Assessing the impact of free-ranging cats (Felis silvestris catus) and dogs (Canis lupus familiaris) on wildlife in a natural urban reserve in Mexico City

Conservation of natural ecosystems embedded in urban contexts is a big challenge because diverse anthropogenic factors continuously impact native biodiversity and ecological processes. One such factor is the pervasive presence of free-ranging predators, especially cats (Felis silvestris catus) and dogs (Canis lupus familiaris), which may severely affect local wildlife. The “Reserva Ecológica del Pedregal de San Ángel” (REPSA), located within the main campus of the Universidad Nacional Autónoma de México (UNAM), in southern Mexico City, is an important natural area that protects a peculiar volcanic spill ecosystem holding native and endemic biodiversity. In 2012, UNAM launched a control program of cats and dogs in REPSA that is still active. To assess the potential impact of cats and dogs on wildlife species, we used live and camera traps coupled with capture-recapture analyses to evaluate changes in the vertebrate community, particularly in the presence and abundance of two medium-size native mammals (Didelphis virginiana and Bassariscus astutus), before (2008–2009) and during (2017–2019) the control program. Results showed that the abundance of dogs decreased between the two periods, but not so the abundance of cats, whereas the native vertebrate diversity increased from the pre-control stage to the control period. Furthermore, we found a negative, non-significant relationship between the abundance of D. virginiana and that of dogs, and a positive, also non-significant relationship between the presence and abundance of B. astutus with the abundance of dogs. We conclude that the control program of free-ranging predators has been beneficial for the conservation of native vertebrates and recommend its continuation and enhancement.


Introduction
The establishment of protected areas that remain relatively unaltered has been one strategy for conserving local biodiversity in urban areas (Mcdonald et al. 2008;Borgström et al. 2013).However, these areas have enormous complexities compared to protected areas in the wild because they present numerous theoretical and practical challenges that imply ecologic, economic, political, social, and cultural pressures that affect the conservation of local biodiversity (Terradas et al. 2011;Gaertner et al. 2017).One of the most conspicuous and detrimental impacts of urban contexts upon native biodiversity is biotic homogenization, due to a combined effect of native species loss and the introduction of exotic species owed to altered natural conditions and human facilitation (McNeely et al. 2001;McKinney 2006;Cadotte et al. 2017).
Urban protected areas are continually threatened with human invasion because they represent attractive areas for urban developments.Even if armored against land-use conversion, urban growth frequently isolates them from other natural areas.Thus, populations of local biota become gradually reduced, increasing their extinction risk (Mcdonald et al. 2008).Furthermore, they are directly and indirectly affected by several anthropogenic stresses and the invasion of accompanying fauna, including dogs, cats, mice, and rats, among others (Miller and Hobbs 2002;Gaertner et al. 2017).
Negative impacts of domestic cats (Felis silvestris catus) and dogs (Canis lupus familiaris) in their diverse categories of human dependency (owned pets, unowned, feral; depending on the author) are well-known in protected and urban areas.These include competition and predation upon local fauna, ecological disturbance, disease transmission, hybridization, public health issues, environmental contamination, animal welfare, and wildlife conservation (Tasker 2007;Baker et al. 2010;Doherty and Ritchie 2017;Loss and Marra 2017).For example, feral cats dwelling islands represent a threat to local biota since they have caused or contributed to recent bird, mammal, and reptile extinctions (Nogales et al. 2004;Medina et al. 2011Medina et al. , 2014)).Also, in mainland, invasive predators have also reduced or impacted vertebrate populations worldwide (Loss et al. 2013;Loss and Marra 2017), being wild mammals one of the most affected taxonomic groups since 96 species have gone extinct or are currently threatened with extinction due to the impact of domestic dogs (Doherty et al. 2017).
Conflicts between exotic and native fauna may become more complicated and intense when it involves charismatic pet species, like domestic cats and dogs, because divergent opinions emerge between stakeholders from the animal rights and biodiversity conservation communities (Loss et al. 2013;McDonald et al. 2015;Gaertner et al. 2017;Loss and Marra 2017).Moreover, these conflicts become exacerbated in areas where the interaction between human activities and wildlife is more intense, such as urban protected areas (Vázquez 2016).Although it is well documented in the scientific literature the damage that cats and dogs infringe on wildlife, the appropriateness and efficacy of the management schemes used to control these predators are still in debate even within the scientific community, particularly in urban contexts (Loyd and DeVore 2010;Loss andMarra 2017, 2018;Munro et al. 2019).The options offered include: (1) the permanent removal of cats and dogs via euthanasia (Trap-Euthanize or TE); (2) capture, sterilization, vaccination, and return of cats/dogs to their colonies (Trap-Neuter-Return or TNR), or the implementation of adoption programs (Hughes and Slater 2002;Longcore et al. 2009;Loyd and DeVore 2010).
The efficacy of methods and interests of dissenting points of view about the rights of pets and wildlife are controversial management issues in many countries.Therefore, it is necessary to carry out studies about the impacts of control programs of cats and dogs on wildlife to help wildlife managers, researchers, decision-makers, and any other stakeholders to improve management decisions.Consequently, the main goal of this study was to assess the changes in the presence of vertebrates and the abundance of medium-sized mammals before (2008)(2009) and during (2017-2019) the implementation of a control program of cats and dogs in the "Reserva Ecológica del Pedregal de San Ángel" (henceforth REPSA), a protected area within Mexico City, the fifth largest city in the world (United Nations 2018).In this contribution, we refer to cats and dogs roaming in the REPSA as free-ranging, which includes feral, stray, and house organisms (Spotte 2014), because REPSA is an open area with a continuous entrance and settlement of individuals from different origins and ages.

Study area
The xerophilous shrub ecosystem is one of the remaining natural areas within the urban area of Mexico City and originated from the eruption of the Xitle volcano 1670 ± 35 years ago (Siebe 2009).In the 1950s, the construction of roads, residential areas, and settlement of the new campus of the Universidad Nacional Autónoma de Mexico (UNAM) accelerated its population growth, reducing the volcano spill ecosystem (Lot and Cano-Santana 2009).In 1983, UNAM created the REPSA to protect the xerophilous shrub, currently covering 237.3 ha (Castillo Argüero et al. 2007).REPSA is in southern Mexico City, at 19°17' N, 99°11' W. The climate is temperate sub-humid, with rainfalls concentrated in the summer (June-October).The annual mean temperature is 15.6 ºC, and the average annual rainfall is 833 mm (Castillo Argüero et al. 2007).
REPSA covers three core and 13 buffer areas.We carried out fieldwork in the two largest cores: the Western Core Area (henceforth WCA) with two buffer areas (A10 & A11), covering 114.42 ha; and the Eastern Core Area (henceforth ECA), which covers 52.44 ha (Zambrano et al. 2016).In each area, we set live traps in a grid arrangement with points separated every 200 m, resulting in 26 points for the WCA and 11 for the ECA (Fig. 1).

Free-ranging cats and dogs
Unfortunately, urban growth and people's behavior harm biodiversity in REPSA.For instance, visitors commonly abandon dogs or cats therein, where they frequently grow up and reproduce.Free-ranging dogs and cats attack or predate upon wildlife (Granados Pérez 2008;Ramos Rendón 2010;Zambrano et al. 2016).Also, at least 35 zoonotic diseases circulate in the area, representing a risk for visitors and the native fauna (Arenas Pérez 2016;Zambrano et al. 2016).
To face these problems, in 2012, the Executive Secretary of REPSA (SEREPSA) and the Facultad de Medicina Veterinaria y Zootecnia (Vet School) launched a program for the control of free-ranging cats and dogs that continues until the publication of this work (Zambrano et al. 2016;Ramírez Velázquez 2017).In this study, we estimated the number of cats and dogs before and during the control program using three sources of data: (i) We carried out capture campaigns of cats and dogs.Captured animals were not released, following the control program protocol.(ii) We set camera traps to identify individuals by color, size, or other physical characteristics.(iii) We gathered and analyzed the data of recorded or captured individuals provided by the SEREPSA staff.With this information, we estimated the density of cats and dogs in each period and zone (WCA and ECA).We performed a paired t-student test for each species and zone mango, or papaya) and salty food (commercial food for cats, tuna, sardine, chicken, beef, or pork) (Castellanos Morales 2006;García Peña 2007;Ramos Rendón 2010).Sampling was carried out monthly during one year in the pre-control and bimonthly for two years in the during-control; therefore, we conducted 12 sampling sessions in both periods.We opened the traps for three consecutive nights, activating them in the afternoon and checking them in the morning.We covered the traps with fabric for camouflaging them and protecting the animals from the sun or rain.We marked all captured animals with a tattoo and earrings in the pre-control and with intradermic AVID® Microchips ID and sometimes with earrings in the during-control.Handling live organisms consisted of immobilizing them physically, in the case of opossums (Fig. S1, Supplementary Information), or chemically with a combination of Ketamine/Xylazine for carnivores, like ringtails, southern spotted skunks, and cats.For the rock squirrel (Otospermophilus variegatus), we used a combination of Ketamine/Dexmetomidine (Fig. S2, Supplementary Information).Animal handling was always carried out by a professional veterinarian or under his/her supervision.We recorded the conventional morphometric measures for each captured individual, in addition to sex, weight, and a clinical examination.We conducted these procedures using the software PAST 3.25 (Hammer et al. 2001), after testing for normality via a Shapiro-Wilk test, to compare density estimates between periods.

Native wildlife
We carried out field sampling to record the presence and estimate the abundance of native wildlife species in two periods, before the control program started (2008)(2009) and during the control (2017-2019), "pre-control" and "during-control," from now on, respectively.
The study area has a complex topography; therefore, to cover all the grid points, it was necessary to apply two sampling methods, namely direct capture and camera trapping, which were analyzed independently given their capture capacity.We analyzed the results at the species level (only mammals) with the first method, while with the second method, we compared the results at the community level.The sampling scheme (Fig. 2) is described below.
Direct capture.To capture medium-sized mammals in REPSA, we set 19 traps in WCA and eight in ECA.We used Tomahawk® live traps (32" L×10" W×12" H and 32" L×10" W×12" H) baited with a mix of sweet (bread with strawberry jam and fruit, such as banana, apple, pear, the relationship between the density of dogs versus the relative abundance of opossums and ringtails via a multivariate linear regression using the 12 monitoring periods.We performed all statistical analyses with PAST 3.25 (Hammer et al. 2001).
Finally, to assess changes in the species diversity between the pre-control and during-control seasons, we used the effective number of species (ENS; Jost 2006), which is suitable for a small sample size (Jost 2006;Chao et al. 2014a).
In ENS, we used two values of order q = 0 and q = 1, where q represents the importance of species' abundance in the calculations of diversity.Therefore, q = 0 does not include abundance and refers only to species richness, and q = 1 includes all species in the proportion of their abundance (Hill 1973;Jost 2006).We carried out the diversity analysis with the iNEXT package in R (Chao et al. 2014b;Hsieh et al. 2016), with 84% confidence intervals (MacGregor-Fors and Payton 2013).

Results
The sampling effort was different for the two seasons.For pre-control, the sampling effort for live traps was 291 and 126 trap nights for WCA and ECA, respectively, whereas for during-control, it was 342 and 144 trap nights, respectively (Table 1).For the WCA, we found significant differences between years in capture success (Kruskal-Wallis: H = 12.13; p = 0.002; d.f.=3; pairwise Mann-Whitney: d.f.=12; p < 0.05), but we did not find significant differences for ECA (Kruskal-Wallis: H = 0.286; d.f.=3; p = 0.866; pairwise Mann-Whitney: d.f.=12; p > 0.05).Capture effort also increased for camera traps from pre-control to duringcontrol, with 387 trap nights in WCA and 135 in ECA for the former, and 532 and 836, respectively, for the latter.
We recorded 17 native species with both methods in the two sampling periods and core areas, including eight mammals, seven birds, and two reptiles (Table 2).For the ECA, we recorded four species with the Tomahawk traps in precontrol and only three in during-control.Nevertheless, we recorded seven in pre-control and 12 in during-control with the camera traps.In total, we recorded eight species in precontrol and 12 in during-control with both methods in this zone.It is important to note that we captured the southern following the guidelines for capture, handling and physical or chemical immobilization of wild mammals (Kreeger et al. 2002; Sikes and the Animal Care and Use Committee of the American Society of Mammalogists 2016).
Camera trapping.In the pre-control, we used analog and digital camera traps (CT) "StealthCam."We set five camera traps in the 11 most difficult points to reach (eight in WCA and three in ECA), shifting them between WCA and ECA every 15-20 days.In the during-control, we set 4-6 cameras (Bushnell Trophy Camera Brown® and Ltl Acorn Hunting Camera Llt 5210 A Series®) following almost the same accommodation as in the pre-control (seven in WCA and three in ECA), with the difference that the cameras were shifting between locations every month (Fig. S3, Supplementary Information).The cameras were programmed to shoot three photos in each event with a minimum of three minutes between detection events.Cameras were attached to trees or shrubs at 80-120 cm from the ground (Rovero and Zimmermann 2016) and baited with tuna fish and a water container.
Data organization and analysis.We identified live-and camera-trapped individuals using scientific literature and local guides (Hortelano-Moncada et al. 2009;Berlanga et al. 2012;Balderas Valdivia et al. 2014).We organized the information in a database, including a unique Id, date of capture (and recaptures, if any), and body measures in the case of live-trapped individuals.For the camera trap data, we considered "independent" records the pictures of individuals of the same species separated by 24 h (Rovero and Zimmermann 2016).In both cases, we calculated the capture effort (total number trap nights) and capture success (total captures/total nights) in each period and area (WCA and ECA) and compared the capture success between periods with a non-parametric ANOVA (Kruskal-Wallis test) and a pairwise Mann-Whitney test.
To assess if mammal abundance changed before and during the control program, we calculated the relative abundance of each species as the number of individuals caught divided by the sampling effort (trap nights) (Cruz-Salazar et al. 2014;Freeman and Beehler 2018).In this way, we considered the potential bias due to the different sampling effort between periods.Then, we tested for normality using the Shapiro-Wilk test and performed a paired t-student test per species and zone (WCA and ECA).Finally, we evaluated

Spilogale angustifrons
Southern Spotted Skunk 1 3 One of the main challenges during the fieldwork was capturing or detecting dogs and cats with both live and camera traps, so we used complementary information provided by SEREPSA.We identified all dog individuals captured or photographed, so we have a complete census of the dog population.Based on this information, we found a significant decrease in dogs between the pre-control and duringcontrol seasons (t = 4.48; d.f.=12; p = 0.046).The estimated density for pre-control was between 0.29 and 0.31 ind/ha for ECA and 0.13-0.14ind/ha for WCA.In during-control, it was between 0.01 and 0.08 ind/ha for ECA and 0.02-0.17ind/ha for WCA (Fig. 3).Conversely, for cats, the density did not show a significant change between periods (t = 0.83; d.f.=12; p = 0.55).Only for the last year (2019) we did not record cats in ECA (Fig. 3).
The analysis of relative abundance (expressed as the number of individuals caught divided by the sampling spotted skunk in both sampling seasons and recorded this species for the first time in a camera trap in during-control in this core area.Additionally, we had the first record of a gray fox in a camera trap (July 2017) after ten years of not registering this species in REPSA.
For the WCA, we captured four species in Tomahawk traps in pre-control and five in during-control, but we recorded seven and 11 species, respectively, with the camera traps.In total, we recorded eight species in pre-control and 12 in during-control with both methods in this zone (Table 2).Incidental captures in Tomahawk traps included a Mexican woodrat (Neotoma mexicana) and a northern black-tailed rattlesnake (Crotalus molossus).The species more frequently captured in live traps were the opossum (> 75% of captures), followed by the ringtail (8-17%; Table 3).Finally, the abundance and effective number of species of mammals (Fig. 4) and vertebrates (Fig. 5) increased from pre-control to during-control, according to the diversity values q = 0 and q = 1.

Discussion
Free-ranging domestic carnivores negatively impact local biotas (Young et al. 2011;Doherty et al. 2016).Previous studies in REPSA have documented the impact of freeranging dogs and cats on native wildlife via diet analyses, disease transmission, or accidents (Granados Pérez 2008; Ramos Rendón 2010; Arenas Pérez 2016;Zambrano et al. 2016).However, this is the first study specifically designed to assess the effectiveness of a control program for cats and dogs over the local fauna.REPSA implemented the control program in 2012, and according to our data, it has been effort) between pre-control and during-control obtained with live traps was only possible for two mammal species with enough data in both periods, namely, the opossum and the ringtail.Our results indicated a significant increase of opossum relative abundance (t=-4.435,d.f.= 11; p = 0.001) from pre-control (0.123) to during-control (0.260) only in WCA after domestic species control.Ringtail relative abundance did not show significant changes between precontrol (0.026) and during-control (0.023) in WCA or ECA (t = 0.265; d.f.= 11; p = 0.796).The multivariate linear regression indicated a non-significant relationship between the relative abundance of D. virginiana and B. astutus with the density of dogs (R 2 = 0.142; d.f.= d.f. 1 =2, d.f. 2 =9; p = 0.221).Individually, we found an inverse, non-significant relationship between the density of dogs and the relative abundance of D. virginiana (r=-0.363;d.f. 1 =2, d.f. 2 =9; p = 0.282), and a positive one, also non-significant, with B. astutus (r = 0.382, d.f. 1 =2, d.f. 2 =9; Fig. 4 Comparison of diversity values (q) for mammal species recorded in camera traps before and during control of freeranging dogs and cats.Shaded areas represent ± 80% CI.For q = 0 small differences were detected between both sampling periods, for q = 1 the change was significant substantially increased worldwide, demonstrating that this is another expression of environmental degradation related to global change (Medina et al. 2011;Young et al. 2011;Duffy and Capece 2012;Hughes and Macdonald 2013;Loss et al. 2013;Doherty et al. 2015;Hughes et al. 2016;Loss and Marra 2017;Gil Alarcón et al. 2018).For example, free-ranging cats have contributed to at least 63 vertebrate extinctions mainly on islands worldwide; therefore, these predators are highly harmful to vertebrates, particularly to native birds (Wi et al. 2013;Loss and Marra 2017).
In REPSA, although the program has been successful in controlling dogs, it has not been so much for cats (Coronel-Arellano et al. 2020).In fact, according to our results, the number of cats has maintained relatively constant during the control program, mainly because it is far more challenging to detect and trap cats than dogs in the rough topography of REPSA, where plenty of refuges and escape routes exist for them in the lava flow.Also, although pet-lovers at the campus protect and constantly feed free-ranging cats and dogs, cats do not depend on human supplementation or waste since they are efficient predators with abundant food sources, including lizards, birds, small mammals, and beneficial for local vertebrate populations.We observed that the vertebrate diversity increased during the control program, recording a higher relative abundance of some species, like opossums, and the presence of bird and mammal species after years of absence, notably the gray fox (Urocyon cinereoargenteus), which was apparently absent for more than a decade and it has been recently recorded (Proceso 2017; Coronel-Arellano et al. 2020).
There is a growing interest to study the impact of introduced species in urban protected areas and green areas worldwide, and Mexico is not an exception (Ramírez-Cruz et al. 2018;Mella-Méndez et al. 2019;Zúñiga-Vega et al. 2019).However, studies about the impact of cats and dogs are still scarce in the country (Elizondo and Loss 2016).Nonetheless, some studies in other countries have found high rates of free-ranging dog occupancy and adverse effects on native species (Young et al. 2011;Hughes and Macdonald 2013;Loss et al. 2013;Doherty and Ritchie 2017;Morin et al. 2018).
In the last years, the number of studies assessing the impact of domestic cats and dogs with different degrees of "wilderness" upon local biota, mainly vertebrates, have Fig. 5 Comparison of diversity values (q) for vertebrate species recorded in camera traps before and during control of free-ranging dogs and cats.Shaded areas represent ± 80% CI.For q = 0 and q = 1 the change was statistically significant 2002); and 0.02-0.11ind/ha in the Serengeti Ecological Region, northwestern Tanzania (Lembo et al. 2008).
Regarding the local fauna, we recorded eight mammal species, where the most abundant was the opossum, followed by the ringtail and the southern spotted skunk, the latter with low numbers in both seasons.We recorded all eight species with the camera traps and only six with the live traps (Table 2).This result largely coincides with a recent study based only on camera trapping in a different portion of REPSA, where they recorded five native mammals, and also dogs and cats (Coronel-Arellano et al. 2020).Interestingly, in our results, the southern spotted skunk was the only species with a higher number of records before the control program (six captures) but currently is one of the most difficult mammals to detect, though it occurs in low numbers in other natural systems too (Farías-González and Vega-Flores 2019).However, it is important to note that we followed two sampling schemes with the same total number of sampling sessions in the pre-control and during-control phases, being more intensive in the former (monthly samplings for one year) and more extensive in the later (bimonthly samplings for two years).It is possible that this imbalance may introduce some bias if the capture/detection of species behaves non-linearly with increases of effort or time, so we reduced this effect by conducting our analysis taking into account the trapping effort.This potential bias, if any, is minimal and do not affect our results, as observed in the equal number of species detected before and during the control program (except for the special case of the gray fox), and also in the fact that the accumulation curves reached the asymptote in the two sampling periods (Figs. 4 and 5).
It is noteworthy to point out that density estimations with camera traps depend on the assumptions and parameters used in the model, particularly the time lapse between records of the same species in the same camera (Rovero and Zimmermann 2016).People have used different temporal thresholds, ranging from 30 min, an hour to 24 h (Sollmann 2018);however, Royle et al. (2009) considered that multiple records in the same day may not represent independent detections, and if this is the case, density would be overestimated.Here, we considered "independent" records the pictures of the same species separated for at least 24 h because the only study regarding the use of space of the ringtails found that the average home range size in REPSA is among the smallest reported in the literature (Castellanos Morales 2006), so the probability of detecting the same individual increases with shorter time lapses.Sadly, there is no information on the movement patterns of opossums in REPSA to make an informed decision, so we decided to be conservative and use the 24-hour lapse acknowledging that this decision may underestimate the true density of this species if individuals hold large, overlapped home ranges.However, even insects (Granados Pérez 2008;Ramos Rendón 2010;Ramírez Velázquez 2017;Negrete-González 2020).
Furthermore, the impact of free-ranging cats and dogs upon native fauna is not only as predators; these species represent a focus of diseases that affect local wildlife and people (Yoak et al. 2016).Indeed, different studies have confirmed the presence of disease agents shared between wild and feral fauna in REPSA.For instance, wild fauna presents a high antibody response to three infectious agents, namely parvovirus, toxoplasma, and rabies, most likely due to the high density of cats and dogs before the control program (Suzán and Ceballos 2005;Pacheco Coronel 2010;Arenas Pérez 2016).In another study, Pacheco Coronel (2010) demonstrated the presence of the fleas Ctenocephalides felis and Echidnophaga gallinacea in both feral and wild animals.In addition, he also found the zoonotic parasites Toxoplasma gondii, Dipylidium caninum, Ancyclostoma caninum, Toxocara cati, and T. canis.Likewise, another study detected the presence of tick and flea-borne pathogens (Bartonella vinsonii subsp.berkhoffii, Ehrlichia canis, and Mycoplasma haemocanis) (Arenas et al. 2019), which represent a public health problem since they cause human diseases, as well as endocarditis and hemolytic anemia in domestic dogs; however, the effects on wildlife are still unknown.Yet, circulation of pathogens within and outside REPSA is bidirectional between introduced and native and fauna since some local species, particularly opossums and ringtails, are frequently observed around houses in the city and feeding from pet trays, so they may also play a role in disease prevalence.
In general, the impact of invasive predators in an ecosystem and the prevalence of imported diseases depends primarily on their abundance rather than their sole presence (Butler et al. 2004;Doherty et al. 2016Doherty et al. , 2017;;Home et al. 2018), hence the relevance of the techniques used to estimate population size (Belo et al. 2015).In this study, we used two complementary sampling methods in the two largest core areas of REPSA, namely live and camera trapping.As mentioned before, monitoring cats was challenging because they occur at low densities, are cryptic, and tend to avoid people (Short and Turner 2005;Fisher et al. 2015).Thus, we found it difficult to trap them, leading to low capture rates and, therefore, to population size estimations with high uncertainty, as has occurred in similar efforts to assess the effectiveness of control programs (Witmer et al. 2005;Fisher et al. 2015;Comer et al. 2018).In contrast, dog density estimations were reliable, ranging from 0.01 to 0.31 ind/ha, depending on the sampling zone and period, which is comparable to the numbers reported in different regions by Hughes and Macdonald (2013), including: 0.04-0.1 ind/ ha in the Bale Mountains National Park, Ethiopia (Atickem et al. 2009); 0.07 ind/ha in the southern part of the Chirisa Safari Area, in northwestern Zimbabwe (Butler and du Toit nothing about the direct effect of them and their interaction in the dynamics of local faunas in REPSA.Therefore, we recommend developing a more integrated management program conformed by an interdisciplinary team of researchers and students.This program should include a regular monitoring scheme of introduced predators and native fauna carried out by biologists and veterinarians, including diet analyses, light and noise pollution to measure their impact on local fauna, but also a robust component of environmental education and awareness campaigns led by social scientists, aiming to preserve one of the very last refuges for wildlife within one of the largest megalopolis in the world.
the ultimate goal of our study was to compare species' densities before and during the control program, so since we used the same criterion in the two phases of the study, this comparison is not affected.
Although we obtained enough data to quantify the effectiveness of the control program only for opossums and ringtails, the population reduction of the top predators has probably benefited other species as well.For instance, the gray fox is currently the most threatened medium-size mammal in REPSA despite being relatively common in the WCA more than a decade ago (Negrete Yankelevich 1991;Castellanos Morales 2006;García Peña 2007;Hortelano-Moncada et al. 2009).Unfortunately, we did not record a single individual in our first sampling period (before the control program) and only one individual in 2018.Based mainly on anecdotic evidence, some biologists believe that the increase of free-ranging dogs drove it to its extirpation due to direct attacks and disease transmission (García Peña 2007;Pacheco Coronel 2010).Other factors may also have impacted the gray fox and other mammals, like habitat destruction due to the expansion of buildings and the use of fences in some areas of the campus that reduce mobility and connectivity, and car accidents, among others.Conversely, the arrival of the single individual recorded in 2018 is unknown.It was possibly informally reintroduced, or it could have arrived by itself from nearby semi-natural areas where the presence of this species is confirmed, for example, the Bosque de Tlalpan park (Negrete-González 2020).Regardless of the way this individual arrived, the fact is that it survived for at least a couple of years (Coronel-Arellano et al. 2020) until it was hit by a car in 2020, confirming that REPSA holds suitable conditions in terms of resource availability and low density of dogs to launch a reintroduction program for the gray fox and possible other extirpated species.On the other hand, the most significant impact of cats is probably not on medium-size mammals but small mammals, reptiles, and birds (Phillips et al. 2007;Medina et al. 2011;Hernandez et al. 2013;Loss et al. 2013;Loss and Marra 2017;Ortiz-Alcaraz et al. 2017), but we require further studies specifically designed to quantify the impacts on these communities.
In conclusion, our results confirm that the control program has positive results for the native biota of REPSA and needs to be maintained and enhanced.Although, the damage that free-range cats or dogs can cause to birds was not directly measured or observed, we provide indirect evidence of the potential damage that introduced predators might have over local fauna.It is important to keep in mind that the abundance of exotic fauna is one among several other disturbing factors that impact native faunas in urban ecosystems, including night light and noise pollution, to mention a couple (Barber et al. 2010;Hölker et al. 2010).We know

Fig. 1
Fig. 1 Location the sampling points in the Eastern (ECA) and Western (WCA) Core Areas of the REPSA.

Fig. 2
Fig. 2 Diagram of the sampling scheme and data analysis

Fig. 3
Fig. 3 Density (individuals per hectare) of cats and dogs in the Eastern (ECA) and Western Core Areas (WCA) of REPSA in different sampling periods

Table 1
Capture effort (number of trap nights) with Tomahawk live traps (TH) for the Eastern (ECA) and Western Core Areas (WCA) in the precontrol and during-control periods of free-ranging dogs and cats

Table 2
Species captured with Tomahawk live traps (TH) and camera traps (CT) for the Eastern (ECA) and Western Core Areas (WCA) in the pre-control and during-control periods of free-ranging

Table 3
Percentage of captured species with Tomahawk live traps (TH) for the Eastern (ECA) and Western Core Areas (WCA) in the pre-control and during-control periods of free-ranging dogs and cats.Ba = Bassariscus astutus; Dv = Didelphis virginiana; Ov = Otospermophilus variegatus; Sa = Spilogale angustifrons; n = number of captures