*2.3. Statistical Analyses*

We conducted General Linear Models (GLMs; Poisson error distribution and log link function) to explore the characteristics of the species that were most frequently studied in urban predation and urban scavenging studies. In order to do this, we only considered those studies that had analyzed the activity of one or more species within the city or surrounding area. We created two sets of models separating information about urban predation and urban scavenging studies. In both sets, the explanatory variables were the number of times each species appeared in the articles found during the bibliographic review. As explanatory variables we included: (a) taxonomic class; (b) taxonomic order; (c) carnivory level; (d) body mass; (e) daily rhythm. In both sets of models, we explored all combinations of the variables, considering only additive effects. To discuss the results, we followed an Information Criterion Approach. We selected the most parsimonious models (i.e., lowest AICc). In addition, we selected all models with a difference of ΔAICc < 2 compared to the most parsimonious model [76]. As the information criterion does not inform about the absolute predictive capacity of the models, we later discarded those selected models with little-relevant explanatory variables (i.e., the values at the 85% confidence interval which overlapped with 0 [76]. We used an 85% confidence interval to avoid the possible loss of valid models by using more restrictive and conservative approaches such as the 95% confidence interval [77]. We explored the overdispersion of selected models using the AER package [78]. Finally, to analyze how well our models fit, we calculated the explained deviance (D2). All statistical analyses were performed in the R v.3.6.1 statistical platform [79].

## **3. Results**

The results using predat\* as the search term returned a total of 32 items, while using scaveng\* offered only five results. Including only the articles found in the enclosed review, we observed that most predation articles were published after the 2000s, and all scavenging articles had been published in the last 10 years, with scavenging remaining in the background in the studies focused on carnivores in cities, although proving to be a nascent

interest as a study topic (Figure 1). By adding the articles obtained with complementary searches (snowball and other sources), we increased the number of studies, obtaining a final dataset of 95 articles for urban predation and 47 articles for urban scavenging (see Tables S1–S4 for details). Regarding the articles focused on urban scavenging, at least 13 of those included in the final list clearly demonstrated that these studies were carried out under experimental conditions, providing the scavengers with carrion (mainly farm animals) instead of evaluating the consumption of refuse scattered in cities.

**Figure 1.** Temporal comparison between urban predation and urban scavenging published studies, based on the total of articles retained strictly from the systematic review, conducted with closed search.

We found that most of the studies analyzed were developed in temperate biomes (68.4% of the predation studies and 50% of the scavenging studies), followed by urban predation studies in Mediterranean forests (11.58%) and by tropical biomes (22.92%) in urban scavenging studies. Regarding the geographical locations, in the case of urban predation, 35 studies were performed in Europe, 32 in America, 17 in Oceania, 6 in Asia and 5 in Africa, with the United States of America (*n* = 20) and Australia (*n* = 13) accumulating the highest number of studies (Figure 2A, S1). On the other hand, urban scavenging studies were carried out in America in 12 cases, 10 in Europe, 9 in Oceania and Asia, and 8 in Africa. As for urban scavenging studies, Australia (*n* = 9) and the United Kingdom (*n* = 5) were the countries with the most published papers (Figure 2B, S2). The population size of the cities studied ranged from 3.54 to 9.1 million for predation studies and 3500 to 9 million for scavenging studies, which produced no significant difference (*t* = 1.71, df = 44.46, *p*-value = 0.095). Regarding the location where the studies were focused, we detected that the majority were conducted in the interior of cities or included both urban and peri-urban areas (Figure 3).

Of the 95 urban predation studies, 93 contained information on at least one species (the remaining two were focused on eggs predation without identifying the predators). In total, we recorded 100 species in the reviewed urban predation studies (62 birds, 28 mammals and 10 reptiles; see Table 1 for a complete list). The most represented orders were Passeriformes (*n* = 23), Accipitriformes (*n* = 16) and Carnivores (*n* = 16). Seven species appeared as predators in five or more articles: *Felis catus* (*n* = 21), *Falco tinnunculus* (*n* = 8), *Tyto alba* (*n* = 8), *Vulpes vulpes* (*n* = 8), *Canis latrans* (*n* = 6), *Canis lupus familiaris* (*n* = 5), and *Falco peregrinus* (*n* = 5) (Figure 2A). Regarding urban scavenging, we registered 49 species (25 birds, 21 mammals, and 3 reptiles) (Table 2), with Carnivores being the most represented group (*n* = 11), followed by Passerines (*n* = 10) and Accipitriformes (*n* = 9). The number of species is lower in comparison with those studies focused on predation, a logical result considering the smaller number of studies found on scavenging. All studies on scavenging (47) contained information on one or more species. Urban scavengers appearing in five

or more different articles, were *Vulpes vulpes* (*n* = 11), and two domestic mammals, *Felis catus* (*n* = 9) and *Canis lupus familiaris* (*n* = 7) (Figure 2B). From the total of articles reviewed on predation, 21 contained sufficient information to calculate species richness (mean and standard deviation: 5.36 ± 5.26; range: 2–20 species per study), while 9 of the scavenging studies contained such information (mean and standard deviation: 7.00 ± 2.54; range: 3–10 species per study).

**Figure 2.** Number of urban predation (**A**) and scavenging (**B**) studies per country. Darker colors show high number of articles found for a given country, ranging from 0 to 20. The circles represent the species most reported in the articles (appearing on 5 or more occasions), ordered from most-reported to least-reported, from the top left corner. Pictures by: Manfredrichter, Ruthmcd, Miles, Wilda3, Skeeze and Ebor.

**Figure 3.** Number of studies conducted in urban and/or peri-urban areas on predation (blue bars) and scavenging (orange bars).




**Table 2.** Complete list of scavengers recorded in all articles reviewed. N is the number of studies that include each species.

In urban predation and scavenging studies, the most recorded species were those with meat-based diets and nocturnal activity (Table 3). In both sets, there were other models that included the explanatory variables carnivorous level and class (Table 3), at only a few tenths above the threshold of 2 ΔAICc. The selected models explained an important fraction of the deviance (0.27 in the case of the urban predation model and 0.29 in the case of the urban scavenging model) (Table 4). Both models showed very low values of over-dispersion (0.63 in the case of the urban predation model and 0.18 in the case of the urban scavenging model).

**Table 3.** AIC-based model selection to assess the characteristics of the scavenger species most recorded. Only models with informative variables are included. Model selected is represented in bold. Number of estimated parameters (K), AICc values, AICc differences (ΔAICc) compared to the highest ranked model (i.e., the one with the lowest AICc), Akaike weights (AICcWt), Cumulative weight (CumWt) and the variability of the models explained by the predictors (D2) are represented.



**Table 4.** Results of the General Linear Models (Poisson error distribution and identity linkage) selected to explain the number of studies in which each species appears. The estimate of the parameters (including the sign), the standard error of the parameters, the degrees of freedom of the models (DF) and variability explained by the models are represented (D2).

## **4. Discussion**

## *4.1. Temporal Trends of Predation/Scavenging Studies in Urban Ecology*

In this review, we explored the scientific literature focused on urban predators and scavengers. A positive observation is that there is a growing interest in these subject as a study topic, as the articles have been increasing in number over the years. To strengthen the result that the topic is gaining interest, it can be noted that most of the articles reviewed had predation and scavenging as their main subject, and they did not relegate this aspect to a secondary interest within other subjects. Moreover, digging further into the general information assessed in the articles published, we can observe that most of the research studied not only peri-urban areas but also the interior of the cities, showing a clearer understanding of the activities of predators and scavengers in the cities themselves [80,81]. However, it is also true that in the case of scavenging studies, a non-negligible amount was carried out under experimental conditions [49,50,82,83], so it would be of interest to devote more effort to understanding the interaction of certain species with actual available refuse. When explaining the reason behind the temporal growth in the number of studies on predation and scavenging in cities, we consider that the general growth of urban ecology, as a relatively new discipline, may be also responsible for this positive trend [59,84]. Nevertheless, the gap detected between the number of articles referring to predators and scavengers reveals that less attention was devoted to urban scavenging, confirming that this topic has been widely overlooked and, to date, is lesser known and understood. To no surprise, our bibliographic search on urban scavenging only returned five articles, and despite having completed the dataset through the "snowball" and the complementary search, the total number of articles referring to urban scavenging was half of that found for urban predation. In any case, we cannot ignore that, in general, studies focused on scavenging ecology are fewer, with an increase only in recent years [62], a reality which may help to explain our result. Clearly, our search on both subjects (e.g., urban predators and scavengers) does not include the total studies on both subjects; however, the review we conducted shows that the number of studies is not related to the relevance that scavenging studies may acquire, attending to their ecological importance and the ecosystem services they might provide in cities [62,85].

## *4.2. The Bias Extant in Urban Predation and Scavenging Studies across Regions*

We confirmed a geographical bias in the reviewed studies across different regions of the world [60], as the information is concentrated in a few specific countries while it remains scarce for other vast regions, especially the less developed ones (see Figure 2). We must address the simple fact that the lower scientific research activity detected in developing countries [86] may contribute to explaining the scarcity of articles focused on predation and scavenging in cities located in less developed countries. Moreover, we observed that a great variety of biomes are clearly underrepresented, even in regions where studies have been conducted, confirming the existence of a second spatial gap. Curiously, the regions which are less represented in the reviewed articles (mainly in Africa, Asia and Latin America) have urban areas that accumulate greater quantities of garbage

and carrion in the streets (their poor management systems prevent them from handling the enormous amount of waste produced [87]), and thus are potentially suitable areas to study the interaction between wildlife and refuse. Additionally, the sites with worst waste collection are sites with potentially higher populations of synanthropic species (such as rats [88]) and semi-domestic animals roaming in the streets, and are thus locations where, hypothetically, more carnivores can thrive [45,89].

## *4.3. Predator and Scavenger Species in Urban Environments*

Considering all the articles reviewed, we recorded a total of 100 predator species (found in 95 studies) and 49 scavengers (found in 47 studies). Some of these species are repeated, and appear in both sets of considered articles, given their opportunistic feeding habits that can lead them to both hunt and scavenge. Despite the greater number of species reported as predators, this may not necessarily be due to a greater presence of predators in cities, but rather to the fact that we only found around half the number of studies on scavenging compared to those on predation, thus reducing the possibility of detecting more species. This reinforces the need to increase research devoted to scavenging in cities. Life history, and ecological and behavioral traits predispose certain species to exploit urban environments [20,24,90,91], and the development of cities and associated impacts can limit or facilitate the successful exploitation of them by wildlife [92,93]. The reviewed studies do not offer sufficient data to analyze the influence of the elements associated with urbanization (i.e., noise, artificial light or human activities) on carnivorous occurrence. However, data extracted from the studies with ample information regarding predator and scavenger richness in cities suggest that the carnivore guild is reduced in urban environments, with an important range of variability. Further research is needed to gain knowledge of how certain characteristics inherent to the structure and environment of cities influence the occurrence of more carnivores. With regard to this, we suggest the inclusion of variables related to noise and acoustic pollution in studies related to urban carnivores, which are practically absent in the reviewed articles, despite their recognized impact on species exploiting cities [15–17]. In any case, we cannot forget that urban areas are normally characterized by lower biodiversity than wildlands [2,5–7], so it is expected that not all of the potential species inhabiting the surrounding areas of each city will be able to successfully exploit these human-dominated ecosystems.

Additionally, our results showed that the most encountered species are those with a highly meat-based diet, and, consequently, less consumption of vegetables or insects. Urban predators find a high availability of food associated with human environments (i.e., pets, pigeons and rats), which aids them in seeing that their trophic needs are easily met [36,94,95]. On the other hand, garbage constitutes an accessible resource which not only attracts obligate scavengers such as black vultures (*Coragyps atratus*) [30], but also seems to serve as a complement to the predators settled in the city. This is corroborated by the fact that most of the species detected in the scavenging studies also appeared on the predator list (Figure 2 and Tables 1 and 2). Regarding other traits pertaining to the carnivores recorded in the studies, most of the reported species showed nocturnal habits, coinciding with hours of less human activity. Thus, our results suggest that urban carnivores take advantage of periods of less human disturbance to satisfy their feeding needs. In addition, although the most frequent species in our review were a medium size (Figure 2), the selected models did not include the body mass variable, which may be attributed to our relatively small sample size.

Regarding taxonomic features, most of the species recorded in the articles are mammals and birds (mainly raptors), which may be explained by the general bias present in urban ecology towards these groups [60]. In the case of predators and scavengers, there were not many options beyond them; however, we still found some articles focused on Asian monitor lizards [96,97], and others that include these reptiles as scavengers among other species [94]. Despite that, the selected models did not include any taxonomic classification, but the model that included the class variable proved to be better than the null

model, being very close to the threshold of 2 points of AICc. This, added to the fact that the presence among the main species detected in both cases includes mammals such as the red fox, and the domestic dog and cat, makes us suspect that there may indeed be a taxonomic tendency among carnivores to adapt to cities. However, available data in the reviewed articles were not sufficient to demonstrate this.

#### *4.4. Conservation Implications*

Regarding our results, we consider that it is necessary to devote more effort to assessing the role that cities may play for scavengers, in order to obtain a more complete understanding of the ecological and conservation implications which may result from the urban predatory and scavenging habits of wildlife [11,33]. In this sense, from a conservation point of view, species management strategies should consider the role of cities for predators and scavengers, due to the short- and long-term influence (both positive and negative) that their presence in cities and the consumption of living prey and refuse may have, at both individual and population levels [28–30,45]. Due to natural habitat loss, the probability of larger mammals colonizing urban environments may increase, as evidenced by the anecdotal but growing data shown for leopards (*Panthera pardus*), tigers (*Panthera tigris*) and lions (*Panthera leo persica*) in India [98–100]. In this sense, the increase in the number of medium- to large-size predators feeding in cities may lead to an increase of potential human–wildlife conflicts, which will ultimately call for a better understanding of their presence in humanized areas. Furthermore, cities should be treated as other habitats in conservation projects relating to endangered carnivores, taking into account that they undoubtedly use this type of ecosystem. Thus, ignoring the pros and cons that result from the species' exploitation of urban environments would be negligent.

Lastly, we must reflect on the multiple interpretations from the nature conservationist point of view that the prominent presence of cats and dogs in both reviews offers. First, the predominance of the same species among the most recorded predators and scavengers is a logical consequence of their close association to humans, who have aided in their expansion wherever they have settled [101,102]. Beyond their presence, growing research shows how these species have a negative impact on native wildlife, thereby converting themselves into conflictive invasive species [103,104]. In the case of cities, the presence of cats and dogs (domestic or feral) may increase spatial-use conflicts with wildlife in urban green areas [105,106], also affecting native mammal, bird, reptile and amphibian communities by predation [107]. Finally, the high densities of these species can threaten public health by increasing the prevalence of diseases such as toxoplasma and rabies [108,109].
