Received: 15 May 2020 Revised: 27 March 2021 Accepted: 26 May 2021

DOI: 10.1111/csp2.477

CONTRIBUTED PAPER

Deforestation, fires, and lack of governance are displacing

thousands of jaguars in Brazilian Amazon

Jorge F. S. Menezes1,2 | Fernando R. Tortato3 | Luiz G. R. Oliveira-Santos1 |

Fabio O. Roque1 | Ronaldo G. Morato2

1
Departamento de Ecologia, Instituto de
Biologia, Universidade Federal do Mato
Abstract
Grosso do Sul, Campo Grande, MS, Brazil The rate of deforestation and the number of large wildfires are increasing in
2
Centro Nacional de Pesquisa e the Amazon. Illegal loggers, miners, ranchers, and farmers all have contrib-
Conservaç~ao de Mamíferos Carnívoros,
uted to this increase. Their activities have dramatic consequences for biodiver-
Instituto Chico Mendes de Conservaç~ao
da Biodiversidade, Estrada Municipal sity, ecological services, and people. In this study, we estimated the number of
Hisaichi Takebayashi, S~ao Paulo, Brazil jaguars affected by deforestation. We focused on the Brazilian Amazon from
3
Panthera, New York, New York August 2016 to December 2019. Further, we analyzed the effects of socio-
Correspondence
geographic determinants of deforestation and state policies. To do so, we used
Jorge F. S. Menezes, Departamento de deforestation data from DETER-B satellite system. The number of jaguars
Ecologia, Instituto de Biologia, within each deforested area was pulled from a previous study, which provided
Universidade Federal do Mato Grosso do
Sul., Av. Costa e Silva s/n , Bairro jaguar abundances for jaguar entire range. We assumed all jaguars within a
Universitario, 79070-900, Campo Grande, deforested area were affected (displaced or killed). To determine the underly-
MS, Brazil.
ing causes of jaguar loss, we regressed the number of jaguars lost per state and
Email: jorgefernandosaraiva@gmail.com
year against the proportion of total forest area within reserves, distance to for-
Funding information est border, and monetary efficiency in cattle production. We estimate a total of
Conselho Nacional de Desenvolvimento
Científico e Tecnologico, Grant/Award
1,422 jaguars have been displaced/killed in recent years (2016: 488, 2017:
Number: 152160/2018-3; Coordenaç~ao de 360, 2018: 268, 2019: 354). Only the proportion of protected area had an effect
Aperfeiçoamento de Pessoal de Nível in reducing jaguar deforestation. We discuss how our work could result in near
Superior, Grant/Award Number:
59/300.135; Fundaç~ao de Amparo à real-time monitoring of jaguar displacement and how policies such as wood
Pesquisa do Estado de S~ao Paulo, Grant/ certification, more efficient cattle production, and centralizing governance
Award Number: 2017/08461-8
may be solutions.

KEYWORDS
conservation, deforestation, DETER-B, fires, Panthera onca

1 | INTRODUCTION event was unprecedented in its coordination. Arsonists

used social apps to synchronize their acts in a single day:
In 2019, the Brazilian Amazon was highlighted on inter- the “day of fire” (Eisenhammer, 2019). This day repre-
national media due to a series of coordinated arsons draw- sents a spike in deforestation in the Amazon forest in its
ing focus to the destruction of the forest by illegal loggers, most conspicuous form, forest fires. However, deforesta-
miners, ranchers, and farmers (Eisenhammer, 2019). This tion itself is not new. The deforestation rate is increasing

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided
the original work is properly cited.
© 2021 The Authors. Conservation Science and Practice published by Wiley Periodicals LLC. on behalf of Society for Conservation Biology

Conservation Science and Practice. 2021;3:e477. wileyonlinelibrary.com/journal/csp2 1 of 10

https://doi.org/10.1111/csp2.477
2 of 10 MENEZES ET AL.

(Diniz et al., 2015) and prediction models suggest an Despite their clear importance to multiple ecological
increase in the coming years (Brando et al., 2020). The services and their use as indicator of environmental
Amazon harbors some of the richest biodiversity on Earth integrity, jaguars are directly threatened by deforestation
(Jenkins, Pimm, & Joppa, 2013). It also serves as a mas- (Romero-Muñoz, Morato, Tortato, & Kuemmerle, 2020).
sive source of heat and moisture in the tropics and plays a The issue is often compounded by deforestation often uti-
crucial role in the global climate system (Werth & lized as open space for cattle ranching and agriculture.
Avissar, 2002). Moreover, it houses 33 million people and Jaguars often hunt cattle (Carvalho, Zarco-Gonzalez,
about 420 indigenous communities that rely directly on Monroy-Vilchis, & Morato, 2015), a behavior that triggers
its resources for sustaining their livelihoods (FAO, n.d.). retaliation hunting by cattle ranchers (Marchini &
Thus, the current deforestation is extraordinarily con- Macdonald, 2012). Thus, both directly and indirectly, the
cerning for the planet's life support system. survival of this species is dependent on the presence of
While there is a clear effect of recent deforestation forest in the Amazon region.
and fires on the landscape, their effects on animals It is essential to understand how fires and changes in
remain difficult to quantify due to the lack of population land use influence the jaguar population in Amazon
densities estimates for most species. An indirect estima- region. To identify strategies for reducing negative
tion can be reached by using jaguars (Panthera onca) as a human impacts on jaguar populations, we estimated the
surrogate for biodiversity. This species is heavily depen- number of jaguars that occurred in deforested areas dur-
dent on forested landscape, showing a strong avoidance ing the period 2016–2019. We assume those animals were
of non-forested areas (Morato et al., 2018) and a world- either displaced or killed by this deforestation, although
wide abundance estimate for this species is available, all- the exact fate is not known. Given that jaguar density is
owing us to measure the likely impact of deforestation on directly associated with forest cover (Jȩdrzejewski
it (Jȩdrzejewski et al., 2018). et al., 2018), we expected that jaguar displacement/killing
Jaguars are also associated with many ecological ser- would increase in 2019. We also analyzed this trend state
vices (Thornton et al., 2015), considered a keystone spe- by state and year by year, to obtain a more detailed pic-
cies (Boron et al., 2016). Jaguars have been shown to ture of the process.
influence understory biomass, with great reduction in Likewise, it is important to understand how fires and
biomass observed when jaguars (and other large preda- deforestation influence jaguar displacement/killing. We
tors) are absent (Estes et al., 2011). Other studies chal- were interested in three social and geographic variables:
lenged this definition of keystone species, showing that distance to the forest border, the proportion of protected
the removal of some jaguars does not cause intense forest, and cattle production efficiency. The first variable
changes in the environment when its population is high is due to an effect known as the Arch of Deforestation
(Wright et al., 1994). Nonetheless, both beforementioned (Becker, 2005). Deforestation in the Amazon seems to
studies agree that the presence of jaguars, even if in low progress from South to North. One of the reasons for that
numbers, is necessary for conserving the environment in phenomena is likely logistic access. It is harder to access
its functional state. areas in the center of the Amazon than on its border.
This species also has great value to conservation plan- Thus, we would expect distance to any forest border to
ning since they can be considered an umbrella species. decrease deforestation and hence jaguar displacement/
Due to its wide ranging behavior, its movement patterns killing. The second variable was chosen based on the
are useful in predicting other species movements and assumption that protected areas prevent deforestation.
therefore is important for locating ideal areas for forest We expect states with most of their forest protected
corridors (Thornton et al., 2015). Tracked jaguars also would have little to no deforestation and thus little jaguar
have been used to identify stepping stones between good displacement/killing. For the third variable we expected
patches of habitat in a fragmented matrix (Cullen Jr, an increase in production efficiency in cattle farms (mea-
Abreu, Sana, & Nava, 2005). Jaguars are also considered sured as yearly monetary profit over monetary costs)
flagship species due to their large size and iconic appear- would decrease jaguar displacement/killing. A study on
ance that allows identification by the public (Caro, deforestation in Cameroon (Epule, Peng, Lepage, &
Engilis, Fitzherbert, & Gardner, 2004). Finally, jaguars Chen, 2014) suggests increasing farmland throughput
have been used to calibrate corridor selection models could reduce the pressure on farms to utilize all land
based on their movements (Zeller et al., 2011). Thus, the available, creating space for protected areas. This release
protection of this species is warranted not only for its of farmland to invest in reserves is also positive for farm-
benefit to the environment, but its benefit as a research ing overall, since it allows the farmers to have a provider
tool to better understand and manage environments for of ecosystem services within their land, ensuring more
biodiversity conservation. indirect benefits such as pollination protection against
MENEZES ET AL. 3 of 10

drought. A study on the cattle production efficiency in (Conde et al., 2010). We excluded regions with selective
Amazon, suggests current farmlands could be considered logging (either regular or conventional) because it was
inefficient, operating at 38% efficiency (Igliori, 2005). found that these species still occur in areas with selective
Lastly, we discussed different strategies to reduce the loss logging (Polisar et al., 2017). With this information, we
of jaguars, with an emphasis on strategies instituted at settled for areas in which we believed we could not find
the landscape scale. jaguars.
Once we had the areas where jaguars were displaced/
killed, we attempted to identify how many jaguars would
2 | METHODS be in these areas. For that, we based our estimation on a
previous spatially explicit estimation of jaguar abundance
2.1 | Background data (Jȩdrzejewski et al., 2018). In this estimation, authors col-
lected density data from 117 camera trap studies con-
To estimate how many jaguars have been killed or dis- ducted in pristine locations in South and Central
placed by the deforestation between 2016–2019, we first America (7 in Argentina, 22 Belize, 13 Bolivia, 15 Brazil,
need to assess which areas have been deforested. We used 4 Colombia, 3 Costa Rica, 1 Ecuador, 7 Guatemala,
satellite data from DETER-B system of the Brazilian 4 French Guiana, 3 Guiana, 17 Mexico, 2 Panama, 1 Para-
Institute for Space Research (Instituto Nacional de guay, 6 Peru and 6 Venezuela), from 2002 to 2014 to gen-
Pesquisas Espaciais [INPE], accessed on January 5th, erate a model of jaguar density (Jȩdrzejewski et al., 2018).
2020). This dataset includes deforestation throughout the For that, Jȩdrzejewski et al. (2018) used density data from
Brazilian Amazon, from August 2016 to December 2019, the studies mentioned above to create a linear model,
and provided polygons representing deforested areas with density as the response variable and a series of envi-
larger than 6.25 ha. This system uses satellite images ronmental variables as explanatory variables (tempera-
from satellites Resourcesat-1, CBERS-4, and AWiFS, ture, net primary productivity, etc.; see Jȩdrzejewski
which allow deforestation updates every 5 days in the et al., 2018 for further methods).
Amazon region, a “real-time” estimate of deforestation However, Jȩdrzejewski et al. (2018) argue this data
instead of previous yearly estimations (Diniz et al., 2015). might not be representative of jaguar density in impacted
These images are processed by a linear spectral mixture regions, since density data reflect regions with pristine
model to infer the fraction of soil, shade, and vegetation environments. To account for that, the authors also
in the image. (Diniz et al., 2015). This percentage of developed a species distribution model of jaguars, using
exposed soil and appearance is then visually classified by further presence records throughout the Americas. This
INPE operators in eight classes: “clearcut deforestation,” second model was also a generalized linear model, with
“deforestation with vegetation,” “mining,” “moderate probability of presence as the response variable, and the
degradation,” “intense degradation,” “burnt scar,” “regu- same environmental variables as explanatory variables.
lar selective logging,” and “conventional (irregular) selec- As a result, this second model could predict a probability
tive logging” (Diniz et al., 2015). Classes are of occurrence for each location given a set of environ-
distinguished based on color tonality, shape, texture, and mental variables. This probability was interpreted as a
context. More information on each class can be found in measure of habitat quality, the higher the probability, the
Diniz et al. (2015). This information is then audited to higher is the quality for the animal. This probability was
reduce human errors (Diniz et al., 2015). DETER-B had then multiplied with the density predicted by the first
an 80% agreement in classification with INPE's PRODES model to generate a final density prediction for a given
system (Diniz et al., 2015), whose accuracy reaches 90% set of environmental variables.
(Pendrill & Persson, 2017). Using those parameters, we After this model was constructed, it was necessary to
attempt to discern which classes of deforestation would spatialize these predictions. Jȩdrzejewski et al. (2018)
lead to jaguar displacement and death. divided the Americas into pixels of 1 km2 and the values
For our analysis, we considered the classes “clearcut of environmental variables for each pixel were obtained
deforestation,” “deforestation with vegetation,” “mining,” from satellite images. Based on that, the density on each
“moderate degradation,” “intense degradation,” “burnt 1 km2 pixel was calculated as explained above. This den-
scar,” to be classes that would displace jaguars. These sity was later multiplied by the area to obtain the esti-
classes either expose the ground or show growth of sec- mated number of jaguars in each 1 km2 pixel. As a result
ondary forest (“deforestation with vegetation”) (Diniz of this process, Jȩdrzejewski et al. (2018) produced a sin-
et al., 2015). Secondary growth is avoided by jaguars in gle map, ranging from Southern USA to the southern end
comparison to tall forests, so we assumed that animals in of South America, covered in 1 km2 grid cells, each with
those areas would be displaced the surrounding region an estimated of the number of jaguars.
4 of 10 MENEZES ET AL.

2.1.1 | Novel methods of total amount of forest pixels, which could be compared
with the number of unprotected forest pixels described
The DETER-B and the Jȩdrzejewski et al. (2018)'s jaguar above.
density map, however, are not immediately comparable. We also measured distance to the border of deforesta-
The DETER-B dataset has a resolution of 6.25 ha com- tion, that is, distance to a point of entry in the forest, to
pared to 1 km2 for the jaguar density data. Thus, we represent that it is logistically harder to deforest in the
downscaled the DETER-B data by creating a grid of center of a continuous piece of forest than at its border.
1 km2 cells and assigned to each cell the proportion of its To do so, we took the binary forest non-forest image from
area covered by one of DETER-B polygons. Following before, and inverted it, to generate an image of potential
that, we multiplied that value by the estimate of abun- entry points. This image was clipped to exclude areas out-
dance of jaguars in that pixel. This multiplication implies side of the states. After that, we calculated the distance of
all jaguars in that area were affected. They were either every forested pixel to the closest non forested pixel.
displaced by human activities or killed; however, the pre- These distances, were aggregated by a scale of 400, using
cise fate is unknown. We only assume such a destructive the mean value. Then, we calculated the mean value of
event in the landscape will remove jaguars from the each pixel within a state per year.
region. Once we had the total number of jaguars killed or Lastly, we took the average efficiency of cattle pro-
displaced, we converted it to a percentage of the total duction for each state from a previous study
number of jaguars in the state. (Igliori, 2005). We used this metric following the hypoth-
To infer the cause of these displacements and deter- esis that more efficient cattle management is less land
mine whether different states perform better than others, intensive, and thus highly efficient properties can afford
we calculated the number of deaths/displacements by to preserve the amount required by law without losing
year and by Brazilian state, using the same methods much profit. As a result, we expect more efficient cattle
above. Further, we measured the relationship between production to lead to lower jaguar displacement.
displacement and socio-geographical variables. We calcu- Once we had measures for each variable, we tested
lated three variables: percentage of unprotected forest, their effect using two linear models. The first considered
distance to the Arch of Deforestation, and cattle information per state and year, having displacement as
efficiency. response variables, and state, year, proportion of covered
We chose percentage of unprotected forest to measure forest outside of protected areas, and average distance to
the remaining forest available to be deforested. Through- edge as independent variables. Meanwhile, we produced
out the year several states had their cover almost entirely a second regression using data averaged across years.
extirpated. Therefore, it might be possible that the state This regression measured displacement/killings per state
has few jaguar displacements because most of its unpro- against efficiency. All analyses were done in R v.3.6.2
tected forests are gone. To measure that, we used satellite (R Core Team, 2019), using packages sf and raster
images of yearly vegetative cover from the MODIS terra (Hijmans, 2020) for spatial analysis and car (Fox &
satellite (MOD44B), from the period of 2016 to 2019 Weisberg, 2019) for type sums of squares in the linear
(Dimiceli et al., 2015). This data had resolution of 250 m, model.
and we cropped the layers to consider only the studied
states in Brazil. After that, we categorized the image in
forest or non-forest, considering areas with more than 3 | RESULTS
50% vegetation cover as forested. We then masked the
image with polygons representing the protected areas, We estimate a total of 1,422 jaguars (1.8% of the popula-
from the World Protected Regions Database (IUCN & tion) have been displaced/killed in recent years (2016:
UNEP, 2009). After masking the image, we were left with 488, 2017: 360, 2018: 268, 2019: 354; Figure 1). As we
a binary map of values 1—(unprotected forest) or 0— expected, we found that in 2019, the trend of reduction
(non-forest areas or protected forests). We aggregated the since 2016 was reversed. On a state by state basis, Para
image by a factor of 400, with each new pixel rep- (n = 537) and Mato Grosso (n = 438) showed the highest
resenting the sum of forested pixels within it. Following number of displacements/killings across all years com-
that, we summed the values of pixels in each state to gen- bined. Maranh~ao state (n = 134) showed a large reduc-
erate a state-wide estimate of unprotected forest pixels. tion in displacement/killings across the years. The largest
We also did a similar process to the original map of forest state in the region, Amazonas (n = 95) showed a small
vs. non-forested areas (before the masking by protected number of displaced/killed jaguars. Only two states show
areas). We aggregated it by a factor of 400 and summed it a decrease in the year 2019, Tocantins and Maranh~ao
the values per state. That produced a state-wide estimate (Table 1). Regarding the socio-geographic variables, we
MENEZES ET AL. 5 of 10

F I G U R E 1 Estimated number of jaguars displaced by fire or deforestation in the Brazilian Amazon per state and per year. Each state
has a set of four bars representing the study years (2016–2019, from left to right). State's shade is proportional to the average amount of forest
cover according to the MODIS44b yearly forest cover map (Dimiceli et al., 2015)

T A B L E 1 Number of jaguars estimated dead and displaced, per found an effect of state (F8,24 = 30.3815, p < .001). In
year and per state contrast, we found no support for an effect of distance to
Year
the border (beta = 0.008 ± 0.006, t = 1.217, p = .235) or
year (beta = 5.365 ± 2.665, t = 2.014, p = .055). Using
State 2016 2017 2018 2019 our second regression, we found no support for an effect
Rondônia 10.43 12.77 14.58 23.14 of cattle efficiency (F1,7 = 0.3501; p = .572).
Acre 2.47 1.61 4.64 6.93
Amapa 1.95 0.15 0.05 0.21
Amazonas 34.90 14.30 23.56 33.83
4 | DISCUSSION
Tocantins 27.58 13.33 17.61 0.00
We present a broad assessment of the dramatic effects
Maranh~ao 81.89 30.06 16.55 5.15 that deforestation and fires can have on the largest cat in
Mato Grosso 113.84 119.97 88.06 116.43 the Americas. A loss of 1,422 jaguars in just 5 years sug-
Roraima 69.80 1.58 10.35 35.59 gests that time is short to avert jaguar decline. This trend
Para 145.51 166.31 92.86 132.86 varies widely among states, with some showing the oppo-
Sum 488.37 360.08 268.26 354.14
site trend of the entire country (e.g., Tocantins and
Maranh~ao states). However, four states (Acre, Amazonas,
Roraima, and Rondônia) showed a trend of increasing
found support for an effect of forest cover (beta = displacement/killing that was not triggered in 2019, but
366.406 ± 141.824, t = 2.584, p = . 016). We also instead is a continuous process for many years. Thus,
6 of 10 MENEZES ET AL.

although there is a clear increase on average in 2019, this machinery and workforce to the border of the jungle than
effect is different among states, which suggest that poli- to its core. That makes it easier to start the deforestation
cies at state levels are critical for conserving jaguars. process from the original borders of this biome. Although
Some of this variation in jaguar loss from state to state the effect of the Arch of Deforestation is visible, we found
is geographic. States that have less unprotected forests no support for an effect of distance to border in the dis-
are less prone to jaguar displacement. This result can be placement of jaguars. Perhaps, jaguars' requirements
interpreted two ways: First, the inverse relationship would already push the species into deeper and more
between proportion of protected forest and number of intact forests, reducing their abundance in places where
displaced jaguars indicates conservation units reduce jag- human activity is prone to start. However, the mecha-
uar displacement/killing, at least to some level. However, nism still requires study to obtain more detailed
this effect is weak, explaining little of the variance in the explanation.
displacement (Figure 2). Much more of the variance is We were surprised to find that cattle efficiency did
captured by states across years, indicating that there are not have a perceivable effect on jaguar displacement, as
more determining factors than protection alone. Alterna- initially hypothesized. Reduction in the use of land with
tively, the same result may stem from a bleaker interpre- increased efficiency has been a long-standing hypothesis
tation. Since, to our knowledge, no new conservation in regards to conservation policy, also known as land
units have been created in the studied years, the reason sparing (Phalan, Onial, Balmford, & Green, 2011). Yet,
for states having large proportion of forest being protec- we found no effect of efficiency on the displacement. Bar-
ted is because they loss most of their unprotected forest. retto, Berndes, Sparovek, and Wirsenius (2013) offer an
For example, Maranh~ao has 25% of its original cover. Of explanation to this conundrum. Increasing efficiency
those 25%, 70% are in protected areas (Celentano often leads to greater consumption of input materials
et al., 2017). The effect of protected areas may be small (e.g., larger grasslands, more roads, an increase in local
overall, but for some states that is all they have left. industries to supply additives), which would also lead to
Some states are associated with the Arch of Deforesta- greater use of the land. Land use is only reduced if land
tion, a region where the agricultural border advances is already scarce, either due to high price of land or due
towards the forest, going from the east and south of the to government enforcement (Barretto et al., 2013; Strass-
Brazilian state of Para towards the west, passing through burg et al., 2012). If enforcement is already underway,
the states of Mato Grosso, Rondônia, and Acre some studies suggest methods that can be used to
(Michalski, Peres, & Lake, 2008). It is easier to transport increase efficiency. Pasture rotation, along with planting

F I G U R E 2 Effect of the proportion of forest cover protected by reserves on the displacement of jaguars. This graph represents an added
variable plot, that is, the relationship between forest cover and displacement after all other effects have been removed (distance, year, and
state)
MENEZES ET AL. 7 of 10

legumes within the pastures, may increase livestock mass The state of Mato Grosso was the second state with the
growth (Latawiec, Strassburg, Valentim, Ramos, & Alves- highest number of jaguars displaced by deforestation and
Pinto, 2014). At a larger scale, municipalities with larger fire (n = 438). In 2019, the state recorded the highest rate
populations tend to handle cattle more efficiently (Igliori, of deforestation in the last eleven years (Valdiones,
2005). Paradoxically, higher education tends to decrease Silgueiro, Cardoso, Bernasconi, & Thuault, 2019). Although
cattle efficiency as municipalities switch to other agricul- the current scenario is unfavorable for habitat protection
tural practices (Igliori, 2005). This suggests the bottleneck for jaguars in Mato Grosso, the state has already been pre-
for the preservation of Brazilian Amazon is enforcement. viously recognized as an example of governance in reduc-
Other activities may contribute to forest preservation ing deforestation (DeFries, Herold, Verchot, Macedo, &
once enforcement is present, but cannot replace it. Some Shimabukuro, 2013; Fearnside, 2003). This reduction in
of the changes in jaguar numbers during the years might deforestation (in 2010) in Mato Grosso was the result of a
result from policy changes. Policy, however, is hard to combination of market forces, policies, enforcement, and
quantify. Currently, most states in Brazil lack a unified improved monitoring (DeFries et al., 2013). One of the
database of permits for deforestation, along with statistics aspects was the increase in enforcement, possible through
of anti-deforestation measures. Thus, it is currently the communication between the national space institute
impossible to analyze the efficiency of current policies, or and the local environmental agency, which automatically
even to compare them across states. We can, however, evaluation whether a farm had been illegally deforested.
suggest some policies that might be effective in fighting Mato Grosso has proven that it is possible to use the gov-
deforestation. ernment's capacity to impose regulations and influence
For example, Tocantins' government implemented a deforestation trends (Fearnside, 2003), although it is
policy of reducing the number of permits for deforesta- unclear what changed after 2011. These measures should
tion in the years 2018–2019 (Caldas & Government of be stimulated to ensure the integrity of habitats for jaguars.
Tocantins, 2019). Maranh~ao has also registered record Another policy is to require certified sustainable tim-
decline in deforestation, much larger than expected by its ber in each state. There are current certification initiatives
own action plan to reduce deforestation (State of for wood trading that ensure sustainable production.
Maranh~ao, 2011). This reduction has been attributed to However, these certifications are often not economically
technical support provided to local farmers by the state advantageous, since the Brazilian market is uninterested
secretary of the environment (SEMA, 2019). However, in paying higher market prices for certified products
the same may result from a lack of forest to cut down. (Dasgupta & Burivalova, 2017). Further, wood loggers
On a negative side, the states of Para and Mato often engage in land grabbing, a process where land with
Grosso lead in the number of jaguars displaced by defor- uncertain governance is exploited without proper licens-
estation. Cumulative deforestation in these states is ing. The ease of this process generates a tragedy of the
clearly related to the expansion of soybean and cattle commons scenario (Hardin, 1968), where the costs of
areas, partially for exportation to China (Fearnside & clear cutting are shared by society (loss of forest land
Figueiredo, 2015).The dynamic of land use changes in without clear ownership) and benefits go to a small group
these states seems to be connected. Advances of soy- (clear cut profits, with no land expense). Thus, several
beans into pastures and forests in Mato Grosso displace studies suggest clear ownership of land would reduce the
ranching activities into the forests of Para (Arima, benefit for clear cutters (Reydon, Fernandes, & Telles,
Richards, Walker, & Caldas, 2011). Moreover, recent 2020; Stabile et al., 2020).
state policies can also account for part of deforestation Although current federal policies are focused on
patterns. For example, Para introduced an ecological tax decreasing enforcement, some of their directives might
on merchandise and services. While it has been capable be used to improve conservation indirectly. The current
of raising funds to direct to municipalities with high Brazilian president was elected under a promise of reduc-
level of deforestation, there was no impact on deforesta- ing bureaucracy. Bureaucracy has been shown to enable
tion rates (Tupiassu, de SL, & Gros-Désormeaux, 2019). deforestation when it is present in the form of shared
A previous study analyzed the distribution of this competencies and governance. For example, in Camer-
money and found that many municipalities were not oon, the division of the previously existing ministry of
investing this extra tax in environmental projects forestry in to a new ministry of forestry and another of
(Tupiassu et al., 2019). There was also no requirement environment generated overlapping competencies (Epule
(only a suggestion) by the state that this money would et al., 2014). Further, when actions required cooperation
be applied toward that end. Thus, the state had no between both, communication speed was reduced by the
means to divest or otherwise enforce this money would extra formalism of inter-ministerial communication
be dedicated to their conceived goal. (Epule et al., 2014). Similar issues have been found across
8 of 10 MENEZES ET AL.

government sectors in multiple countries, often dubbed conservation strategies and understand the net impacts of
the “curse of common competencies” (Steytler, 2003). anthropogenic activities on jaguar at large scales.
Brazil might face similar or perhaps higher levels of frag- DETER-B images scan the amazon every 5 days, with
mentation of authority. Its system of environmental pro- expectation to increase to every 2 days in near future
tection is divided in two levels. At the federal level, (INPE, 2016), and are capable of identifying small scale
Brazilian Institute of the Environment and Renewable deforestations (<7 ha). Because a measurement can be
Natural Resources (IBAMA) is responsible for licensing obtained weekly, it is now possible to detect and thus visit
private environmental projects. However, interested a deforestation area while its deforestation is still happen-
parties also have to pursue licensing with the local state ing or even when it has just begun. However, enforcement
environment agencies, which often have local environ- incursions are often expensive, and the government might
mental laws and independent mechanisms of enforce- not have enough resources to visit all deforestation areas.
ment. Previous studies recommend ameliorating these Using the same multiplication of rasters employed in this
issues by centralizing authority in a single agency (Epule study, it is possible to obtain daily estimation of jaguars dis-
et al., 2014), which would be in alignment with the cur- placed/killed in each deforestation area. This could be used
rent drive to reduce bureaucracy. to prioritize visits by enforcement agents, preferring areas
The success or failure of these government measures with larger number of jaguars. With an indication of how
is usually assessed based on deforestation rates. In this many jaguars are being lost in each deforestation hotspot,
study, we present a new metric, based on the displace- they can opt to prioritize regions where more jaguars are
ment/killing of a charismatic species and with a recog- lost. Further, the same system can be used to produce daily
nized role in the maintenance of Amazonian biodiversity. estimates of how many jaguars are being killed/displaced.
Despite showing a significant reduction in jaguar num- This estimation of numbers of jaguar killed or displaced
bers, our estimation is likely conservative. Not only do we can also be used to provide a live tracker online, to inform
assume that all animals in a deforestation polygon are dis- the public of losses.
placed/killed, but we have not considered displacement Jaguars can be conserved through a wide range of
and killing for other reasons. For instance, as a secondary strategies and mechanisms, from government-designated
effect, deforestation increases the access to formerly protected areas to economic measures, such as reducing
remote areas facilitating poaching and retaliation hunting meat consumption, or to promoting wildlife friendly beef
(Romero-Muñoz et al., 2020). These factors combined certification (Romero-Muñoz et al., 2020). We hope the
were responsible for the drastic jaguar population decline somber estimation that around 300 jaguars are lost yearly
in the southern portion of the species distribution will push the implementation of potential conservation
(e.g., Romero-Muñoz et al., 2018). A possible example is measures forward.
Iguaçu National Park, while the species density was esti-
mated to be 2.96 individuals/100 km2 in 1995 (Crawshaw, ACKNOWLEDGMENTS
1995), about 10 years later the jaguar estimated density We thank the National Council for Scientific and Tech-
was 0.545 individuals/100 km2 (Paviolo et al., 2008). nological Development (grant: 152160/2018-3), the S~ao
Deforestation and fire rapidly transform source habitats Paulo Research Foundation (2017/08461-8), the Coordi-
for jaguars into sink habitats, with resource restriction nation for the Improvement of Higher Education Person-
and increased risk (Romero-Muñoz et al., 2018). With nel and the Foundation for support to the Development,
this, we can assume that the numbers presented here rep- teaching, science and technology of the state of Mato
resent a starting value for estimating the subsequent Grosso do Sul (combined grant number 59/300.135) for
impact of deforestation and fire on the jaguar population. providing support and scholarships for the researchers in
Given the circumstances that information on jaguars this letter. We thank Włodzimierz Jędrzejewski for kindly
is scarce on the scale proposed by this study, we assume provide the jaguar density data from his study.
some caveats, such as not offering a measure of error
associated with each pixel and considering the densities CONFLICT OF INTEREST
as constant over the period from 2015 to 2019, since The authors have no conflict of interest to report.
Jȩdrzejewski et al. (2018) density estimates covered 2002–
2014. However, we still believe the information presented AUTHOR CONTRIBUTIONS
here could not be gathered through other means, and a Jorge F. S. Menezes wrote the first draft and performed
potential loss of 1.8% in the jaguar population deserves to all analysis. Fernando R. Tortato sourced the jaguar data,
be noted, even if only as a warning. Luiz G. R. Oliveira-Santos, Fabio O. Roque and Ronaldo
The approach and data presented here, despite G. Morato had the idea to make the paper and contrib-
some limitations, can be built on to further improve uted with comments on the manuscript.
MENEZES ET AL. 9 of 10

E TH IC S ST A T EME N T Crawshaw, P. G. J. (1995). Comparative ecology of ocelot (Felis

All data in this paper was derived from other previously pardalis) and jaguar (Panthera onca) in a protected subtropical
published studies. No animal or human experiment was Forest in Brazil and Argentina. University of Florida.
Cullen, L., Jr., Abreu, K. C., Sana, D., & Nava, A. F. D. (2005). Jag-
involved.
uars as landscape detectives for the upper Parana River corri-
dor, Brazil. Natureza & Conservaçao,~ 3, 43–58.
DATA AVAILABILITY STATEMENT Dasgupta, S., & Burivalova, Z. (2017). Does forest certification really
All data used is derived from previously published work. mongabay. Retrieved from https://news.mongabay.com/
sources from other authors. The current authors have no 2017/09/does-forest-certification-really-work/.
right over it. DeFries, R., Herold, M., Verchot, L., Macedo, M. N., &
Shimabukuro, Y. (2013). Export-oriented deforestation in Mato
ORCID Grosso: Harbinger or exception for other tropical forests? Philo-
Jorge F. S. Menezes https://orcid.org/0000-0003-2224- sophical Transactions of the Royal Society B: Biological Sciences,
368, 20120173. https://doi.org/10.1098/rstb.2012.0173
9758
Dimiceli, C., Carroll, M., Sohlberg, R., Kim, D. H., Kelly, M., &
Fernando R. Tortato https://orcid.org/0000-0003-1901-
Townshend, J. R. G. (2015). MOD44B MODIS/Terra Vegetation
5037 Continuous Fields Yearly L3 Global 250m SIN Grid V006.
Fabio O. Roque https://orcid.org/0000-0001-5635-0622 Diniz, C. G., Souza, A. A. A., Santos, D. C., Dias, M. C.,
Ronaldo G. Morato https://orcid.org/0000-0002-8304- Luz, N. C. d., Moraes, D. R. V. d., … Gomes, A. R. (2015).
9779 DETER-B: The new Amazon near real-time deforestation
detection system. IEEE Journal of Selected Topics in Applied
R EF E RE N C E S Earth Observations and Remote Sensing, 8, 3619–3628.
Arima, E. Y., Richards, P., Walker, R., & Caldas, M. M. (2011). Sta- Eisenhammer, S. (2019). “Day of Fire”: Blazes ignite suspicion in
tistical confirmation of indirect land use change in the Brazil- Amazon town. Novo Progresso: Reuters.
ian Amazon. Environmental Research Letters, 6, 024010. Epule, E. T., Peng, C., Lepage, L., & Chen, Z. (2014). Policy options
Barretto, A. G. O. P., Berndes, G., Sparovek, G., & Wirsenius, S. towards deforestation reduction in Cameroon: An analysis
(2013). Agricultural intensification in Brazil and its effects on based on a systematic approach. Land Use Policy, 36, 405–415.
land-use patterns: An analysis of the 1975-2006 period. Global https://doi.org/10.1016/j.landusepol.2013.09.004
Change Biology, 19, 1804–1815. Estes, J. A., Terborgh, J., Brashares, J. S., Power, M. E., Berger, J., …
Becker, B. K. (2005). Geopolítica da Amazônia. Estudos Avançados, David, A. (2011). Trophic downgrading of planet Earth. Science,
19, 71–86. 333, 301–306. Retrieved from http://www.ncbi.nlm.nih.gov/
Boron, V., Tzanopoulos, J., Gallo, J., Barragan, J., Jaimes-Rodriguez, L., pubmed/2176474.
Schaller, G., & Payan, E. (2016). Jaguar densities across human- FAO. (n.d.). At home in the Amazon: Protecting biodiversity and live-
dominated landscapes in Colombia: The contribution of unpro- lihoods together. Retrieved from http://www.fao.org/in-action/
tected areas to long term conservation. PLoS One, 11, 1–14. at-home-in-the-amazon/en/.
Brando, P. M., Soares-Filho, B., Rodrigues, L., Assunç~ao, A., Fearnside, P. M. (2003). Deforestation control in Mato Grosso: A
Morton, D., Tuchschneider, D., … Coe, M. T. (2020). The gathering new model for slowing the loss of Brazil's Amazon forest.
firestorm in southern Amazonia. Science advances, 6, eaay1632. AMBIO: A Journal of the Human Environment, 32, 343–345.
Caldas, T., & Government of Tocantins. (2019). Tocantins reduz em https://doi.org/10.1579/0044-7447-32.5.343
16% o desmatamento entre 2018 e 2019. Retrieved from https:// Fearnside, P. M., & Figueiredo, A. M. (2015, 229). China's influence
portal.to.gov.br/noticia/2019/11/20/tocantins-reduz-em-16-o- on deforestation in Brazilian Amazonia: A growing force in the
desmatamento-entre-2018-e-2019/. state of Mato Grosso. In R. Ray, K. Gallagher, A. L opez, & C.
Caro, T., Engilis, A., Fitzherbert, E., & Gardner, T. (2004). Prelimi- Sanborn (Eds.), BU global economic governance initiative discus-
nary assessment of the flagship species concept at a small scale. sion papers 2015-3. Boston: Boston University.
Animal Conservation, 7, 63–70. Fox, J., & Weisberg, S. (2019). An R companion. Thousand Oaks,
Carvalho, E. A. R., Zarco-Gonzalez, M. M., Monroy-Vilchis, O., & CA: Sage.
Morato, R. G. (2015). Modeling the risk of livestock depredation by Hardin, G. (1968). The tragedy of the commons. Science, 162, 1243–
jaguar along the Transamazon highway, Brazil. Basic and Applied 1248.
Ecology, 16, 413–419. https://doi.org/10.1016/j.baae.2015.03.005 Hijmans, R. J. (2020). raster: Geographic data analysis and model-
Celentano, D., Rousseau, G. X., Muniz, F. H., Varga, I. . D., ing. https://CRAN.R-project.org/package=raster.
Martinez, C., Carneiro, M. S., … Martins, M. B. (2017). Towards Igliori, D. C. (2005). Determinants of technical efficiency in agricul-
zero deforestation and forest restoration in the Amazon region ture and cattle ranching: A spatial analysis for the Brazilian
of Maranh~ao state, Brazil. Land Use Policy, 68, 692–698. http:// Amazon. Environmental Economy and Policy Research Discus-
www.sciencedirect.com/science/article/pii/S0264837717302296 sion Papers, 31927.
Conde, D. A., Colchero, F., Zarza, H., Christensen, N. L., INPE. (2016). Monitoramento da Floresta Amazônica Brasileira por
Sexton, J. O., Manterola, C., … Ceballos, G. (2010). Sex matters: Satélite – Projeto PRODES. S~ao José dos Campos. INPE.
Modeling male and female habitat differences for jaguar con- IUCN, UNEP. (2009). The World Database on Protected Areas
servation. Biological Conservation, 143, 1980–1988. https://doi. (WDPA). Cambridge: UNEP-WCMC Retrieved from http://
org/10.1016/j.biocon.2010.04.049 www.wdpa.org/
10 of 10 MENEZES ET AL.

Jȩdrzejewski, W., Robinson, H. S., Abarca, M., Zeller, K. A., Romero-Muñoz, A., Torres, R., Noss, A. J., Giordano, A. J.,
Velasquez, G., Paemelaere, Evi A. D., … Quigley, H. (2018). Quiroga, V., Thompson, J. J., … Kuemmerle, T. (2018). Habitat
Estimating large carnivore populations at global scale based on loss and overhunting synergistically drive the extirpation of jag-
spatial predictions of density and distribution – Application to uars from the Gran Chaco. Diversity and Distributions, 25,
the jaguar (Panthera onca). PLoS One, 13, 1–25. 176–190.
Jenkins, C. N., Pimm, S. L., & Joppa, L. N. (2013). Global patterns SEMA. 2019. Desmatamento no Maranh~ao cai 15% no período
of terrestrial vertebrate diversity and conservation. Proceedings 2018/2019, diz relat orio do INPE. Retrieved from https://www.
of the National Academy of Sciences of the United States of sema.ma.gov.br/desmatamento-no-maranhao-cai-15-no-
America, 110, E2603–E2610. periodo-20182019-diz-relatorio-do-inpe/.
Latawiec, A. E., Strassburg, B. B. N., Valentim, J. F., Ramos, F., & Stabile, M. C. C., Guimar~aes, A. L., Silva, D. S., Ribeiro, V.,
Alves-Pinto, H. N. (2014). Intensification of cattle ranching pro- Macedo, M. N., Coe, M. T., … Alencar, A. (2020). Solving
duction systems: Socioeconomic and environmental synergies Brazil's land use puzzle: Increasing production and slowing
and risks in Brazil. Animal, 8, 1255–1263. https://doi.org/10. Amazon deforestation. Land Use Policy, 91, 104362. https://doi.
1017/S1751731114001566 org/10.1016/j.landusepol.2019.104362
Marchini, S., & Macdonald, D. W. (2012). Predicting ranchers' State of Maranh~ao. (2011). Plano de Aç~ao Para Prevenç~ao e Con-
intention to kill jaguars: Case studies in Amazonia and Panta- trole do Desmatamento e das Queimadas no Estado do
nal. Biological Conservation, 147, 213–221. https://doi.org/10. Maranh~ao (PPCD-MA).
1016/j.biocon.2012.01.002 Steytler, N. (2003). The powers of local government in decentralized
Michalski, F., Peres, C. A., & Lake, I. R. (2008). Deforestation systems: dispelling “the curse of common competencies.”
dynamics in a fragmented region of southern Amazonia: Evalu- 2, Local Government Working Paper Series.
ation and future scenarios. Environmental Conservation, 35, 93– Strassburg, B. B. N., Rodrigues, A. S. L., Gusti, M., Balmford, A.,
103. https://www.cambridge.org/core/article/deforestation- Fritz, S., Obersteiner, M., … Brooks, T. M. (2012). Impacts of
dynamics-in-a-fragmented-region-of-southern-amazonia- incentives to reduce emissions from deforestation on global
evaluation-and-future-scenarios/ species extinctions. Nature Climate Change, 2, 350–355.
4D7B98329D8FEDE55B71644B7F77A67A Thornton, D., Zeller, K., Rondinini, C., Boitani, L., Crooks, K. R.,
Morato, R. G., Connette, G. M., Stabach, J. A., de Paula, R. C., Burdett, C., … Quigley, H. (2015). Assessing the umbrella value
Ferraz, K. M. P. M., Kantek, D. L. Z., … Leimgruber, P. (2018). of a range-wide conservation network for jaguars (Panthera
Resource selection in an apex predator and variation in onca). Ecological Applications, 26, 1112–1124.
response to local landscape characteristics. Biological Conserva- Tupiassu, L., de SL, F. L. P., & Gros-Désormeaux, J.-R. (2019). ICMS
tion, 228, 233–240. https://doi.org/10.1016/j.biocon.2018.10.022 Ecol ogico e desmatamento nos municípios prioritarios do
Paviolo, A., De Angelo, C. D., Di Blanco, Y. E., & Di Bitetti, M. S. estado do Para. Revista Direito GV, 15, e1928.
(2008). Jaguar Panthera onca population decline in the Upper Valdiones, A. P., Silgueiro, V., Cardoso, B., Bernasconi, P., & Thuault, A.
Parana Atlantic Forest of Argentina and Brazil. Cambridge Univer- (2019). Características do desmatamento na amazônia mato- grossense
sity Press, 42, 554–561. Retrieved from https://www.cambridge. em 2019. Retrieved from https://www.icv.org.br/drop/wp-content/
org/core/article/jaguar-panthera-onca-population-decline-in- uploads/2019/12/2019AnaliseDesmatamentoAmazoniaMT.pdf.
the-upper-parana-atlantic-forest-of-argentina-and-brazil/932 Werth, D., & Avissar, R. (2002). The local and global effects of Ama-
ED14C7080FB5998537FD305DC0B28. zon deforestation. Journal of Geophysical Research: Atmo-
Pendrill, F., & Persson, U. M. (2017). Combining global land cover spheres, 107, LBA 55-1-LBA 55-8. https://doi.org/10.1029/
datasets to quantify agricultural expansion into forests in Latin 2001JD000717
America: Limitations and challenges. PLOS ONE, 12, e0181202. Wright, S. J., Gompper, M. E., & Deleon, B. (1994). Are large preda-
Retrieved from http://www.globeland30.org. tors keystone species in neotropical forests?. The Evidence from
Phalan, B., Onial, M., Balmford, A., & Green, R. E. (2011). Reconcil- Barro Colorado Island. Oikos, 71, 279–294.
ing food production and biodiversity conservation: Land shar- Zeller, K. A., Nijhawan, S., Salom-Pérez, R., Potosme, S. H., &
ing and land sparing compared. Science, 333, 1289–1291. Hines, J. E. (2011). Integrating occupancy modeling and inter-
https://doi.org/10.1126/science.1208742 view data for corridor identification: A case study for jaguars in
Polisar, J., de Thoisy, B., Rumiz, D. I., Santos, F. D., McNab, R. B., Nicaragua. Biological Conservation, 144, 892–901. https://doi.
Garcia-Anleu, R., … Venegas, C. (2017). Using certified timber org/10.1016/j.biocon.2010.12.003.
extraction to benefit jaguar and ecosystem conservation. Ambio,
46, 588–603.
R Core Team. (2019). R: A language and environment for statistical
computing. Vienna, Austria: R Foundation for Statistical Com-
puting. Retrieved from https://www.r-project.org/ How to cite this article: Menezes, J. F. S.,
Reydon, B. P., Fernandes, V. B., & Telles, T. S. (2020). Land gover- Tortato, F. R., Oliveira-Santos, L. G. R., Roque, F.
nance as a precondition for decreasing deforestation in the Bra- O., & Morato, R. G. (2021). Deforestation, fires, and
zilian Amazon. Land Use Policy, 94, 104313. https://doi.org/10.
lack of governance are displacing thousands of
1016/j.landusepol.2019.104313
Romero-Muñoz, A., Morato, R. G., Tortato, F., & Kuemmerle, T.
jaguars in Brazilian Amazon. Conservation Science
(2020). Beyond fangs: Beef and soybean trade drive jaguar and Practice, 3(8), e477. https://doi.org/10.1111/
extinction. Frontiers in Ecology and the Environment, 18, 67–68. csp2.477