Forensic Science International 236 (2014) 1–9

Contents lists available at ScienceDirect

Forensic Science International journal homepage: www.elsevier.com/locate/forsciint

Species composition of forensically important blow flies (Diptera: Calliphoridae) and flesh flies (Diptera: Sarcophagidae) through space and time Heike Fremdt a,b,*, Jens Amendt a a b

Institute of Forensic Medicine, Goethe-University Frankfurt, Kennedyallee 104, D-60596 Frankfurt am Main, Germany Department of Aquatic Ecotoxicology, Goethe-University Frankfurt, Max-von-Laue-Strasse 13, D-60438 Frankfurt am Main, Germany

A R T I C L E I N F O

A B S T R A C T

Article history: Received 11 February 2013 Received in revised form 13 November 2013 Accepted 11 December 2013 Available online 19 December 2013

Weekly monitoring of forensically important flight-active blow flies (Diptera: Calliphoridae) and flesh flies (Diptera: Sarcophagidae) was performed using small baited traps. Sampling took place in two rural, one suburban and two urban habitats in and around Frankfurt (Main), Germany, lasting two years and eight months. Highest values for species richness and Chao–Shen entropy estimator for Shannon’s index in both families were found at the urban sites, peaking during summer. Space–time interaction was tested and found to be significant, demonstrating the value of a statistical approach recently developed for community surveys in ecology. K-means partitioning and analysis of indicator species gave significant temporal and habitat associations of particular taxa. Calliphora vicina was an indicator species for lower temperatures without being associated with a particular habitat. Lucilia sericata was an indicator for urban sites, whereas Lucilia ampullacea and Lucilia caesar were indicators for rural sites, supplemented by the less frequent species Calliphora vomitoria. Sarcophagidae were observed during a clearly shorter period of year. Sarcophaga subvicina þ Sarcophaga variegata was found to be an indicator for urban habitats during summer as well as Sarcophaga albiceps for rural habitats. A significant association of Sarcophaga caerulescens to rural habitats as well as one of Sarcophaga similis to urban habitats was observed. ß 2013 Elsevier Ireland Ltd. All rights reserved.

Keywords: Diversity Forensic entomology Habitat association Seasonal pattern Space–time interaction

1. Introduction Necrophagous flies play a prominent role as potential vectors of human [1] as well as livestock-disease [2], but are also one of the main actors in forensic entomology [3]. A main task of forensic entomology is to estimate the minimum time since death and here, Calliphoridae and Sarcophagidae species are especially important, because they may colonise a body very early, i.e. most closely to the time of death [4]. Compared to other necrophagous flies, the duration of development of their immature stages is best understood. But our knowledge about the spatial and temporal distributions of these specialised taxa is still poor, which is regrettable for other important questions as, e.g. the interpretation of a delayed or even failed colonisation of a murder victim due to special phenological features of the relevant species or the proof of

* Corresponding author at: Institute of Forensic Medicine, Goethe-University Frankfurt, Kennedyallee 104, D-60596 Frankfurt am Main, Germany. Tel.: þ49 69 6301 7597; fax: þ49 69 6301 5882. E-mail addresses: [email protected], [email protected] (H. Fremdt). 0379-0738/$ – see front matter ß 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.forsciint.2013.12.010

a postmortal relocation of a body from one habitat to another by the occurrence of certain taxa on the cadaver. Unfortunately, there is a lack of appropriate detailed long-term studies. So far, data for occurrence and activity patterns of necrophagous insects come mainly from succession studies on pig models [5,6], but also smaller vertebrate carcasses, e.g. dogs [7], rats [8], rabbits [9] and chickens [10]. Those succession studies usually just cover selected seasons of the year or an even smaller window of time. The aim of such experiments is to analyse the change of a decomposing cadaver [11] and the related variation of its necrophagous fauna [12]. The phenology and habitat association for certain ecological guilds or insect families is not covered accurately by this approach. A collection of early colonising necrophagous Diptera with baited traps or nets [13–16] is a more systematic way of assessing phenology and habitat association and documents the present fauna of a certain habitat much more precisely. The systematic surveillance of fly species produces a bulk of data to be analysed. So far, surveillance data of fly species has been analysed using descriptive methods (e.g., [17]) or by testing spatial and temporal variability independently of each other [15]. Since those data lack replication, one cannot test a space–time interaction by statistical methods like a classical ANOVA [18].

H. Fremdt, J. Amendt / Forensic Science International 236 (2014) 1–9

2

Conclusions concerning the spatio-temporal occurrence pattern of important species may only be drawn with caution, since a space– time interaction can not be assumed to be significant, even if temporal and spatial effects are tested to be so. To understand the ecological correlates of species, one needs to test whether spatial structures change during time and whether temporal structures differ from site to site. This has recently become possible by the use of an ANOVA-model developed by Legendre et al. [18]. The present study surveyed the activity of necrophagous Calliphoridae and Sarcophagidae at five sites in and around Frankfurt am Main, Germany, representing different kinds of habitats. The experiment lasted over a period of two years and eight months. The method of Legendre et al. [18] for testing the space–time interaction of a community was adapted for use for the first time in forensic entomology, to analyse habitat associations and temporal patterns of the detected species.

2. Materials and methods

Fig. 1. Catope-watertrap modified after Ho¨ller et al. [19].

2.1. Experimental design Catope-watertraps were built following the method of Ho¨ller et al. [19] with the sole modification of placing the bait on top of a small tube protruding from the trapping liquid (Fig. 1). This allows trapping of insects on the wing by attracting them to a bait without allowing them a real opportunity for landing, which finally leads to liquid drowning. The trapping liquid consisted of water including odourless soap to eliminate surface tension. If temperatures below freezing were expected, glycerine was added to avoid freezing and enable sampling. Traps were emptied at intervals of 7 days by pouring all contents through a fine-meshed sieve. Insects were transferred to tubes with forceps and stored in 95% EtOH until identification. Ca. 10 g beef liver in combination with ca. 500 ml dimethyl trisulfide (1:10 in MeOH; Sigma–Aldrich) was used as bait and was renewed every week after removing the insects. Traps were placed approximately 1 m above ground to prevent crawling insects from entering. Sampling was conducted at five sites, representing two urban, one suburban, one semi-natural and one rural habitat (Table 1, Fig. 2). The survey lasted from September 1st 2008 to May 3rd 2011. Thus, there were 139 sampling occasions in total. Keys by Rognes [20] and Pape [21] were used for morphological identification of Calliphoridae and Sarcophagidae, respectively. Since females of the Sarcophaga carnaria-group can not be separated by morphological characters, the barcoding region was sequenced according to Folmer et al. [22]. For a detailed protocol see Fremdt et al. [23]. Specimens were identified by editing the sequences and aligning them with those published by Jordaens et al. [24] using Sequence analysis software (Applied Biosystems, Version 2.1.2) and Sequence Navigator software (Applied Biosystems, Version 1.0.1).

2.2. Data analysis All statistics were performed using R [25]. To prevent an interference of data calculated for Calliphoridae by those of Sarcophagidae and vice versa, families were analysed separately. That way, significant patterns inbetween a family were not masked and accurate family specific occurrence patterns were given. Species richness, Shannon’s index [26] and Pielou’s eveness index [27] were calculated as diversity indices. Species richness S is defined by S ¼ q and is simply a count of species being present at a sampling site [28]. Shannon’s index or Shannon’s entropy H is widely used to measure the biological diversity and is defined by q X pi logð pi Þ; H¼ i¼1

where pi is the relative frequency of species i [28]. The minimum value H ¼ 0 is given for q ¼ 1 and H increases with the number of species. For a given number of species, e.g. q ¼ 10, H increases to its maximum H ¼ logðqÞ ¼ logð10Þ when species are distributed equally, and decreases with an increasing unequability, i.e. a dominance by few species. Since sampling design and species abundance patterns suggested that not all species present at a location were actually detected during a sampling occasion, the ˆ was Chao–Shen entropy estimator for Shannon’s index (H) calculated rather than just Shannon’s index solely, since it accounts for unseen species (for a formal definition see Chao and Shen [29]). The eveness of species quantifies the numerical equitability of a community and is calculated by Pielou’s eveness index J [28]. It is defined by

Table 1 Description of sampled sites. Location

Latitude (þ50 )

Longitude (þ8 )

Description and definition

1

70 34.4700 N

390 33.0200 E

2

60 58.7400 N

380 23.6000 E

3

50 37.8600 N

390 56.6500 E

4

40 46.7500 N

400 5.6700 E

5

40 34.7600 N

400 15.9300 E

An urban habitat in the northern city. Extension of a park. Adjoined to office buildings of the JWG-University as well as private houses. An urban habitat in the northwestern city. Surrounded by office buildings and few private gardens with mowed lawn and few hedgerows, mostly consisting of Rubus-bushes. A suburban habitat in the southwestern city. Private houses and few gardens with few commercial activities. A semi-natural habitat to the south of the city. Mixed forest. Abundantly covered with vegetation and adjoining to allotment gardens. A rural habitat to the south of the city. Mixed forest. In relation to location 4 a little less abundantly covered with vegetation and with little human interference.

H. Fremdt, J. Amendt / Forensic Science International 236 (2014) 1–9

3

between a species and a group or a combination of groups as a measure of the association and uses a permutation test to test the statistical significance of the association with the highest indicator value [32,33]. The indicator value is a product of two components: the positive predictive value A and the sensitivity B. In the present study, A is the probability that a particular space–time object belongs to the target group if the species has been found, and B is the probability of species presence in objects of the group [32]. Finally, the function strassoc of the indicspecies package was used to derive confidence intervals of indicator values as well as values of A and B by bootstrapping [32]. Numerically dominant species and indicator species of each group allowed for an interpretation of the spatio-temporal pattern. 3. Results

Fig. 2. Map of sampled sites according to geographical coordinates (cf. Table 1).

J¼

H ¼ Hmax

! q X  pi logð pi Þ logðqÞ

During the 139 sampling occasions a total of 14,355 Calliphoridae and 505 Sarcophagidae were collected. Those were identified to 17 species in five subfamilies of Calliphoridae and 13 species in one subfamily of Sarcophagidae (Table 2). During the whole sampling period Lucilia sericata (38.01%) represented the species with largest number of specimens in family Calliphoridae, followed by Calliphora vicina (33.47%) and L. ampullacea (9.85%; Fig. 3). The most frequent species was C. vicina except for location 3, where L. sericata occurred more frequently. The blow flies Bellardia pandia, Lucilia silvarum, Melinda viridicyanea and Pollenia

i¼1

which compares species diversity to its corresponding maximum value [28]. For reasons of simplicity, all values are given for seasons instead of weeks (Figs. 4 and 5). According to the meterological seasons, March–May were defined to be spring, June–August to be summer, September–November to be autumn and December–February to be winter. Thus, diversity indices show pooled data of 3 years in autumn and winter, for almost 3 years in spring and only 2 years in summer. The presence of a space–time interaction was tested using Model 5 of the STI package for R, which is described by Legendre et al. [18] as a two-way ANOVA crossed design with interaction under-fitted. This model is recommended particularly for data without replication (see Legendre et al. [18] for details of the model). To approximate a normal distribution, the abundance data was log(y þ 1) transformed before being subjected to Model 5. If the hypothesis of no interaction was rejected, similar space–time objects were grouped in the next step using K-means partitioning [18]. Via this particular method of cluster analysis objects of a data set are partitioned into a given number of groups by an iteration of the following steps: An object is assigned to the group whose mean is closest to it, then new means of groups are calculated before the next assignment step is performed [28]. Hence, the objects in a group have higher similarity to one another than to the objects in the other groups. Since the raw data is frequency data, for which an Euclidean distance is unsuitable, raw data were Hellinger transformed first, to gain a distance matrix among the space– time objects [28]. The best partition was chosen using the cascadeKM function of the vegan package with the simple structure index (ssi) as the determination criterion [30]. Several elements influencing the interpretability of a particular partition are combined multiplicatively by the ssi [31]. The number of groups as an input parameter for K-means partitioning was chosen as indicated to be best by the highest ssi [27]. Groups were subsequently connected to the raw data to identify indicator species by using the indicspecies package for R [32,33]. The function multipatt calculates indicator value indices

Table 2 List of Calliphoridae and Sarcophagidae detected during the experiment with percentages of sampling occasions at which species were present at each site. For description of locations cf. Table 1. location Calliphoridae Calliphorinae B. pandia (WALKER 1849) C. vicina ROBINEAU-DESVOIDY 1830 C. vomitoria (LINNAEUS 1758) C. mortuorum (LINNAEUS 1761) Chrysomyinae P. regina (MEIGEN 1826) P. terraenovae (ROBINEAU-DESVOIDY 1830) Luciliinae L. ampullacea VILLENEUVE 1922 L. caesar (LINNAEUS 1758) L. illustris (MEIGEN 1826) L. sericata (MEIGEN 1826) L. silvarum (MEIGEN 1826) Melanomyinae M. viridicyanea (ROBINEAU-DESVOIDY 1830) Polleniinae P. angustigena WAINWRIGHT 1940 P. hungarica ROGNES 1987 P. pediculata MACQUART 1834 P. rudis (FABRICIUS 1794) P. vagabunda (MEIGEN 1826 Sarcophagidae Sarcophaginae S. africa (WIEDEMANN 1824) S. melanura MEIGEN 1826 S. noverca RONDANI 1860 S. vagans MEIGEN 1826 S. argyrostoma (ROBINEAU-DESVOIDY 1830) S. sexpunctata (FABRICIUS 1805) S. similis MEADE 1876 S. albiceps MEIGEN 1826 S. caerulescens ZETTERSTEDT 1838 S. carnaria (LINNAEUS 1758) S. subvicina ROHDENDORF 1937 S. variegata (SCOPOLI 1763) S. incisilobata PANDELLE 1896

1

2

3

4

5

5 64.7 0.7 2.2

0 63.3 1.4 0

1.4 41 3.6 0

0 52.5 15.8 0.7

0 39.6 14.4 2.2

5.8 14.4

2.9 22.3

6.5 12.9

4.3 1.4

2.9 4.3

39.6 29.5 20.9 43.2 3.6

33.8 15.1 15.8 33.1 0.7

2.2 8.6 12.2 44.6 1.4

26.6 23 8.6 18.7 0

21.6 18.7 2.9 6.5 0

5.8

1.4

0

0

0

0.7 4.3 11.5 5.8 0.7

0 0 4.3 0.7 0

0 2.2 5.8 4.3 0

0 1.4 1.4 4.3 0

0 2.2 0 2.9 1.4

1.4 0 2.2 0.7 0.7 0 5 3.6 0 7.2 22.3 11.5 0.7

0.7 0.7 0 0 0 3.6 1.4 6.5 0.7 7.9 10.8 11.5 0.7

3.6 0 0 0 0 0 0.7 6.5 0 4.3 6.5 12.2 0.7

0 0 0 0 0 0 0 6.5 3.6 2.9 0 2.9 0

0 0 0 0 0 0 0 2.9 4.3 3.6 2.2 5.8 0

4

H. Fremdt, J. Amendt / Forensic Science International 236 (2014) 1–9

Fig. 3. Relative abundance of (A) Calliphoridae and (B) Sarcophagidae in total. Species representing below 0.4% of Calliphoridae and 1% of Sarcophagidae were pooled.

angustigena were present at the urban and suburban sites, exclusively (Table 2). In the Sarcophagidae, Sarcophaga subvicina (35.65%), Sarcophaga variegata (29.11%) and Sarcophaga albiceps (13.07%) were numerically dominant (Fig. 3). S. variegata was the commonest species at location 2, 3 and 5, whereas S. subvicina and S. albiceps showed highest frequency at locations 1 and 4, respectively. Of the 13 species only S. albiceps, Sarcophaga caerulescens, S. carnaria, S. subvicina and S. variegata were observed at the rural locations (Table 2). 3.1. Diversity Total species richness for both, Calliphoridae and Sarcophagidae, was highest at the urban sites. At location 1, a total of 17 Calliphoridae species was found, whereas total species richness was relatively uniform at the remaining sites with 12 and 13 species (cf. Table 2). For the Sarcophagidae a decreasing number of species was obtained along the urban-rural gradient. Whereas 10 species were found at urban location 1 and 2, only 7 species were obtained at the suburban site, decreasing to only 4 and 5 species in the forest habitats 4 and 5, respectively (cf. Table 2). For reasons of simplicity, the seasonal splitting of species richness is shown for pooled data of all years (Figs. 4 and 5), i.e. values for autumn and winter are of 3 years, values for spring for almost 3 years and values for summer for only 2 years. No general gradient from urban to rural sites or vice versa was visible and no

temporal gradient was obtained from this data, either. Those data reflect the local and seasonal patterns of diversity, although some annual characteristics were determined to be masked. For Calliphoridae, peaks of species richness were detected at location 1 and 3 in summer and at remaining sites in spring as well as in summer (Fig. 4). A closer look at the yearly separated data showed that in fact species richness was highest at all sites in summer within one year. One single unusual high count of species at sites 2 (10 species), 4 (11 species) and 5 (10 species) in spring caused the before mentioned shift in the general pattern. At site 5 number of species was found to be almost continuously 4 during the first year with only 1 species present during winter. It is particularly striking that the Chao–Shen entropy estimator for pooled data suggests an increase of species evenness at location 1–4 and a decrease at location 5 from spring to summer (Fig. 4). Except of location 5 (1.42 in spring) Chao–Shen’s entropy estimator was highest in summer (1.84 at location 1, 1.52 at location 2, 0.7 at location 3 and 1.39 at location 4; Fig. 4), but Pielou’s eveness was only highest in summer at the urban sites, whereas it increased over the year at the rural sites (data not shown). At the suburban site pooling of data masks, that Chao-Shen’s entropy estimator (Hˆ ¼ 0:70) and Pielou’s eveness index (J ¼ 0:39) peaked in spring of the first year, whereas both values increased steadily from spring to autumn during the next year (Hˆ ¼ 1:00; J ¼ 0:62 in autumn). In Sarcophagidae species richness clearly peaked at all sites in summer (10 species at locations 1 and 2, 7 species at location 3, 4 species at location 4 and 5 species at location 5; Fig. 5). With exception of site 5, Chao–Shen entropy estimator was highest in summer as well (1.47 at location 1, 1.70 at location 2, 1.44 at location 3 and 1.15 at location 4), which is confirmed by the annual data. Location 5 appears to be a special case, giving the highest value for Chao–Shen entropy estimator in spring (Hˆ ¼ 1:95), what suggests a lower eveness in summer with richness increasing only slightly. These data mask a total absence of Sarcophagidae at this site in spring of year 2. Nevertheless, when Sarcophagidae were present in spring, Pielou’s eveness index (data not shown) as well as Chao–Shen entropy estimator were highest then. In general, Sarcophagidae were much less abundant than Calliphoridae and can be characterised by a clear association with the summer months. If any, the lesser number of species during the remaining seasons varied strongly inbetween the years. 3.2. Spatio-temporal pattern of Calliphoridae The space–time interaction was tested to be highly significant (R2 = 0.218, P = 0.001 after 999 permutations). To obtain spatiotemporal patterns, a K-means partition into seven groups was conducted, since this showed the highest ssi. A space–time map visualises the results (Fig. 6). During autumn (sampling occasions 01–13 of each year) most space–time objects, especially those at the urban and suburban location, were assigned to group 6. At locations 1 and 2, mainly group 3 completed the pattern, whereas group 4 and group 7 undertook this part at location 3 and locations 4 and 5, respectively. Groups 1 and 6 almost exclusively framed the period of no Calliphoridae presence during the winter months (sampling occasions 14–26). The change from spring (sampling occasions 27– 39) to summer (sampling occasions 40–52) at the urban sites was characterised by a shift of dominance from groups 1 and 6 to group 2, appearing earlier at location 1. Group 4 represented most of the sampling occasions from mid spring to mid autumn at location 3, whereas at the rural sites a temporal gradient from group 1 over groups 6 and 3 to group 7 was characteristic. Particularly in the summer months, locations were characterised by groups, which were mainly restricted to the respective sites. This was group 2 at the urban sites, group 4 at the suburban site and group 7 at the

H. Fremdt, J. Amendt / Forensic Science International 236 (2014) 1–9

5

Fig. 4. Maps of Calliphoridae diversity in (A) spring, (B) summer, (C) autumn and (D) winter. The locations are visualised according to their geographical coordinates (cf. Table 1 and Fig. 2). The bubble sizes visualise the species richness with the smallest bubble representing 1 and the biggest bubble representing 13, whereas numbers give the values of Chao–Shen entropy estimator for Shannon’s index. Note that data were pooled of 3 years for autumn and winter, of almost 3 years for spring and only of 2 years for summer.

rural sites. Except for location 3, a framing of such a summerly prominent and site specific group by few records of group 3 could be seen. Group 5 was the only group irregularly scattered through the time axis, though it was restricted to the rural locations 4 and 5. By describing the groups due to the indicator values, as well as the numerical dominance, the spatio-temporal pattern of the species themselves can be obtained. During the winter months, C. vicina was the commonest species by far and only joined time by time by single occurrences of some common species in principle (group 6) or of some rare species (group 1). Since C. vicina was the commonest species in general, reflected by the fact that it is associated with six groups, and other species in groups 1 and 6 were too sparse, no indicator species was found for these two groups. Though the unspecific association of C. vicina caused a small positive predictive value for group 6 (i.e. if C. vicina was present, there was only a relatively low probability, that this object belonged to the target group), the sensitivity value (B) was 1, i.e. C. vicina was present in objects of group 6 with a probability of 100%. A high abundance of C. vicina in combination with one of L. sericata was characteristic for the urban locations, mostly notable in summer (group 2). Due to very high values of A and B, this combination was found to be a significant indicator covering 100% of objects belonging to the target object group. At the suburban site, the summer fauna was strongly dominated by L. sericata alone, being a highly significant indicator for group 4. Finally, the highest summerly abundance at the rural site was shown by L. ampullacea

and L. caesar. The combined presence of both, as well as L. caesar solely were found to be significant indicators of group 7 and covered 92% of the objects belonging to this group. During the periods of transition to increasing blow fly abundance in spring and to decreasing blow fly abundance in autumn, the fauna was, at all sites except of location 3, again dominated by C. vicina, but with a high abundance of L. ampullacea (group 3). Due to the relatively low values of A, no significance was obtained for this group, although maximum values of B were found for both species. Though no temporal pattern was obtained, group 5 was found at the rural sites, exclusively, indicating high abundance values of C. vicina followed by those of Calliphora vomitoria and L. caesar, whereas C. vomitoria was a highly significant indicator species for this group. 3.3. Spatio-temporal pattern of Sarcophagidae Generally, Sarcophagidae were observed on much fewer sampling occasions. Thus, the time period during harsh temperatures, when no Sarcophagidae were found, lasted at least 25 weeks, i.e. almost half a year. The space–time interaction was again tested to be highly significant (R2 = 0.242, P = 0.001 after 999 permutations). Highest ssi was found for a K-means partition into eight groups. The spatio-temporal pattern is shown in Fig. 7. Group 7 was set to be a special group indicating a lack of Sarcophagidae occurrence, when Sarcophagidae were found at

6

H. Fremdt, J. Amendt / Forensic Science International 236 (2014) 1–9

Fig. 5. Maps of Sarcophagidae diversity in (A) spring, (B) summer, (C) autumn and (D) winter. The locations are visualised according to their geographical coordinates (cf. Table 1 and Fig. 2). The bubble sizes visualise the richness with the smallest bubble representing 1 and the biggest bubble representing 10, whereas numbers give the values of Chao–Shen entropy estimator for Shannon’s index. Note that data were pooled of 3 years for autumn and winter, of almost 3 years for spring and only of 2 years for summer.

other sites in the same sampling occasion. This makes location 1 the site with the longest Sarcophagidae occurrence, whereas at locations 3 and 5 only few Sarcophagidae were found in year 2 and year 1, respectively. However, at the urban and suburban locations group 3 dominated the pattern during summer months, showing a temporal occurrence-gradient from urban to suburban. At location 1 group 3 was displaced by group 2 during autumn, whereas only very few Sarcophagidae were observed at the other sites during the same period. Group 4 was characteristic for the rural sites, whereas this pattern appeared earlier at location 5 than at location 4. An irregular temporal pattern was found to be typical for the four additional groups, being found sparse. Nevertheless, it is worth mentioning that group 1 was restricted to the rural sites, group 6 to the (sub)urban sites, whereas groups 5 and 8 were found everywhere except location 4. Interpreting those patterns due to the numerical dominance and indicator values of the species, S. variegata and S. subvicina appeared to undertake the role of C. vicina in Calliphoridae: Both were associated to three groups, describing together approximately half of the sampling occasions. The species composition at the urban and suburban sites were characterised by a high abundance of both, arising later at the suburban site (group 3). This species combination was found to be a significant indicator for this group, due to very high values of A and B, and covered 100% of objects. At location 1, abundance of S. variegata decreased during autumn, resulting in a numerical dominance of S. subvicina alone (group 2).

S. albiceps and S. variegata were the commonest summerly species at the rural sites, showing that the presence of S. albiceps alone covered 100% of objects belonging to group 4. Altogether, the spatio-temporal pattern in Sarcophagidae was not that obvious as in Calliphoridae. The relatively high amount of four groups appearing without being strictly associated with a particular spatial or temporal occasion reflected that fact. However, since groups 5 and 8 were detected at every site with the exception of location 4, an habitat association was rejected for S. carnaria and S. subvicina (group 5) as well as S. variegata (group 8). Due to their group associations, an association with rural habitats was assumed for S. caerulescens (group 1), whereas Sarcophaga similis appeared to occur at urban and suburban sites (group 6). 4. Discussion General patterns in spatio-temporal abundances must be obtained to prove statistically whether fly species show habitat or seasonal associations. Due to the long sampling period and large number of target species, a huge data set resulted. A descriptive analysis of richness and diversity indices was used to explore the data and to obtain first patterns, such as, e.g., an association with urban locations by Calliphoridae as well as Sarcophagidae. In this regard, some rare species, Bellardia pandia, Melinda viridicyanea and Lucilia silvarum (Diptera: Calliphoridae), as well as Sarcophaga africa, S. noverca, and S. sexpunctata (Diptera: Sarcophagidae), were

H. Fremdt, J. Amendt / Forensic Science International 236 (2014) 1–9

7

Fig. 6. Spatiotemporal map visualising the K-means partition of Calliphoridae occurrences into seven groups. The x-axis shows weeks of sampling. Note that the sampling was started in September (sampling week 1). A description of locations can be found in Table 1. Symbols for groups are: 1, blue; 2, red; 3, yellow; 4, dark red; 5, green; 6, dark green; 7, orange.

found to be clearly associated with the urban locations. Since rare species may have an edge over usually dominating species in spatial or temporal occasions, in which the environmental conditions disadvantage the common species, these findings are of particular interest to forensic entomology, because the presence of rare species may prove the place or time of the colonisation. Even species which are not obligate colonisers of carrion can be of forensic importance, as shown recently for L. silvarum [23] and M. viridicyanea [6]. Despite their ecological relevance, rare species are often ignored in forensic entomological research [15,16,34,6], since the statistical methods applied, such as canonical correspondence analysis, are sensitive to those species [15]. Testing an assumed effect of rare species on space–time interaction is only recently possible using the STI model proposed by Legendre et al. [18] by considering the whole data set. The applied ANOVA-model is recommended for data without replication, which is of great importance especially for this kind of survey study. The problem of replicates required for classical ANOVA, might be handled by combining neighbouring sampling sites into strata [35]. Mainly at urban locations this is not practicable for studies surveying necrophagous species, because an odour nuisance results from the bait and species composition of locations separated only by short distances may differ, which would be masked when pooling those data. However, since the experimental design of survey studies varies, there is no

recommended way to analyse the data in general. Because variability through space as well as time is of interest in survey studies, the STI model proposed by Legendre et al. [18] appears to be a fundamental method and should be applied at least supplementarily to further statistical methods, which allow one to analyse certain aspects more closely. The STI test was followed by a K-means partitioning, for which the raw data was Hellinger transformed to avoid the double zero problem [28], i.e. concluding a resemblance of two zero objects due to the absence of species in both. Thus, the distances between space–time objects were used rather than the raw abundance data, i.e. objects were grouped according to their similarity to each other and raw data was used again to conduct the analysis of indicator species. Since frequent and numerous species are detected to be significant indicator species, assumptions concerning habitat association or temporal shifts in these characteristics are only valid, if an interpretation of the raw data combined with the results of the statistical analysis are given. Habitat associations of certain species, such as Lucilia ampullacea, L. caesar and C. vomitoria to rural sites and L. sericata to urban sites were obtained to be significant by indicator species analysis. Lucilia species occurred during a shorter period as Calliphora species, restricted mainly to the warm summer months. That reflects higher development thresholds of Lucilia species [3], causing a later appearance of adults within a year as well as an

8

H. Fremdt, J. Amendt / Forensic Science International 236 (2014) 1–9

Fig. 7. Spatiotemporal map of the K-means partition of Sarcophagidae occurrences into eight groups. The x-axis shows weeks of sampling. Note that the sampling was started in September (sampling week 1). A description of locations can be found in Table 1. Symbols for groups are: 1, blue; 2, red; 3, yellow; 4, dark red; 5, green; 6, dark green; 7, light blue; 8 orange.

earlier disappearance, since larvae will diapause rather than continue development, if the environmental conditions become harsh [36]. C. vicina was found to be the least temperature sensitive species without a clear association to a particular habitat and was present in the bulk of sampling occasions. This is in accordance with the findings of Hwang and Turner [15], who concluded a great importance of C. vicina in legal investigations. Sarcophagidae are mentioned to be of great importance for PMI estimations in forensic entomology (e.g. [3]). Nevertheless, identification on species level has often been omitted in surveillance studies (e.g. [8,16]), ignoring the potential benefit for forensic estimations. Since many Sarcophaga species show a high morphological similarity to other taxa, especially of the same subgenus, identification is often difficult and sometimes only possible in males [24]. Recently published reference data of DNAsequences of the barcoding region for many Sarcophagidae tackle this problem [24]. Questionable females can be identified now by molecular methods and may solve uncertainties due to problems in morphological identification. One of the few survey studies taking into account for Sarcophaga spp. Hwang and Turner [15] recorded S. carnaria but not S. variegata. This may be traced back to the fact, that flies were identified by morphological character sets, which prohibits a separation of those species in females (see [21]), but the authors did not discuss this point. Since S. variegata was the second commonest species in the present study, followed by S. carnaria, this is either the first record of S. variegata in a survey

study or just the first inclusion of this species in the analysis. However, S. variegata was recorded in each location just as S. carnaria and S. subvicina. The latter findings are again in contrast with those of Hwang and Turner [15], who obtained an association of S. carnaria and S. subvicina to urban and grassland habitats. A higher synanthropy of Sarcophaga species in higher latitudes as found for the blow fly Lucilia silvarum [37] may cause those diverging habitat associations, but further investigations are required to confirm this hypothesis. In conclusion, the present study is the first to analyse the spatiotemporal abundance pattern of Calliphoridae and Sarcophagidae considering a space–time interaction. Rare species were included in the analysis, showing habitat associations of single species. Temporal associations were obtained mainly for the more frequent species and tested to be significant. For the first time, ecological descriptions of species concerning temporally changing habitat associations were confirmed statistically. The results are of particular use for forensic entomology in the geographical region of Central Europe, because the data obtained allow for the calculation of the likelihood of a particular species composition for a space–time object of interest, strengthening a juridical estimation. Additionally, this built the foundation for subsequent research on the temporal or spatial abundance of important indicator species; using selected conditional factors such as temperature to model the temporal occurrence pattern of an indicator species like C. vicina will be an essential step towards an

H. Fremdt, J. Amendt / Forensic Science International 236 (2014) 1–9

understanding of potential environmental factors influencing the colonisation of a corpse by insects. Acknowledgements We thank the staffs of the Botanical Garden and the Senckenberg research facility Arthur von Weinberg-Haus for providing us the possibility of using their property as sampled sites. Stephanie Jungnickel and Megan Critser Bourguignon are greatly acknowledged for editing the language of the manuscript. Finally, we thank the anonymous reviewers for their helpful and critical advices improving the quality of the manuscript. References [1] P. Nuorteva, Die Rolle der Fliegen in der Epidemiologie der Poliomyelitis, Anzeiger fu¨r Scha¨dlingskunde 36 (1963) 149–155. [2] F. Zumpt, Myiasis in Man and Animals in the Old World: A Textbook for Physicians, Veterinarians and Zoologists, Butterworth, London, 1965. [3] J. Amendt, R. Krettek, R. Zehner, Forensic entomology, Naturwissenschaften 91 (2004) 51–65. [4] J. Amendt, C. Richards, C. Campobasso, R. Zehner, M. Hall, Forensic entomology: applications and limitations, Forensic Sci. Med. Pathol. 7 (2011) 379–392. [5] S. Matuszewski, D. Bajerlein, S. Konwerski, K. Szpila, Insect succession and carrion decomposition in selected forests of Central Europe. Part 1: pattern and rate of decomposition, Forensic Sci. Int. 194 (2010) 85–93. [6] C. Prado e Castro, A. Serrano, P. Martins da Silva, M. Garcı´a, Carrion flies of forensic interest: a study of seasonal community composition and succession in Lisbon, Portugal, Med. Vet. Entomol. 26 (2012) 417–431. [7] H. Reed, A study of dog carcass communities in Tennessee, with special reference to the insects, Am. Midl. Nat. 59 (1958) 213–245. [8] E. Kentner, B. Streit, Temporal distribution and habitat preference of congeneric insect species found at rat carrion, Pedobiologia 34 (1990) 347–359. [9] B. Bourel, V. He´douin, L. Martin-Bouyer, A. Becart, G. Tournel, M. Deveaux, D. Gosset, Effects of morphine in decomposing bodies on the development of Lucilia sericata (Diptera: Calliphoridae), J. Forensic Sci. 44 (1999) 354–358. [10] I. Arnaldos, E. Romera, M. Garcı´a, A. Luna, An initial study on the succession of sarcosaprophagous Diptera (Insecta) on carrion in the southeastern Iberian peninsula, Int. J. Legal Med. 114 (2001) 156–162. [11] M. Johnson, Seasonal and microseral variations in the insect populations on carrion, Am. Midl. Nat. 93 (1975) 79–90. [12] R. Denno, W. Cothran, Niche relationship of a guild of necrophagous flies, Ann. Entomol. Soc. Am. 68 (1975) 741–754. [13] H. Haschemi, Untersuchungen zur Biotopbindung von Lucilia-Arten (Dipt. Calliphoridae), Justus-Liebig-University, 1981 (Ph.D. Thesis). [14] H. Schumann, The occurrence of Diptera in living quarters, Angew. Parasitol. 31 (1990) 131–141. [15] C. Hwang, B. Turner, Spatial and temporal variability of necrophagous Diptera from urban to rural areas, Med. Vet. Entomol. 19 (2005) 379–391.

9

[16] A. Brundage, S. Bros, J. Honda, Seasonal and habitat abundance and distribution of some forensically important blow flies (Diptera: Calliphoridae) in Central California, Forensic Sci. Int. 212 (2011) 115–120. [17] J. MacLeod, J. Donnelly, Some ecological relationships of natural populations of Calliphorine blowflies, J. Anim. Ecol. 26 (1957) 135–170. [18] P. Legendre, M. de Ca´ceres, D. Borcard, Community surveys through space and time: testing the space–time interaction in the absence of replication, Ecology 91 (2010) 262–272. [19] C. Ho¨ller, T. Hoffmeister, P. Christiansen-Weniger, Eine neue Wasserfalle fu¨r Insekten zur Pru¨fung der Attraktivita¨t von Semiochemikalien, Mitteilungen aus der Biologischen Bundesanstalt fu¨r Land- und Forstwirtschaft (1990) 481. [20] K. Rognes, Blowflies (Diptera, Calliphoridae) of Fennoscandia and Denmark, Volume 24 of Fauna Entomologica Scandinavica, Brill, Leiden, 1991. [21] T. Pape, The Sarcophagidae (Diptera) of Fennoscandia and Denmark, Volume 19 of Fauna Entomologica Scandinavica, E., Brill, Leiden, 1987. [22] O. Folmer, M. Black, W. Hoeh, R. Lutz, R. Vrijenhoek, DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates, Mol. Mar. Biol. Biotechnol. 3 (1994) 294–299. [23] H. Fremdt, K. Szpila, J. Huijbregts, A. Lindstro¨m, R. Zehner, J. Amendt, Lucilia silvarum Meigen, 1826 (Diptera: Calliphoridae) – a new species of interest for forensic entomology in Europe, Forensic Sci. Int. 222 (2012) 335–339. [24] K. Jordaens, G. Sonet, R. Richet, E. Dupont, Y. Braet, S. Desmyter, Identification of forensically important Sarcophaga species (Diptera: Sarcophagidae) using the mitochondrial COI gene, Int. J. Legal Med. 127 (2013) 491–504. [25] R Development Core Team, R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, 2012. [26] J. Hausser, K. Strimmer, Entropy: Entropy and Mutual Information Estimation. R Package Version 1.1.7., 2012. [27] J. Oksanen, F. Blanchet, R. Kindt, P. Legendre, P. Minchin, R. O’Hara, G. Simpson, P. Solymos, M. Stevens, H. Wagner, Vegan: Community Ecology Package. R Package Version 2.0-5, 2012. [28] P. Legendre, L. Legendre, Numerical Ecology, 3rd English ed., Elsevier, Amsterdam, 2012. [29] A. Chao, T.-J. Shen, Nonparametric estimation of Shannon’s index of diversity when there are unseen species in sample, Environ. Ecol. Stat. 10 (2003) 429– 443. [30] D. Borcard, F. Gillet, P. Legendre, Numerical Ecology with R, Springer, New York, 2011. [31] S. Dolnicar, K. Grabler, J. Mazanec, A tale of three cities: perceptual charting for analyzing destination images, CAB International (1999) 39–62. [32] M. De Ca´ceres, P. Legendre, Associations between species and groups of sites: indices and statistical inference, Ecology 90 (2009) 3566–3574. [33] M. De Ca´ceres, P. Legendre, M. Moretti, Improving indicator species analysis by combining groups of sites, Oikos 119 (2010) 1674–1684. [34] D. Martı´n-Vega, A. Baz, Sarcosaprophagous Diptera assemblages in natural habitats in central Spain: spatial and seasonal changes in composition, Med. Vet. Entomol. 27 (2013) 64–76. [35] P. Legendre, B. McArdle, Comparison of surfaces, Oceanol. Acta 20 (1997) 27–41. [36] J. Cragg, P. Cole, Diapause in Lucilia sericata (Mg.) Diptera, J. Exp. Biol. 29 (1952) 600–604. ¨ kologische Untersuchungen an Schmeißfliegen (Calliphoridae), [37] H.-A. Steinborn, O Drosera 81 (1981) 17–26.