Wednesday, August 21, 2019

Trophic ecology

Understanding the trophic structure of a community is indispensable to understanding the underlying ecosystem in its entirety. Unfortunately, basic information on animal diet is often incomplete or simply missing. This is not a surprise as the assessment of animal diet is rather difficult. Very often direct observation of feeding is not possible. Species are elusive, small, rare or live in regions that are inaccessible. Another issue is variability of diets within species, populations or even between life stages. And to add more complexity to that, diet is dynamic, i.e. it fluctuates with seasons, day time, geography, and ecosystems. 

As usual, our knowledge about diet is far more detailed when it comes to vertebrates, specifically larger predators. However, once the focus shifts toward e.g. the trophic ecology of many arthropod species one quickly finds out that not much is covered by the literature, because it has not yet been studied as a consequence of the difficulties described above. 

For quite some years stable isotope analysis has been used to reconstruct diets, to characterize trophic relationships, to elucidate patterns of resource allocation, and for construction of food webs. Isotopes of a given element differ in the number of neutrons they contain. Everybody might have heard about radioactive isotopes which are those that spontaneously decay into other element isotopes. Stable isotopes are those which do not, but they might be the result of such a decay. Nonradiogenic stable isotopes are most useful for ecological studies, and particularly those of the light elements: H, C, N, O and S, as those are the major constituents of organic materials, and for which there are large relative mass differences between isotopes of the same element. Stable isotope ratios are usually measured using stable isotope mass spectrometry and currently the most commonly used isotopes are δ¹³C and δ¹⁵N. 

Similarly, DNA barcodes have much utility in revealing trophic interactions among organisms and determining diet items although they often can only provide a snapshot of mostly local, present-day interactions. Thus, stable isotope analysis can complement DNA barcoding techniques by adding coarser but more integrated averages of food web structure across space and time. In addition it allows for tracing nutrient fluxes in changing organismal communities. In other words, DNA barcodes allow us to pinpoint the exact composition of a species’ diet while both δ¹³C and δ¹⁵N isotopes can be used to trace the structure (and its change over time) of the underlying food webs, with δ¹⁵N showing enrichment between trophic levels and δ¹³C distinguishing between prey groups. 

A recent study on a particular butterfly group shows how stable isotope analysis and DNA metabarcoding can be combined to better assess diet and trophic position. The family Lycaenidae (Gossamer-winged butterflies) is the second largest of all butterflies (about 6000 species) with various life histories and diets. About 75% of the species in this group are known to associate with ants and these associations can be mutualistic, parasitic, or predatory. In some cases caterpillars produce nutritious secretions from specialized organs to reward ants in exchange for their protection against predators and parasitoids.

However, the larval diets of many lycaenid butterflies remain poorly characterized, likely because the caterpillars are camouflaged, difficult to rear in captivity and often live in inaccessible microhabitats such as flower buds or subterranean ant nests. For many species, no records exist for direct observations of feeding behaviors, leaving researchers to rely on indirect evidence to infer larval diets. One such species is Anthene usamba (previously A. hodsoni), a lycaenid butterfly found in the savannas of eastern Africa.

An earlier study using comparative microbiome analysis showed that adults in contrast to caterpillars were aphytophagous. Now, metabarcoding of caterpillar gut content chloroplast 16S rRNA found a match to the acacia tree (Vachellia drepanolobium) the caterpillars are known to live on. Stable isotope analysis confirmed that Anthene usamba gets most of its carbon directly or indirectly from its host plant, rather than from surrounding grasses. Although the feeding behaviour of the larvae remains unknown it can be clearly stated that they feed on their host tree.

There is a lot that speaks for the complementary use of stable isotope analysis and DNA barcoding in some dietary studies. There are individual limitations to both methods, e.g. analysis of diets with stable isotope analysis requires prior knowledge of the isotopic signatures of potential food sources. They also are not useful when the number of potential food sources exceeds the number of isotopes available. DNA-based methods such as metabarcoding are subject to various biases especially when PCR is involved. They highly depend on well-parameterized reference libraries and it is not possible to quantify individual contributions to complex DNA mixtures especially in a sample with different levels of degradation (here through differential digestion).

As there are so many species for which we have no clue on what they are feeding on it is reassuring to know that there are a number of methods available that can help us to learn more.

Friday, August 16, 2019


"That’s God’s signature. God’s signature is never a forgery."
Eddie Joe Lloyd, quoted in the New York Times (26 August 2002). Lloyd had been sent to prison 17 years before and was released in 2002 when DNA testing showed a mismatch between his DNA profile and the profile developed from evidence at the crime scene.

The first DNA profiling techniques were developed in the mid-1980s. The technology was named DNA fingerprinting in reference to the well-established forensic method that was rarely questioned in the courts. Over the years methodologies changed and were refined. By the mid-1990s criminal justice systems started using short tandem repeat (STR) technology, which is based on hypervariable DNA sequences of relatively short length. The STR system became the new gold standard in DNA profiling. It is widely used to compare criminal suspects' profiles stored in massive databases (e.g. the FBI's Combined DNA Index System - CODIS) with DNA trace evidence to assess the likelihood of their involvement in a crime.

As technology has progressed, forensic scientists are exploring other applications of DNA-based methods in crime scene investigation. Especially DNA barcoding and metabarcoding are about to revolutionize other aspects of forensic science.

Forensic entomology

The first recorded incident where insects were used in a criminal investigation was in 13th-century China as described in Sung Tzu's book called The washing away of wrongs. When a farmer was found murdered in a field with a sharp weapon, all the suspects were told to place their sickles on the ground. Only one sickle attracted blow flies to the trace amount of blood hidden to the naked eye which resulted in the confession by the murderer.

The study of insects in criminal investigation is called forensic entomology. It is mostly known for its use to estimate the time of death. A plethora of insect species are attracted to a decomposing body and may lay eggs in it. The succession waves in which different insect species colonize a corps depend on its state of decomposition which in turn means that the identification of an insect obtained on a decomposing body will allow for a relative precise estimate of the time of death. Insects colonize a corps in successive waves, and each has its own unique life cycle. Many kinds of insects will flock to a decomposing body, but the most common found on a corpse are flies and beetles. Flies, particularly blow flies, can find dead flesh within minutes. Their maggots do the majority of the eating and are responsible for much of a corpse's decay. Beetles, on the other hand, will typically move in once a corpse has dried out. But there is more. Forensic scientists can sometimes determine if a corpse has been moved simply by studying the insect population and their larval stages.

All these methods require accurate species identification which is not as easy at it would seem. Forensic entomologists are often confronted with large amounts of partial insect remains or impossible to identify early life stages, such as eggs and larvae. In the latter case a scientist would try to incubate and raise insects until distinguishable features become apparent. Aside from the fact that this is not always possible criminal investigations are loosing valuable time during the wait for an insect to mature. 

A recent study shows very clearly that with a proper reference library and the right technology DNA barcoding enables forensic entomologists to rapidly analyze hundreds of bulk samples obtained from corpses and provide precise and reliable species assignments. In collaboration with the Bavarian State Criminal Police Office, colleagues at the Bavarian State Collection of Zoology used an insect reference library they build over years (together with our institute) to identify the content of 30 metabarcoded bulk samples obtained from the morgue. Given today's cost for high throughput sequencing and the fact that all samples can be done in a single run I am estimating costs of about US$80 per bulk sample in this particular case. All of a sudden regular DNA-based identification of massive amounts of insect remains becomes feasible and applicable in criminal investigations. The only caveat in many parts of the world is the incompleteness of local DNA barcode reference libraries. However, it seems rather feasible to build dedicated datasets based on expert identified material. That being said, countries with very active insect barcoding programs such as Canada, Finland, Germany, Norway and other members of the International Barcoding of Life Project (iBOL) could explore opportunities already today.

Forensic soil analysis

Soil is commonly encountered as trace evidence in criminal cases, i.e. mud sticking to footwear, tires and shovels, soil splash marks on vehicles, and traces left on clothes, the floor or in the trunk of vehicles. Those soil samples can be compared to samples from known locations, where an offence is thought to have occurred, thereby establishing a link between a suspect or a victim and a crime scene. In an investigative process, where for example the crime scene is unknown, soil trace evidence can also give valuable information on geographic origin or provenance and help narrow the search for a location. 

Forensic scientists often use inorganic soil properties such as colour, consistency, structure, texture, segregations/coarse fragments (charcoal, ironstone or carbonates), and abundance of roots/pores to aid the identification of soil materials. They are following strict conventions and sophisticated systematic procedures (see figure on the left from Fitzpatrick 2013).

However, soil contains a lot of DNA from the living organisms that populate it but also in form of environmental DNA (eDNA). Consequently, forensic scientists have begun considering biological material for the characterization of soil types. Pollen grains and grass spores are preserved in soil samples over a longer period of time and can be used to identify surrounding flora. Newer  forensic studies also show that a DNA profile of the soil bacterial community DNA in small samples of soil recovered from crime scenes can be matched with representative profiles of a suspect. A new study now goes further and explores whether soil eDNA could be used to predict a sample’s origin along environmental gradients (light, soil moisture, pH and nutrient status), origin in terms of habitat types (e.g., forest, heathland and rotational field), and in terms of geographic origin. A Danish research group utilized data collected in a nation-wide survey of biodiversity in Denmark to establish Ellenberg Indicator Values (EIV). EIVs are based on an ordinal classification of plants according to the position of their ecological niche along an environmental gradient. They were initially applied to the flora of Germany as a model of bioindication. Each plant species is assigned an EIV and the community or site EIV is calculated as an average of all the indicator values. The only issue is that EIVs are only available for Central Europe and the UK. Work elsewhere would require the use of species scores from ordination of large and representative vegetation datasets. The study investigated the potential for constructing predictive models of environmental properties, habitat types and geographic origin based on soil eDNA. The colleagues found that variation in soil eDNA can predict environmental conditions and most habitat types but not geographic provenance. Model predictions for two mock crime scenes corresponded well with the actual EIVs at the site. This shows that an eDNA approach can become be a useful investigative tool in crime scene cases, however, at this point in time only for those without the need for the strict and validated procedures necessary to be used as evidence in court. There is still a lot of work left to be done.

DNA-based forensics science has come a long way from the first profiles developed in the 1980's  and it has matured into a powerful tool for both catching criminals and exonerating innocent people. Nevertheless, it has not reached its full potential as new technologies provide applications in trace evidence analysis.

Countless CSI TV series might make us believe that solving a crime is only a matter of having the best high-tech at your fingertips and that catching a culprit is only a DNA sequence away. Nothing could be further from the truth. One risk lies in the complexity of the statistical methods used to analyze DNA samples and their interpretation. Attorneys and judges, who must understand how they work to reliably assess their validity in court are facing a challenge. Methods for interpreting DNA evidence are inconsistent, potentially leading to biased verdicts on the identities of DNA donors. Therefore, it is paramount to ensure that the science underlying the analysis used to make decisions in court remains transparent and validated by the broader scientific community.