Lead contamination in raptors in Europe: A systematic review and meta-analysis ☆ Science of the Total Environment

Lead contamination is a widely recognised conservation problem for raptors worldwide. There are a number of studies in individual raptor species but those data have not been systematically evaluated to understand raptor-wide lead exposure and effects at a pan-European scale. To critically assess the extent of this problem, we performed a systematic review compiling all published data on lead in raptors (1983 – 2019) and, through a meta-analysis, determined if there was evidence for differences in exposure across feeding traits, geographical regions, between hunting and non-hunting periods, and changes over time. We also reviewed the impact of leadonraptorsandthelikelymainsourceofexposure.Weexamined114studiesthatwereunevenlydistributed in terms of time of publication and the countries in which studies were performed. Peer-reviewed articles re-porteddata for 39 raptor species but very few species were widely monitored across Europe. Obligate (vultures) and facultative scavengers (golden eagle,common buzzard and white-tailed sea eagle) accumulated the highest lead concentrations in tissues and generally were the species most at risk of lead poisoning. We found no evi- dence of a spatial or decadal trend in lead residues, but we demonstrated that high lead blood levels relate to hunting season. Exposure at levels associated with both subclinical and lethal effects is common and lead from ri ﬂ e bullets and shot is often the likely source of exposure. Overall, our review illustrates the high incidence and ubiquity of lead contamination in raptors in Europe. However, we did not ﬁ nd studies that related exposure to quantitative impacts on European raptor populations nor detailed studies on the impact of mitigation measures. Such information is urgently needed and requires a more harmonised approach to quantifying lead con- tamination and effects in raptors across Europe.


Introduction
Lead is one of the most toxic heavy metals affecting all living organisms, including humans and wildlife. It is widely distributed and persistent in the environment causing problems worldwide (Burger, 1995). Birds are one of the taxa most affected by lead and exposure and effects have been extensively studied and documented over the last half century (Benson et al., 1974;Hernández et al., 1988;Pain et al., 1993aPain et al., , 2009Pain et al., 1995;Fisher et al., 2006). Trends in the exposure to and effects of lead contamination in birds have been seen to reflect those in humans (Pain et al., 2019) and so raptors can be valuable indicators of environmental pollution as well as warning system of potential hazards to human health. Raptors are long-lived apex predators which accumulate contaminants, are distributed across large geographical areas and are relatively easy to sample in order to obtain nondestructive materials. They can, therefore, be used to track spatiotemporal trends of pollutants as well as to identify adverse effects (Furness, 1993;Gómez-Ramírez et al., 2014).
In general, exposure to lead can result from numerous sources such as lead-based gasoline, fishing sinkers, mining activities and industry. However, lead from ammunition used by hunters has been described as the most important source affecting birds and raptors in particular (Krone, 2018). Raptors ingest lead in their food in the form of gunshot and bullet fragments that are present in the viscera of prey and scavenge (including ingested shot) or embedded in tissues. Although raptors can eliminate lead via regurgitation of pellets, their digestive process facilitates rapid dissolution and absorption of lead into the bloodstream. It is transported around the body, reaching all organs and tissues including the liver and kidneys, bones and growing feathers (Pain et al., 2019). While bones retain elevated lead levels for long periods (years), thereby providing a measure of lifetime exposure, the half-life of lead is shorter in soft tissues (weeks to months) and blood (around two weeks) (Pain, 1996). Depending on the levels reached in the different organs, lead can cause effects that range from subclinical to lethality. Lead affects the vascular, nervous, renal, immune and reproductive systems, haematological parameters, and also impacts behaviour and survival (Eisler, 1988;De Francisco et al., 2003;Franson and Pain, 2011). When exposure results in acute toxicity, birds may die suddenly yet appear to be in good physical conditions (Krone, 2018).
Given the available evidence of the environmental impacts of lead, several countries have implemented national regulations to ban lead in ammunition. In Europe, the Convention on the Conservation of European Wildlife and Natural Habitats was the first to phase out lead shot ammunition in 1991 (Bern Convention). Subsequently, various International Agreements, Resolutions and Guidelines were adopted by the European Union (EU) and associated countries such as Norway and Switzerland. These led Member States to develop and implement their own regulations on the use of lead in rifle and shotgun ammunition (reviewed in Mateo and Kanstrup, 2019) and there is no harmonised legislation across the EU (ECHA, 2018; Mateo and Kanstrup, 2019). As a result, the European Chemicals Agency called for the need for EU-wide action to address the environmental risk of lead across all Member States, the aim of which would be to protect not only wildlife but also human health (ECHA, 2017).
Recent publications have suggested that lead is an important conservation problem for raptors (Pain et al., 2019;Plaza and Lambertucci, 2019). However, while there are a large number of studies on lead contamination in individual raptor species, typically in single countries, there has been no systematic evaluation of the available data on exposure and effects in raptors of lead at a pan-European scale. The aim of this systematic review was to compile all published data reporting lead contamination in European raptors and, through a meta-analysis, analyse spatio-temporal trends in contamination and explore how lead concentrations differ between species with different feeding traits and between hunting and non-hunting seasons. We also aimed to examine the scale of subclinical and lethal effects and evaluate the contribution of rifle and shotgun ammunition as a source of lead exposure.

Material and methods
This systematic review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Moher et al., 2009).

Data sources and search strategy
We examined papers from the first published on the subject (1983) up to February 2019. Four different search engines were used: (1) Web of science (www.webofscience.com); (2) PubMed (www.ncbi.nlm.nih. gov/pubmed); (3) ScienceDirect (www.sciencedirect.com); and (4) Scopus (www.scopus.com). First, we performed two general searches without restriction on the year of publication and including all titles, abstracts and keywords. We used the following terms: (1) Lead OR plumbism OR Pb AND raptor* OR "bird* of prey" OR falcon* OR accipitr* OR strigiformes; (2) Lead OR plumbism OR Pb AND raptor* OR "bird* of prey" OR vulture* OR hawk* OR owl* OR eagle*. We then made additional searches using the words ammunition, heavy metals and raptors. We also examined the reference list of the main reviews on heavy metals in raptors to identify additional papers that had been missed.

Criteria for selection and eligibility of data
The study design and inclusion/exclusion criteria are defined in SI Appendix A1. The eligibility criteria are illustrated in the PRISMA flow diagram (SI Fig. A1).
The search returned more than 10,000 publications, from which we removed irrelevant (for example where "lead" was used as a verb) and non-English papers, duplicates and grey literature. Book chapters, technical reports and conference proceedings were also excluded (SI Fig. A1). This left 293 publications and we screened their title and abstracts, removing 179 because they covered non-European or nonraptor species (SI Fig. A1). We fully reviewed the remaining 114 papers, including 10 reviews and 1 modelling study. For our qualitative analysis, we included all 114 papers as our objective was to evaluate trends in the publications; the reviews and modelling studies were only included in this first part of the analysis (Section 2.3). We then conducted a quantitative meta-analysis and only included publications containing data that reported raw or statistical summary data on lead concentrations in raptors (n = 46 publications; see SI Excel file). We excluded publications in which: lead concentrations were only recorded per pool of species; lead contamination was reported only as pellets or bullets detected by X-ray; results were only expressed in graphical form; data only consisted of non-detected values; mean and standard deviation was not clearly shown or could not be estimated (see Section 2.4.2). We determined whether any papers overlapped in the data presented and found two such publications. We excluded one of these from the meta-analysis (SI Excel file) but included it in the qualitative analysis as it provided relevant new information. Finally, we also checked for papers that reported potential data outliers that would excessively influence our meta-analysis and excluded them from the analyses (see Section 2.5).

Data extraction
Extraction of data from the final set of studies (n = 114) was conducted by one reviewer (LM) using standardised forms (see SI Excel file). Reviews and modelling papers were only considered to gather general information and were excluded from the rest of the analyses. Key information was categorized as: (1) general information: year of publication and country (including 114 publications in the analysis); (2) individual information: species, source of animals (i.e., freeranging individuals, individuals admitted to wildlife rehabilitation centres, museum specimens, captive birds) (103 publications); (3) sample information: type of sample analysed for lead (103 publications); (4) lead information: concentrations reported (46 publications); (5) source of contamination: presumed or confirmed lead source and use of isotopes (103 publications); (6) health effects of lead (103 publications).

Description of variables
For the first part of the study (qualitative analysis), we analysed trends in publication, and when examining temporal trends, we used the year of publication of the paper. Other factors that we considered were country, species, types of matrix, lead levels above threshold concentrations, occurrence of subclinical health effects and mortality, and source of lead.
For the meta-analysis, we defined year as the year the study was undertaken (not published) and we selected the mid-year of the timespan in long-term studies. We then pooled studies into four decades: 1983-1989; 1990-1999; 2000-2009 and 2010-2019. With regards geographical region, we pooled countries into four European regions: West (The Netherlands, Switzerland, Germany, Austria, Belgium, England and France), North (Norway, Finland and Sweden), South (Spain, Portugal and Italy) and East (Hungary, Czech Republic and Poland). For the meta-analysis, we pooled species into two feeding trait groups: scavengers (including obligate and facultative) and nonscavengers (including the rest of birds of prey and owls) (see species classification in SI Table A1).
Nearly 60% of the studies we reviewed involved the analysis of lead in free-ranging birds trapped in the wild, while the rest were from animals sampled in captivity (30% wildlife rehabilitation centres, <7% zoos, <7% veterinary clinics and <5% museums). Half of the individuals analysed were found dead in the field or were euthanized for medical reasons. The other half was sampled alive. To avoid possible bias associated with individual origin, we checked each study for where and when individuals were sampled. In captivity (including wildlife rehabilitation centres, zoos and veterinary clinics), almost all individuals were sampled either on arrival or within the first weeks of admittance to the centre. Individuals maintained in captivity for long periods before being sampled were removed from the analysis to avoid bias. We also screened statistically to determine whether provenance and the time of sampling of the bird (i.e., on arrival, within the first weeks, time of screening unknown) affected lead levels. As we did not find any significant effect (all cases P > 0.05) we therefore removed this parameter from further analysis.

Lead calculations
We extracted the published data on lead concentrations by species and by study (SI Table B1). To compare lead concentrations between studies, we converted all tissue concentrations so that they were expressed as mg/kg dry weight (dw). Following Krone (2018), we used the following correction factors: 1 μg g −1 ww = 4.6 μg g −1 dw for blood, 3.1 μg g −1 dw for liver, 4.3 μg g −1 dw for kidney and 1.2 μg g −1 dw for bone.
Lead concentrations were included in statistical models when the mean and variance (standard deviation; sd) from each study were given or could be derived. For studies that did not report the mean and sd but instead provided the median, range (minimum-maximum) and sample size, we estimated the mean and sd using the formulas described by Hozo et al. (2005). Studies not reporting the mean and sd, nor the median, range and sample size, were excluded from the metaanalysis (see SI Excel file).

Meta-analysis and statistics
We first ran forest plots to visualise how mean lead concentrations and confidence intervals (CIs) varied between the species covered in each study. The lead concentration data for individual species were not homogeneously distributed across decades and/or geographic region. Therefore, for the purposes of statistical evaluation of the data in the meta-analysis, we pooled species by feeding trait (see Section 2.4.1).
We used Metafor (Viechtbauer, 2019), a R package (R version 3.5.3, R Core Team, 2015), to estimate the differences in mean lead concentrations across feeding traits, decades and geographical regions. When examining lead concentrations in blood, we also explored differences between sampling seasons (4 levels: hunting, non-hunting, unknown and year-round). Lead concentrations were included as the response variable and feeding trait, decade, geographical region and sampling season (model for blood only) as explanatory variables; geographic region was not included in the model for bone lead because almost all the studies used were on raptors from southern Europe. We used multilevel random-effects models with REML (Restricted Maximum Likelihood) to account for within study variance caused by change and sampling error, and for between study variance caused by heterogeneity (Viechtbauer, 2010). We conducted four separate meta-analyses with the exposure measures being lead concentrations in: (i) liver, (ii) kidney, (iii) blood and (iv) bone.
The significance of the explanatory variables included in the models was assessed using the Akaike's Information Criterion (AIC) (Burnham and Anderson, 2002). We first included all variables in the models but later removed non-significant variables in a backward stepwise procedure. We ranked models using AIC and Akaike weight (w) (likelihood that a given model is the best among all candidate models). We selected the model with the greatest w and lowest AIC value as the model that best fitted the data without overparameterization (SI Table A2).
We computed the heterogeneity statistic I 2 which estimates (in percent) how much of the observed variation can be attributed to the actual difference between studies, rather than within-study variance (Higgins et al., 2003). We also computed the R 2 statistic, which is the amount (in percent) of the variation that is accounted for by the variables included in the model (Viechtbauer, 2010). The models for all four models (liver, kidney, blood and bone) had high heterogeneity (I 2~9 9%) indicating that most of the variation in lead levels was due to differences between rather than within individual studies (Higgins et al., 2003). The models explained circa 30% (R 2 ) of the variation in lead concentrations in the case of in liver, kidney and blood, and 15% in the case of bone.
To identify studies that were potential outliers and contributed excessively to model heterogeneity, we applied the Baujat plot and a set of diagnostic tools (Quintana, 2015;Viechtbauer and Cheung, 2010).
The Baujat plot horizontal axis illustrates study heterogeneity and the vertical axis the influence of the study. Studies falling on the top right quadrant of the plot were rejected (see SI Excel file), as these exerted an excessive contribution to both factors. To investigate whether the studies included in our meta-analysis were a representative sample of all studies conducted on lead contamination in raptors, we checked for publication bias using the funnel and Egger's test (Egger et al., 1997). We found evidence of publication bias (Egger's test P < 0.001) for liver, kidney and blood (SI Figs. A2-A4). Publication bias dictates that studies with non-significant results are less likely to be published (Quintana, 2015). However, as our measure of interest was the concentrations of lead accumulated by raptors and our main aim was to study the variables that explained variation in residue magnitude, we did not expect publication bias to be a significant matter of concern.
Finally, Spearman's rank correlation tests were used to calculate correlations between lead concentrations in liver, kidney, blood and bone.

Overview of lead publications in Europe
Of the 114 papers describing lead contamination in raptors in Europe, almost half (46%) were published between 2010 and 2019, while~34% were published between 2000 and 2009,~12% from 1990 to 1999 and < 8% from 1983 to 1989 (SI Fig. A5).
The number of studies and species monitored per country are shown in Table 1. The studies included information of 39 species consisting on 29 birds of prey (4 obligate scavengers, 5 facultative scavengers and 20 non-scavengers) and 10 owls. Among the birds of prey, the common buzzard (Buteo buteo) (23 papers), the white-tailed sea eagle (Haliaeetus albicilla) (20 papers) and the griffon vulture (Gyps fulvus) (15 papers) were the most studied species. The most studied owls were the barn owl (Tyto alba) and tawny owl (Strix aluco) (12 papers each), then the Eurasian eagle owl (Bubo bubo) and the little owl (Athene noctua) (11 papers each) (SI Table A1). The predominant species as well as the number of species studied often varied between countries (Table 1; SI Table A1). The species in which lead has been monitored most widely across Europe are the common buzzard and the Eurasian sparrowhawk (Accipiter nisus) (8 countries each), then the barn owl (7 countries) (SI Fig. A6).

Methods of lead analyses
Studies analysing lead concentrations in birds used a variety of different matrices and some studies analysed multiple matrices. The most commonly analysed matrix was the liver (n = 53 studies), followed by blood (n = 36), kidney (n = 34), bone (n = 26); other organs and sample types included lung, stomach, brain, intestine and heart (n = 34), feathers (n = 24), pellets (n = 9) and eggs (n = 7). The extent to which each matrix has been used has varied widely over time. The most marked change in the last decade has been the increase (~55%) in analysis of non-destructive samples, in particular blood and feathers, although liver has been the most extensively analysed matrix overall (SI Fig. A7).
There was a significant positive correlation between lead concentrations in the liver and kidney across all species combined (Spearman's correlation coefficient: r = 0.66, P < 0.001, n = 25; SI Fig. A8). There were no other significant correlations in lead concentrations between matrices (SI Table A3).

Lead concentrations in samples
All data on lead concentrations in different matrices and species are given in SI Table B1.
The griffon vulture, an obligate scavenger, had the highest mean lead concentrations in liver and kidney (Figs. 1-2).
Facultative scavengers (common buzzard, golden and white-tailed sea eagle) had the next highest concentrations. Interestingly a single study showed that the marsh harrier (Circus aeruginosus) had the highest lead blood levels ( Fig. 3) but the next recorded highest concentrations were in the griffon vulture.
The golden eagle (Aquila chrysaetos), the Eurasian eagle owl and the common buzzard were the species with the highest levels of bone lead (Fig. 4).
The most parsimonious model for liver lead included feeding trait, geographic region and decade (QM 7 = 33.07, P < 0.0001). The model for kidney lead also included feeding trait and decade but not geographic region (QM 3 = 12.83, P < 0.01), while the best model for blood lead included feeding trait and sampling season (QM 4 = 18.55, P < 0.001) (SI Table A2). Lead concentrations in all three matrices were higher in scavengers than non-scavengers (P < 0.01; Fig. 5; SI Tables A4-A6). Blood lead concentrations were higher in birds sampled during the hunting season than in those sampled in the non-hunting season, year-round or at an unknown time (P < 0.01 all, SI Table A7). Although geographic region was retained as a variable in best model for liver lead, post-hoc comparisons did not indicate any significant difference between individual regions (SI Fig. A9, Table A8). There was no overall consistent temporal trend across the decades in mean lead  Table 1 Overview of papers published per country, raptor groups monitored and most studied raptor species. Asterisk shows countries that have equal number of studies for more than two species. Species classification can be found in SI Table A1.   Table A9). The best model explaining lead variation in bone differed from those for the other matrices in that it was the null model closely followed by the one including feeding trait, geographic region and decade (ΔAIC < 2) (SI Table A2). We did not continue further with any statistical analysis because the model was non-significant and explained little of the variation in bone lead (QM 5 = 8.2, P > 0.05). However, we included bone concentrations in Figs. 4 and 5 and SI Figs. A5 and A6 to enable visualisation of the distribution of data.

Subclinical effects
In total, there were 226 species-specific datasets in the 114 papers reviewed. Just over half (51.8%) reported lead concentrations that exceeded subclinical threshold values (defined in Section 2.4.1) in at least some individual birds (SI Table B1), although in some cases, these included reports for multiple sample matrices (i.e., blood, liver, kidney and bone) from the same individuals. The species with the highest prevalence of exceedances (but excluding single case study reports) were the bearded vulture (Gypaetus barbatus) (100% of n = 3 datasets), the griffon vulture (92% of n = 13), the red kite (Milvus milvus) (75% of n = 8), the common buzzard (75% of n = 20) and the white-tailed sea eagle (67% of n = 18) (SI Table B1). The main source and route of exposure was ingestion of lead shot or ammunition (31 studies) although this was only presumed (with no confirmation) in most cases. Another fourteen studies found lead shot (9) or lead fragments (5) in the gastrointestinal tract of birds while four reported lead in regurgitated pellets. Nine studies undertook isotopic analyses and ten other studies found embedded shot in muscles of birds, suggesting a non-ingestion source of contamination but one that was still associated with hunting (SI Table B1). Mining and urban/industrial pollution were the other most frequently cited (but typically presumed) sources of lead contamination (12 studies).
A number of studies have examined associations between lead concentrations and subclinical effects. We found seven studies that determined if there was evidence that lead contamination could affect a range of effects on biomarkers in raptors. These included reports of adverse effects on oxidative stress, enzymatic activity, DNA damage and blood biochemistry ( Table 2). The potential effect of lead on reproductive success in raptors has also been examined and the occurrence and nature of any effects has varied between studies (Table 2).

Lethality
Thirty-six (36) studies, encompassing 14 species, reported lead residues that exceeded the concentration thresholds for lethal lead contamination and lead-induced mortalities (Table 3). Most were from the last two decades (2000-09: 47.2% of studies; 2010-19: 33.3%) with only five (11%) conducted between 1990 and 1999 and three (8%) between 1983 and 1989 ( Table 4). The proportion of studies on each species that reported exceedance of lethal concentration thresholds or mortality was calculated. The highest proportions were for bearded vulture, red kite, white-tailed sea eagle, golden eagle and common buzzard (Table 3). We also calculated the percentage of individuals in each of those studies that had lead concentrations above the lethal threshold level (Table 4). This was 2% and 40% for the bearded vulture (two studies only) while the median (range) was 12% (2%-24%) for the red kites (five studies), 26% (9%-100%) for the white-tailed sea eagle (nine studies), 24% (7%-43%) for the golden eagle (five studies) and 4.5% (1%-100%) for the common buzzard (eight studies). Those studies that reported the presence of lead in the gastrointestinal tract are also detailed in Table 4.

Overview of lead publications in Europe
Our review of 36 years of research shows that 16 of 44 European countries have reported data on lead concentrations in raptors. The highest number of publications is from Spain. This perhaps reflects the fact that Spain's geographical position and geophysical diversity provides a matrix of suitable resources and habitats for raptors, of which there are some 26 bird of prey and 6 owl species; Spain supports a good proportion of the wintering populations of many species (García-Fernández et al., 2008). In addition, hunting is a widespread activity in Spain and constitutes an important conservation problem (Moleón et al., 2011;Mateo et al., 2013). Although there have been fewer (7) published studies reporting data from France, they contain data on lead concentrations in more raptor species (24) than reported by any other European country (Table 1). Monitoring of lead in raptors is relatively sparse in northern Europe (data for 6 bird of prey and 2 owl species) while only Poland in eastern Europe has monitored lead in a large number of species (12 birds of prey and 5 owl species). These results indicate that there is a major geographical bias in the reporting of lead contamination in raptors and this effectively hampers assessment of temporal and spatial trends in lead contamination at a pan-European scale. We also note that almost half of the information available about lead contamination in European raptors has been published in the last two decades. This publications trend has also been reported in other avian studies (e.g., Plaza and Lambertucci, 2019).

Raptor species used to monitor lead concentrations
The common buzzard, the griffon vulture and the white-tailed sea eagle are the species that have been most frequently monitored for lead contamination (SI Table A1) but such monitoring has not been geographically widespread across Europe, except perhaps for the common buzzard (data from 8 countries; SI Fig. A6). A previous assessment of the state of contaminant monitoring using raptors noted the need for a harmonised (species and matrices analysed) sampling strategy (Gómez-Ramírez et al., 2014;Espín et al., 2016) if spatial and temporal trends were to be detected at a pan-European scale. It has recently been argued that this could be achieved for lead through monitoring of the common buzzard or by monitoring the golden and white-tailed sea eagle in combination (Badry et al., 2020). The current geographical spread of available data on lead contamination is similar for common buzzard and for the golden/white-tailed sea eagle combination but both still only encompass some 20% of European countries and less than 30% of EU Member States. Given regulation of lead and other contaminants is centralised across the EU through such bodies as ECHA, there is clear need for wider-scale harmonised monitoring of lead to understand how environmental concentrations are changing and the associated impacts on raptors across Europe.

Use of matrices
Liver has been most extensively used for monitoring exposure to lead in recent decades, followed by blood, kidney and bone. Liver and kidney are useful proxies for medium-term exposure and bone constitutes a long-term depot for lead and therefore is used as a proxy for lifetime exposure (Pain et al., 2005;Rodriguez-Ramos Fernandez et al., 2011). Such analyses provide post-mortem confirmation of exposure and exceedance of threshold concentrations, but they can only be used in passive monitoring. Such monitoring comprised 50% of the studies we reviewed.
Feathers and blood can be sampled from live birds as part of active monitoring programmes. In the last 10 years, feathers have been increasingly analysed as a non-(or minimally-) invasive matrix and the number of such studies is likely to grow further. This is perhaps because feathers can be sampled opportunistically (during other sampling activities or ringing) and without disturbing the bird as shed feathers can be collected from the vicinity of the nests. Feathers can also be taken from dead birds and so the same sample type can be collected across both active and passive monitoring activities. Feathers act as an archive of exposure during the period of feather growth and thus reflect chronic levels of contamination (Burger, 1993). In the case of lead, exposure may be intermittent (e.g., when lead shot is ingested) and feather segments may be used to quantify variability in exposure (Rodriguez-Ramos Fernandez et al., 2011;Ganz et al., 2018b). However, feathers are moulted and re-grown during the summer months when, compared to winter, fewer game animals are shot. Thus, seasonal exposure peaks may be missed by feather analyses. Furthermore, there remains some uncertainty as to how lead is distributed between different feather sections and the extent to which external contamination may interfere with the analytical determination of internal lead concentrations (Cardiel et al., 2011;Ganz et al., 2018b). This may explain why blood remains the currently most widely used sample taken during active sampling and is frequently used for diagnosing poisoning in living birds.
Although eggs have been long used for monitoring several other types of contaminants such as persistent organic pollutants, perfluorinated compounds and mercury (Espín et al., 2016), very few studies have used eggs to monitor lead concentrations. This is because maternal transfer of lead to eggs is low (Walsh, 1990;Furness, 1993). Similarly, regurgitated pellets have been rarely used in monitoring studies even though they provide spatio-temporal information about both lead ingestion and overall diet (Mateo et al., 1999). Bullet fragments and shotgun pellets in regurgitated castings can be used as indicators of the ingestion of ammunition fragments by birds and can be used to non-invasively monitor the compliance of regulations with regards the use of lead-free ammunition.

Correlation in lead concentrations between matrices
We examined the available data from the publications we reviewed to determine if lead concentrations in different matrices were correlated with each other across multiple species and so whether concentrations in one sample type could be used to predict concentrations in another. Quantification of such relationships may facilitate future capability to compare data between studies that have measured residues in different sample types. We found there was a significant correlation between liver and kidney lead concentrations across species and that concentrations tended to be higher in the kidney. Our findings are consistent with those of other studies on individual species that have reported a similar association and higher levels in kidney than in liver (Esselink et al., 1995;Ek et al., 2004;Helander et al., 2009). The transfer Table 3 Number and proportion (%) of studies reporting lead residues that exceeded the concentration thresholds for lethal lead contamination and lead-induced mortalities.

Species
Total studies  of lead into bone is slower than that to the soft tissues and bone lead is thought to reflect long-term chronic accumulation (Fisher et al., 2006). While some studies have reported positive correlations between bone and liver or bone and kidney lead concentrations (Esselink et al., 1995;Ganz et al., 2018a;Ishii et al., 2018), we found no such relationships nor any statistically significant association between levels in blood and in soft tissues or bone. This likely reflects the differences in the pharmacokinetics of lead between different body compartments. However, our results suggest that either the liver or kidney may be a suitable sample type for monitoring medium term exposure to lead across different raptor species. With additional data collection, it would be possible to generate read-across values between the two sample types, enabling comparison between studies reporting only liver or kidney lead concentrations.

Lead concentrations in raptors
4.4.1. Feeding traits associated with high exposure to lead It has been argued that scavengers are at greater risk than active hunters of ingesting lead particles from spent ammunition as they feed on the unretrieved carcasses of hunted animals (Krone, 2018). Our findings are consistent with this concept insomuch that, within the studies we reviewed, obligate and facultative scavenger species generally had the highest lead concentrations in liver, kidney, blood and bone.
Of the obligate scavengers, the griffon vulture, which is the most abundant vulture in south Europe (BirdLife International, 2017), has been the only species in which lead contamination has been studied extensively (Table 3). The highest reported average liver, kidney and blood lead concentrations in the studies we reviewed were reported in this species, although we did not find extensive reporting of individuals exceeding lethal threshold concentrations. Other obligate scavenging birds of prey in Europe include the bearded vulture, Egyptian vulture (Neophron percnopterus) and cinereous vulture (Aegypius monachus), but they have been less well studied (Donázar et al., 2002;Hernández and Margalida, 2008;Gangoso et al., 2009;Hernández and Margalida, 2009;Berny et al., 2015;Ganz et al., 2018a;Pikula et al., 2013). However, Berny et al. (2015) found high lead liver concentrations in 8 bearded vultures and suggested lead may be a conservation problem for this scavenger because its highly acidic gastric juices, designed to facilitate digestion of bone, also enhances absorption of lead. Hernández and Margalida (2009) reported lead concentrations above lethal threshold levels in individual bearded vultures, albeit a small number (Table 4), and showed that lead concentrations were higher in individuals during the hunting than the non-hunting season. Lead concentrations in Egyptian vultures were also found to peak during the hunting season (Gangoso et al., 2009), suggesting that lead from hunting ammunition is a likely important contaminant source in both species.
Facultative scavenging species are also exposed to and accumulate relatively high lead concentrations. This is particularly true for the golden eagle, common buzzard and white-tailed sea eagle. All three are known to scavenge carcasses, including those of unretrieved shot animals, but the extent to which they do this may have been underestimated (Selva et al., 2005;Blázquez et al., 2009;Sánchez-Zapata et al., 2010). Five of the 11 studies on golden eagles that we reviewed were from Switzerland and bone lead concentrations in Swiss eagles were the highest of those reported in any species. The lead isotope signatures in those birds were similar to that of hunting ammunition Ganz et al., 2018a) and the authors concluded there was a high risk of lead exposure in golden eagles in the Swiss Alps. There is less information for lead contamination in golden eagles elsewhere in Europe.
Compared with the golden eagle, less is known about lead exposure in the common buzzard and we did not find any published studies that reported lead isotope signatures in this species. However, opportunistic scavenging has been presumed to be the cause of lead contamination in buzzards (Battaglia et al., 2005;Carneiro et al., 2014). Furthermore, three studies on common buzzards have reported elevated liver and kidney lead levels that were of similar magnitude to those found in (obligate scavenging) vultures (Pain and Amiard-Triquet, 1993;Naccari et al., 2009;Castro et al., 2011). Scavenging of shot animals is likewise thought to account for high levels of lead accumulation in white-tailed sea eagles and marsh harriers. In a study on white-tailed sea eagles, the lead isotope ratios differed between individuals with lethal lead concentrations and those with only background lead concentrations (Helander et al., 2009). The authors concluded that lead in individuals with lethal concentrations likely originated from lead ammunition sources, thus suggesting that scavenging was a likely route of exposure. In marsh harriers from wetlands in France (Pain et al., 1993a) and Spain (Mateo et al., 1999), blood lead concentrations were among the highest reported for any raptor. This high level of exposure was also attributed to their scavenging of carcasses or capturing injured (shot) prey as shot was found in the pellets regurgitated by the marsh harriers.
Overall, these studies demonstrate that not only obligate but also facultative scavengers are vulnerable to lead exposure and poisoning.  were considered when calculating the %; b when some of the same individuals had lethal lead levels in multiple matrices, this was taken into account indicated underneath the % value; c % of individuals with Pb levels > 20 mg/kg was not reported. 3 Considered lethal threshold. 4 Showing acute poisoning symptoms. 5 Lethal threshold concentration used in the study was 60 (not 50) mg/kg. However, it is notable that among the studies we reviewed, some reported only low lead concentrations in facultative scavenging species such as common buzzard, white-tailed sea eagle and marsh harriers (Figs. 1-4). This may reflect differences between studies in the extent of opportunistic scavenging exhibited by individuals and is likely a function of the intensity of hunting activity and subsequent availability of scavenge.

Spatio-temporal trends in lead residues
The use of lead-based ammunition for hunting has been regulated in 23 European countries, with some countries adopting total (e.g., The Netherlands and Denmark) and others partial (see Mateo and Kanstrup, 2019) bans. However, the effectiveness of such mitigations on reducing lead contamination in raptors has not been widely examined. Our review indicated that lead can still be found in high concentrations in European raptors, but concentrations vary markedly between species in different regions. This variation may reflect differences between countries in the scale and effectiveness of mitigation measures. Mateo et al. (2007) reported a decrease in the ingestion of lead shot by the Spanish imperial eagles in Doñana, an important Spanish wetland where legislation against the use of lead shot was adopted in 2001. In contrast, Helander et al. (2009) found that the proportion of white-tailed sea eagles poisoned by lead did not differ before and after a partial ban of lead shot in Sweden and suggested that this was because the ban did not cover coastal areas utilized by this species. Overall, our review found no clear evidence that lead concentrations in raptors have decreased over time across Europe. Although visually, there appeared to be some difference in lead concentrations between regions (SI Fig. A9), these differences were not statistically significant. However, data were scant for the northern and eastern regions and further studies are needed to investigate if there are, in fact, regional differences and, if so, what are the causes.

Influence of hunting season on lead concentrations
When evaluating blood lead concentrations, we found seasonal trends in risk of lead poisoning with blood lead levels higher during the hunting than the non-hunting season. Previous studies have already reported such an association (e.g., Berny et al., 2015;Carneiro et al., 2014;Gangoso et al., 2009;Mateo et al., 1999). However, our review provides the first systematic evaluation and evidence of this trend using data from 10 species. The generally high blood lead concentrations during the hunting season are probably associated with the consumption of lead-shot individuals of small or large game species and implicate spent lead ammunition as an important factor for the uptake of lead in raptors. Among the studies reviewed, almost all were from Spain and other southern countries. Thus, we were not able to evaluate how patterns for blood lead concentrations were related to hunting activities across different regions of Europe. The studies incorporated in the analysis were published in a range of years from the 1990s to the 2010s indicating that hunting is still a current important conservation problem for raptors and that future mitigation measures are necessary to avoid high lead exposure or lead poisoning. 4.5. Effects of lead 4.5.1. Threshold for subclinical exposure More than half of the studies reviewed reported lead concentrations in raptors that exceeded the subclinical threshold for harm defined by Franson and Pain (2011). This threshold has been widely used to interpret the toxicological significance of lead contamination in raptors. However, both higher [e.g., 0.3 mg/kg dw (Pain et al., 1993a(Pain et al., , 1993b] and lower [0.15 mg/kg dw (Espín et al., 2015;Gómez-Ramírez et al., 2011;Martínez-López et al., 2004); 0.10 mg/kg dw (Gómez-Ramírez et al., 2011;Espín et al., 2014aEspín et al., , 2014b] threshold levels for blood have been considered as subclinical in some studies (SI Table B1). In fact, critical thresholds may be species-specific as it has been shown that some raptor species are more sensitive to lead than others (Ecke et al., 2017;Pain et al., 2019). For instance, the griffon vulture is widely recognised as relatively tolerant to lead (García-Fernández et al., 2008;Mateo-Tomas et al., 2016) and Espín et al. (2015) showed that Eurasian eagle owls were more affected than griffon vultures with similar blood lead residues. Although 12 out of 15 studies in griffon vultures we reviewed showed some individuals exceeding subclinical thresholds (SI Table B1), it is possible that this may overestimate the likelihood of effects in this species. Other species in which a high proportion (60%) of studies indicated individuals exceeding subclinical threshold levels included the common buzzard, the red kite and the white-tailed sea eagle (SI Table B1) and these may be species particularly at risk of subclinical effects although species-specific thresholds for these species are not defined.

Subclinical and clinical effects
Exposures to lead that result in subclinical effects are of particular concern as such effects are often hard to recognise in free-living birds and their effects on populations remain unknown. In our review, we have highlighted studies that reported a relationship between lead residues in birds and biomarker responses. These studies provide evidence that antioxidant enzyme activity can be used as a biomarker of heavy metal exposure and effects in raptors (Martínez-López et al., 2004;Gómez-Ramírez et al., 2011;Espín et al., 2014aEspín et al., , 2014bEspín et al., , 2015. They demonstrated for the first time that low lead levels in blood or tissues (between "background" and subclinical concentrations) are associated with effects on the antioxidant system and that such effects may occur below the subclinical threshold concentrations suggested by Franson and Pain (2011).
Exposure to lead has also been reported to affect reproduction (sperm motility, organ development, egg hatching rate) in non-raptor species such as red-legged partridges (Alectoris rufa) and domesticated pigeons (Columba livia domestica) (Pain et al., 2019). Less is known about the potential reprotoxic effects of lead in raptors (Table 2). Although Gil-Sánchez et al. (2018) did observe an apparent direct negative relationship between high lead concentrations and the number of fledglings per breeding attempt in Bonnelli's eagle (Aquila fasciata), they could not suggest a likely mechanism of action. Pain et al. (1999) showed that lead did not affect shell thicknesses and suggested that any pathway by which lead might affect reproductive success would be through direct effects on the parents that resulted in impaired incubation or nestling care. Further studies are needed to establish how lead affects reproductive success and whether this is largely or exclusively driven by impacts on parent behaviour and condition.
In this review, we found a number of studies that related behavioural changes in raptors to chronic exposure to lead. For instance, Krone et al. (2009) recorded the long-term activity of a white-tailed sea eagle and noted a change in daily movement and activity patterns that was associated with clinical lead intoxication caused by incidental ingestion of lead fragments from a rifle bullet. Berny et al. (2015) found an association between lead concentrations and the proportion of trauma/electrocution. They found that long-term low lead levels (even below subclinical levels) could impair flight capabilities and individuals were more likely to hit obstacles. Ecke et al. (2017) found that blood lead concentrations higher than 2.5 μg/dl in golden eagles impaired flight performance in terms of decreased height and movement rate. These studies suggest that lead may well have the capacity to affect individual fitness in raptors. Such effects are rarely detected by conventional monitoring programmes examining lead contamination in raptors and the impacts of such exposure on raptor populations is likely to be underestimated.

Mortality
Lead poisoning has been identified as an important cause of death for wildfowl (Pain et al., 2019). Lead poisoning in raptors is less wellstudied but individuals of many species have been reported as dying of lead poisoning (Mateo, 2009;Pain et al., 2009Pain et al., , 2019. In our review, we included all studies that reported lead-related mortality in different raptor species. By specifying both the proportion of studies in which mortality was reported and the proportion of individuals per study that had lead residues above the lethal threshold level, we were better able to evaluate the scale of lead poisoning in European raptors (Tables 3 & 4). Mortality was reported in 14 of 39 raptor species covered in the studies that we reviewed. Our review of the mortality data (Tables 3 & 4) suggests that the bearded vulture, the red kite, the white-tailed sea eagle and the golden eagle may be the species most at risk of being poisoned. Interestingly, reports of mortalities in griffon vultures were less prevalent than for these other species (Table 3), even though griffon vultures often had the highest accumulated lead residues of all species. This is consistent with the concept that this species may be relatively tolerant of lead (Section 4.5.1), although those studies reporting the highest lead levels in griffon vultures also reported that there were mortalities.
Overall, lead continues to cause mortality in many raptor species, as determined by diagnosed clinical cases and from the exceedance of lethal threshold levels. Circa 75% of all studies reporting lead poisoning are from the last 20 years (Table 4). This increase most likely reflects a rise in awareness of lead poisoning in raptors and a resultant increase in the number of investigations performed. In addition, it is important to consider that reported mortality events may be a small proportion of the real lead-related mortality that happens in the wild.

Identifying the sources of lead contamination
Most of the articles reporting lead concentrations above the thresholds suggested that hunting ammunition was the source. In many studies, the source of lead was only presumed but such attribution is difficult to confirm. One method has been to search for ammunition fragments and shot in the gastrointestinal tract (MacDonald et al., 1983;Lumeij et al., 1985;Mateo et al., 1997;Andreotti et al., 2017). Other studies have also reported lead ammunition/shot embedded in tissues but this represents a different exposure pathway and is associated with accidental or deliberate shooting of individuals; tissue concentrations associated with embedded shot are mostly lower than those caused by ingested lead (SI Table B1; Ganz et al., 2018a;Plaza and Lambertucci, 2019; but also Berny et al., 2015). Another attribution method that is gaining popularity is the analysis of isotopic signatures. This involves comparing the lead isotope signatures of raptor tissues with those of lead ammunition. For example, in red kites, isotope signatures for body tissues were similar to those for lead shot retrieved from regurgitated pellets (Pain et al., 2007). Other studies have used isotopic signatures more indirectly. For instance, hunting ammunition was inferred as the likely cause of lead poisoning in four vulture species because the lead isotopes in the vultures were different to those in soil, including soils from mining areas (Berny et al., 2015;Madry et al., 2015). Other studies using isotopes on griffon vultures and white-tailed sea eagles distinguished between background levels of lead that were derived from natural sources and elevated levels that were assumed to have come from ammunition sources (Helander et al., 2009;Mateo-Tomas et al., 2016). The characterisation and use of isotope signals for environmental samples and hunting ammunition may be a key tool for confirming when lead ammunition/shot is the cause of lead poisoning in raptors. Reliable quantification of the number of cases of poisoning that are due to ammunition lead would enable assessment of the importance of this issue across Europe.
Besides ammunition, some studies have hypothesised that highly polluted environments may be the source of lead exposure in the raptors that they studied. However, the number of these studies is small. They have generally been focused on mining areas (Baos et al., 2006;Gómez-Ramírez et al., 2011;Espín et al., 2014bEspín et al., , 2015 or where raptors are in close proximity to industrialised (García-Fernández et al., 1997; Espín et al., 2014b) or highly-polluted (i.e., solid-waste incinerator; Blanco et al., 2003) areas.

Conclusions: key findings and recommendations
Despite an increasing amount of research on lead in raptors, there is a geographical distribution bias in publications originating from Europe. Most studies come from western and southern Europe and few are from eastern and northern countries. Furthermore, few raptor species are widely monitored for lead. As a result, we had to pool our analyses by feeding trait, geographical region and across decades. We also found that there are large-scale variation in lead concentrations within and between raptor species. However, through the meta-analysis we were able to conclude four main points: (1) scavengers, both obligate and facultative species, are more prone to lead contamination than nonscavengers including birds of prey and owls, (2) lead contamination in raptors is still widely detected across Europe despite partial bans on the use of lead in ammunition and shot, (3) there is a seasonal peak in blood lead concentrations related to hunting season in southern European countries, (4) the levels of exposure in several species are generally relatively high and exceedance of subclinical threshold levels is widespread.
By conducting this meta-analysis, we have identified a range of gaps in information provision and knowledge. We make a number of recommendations about future approaches that are needed if our understanding of the impacts of lead on raptors is to be improved. These are: • Monitoring the same species across different monitoring schemes and countries. Suitable candidate species have been suggested elsewhere (Badry et al., 2020). In agreement, we suggest the common buzzard and the golden/white-tailed sea eagle combination as potential species to monitor lead trends in Europe. Our review indicates these species are more prone to lead poisoning and are also most widely monitored across Europe. Their distribution is also suitable covering most of Europe. • Monitoring the same sample matrices across schemes. This review shows that lead levels in kidney and liver are highly correlated and these matrices could be used interchangeably to some extent, although this would require further studies to generate speciesspecific read-across values. Monitoring of lead using non-invasive samples, such as feathers, has the potential to increase our breadth of knowledge about exposure of raptors to lead but further studies are required to determine adverse-effects threshold values. • This review shows that there is a scarcity, and in some cases complete absence, of data on lead exposure and monitoring in the northern and eastern Europe, and increased sampling and measurements are recommended for these regions. • As hunting ammunition is among the most important causes of lead poisoning in raptors, publications should always specify when samples are collected (winter, breeding, hunting, non-hunting season). • Lead-based ammunition is a well-recognised source of lead exposure and poisoning in raptors, yet it is difficult to confirm. More routine measurement of tissue isotopic signatures, searching for ammunition fragments in the gut, and non-invasive monitoring of regurgitated pellets would all enable quantification of the true scale of exposure. Where possible, we recommend analysis of lead isotopic ratios as well as determination of total lead concentrations. • When reporting results, studies need to publish the data for individuals, either as part of the supporting information of the paper or preferentially as published datasets that can be downloaded from data archives. Metadata should include information of the quality assurance measurements associated with the lead concentration data. Provision of such information, and inclusion within publications of the full range of summary statistics (mean, median, variance and range of values) would enhance data comparability and the value of such studies for future systematic evaluations.
• When mortality or subclinical effects are reported, it is crucial to provide individual information on the tissue lead levels that are related to the effects. • This review shows that there are few studies on the subclinical effects of lead in raptors and the consequences and significance of subclinical exposure remains largely unknown. We highlight the need for such studies which should focus on subclinical effects that may directly or indirectly affect survival and reproduction.
Finally, it is evident from this review that lead-induced mortalities and subclinical effects occur in European raptors. However, the impacts on population demography appear to have been little studied. This is in stark contrast to studies on species such as the Californian condor (Gymnogyps californianus) in North America which have resulted in real understanding of the population impacts of lead and the need for, and effectiveness of, mitigation options in this species (Finkelstein et al., 2012;Herring et al., 2018). In our review, we did not find any studies that related exposure to lead in raptors to predicted quantified impacts on European raptor populations. While the death of individual raptors from lead poisoning is clearly undesirable (and in some cases may be sufficient to elicit mitigation), an understanding of the risk at the population level is typically the evidence that guides the need for mitigation. Improved assessment of the extent and scale of lethal and subclinical effects, modelling of how such effects may alter population demography, and evidence of how bans (where these have been implemented) have reduced exposure, are all needed. Such evidence at a pan-European scale is likely to require increased harmonisation between national contaminant raptor monitoring schemes in Europe (Gómez-Ramírez et al., 2014).

Funding sources
This paper arises from work carried out by the European Raptor Biomonitoring Facility (ERBF) COST Action (CA16224). A short-term scientific mission was granted to Laura Monclús to investigate published data on lead monitoring in European raptors in the Leibniz Institute for Zoo and Wildlife Research in Berlin, Germany. Richard Shore was supported by the Natural Environment Research Council National Capability UK-SCAPE programme [NERC grant reference number NEC06550].

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.