Compliance Assessment of Public Commitment on

Amazon Cattle Ranching Marfrig Global Foods S.A.

Report No.: Z0520642, Rev. 0

Date: 27 April 2015

Det Norske Veritas

Av. Alfredo Egydio de Souza Aranha,100

Centro Administrativo Santo Amaro – CASA

Bloco D – 3º andar

04726-170 – São Paulo, SP

Brasil

Tel.: +55 11 3305 3305

Fax: +55 11 3305 3313

Report Title: Compliance Assessment of Public Commitment on

Amazon Cattle Ranching

Client: Marfrig Global Foods S.A.

Contact Person: Mathias Almeida

Issuance date: 27 April 2015

Project No.: 1-MDZ505,0

Unit: Business Assurance

Report No.: Z0520642, Rev. 0

Prepared by: Verified by: Felipe Lacerda Antunes Juliana Scalon

1

To Marfrig Global Foods

I) Introduction

Since 2007 Greenpeace studies the behavior of the supply chain of livestock in the Amazon. After a

long investigative process, in 2009, the organization launched the report “Slaughtering the Amazon”,

which pointed the relationship between meat processors involved in illegal deforestation and forced

labor with cutting edge products sold on the international market, like bags and sports shoes. Since

then, meatpackers Marfrig, Minerva and JBS made a public commitment to exclude from their list of

suppliers farms who deforested the Amazon rainforest after October, 2009, besides those that use

labor analogous to slavery or are located in indigenous lands and conservation units. The public

commitment establishing criteria for purchasing cattle on properties located in the Amazon Biome is

called "Minimum Criteria for Industrial Scale Cattle Operations in the Brazilian Amazon Biome”.

II) Objective

DNV GL was hired to carry on an audit in order to independently evaluate Marfrig information and

processes to identify whether it met the criteria given in the above public commitment, covering the

period January 1st to December 31st 2014.

III) Audit period

The audit was conducted in the period from March 5th to April 8th 2015, covering the purchase of animals

between January 1st and December 31st 2014.

IV) Company description and Cattle Purchase process

Marfrig Global Foods is a global food company based on beef cattle, sheep, swine, poultry and fish. It

operates in the food service and retail segments, through an operating platform comprising production,

commercial and distribution units installed in 16 countries. In Brazil, the company has 18 units, 5 of

which buyers of cattle from farms located within the Amazon Biome: Tangara da Serra / MT; Paranatinga

/ MT; Rolim de Moura / RO; Chupinguaia / RO and Tucumã / PA.

A site visit to Rolim de Moura/ RO unit checked the application of the " Procedure of Socio-environmental

Control of Purchase of Cattle", code DSGPSC 001, issuance date May 2012, Revision 00, applicable to all

units, with the description of the steps required for the purchase of animals, which is followed by the

cattle purchase and sustainability team sectors, namely:

1. Consult of IBAMA embargoed list;

2. Consult of the Forced Labor list of the MTE (Ministry of Labor and Employment);

3. Request of environmental license or CAR (Rural Environmental Register);

4. Request of CCIR (Certificate of Registration of Rural Property) copy;

5. Request of information of production phases of property (breeding/rearing/fattening);

2

6. Request of information about cattle feed system (grass/feedlot);

7. Consult of list of compliance/ non-compliance from the geospatial monitoring company, for

Amazon biome units.

The cattle purchase team (2 buyers), the sustainability supervisor, the sustainability analyst and the field

technician from Rolim de Moura unit have demonstrated to know all instructions of the social-

environmental procedures for cattle purchasing, having the training records filed in a folder at the

sustainability department.

All Marfrig units use the module Company of APIS system and/or TAURA system for cattle purchase

operations. Despite the automatic blocking in the system APIS/TAURA of the suppliers that are on the

IBAMA embargoed list, it was found in the audited units that cattle purchase department and/or

sustainability department also make queries at IBAMA site in every purchase about the status of the

supplier (by consulting through his Tax ID number – CPF/CNPJ).

The information about the SNCR/CCIR and CAR/Environmental Licenses are obtained after contact of

cattle buyers with farmers.

For units that purchase from suppliers located in the Amazon Biome, in addition to the information about

suppliers contained in the APIS_MARFRIG System / TAURA System and corporate controls, there are also

geographically-referenced information analysis of farms in this biome. Marfrig has a contract with the

geospatial monitoring company, that uses geospatial devices and geographical information system (GIS)

to draw the polygons, allowing to correlate data related to the properties location with deforestation

areas, conservation areas and indigenous land.

For the survey of the geographically-referenced point in the Amazon Biome, Marfrig has a contract with

the expert company that uses geospatial and geographic information system (GIS). On April 8th 2015

this company was visited and the procedures of data collection were presented to the DNV GL auditor.

For each unit of the Amazon Biome there is at least one Marfrig technician in the field visiting the

properties of farmers equipped with an automatic GPS that monitors the entire route up to the properties

and also the coordinates of the gate and lairage. This technician also takes photographs of the visited

farms and documents information collected from them. From this information and farm maps the expert

company performs analysis of each property (overlap with Conservation Units, indigenous lands, new

deforestation).

The contractor verifies data with INPE (National Institute of Space Research) through DETER (detection

program of deforestation focus in real time) and PRODES (program of calculation of deforestation in the

amazon), to evaluate occurrence of new deforestation in the supplier farms.

With DETER, INPE publishes each three months a statement of alerts, which indicate areas in process of

deforestation through progressive forest degradation.

3

PRODES measures the annual rates of clearcut since 1988. Since it is more detailed and depends on the

climatic conditions of the dry season for the acquisition of cloud free images, taken between May and

September, this is done only once a year, with release scheduled for December of each year. After that,

remaining data is released.

V) Procedures

Phase 1 - Process sampling and tests at the slaughterhouse.

Step 1 - Selecting sampling

Verification is limited to the scope of the reference protocol used ("Minimum Criteria for Industrial Scale

Cattle Operations in the Brazilian Amazon Biome", Greenpeace) and scope of pre-defined evaluation.

Work sampling is based on verification of documents, the information gathered in the interviews and

check the physical conditions (done randomly). The reviewed documents are described in table 1 below.

4

Table 1. List of analyzed documents

Name of document Data range / code and version Evaluate

d (Y/N)

Audit Work Plan Verification Protocol DNV GL / March 4th 2015 Y

Procedures Procedure of Socio-environmental Control of Cattle

Purchase, 001 DSGPSC code, date of issue 05/2012,

Revision 00.

Y

Registers - Documents of suppliers (CAR/LAU) and printouts of each

step.

- Polygons Information / April 8th 2015

Y

Monitoring System System of the Contracted company in geospatial

monitoring

Y

Blocking System System APIS, and/or system TAURA / verified during audit Y

List of suppliers For the period of January 1st to December 31st 2014:

- Information audit Rolim de Moura with statistics

- Information audit Tangará da Serra with statistics

- Information audit Chupinguaia with statistics

- Information audit Paranatinga with statistics

- Information audit Tucumã with statistics

Y

Public embargoed

properties list– IBAMA

Consult to website:

http://servicos.ibama.gov.br/ctf/publico/areasembargadas

/ConsultaPublicaAreasEmbargadas.php

Y

Public Listo f people, /

companies, MTE,

forced work

List of employers Interministerial Ordinance n. 02 of 12

May 2011. Update of December 26th 2014

Y

Listo f suppliers:

compliant and non-

compliant

Copy of non-compliant MT - 2014 12 29 to 2015 01 04

Copy of non-compliant RO - 2014 12 29 to 2015 01 04

Copy of non-compliant PA - 2014 12 29 to 2015 01 04

Y

Marfrig Workplan 2014 2014 – Work Plan Marfrig Global Foods: minimal criteria for cattle

operations and beef products on industrial scale in the amazon

biome

Y

This evaluation was conducted through: examination of documentation, access to data systems,

evaluation of environmental conditions of work and interviews with employees, using as a reference

documents:

"Minimum Criteria for Industrial Scale Cattle Operations in the Brazilian Amazon Biome", from

Greenpeace.

“Term of Reference for a Third Party Assessment of the Amazon Cattle Public Commitment”

5

The methodology defined for this assessment was applied in all units that buy cattle from suppliers

inside the Amazon Biome: Tangará da Serra/MT, Paranatinga/MT, Rolim de Moura/RO, Chupinguaia/RO

and Tucumã/PA.

During the document review carried on before the visits, for the period January 1st to December 31st

2014, 10% of the cattle suppliers located in the Amazon Biome was randomly sampled for each unit,

which resulted in:

- 43 of 425 active suppliers of Tangará da Serra/MT, representing 17% of the volume of

slaughtered cattle in the unit for the period;

- 12 of 118 active suppliers of Paranatinga/MT, representing 8% of the volume of slaughtered

cattle in the unit for the period;

- 138 of 1375 active suppliers of Rolim de Moura/RO, representing 17% of the volume of

slaughtered cattle in the unit for the period;

- 40 of 398 active suppliers of Chupinguaia/RO, representing 10% of the volume of slaughtered

cattle in the unit for the period;

- 48 of 476 active suppliers of Tucumã/PA, representing 15% of the volume of slaughtered cattle

in the unit for the period.

A sample of 10% of cattle purchases in slaughtered volume of the units that buy cattle from properties

within the Brazilian Amazon Biome, had been defined by the signed industries and Greenpeace. However,

Marfrig’s system does not allow the authomatic generation of total purchase operations in Amazon Biome.

In this sense, a sampling of 10% of the suppliers was defined. DNV GL considers this sampling larger

than the one based on purchase operations, since for each sampled supplier 100% of the purchases was

indirectly checked for the period, which represents more than a 10% of the purchase operations.

At Promissão/SP, the geospatial systems of satellite monitoring was presented and demonstrated. This is

performed by a contracted company. At Rolim de Moura/RO, were verified cattle purchases for Rolim de

Moura/RO and Chupinguaia/RO, its evidences and suppliers documentation and identification of cattle

origin. The same sustainability team manages Rolim de Moura/RO and Chupinguaia/RO, therefore it was

possible to assess documents and procedures of Chupinguaia while the audit was performed in Rolim de

Moura. For the other units (Tangará da Serra/MT, Paranatinga/MT and Tucumã/PA), sampled suppliers

were also assessed against the IBAMA (Environmental Agency) embargoed areas list and the list of

forced labor of the Ministry of Labor and Employment and the list of compliance/non-compliance

(geospatial monitoring regarding new deforestation, conservation units and indigenous lands).

The bovine by-products (e.g. leather, horn, blood) were indirectly included in the scope of this audit,

through sampling of active direct suppliers.

Step 2 – Test of cattle purchase

For all units of the Amazon Biome, the sampled suppliers (10 % of the suppliers of the period January 1st

to December 31st 2014) were compared with the embargoed properties lists of IBAMA, forced labor list of

MTE, and the list of the contracted company in geospatial monitoring (last list available in 2014) .

6

The documentation of all sampled purchases was also analyzed in order to verify if all procedures were

completed. Printouts of computer screens (printscreens) of each phase were analyzed, including the

analysis of IBAMA and MTE sites and lists of non-compliant geospatial monitoring company.

Through documentation analysis, interviews with staff and access to Marfrig's cattle purchase system, it

was verified that all units of Marfrig follow the company's policy of not acquiring cattle from farms

included in the list of areas embargoed areas by Decrete IBAMA n. 19 of 2 July 2008 and Decree No.

6.321 of December 21, 2007 and its available updates.

In the sampling carried out for the units of MT, PA and RO inside the Amazon Biome, it was checked

purchases from 281 direct suppliers from January 1st to December 31st 2014, and it was not observed

any purchase from a supplier that is in the IBAMA embargoed list.

Also none of the suppliers of the sampled units in MT, RO and PA appeared in the Forced Labor List of

the MTE, as confirmed against the last list dated December 26th 2014.

None of the 281 suppliers sampled in the units of MT, PA and RO was noncompliant at the slaughter date,

according to the list of geospatial monitoring, i.e.: there were no slaughtered animals from farms that

presented PRODES and / or DETER and / or overlap with Indigenous Lands and / or conservation units.

Besides the sampling mentioned above, there were simulation "in situ" of purchases for 45 producers

with one or more restrictions – that are part of the IBAMA embargoed list, forced labor list, and from the

list of non-compliance with geospatial monitoring, randomly chosen within the period between January

1st to December 31st 2014, which has 1228 farmers listed in the units of Rolim de Moura and

Chupinguaia, as well as checking their presence in the IBAMA embargoed list and forced labor list. The

purpose of this additional sampling was to determine whether there was slaughter of non-compliant

suppliers regarding geospatial monitoring related to new deforestation ( PRODES, DETER), indigenous

lands, units of environmental conservation and presence in the IBAMA embargoed list and forced labor

list.

It was not identified any cattle purchase from properties of any of the lists. It was also possible to verify

that these suppliers are automatically blocked by the system.

From January 2013 Marfrig implemented an RFI (Request for Information) tool by which farmers that

acquired cattle from third parties (indirect suppliers) shall inform the origin of the animals - Property,

Municipalty, State, Owner, CPF and CNPJ. The sustainability department consults the lists of IBAMA and

forced labor to verify that these indirect suppliers are not listed. However, according to the Terms of

Reference for 2014, no audit was carried out for the indirect suppliers, since it is not possible to check

the data, since there is no general solution for the matter. Even though, Marfrig keeps applying RFI

(Request for Information) methodology to map its indirect suppliers and possible irregularities with

deforestation and forced labor. It was also evidenced contacts with the Prosecutor Dr. Daniel César

Azeredo Avelino from Para Public Attorney and with the Legal Amazon Workgroup from the Federal Public

7

Attorney about proposals, solutions and status of Marfrig. There were two meetings during 2014.

Additional information are presented in Annex I, where the Work Plan by the company in 2014 was

evaluated.

It was evidenced the traceability procedure, POPSAU 011.9/SIF 4334 from May 2010, revision of July

10th 2014.

The animals are transported from supplying farms, always accompanied by the Transit Animal Certificate

– GTA, as ruled by the Brazilian Ministry of Agriculture. When they reach the slaughterhouse, an

employee checks GTA information and feeds the APIS / TAURA systems. Then the cattle are sent to the

corral, respecting the capacity of each corral. In each corral, the Veterinary Inspector from the Brazilian

Ministry of Agriculture (SIF) performs the checking and completes the documentation, reporting rancher

data, city, property, number of GTAs and number of animals (males and females).

Step 3 – Blocking system test

In the sampling carried out on the list provided by Marfrig, of 281 suppliers from all units of the Amazon

Biome, for the period January 1st to December 31st 2014, none of them were on the IBAMA embargoed

list or on the Forced Labor List as per consultations during the audit.

Besides the sampling mentioned above, there were simulation "in situ" of purchases for 45 producers

with one or more restrictions – that are part of the IBAMA embargoed list, forced labor list, and from the

list of non-compliance with geospatial monitoring, randomly chosen within the period between January

1st to December 31st 2014 which has 1228 farmers listed in the units of Rolim de Moura and Chupinguaia,

as well as checking their presence in the IBAMA embargoed list and forced labor list. The purpose of this

additional sampling was to determine whether there was slaughter of non-compliant suppliers regarding

geospatial monitoring related to new deforestation (PRODES, DETER), indigenous lands, environmental

conservation units and presence in the IBAMA embargoed list and forced labor list, as well as testing the

blocking system for these suppliers.

In all simulations the cattle purchase system APIS_MARFRIG did not allow the purchase, accusing the

producer as "not registered".

For cases of suppliers present in IBAMA embargoed list for a property different from that in which Marfrig

want to buy cattle, the buyer can request an unlock. The request is made by the Sustainability

Department, then passed it on to the Corporate Sustainability department, to review the case. Each non-

compliance is analyzed by means of more refined techniques tools, evaluating historical images that can

identify false positives. If proven that the property in question is not non-compliant, the request is

released by the Corporate Sustainability to the Cattle Purchase department. All files and steps of this

procedure are documented and archived.

The 178 farmers sampled in the units Rolim de Moura and Chupinguaia were tested in the purchasing

system. The remaining 103 suppliers of 3 other units within the Brazilian Amazon biome were checked in

8

the list of IBAMA, MTE, list of compliant / non-compliant regarding geospatial monitoring and it was

evidenced that all were able to provide cattle to the units at slaughter.

In all simulations the cattle purchase system APIS_MARFRIG, which operates automatically, did not allow

to buy, accusing the producer as "Unregistered" to suppliers on the list of IBAMA, MTE and non-

compliant monitoring.

Phase 2 – Geospacial monitoring company.

Step 1 – Procedures verification

On April 8th 2015 DNV GL assessed the geospatial monitoring company, and all data capturing processes

were presented. During the visit DNV GL could confirm that the company is able to conduct the

described activities, considering: Social contract, Technical responsibility, qualified personnel. It was

verified the ART record (Technical Responsibility Note) related to the activities carried out on 2014, the

scope description done by the company during the monitoring process, identifying the “products”, as well

as the Project documents describing the scope activities, the operational plan and the technical

description related to access procedures, treatment and database use related to the criteria used in the

social-environmental analysis.

Box 1 – Agrotools data

Object Geoprocessing, images processing and commerce; Geographical Information

System development and integration, administrative and technical applications

from digital cartography or any other source; development and structure of data

administration systems related to environment, urban emergencies, public and

private services network, infra-structure and urban equipment, bathymetry,

oceanography, maritime navigation, natural resources and any other information

that has spatial expression and needs its elements administration, multi-

discipline or multiple projects application integration.

CNAE 71.19-7-01

Involved

professionals

- Thiago Dias Mancilha, Computation Engineer, Technology and Innovation

Director, ART CREA-SP 92221220140517762;

- Diego Garcia Paiva, Geographer, ART CREA-SP 92221220140042761

Step 2 – Monitoring criteria assessment

The process starts when Agrotools receive data from Marfrig and validate it. In case there is any data

discrepancy, Agrotools send to Marfrig an “Anomaly Report”. After that the farm is registered in the

system and a GeoID is generated, the property geographical identity. After that the geometry map (farm

boundaries) is generated, and the social-environmental criteria is assessed, reporting any overlap when

that happens. The maps are developed based in official monitored, reported and verified information,

and Agrotools evaluates all information carefully before generating the geometry map. Weekly Agrotools

send to Marfrig the updated monitoring reports – list of compliant farms and list of non-compliant farms.

When asked by Marfrig special cases may be assessed and reported with more details.

9

Some properties’ polygons (farm boundaries) were consulted during the visit, both from the compliant

list and the non-compliant list, in order to check the overlapping with PRODES, DETER, indigenous lands

and conservation units. From the 16 polygons consulted, all shown results in line with the respective

lists.

VI) Audit results

It was not identified any purchase transaction that did not fulfill all points of public commitment.

1. Acess to information

The reviewed documents are described in table 1 (page 4).

2. Non compliance

There were zero non compliances found during this audit.

VII) Audit limitations

There were no limitations to the information during the auditing.

VIII) Conclusions

The assessment of compliance with the “Greenpeace: Minimum Criteria for Industrial Scale Cattle

Operations in the Brazilian Amazon Biome” occurred through audit at Promissão unit on March 5th to 6th

2015, at Rolim de Moura unit on March 10th to 11th 2015, and at the geospatial monitoring company on

April 8th 2015 .

Purchases of units Paranatinga / MT, Rolim de Moura / RO, Tangara da Serra / MT, Chupinguaia / RO and

Tucumã / PA were evaluated and based on details of the previous chapter of this report, with the

selected highlight of the following key findings of this evaluation regarding adherence to the "minimum

criteria for Operations with Cattle and Beef Products in Industrial Scale in the Amazon Biome -

Greenpeace":

1. In the random sampling carried out in all the units during the desk review, it was not observed

any supplier that was included in the IBAMA embargoed list as per consultations held during the

audit at the website

http://servicos.ibama.gov.br/ctf/publico/areasembargadas/ConsultaPublicaAreasEmbargadas.ph

p, nor any supplier that was included in the MTE’s Forced Labor list as consultations in the days

of the audit on the website http://portal.mte.gov.br/trab_escravo/portaria-do-mte-cria-cadastro-

de-empresas-e-pessoas-autuadas-por-exploracao-do-trabalho-escravo.htm.

2. Similarly it was shown that Marfrig did not buy cattle from suppliers sampled from the list of

non-compliant producers regarding the geospatial monitoring units of the Amazon Biome in

DETER, PRODES, conservation areas and indigenous lands;

10

3. The cattle purchasing team practices a systematic daily verification of the forced labor list

(Ministry of Labour) and embargoed areas list (IBAMA), as described in Procedure of Socio-

environmental Control of Cattle Purchase DSGPSC 001 code, date of issue 05/2012 , revision 00,

with the description of the steps required for the cattle purchase;

4. Cattle purchase use of the computerized system APIS_MARFRIG / TAURA that automatically

blocks the acquisition of properties that have cattle in the lists of IBAMA or MTE by use of forced

labor list;

5. Information about suppliers of the Amazon Biome regarding georeferenced coordinates of its

properties. In the sampling performed, 100% of the active direct suppliers located in the Amazon

Biome had farm boundaries, showing the polygon of the property;

6. 100% of sampled suppliers from Mato Grosso and Para have CAR (Rural Environmental Registry),

LAR (Rural Environmental License) or LAU (Unique Environmental License);

7. Existence of performance management and qualification of the field technician and relationship

with ranchers, through visits that the technician performs on the properties;

8. Marfrig presented the Table of Polygons (total compliant suppliers with the property limits -

polygons) at the units of the Amazon Biome, until December 31st 2014, for direct suppliers. The

maps are developed based in official monitored, reported and verified information, and Agrotools

evaluates all information carefully before generating the geometry map:

Tangará da Serra: 100% of the suppliers.

Paranatinga: 100% of the suppliers.

Rolim de Moura: 100% of the suppliers.

Chupinguaia: 100% of the suppliers.

Tucumã: 100% of the suppliers.

9. Although there was the option to perform the audit only through documents, Marfrig also opted

to perform a step in situ (in units of the Amazon Biome) to demonstrate the procedures

performed and the practical knowledge of local staff. This allowed visualization of the information

flow between the Corporate Sustainability team with the local team (Sustainability and Cattle

Purchase), as well as the company responsible geospatial for monitoring.

São Paulo, 27 April 2015

Felipe Lacerda Antunes – Auditor

11

ANNEX I –Complementary Verification of 2014 MARFRIG GLOBAL FOODS Work Plan

12

DNV GL confirms that meetings were held with the Prosecutor Dr. Daniel Azeredo and members from Legal Amazon Work Group, the meeting agenda were

Agenda Federal Attorney/Legal Amazon Workplan:

Indirect Suppliers– Target 1.4.1 and 5.5.1 – Indirect suppliers deforestation free

Actual situation related to indirect suppliers;

Activities implemented by Marfrig (RFI);

RFI tool and partial results presentation;

Presentation of four proposals and solutions suggested by the company, focusing on the Transit Animal Guide;

Indigenous people - 2.2.2 and 2.2.3 – List of Embargoed Areas (Indigenous Lands)

Actual situation related to Indigenous Lands;

Activities implemented by Marfrig, Geospatial Monitoring and consultation to public entities;

Presentation of proposals and solutions suggested by the company, focusing on the development of a list of embargoed areas due to indigenous conflicts / irregular invasion of Indigenous Lands;

Land conflicts - 4.1.2 – List of Embargoed Areas (Land conflicts/blocked CCIR)

Actual situation related to land conflicts;

Activities implemented by Marfrig, Land Legal Document Requests and consultation to public

entities;

Presentation of proposals and solutions suggested by the company, focusing on the development of a list of embargoed areas due to land conflicts and blocked CCIRs due to land conflicts and violence;

Other general discussion:

CAR consultation and transparency by SiCAR when publishing information;

Creation of a specific geospatial indicator for the Beef Chain

Consultation and transparency related to blocked CCIRs, authenticity of issued documents and improvement in search tools.

Met targets – Deadline December/2014

Target 1.1.1 – Met (all suppliers have 100% maps, as confirmed by the sampling applied during the audit);

Target 1.3.1 – Met (all informed indirect suppliers were checked at IBAMA list)

Target 3.2.1 – Met (all informed indirect suppliers were checked at MTE list)

Target 5.1.1 – Met (62,29% from the suppliers presented to Marfrig complete information about the

indirect suppliers)

Constant Targets

For these targets that already had implemented actions, DNV GL could verify during the audit that all actions are implemented and no mistakes were found.

13

SUMMARY TABLES

1. Total purchases and sampling 2014

1 - Total purchase and sampling (2014)

Total cattle purchase operation from the Amazon Biome by Marfrig from January 1st to

December 31st 2014

Total cattle purchase operation sampled for the assessment

2792* 281*

* Marfrig’s system does not allow the authomatic generation of total purchase operations in

Amazon Biome. In this sense, a sampling of 10% of the suppliers was defined. DNV GL considers

this sampling larger than the one based on purchase operations, since for each sampled supplier 100% of the purchases was indirectly checked for the period.

.

2. Non conformities 2014

2 – Non conformity(ies) (2014)

Criteria Total non-conformities

% non conformities related to total purchase 2014

% non conformities related to total sampled purchase operations

Cattle purchase operations from properties with an identified deforestation

after Oct/2009.

0 0 0

Blocked properties due to indigenous land invasion

0 0 0

Blocked properties due to

environmental conservation unit invasion

0 0 0

Blocked properties due to presence at MTE list

0 0 0

Blocked properties due to presence at IBAMA list

0 0 0

3. Blocking test result

3 – Blocking test

Total number of

purchase simulation tests in the system

Conformity Non-

conformity

IBAMA 11 11 0

MTE 18 18 0

GEO (Prodes, DETER, TI e UC)

16 16 0

14

DNV GL Contacts Felipe Lacerda Antunes – Auditor

Juliana Scalon – Reviewer

Kátia Nogueira – Comercial Manager

Det Norske Veritas

Av. Alfredo Egydio de Souza Aranha,100

Centro Administrativo Santo Amaro – CASA

Bloco D – 3º andar

04726-170 – São Paulo, SP

Brasil

Tel.: +55 11 3305 3305

Fax: +55 11 3305 3313

Vision

To have a global impact for a safe and sustainable future

Values

We never compromise on QUALITY or our INTEGRITY

We care about our CUSTOMERS and provide solutions that improve their business results

We create a working environment where PEOPLE are challenged to do their best

We acquire and share KNOWLEDGE and apply it to enhance value for our customers

We build our future through PROFITABLE operations, innovation and new business opportunities

We succeed by LIVING OUR VALUES

www.dnvgl.com