Marginal Models Via GLS: A Convenient Yet Neglected Tool for the Analysis of Correlated Data in the Behavioural Sciences

Behavioural research often produces data that have a complicated structure. For instance, data can represent repeated observations of the same individual and suffer from heteroscedasticity as well as other technical snags. The regression analysis of such data is often complicated by the fact that the observations (response variables) are mutually correlated. The correlation structure can be quite complex and might or might not be of direct interest to the user. In any case, one needs to take correlations into account (e.g. by means of random‐effect specification) in order to arrive at correct statistical inference (e.g. for construction of the appropriate test or confidence intervals). Over the last decade, such data have been more and more frequently analysed using repeated‐measures ANOVA and mixed‐effects models. Some researchers invoke the heavy machinery of mixed‐effects modelling to obtain the desired population‐level (marginal) inference, which can be achieved by using simpler tools – namely marginal models. This paper highlights marginal modelling (using generalized least squares [GLS] regression) as an alternative method. In various concrete situations, such marginal models can be based on fewer assumptions and directly generate estimates (population‐level parameters) which are of immediate interest to the behavioural researcher (such as population mean). Sometimes, they might be not only easier to interpret but also easier to specify than their competitors (e.g. mixed‐effects models). Using five examples from behavioural research, we demonstrate the use, advantages, limits and pitfalls of marginal and mixed‐effects models implemented within the functions of the ‘nlme’ package in R.     

Plasma Midregional Pro-Adrenomedullin Improves Prediction of Functional Outcome in Ischemic Stroke

Background

To evaluate if plasma levels of midregional pro-adrenomedullin (MR-proADM) improve prediction of functional outcome in ischemic stroke.

Methods

In 168 consecutive ischemic stroke patients, plasma levels of MR-proADM were measured within 24 hours from symptom onset. Functional outcome was assessed by the modified Rankin Scale (mRS) at 90 days following stroke. Logistic regression, receiver operating characteristics (ROC) curve analysis, net reclassification improvement (NRI), and Kaplan-Meier survival analysis were applied.

Results

Plasma MR-proADM levels were found significantly higher in patients with unfavourable (mRS 3–6) compared to favourable (mRS 0–2) outcomes. MR-proADM levels were entered into a predictive model including the patients'' age, National Institutes of Health Stroke Scale (NIHSS), and the use of recanalization therapy. The area under the ROC curve did not increase significantly. However, category-free NRI of 0.577 (p<0.001) indicated a significant improvement in reclassification of patients. Furthermore, MR-proADM levels significantly improved reclassification of patients in the prediction of outcome by the Stroke Prognostication using Age and NIHSS-100 (SPAN-100; NRI = 0.175; p = 0.04). Kaplan-Meier survival analysis showed a rising risk of death with increasing MR-proADM quintiles.

Conclusions

Plasma MR-proADM levels improve prediction of functional outcome in ischemic stroke when added to the patients'' age, NIHSS on admission, and the use of recanalization therapy. Levels of MR-proADM in peripheral blood improve reclassification of patients when the SPAN-100 is used to predict the patients'' functional outcome.     

Pre-adaptive shift of a native predator (Araneae, Zodariidae) to an abundant invasive ant species (Hymenoptera, Formicidae)

Invasive species often displace native species and can affect ecological processes in invaded habitats. If invasive species become abundant, changes in prey availability may be particularly harmful to specialist predators. The Argentine ant, Linepithema humile Mayr, is an important invasive species on nearly all continents. Spiders of the genus Zodarion are specialised ant-eating predators native to the Mediterranean yet it is unknown if they can exploit invasive ant species. Here we studied spatial and temporal abundance of this invasive ant and the native spider, Zodarion cesari Pekár, during 4?years in four citrus groves. Circadian activity of both spiders and ants, and capture efficiency and prey specificity of the predator were also evaluated. The abundance of Z. cesari was strongly correlated to L. humile abundance. The predatory activity of spiders varied seasonally with differences on the relative frequency of spiders capturing ants depending on the time of the year. In laboratory, Z. cesari displayed most efficient capture upon the native ant Tapinoma nigerrimum (Nylander) and the invasive ant L. humile in comparison with five other native ant species. These results demonstrate that the native spider Z. cesari is successfully exploiting the invasive ant species L. humile and is likely a locally monophagous predator. We suggest that Z. cesari shifted away from native T. nigerrimum post invasion as both ant species are phylogenetically related and of similar size.     

Ser-His catalyses the formation of peptides and PNAs

The dipeptide seryl-histidine (Ser-His) catalyses the condensation of esters of amino acids, peptide fragments, and peptide nucleic acid (PNA) building blocks, bringing to the formation of peptide bonds. Di-, tri- or tetra-peptides can be formed with yields that vary from 0.5% to 60% depending on the nature of the substrate and on the conditions. Other simpler peptides as Gly-Gly, or Gly-Gly-Gly are also effective, although less efficiently. We discuss the results from the viewpoint of primitive chemistry and the origin of long macromolecules by stepwise fragment condensations.     

Different hunting strategies of generalist predators result in functional differences

The morphological, physiological, and behavioural traits of organisms are often used as surrogates for actual ecological functions. However, differences in these traits do not necessarily lead to functional differences and/or can be context–dependent. Therefore, it is necessary to explicitly test whether the surrogates have general ecological relevance. To investigate the relationship between the hunting strategies of predators (i.e., how, where, and when they hunt) and their function, we used euryphagous spiders as a model group. We used published data on the diet composition of 76 spider species based on natural prey and laboratory prey acceptance experiments. We computed differences in the position and width of trophic niches among pairs of sympatrically occurring species. Pairs were made at different classification levels, ranked according to the dissimilarity in their hunting strategies: congeners, confamiliars (as phylogenetic proxies for similarity in hunting strategy), species from the same main class of hunting strategy, from the same supra-class, and from different supra-classes. As for niche position computed from the natural prey analyses, species from the same class differed less than species from different classes. A similar pattern was obtained from the laboratory studies, but the congeners differed less than the species from the same classes. Niche widths were most similar among congeners and dissimilar among species from different supra-classes. Functional differences among euryphagous spiders increased continuously with increasing difference in their hunting strategy. The relative frequency of hunting strategies within spider assemblages can, therefore, influence the food webs through hunting strategy-specific predator–prey interactions.     

Lysozyme Effect on Oleic Acid/Oleate Vesicles

We have analysed by means of turbidimetric, dynamic light scattering (DLS), and fluorimetric techniques the effect of lysozyme on negatively charged oleic acid/oleate vesicles. The addition of lysozyme brings about a decrease in optical density of the vesicle population, which finally results in a size distribution of oleate vesicles shifted toward smaller mean diameters. On the contrary, (a) when phosphatidylserine vesicles were used, lysozyme induces an increase of turbidity and a shift toward larger vesicle sizes; and (b) the addition of histone H1 or poly-L-lysine produces an aggregative behavior both in oleate and in phosphatidylserine vesicles. Experiments carried out with calcein-containing vesicles indicate that the observed changes in the lysozyme/oleate system occur with partial leakage of the vesicle content. All this is taken to suggest that the interaction between lysozyme and oleate vesicles is of quite specific nature, and certainly not just due to electrostatic interactions.     

Life history of the spider parasitoid Zatypota percontatoria (Hymenoptera: Ichneumonidae)

The life history of polysphinctine parasitoids is poorly understood. As a result, their effect on their host has not been evaluated. Here, we present the phenology of Zatypota percontatoria, an ectoparasitoid wasp of theridiid spiders, using rich field, semi‐field and laboratory investigations of selected life‐history parameters. The wasps overwintered as larva attached to the spider abdomen and the imagos of the first generation emerged at the beginning of May. The sex ratio (F/M) of overwintered wasp larvae was 1.4. The wasp spent an average of 6.4 days in the egg stage, 27.67 days in the larval stage and 9.6 days in the pupal stage. Adult longevity was on average 14.34 days. Oviposition occurred on average 8.44 days following maturity. The wasp laid on average 0.35 eggs/day. The total fecundity was on average 7.4 eggs per wasp and decreased with age (data from unmated females). The developmental rate increased with experimental temperature up to 28 °C. The lower developmental threshold for pupae was estimated at 6.6 °C. The sum of effective temperatures for pupae was estimated at 157.8 degree days. Using daily temperature records from the Czech Republic over the last 41 years, we estimated an average of 3.5 generations of Z. percontatoria per year. Mortality was 48% in semi‐field conditions and 23% in the laboratory. The incidence of mortality was highest during egg and larval stages caused by the detachment of eggs or larvae during spider molting, the mortality of the spider‐host and egg deposition on the spider prosoma.