Thierry Onkelinx

The analysis of spatio-temporal forest changes (1775–2000) in Flanders (northern Belgium) indicates habitat-specific levels of fragmentation and area loss

IntroductionSpatio-temporal forest changes can have a progressive negative impact on the habitat of species that need forest continuity, i.e. the continuous presence of forest. Long-term species data that demonstrate such an impact are often not available. Instead we applied a spatial analysis on maps of the historical and present-day forests, by calculating landscape indices that explain forest plant species diversity.MethodsWe digitized for this purpose, forests in Flanders (northern Belgium, ~13,500 km2) at four time slices (1775, 1850, 1904–1931, 2000) and created a map of forest continuity in 2000. The ecological relevance of the analysis was further enhanced by a site classification, using a map of potential forest habitat types based on soil–vegetation relationships.ResultsOur results indicated that, between 1775 and 2000, forests occupied 9.7–12.2 % of the total study area. If continuity was not taken into consideration, forest fragmentation slightly increased since 1775. However, only 16 % of the forest area in 2000 remained continuously present at least since 1775 and is therefore called ancient forest (AF). Moreover, connectivity of forest that originated after 1775, called recent forest, was low and only 14 % was in physical contact with AF. The results were habitat-specific as forest on sites that are potentially suitable for a high number of slow-colonizing species, e.g. ancient forest plants, were affected most.ConclusionWe discuss that a GIS analysis of this kind is essential to provide statistics for forest biodiversity conservation and restoration, in landscapes with a dynamic and heterogeneous forest cover.

Seabird avoidance and attraction at an offshore wind farm in the Belgian part of the North Sea

Through before–after control-impact designed ship-based seabird surveys, seabird displacement occurring after the installation of an offshore wind farm at the Belgian Bligh Bank in 2010 was studied. Results demonstrate that northern gannet (Morus bassanus), common guillemot (Uria aalge) and razorbill (Alca torda) avoided the wind farm area, and decreased in abundance with 85, 71 and 64%, respectively. Lesser black-backed gull (Larus fuscus) and herring gull (Larus argentatus) were attracted to the wind farm, and their numbers increased by a factor 5.3 and 9.5. Other gull species too were found to frequent the turbine-built area, most notably common gull (Larus canus), black-legged kittiwake (Rissa tridactyla) and great black-backed gull (Larus marinus). The ecological incentives behind the observed attraction effects are still poorly understood, but on top of the increase in roosting possibilities it is plausible that offshore wind farms offer enhanced feeding opportunities. Importantly, attraction of seabirds to offshore wind farms implies an increased collision risk.

Application of the Ancient Forest Concept to Potential Natural Vegetation Mapping in Flanders, A Strongly Altered Landscape in Northern Belgium

Construction of potential natural vegetation (PNV) poses particular challenges in landscapes heavily altered by human activity and must be based on transparent, repeatable methods. We integrated the concept of ancient forest (AF) and ancient forest species (AFS) into a four-step procedure of PNV mapping: 1) classification of forest vegetation relevés; 2) selection of those vegetation types that can serve as PNV units, based on AF and AFS; 3) merging of selected vegetation types into five PNV units that can be predicted from a digital morphogenetic soil map; 4) mapping of three additional PNV units based on additional environmental data. The second step, concerning the selection of reference forest vegetation, is of particular interest for PNV construction in Flanders (northern Belgium), where forest cover has been subject to temporal disruption and spatial fragmentation. Among the variety of extant forest recovery states, we chose as PNV units those vegetation types for which a high proportion of relevés had been located in AF and that contained many AFS. As the frequency of AFS depends on site conditions, we only compared and selected vegetation types that are found on similar sites according to average Ellenberg indicator values. While succession is irrelevant for the definition of PNV, colonization rates of AFS can be used to estimate the time required for PNV to be restored in a site.

Working with population totals in the presence of missing data comparing imputation methods in terms of bias and precision

AbstractMissing observations in water bird censuses are commonly handled using the Underhill index or the birdSTATs tool which enables the use of TRIM under the hood. Multiple imputation is a standard technique for handling missing data that is rarely used in the field of ecology, but is a well known statistical technique in the fields of medical and social sciences. The purpose of this paper is to compare these three methods in terms of bias and variance. The bias in the Underhill method depends on the algorithm and starting values. birdSTATs and multiple imputation are unbiased in the case of missing values that are missing completely at random; more missing values implies less information, and so wider confidence intervals are expected as the missingness increases. The Underhill method and birdSTATs tool underestimate the variance; omitting data from a complete dataset and applying the Underhill index or birdSTATs tool results in smaller confidence intervals. Multiple imputation with an adequate imputation model provides wider confidence intervals. Biased parameter estimates with underestimated variance can potentially lead to incorrect management and policy conclusions. Hence, we dissuade the use of Underhill indices or the birdSTATs tool to handle missing data, rather we suggest that multiple imputation is a more robust alternative, even in suboptimal conditions.ZusammenfassungGesamtbestandszahlen trotz fehlender Daten – ein Vergleich von Imputationsmethoden hinsichtlich systematischer Abweichungen und Genauigkeit Fehlende Beobachtungen bei Wasservogelzählungen werden üblicherweise gehandhabt, indem der Underhill-Index oder birdSTATs angewendet werden. Letzteres nutzt TRIM. Multiple Imputation ist eine Standardmethode für die Handhabung fehlender Daten, die in der Medizin und in den Sozialwissenschaften wohlbekannt ist, in der Ökologie jedoch kaum angewendet wird. Das Ziel dieses Artikels ist es, diese drei Methoden hinsichtlich systematischer Abweichungen und Varianz zu vergleichen. Systematische Abweichungen beim Underhill-Index hängen vom Algorithmus und von den Ausgangswerten ab. birdSTATs und multiple Imputation sind frei von systematischen Fehlern, falls Daten vollkommen zufällig fehlen (MCAR). Fehlen mehr Werte, bedeutet dies weniger Information, und folglich erwarten wir umso größere Konfidenzintervalle, je mehr Werte fehlen. Der Underhill-Index und birdSTATs unterschätzen allerdings die Varianz. Lässt man aus einem an sich kompletten Datensatz Daten aus und wendet den Underhill-Index oder birdSTATs an, werden die Konfidenzintervalle kleiner. Multiple Imputation mit einem geeigneten Imputationsmodell liefert hingegen größere Konfidenzintervalle. Systematisch abweichende Parameterschätzungen mit unterschätzter Varianz führen möglicherweise zu falschem Management und Leitlinienabschlüssen. Daher raten wir vom Gebrauch des Underhill-Index oder birdSTATs zur Handhabung fehlender Daten ab. Multiple Imputation ist selbst unter suboptimalen Bedingungen eine robustere Alternative.

Effects of anthelmintic treatment and feed supplementation on parasite infections and morbidity parameters in Cambodian cattle

Helminth infections are the cause of morbidity in Cambodian cattle but other factors such as nutritional deficiencies and concurrent diseases may enhance the effects of parasites. The present study aimed to investigate the impact of anthelmintic treatment, feed supplementation, or both on gastrointestinal strongyle (GIS) and trematode infections as well as on morbidity parameters in Cambodian village cattle. At the beginning of the dry season, cattle populations in six villages were randomly assigned to a group: (A) receiving anthelmintic treatment (ivermectin+clorsulon) at week 0; (P) feed pellet supplementation during week 0-13 or both (AP). On five visits (week 0-29), faecal and blood samples were obtained for parasitological examination and haematocrit determination, respectively. Body condition (BCS), hind quarter fouling (HQFS), diarrhoea (DS), and conjunctiva colour (FAMACHA©) were scored and heart girth circumference was determined. To investigate the impact of treatment over time (week 0-29), a mixed model was used with treatment, time, and their interaction as fixed effects, and animal and village as random factors. At baseline, the proportion of GIS positive animals was high (67.9%), whereas trematode infections were low (Paramphistomum: 8.8%; Fasciola: 2.6%). Very thin to emaciated cattle (BCS 1-2) were more prevalent (11.4%) and FAMACHA© scores of ≤3 or below (65.8%) less prevalent than in an earlier study in the region. A Time ⨯ Treatment interaction was present for faecal egg counts (FEC) of GIS, GIS prevalence (both p<0.0001), PCV (p=0.0034), DS (p=0.0086) and HQFS (p=0.0241). For GIS FEC, treatment groups differed at a specific time point, with levels of treatment group P being higher than in A at week 6 (p=0.0054). For Paramphistomum prevalence as well as FAMACHA© scoring, heart girth and BCS, the interaction between treatment and time was not significant, yet, time in itself had a significant impact on all (p<0.0001). The beneficial effects of protein supplementation were unclear from the current study.

Reply to the comment on 'Working with population totals in the presence of missing data comparing imputation methods in terms of bias and precision' by Bogaart et al.

Bogaart et al. (2017) indicate in their comment that Onkelinx et al. (2017b) misinterpreted some aspects of the Underhill index (UIndex), Species Trends Analysis Tool for birds (birdSTATs) and TRends and Indices for Monitoring data (TRIM), and, as a consequence, do not sufficiently acknowledge the quality of those methods. We agree that some operational choices and underlying assumptions were not fully clear to us. However, if the documentation was incomplete and/or if variants existed, we made choices and filled in gaps, always in favour of Underhill and TRIM, to test our approach as thoroughly as possible and to guarantee a balanced comparison. In fact, in our paper we acknowledge that under certain circumstances, Underhill and TRIM can work properly (see the results and discussion of our paper), but our main point is that multiple imputation covers a broader range of situations and assumptions, and hence it is more flexible and robust. In practice, we cannot always be sure that the assumptions of Underhill or TRIM are valid. In these circumstances, a method that proves to be more robust is preferable. For instance, Bogaart et al. (2017) mention in their fifth point that it is still an open question within the eco-statistical community whether a negative binomial or a quasi-Poisson distribution is more appropriate. However, with multiple imputation, you can make your own choice according to the context or theoretical insight. With an analytical approach, for a new model, the source code needs to be adapted. In addition, in our paper, we caution that an appropriate model must be carefully selected, and we demonstrate what happens with a less appropriate model. In the following, we reply in greater depth to the statements in Bogaart et al. (2017). Each number corresponds to their numbering.