Accounting for uncertainty from zero inflation and overdispersion in paleoecological studies of predation using a hierarchical Bayesian framework

Abstract

The effects of overdispersion and zero inflation (e.g., poor model fits) can result in misinterpretation in studies using count data. These effects have not been evaluated in paleoecological studies of predation and are further complicated by preservational bias and time averaging. We develop a hierarchical Bayesian framework to account for uncertainty from overdispersion and zero inflation in estimates of specimen and predation trace counts. We demonstrate its application using published data on drilling predators and their prey in time-averaged death assemblages from the Great Barrier Reef, Australia. Our results indicate that estimates of predation frequencies are underestimated when zero inflation is not considered, and this effect is likely compounded by removal of individuals and predation traces via preservational bias. Time averaging likely reduces zero inflation via accumulation of rare taxa and events; however, it increases the uncertainty in comparisons between assemblages by introducing variability in sampling effort. That is, there is an analytical cost with time-averaged count data, manifesting as broader confidence regions. Ecological inferences in paleoecology can be strengthened by accounting for the uncertainty inherent to paleoecological count data and the sampling processes by which they are generated.

Community evenness and sample size affect estimates of predation intensity and prey selection: A model-based validation

“…Inferred predation intensity (PI T.inf ) for the drawn sample is calculated as a ratio of the number of attacked individuals and the total number of individuals (i.e., 100 in the first draw). We kept the step size as 100 to gain an accurate representation of predation intensity and to avoid the issues related to insufficient sample size (Kosloski et al 2008; Dietl and Kosloski 2013; Smith et al 2022). The exact process is repeated 30 times without replacement until all the individuals from the assemblage are sampled.…”

“…In contrast to the bulk collection, targeted sampling of specific size class or taxon impacts inferred predation intensities (Kowalewski and Hoffmeister 2003; Kosloski et al 2008; Ottens et al 2012; Hattori et al 2014; Chattopadhyay et al 2016; Hausmann et al 2018). Theoretical investigations also demonstrated the effect of sample size on inferred predation intensity (Smith et al 2018, 2022). Analytical techniques to evaluate and compare predation measures across groups often impact the inferences (Kowalewski 2002; Leighton 2002; Grey et al 2006; Stafford and Leighton 2011; Dietl and Kosloski 2013; Smith et al 2018; Budd and Mann 2019).…”

“…A few studies have examined the impact of size-specific sampling on inferred predation intensities (Chattopadhyay et al, 2016; Hattori et al, 2014; Hausmann et al, 2018; Kowalewski and Hoffmeister, 2003; Ottens et al, 2012; Visaggi and Kelley, 2015). Smith et al (2021) demonstrated the effects of overdispersion and zero inflation (e.g., the poor model fits) using count data of predation traces. Aspects of a specific community such as evenness, selectivity of predation, and sampling intensity may influence the estimates of predation intensity.…”

Assessing the utility of death assemblages as reference conditions in a common benthic index (M-AMBI) with simulations

“…Before death assemblages and other sources of geohistorical data can fulfill this potential, it is critical to evaluate the consequences of taphonomic processes that may bias the accumulation of individuals. Though there is a robust literature on these processes and biases (e.g., Cummins et al 1986; Davies et al 1989; Kowalewski et al 1998; Behrensmeyer et al 2005; Lockwood and Chastant 2006; Olszewski and Kidwell 2007; Kosnik et al 2009; Powell et al 2011; Kidwell and Tomašovỳch 2013; Tomašovỳch et al 2016; Smith et al 2021), they have not been considered with respect to the use of benthic indices like M-AMBI (Smith et al 2020).…”

“…Absolute abundances of taxa in death assemblages are often skewed with respect to abundances in living assemblages, although rank- order abundance in the death assemblage tends to be more resistant to taphonomic bias than absolute abundance when the environment is stable (e.g., Kidwell 2002, 2007). Many factors can influence abundances in the death assemblage (e.g., Cummins et al 1986; Kidwell 2002, 2007; Olszewski and Kidwell 2007; Kosnik et al 2009; Kidwell and Rothfus 2010; Powell et al 2011; Olszewski 2012; Tomašovỳch et al 2016, 2018; Smith et al 2019, 2021), including differences in the life spans of species and the consequent rate at which dead individuals enter the death assemblage (Cronin et al 2018). Thus, the relationship between preservation of absolute abundance and EcoQS is likely to be variable and dependent on characteristics of particular species in an assemblage.…”