Epidemiology of Animal Poisonings in the Canary Islands (Spain) during the Period 2014–2021

Animal poisoning is one of the greatest conservation threats facing wildlife. In a preliminary study in the oceanic archipelago of the Canary Islands, we showed that the degree of threat from this circumstance was very high-even higher than that reported in other regions of continental Europe. Consequently, a legal framework for the effective prosecution of the crime of wildlife poisoning came into force in 2014 in this region. We present the results of the investigation of 961 animals and 84 baits sent to our laboratory for the diagnosis of animal poisonings during the period 2014–2021. We were able to identify poison as the cause of death in 251 animals and 61 baits. Carbofuran stands out as the main agent used in this archipelago. We have also detected an increasing tendency to use mixtures of several pesticides in the preparation of baits. The entry into operation of two canine patrols has led to the detection of more dead animals in the wild and a greater number of poisoned animals. The percentage of poison positives is significantly higher in areas with lower population density, corresponding to rural environments, as well as in areas with greater agricultural and livestock activity.


Introduction
Poison remains one of the most important conservation threats faced by many wildlife species, and it also affects domestic animals frequently [1][2][3][4][5][6][7][8][9][10][11][12]. There are numerous chemicals that can affect animals, but perhaps the group of agricultural pesticides is the most important [13]. These compounds, commonly used and widespread in global food production, cause many accidental poisonings in non-target species [5,7,13,14], and accidental poisonings by these substances in humans are also frequently described [15,16]. Among the pesticides, rodenticides stand out, since they are directed against higher vertebrates-rodentsand can easily reach other non-target species that share their habitat [17]. Numerous studies

Sampling
Blood and liver samples from 961 animals received in our laboratory for diagnosis of possible poisoning were analyzed between January 2014 and June 2021. During this period, we also received 84 meat baits. Occasionally, we also received other samples, such as gastric contents, degraded carcasses, cadaveric fauna, and insects collected at the scene of the incident. When necessary, these samples were used to clarify results found in blood or liver. However, none of the results we present were obtained solely from these ancillary samples, so cases where blood or liver were not available have not been included. It is noteworthy that 424 animals belong to species or subspecies endemic to the Canary Islands, highlighting 307 birds of prey and 74 lizards (Gallotia spp.), the latter being critically endangered. The animals correspond to a total of 753 incidents investigated, giving an average of 1.3 animals affected per incident (although they ranged from one to 20 animals per incident). The animals investigated belonged to 63 different species, including 44 dogs and 49 cats. Table 1 (Results and Discussion section) lists the species in which at least one chemical as the cause of death was detected. All animals were collected in the context of investigations of possible environmental crimes by environmental agents and were transferred to the wildlife rehabilitation centers of Tafira (in Gran Canaria) or La Tahonilla (in Tenerife) where liver samples for toxicological analysis were collected during necropsy. Once collected, the samples were kept frozen until they were transferred to the Clinical and Analytical Toxicology Service (SERTOX) of the University of Las Palmas de Gran Canaria (ULPGC, Canary Islands, Spain), where they remained frozen at −24 • C until their analysis. Most of the animals were found dead in the field or in urban areas or died while in veterinary facilities. No animals were sacrificed for the purposes of this study, and no experiments were performed on or with live animal samples.

Analytical Method
The methodology used for blood and liver samples allowed the search respectively for 360 and 351 compounds highly toxic to animals and has been fully validated according to international guidelines [38,39] and previously published by our group [3,35,36,40]. The complete list of analytes, as well as the technique used for their quantification, can be found in Appendix A. The extraction of blood and liver are based on the QuEChERS technique, although in our methods, a miniaturization of this technique has been performed, allowing the use of only 250 µL of blood [35,36] or 1 g of liver [3,40], without requiring any additional purification step, nor any change of solvent. For the rest of the samples (baits, gastric and intestinal contents, decomposing carcasses), a solid-liquid extraction was usually used, according to a procedure also previously published by our group [4,37], although in the case of these auxiliary samples, we consider the analyses to be semi-quantitative only.
The quantitative analysis was carried out in all cases by a combination of two complementary analyses, one by gas chromatography coupled to triple quadrupole mass spectrometry (GC-MS/MS) for the analysis of the more volatile compounds (mainly persistent organic pollutants and some less polar pesticides) and another analysis by liquid chromatography coupled to triple quadrupole mass spectrometry (LC-MS/MS) for pharmaceuticals, rodenticides, and more polar pesticides. For GC-MS/MS, an Agilent 7890B gas chromatograph (Agilent Technologies, Palo Alto, CA, USA) was used, which was equipped with two Agilent J&W HP-5MS (5% cross-linked phenyl-methyl-polysiloxane, Agilent Technologies) fused silica capillary columns, with a total length of 30 m (15 + 15 m), a film thickness of 0.25 µm, and 0.25 mm diameter each. The reason for using two columns joined by a purged joint was to allow the application of the backflushing technique that reduces background noise and prolongs the lifetime of the column. He 5.0 (99.999%) was used as the carrier gas, and N 2 6.0 (99.9999%) was used as the collision gas. Gases were from Linde (Dublin, Ireland). For LC-MS/MS, an Agilent 1290 Infinity II UHPLC instrument (Agilent Technologies, Palo Alto, CA, USA) was used. It was equipped with an InfinityLab Poroshell 120 column (2.1 mm × 100 mm, 2.7 µm), coupled to an in-line filter and a UHPLC guard column with the same characteristics as the analytical column. The detection and quantification of GC-separated compounds was performed with an Agilent 7010c mass spectrometer, and that of LC-separated compounds with an Agilent 6460 electrospray Jet Stream (AJS-ESI) mass spectrometer (Agilent Technologies, Palo Alto, CA, USA), which were both operated in dynamic multiple reaction monitoring (dMRM) mode, in positive (GC-MS/MS), and positive and negative (LC-MS/MS) polarities. All the technical details of the extractions, the programming of the equipment, and the results of the validation of the techniques can be found in the cited references [3,4,[35][36][37].
All quantifications were performed against calibration curves prepared in matrix, blood, or liver as appropriate with a mixture of the 360 individual chemicals and the deuterated compounds used as procedural internal standards. All standards were of the highest purity available (93.1% to 99.8%) and purchased from various suppliers (A2S-Analytical Standard Solutions (Staint Jean D'Illac, France), Sigma-Aldrich (Augsburg, Germany), CPA Chem (Stara Zagora, Bulgaria), European Pharmacopoeia Reference Standards (Strasbourg, France), Accustandard (New Haven, CT, USA), and Dr. Ehrenstorfer (Augsburg, Germany)). The calibration curves were prepared with a minimum of 6 points, covering the range up to 2 µg/mL.

Statistical Analyses
All statistical analyses were performed with GraphPad Prism v9.2 software (GraphPad Software, San Diego, CA, USA). The distribution of the variables included in this study was evaluated using the Kolmogorov-Smirnov test. Given the nature of the cases investigated (animals due to poisoning), the concentrations of most of the chemicals detected did not follow a normal distribution, so the results presented in Table 2 are expressed in terms of median and range. For the same reason, we employed nonparametric tests to check for statistical differences between the concentrations found in the groups of animals, using the Toxics 2021, 9, 267 5 of 19 median rather than the mean. For the study of determinants of poisoning, the dichotomous categorization (0/1) was used for the absence or presence of substances in a concentration that was compatible with the poisoning of the animals, regardless of whether one or more substances were involved. The association study between the different determinants and the outcome (poisoning vs. non-poisoning) was evaluated with the Chi-square test (χ 2 ). A p-value of less than 0.05 (two-tailed) was considered statistically significant in all the statistical analyses.

Results and Discussion
We studied the incidence of wildlife poisonings in the period from January 2014 to June 2021. We finally included liver or blood samples from 961 deceased animals and 84 baits. A total of 312 animals were referred with strong suspicion of poisoning, and of these, the presence of poisoning was confirmed in 223 animals (71.5%). The remaining 649 animals were sent to rule out the presence of poison, among other possible causes of death, and the poisoning was confirmed in 29 of them (4.5%). All baits were initially classified as intentionally supplied in the environment, and the presence of some type of poison was confirmed in 61 of them (72.6%). The total number of positive identifications represented 29.9% of the samples submitted during this period. The number of deceased animals in incidents where a chemical was detected in toxic concentrations was significantly higher than in negative cases (mean 2.53 ± 0.28 vs. 1.31 ± 0.15, p < 0.0001), as has been described in other works at the international level [6,7,[40][41][42][43], and in the previous studies conducted in the Canary Islands [4,44].
In Table 1, we show the number of positive cases distributed by species and type of toxicants involved. The wildlife species with the highest number of positive cases was Columba livia, with 92.6% of the specimens submitted positive for poison (25/27). This can be explained by the fact that only pigeons for which a strong suspicion of poisoning had been received by our laboratory, in contrast with what occurred with other species. We also identified poisoning in a high number of Falco tinnunculus canariensis (32/108), Gallotia spp.  [4], these two species represented the highest percentage in the total diagnoses. In the period we now present, the percentage of positives among domestic animals has fallen appreciably, while the percentage of positives among wild species has remained fairly similar over the years [4].
Seventeen different toxicants were detected, and most of them were pesticides, except for amobarbital, which was identified as the probable cause of death in one case (Corvus corax canariensis). According to our results, in the positive cases, a mean of 1.54 ± 0.38 toxicants per incident were detected. The most frequently detected chemical was carbofuran, which was present in 97 animals and 24 baits (38.7% of positive cases). This percentage is very similar to that previously reported in the Canary Islands [4], which would indicate that the habit of using this compound to get rid of nuisance animals has not diminished in the Canary Islands. It should be noted that carbofuran has been banned in the EU since 2007. Almost 15 years after its withdrawal from the market, there still seems to be an important stock of this compound in private farms in the archipelago, which has not been handed over to the authorities and which continues to be used illegally. The use of aldicarb, also banned in the EU for almost two decades [45], does not seem to have decreased either, as it was detected in almost 10% of the positive cases (in the previous period, it was present in 12.2% of the positive cases [4]). In addition, the percentage of cases in which the cause of death was an anticoagulant rodenticide has remained practically unchanged between both periods (29.9% in 2014-2020 vs. 29.1% in 2010-2013). However, regarding the rest of the detected compounds, we did find differences between both periods, although this was not too important. We detected a greater variety of substances involved (17 vs. 14) as well as a greater involvement of other acetylcholinesterase inhibitor insecticides, such as chlorpyrifos, pirimiphos methyl, dimethoate, oxamyl, or methomyl. We were struck by the fact that unlike what happened in the previous study period, in these years, we have detected an increase in the cases in which several poisons are detected simultaneously, which became more evident in the study of baits. In 35.8% of the cases, the baits were prepared with between three and five different compounds, all of them commonly used in Canary Island agriculture in the past but mostly also substances currently banned in the EU. This could be due to the fact that stocks of the more potent poisons that have traditionally been used may be running low in some places, and poisoners seek to maximize the effectiveness of the baits they prepare by mixing different substances of lower toxicity than the compounds previously used.
In Figure 1, we present in graphical form a comparison of the distribution of poisons that have affected wild and domestic fauna in the Canary Islands in both study periods; the present study corresponds to the investigations after the regional law against the poison in natural areas came into force [34], as compared to our previous report, in which the cases were received quite informally [4]. A slight change in the pattern of use of poisons is evident, particularly regarding baits. Possibly, this indicates that in certain parts of the archipelago, carbofuran and aldicarb have been depleted, and mixtures of other compounds that are assumed to be less potent individually are beginning to be used. cases were received quite informally [4]. A slight change in the pattern of use of poisons is evident, particularly regarding baits. Possibly, this indicates that in certain parts of the archipelago, carbofuran and aldicarb have been depleted, and mixtures of other compounds that are assumed to be less potent individually are beginning to be used. With respect to the proportion of poison-positive cases among the total number of cases referred during this period, we found significant differences depending on whether they were wild or domestic animals. As shown in Figure 2, the percentage of positives among wildlife barely reached 20%, while in the case of dogs and cats, this percentage was more than double. This finding is not surprising, since it has been described that cats, and mainly dogs, when sick become much more visible to humans than wild animals, which, in these same circumstances, tend to seek refuge in their nests and burrows, and many of them are never found [46]. We highlight this fact because the cases of poison in wildlife could be much higher than what we have officially recorded from the samples submitted to us, because dying animals that hide are probably never found. As can be seen in the graph, in the baits analyzed, the percentage of positive identifications was very high (75%). This finding is also very logical, since when baits are sent to us, their appearance and location attract attention in most cases, and there are strong suspicions that they have been laced with poison, and virtually all these samples are submitted to the laboratory for investigation. Additionally, the poisons of interest are expected to be more concentrated and perhaps also more stable in the bait source compared to biological samples, so even if there is degradation, it is more likely that high enough concentrations will remain to allow detection. With respect to the proportion of poison-positive cases among the total number of cases referred during this period, we found significant differences depending on whether they were wild or domestic animals. As shown in Figure 2, the percentage of positives among wildlife barely reached 20%, while in the case of dogs and cats, this percentage was more than double. This finding is not surprising, since it has been described that cats, and mainly dogs, when sick become much more visible to humans than wild animals, which, in these same circumstances, tend to seek refuge in their nests and burrows, and many of them are never found [46]. We highlight this fact because the cases of poison in wildlife could be much higher than what we have officially recorded from the samples submitted to us, because dying animals that hide are probably never found. As can be seen in the graph, in the baits analyzed, the percentage of positive identifications was very high (75%). This finding is also very logical, since when baits are sent to us, their appearance and location attract attention in most cases, and there are strong suspicions that they have been laced with poison, and virtually all these samples are submitted to the laboratory for investigation. Additionally, the poisons of interest are expected to be more concentrated and perhaps also more stable in the bait source compared to biological samples, so even if there is degradation, it is more likely that high enough concentrations will remain to allow detection. During these years, the diagnosis of death by poisoning has been made based either on the comparison of the liver concentrations found with the data available in the literature, on the calculation of the toxic dose from the blood concentration and the apparent volume of distribution of the poison found (when these data were available), and on the clinical or presumptive findings of poisoning collected in the files by the environmental police agents or by the veterinarians who attended the animals. It was not always possible to attribute the death of the animal to the toxic substances found, which was either because the concentrations found did not seem sufficiently high or because the advanced state of degradation of the samples did not allow knowing if there was degradation of the chemical substances decreasing their concentration or because there were no reference data. Therefore, the doubtful cases have not been included as positive in this study. However, although the outcome does not change once a certain threshold concentration of a chemical is exceeded that is considered potentially lethal, we do consider that the finding of extremely high concentrations in liver would reflect exposure to massive doses of the poison, probably after ingestion of a bait, thus indirectly suggesting a pattern of intentionality in the poisoning. Obviously, intentionality cannot be inferred from concentrations alone, so we only state this as a working hypothesis based on our experience. According to this hypothesis, we wanted to compare whether there were also differences in the type of poisons affecting one or the other type of animal (wildlife vs. domestic) and the concentrations found for each of them in both types of animals ( Table 2).
We found significant differences in the concentrations of some toxicants. Thus, the most striking case is that of aldicarb, which presented a median value about 40 times higher in the series of wildlife than in that of domestic animals (p < 0.05), with several animals of different incidents presenting massive concentrations of this substance in their liver. Something similar was observed with brodifacoum, which also presented significantly higher concentrations in the livers of wild animals than in domestic animals (p < 0.0001). In the first case, the use and even possession of aldicarb has been prohibited for almost 20 years [45], so that accidentality could be ruled out in all cases. In the case of brodifacoum, there is a limitation for its outdoor use in the agricultural environment, so this result was to some extent surprising. The high concentrations found in wild animals compared to dogs and cats probably suggests a pattern of intentionality, at least in part of During these years, the diagnosis of death by poisoning has been made based either on the comparison of the liver concentrations found with the data available in the literature, on the calculation of the toxic dose from the blood concentration and the apparent volume of distribution of the poison found (when these data were available), and on the clinical or presumptive findings of poisoning collected in the files by the environmental police agents or by the veterinarians who attended the animals. It was not always possible to attribute the death of the animal to the toxic substances found, which was either because the concentrations found did not seem sufficiently high or because the advanced state of degradation of the samples did not allow knowing if there was degradation of the chemical substances decreasing their concentration or because there were no reference data. Therefore, the doubtful cases have not been included as positive in this study. However, although the outcome does not change once a certain threshold concentration of a chemical is exceeded that is considered potentially lethal, we do consider that the finding of extremely high concentrations in liver would reflect exposure to massive doses of the poison, probably after ingestion of a bait, thus indirectly suggesting a pattern of intentionality in the poisoning. Obviously, intentionality cannot be inferred from concentrations alone, so we only state this as a working hypothesis based on our experience. According to this hypothesis, we wanted to compare whether there were also differences in the type of poisons affecting one or the other type of animal (wildlife vs. domestic) and the concentrations found for each of them in both types of animals ( Table 2).
We found significant differences in the concentrations of some toxicants. Thus, the most striking case is that of aldicarb, which presented a median value about 40 times higher in the series of wildlife than in that of domestic animals (p < 0.05), with several animals of different incidents presenting massive concentrations of this substance in their liver. Something similar was observed with brodifacoum, which also presented significantly higher concentrations in the livers of wild animals than in domestic animals (p < 0.0001). In the first case, the use and even possession of aldicarb has been prohibited for almost 20 years [45], so that accidentality could be ruled out in all cases. In the case of brodifacoum, there is a limitation for its outdoor use in the agricultural environment, so this result was to some extent surprising. The high concentrations found in wild animals compared to dogs and cats probably suggests a pattern of intentionality, at least in part of the cases. Even more so considering that with another frequently detected rodenticide, bromadiolone, just the opposite is true. Bromadiolone levels were significantly higher in the livers of intoxicated pets (Table 2). This second-generation anticoagulant is a legal and very commonly used rodenticide, both in urban and rural areas, including agriculture and livestock. It is more than likely that most of the cases involving this toxicant, both in domestic and wild animals, are due to accidental poisoning, probably secondary to the ingestion of poisoned rodents, rather than to the ingestion of baits. With respect to the other chemical substance for which there was a significant difference between wild and domestic animals, permethrin, it should be noted that this is a compound of particularly high toxicity to cats [47], which is the species with the highest liver concentrations. It is true that the literature describes that some cat poisonings are due to the accidental application of flea products labeled for dogs. It might be reasonable for well-meaning people to apply permethrin-containing dog products to cats with the goal of helping rather than harming them. However, it should also be noted that permethrin was also identified in two of the baits analyzed, so, at least in some cases, the high concentrations found could also point to a pattern of intentionality aimed at eliminating stray cats. Finally, some compounds were only detected in wildlife specimens, such as amobarbital, dimethoate, fenamiphos, flocoumafen, and pirimiphos methyl, while others were only detected in domestic animals, such as alpha chloralose, chlorpyrifos, imidacloprid, methiocarb, and tetramethrin, although some of the chemicals were detected only sporadically. Among them, it is worth highlighting the detection of amobarbital, since it represents the first case of poisoning by barbiturates recorded in the Canary Islands, unlike what has been reported recently for mainland Spain, where barbiturates were involved in up to 3.4% of the poisonings detected [31].
During this period, the number of positive cases per year has not shown a downward trend (Figure 3), remaining stable. What has increased is the number of samples received in our laboratory to rule out cases of poisoning, as this increase has been very noticeable from 2017. We previously pointed out that one of the biggest problems in the investigation of wildlife poisoning is that a large part of the cases may never be detected due to the elusive behavior of most species when they are seriously ill [46]. Precisely because of this, and in the context of the Canarian strategy against poison [34], two canine patrols trained in the detection of poisons and carcasses have come into operation: the first one was on the island of Gran Canaria, which began operating in 2017, and the second one came into operation in 2020 and covers the islands of Fuerteventura and Lanzarote. As can be seen in Figure 3, these milestones coincide with respective increases in the receipt of samples and the corresponding increase in the detection of poisoning cases. Even so, although there was a quantitative increase in the number of positive cases identified, this did not alter the proportion of cases in relation to the total number of samples received in the laboratory, remaining approximately the same. Only in the last period, from January 2020 onwards, has there been a slight increase in the number of positive cases detected, but it is too early to conclude whether this trend will continue over time. Future studies will test whether canine patrols contribute effectively to the visibility of wildlife poisoning cases, as has been described in other regions [46]. Toxics 2021, 9, x FOR PEER REVIEW 10 o The Canary Islands include eight inhabited islands, five islets, eight rocks, and sea. The animals received in our service correspond only with the eight inhabited isla although the smallest of all, La Graciosa, is usually considered together with Lanzar on which it has depended administratively until very recently (2018). Figure 4 repres how the cases received have been distributed in relation to the island on which the i dent occurred. As can be seen, most of the cases were recorded on the island of Gran naria, which is where our laboratory is physically located. Probably, since it is an ar pelago, this is due to logistical reasons, since it was easier to send the samples, especi during the first few years of operation. It is also noteworthy that one of the canine patr the first to become operational, is also based on this island. Although this patrol oper throughout the archipelago, it is true that the highest rate of interventions occurs on island of Gran Canaria, so this undoubtedly influences this difference with respect to rest of the archipelago. From the islands of Fuerteventura and, to a much lesser ext Lanzarote, a good number of cases have also been received, especially in the last years, also coinciding with the entry into operation of the second canine patrol, whic physically located on the island of Fuerteventura. Our results indicate that dogs trai in the detection of poisons are a valuable aid in detecting this serious environmental p lem. With respect to the cases coming from the islands farthest from our laboratory Gomera, La Palma, and El Hierro), the high percentage of positives is striking, reach 100% of those sent from the island of El Hierro. From our point of view, this reflects logistical problems that have existed during these first years of operation of the Can Islands anti-poison strategy, which has meant that the cases sent for investigation f these islands have been meticulously selected. The case of the island of Tenerife is n worthy, since it is the most populated island of the archipelago, the one with the grea agricultural and livestock activity, and one of the islands with the greatest biodiver However, the low rate of samples investigated on this island is surprising. This situa will probably change significantly in the next few years, since a third canine patrol is pected to start operating, which will be located on this island, and which will undoubte help to make visible the cases of poisoning that we believe are going unnoticed there. The Canary Islands include eight inhabited islands, five islets, eight rocks, and the sea. The animals received in our service correspond only with the eight inhabited islands, although the smallest of all, La Graciosa, is usually considered together with Lanzarote, on which it has depended administratively until very recently (2018). Figure 4 represents how the cases received have been distributed in relation to the island on which the incident occurred. As can be seen, most of the cases were recorded on the island of Gran Canaria, which is where our laboratory is physically located. Probably, since it is an archipelago, this is due to logistical reasons, since it was easier to send the samples, especially during the first few years of operation. It is also noteworthy that one of the canine patrols, the first to become operational, is also based on this island. Although this patrol operates throughout the archipelago, it is true that the highest rate of interventions occurs on the island of Gran Canaria, so this undoubtedly influences this difference with respect to the rest of the archipelago. From the islands of Fuerteventura and, to a much lesser extent, Lanzarote, a good number of cases have also been received, especially in the last two years, also coinciding with the entry into operation of the second canine patrol, which is physically located on the island of Fuerteventura. Our results indicate that dogs trained in the detection of poisons are a valuable aid in detecting this serious environmental problem. With respect to the cases coming from the islands farthest from our laboratory (La Gomera, La Palma, and El Hierro), the high percentage of positives is striking, reaching 100% of those sent from the island of El Hierro. From our point of view, this reflects the logistical problems that have existed during these first years of operation of the Canary Islands antipoison strategy, which has meant that the cases sent for investigation from these islands have been meticulously selected. The case of the island of Tenerife is noteworthy, since it is the most populated island of the archipelago, the one with the greatest agricultural and livestock activity, and one of the islands with the greatest biodiversity. However, the low rate of samples investigated on this island is surprising. This situation will probably change significantly in the next few years, since a third canine patrol is expected to start operating, which will be located on this island, and which will undoubtedly help to make visible the cases of poisoning that we believe are going unnoticed there. Finally, we wanted to study the main determinants of the pattern of poison use in the Canary Islands, using the available variables, mainly in relation to parameters related to the species, land use, and population density, as has been established in other research studies [21].
First, we studied the influence of habitat (urban vs. rural), and population density ( Figure 5). We found that the number of animals referred from cities is significantly higher than those referred from smaller towns (based on the number of inhabitants of the municipality in which the carcass was found, cut-off point = 19,657 inhabitants (median value)), but the percentage of positive cases is the inverse, being significantly higher in animals referred from rural localities ( Figure 5, left). Something very similar occurred with the population density (cut-off point = 161 inhabitants/km 2 ). The number of cases referred from less densely populated areas was lower, which was probably because it is more difficult to find the carcasses, but the percentage of positive cases among the animals referred from these areas (43.1%) was significantly higher than those referred from more populated areas (34.1%) (Figure 5, right). This finding seems quite logical to us, given that most of the chemical compounds we have detected are agricultural pesticides, so their availability in rural areas should be greater. Finally, we wanted to study the main determinants of the pattern of poison use in the Canary Islands, using the available variables, mainly in relation to parameters related to the species, land use, and population density, as has been established in other research studies [21].
First, we studied the influence of habitat (urban vs. rural), and population density ( Figure 5). We found that the number of animals referred from cities is significantly higher than those referred from smaller towns (based on the number of inhabitants of the municipality in which the carcass was found, cut-off point = 19,657 inhabitants (median value)), but the percentage of positive cases is the inverse, being significantly higher in animals referred from rural localities ( Figure 5, left). Something very similar occurred with the population density (cut-off point = 161 inhabitants/km 2 ). The number of cases referred from less densely populated areas was lower, which was probably because it is more difficult to find the carcasses, but the percentage of positive cases among the animals referred from these areas (43.1%) was significantly higher than those referred from more populated areas (34.1%) (Figure 5, right). This finding seems quite logical to us, given that most of the chemical compounds we have detected are agricultural pesticides, so their availability in rural areas should be greater.
We also explored the relationship with agricultural and livestock activity in the municipalities where the carcasses that were sent to our laboratory were found ( Figure 6). First, with respect to agricultural activity, we found that there was a relationship, with the percentage of positive cases being significantly higher in those municipalities with greater agricultural activity. Several cut-off points were used to calculate this (number of cultivated hectares, cultivated area surface (%), and cultivated area per inhabitant), and with all of them, statistical significance was maintained (Figure 6, left). We also find this result logical for the same reason discussed above: the availability of agricultural products, whether permitted or not, is closely related to agriculture. According to our results, the same is true for livestock activity, since the situation is repeated: the percentage of positive cases in municipalities with more livestock is significantly higher than in those with less livestock activity. This significance was maintained for all the cut-off points tested (total number of livestock in the municipality; livestock density (number of heads/hectare); and livestock density/population density) (Figure 6, right). We also explored the relationship with agricultural and livestock activity in the municipalities where the carcasses that were sent to our laboratory were found ( Figure 6). First, with respect to agricultural activity, we found that there was a relationship, with the percentage of positive cases being significantly higher in those municipalities with greater agricultural activity. Several cut-off points were used to calculate this (number of cultivated hectares, cultivated area surface (%), and cultivated area per inhabitant), and with all of them, statistical significance was maintained (Figure 6, left). We also find this result logical for the same reason discussed above: the availability of agricultural products, whether permitted or not, is closely related to agriculture. According to our results, the same is true for livestock activity, since the situation is repeated: the percentage of positive cases in municipalities with more livestock is significantly higher than in those with less livestock activity. This significance was maintained for all the cut-off points tested (total number of livestock in the municipality; livestock density (number of heads/hectare); and livestock density/population density) (Figure 6, right). In conjunction with our previous report of poisonings in the Canary Islands [4], our results indicate that the incidence of poisonings in this archipelago is very high and probably higher than in other European regions [7,9,9,14,41,48]. Moreover, the profile of toxi- In conjunction with our previous report of poisonings in the Canary Islands [4], our results indicate that the incidence of poisonings in this archipelago is very high and probably higher than in other European regions [7,9,14,41,48]. Moreover, the profile of toxicants that we have found suggests that many of these poisonings occur intentionally, given the high prevalence of substances whose use in agriculture would be illegal throughout the EU. Since this practice is highly detrimental to biodiversity, as well as a major public health problem, it is necessary that the authorities enact effective measures on the marketing of toxic chemicals, the control of stocks of banned chemicals, the implementation of educational programs and the effective criminal prosecution of poisoners to prevent, or at least minimize, the incidence of this harmful practice. From the literature, it can be assumed that the numbers reported in those studies only represent an approximation of the actual incidence of wildlife mortality, because it has been estimated that less than 10% of poisoning cases are detected and sent to a forensic laboratory [49]. This is especially relevant for wildlife because sick animals are often less visible and many die in nests, burrows, or inaccessible locations. The presence of canine patrols probably increases detection rates, as we have found in this study, but it is still quite likely that a good portion of cases will go undetected, particularly regarding wildlife. Whatever the case, our findings indicate that the actual incidence of poisoning mortality in the Canary Islands is very high and certainly higher than in other European regions.
We have observed a very slight decreasing trend in the use of prohibited substances, which is much lower than the progressively decreasing annual trend reported in other regions [50,51]. Carbofuran, aldicarb, and other banned AChE inhibitors were used extensively in agriculture in the Canary Islands, which are mainly associated with the cultivation of banana and other export vegetables. It is likely that there are still significant stocks of these banned substances on many farms, although there is also the possibility that they are still being acquired on the black market [50].
Although numerous measures have been taken to correct this problem in this region, it is probably too early to verify their efficacy. The authorities should take different measures to correct the circumstances that motivate the intentional poisoning of animals to curb, or at least minimize, this serious problem that seriously threatens biodiversity, animal welfare, and public health.