BookLiterature Review

# Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach

Authors:
... De acuerdo con Burnham y Anderson (2002), los modelos con un Δ i <2 tienen un soporte sustancial y deben ser incluidos para futuras consideraciones. Utilizando el enfoque de la inferencia multimodelo se calcularon el modelo promedio y los límites inferior y superior del intervalo de confianza promedio: Ec. 7 ...
... El de Richards es el que mejor se ajusta a los datos de hembras (-logL = 2.43) y machos (-logL = 2.1); sin embargo, utilizando el criterio de información de aic c , el modelo de Brouwer y Griffiths se acepta como el mejor para las hembras y el de Gompertz para los machos, debido a la penalización del parámetro extra del de Richards. Para ambos sexos, la diferencia de aic c del "modelo ganador" y el resto de los modelos presentan una Δ i ≤ 2. De acuerdo con el criterio de inferencia multimodelo (Burnham y Anderson 2002), no existe uno "claramente ganador", por lo que es necesario incluir los cinco para estimar el global. ...
... Resultados del ajuste y la selección de los cinco modelos utilizados en el estimado de la l 50 para cada sexo del ostión de roca Striostrea prismatica (Crassostrea iridescens) el valor estimado de l 50 es menor (Vélez-Barajas 1996). Se ha observado en otros recursos que las discrepancias entre parámetros estimados se pueden deber a múltiples factores: diferencias fenotípicas de los stocks a lo largo de un gradiente latitudinal o temporal (Richards et al. 1990, Zhu et al. 2011), a la calidad y la estructura de los datos (Trippel y Harvey 1991, Chen y Paloheimo 1994, Punt et al. 2006) y la elección del modelo (Burnham y Anderson 2002, Zhu et al. 2011. En este sentido, es importante uniformar criterios para minimizar las diferencias por errores de proceso y evaluar la eficiencia y la parsimonia de diferentes modelos a partir del criterio de información dados los datos. ...
Article
El ostión de roca Striostrea prismatica es una especie de valor comercial presente en la zona tropical de la costa del Pacífico mexicano. La Carta Nacional Pesquera reconoce esta especie como Crassostrea iridiscens (Hanley 1854); sin embargo, éste es un sinónimo de S. prismatica (Huber 2012). La talla media de madurez sexual (l 50) se estimó a partir de seis modelos sigmoides para 83 hembras y 121 machos, los organismos fueron muestreados en el litoral de Nayarit durante el periodo 2008-2010. La selección del mejor mo-delo se realizó con base en el criterio de información de Akaike corregido (aic c). El mejor modelo para las hembras fue el de Brouwer y Griffiths (l 50 = 9.1 cm de longitud total, lt), y para los machos el de Gompertz fue el más aceptado (l 50 = 9 cm lt). En ambos sexos, la diferencia entre aic c fue menor a dos. Acorde con la inferencia multimodelo, los cinco modelos se utilizaron para estimar el global por sexo. En hembras, el valor de fue 9.0 cm y el de 8.8 a 9.1 cm y en machos fue 8.9 cm, de 8.7 a 9.0 cm. Los valores de ambos sexos coinciden con aquellos registrados en la Carta Nacional Pesquera. En el presente trabajo no hay un claro "modelo ganador", por lo que la inferencia multimodelo es útil en el cálculo de un modelo global. Palabras clave: Modelos sigmoideos, selección de modelos, criterio de información, modelo global. Multi-model inference: Estimation of mean size at sexual maturity for the tropical rocky oyster Striostrea prismatica from Nayarit, Mexico The rocky oyster Striostrea prismatica is an important commercial species found off shores in the Mexican Pacific. The National Fishery Chart recognizes this species as Crassostrea iridiscens (Hanley 1854); although it isa synonym of S. prismatica (Huber 2012). The mean size at sexual maturity (l 50) was estimated using six sigmoid models for 83 females and 121 males, those organisms were sampled along the coast of Nayarit from 2008 to 2010. The best model was selected using the bias-corrected Akaike information criterion (aiC C). The best model for females was Brouwer and Griffiths (l 50 = 9.1 cm total length, Lt), and the Gompertz model for males (l 50 = 9 cm Lt). For both sexes differences among aiC C was less than two. According to the multimodel inference approach, the model-average for each sex was estimated, using all five models. For females, was 9.0 cm and varied from 8.8 to 9.1 cm, while for males was 8.9 cm, was 8.7 to 9.0 cm. For both sexes values of agree with the value reported in the National Fishery Chart. In the present work there is no clear "winner model", but nonetheless the multimodel inference approach is a useful method for estimating the mean size at sexual maturity.
... La inferencia multimodelo postula que cuando los datos soportan evidencia para más de un modelo, en lugar de estimar los parámetros a partir de sólo "el mejor modelo", es posible estimarlos a partir de varios modelos e incluso todos los modelos considerados. Obtener el valor promedio de la variable predicha, a partir de varios modelos, es una ventaja para alcanzar una inferencia robusta que no esté condicionada a un sólo modelo y que integra la aportación ponderada de cada modelo ajustado (Burnham y Anderson 2002). El objetivo de la presente investigación es utilizar el enfoque de teoría de información y estimar la talla promedio de madurez sexual (l 50 ) en hembras de O. libertate con la técnica de la inferencia multimodelo. ...
... Los modelos con un D i < 2 son estadísticamente viables para estimar la variable de respuesta dado el conjunto de datos (Burnham y Anderson 2002). ...
... Utilizando el criterio de decisión de aiC, los modelos de Lysack, White y Brouwer y Griffiths se aceptan como los mejores (47.62); sin embargo, el modelo de Gompertz presenta una Δ i ≤ 1.24. De acuerdo con el criterio de inferencia multimodelo (Burnham y Anderson 2002), los cuatro modelos deben de ser considerados para el cálculo del global . La inferencia multimodelo es una herramienta muy poderosa en la toma de decisiones, pues permite confrontar diferentes hipótesis (modelos) e incluir las más plausibles en un modelo global incorporando la información de todos los modelos. ...
Article
... We, therefore, have to revert to the subject of selection an optimal total number of knots to avoid possible data under-/overfitting without reducing the B-spline model generality. This problem can be addressed by using some model-selection techniques, like well-established MSE-based information criteria (ICs) [27,28]. These techniques provide an objective way to compare performance of the B-spline models with various numbers of knots to select a model with optimal parametric complexity (in the sense of minimizing the MSE with a minimal number of knots distributed in an appropriate manner). ...
... The section also describes the essentials of the AIC and BIC which are used as the theoretical basis of the knot number optimization applied in this paper. For the sake of brevity, we will omit any deep mathematical and philosophical backgrounds behind these criteria and, instead, we direct the interested reader to several excellent textbooks for a general overview [27,28,35,36]. ...
...     maxl n ( | ) / 2 ln RSS / data n n   L [27,28]. Therefore, recalling that in regression analysis the MSE is defined as the RSS divided by the number of degrees of freedom, Eq. (7) can be expressed in terms of MSE as ...
Preprint
Full-text available
Currently, a Kramers-Kronig consistent B(asis)-spline representation of the dielectric function is an efficient and widely used method for accurate modeling of the material optical functions in ellipsometric data analysis. However, the B-spline approach to the dielectric function modeling should include an appropriate and user-independent way of a knot vector generation, i.e., the proper selection of the number and locations of knots. In this paper, we advocate for a systematic approach which combines a specific knot allocation method, based on so-called "Integral Span", a slope-weighting factor, with a selection of optimal number of knots using the Akaike and Bayesian Information Criteria, two statistical estimators, thereby replacing an intuitive and time-consuming "trial-and-error" strategy. The proposed hybrid approach is used to optimize the B-spline models for an epitaxial cobalt disilicide (CoSi 2) thin film and a crystalline silicon substrate (c-Si).
... Each comment was processed with Python 3.8 script 9 . Firstly, all of the hyperlinks were removed from the comments and the comments were part-of-speech tagged with the use of the NLTK POS tagger. ...
... 8 The reason behind this is that the YouTube search algorithm is not straightforward and may decide that a video is relevant even if there is no keyword in the title. 9 The following packages were used: string, NLTK, AFINN, collections, NumPy, and pandas. and non-alphabetic words were removed. ...
... Firstly, I tested the relationship between humanlikeness of robots and general sentiment scores and examined how the relationship changes for particular ABOT subscales of humanlikeness (H1). Therefore, in order to characterize the shape of the relationship and test which model better describes the relationship (linear, quadratic or cubic), I performed polynomial curve fitting and tested the goodness of fit of each regression model using the Akaike Information Criterion (AIC; see [9]). AIC allows us to compare models of varying complexity, penalizing a higher number of parameters. ...
Article
Full-text available
The uncanny valley hypothesis (UVH) suggests that almost, but not fully, humanlike artificial characters elicit a feeling of eeriness or discomfort in observers. This study used Natural Language Processing of YouTube comments to provide ecologically-valid, non-laboratory results about people’s emotional reactions toward robots. It contains analyses of 224,544 comments from 1515 videos showing robots from a wide humanlikeness spectrum. The humanlikeness scores were acquired from the Anthropomorphic roBOT database. The analysis showed that people use words related to eeriness to describe very humanlike robots. Humanlikeness was linearly related to both general sentiment and perceptions of eeriness—-more humanlike robots elicit more negative emotions. One of the subscales of humanlikeness, Facial Features, showed a UVH-like relationship with both sentiment and eeriness. The exploratory analysis demonstrated that the most suitable words for measuring the self-reported uncanny valley effect are: ‘scary’ and ‘creepy’. In contrast to theoretical expectations, the results showed that humanlikeness was not related to either pleasantness or attractiveness. Finally, it was also found that the size of robots influences sentiment toward the robots. According to the analysis, the reason behind this is the perception of smaller robots as more playable (as toys), although the prediction that bigger robots would be perceived as more threatening was not supported.
... The model selection was conducted following an information-theoretic approach (Burnham and Anderson, 2002), using the R package "MuMIn" (Bartoń, 2019). For each analysis, we kept models with delta corrected Akaike Information Criteria ( AICc) <2 (e.g., Rayner et al., 2007) to calculate the standardized fully averaged coefficient at 95% confidence interval. ...
... For each analysis, we kept models with delta corrected Akaike Information Criteria ( AICc) <2 (e.g., Rayner et al., 2007) to calculate the standardized fully averaged coefficient at 95% confidence interval. The relative importance of predictors in each model was calculated by the sum of the Akaike weights (Burnham and Anderson, 2002). The predictor with coefficient furthest away from 0 is the most robust explanatory variable, where 95% confidence intervals determine the confidence in the direction of the variable's effect (Grueber et al., 2011). ...
Article
Full-text available
To avoid risks, organisms must recognize threatening heterospecies from non-threatening ones via acoustic cues from a distance. With land-use change, humans have encroached considerably into natural areas. Therefore, it is beneficial to animals to use acoustic cues to discriminate between different levels of threats posed by humans. Our study aims at testing this discriminatory ability in Asian elephants ( Elephas maximus ), animals that have been for long history subjected to human interaction. We tested whether eighteen semi-captive elephants could discriminate between voices of their own mahouts (i.e., who take care of the elephants exclusively) and of other mahouts (unfamiliar individuals). The results showed that elephants responded successfully to the commands from their own mahouts, with an average response rate as high as 78.8%. The more years the mahouts had been as their caretakers, the more the elephant showed active responses toward the commands. Female elephants responded to the commands more frequently and faster than males. Also younger elephants responded more frequently and faster than older elephants. We argue that Asian elephants can discriminate between familiar and unfamiliar humans by acoustic cues alone. Proximity with humans may be a factor, as fundamental as domestication, for animals to develop heterospecies discriminatory ability.
... Additive and interactive effects were evaluated; however, due to limited degrees of freedom, only models with up to three factors were considered. Akaike Information Criterion corrected for small sample size (AICc) was used to identify the best model to parsimoniously explain the variation in concentrations of contaminants (Akaike, 1974;Burnham & Anderson, 2002). We selected the model with the lowest AICc score as the most parsimonious model, unless a simpler model existed with a ΔAICc score less than or equal to 2 compared to the more complex model (Burnham & Anderson, 2002). ...
... Akaike Information Criterion corrected for small sample size (AICc) was used to identify the best model to parsimoniously explain the variation in concentrations of contaminants (Akaike, 1974;Burnham & Anderson, 2002). We selected the model with the lowest AICc score as the most parsimonious model, unless a simpler model existed with a ΔAICc score less than or equal to 2 compared to the more complex model (Burnham & Anderson, 2002). For the best fit model for each contaminant class, multiple comparisons between estimated means were conducted using the Sidak adjustment (emmeans package; Lenth, 2020). ...
Article
Full-text available
River otters ( Lontra canadensis ) are apex predators that bioaccumulate contaminants via their diet, potentially serving as biomonitors of watershed health. They reside throughout the Green-Duwamish River, WA (USA), a watershed encompassing an extreme urbanization gradient, including a US Superfund site slated for a 17-year remediation. The objectives of this study were to document baseline contaminant levels in river otters, assess otters’ utility as top trophic-level biomonitors of contaminant exposure, and evaluate the potential for health impacts on this species. We measured a suite of contaminants of concern, lipid content, nitrogen stable isotopes (δ ¹⁵ N), and microsatellite DNA markers in 69 otter scat samples collected from twelve sites. Landcover characteristics were used to group sampling sites into industrial (Superfund site), suburban, and rural development zones. Concentrations of polychlorinated biphenyls (PCBs), polybrominated diphenyl ether flame-retardants (PBDEs), dichlorodiphenyl-trichloroethane and its metabolites (DDTs), and polycyclic aromatic hydrocarbons (PAHs) increased significantly with increasing urbanization, and were best predicted by models that included development zone, suggesting that river otters are effective biomonitors, as defined in this study. Diet also played an important role, with lipid content, δ ¹⁵ N or both included in all best models. We recommend river otter scat be included in evaluating restoration efforts in this Superfund site, and as a potentially useful monitoring tool wherever otters are found. We also report ΣPCB and ΣPAH exposures among the highest published for wild river otters, with almost 70% of samples in the Superfund site exceeding established levels of concern.
... The 'Akaike weight' is considered as the weight of evidence in favor of model i being the actual best model of the available set of models (Burnham and Anderson, 2002;Burnham et al., 2011). We also estimated the relative importance (r i ) of predictor variables x j in the set of models with the importance function of the MuMin package (Package 'MuMIn'), which sums the Akaike weights across all the models in the set where variable j occurs. ...
... Thus, the relative importance of variable j is reflected in the sum w + (j). The larger the w + (j) the more important variable j is, relative to the other variables (Burnham and Anderson, 2002). Predictors with ri > 0.6 were considered strongly related to response variables. ...
Article
The decline of pollinators is a widespread problem in today's agriculture, affecting the yield of many crops. Improved pollination management is therefore essential, and honey bee colonies are often used to improve pollination levels. In this work, we applied a spatially explicit agent-based model for the simulation of crop pollination by honey bees under different management scenarios and landscape configurations. The model includes 1) a representation of honey bee social dynamics; 2) an explicit representation of resource dynamics; 3) a probabilistic approach to the foraging site search process; and 4) a mechanism of competition for limited resources. We selected 60 sample units from the rural landscape of the Chilean region with the largest apple-growing area and evaluated the effectiveness of different pollination strategies in terms of number of visits and number of pollinated flowers per hectare of apple crops. Finally, we analyzed how the effects of these practices depended on the structure of adjacent landscapes. Higher colony density per hectare in the focal crop increased the number of honey bee visits to apple inflorescences; however, the effects were nonlinear for rates of pollinated flowers, suggesting that there is an optimum beyond which a greater number of honey bees does not signify increased levels of crop pollination. Furthermore, high relative proportions of mass flowering crops and natural habitats in the landscape led to a decrease in honey bee densities in apple fields in landscapes with high relative cover of apple orchards (dilution effect). Our results indicate that for optimal crop pollination, strategies for management of pollinator species should consider the modulating effects of the surrounding landscape on pollination effectiveness. This model could thus be a useful tool to help farmers, beekeepers, and policy-makers plan the provision of pollination services, while also promoting the biodiversity and sustainability of agroecosystems.
... Based on the AICc values, the "best" model settings were selected by the function model.avg, whereas models with ΔAIC≥7 were considered to result in less explanatory power and were excluded from observations [8,46]. ...
... We used multi-model inference with the information criterion, corrected for small sample size (AICc) and predictor relative importance to rank our subset models in the package MuMIn (model.avg function) [46,50]. ...
Article
Full-text available
The European roller (Coracias garrulus) is an endangered species whose breeding in Serbia depends almost entirely on nest boxes. The aim of the present study was to assess the influence of prey availability and foraging habitat characteristics on nest-box occupancy and breeding parameters. Data from 20 roller foraging sites over 5 breeding seasons were used in a set of linear regression models to evaluate which factors affect the diversity and biomass of roller prey, as well as nest-box occupancy and breeding parameters. Our analyses revealed that prey availability parameters were significantly affected by the grazing regime and biophysical parameters. An area under grassland negatively affected nest-box occupancy, clutch size and fledging success. In contrast, grazing intensity showed positive effects. Although grazing negatively affected prey diversity and quantity, it potentially increased the likelihood of a successful hunt by forming short vegetation. These results indicate that the habitat characteristics linked to the ability of the species to hunt successfully should be considered when installing nest boxes as a part of the broader management of roller breeding sites. Furthermore, nest-box installation in open agricultural habitats other than grasslands should be considered in the conservation strategy for the species in Serbia.
... For each individual model analysis, we performed the step-wise model building approach to construct a set of cumulative link models (CLMs; [24]), first starting with a null model (no covariates) and global model (models including all covariates), and then building subsequent models by excluding covariates with the least support based on p-values (higher p-value, lower support). Models were ranked using AICc [25], and model averaging of the top models (∆AICc < 2) was performed. Model convergence was evaluated through parameter assessment accuracy to the global model of every model set [25]. ...
... Models were ranked using AICc [25], and model averaging of the top models (∆AICc < 2) was performed. Model convergence was evaluated through parameter assessment accuracy to the global model of every model set [25]. All global models showed a reported error log-likelihood value below 10 −10 , which indicated accurate parameter estimates. ...
Article
Full-text available
Snakes are commonly associated with feelings of anxiety or disgust, and snake conservation is often hindered by negative attitudes and perceptions. Although global snake populations are generally in decline, snakebite envenomation (SBE) continues to be recognized as a serious public health issue, particularly in rural areas of tropical and subtropical countries. Data on SBE, a neglected tropical disease, are lacking, and Indonesia, a hotspot of venomous snake diversity, has no snake bite reporting system. We analyzed 127 survey results in Banten, Indonesia with the aim of documenting SBE and exploring the relationships between respondents’ experiences, attitudes, and behaviors toward snakes. Nine percent of respondents had experienced SBE, and knowledge of SBE incidents was associated with negative attitudes toward snake populations, with negative attitudes toward snakes associated with a higher likelihood of anti-conservation behavior. Women were more fearful than men, and women and those with lower education levels were more negative toward pythons (Malayopython reticulatus), suggesting that increased knowledge may aid in snake conservation efforts. Universally negative risk beliefs and attitudes toward venomous snakes indicate a need to reduce the threat of SBE in our study area.
... A model averaging approach was used to identify significant variables (Burnham & Anderson 2002) describing the variation of SVL. For this purpose, a set of candidate models with all possible combinations of variables was generated based on the full model and the Akaike Information Criterion with a correction for small sample sizes (AICc) was calculated. ...
... For this purpose, a set of candidate models with all possible combinations of variables was generated based on the full model and the Akaike Information Criterion with a correction for small sample sizes (AICc) was calculated. We considered candidate models with a ΔAICc < 4 to the best model to have em- pirical support (Burnham & Anderson 2002) and used them to calculate averaged model parameters, p-values and 95% confidence-intervals. We used this approach also with body mass, SMI and sexual dimorphic traits (CF and T) as response variable, including SVL and interaction of SVL and pond type as additional fixed effects (see Table 3 for full models). ...
Article
Full-text available
Ponds in agricultural landscapes are often used by amphibians as breeding habitat. However, the characteristics of agricultural ponds and especially the surrounding area are usually said to be suboptimal for many amphibian species. Using suboptimal habitats might allow a species' survival and reproduction, but can have negative consequences at the individual and population level. In the present study, we investigated Palmate Newt (Lissotriton helveticus) populations from an intensive wine-growing region in southern Germany and compared them with populations located in a nearby forested area in terms of biometric traits, age and genetic structure. By analyzing over 900 adult newts from 11 ponds, we could show that newts reproducing in forest ponds were larger than newts reproducing in agricultural ponds. We did not find differences in the newt age and growth rate between habitat types. Therefore, differences in the body size of newts might already existed in larvae and/or juveniles, what might be related to a lower habitat quality for larvae and/or juveniles in the agricultural landscape. Body mass, body condition and sexual dimorphic traits (length of the caudal filament and max. height of the tail) correlated with body size, but no additional effect of the habitat type was found. The analysis of microsatellites revealed a higher genetic diversity in forest ponds. However, no clear sign of inbreeding was observed in any agricultural population, suggesting some degree of gene flow between them. We conclude, that agricultural ponds can be suitable habitats for the Palmate Newt and that conservation effort should aim to preserve them. The observed effects on body size indicate the need to increase the quality of the aquatic and terrestrial habitat for early life stages of this newt species in agricultural landscapes.
... We selected the bestsupported GLMMs according to the lowest Akaike's information criterion for a small sample size (AICc) ( Table 2). A difference in AICc (ΔAICc) of more than two was considered as evidence for a significant difference between models (Burnham & Anderson 2002). We performed a Pearson correlation to assess the relationship between the number of vole burrow holes and the number of Red Kites at communal roosts, and the relationship between the number of GPS fixes and the overall size of the home range. ...
... GLMMs were used to explore the number of communal roosts, home ranges (95%), and core areas (50%) of Red Kites during the winter in the natal area between 2015 and 2021. Models shown are those with ΔAIC < 4 (2-4 providing 'plausible' evidence according toBurnham & Anderson 2002). Each model is ranked according to differences in Akaike's Information Criterion adjusted for small sample sizes (ΔAICc), logLikelog-likelihood. ...
... DAIC was the difference between the AIC of each of the remaining models and the AIC of the 'best model'. The conventional rules of thumb for interpreting the DAIC values are: values <2 indicate that the candidate model is almost as good as the best model, values 4-7 indicate considerably less support for the candidate model and values >10 indicate that there is no support for this model providing the best fit to the data (Burnham & Anderson, 2002;Fabozzi, Focardia, Rachey & Arshanapalli, 2014). ...
Preprint
Full-text available
Verb-marking errors such as 'she play football' and 'Daddy singing' are a hallmark feature of young children's speech. We investigate the proposal that these errors are input-driven errors of commission, arising from the high relative frequency of subject+unmarked verb sequences in well-formed child-directed speech. We test this proposal via a pre-registered corpus analysis, and ask at what level the effects occur: is it the relative frequency of specific subject+unmarked verb sequences in the input that is important, or is it simply that verbs become entrenched, such that their frequency of appearance with any third-person singular subject accounts for errors? We find that the best predictor of children's verb-marking errors is the relative frequency of unmarked forms of specific subject+verb sequences. Our results support the proposal that children's apparent omissions of certain grammatical morphemes are in fact input-driven errors of commission and provide insight into the mechanisms by which this occurs.
... The feeding station where each vulture was captured was included as a random effect to account for other characteristics at these feeding sites not otherwise included as explanatory variables. We used multimodel inference to select the best models, i.e., the one with the lowest corrected Akaike's Information Criterion (AICc) and those within 4 units of AICc (Burnham & Anderson, 2002). The MuMIn R package was used for these calculations as well as for obtaining the variance explained by the full model (i. ...
Article
Veterinary pharmaceuticals, including antibiotics, are emerging contaminants of concern worldwide. Avian scavengers are exposed to pharmaceuticals through consumption of livestock carcasses used for feeding wildlife for conservation purposes at supplementary feeding stations. Here we tested the hypothesis that griffon vultures (Gyps fulvus) would be more exposed to antibiotics (i.e., quinolones) when feeding on livestock carcasses from intensive farming than when they rely on carcasses from extensive farming or wild animals. We sampled 657 adult griffon vultures captured between 2008 and 2012. In addition, we sampled tissues from domestic livestock supplied at feeding stations in the study area between 2009 and 2019; pig (n = 114), sheep (n = 28), cow (n = 1) and goat (n = 2). Samples were analysed by liquid chromatography with electrospray ionization mass spectrometry (LC-ESI-MS). Quinolones were detected in plasma from 12.9% of the griffon vultures analysed. Quinolone prevalence in griffon vultures varied significantly among feeding stations but was also affected by the total amount of carcasses supplemented, especially the mass of pig carcasses. These results aligned with a 21.1% quinolone prevalence in pig carcasses sampled at feeding stations, wherein enrofloxacin and ciprofloxacin levels of up to 3359 ng/g and 1550 ng/g, respectively, were found. Given enrofloxacin pharmacokinetics in pig tissues, 5.3% of the analysed pigs may have died no more than one day after treatment. Quinolone presence in vultures was negatively associated with blood lead levels, which mostly originates from lead ammunition and indicates a higher consumption of game animal carcasses. Carcass disposal for feeding avian scavengers must always assess and manage the risks posed by veterinary pharmaceuticals, especially when livestock provided may have died soon after treatment.
... The minimum adequate model within a ΔAIC c < 2 was considered the best model. 50 We calculated AIC c weights for all available models. MeHg and THg concentrations were Ln(log)-transformed to ensure normality. ...
Article
Marine predators are monitored as indicators of pollution, but such trends can be complicated by variation in diet. Glaucous-winged gulls (Larus glaucescens) have experienced a dietary shift over the past century, from mainly marine to including more terrestrial/freshwater inputs, with unknown impacts on mercury (Hg) trends. We examined 109-year trends in total mercury (THg) and methylmercury (MeHg) concentrations in glaucous-winged gull feathers (1887− 1996) from the Salish Sea. Adult flank feathers had higher MeHg concentrations than immature feathers, and males head feathers had higher THg concentrations than females. Overall, we found no evidence of a trend in feather MeHg or THg concentrations over time from 1887 to 1996. In the same individuals, δ 15 N, δ 13 C, and δ 34 S declined over time in gull feathers. In comparison, egg THg concentrations declined from 1970 to 2019 in two species of cormorants, likely reflecting decreases in local Hg sources. We conclude that diet shifts through time may have countered increased Hg deposition from long-range transport in glaucous-winged gulls. The lack of Hg trends over time in glaucous-winged gull feathers provides additional support that these gulls have decreased the amount of marine forage fish in their diet.
... The reduced 85% CI was used instead of traditional 95% CI, because the former is more appropriate when conducting AIC-based model selection (Arnold, 2010;Powell and Gale, 2015). Model parameters were considered to be informative if the 85% CI excluded 0, and uninformative if they included 0. Model fit was examined by calculating the marginal (considers variance of only the fixed effects) and conditional R 2 (considers variance of both the fixed and random effects) for the most parsimonious model (Burnham and Anderson, 2002;Faraway, 2016). Data filtering, ANOVA tests, and Tukey tests were conducted in R, version 3.6.1 (R Development Core Team, 2019), and mixed effect modeling was performed in the R packages nlme (Pinheiro et al., 2020) and MuMIn (Bartoń, 2019). ...
... To consider the influential questions across the CLOs, models with loadings smaller than an absolute value of 0.40 were removed from the latent structure (Field et al., 2012). Various relative fit indicessuch as Akaike Information Criterion (AIC; Akaike, 1974), sample size adjusted AIC (AICc; Burnham & Anderson, 2002), Bayesian Information Criterion (BIC; Schwarz, 1978), Sample-Size Adjusted BIC (SABIC; Sclove, 1987), and Hannan-Quinn (HQ;Hannan & Quinn, 1979) criterionwere then used to provide robust evidence for the model comparison. Models with smaller values for each of the different fit indices were considered to provide a better fit. ...
Article
Full-text available
Well-designed online courses enhance learning experiences and allow effective development of learners' skills and knowledge. A critical factor contributing to the design of online courses in the higher education settings are well-defined learning objectives that align with course assessments and learning activities. While there are several introspective instruments to evaluate course designs, with the broader adoption of educational technologies and digital tools, there is a wealth of data that offers insights on the alignment of learning objectives to assessments. Such data has paved the way for evidence-based methods of investigating course effectiveness within higher education. This study outlines a methodology for designing and evaluating the alignment between course learning objectives and assessment activities at scale, utilising a combination of learning analytics and measurement theory approaches, more specificially exploratory multi-dimensional item response theory (MIRT) models. We demonstrate the proposed methodology within a professional development MOOC on leadership skills development, where we evaluate the alignemnt between course objectives and reflective writing assessments activities. Our results suggested that the alignment of the existing course objectives to assessment activities can be improved, showing the practical value of the proposed approach. The theoretical and practical implications of this research are further illustrated.
... where 6(c − 1) is the number of parameters in c − 1 halo components plus the background component and logL is the log-likelihood for the best fit parameters. The relative strengths of AIC and BIC is widely debated, although both are founded on powerful theorems (Lahiri 2001;Konishi & Kitagawa 2008;Burnham & Anderson 2002;Kass & Wasserman 1995;Everitt et al. 2011). The BIC has a well-accepted valuation for relative model merit: one model is strongly (very strongly) favored over another when ∆(BIC) > 6 (> 10) (Kass & Raftery 1995). ...
Preprint
Full-text available
Dark matter simulations require statistical techniques to properly identify and classify their halos and structures. Nonparametric solutions provide catalogs of these structures but lack the additional learning of a model-based algorithm and might misclassify particles in merging situations. With mixture models, we can simultaneously fit multiple density profiles to the halos that are found in a dark matter simulation. In this work, we use the Einasto profile (Einasto 1965, 1968, 1969) to model the halos found in a sample of the Bolshoi simulation (Klypin et al. 2011), and we obtain their location, size, shape and mass. Our code is implemented in the R statistical software environment and can be accessed on https://github.com/LluisHGil/darkmix.
... accounting for both between and within-study heterogeneity) rather than those with only fixed effects (i.e. only accounting for within-study heterogeneity), as they fitted the data better in terms of the Akaike information criterion (AIC) (Burnham and Anderson, 2002). Therefore, both within and between-study heterogeneities were included in the models. ...
Article
Full-text available
Isotope-based approaches to study plant water sources rely on the assumption that root water uptake and within-plant water transport are non-fractionating processes. However, a growing number of studies have reported offsets between plant and source water stable isotope composition for a wide range of ecosystems. These isotopic offsets can result in the erroneous attribution of source water used by plants and potential overestimations of groundwater uptake by the vegetation. We conducted a global meta-analysis to quantify the magnitude of these plant source water isotopic offsets and explored whether their variability could be explained by either biotic or abiotic factors. Our database compiled 112 studies spanning arctic to tropical biomes that reported the dual water isotope composition (δ2H and δ18O) of plant (stem) and source water, including soil water (sampled following various methodologies and along a variable range of depths). We calculated plant source 2H offsets in two ways: a line conditioned excess (LC-excess) that describes the 2H deviation from the local meteoric water line and a soil water line conditioned excess (SW-excess) that describes the deviation from the soil water line, for each sampling campaign within each study. We tested for the effects of climate (air temperature and soil water content), soil class, and plant traits (growth form, leaf habit, wood density, and parenchyma fraction and mycorrhizal habit) on LC-excess and SW-excess. Globally, stem water was more depleted in 2H than in soil water (SW-excess < 0) by 3.02±0.65 ‰ (P < 0.05 according to estimates of our linear mixed model and weighted by sample size within studies). In 95 % of the cases where SW-excess was negative, LC-excess was negative, indicating that the uptake of water that had not undergone evaporative enrichment (such as groundwater) was unlikely to explain the observed soil–plant water isotopic offsets. Soil class and plant traits did not have any significant effect on SW-excess. SW-excess was more negative in cold and wet sites, whereas it was more positive in warm sites. The climatic effects on SW-excess suggest that methodological artefacts are unlikely to be the sole cause of observed isotopic offsets. Our results would imply that plant source water isotopic offsets may lead to inaccuracies when using the isotopic composition of bulk stem water as a proxy to infer plant water sources.
... We censored birds after they went missing from the site or transmitters were believed to be no longer functional. We used Akaike's Information Criterion (AIC c ) to rank the models (Burnham and Anderson 2002). Analysis was preformed separately for each year since annual weather conditions, sample size, and mortality patterns were substantially different. ...
Article
Northern bobwhites (Colinus virginianus) have experienced range-wide population declines. Recent harvest data from northeastern Colorado suggests bobwhite populations have declined and populations can be sensitive to adult nonbreeding season survival. We monitored 157 bobwhites in northeastern Colorado, on the northern periphery of the species’ range, for the 2019-2020 and 2020-2021 nonbreeding seasons to estimate nonbreeding season survival and habitat selection. We used known-fate survival models to assess any variation in survival between the winter stages of early-winter, mid-winter, and late-winter, as well as sex, age class, and mass at the time of capture. Survival varied among winter stages, with the period having the lowest survival corresponding with the harshest weather conditions. We also monitored bobwhite habitat selection by performing weekly covey habitat surveys to collect vegetation cover percentages, micro-climate variables, vegetation structure variables, and plant species richness estimates at used and random sites. We used stepwise backward selection modeling to determine if any variables were being selected disproportionate to their availability. Bobwhites chose locations with higher levels of visual obstruction, percent bare ground and percent litter cover Winter survival was low in our study and dependent on weather conditions and predation. Habitat management has the potential to reduce the risk of nonbreeding season mortality for bobwhites, which is critical to sustain local populations at the northwestern edge of the species’ range. Our research aims to provide demographic and habitat selection data to managers to assist them in management action decision making. Advisors: Larkin Powell and Adam Behney
... AICc scores provided us with the likelihood of a model given a dataset while minimizing model complexity. Models within two AICc values of the "best model" were deemed equally good (Burnham & Anderson, 2003;Wheeler et al., 2011). We compared these hypotheses-based models with a null model (i.e., intercept-only, involving modeling each PC value against a constant). ...
Article
Full-text available
Pigmentation is one of the most striking examples of diversity in the natural world. Specifically, primate pelage (hair/fur) pigmentation provides a substrate for selection, and is tied to functions such as crypsis, signaling, and thermoregulation. However, most of our hypotheses on the forces responsible for primate pelage pigmentation are based solely on macro-evolutionary studies. Here, we test how distinct hypotheses around climate, age, and sex might be influencing pelage color variation in a wild primate population. We characterize pelage color and pattern variation within a population of wild diademed sifakas (Propithecus diadema) exhibiting striking diversity in coloration (melanic to tri-colored). Our approach jointly assesses climate and pelage variation across the region. We score pelage using a semi-quantitative methodology (n = 87 individuals). We then test if pelage variation is associated with climatic or demographic factors (i.e., sex–class, age–class) across the Tsinjoarivo forest, Madagascar. We find darker bodies and greater facial complexity in colder and more fragmented forests. We conclude that three explanations may explain the phenotypic pattern: isolation by distance, an environmental gradient, or local adaptation. Although it is difficult to reach a causal conclusion without genetic data, each scenario signals the need for enhanced conservation of diademed sifakas in the Tsinjoarivo forest. More studies on primate pigmentation in wild populations will be needed to contextualize if this pattern is exceptional or typical. We argue that primate color in wild populations may also foretell either adaptation or maladaptation (i.e., via environmental mismatch).
... We evaluated a third model considering the additive effect of both N-Census and OMA. The AIC c was used for model selection (Burnham and Anderson 2002). The Catalan Griffon Vulture census is only performed every decade and it was provided by Del Moral (2009) For each grid, we only have two observations of the dependent variable and we assumed that number of pairs changed linearly through the period. ...
Article
Full-text available
Globally, vultures are one of the most threatened of all groups of birds. European vulture populations are benefited by several anthropogenic food sources such as landfills. Current European Union directives aim to decrease the amount of organic matter dumped in landfills, reducing this important food source for some vulture species. In this context, we assessed the effect of the reduction of organic waste available and accessible for scavengers in a landfill on the visitation probability and abundance of a local Eurasian Griffon Vulture Gyps fulvus population in Central Catalonia (NE Iberian Peninsula), using a long-term dataset of captured-marked-recaptured individuals in the period 2012–2018. Our results indicated a decrease in the visitation probability due to a significant reduction of organic matter dumped into the landfill after a waste treatment centre was built (0.82 to 0.76) that may cause a permanent emigration of vultures in response to food reduction. However, the estimated annual abundance of vultures tended to grow over time due to the positive trend that regional vulture populations have experienced in recent decades. These results suggest that population processes occurring at regional scales are more relevant to vulture populations than local waste management measures. A reduction in locally available food can make a site less attractive, but species with high dispersal capacity such as vultures may overcome this issue by moving to other suitable sites. Although Griffon Vultures obtain most of the food from domestic and wild ungulates, a regional application of European directives could threaten an important alternative feeding source, especially in food shortage seasons where landfills could be supporting the energetic requirements of the species. Conservation strategies should be planned to counteract the possible negative effects of new European directives on scavenger populations.
... To evaluate the equidistance hypothesis, the fitted linear mixed models were compared using the Akaike Information Criterion (AIC), following the method of Burnham and Anderson (2002), based on the evidence ratio, defined by: ...
Article
When the gradation of animal welfare is scored through ordinal scales and equidistant tags are used, empirical data between tags tend to be non-equidistant. Ordinal rate scales (ORS) and visual analogue scales (VAS) were tested for the assessment of contact dermatitis on breast and abdominal areas (CD), footpad dermatitis (FP), hock burns (HB) and bird soiling (BS) in broiler chickens. Calculations regarding the inter-rater reliability, the correlation between VAS and ORS and amongst the welfare indicators measured with both scales, and the equidistance of ORS categories in relation to values measured using VAS, were made. A total of 1,303 broiler chickens from 10 flocks was assessed on-farm by three raters using both scales. Inter-rater reliabilities of CD and HB were higher when using VAS compared with ORS, but FP was lower. Correlations between scales varied between 0.90-0.97 and 0.77-0.95 (P<0.001), considering mean and individual values. Low-to-moderate correlations were observed between the four indicators using the scales. Tags on VAS that best represented ORS were non-equidistant. Results suggest both scales were reliable assessing the selected broiler chicken welfare indicators.
... Traditionally the past dynamic of lineages was inferred using the fossil record (e.g., Stanley 1979), although recent progress have led to estimate diversification rates from dated phylogenies (e.g., Harvey et al. 1994, Morlon et al. 2011, Rabosky et al. 2013). Numerous issues have arisen regarding the analyses of the fossil record: (1) using discrete time bins, (2) analyzing only first and last appearances of taxa, thus ignoring other occurrences if available while first and last occurrences cannot be assumed to reflect the true times of speciation/origination and extinction (Strauss andSadler 1989, Marshall 1990), (3) the inability to perform model testing against over-parameterization, which limits models' robustness (Burnham and Anderson 2002), (4) the poor or sole incorporation of extant taxa in the analyses, which in both cases represents only a small fraction of the total diversity that has existed since their origination (e.g., Raup 1986). ...
Article
Deciphering the timing of lineage diversification and extinction has greatly benefited in the last decade from methodological developments in fossil-based analyses. If these advances are increasingly used to study the past dynamics of vertebrates, other taxa such as insects remain relatively neglected. Our understanding of how insect clades waxed and waned or of the impact of major paleoenvironmental changes during their periods of diversification and extinction (mass extinction) are rarely investigated. Here, we compile and analyze the fossil record of Plecoptera (1,742 vetted occurrences) to investigate their genus-level diversification and diversity dynamics using a Bayesian process-based model that incorporates temporal preservation biases. We found that the Permian-Triassic mass extinction has drastically impacted Plecoptera, while the Cretaceous Terrestrial Revolution corresponds with a turnover of plecopteran fauna. We also unveiled three major gaps in the plecopteran fossil record: the Carboniferous-Permian transition, the late Early Cretaceous, and the late Cenomanian to Bartonian, which will need to be further investigated. Based on the life history of extant Plecoptera, we investigate the correlations between their past dynamic and a series of biotic (Red Queen hypothesis) and abiotic (Court Jester hypothesis) factors. These analyses highlight the major role of continental fragmentation in the evolutionary history of stoneflies, which is in line with phylogeny-based biogeographic analyses showing how vicariance drove their diversification. Our study advocates analyzing the fossil record with caution, while attempting to unveil the diversification and extinction periods plus the likely triggers of these past dynamics of diversification.
... Alternative models for each response variable were compared using their associated Akaike Information Criterion (AIC) values, where lower AIC values indicate greater empirical support for a model. Differences in AIC values (∆AIC) < 2, between 4 and 7, and >10 respectively indicate negligible, moderate and strong empirical support between alternative models (Burnham & Anderson 2002). As the 1980 density in the unthinned control stand was unavailable, we: (i) fitted a regression for the relationship between tree density post-thinning in 1980 and density of live stems ≥10 cm DBH in 2009 (df =13, R 2 = 0.92) for all but the unthinned stand and then: (ii) back-estimated tree density in 1980 from the fitted model and the estimated density of the unthinned stand in 2009. ...
Article
Full-text available
Background: The tree stem density which optimises merchantable timber yield (volume per unit area) is unknown for most of New Zealand's indigenous tree species. While moderate thinning of even-aged stands can promote yield, intense thinning may decrease yield by creating space that cannot be filled by residual trees, increasing tree mortality or reducing tree height. We quantified the effects of density on silver beech (Lophozonia menziesii (Hook.f.) Heenan & Smissen) tree growth, height and mortality, identified the density leading to optimal merchantable yield and assessed if this density varied with stand age.
... Model-averaged parameter estimates (i.e., weighted averages of model coefficients across all possible models) are shown in Table 4, sorted by parameter importance (i.e., the sum of the evidence weights across all models). All three main effects have high importance values, whereas the importance values for the three interactions are lower than those for the main effects and also below the recommended threshold of 0.8 [97]. This suggests that the two significant interactions in the best model (FOV by weight as well as pixel density by weight) are of limited importance. ...
Article
Full-text available
Distances are commonly underperceived in virtual reality (VR), and this finding has been documented repeatedly over more than two decades of research. Yet, there is evidence that perceived distance is more accurate in modern compared to older head-mounted displays (HMDs). This meta-analysis of 131 studies describes egocentric distance perception across 20 HMDs, and also examines the relationship between perceived distance and technical HMD characteristics. Judged distance was positively associated with HMD field of view (FOV), positively associated with HMD resolution, and negatively associated with HMD weight. The effects of FOV and resolution were more pronounced among heavier HMDs. These findings suggest that future improvements in these technical characteristics may be central to resolving the problem of distance underperception in VR.
... (2) AIC = n × log SSE n + 2 × p. smaller datasets, and which is used herein (Hurvich and Tsai 1989). The AIC c calculation is shown in Eq. 3 AIC c should be used when the ratio of training samples ( n ): number of parameters ( p ) is less than 40 (Burnham and Anderson 2002), i.e., when there are less than 40 samples of data for every training parameter in the model. The AIC c approaches and converges to the AIC value as the number of training samples approaches infinity. ...
Article
Full-text available
To validate the application of machine learning (ML) to rock engineering practice, it is crucial that algorithm developers use appropriate methods to quantify how closely the ML reproduces the observed rock mass deformation. Input variable selection (IVS) is one approach that examines how ML uses the given data, or inputs, to forecast rock mass behavior. Three IVS methods were developed for two convolutional neural network (CNN) architectures that predict tunnel liner yield at the Cigar Lake Mine, which exhibits time-dependent squeezing deformation. One model architecture focused on accurately predicting the higher tunnel liner yield classes, while the second architecture prioritized prediction accuracy across all tunnel liner yield classes. The three IVS methods investigated herein were channel activation strength (CAS), input omission (IO), and partial correlation (PC). The IO and PC approaches proposed are novel approaches proposed for CNNs using a spatial and temporal geomechanical dataset. Performance of all models was compared using the corrected Akaike information criterion (AICc), where lower values indicate better performance. Each IVS method was used to produce a unique ranking for each model architecture and training/testing data split: CAS produced an activation ranking, IO produced an Omission Ranking, and PC produced a correlation ranking. The activation rankings showed that the geology input had the lowest activation strength in the CNN relative to the other inputs (ground freezing, primary installed support class, and radial tunnel displacement). Geology had the highest omission ranking, resulting from it having the most negative impact on performance as compared to the other inputs when it was omitted from the models entirely. The PC approach, using the Correlation Rankings, found that the highest model performances were reached when the most recent radial tunnel displacement was added into the pool of candidate inputs. The three IVS approaches and their respective rankings proved to be useful for analyzing the CNN inputs in terms of importance and confirming underlying assumption about the deformation mechanics at Cigar Lake Mine. Collectively, the IVS analyses indicated that all of the available digitized inputs for the Cigar Lake Mine CNNs are needed to produce good model performances. Each IVS method revealed different insights into this CNN development. Undertaking IVS for ML developed using geomechanical datasets allows for verification of the algorithms and thereby a better understanding of the nuance of the rock mass deformation. At Cigar Lake Mine, these findings may be used to assist in forecasting the schedule and budget for ground support rehabilitation.
... We used the glmer function in the lme4 package (Bates et al. 2015) to run logistic regression models. We identified the best fit and most parsimonious model predicting emergence success, using small sample-corrected Akaike information criterion (Burnham and Anderson 2002). Given a strong divergence in emergence success between FTR and STR through time, all candidate models, except for the null model, included the main effects of month and temperature, as well as their interactions. ...
Article
Full-text available
Megachile rotundata (F.) is an important pollinator of alfalfa in the United States. Enhancing landscapes with wildflowers is a primary strategy for conserving pollinators and may improve the sustainability of M. rotundata. Changing cold storage temperatures from a traditionally static thermal regime (STR) to a fluctuating thermal regime (FTR) improves overwintering success and extends M. rotundata’s shelf life and pollination window. Whether floral resources enhance overwintering survival and/or interact with a thermal regime are unknown. We tested the combined effects of enhancing alfalfa fields with wildflowers and thermal regime on survival and macronutrient stores under extended cold storage (i.e., beyond one season). Megachile rotundata adults were released in alfalfa plots with and without wildflower strips. Completed nests were harvested in September and stored in STR. After a year, cells were randomly assigned to remain in STR for 6 months or in FTR for a year of extended cold storage; emergence rates were observed monthly. Macronutrient levels of emerged females were assessed. FTR improved M. rotundata survival but there was no measurable effect of wildflower strips on overwintering success or nutrient stores. Timing of nest establishment emerged as a key factor: offspring produced late in the season had lower winter survival and dry body mass. Sugars and glycogen stores increased under FTR but not STR. Trehalose levels were similar across treatments. Total lipid stores depleted faster under FTR. While wildflowers did not improve M. rotundata survival, our findings provide mechanistic insight into benefits and potential costs of FTR for this important pollinator.
... Standardized lagged association rates (SLAR) were run in SOCPROG to estimate temporal stability in dyads, that is, the probability that a dyad associated at one time remains associated after a time lag [61,71]. Four exponential models were fit to the output data; the models include combinations of the duration of associations (long term ("constant companions", CC), short term ("casual acquaintances", CA), and disassociation [61,71], with the best-fit model determined using the lowest quasi-Akaike Information Criterion (QAIC) [72]. Precision was estimated using jackknife resampling, wherein one observation is omitted during repeated resampling [61,73,74]. ...
Article
Full-text available
The social structure of estuarine-resident bottlenose dolphins is complex and varied. Residing in habitats often utilized for resource exploitation, dolphins are at risk due to anthropogenic pressures while still federally protected. Effective conservation is predicated upon accurate abundance estimates. In North Carolina, two estuarine-resident stocks (demographically independent groups) of common bottlenose dolphin have been designated using spatiotemporal criteria. Both stocks are subjected to bycatch in fishing gear. The southern North Carolina estuarine stock was estimated at <200 individuals from surveys in 2006, which is outdated per US guidelines. Thus, we conducted a new capture-mark-recapture survey in 2018, identifying 547 distinct individuals, about three times higher than the prior abundance estimate. We compared those individuals to our long-term photo-identification catalog (1995-2018, n = 2,423 individuals), matching 228 individuals. Of those 228, 65 were also included in the 2013 abundance estimate for the northern North Carolina estuarine stock. Using sighting histories for all individuals in the long-term catalog, we conducted a social network analysis, which is independent of a priori stock assignments. The three primary clusters identified were inconsistent with current stock designations and not defined by spatiotemporal distribution. All three clusters had sighting histories in the estuary and on the coast, however, that with the highest within-cluster associations appeared to use estuarine waters more often. The within-cluster association strength was low for one cluster, possibly due to only part of that cluster inhabiting the southern North Carolina estuarine system. Between-cluster differences occurred in infestation rates by the pseudostalked barnacle, Xenobalanus globicipitis, but that did not predict clusters. We suggest the need to re-evaluate the stock structure of estuarine-resident common bottlenose dolphins in North Carolina and currently have insufficient information to assign an abundance estimate to a currently designated stock.
... Generalized linear models (GLM) with categorical and continuum predictors were used as well. The combinations of predictors that optimally explained δ 15 N in plant leaves were selected using the corrected Akaike's information criterion (AICc) [40]. The values of δ 13 C and δ 15 N in each sample of soil, litter, or plant leaves of one taxon on each sample plot or the value of another feature on a sample plot were the units for statistical analysis. ...
Article
Full-text available
δ13C and δ15N were analyzed in the leaves of Pinus sylvestris L. and Betula spp. under the conditions of severe heavy metal (Zn, Cu, Cd, and Pb) contamination. Twenty-seven plots located near the Karabash copper smelter (Russia) were studied. No reliable correlation of 13C in tree leaves with the level of pollution was observed. δ15N, both in Pinus sylvestris and Betula spp., increased similarly in polluted areas. δ15N was increased by 2.3‰ in the needles of Pinus sylvestris and by 1.6‰ in the leaves of Betula spp. in polluted plots compared to the background ones. The probable reasons for the increase in δ15N were estimated using multiple regression. The regression model, which includes two predictors: δ15N in the humus horizon and the occurrence of roots in the litter, explains 33% of the total variability of δ15N in leaves. Thus, in ecosystems polluted with heavy metals, the state of trees is determined not only by the direct toxic effects of heavy metals but also by indirect ones associated with the features of plant mineral nutrition. This fact opens the way to the search for opportunities to control the state of plants in disturbed ecosystems by regulating the content of mineral nutrition elements.
... Here, indices 1 and 2 refer to qPCR pla and caf, respectively, and cov denotes the pairwise covariance between diagnostic outcomes [22]. We assess whether the inclusion of a covariance factor affected the fit to the data, using DIC as an indicator of fit [23], and whether estimated test characteristics were robust to departures from assumptions of independence. ...
Article
Full-text available
During outbreaks, the lack of diagnostic “gold standard” can mask the true burden of infection in the population and hamper the allocation of resources required for control. Here, we present an analytical framework to evaluate and optimize the use of diagnostics when multiple yet imperfect diagnostic tests are available. We apply it to laboratory results of 2,136 samples, analyzed with 3 diagnostic tests (based on up to 7 diagnostic outcomes), collected during the 2017 pneumonic (PP) and bubonic plague (BP) outbreak in Madagascar, which was unprecedented both in the number of notified cases, clinical presentation, and spatial distribution. The extent of these outbreaks has however remained unclear due to nonoptimal assays. Using latent class methods, we estimate that 7% to 15% of notified cases were Yersinia pestis -infected. Overreporting was highest during the peak of the outbreak and lowest in the rural settings endemic to Y . pestis . Molecular biology methods offered the best compromise between sensitivity and specificity. The specificity of the rapid diagnostic test was relatively low (PP: 82%, BP: 85%), particularly for use in contexts with large quantities of misclassified cases. Comparison with data from a subsequent seasonal Y . pestis outbreak in 2018 reveal better test performance (BP: specificity 99%, sensitivity: 91%), indicating that factors related to the response to a large, explosive outbreak may well have affected test performance. We used our framework to optimize the case classification and derive consolidated epidemic trends. Our approach may help reduce uncertainties in other outbreaks where diagnostics are imperfect.
... [61] with the function glm.nb in the MASS package [62]. The adopted distribution was selected by the Akaike Information Criteria (AIC) [63]. ...
Article
Full-text available
... Sixth, only 5 goodness-of-fit measures were applied to compare model usefulness (Fig. 10). Other indicators (e.g., Hannan-Quinn criterion [57] and minimum description length [58] ) could be employed in future studies. ...
Article
Background: The negative impacts of COVID-19 (ImpactCOVID) on public health are commonly assessed using the cumulative numbers of confirmed cases (CNCCs). However, whether different mathematical models yield disparate results based on varying time frames remains unclear. This study aimed to compare the differences in prediction accuracy between 2 proposed COVID-19 models, develop an angle index that can be objectively used to evaluate ImpactCOVID, compare the differences in angle indexes across countries/regions worldwide, and examine the difference in determining the inflection point (IP) on the CNCCs between the 2 models. Methods: Data were downloaded from the GitHub website. Two mathematical models were examined in 2 time-frame scenarios during the COVID-19 pandemic (the early 20-day stage and the entire year of 2020). Angle index was determined by the ratio (=CNCCs at IP÷IP days). The R2 model and mean absolute percentage error (MAPE) were used to evaluate the model's prediction accuracy in the 2 time-frame scenarios. Comparisons were made using 3 visualizations: line-chart plots, choropleth maps, and forest plots. Results: Exponential growth (EXPO) and item response theory (IRT) models had identical prediction power at the earlier outbreak stage. The IRT model had a higher model R2 and smaller MAPE than the EXPO model in 2020. Hubei Province in China had the highest angle index at the early stage, and India, California (US), and the United Kingdom had the highest angle indexes in 2020. The IRT model was superior to the EXPO model in determining the IP on an Ogive curve. Conclusion: Both proposed models can be used to measure ImpactCOVID. However, the IRT model (superior to EXPO in the long-term and Ogive-type data) is recommended for epidemiologists and policymakers to measure ImpactCOVID in the future.
... All models for which cumulative Akaike weight remained below 0.95 were retained to constitute a 95% confidence interval set of models (Burnham & Anderson, 2002). Climate variables estimates were averaged among selected models and weighted by models AIC. ...
Preprint
Understanding boreal/hemi-boreal forest growth sensitivity to seasonal variations in temperature and water availability provides important basis for projecting the potential impacts of climate change on the productivity of these ecosystems. Our best available information currently comes from a limited number of field experiments and terrestrial biosphere model (TBM) simulations of varying predictive accuracy. Here, we assessed the sensitivity of annual boreal/hemi-boreal forest growth in Canada to yearly fluctuations in seasonal climate variables using a large tree-ring dataset and compared this to the climate sensitivity of annual net primary productivity (NPP) estimates obtained from fourteen TBMs. We found that boreal/hemi-boreal forest growth sensitivity to fluctuations in seasonal temperature and precipitation variables changed along a southwestern to northeastern gradient, with growth limited almost entirely by temperature in the northeast and west and by water availability in the southwest. We also found a lag in growth climate sensitivity, with growth largely determined by the climate during the summer prior to ring formation. Analyses of NPP sensitivity to the same climate variables produced a similar southwest to northeast gradient in growth climate sensitivity for NPP estimates from all but three TBMs. However, analyses of growth from tree-ring data and analyses of NPP from TBMs produced contrasting evidence concerning the key climate variables limiting growth. While analyses of NPP primarily indicated a positive relationship between growth and seasonal temperature, tree-ring analyses indicated negative growth relationships to temperature. Also, the positive effect of precipitation on NPP derived from most TBMs was weaker than the positive effect of precipitation on tree-ring based growth: temperature had a more important limiting effect on NPP than tree-ring data indicated. These mismatches regarding the key climate variables limiting growth suggested that characterization of tree growth in TBMs might need revision, particularly regarding the effects of stomatal conductance and carbohydrate reserve dynamics.
Chapter
Carnivore species are believed to exert strong competitive pressure on each other, resulting in adaptations to allow for niche separation through resource partitioning. However, factors that promote ecological separation among species in tropical forests are difficult to explain and are poorly understood because robust field studies are lacking. We examined spatial, temporal and morphological segregation between tropical carnivores in a protected forest in north‐central Thailand. Sympatric spatial overlap was calculated from radio‐telemetry data of 38 individuals from six species (5 yellow‐throated martens, Martes flavigula , 20 leopard cats, Prionailurus bengalensis , 2 Asiatic golden cats, Catopuma temminckii , 4 clouded leopards, Neofelis nebulosa , 5 binturongs, Arctictis binturong , and 2 dholes, Cuon alpinus ) in the same study area. Spatial overlap was then correlated with 14 independent variables (i.e. skull and dental morphology, body mass, habitat use and activity patterns) compared among the six species. We predicted that carnivores with differing morphology and activity patterns would exhibit more spatial overlap because these species would compete less for prey resources. Our statistical analyses indicated that lower mean carnassial length and activity patterns in closed habitat cover were significantly correlated ( p < 0.05) with species spatial overlap. Binturongs appeared to have the greatest amount of spatial overlap with other species of carnivores, whereas dholes had the least spatial overlap; also, dholes and yellow‐throated martens tended to be more active in open habitats and during diurnal time periods, whereas clouded leopards and Asiatic golden cats were more active in closed cover and were more arrhythmic in activity. Although these results provide useful information on carnivore coexistence, we recommend that future studies monitor larger sample sizes of carnivore species over the same time period to provide more robust statistical analyses. In addition, we suggest that future research on carnivore coexistence evaluates the impacts of anthropogenic activity on study results.
Book
Full-text available
Introdução, caracterização e classificação de modelos, amostragem para modelagem florestal, transformação de variáveis, transformação box-cox, ajuste e construção de modelos, eliminação de valores discrepantes, escolha do melhor modelo, construção de modelos lineares, análise de covariância, avaliação e validação de modelos, comparação entre modelos, modelos especiais de regressão, modelos alométricos florestais (relação hipsométrica, afilamento do tronco, volume individual do tronco, fator de forma, casca, diâmetro e volume sem casca, diâmetro da copa, volume da copa), modelos de crescimento e produção, modelos de povoamento completo, índices de sítio, índices de competição, tabelas de crescimento e produção, modelos de classe de diâmetro, modelos de distribuição para florestas equiâneas.
Article
Purpose: This article summarizes a number of presentations from a session on "Radiation and Circulatory Effects" held during the Radiation Research Society Online 67th Annual Meeting, October 3 - 6 2021. Materials and methods: Different epidemiological cohorts were analyzed with various statistical means common in epidemiology. The cohorts included the one from the U.S. Million Person Study and the Canadian Fluoroscopy Cohort Study. In addition, one of the contributions in our article relies on results from analyses of the Japanese atomic bomb survivors, Russian emergency and recovery workers and cohorts of nuclear workers. The Canadian Fluoroscopy Cohort Study data were analyzed with a larger series of linear and nonlinear dose-response models in addition to the linear no-threshold (LNT) model.Results and Conclusions: The talks in this symposium showed that low/moderate acute doses at low/moderate dose-rates can be associated with an increased risk of CVD, although some of the epidemiological results for occupational cohorts are equivocal. The usually only limited availability of information on well-known risk factors for circulatory disease (e.g. smoking, obesity, hypertension, diabetes, physical activity) is an important limiting factor that may bias any observed association between radiation exposure and detrimental health outcome especially at low doses. Additional follow-up and careful dosimetric and outcome assessment are necessary and more epidemiological and experimental research is required. Obtaining reliable information on other risk factors is especially important.
Article
Full-text available
In contrast to old-growth forests, early-successional stands remain understudied despite potentially harbouring species of conservation interest. With this work, focused on hazel grouse Tetrastes bonasia , a cryptic and indicator species known to select for close-to-natural forests, we evaluated winter densities, home range, microhabitat selection and diet, combining DNA-based mark-recapture and metabarcoding from faecal samples. In total, 216 droppings, collected over 2 years along forest transects in the Italian Alps, were successfully genotyped and 43 individuals were identified. Density estimates were similar to values reported by other studies in the Alps with an average of 4.5 and 2.4 individuals/km ² in the first and second study year, respectively, and mean home ranges estimated at 0.95 km ² . According to habitat selection models and eDNA-based diet analysis, hazel grouse selected early-succession secondary-growth forests formed after the abandonment of traditional agropastoral activities. These forests, mostly composed of hazel Corylus avellana, Norway spruce Picea abies and Sorbus spp., provided winter food resources and shelter. The diet analysis also highlighted forest arthropods as a non-negligible source of food. Birds avoided areas subject to intensive browsing by ungulates; small forest roads seasonally closed to traffic had positive influence on hazel grouse (i.e. higher abundance of droppings), while roads open to traffic had no effect. Importantly, despite the high coverage of mature forest habitats of Community Interest (53% of our study area), droppings were more abundant in non-listed early-succession secondary forests with similar plant composition. Our results suggest that forest succession after agropastoral abandonment may be beneficial for some forest birds of conservation interest, while acknowledging its negative effects on the previous grassland biodiversity. Graphical abstract
Article
Full-text available
Male and female reproductive behaviour is typically synchronised. In species such as those in the family Cervidae , reproductive timing is often cued by photoperiod, although in females, it can be dependent on body condition. When a species is introduced to a novel environment, the environment changes, or responses of the sexes to such cues differ, asynchronous reproductive behaviour between males and females may occur. We investigated the seasonality of reproductive behaviour in introduced chital deer in northern Queensland by examining male antler phase in relation to female conception rates. We then analysed the influence of different variables likely to affect the timing of male and female reproductive physiology. The lowest percentage of chital in hard antler in any 1 month in this study was 35% (Fig. 1), but the average value was closer to 50%, thus there was a seasonal peak in antler phase linked with photoperiod. Females conceived at any time of year, but were strongly influenced by the amount of rainfall 3 months prior to conception. This resulted in varying conception peaks year-to-year that often did not correspond to the male’s peak in hard antler. In this system, a proportion of males and females were physiologically and behaviourally ready to mate at any time of the year. We predict that differences in the timing of the peaks between the males and females will lead to increased reproductive skew (variation in reproductive success among individual males). This pattern may select for different mating strategies or physiological mechanisms to increase reproductive success.
Article
Geographical location, landscape heterogeneity, and their position at the edge of Quaternary glaciations are the major reasons why the river basins of the Western Carpathians are of the utmost importance for research on European aquatic biodiversity. Moreover, unlike streams, springs have several features determining their insularity in the river continuum. This study aimed to uncover the mechanism forming the diatom assemblages in the karst springs environment, using a two-dimensional approach, including taxonomic and trait data. Thus, (1) the congruence between observed and expected values of null models of taxonomic/functional diversity and functional dispersion and (2) Generalized mixed models and Variation partitioning were used to assess how taxonomic/functional diversity and composition were associated with the local conditions, regional affiliation and spatial springs distance. In 64 rheocrene karst springs (10 karst mountain ranges) of the Western Carpathians, 138 epilithic diatom taxa were found. Null model analyses did not uncover any clear evidence of non-random events forming the spring diatom community. Although, the observed functional diversity did not differ from the null model for most springs, most observed values were lower than the mean value of the null model, suggesting that, except for random events, environmental filtering was an important driver of community assembly. Substratum type, spring connectivity and phosphorus concentration drove the value of taxonomic richness, whereas functional diversity was controlled by silica and bicarbonate concentration as well as spring size. In both models, affiliation to the mountain range (defined as a random factor) was significant. The Variation Partitioning Analysis showed that diatom assemblage composition was mainly determined by local and regional environmental conditions, whilst the distance between springs had only a weak influence.
Thesis
In this thesis I assess the ability of biodiversity to provide a functioning pest control ecosystem service to control moth pest species in UK apple orchards. I assess the ability of four types of farm management: organic, Linking Environment and Farming (LEAF), integrated pest management (IPM) and conventional, to measure the ability of pest predation from birds, and the impact that predation has on apple yields. I firstly describe the history and the landscape of the study area, an overview of the methods used and the farming systems that the field study and experiments took place on in Chapter 2. In Chapter 3 I assess farmland biodiversity by monitoring birds and butterflies as indicator species of biodiversity, to understand if farm management impacts biodiversity levels. Biodiversity was highest on organic orchards, which supports the plethora of studies in the literature. Using this information of biodiversity levels on orchard management types, in Chapter 4 I investigate whether this biodiversity supports a pest control service, and to a natural pest control service compares to a synthetic alternate used on non-organic orchards, through using a sentinel prey experiment in field. Pest control services were greater on organic farms, and followed the same patterns as insectivorous bird abundance, species richness, diversity, and density. This chapter also compares moth pest levels to understand the pest pressures across farms, which harbour different pest control strategies and showed that moth pest levels were broadly similar across all farm management types. Finally, in Chapter 5 I compare the farm management options available to famers, both the natural pest control system and the synthetic control system, using economic valuation methods. Although a natural pest control service from birds is present on organic orchards (Chapter 4), the yield per hectare increased significantly on non-organic orchards (expect LEAF) but is found to be in-different to yield value per hectare of organic orchards in variable scenarios. Importantly, the synthetic alternative to a pest control service available from wild insectivorous birds was found to be an insignificant farm management variable that impacts apple yield and yield value on non-organic orchards.
Article
Full-text available
Background Animal personality has emerged as a key concept in behavioral ecology. While many studies have demonstrated the influence of personality traits on behavioral patterns, its quantification, especially in wild animal populations, remains a challenge. Only a few studies have established a link between personality and recurring movements within home ranges, although these small-scale movements are of key importance for identifying ecological interactions and forming individual niches. In this regard, differences in space use among individuals might reflect different exploration styles between behavioral types along the shy-bold continuum. Methods We assessed among-individual differences in behavior in the European hare (Lepus europaeus), a characteristic mammalian herbivore in agricultural landscapes using a standardized box emergence test for captive and wild hares. We determined an individuals’ degree of boldness by measuring the latencies of behavioral responses in repeated emergence tests in captivity. During capture events of wild hares, we conducted a single emergence test and recorded behavioral responses proven to be stable over time in captive hares. Applying repeated novel environment tests in a near-natural enclosure, we further quantified aspects of exploration and activity in captive hares. Finally, we investigated whether and how this among-individual behavioral variation is related to general activity and space use in a wild hare population. Wild and captive hares were treated similarly and GPS-collared with internal accelerometers prior to release to the wild or the outdoor enclosure, respectively. General activity was quantified as overall dynamic body acceleration (ODBA) obtained from accelerometers. Finally, we tested whether boldness explained variation in (i) ODBA in both settings and (ii) variation in home ranges and core areas across different time scales of GPS-collared hares in a wild population. Results We found three behavioral responses to be consistent over time in captive hares. ODBA was positively related to boldness (i.e., short latencies to make first contact with the new environment) in both captive and wild hares. Space use in wild hares also varied with boldness, with shy individuals having smaller core areas and larger home ranges than bold conspecifics (yet in some of the parameter space, this association was just marginally significant). Conclusions Against our prediction, shy individuals occupied relatively large home ranges but with small core areas. We suggest that this space use pattern is due to them avoiding risky, and energy-demanding competition for valuable resources. Carefully validated, activity measurements (ODBA) from accelerometers provide a valuable tool to quantify aspects of animal personality along the shy-bold continuum remotely. Without directly observing—and possibly disturbing—focal individuals, this approach allows measuring variability in animal personality, especially in species that are difficult to assess with experiments. Considering that accelerometers are often already built into GPS units, we recommend activating them at least during the initial days of tracking to estimate individual variation in general activity and, if possible, match them with a simple novelty experiment. Furthermore, information on individual behavioral types will help to facilitate mechanistic understanding of processes that drive spatial and ecological dynamics in heterogeneous landscapes.
Article
Full-text available
Context Large-scale programs for eradication of pest mammals are confronted with the challenge of managing reinvasion. Exploiting high-elevation landscape features that naturally limit the rate of reinvasion is a strategy that is presumed to improve the success of such initiatives, however, the efficacy of doing so has not yet been investigated. Objectives We aimed to assess whether high-elevation landforms limit the movements of 10 species of invasive small mammal in New Zealand to such a degree that they could be exploited in landscape-scale eradication programmes. Methods We determined the upper elevation limits of species’ distributions, and made spatial predictions based on occupancy models. We applied these in concert to a 310,000 ha area of rugged mountainous environments and identified landforms that function as dispersal barriers to each species of interest. We validated our predictions with existing presence/absence and GPS movement data, and tested our predictions of high-elevation landform barriers with the GPS movement data of a sample of European hedgehogs ( Erinaceus europaeus ). Results We found that the extent of barriers which limited movement ranged from widespread (5/10 species), to localised, (3/10 species) to limited (2/10 species). Our predictions of hedgehog movement barriers were strongly supported by GPS movement data of 26 hedgehogs that were tracked in the study area. Conclusions Our findings show there is enormous potential to advance landscape-scale eradication of invasive small mammals in areas adjacent to high-elevation landforms by identifying and exploiting landscape features that limit the movement of target species in the strategies of eradication programmes.
Thesis
Full-text available
As climate change proceeds, ecosystems are changing rapidly. Marine ecosystems are complex and difficult to monitor, making the prediction of future changes a daunting task. Seabirds are often suggested as potential bioindicators, yet the development of their data for general predictive use is rare. I, using the data of many collaborators, examined the potential use of Arctic Tern (Sterna paradisaea) diet and demographic data as bioindicators in the rapidly warming Gulf of Maine (GOM) ecosystem. The collapse of the breeding colony at Machias Seal Island, formerly the largest in North America, was primarily driven by unmitigated egg predation by large Larus gulls, not a decline in Atlantic herring (Clupea harengus). At the metapopulation scale, reproductive success of Arctic Terns was subject to multiple environmental interactions. The proximity of a colony to shore was associated with higher diet quality, but also greater predator pressures. Interactions between diet quality, weather, and predator behaviour strongly influenced the reproductive success of terns. At the ecosystem scale, the diets of Arctic and Common Terns (S. hirundo) could be combined with environmental data to describe and track ecosystem states. Arctic Terns were more likely to consume white hake (Urophycis tenuis) and small marine invertebrates, but their dietary trends also tracked measures of the herring stock and fishery. Common Terns were more likely to target high-lipid fish, but they showed stronger spatial trends that limited regional inferences. Finally, I estimated rates of survival, dispersal, and return rates. Arctic Terns have strong dispersal behaviour and regularly leave the major colonies of the GOM, but adult survival has not changed, despite major warming. Multiple types of tern data indicated that the weakest period of the past 30 years was between 2004-2012, when salinity was reduced and small copepods like Centropages typicus were less abundant. Arctic Terns were able to react to major declines in habitat quality, and have adjusted well to recent warming. Although Arctic Terns are affected by both top-down and bottom-up forcing, their diet and demographic data are useful as indicators. Terns could function well as ecosystem, guild-, or site-specific indicators, depending on the desired use.
Article
Full-text available
Most particle detectors are based on the hypothesis that particles are emitted randomly upon nuclear decay. In the present work, we tested the hypothesis of the existence of correlation in the random trajectories of alpha particles emitted from 241\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${}^{241}$$\end{document}Am source and the null hypothesis of random trajectories. The trajectories were clued through the registration of track in a solid-state nuclear track detector. The experimental parameters were optimized to identify the possible sources of correlation in the track registration and the detector conditions upon exposure and etching to avoid misleading results. The optimization included authentication of linearity in registration efficiency with exposure time to prevent coalescence of registered tracks. The statistical inference processes were based upon adaptive quadrates analysis of the spatial data, and entropy and divergence analysis of the quadrate data together with the null hypothesis of Poisson distribution of random trajectories. The clustering and dispersion analysis were performed with central deviation tendency, empirical K-function, radial distribution analysis, and proximity Analysis. Results showed a pattern of gained information within the registered tracks that may be attributed to the alteration in the alpha particles’ trajectories induced by the strong electric field due to atoms in the source compound and encapsulation film.
Article
Statistical model comparison has become common in historical biogeography, enabled by the R package BioGeoBEARS, which implements several models in a common framework, allowing models to be compared with standard likelihood‐based methods of statistical model comparison. Ree and Sanmartín (Journal of Biogeography, 45, 741–749, 2018) critiqued the comparison of Dispersal–Extinction–Cladogenesis (DEC) and a modification of it, DEC+J, which adds the process of jump dispersal at speciation. DEC+J provides highly significant improvements in model fit on most (although not all) datasets. They claim that the comparison is statistically invalid for a variety of reasons. I analyse the key claims made by the critique. Simulated data. Simulated data. Likelihood calculations are checked by comparison between programs and by‐hand calculations, and by summing likelihoods across all possible datasets. Model adequacy of DEC versus DEC+J is checked by a simulation/inference experiment. Mistakes in the critique's example likelihood calculations are demonstrated. DEC+J fits better on datasets because the DEC model is statistically inadequate in the common situation when most species have geographical ranges of single areas; the DEC model requires long residence times of multi‐area ranges, and when these are not observed, a model that does produce such data patterns, such as DEC+J, prevails. More fundamentally, statistical comparison of DEC and DEC+J produces identical log‐likelihood differences to statistical comparison of two submodels of ClaSSE where extinction rates are fixed to 0. DEC fails a basic model adequacy check for understandable reasons, while DEC+J does not. As Ree and Sanmartín recommend ClaSSE models as valid for comparison, the comparison of DEC and DEC+J is statistically valid according to their own criteria.
Article
The regions between the tropics provide climatic characteristics that contribute to the formation of unique vegetation typologies, the Tropical Forests. It is estimated that these forests are home to about 50 000 tree species. The Atlantic Forest, located mostly on the Brazilian coast is an example of the richness of phanerophytes inside an important hotspot for biodiversity. Understanding the relationship between these edaphoclimatic factors and the richness of tree species in the Atlantic Forest is important to establish conservation strategies in this threatened biome. Our main objective was to analyse the relationship between abiotic variables and tree species richness in the southern region of the state of Bahia, Brazil. In this sense, we used data from tree species and abiotic factors (extracted from the WorldClim and FAO platforms) in a selection of mixed generalized linear multimodels. Variables related to temperature and precipitation in the driest periods influenced the richness of tree species in the study region. Whereas temperature seasonality and minimum temperature of the coldest month influenced negatively the values for extrapolated richness the maximum temperature of the hottest month exhibited a positive influence for richness indexes. These results emphasize that the maintenance of the arboreal richness in tropical zones is directly dependent of the climatic seasonality. In the present study, we point out the importance of climatic factors for the maintenance of tree species in a region that has expressive numbers of tree species per hectare, one of the world records in this index.
ResearchGate has not been able to resolve any references for this publication.