Inter-generational loss of information relating to marine resource exploitation leads to shifting baselines (Pauly, 1995; Pinnegar and Engelhard, 2007), which have direct consequences for fisheries management and livelihoods opportunities. Historical data provide a means of regaining that information (Jackson et al., 2011; Cardinale et al., 2015; Fortibuoni et al., 2017) and are increasingly incorporated into assessments of change (Lotze and Milewski, 2004; Eddy et al., 2010; Gianelli and Defeo, 2017). Through improved knowledge of past environmental states and resource use dynamics, there is demonstrable evidence that historical data brings value to modern marine policy and management in both national and regional contexts (Jackson et al., 2011; Engelhard et al., 2016; Fortibuoni et al., 2017).

Despite this, historical data are not commonly incorporated into such frameworks for several reasons. Most notably, historical data are difficult to collect. Records may exist in a variety of languages, accessible only as physical documents, widely dispersed among archives and institutions, require special permissions to access, or occur in obfuscating formats (Lotze and Milewski, 2004; Rose et al., 2009; McClenachan et al., 2012). Furthermore, the challenge of standardising unfamiliar data, such as non-metric weights and pre-decimal currencies, to enable interpretation in modern contexts, presents a barrier to their use (Lotze and Milewski, 2004; Bainbridge and Hulme, 2014; Tesfamichael et al., 2014). Funding barriers also exist, which hinder government agencies in addressing these challenges (McClenachan et al., 2012).

The inherent challenges of accessing and analysing historical data are particularly germane for agencies concerned with fisheries management in the Pacific region (Gillett and Tauati, 2018). Historical data of relevance to fisheries were rarely published or disseminated and, where extant, records are less accessible from within the region than from outside (Flores, 1984; Blanchet, 1990). Lack of coherent information policies (SPC, 1988; Blanchet, 1990) coupled with poor conservation of public records (Bell, 2003; Rose et al., 2009) has generated an abundance of isolated reports which are now only available from repositories of formal colonial authorities and international agencies. Compiling existing information into a usable, inter-operable data format, supporting multi-disciplinary use, has long been a regional priority, particularly for fisheries supporting livelihoods (SPC, 1988; Blanchet, 1990; Halford et al., 2021).

For a substantial part of the Pacific region, livelihoods are partly or wholly dependent on fishing for food security and income generation (Dalzell et al., 1996; Gillett and Tauati, 2018; Andrew et al., 2019). While artisanal fishing for subsistence has occurred for millennia (Swadling, 1976, 1977; Szabó and Amesbury, 2011), escalation of exploitation for both subsistence and export markets has occurred in the last several decades or centuries (Dalzell et al., 1996; Gillett and Lightfoot, 2001). Since the introduction of colonial capitalist economies to the region (ca. 1800s) (Cariño and Monteforte, 2009), export-driven mother-of-pearl (MoP) fisheries, which target pearl oysters (Pinctada spp.), greensnail (Turbo marmoratus), and trochus (Rochia nilotica) for their nacreous shells, have made and continue to make important contributions to household earnings (Hawes et al., 2011; Purdy et al., 2017; Vieira et al., 2017; Gillett et al., 2020; Purcell et al., 2021). With past exploitation inevitably influencing the status of present-day populations of these commodities (Berzunza-Sanchez et al., 2013), resulting in local depletion in some cases (Chesher, 1980; Kelso, 1996; Kile, 2000; Pakoa et al., 2014), there is scope for historical data to provide valuable insight into the scale, nature, and timing of human influences on MoP fisheries.

To support the use of historical data by agencies concerned with managing MoP fisheries, we present a quantitative, standardised, and quality-validated dataset covering over 130 years (1888–2019) of MoP exports from a key producer in the Pacific region; Papua New Guinea (PNG). Both the size and economic importance of MoP fisheries in PNG (Purdy et al., 2017; Vieira et al., 2017; Gillett et al., 2020) have motivated the collaborative effort presented here, which aims to raise awareness and expand access to existing historical information of consequence to national and regional marine policy and management (Anon, 2017; SPC, 2019). Specifically, weight, value, and value tonne⁻¹ of MoP exports from PNG are presented as time-series from the onset of commercial fishing in 1888 to present (2019).

In working with historical data, two aspects of selection bias must be addressed: (1) the degree to which collected data represent the wider record collection from which records were drawn; (2) the degree to which the collected data reflect actual history (Inwood and Maxwell-Stewart, 2020). Independently constructed time-series for MoP fisheries of PNG and neighbouring Pacific nations allows quantitative assessment of such biases, permitting quality-validation of our dataset.

Four-decades ago, Glucksman and Lindholm (1982) constructed time-series for MoP exports from PNG between 1948 and 1976. Their data, from undisclosed sources, were very highly correlated (Pearson's distances: d_cor ≤ 0.05, r > 0.99, and p < 0.001) with our dataset for the period of overlap, showing near-identical trends and remarkably similar magnitudes (Supplementary Figure 1). Agreement between these two successive efforts demonstrates replicability and implies fairly homogenous data among records for this period.

In contrast, heterogeneous data among records for MoP exports from PNG between 1980 and 1993 is a known problem (Kailola, 1995). By presenting the mean ± 95% CI of collected data for this period, our dataset accurately reflects the variability among accessible records, thus, managing potential bias from record selection.

Comparison with the FAO Global Production database shows our dataset largely agrees with FAO's “best scientific estimates” (Garibaldi, 2012) of trochus and pearl oyster production for PNG (Figure 2). The reported weight of trochus production by FAO for periods 1953 to 1968 and 1985 to 2018 were highly correlated (Pearson's distance: d_cor0.55, r = 0.85, and p < 0.001) with the weight of trochus exports presented in our dataset. In comparison, the weight of pearl oyster production is reported by FAO for only a single period (1988–2018) and had a moderate correlation (Pearson's distance: d_cor 0.85, r = 0.64, and p < 0.001) with the weight of pearl oyster exports presented in our dataset. While disagreement with FAO estimates for missing data is understandable, the substantially greater weight of pearl oyster exports in 2011 and 2012 (Figure 2B) likely results from pearl oyster exports of aquaculture origin which should be excluded from the FAO data (Garibaldi, 2012).

To gauge whether our dataset accurately reflects the trends of MoP fisheries in the Pacific region, comparison was made to the Australian trochus fishery, which operated in the Great Barrier Reef region (Nash, 1985). The weight and value tonne⁻¹ of trochus production from the Australian fishery was reported, almost continuously, from 1912 to 1962; these data show similar (Pearson's distances: d_cor ≤ 1.05, r ≥ 0.45, and p ≤ 0.002) trends to those observed for MoP exports from PNG (Supplementary Figure 2).

The series of comparisons presented here, while not exhaustive, validate that our dataset provides a reliable indication of temporal change in MoP exports from PNG over many decades, consistent with past studies, international agency estimates, and regional trends in production and value tonne⁻¹. Since our dataset provides unprecedented coverage for MoP fisheries in the Pacific region, it is unreasonable to expect validation of each datum. By making the dataset freely available, however, we encourage further validation of our dataset as more information on MoP fisheries in the Pacific region becomes available.

Effective management of commercial fisheries is of salience to the future of coastal communities in the Pacific region, where a substantial part of the population is partly or wholly dependent on fishing for income generation (Dalzell et al., 1996; Sulu et al., 2015). With extensive reef systems and abundant MoP resources, PNG is regarded as a main regional exporter of MoP (Gillett et al., 2020), where the harvest and sale of MoP is estimated to support 20–30% of the coastal population (Purdy et al., 2017), accounting for as much as 75% of income in some areas (Vieira et al., 2017). Expanding coastal human population and increasing pressure on marine resources emphasise the need for accessible fisheries information for analysis and to support effective fisheries management in the region (SPC, 2019). Since little attention has been given to MoP fisheries in the last 25 years (Gillett et al., 2020) and historical information is important in guiding appropriate policy (Jackson et al., 2011; Engelhard et al., 2016; Fortibuoni et al., 2017), our dataset provides a basis for raised awareness and improved management of MoP fisheries within both national (Anon, 2017) and regional contexts (SPC, 2019; Gillett et al., 2020). Our dataset presents and facilitates access to MoP trade information, which can be used to estimate and analyse fisheries production, bioeconomic trends, shocks, and their potential drivers (Gephart et al., 2017; Gianelli and Defeo, 2017). The dataset presented here could further be evaluated against other economic and social variables as well as ecological information (Barausse et al., 2011; Papetti et al., 2013; Haimovici and Cardoso, 2016; Gianelli and Defeo, 2017) to inform development of broader marine policy and management strategies. Considerations for the use and interpretation of our dataset in addressing the above opportunities are outlined below.

A concern in the use of historical data for monitoring fisheries is the existence of a latent bias arising from improvements in fisheries reporting systems and regulatory systems. This “presentist bias” (Zeller and Pauly, 2018) has potentially serious consequences when assessing the status of fisheries or in interpreting resource use dynamics. In the case of MoP exports from PNG, procedures for collection of export data have remained largely unchanged since their inception, relying on export declarations validated through visual inspection. Prior to 1899, however, we acknowledge that presentist bias led to underestimation of weight and value of MoP exports from PNG. This is partly because records contain minimal economic data before 1899, for German New Guinea, when a private company functioned as both the administrative authority and an economic competitor. With this conflict of interest, an accurate disclosure of trade information from economic rivals could not be expected and, understandably, such information is scant (Sack and Clark, 1979). Additionally, in British New Guinea, it was not until 1894 that inspectors were instated and ports of export constituted for MoP fisheries (British New Guinea, 1896). Prior to this, “considerable yields of pearl-oyster” appropriated from PNG went unreported, but it is impossible to fix the amount (British New Guinea, 1888, 1892). Caution is therefore advised when drawing conclusions from our dataset for MoP exports during the 19th century, because available data provides a known underestimation.

For most nations in the Pacific region, data on MoP fisheries production is virtually non-existent because artisanal fishing activities of coastal communities are generally not monitored by government agencies (Gillett and Lightfoot, 2001; Govan, 2013; Zeller et al., 2014). In lieu of definitive production data, export data are commonly used by management agencies to estimate production of MoP fisheries (e.g., Lasi, 2010; Gillett et al., 2020). Our analysis (Quality-validation of the dataset section) confirmed that MoP exports from PNG were reflective of national (Figure 2) and regional (Supplementary Figure 2) production trends validating such application, with the following considerations for unreported harvest. First, it must be recognised that pearl oysters, greensnail, and trochus are also harvested for subsistence (as a protein source) (Glucksman and Lindholm, 1982) and, to a small extent, for the domestic shell trade (Simard et al., 2019). Second, the proportion of MoP production ultimately exported is further reduced by shells rejected as part of quality control (i.e., undersized, oversized, or damaged shell) (Kelso, 1996). Third, export tariffs have incentivised commercial operators to under-declare export volumes (SPC, 1997; Gillett et al., 2020). Fourth, exports do not account for illegal, unreported, and unregulated (IUU) harvests by foreign fishing vessels, with poaching of MoP from territorial waters a known issue for much of the period covered by our dataset (Bach, 1955; Christensen, 2016). The extent to which these four factors contributed to a disparity between production and export of MoP in PNG is not precisely known; unreported harvest between 1980 and 1995 was approximated at 25–30% of the exported volume (ICECON, 1997; SPC, 1997) and a correction of 25% is currently adopted at a regional level when adjusting export data to estimate production (Gillett et al., 2020). Adoption of a similar correction when using our dataset to estimate MoP production should be given consideration until more precise estimates of error are established by clear scientific evidence (Garcia, 1994).

In addition to unreported harvest, the potential contribution of aquaculture to exports has implications for estimating fisheries production of pearl oysters. Neither greensnail nor trochus were ever cultured commercially in PNG (Kailola, 1995; Gillett et al., 2020), but pearl oysters were. During the 1960–70s, and more recently since 1997, pearl oysters were cultured commercially to support pearl production in PNG (George, 1978; IPA, 2021). Since records failed to identify the origin of pearl oyster exports, aquaculture could lead to an overestimation of fisheries production. The small volume of pearl oyster exports, however, suggests that aquaculture contributed, at most, only 4.9% of total MoP exports by weight during these periods (Figure 1), although actual contribution is likely less since the artisanal fishery was active and contributing to pearl oyster exports concurrently (Glucksman and Lindholm, 1982).

A limited aquaculture contribution would imply that export data provides a good indication of the minimal MoP fisheries production which has occurred in PNG, for reasons discussed above. Such information provides a much-needed overview in the case of data-poor fisheries (Govan, 2013), and several studies have demonstrated that export data provide an accurate understanding of relative temporal change in commercial fisheries production (e.g., Clarke, 2004; Schwerdtner Máñez and Ferse, 2010; Gianelli and Defeo, 2017; Plagányi et al., 2017). The significant correlation between Australian trochus production and MoP exports from PNG presented above certainly indicates that this holds true for MoP fisheries, as suggested elsewhere (ICECON, 1997; Gillett et al., 2020). It would be remiss, however, to not offer a few words of caution when relating export trends to production.

Trends in MoP exports, such as those shown in Figure 1A, can be difficult to interpret as they may reflect both biological and economic factors (Nash, 1985; Foale, 2008). For example, reduced export volume in a given year can reflect poor market price (SPC, 1960), rather than a depletion of stocks. Atypical of most commercial fisheries commodities, MoP is an inert product having a long-lasting (i.e., years) shelf-life (Glucksman and Lindholm, 1982; Foale, 2008). This permits stockpiling of MoP for liquidation when more favourable market prices arise and, thus, exports for any given year may reflect both current and past production. To better enable the users of our dataset to address potential economic factors influencing exports, the value tonne⁻¹ of MoP exports was calculated and included in our dataset. Whilst in-country processing of MoP buttons would invariably impact the value tonne⁻¹ of MoP exports, this is a relatively recent development, which occurred irregularly between 1992 and 2014 (ICECON, 1997; Gillett et al., 2020).

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: USC Research Bank [DOI 10.25907/00080] and Research Data Australia [https://researchdata.edu.au/mother-of-pearl-1888-2019/1734519].

Records catalogued by the National Library of Australia had the following accession numbers (Bib ID):

395064: New Guinea gazette.

Records catalogued by the Pacific Community (SPC) Library had the following accession numbers (Bib ID):

18399: SPC (1997). Workshop on trochus resource assessment, management and development report and selected papers. Integrated Coastal Fisheries Management Project Technical Document No. 13. Noumea, New Caledonia: South Pacific Commission (SPC). Digitised and available from: http://opac.spc.int/cgi-bin/koha/opac-detail.pl?biblionumber=18399

Records catalogued by the Staats- und Universitätsbibliothek Bremen had the following accession numbers (Uniform Resource Locator):

urn:nbn:de:gbv:46:1-14372: Jahresbericht über die Entwickelung der Schutzgebiete in Afrika und der Südsee: im Jahre (1899–1907). Digitised and available from: https://nbn-resolving.de/urn:nbn:de:gbv:46:1-14372.

NS, TM, and JK collected relevant records. NS extracted data from records and constructed the dataset. TM and JK verified the dataset. NS and TM performed the statistical analysis for validation. NS wrote the manuscript with support from TM, JK, and PS. All authors contributed to conception and design of the project, manuscript revision, read, and approved the submitted version.

This work was supported by the Australian Centre for International Agricultural Research (ACIAR) and the National Fisheries Authority (NFA) within ACIAR Projects FIS/2014/060 and FIS/2019/122 led by PCS at the University of the Sunshine Coast.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

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