ARCHAEOLOGY AT RAS MUARI: SONARI, A BRONZE AGE FISHER-GATHERERS SETTLEMENT AT THE HAB RIVER MOUTH (KARACHI, PAKISTAN)

This paper describes the results of the surveys carried out along Ras Muari (Cape Monze, Karachi, Sindh) by the Italian Archaeological Mission in Lower Sindh and Las Bela in 2013 and 2014. The surveyed area coincides with part of the mythical land of the Ichthyophagoi, mentioned by the classical chroniclers. Many archaeological sites, mainly scatters and spots of fragmented marine and mangrove shells, were discovered and AMS dated along the northern part of the cape facing the Hab River mouth. The surveys have shown that fisher and shell gatherer communities temporarily settled in different parts of the headland. They began to exploit the sea resources during the Neolithic. However, the most important discovery consists of a unique fishers’ settlement with rectangular stone-walled structures located on a limestone terrace near Sonari (SNR-1), the first ever found along the northern coast of the Arabian Sea. The AMS dates show that it was settled mainly during the first half of the third millennium cal bc when the Indus Civilisation flourished in the area. Considering the importance of the discovery, all the material culture remains from the Sonari sites have been described and analysed in detail and, whenever possible, framed into the different phases of environmental changes and human adaptation to the coastal environment that have been interpreted thanks to a good series of AMS dates from marine and mangrove shells.


Paolo Biagi
This paper presents and discusses the research carried out in  and  by the Italian Archaeological Mission in Sindh and Las Bela (Balochistan, Pakistan), part of an archaeological project conducted in the region since . More precisely it describes the results of the surveys conducted along the northern side of Ras Muari (Cape Monze, Karachi) between the present village of Sonari at the Hab mouth, in the north, and the Baloch coastal settlement of Mubarak, a few kilometres north of the cape, in the south ( fig ).  This territory had never been studied in detail before the s. Previous surveys were carried out by Professor A R Khan of the Institute of Geography, Karachi University, during the second half of the s. His surveys led to the discovery of many archaeological sites of different ages, from the Palaeolithic to the Buddhist and Muslim periods,  among which is the prehistoric settlement of Sonari (SNR-). Sonari is the only Bronze Age fisher-gatherer village ever discovered along the northern coast of the Arabian Sea with rectangular stone structures and net sinkers. The site is so far unique, and very important for the study of the Bronze Age of a region located at the south-western edge of the spread of the Indus Civilisation, where different archaeological aspects from the Indian subcontinent, the Arabian/Persian Gulf and the Gulf of Oman interacted.  The scope of the  and  research was also to fill a chronological gap in the archaeology of the territory, which extends between Lake Siranda in the west and the Indus Delta in the east, by means of a systematic programme of accelerated mass spectrometry (AMS) dates. To date, more than  radiocarbon dates have been obtained from samples collected from the coastal region, mainly from shell middens and scatters of mangrove and marine shells recovered during the surveys conducted in the last twenty years.  Our surveys were conducted on foot in August  and January  by two of the present authors (PB and RN). They were repeated systematically four different times following the same track: starting from the Hab mouth, where the present fishers' village of Sonari is located, walking from north-east to south-west across small fans and slope debris along the ridge that runs at the northern edge of the alluvial plain where the saddle inside which the prehistoric settlement of Sonari SNR- is located, and then south-west towards the village of Mubarak and the Arabian Sea coast.
The surveys led to the discovery of many archaeological sites, most of which consist of concentrations of shell fragments lying at the top of wind-borne sediments and distributed over an area of c .km  at altitude -m asl. The presence of scatters of marine and mangrove shells had already been reported by W T Blanford in the late s  and O Schmieder in the s.  It is important to note that the recorded shell features cannot be due to natural factorsthey must be attributed to the human exploitation of past marine and mangrove environments during different ages. In fact, most shells have been intentionally fragmented by hard hammering to extract meat in a way that cannot be due to natural causes or animal predation. Moreover, the shell scatters consist of circular/oval concentrations, never alignments as would be the case if they had been deposited by sea wave/tide action.
Each archaeological site, irrespective of its age, was recorded by global positioning system (GPS) device and precisely positioned with the help of Soviet military maps and high magnification satellite photographs. One specimen of marine and mangrove shell species was collected from each site for identification. Whenever available, one single adult specimen of Terebralia palustris or Telescopium telescopium gastropod was selected for dating and defining where mangroves flourished around the cape. If mangrove shells were not present coastline, and c .km south of the seasonal stream along which Mubarak village is located, a continuous horizontal layer of Lunella coronata marine shells was observed along natural erosion of a stream at c m of depth from the topsoil and m of altitude. One shell specimen was collected and dated to ± cal BP (RMR-: GrA-) (fig , top). This is one of the most recent radiocarbon dates obtained from our Las Bela/Indus Delta survey. It deserves further investigation because of the absence of archaeological sites of this period in the cape that makes the presence of this shell horizon intriguing. Increasing evidence of eustatic/tectonic changes along the sea side that might have affected archaeological sites' visibility even in historic times has been demonstrated for the coast east of the Indus Delta, Gujarat in particular.  Apart from unique geomorphologic characteristics, archaeological aspects should also be considered in their settings, as documented by G F Dales.  These include uplifted coastal and riverine Indus Civilisation ports such as Sutkagen-dor and Sotka-koh, at present located several kilometres from the coastline in the Gwadar-Ormara region of Makran, and Mai Gondrani Buddhist caves that open in vertical walls of faults in Las Bela Valley.  These geological conditions (crustal instability and strong uplift) should be taken into account when considering the changing past relationships between the archaeological sites of the area and the Hab mouth mangrove forest. Affecting Hab water discharge, together with climatic change, these movements might have caused the decline of mangrove environments and consequently the availability of its food and wood resources, until the very recent disappearance of this unique ecosystem (GrA-: ± BP on T telescopium).

THE  AND  SURVEYS Renato Nisbet and Paolo Biagi
The main scopes of the  and  surveys were to record and reconstruct the history of human presence and interference in an almost unknown territory of coastal Sindh and Las Bela. Therefore, the following strategy has been adopted: ) record and precisely locate each archaeological site by GPS device;  ) radiocarbon date the periods of human settlement by means of mangrove and marine shells recovered from the sites; ) define the presence and death of mangroves at the Hab mouth and their different periods of human exploitation; and ) study the Bronze Age fisher-gatherers village of Sonari (SNR-) and its eventually related archaeological evidences. The discovery of SNR- is very important since so far we do not have any evidence of other prehistoric fisher settlements along the northern coast of the Arabian Sea, despite great emphasis given to their presence in different periods.  The territory discussed in this paper is part of the mythical land of the Ichthyophagoi (fish-eaters) described by ancient Greek and Roman authors,  according to whom Las Bela was settled by groups of Oritae. However, we have scarce information regarding the locations and characteristics of their villages  and their eventual relationship with other sites of the interior. This absence of evidence contrasts with that of the many Indus Civilisation and later port towns known from neighbouring Gujarat in India.  Moreover, the coast west of Karachi is considered to be one of the important north Arabian Sea sources from which large oceanic shells were exploited and traded to supply the artisan workshops of the Indus cities.  Our surveys led to the discovery of many archaeological sites. They mainly consist of man-made heaps or scatters/spots of decoloured, fragmented marine and mangrove shells, samples of which were AMS dated from most sites (fig , bottom; tables  and ).
The most recent AMS date (± BP: GrA-) was obtained from a small scatter of Meretrix sp, Cyprea sp and T telescopium fragments (SNR- on T telescopium, see table ). The result shows that a mangrove environment was still flourishing around the Hab's mouth during the seventeenth to nineteenth centuries AD, though we do not know exactly when it began to disappear.
Close to site SNR-, along the foot of the ridge, five Islamic tombs were discovered c m from the south-westernmost limit reached by the Hab's floods (SNR-: fig , top). The tombs are roughly rectangular, except for one that is almost oval and oriented east-west. They measure c . × .m, and are delimited by medium-sized, limestone blocks (fig , bottom). They were numbered from A to E. Inside and around them a few grave goods were recovered. They consist of two conjoining sherds of one small Chinese porcelain cup painted in blue, probably attributable to the sixteenth century AD,  from graves  (table ).
A scatter composed exclusively of fragments of marine and mangrove shells was found c m south-west of the graveyard (SNR-). A few pieces of Turbinella pyrum, a large oceanic shell often exploited in prehistoric and historic times for the manufacture of different types of artefacts,  were collected from its surface. The site was dated to ± BP (GrA- on T palustris).
SNR- was discovered c m west of SNR- on the lower part of the slope that takes to the saddle inside which the settlement of SNR- was found (fig , top). The site consists of irregular spots of marine and mangrove shell fragments scattered over a surface c m in diameter. Two different dates suggest that the site was settled during different periods. In fact, one Lunella coronata specimen yielded the result of ± BP (GrA-), while one T palustris gastropod was dated to ± BP (GrA-). The assays show that while part of the midden accumulated around the end of the Bronze Age, a mangrove environment was still flourishing in the surroundings during the third to fifth centuries AD.
Another scatter of T telescopium and T palustris fragments recorded c m southsouthwest of SNR-, along the edge of the same ridge (SNR-), was dated to ± BP (GrA-). It shows that a mangal environment was present in the area also during the Bronze Age.
Moving farther south-west, five more shell spots were discovered. Site SNR- is an eroded midden located c m from SNR- (fig , bottom)  of marine and mangrove shells, dated to ± BP (GrA- on T palustris). Some m west of the site, a heap of Lunella coronata shells (SNR-), c .m in diameter, was dated to ± BP (GrA-). It was found close to another scatter of Lunella coronata and Meretrix shells (SNR-). SNR- is a spot of Turbo bruneus marine shells found on the surface of an eroded terrace. It was dated to ± BP (GrA-).
SNR- is a heap of marine gastropods, c .m in diameter, while SNR- is another small spot of Meretrix marine shells dated to ± BP (GrA-).
Other sites were recorded inside the saddle, the most important of which is the prehistoric fisher-gatherer village of SNR- discovered by A R Khan in the late s ( fig ). The settlement is a C-shaped feature made of collapsed limestone blocks well-sheltered inside the saddle that makes it invisible from both the sea and the Hab plain ( fig ). The site is c m long (north-south) and m wide (east-west). Its north, concave, side faces the Arabian Sea.
SNR- is composed of at least six rectangular rooms (or cabins) of different size delimited by limestone blocks. Their floors are covered with Meretrix and a few other shells that were sampled for radiocarbon dating (fig ; tables  and ). Most rooms are oriented east-west, only one is north-south (SNR-B). The rooms SNR-A, SNR-A and SNR-A are adjacent to each other. The larger feature SNR-A was built west of SNR-A and SNR-A, north of the previous one following the same orientation. Most probably, more rooms are buried in the eastern part of the village below the stone rubble.
Just in front of the concave side of the settlement, three shell heaps were recorded of different size and thickness, made of fragments of mangrove and marine shells, called SNR-AC, AC and AC (table ). Regarding these features, it is important to note that around the end of the eighteenth century Lieutenant R Hughes-Buller reported the presence of small shell heaps along the coast of Makran, related to the production of lime from shells, whose scope was to lime cotton nets.  This process is accurately described by M I Siddiqi in his article on the fishers of West Pakistan, in which he points out that in Las Bela lime was obtained exclusively from T telescopium shells.  Therefore, we suggest that the SNR- shell heaps might be related to a similar activity that took place mostly during the Bronze Age.
Two more points were taken by GPS within site SNR-. The first marks the presence of a microlithic backed point (SNR-) surrounded by fragments of T telescopium shells, one specimen of which was dated to ± BP (GrA-). The second is a bladelet core made of dark reddish brown Gadani chert (SNR-) ( fig ).
Telescopium telescopium (Horn snail) GrA- ± AD-: : : AD-: : :  .  Simple, double-notched flat pebbles of various sizes are the commonest stone implement associated with fishing.  Their size variability has been related to different fishing techniques. The general impression is that small sized items and weight are associated with casting or 'beach seine net', while specimens of a larger size and weight belong to gill nets or large 'seine nets'. However, this simplistic subdivision makes problematic the attribution of medium size and weight specimens.  Net sinkers, obtained from flat oval pebbles notched roughly in the middle of their long sides, are known from many Omani coastal sites of the Saruq and Bandar Jissah facies, dated between c  and  BP,  though they are known also from more recent Bronze Age sites.  Their form and weight variability is believed to show some chronological significance,  though some authors suggest that their different typology might derive from certain local traditions.   (table ). Most artefacts are less than c g, ie lighter than those reported from the Arabian/Persian and Oman Gulfs, Bronze Age Umm an-Nar period.  They can be compared with a group of medium-sized sinkers from the Middle Holocene coastal site of Ra's al-Khabbah (KHB-) in Oman.  The Sonari net sinkers consist of flat pebbles with two opposed notches knapped in the middle of the long sides, around which a string can be firmly tied. These objects are often reported in the literature as 'net weights', 'notched pebbles', 'fishing weights' or 'notched weights'. Indeed, the blanks have been carefully selected among the limestone beach pebbles that abound along the shores of Cape Monze, the sole modification being represented by the two opposed, bifacial notches knapped by hard hammering.
The size variability of the Sonari net sinkers can be explained either as consequent to different, albeit contemporaneous, fishing strategies adopted by the same community or as a chronological proxy implying multiple frequentations of the site. Therefore, it is necessary to examine their significance in better detail.
These stone objects were illustrated and described for the first time by C Rau towards the end of the nineteenth century in his book devoted to prehistoric fishing, where he // Very weathered pebble flake, retouch traces, side scraper?
Oval pebble flake with centripetal flaking on one face, net sinker?  , no. ). The ethnographic parallels provided by Rau are an undisputable proof of the function of these stone artefacts. The same author noticed another characteristic of the stone sinkers, ie their size variability, most probably related to their different use and fishing techniques.  H-P Uerpmann and M Uerpmann, describing stone implements from the prehistoric fishers' sites of Oman,  distinguished two basic types according to their manufacture technology. They called them N-type ('Normal', ie made from oval pebbles notched on their longer sides) and R-type ('Retouched', ie with a retouch shaping the whole outline of the pebble on both faces and notches) respectively, assuming that their difference is related to two different functions and fishing techniques. All the Sonari specimens, regardless of their size, fit into the N-type category, except for one atypical item (fig , no. ).
Similar artefacts are known from the Neolithic site of Sha'ar Hagolan, and many other sites in Israel, dated from the Upper Palaeolithic to the Early Bronze Age.  Here, stone net sinkers in the form of notched pebbles have been related to freshwater fishing with 'throwing nets' or fishing lines or rods. This fishing gear was apparently first used by Upper Palaeolithic and Epipalaeolithic hunters and gatherers, though their use was continued by sedentary Neolithic communities.  Moreover, it has been suggested that differences in the form of weights, size and the location and size of the notches may indicate different fishing methods, probably related to tying. Their occurrence within archaeological sites is considered to be consequent to the discard of the net.  According to Siddiqi,  who studied the fishers' settlements of the south-western coast of Pakistan in the s, 'the most promising fishing grounds are situated at the mouths of the rivers, where the small fish congregate and attract the larger ones'. The same author reports that stones equipped the gears at the fishers' settlements of Makran and Las Bela,  although he did not provide us with any description of the net weights.
Moreover, stone net sinkers are part of the material culture of the Jōmon huntergatherers of prehistoric Japan,  who were also skilled fishers since the Initial Jōmon period (ninth to seventh millennia BP). Ichthyologic studies have shown that the Jōmon fishers captured salmon and trout running upstream during the autumn, carp in the freshwater zone, black sea bream and sea bass in the estuarine water and, most interestingly, bonito and tuna in the off-shore zone.  Y Kondo provided a very interesting spatial analysis of the archaeological contexts (house, midden, pit, etc.) where Jōmon net sinkers have been recorded,  discovering that disposal loci can be classified into settlement (on-site) and non-settlement (off-site). The same author observed a significant correlation between fishing potential and standardised weight of sinkers,  interpreting the variability of sinkers in function of fishing conditions such as water depth, target fish and expected fish catch. The variability of net fishing strategies is explained as a means to effectively exploit the aquatic resources available close to settlements in accordance with their population size and location.  Therefore, we can confidently interpret our net weight assemblage from Sonari as evidence of fishing most probably related to different fish catches.
Other Sonari stone objects consist of several pebble flakes with retouched sides or edges, resembling large scrapers (fig , no. ; fig , nos , ,  and ). They have never been reported or described before from any other Pakistani Arabian Sea site, except for one limestone specimen from Daun- shell midden (Las Bela), AMS dated to the midseventh millennium BP.  These tools resemble the so-called 'heavy duty scraping tools' that are known from many Late Stone Age sites of Oman whose subsistence economy was based on the exploitation of marine and mangrove resources.  Sonari also yielded some pebbles with traces of hammering and pitting in the form of shallow, rounded cup marks ( fig , nos  and ; fig ), These tools have often been described as anvil stones and hammers, and have been retrieved from shell middens of both the Arabian Sea coasts.  Pebbles with pecked round grooves roughly in the centre of both surfaces from the Omani shell middens have been interpreted as crushing stones  for breaking the hard shell of mangrove gastropods to extract their meat.  Finally, there are two stone tools on cobbles that can be interpreted as lower parts of grinding implements. They are made from sandstone and present one worn, concave working platform (fig , no. ).
To sum up, the small assemblage of stone tools described above can be attributed to the activities of a community of fisher-gatherers that exploited sea and mangrove environments. On the basis of the typology of the associated potsherds and the radiocarbon dates it can be attributed to the Bronze Age (figs  and ).

THE KNAPPED STONE ASSEMBLAGE Paolo Biagi
A few knapped stone artefacts were collected from the sites located on the Sonari terrace. Most specimens come from a spot called SNR-bis, c m east of SNR- (table )    The knapped stone assemblage from Sonari consists of very few artefacts. However, some considerations can be made on their typology, function and the raw material exploited to produce them. Most specimens are made from Gadani dark reddish brown chert,  though Rohri and Las Bela chert artefacts are also present. Gadani chert is easy to identify. It was exploited for making tools between the beginning of the Holocene  and most probably the Indus period, as is shown by a radiocarbon date obtained from Ras Gadani (GrN-, see above). However, during this latter period the exploitation of Gadani chert seems to have partly ceased and been substituted by that of exogenous, better quality raw material extracted from the chert mines of Lower and Upper Sindh.  As far as we know, the present distribution radius of Gadani artefacts  extends roughly from the Mulri Hills at the eastern outskirts of Karachi, in the east, to Lake Siranda (Las Bela), in the north-west.  The presence of two cores, from SNR- (fig , no. ) and SNR- (fig , no. ) respectively, shows that at least some of the artefacts were produced on the spot. The backed point of no.  in fig  is atypical for a Bronze Age lithic assemblage, and might be better related to an earlier occupation, most probably of the seventh millennium BP. This attribution is suggested by the recovery of one comparable tool type, made from the same raw material, from the impressive Lake Siranda shell midden SRN-, whose earliest occupation is radiocarbon dated between ± BP (GrM-)  and ± BP (GrA-), both on T palustris.  However, we have to consider that we know very little of the typology and technology of the knapped stone assemblages of the fisher-gatherers of this period, and most Gadani chert artefacts from the Sonari sites are indeed microlithic specimens. The presence of one regular bladelet of non-local chert from the surface of one of the SNR- structures is also important. Most probably it comes from Lower or Upper Sindh sources, among which are the Rohri Hills, Ongar and Jhimpir,  that undoubtedly supplied the Bronze Age Indus sites located around Sonari, including Pir Shah Jurio  and Balakot. 


The general impression, based on the study of the knapped stone artefacts at present underway from the Lake Siranda sites in Las Bela,  the Tharro Hills and Kot Raja Manjera, in Lower Sind,  suggests that important technological changes took place at the turn of the Chalcolithic in the study region, when pressure technique was introduced for the manufacture of regular blades and bladelets and new, exotic raw materials started to be exploited for the first time during the Amri Culture period. 

THE POTTERY Michela Spataro
A few small potsherds come from the surface of three radiocarbon-dated sites (SNR-, SNR- and SNR-: tables  and ; fig , bottom), two of which from structure SNR-A ( fig , nos  and ). Seventeen ceramic potsherds from a maximum of sixteen vessels (table ) and three porcelain samples were analysed.
All samples were sent for polished thin sectioning. The thin sections were analysed with a Zeiss Axiophot polarised microscope and by variable pressure scanning electron microscopy-energy dispersive X-ray spectrometry (SEM-EDX).  EDX analyses were carried out on four different areas of each sample at × magnification (each covering an area of c . × .mm). The SEM was used at a pressure of Pa with a kV accelerating voltage; the samples were analysed uncoated at a -mm working distance. Ten elements (Na, Mg, Al, Si, K, Ca, Ti, Mn, Fe and Pb) were detected and measured. The results were converted into oxide percentages, which were normalised (oxygen by stoichiometry) to take into account the fact that oxygen and carbon are not measured.
A variety of recipes and raw materials were used to make the seventeen ceramic samples, which were divided into eleven fabric groups (see table  for detailed descriptions).
No fabric group appears at more than one site.
There are therefore significant differences between the ceramic fabrics, as expected from pottery made in different periods. Unfortunately, on the basis of the fragmented status and the weathered surfaces of the sherds, it is not possible to trace correlations between shape, surface treatments and paste used to make the ceramics. In some instances, ceramics were not tempered (Fabrics ,  and , eg fig c), while others were probably Grog was added to a couple of Bronze Age sherds (SNR and ) and to a body sherd with grooved decoration (SNR). Grog inclusions varied, some were calcareous, others were not (eg in sample SNR), some were more highly-fired than the matrix of the sherds to which they were added, some included scattered coarse and fine quartz. Sample SNR included second generation grog (grog in grog; see fig f),       suggesting that this technical choice was an ongoing tradition. In prehistory, the use of a specific temper type is often linked to cultural tradition rather than for functional reasons, so the presence of a grog fragment within another grog fragment suggests a tradition of grog-tempering within the same cultural group. The only levigated pot (SNR, Fabric ) is a body sherd (fig , no. ), which was also red-painted (fig d) with a paint made by mixing a fine clay and iron oxides (the paint is not visible to the naked eye).
The sands inclusions in the potsherd fabrics, some of which were deliberately added to the clay and others were naturally occurring in the raw materials, represent a variety of minerals that come from metamorphic and igneous settings. As they are fine-grained, they might have been transported over long distances from the original rock outcrops. The Sonari sites are located in proximity of the Hab mouth, whose lower reaches cross Alveolina limestone. The local geology also includes Nari and Gáj formations and Kojak shales.  The Nari (Oligocene to early Miocene) and Gáj (Miocene) formations are consolidated rocks that include igneous rocks (altered basalt, andesite, diorite and granite); metamorphism is also recorded.  Amphibole and the igneous inclusions identified in the potsherds are too fine to be attributed to a specific rock formation; in addition, amphibole occurs in different igneous rock formations (eg granite, andesite, diorite, etc.). On these bases, the potsherds could be of local origin, but the geology is also homogenous for a long stretch of the Hab.  The chemical compositions of the sherds confirm the mineralogical groupings (table ). For example, four body sherds (samples SNR, ,  and ) were found at the same spot (SNR-), and they were attributed to three fabrics (table ): they have three different chemical compositions, as seen in principal components analysis (PCA) (fig ). Other sherds found together have a very similar chemical composition, suggesting that they were made from the same raw materials. The samples of Fabric  (SNR,  and ) could come from the same pot ( fig ), as suggested by the mineralogical analysis. Two pots (SNR and ), which were made with a clay richer in muscovite and biotite micas than the other pots, are also chemically different from the others. Although mineralogically different, the dark bluish/grey vessel (SNR), is chemically similar to samples attributed to Fabric , found at SNR-.
Finally, the Chinese porcelain sherds are very different from each other; although they have both a fine paste, the small blue and white bowl (SNR, ) has a paste rich in fine quartz with recurrent voids. The glaze is .-.mm thick and lime-alkaline (fig g). The white porcelain fragment (SNR) has a finer paste, with few quartz inclusions, occasional iron oxides and no voids. It was glazed with a fine, c .mm thick, lead glaze (fig h).

DISCUSSION
The surveys carried out by the Italian Archaeological Mission during the last twenty years in the territory stretching from Sonmiani and Siranda lagoons in Las Bela to the Indus Delta, in Lower Sindh, led to the discovery of many archaeological sites, most of which  Table 7. Sonari: SEM-EDX compositional results of the fabrics of the Sonari sherds. Average of four bulk analyses in the grey rows, and standard deviation in the white rows; s.d. = standard deviation; -= below detection limit (analysis by M Spataro).
consist of shell middens and scatters/spots of mangrove and marine shells that have been systematically recorded and sampled for radiocarbon dating. The project led to the first reconstruction of the time, pace and settlement pattern related mainly to the human exploitation of mangal and marine resources along this part of the north Arabian Sea coastline, an almost terra incognita from both archaeological and palaeoenvironmental points of view.
At present we have at our disposal more than  radiocarbon dates, most of which are from T palustris and T telescopium samples.  They show that mangroves flourished during the Holocene in virtually every estuary of freshwater courses in this part of the north Arabian Sea coast whenever suitable environmental conditions were available in term of salinity and temperature, attracting human groups that exploited their resources. Their late Holocene and sub-recent disappearance has been observed along the coasts of Sindh, Balochistan and the Persian/Arabian Gulf.  This fact is connected with the fractionation and later disappearance of the Bronze Age Indus Civilisation that was caused by monsoon weakening and the establishment of arid conditions that started to prevail around the end of the third millennium cal BC. 


The earliest indication of the presence of mangrove swamps comes from the Mulri Hills (MH-: GrA-: ± BP on T palustris), a low limestone range located at the eastern outskirts of Karachi, at that time under direct influence of the Malir and some Indus channels, while a slightly more recent date comes from a shell midden discovered along the eastern shore of Lake Siranda (Las Bela) (SRN-: GrA-: ± BP on T palustris). Both these dates are from single adult fragments of T palustris shell. However, it is important to remark that a piece of marine bivalve from KDJ-, a site located along the southern bank of the Kadeji River at its confluence with the Mol, yielded a result c  years earlier (KDJ-: GrA-: ± BP). Though the date comes from a marine shell, its negative δ  C value of −. shows that it grew in a mangrove environment, suggesting that mangrove swamps were already flourishing at the Malir mouth, and more broadly the Karachi Gulf, around the end of the ninth millennium BP.  The earliest evidence of human exploitation of the coastal resources at Sonari comes from two small scatters of Meretrix bivalves, radiocarbon dated to the second half of the seventh millennium BP (SNR-: GrA: : ± BP, and SNR-: GrA-: ± BP). These results raise the question of the time mangroves made their first appearance around the Hab mouth ( fig ), given that the oldest T palustris dates available from the area fall between the middle of the fifth and the end of the fourth millennium BP (from SNR- to SNR-W: see table ). Moreover, the results show that mangroves were still present around Sonari when they were already disappearing from the coast of Las Bela, as suggested by the assays obtained from Lake Siranda and the Bay of Daun.  The SNR- date matches well with that obtained from the Chalcolithic/Bronze Age mound of Balakot  (BLK-: GrA-: ± BP on T palustris),  along the course of the Winder, suggesting the existence of a mangrove swamp exploited by its inhabitants most probably along the eastern coast of the Sonmiani Lagoon. Two slightly later dates were obtained from the small Indus Civilisation settlement of Pir Shah Jurio ( fig ; table ). The site is located on a protruding terrace surrounded by alluvium and the traces of an old riverbed, along the left, eastern, bank of the Hab, c .km north-east of its mouth and km north of Sonari (fig , top).  Finally, two results from Sonari (SNR- and SNR-: see table ), one from Daun (Daun-: GrN-: ± BP on T palustris) and one from the northern shore of Lake Siranda (SRN-, GrA-: ± BP on T palustris) are the only historical dates so far available from Pakistani palaeo-mangroves.
As reported above, the Sonari sites yielded a small number of stone artefacts related to fishing. They consist almost exclusively of net sinkers, while fish hooks, harpoons and fish bones are missing. Their absence is most probably due to the very arid climatic conditions of the cape and salinity that have destroyed both metal artefacts and bones. However, the presence of notched stone net sinkers is very important. These artefacts are very rare in Las Bela. So far, Bronze Age specimens are known only from two sites discovered along the shores of the Bay of Daun, namely Daun  and Daun ,  where ten shell middens have been radiocarbon dated to the Mature Indus period, and two more to the end of the same civilisation.  They are not represented from the Chalcolithic and Bronze Age mound of Balakot, along the Winder course, in the northern part of the Kurkhera Plain, where fish bones and mangrove shells are numerous.  Stone net sinkers are also very rare from the shell middens discovered along the shores of the ancient Siranda lagoon, in which only four specimens have been recovered,  three of which from Neolithic sites and one dated to the beginning of the Chalcolithic (SRN-).  According to the available evidence, we can suggest that fishing was not the primary activity practised by Neolithic to Bronze Age groups that seasonally settled along the shores of Daun and Siranda. These sites show that their main subsistence economy was the collection of mangrove shells and, to a lesser extent, bivalves that live also in mangrove waters. Only at Sonari was a different pattern recognised. Here, fishing was undoubtedly practised as well as shellfish gathering. The location of the site and the material used for the constructions of the SNR- village stone structures are also important. The site is definitively well hidden inside a wide saddle c m from the present seashore. Moreover, the exposed features are made of local limestone blocks and not bricks, which are widely considered the commonest construction material used during the Indus Civilisation. Though of quite a different age, the only known parallels are the small, historical stone fishers' structures on the hills near Gwadar, along the coast of Makran. On their surface, Siddiqi reports the presence of potsherds, flints, grinding stones and lime fragments, as well as Greek and Bactrian coins associated with bones and shells. 


Our data would suggest that fishing started to be more intensively practised in the area during the Bronze Age. The Sonari net sinkers are all typologically the same, though their weight differs. They would suggest that coastal fishing took place during the winter months, at low-tide in shallow waters along a narrow zone of the rocky coastline of this part of the cape,  considering the great danger represented by the Arabian Sea waters, especially in the summer months when the seasonal monsoon blows from the south-east and fishing is not practised even today.
Finally, two conjoined fragments of ostrich egg from SNR- are also important to report.  This discovery is unique in the archaeology of the region, since we still know very little of the presence of ostriches along the Arabian Sea coasts during the Holocene, and our knowledge has not improved during the last thirty years.  To conclude, SNR- is the only Bronze Age fisher-gatherer site with stone structures so far discovered along the northern coast of the Arabian Sea.  It was settled mainly between the last centuries of the fourth and the first centuries of the third millennium cal BC; that is, between the end of the Chalcolithic and the beginning of the Bronze Age. One radiocarbon result obtained from structure SNR-D (GrA-: ± BP on T telescopium) shows that it was inhabited again a few centuries later. According to the radiocarbon chronology, SNR- is slightly older than the small Indus settlement of Pir Shah Jurio (PSH).
A ceramic bangle from the latter site is petrographically and chemically similar to the vessel found at SNR-,  suggesting a similar provenance ( fig ). However, most probably the two sites were settled in different periods and were not complementary to each other. This evidence raises interesting questions regarding the origins of SNR-'s inhabitants, the radius of their seasonal movements along the north Arabian Sea coast and, more broadly, about their systems of production, of which fishing, gathering, farming and probably hunting were part. 