Humans have increasingly been affecting the environment since their origin, causing direct and indirect damage to living beings around them (Kowalewski et al., 2000; Millenium Ecosystem Assessment [MEA], 2005; Ewers and Didham, 2006; Diaz and Rosenberg, 2008; Halpern et al., 2008; Barnosky et al., 2011; Conservation Paleobiology Workshop, 2012; Dirzo et al., 2014; Boivin et al., 2016). Conservation studies have established baselines for sustainable conditions prior to human impact, or where there is less environmental interference (Monsarrat et al., 2019). Therefore, it is possible to understand how the environments used to be, which species existed, and how they responded to variations; in this way, species management can be more effective and clearer (Willis and Birks, 2006; Dietl and Flessa, 2009, 2011; Kowalewski, 2017; Kusnerik et al., 2018; Dietl, 2019; Albano et al., 2020). The name given to this science is conservation paleobiology. It uses paleontology techniques and theories to unravel conservation issues, restoration of biodiversity, and ecosystem services. It aims to develop a long-term scale (Dietl and Flessa, 2009, 2011); and studies using archeological constructions are increasingly common, as they are the geo-archives of past species data (e.g., Mannino and Thomas, 2002; Cannarozzi and Kowalewski, 2019; Fossile et al., 2019; Mendes et al., 2019).

One such construction is the ancient shell mounds, locally named “sambaquis” (from “Tupi-Guarani,” tamba = shell, ki = huddle), which are distributed plentifully all along the Brazilian coast. These sambaquis were built by indigenous groups of fisher-gardeners between circa 10,000 and 1,000 years ago, and are composed predominantly of mollusk shells (Lima, 1999; Figuti, 2005; Schmitz, 2006; Wagner, 2009; Ferrasso et al., 2016; Silva et al., 2017; Scheel-Ybert and Boyadjian, 2020). Presently, the sambaquis are interpreted as funerary monuments and ceremonial spaces, which signifies that the mollusks were collected principally as building material and their role as food was secondary—the shells are great building material and food supplements at once (DeBlasis et al., 1998, 2007; Scheel-Ybert and Boyadjian, 2020). The use of these Brazilian shell mounds in conservation paleobiology, as a baseline for when human activities were less expressive, has already showed that it is possible to establish a taxonomic census of species that occurred in past environments (Fossile et al., 2019; Mendes et al., 2019), but no effort has been made yet to evaluate the body-size difference between the valves of present-day mollusks and of sambaquis for conservation purposes. These information from the past can help us conserve the environment today; in other words, the past is also the key to the present and the future (Dietl and Flessa, 2017). The fisher-gardeners may have acted as researchers of the past, although they lacked discernible scientific methods.

However, working with research concerning the comparison of the valves’ body size (an intrinsic characteristic of mollusks) from the sambaquis and of the current populations proved to be a convoluted task. Although the body size of the valves contains relevant information about what the environment was like in their lifetime (Fiori and Cazzaniga, 1999; Mannino and Thomas, 2002; Campbell, 2008; Álvarez-Fernández et al., 2011), the information acquired from the sambaquis may be biased, because they were made by humans. There is no real evidence from sambaquis in south Brazil about how the gathering was done by the fisher-gardeners, but factors such as cultural preferences, food taboos, discarded materials, and technique can affect the data that was collected (Silva et al., 2017; Mendes et al., 2019). For the result to be as reliable as possible, other important questions, such as the type of data and tests that are appropriate for this comparison, must be addressed.

For these reasons, this research aims to assess whether it is possible to identify differences in body size between the Amarilladesma mactroides (Reeve, 1854) valves found in two sambaquis, and in two transects in the foreshore areas of the state of Rio Grande do Sul in the southernmost part of Brazil. Furthermore, we verify the existence of significant differences between valve sizes of the sambaquis and the current mollusk population at distinct swash zone levels (from one of the current transects only). These analyses were made using parametric (t-test, tt), non-parametric (Mann-Whitney test, mw) analyses, and Bootstrap (bt) techniques for the following scenarios: normal comparison (equal sample numbers) and decreasing the sample number (n = 100, n = 50); comparison with valves bigger than average, and comparison of the valves of sambaquis with those from different parts of the swash zone, which is the beach area that is covered and exposed, alternatively, by the uprush and backwash of the water and is thus limited by it (Masselink and Puleo, 2006; Hughes and Baldock, 2020). Therefore, we aim to contribute to conservation paleobiology by assessing the problems enumerated above and trying to obtain clues regarding how groups of ancient fisher-gardeners exploited these shells (and the swash zones they were collected in).

The sample size for comparing old (from sambaquis) and recent (from the Tramandaí-Palmares do Sul, and Altair shipwreck-Sarita Lighthouse transects) valves was 233, and 177 for valves larger than average (mean sambaquis: 45.95; mean of recent valves: 44.30). The sample size was 74 for comparing body size between the valves of the shell mound, with valves from all levels of the swash zone, from the Altair shipwreck-Sarita Lighthouse transect. When we refine for the different zones, the values are as follows: n = 31 in the comparison shell mound and upper-swash zone; n = 30 in sambaqui and middle level; and n = 13 in the sambaqui and lower swash zone level.

The Shapiro-Wilk test showed that the data from the normal comparison between the body size of shell mound valves (p < 0.001) with the current ones (from the two transects) (p < 0.001) do not have normality, even when performing the Box-Cox conversion; therefore, we did not transform the data that were not normal. When we reduced the sample size to 100 and 50, the body size data for shell mound valves were normal (p = 0.28, and p = 0.71, respectively), but this was not the case for recent valves (p = 0.01 and p = 0.004). Comparison data with sambaquis and recent valves above the mean were also abnormal (p = 0.0004 and p = 0.01). In comparing sambaquis valves with those of all levels in the swash zone (from Altair shipwreck-Sarita Lighthouse transect), the data showed a normal distribution (p = 0.66 and p = 0.11). For the comparisons with the different levels of the swash zone, the data from the shell mounds were all normal (upper: p = 0.37; middle: p = 0.30; lower: p = 0.92) and the data on the current valves had normality only at the upper and lower levels (upper: p = 0.77; middle: p = 0.04; lower: p = 0.09).

There is a significant difference (p < 0.05) between the means (the body size is greater in the sambaquis) of the Amarilladesma mactroides valves present in the shell mounds and those in the current coastal plain in all statistical tests, even when the sample number is reduced (Table 1 and Figure 3). However, when comparing the size of valves with a body size greater than average, we obtained a significant difference only in the Mann-Whitney test (tt, p = 0.24; mw, p = 0.04; bt, p = 0.25) (Table 1 and Figure 3). The means do not differ significantly when the body sizes of sambaqui are compared with those of the Altair shipwreck-Sarita Lighthouse transect (tt, p = 0.21; mw, p = 0.28; bt, p = 0.21) (Table 1 and Figure 4A). Thus, when refining the comparisons for each level of the swash zone (Figure 4B), there is no significant difference in the means when comparing sambaquis valves with the upper level (tt, p = 0.84; mw, p = 0.63; bt, p = 1), or with the lower level (tt, p = 0.16; mw, p = 0.44; bt, p = 0.15); there is a significant difference only a when comparing sambaqui valves with the middle level (tt, p = 0.004; mw, p = 0.004; bt, p = 0.005) (Table 1 and Figure 4A).

Body size is an intrinsic characteristic of mollusk valves related to the physiology, lifestyle, and ecology of organisms, and is controlled by abiotic and biotic factors (Campbell, 2008). Specifically, from the body size of fossil valves or those present in archeological sites, relevant information can be obtained regarding paleoenvironmental conditions (Balbo et al., 2011), population size estimates of ancient fisher-gardener groups (Klein and Steele, 2013), the establishment of baselines, and contributions to species management (Kusnerik et al., 2018). However, when comparing this parameter at different times, it is necessary to take some precautions, such as knowing which data and methods to use, the location and its history, considering processes that can lead to biased data, and similar environmental conditions; however, such information is not always available.

An important problem for comparing body size between recent and ancient Amarilladesma mactroides valves is being unsure of how and in which areas of the beach the fisher-gardeners collected or discarded unwanted shell sizes. Ancient populations from different areas of the world collected the shells with their bare hands or simple tools in areas close to the camp, where there were banks of abundant organisms (reviewed in Waselkov, 1987). However, there are no reports or studies on how this was done by the people who collected from sambaquis at the North Coast of CPRS. Thus, the approach taken here uses body size and conservation paleobiology to provide information on how fisher-gardeners exploited these mollusks (Campbell, 2008; Cannarozzi and Kowalewski, 2019).

Interestingly, the statistical tests—and even the t-test that requires data normality, unlike the Mann-Whitney and Bootstrap—had significantly similar results to at least one of the non-parametric tests, especially the Bootstrap approach. Studies suggest that sometimes, due to moderate deviations from normality and large and equal sample size in comparatives, the t-test works without normality (Bartlett, 1935; Kowalewski and Novack-Gottshall, 2010; Poncet et al., 2016); however, there are also studies that show otherwise (Kim and Park, 2019). The t-test makes this dubious, and there are comparisons to non-normal data with small sample sizes; therefore, this test is not indicated here. Additionally, the Mann–Whitney test, a non-parametric rank-based test, is not the best to use, as it consists of a loss of statistical power, increasing the chance of false results (Kowalewski and Novack-Gottshall, 2010); further, there are data with normality. Alternatively, the Bootstrap is a resampling method and there is little or no loss of statistical power (Kowalewski and Novack-Gottshall, 2010). The resulting sample from the bootstrap estimates, even with non-normal data, generally form a Gaussian distribution and are more robust and valid for various conditions (Efron and Tibshirani, 1985). Hence, it is indicated to a greater extent in this paper.

Considering all this, and despite having found a difference between the means of the body size of Amarilladesma mactroides valves from the sambaquis and current population (even with the decrease in the sample size), this result may not be credible and indeed, may be biased, which makes it impossible to discern whether this change in body size is due to human or environmental changes. This is due to the fact that when a more refined comparison was made between the different swash zones (upper, middle, and lower zones) of the Altair shipwreck-Sarita Lighthouse transect, we observed no difference in means between the lower and the upper zones, where the larger organisms stay during winter and summer, respectively, and/or where larger organisms are found at low tide (Fiori and Morsán, 2004; Bergonci and Thomé, 2008).

Thus, we perceive that the groups of fisher-gardeners chose larger shells, possibly for the construction of the sambaquis, when we consider that the organisms of Amarilladesma mactroides were distributed in this manner in the past. Nevertheless, it remains unclear whether or not they fed only on larger shells and discarded the smaller ones. This, however, seems unlikely due to the energy expended in collecting and searching (Waselkov, 1987; Campbell, 2008), since Amarilladesma mactroides is infaunal. Alternatively, they may have collected all the shell sizes and fed on them, using only the biggest ones to make the shell mounds to preserve energy. Finally, they may have collected the larger shells seasonally, at only one level of swash (e.g., Waselkov, 1987; Andersen, 2000; Bailey and Craighead, 2003; Colonese et al., 2011; Cannarozzi and Kowalewski, 2019).

Therefore, when working with the body size of shellfish valves in sambaquis, one must be careful to discern whether the results are biased toward a better similarity with current data, as attempted here. However, it would be valuable to carry out samplings in the different swash zones on the Arroio do Sal, the location of the sambaquis, as it presents the most accurate conditions for comparing the organisms of the present and the past. Additionally, to conduct more expeditions to ascertain whether there are Amarilladesma mactroides valves preserved in the depositional barriers (retro barrier, lagoonal systems) on the coastal plain (e.g., Fiori et al., 2019; Frontini et al., 2021), we would have to use older samples while comparing the body size to prevent artificial biases. As a precaution, it would be important to investigate the body size and horizontal distribution of Amarilladesma mactroides before and after the mass mortality that occurred between 1993 and 2004 (Fiori and Cazzaniga, 1999; Dadon, 2005; Vázquez et al., 2016) because it may change those parameters; however, mass mortalities could also exist continually through time.

The Brazilian shell mounds have been destroyed since the invasion by the Europeans in the 16th century, as they were used in lime making, as mortar in constructions, fertilizer, and in road paving (Lima, 1999). This exploration for industrial purposes ended in the second half of the 20th century, when sambaquis began to be legally protected (Lima, 1999). However, they are still depredated in the present day because, in most cases, there is neither the implementation of in loco protective measures, nor any contact with the local population and visitors (e.g., Costa, 2019; Guimarães et al., 2021), which is why the “know to protect” phrase is important. Hence, the results herein can help future conservation paleobiology studies using sambaqui to access more archeological and paleontological information. This may improve the importance of these mounds and bring them to the current and accessible media, where the population are aware of their responsibility to protect them.

In conclusion, we use conservation paleobiology procedures to assess the problems brought about when comparing the body size of recent and past mollusk valves that are present in shell mounds, in an attempt to show that the choice of tests and data collection sites are essential in this regard. Additionally, we are able to propose that the groups of ancient fisher-gardeners exploited the larger shells for the construction of shell mounds in the upper and lower part of the swash zone; hence, the body size data from sambaquis are biased and is not reliable to infer morphometrics of past populations of mollusks in these shell mounds from the southern Brazilian coast. This research contributes to the existing information on paleontological and archeological areas of the CPRS, and to future worldwide studies that may use shell mounds as geo-archives. Thus, the sambaquis will have increased attention and protection through the responsible agencies, as currently they are somewhat degraded and depredated for not having in loco protection.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

AA and MR designed the study and analyzed the data. FC, FE, and EB helped with field exploration and data collection. RP helped by sharing live datasets. All authors were involved in data interpretation, drafting of the manuscript, and read and approved the final manuscript.

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.