Sampling and collector biases as taphonomic filters: an overview

Abstract

Sampling (or sample) bias is a widespread concern in scientific research, across several disciplines. The concept of sampling bias originated in statistical studies. The consequence of a biased sample is that scientists will conclude about a population different from their target. In paleontology, sampling bias is typically related to fieldwork context. Human factors, known as sullegic (e.g. collection method, historic resampling) and trephic (transport, and curatorial processes) factors can generate bias. Other factor is the ugly fossil syndrome (i.e. choosing based on completeness of the specimens, or according to the researcher interest). Thus, sampling implies information loss. Biased samples add artificial results and can be considered an additional taphonomic filter. Therefore, sampling bias and the collector role and choices are frequently linked and almost indistinguishable. Compared to the treatment of this topic in other research fields, little related discussion has been held in vertebrate paleontology, especially regarding what happens at the interface between the biosphere, lithosphere, and anthroposphere, and during the transition between the anthroposphere and the patrisphere (museums). Numerous questions still arise. As a community, we must pay attention, to minimize the loss of information, from field activities to cataloging.

Key words
Paleoecology; paleontological collections; sampling effect; vertebrate paleontology; vertebrate taphonomy

INTRODUCTION

‘Sampling bias’ is a term widely used in diverse research areas, from economics and health to sociology and ecology, and in practically all the natural sciences. In statistics, the preferential selection of specimens of one group over another, generates non-real, erroneous, and tendential data in a phenomenon known as ‘sampling selection bias’. This trend is generally linked to non-random data sampling (Bareinboim et al. 2014, Qin 2017).

In paleontological literature, sampling bias can be found in all areas of specialization, including paleobotany, invertebrate and vertebrate paleontology, macro- and micropaleontology, from both marine and continental environments. In paleontology, the concept is mainly associated with evolutionary or paleobiodiversity studies (Mannion et al. 2013), often linked to the discontinuity of the geological record, the quality of the fossil record (Raup 1972, Benton et al. 2000, Kidwell & Holland 2002, McGowan & Smith 2011, Mannion et al. 2013), and possible distortions distributed across various environments in space and time (Alroy 2010, Smith & McGowan 2011). However, the term can also be associated with the ‘privileged selection’ of certain fossil specimens that are more complete, better preserved, more easily recognizable from an anatomical perspective, or more readily taxonomically identifiable (Tang 2000), with possible repercussions in terms of paleoecological reconstructions, leading to further information loss (Clark & Kietzke 1967, Behrensmeyer & Kidwell 1985, Behrensmeyer & Miller 2012, Behrensmeyer et al. 2018) and impacting the constitution and quality of collections (Fara et al. 2005).

Rather than seeking a solution to the problem of sampling bias, this overview offers a point of reflection on the role of the researcher as the cause of the problem. It draws upon the concept of bias, including sampling bias, in vertebrate paleontology with a focus on vertebrate taphonomy. The emphasis on selection by the collector (professionals and enthusiasts) positions the collector role as an additional taphonomic filter, an idea that is present in some form in the literature (e.g. Clark & Kietzke 1967, Behrensmeyer & Kidwell 1985, Behrensmeyer & Miller 2012), though it has been given little consideration.

SAMPLING BIAS

The concept of biased, which implies its opposite, unbiased, exists in socioeconomic, health, physical and natural, and mathematical and statistical research (Berk 1983, Winship & Mare 1992, Breslow 2003, Hernán et al. 2004, Dibble et al. 2005, Fara et al. 2005, Zar 2010, Qin 2017).

The term originated in the mathematical statistical environment. It dates back at least 120 years (Zar 2010) according to a review by Miller (2004) (online and open source) regarding the use of some common mathematical terms in non-mathematical fields. In a comparison of investigation methods used in the physical and statistical economic sciences, Bowley (1897) found that even the most precise measurements have accuracy limits and thus, the potential for bias; like initial errors, biases affect the results, albeit minimally. Currently, terms such as ‘sampling bias’, ‘biased/unbiased sample’, and ‘selection bias’ can be found in scholarly literature spanning a wide range of fields. As of November 2023, a simple generic Google Scholar search for ‘sampling bias’, ‘biased sample’, and ‘selection bias’, without any specific search filters, returned over 6.7 million, 6 million, and 5.9 million results, respectively, within a few seconds. These results indicate the degree to which sampling bias is widely present and debated.

A bias is a deviation, distortion, or tendency. Therefore, the commonly accepted meaning of ‘sampling bias / sample selection bias’ is a tendency that occurs when a researcher usually inadvertently performs non-random sampling by favoring one population segment over another, thus obtaining data that are not representative of the entire population (Bareinboim et al. 2014, Qin 2017). In this context, in agreement with Zar (2010), statisticians defined ‘population’ as a group of measurements or data from which to draw conclusions.

According to Heckman (1979), sampling errors arise because of self-selection or due to researchers and analysts decisions. This is common in the health field, and even more when conducting market surveys, since the method is financially convenient and cost-effective, directing the survey only to a certain area or a specific part of the population (Qin 2017). Therefore, the risk of directed sampling is high, as is the likelihood of obtaining distorted results. Numerous generic and specific models have been proposed over the years to minimize error (e.g. Heckman 1976, 1979, Ponder et al. 2001, Breslow 2003, Bourguignon et al. 2007, Cortes et al. 2008, Pearl 2009, Zar 2010, Bareinboim et al. 2014, Breen 2015, Qin 2017).

Recent examples are related to the novel coronavirus pandemic (World Health Organization 2020, Zhu et al. 2020). Among the problems, the real risk estimate detected was the between test time interval role in determining epidemiological evolution, especially at the beginning of the phenomenon (Du et al. 2020, Kobayashi et al. 2020). The different sources used for data collection also favoured possible distortions (e.g. as different jurisdictions counted cases differently), resulting in a non-real estimate (Nishiura et al. 2020). Data collection is also linked to the size of the population/city and the relative efficiency of the health and diagnostic systems (Du et al. 2020). It may also be linked to the type of analysis, which may not be representative of the target population (Wynants et al. 2020). The latter exists in the medical field, and various models of bias prediction and correction as well as procedural protocols have been proposed to reduce this error (e.g. Barton et al. 2020, Kong et al. 2020, Wynants et al. 2020).

In ecological studies involving aggregate data (Rousson et al. 2017), distorted results are rightly linked to the sampling method. This is especially true of ecological survey methodologies (i.e. area- and individual-level), which have a wide range of applications, including in epidemiological and sociological studies (e.g. Berk 1983, Winship & Mare 1992, Hernán et al. 2004, Wakefield 2004, Rousson et al. 2017), and those more closely related to the natural sciences, such as studies of a purely environmental nature and research concerned with the conservation and preservation of biodiversity, biogeography, and animal behavior (e.g. Reddy & Dávalos 2003, Biro & Dingemanse 2009, Leitão et al. 2011, Stuber et al. 2013, Zhang & Zhu 2018). In the case of animal behavior, where there are clear repercussions for the distribution analyses of a given taxon, scholars are aware (Réale et al. 2007) that individuals personality and temperament can be sources of distortion (Biro & Dingemanse 2009, Stuber et al. 2013, Michelangeli et al. 2016).

University and museum collections, widely used in biodiversity studies, have also been affected by sampling bias (Ponder et al. 2001, Thomas 2010). In studies on environmental conservation and biodiversity (Reddy & Dávalos 2003), sampling is often in the areas closest to inhabited sites or those that are more easily accessible. In this case, the sampling of certain regions is much fuller and more detailed, even with the amount of information available (Ponder et al. 2001). Cooper et al. (2019) examined extant mammal and bird collections from five major museums sampled over the past 250 years and observed that in both classes, the number of male specimens (accounting for approximately 60% of the materials present in the collection) exceeded the number of females (approximately 40%), revealing preferential selection linked to the animal sex. According to Cooper et al. (2019), this difference cannot be attributed solely to human action (linked to ethics) or to males being more conspicuous than females due to secondary sexual characteristics (e.g. ornamentations such as horns or showy plumage). Rather, a potential explanation may be that mammalian males tend to move around more than females to gain and maintain territorial control, thus making them more susceptible to capture. Hence, gendered behavioral differences within a given species play an important role during sampling (Biro & Dingemanse 2009, Cooper et al. 2019).

Botanical collections and plant genebanks suffer similar problems. For example, in a study on the geographical representativeness of Bolivian wild potatoes in genebanks, Hijmans et al. (2000) identified four distortions: species bias, the preferential collection of one species over another linked to the researcher choices, to a greater abundance of one in relation to the other; species-area bias due to under- or oversampling of a given species in the area under examination; hotspot bias, referring to insufficient or excessive sampling in a given region, based on its genetic diversity; and infrastructure bias or oversampling in the regions closest to communication routes or urban centers, which Ponder et al. (2001) also subsequently observed.

Like naturalistic collections and genebanks, paleontological collections can also be directed. Selective sampling has been found to be present and influential in the constitution of paleontological collections, particularly the most ancient ones (Thomas 2010). Fossil research has been preoccupied with the search for the ‘most beautiful fossil’, the most complete or best-preserved specimen. This race for the fossil reached its peak during the so called ‘Bone War’ between Edward Cope and Othniel Marsh during 19^th century (Skelton 1984), in which rivalry between the two scientists led to the discovery of over 100 species of dinosaurs (but also to the espionage and destruction of newly discovered and not yet studied fossils: Wallace 1999). This is not that far from the phenomenon that Tang (2000) termed as ‘Ugly Fossil Syndrome’, which we discuss in the following section.

Bias in vertebrate paleontology

Paleontologists strive to understand the evolutionary dynamics and biodiversity of the past, given those factors’ strong influence on the quality, richness, and completeness of scientific collections (Sepkoski Jr. et al. 1981, Mannion et al. 2013). Over the last five decades, numerous studies have aimed to understand the fossil record (e.g. Raup 1972, Sepkoski Jr. 1975, Sepkoski Jr. et al. 1981, Signor 1978, Signor & Lipps 1982, Alroy et al. 2001, McGowan & Smith 2011, Mannion et al. 2013, Lukic-Walther et al. 2019, Dean et al. 2020), which has been accepted as incomplete. It is also known that the quality decreases as one moves further into the past, from the most recent deposits to the oldest (Raup 1972, Benton et al. 2000), with a marked difference between invertebrate (with mineralized structure or shell) and vertebrate fossils, due to mineralogy of the skeletal elements, anatomy, and related to geography and depositional environmental conditions, for example (e.g. Vermeij & Leighton 2003, Smith & McGowan 2011, Krone et al. 2024). Almost two centuries ago, Charles Darwin noted that the geological record influences our knowledge and understanding of evolution (Darwin 1859). Apart from its incompleteness, we must also consider heterogeneity in outcrops distribution, extension, and location (Crampton et al. 2003, Smith & McGowan 2007) (Figure 1), which underlines what was previously introduced, highlighting the effort invested in collection (Crampton et al. 2003) and the objective of exploration (i.e. geological, or paleontological survey: Whitaker & Kimmig 2020) (Figure 1). Crampton et al. (2003) suggested that the outcrop area can be a reasonably reliable estimator of rock volume bias, albeit independent of the tectonic regime and in the absence of robust data reflecting the collection effort.

Figure 1
Conceptual map of the factors determining biases in the fossil record and palaeontological collections. Adapted from Clark & Kietzke (1967), Wolff (1975), Behrensmeyer & Kidwell (1985), Fernández-López (1991), Tang (2000), Crampton et al. (2003), Smith & McGowan (2011), Behrensmeyer & Miller (2012), and Whitaker & Kimmig (2020).

In his analysis of taxonomic diversity during the Phanerozoic Eon, Raup (1972) identified nine filters that affect our knowledge: anatomical, biological, ecological, sedimentary, preservation, diagenetic, metamorphic, vertical, and human. The first three refer to the pre-burial period, while the next five are of a basic geological nature. According to Benton & Harper (2009), the human filter (Raup 1972) acknowledges the necessity of human intervention, as the fossil, once recognizable on the surface, must be found and collected, then prepared, catalogued, and deposited in a collection.

In addition to the geological record’s completeness and real content, another determining factor is linked to Tang’s (2000) ugly fossil syndrome (Figure 1). Tang (2000) highlighted some differences between macro- and micropaleontology in terms of collection and preparation methods, specifically a natural attraction to those fossils that are more ‘beautiful’, complete, best-preserved, and in some cases, the most easily recognizable, both anatomically and taxonomically. This observation certainly applies to macropaleontology but is less applicable to micropaleontology (though not non-existent) since innumerable fossilized microorganisms can be present in a single sample in block or in bulk. Researchers’/collectors’ level of preparation and experience are equally crucial in terms of ease of recognition in the field (Shipman 1981, Lyman 1994). These factors can contribute to paleontological collections comprising well-preserved and better-studied specimens as opposed to those with worse and more fragmentary preservation (Fara et al. 2005, Thomas 2010, Marramà et al. 2016) as well as to databases widely used in paleobiodiversity analyses (e.g. Mannion et al. 2013, Dean et al. 2020).

However, this trend was not always prevalent. ‘Less beautiful’ fragmentary specimens can be investigated with less scruple and more destructive techniques, which may yield information that would otherwise be difficult to access (Thomas 2010). For example, Melo et al. (2019) pointed out that where non-invasive analyses (e.g. CT scanning) were not appropriate due to the type of fossil preservation, destructive techniques unearthed evidence of hypsodonty in early Carnian non-mammalian cynodonts (Late Triassic), which was previously only known to exist in Mammaliaformes. As Mannion et al. (2013) noted, any attempt to read the fossil record entails a high risk of obtaining models that are partially or entirely affected by sampling bias.

Bias in vertebrate taphonomy

Just over eighty years ago, Efremov (1940) introduced the term ‘taphonomy’ as a new branch of paleontology, incorporating Johannes Weigelt’s 1927 biostratinomic concepts (‘biostratonomy’, back then: Weigelt 1989). Weigelt’s goal was to study the passage of animal remains from the biosphere into the lithosphere (Olson 1980). In this vein, Müller (1963 apud Lyman 1994) coined the term fossildiagenese to refer to the chemical-physical processes that continue to modify once buried organic remains. The definition of ‘taphonomy’ has since evolved to the modern meaning Behrensmeyer & Kidwell (1985) proposed: the study of preservation processes and the extent to which they influence the fossil record through biological information loss and the acquisition of taphonomic information (Fernández-López 1991) (Figure 1). Any post-mortem event or phenomenon, whether pre- (biostratinomic) or post-burial (fossildiagenetic), involving alteration of the carcass’ initial condition (e.g. through necrophagy and scavenging, skeletonization, transport, breakage, or chemical dissolution) has as a consequence a reduction in the amount of biological information available (Efremov 1940, Clark & Kietzke 1967, Behrensmeyer & Kidwell 1985, Behrensmeyer et al. 2000, Behrensmeyer & Miller 2012). The researcher or ideally the taphonomist role is integral to minimize this natural distortion and provide as accurate a paleobiological and paleoecological reconstruction as possible (Lawrence 1968, 1971). This is even more crucial if we consider the great contribution taphonomic studies makes to other disciplines such as paleoecology, paleobiogeography, and paleobiology (Shotwell 1955, Behrensmeyer & Kidwell 1985, Behrensmeyer et al. 2000, Behrensmeyer & Miller 2012).

As discussed regarding the abovementioned research areas, the topography of the area, the environmental context, proximity to a city or communication routes, and the main purpose of the research or sampling activity, among others, are all sources of bias. Additionally, classical, and guiding works related to taphonomic studies acknowledge the existence of problems arising from the figure of the researcher. Clark & Kietzke (1967) proposed an examination of the following factors that can generate biases in a collection (Figure 1): biotic factors, which are strictly linked to the life of organisms and species, the distribution and density of organisms in a specific area, and the osteological structure (implying that certain bones in a skeleton are more fragile than others, with less chance of good preservation); thanatic factors, which are related to the causes and place of death; and perthotaxic factors, which are due to biotic or abiotic phenomena (e.g. scavenging, climate, and exposure). The latter two are constituents of Weigelt’s (1989) biostratinomic concept. Furthermore, Clark & Kietzke (1967) identified taphic (burial) factors related to sediment type, depositional rate, lithostratigraphic pressure, the action of roots and fossorial organisms, and anataxic factors, which are events following fossilization that led to the destruction of the remains such as fossil exposure and transport. Once the specimen is found, further damage and loss of information were identified by Clark & Kietzke (1967) and included in sullegic (collecting) and trephic (curatorial) factors that are fundamentally linked to human action. These factors include ‘the Chevrolet effect’ (Flessa et al. 1992), and ‘cultural filter’ (Reed 1963 apud Faith & Lyman 2019). These are linked to sampling and prospecting methods, successive sampling activities over the years (with different results), the collector physical and mental state (external disturbance, e.g. mosquitoes, physical fatigue, and/or less training and experience), post-harvest transport, and laboratory preparation. In a study on actualistic taphonomy, Andrews & Cook (1985) proposed similar considerations. Furthermore, Voorhies (1969) admitted that his results may be biased, having washed only a third of the sediments sampled from the study area. This is because, depending on the method used (e.g. ordinary quarrying or screen washing), the results obtained can be significantly different, for example in the relationship between proximal (larger) and medial-distal (smaller) phalanges. Furthermore, based on experience with invertebrate samples, Rothfus (2004) and Ritter et al. (2016) identified multi-operator biases. Evaluating taphofacies and the replicability of statistical tests involving taphonomic signatures (e.g. fragmentation of the valves) observed by multiple operators, both Rothfus (2004) and Ritter et al. (2016) highlighted some differences in the obtained results. Finally, among the collection factors, multi-operator bias is not far removed from resampling proposed by Clark & Kietzke (1967).

In the case of bones accumulated after hydraulic transport, the difference in the frequency at which various disjointed bone elements appear in an outcrop and then in a collection of that outcrop might include a taphonomic bias, as has been a common knowledge for some time (Leidy 1869 apud Voorhies 1969). Voorhies (1969) attributed this difference to susceptibility during the hydraulic transport of individual bones and schematized it in the ‘Voorhies groups’ (Behrensmeyer 1975). This is directly linked to the collection of information, such as the orientation of the individual bone elements with respect to the plane (transport and/or preferential alignment), the reciprocal position of one bone with respect to another (accumulation), and their arrangement with respect to the sedimentary interval containing the bone (pre-burial biotic action) (Toots 1965, Voorhies 1969, Fiorillo 1989). Given the importance of this information for both paleoenvironmental and paleogeographic reconstructions (Toots 1965), an incomplete reading of (or worse, the failure to read) this information can generate bias in reconstructions.

In the mid-1970s, while valuating the adequacy of a collection of vertebrate remains for paleoecological analysis, Wolff (1975) noted that this is linked to the exposure of the outcrop. It was therefore initially superficial, making it necessary to excavate, wash and sift the sediment given that small bones and teeth, especially in micro-mammals, have similar dimensions to the sediment and are therefore equally transportable and more difficult to identify in the field (Behrensmeyer 1975). During the sample sieving phase, the aperture mesh size plays a decisive role and can alter the results (McKenna 1962, Wolff 1973, 1975, Behrensmeyer 1975, Korth 1979, Lyman 1994, 2008, Kidwell 2002, see also Pike 1993, Kowalewski & Hoffmeister 2003). This was the case of a Maastrichtian bonebed with hadrosaurids from Catalonia (Spain), which was apparently monodominant (Battista et al. 2014). Sifting and a careful review of the finer portion revealed richer non-monotaxic fauna with fragments of dinosaur eggshells, amphibians, fish, lepidosaurs, and crocodiles (Fondevilla et al. 2017, and references therein). Therefore, the sampling method and choices, also in relation to the size of the bone elements (small-body vs large-body remains), are crucial for the purposes of any reconstruction (see Faith & Lyman 2019, and references therein).

DISCUSSION

As several authors have pointed out, especially in studies on biodiversity and its preservation, a given species’ distribution models are often based on pre-existing datasets. However, this can be unbalanced owing to numerous factors including the sample size, incorrect data recording, and mismatched scales (see Reddy & Dávalos 2003, Crampton et al. 2003, Leitão et al. 2011). By contrast, historically known but relatively uncommon events, such as the stranding of cetaceans in some regions, have been recorded with greater accuracy (Sheldrick 1976), supporting more reliable datasets for studies on the distribution of the group and those pertaining to the taphonomic field (Pyenson 2011). These datasets are based on information available in various collections. However, because they are also based on publications, they vary according to the research group’s specialization, and there may be a tendency to favor certain specimens over others or decide based on a specific taxonomic group’s ‘charisma’, as Whitaker & Kimmig (2020) deduced, compounded with Tang’s (2000) claims. For example, during field activities, there are relatively few students who, for example, when analyzing a given outcrop from a biostratigraphic standpoint, collect everything, especially when the rock is relatively easily eroded and thus offers greater ease of excavation and collection. Typically, the collection is composed primarily of well-preserved or relatively ‘complete’ specimens. This was observed directly in the field (F. Battista, pers. obs.) in an outcrop of Rosso Ammonitico, a Toarcian reddish shales and marls to marly limestone formation rich in ammonites, from the central-northern Italian region of the Apennines (Petti & Falorni 2007). This formation, as well as others from the Jurassic of the Italian Apennines, is characterized by an abundance of phylloceratid ammonites, which possess little biostratigraphic power. Once the students learned this, they tended to leave many of these specimens in the field, producing high concentration of exhumed, anthropologically reworked, and discarded fossil remains. On the other hand, the opposite has also been observed in deposits of Pliocene sands and clays rich in bivalves and gastropods in known and studied sections; the derived collection and the distribution of taxa in the final columnar section were composed primarily of the most complete specimens and those with the most conspicuous ornamentation (U. Nicosia, pers. comm.). Similar selection has been observed even in vertebrate paleontology field works. During excavation in a bonebed featuring hadrosaurid dinosaurs of the Catalan Maastrichtian, it was observed that students with more experience than first-year undergraduates also tended to not collect everything and excluded indeterminate centimetric fragments (which are always abundant) in favor of whole remains, such as centra, neural arches, phalanges, and femora (F. Battista, pers. obs.). These are just a few among countless examples that demonstrate the importance of the researcher role, which falls under sullegic factors, as defined by Clark & Kietzke (1967), while also exemplifying Tang’s (2000) ugly fossil syndrome. Thus, graduate, master, and doctoral students behaved similarly, representing an important filter and producing bias during sampling.

Therefore, data collection must be scrupulous and meticulous to minimize the margin of error and the distortion of results. Taphonomic studies also have this requirement, especially given that taphonomic analyses supply information for paleobiological, paleoenvironmental, and paleoecological reconstructions. Should fieldwork and paleontological collections have an express taphonomic purpose (see Fara et al. 2005, Pickering & Brain 2010, Marramà et al. 2016) to obtain richer, more comprehensive data? Differentiating between geological and paleontological surveys, Whitaker & Kimmig (2020) emphasized the ‘why’ behind the expedition as one of the determining factors of the type and quality of the resulting collection. During a geological survey (regardless of its main purpose), the recognition and collection of fossil remains are fundamentally linked to the utility of the relative dating of the outcrop. Even in such a case, the identification of a zone marker specimen, whether fragmented or whole, carries much more weight than an indeterminate one. This difference is more pronounced if we compare marine and continental deposits. During a geological survey, for example, finding a fragment of the ammonite Hildoceras bifrons in Jurassic marine deposits (a marker of the homonymous Toarcian Biozone: Gradstein et al. 2020) can be drastically more informative (form a biostratigraphic point of view) than the discovery of many diaphyseal long bone fragments in contemporary soils in a continental environment. On the contrary, the discovery of a whole or fragmented fossil bone, during a paleontological survey, pushes the researcher to perform more in-depth analysis of the outcrops and the entire investigated area, in either continental or marine settings.

Another factor that can contribute to the loss of taphonomic data and thus possibly to distorted paleoecological and paleoenvironmental reconstructions, is the fact that several different researchers and/or institutions may sample the same outcrop over the years. These researchers do not necessarily belong to the same institution and, even when they do, they may not necessarily use the same methodologies (indeed, an outcrop can change over time due to the action of nature and/or man). Clark & Kietzke (1967) included this among the sullegic factors. It may be the case for the fossiliferous Faixa Nova locality, in the city of Santa Maria (Rio Grande do Sul State, Brazil: Da-Rosa 2004), where three different research teams (Pontifícia Universidade Católica do Rio Grande do Sul, Universidade Federal do Rio Grande do Sul, and Universidade Federal de Santa Maria) collected parts of a fossiliferous association comprising aetosaurine aetosaurs and hyperodapedontine rhynchosaurs at different times (Desojo & Ezcurra 2011, Paes-Neto et al. 2021a). The understanding that these three collections included parts of the same specimens followed the realization that the fragments of some bones that had been separated into different collections fit together perfectly. However, the taphonomic history of this association has not been investigated, despite the fossil richness and the size of the outcrop (investigation has been partial and only with respect to other outcrops and localities: Holz & Barberena 1994).

The influence of randomness in any collection is also notable, as it is related to the prospecting method or looking at the outcrop (see Clark & Kietzke 1967) and subsequent collection techniques (e.g. Wolff 1975). For example, until the mid-1990s, little was known about the diversity of non-mammalian cynodonts (Azevedo et al. 1990) and basal dinosaurs (Kellner & Campos 1999) from Middle and Upper Triassic deposits of the Rio Grande do Sul (southern Brazil). During a fieldtrip, one of the authors (CLS) found a new fossiliferous deposit containing hundreds of non-mammalian cynodont remains (Abdala et al. 2001, Battista et al. 2023, 2024) and several archosaurian remains (Battista et al. 2023, 2024). Since then, in less than 30 years, knowledge about the Triassic fauna of Rio Grande do Sul has completely changed (Schultz et al. 2020). At present, we can consider this Brazilian region as the veritable cradle of the lineages that led to the origin of the first mammals and dinosaurs (Kellner & Campos 1999, Langer et al. 2018, Kerber et al. 2023; among others). Consequently, if the faunal analysis proposed by Azevedo et al. (1990) were performed today, the result would be completely different. Shotwell (1955) highlights the importance of the sample size to understand and reconstruct the paleoecology and to determine if recognized taxa belong to a proximal or distal community. In this sense, the new Triassic deposit described by Abdala et al. (2001) represents a good example of anthropogenic selection (‘non-representativeness’ in Figure 2) as taphonomic filter. In fact, the new site, belonging to Santa Cruz Sequence (for details, see Battista et al. 2024), has been object of taphonomic analysis by Bertoni-Machado & Holz (2006), in which less than 100 bone elements have been reported, while more than 2300 elements (either complete or fragmentary elements) had already been sampled (Table I), at the time, from the same outcrop by three different institutions (for more details, see Battista et al. 2023). Consequently, the taphonomic reconstruction proposed by Bertoni-Machado & Holz (2006) results incomplete, as well as will results incomplete any analysis focused on the biodiversity of this outcrop if consider the three collections in a separate way (for details, see Battista et al. 2024). As also highlighted by Battista et al. (2024), a further problem was due to the incompleteness of the sampling information (for example, stratigraphic provenance and photographic material). The absence of this information, as well as unlabeled fragmented specimens that have been (involuntarily) separated within the collections (Battista et al. 2023), can also be included in both the trephic and sullegic factors (Figure 2) defined by Clark & Kietzke (1967).

Figure 2
Conceptual map of further anthropogenic biases that lead to the loss, even definitive, of taphonomic information and consequent biased results.

Table I
Number of bone elements collected between 2002 and 2003, and studied by Bertoni-Machado & Holz (2006) in their taphonomic analysis of the Santa Cruz’ outcrop (southernmost Brazil) in comparison with the quantity of already available material from the same outcrop, collected between 1995 and 2003, and kept in paleontological collections of Universidade Federal do Rio Grande do Sul (UFRGS), Museu de Ciências Naturais do Rio Grande do Sul (MCN/SEMA-RS), and Pontifícia Universidade Católica do Rio Grande do Sul (PUC-RS). Asterisk indicates that this amount includes the elements studied by Bertoni-Machado & Holz (2006).

	Bertoni-Machado & Holz (2006)	UFRGS	MCN/SEMA-RS	PUC-RS	Total (UFRGS+MCN+PUC)
Number of elements	53	662*	1358	283	2303
%	2.3	28.7*	59	12.3

From the sedimentary deposit degree of cementation to its fossil richness, which may or may not favor a certain ease in terms of recognizing fossil remains in the field, we have seen why the researcher’s role is important. The objective of the survey, the collection method, and the mode of transport to the laboratory are just some examples of how researchers can cause information alteration (e.g. Clark & Kietzke 1967, Wolff 1975, Flessa et al. 1992).

Once in the laboratory, another phase of fossils’ post-mortem life begins (Clark & Kietzke 1967, Whitaker & Kimmig 2020). Which fossils were prepared and studied first? Are they all ‘the same’ in our eyes? Are the spoils of collection prepared for cataloguing and archiving in the laboratory under the same conditions (i.e. they did not suffer further damage during transport) under which they were collected? Preparation techniques have evolved over time, and numerous materials that have been collected for many years prior are now prepared using previously non-existent techniques and/or tools, as well as the well-established use of the computed tomography (e.g. Ketcham & Carlson 2001, von Baczko et al. 2023, Caratelli et al. 2024; among others). This re-examination of known materials revealed new information that has changed how some materials are perceived (e.g. Brink et al. 2015, Francischini et al. 2018, Aureliano et al. 2020, Paes-Neto et al. 2021b), an aspect that is becoming increasingly important in taphonomic and biodiversity analyses (Mannion & Upchurch 2011, Fernández et al. 2017). An example of how researchers’ paradigms have changed over time can be found in Jurassic Park, a popular science fiction film that drew public interest to paleontology and technological innovations. The application of new digital techniques during the film’s production completely changed the way dinosaurs are viewed. In summary, new ways of looking at the same fossils can lead to different conclusions. However, how many ‘ancient’ fossils (collected decades ago and stored) that are not famous dinosaurs have undergone this process? As trivial as these questions may seem, they represent problems that can affect the quality of our collections and reconstructions. Furthermore, as Benton & Harper (2009, p. 71) noted in their book Introduction to Paleobiology and the Fossil Record, that there is always the risk that, once collected, the fossils will spend a long period of time sitting in a room, where they could be forgotten and eventually discarded.

Regarded as paleontology’s sister discipline, archaeology also acknowledges these questions, as both fields share many commonalities with taphonomy, particularly in terms of data acquisition techniques (e.g. Fernández-Jalvo & Marin-Monfort 2008, Pesquero et al. 2013). Archaeologists who use museum collections must deal with ancient archaeofaunas that have been imperfectly collected and documented (under modern protocols), and it may be tempting to deem these materials as worthless because of their limitations. Many zooarchaeologists, like others in their discipline, engage in ‘archaeology of archaeology’ by scrupulously studying incomprehensible field notes, complaining of uncertain or lost origins, and repairing (when possible) bones and teeth damaged by post-excavation ‘cure’ (Gifford-Gonzalez 2018). This is also applicable to paleontology and taphonomy and can constitute a serious problem for the resulting reconstructions.

Finally, another important aspect that should not be underestimated, and one that Whitaker & Kimmig (2020), among other scholars, also mentioned, concerns researchers’ financial resources. Limited funds often lead to truncated fieldwork, which can sometimes result in approximation or information paucity (e.g. lack of block orientation and approximate stratigraphic position). This raises a concerning doubt: Is ‘slow’ fieldwork/collection, that is, concentrated in a single outcrop and exhaustive of all collectible information, superior to more expeditious work that is richer in fossils but relies heavily on summary and, in some cases, incomplete information? Unfortunately, the economic aspect is becoming more influential, affecting researchers’ choices and favoring the generation of bias due to human actions.

Summarizing, during their taphonomic history, buried remains can be exhumed, undergo further modifications, and possibly be reinterred to be discovered again as fossils in a ‘round trip’ exchange between the biosphere and the lithosphere, according to Behrensmeyer & Kidwell (1985). Such remains may (or may not) have already undergone diagenetic processes, but each passage between biosphere and lithosphere, as previously mentioned, can contribute to further biological (up to taphonomic) information loss. And this can also be applied to the moment of their discovery. Therefore, why are the so-called sullegic and trephic factors proposed by Clark & Kietzke (1967) traditionally excluded from taphonomy stricto sensu or, at least, rarely considered as filters? After all, prospection and sampling (sullegic), as well as transport and curatorial activity (trephic) should represent the passage of a fossil specimen from lithosphere to anthroposphere and patrisphere (sensu Behrensmeyer et al. 2018), thus, the last ‘sieve’ figured by Behrensmeyer & Kidwell (1985) and Behrensmeyer & Miller (2012), preceding the paleobiological and paleoecological reconstruction. Our choices, starting with sampling and selection in the outcrop and the laboratory, can have important repercussions for paleoecological reconstructions. They can alter fossils’ integrity and level of fragmentation and may reduce the amount of available taphonomic information. Hence, our (in)voluntary sampling choices can be a source of both biological and taphonomic data loss.

Despite various attempts to propose systemic collection protocols and methods over the years, starting from the birth of taphonomy, paleontological collections (and associated taphonomic studies) continue to be severely distorted and strongly influenced by collector choices. Assuming the best possible conditions (ease of excavation and fossil abundance), has there been a tendency over the years to place greater emphasis on certain types of fossils? Furthermore, given that multiple people (researchers and professionals or enthusiasts and fossil hunters) can sample from the same outcrop, do results differ in terms of data quantity and quality, as well as according to the outcrop’s biodiversity and fossil richness? Is this because the researcher role is not yet fully understood as a taphonomic filter? If our post-sampling actions can contribute to modifying the state of the fossil and to the loss of useful information for the purposes of palaeoecological reconstructions, shouldn’t they be understood as a new phase?

CONCLUSIONS

Human action and selection (e.g. sampling method, mode of transportation, and preparation) can influence paleontological collections, as several authors have underlined in the last 60 years.

It is accepted that the sedimentary and fossil records are incomplete and affected by natural biases, that are independent of researchers, which generates difficulties in studying paleobiodiversity and the distribution of taxa over time. Statistical models have been and will continue to be proposed to reduce error. Even so, in addition to all the factors that can generate biases, to what extent does a paleontological collection reflect the real quantitative relationship of fauna as a function of the findings randomness? The answer may not be helpful in taphonomic terms; to propose reconstruction of a given outcrop taphonomic history, the completeness of the available data in a paleontological collection does not depend exclusively on traditional biases but rather on the number of researchers and institutions acquiring the data, the amount of time between successive samplings, what is sampled and the sampling method, and the sampler experience. Is it necessary, then, to propose a new term for the study of what happens at the interface between the biosphere, the lithosphere, and the anthroposphere and during the transition between the anthroposphere and the patrisphere (museums), as Behrensmeyer et al. (2018) have argued? Or is what we have sufficient, provided the current notion of taphonomy acknowledges the weight of researchers’ (in)voluntary choices? Alternatively, should we review our approach as researchers?

Paleontological collections cannot be viewed as static sets of information that only permit the addition of new information derived from the discovery and study of new fossils. Rather, they must be continuously reevaluated, considering all the previously discussed aspects, so that all conclusions can be verified in light of any updates.

To summarize, this discussion begs other questions that go beyond the scope of this work. However, we believe that these questions are worth considering, given that our work can determine whether important taphonomic information is lost and has the potential to generate biased paleontological collections. Is taphonomically-oriented work necessary to obtain all the relevant information? Does one have to be a taphonomist to achieve complete data collection that enhances our understanding of the genesis of the deposit and ultimately enriches the collection? The answers may not be univocal and may generate further debate; however, researchers’ actions should be viewed as another filter, since researchers themselves are a confounding factor to taphonomic analysis and palaeoecological reconstruction.

ACKNOWLEDGMENTS

This work was supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) via grants awarded to Francesco Battista (CNPq proc.: 168678/2018-7) and Cesar L. Schultz (307711/2017-0, 311251/2021-8, 406902/2022-4), as a result of their participation in the TAAS 2021 – 2^nd Workshop on Actualistic Taphonomy (Brazil, 2021), supported by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) (proc.: 8887.470844/2019-00) and CNPq (403577/2019-5). The authors are deeply grateful to the organizers of the event. The Editor-in-Chief Alexander Kellner, the editorial assistant Daniel Sant’Anna, Fernando Archuby (Universidad Nacional de La Plata, Argentina) and two other anonymous reviewers are warmly thanked. We would also like to extend our warmest gratitude to Umberto Nicosia, Eva Sacchi (Sapienza Università di Roma, Italy), Fernando Erthal, Heitor Francischini, Paula Dentzien-Dias, Tomaz Melo, André Barcelos, Matias Ritter (Universidade Federal do Rio Grande do Sul, Brazil), Voltaire Paes-Neto (Universidade Federal do Pampa, Brazil), Paolo Citton, Martina Caratelli (CONICET - Consejo Nacional de Investigaciones Científicas y Técnicas, Argentine), Kenneth Angielczyk (Field Museum of Natural History, Chicago, USA), Flávio Pretto (Universidade Federal de Santa Maria, Santa Maria, Brazil), Ana Maria Ribeiro, Jorge Ferigolo (Museu de Ciências Naturais/SEMA, Porto Alegre, Brazil), and all the TaphCon2020 (November 2020) and TAAS 2021 (July 2021) participants for their helpful comments on the topic of this paper. Anna K. Behrensmeyer (National Museum of Natural History, Washington D.C., USA) and Yolanda Fernández-Jalvo (Museo Nacional de Ciencias Naturales, Madrid, Spain) are strongly acknowledged for interesting discussion on the topic which took place during the 6^th ICAZ-9^th TAPHOS meeting (June 2022).

REFERENCES

• ABDALA F, RIBEIRO AM & SCHULTZ CL. 2001. A rich cynodont fauna of Santa Cruz do Sul, Santa Maria Formation (Middle-Late Triassic), southern Brazil. Neues Jahrb Geol P M 11: 669-687.
• ALROY J. 2010. Geographical, environmental and intrinsic biotic controls on phanerozoic marine diversification. Palaeontology 53: 1211-1235.
• ALROY J ET AL. 2001. Effects of sampling standardization on estimates of Phanerozoic marine diversification. Proc Natl Acad Sci 98: 6261-6266.
• ANDREWS P & COOK J. 1985. Natural modifications to bones in a temperate setting. Man (New Series) 20: 675-691.
• AURELIANO T, GHILARDI AM, SILVA-JUNIOR JCG, MARTINELLI AG, RIBEIRO LCB, MARINHO T, FERNANDES MA, RICARDI-BRANCO F & SANDER PM. 2020. Influence of taphonomy on histological evidence for vertebral pneumaticity in an upper cretaceous titanosaur from South America. Cret Res 108: 104337.
• AZEVEDO SAK, SCHULTZ CL & BARBERENA MC. 1990. Novas evidências bioestratigráficas e paleoecológicas na análise da evolução explosiva dos rincossauros do Triássico. Paula-Coutiana 4: 23-31.
• BAREINBOIM E, TIAN J & PEARL J. 2014. Recovering from selection bias in causal and statistical inference. In: Brodley CE & Stone P (Eds), Proceedings of the AAAI Conference on Artificial Intelligence 28(1), AAAI Press, Menlo Park, CA, p. 2410-2416.
• BARTON LM, DUVAL EJ, STROBERG E, GHOSH S & MUKHOPADHYAY S. 2020. COVID-19 autopsies, Oklahoma, USA. J Clin Pathol 153: 725-733.
• BATTISTA F, FONDEVILLA V, RAZZOLINI N, VILA B, SELLÉS AG & GALOBART À. 2014. The L’Espinau locality (Upper Maastrichtian, NE Iberian Peninsula): first data from a new hadrosauroid bonebed in Europe. In: Marmi J, Oms O, Vila B, Galobart À, Estrada R & Dinarès-Turell J (Eds), Reconstructing the Terrestrial End-cretaceous Paleoenvironments in Europe (Tremp, 2014): Paleontologia i Evolució, Memòria Especial, 7, p. 57.
• BATTISTA F, RIBEIRO AM, ERTHAL F & SCHULTZ CL. 2023. Post-collection taphonomy, sampling effects and the role of the collector in palaeontological collections: A case study from an early Late Triassic bone accumulation in southernmost Brazil. Acta Palaeontol Pol 68: 359-372.
• BATTISTA F, RIBEIRO AM, MELO TP & SCHULTZ CL. 2024. A reinterpretation of taphonomy and palaeoecology of the early Carnian Santacruzodon Assemblage Zone type-locality from the Santa Maria Supersequence, Brazil. Palaeogeogr Palaeoclimatol Palaeoecol 639: 112080.
• BEHRENSMEYER AK. 1975. The taphonomy and paleoecology of Plio-Pleistocene vertebrate assemblages east of Lake Rudolf, Kenya. Bull Mus Comp Zool 146: 473-578.
• BEHRENSMEYER AK, DENYS C & BRUGAL J-P. 2018. What is taphonomy and what is not? Hist Biol 30: 718-719.
• BEHRENSMEYER AK & KIDWELL SM. 1985. Taphonomy’s contributions to paleobiology. Paleobiology 11: 105-119.
• BEHRENSMEYER AK, KIDWELL SM & GASTALDO RA. 2000. Taphonomy and paleobiology. Paleobiology 26: 103-147.
• BEHRENSMEYER AK & MILLER JA. 2012. Building links between ecology and paleontology using taphonomic studies of recent vertebrate communities. In: Louys J (Ed), Paleontology in Ecology and Conservation, Springer-Verlag, Berlin and Heidelberg, p. 69-91.
• BENTON MJ & HARPER DAT. 2009. Introduction to Paleobiology and the Fossil Record. Wiley-Blackwell, Chichester, 608 p.
• BENTON MJ, WILLIS MA & HITCHIN R. 2000. Quality of the fossil record through time. Nature 403: 534-537.
• BERK RA. 1983. An introduction to sample selection bias in sociological data. Am Sociol Rev 48: 386-398.
• BERTONI-MACHADO C & HOLZ M. 2006. Biogenic fossil concentration in fluvial settings: An example of a cynodont taphocoenosis from the Middle Triassic of southern Brazil. Rev Bras Paleontolog 9: 273-282.
• BIRO PA & DINGEMANSE NJ. 2009. Sampling bias resulting from animal personality. Trends Ecol Evol 24: 66-67.
• BOURGUIGNON F, FOURNIER M & GURGAND M. 2007. Selection bias corrections based on the multinomial logit model: Monte Carlo comparisons. J Econ Surv 21: 174-205.
• BOWLEY AL. 1897. Relations between the accuracy of an average and that of its constituent parts. J R Stat Soc 60: 855-866.
• BREEN R. 2015. Selection bias, Statistics of. In: Wright JD (Ed), International Encyclopedia of the Social & Behavioral Sciences, 2nd Edition: Elsevier, p. 414-418.
• BRESLOW NE. 2003. Are statistical contributions to medicine undervalued? Biometrics 59: 1-8.
• BRINK KS, MADDIN HC, EVANS DC & REISZ RR. 2015. Re-evaluation of the historic Canadian fossil Bathygnathus borealis from the Early Permian of Prince Edward Island. Can J Earth Sci 52: 1109-1120.
• CARATELLI M, CITTON P, ARCHUBY FM & PIGNATTI J. 2024. Paleoecology of the foraminifer Acruliammina longa (Tappan, 1940) from the upper Hauterivian of the Neuquén Basin (northern Patagonia, Argentina). Cret Res 158: 105857.
• CLARK J & KIETZKE KK. 1967. Paleoecology of the lower Nodular Zone, Brule Formation, in the Big Badlands of South Dakota. In: Clark J, Beerbower JR & Kietzke KK (Eds), Oligocene Sedimentation, Stratigraphy, Paleoecology and Paleoclimatology in the Big Badlands of South Dakota: Fieldiana Geology Memoirs, 5, p. 111-155.
• COOPER N, BOND AL, DAVIS JL, PORTELA MIGUEZ R, TOMSETT L & HELGEN KM. 2019. Sex biases in bird and mammal natural history collections. P R Soc B 286: 20192025.
• CORTES C, MOHRI M, RILEY M & ROSTAMIZADEH A. 2008. Sample selection bias correction theory. In: Freund Y, Györfi L, Turán G & Zeugmann T (Eds), Algorithmic Learning Theory, ALT 2008, Lecture Notes in Computer Science, 5254, Springer-Verlag, Berlin and Heidelberg, p. 38-53.
• CRAMPTON JS, BEU AG, COOPER RA, JONES CM, MARSHALL B & MAXWELL PA. 2003. Estimating the rock volume bias in palaeodiversity studies. Science 301: 358-360.
• DA-ROSA ÁAS. 2004. Sítios fossilíferos de Santa Maria, RS, Brasil. Cienc Nat 26: 75-90.
• DARWIN C. 1859. On the Origin of Species. John Murray, London, 516 p.
• DEAN CD, CHIARENZA AA & MAIDMENT SCR. 2020. Formation binning: a new method for increased temporal resolution in regional studies, applied to the Late Cretaceous dinosaur fossil record of North America. Palaeontology 63: 881-901.
• DESOJO JB & EZCURRA MD. 2011. A reappraisal of the taxonomic status of Aetosauroides (Archosauria, Aetosauria) specimens from the Late Triassic of South America and their proposed synonymy with Stagonolepis. J Vertebr Paleontol 31: 596-609.
• DIBBLE HL, RACZEK TP & MCPHERRON SP. 2005. Excavator bias at the site of Pech de l’Azé IV, France. J Field Archaeol 30: 317-328.
• DU Z, XU X, WU Y, WANG L, COWLING BJ & ANCEL MEYERS L. 2020. Serial interval of COVID-19 among publicly reported confirmed cases. Emerg Infect Dis 26: 1341-1343.
• EFREMOV IA. 1940. Taphonomy: new branch of paleontology. Pan-American Geologist 74: 81-93.
• FAITH JT & LYMAN RL. 2019. Paleozoology and Paleoenvironments: Fundamentals, Assumptions, Techniques. Cambridge University Press, Cambridge, 398 p.
• FARA E, SARAIVA AAF, CAMPOS DA, MOREIRA JKR, SIEBRA DC & KELLNER AWA. 2005. Controlled excavations in the Romualdo Member of the Santana Formation (Early Cretaceous, Araripe Basin, northeastern Brazil): stratigraphic, palaeoenvironmental and palaeoecological implications. Palaeogeogr Palaeoclimatol Palaeoecol 218: 145-160.
• FERNÁNDEZ FJ, MONTALVO CI, FERNÁNDEZ-JALVO Y, ANDREWS P & LÓPEZ JM. 2017. A re-evaluation of the taphonomic methodology for the study of small mammal fossil assemblages of South America. Quaternary Sci Rev 155: 37-49.
• FERNÁNDEZ-JALVO Y & MARIN-MONFORT MD. 2008. Experimental taphonomy in museums: Preparation protocols for skeletons and fossil vertebrates under the scanning electron microscopy. Geobios 41: 157-181.
• FERNÁNDEZ-LÓPEZ SR. 1991. Taphonomic concepts for a theoretical biochronology. Spanish J Paleontol 6: 37-49.
• FIORILLO AR. 1989. An experimental study of trampling: implications for the fossil record. In: Bonnichsen R & Sorg MH (Eds), Bone Modification, Center for the Study of the First Americans, University of Maine, Orono, p. 61-72.
• FLESSA KW, KOWALEWSKI M & WALKER SE. 1992. Post-collection taphonomy: shell destruction and the Chevrolet. Palaios 7: 553-554.
• FONDEVILLA V ET AL. 2017. Geology and taphonomy of the L’Espinau dinosaur bonebed, a singular lagoonal site from the Maastrichtian of South-Central Pyrenees. Sediment Geol 355: 75-92.
• FRANCISCHINI H, DENTZIEN-DIAS P & SCHULTZ CL. 2018. A fresh look at ancient dungs: the Brazilian Triassic coprolites revisited. Lethaia 51 389-405.
• GIFFORD-GONZALEZ D. 2018. An Introduction to Zooarchaeology. Springer, Cham, 604 p.
• GRADSTEIN FM, OGG JG, SCHMITZ M & OGG G. 2020. Geologic Time Scale 2020. Elsevier, 1390 p.
• HECKMAN JJ. 1976. The common structure of statistical models of truncation, sample selection and limited dependent variables and a simple estimator for such models. Ann Econ Soc Meas 5: 475-492.
• HECKMAN JJ. 1979. Sample selection bias as a specification error. Econometrica 47: 153-161.
• HERNÁN MA, HERNÁNDEZ-DÍAZ S & ROBINS JM. 2004. A structural approach to selection bias. Epidemiology 15: 615-625.
• HIJMANS RJ, GARRETT KA, HUAMÁN Z, ZHANG DP, SCHREUDER M & BONIERBALE M. 2000. Assessing the geographic representativeness of genebank collections: the case of Bolivian wild potatoes. Conserv Biol 14: 1755-1765.
• HOLZ M & BARBERENA MC. 1994. Taphonomy of the south Brazilian Triassic paleoherpetofauna: pattern of death, transport and burial. Palaeogeogr Palaeoclimatol Palaeoecol 107: 179-197.
• KELLNER AWA & CAMPOS DA. 1999. Vertebrate paleontology in Brazil – a review. Episodes 22: 238-251.
• KERBER L, ROESE-MIRON L, BUBADUÉ JM & MARTINELLI AG. 2023. Endocranial anatomy of the early prozostrodonts (Eucynodontia: Probainognathia) and the neurosensory evolution in mammal forerunners. Anat Rec 307: 1442-1473.
• KETCHAM RA & CARLSON WD. 2001. Acquisition, optimization and interpretation of X-ray computed tomographic imagery: applications to the geosciences. Comput and Geosci 27: 381-400.
• KIDWELL SM. 2002. Mesh-size effects on the ecological fidelity of death assemblages: a meta-analysis of molluscan live-dead studies. Geobios 35: 107-119.
• KIDWELL SM & HOLLAND SM. 2002. The quality of the fossil record: implications for evolutionary analyses. Annu Rev Ecol Syst 33: 561-588.
• KOBAYASHI T ET AL. 2020. Communicating the risk of death from novel coronavirus disease (COVID-19). J Clin Med 9: 580. https://doi.org/10.3390/jcm9020580.
  » https://doi.org/10.3390/jcm9020580
• KONG W-H, LI Y, PENG M-W, KONG D-G, YANG X-B, WANG L & LIU M-Q. 2020. SARS-CoV-2 detection in patients with influenza-like illness. Nat Microbiol 5: 675-678.
• KORTH WW. 1979. Taphonomy of microvertebrate fossil assemblages. Ann Carnegie Mus 48: 235-285.
• KOWALEWSKI M & HOFFMEISTER AP. 2003. Sieves and fossils: effects of mesh size on paleontological patterns. Palaios 18: 460-469.
• KRONE IW, MAGOULICK KM & YOHLER RM. 2024. All the Earth will not remember: how geographic gaps structure the record of diversity and extinction. Paleobiology 50: 214-225.
• LANGER MC, RAMEZANI J & DA-ROSA ÁAS. 2018. U-Pb age constraints on dinosaur rise from south Brazil. Gondwana Res 57: 133-140.
• LAWRENCE DR. 1968. Taphonomy and information losses in fossil communities. Geol Soc Am Bull 79: 1315-1330.
• LAWRENCE DR. 1971. The nature and structure of paleoecology. J Paleontol 45: 593-607.
• LEIDY J. 1869. The extinct mammalian fauna of Dakota and Nebraska: Including an account of some allied forms from other localities, together with a synopsis of the mammalian remains of North America. J Acad Nat Sci Phila, 7, 472 p.
• LEITÃO PJ, MOREIRA F & OSBORNE PE. 2011. Effects of geographical data sampling bias on habitat models of species distributions: a case study with steppe birds in southern Portugal. Int J Geogr Inf Sci 25: 439-454.
• LUKIC-WALTHER M, BROCKLEHURST N, KAMMERER CF & FRÖBISCH J. 2019. Diversity patterns of nonmammalian cynodonts (Synapsida, Therapsida) and the impact of taxonomic practice and research history on diversity estimates. Paleobiology 45: 56-69.
• LYMAN RL. 1994. Vertebrate Taphonomy. Cambridge University Press, Cambridge, 524 p.
• LYMAN RL. 2008. Quantitative Paleozoology. Cambridge University Press, Cambridge, 348 p.
• MANNION PD, BENSON RBJ & BUTLER RJ. 2013. Vertebrate palaeobiodiversity patterns and the impact of sampling bias. Palaeogeogr Palaeoclimatol Palaeoecol 372: 1-4.
• MANNION PD & UPCHURCH P. 2011. A re-evaluation of the ‘mid-Cretaceous sauropod hiatus’ and the impact of uneven sampling of the fossil record on patterns of regional dinosaur extinction. Palaeogeogr Palaeoclimatol Palaeoecol 299: 529-540.
• MARRAMÀ G, BANNIKOV AF, TYLER JC, ZORZIN R & CARNEVALE G. 2016. Controlled excavations in the Pesciara and Monte Postale sites provide new insights about the palaeoecology and taphonomy of the fish assemblages of the Eocene Bolca Konservat-Lagerstätte, Italy. Palaeogeogr Palaeoclimatol Palaeoecol 454: 228-245.
• MCGOWAN AJ & SMITH AB. 2011. Comparing the Geological and Fossil Records: Implications for Biodiversity Studies. Geological Society, London, Special Publications 358, 247 p.
• MCKENNA MC. 1962. Collecting small fossils by washing and screening. Curator 5: 221-235.
• MELO TP, RIBEIRO AM, MARTINELLI AG & SOARES MB. 2019. Early evidence of molariform hypsodonty in a Triassic stem-mammal. Nat Commun 10: 2841. https://doi.org/10.1038/s41467-019-10719-7.
  » https://doi.org/10.1038/s41467-019-10719-7
• MICHELANGELI M, WONG BBM & CHAPPLE DG. 2016. It’s a trap: sampling bias due to animal personality is not always inevitable. Behav Ecol 27: 62-67.
• MILLER J. 2004. Earliest known use of some of the words in mathematics. https://jeff560.tripod.com/b.html Accessed on April 2021.
  » https://jeff560.tripod.com/b.html
• MÜLLER AH. 1963. Lehrbuch der Paläozoologie. Band 1: Allgemeine Grundlagen. Gustav Fischer Verlag, Jena, 387 p.
• NISHIURA H, LINTON NM & AKHMETZHANOV AR. 2020. Serial interval of novel coronavirus (COVID-19) infections. Int J Infect Dis 93: 284-286.
• OLSON EC. 1980. Taphonomy: its history and role in community evolution. In: Behrensmeyer AK & Hill AP (Eds), Fossils in the Making: Vertebrate Taphonomy and Paleoecology, University of Chicago Press, Chicago, p. 6-19.
• PAES-NETO VD, DESOJO JB, BRUST ACB, SCHULTZ CL, DA-ROSA ÁAS & SOARES MB. 2021a. Intraspecific variation in the axial skeleton of Aetosauroides scagliai (Archosauria: Aetosauria) and its implications for the aetosaur diversity of the Late Triassic of Brazil. An Acad Bras Cienc 93: e20201239. https://doi.org/10.1590/0001-3765202120201239.
  » https://doi.org/10.1590/0001-3765202120201239
• PAES-NETO VD, DESOJO JB, BRUST ACB, RIBEIRO AM, SCHULTZ CL & SOARES MB. 2021b. The first braincase of the basal aetosaur Aetosauroides scagliai (Archosauria: Pseudosuchia) from the Upper Triassic of Brazil. J Vertebr Paleontol 41: e1928681. https://doi.org/10.1080/02724634.2021.1928681.
  » https://doi.org/10.1080/02724634.2021.1928681
• PEARL J. 2009. Causality: models, reasoning, and inference - 2nd Edition. Cambridge University Press, Cambridge, 464 p.
• PESQUERO MD, ALCALÁ L & FERNÁNDEZ-JALVO Y. 2013. Taphonomy of the reference Miocene vertebrate mammal site of Cerro de la Garita (Concud, Teruel, Spain): Lethaia 46: 378-398.
• PETTI FM & FALORNI P. 2007. Rosso Ammonitico. In: Cita MB, Abbate E, Aldighieri B, Balini M, Conti MA, Falorni P, Germani D, Groppelli G, Manetti P & Petti FM (Eds), Carta Geologica d’Italia 1: 50.000, Catalogo delle formazioni - Unità tradizionali (1) Fascicolo VI. Quaderni del Servizio Geologico d’Italia, Serie III, v. 7, p. 140-149.
• PICKERING TR & BRAIN CK. 2010. What taphonomically oriented research at Swartkrans caves reveals about early hominid behavior. J Taphonomy 8: 215-232.
• PIKE EM. 1993. Amber taphonomy and collecting biases. Palaios 8: 411-419.
• PONDER WF, CARTER GA, FLEMONS P & CHAPMAN RR. 2001. Evaluation of museum collection data for use in biodiversity assessment. Conserv Biol 15: 648-657.
• PYENSON ND. 2011. The high fidelity of the cetacean stranding record: insights into measuring diversity by integrating taphonomy and macroecology. P R Soc B 278: 3608-3616.
• QIN J. 2017. Biased Sampling, Over-identified Parameter Problems and Beyond. Springer Singapore, Singapore, 624 p.
• RAUP DM. 1972. Taxonomic diversity during the Phanerozoic. Science 177: 1065-1071.
• RÉALE D, READER SM, SOL D, MCDOUGALL PT & DINGEMANSE NJ. 2007. Integrating animal temperament within ecology and evolution. Biol Rev 82: 291-318.
• REDDY S & DÁVALOS LM. 2003. Geographical sampling bias and its implications for conservation priorities in Africa. J Biogeogr 30: 1719-1727.
• REED CA. 1963. Osteo-archaeology. In: Brothwell D & Higgs ES (Eds), Science in Archaeology, Basic Books, New York, p. 204-216.
• RITTER MN, FRANCISCHINI H, KUHN LA, DA LUZ NC, MICHELS FH, DE MORAIS ALM, PAIM PAV, XAVIER PLA & DE FRANCESCO CG. 2016. El sesgo del operador en la replicabilidad de los estudios tafonómicos comparativos. Rev Bras Paleontolog 19: 449-464.
• ROTHFUS TA. 2004. How many taphonomists spoil the data? Multiple operators in taphofacies studies. Palaios 19: 514-519.
• ROUSSON V, ROSSELET PC & PACCAUD F. 2017. Selection bias in ecological studies. Public Health 153: 103-106.
• SCHULTZ CL, MARTINELLI AG, SOARES MB, PINHEIRO FL, KERBER L, HORN BLD, PRETTO FA, MÜLLER RT & MELO TP. 2020. Triassic faunal successions of the Paraná Basin, southern Brazil. J S Am Earth Sci 104: 102846. https://doi.org/10.1016/j.jsames.2020.102846.
  » https://doi.org/10.1016/j.jsames.2020.102846
• SEPKOSKI JR JJ. 1975. Stratigraphic bias in the analysis of taxonomic survivorship. Paleobiology 1: 343-355.
• SEPKOSKI JR JJ, BAMBACH RK, RAUP DM & VALENTINE JW. 1981. Phanerozoic marine diversity and the fossil record. Nature 293: 435-437.
• SHELDRICK MC. 1976. Trends in the strandings of Cetacea on the British Coasts 1913-72. Mammal Rev 6: 15-23.
• SHIPMAN P. 1981. Life History of a Fossil: An Introduction to Taphonomy and Paleoecology. Harvard University Press, Cambridge, 222 p.
• SHOTWELL JA. 1955. An approach to the paleoecology of mammals. Ecology 36: 327-337.
• SIGNOR PW. 1978. Species richness in the Phanerozoic: an investigation of sampling effects. Paleobiology 4: 394-406.
• SIGNOR PW & LIPPS JH. 1982. Sampling bias, gradual extinction patterns and catastrophes in the fossil record. In: Silver LT & Schultz PH (Eds), Geological Implications of Impacts of Large Asteroids and Comets on the Earth. Geol S Am S 190: 291-296.
• SKELTON L. 1984. Kansas skirmishes in the Cope/Marsh war. Earth Sci Hist 3: 117-122.
• SMITH AB & MCGOWAN AJ. 2007. The shape of the Phanerozoic marine palaeodiversity curve: how much can be predicted from the sedimentary rock record of western Europe? Palaeontology 50: 765-774.
• SMITH AB & MCGOWAN AJ. 2011. The ties linking rock and fossil records and why they are important for palaeobiodiversity studies. In: McGowan AJ & Smith AB (Eds), Comparing the Geological and Fossil Records: Implications for Biodiversity Studies, Geological Society, London, Special Publications 358, p. 1-7.
• STUBER EF, ARAYA-AJOY YG, MATHOT KJ, MUTZEL A, NICOLAUS M, WIJMENGA JJ, MUELLER JC & DINGEMANSE NJ. 2013. Slow explorers take less risk: a problem of sampling bias in ecological studies. Behav Ecol 24: 1092-1098.
• TANG CM. 2000. Ugly fossil syndrome. Palaios 15: 175-176.
• THOMAS J. 2010. Collections universitaires et Paléontologie. La Lettre de l’OCIM 129: 28-35.
• TOOTS H. 1965. Random orientation of fossils and its significance. Rocky Mt Geol 4: 59-62.
• VERMEIJ GJ & LEIGHTON LR. 2003. Does global diversity mean anything? Paleobiology 29: 3-7.
• VON BACZKO MB, CARDILLO A, ULLOA-GUAIQUIN K, DESOJO JB & PAULINA-CARABAJAL A. 2023. Paleoneurology and neuroanatomical notes of the South American rauisuchian Saurosuchus galilei from the Ischigualasto Formation, San Juan, Argentina. Anat Rec 307: 1239-1253.
• VOORHIES MR. 1969. Taphonomy and population dynamics of an early Pliocene vertebrate fauna, Knox County, Nebraska. Contrib Geol 1: 1-69.
• WAKEFIELD J. 2004. Ecological inference for 2 × 2 tables (with discussion). J R Statist Soc A 167: 385-445.
• WALLACE DR. 1999. The Bonehunters’ Revenge: Dinosaurs, Greed, and the Greatest Scientific Feud of the Gilded Age. Houghton Mifflin Company, Boston, 366 p.
• WEIGELT J. 1989. Recent Vertebrate Carcasses and Their Paleobiological Implications. English translation of original Weigelt’s 1927, by J. Schaefer, University of Chicago Press, Chicago, 204 p.
• WHITAKER AF & KIMMIG J. 2020. Anthropologically introduced biases in natural history collections, with a case study on the invertebrate paleontology collections from the middle Cambrian Spence Shale Lagerstätte. Palaeontol Electron 23: a58. https://doi.org/10.26879/1106.
  » https://doi.org/10.26879/1106
• WINSHIP C & MARE RD. 1992. Models for sample selection bias. Annu Rev Sociol 18: 327-350.
• WOLFF RG. 1973. Hydrodynamic sorting and ecology of a Pleistocene mammalian assemblage from California (U.S.A.). Palaeogeogr Palaeoclimatol Palaeoecol 13: 91-101.
• WOLFF RG. 1975. Sampling and sample size in ecological analyses of fossil mammals. Paleobiology 1: 195-204.
• WORLD HEALTH ORGANIZATION. 2020. WHO Director-General’s opening remarks at the media briefing on COVID-19 - 11 March 2020. https://www.who.int/director-general/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020 Accessed on May 2020.
  » https://www.who.int/director-general/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020
• WYNANTS L ET AL. 2020. Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal. BMJ 369: m1328. https://doi.org/10.1136/bmj.m1328.
  » https://doi.org/10.1136/bmj.m1328
• ZAR JH. 2010. Biostatistical Analysis - 5th Edition. Pearson, Upper Saddle River, 960 p.
• ZHANG G & ZHU A-X. 2018. The representativeness and spatial bias of volunteered geographic information: a review. Ann GIS 24: 151-162.
• ZHU N ET AL. 2020. A novel coronavirus from patients with pneumonia in China, 2019. N Engl J Med 382: 727-733.

Publication Dates

• Publication in this collection
  22 Nov 2024
• Date of issue
  2024

History

• Received
  17 Nov 2023
• Accepted
  11 Aug 2024