1. Introduction
Determining time of death is a significant element of solving crimes, yet it is a difficult piece of information to estimate using existing, and often fragmented, evidence collected during a death scene investigation. Patterns of postmortem insect succession and development are common tools used to help narrow the estimated time since death, a time range commonly referred to as the postmortem interval (PMI) [
1,
2]. Indeed, insects provide valuable clues towards narrowing the estimate of the time since death, however, the term PMI can be misleading as it implies the entire time elapsed since death, and not just the insect activity [
3]. Larval dipteran evidence provides information on the length of time those insects have been present on a body postmortem (i.e., how long the resource has been colonized) which, excluding cases of myiasis, does not always correspond to the actual time since death [
3,
4]. For this reason, forensic entomologists often use the term minimum postmortem interval (PMI
min) or the minimum amount of time elapsed since death as indicated by insect development [
1]. Death scenes can be highly variable in their circumstances and insect colonization patterns immediately following death cannot always be assumed. For example, the arrival of insects can be delayed if a death occurs indoors [
5], if the decedent is wrapped in cloth or like material [
6], if the remains are altered by chemicals [
7], or if the remains are burned [
8]. Therefore, we use the term post-colonization interval (PCI) here when we refer to the time from insect colonization until discovery of the body [
3].
The primary insect colonizers of decomposing bodies are blow flies (Diptera: Calliphoridae) [
9,
10]. The greatest influence on blow fly larval development is temperature, since insects are poikilothermic [
1]. Therefore, the amount of time required for calliphorid larvae to develop depends on environmental temperature. In order to standardize development time for temperature dependent development, temperature data are commonly converted to accumulated degree hours (ADH) using thermal summation models [
11,
12,
13]. However, there are many biotic and abiotic factors that can influence larval development, thus making it difficult to accurately estimate the initial time of colonization [
14]. For example, rain [
15], drugs [
16,
17,
18], predation [
19], larval sex [
20], geographic location and population genetics [
21,
22], tissue type [
23], and bacteria [
24] can all significantly affect larval development by either increasing or decreasing the time required for carrion flies to complete their life cycle.
With all of this potential for variability at death scenes, the Daubert standard [
25] and the 2009 National Academy of Sciences Report [
26] suggest that laboratory and field variability studies are important to understanding how different factors affect insect development, such as the latter in this paper. Establishing a better understanding of the naturally occurring variability and factors that influence development is necessary to improve evidence collection and analysis techniques that can estimate any indexed concept, such as PCI or PMI, as accurately as possible [
3,
27]. Investigators at the scene may only collect the largest larval specimens from a body, on the assumption that they are the most developed, and, therefore, could be missing essential information for developing a more concise PMI estimate—even though protocols suggest collecting larvae of variable sizes [
1,
2,
4,
28]. Studies have shown that a larval mass can contain multiple blow fly species with abundances that change over time [
29]. By only collecting the largest specimens, entire species that are potentially further along in their life cycle (i.e., older) but smaller in size could go unnoticed and uncollected, and thus an inaccurate estimate of the PCI would be generated based solely on the largest but perhaps not oldest larvae. For example, at 25 °C, a 10 mm
Cochliomyia macellaria (Fabricius) (Diptera: Calliphoridae) larva would indicate 58 h [
30] while a 9 mm
Phormia regina (Meigen) (Diptera: Calliphoridae) larva would indicate 66 h [
31]. Field studies like this one are important for investigating how different larval sizes (and within an instar) may lead to variable PCI estimates.
Another common practice in forensic entomology is to use the nearest, certified weather station to the death scene and previously published larval development data in order to determine insect developmental time [
1,
2]. While a weather station is the best source for historically accurate temperature data, it cannot be assumed that the station accurately reflects the microhabitat temperatures of a scene that is not near the said weather station [
14,
32]. Temperature variation between the death scene and weather station could potentially affect the accuracy of PCI estimates.
The objective of this study was to conduct a field study using multiple vertebrate carcasses to assess variation in PCI estimates derived from different temperature sources and published developmental data for two Calliphoridae species. The goal was not to build predictive models for PMI based on calliphorid life history traits, as has been done in other important studies [
33,
34,
35], but rather, to explore how collecting multiple larval sizes of calliphorid species and using temperature data from multiple sources affected the range of PCIs that were calculated using different data sources. Our overall goal was to explore how multiple sources of variation in larval size and sources of temperature (as discussed by Amendt [
4]) affected PCI calculations/estimates in a field study. ADH ranges were derived from temperature data sets acquired from instruments located at varying distances from the carcasses (i.e., individual carcass (0.90 m), microhabitat (0.90 m average of six carcasses), local (1.63 km), regional (7.61 km)), and used to calculate PCI estimates in order to evaluate how temperature sources influence ADH associated PCI estimates. We hypothesized that different Calliphoridae species would colonize the resource at different times resulting in larval size variation on and among carcasses, and that the temperature source nearest the carcasses (microhabitat) would provide temperatures most similar to what the larvae experience. Therefore, we predicted that different species, and larval sizes, would yield different PCI estimates and that the data from the closest temperature source to the carcasses would result in the most accurate PCI estimate. However, we also predicted that PCI estimates, regardless of temperature source utilized, would differ from the actual PCI due to differences in the published laboratory conditions (where developmental data were generated) and the field study conditions.
4. Discussion
Recently, the forensic science community has increased efforts to recognize the amount of variability in specific disciplines (e.g., trace evidence) and develop error rates [
26]. Yet, it is still relatively unknown how the accuracy of using a calliphorid developmental data set to estimate a PCI varies depending on temperature source. Utilizing weather station data distant from the scene can be problematic, as possible temperature variations between the location of the station and the death scene could result in an inaccurate calculation of insect developmental time [
14,
32]. Archer [
59] examined the accuracy of collecting scene temperatures after body removal and retrospectively correcting weather station data. In that study, often the minimum PMI estimates improved following correction, but temperature collection after body discovery is not always possible. Archer [
59] urged caution when correcting data, since weather can change significantly over a short period of time. Improvement of PMI estimates was highly variable among correlation periods and in a few instances, temperature correction was associated with a decrease in estimated decomposition time accuracy. Additionally, in that excellent study only hypothetical insect data were analyzed [
59].
Monthei [
51] conducted a similar study to ours in Virginia in which accumulated degree days calculated from temperatures at the carcass (ambient, inside the cage, head, thorax, abdomen) and from two weather stations at different distances from the site (5.63 and 10.46 km) were used along with two
P. regina developmental data sets [
31,
49] to estimate PMIs. The author reported that, when using Byrd and Allen’s data set [
31], the furthest weather station from the study site produced the most accurate PMI estimate. When utilizing Anderson’s data set [
49], however, temperatures from the head of the carcass produced the most accurate PMI estimate. Monthei postulated that this could be due to the fact that Anderson’s data came from larvae reared in masses while Byrd and Allen kept a ratio of 1.5 larvae per 1.0 g of pork [
51]. Indeed, the role of larval mass heat from thermogenesis may both increase and stabilize temperatures that are best reflected in temperature conditions from weather stations; this would be a fruitful area of research comparing multiple locations in different geographic regions.
Contrary to our hypothesis, using temperature sources nearer to the carcasses did not result in increased expected ADH range accuracy or the associated estimates of the PCI. ADH ranges determined using published developmental data sets were surprisingly more similar to the actual third instar ADH ranges when weather station temperatures were used rather than temperatures from the scene. The expected ADH ranges under-represented the actual range when microhabitat temperatures were used: the ADH range using microhabitat temperatures was greater than the weather station ranges. This wider ADH range at the microhabitat may be explained by the temperature loggers being located closer to the ground (and thus exposed to increased radiant and reflective heat) than the weather stations, resulting in higher temperatures and thus more rapid accumulation of degree hours [
1]. The greatest difference in temperatures between the sources occurred during the afternoon, when temperatures were highest. Additionally, developmental time is longer for
C. macellaria and
P. regina under cyclic temperatures than at a constant temperature [
19]. This could explain why the expected ADH ranges under-represented the actual ADH ranges.
Regardless of the temperature source used to calculate ADH ranges for use in PCI estimates, we hypothesized that the artificial circumstances (i.e., controlled growth chamber) of the developmental data sets would result in differences in the estimated PCIs [
30,
31] compared to the actual PCIs based on 12 h interval field observations and sample collections [
1]. Constant temperature conditions and other variables (e.g., humidity) of lab conditions are very different from field conditions where temperatures naturally fluctuate and development can be influenced by additional abiotic variables (e.g., precipitation). Our findings suggest that there were fewer extreme fluctuations in temperature from the weather stations than at the microhabitat, making it more similar to the laboratory setting where conditions were constant. These two sources of error—utilizing expected ADH ranges from data sets from conditions different than those experienced by larvae at the scene, and weather station temperatures that vary from those at the scene—may be canceling each other in PCI estimates: one may overestimate while the other underestimates development time [
60].
Another notable finding was the variation in species and third instar length. This supports the importance of thorough evidence collection in order to utilize as much information available from a death scene as possible. The largest larvae may be the most developed and therefore an initial colonizer, but the presence of smaller third instars, of either the same or different species, could indicate multiple oviposition events that could help develop a more complete timeline of the post-colonization activity. Third instars were present for a wide range of time (hours), yet the relative abundances of this life stage compared to other instars shifted throughout decomposition. There was an increase in relative abundance of first and second instars at 96 h and then an increase in abundance of third instar
P. regina at 120 h, suggesting a second oviposition event occurred later in the decomposition process [
13]. Considering smaller third instars of another species collected from a body could be informative, as different species colonize at different times following death. Further,
P. regina is known as a later colonizer during carrion decomposition, compared to
L. coeruleiviridis,
C. macellaria, and
Chrysomya rufifacies (Macquart) (Diptera: Calliphoridae) [
29], and prefers a resource that has been previously colonized [
61]. We initially observed a high average relative abundance of
C. macellaria third instars (59.51%) early on (48 h) followed by an increase in
P. regina third instar average relative abundance (55.01% at 72 h) (
Figure 4). Using the 50th and 90th percentile lengths of third instars was also effective at indicating the lower ADH and PCI range, but failed to encompass the upper ADH and PCI ranges. A possible explanation could be found in the preservation method. Specimens were frozen and then placed in ethanol, which could have led to shrinkages and underestimated ranges based on larval length [
62]. It would be useful to replicate this study and compare PCI estimates derived from larval body length among different preservation methods. This knowledge would be valuable as experience and technique varies among investigators, and it is not uncommon for an entomologist to receive poorly preserved specimens.
Finally, the Calliphoridae collected during our study are similar to the species reported for a previous study conducted in the same geographic location [
63]. This previous study also used swine carcasses and found
P. regina and
L. coeruleiviridis as colonizers (evident by larvae present on the carcasses) with additional blow fly species present as adults:
C. macellaria, Hydroatea leucostoma (Fabricius) (Diptera: Calliphoridae), and
Pollenia rudis (Fabricius) (Diptera: Calliphoridae) [
63].
Phormia regina is considered a cold weather species, while
C. macellaria is more abundant in warmer regions, however, both are common postmortem colonizers [
30,
31]. Thus, our findings are consistent with previous descriptions of the distribution and the ecology of the species that we collected in this study.