg.): angle between the point and the shank. The outline of the point and the shank was further subdivided into slices taken along the long axis (100 slices). For each slice, the mean of the length and width coordinates of all the vertexes was calculated. Two straight lines were fitted to best approximate the values of the mean coordinates of all the slices of the point and shank, respectively. The angle between the two fitted lines represents the angle between the point and the shank. Measured on all complete and refitted items.

Bend thickness (T; mm): distance on the z-axis between the most external vertices of the transversal cross-section of the bend at its lower point. Measured for all items.

Bend height (H; mm): distance on the x-axis between the lower and the higher vertices of the transversal cross-section of the hook bend at its lower point. Measured for all items.

Bend size (BS; mm): the square root of the product of bend thickness and height. Measured for all items.

The measurement and the 3D digital models of the hooks are available in “S1 Table”.

Hook size results.

The JRD hook sample is too small to allow more than descriptive statistical analysis. The hooks vary considerably in all size dimensions and ratios (Fig 6, Table 2). Despite this variability, average maximum length of complete hooks declines over time (Fig 4; Table 2). The hooks from Layer 3b are larger on average than the hooks from Layer 3a, while hook #13, the only example from Layer D-8 (equivalent to Layer 3c), is the earliest and the largest of all. A similar trend is observed in the gape and bite dimensions and, although weaker, in bend size. Finally, high variability in the point angle is observed both within and among the layers (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Measurements of the JRD bone hooks by layer.

https://doi.org/10.1371/journal.pone.0257710.t002

Morphological characters of fish hooks

Next, we describe each hook’s overall shape. On the most robust level, we categorize the hooks according to the modern-day division of J-shaped and circular categories [76,90,91]. We then zoom in to characterize important, yet variable features of the JRD hooks including barbs that prevent the fish from escaping after it is captured and line attachment features. We do so by recording the form, location and presence or absence of barbs, grooves and knobs on the shank, and the point of the hook.

The most striking observation regarding the JRD hooks is their variability (Figs 3 and 4; Table 1). Variability is evident in all aspects of hook morphology–their size, angle and the presence and location of features and even the solutions for line attachment. The shape, size range and angles of the fish hooks fit well within the range of variation of modern fish hooks, but a few features found in the JRD hooks are not observed in present day hooks, nor in any early prehistoric hooks that we know of (see below).

In regard to their general morphology (Figs 3 and 4; Table 1), most of the complete JRD hooks belong to the J-type (#2, 3, 4, 11 and 13), while two hooks (#1 and 12) are more circular in shape, although not fully circular by modern criteria. Although no off-set hooks (where the plane of the shank deviates sideways from that of the point; [90] were found at JRD (or any Natufian site), the bended natural morphology of the long bones used as blanks for the production of hooks #1 and 3 may have served a similar purpose.

The most notable features on the JRD hooks include barbs and line attachment modalities (Table 1).

Functional analysis of bone fish hooks

Functional analysis results.

Reconstructing hook reduction sequences was challenging as no unfinished hooks, roughouts or manufacturing waste were identified at the site. Moreover, evidence left by intense sharpening and/or the prolonged use of the hooks had often worn off anatomical or technological features. Nevertheless, major trends could be identified. The technological analysis of the JRD hooks identified the following manufacturing phases (for hook parts terminology see Fig 5):

1. Extraction of the blanks: Rectangular blanks were extracted from the bone diaphysis through longitudinal and transversal cutting (Figs 7C, 7F, 8B, 8O and 9C) and roughly shaped through scraping and/or grinding (Figs 7F, 7I, 8B, 8O and 9C).
2. Preparation of the bend: drilling (probably using a flint? bit) was used to initiate the concave side of the bend of the hook (Figs 7B, 7G, 8I and 9F). When necessary, the area of the perforation was regularized by scraping or grinding (Figs 7K, 8I and 9F). It was not possible to determine whether the hook was perforated before or after the extraction of the blank as roughouts/manufacturing waste were not recovered at JRD.
3. Creation of the shank/throat: the shank, bend, point and the resulting throat of the hook were carefully regularized using longitudinal scraping. On the basis of our experiments, we can suggest that both flint and basalt tools were used, the first leaving very regular striations. Precise strokes were always applied along the shank (#4, 11, 6, 7, 14 and 13; Fig 4), the outer side of the bend, and the throat of the hooks. Sometimes scraping of the shank created a sinuous profile or left chatter marks, which are typical microscopic markers of this technique (Fig 7F and 7I).
4. Creation of the point/barbs: Points were sharpened through regular scraping (Fig 7J). In a few cases, barbs were created through lateral transversal grooving (#1, 2, 3 and 5; Fig 7C and 7K).
5. Line attachment features on the shank. Most of the features on the hooks were created by grooving. Grooves were created around the full circumference of the distal end of the shank of four hooks (#2, 4, 6 and 12) to create a small globular head (Fig 8K and 8O) or in three cases, on the lateral side to create single or double notches (e.g., #1 and 3 and 11; Figs 7M and 8B–8D). In one case, a lateral knob was also part of the hook’s line attachment system (Fig 7I), while the shank was worked into a point on another (Fig 9A).
6. Technological adjustments on the bend. In two cases, deep transversal incisions were made across the bend of the hooks using a stone tool (#4 and 9). These likely served to connect a cord for a lure (see below; Figs 8M and 10D–10F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7.

Technological traces and functional modifications on archaeological hooks (a) hook #1, h) hook #2, l) hook #3. b) Detail of the marks left during the preparation of the bend through drilling; c) detail of the barb and transversal grooving associated with its production; d) surface alteration associated with the decomposition of organic residues, still preserved on the other side of the hook; e) dashed lines indicate changes of color according to the disposition of the organic residues; f) scraping traces associated with the production of the shaft; g) perforation marks; i) j) k) transversal grooving applied to produce the barb; m) detail of the top of the shank. The arrows indicate technological mistakes produced by a lithic tool for creating a double knob for line attachment.

https://doi.org/10.1371/journal.pone.0257710.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8.

Technological traces, use wear traces and residues on archaeological hooks (a) hook #14; f) hook #16, h) hook #16; j,p) hook #4). b) longitudinal cutting marks produced when extracting the rectangular blank from the bone diaphysis; c, d) details of functional features (rounding, striations and residues) related to the attachment of the line; e) close-up photos of the line attachment system of an ethnographic lure hook with details and distribution of use traces and residues (hook # 25.445) from the Museum of Archaeology and Anthropology (UK); g) transversal cutting marks; i) marks associated with drilling used to create the concave side of the bend of the hook; k) rounding observed on the knob and particles of residual brownish adhesive matter inside the notch produced for creating the knob; l) particles of residual brownish adhesive matter; m) deep transversal incisions across the bend of the hook likely used to connect a cord for a lure. Note patches of residual brownish adhesive matter; n) short oblique and superimposed striations characterized by an initial depression and rough bottom, possibly fish tooth marks; o) rounding on the knob. Note also the longitudinal cutting marks left during the preparation of the hook; r) functional reconstruction of the hook according to archaeological use wear and residues.

https://doi.org/10.1371/journal.pone.0257710.g008

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Technological traces, use wear traces and residues on archaeological hook #13.

a) functional reconstruction of the hook according to archaeological use wear and residues; a) rounding developed on the tip of the shaft; b) bright polish with flat topography and fine transversal striations identified on the upper part of the shank; c) longitudinal cutting marks produced to extract rectangular blanks from the bone diaphysis; d,e) particles of residual brownish adhesive matter with clear transversal features; f) marks associated with drilling used to create the concave side of the bend of the hook and subsequent abrasion.

https://doi.org/10.1371/journal.pone.0257710.g009

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10.

Technological traces, use wear traces and residues on archaeological hook #9 a,b,c) details of a plant fiber identified inside the notch on the top of the shank; d-f) animal hair recovered inside the groove on the bend of the hook.

https://doi.org/10.1371/journal.pone.0257710.g010

All of the JRD bone fish hooks show wear traces related to prolonged use as well as residues connected to line attachment features and when present the use of lures. At low magnification, rounding is observed on the hook’s shanks and especially inside the notches created by the knobs (Figs 7M and 8C). Use-traces are also visible across the knobs (Figs 8B, 9C and 9D), inside the shank grooves (Fig 7I), across the bends (Figs7K, 8D and 11D), along the upper part of the shanks, and on the tips (Fig 10A). Fine striations, transversally oriented to the main axis of the hook located on the shanks (Fig 9C and 9D) and line attachment features (Fig 7D, 7E and 7I) relate to the connection and securing of the line.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 11. Grooved pebbles from JRD.

a) Selected grooved pebbles; b) 3D models of the six analyzed pebbles. Numbers are according to site data catalog.

https://doi.org/10.1371/journal.pone.0257710.g011

At high magnification patches of bright polish with flat topography and fine transversal striations have been identified on the upper parts of the shank (Fig 9B). On hooks #1, 4, 9, 13 and 14, rounding and striations on the line-attachment features are associated with residues of clear feature and orientation (Figs 7D, 7E, 8C, 8D, 8K, 8L, 9D, 9E, 10B and 10C). In particular, three types of residues have been identified: (a) particles of yellowish/brownish adhesive matter, rich in charcoal inclusions, and characterized by a flattened profile and transversal impressions (Figs 8C, 8D, 8L, 9D and 9E); (b) plant fibers (Fig 10B and 10D); and (c) animal hair (Fig 10E and 10F). Wear traces and yellowish/brownish residues are remarkably well preserved on the shank of hook #13, suggesting that adhesive was applied to the fibers to better secure the hook to the line (Fig 9D and 9E). Hooks #3, 4, 9 and 11 bear similar traces of line attachment method and residues (Figs 8C, 8D, 8K, 8L, 10B and 10C). A plant fiber was found inside the notches that form the line-attachment of hook #9 (Fig 10B and 10C). On hook #1, transversally oriented post-depositional bacterial modifications of the surface of the shank are associated with spots of brownish residues suggesting the coiling of threads around the shank (Fig 7D and 7E).

Interestingly, a few particles of adhesive and one long hair fiber were identified on the bends of hook #4 and 9 respectively (Figs 8M, 8Q and 10D–10F). The combination of such residues and evidence for the technological adjustment of the bend through deep transversal incisions hints at the use of lures at JRD (Fig 8R).

Use-wear traces are less developed on the points and the barbs than on other hook parts probably due to resharpening of the hook point after prolonged use. Along with traces and residues produced by prolonged use, the shaft of hook #9 is marked by short oblique/transversal striations, sometimes superimposed, and characterized by an initial depression, rough bottom and irregular walls. These traces resemble fish tooth marks such as the ones documented on experimental as well as ethnographic hooks (Fig 8E; [25,112]).

Materials

Ten small, grooved pebbles were recovered at JRD. These pebbles are less than 3 cm in maximal diameter and are bisected by a single or, in one case, double groove (Table 3; Fig 11).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Size measurements of the JRD grooved pebbles.

The JRD grooved pebbles are reposited at the Prehistory Laboratory, Tel Hai College, Israel. The 3D grooved pebble models are available (wrl format) at https://doi.org/10.7910/DVN/VTKRQW.

https://doi.org/10.1371/journal.pone.0257710.t003

Six of the ten grooved stones are produced on smooth basalt pebbles and four on limestone. The mean maximal dimension of the pebbles is 17.4 mm (s.d. 4.5 mm), the width is 13.4 mm on average (s.d. 4.1 mm) and the mean circumference is 52.0 mm (s.d. 14.0). The average weight is 4.8g with a very large s.d. of 4.2g reflecting a wide weight range from 1 to 14.5g. Sample size is not large enough to allow statistical analysis, but the distribution shows a concentration of pebbles around 5g in weight with both small and large outliers. The grooves run either longitudinally, or more commonly, transversally to the main axis of the object (Fig 11), and one case shows a double parallel groove (pebble #110; Fig 11).

Six of the ten grooved pebbles found at JRD were analyzed in detail for the current study (Fig 11; Table 3). Two of the analyzed pebbles are basalt and four are limestone.

Archaeological context

Of the ten grooved pebbles from JRD, four were identified during excavation and six were found during sediment sorting. Six pebbles originated from Layer 3a, the latest Natufian layer at JRD (Fig 2). Two pebbles were obtained from Layer 3c and a single pebble is from the lower Natufian Layer D6 in Area D (equivalent to 3b). Thus, 90% of the grooved pebbles originate from the Natufian layers of the site. The only exception is pebble #106, which was found in Layer 5. The groove around pebble #106 is not as deep as the grooves on the other pebbles and it may represent an unfinished product (Fig 12A). Importantly, with the exception of pebble #106, the archaeological context of the small grooved pebbles matches that of the bone fish hooks.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 12.

Technological modifications (grooves with variables depths) of a) pebble #106, d) pebble #105, g) pebble #104. Details of the incised grooves with rare technological errors indicated as shallow lines parallel to the main groove. i-k) Microscopic features on the surface of pebble #100. The groove is particularly deep (i, k) and technological errors are evident. Linear polish was also observed within the groove (m, n). f-h) Microscopic features on the surface of pebble #103. o,q) Details of the groove of pebble #103.

https://doi.org/10.1371/journal.pone.0257710.g012

Functional analysis of the small grooved pebbles

Results.

Other than limestone pebble #105 which was chemically altered by post-depositional erosion (Fig 12E and 12F), the surfaces of the small grooved pebbles were well preserved (Table 3). As is typical, the identification of use-wear traces on natural, unmodified stone surfaces was challenging [119].

Technological traces are divided into tool preparation marks (e.g., incision of the groove) and manufacturing errors. Grooves vary in depth and follow the longitudinal (pebbles #100, 101 and 103) or vertical axis (pebbles #104, 105 and 106; Fig 12). Variation in the depth of grooves on the superior and inferior faces is sometimes observed on the same implement (Fig 12B, 12C, 12H and 12I). The depth of the groove is not correlated with the raw material, as a range of depths values were observed on both limestone and basalt. The grooves were produced by the sharp edges of stone tools, as suggested by their irregular outlines and the variation in depth (Fig 12E, 12F, 12K and 12L). Technological errors, such as narrow, shallow micro-grooves oriented parallel to the main groove, are common (Fig 12K, 12L, 12P and 12Q). These were likely created when the stone tool slid out of the main groove during manufacture.

The studied pebbles show limited use-wear traces. This may be related to their short duration of use and/or the nature of their use. For example, if the small grooved pebbles were used only briefly as weights connected to a line of soft material, then only minimal traces are expected. In addition, one limestone specimen was subjected to post-depositional chemical alteration of its surface, making observation challenging (Fig 12A).

The only wear connected to use are lines of polish located inside and parallel to the groove (Figs 12M, 12N and 13H) on two of the basalt pebbles (#100 and 106; Figs 12J and 13F). No polish was observed on the limestone specimens, which are also slightly smaller than the basalt pebbles This pattern may be related to the nature of the raw material itself. Another study [120] has shown that polish forms much more slowly on limestone than on chert and usually presents a much less shiny surface.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 13. Some residues observed on small pebbles #103, 105 and 106.

a) dark spots, possibly oxides, on pebble #105; b) a shell fragment on pebble #106; c-d) smashed shell powder mixed with sediment on pebble #103; e) pebble #106; f) pebble #103; g) polished area with randomly oriented striations; h) linear bands of polish located in the interior of the groove; i-j) fragments of vegetal fibers under plane polarized and crossed polarized light from pebble #103. Scale bars measure 100 μm.

https://doi.org/10.1371/journal.pone.0257710.g013

Methods

The preliminary zooarchaeological data presented here includes the identified fauna from the 2016 excavation season (NISP = 413). The 2016 assemblage includes fauna from Layers 3–0, 3a, 3b, 3c, 4 and 5 and thus the samples for each individual layer are small (NISP = 11, 145, 49, 15 119, and 74 respectively). Although the hooks and grooved stones are found only in the Natufian Layers 3a 3b & 3c, we present data from all of the site’s layers to highlight differences in the importance and composition of the assemblage that may be related to technological change.

This report focuses exclusively on the fish remains. Importantly, although the fish bones from the sample have been recorded in the database, only general preliminary identifications have been made to date. The three fish families identified thus far (Cyprinidae, Cichlidae and probable Salmonidae) can be distinguished based on the morphology of their vertebrae, teeth and head elements. The identification of species within these families is more challenging and only possible for certain elements (e.g., axis and atlas, teeth). Although most vertebrae cannot be identified beyond the family level, the maximum diameter of the centrum provides a rough proxy for fish body-size, that can further narrow down the potential species. For example, using modern fish referents Zohar et. al. [49] determined that Cyprinid vertebrae centra with diameters wider than 3.5mm can belong to only to fish great than 220 mm in length and thus must belong to one of three large Cyprinid species (Luciobarbus longiceps, Casiobarbus canis, Capoeta damascina) inhabiting the Upper Jordan Valley.

Results

Because it was retrieved from water-logged deposits, the JRD fauna is in an excellent state of preservation. The sample includes fauna recovered from 2 mm mesh that was later washed and picked to recover small and delicate elements including tiny rodent and fish bones. The JRD fauna is comprised of diverse terrestrial and aquatic species represented by ungulates, carnivores and small game animals such as snakes, tortoises, turtles, birds, hares and fish. The ungulates that dominate Natufian assemblages at other sites are unusually rare at JRD, especially in Layer 3–0, 3a, 3b and 3c (<7%). Instead, in terms of raw frequency counts (number of identifiable specimens-NISP), fish is the most abundant taxon in every layer at the site. Nevertheless, there are fundamental differences in the relative abundance of fish across the layers, with the assemblages from Layers 3–0, 3a and 3b comprised of more than 80% fish, compared to only 27% fish in Layer 4 and more intermediate frequencies of fish in Layer 3c (55%) and 5 (45%; Fig 14A). Thus far, members of the Cyprinid (vertebrae, teeth and pharyngeal bones) and Cichlid families have been positively identified in the 2016 assemblage. Cyprinids are by far the most common family in all layers and include both large and small individuals. Teeth from the Cyprinid species Luciobarbus longiceps, Casiobarbus canis and Acanthobrama hulensis have been identified thus far. A few Salmonids (e.g., trout) may also be present (see also [121], but these identifications must be confirmed with an appropriate comparative assemblage.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 14.

a) Relative abundance of fish at JRD by arcaheological layer; b) The proportion of fish with vertebral centrum diameters greater and less than 3.5 mm. c) Frequency distribution of fish vertebral centra diameters in the combined Natufian layers (3a, 3b, 3c).

https://doi.org/10.1371/journal.pone.0257710.g014

The fish from the Natufian layers are larger in size than those from Layers 4 and 5. Between 45–65% of the fish vertebrae from Layers 3–0, 3a, 3b and 3c (52% for the combined Natufian layers) at JRD are larger than 3.5 mm in diameter, while only 25% in Layer 4 and 40% in Layer 5, are larger than 3.5 mm in diameter, (note that the sample sizes for the latter two layers are very small; Fig 14B). The combined distribution of vertebral centra diameters for the Natufian layers (3a, 3b, 3c) shows that they derive from fish with wide-ranging body-sizes (Fig 14C).

Features of the fish hooks

Fish hook assemblage diversity

Perhaps the most important aspect of the JRD fish hook assemblage is its substantial variability which is even more significant given the very old age of the assemblage. There are no two similar hooks at the site. Hooks vary in size, style, morphology, features and line attachment methods (Table 1) and this variation occurs both within and between the archaeological horizons. The process of hook manufacture at JRD indicates familiarity with the technological sequence, the application of different manufacturing techniques, the use of tools made of different materials, and different scales of work ranging from initial shaping to the production of very fine barbs and points. The variation in features and in technological solutions for the same need (i.e., the attachment of the line to the shaft) indicates a high level of technological sophistication, innovation and dexterity. Some of the morphological variability might reflect different individual skills within the community.

The size and shape of fish hooks and their features impacts their functionality (see above) and, consequently, the captured prey and the angling technique. In modern fishing, hooks are produced to target specific size and species of fish to maximize the fishing success. Thus, the variability in hook dimensions and functional features also suggests the intention to capture a range of sizes and species of fish with different behaviors. This reflects deep knowledge of fish behavior and suggests that the Natufian visitors to JRD intentionally targeted a broad spectrum of fish taxa with different sizes and behaviors supporting a region-wide trend toward broad exploitation of small-bodied prey at the end of the Epipalaeolithic [127–132].

Line and hook technology

Use wear and residue data from the grooved pebbles at JRD indicates that the pebbles were intentionally modified and then affixed to a line made of soft material of botanic origin. Manufacturing marks indicate that the grooves (clearly marked except in two cases), were made using stone tools. The presence of several polish marks inside the grooves of three artifacts confirms that lines were attached to the grooves. This is further supported by the discovery of two fragments of vegetal fibers on a sample extracted from the groove of one pebble. The lines were tied to the small stones which were probably used as weights. The small size and weight of these stones (<15 g) is within the range of weights typically used for line (<30 grams; [133] rather than net fishing.

The JRD layers are littered with small limestone and basalt pebbles suitable in size and shape to function as line weights, though most were not modified by grooving. Some of the limestone pebbles show possible evidence for limited grooving, but the grooves are equivocal. The importance of the line weights to fishing at JRD is reflected in the careful selection of pebble blanks of a particular shape and size and investment in the creation of the groove around their circumference to secure the attachment weight to the line.

Although larger unmodified stones that likely served as net sinkers are abundant in both pre-Natufian and Natufian layers at JRD (especially Layer 4), with the exception of one poorly made example from Layer 5, small grooved pebbles make their first appearance in an Early Natufian context. The simultaneous appearance of the first bone hook in the earliest Natufian layer at JRD, suggest that the two acted together to provide what was already a sophisticated line and hook technology. Line weights are a crucial part of line fishing equipment as when combined with floats, they enable the bait to be positioned at the desired location underwater. Floats, such as porcupine quills [134] are typically made of perishable materials and are not typically found in archeological context. During the later stages of the Natufian, line and hook fishing continued to be frequently practiced as evident from the presence of both grooved pebbles and bone fish hooks at JRD.

Fishing at JRD

Preliminary faunal data from JRD reveal changes in species composition and average fish body size that may be related to the appearance of line and hook technology in Natufian Layer 3c (Fig 2). In particular, fish are even more dominant in the Layer 3–0, 3a and 3c assemblages (Fig 2) than in the earlier layers at JRD or at any other Epipaleolithic site for that matter [49,55,132]. In addition, although samples are small, the proportion of larger fish in the assemblage increases in all of the Natufian assemblages (Layer 3a, 3b, 3c) in comparison to Layers 4 and 5. These changes suggest both that fishing was even more important at JRD during the Natufian than it already was in earlier layers and that hook and line technology enabled the targeted capture of larger fish than previously used fishing techniques. Nevertheless, small fish also remain abundant in all of the Natufian layers and given the location of the site next to the lakeshore and frequent inundations of the lake throughout history, it is difficult to distinguish how much of this is related to a change in the relative abundance of natural versus culturally deposited fish remains and how much can be ascribed to a change in human behavior. The difficult task of distinguishing natural from culturally deposited fish remains will require larger sample sizes, more comparative analyses, the study-of fish body-parts and ideally, the study of a natural assemblage originating close to the site (cf. [49]). A comparison of natural and cultural deposits from the site of Ohalo II located on the shore of the Sea of Galilee shows a significantly greater abundance of large Cyprinids in culturally deposited fish assemblages than those in a natural excavation 150 m from the site [49]. Thus, the greater proportion of large fish in the Natufian layers, may also suggest more significant human contribution to layer formation compared to earlier periods and thus more intensive activity at the site.

Change in fishing technology and fishing strategies over time

Even though it is the largest collection of fish hooks for its early date, the JRD sample is small for statistical study, particularly when only complete hooks are included, or the hooks are separated by layers. Therefore, we can only suggest trends in technological evolution and fishing strategies over time. First, and most importantly, the association between fish hooks, line fishing weights and fish taxa in the Natufian layers reveals that even though there is evidence for fishing throughout the Epipaleolithic, line and hook fishing does not begin until the Natufian period. All layers at JRD are rich with evidence for fishing including fish bones, notched limestone pebbles identified as net sinkers, basalt cobbles of similar size that may have served the same purpose and a unique flint tool assemblage rich in burins and scrapers, that may also have been used for fishing and fish processing activities. Despite this, apart from the featureless hook #13, which deviates from all other hooks in its size and morphology (Figs 3 and 4; Table 2), all hooks originate from the two later Natufian layers of the site (Layers 3a and 3b) or the contact between Layer 3a and the Early Neolithic Layer 3–0. This pattern is supported by other south Levantine sites where fish hooks do not appear until the Natufian. Importantly, advanced hook technology and small weights become common at JRD only in Natufian Layer 3b. The assemblage features the first barbs and line attachment features in the region, although many of these disappear in later layers at the site. Finally, the decline in average maximum length of the fish hooks across the Natufian layers may relate to the targeting of smaller-bodied fish or a less visible shift in other fishing technologies used at the site (i.e., traps may have been used for catching larger bodied fish). Nevertheless, the size decline does not agree with the preliminary faunal analysis where fish become larger over time. Although large hooks are not very effective for capturing small fish, small hooks are capable of capturing a wide variety of fish sizes. Hook size may also relate more to the type of bait or lure chosen and the fishing conditions (smaller hooks are less likely to get tangled in vegetation and rocks) to target a particular type of fish. These hypotheses require larger samples of fish hooks and fish bones for validation.

The appearance of line and hook technology at JRD and in the Natufian more generally, coincides with evolutionarily significant changes in human diets and associated technologies that typify the end of Epipaleolithic in this part of the world [127,128,130,132,135]. Although diets gradually broadened from the Middle Paleolithic onward in the southern Levant, this trend accelerates in the Epipaleolithic and peaks in the Natufian and following Pre-Pottery Neolithic A periods as people became more tethered to the landscape, populations grew, and local large game resources became more exhausted [127,129,136]. Not surprisingly, a parallel trend occurs in prey acquisition technology. Evidence for technology (bow and arrow, nets, traps, and hook and line technology) suitable for overcoming prey with variable characteristics also diversifies in the Epipaleolithic [4,37,52,54,57,60]. Although many of these new technologies, like the line and hook, were more complex and costly to manufacture and maintain, they ultimately lowered the cost of the acquisition of smaller-bodied, difficult to catch species like the fish from JRD. Thus, it became more efficient to forage for smaller-bodied taxa which enabled humans to better face the challenges of sedentism. It is quite expected that the diversification of prey technologies would occur hand in hand with similar changes in animal prey as humans became more proficient at exploiting new niches and prey types [137].

Conclusion

The recovery of twenty complete and fragmentary bone fish hooks and a collection of grooved stones at JRD enabled a detailed study of their manufacture and use based on technological, use wear and residue analyses. The analyses illuminated several groundbreaking innovations in fishing technology that were not previously visible at such an early date. These include the use of inner and outer barbs, artificial baits (lures), diverse line attachment features including adhesives and stratigraphic associations between fish hooks and grooved stones. These many innovations, the variety of combinations in which they appear, and the multiple steps required to manufacture the many components (weights, hooks, lines, adhesives and lures) and then assemble them into an integrated line and hook technology attests to the emergence of complex fishing technology by the Late Epipaleolithic in which many features of modern line fishing had already appeared. This sophisticated technology attests to deep knowledge of fish behavior, ecology and acquisition strategies and fits into a larger pattern of technological and resource diversification at the end of the Pleistocene in the Levant immediately preceding the Neolithic period.