Appendix 3: Solomon Restricted Vegetation Assessment

MARCH 2015

FORTESCUE METALS GROUP LTD SOLOMON RESTRICTED VEGETATION ASSESSMENT SUPPLEMENTARY REPORT

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

Document Status

Approved for Issue Rev Author Reviewer Date Name Distributed to Date S Grein A C Pa rker M Macdonald 19/01/2015 R Young 19/01/2015 T Edwards 0 M Macdonald S Grein 11/03/2015 S Grein A Barker 11/03/2015 1 M Macdonald S Grein 13/03/2015 S Grein A Barker 13/03/2015

ecologia Environment (2015). Reproduction of this report in whole or in part by electronic, mechanical or chemical means including photocopying, recording or by any information storage and retrieval system, in any language, is strictly prohibited without the express approval of Fortescue Metals Group Limited.

Restrictions on Use This report has been prepared specifically for Fortescue Metals Group Limited. Neither the report nor its contents may be referred to or quoted in any statement, study, report, application, prospectus, loan, or other agreement document, without the express approval of Fortescue Metals Group Limited.

ecologia Environment Level 8, Carillon City Office Tower 207 Murray Street PERTH WA 6000 Phone: +61 (0) 8 6180 4450 Fax: +61 (0) 8 6180 4451 Email: [email protected]

March 2015

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

EXECUTIVE SUMMARY Fortescue Metals Group (Fortescue) is investigating options to expand the Solomon Hub. Baseline biological surveys have been completed to support environmental approval of this expansion. Several baseline flora and vegetation surveys have previously been conducted within and surrounding the Solomon Hub. Fortescue commissioned ecologia Environment (ecologia) in 2014 to consolidate the results of these surveys, verify existing mapping and survey any gaps identified. ecologia conducted a two phase Level 2 flora and vegetation assessment within these areas within the Solomon Hub. Two vegetation units described in the assessment is considered regionally restricted:  CddAdTw (Corymbia deserticola subsp. deserticola and Corymbia hamersleyana low sparse woodland over Hakea chordophylla tall isolated shrubs over Acacia dictyophleba mid sparse shrubland over Acacia tumida var. pilbarensis low sparse shrubland over Triodia wiseana open hummock grassland) represented by 343.0 ha, all within Solomon Hub; and  EllGwTw2 (Eucalyptus leucophloia subsp. leucophloia low open woodland over Grevillea wickhamii mid sparse shrubland over Triodia wiseana hummock grassland) represented by 82.6 ha within Solomon Hub and 4.6 ha outside Solomon Hub. A new regional floristic analysis was conducted for the Solomon Hub which included data collected at 4,254 quadrats from 33 flora and vegetation assessments. CddAdTw The new regional floristic analysis split the 10 quadrats representing CddAdTw amongst three new groups, in association with 22 quadrats from three additional vegetation units (five from ChAiTw4, two from EllApTw1, and 15 from EllHcTw2). However all 22 of these quadrats are also from within Solomon Hub, so no additional areas were identified in the wider region which may support CddAdTw. Based on the relationship between records of CddAdTw and environmental variables such as geology, land systems, topography and depth to regolith, a MaxEnt distribution model was developed which identified three areas with environmental conditions potentially suitable for vegetation similar to CddAdTw. The most prospective, but smaller of these areas is immediately north of Solomon Hub. The other two potentially larger occurrences of most suitable habitat are located further away, near the eastern boundary of Karijini National Park, between Dales Gorge and the Great Northern Highway (Figure 3.3). This area is located to the west of the Fig Tree flora and vegetation assessment study area. EllGwTw2 The new regional floristic analysis maintained the five quadrats representing EllGwTw2 in one coherent group, separate from quadrats from other vegetation units. However, as two of these quadrats represent resampling the same location, there are only four unique known locations of vegetation unit EllGwTw2. The other three locations of vegetation unit EllGwTw2 are from outside Solomon Hub, and areas supporting vegetation unit EllGwTw2 are mapped based on previous vegetation mapping and aerial imagery. In addition to the 4.3 ha of vegetation unit EllGwTw2 previously mapped outside Solomon Hub, an additional two areas were mapped based around EllGwTw2 quadrat locations, representing an additional 217.9 ha of this vegetation unit. Therefore the 82.6 ha of EllGwTw2 within Solomon Hub represents approximately 27.1% of the mapped distribution of this vegetation unit (total area of EllGwTw2 is now 304.8 ha).

March 2015 i

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 ii

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

1 INTRODUCTION Fortescue Metals Group (Fortescue) is investigating options to expand the Solomon Hub. Baseline biological surveys have been completed to support environmental approval of this expansion. Solomon Hub is located 40 km north of Tom Price in the Pilbara region (Western Australia). Several baseline flora and vegetation surveys have previously been conducted within and surrounding the Solomon Hub. Fortescue commissioned ecologia Environment (ecologia) in 2014 to consolidate the results of these surveys, verify existing mapping and survey any gaps identified. ecologia conducted a two phase Level 2 flora and vegetation assessment within these areas within the Solomon Hub (ecologia 2014e). Seventy vegetation units were described across the Solomon Hub. Two vegetation units described in the assessment are considered regionally restricted:  CddAdTw: Corymbia deserticola subsp. deserticola and Corymbia hamersleyana low sparse woodland over Hakea chordophylla tall isolated shrubs over Acacia dictyophleba mid sparse shrubland over Acacia tumida var. pilbarensis low sparse shrubland over Triodia wiseana open hummock grassland; and  EllGwTw2: Eucalyptus leucophloia subsp. leucophloia low open woodland over Grevillea wickhamii mid sparse shrubland over Triodia wiseana hummock grassland. Based on floristic analysis with data from 19 other flora and vegetation assessments in the region, the currently known extent of vegetation unit CddAdTw is restricted to 343.0 ha, with no known occurrence outside Solomon Hub, and EllGwTw2 is restricted to 82.6 ha, with 4.3 ha outside Solomon Hub. To provide further information on the possible regional extent of these vegetation units outside the Solomon Hub, this assessment was undertaken to supplement the existing information on the significance and potential regional distribution of vegetation units CddAdTw and EllGwTw2. This supplementary Assessment incorporates the following aspects: 1. Reassess the Solomon Hub regional floristic analysis (ecologia 2014e); Reassessment of the floristic analysis from the Solomon Hub flora and vegetation assessment to determine if the division between vegetation units CddAdTw and EllGwTw2 and other adjacent vegetation units is justified by the composition of the quadrats representing each of the vegetation units. 2. New regional floristic analysis; Complete a new regional floristic analysis incorporating an expanded dataset with floristic data, particularly from other areas that have outcropping Channel Iron Deposit (CID)/pisolitic geology, that may support similar vegetation to that represented by vegetation units CddAdTw and EllGwTw2. 3. Distribution modelling; and Develop distribution models using locations of quadrats representing the restricted vegetation units CddAdTw and EllGwTw2 incorporating suitable environmental attributes to identify other areas which share similar environmental conditions and therefore represent potentially suitable habitat for vegetation units CddAdTw and EllGwTw2. 4. Map potential distribution. Prepare maps showing potentially suitable habitat for the restricted vegetation units CddAdTw and EllGwTw2 and provide associated spatial data adequate for planning a field assessment to

March 2015 1

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

attempt to locate similar vegetation outside Solomon Hub and acquire additional quadrat data to incorporate into a future floristic analysis.

March 2015 2

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

2 METHODOLOGY

2.1 SOLOMON HUB FLORISTIC ANALYSIS The regional floristic analysis conducted for the Solomon Hub flora and vegetation assessment (ecologia 2014e) included data collected at 2,234 sites (2,500m2 quadrats) from 20 flora and vegetation assessments (listed in Table 2.1). The sites were compared using the Bray‐Curtis coefficient association matrix and displayed as a dendrogram using the PATN software package.

2.2 NEW REGIONAL FLORISTIC ANALYSIS A total of 4,254 quadrats were included in the new regional analysis. These quadrats are from 33 flora and vegetation assessments from the Pilbara (Table 2.1, Figure 2.1). The analysis was conducted in a method consistent with the Solomon Hub floristic analysis. Table 2.1 – Flora and vegetation assessments included in regional floristic assessment

Included in Solomon Number of Report Hub regional floristic Reference sites analysis? Solomon Hub Flora and Vegetation Assessment 281 Yes ecologia (2014e) Central Pilbara Project Mine Level 2 Flora and 203 Yes Ecoscape (2012a) Vegetation Assessment Central Pilbara Project Rail Vegetation and Flora 264 Yes ecologia (2012a) Assessment Flora and Vegetation Assessment, Solomon Project & 94 Yes Coffey (2010a) Investigator Flora and Vegetation Assessment, Solomon Rail 264 Yes Coffey (2010b) Project Investigator Project Flora and Vegetation Assessment 36 Yes ecologia (2014b) Level Two Flora and Vegetation Assessment, Firetail 25 Yes Ecoscape (2010a) Mining Area Mt MacLeod Flora and Vegetation Assessment 54 Yes ecologia (2014c) Solomon Project Rail Re‐alignment Flora & Vegetation 13 Yes Ecoscape (2010b) Assessment Solomon Project: Kings Flora and Vegetation 124 Yes ENV (2010b) Assessment Solomon South Flora and Vegetation Assessment 20 Yes ecologia (2014f) Stingray Flora and Vegetation Assessment 51 Yes ecologia (2014g) Solomon Regional Flora Assessment 4 Yes Coffey (2010c) Delphine Level 2 Flora and Vegetation Assessment 162 Yes Ecoscape (2012b) Eliwana and Flying Fish Level 2 Flora and Vegetation 161 Yes Ecoscape (2012c) Assessment Level 2 Flora and Vegetation Assessment – Western 150 Yes Astron (2013) Hub (The Edge) Mt Farquhar Level 2 Flora and Vegetation Assessment 60 Yes Ecoscape (2012d) Turner Syncline Flora and Vegetation Assessment 223 Yes ecologia (2012d) Turee Flora and Vegetation Assessment 306 No ecologia (2012c) Prairie Heights Flora and Vegetation Assessment 50 No ecologia (2014d) Indibiddy East and West Flora and Vegetation 92 No Ecoscape (2014) Assessment Fig Tree Flora and Vegetation Assessment 61 No ecologia (2014a) Nyidinghu [Mine] Flora and Vegetation Assessment 281 No Cardno (2012a)

March 2015 3

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

Included in Solomon Number of Report Hub regional floristic Reference sites analysis? Nyidinghu Rail Spur Flora and Vegetation Assessment 208 No Cardno (2012b) Stage A Rail Corridor 96 No Biota (2004a) Stage B Rail Corridor 217 No Biota (2004b) Cloudbreak Flora and Vegetation Assessment 27 No ENV (2011b) Christmas Creek Flora and Vegetation Assessment 20 No ENV (2010a) Christmas Creek Airstrip Flora, Vegetation and Fauna 15 No ENV (2011a) Assessment Christmas Creek Life of Mine Flora and Vegetation 73 No ENV (2012) Assessment Roy Hill Flora and Vegetation Assessment 255 No ecologia (2009) Kutayi Flora and Vegetation Assessment 96 No ecologia (2013) North Star Flora and Vegetation Assessment 272 No ecologia (2012b) Total 4,303

2.3 DISTRIBUTION MODELS The MaxEnt program (version 3.3.3) uses a machine‐learning approach to maximum entropy modelling of habitat suitability, with the advantage of being applicable for use with presence records and does not require absence data (Phillips et al. 2004; Phillips et al. 2006), making it suitable for incorporating records from multiple sources. Two separate models were developed, one for each of the two restricted vegetation units CddAdTw and EllGwTw2. All known records of vegetation unit CddAdTw and EllGwTw2 were included in the MaxEnt models with the following environmental variables:  Geology (1:250;000);  Land systems (Van Vreeswyk et al. 2004);  Digital Elevation Model (DEM) (Shuttle Radar Topography Mission (SRTM) derived DEM Version 1), approximately 90 x 90 m pixel size;  Regolith Depth (Wilford et al. 2014);  Topographic Position Index (Gallant and Austin 2012) ; and  Multi‐resolution Valley Bottom Floor (MrVBF) (Gallant et al. 2012). The MaxEnt models were run across an area approximately 450 x 160 km, centred around the Solomon Hub plus approximately 50 to 220 km (Figure 3), with 15 replicate models and a randomly selected 25% of examples withheld for each replicate to be used to test the models’ predictive power. The average test area under the receiver operating characteristic (ROC) curve (AUC) for the 15 replicates is 0.916 ± 0.030 (± standard deviation) for the CddAdTw model and 0.688 ± 0.192 (± standard deviation) for the EllGwTw2 model. The ROC measures model performance by plotting sensitivity (true positive rate) against 1 ‐ specificity (false positive rate) of the model. An AUC of 0.5 represents a random model, and an AUC of 1 represents a perfect model, but because the use of ROC curves with presence‐only data “negative examples” include all cells with no recorded occurrences, the maximum AUC is less than one, and is smaller the more wide ranging the species distribution (Phillips et al. 2004). The models included subsample replication, a maximum of 5,000 iterations, and previous flora and vegetation survey areas included to represent sampling bias and increased likelihood of sampling these areas (Elith et al. 2011). For all other settings, default values were retained.

March 2015 4

400000 600000 800000 Pindanland Legend McLarty CddAdTw 2 EllGwTw Roebourne Solomon Hub Mackay Flora and vegetation assessments in the floristic analysis Conservation Reserve

Chichester 7600000

Cape Range Hamersley Fortescue K Ashburton 0 30 60 Trainor Kilometres

7400000 Augustus Absolute Scale - 1:2,250,000 Figure: 2.1 Drawn: MJM Project ID: 1629 Locations of flora and vegetation assessments Date: 19/01/2015 Coordinate System included in the new floristic analysis Name: GDA 1994 MGA Zone 50 Projection: Transverse Mercator Datum: GDA 1994 A4

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 6

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

3 RESULTS

3.1 SOLOMON HUB FLORISTIC ANALYSIS

3.1.1 Vegetation Unit CddAdTw The relevant part of the dendrogram from the Solomon Hub regional floristic analysis, incorporating 2,234 quadrats from 20 flora and vegetation assessments, is shown in Figure 3.1. The most similar vegetation unit to CddAdTw is ChAiTw2 (Corymbia hamersleyana and Eucalyptus leucophloia subsp. leucophloia low sparse woodland over Acacia inaequilatera and Grevillea wickhamii tall sparse shrubland over Triodia wiseana hummock grassland) which has similar composition and occurs on a similar landform of low undulating hills but with a different combination of dominant taxa (Table 3.1). While both vegetation units contain a high frequency of Corymbia hamersleyana (90% and 100% respectively), only CddAdTw has a high frequency of Corymbia deserticola subsp. deserticola (100% and 5%, respectively). Similarly, both vegetation units have a high frequency of Acacia inaequilatera and Grevillea wickhamii, but only CddAdTw has a high frequency of Acacia dictyophleba). Other taxa with relatively high frequency in CddAdTw but not ChAiTw2 quadrats are: Acacia tumida var. pilbarensis, Eucalyptus gamophylla and Acacia tenuissima. These differences are considered sufficient to justify the discrimination of these two vegetation units.

Table 3.1 – Species composition of CddAdTw and ChAiTw2 from Solomon Hub (ecologia 2014e) Vegetation Unit 2 Taxon CddAdTw ChAiTw Quadrats Percent Quadrats Percent Corymbia deserticola subsp. deserticola 10 100 1 5 Corymbia hamersleyana 9 90 20 100 Acacia dictyophleba 9 80 0 0 Grevillea wickhamii 8 80 8 40 Acacia inaequilatera 6 60 14 70 Triodia wiseana 10 100 20 100 Eucalyptus leucophloia subsp. leucophloia 2 20 8 40 Hakea lorea subsp. lorea 5 50 5 25 Acacia pyrifolia 1 10 7 35 Hakea chordophylla 5 50 3 15 Acacia tumida var. pilbarensis 7 70 0 0 Eucalyptus gamophylla 6 60 1 5 Acacia pruinocarpa 0 0 4 20 Acacia tenuissima 4 40 0 0 Grevillea pyramidalis 0 0 4 20 Jasminum didymum subsp. lineare 3 30 1 5 Acacia hilliana 3 30 0 0 Eremophila longifolia 0 0 3 15 Gompholobium oreophilum 3 30 0 0 Senna artemisioides subsp. oligophylla 1 10 2 10 Acacia atkinsiana 2 20 0 0 Acacia bivenosa 2 20 0 0 Aristida holathera var. latifolia 2 20 0 0 Eriachne mucronata 0 0 2 10 Senna artemisioides subsp. helmsii 0 0 2 10 Acacia acradenia 0 0 1 5 Acacia adoxa var. adoxa 1 10 0 0

March 2015 7

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

3.1.2 Vegetation Unit EllGwTw2 The relevant part of the dendrogram from the Solomon Hub regional floristic analysis, incorporating 2,234 quadrats from 20 flora and vegetation assessments, is shown in Figure 3.2. The most similar vegetation unit to EllGwTw2 are vegetation units EllAaTw1 and EllAbTw3. Vegetation unit EllAaTw1 (Eucalyptus leucophloia subsp. leucophloia low sparse woodland, over Acacia arida, Acacia bivenosa and Senna glutinosa subsp. glutinosa mid sparse shrubland, over Triodia wiseana open hummock grassland) has similar composition with occurrences on a similar landform of low undulating hills (some occurrences are adjacent to the areas mapped as EllGwTw2 and other occurrences on low hills in the Solomon southern borefield) but with a different combination of dominant taxa (Table 3.2). While all three vegetation units (EllGwTw2, EllAaTw1 and EllAbTw3 contain high frequencies of Eucalyptus leucophloia subsp. leucophloia (100%, 96% and 100% respectively), but EllGwTw2 has a high frequency of Corymbia deserticola subsp. deserticola (60%), which is not present in either EllAaTw1 or EllAbTw3. Similarly, vegetation unit EllGwTw2 has a high frequency of Grevillea wickhamii (80%) and Gompholobium oreophilum (60%), neither of which is present in vegetation units EllAaTw1 or EllAbTw3. These differences are considered sufficient to justify the discrimination of these three vegetation units.

Table 3.2 – Species composition of EllGwTw2 and EllAaTw1 from Solomon Hub (ecologia 2014e) Vegetation Unit 2 1 3 Taxon EllGwTw EllAaTw EllAbTw Quadrats Percent Quadrats Percent Quadrats Percent Triodia wiseana 5 100 27 96.43 13 100 Eucalyptus leucophloia subsp. leucophloia 5 100 27 96.43 12 92.31 Acacia pruinocarpa 1 20 8 28.57 7 53.85 Acacia citrinoviridis 0 0 12 42.86 4 30.77 Acacia bivenosa 0 0 6 21.43 9 69.23 Senna glutinosa subsp. glutinosa 0 0 12 42.86 2 15.38 Acacia arida 0 0 10 35.71 1 7.69 Eriachne mucronata 0 0 3 10.71 7 53.85 Triodia sp. Robe River 0 0 10 35.71 0 0.00 Acacia maitlandii 2 40 2 7.14 3 23.08 Cymbopogon ambiguus 0 0 1 3.57 6 46.15 Hakea chordophylla 0 0 3 10.71 4 30.77 Acacia synchronicia 0 0 5 17.86 1 7.69 Corymbia deserticola subsp. deserticola 3 60 2 7.14 0 0 Petalostylis labicheoides 2 40 2 7.14 1 7.69 Acacia ancistrocarpa 0 0 4 14.29 1 7.69 Grevillea wickhamii 4 80 0 0 0 0 Acacia monticola 1 20 1 3.57 2 15.38 Acacia aptaneura 0 0 0 0 4 30.77 Jasminum didymum subsp. lineare 0 0 0 0 4 30.77 Ptilotus obovatus 0 0 2 7.14 2 15.38 Ptilotus rotundifolius 0 0 3 10.71 1 7.69 Senna glutinosa 0 0 0 0 4 30.77 Gompholobium oreophilum 3 60 0 0 0 0 Acacia atkinsiana 2 40 0 0 1 7.69 Keraudrenia nephrosperma/velutina 2 40 0 0 1 7.69 Dodonaea lanceolata 1 20 0 0 2 15.38 Hakea lorea subsp. lorea 1 20 1 3.57 1 7.69 Acacia aneura 0 0 3 10.71 0 0 Acacia inaequilatera 0 0 3 10.71 0 0 Corymbia hamersleyana 0 0 0 0 3 23.08

March 2015 8

Figure 3.1 Regional dendrogram from Solomon Hub (ecologia 2014e) Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report 0.0000 0.5468 1.0937 1.6405 2.1873 2 1873 0.0000 0.5468 1.0937 1.6405 Vegetation unit A-KINGS-FL-TY001 A-KINGS-FL-TY005 A-KINGS-FL-VOK010 A-KINGS-FL-VOK002 K-EL_FF-FL-WH12048 K-DE-FL-WH12050 K-EL_FF-FL-WH12153 K-SM_IV-FL-SERN-BSR05 K-SM-FL-SERN-BSR15 ChAiTw1 A-KINGS-FL-TY009 A-KINGS-FL-TY010 A-KINGS-FL-TY012 A-KINGS-FL-TY014 A-KINGS-FL-TY013 A-KINGS-FL-VOK020 A-KINGS-FL-VOK022 A-KINGS-FL-VOK030 A-KINGS-FL-VOK027 K-SM-FL-TRIN004 A-KINGS-FL-VOK004 K-SM_IV-FL-SERN-BSR04 A-KINGS-FL-VOQ045 ChApTe2 A-KINGS-FL-VOQ046 I-SR-FL-02RAILREL02 I-SR-FL-11RLREL2 I-SR-FL-11RLREL3 I-SR-FL-11RAIL102 K-SM_IV-FL-KR124 I-SR-FL-11RAIL105 I-SR-FL-11RAIL106A I-SR-FL-11RAIL169 Not within study area I-SR-FL-12RAIL104 K-SM_IV-FL-KR121A I-SR-FL-12RAIL108 I-SR-FL-12RAIL120a I-SR-FL-12RAIL120C I-SR-FL-12RAILRLC A-KINGS-FL-TY003 A-KINGS-FL-TY004 A-KINGS-FL-VOQ043 A-KINGS-FL-VOQ032 A-KINGS-FL-VOQ010 A-KINGS-FL-VOQ016 A-KINGS-FL-VOQ101 I-SR-FL-11RAILREL5 A-KINGS-FL-VOK001 B-TE-FL-EST23A ChAiTw2 A-KINGS-FL-VOK008 A-KINGS-FL-VOQ034 A-KINGS-FL-VOQ047 A-KINGS-FL-VOQ029 A-KINGS-FL-VOK007 A-KINGS-FL-VOQ042 K-EL_FF-FL-WH12142 A-KINGS-FL-VOQ028 A-KINGS-FL-VOQ033 A-KINGS-FL-VOQ100 A-KINGS-FL-VOQ011 A-KINGS-FL-VOQ017 A-KINGS-FL-VOQ021 A-KINGS-FL-VOQ025 A-KINGS-FL-ZION009 CddAdTw A-KINGS-FL-ZION011 A-KINGS-FL-VOQ030 K-SM-FL-VOQ30 J-SLOM-FL-136 K-SM-FL-VOQ21 A-KINGS-FL-TY007 A-KINGS-FL-VOK009 K-SM-FL-VOQ29 A-KINGS-FL-TY011a A-KINGS-FL-TY011 K-SM-FL-TRIN011 A-KINGS-FL-VOQ036 K-SM-FL-VOQ31 I-SR-FL-07RAIL035 ChAiTw4 K-CM-FL-s21030 J-CP-FL288 I-SR-FL-11RAIL106 K-SM-FL-VOQ43 I-SR-FL-11RAIL1101 K-SM_IV-FL-KR120 K-SM_IV-FL-KR121 K-SM_IV-FL-KR-BS-R1 K-SM_IV-FL-KR115 B-TE-FL-EST06A MN

March 2015 Figure 3.2 Regional dendrogram from Solomon Hub (ecologia 2014e) Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report B-TE-FL-EST36B MN I-SR-FL-02RAILREL03 A-KINGS-FL-VOQ040 B-TE-FL-EST 47BQ B-TE-FL-EVQ12 MN J-MM-FL-127 B-TE-FL-EST 07AQ EllAaTw1 J-MM-FL-136 B-TE-FL-EST 11BQ B-TE-FL-EST14 MN B-TE-FL-EST45 MN J-MM-FL-134 J-TS-FL-109 B-TE-FL-EST13A MN K-DE-FL-WH12249 B-TE-FL-EST22B B-TE-FL-EST26 MN B-TE-FL-EST30 MN B-TE-FL-EST30B B-TE-FL-EST35A B-TE-FL-EST47A B-TE-FL-EST 14AQ B-TE-FL-EST26B B-TE-FL-EST 18BQ K-DE-FL-WH12304 B-TE-FL-EVQ38 Not within study area B-TE-FL-EVQ45 B-TE-FL-EVQ46 B-TE-FL-EST41B B-TE-FL-EST49B B-TE-FL-EVQ04 B-TE-FL-EST 44AQ J_SR_FL_18 B-TE-FL-EST19B MN J_SR_FL_07 J_SR_FL_09 J_SR_FL_04 EllAbTw3 J_SR_FL_05 K-CM-FL-CP11017 K-DE-FL-WH12262 K-MF-FL-WH12133 K-MF-FL-WH12113 J_SR_FL_01 J_SR_FL_02 A-KINGS-FL-VOQ038 K-SM-FL-VOQ38 EllGwTw2 J-CP-FL249 J-SY-FL-251 J-SY-FL-272 A-KINGS-FL-TY003a A-KINGS-FL-VOK013 A-KINGS-FL-VOQ027 A-KINGS-FL-VOQ022 A-KINGS-FL-VOQ024 A-KINGS-FL-VOQ023 A-KINGS-FL-VOK017 A-KINGS-FL-VOK026 A-KINGS-FL-VOK023 ChApTe1 A-KINGS-FL-VOK029 I-SR-FL-KR008 A-KINGS-FL-VOQ044 I-SM-FL-SERN-BSR07 J-SLOM-FL-18 J-SLOM-FL-21 J-SLOM-FL-264 J-SLOM-FL-376 K-SM-FL-VOQ28 K-SM-FL-VOQ46 I-SM-FL-SERN108 I-SM-FL-SERN109 J_SR_FL_06 J_SR_FL_08 J-SLOM-FL-79 J-SLOM-FL-302 K-MF-FL-WH12058 K-DE-FL-WH12098 K-MF-FL-WH12124 I-SR-FL-06RAIL007a ChAtpTw I-SR-FL-06RAIL011 I-SR-FL-06RAIL010 J-SLOM-FL-117 J-SLOM-FL-23 J-SLOM-FL-3 J-SLOM-FL-4 J-SLOM-FL-13 K-MF-FL-WHJN02 J-SLOM-FL-248 J-SLOM-FL-362 I-SR-FL-07RAIL036 J-SLOM-FL-36

March 2015

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

3.2 NEW REGIONAL FLORISTIC ANALYSIS

3.2.1 Vegetation Unit CddAdTw The new dendrogram, incorporating 4,254 quadrats from 33 flora and vegetation assessments has split the 10 quadrats comprising vegetation unit CddAdTw into three separate, but reasonably closely related clusters (the relevant section of the dendrogram is provided in Figure 3.3). The first group of four CddAdTw quadrats clusters with five quadrats of vegetation unit ChAiTw4 (Corymbia hamersleyana low sparse woodland, over Acacia inaequilatera tall sparse shrubland over Senna glutinosa subsp. glutinosa mid sparse shrubland over Triodia wiseana hummock grassland), which is an adjacent cluster in the Solomon Hub floristic analysis (ecologia 2014e, Figure 3.1). The second group of four CddAdTw quadrats has clustered with two quadrats of vegetation unit EllApTw1 (Eucalyptus leucophloia subsp. leucophloia low open woodland, over Acacia pruinocarpa mid sparse shrubland, over Triodia wiseana open hummock grassland and isolated Eriachne mucronata tussock grasses). The third cluster consists of the remaining two quadrats of CddAdTw and 15 quadrats from the vegetation unit EllHcTw2 (Eucalyptus leucophloia subsp. leucophloia and Corymbia hamersleyana low sparse woodland over Hakea chordophylla mid sparse shrubland over Acacia hilliana and Acacia adoxa var. adoxa low sparse shrubland over Triodia wiseana hummock grassland). All 22 quadrats from vegetation units ChAiTw4, EllApTw1, and EllHcTw2 are from similar landforms within Solomon Hub: that is none of the 22 quadrats associated with CddAdTw quadrats in this new floristic analysis are from any of the 3,569 quadrats outside the Solomon Hub.

3.2.2 Vegetation Unit EllGwTw2 The new floristic analysis, incorporating 4,254 quadrats from 33 flora and vegetation assessments has maintained the coherence of the 5 quadrats comprising vegetation unit EllGwTw2 (the relevant section of the dendrogram is also provided in Figure 3.3). However, as two of these quadrats represent resampling the same location (K‐SM‐FL‐VOQ38 and A‐KINGS‐FL‐VOQ038) there are only four unique known locations of vegetation unit EllGwTw2. This location (represented by quadrats K‐ SM‐FL‐VOQ38 and A‐KINGS‐FL‐VOQ038) is within Solomon Hub. The other three locations of vegetation unit EllGwTw2 are from outside Solomon Hub. Locations of EllGwTw2 quadrats are provided in Table 3.3, and mapped in Figure 3.4. The area of contiguous vegetation around these quadrats was mapped and is provided in Table 3.3. The entire dendrogram from the new floristic analysis can be provided on request.

Table 3.3 – Locations of EllGwTw2 quadrats Area inside Area outside Quadrat Easting Northing Location Source Solomon Hub Solomon Hub (ha) (ha) J‐CP‐FL‐249 557758 7571128 Central Pilbara Rail ecologia (2012a) ‐ 4.3 J‐SY‐FL‐251 555936 7525588 Stingray ecologia (2014g) ‐ 146.9 J‐SY‐FL‐272 559728 7523606 Stingray ecologia (2014g) ‐ 70.9 K‐SM‐FL‐VOQ38 581801 7550429 Solomon Coffey (2010a) 82.6 ‐ A‐KINGS‐FL‐VOQ038 581801 7550429 Solomon ENV (2010b) N/A (Duplicate) ‐ Total 82.6 222.2 Datum: GDA 1994 MGA Zone 50

March 2015 11

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 12

Figure 3.3 Dendrogram of new regional floristic analysis Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

Vegetation Unit 0.0000 0.6122 1.2244 1.8366 0.0000 0.6122 1.2244 1.8366 (ecologia 2014e) ChAiTw1 A‐KINGS‐FL‐TY001 ChAiTw1 A‐KINGS‐FL‐TY005 ChAiTw1 A‐KINGS‐FL‐VOK010 ChAiTw1 A‐KINGS‐FL‐VOK002 ChAiTw1 K‐EL_FF‐FL‐WH12048 ChAiTw1 K‐DE_FL‐WH12050 ChAiTw1 K‐EL_FF‐FL‐WH12153 ChAiTw3 K‐DE_FL‐WH13057 ChAiTw3 K‐SM‐FL‐VOQ34 ChAiTw1 K‐SM_IV‐FL‐SERN‐BSR05 ChAiTw1 A‐KINGS‐FL‐TY009 ChAiTw1 A‐KINGS‐FL‐TY010 ChAiTw1 A‐KINGS‐FL‐TY012 ChAiTw1 A‐KINGS‐FL‐TY014 ChAiTw1 A‐KINGS‐FL‐TY013 ChAiTw1 A‐KINGS‐FL‐VOK020 ChAiTw1 A‐KINGS‐FL‐VOK022 ChAiTw1 A‐KINGS‐FL‐VOK030 ChApTe2 A‐KINGS‐FL‐VOK004 ChApTe2 K‐SM_IV‐FL‐SERN‐BSR04 ChApTe2 A‐KINGS‐FL‐VOQ045 ChApTe2 A‐KINGS‐FL‐VOQ046 ChApTe2 I‐SR‐FL‐02RAILREL02 ChApTe2 I‐SR‐FL‐11RLREL3 ChApTe2 I‐SR‐FL‐11RLREL2 ChAiTw2 A‐KINGS‐FL‐TY003 ChAiTw2 A‐KINGS‐FL‐TY004 ChAiTw2 A‐KINGS‐FL‐VOQ043 ChAiTw2 A‐KINGS‐FL‐VOQ032 ChAiTw2 A‐KINGS‐FL‐VOQ010 ChAiTw2 A‐KINGS‐FL‐VOQ016 ChAiTw2 A‐KINGS‐FL‐VOQ101 ChAiTw2 A‐KINGS‐FL‐VOK001 ChAiTw2 B‐TE‐FL‐EST23A ChAiTw2 A‐KINGS‐FL‐VOK008 ChAiTw2 A‐KINGS‐FL‐VOQ034 ChAiTw2 A‐KINGS‐FL‐VOQ047 ChAiTw2 A‐KINGS‐FL‐VOQ029 ChAiTw2 A‐KINGS‐FL‐VOK007 ChAiTw2 A‐KINGS‐FL‐VOQ042 ChAiTw2 K‐EL_FF‐FL‐WH12142 ChAiTw2 A‐KINGS‐FL‐VOQ028 ChAiTw2 A‐KINGS‐FL‐VOQ100 ChAiTw2 A‐KINGS‐FL‐VOQ033 ChAaTw J‐SLOM‐FL‐304 EllHcTw3 B‐TE‐FL‐EST 15AQ J‐NS‐FL‐1321_122 EllApTw1 I‐SM‐FL‐FT33 ChAiTw1 A‐KINGS‐FL‐VOK027 ChAiTw1 K‐SM‐FL‐TRIN004 ChAiTw4 A‐KINGS‐FL‐VOQ036 ChAiTw4 K‐SM‐FL‐VOQ31 G‐CC‐FL‐BBg_01 ChAiTw4 J‐CP‐FL288 J‐NS‐FL‐1321_275 CddAdTw A‐KINGS‐FL‐VOQ030 CddAdTw K‐SM‐FL‐VOQ30 CddAdTw J‐SLOM‐FL‐136 CddAdTw K‐SM‐FL‐VOQ21 ChAiTw4 A‐KINGS‐FL‐TY007 ChAiTw4 A‐KINGS‐FL‐VOK009 ChAiTw4 K‐SM‐FL‐VOQ29 ChAiTw4 A‐KINGS‐FL‐TY011 ChAiTw4 K‐SM‐FL‐TRIN011 EllHcTw1 A‐KINGS‐FL‐VOQ026 EllHcTw1 K‐SM‐FL‐VOQ26 EllHcTw1 J‐SLOM‐FL‐172 EllHcTw1 A‐KINGS‐FL‐ZION010 EllHcTw4 J‐SLOM‐FL‐380 EllHcTw4 J‐SLOM‐FL‐80 EllHcTw4 J‐SLOM‐FL‐67 EllHcTw4 I‐SR‐FL‐05RAIL033

March 2015 Figure 3.3 Dendrogram of new regional floristic analysis Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report EllHcTw4 J‐SLOM‐FL‐75M EllAbTw4 I‐SM‐FL‐SERN019 EllAiTw K‐SM_IV‐FL‐KR004 EllHcTw4 J‐SLOM‐FL‐309 J‐NS‐FL‐1321_159 EllAbTw4 J‐SLOM‐FL‐2 ChAiTw2 I‐SR‐FL‐11RAILREL5 J‐NS‐FL‐1321_027 J‐NS‐FL‐1321_113 J‐NS‐FL‐1321P2_028 ChAiTw4 I‐SR‐FL‐11RAIL106 H‐NY‐R‐FL‐C081 J‐KI‐FL‐KI_007 Not within study area J‐TS‐FL‐140 J‐NS‐FL‐1321_139 EgAaTe3 I‐SR‐FL‐05RAIL032 ChAiTw4 I‐SR‐FL‐11RAIL1101 ChAiTw4 K‐SM_IV‐FL‐KR120 ChAiTw4 K‐SM_IV‐FL‐KR121 ChAiTw4 K‐SM_IV‐FL‐KR‐BS‐R1 Not within study area B‐TE‐FL‐EST06A MN Not within study area B‐TE‐FL‐EST07B Not within study area J‐TS‐FL‐29 Not within study area K‐DE_FL‐WH12092 Not within study area K‐EL_FF‐FL‐WH12199 Not within study area K‐SM_IV‐FL‐PLDK05 ChAaTw K‐DE_FL‐WH12303 EllAaTw1 B‐TE‐FL‐EST22B EllAaTw1 B‐TE‐FL‐EST26 MN EllAaTw1 B‐TE‐FL‐EST30 MN EllAaTw1 B‐TE‐FL‐EST30B EllAaTw1 B‐TE‐FL‐EST13A MN EllAaTw1 K‐DE_FL‐WH12249 EllAaTw1 B‐TE‐FL‐EST35A EllAaTw1 B‐TE‐FL‐EST47A EllAaTw1 B‐TE‐FL‐EST 07AQ EllAaTw1 J‐MM‐FL‐136 EllAaTw1 B‐TE‐FL‐EST 11BQ EllAaTw1 B‐TE‐FL‐EST14 MN EllAaTw1 B‐TE‐FL‐EST45 MN EllAaTw1 J‐MM‐FL‐134 EllAaTw1 J‐TS‐FL‐109 EllAbTw3 B‐TE‐FL‐EST19B MN EllAbTw3 J_SR_FL_07 EllAbTw3 J_SR_FL_09 Not within study area B‐TE‐FL‐EST 14AQ Not within study area B‐TE‐FL‐EST26B Not within study area B‐TE‐FL‐EST 18BQ Not within study area K‐DE_FL‐WH12304 Not within study area B‐TE‐FL‐EVQ38 Not within study area B‐TE‐FL‐EVQ45 Not within study area B‐TE‐FL‐EVQ46 Not within study area B‐TE‐FL‐EST41B Not within study area B‐TE‐FL‐EST49B Not within study area B‐TE‐FL‐EVQ04 EllAaTw1 A‐KINGS‐FL‐VOK005 EllAaTw1 A‐KINGS‐FL‐VOK006 EllAaTw1 B‐TE‐FL‐EST10B EllAaTw1 A‐KINGS‐FL‐VOQ031 EllAaTw1 B‐TE‐FL‐EVQ35 MN EllAaTw1 B‐TE‐FL‐EST11A EllAaTw1 B‐TE‐FL‐EST16B EllAaTw1 B‐TE‐FL‐EST36B MN EllAaTw1 I‐SR‐FL‐02RAILREL03 EllAaTw1 A‐KINGS‐FL‐VOQ040 EllGwTw2 A‐KINGS‐FL‐VOQ038 EllGwTw2 K‐SM‐FL‐VOQ38 EllGwTw2 J‐CP‐FL249 EllGwTw2 J‐SY‐FL‐251 EllGwTw2 J‐SY‐FL‐272 CddAdTw A‐KINGS‐FL‐VOQ011 CddAdTw A‐KINGS‐FL‐VOQ017 CddAdTw A‐KINGS‐FL‐VOQ021 CddAdTw A‐KINGS‐FL‐VOQ025 EllApTw1 J‐SY‐FL‐284

March 2015 Figure 3.3 Dendrogram of new regional floristic analysis Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report EllAbTw1 A‐KINGS‐FL‐VOQ015 EllAdTw2 A‐KINGS‐FL‐VOK011 EllAdTw2 A‐KINGS‐FL‐VOQ018 EllAdTw2 A‐KINGS‐FL‐ZION003 EllAdTw2 A‐KINGS‐FL‐ZION041 EllAdTw2 A‐KINGS‐FL‐ZION008 EllAdTw2 A‐KINGS‐FL‐VOK012 EllAdTw2 J‐SLOM‐FL‐360 EllAdTw2 J‐SLOM‐FL‐343 EllAdTw2 K‐SM_IV‐FL‐KR005 EllAdTw2 A‐KINGS‐FL‐VOK016 EllAdTw2 A‐KINGS‐FL‐ZION001 EllHcTw3 K‐CM‐FL‐s21020 EllHcTw3 K‐CM‐FL‐s21064 EllAdTw2 I‐SM‐FL‐FT27 EllAdTw2 I‐SM‐FL‐FT36 EllAdTw2 I‐SM‐FL‐FT29 ChAaTw J‐SLOM‐FL‐342 ChAaTw A‐KINGS‐FL‐ZION013 ChAaTw A‐KINGS‐FL‐ZION021 ChAaTw A‐KINGS‐FL‐ZION022 ChAaTw A‐KINGS‐FL‐ZION015 ChAaTw J_SR_FL_17 ChAaTw I‐SM‐FL‐FT07 EllHcTw2 A‐KINGS‐FL‐ZION007 EllHcTw2 I‐SM‐FL‐FT32 ChAaTw A‐KINGS‐FL‐ZION014 ChAaTw A‐KINGS‐FL‐ZION018 ChAaTw A‐KINGS‐FL‐ZION023 ChAaTw A‐KINGS‐FL‐ZION026 ChAaTw A‐KINGS‐FL‐ZION012 ChAaTw A‐KINGS‐FL‐ZION017 ChAaTw A‐KINGS‐FL‐ZION025 EllHcTw2 A‐KINGS‐FL‐ZION004 EllHcTw2 A‐KINGS‐FL‐ZION006 ChAaTw A‐KINGS‐FL‐ZION030 ChAaTw I‐SR‐FL‐06RAIL008 ChAaTw I‐SM‐FL‐FT11 ChAaTw I‐SM‐FL‐FT24 ChAaTw J_SR_FL_12 ChAaTw I‐SM‐FL‐FT25 ChAaTw I‐SM‐FL‐FT31 ChAaTw J_SR_FL_13 ChAaTw J‐SLOM‐FL‐341 ChAaTw J‐SLOM‐FL‐31 ChAaTw J‐SLOM‐FL‐6 ChAaTw A‐KINGS‐FL‐ZION029 ChAaTw B‐TE‐FL‐EST15A ChAaTw K‐DE_FL‐WH13066 EllGwTw1 I‐SM‐FL‐FT03 EllGwTw1 I‐SM‐FL‐FT15 EllGwTw1 J_SR_FL_20 EllGwTw1 J_SR_FL_11 EllGwTw1 J_SR_FL_03 EllGwTw1 I‐SM‐FL‐FT28 EllGwTw1 J_SR_FL_19 EllGwTw1 K‐SM_IV‐FL‐KR001 CddAdTw A‐KINGS‐FL‐ZION009 CddAdTw A‐KINGS‐FL‐ZION011 EllHcTw2 J_SR_FL_15 EllHcTw2 A‐KINGS‐FL‐ZION019 EllHcTw2 A‐KINGS‐FL‐ZION027 EllHcTw2 J_SR_FL_16 EllHcTw2 A‐KINGS‐FL‐VOK021 EllHcTw2 K‐SM‐FL‐VOQ47 EllHcTw2 J_SR_FL_10 EllHcTw2 K‐CM‐FL‐CP11072 EllHcTw2 A‐KINGS‐FL‐VOK024 EllHcTw2 A‐KINGS‐FL‐VOK025 EllHcTw2 K‐SM‐FL‐VOQ39 EllHcTw2 A‐KINGS‐FL‐VOK028 EllHcTw2 A‐KINGS‐FL‐VOQ013 EllHcTw2 A‐KINGS‐FL‐ZION020 EllHcTw2 A‐KINGS‐FL‐ZION024 EllHcTw2 I‐SM‐FL‐FT05

March 2015

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 16

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

3.3 DISTRIBUTION MODELS

3.3.1 Vegetation Unit CddAdTw The MaxEnt model of the distribution of potentially suitable habitat for vegetation unit CddAdTw returned several areas of high habitat suitability outside Solomon Hub. The most likely area identified is adjacent to the north of Solomon Hub (Figure 3.5). Two more extensive areas of potentially suitable habitat are present either side of the eastern boundary of Karijini National Park between Dales Gorge and the Great Northern Highway. These areas area also close to the western boundary of the Fig Tree flora and vegetation study area, which was included in the floristic analysis, but none of the quadrats matched CddAdTw. The analysis of the contribution of the environmental variables incorporated into the model shows that the Multi‐resolution Valley Bottom Flatness layer provided the most information for the model, representing 44.1% of model performance. Regolith depth and Geology (1:250,000), were the next most important variables, contributing 21.4% and 21.2% of model performance. Land systems provided 10.6% of the model and the Topographic Position Index and Digital Elevation Model provided 2.4% and 0.3% of the model performance respectively. The jackknife analyses of the contribution of environmental variables to the model show that the model performance is most reduced if the Geology (1:250:000) layer is excluded from the model, suggesting it provides the most information not included in the other variables. Next in contributing unique information to the model are Land system, Depth to Regolith and Multi‐resolution Valley Bottom Flatness variables. The variables with the highest gain across training, test and AUC of test data is Land systems and Geology (1:250,000). The raw output from the CddAdTw MaxEnt model is provided in Appendix A.

3.3.2 Vegetation Unit EllGwTw2 The MaxEnt model of the distribution of potentially suitable habitat for vegetation unit EllGwTw2 was based on the four known locations (quadrats) of this vegetation unit. While MaxEnt is capable of producing useful results with sample sizes as small as five (Hernandez et al. 2006), the AUC of 0.688 (± 0.192 standard deviation) for the EllGwTw2 model is not considered sufficient to predict new locations, given that the AUC for a random model is 0.5. The raw output form the EllGwTw2 MaxEnt model is provided in Appendix B.

March 2015 17

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 18

560000 580000 600000 620000 640000

7580000 Legend

2 ) EllGwTw quadrat

2 EllGwTw mapped in (ecologia 2014e)

J-CP-FL-249 2 ) EllGwTw new polygons

Solomon Hub and Regional Vegetation Mapping (ecologia 2014e)

Solomon Hub

Fig Tree 7560000

A-KINGS-FL-VOQ038 K-SM-FL-VOQ38 )) 7540000 K J-SY-FL-251 ) 0 5 10 J-SY-FL-272 ) Kilometres Absolute Scale - 1:250,000

7520000 Figure: 3.4 Drawn: MJM Project ID: 1633 Date: 13/03/2015 2 EllGwTw quadrats and polygons Coordinate System Name: GDA 1994 MGA Zone 50 Projection: Transverse Mercator Datum: GDA 1994 A3 400000 500000 600000 700000 800000 Legend Roebourne CddAdTw

Solomon Hub

Fig Tree

ConservationRoebourne Reserve

Chichester MaxEnt target areas

MaxEnt model probability High : 0.96 7600000 Low : 0.00

Fortescue Cape Range Hamersley 7500000 K Ashburton 0 25 50

Kilometres Absolute Scale - 1:1,250,000 7400000 Augustus

Figure: 3.5 Drawn: MJM Project ID: 1633 Date: 19/01/2015

CddAdTw MaxEnt model output Coordinate System Name: GDA 1994 MGA Zone 50 Projection: Transverse Mercator Datum: GDA 1994 A3

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

4 DISCUSSION

4.1 SOLOMON HUB FLORISTIC ANALYSIS

4.1.1 Vegetation Unit CddAdTw The species composition of the quadrats that represent vegetation unit CddAdTw and its most similar vegetation unit ChAiTw2 (as defined in the Solomon Hub floristic analysis, ecologia 2014e) are similar, sharing a relatively high frequency of several dominant species, including Corymbia hamersleyana, Acacia inaequilatera, Grevillea wickhamii and Triodia wiseana (ecologia 2014e). However, CddAdTw also consistently has higher frequency of several other dominant species, most notably Corymbia deserticola subsp. deserticola, Acacia dictyophleba, Acacia tumida var. pilbarensis, Eucalyptus gamophylla and Acacia tenuissima. The relatively consistent nature of these differences provides sufficient justification for retaining the distinction between the vegetation units, consistent with the detail required for NVIS Level V description (ESCAVI 2003).

4.1.2 Vegetation Unit EllGwTw2 The species composition of the quadrats that represent vegetation unit EllGwTw2 and its most similar vegetation units EllAaTw1 and EllAbTw3 (as defined in the Solomon Hub floristic analysis, ecologia 2014e) are similar, sharing a relatively high frequency of the dominant species, Eucalyptus leucophloia subsp. leucophloia and Triodia wiseana (ecologia 2014e). However, EllGwTw2 also consistently has higher frequency of several other dominant species, most notably Grevillea wickhamii and Gompholobium oreophilum. The relatively consistent nature of these differences provides sufficient justification for retaining the distinction between the vegetation units, consistent with the detail required for NVIS Level V description (ESCAVI 2003).

4.2 NEW REGIONAL FLORISTIC ANALYSIS

4.2.1 Vegetation Unit CddAdTw The new regional floristic analysis split the 10 quadrats representing CddAdTw amongst three new groups, in association with 22 quadrats from three additional vegetation units (five from ChAiTw4, two from EllApTw1, and 15 from EllHcTw2). However all 22 of these quadrats are also from within Solomon Hub, so no additional areas were identified in the wider region which may support CddAdTw.

4.2.2 Vegetation Unit EllGwTw2 The new regional floristic analysis maintained the five quadrats representing EllGwTw2 in one coherent group, separate from quadrats from other vegetation units. However as one two of these quadrats represent resampling of the same location, there are only four unique known locations of vegetation unit EllGwTw2. The other three locations of vegetation unit EllGwTw2 are from outside Solomon Hub, and areas supporting vegetation unit EllGwTw2 are quantified based on previous vegetation mapping and aerial imagery. In addition to the 4.3 ha of vegetation unit EllGwTw2 previously mapped outside Solomon Hub, there are an additional two areas, representing an additional 217.9 ha of this vegetation unit. Therefore the 82.6 ha of EllGwTw2 within Solomon Hub represents approximately 27.1% of the mapped distribution of this vegetation unit.

March 2015 21

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

4.3 DISTRIBUTION MODELS

4.3.1 Vegetation Unit CddAdTw Based on the relationship between records of CddAdTw and environmental variables such as geology, land systems, topography and depth to regolith, the MaxEnt model has identified three areas with environmental conditions potentially suitable for vegetation similar to CddAdTw. The most prospective, but smaller of these areas is immediately north of Solomon Hub. The other two potentially larger occurrences of most suitable habitat are located further away, near the eastern boundary of Karijini National Park, between Dales Gorge and the Great Northern Highway (Figure 3.3). This area is located to the west of the Fig Tree flora and vegetation assessment study area. Field assessment of these potentially suitable areas would be required to determine if similar vegetation to CddAdTw occurs. Such field assessment would need to be conducted consistent with relevant guidance (EPA 2004), particularly in quadrat size and survey timing. Data from any such field assessment would then require to be analysed in conjunction with existing data from CddAdTw quadrats and other Pilbara data to determine floristic relationships.

4.3.2 Vegetation Unit EllGwTw2 The relatively small number of records of vegetation units EllGwTw2 did not allow the development of a robust distribution model, with an AUC of 0.688 (± 0.192 standard deviation) for the EllGwTw2 model which is not considered sufficient to predict new locations, given that an AUC of 0.5 is expected for a random model.

March 2015 22

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

5 CONCLUSIONS CddAdTw The revised regional floristic analysis did not identify any additional locations of vegetation unit CddAdTw, however the distribution model identified potentially suitable habitat for the vegetation unit at three additional areas outside the Solomon Hub, based primarily on land system, geology, regolith and topography. EllGwTw2 The new regional floristic analysis confirmed the original (ecologia 2014e) regional floristic analysis, but additional mapping around two of the three known locations outside Solomon Hub delineated an additional 217.8 ha, which gives a total area of vegetation unit EllGwTw2 305.0 ha, of which 82.6 ha (27.1%) is within Solomon Hub.

March 2015 23

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 24

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

6 REFERENCES

Astron. 2013. Level 2 Flora and Vegetation Assessment Western Hub (The Edge). Unpublished report for Fortescue Metals Group Limited. Biota Environmental Sciences. 2004a. Fortescue Metals Group Stage A Rail Corridor. Unpublished report prepared for Fortescue Metals Group. December 2004 Biota Environmental Sciences. 2004b. Fortescue Metals Group Stage B Rail Corridor, Christmas Creek, Mt Lewin, Mt Nicolas and Mindy Mindy Mine Areas. Unpublished report prepared for Fortescue Metals Group. December 2004 Cardno. 2012a. Nyidinghu Flora and Vegetation Assessment. Unpublished Report for Fortescue Metals Group Ltd. Cardno. 2012b. Nyidinghu Rail Spur Flora and Vegetation Assessment. Unpublished Report for Fortescue Metals Group Ltd. Coffey. 2010a. Flora and Vegetation Assessment, Solomon Project and Investigator. Unpublished report prepared for Fortescue Metals Group Ltd. Coffey Environments. Coffey. 2010b. Flora and Vegetation Assessment, Solomon Rail Project. Unpublished report prepared for Fortescue Metals Group Ltd. Coffey Environments. Coffey Environments. 2010c. Solomon Regional Flora Assessment. Unpublished Report for Fortescue Metals Group Ltd. ecologia Environment. 2009. Roy Hill 1 Vegetation and Flora Assessment. A report by ecologia Environment for Hancock Prospecting Pty Ltd. ecologia. 2012a. Central Pilbara Project ‐ Rail Vegetation and Flora Assessment. Unpublished Report for Fortescue Metals Group Ltd. ecologia Environment. ecologia Environment. 2012b. North Star flora and vegetation assessment. Unpublished report prepared for Fortescue Metals Group Limited. ecologia. 2012c. Turee Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Group Ltd. ecologia Environment. ecologia. 2012d. Turner Syncline Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Group Ltd. ecologia Environment. ecologia. 2013. Kutayi Flora and Vegetation Assessment. Unpublished report by ecologia Environment for Fortescue Metals Group Ltd. ecologia. 2014a. Fig Tree Flora and Vegetation Assessment. Unpublished report by ecologia Environment for Fortescue Metals Group Ltd. ecologia. 2014b. Investigator Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Ltd. ecologia Environment. ecologia. 2014c. Mt MacLeod Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Group Ltd. ecologia Environment. ecologia. 2014d. Prairie Heights Flora and Vegetation Assessment. Unpublished report by ecologia Environment for Fortescue Metals Group Ltd. ecologia. 2014e. Solomon Hub Flora and Vegetation Assessment. Unpublished report by ecologia Environment for Fortescue Metals Group Ltd.

March 2015 25

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report ecologia. 2014f. Solomon South Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Group Ltd. ecologia Environment. ecologia. 2014g. Stingray Flora and Vegetation Assessment. ecologia Environment. Ecoscape. 2010a. Level Two Flora and Vegetation Assessment, Firetail Mining Area. Unpublished report for Fortescue Metals Group Ltd. Ecoscape. Ecoscape. 2010b. Solomon Project Rail Re‐alignment, Flora and Vegetation Assessment. Ecoscape Australia. Ecoscape. 2012a. Central Pilbara Project Level 2 Flora and Vegetation Assessment. Unpublished Report for Fortescue Metals Group Limited. Ecoscape. Ecoscape. 2012b. Delphine Level 2 Flora and Vegetation Survey. Unpublished Report for Fortescue Metals Group Limited. Ecoscape. 2012c. Eliwana and Flying Fish Level 2 Flora and Vegetation Assessment. Unpublished Report for Fortescue Metals Group Limited. Ecoscape. Ecoscape. 2012d. Mt Farquhar Level 2 Flora and Vegetation Survey. Unpublished Report for Fortescue Metals Group Limited. Ecoscape. 2014. Indibiddy East and West Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Group Ltd. Elith, J., Phillips, S. J., Hastie, T., Dudik, M., Chee, Y. E., and Yates, C. J. 2011. A statistical explanation of MaxEnt for ecologists. Diversity and distributions. 17:43‐57. ENV. 2010a. Christmas Creek Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Group Ltd. Env Australia. ENV. 2010b. Solomon Project: Kings Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Group Ltd. ENV Australia. ENV. 2011a. Christmas Creek Airstrip Flora, Vegetation and Fauna Assessment. Unpublished report for Fortescue Metals Group Ltd. Env Australia. ENV. 2011b. Cloudbreak Flora and Vegetation Assessment. Unpublished report for Fortescue Metals Group Ltd. ENV Australia. ENV. 2012. Christmas Creek Life of Mine Flora and Vegetation Assessment. Unpublished Report by ENV Autralia for Fortescue Metals Group Ltd. EPA. 2004. Guidance for the Assessment of Environmental Factors (in accordance with the Environmental Protection Act 1986). Terrestrial Flora and Vegetation Surveys for Environmental Impact Assessment in Western Australia. No. 51. June. Environmental Protection Authority, Western Australia. ESCAVI. 2003. Australian Vegetation Attribute Manual: National Vegetation Information System (NVIS). Executive Steering Committee for Australian Vegetation Information System (ESCAVI), Australian Government Department of Environment and Heritage, Canberra. Gallant, J. and Austin, J. 2012. Topographic Position Index (3" resolution) derived from 1" SRTM DEM‐ S. v3. CSIRO Australia. Gallant, J., Dowling, T., and Austin, J. 2012. Multy‐resolution Valley Bottom Flatness (MrVBF, 3" reolution) ‐ v2. CSIRO Australia.

March 2015 26

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

Hernandez, P. A., Graham, C. H., Master, L. L., and Albert, D. L. 2006. The effect of sample size and species characteristics on performance of different species distribution modeling methods. Ecography. 29:773‐785. Phillips, S. J., Anderson, R. P., and Schapire, R. E. 2006. Maximum entropy modeling of species geographic distributions. Ecological Mmodelling. 190:231‐259. Phillips, S. J., Dudik, M., and Schapire, R. E. 2004. A maximum entropy approach to species distribution modeling. Proceedings of the twenty‐first international conferecne on machine learning:655‐662. Van Vreeswyk, A. M. E., Payne, A. L., Leighton, K. A., and Hennig, P. 2004. An inventory and condition survey of the Pilbara region, Western Australia. Technical Bulletin No. 92. Department of Agriculture, Western Australia. Wilford, J., Searle, R., Thomas, M., and Grundy, M. 2014. Soil and Landscape Grid National Soil Attribute Maps ‐ Depth of Regolith (3" resolution) ‐ Release 1. v3. CSIRO Australia.

March 2015 27

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 28

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

APPENDIX A CddAdTw MaxEnt Model Output

March 2015 29

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 30

Replicated maxent model for CddAdTw Replicated maxent model for CddAdTw

This page summarizes the results of 15 split-sample models for CddAdTw, created Fri Jan 16 11:05:38 WST 2015 using Maxent version 3.3.3k. The individual models are here: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]

Analysis of omission/commission

The following picture shows the test omission rate and predicted area as a function of the cumulative threshold, averaged over the replicate runs. The omission rate should be close to the predicted omission, because of the definition of the cumulative threshold.

The next picture is the receiver operating characteristic (ROC) curve for the same data, again averaged over the replicate runs. Note that the specificity is defined using predicted area, rather than true commission (see the paper by Phillips, Anderson and Schapire cited on the help page for discussion of what this means). The average test AUC for the replicate runs is 0.916, and the standard deviation is 0.030.

file:///R|/...mon%20Restricted%20Vegetation%20Supplementary%20Report/03%20Reference/MaxEntOutput/CddAdTw1/CddAdTw.html[19/01/2015 11:47:01 AM] Replicated maxent model for CddAdTw

Pictures of the model

The following two pictures show the point-wise mean and standard deviation of the 15 output grids. Other available summary grids are min, max and median.

file:///R|/...mon%20Restricted%20Vegetation%20Supplementary%20Report/03%20Reference/MaxEntOutput/CddAdTw1/CddAdTw.html[19/01/2015 11:47:01 AM] Replicated maxent model for CddAdTw

Response curves

These curves show how each environmental variable affects the Maxent prediction. The curves show how the logistic prediction changes as each environmental variable is varied, keeping all other environmental variables at their average sample value. Click on a response curve to see a larger version. Note that the curves can be hard to interpret if you have strongly correlated variables, as the model may depend on the correlations in ways that are not evident in the curves. In other words, the curves show the marginal effect of changing exactly one variable, whereas the model may take advantage of sets of variables changing together. The curves show the mean response of the 15 replicate Maxent runs (red) and and the mean +/- one standard deviation (blue, two shades for categorical variables).

In contrast to the above marginal response curves, each of the following curves represents a different model, namely, a Maxent model created using only the corresponding variable. These plots reflect the dependence of predicted suitability both on the selected variable and on dependencies induced by correlations between the selected variable and other variables. They may be easier to interpret if there are strong correlations between variables.

file:///R|/...mon%20Restricted%20Vegetation%20Supplementary%20Report/03%20Reference/MaxEntOutput/CddAdTw1/CddAdTw.html[19/01/2015 11:47:01 AM] Replicated maxent model for CddAdTw

Analysis of variable contributions

The following table gives estimates of relative contributions of the environmental variables to the Maxent model. To determine the first estimate, in each iteration of the training algorithm, the increase in regularized gain is added to the contribution of the corresponding variable, or subtracted from it if the change to the absolute value of lambda is negative. For the second estimate, for each environmental variable in turn, the values of that variable on training presence and background data are randomly permuted. The model is reevaluated on the permuted data, and the resulting drop in training AUC is shown in the table, normalized to percentages. As with the variable jackknife, variable contributions should be interpreted with caution when the predictor variables are correlated. Values shown are averages over replicate runs.

Variable Percent contribution Permutation importance 1629landsystem90mmask 33.6 10.6 1629250geology90mmask 32.2 21.2 2_1629vbf90mmask 18.4 44.1 2_1629regolithdepth90mmask 11.5 21.4 2_1629dem90mmask 2.6 0.3 2_1629tpi90mmask 1.7 2.4

The following picture shows the results of the jackknife test of variable importance. The environmental variable with highest gain when used in isolation is 1629landsystem90mmask, which therefore appears to have the most useful information by itself. The environmental variable that decreases the gain the most when it is omitted is 1629250geology90mmask, which therefore appears to have the most information that isn't present in the other variables. Values shown are averages over replicate runs.

file:///R|/...mon%20Restricted%20Vegetation%20Supplementary%20Report/03%20Reference/MaxEntOutput/CddAdTw1/CddAdTw.html[19/01/2015 11:47:01 AM] Replicated maxent model for CddAdTw

The next picture shows the same jackknife test, using test gain instead of training gain. Note that conclusions about which variables are most important can change, now that we're looking at test data.

Lastly, we have the same jackknife test, using AUC on test data.

Command line to repeat this species model: java density.MaxEnt nowarnings noprefixes -E "" -E CddAdTw responsecurves jackknife "outputdirectory=R:\100 Projects\15.0-2014-'15\1629 FMG Solomon Restricted Vegetation Supplementary Report\03 Reference\MaxEntOutput\CddAdTw1" "samplesfile=R:\100 Projects\15.0-2014-'15\1629 file:///R|/...mon%20Restricted%20Vegetation%20Supplementary%20Report/03%20Reference/MaxEntOutput/CddAdTw1/CddAdTw.html[19/01/2015 11:47:01 AM] Replicated maxent model for CddAdTw

FMG Solomon Restricted Vegetation Supplementary Report\03 Reference\Restricted Veg Unit Quadrat Locations CddAdTw DDddd.csv" "environmentallayers=G:\Projects\15.0-2014-'15 projects\1629 FMG Solomon Restricted Vegetation Supplementary Report\Map Data\ASCII" randomseed nowriteclampgrid randomtestpoints=25 "biasfile=G:\Projects\15.0-2014-'15 projects\1629 FMG Solomon Restricted Vegetation Supplementary Report\Map Data\Bias\bias_all_dissolve_wgs84_clipptr3.asc" replicates=15 replicatetype=subsample nooutputgrids maximumiterations=5000 biastype=3 -N 1629dem90mmask -N 1629regolithdepth90mmask -N 1629tpi90mmask -N 1629vbf90mmask -t 1629250geology90mmask -t 1629landsystem90mmask

file:///R|/...mon%20Restricted%20Vegetation%20Supplementary%20Report/03%20Reference/MaxEntOutput/CddAdTw1/CddAdTw.html[19/01/2015 11:47:01 AM]

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

APPENDIX B EllGwTw2 MaxEnt Model Output

March 2015 37

Fortescue Metals Group Ltd Solomon Restricted Vegetation Assessment Supplementary Report

This page has been left blank intentionally.

March 2015 38

Replicated maxent model for EllGwTw2 Page 1 of 5

Replicated maxent model for EllGwTw2

This page summarizes the results of 15 split-sample models for EllGwTw2, created Mon Mar 09 16:54:24 WST 2015 using Maxent version 3.3.3k. The individual models are here: [0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]

Analysis of omission/commission

file:///G:/Projects/15.0-2014-'15%20projects/1633%20FMG%20Solomon%20Restrict... 10/03/2015 Replicated maxent model for EllGwTw2 Page 2 of 5

Pictures of the model

The following two pictures show the point-wise mean and standard deviation of the 15 output grids. Other available summary grids are min, max and median.

file:///G:/Projects/15.0-2014-'15%20projects/1633%20FMG%20Solomon%20Restrict... 10/03/2015 Replicated maxent model for EllGwTw2 Page 3 of 5

Response curves

Analysis of variable contributions

The following table gives estimates of relative contributions of the environmental variables to the Maxent

file:///G:/Projects/15.0-2014-'15%20projects/1633%20FMG%20Solomon%20Restrict... 10/03/2015 Replicated maxent model for EllGwTw2 Page 4 of 5

model. To determine the first estimate, in each iteration of the training algorithm, the increase in regularized gain is added to the contribution of the corresponding variable, or subtracted from it if the change to the absolute value of lambda is negative. For the second estimate, for each environmental variable in turn, the values of that variable on training presence and background data are randomly permuted. The model is reevaluated on the permuted data, and the resulting drop in training AUC is shown in the table, normalized to percentages. As with the variable jackknife, variable contributions should be interpreted with caution when the predictor variables are correlated. Values shown are averages over replicate runs.

Variable Percent contribution Permutation importance 1629landsystem90mmask 51.7 51.2 1629250geology90mmask 45.7 45.8 2_1629dem90mmask 2 1.5 2_1629vbf90mmask 0.3 0 2_1629tpi90mmask 0.2 1.4 2_1629regolithdepth90mmask 0.1 0.1

The following picture shows the results of the jackknife test of variable importance. The environmental variable with highest gain when used in isolation is 1629landsystem90mmask, which therefore appears to have the most useful information by itself. The environmental variable that decreases the gain the most when it is omitted is 1629landsystem90mmask, which therefore appears to have the most information that isn't present in the other variables. Values shown are averages over replicate runs.

Lastly, we have the same jackknife test, using AUC on test data.

file:///G:/Projects/15.0-2014-'15%20projects/1633%20FMG%20Solomon%20Restrict... 10/03/2015 Replicated maxent model for EllGwTw2 Page 5 of 5

Command line to repeat this species model: java density.MaxEnt nowarnings noprefixes -E "" -E EllGwTw2 responsecurves jackknife "outputdirectory=G:\Projects\15.0-2014-'15 projects\1633 FMG Solomon Restricted Vegetation Assessment (EllGwTw2)\Map Data\MaxEntEllGwTw2Output" "samplesfile=R:\100 Projects\15.0-2014-'15\1633 FMG Solomon Restricted Vegetation Assessment (EllGwTw2)\03 Reference\EllGwTw2 DDdddd for MaxEnt.csv" "environmentallayers=G:\Projects\15.0-2014-'15 projects\1629 FMG Solomon Restricted Vegetation Supplementary Report\Map Data\ASCII\maxent.cache" randomseed nowriteclampgrid randomtestpoints=25 "biasfile=G:\Projects\15.0-2014-'15 projects\1633 FMG Solomon Restricted Vegetation Assessment (EllGwTw2)\Map Data\Bias\bias_all_dissolve_wgs84_clipptr3.asc" replicates=15 replicatetype=subsample nooutputgrids maximumiterations=5000 biastype=3 -t 1629250geology90mmask -t 1629landsystem90mmask

file:///G:/Projects/15.0-2014-'15%20projects/1633%20FMG%20Solomon%20Restrict... 10/03/2015