Abstract

The Pramollo Basin (Italy-Austria) is one of the richest body and trace fossil sites of the Alps, and exhibits a well-preserved Permian–Carboniferous fluvio-deltaic to marginal-marine sedimentary succession. Despite the exceptionally abundant and well-preserved ichnological heritage, the trace fossils of the Pramollo Basin are not well studied, particularly those of Permian units.

This study focuses on the ichnofauna of the Val Dolce Formation (Permian; partly Asselian to partly Sakmarian), with the goal of documenting its ichnological heritage and reconstructing its paleoenvironment. These research questions are addressed by applying network theory, an emerging field of complexity science that focuses on web-like systems made of interconnected entities. An ichnological system can be seen as a set of interlinked ichnotaxa, the topology of which depends on the organism-environment interactions. In addition, traditional paleontological and sedimentological observations are used to reconstruct the paleoenvironment.

The following ichnotaxa are documented from the Val Dolce Formation: Archaeonassa isp., Curvolithus simplex, Cylindrichnus isp., Helminthoidichnites tenuis, Nereites missouriensis, Planolites isp., Phymatoderma isp., Pramollichnus pastae, Psammichnites plummeri, Taenidium isp., and Zoophycos isp. Network analysis indicates that the Val Dolce ichnological system is structured, with ichnotaxa organized in environment-driven ichnoassociations: Cylindrichnus-Planolites (proximal delta front), Phymatoderma-Zoophycos (prodelta with dysoxic porewaters), Cylindrichnus-Helminthoidichnites-Curvolithus-Zoophycos (distal delta front–proximal prodelta), and Helminthoidichnites-Taenidium-Curvolithus-Nereites-Zoophycos (prodelta). Furthermore, the delta front–prodelta gradient is accompanied by increasing bioturbation intensity and diversity, reflecting the decreasing intensity of major environmental stressors (hydrodynamics, freshwater input, turbidity). Centrality measures of network analysis allow the topological position of traces to be discerned within the studied system, detecting the paleoenvironmental resolution of individual ichnotaxa.

As intersections of sets can be described by networks, the studied ichnoassociations can be considered as occupying intersecting behavioral niches. In analogy with the concept of a Hutchinsonian niche, an ichnotaxon’s niche exists in a multidimensional abstract space defined by environmental parameters, which are expressed as spatial variables in the paleolandscape. Consequently, ichnoassociations are not just association patterns, but represent spatial, environmental, and topological entities. This approach allows the reconstitution of spatial relationships between the geographical ranges of ichnotaxa and ichnoassociations, providing information on the physical arrangement of different subenvironments, that is, the structure of the paleoenvironment.

INTRODUCTION

“In addition to that branch of geometry which is concerned with magnitudes, and which has always received the greatest attention, there is another branch, previously almost unknown, which Leibniz first mentioned, calling it the geometry of position”

(Euler, 1736, Solutio problematis ad geometriam situs pertinentis, translation by Biggs et al., 1976, p. 3–8).

The process of describing real-world phenomena as networks, sets of nodes connected by links, is rooted in the eighteenth century, when Euler (1736) used an abstract representation of the city of Königsberg (Lewis, 2009; Barabási, 2002). Since then, network theory has emerged as an efficient framework for describing and analyzing complex systems, from food webs to the Internet. Many natural and artificial systems can be seen as web-like groups of interconnected entities, the topology of which can be analyzed using graphical and mathematical tools. Consequently, network theory finds application in the most disparate disciplines, including biology, computer science, and physics (Réka and Barbási, 2002). The social sciences in particular have greatly benefited from the network approach, according to which individuals are embedded in webs of interpersonal relationships (Borgatti et al., 2009). The corresponding graphs have been considered “geographies of emotion” because they map social relationships, while a vast set of analytical tools allows researchers to explore network topology, recognize groupings, and discern critically important individuals (Wassermann and Faust, 1994).

Paralleling sociology, network theory has inspired a new approach in the study of fossil and modern traces (ichnology) (Baucon and Felletti, 2013b). Because traces such as burrows, footprints, and borings are manifestations of biological behavior (Bromley, 1996; Seilacher, 2007; Buatois and Mángano, 2011; Knaust, 2012a; Baucon et al., 2012), ichnosites are behavioral networks recording the biological response to the surrounding environment. Ichnological networks are therefore “geographies of behavior” that can shed light on organism-environment relationships in fossil and modern ecosystems.

In light of these assumptions, this paper aims to apply network theory for analyzing the ecosystem of the Pramollo Basin (Italy-Austria), known as an ichnolagerstätte for the exceptional quality and quantity of Paleozoic trace fossils (Baucon and Carvalho, 2008). In particular, the focus is on the ichnology of the Val Dolce Formation (lower Permian), which has been comparatively less studied with respect to the underlying Carboniferous units (Fig. 1). In this regard, systematic studies on the trace fossils of the Val Dolce Formation and their paleoenvironmental significance are lacking, with particular regard to the physical arrangement of the environment (i.e., environment structure). Specifically, three major questions are posed. (1) What are the trace fossils of the Val Dolce Formation? (2) What was the paleoenvironment of the Val Dolce Formation? (3) How was the paleoenvironment structured?

GEOLOGICAL FRAMEWORK

The Pramollo Basin is located in the Carnic Alps of Italy and Austria and exhibits a Carboniferous to Permian succession (partly Moscovian–Artinskian) referred to as the Pontebba Supergroup (Fig. 2), which unconformably overlies the lower Carboniferous basement (Venturini, 1990, 1991, 2002a). The evolution of the Pramollo Basin is framed within the transtensional tectonic regime that characterized the southern Alps during the late Paleozoic, opening as a pull-apart basin from the Moscovian to the Kasimovian (Venturini et al., 1982; Venturini, 1983; Vai, 1991). The corresponding fault-delimited trough was initially filled with coarse conglomerates (Bombaso Formation), followed by a prevailing siliciclastic influx with minor limestone deposits (Pramollo Group). These units are overlain by carbonates and minor siliciclastic deposits (Rattendorf and Trogkofel Groups). According to petrographic and sedimentological studies (Fontana and Venturini, 1982; Venturini, 1990, 1991), the source area of the Bombaso Formation was dominated by carbonates and siliciclastics, whereas the Pramollo and Rattendorf Groups were related to the erosion of a phyllitic basement located northwest of the Pramollo Basin.

Overall, the Pontebba Supergroup reflects fluvio-deltaic and marginal-marine deposition strongly controlled by tectonics (Venturini, 1983). Consequently, the Pramollo Basin is one of the few peri-Tethyan sites recording marine and deltaic conditions (Vai and Venturini, 1997; Venturini, 2002b).

This peculiarity is accompanied by a rich body fossil heritage that has been documented since the nineteenth century (Venturini, 1991). The fossil record of the Pontebba Supergroup includes brachiopods, trilobites, cartilaginous fishes, corals, bryozoans, fusulinids, crinoids, and a well-preserved paleoflora (Venturini, 2006). Although the trace fossil record of the Pramollo Basin has been known since the 1960s (Selli, 1963; Vai et al., 1979; Barbiero et al., 1990; Mietto et al., 1985; Conti et al., 1991), the paleontological site has been qualified only recently as anichnolagerstätte for its exceptional trace fossil content (Baucon and Carvalho, 2008). For these reasons, Pramollo and its surroundings constitute one of the most spectacular trace and body fossil sites of the Alps (Venturini, 2006; Baucon and Carvalho, 2008).

This paper focuses on the Val Dolce Formation, which received less ichnological attention with respect to the Pramollo Group. There is general consensus that the Permian-Carboniferous boundary is near the base of the Val Dolce Formation, which is mostly of Asselian age (Vai and Venturini, 1997; Forke, 2002); some extend it to the basal Sakmarian (Forke, 2002). The Val Dolce Formation, known as Grenzland Formation to German researchers (Forke, 2002), is well exposed in the vicinity of Cason di Lanza, where it is represented by siliciclastic deposits with minor bioclastic limestones. This unit is organized in shallowing-upward cycles that commonly comprise brachiopod-rich pelites, trough, planar, and hummocky stratified sandstones, and quartz conglomerates (Vai and Venturini, 1997; Venturini, 1990; Fig. 3). Table 1 includes a general facies description; more detailed sedimentological analyses are available elsewhere (Venturini, 1991; Vai and Venturini, 1997; Forke, 2002). In accord with the general setting of the Pontebba Supergroup, the Val Dolce Formation represents fluvio-deltaic to marginal-marine deposition (Venturini, 1990).

MATERIALS AND METHODS

A field survey was done in 2012 to define the ichnofauna of the Val Dolce Formation (Fig. 1). Based on optimal outcrop quality, the stream intersecting Pian di Lanza (Fig. 1) was selected for erecting a standard stratigraphic section in which trace fossil distribution was noted (Fig. 4). This section records the lower and middle part of the Val Dolce Formation, although the contact with the underlying lower Pseudoschwagerina Formation is not exposed. Nevertheless, comparison with previously studied stratigraphic sections (Forke, 2002; Venturini, 1990; Schönlaub and Forke, 2007) shows that the Pian di Lanza section is one of the most complete exposures of the Val Dolce Formation. Because the outcrop quality of the Pian di Lanza section does not allow bed-by-bed observation of trace fossils, the sampling unit of this study is the bed set, i.e., a group of superposed beds with similar characteristics (McKee and Weir, 1953; Campbell, 1967; Reineck and Singh, 1986).

This data set represents the source for drawing the ichnonetwork, following the methodology described in Baucon and Felletti (2013b) and Baucon et al. (2014). Specifically, source data are the presence or absence of each ichnotaxon per sampling unit. Following Pickerill and Brenchley (1975), the source data for drawing an ichnonetwork therefore consist of a set of trace fossil assemblages. In addition to the presence or absence of each ichnotaxon, bioturbation intensity (referred to as BPBI; Miller and Smail, 1997), body fossils, lithology, and physical sedimentary structures have been described for each sampling unit. Sedimentological features have been synthesized as facies; see Table 1 for facies codes.

The process for drawing an ichnonetwork is based on the association relationships between pairs of ichnotaxa. The idea is to represent ichnotaxa as nodes, and connect those pairs that occur together. The strength of each association relationship is described by link weight, which corresponds to the Jaccard index (Hammer and Harper, 2006; Jaccard, 1901, 1912). The Jaccard index of ichnotaxon A and ichnotaxon B is the proportion between the number of associated occurrences and the number of samples with one ichnotaxon or the other (Hammer and Harper, 2006; Jaccard, 1901, 1912): J(A, B) = number of samples with A and B/number of samples with A or B.

In other words, the Jaccard index can be seen as a probabilistic measure of association, indicating the probability of finding both ichnogenera if at least one of them is present in the sample (Baucon and Felletti, 2013b; Hammer and Harper, 2006; Jaccard, 1901, 1912). For example, the Jaccard index equals 0 if the traces A and B are never found together; it equals 1 if A and B co-occur in all samples.

In practical terms, free open-source software allows the derivation of an ichnonetwork from stratigraphic data in three steps (Fig. 5).

  1. Store stratigraphic data into a spreadsheet (i.e., LibreOffice Calc) with ichnotaxa in columns and samples in rows.

  2. Calculate the Jaccard index for each ichnotaxon pair. Statistical software (i.e., PAST; Hammer et al., 2001) is used to calculate and save results in matrical form (adjacency matrix), according to which each entry is the Jaccard index for a given pair of ichnotaxa.

  3. Render the adjacency matrix as a network. Network analysis software (i.e., Gephi; Bastian et al., 2009) is used to draw and analyze the network and perform network analysis.

The process for obtaining an ichnonetwork from a stratigraphic log with free and open-source software has been described (Baucon and Felletti, 2013b; Baucon et al., 2014).

RESULTS

Trace Fossils of the Val Dolce Formation

This section describes the types of trace fossils of the Val Dolce Formation, named after a well-established nomenclature based on morphological characters (Knaust, 2012a). According to this practice, trace fossils are placed into taxonomical units (ichnotaxa); two ranks of ichnotaxa are generally used, ichnogenus (igen.) and ichnospecies (isp.) (Bromley, 1996). Table 2 summarizes the major ichnological terms used in this paper.

During field surveys, 11 ichnogenera were recognized in the study area. With the exception of Taenidium isp., Nereites missouriensis, and Phymatoderma isp., all of the observed ichnotaxa have received ichnotaxonomic treatment in the previous ichnological review of the Pramollo Basin (Baucon and Carvalho, 2008). The vertebrate footprint Limnopus, previously reported from the same study area (Mietto et al., 1985), probably belongs to the Val Dolce ichnofauna, although further records are needed to confirm this hypothesis.

This section documents the ichnofauna noted during the field survey, providing emended descriptions of the morphology, behavior, tracemaker, and environmental range of each ichnotaxon.

Archaeonassa isp.

Winding trails consist of a central furrow and two parallel, symmetric lobes. Trails are preserved as convex epireliefs or concave hyporeliefs. Trail width is 0.8–1 cm. In light of the comparison with modern gastropod trails (Turner, 1978; Baucon and Felletti, 2013a), Archaeonassa is interpreted as a surface trail produced by locomoting invertebrates. The symmetry of the lobes may indicate a tracemaker with a symmetric body plan, although symmetric trails may also be produced by gastropods (Baucon and Felletti, 2013a, fig. 9a). Archaeonassa is reported from both marginal-marine and continental environments (Buatois and Mángano, 2002; Knaust et al., 2012).

Curvolithus simplex

Unbranched, straight or winding structures consist of a central lobe flanked by two narrower lateral lobes (Fig. 6). Burrow width is 0.3–1.5 cm. The ichnogenus is interpreted as the locomotion trace of infaunal carnivores, including flatworms, nemerteans, and gastropods (Buatois et al., 1998). Curvolithus is a typical component of shallow marine and, especially, deltaic environments with rapid sand deposition in quiet settings (Curvolithus ichnofacies; Lockley et al., 1987; Tonkin, 2012).

Cylindrichnus isp.

Cylindrichnus isp. are dominantly vertical, curved unbranched burrows with circular cross sections. Traces are preserved as full reliefs. Lining is concentrically laminated (Fig. 6). Burrow diameter is 0.3–1 cm (Figs. 6 and 7). There is no consensus on the trophic strategy of Cylindrichnus, which is interpreted as the burrow of an ambush predator, a suspension feeder, or a detritus feeder (Głuszek, 1998; Sarkar et al., 2009). However, a strong analogy exists between Cylindrichnus and the burrow of the modern terebellid polychaete Amphitrite ornata, which feeds on detritus and bacterial cultures (Głuszek, 1998). Cylindrichnus is frequently reported from storm-influenced environments, ranging from middle shoreface to middle offshore (Sarkar et al., 2009). Cylindrichnus is also common in, but not exclusive to, brackish-water settings (Gingras et al., 2012a, 2012b). In this context, Cylindrichnus typically characterizes delta front settings, including mouth bars (Tonkin, 2012).

Helminthoidichnites tenuis

Helminthoidichnites tenuis is represented by horizontal, irregularly winding traces without branches or self overcrossings. Weathered specimens show dark fill. Burrow diameter is 0.1–0.2 cm (Fig. 7). The presence of dark fill suggests sediment processing, indicative of deposit feeding. Helminthoidichnites is an environment-crossing ichnogenus, being reported in continental and marine settings; typical producers include nematodes and insects (Uchman et al., 2009).

Nereites missouriensis

Simple horizontal burrows consist of a central meniscate ribbon surrounded by a lobate zone of reworked sediment (Fig. 8). The menisci can be widely spaced (thick meniscate form sensu Uchman, 1995) or tightly packed, resulting in a homogeneous string (fecal ribbon sensu Uchman, 1995). The lobate zone has a low contrast with respect to the host sediment (Scalarituba preservation; Seilacher, 2007); therefore this ichnotaxon is easily confused with Palaeophycus or Taenidium. Burrow width is 0.4–0.8 cm. Trace morphology is consistent with the ichnospecies N. missouriensis, represented by horizontal, unbranched burrows with a central meniscate ribbon surrounded by a lobate zone (Uchman, 1995).

Two models have been invoked for explaining the morphology of Nereites.

  1. In the Pascichnial model, a worm-like organism fed successively within each lobe and left a fecal string preserved as the central ribbon (Uchman, 1995; Martin and Rindsberg, 2007). Pascichnial forms of Nereites are generally documented from shelf to deep-sea environments (i.e., N. missouriensis, N. irregularis; Seilacher, 2007; Uchman, 1995).

  2. In the Repichnial model, an arthropod producer moved at the surface, producing lobed pressure-release structures with legs (Martin and Rindsberg, 2007). This model accounts for modern xiphosurans (Martin and Rindsberg, 2007) and hermit crabs (Baucon and Felletti, 2013a) inhabiting intertidal and shallow subtidal settings, respectively. Similar environmental settings are attributed to Ordovician forms of Nereites jacksoni (Neto de Carvalho and Baucon, 2010).

The presence of a well-developed fecal string is consistent with a combination of locomotion and feeding (pascichnia) rather than locomotion only (repichnia). N. missouriensis is aeurybathic form, documented from shelf (Seilacher, 2007) and turbiditic settings (Uchman, 1995). In deltaic settings, N. missouriensis is reported from the delta front-prodelta transition (Carmona et al., 2009). Modern Nereites ichnofabrics, comprising N. missouriensis, have been reported from deep-sea softgrounds and soupgrounds influenced by upwelling (Wetzel, 2002).

Planolites isp.

Simple, horizontal, unbranched burrows lack a wall lining, the fill similar to the host rock. Burrow width is 0.3–0.5 cm. The bioprint is similar to that of Pramollichnus, suggesting similar worm-like trace makers. Planolites is regarded as an environment-crossing ichnogenus corresponding to deposit-feeding strategies (Pemberton and Frey, 1982).

Phymatoderma isp.

Horizontal or inclined burrows have multiple palmate lobes. The fill contrasts in color with the host rock; pelleted fill is rarely preserved. Burrow width is 0.4–0.8 cm (Fig. 8A). The ichnogenus Phymatoderma has a complicated taxonomic history, and has often been described as Chondrites (Fu, 1991; Miller, 2011; Izumi, 2012). Phymatoderma is distinguished from Chondrites by the manifest pelletal fill, the poor definition of burrow contours, and the narrower branching angle (Bromley and Ekdale, 1984; Fu, 1991; Seilacher, 2007; Gerard and Bromley, 2008; Miller, 2011; Izumi, 2012). In addition, the burrow diameter of Chondrites is generally smaller (3 mm or less) than that of Phymatoderma (Izumi, 2012).

Phymatoderma is interpreted as a deposit-feeding burrow (Uchman and Gaździcki, 2010). Analysis of large Pliocene specimens (Miller and Vokes, 1998) revealed that the Phymatoderma producer fed on surface sediments and excreted fecal pellets at depth as a food cache. This behavior allowed the tracemaker to revisit tunnels successively, in response to a fluctuating trophic regime. Carbon isotope data confirm that the Phymatoderma producer ingested the surface sediments and subsequently excreted fecal pellets at depth (Izumi, 2012). Phymatoderma occurs mostly in outer shelf or deeper environments (Uchman and Gaździcki, 2010), especially from deep glaciomarine settings (Lima and Netto, 2012). The ichnogenus is interpreted to reflect oxygenated bottom water (Lima and Netto, 2012; Izumi, 2012), but is commonly found within black shale facies, indicating low-oxygen porewater.

Pramollichnus pastae

Crescentic structures result from juxtaposed spaghetti-like burrows that connect two apical points (Baucon and Carvalho, 2008, fig. 17); tight meanders may be present (Venturini, 2006, fig. 9.3c). The width of an individual burrow is 0.3–0.7 cm (Fig. 8). Pramollichnus is interpreted as a deposit-feeding structure produced by worm-like organisms (Baucon and Carvalho, 2008). In the study area, Pramollichnus occurs at the top of hummocky cross-stratified intervals, therefore reflecting opportunistic colonization of storm-deposited sandbodies. Current data extend the previous record of Pramollichnus (Baucon and Carvalho, 2008) from the Carboniferous into the Permian.

Psammichnites plummeri

Psammichnites plummeri is represented by horizontal, unbranched traces with a median string. Traces follow a straight, winding, meandering, or looping course. Trace width is 0.8–1.2 cm (Fig. 9). Psammichnites is interpreted as the trace of an infaunal deposit feeder (probably a mollusc) with a snorkel-like device (Mángano et al., 2002; Seilacher, 2007; Seilacher and Hagadorn, 2010; Alonso-Muruaga et al., 2013). Psammichnites is a common component of the Psilonichnus and Cruziana ichnofacies, being reported from intertidal flats (Mángano et al., 2002; MacEachern et al., 2012). In addition, Psammichnites is documented from deltaic environments as a component of the Cruziana ichnofacies (Mángano et al., 2002; Alonso-Muruaga et al., 2013).

Taenidium isp.

Horizontal, unbranched, unwalled burrows consist of dark, arcuate menisci alternating with thinner menisci that are concordant in texture with the host rock. Burrow width is 0.4–1 cm. The nature of the fill suggests that the tracemaker backfilled its burrow by excreting digested sediment (ingestion and excretion backfill sensu Baucon et al., 2014). For this reason, a deposit-feeding strategy is likely.

The taxonomy of meniscate traces is disputed, in particular the distinction between Beaconites and Taenidium. According to Keighley and Pickerill (1994), Beaconites is walled, Taenidium is not. For this reason, the studied structure is attributed to Taenidium, which is typically documented from continental environments but appears also in marine settings (Uchman et al., 2013; Buatois and Mángano, 2011).

Zoophycos isp.

Helically coiled spreite structures consist of a causative burrow revolving around a central axis. The outline is circular. Two morphotypes are distinguished on the basis of the absence or presence of lobes (Figs. 10A, 10B). Specimens without lobes often present a well-preserved apex (Fig. 10C). Lobate specimens commonly show a marginal tube contrasting in color to the host rock (Fig. 11), suggesting the presence of oxidized iron minerals. Trace fossil diameter is 15–30 cm. Several hypotheses have been proposed for the ethology of the Zoophycos producer, including deposit feeding, detritus feeding, caching, refuse dump, and gardening of microorganisms (Bromley, 1991; Löwemark and Schäfer, 2003). Although general consensus is yet to be reached, the cache model conveniently explains the major features of the ichnogenus (Bromley, 1991; Baucon and Avanzini, 2008; Buatois and Mángano, 2011). According to this model, the Zoophycos producer fed at the sediment-water interface and excreted fecal pellets at depth in order to maintain a food source for periods of scarce nutrient flux (Bromley, 1991). In analogy with Zoophycos rhodensis (Bromley and Hanken, 2003), the function of the marginal tube of lobed specimens could be linked with chemosymbiosis.

Zoophycos shifted from shallow to deep settings across the Phanerozoic (Bottjer et al., 1988; Baucon and Avanzini, 2008; Neto de Carvalho and Rodrigues, 2003), commonly being associated with quiet-water conditions and dysoxia (Martin, 2004; Löwemark and Schäfer, 2003; MacEachern et al., 2012). Recent Zoophycos are particularly abundant under upwellings (Löwemark and Schäfer, 2003). Evidence from Quaternary sediments relates Zoophycos to seasonal fluctuations in food supply deriving from monsoonal climate (Löwemark et al., 2006). In this regard, recent Zoophycos were produced during times of strong summer monsoon, in particular during interglacial times (Wetzel et al., 2011).

Ichnonetwork Analysis

A network is a set of nodes (or vertices) connected by links (or edges) of a specified type, such as social, trophic, or co-occurrence relationships (Borgatti and Halgin, 2011). For example, a friendship network can be described by persons (nodes) linked by friendship relationships (links), while trophic networks comprise biological taxa connected by trophic relationships (who eats whom). These examples confirm that the strength of the network approach is its simplicity and generality, as it can describe almost any system having multiple components (Krause et al., 2009). As anticipated by Euler (1736), the reciprocal position of these components, as defined by their links, plays a central role in understanding the studied system. In this regard, network theory provides measures to describe the position of individual components and the architecture of the entire system, which can be related to real phenomena (Borgatti and Halgin, 2011). Regardless of the network type, structure is central in network analysis (Borgatti et al., 2009).

Network theory has been demonstrated as an extremely powerful tool in the study of fossil and recent traces (Baucon and Felletti, 2013b; Baucon et al., 2014). Ichnological systems can be described as behavioral networks (ichnonetworks) consisting of mutually interconnected elements. In this case, nodes represent ichnotaxa, while links represent the association relationships between ichnotaxon pairs. In other words, the presence of a link between a pair of ichnotaxa indicates that those ichnotaxa co-occur at least in one sample. Though linked ichnotaxa co-occur in the same sampling unit, they were not necessarily emplaced at the same time. Linked traces may even pertain to different ichnocoenoses, and on further analysis may eventually indicate different environmental settings. However, in the Pian di Lanza section, preservation quality and crosscutting relationships do not evidence palimpsesting. In addition, the environmental preferences of co-occurring ichnotaxa do not indicate contrasting environmental scenarios.

This section provides an overview of the topographical features of the ichnonetwork under study, derived from the stratigraphic section of the Pian di Lanza stream (Fig. 1).

The corresponding ichnonetwork (Fig. 12; Supplemental File 11) shows ichnotaxa, association relationships, and their intensity. Although Archaeonassa, Pramollichnus, and Psammichnites are found in the study area, they have been not reported from the Pian di Lanza section and for this reason are not described in the ichnonetwork.

The studied ichnonetwork has a single component, i.e., all pairs of nodes are reachable through a continuous chain of links (connected network sensu Wassermann and Faust, 1994). Edges have variable weight, displaying strong association relationships between Zoophycos and Curvolithus, Curvolithus and Palaeophycus, Helminthoidichnites and Palaeophycus, and Zoophycos and Helminthoidichnites.

Graphic inspection shows that not all node pairs are directly connected (e.g., Phymatoderma and Cylindrichnus; Fig. 12), so some ichnotaxa never co-occur with some others. For this reason, the ichnonetwork is incomplete as only a part of the maximum possible number of links is present. Because two nodes can be linked by no more than one link, an ichnonetwork can have only a given number of links, determined by the number of nodes (Wassermann and Faust, 1994). It is possible to quantify the completeness of the studied ichnonetwork by computing graph density, i.e., the ratio of the links present to the maximum possible (Wassermann and Faust, 1994). According to this measure, only 53.6% of the possible links are present.

In light of this result, it is interesting to see if some nodes (ichnotaxa) have more connections (association relationships) than others. The number of links incident with a given node is considered; this measure, known as node degree (Wassermann and Faust, 1994), can be seen as the number of nodes directly connected with a given node (Wassermann and Faust, 1994). In a friendship network, degree shows the number of friends of a given individual. In ichnological terms, degree indicates the number of ichnotaxa that are directly associated with a certain ichnotaxon. Zoophycos, Helminthoidichnites, and Curvolithus are among the ichnotaxa with the highest degree, while Planolites and Phymatoderma are associated only with one ichnotaxon each (degree = 1). Degree is represented by node size in Figure 12.

Node degree provides information on local network structure, because it quantifies the number of direct connections, but it ignores any indirect link. For example, Curvolithus and Phymatoderma are not directly connected; it takes at least two steps to get from one to the other. This example show the concept of path length, which is the number of steps (links) necessary to get from one node to another (Scott, 2000). As many paths are likely to connect a pair of nodes, it is important to consider the shortest path length among them (geodesic distance). The concept of distance allows us to discern the structural position of a node within the network by calculating how close a node is to all other nodes (node closeness; Wassermann and Faust, 1994). More specifically, closeness is the inverse of the sum of the shortest paths (or distances) from a given node to the others (Wassermann and Faust, 1994). Accordingly, closeness lets zones be discerned with the network, in the same way that central bus stations are recognized because they are at only a few stops from the others. Based on their low closeness scores, Phymatoderma and Planolites are at the periphery of the studied network because they are far from the other nodes. In contrast, the center of the network is occupied by Zoophycos, Curvolithus, Helminthoidichnites, and Cylindrichnus.

In light of these results, a geography of the studied ichnonetwork is recognized. Accordingly, the structure of the network can be better understood if nodes are arranged in a way that highlights their topological geography. Nodes can be arranged manually, although there are many algorithms to display nodes in order to facilitate interpretation. Among these, the Hu (2005) method is a force-directed algorithm that models the graph-drawing problem by a physical system of bodies with forces acting among them. The resulting layout finds a good placement of nodes by minimizing the energy of the system (Hu, 2005); changing the layout does not change the connection pattern of the network, so Figures 13 and 14 are topologically equivalent.

Visual inspection of Figure 14 reveals that some nodes act as a bridge and connect central to peripheral areas of the networks. This subject can be approached by considering node “betweenness,” which measures the extent to which a particular node is between the others (Scott, 2000). In more formal terms, this measure is proportional to the number of shortest paths passing through a node (Wassermann and Faust, 1994). The communication of indirectly connected nodes depends on nodes with high betweenness because a great number of shortest paths are going through them (Boccaletti et al., 2006). Zoophycos and Cylindrichnus are the ichnotaxa of highest betweenness, being the intermediaries of the network (Fig. 15). In other words, Zoophycos and Cylindrichnus tend to co-occur with traces that are not associated with each other.

The ichnological significance of betweenness is important, because it is related both to the paleoenvironmental resolution of the ichnotaxon and the existence of network inhomogeneities.

  1. Ichnotaxa of high betweenness are likely to be poor environmental indicators. If a node is between many others, it bridges different structural islands of the network (bridge ichnotaxa sensu Baucon and Felletti, 2013b).

  2. The existence of bridges implies that the network texture is not as regular as a lattice, i.e., it is structured. Nodes with nonzero betweenness connect areas of the network that would otherwise be sparsely connected or not connected at all (Martín González et al., 2010). For example, in a friendship network, nodes having high betweenness are hubs between otherwise disconnected groups; this could be exemplified by rugby players and ballet dancers with a single common friend.

The structured organization of the network is also suggested by a clustering coefficient, which offers a measure of the interconnectivity between nodes linked to a given node (Ravasz et al., 2002; Boccaletti et al., 2006). In sociological terms, the clustering coefficient of a node indicates how likely that a person’s friends are friends of each other. In ichnological terms, clustering coefficient of a node is maximum (1) if the node is at the center of a fully interconnected cluster, i.e., when all the traces associated to a given one are also associated (Fig. 16).

Consequently, if the network structure is modular, the clustering coefficient averaged over all nodes will be high (Ravasz et al., 2002). In this context, the considerable average clustering coefficient (0.772) of the studied network is suggestive of community structure. This property does not refer to the concept of biological community, but is a network-related property. In mathematical terms, community structure implies high concentrations of edges within special groups of nodes, and low concentrations between these groups, which are termed communities or modules (Fortunato, 2010). In the social context, friendship circles exemplify the concept of community, while in ichnological terms, communities are cohesive groups of closely associated ichnotaxa. For this reason, following Pickerill and Brenchley (1975) and Legendre (2005), network communities represent trace fossil associations (see also Baucon and Felletti, 2013b; Baucon et al., 2014).

Communities have important implications for network analysis, because they represent groups of nodes that are likely to share common properties (Fortunato, 2010). In ichnological networks, environmental parameters are the properties shared among ichnoassociation members. Because biologic behavior depends on the surrounding environment, the formation of groups of associated behaviors (i.e., ichnoassociations) is likely to be driven by environmental properties. In this regard, crosscutting relationships and preservation style show that the association relationships observed on the field do not represent the superimposition of different ichnocoenoses. For this reason, each ichnoassociation is likely to represent a relatively stable and well-defined set of environmental conditions.

Although the concept of community is intuitive, community detection is not straightforward and many algorithms have been proposed to approach the problem (Boccaletti et al., 2006; Fortunato, 2010). In this regard, the concept of the clique is among the most immediate and stringent (Boccaletti et al., 2006). Specifically, a clique is a subset of nodes in which every possible pair of nodes is directly connected by a link, and the clique is not contained in any other (Scott, 2000; Fig. 12D). This definition implies that cliques may partially overlap and that a clique has the maximum possible link density (Boccaletti et al., 2006; Yan and Gregory, 2009). For example, in a friendship network, a clique is a group of persons all of whom are linked by friendship. In ichnology, a clique consists of a group of traces all of which are associated with one another. In ichnology it is necessary to relax the recommendation of Wassermann and Faust (1994) by considering cliques formed by node pairs (Baucon et al., 2014). In light of these assumptions, four cliques have been found in the studied ichnonetwork (Fig. 17): C1Cylindrichnus-Planolites, C2Phymatoderma-Zoophycos, C3Cylindrichnus-Helminthoidichnites-Curvolithus-Zoophycos, and C4Helminthoidichnites-Taenidium-Curvolithus-Nereites-Zoophycos. These cohesive subgroups represent ichnoassociations.

DISCUSSION

The topological attributes described in the previous section derive from the paleoenvironment, to the measure that it influenced the biological behavior recorded by trace fossils. These attributes account for various domains of scale ranging from individual nodes to the entire network: (1) network-level scale; (2) group-level scale; (3) node-level scale.

These different observation levels reflect different scales of organization. Network-level attributes (i.e., average clustering coefficient) refer to the ichnological system as a whole, and therefore consider the system at the level of the ichnosite; group-level attributes (i.e., network communities) account for ichnoassociations, corresponding to subenvironments; node-level attributes (i.e., node degree) refer to the role of individual ichnotaxa within the ichnological system.

Network-Level Scale

General features of the studied ichnonetwork and of the Val Dolce Formation show the ichnological signature of deltaic environments.

  1. Dominance of deposit-feeding behaviors. Deltaic environments are characterized by high suspended loads that clog filter-feeding structures and therefore preclude suspension feeding (MacEachern et al., 2005; Buatois and Mángano, 2011). In the Val Dolce Formation, the absence of suspension-feeding structures is documented even in sandy facies (i.e., facies S1), where suspension-feeding behaviors would be expected in nondeltaic analogues (MacEachern et al., 2005).

  2. Presence of opportunistic trophic generalists. Harsh environments are commonly inhabited by unspecialized r-selected animals that are adapted to high levels of stress (Buatois and Mángano, 2011). The studied unit comprises opportunistic, facies-crossing ichnotaxa such as Planolites, Helminthoidichnites, and Cylindrichnus.

  3. Evidence of freshwater influence. The Val Dolce ichnofaunas are commonly (but not exclusively) characterized by forms typically found in marine environments: a mixture of horizontal and vertical structures from the Cruziana and Skolithos ichnofacies, dominance of infaunal traces, monospecific assemblages, and small size. These features are typical of brackish-water conditions (Pemberton et al., 1982; Buatois and Mángano, 2004, 2011; Gingras et al., 2012a, 2012b).

Network texture has no abrupt discontinuities, as manifested by the existence of a single connected component in which all nodes are reachable through a continuous chain of links. As trace fossil distribution is driven by the environment, this means that individual ichnotaxa are part of the same environmental continuum. Evidence of an environmental continuum at a broad scale does not exclude fine fluctuations in environmental conditions. Patchy distribution of traces and fluctuating bioturbation index are commonly recognized at the outcrop scale, suggesting local variation in environmental variables.

Similarly, at an intermediate scale, the continuum of the network is not as homogeneous as a regular lattice. Nodes have varying numbers of connections (degree) and the network is characterized by a relatively high average clustering coefficient. As the clustering coefficient of a node is related to its interconnectivity pattern, the average clustering coefficient of the network is a measure of its potential modularity (Ravasz and Barabási, 2003; Solé and Valverde, 2004). This aspect bridges the network-level scale of analysis to the scale of groups, the ichnological significance of which is to be found.

Ichnological Terms for the Group-Level Scale

Ichnologists as well as ecologists are interested in finding groups of taxa that are significantly found together in order to synthesize environmental features or to predict environmental characteristics (Legendre, 2005). The process of finding groups of associated taxa can be seen as a pattern-finding process in which a set of information (source data) is generalized into a number of categories (groups of taxa). In this paper, the elements of this process are the following.

  1. Source data. The Pian di Lanza stratigraphic section provides the source data for drawing the ichnonetwork. Specifically, source data are the trace fossils found in each sampling unit.

  2. Generalization rule. The concept of clique is the generalization rule used for finding groups of associated taxa.

  3. Generalized groups of taxa. Source data are generalized into four groups of associated ichnotaxa, i.e., cliques C1–C4.

This study follows Pickerill and Brenchley (1975) in naming source data as trace fossil assemblages (ichnoassemblages) and generalized groups of associated taxa as trace fossil associations (ichnoassociations). However, the lack of consensus on the concept of ichnoassociation requires a more detailed terminological discussion. The term association has been applied loosely to trace fossils (Buatois and Mángano, 2011), although Häntzschel (1955) recognized the value of the Spuren-Vergesellschaftung (trace fossil association) for the characterization of sedimentary environments (Baucon et al., 2012).

From a terminological perspective, each sample of the source data is an ichnoassemblage because it consists of the trace fossils found in a rock unit. The term ichnoassemblage embraces all the trace fossils occurring within a single unit of rock, being equivalent to the assemblage of body fossils (Bromley, 1996; see also McIlroy, 2004; Buatois and Mángano, 2011). In this regard, an assemblage is defined as the collected sample and is purely observational, whereas an association is the recurrent association of taxa across a group of assemblages (Pickerill and Brenchley, 1975). A similar view was shared by Brenchley and Harper (1998, p. 230), according to whom “…an assemblage is a collection of fossils made from a single horizon or bed and an association is recognized by the presence of the same species in several assemblages.”

An association can therefore be regarded as the generalization of many assemblages (Pickerill and Brenchley, 1975; Brenchley and Harper, 1998; Clarkson, 2008). Consequently, each of the generalized groups of ichnotaxa described herein (cliques C1–C4) is a trace fossil association. This concept of association is widely accepted (e.g., Pickerill and Brenchley, 1975; Brenchley and Harper, 1998; Clarkson, 2008), although some exceptions are documented. For example, Bromley (1996) used the term trace fossil association in a more general sense to describe any group of traces (i.e., assemblages, suites, ichnocoenoses). In addition, the term ichnoassociation has been used with several specific connotations, ranging from the spatial to the temporal and/or ecological connotation (Buatois and Mángano, 2011; e.g., Savary et al., 2004).

Despite these important exceptions, we follow the concept of ichnoassociation as the generalization of many ichnoassemblages because it fits very well with the concept of association in paleontology and ecology. Clarkson (2008) summarized previous paleontological literature by indicating that an assemblage refers to a single sample from a particular horizon, whereas an association refers to a group of assemblages, all showing similar recurrent patterns of species. In ichnology, a similar concept was introduced by Gàmez Vintaned and Mayoral Alfaro (1992) and Gàmez Vintaned and Liñan (1996), defining an ichnoassociation as the ichnotaxon or the ichnotaxa that are recurrent along a set of ichnocoenoses. Similarly, the operational definition of association in ecology fits with the herein-adopted concept of association: “an association is simply a group of species (or taxa pertaining to some other systematic category) recognized as a cluster following the application of a clearly stated set of rules” (Legendre, 2005, p. 227). For example, the groups of ichnotaxa C1–C4 can be regarded as trace fossil associations because they have been recognized as clusters following the application of the concept of clique.

In sum, based on the paleontological and ecological concepts of association (Pickerill and Brenchley, 1975; Brenchley and Harper, 1998; Legendre, 2005; Clarkson, 2008), ichnoassociations are here considered as the generalization of many ichnoassemblages.

Group-Level Scale

Paralleling community ecology (Legendre, 2005), our aim is to find the ecological requirements common to the members of each ichnoassociation. These ecological requirements allow us to define a corresponding depositional subenvironment, which is named after the well-established process-based zonation of deltas (Reading, 1996). We describe the succession of bioturbation, bioturbation intensity, bioturbation distribution, ichnodiversity, tiering structure, and primary sedimentological features (Table 2) for each ichnoassociation.

Based on these assumptions, we provide an environmental interpretation of the clique-based ichnoassociations of the Pian di Lanza section, named after the characterizing ichnotaxa; clique names are also reported.

Cylindrichnus-Planolites Ichnoassociation (C1)

This ichnoassociation is characterized by low ichnodiversity and patchy, low bioturbation intensity (BPBI 2–3). Monospecific manifestations of Cylindrichnus (Figs. 18A–18C) are more common than the presence of both components of the ichnoassociation. Specifically, 420 cm (6 samples) of the Pian di Lanza section are characterized by Cylindrichnus only, while 260 cm (2 samples) have both Cylindrichnus and Planolites. A single monospecific sample with Planolites (thickness 20 cm) is also reported (Fig. 4). Tiering structure is difficult to interpret, because crosscutting relationships between Cylindrichnus and Planolites have not been discerned.

Low ichnodiversity is commonly associated with stressed settings (Buatois and Mángano, 2011) and, in deltaic environments, monospecific assemblages are usually indicative of high energy, high sedimentation rate, and lowered salinity (Tonkin, 2012). In particular, dominance of deposit-feeding strategies and absence of suspension-feeding behaviors suggest noxious conditions for suspension feeders. This phenomenon may be explained by quiet waters with low sedimentation rate, providing low concentrations of suspended nutrients, or high suspended loads, clogging filter-feeding structures (MacEachern et al., 2005; Buatois and Mángano, 2011; Tonkin, 2012). The latter hypothesis is more convenient, as ichnological and sedimentological evidence (facies S1) points to a high-energy setting. Vertical or steeply inclined forms of Cylindrichnus have been interpreted as indicators of high hydrodynamic energy (Frey, 1970). Cylindrichnus is commonly found in association with shifting substrates and event-bed deposition (Tonkin, 2012; Fig. 19).

In line with previous work on brackish-water ichnofaunas (Buatois and Mángano, 2011; Pemberton et al., 1982; Gingras et al., 2012a, 2012b), this ichnoassociation displays typical signatures of freshwater input in marine settings: marine forms (i.e., Cylindrichnus), dominance of infaunal traces rather than epifaunal trails, simple structures produced by trophic generalists, variable abundance, low ichnodiversity, and monospecific occurrences.

In sum, the ecological requirements of this ichnoassociation are high suspended load, freshwater input into seawater, and relatively high hydrodynamics. These features are characteristically associated with the delta front, the area where sediment-laden fluvial currents enter the basin and interact with basinal processes (Reading, 1996). More specifically, significant freshwater input and hydrodynamics are indicative of proximal bar subenvironments (i.e., bar back and bar crest areas; Reading, 1996).

Sedimentological features are in accord with this interpretation, as ichnoassociation C1 occurs mainly with the sandy facies S1 (Table 1). Facies S1 shows evidences of traction (parallel laminae in sandstone) and storm deposition (hummocky cross-stratification; Fig. 19A). For this reason, the maximum depth of this ichnoassociation coincides with storm wave base, which commonly ranges between 25 and 50 m in modern environments (Immenhauser, 2009). These features are consistent with delta front settings (Nichols, 2009), supporting the environmental interpretation of ichnoassociation C1.

In the Pian di Lanza section, an individual sample with Cylindrichnus and Planolites is reported from facies S2. However, that specific sample presents transitional features with facies S1.

Phymatoderma-Zoophycos Ichnoassociation (C2)

Low ichnodiversity and homogeneous, moderate to high intensity of bioturbation (BPBI 3–5) characterize this ichnoassociation (Fig. 20A). Zoophycos and Phymatoderma are likely to have occupied a deep tier because they crosscut background bioturbation, which is likely to represent the shallow mixing zone (Bromley, 1996). Because Zoophycos commonly overprints Phymatoderma (Fig. 11B), it must have occupied the deepest tier.

In contrast to the Cylindrichnus-Planolites ichnoassociation, there is no evidence of freshwater influence. Homogeneous, high bioturbation intensities are commonly regarded as signatures of normal-marine salinity (Buatois and Mángano, 2011). This interpretation is in line with the distribution of the individual components of the ichnoassociation. Zoophycos is commonly associated with quiet-water environments with low sedimentation rates, while Phymatoderma occurs mostly in outer shelf or deeper environments (Uchman and Gaździcki, 2010; Lima and Netto, 2012). Consequently, the Zoophycos-Phymatoderma ichnoassociation is thought to represent seafloors below the fairweather wave base.

Though freshwater influence is insignificant, availability of oxygen seems to be the limiting factor. In the oxygenation model of Ekdale and Mason (1988), intense bioturbation indicates that the bottom seawater was sufficiently oxygenated for supporting infaunal communities; at the same time, the predominance of burrows connected to the seafloor (Phymatoderma and Zoophycos) suggests that porewaters were poorly oxygenated. A permanent connection with the seafloor allows the exploitation of dysoxic substrates by ventilating the burrow with oxic water (Ekdale and Mason, 1988; Baucon and Felletti, 2013a). Nevertheless, caution must be exercised in interpreting ichnodiversity because intense bioturbation by deep-tier burrowers making Zoophycos and Phymatoderma may have obliterated shallower tier structures.

Bioturbation intensity, ichnodiversity, and the presence of Zoophycos indicate that this ichnoassociation is attributable to the Zoophycos ichnofacies, typical of quiet marine environments (Bromley, 1996). The deepest tier of sediments of the Zoophycos ichnofacies is commonly characterized by the Chondrites-Zoophycos ichnoguild, which is represented by permanent, deep-tier, deposit-feeding structures and it is typically documented from low-oxygen environments (Bromley, 1996; Martin, 2004). Being permanent, deep-tier, deposit-feeding structures, Phymatoderma and Zoophycos present functional features similar to those of the Chondrites-Zoophycos ichnoguild; therefore, they are interpreted to reflect low-oxygen porewaters. Furthermore, the shift from Phymatoderma-Zoophycos ichnoassociation to the rest of the network shows an increase in ichnodiversity and a shift from dominant surface deposit feeding to mixed surface and subsurface deposit feeding, according to the behavioral interpretation provided herein.

  1. The ichnoassociation C2 is characterized by surface deposit-feeding structures (Phymatoderma and Zoophycos).

  2. Ichnoassociation C3 is also characterized by surface deposit-feeding structures (Zoophycos, Cylindrichnus), subsurface deposit-feeding structures (Helminthoidichnites), and locomotion traces (Curvolithus).

  3. Ichnoassociation C4 presents surface deposit-feeding traces (Zoophycos), subsurface deposit-feeding burrows (Taenidium, Nereites, Helminthoidichnites), and locomotion traces (Curvolithus).

A similar shift from subsurface deposit feeding to surface feeding accompanies declining levels of oxygen in modern environments, along with decreasing burrow diameter and ichnodiversity (Smith et al., 2000). In addition, Phymatoderma and Zoophycos are commonly (but not exclusively) reported from substrates with lowered interstitial oxygen (Izumi, 2012; Bromley, 1996).

The trophic strategies represented by Zoophycos and Phymtoderma suggest a seasonally fluctuating environmental regime. A cache model has been proposed for both Zoophycos and Phymatoderma (Miller and Vokes, 1998; Miller and D’Alberto, 2001; Uchman and Gaździcki, 2010; Wetzel et al., 2011); according to this interpretation, during periods of high nutrient flux the tracemakers fed on surface sediment and excreted fecal pellets at depth, in order to revisit them during periods of reduced food supply.

In sum, the Phymatoderma-Zoophycos ichnoassociation indicates normal-marine salinity, quiet hydrodynamics, low-oxygen porewaters, oxic to dysoxic bottom waters, and seasonal fluctuations in nutrient flux. Overall, these features are compatible with the prodelta, the part of the delta unaffected by wave or tidal processes (Reading, 1996).

Seasonality of food flux and low-oxygen conditions can be explained by monsoons, which are seasonal reversals of atmospheric circulation (Jacques et al., 2013). Marine productivity is seasonally enhanced by monsoons, which control river runoff and upwelling (Jyothibabu et al., 2008). As a consequence, increased deposition of organic carbon may have led to low-oxygen conditions. This hypothesis fits well with the record of recent Zoophycos, which has been reported from monsoon-influenced areas with upwelling and seasonal pulses in nutrient flux (Wetzel et al., 2011). Although no direct link has been established, Phymatoderma is abundant within the monsoon-influenced Posidonia Shale (Röhl et al., 2001; Izumi, 2012).

Despite the coherent explanation offered by monsoonal climate, other climatic phenomena (e.g., summer water stratification; Eldridge and Morse, 2008; Tyson and Pearson, 1991) are able to produce low-oxygen conditions, and further studies are required to test for the influence of monsoons in the Val Dolce Formation.

The ichnological evidence for quiet, fully marine settings with poorly oxygenated porewaters is corroborated by sedimentological and paleontological features. The Phymatoderma-Zoophycos ichnoassociation is reported only from the pelitic facies P1. Prodeltaic settings are commonly characterized by pelitic sediments, because the current from the river is dissipated away from the channel mouth and wave energy decreases with depth (Nichols, 2009).

The Phymatoderma-Zoophycos ichnoassociation is commonly found with low-oxygen indicators (pyrite, well-preserved body fossils; Fig. 19B). These features are in accord with the dysoxic prodelta suggested by ichnoassociation C2.

Cylindrichnus-Helminthoidichnites-Curvolithus-Zoophycos Ichnoassociation (C3)

With respect to the partially overlapping ichnoassociation C1, this ichnoassociation marks an increase in ichnodiversity, intensity (BPBI 3–5) and homogeneity of bioturbation. These aspects are likely to reflect the decreasing role of the major environmental stressors (i.e., freshwater input, hydrodynamics, suspended load). This ichnoassociation covers a gradient in bioturbation homogeneity and intensity (Figs. 18B, 18C) ranging from sparse and moderate (BPBI 3) bioturbation to homogeneous and intense (BPBI 5) bioturbation. The homogeneous and intense bioturbation is typical of fully marine conditions, in contrast with the sparse bioturbation characterizing brackish settings (Buatois and Mángano, 2011). For this reason, this ichnoassociation is interpreted to have covered an environmental gradient ranging from freshwater-influenced to fully marine conditions (Fig. 18).

The moderate diversity of deposit-feeding strategies indicates that nutrients were widely available onto and within the sediment, and so relatively low hydrodynamic energy is necessary to settle fine organic particles.

The ichnofauna shows no evidence of low-oxygen conditions. Although the water remained oxic, it was evidently low enough to avoid total oxidation of the organic matter exploited by deposit feeders. The presence of Zoophycos could indicate seasonal fluctuations in food flux (Wetzel et al., 2011; Miller and D’Alberto, 2001).

According to crosscutting relationships, Helminthoidichnites and Curvolithus occupied the shallowest tier, Cylindrichnus occupied an intermediate tier, and Zoophycos the deepest one.

Overall, the ecological requirements of this ichnoassociation are freshwater-influenced to fully marine conditions, moderate to low hydrodynamics, oxic bottom waters and seasonally fluctuating nutrient flux. These features are coherent with distal delta front (i.e., distal bars) and, especially, prodeltaic settings, where influence of wave and/or tidal processes is minimal (Reading, 1996).

Ichnoassociation C3 has a wider facies distribution than others, being reported from facies S1, S2, and P1. The sedimentological features of facies S1 (fine- to medium-grained sandstone, planar lamination, hummocky cross-stratification) are consistent with delta front settings. Sedimentological and paleontological features of facies P1 (siltstone and sandy siltstone, locally abundant marine body fossils) are consistent with prodelta settings. Because the sandy facies S2 is interbedded in thick pelitic sequences, it is tentatively interpreted as high-energy events in prodeltaic settings. The depositional process is difficult to ascertain because bioturbation is very intense, commonly obliterating physical sedimentary structures.

In light of these observations, facies analysis supports the environmental interpretation of the ichnoassociation ranging from distal delta front to prodeltaic settings.

Helminthoidichnites-Taenidium-Curvolithus-Nereites-Zoophycos Ichnoassociation (C4)

This ichnoassociation is characterized by homogeneous intense bioturbation (BPBI 5) and moderately diverse ichnofauna. Based on crosscutting relationships, Helminthoidichnites, Taenidium, and Curvolithus occupied the shallowest tier, Nereites occupied an intermediate tier, and Zoophycos occupied the deepest one. Because the backfill of Nereites and Taenidium shows that no connection was maintained with the seafloor, interstitial waters are interpreted as sufficiently oxygenated (Buatois and Mángano, 2011; Ekdale and Mason, 1988).

In light of the common ichnological features, ichnoassociations C3 and C4 had similar environmental significance; nevertheless, ichnoassociation C4 is likely to correspond to more distal environments on the basis of its more intense and homogeneous bioturbation. Unlined burrows would collapse in high-energy environments with shifting substrates; therefore, the abundant occurrences of the unlined ichnospecies N. missouriensis support a quiet environment. N. missouriensis is reported from deep-sea settings and can indicate fluctuation in nutrient flux (Wetzel and Uchman, 2012); shallower occurrences of the ichnospecies are also known (e.g., delta front–prodelta transition—Carmona et al., 2009; estuarine settings—A.K. Rindsberg, 2014, personal commun.).

In addition, the small size of Curvolithus (width < 0.5 cm; Fig. 19D) indicates suboptimal conditions for the producer. As Curvolithus is commonly regarded as an indicator of rapid sand deposition (Tonkin, 2012), reduction in size supports the idea of a somewhat quieter, more distal environment.

In conclusion, this ichnoassociation could represent an oxic prodeltaic environment characterized by dominant marine influence. This interpretation is supported by the fine grain size (facies P1) and the presence of marine body fossils (Fig. 19D).

In light of neoichnological evidence (Wetzel, 2002; Wetzel et al., 2011; Wetzel and Uchman, 2012), the presence of Zoophycos and Nereites is compatible with monsoon-influenced climate, although it does not constitutes conclusive proof; further studies are required to test the idea of monsoonal climate in the Val Dolce Formation.

Facies analysis supports the interpretation of a prodeltaic setting as this ichnoassociation is reported from facies S2 and P1.

Unburrowed Units

Sediment units without distinct burrows occur as unbioturbated patches within sandy facies (facies S1, S3), and as largely unburrowed sediments. The lack of distinct burrows may indicate lack of bioturbation, but also total reworking of the sediment or low sediment cohesiveness (Bromley, 1996; Lobza and Schieber, 1999; Buatois and Mángano, 2011).

In the study site, conglomerates (facies C1, C2) are unburrowed, probably because their high-energy setting prevented benthic colonization. In light of their sedimentological attributes, facies C1 is interpreted to represent distributary channels, mostly pertaining to the river-dominated delta plain, while facies C2 could represent beach deposits (Venturini, 1990). Pelitic units (facies P1) commonly lack distinct burrows and are locally characterized by mottled textures. For these reasons, the absence of distinct burrows is ascribed to the total reworking of the sediment, which is typical in the predominantly quiescent setting of the distal delta front (Tonkin, 2012). These observations are in accord with the interpretation of similar facies of the Pontebba Supergroup (Venturini, 1990). In contrast, unburrowed sandy units (facies S1, S3) show clear evidence of lamination and therefore indicate the virtual lack of biogenic reworking. A low bioturbation index is a common feature of proximal delta front environments (Tonkin, 2012), as suggested by the sedimentological interpretation of similar facies in the Pontebba Supergroup (Venturini, 1990).

The measured section does not have limestone layers, although they are present nearby. Distinct burrows have never been recovered from the limestone facies, although indistinct bioturbational structures may be present. In light of their sedimentological features, these limestone facies has been attributed to shallow, open-sea environments below the fairweather wave base (Sanders and Krainer, 2005; Venturini, 1990).

Geographical Ranges and the Structure of the Paleoenvironment

According to the principles of biogeography, each species has a geographical range, i.e., a geographical area within which it can be found (Brenchley and Harper, 1998; Franklin, 2009). This assumption can be extended to trace fossils: all ichnotaxa had a geographical range and, similarly, ichnoassociations were also characterized by a geographical range.

What were the geographical ranges of the trace fossils of the Pian di Lanza ichnonetwork? Distribution maps (Franklin, 2009) would provide the ideal answer by representing geographical ranges as colored areas superimposed on the paleogeographical map of the study site. However, this solution is difficult to apply in practice because it requires a degree of data resolution that is not available in the rock record. Alternatively, geographical ranges could be represented by spatial relationships, i.e., providing information about the intersection or disjunction of different geographical ranges. This idea supports the ichnonetwork approach because the association relationships described in an ichnonetwork correspond to the spatial relationships that existed in the paleolandscape. Trace fossils are almost invariably found in situ (Buatois and Mángano, 2011).

The spatial relationships between different geographical ranges can be therefore derived from the topology of an ichnonetwork, allowing us to answer to questions such as Did the range of a given ichnotaxon overlap with the range of another? Did the distribution range of a given ichnoassociation grade into the range of another ichnoassociation? Which geographical range bridged otherwise separate ranges?

The importance of understanding the spatial relationships between different geographical ranges is in the environmental significance of traces. Traces are evidence of behavior, and geographical ranges of ichnotaxa and ichnoassociations delimit areas with specific environmental conditions. For this reason, the spatial relationships between geographical ranges provide information on the physical arrangement of the environment, i.e., the structure of the paleoenvironment. The environmental structure can be derived from an ichnonetwork by considering the spatial and ecological significance of the major architectural elements of an ichnonetwork.

Nodes

A node is an ichnotaxon, the geographical range of which corresponds to a specific set of environmental conditions. This idea can be better understood by representing the potential environment of an individual ichnotaxon as an abstract space, the dimensions of which correspond to the major environmental factors. In parallel with the concept of Hutchinsonian niche (Holt, 2009), the environmental requirements of a given ichnotaxon define its environmental range (niche). This concept is exemplified by Figure 21A, which shows a two-dimensional space defined by two axes, salinity and depth. In this abstract space, the environmental niche of an ichnotaxon is defined by specific values of the environmental factors. Although the example is two-dimensional, the concept of niche can be extended to any n-dimensional space.

Links

Links are association relationships between traces and linked nodes are associated traces. Given that traces are almost invariably in situ (Buatois and Mángano, 2011), the geographical ranges of associated traces intersected; for the same reason, disjoint nodes corresponded to nonintersecting (disjunct) geographical ranges. In the niche perspective, the association of two or more ichnotaxa is verified where their niches meet (Fig. 21A) because the intersection of different niches comprises the environmental requirements common to the corresponding ichnotaxa. In more formal terms, an ichnonetwork can be seen as a representation of niche intersections, provided that links do not represent the superimposition of different ichnocoenoses (for a discussion of the concept of ichnocoenose, see Davitashvili, 1945; Lessertisseur, 1955; Ekdale et al., 1984; Pickerill, 1992; McIlroy, 2004). In other words, an ichnonetwork can be seen as an intersection graph, with nodes corresponding to ichnotaxa and links corresponding to niche intersections. Niches can also be seen as an intersection representation, i.e., a family of sets corresponding to the nodes so that nodes are linked only if their assigned sets intersect (Cranston et al., 2012). This demonstrates that an ichnonetwork describes association relationships as derived from environmental parameters.

Communities

Network communities are ichnoassociations, the geographical range of which is likely to represent a specific depositional environment. Network communities are groups of nodes forming distinct structural areas, being characterized by high concentrations of links within these special groups of nodes, and low concentrations between these groups (Newman, 2004; Fortunato, 2010). Given that linked nodes are intersecting niches, ichnoassociations are environmentally distinct groups of traces. This feature is in agreement with the definition of depositional environment, intended as a part of the Earth’s surface that is physically, chemically, and biologically distinct from adjacent areas (Selley, 1970). Interpretation of network communities as ichnoassociations conforms well with a central tenet of ichnology, i.e., that trace fossil associations tend to be environmentally restricted (Seilacher, 1964, 1967; Buatois and Mángano, 2011). For the same reason, communities that share at least one node (overlapping communities; see Evans, 2010) represent similar environments. Overlapping communities are ichnoassociations that share at least one ichnotaxon (overlapping ichnoassociations) and therefore indicate similar environmental parameters, being likely to present neighboring geographical ranges.

These general observations allow us to understand the environmental structure of the Val Dolce Formation. Crosscutting relationships and bioturbation style do not evidence superimposition of different ichnocoenoses at the scale of the sampling unit, allowing us to interpret linked nodes (i.e., associated traces) as spatially and ecologically compatible traces. Because associations relationships are mapped as links, the geographical ranges of linked nodes (e.g., Zoophycos and Phymatoderma; Fig. 14) were intersecting. Conversely, the geographical ranges of topologically disjunct nodes (e.g., Planolites and Phymatoderma; Fig. 14) were disjunct in space. Similarly, ichnoassociations that share at least one ichnotaxon (overlapping ichnoassociations) represent similar environmental conditions, and therefore their geographical ranges were likely to be neighboring in space (e.g., ichnoassociations C1 and C3; Fig. 17).

A more specific knowledge of the environmental structure is provided by considering the rule that grouped ichnotaxa into trace fossil associations. For this reason, an individual clique-based ichnoassociation is represented by the intersection of the niches of its constituting members. For example, the niches of ichnotaxa a, b, and c mutually intersect (Fig. 21A) and therefore form a clique, i.e., an ichnoassociation (Fig. 21B). Because an intersection is equal to or smaller than the intersecting niches, an ichnoassociation delimits an environmental range that is equal to or narrower than the niches of the ichnoassociation members (i.e., Fig. 21A). An ichnoassociation is likely to correspond to a depositional environment.

The topological relationships between ichnoassociations represent the spatial relationships between depositional environments. In this regard, two major clique to clique relationships are distinguished, overlap and disjointness.

Overlapping cliques are ichnoassociations with shared ichnotaxa, and so they have an environmental affinity. However, a more specific significance is revealed by comparing the niche (Fig. 22A) and ichnonetwork perspectives (Fig. 22B). Overlapping cliques share common nodes (e.g., node e in Fig. 22B), the niches of which bridge the environmental ranges of the overlapping cliques (e.g., a-b-e and c-d-e in Fig. 22A). This means that a continuous environmental gradient connects overlapping ichnoassociations, because cliques are ichnoassociations. For this reason, it is likely that the geographical ranges of overlapping cliques were separated by bioturbated areas with transitional ichnological and environmental features. However, given that a clique by definition cannot be contained in another, the geographical ranges of overlapping clique-based ichnoassociations cannot overlap in space.

Several overlapping relationships are seen in the studied ichnonetwork: for example, ichnoassociation C1 overlaps with ichnoassociation C3, which overlaps with ichnoassociation C4 (Fig. 17A). The environment of clique C1 graded into the environment of clique C3, which, in turn, was transitional into the environment of clique C4. This hypothesis conforms with the environmental interpretation of ichnoassociations C1, C3, and C4, indicating proximal delta front, distal delta front to proximal prodelta, and prodelta, respectively. Similarly, the overlap pattern of ichnoassociation C2 may be part of an oxygenation gradient.

If two cliques are disjoint in the ichnonetwork, an environmental gap separates the corresponding environmental ranges (Figs. 23 and 24). This gap may be (1) empty: if there are no niches covering the gap between two ichnoassociations, there are no continuous environmental corridors among the disjoint ichnoassociations (Fig. 23A). In other words, there are no bioturbated environments bridging the environments represented by disjoint cliques. This configuration corresponds, for example, to a network with disconnected components (Fig. 23B). Theoretically, disconnected components may be explained also by adjacency in abstract space, although such sharp boundaries are unlikely in nature (Erdös et al., 2011). Conversely, the gap may be (2) covered. If, in abstract space, two disjoint ichnoassociations are connected through a continuous chain of niches, these constitute the environmental corridors bridging the disjoint ichnoassociations (Fig. 24A). In other words, the environments represented by disjoint ichnoassociations are distinct, but they grade into each other through a set of bioturbated environments. The ichnonetwork equivalent of Figure 24A is represented by two disjoint cliques, both overlapping with a third one (Fig. 24B). In the studied network, the nonoverlapping cliques C1 and C4 provide an analog, being connected by clique C3. Their environmental interpretation is congruent with this assumption, as C1 and C4 represent the proximal delta front and the prodelta, respectively. These deltaic environments are separate zones of the same gradient, being connected by a transitional zone that corresponds to clique C3. Although spatial relationships and paleoenvironmental structure are described in the previous lines, a more synthetic representation is needed. Network analysis provides a practical tool to this aim, i.e., clique graphs. The nodes of a clique graph represent the cliques of the original network and links represent the way they overlap (Evans, 2010). In ichnological terms, the clique graph of an ichnonetwork maps the ichnoassociations of the original ichnonetwork as nodes, whereas links connect ichnoassociations with shared ichnotaxa (overlapping ichnoassociations). Consequently, a link connects neighboring environments. The clique graph of the Pian di Lanza ichnonetwork is shown in Figure 17B, presenting ichnoassociations and their environmental interpretation.

Node-Level Scale

The role of individual ichnotaxa (node-level attributes) is quantified by standard measures of node centrality (Boccaletti et al., 2006), i.e., degree, closeness, and betweenness.

Node degree describes the local connectivity pattern of a node by measuring the number of incident links. Consequently, high-degree nodes such as Curvolithus and Zoophycos are associated with many other ichnotaxa (Fig. 12). Nevertheless, degree is a local measure of centrality, while nodes of similar degree may play different roles within the studied system. In the studied system, this aspect is exemplified by Zoophycos and Curvolithus, which have similar node degrees (Fig. 12), but different betweenness (Fig. 15).

Node betweenness describes the importance of a node as a connector between different parts of the network (Martín González et al., 2010), and therefore recognizes those behaviors that can compete in different environmental contexts. Betweenness measures the reliability of an ichnotaxon as a paleoenvironmental indicator; the higher the betweenness, the lower the environmental specificity. The studied ichnonetwork is generally consistent with global patterns of bioturbation. The bridge (high betweenness) ichnotaxa Zoophycos and Cylindrichnus have a characteristically wide environmental range at a global scale (Bottjer et al., 1988; Tonkin, 2012; Głuszek, 1998; Baucon and Avanzini, 2008), while low betweenness characterizes Curvolithus and Nereites, which are keystone ichnogenera of environmentally sensitive ichnofacies (Buatois et al., 1998; Tonkin, 2012; MacEachern et al., 2012). Despite these results, betweenness describes the role of an ichnotaxon within the boundaries of the studied system, and so global patterns of bioturbation are not necessarily respected in all cases (i.e., low betweenness of the generalist Planolites).

The significance of betweenness can be approached by considering the niche perspective. In this approach, each ichnotaxon is represented by a node with a corresponding niche (Fig. 21). Therefore, the niches of high-betweenness ichnotaxa (bridge ichnotaxa) are expected to be between many other niches. Similarly, the niches of bridge ichnotaxa tend to cover the abstract space separating disjoint ichnoassociations (Fig. 22). Conversely, low-betweenness and high-clustering ichnotaxa are likely to be environment specific, playing a role similar to that of species that reflect some measure of the character of the habitat (indicator species; Anas et al., 2013).

In the original paleolandscape, the spatial distribution of bridge ichnotaxa is expected to mirror the abstract space in which their niches are defined. Each site of the landscape is characterized by specific environmental parameters that correspond to the axes of the abstract space. For example, the spatial distribution of Figure 25A corresponds to the ichnonetwork and the abstract space represented in Figure 22. In this regard, the high-betweenness ichnotaxon (node e in Fig. 22B) covers the abstract space separating disjoint ichnoassociations (Fig. 22A) and, at the same time, it bridges two separate areas of the landscape (Fig. 25A). A similar spatial configuration explains coherently the high-betweenness of Cylindrichnus and Zoophycos, which occupy similar topological positions within the Val Dolce ichnonetwork.

This idea establishes the connection between the abstract space and the topographical one, from which the topology of the ichnonetwork derives. Consequently, ichnoassociations, as defined in this study, are environmental (abstract space), topographical (landscape space), and topological (network space) entities.

In parallel to Figure 25A, which describes overlapping ichnoassociations, Figure 25B represents the spatial manifestation of disjoint ones. In other words, Figure 25B is the equivalent of Figure 23. Network topology is influenced both by spatial relationships and by environmental niche of individual ichnotaxa. The disjoint ichnoassociations of Figure 25B reflect distinct spatial and environmental ranges.

The aforementioned examples (Figs. 25A, 25B) show cliques with homogeneous fabric. However, structural inhomogeneities may occur if several ichnotaxa with different environmental affinities reach their distributional limits within the same zone. For example, if the upper and lower distributional limits of several ichnotaxa converge in the same belt, a clique will include all of these ichnotaxa (Fig. 25C). Such a transition belt is analogous to the concept of ecotone, which is “a stress line connecting points of accumulated or abrupt change” (Livingston, 1903; Clements, 1905, p. 277). In other words, an ecotone indicates a narrow transition zone between two adjacent ecosystems with mixed characteristics of the two adjacent ecosystems (Livingston, 1903; Di Castri et al., 1988; Dutoit et al., 2007; Basset et al., 2012; Erdös et al., 2011). Similarly, a peak in ichnodiversity is reached within the aforementioned belt (Fig. 25C), since it contains a mixture of traces from neighboring zones. The corresponding ichnonetwork, shown in Figure 25C, will consist of a single clique, although finer levels of organization are seen in the spatial distribution of traces. Shallow ichnotaxa (a, b) and deep ichnotaxa (c, d, e) are clearly distinguished in the environmental distribution of Figure 25C. Consequently, we might question whether it is possible to recognize these finer patterns of organization from ichnonetwork analysis.

The answer can be found by contrasting homogeneous trace distributions (Fig. 25D) with a convergent pattern (Fig. 25C). Link weight measures the probability of co-occurrence of ichnotaxa pairs; therefore, if ichnotaxa are homogeneously distributed (i.e., no preferential association patterns), link weights will be approximately equal (Fig. 25D). However, if ichnotaxa are heterogeneously associated within the same ichnoassociation, link weights will be different (Fig. 25C).

The latter scenario is consistent with the clique C3 edge weights, which are not homogeneously distributed, but account for association relationships of different strength. For example, the edge connecting Zoophycos with Curvolithus has a higher weight than the edge between Zoophycos and Cylindrichnus (Fig. 17). Consequently, finer patterns of association can be found within the ichnoassociation C3. In order to better visualize these patterns, clique C3 is isolated from the rest of the network and the corresponding subnetwork is filtered by average edge weight (Fig. 26A). The filtered subnetwork consists of two cliques, Cylindrichnus-Helminthoidichnites and Curvolithus-Zoophycos-Helminthoidichnites. This weight pattern can be coherently explained with the convergent distribution depicted in Figure 25C. More specifically, the ichnosubassociation Cylindrichnus-Helminthoidichnites, which is frequently reported from the field (Fig. 26B), is interpreted to be more proximal in light of the lower ichnodiversity and the environmental affinity of individual ichnotaxa.

While clique C3 is markedly heterogeneous with respect to edge weight, a more homogeneous situation is depicted by clique C4. The links between Helminthoidichnites-Curvolithus-Nereites-Zoophycos have relatively homogeneous weights (Fig. 26C). Only the weight of the Nereites-Zoophycos link is lower than clique average, but the difference is not significant (clique average weight 0.392; Nereites-Zoophycos weight 0.363). In contrast, Taenidium has weaker links, possibly a result of being a rarer ichnotaxon.

From the Outcrop to the Model and Back

The results of the aforementioned analysis can be synthesized as a model in which individual traces and ichnoassociations are framed within a paleoenvironmental setting. Toward this aim, topological patterns provide the constraints for reconstituting the spatial and environmental relationships of both ichnoassociations and individual ichnotaxa. Interpretation of ichnological features offers a perspective on the environmental processes controlling trace distribution.

The paleoenvironmental model of the lower Permian Val Dolce Formation (Fig. 27) describes the geographical range of each ichnotaxon and its relation with the depositional environment. Similarly, the paleoenvironmental model depicts the environmental significance of ichnoassociations and the overall organization of the ichnosite.

This model finds potential application for understanding the paleoenvironment from the field occurrence of trace fossils. However, we might question whether the paleoenvironmental model is coherent.

In order to answer this question, the model is tested against source data. More specifically, the idea is to describe the model (Fig. 27) as an ichnonetwork (model ichnonetwork) and compare it to the Pian di Lanza ichnonetwork. The process for describing the model as an ichnonetwork consists of representing ichnotaxa as nodes and drawing a link between those ichnotaxa that are interpreted to have overlapping environmental ranges. Figure 28A shows the interpreted geographical ranges (corresponding to those shown in Fig. 27) from which the model ichnonetwork (Fig. 28B) is derived.

This process is conceptually analogous to the process of drawing an ichnonetwork from neoichnological data (Baucon and Felletti, 2013b), with the sole exception that links of the model ichnonetwork are unweighted. Ichnotaxa with overlapping environmental ranges are associated, and, in an ichnonetwork perspective, associated ichnotaxa are nodes connected by a link. In order to compare the model ichnonetwork with the Pian di Lanza ichnonetwork, the numbers of nodes and links are considered. In addition, the following network metrics are considered (Table 3).

Network density is the ratio of the number of links present in a network to the maximum possible. Network density goes from 0 (no links being present) to 1 (all possible links being present; Wassermann and Faust, 1994).

Network diameter is the maximum shortest path length of the network (Boccaletti et al., 2006).

Average path length (also known as average shortest path length or characteristic path length) is the mean of geodesic lengths over all couples of nodes (Boccaletti et al., 2006).

Average degree summarizes the degrees of all the nodes in a network by reporting the degree averaged over all nodes (Wassermann and Faust, 1994; Scott, 2000; Boccaletti et al., 2006).

Average clustering coefficient is a measure of the potential modularity of a network (Solé and Valverde, 2004; Ravasz and Barabási, 2003) and consists of the clustering coefficient averaged over all nodes (Ravasz et al., 2002).

The model ichnonetwork presents the same topological features of the Pian di Lanza ichnonetwork, being virtually identical to it (Table 3). For this reason, the paleoenvironmental model of the lower Permian Val Dolce Formation (Fig. 27) is consistent with the Pian di Lanza ichnonetwork, coherently explaining and synthesizing the ichnological features of the source data.

Advantages of Ichnonetwork Analysis

Ichnonetwork analysis allows us to reconstruct the paleoenvironment of the Val Dolce Formation, here interpreted as a fluvio-deltaic system; this interpretation is similar to that provided by facies analysis (e.g., Venturini, 1990; Forke, 2002). In addition, more commonly employed methods of ichnology (i.e., ichnofacies and ichnofabric analysis) are demonstrated to be efficient tools in reconstructing the paleonvironment based on the trace fossil record (McIlroy, 2008; Gingras et al., 2011). An important factor is then the potential advantages of ichnonetwork analysis. Here we evidence the differences and similarities of ichnonetwork analysis with respect to other approaches, i.e., previous ichnological methods, quantitative ichnological methods, and facies analysis.

Among its most characterizing features, the quantitative nature of ichnonetwork analysis contrasts with the qualitative approach of most ichnological studies. In parallel with geology (Krumbein, 1960; Merriam, 2004), ichnology began as mainly a descriptive science, being characterized by concepts and features difficult to describe in numerical terms. Therefore, one of the advantages of ichnonetwork analysis with respect to qualitative approaches is to provide a quantitative description of an ichnosite. This does not imply a greater degree of scientificity with respect to qualitative approaches, because quantitative data and their analysis do not necessarily lead to valid or reliable results (Hubbert, 1974; Merriam, 2004). Quantification allows observations to be characterized synthetically and objectively (Borradaile, 2010).

The ichnonetwork approach fits with this perspective because an ichnonetwork describes quantitatively the relationships among ichnotaxa, synthesizing objectively the source data into associations. For example, the studied ichnonetwork (Fig. 12) displays the same association relationships of the Pian di Lanza stratigraphic log (Fig. 4). Because an ichnonetwork consists of quantitative data, different ichnonetworks can be compared objectively, comparing data sets from different settings. For example, ichnonetwork analysis would allow quantitative comparison of ichnosites from different depositional, geographical, and chronostratigraphic settings, including neoichnological and paleoichnological data.

Despite the qualitative trend in ichnology, a number of quantitative ichnological studies focused on statistics, computer modeling, imaging and multivariate analysis (Knaust, 2012b, and references therein). Specifically, Knaust (2012b) described five major topics in the statistical analysis of traces: (1) trace fossil morphology, (2) distribution analysis and modeling of trace fossils, (3) analysis of benthic communities, (4) changes in trace fossil composition and amount of bioturbation through the stratigraphic record, and (5) bioturbation intensity, trace fossil distribution, and lateral persistence of ichnofabrics. In addition, computer modeling is employed for theoretical and applied studies, whereas imaging techniques are used to characterize individual trace fossils and bioturbation-induced porosity (Knaust, 2012b).

Ichnonetwork analysis differs from these quantitative approaches in that it focuses on the association relationships between ichnotaxa. A similarity is found with cluster analysis, a multivariate technique that has already found application in ichnology by considering co-occurring traces (Bjerstedt, 1988; Draganits et al., 2001; Davies et al., 2006; Minter and Braddy, 2009; Smith et al., 2008; Pervesler et al., 2011; Knaust, 2012b). However, most ichnological applications of cluster analysis focus on Q-mode analyses, i.e., localities are grouped on the basis of the co-occurring ichnotaxa (e.g., Smith et al., 2008; Minter and Braddy, 2009; Davies et al., 2006). Specifically, Q-mode analyses define the relationships between the samples based on the variables (Randazzo and Baisley, 1995; see also Hammer and Harper, 2006). In contrast, ichnoassociation-finding methods of ichnonetwork analysis (Baucon and Felletti, 2013b; Baucon et al., 2014; this study) group ichnotaxa on the basis of the association relationships, being more similar to R-mode analyses (e.g., Pervesler et al., 2011). R-mode analyses compare the relationships between the variables on the basis of all samples (Randazzo and Baisley, 1995; see also Hammer and Harper, 2006).

According to this perspective, cluster analysis differs from ichnonetwork analysis because its focus is narrower, i.e., finding groups of ichnotaxa. For this reason, cluster analysis is comparable to the ichnoassociation-finding methods of ichnonetwork analysis. Cluster analysis is already among the standard community-finding methods of network science (Fortunato, 2010) and has already found application in ichnonetwork analysis (Baucon and Felletti, 2013b).

Ichnonetwork analysis is not an alternative to previous methods of ichnological analysis, but integrates them with a quantitative approach. Ichnonetwork analysis takes advantage of the ichnological principles that makes trace fossils useful in paleoenvironmental reconstitution: among others, that trace fossils are commonly preserved in rock units that are otherwise unfossiliferous, commonly have long stratigraphic ranges, narrow environmental ranges, and are rarely transported (Bromley, 1996; Buatois and Mángano, 2011).

Similarly, ichnonetwork analysis is easily integrated with observations on physical structures (Fig. 4). In this regard, trace fossils often supply evidence of sedimentological conditions that is superior to information gained only by the study of physical structures (Howard, 1975). Ichnonetwork analysis shares this advantage with traditional ichnological methods. For example, this study provided information about salinity and oxygenation at a finer resolution than that offered by the study of physical sedimentary structures only.

At present, the ichnonetwork approach has some limitations. An ichnonetwork does not map bioturbation intensity, which is an important ecological indicator (Reineck, 1967; Taylor and Goldring, 1993; Miller and Smail, 1997; Gingras et al., 2011). In addition, an ichnonetwork does not represent information about lithofacies and body fossils, both of which are important in interpreting the paleoenvironment. In this paper, this information has been integrated with the ichnological data by semiquantitative observations on bioturbation intensity and by qualitative observation on lithofacies and body fossils (Fig. 4). However, an alternative approach would be to represent quantitatively this information as nodes, describing a multinode network.

Thanks to the features described here, ichnonetwork analysis provides information that cannot be easily gathered in another framework. An ichnonetwork describes quantitatively an ichnosite, synthesizing visually the relationships between ichnotaxa. Accordingly, the organization of the ichnosite can be explored at various domains of scale by graphical and analytical methods. Network measures describe quantitatively the role played by individual ichnotaxa within the ichnosite. For example, node betweenness measures quantitatively the reliability of an ichnotaxon as a paleoenvironmental indicator; average clustering coefficient measures the tendency to form ichnoassociations; and node layout tools allow nodes to be positioned so that patterns of association are evidenced. At the group scale, community-finding tools allow us to detect ichnoassociations objectively. In addition, ichnonetwork analysis provides information on the geographical ranges of ichnotaxa and ichnoassociations, testing the coherence of the interpreted environmental model.

CONCLUSIONS

This study documents the ichnological heritage of the Val Dolce Formation (Pontebba Supergroup) and applies ichnonetwork analysis to recognize organism-environment interactions. A clear pattern found is the increase in bioturbation intensity and ichnodiversity with increasing distance from the paleoriver mouth. Specific ichnoassociations, revealed by ichnonetwork analysis, mark different subenvironments within deltaic depositional setting. Hydrodynamic energy, freshwater input, and oxygen are the main controlling factors on the Val Dolce ichnofauna.

Some groups of co-occurring traces in the study area have no correspondence in the studied ichnonetwork. For example, isolated slabs with Pramollichnus and Psammichnites are commonly found in the easternmost part of the study area. This aspect is explained by the limits of the studied stratigraphic section, which lacks the uppermost part of the Val Dolce Formation. For this reason, further research is needed to identify and study a section representing the final part of the Val Dolce Formation. In addition, this example shows that field occurrences may differ from generalized patterns, requiring the use of different names (i.e., ichnoassemblage and ichnoassociation) to indicate these different aspects.

Ichnonetwork analysis is a powerful approach for studying fossil ichnological systems, providing a rich set of graphical and statistical tools for paleoenvironmental analysis. Nevertheless, some questions remain on monospecific ichnoassociations. An adjacency matrix based on the Jaccard index does not well represent monospecific ichnoassociations, because all elements on the main diagonal are equal to 1. Although the Jaccard index provided robust results, further investigation is required to define a framework to identify significant monospecific associations.

Identification of superfamilies of ichnonetworks may be useful in finding general patterns in ichnological systems. In line with other applications of network theory, different network-forming phenomena may leave specific architectural signatures in the corresponding graphs. In this context, global network measures (Solé and Valverde, 2004) and network motifs (Milo et al., 2002) may be useful approaches for classifying networks.

This paper quantitatively approaches the concept of ichnoassociation, which is demonstrated to have spatial, topological, and environmental aspects. This result shows that ichnoassociations arise from simple association rules based on environment, and therefore raises a question: Are ichnoassociations artificial categories that describe natural patterns, or are they emergent properties of the ecosystem?

We thank A.K. Rindsberg and an anonymous reviewer for helpful comments that improved the manuscript. We also thank Silvio Cosetti and Lucia Cossetti and the staff of the Cason di Lanza hut for hospitality and field assistance, and Paola and Gerardo for discovering trace fossil–rich sites and bringing them to our attention. Walter Francescut documented the ichnological importance of the Val Dolce Formation. We are grateful to José Gamez Vintaned and Dirk Knaust for discussions on the concept of ichnoassociation and to Ricard Solé Vincente for help on networks. We thank Giuseppe Muscio for solving legal issues on trace fossil collecting.

1Supplemental File. Ichnonetwork of the Val Dolce Formation saved in Gephi format. Gephi is needed to open the supplemental file; it can be dowloaded at http://gephi.github.io/. If you are viewing the PDF of this paper or reading it offline, please visit http://dx.doi.org/10.1130/GES00994.S1 or the full-text article on www.gsapubs.org to view the Supplemental File.