Weighted averaging partial least squares regression (WA-PLS): an improved method for reconstructing environmental variables from species assemblages

Weighted averaging regression and calibration form a simple, yet powerful method for reconstructing environmental variables from species assemblages. Based on the concepts of niche-space partitioning and ecological optima of species (indicator values), it performs well with noisy, species-rich data that cover a long ecological gradient (>3 SD units). Partial least squares regression is a linear method for multivariate calibration that is popular in chemometrics as a robust alternative to principal component regression. It successively selects linear components so as to maximize predictive power. In this paper the ideas of the two methods are combined. It is shown that the weighted averaging method is a form of partial least squares regression applied to transformed data that uses the first PLS-component only. The new combined method, ast squares, consists of using further components, namely as many as are useful in terms of predictive power. The further components utilize the residual structure in the species data to improve the species parameters (optima) in the final weighted averaging predictor. Simulations show that the new method can give 70% reduction in prediction error in data sets with low noise, but only a small reduction in noisy data sets. In three real data sets of diatom assemblages collected for the reconstruction of acidity and salinity, the reduction in prediction error was zero, 19% and 32%.     

Matching diatom assemblages in lake sediment cores and modern surface sediment samples: the implications for lake conservation and restoration with special reference to acidified systems

Restoration goals for damaged freshwater habitats can bedefinedaccording to ecological as well as to chemical criteria. Fordisturbed lakes, the sediment microfossil record can be usedtoselect potential modern analogue sites as possible restorationtarget ecosystems.Fossil diatom assemblages in two acidified lakes (Round LochofGlenhead and Loch Dee) in Galloway, Scotland, were comparedfloristically with modern surface sediment samples from ca. 200lakes in Britain, Ireland, Sweden and Norway using numericaltechniques. Mean squared Chi-squared dissimilarity (SCD)valuesbased on between sample Chi-square distance measures were usedtocompare samples.Space-for-time substitution using diatom assemblage matchingtechniques identified several modern analogue sites withHebrideanLoch Teanga and Irish Lough Claggan possessing modern diatomflorasmost similar to those which existed at the Round Loch ofGlenheadand Loch Dee before acidification.From the point of view of atmospheric pollution, the mostcloselymatching modern analogue sites were not necessarily in themostpristine regions. Some analogues occurred in UK regions ofmoderateor low acid deposition and modern diatom assemblages inatmospherically cleaner mid Norway were generally less similarfloristically.It is argued that identification of modern analogue sitesraisesthe possibility of using time-space substitution of closelymatchedmodern and fossil samples to infer whole lake ecosystems.Diatoms are however poor indicators of some water chemistryvariables and the two closest matched modern analogue siteshavetoo high calcium concentrations making faunistic comparisonsquestionable.Identification of good modern analogue lakes can be improvedbyusing selection criteria, other than diatoms, to pre-selectsites.Screening inappropriate sites according to water chemistry andbasin features combined with a larger biological database ofmodernand fossil samples offers a promising way of refining theselectionprocesses.Despite necessary refinements, modern analogue matching canpotentially identify whole lake ecosystems that can serve asbiological target communities for currently disturbed sites.Beingbased on biological rather than chemical criteria, thisapproachdoes not rely on species-water chemistry transfer functions.It istherefore directly relevant to lake conservation andrestorationobjectives and offers an alternative method for reconstructing lakepalaeo-environments.     

COASTAL DIATOM–ENVIRONMENT RELATIONSHIPS FROM THE GULF OF FINLAND,BALTIC SEA1

Eutrophication of the Baltic Sea has become a serious concern in recent decades. To provide a potential means for quality assessments of coastal waters in this area, we collected a data set of 49 embayments in the Gulf of Finland, and explored the relationship between surface sediment diatom assemblages and 15 environmental variables, with special emphasis on nutrients. Total dissolved nitrogen, total phosphorus, depth, and salinity all accounted for significant and independent fractions of variation in the diatom data and explained 34% of the total variation. There were clear changes in diatom assemblage structures along the nutrient gradients. Although these changes were gradual, we could identify a number of taxa that were more abundant in a particular nutrient environment. These taxa could be used as potential indicators of the quality of coastal waters in the Baltic Sea. Diatom assemblages that were least affected by nutrient enrichment included a variety of benthic species and a relatively high species richness. Small planktonic taxa such as Cyclotella atomus Hustedt, Cyclotella meneghiniana Kützing and Thalassiosira pseudonana Hasle and Heimdal were good indicators of highly elevated nutrient concentrations (>600 lg·L^?1 total dissolved nitrogen and 60 lg·L^?1 total phosphorus) together with low species richness. The first appearance of these small planktonic taxa in regular monitoring could be used as an early warning sign for deteriorating water quality. Diatoms could be applied to water quality classification and monitoring purposes in the coastal waters of the Baltic Sea area using techniques such as weighted‐averaging regression and calibration.     

Validation of ecological status concepts in UK rivers using historic diatom samples

We assessed the feasibility of using herbarium specimens to validate reference conditions in the UK by comparing diatom community composition of river sites with both recent and historic diatom samples. The question of substrate specificity was addressed by comparing epilithon (stone-derived) and epiphyton (plant-derived) samples from a number of rivers. No significant differences were found between the Trophic Diatom Index (TDI), species richness, species diversity, and percentage of motile valves between paired diatom samples (epilithic and epiphytic) from contemporary samples. Significant differences were recorded between a number of indices derived from analysis of the historic diatom samples on plant material sampled pre-1930 compared with diatoms from stones collected post-1990 from the same river location. The TDI, mean species richness, and species diversity and percentage of motile valves and nutrient tolerant valves were all significantly greater in the contemporary samples (p ≤ 0.05). The percentage of nutrient sensitive valves was significantly lower in the contemporary samples (p ≤ 0.05).The relative abundance of Achnanthidium minutissimum and Cocconeis placentula var. lineata was significantly greater on the herbarium material compared to matched contemporary samples. Calculated values for the TDI (43 ± 3) expected at reference conditions were similar to the observed TDI values derived from herbarium material (44 ± 12) showing no significant deviation in ecological status.     

The relationship between diatoms and surface water quality in the Høylandet area of Nord-Trøndelag, Norway

Although the ecological effects of surface wateracidification are now well researched, factorscontrolling the abundance and occurrence of aquaticorganisms in unpolluted acid-sensitive systems arepoorly known. The Høylandet region in central Norwayexperiences relatively low levels of atmosphericpollution and its surface waters, although acid, arenot significantly acidified. Hence lakes and streamsin this region were selected to study the influence ofwater chemistry on diatom algae. Relationships betweenthe two were explored using the multivariate techniqueof canonical correspondence analysis (CCA). Theprincipal water chemistry variables influencingspecies composition of periphytic diatoms were foundto be pH and water colour. Furthermore, therelationship between species abundance and pH wassufficiently strong to enable reconstruction of wateracidity from diatom data. Establishing the nature ofaquatic communities in atmospherically clean butgeologically sensitive regions is an important meansof identifying control systems against which therecovery of acidified lakes in polluted regions can beassessed. The Høylandet region has the potential toprovide a Europe-wide control system of this naturebut much further work is required to follow up andextend the results of this preliminary study.     

A Tiered Analytical Approach for Investigating Poor Quality Emergency Contraceptives

Reproductive health has been deleteriously affected by poor quality medicines. Emergency contraceptive pills (ECPs) are an important birth control method that women can use after unprotected coitus for reducing the risk of pregnancy. In response to the detection of poor quality ECPs commercially available in the Peruvian market we developed a tiered multi-platform analytical strategy. In a survey to assess ECP medicine quality in Peru, 7 out of 25 different batches showed inadequate release of levonorgestrel by dissolution testing or improper amounts of active ingredient. One batch was found to contain a wrong active ingredient, with no detectable levonorgestrel. By combining ultrahigh performance liquid chromatography-ion mobility spectrometry-mass spectrometry (UHPLC-IMS-MS) and direct analysis in real time MS (DART-MS) the unknown compound was identified as the antibiotic sulfamethoxazole. Quantitation by UHPLC-triple quadrupole tandem MS (QqQ-MS/MS) indicated that the wrong ingredient was present in the ECP sample at levels which could have significant physiological effects. Further chemical characterization of the poor quality ECP samples included the identification of the excipients by 2D Diffusion-Ordered Nuclear Magnetic Resonance Spectroscopy (DOSY ¹H NMR) indicating the presence of lactose and magnesium stearate.     

Comparison of diatom community structure from epilithon and fish guts: implications for inferring past changes in water quality

The sampling skills of three common European fish species (Barbus barbus, Chondrostoma nasus and Squalius cephalus) were tested to assess their potential as biomonitors of past changes in river water quality through the analysis of diatoms in fish guts. The study was performed on three rivers with different chemical and physical conditions. Comparison of similarity indices revealed low diatom assemblage resemblance between epilithic samples and samples collected from fish guts at the same locality. In contrast, a mixed-effect linear model identified significant differences between locations through comparison of diatom-based water quality indices. Among fish species, diatom indices calculated from gut samples of B. barbus and C. nasus were in most cases not significantly different from those from epilithic samples while those from S. cephalus were often significantly different. The results of the study demonstrate that diatom analysis of fish guts provides a clear distinction between eutrophic and hypereutrophic rivers. Finer nutrient variations within hypereutrophic conditions were also found to be significant for some diatom indices. In this context, choice of appropriate diatom water quality index proved to be crucial in maximizing the sensitivity of the method.     

Invertebrate communities and hydrological variation in Cairngorm mountain streams

Macroinvertebrates, discharge and 16 chemical variables were monitored over a 14-year period in four small streams (catchment area <15 km²) in the Cairngorm mountains, Scotland. Canonical Correspondence Analysis (CCA) was used to assess relationships between invertebrates and environmental conditions on the day of sampling, average conditions over the preceding 1, 2 and 3-month periods and indices of hydrological and hydrochemical variation over preceding monthly intervals.CCA detected subtle inter-catchment differences in invertebrate community structure, with catchments separated along axes representing streamwater calcium, alkalinity and total organic carbon concentrations. Invertebrate communities varied seasonally, with spring, summer and autumn samples separated along CCA axes representing temperature, orthosilicate and discharge. Hydrochemically, spring was the most variable season, characterised by increased frequency of both high and low flow events and acid, snowmelt episodes. In two of the streams, invertebrate community structure varied more in spring than in other seasons.CCA ordinations using indices of hydrological and hydrochemical variation over preceding time periods were more successful (increased eigenvalues) at explaining temporal variation in invertebrate community structure than those using conditions on the day of sampling or average conditions over preceding time periods. For one of the catchments, 40% of the seasonal and between-year variation over the 14-year period could be explained by the frequency of high and low flow events, maximum and minimum water temperatures and acid episodes in the two months prior to the invertebrate samples being collected. The single most important flow parameter (longest CCA arrow) was the frequency of high flow events greater than three times the median discharge.No significant trends in invertebrate community composition were found in any of the streams over the 14-year period so, despite the apparent importance of hydrological and hydrochemical variation, communities appeared stable over the long-term.     

Uncertainty in ecological status assessments of lakes and rivers using diatoms

The EU’s Water Framework Directive requires all surface water bodies to be classified according to their ecological status. As biological communities show both spatial and temporal heterogeneity, expressions of ecological status will, inevitably, have an element of uncertainty associated with them. A consequence of this environmental heterogeneity is that there is a risk that status inferred from one or more samples is different to the true status of that water body. In order to quantify the scale of temporal uncertainty associated with benthic diatoms, replicate samples were collected from sites across the ecological status gradient in lakes and rivers in the UK. Variability (expressed as standard deviation of temporal replicate samples from a single site) could be described using a polynomial function and this was then used to calculate the risk of placing a water body in the wrong ecological status class. This risk varied depending on the distance from the class boundaries and the number of replicates. Based on these data, we recommend that ecological status is determined from a number of samples collected from a site over a period of time.     

The importance of dispersal related and local factors in shaping the taxonomic structure of diatom metacommunities

To date, little is known about the relative importance of dispersal related versus local factors in shaping microbial metacommunities. A common criticism regarding existing datasets is that the level of taxonomic resolution might be too coarse to reliably assess microbial community structure and study biogeographical patterns. Moreover, few studies have assessed the importance of geographic distance between habitats, which may influence metacommunity dynamics through its effect on dispersal rates. We applied variation partitioning analyses to 15 separate regional datasets on diatoms found in lakes in Eurasia, Africa and Antarctica. These analyses quantified the relative contributions of dispersal related and local factors in determining patterns of taxonomic turnover at the species and at the genus level. In general, results were similar at both taxonomic levels. Local environmental factors accounted for most of the explained variation (median=21%), whereas dispersal related factors were much less important (median of significant fractions=5.5% variation explained) and failed to significantly explain any variation, independent of the environmental variables, in the majority of the datasets. However, the amount of variation explained by dispersal related factors increased with increasing geographic distance and increasing taxonomic resolution. We extrapolated our regional scale observations to the global scale by combining the regional datasets into a global dataset comprising 1039 freshwater lakes from both hemispheres and spanning a geographic distance of over 19 000  km. At this global scale, taxonomic turnover was lowest in highly connected habitats, once environmental factors were partialled out. In common with many other studies of macro-organisms, these analyses showed that both dispersal related and local variables significantly contribute to the structure of global lacustrine diatom communities.