draft - university of toronto t-spacedraft 1 fragmentation dynamics in an abies religiosa forest of...

46
Draft Fragmentation dynamics in an Abies religiosa forest of central Mexico Journal: Canadian Journal of Forest Research Manuscript ID cjfr-2019-0235.R3 Manuscript Type: Article Date Submitted by the Author: 27-Dec-2019 Complete List of Authors: Montoya Pérez, Laura; Colegio de Postgraduados, Botánica Guzmán-Plazola, Remigio; Colegio de Postgraduados, Plant Health- Phytopathology Program; López-Mata, Lauro; Colegio de Postgraduados, Botanica Keyword: <i>Abies religiosa</i>, habitat fragmentation, supervised classification, SPOT, forest openings Is the invited manuscript for consideration in a Special Issue? : Not applicable (regular submission) https://mc06.manuscriptcentral.com/cjfr-pubs Canadian Journal of Forest Research

Upload: others

Post on 17-Feb-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

  • Draft

    Fragmentation dynamics in an Abies religiosa forest of central Mexico

    Journal: Canadian Journal of Forest Research

    Manuscript ID cjfr-2019-0235.R3

    Manuscript Type: Article

    Date Submitted by the Author: 27-Dec-2019

    Complete List of Authors: Montoya Pérez, Laura; Colegio de Postgraduados, BotánicaGuzmán-Plazola, Remigio; Colegio de Postgraduados, Plant Health-Phytopathology Program; López-Mata, Lauro; Colegio de Postgraduados, Botanica

    Keyword: Abies religiosa, habitat fragmentation, supervised classification, SPOT, forest openings

    Is the invited manuscript for consideration in a Special

    Issue? :Not applicable (regular submission)

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    1 Fragmentation dynamics in an Abies religiosa forest of central Mexico

    2

    3 Montoya, E. 1

    4 Guzmán-Plazola, R.A.2

    5 López-Mata, L. 1

    6

    7 1 Botany Program, Colegio de Postgraduados Campus Montecillo, 56230, Montecillo,

    8 Texcoco, State of Mexico, Mexico

    9 2 Plant Health-Phytopathology Program, Colegio de Postgraduados Campus Montecillo,

    10 56230, Montecillo, Texcoco, State of Mexico, Mexico

    11

    12 Correspondence:

    13 Remigio Anastacio Guzmán-Plazola, Colegio de Postgraduados Campus Montecillo,

    14 56230, Montecillo, Texcoco, State of Mexico, Mexico

    15 Email: [email protected]

    16

    17 Summary

    Page 1 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    18

    19 Abies forests of Mexico are relicts of the boreal forests that advanced southwards during

    20 glaciation periods. Mexico is a center of diversification of the Abies genus, since in its

    21 territory there are eight species, and six of them are endemic. The Abies religiosa forests

    22 nearby Mexico City are subject to a process of deterioration. We analyze the fragmentation

    23 dynamics of the A. religiosa forest in the northern region of the Sierra Nevada, México. Land

    24 cover change detection was done by means of high-resolution images acquired by the

    25 SPOT satellite in 2005, 2010, 2015 and 2018. Habitat fragmentation was observed, with a

    26 decrease in dense Abies masses size. The area covered by Abies decreased by 22.9%.

    27 The area occupied by forest openings increased 3% from 2005 to 2010 and then decreased

    28 by 1.8 and 1.6% in the following periods. Other Forest Cover increased in frequency and

    29 size, which warns of a change process towards that patch type, which increased 23.3%. The

    30 formation of increasingly smaller and isolated remnants of A. religiosa forest in the Sierra

    31 Nevada can lead to the loss of this vegetation relict and its replacement by other types of cover

    32 in the short term.

    33

    Page 2 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    34 Key words: Abies religiosa, forest openings, habitat fragmentation, supervised

    35 classification, SPOT.

    36

    37 Introduction

    38

    39 Abies forests of Mexico are extensive boreal forest relicts that advanced southwards during

    40 glaciation periods, when cold climates descended to tropical latitudes (Rzedowski and

    41 McVaugh, 1966). They are located in mountains with altitudes between 2,400 and 3,500 m

    42 a.s.l. (Rzedowski, 1998). Flores et al. (1971) estimated that they occupied 0.16% of the

    43 surface of Mexico, however, their area of occurrence has been declining since the end of the

    44 20th century.

    45

    46 In the Trans-Mexican Volcanic Belt (Ferrusquía-Villafranca, 1993) region, known as Sierra

    47 Nevada, there are areas covered by Abies religiosa (Kunth Schltdl. et Cham.) that are part of the

    48 Protected Natural Area (PNA) Izta-Popo-Zoquiapan National Park. It is located adjacent to the

    49 Valley of Mexico (one of the largest and most populated metropolises in the world), so the

    50 pressure on these forests is very noticeable. It is a landscape composed of forest patches of

    Page 3 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    51 various sizes, composition, structure and stages of development, shaped by anthropogenic and

    52 natural disturbances. The most important disturbances in these areas are deforestation, land

    53 openings for agriculture, grazing, fires and selective extraction of timber (Manzanilla, 1974;

    54 Velázquez, 1994; Challenger, 1998) and increased habitat vulnerability to stresses such as air

    55 pollution, pests and diseases (Fenn et al., 2002; Alvarado and Hernández, 2002). These

    56 disturbances exert a strong influence on the Abies forest landscape and may contribute to

    57 habitat losses, increased fragmentation and biodiversity loss by leading to a feedback loop of

    58 fragmentation dynamics that facilitates access to forest habitats and hence to further

    59 disturbances (Broadbent, 2008). Habitat fragmentation reduces biodiversity by 13-75% and

    60 impairs the key functions of the ecosystem by decreasing biomass and altering nutrient cycles

    61 (Ruete, 2016). It involves reduction in patch size, greater isolation of the patches and, often,

    62 changes in habitat quality (Berg, 1997). There is also growing evidence that they can lead to

    63 indirect chain effects (Komonen et al., 2000) like genetic changes associated with isolation

    64 and population size reduction (Geburek and Myking, 2018), formation of critical areas of

    65 geomorphic activity, driven by human activity, natural disturbances and the hydrological

    66 regime (Pawlik et al., 2019).

    67

    Page 4 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    68 Precise identification of forest openings is essential for forest ecosystems characterization

    69 (Chianucci et al., 2016) as they represent an important landscape fraction (Drössler and Lüpke,

    70 2005). Remote sensing techniques provide a unique way for obtaining estimates on extense areas,

    71 but their application is limited by the spectral, spatial and temporal resolution (Senécal et al., 2018;

    72 Zhirin et al., 2019) available from these systems, which is often not adequate to meet regional or

    73 local objectives. From high-resolution images it is possible to more precisely quantify changes in

    74 coverage and even the occurrence of small openings.

    75

    76 The analysis of the variations in forest coverage provides key information to achieve an adequate

    77 understanding of the fragmentation dynamics of the A. religiosa forests in Mexico. In this study,

    78 results of the analysis of the spatial and temporal dynamics of the different types of land cover in

    79 the Abies religiosa forest of the Izta-Popo-Zoquiapan National Park are reported. Our report is

    80 based on the supervised classification of high spatial resolution images acquired by the SPOT

    81 satellite in the years 2005, 2010, 2015 and 2018, in order to document the effect of anthropogenic

    82 pressure on the Abies forest ecosystem over a 13 year period.

    83

    84

    Page 5 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    85 Materials and methods

    86

    87 This work was carried out in the A. religiosa forest of the Sierra Nevada physiographic

    88 region, which is part of the Izta-Popo-Zoquiapan National Park PNA. This type of vegetation

    89 develops between 3,100 and 3,500 m a.s.l., mainly on slopes greater than 40% (Sánchez-

    90 González and López-Mata, 2003). This forest is located between the coordinates 19° 26'

    91 42.36" and 19° 28' 20.09" N and 98° 44' 59.74" and 98° 46' 42.78" W. It has an annual

    92 average temperature that varies between 10 and 12º C and an annual rainfall of 900 to 1000

    93 mm (Flores et al., 2011).

    94

    95 The presence of the A. religiosa forest was verified in one of the polygons reported by the

    96 National Institute of Statistics Geography and Informatics (INEGI) on the map "Uso de Suelo

    97 y Vegetación", scale 1: 250,000. This polygon belongs to the Ejidos Unidos de la Montaña

    98 section and was selected as the study area because it is the most compact and

    99 representative area of the forest under study (Figure 1). Additionally, from supervised

    100 classification results and from field observations, the limits of the base polygon were

    Page 6 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    101 modified to build a study polygon. This resulting polygon has a total area of 1,788.8 ha

    102 (Figure 1).

    103

    104 Figure 1. Polygon of the Abies religiosa forest in the Sierra Nevada, State of Mexico (INEGI,

    105 2017) and polygon modified by the authors from the supervised classification of a SPOT

    106 scene of 2018.

    107

    108 Spatial analysis

    Page 7 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    109 Images from the Satellite Pour l'Observation de la Terre (SPOT) acquired in the years 2005,

    110 2010, 2015 and 2018 were used; all of them correspond to the dry season (December-

    111 March) and had no cloud obstruction over the study area. They were taken with an angle

    112 outside the nadir of 25.7° (Table 1). SPOT 5 and 6 generates a panchromatic image with a

    113 spatial resolution of 2.5 m and four images with a spatial resolution of 10 m. SPOT 7

    114 generates a panchromatic image with a spatial resolution of 1.5 m and four images with a

    115 spatial resolution of 6 m. The SPOT images were acquired by the reception station in Mexico

    116 (ERMEX). The images had a 3A processing level. This level includes orthorectification and

    117 the radiometric correction of errors originated from the differences in sensitivity between

    118 sensor elements, as well as geometric corrections.

    119

    Table 1. Satellite images used for analyzing coverage dynamics in the Abies

    religiosa forest of the Sierra Nevada, State of Mexico, Mexico.

    Year Date Satellite Type* K/J

    2005 December 25th SPOT 5 P 589/311

    2005 December 25th SPOT 5 M 589/311

    2010 March 28th SPOT 5 P 589/311

    2010 March 28th SPOT 5 M 589/311

    2015 January 28th SPOT 6 P 589/311

    2015 January 28th SPOT 6 M 589/311

    2015 February 10th SPOT 7 P 589/312

    Page 8 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    2015 February 10th SPOT 7 M 589/312

    2018 March 4th SPOT 7 P 589/312

    2018 March 4th SPOT 7 M 589/312

    *P= Panchromatic; M= Multiespectral

    120

    121 Image Fusion

    122 The IHS algorithm was implemented in the System for Automated Geoscientific Analyzes

    123 (SAGA) version 2.0.6. It uses the high-resolution panchromatic band (e.g. December 25th

    124 image, Type P, Table 1) and the low-resolution images (e.g. December 25 image, Type M,

    125 Table 1) to create multispectral images of panchromatic resolution. This algorithm was

    126 based on the resampling of the image from which it allows to match raster data layers that

    127 have different cell sizes before overlaying them. The nearest neighbor method was used as

    128 a resampling method to determine what data value should be assigned to the new cell. This

    129 method assigns to the new cell the value of its closest neighbor cell in the original data layer

    130 (Johnston, 1998). The spatial information of the panchromatic image (1.5 or 2.5 m spatial

    131 resolution) was added to the visual information of the multispectral image (6 or 10 m spatial

    132 resolution). Spectral characteristics of the original data were preserved in the resulting high-

    133 resolution image. A detailed description of the IHS fusion technique is given in Zhang and

    Page 9 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    134 Hong (2005). After the fusion process, the images of the years 2005 and 2010, with a spatial

    135 resolution of 2.5 m, were resampled to a resolution of 1.5 m to make them spatially

    136 comparable with those of 2015 and 2018. The resampling was carried out using the

    137 RESAMPLE module of Idrisi v. Terrset (Eastman, 2016).

    138

    139 Geometric transformation

    140 To carry out the corregistration of the SPOT images, four ground control points (GCP) were

    141 identified in the area that covers the scene. These points were accurately located in all

    142 images (Jensen, 1996). A first-degree polynomial and the nearest neighbor resampling

    143 method were applied, using the RESAMPLE command of the IDRISI software v. Terrset.

    144 The software calculates the mathematical corrections necessary to perform the

    145 corregistration and calculates the output pixel position from the ground control points (GCP).

    146 These corrections are evaluated as mean square error (pixel fraction), which refers to the

    147 distance between the input location (source) of a GCP and the transformed location for the

    148 same GCP.

    149

    150 Supervised classification

    Page 10 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    151 Three land patch types were defined by field observations: 1) Abies religiosa, 2) Forest

    152 openings, and 3) Other Forest Cover (Figure 2). The coordinates of the field observation

    153 sites served as the basis for the digitization of the training sites corresponding to each patch

    154 type. By means of the QGIS software v. 2.18 (QGIS, 2018), a vector file of the training areas

    155 was generated to classify the fused multispectral images. Seventy percent of these polygons

    156 were randomly chosen as input of the maximum likelihood classification algorithm

    157 (Chuvieco, 1996). This procedure was carried out for each study year. The separability of

    158 patch types was analyzed (Franklin, 2018).

    Page 11 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft159160 Figure 2. General aspect of the areas used to generate the training sites. The location of the

    161 letters in the upper left image corresponds to the location of the photographs taken during

    162 field trips in the Abies religiosa forest of the Sierra Nevada, Mexico. The upper left image

    163 has coordinates UTM 14N, Datum WGS84. A = forest openings formed by Abies religiosa

    164 cuttings; B = forest openings induced from the creation of a path; C = closed canopy in the

    165 Abies forest; D = forest openings corresponding to Pinus reforestation; E and F = area with

    166 Other Forest Cover, mainly Arbutus, Cupressus, Pinus and Quercus.

    Page 12 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    167

    168 Accuracy of classification results

    169 The accuracy of the classification was statistically evaluated from a confusion matrix

    170 generated from the input data and the supervised classification results. This analysis was

    171 carried out using SAGA software v. 2.18. The remaining 30% of the polygons from field

    172 observations were used as ground truth. Additionally, the Kappa index (Landis and Koch,

    173 1977), the global accuracy and the average reliability (Stehman, 1997) were calculated for

    174 each of the patch types used in the classification.

    175

    176 Temporal analysis

    177 A series of images taken by the SPOT satellite in 2005, 2010, 2015 and 2018, corresponding

    178 to the same study area were used. Each of these images was used in a classification

    179 process based on the aforementioned thematic classes.

    180

    181 Change detection

    182 By means of the CROSSTAB command of IDRISI v. Terrset, pixel based changes in the

    183 cover of the Abies forest occurring in the periods 2005-2010, 2010-2015 and 2015-2018

    Page 13 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    184 were estimated. Based on these results, a flow chart was made which indicates direction

    185 and magnitude of each change through the three periods considered.

    186

    187 Results

    188

    189 Image fusion

    190 Four multispectral images with 1.5 m spatial resolution corresponding to the years 2005,

    191 2010, 2015 and 2018 were generated. An example of the sharpness achieved by this

    192 process is shown in the upper left corner of Figure 2. In this case a false color image is

    193 shown, product of the mixture of the blue, green and red bands derived from the fusion

    194 process. The results of the corregistration of the previous images with the 2018 scene had

    195 a square mean error of 0.004, 0.009 and 0.007 pixel fraction for the years 2005, 2010 and

    196 2015, respectively.

    197

    198 Supervised classification

    199 In general, the Forest Openings class has the highest reflectance values, while the Abies

    200 and Other Forest Cover classes had more similar patterns to each other, but their standard

    Page 14 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    201 deviations did not overlap in most bands (Figure A in supplementary material). In the SPOT

    202 scene corresponding to the year 2005, it is observed that the mean reflectance values and

    203 standard deviations of the patch types for the blue (B), red (R) and infrared (IR) bands are

    204 separable, which allowed a percentage of correct classification of 98% (Figure A and Table

    205 A in supplementary material). The reflectance patterns in the green (G) and IR bands of the

    206 2010 scene show no overlap in their mean reflectance values and standard deviations for

    207 each patch type; this allowed a percentage of correct classification of 95% (Figure A and

    208 Table A in supplementary material). The 2015 image presents average reflectance values

    209 and standard deviations of the B and IR bands that are different for each patch type, which

    210 allowed 80% correct classification in this period (Table A in supplementary material). In

    211 2018, mean values and standard deviations of reflectance do not overlap, except for the red

    212 band; this allowed percentages of correct global classification of 89% (Figure A and Table

    213 A in supplementary material). Four images classified into the three patch types were

    214 obtained (Figure 3). Kappa values for the different patch types are indicated in Table A of

    215 supplementary material.

    Page 15 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    216

    Page 16 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    217 Figure 3. Results of the supervised classification of the SPOT images of the years 2005

    218 (December), 2010 (March), 2015 (February) and 2018 (March) in the area of the A. religiosa

    219 forest of the Sierra Nevada, Mexico.

    220

    221 Temporal dynamics

    222 The increase in the number of A. religiosa fragments followed a linear trend between 2005

    223 (1,249 patches) (Figures 4a and 5) and 2015 (19,552 patches) (Figure 4c), but this trend did

    224 not continue towards 2018, when the graph acquired an inverted “J” shape, in which

    225 increasingly smaller patches dominated, with very few large patches. At the end of the period

    226 under analysis, the densest mass of this forest was reduced from 87.1 to 58.3% of total

    227 surface (Figures 4d and 5) (21,653 patches) (Figure 5). The increase in the number of Other

    228 Forest Cover patches followed a quasi-exponential trend during the whole period under

    229 study (Figure 5), where the number of fragments increased from 15,816 in 2005 to 36,116

    230 in 2018. The formation of forest clearings had a similar trend to that of Other Forest Cover

    231 between 2005 and 2015, but its trend declined towards 2018, at a greater extent than the

    232 number of fragments of A. religiosa.

    Page 17 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    233234 Figure 4. Patch size structure of the different cover classes in the Abies religiosa forest of

    235 the Sierra Nevada, Mexico, during the period from December 2005 to March 2018. Green

    236 bars = Abies religiosa. Grey bars = Forest openings. Red bars = Other Forest Cover. Vertical

    237 lines below the bars indicate the existence of data for patches with the Ln(area) indicated.

    Page 18 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    238

    239 Figure 5. Patch frequency of different types of land coverage from 2005 to 2018 in the Abies

    240 religiosa forest of the Sierra Nevada, Mexico.

    241

    242 The analysis of the transitions between the different patch types over time indicates that in

    243 2005 the A. religiosa class covered 1,564.4 ha (87.5% 2.1) of the study area. In 2010 a

    244 total of 1,281.6 ha (71.7% 2.4) of this patch type was detected. In 2015, a decrease to

    245 1,151.6 ha (64.4% 10.7) was observed. In 2018, the Abies forest represented a coverage

    Page 19 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    246 of 1,154.3 ha (64.5% 6.5). The sources of changes for all patch types are reported in

    247 Figures 6 and 7.

    248249 Figure 6. Dynamics of land coverage changes from 2005 to 2018 in the Abies religiosa forest

    250 in the Sierra Nevada, Mexico. * Data in each cell indicate estimated surface in hectares.

    251

    252 The area with forest openings (including some crop fields) covered 101.5 ha (5.7% 0.1) of

    253 the study area in 2005. In 2010 an increase to 155 ha (8.7% 0.1) was detected. In 2015,

    254 the openings area represented 123.4 ha (6.9% 0.1 of the total area). In 2018, 95.5 ha

    255 (5.3% 0.1) of surface covered by forest openings was identified (Figures 6 and 7).

    Page 20 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    256257 Figure 7. Vegetation coverage changes from 2005 to 2018 in the Sierra Nevada, Mexico.

    258 the Sierra Nevada, Mexico.

    259

    260 In 2005, the area covered by species of other genera (among them: Arbutus, Cupressus,

    261 Pinus, Quercus) in the A. religiosa forest represented 122.9 ha (6.9% 0.4). In 2010 the

    262 area increased to 352.1 ha (19.7% 5.5), and in 2015 it covered 513.7 ha (28.7% 15.8 of

    263 the study area). In 2018, the area with this patch type increased to 538.9 ha (30.1% 11.1

    264 of the total area) (Figures 6 and 7).

    265

    Page 21 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    266 From Figure 6 it can be drawn that the decrease in A. religiosa forest cover between 2005

    267 and 2018 was not compensated by other forest types that became A. religiosa. Also, there

    268 was a sharp drop between 2005 and 2010 and an apparent stabilization in the following

    269 periods.

    270

    271 Discussion

    272

    273 Our results suggest that a process of fragmentation of the Abies forest is increasing in the

    274 Izta-Popo-Zoquiapan National Park, which is causing coverage loss as it is replaced by other

    275 genera (Arbutus, Cupressus, Pinus and Quercus, among others) (Figures 6 and 7).

    276 However, area estimations in our work should be considered as approximate since the errors

    277 in the classification (Stehman, 1997) in some cases were significant with respect to the

    278 calculated percentages of change, particularly in the changes from Other Forest Cover to

    279 A. religiosa (Table A in supplementary material). These errors could be caused by the

    280 relative similarity in the spectral reflectance of these coverage classes. Nevertheless, a clear

    281 trend towards habitat fragmentation was detected (Figures 5 and 6). These quantifications

    282 were possible because the data used in this study come from high resolution remote sensing

    Page 22 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    283 images, which allowed the identification of forest openings with a minimum area of 2.25 m2

    284 (Figures 3 and 4B), unlike other detection methodologies based on satellite images that

    285 often do not detect small clearings (Lobo and Dalling, 2014). Since these smaller openings

    286 contribute to a much larger landscape fraction, the assessment of how disturbance

    287 influences the lower strata (shrubs and herbaceous) requires the inclusion of a wide range

    288 of forest opening sizes, including the smallest ones (~10 m2) and for this, high spatial

    289 resolution images are required, which avoid underestimation (Chianucci et al., 2016).

    290

    291 In the Abies forests, the communities are usually so closed that treetops create limited

    292 lighting conditions and inhibit the growth of the lower shrubs or herbaceous plants (Diaci,

    293 2002). When disturbance events occur in these forests, whose species are very shade

    294 tolerant, they open up canopy gaps that generally allow the presence and proliferation of

    295 less tolerant species (Barden, 1979) (e.g. Pinus). This suggests that several disturbance

    296 events (natural and induced) have indeed occurred in the A. religiosa forest of the Sierra

    297 Nevada, since, among the species identified in the areas not covered by Abies during our

    298 field trips, there are species that belong to established secondary vegetation after partially

    299 or totally removing the primary vegetation coverage.

    Page 23 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    300

    301 It has been suggested that a forest opening created by logging allows to maintain the

    302 ecological integrity since the forest dynamics is activated by the fall of the trees and the

    303 improved use imitates the natural regime of disturbances (Lertzman et al., 1996) in a faster

    304 way. However, our results indicate that the A. religiosa forest has decreased by 22%

    305 (Figures 6 and 7); also, the dynamics of natural regeneration is being affected by other

    306 factors, among which reforestations with different species than A. religiosa stand out, as well

    307 as a slow process of natural regeneration in the fragments, which can take several decades

    308 (Esseen, 1994).

    309

    310 Fragmentation and edge effects could increase the richness of tree species in forests

    311 dominated by Abies. This offers new possibilities for other species to be established

    312 (Lehvävirta et al., 2014). Pine plantation has increased in areas dominated by A. religiosa

    313 (Figure 2D). In our field trips we observed sites with felled Abies individuals, resins or rotten

    314 trees, and trees with exposed roots, as well as several stumps and dead individuals standing

    315 (Figure 2A) which makes us think that the replacement of the Abies trees through the

    316 introduction of pine is generating a different forest cover that modifies forest conditions and

    Page 24 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    317 allows the development of other types of vegetation. This demands a redesign of

    318 reforestation campaigns and other relevant activities, to implement conservation strategies

    319 for this vegetation relict. Our data support these ideas since the shapes of the size structure

    320 observed for forest openings in the first years of records (2005-2010) corresponds to a

    321 closed canopy with uniform or continuous type coverage. However, as of 2015, this structure

    322 changed drastically due to the disturbances of an exploited forest, which generated a

    323 coverage characterized by presenting larger proportions of small and intermediate openings

    324 and smaller areas of large patches of vegetation, which resulted in increased canopy

    325 discontinuity.

    326

    327 Abies religiosa is considered a priority species for conservation according to the

    328 management plan of the PNA Izta-Popo-Zoquiapan National Park. The most important

    329 criterion given by this category is the feasibility of recovering and managing it, as well as the

    330 additional effects that direct conservation would cause, such as the conservation of other

    331 species or the habitat itself. The management plan of this PNA suggests that studies be

    332 carried out to define measures aimed at their conservation; among them, distribution maps

    333 preparation and population monitoring, as well as studies of the ecological processes

    Page 25 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    334 involved in the dynamics of this ecosystem, and the environmental conditions and social

    335 processes involved, since this information is required as support in decision making.

    336

    337 There is no doubt that the Abies forest is in a state that could be defined as extremely fragile,

    338 given the particular conditions that it requires for its development, as well as the intensive

    339 use that the farmers exert over the individuals of this plant community, either through

    340 extraction of wood (Figure 2a) or by extending the agricultural frontier (Figure 1). The

    341 populations in the isolated fragments are at greater risk of disappearing since they are

    342 getting smaller and natural disturbances can eliminate them. The high vulnerability of A.

    343 religiosa forests to climate change must also be recognized, given their high degree of

    344 fragmentation and critical altitudinal limit (Pineda-López et al., 2013). On the other hand, it

    345 is of vital importance the conservation of the habitat of these trees through a correct zoning

    346 of the PNA since an extensive network of trails was built and maintained which it is used

    347 with no restrictions and promote tree extraction even in well-conserved areas of this forest.

    348 These roads can have direct ecological effects on the ecosystem, such as habitat alteration,

    349 greater edge effect, changes in water balance, greater fragmentation, dispersion of exotic

    Page 26 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    350 species and changes in species behavior and greater vulnerability to external factors

    351 (Marcantonio et al., 2013; Vanbianchi et al., 2017; Pawlik, 2019).

    352

    Page 27 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    References

    Alvarado-Rosales, D. and Hernández-Tejeda, T. 2002. Decline of sacred fir in the Desierto

    de los Leones National Park. In: Fenn, M., L. I. Bauer and T. Hernández-Tejeda (eds.).

    Urban Air Pollution and Forests: Resources at Risk in the Atmospheric Air Basin of Mexico.

    Springer-Verlag, New York. Pp. 243-260.

    Barden, L. S. 1979. Tree replacement in small canopy gaps of a Tsuga canadensis forest in

    the Southern Appalachians, Tennessee. Oecologia. 44(1):141-142.

    doi:10.1007/BF00346412.

    Berg, Å. 1997. Diversity and abundance of birds in relation to forest fragmentation, habitat

    quality and heterogeneity. Bird Study. 44(3): 355–366. doi:10.1080/00063659709461071

    Broadbent, E., Asner, G., Keller, M., Macedo, J.N. 2008. Forest fragmentation from

    deforestation and selective logging in the Brazilian Amazon. Biological

    Conservation 141(7):1745-1757. doi: 10.1016/j.biocon.2008.04.024

    Challenger, A. 1998. Utilización y conservación de los ecosistemas terrestres de

    México. Pasado, presente y futuro. Comisión Nacional para el Conocimiento de la

    Page 28 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

    https://www.researchgate.net/deref/http%3A%2F%2Fdx.doi.org%2F10.1016%2Fj.biocon.2008.04.024?_sg%5B0%5D=U9GVBv0k9AamZeyJbCaWOsWD3WtZ_QfdlgNXLPvc1BQAcWI77e6wW43VmT-GPqyG19ZkMIlXcLePJ0cJq05Edso0vw.dzKOjLqBJDtGg1vQFqaw5l0K5GOS6xOhkQHXgYe14W6tAganXX8_sK6BLxIMDjPXGijKS_5llPeHSlGZ_kLm2w&_sg%5B1%5D=Qz5jk6VY5kXh8kpG7OZZkynQcGuCDsEgXKpBKbkQKRLHLYqDuwQwmhAT_VvYeU1JxswLd4V0GKbr.M1oqOWfs7xYVPgTlg35GycERR2dIWOmsOtzIsevaV9PKSsibgL26qp8jj8qcjdainZC1423h7SRNFkISmKxgAg

  • Draft

    Biodiversidad, Instituto de Biología, Universidad Nacional Autónoma de México, and

    Agrupación Sierra Madre, S.C. México. Pp 847.

    Chianucci, F., Disperati, L., Guzzi, D., Bianchini, D., Nardino, V., Lastri, C., Rindinella,

    A. and Corona, P. 2016. Estimation of canopy attributes in beech forests using true

    colour digital images from a small fixed-wing UAV. International Journal of Applied

    Earth Observation and Geoinformation. 47: 60-68. doi:10.1016/j.jag.2015.12.005.

    Chuvieco, E. 1996. Fundamentos de Teledetección espacial, 3ra edición revisada.

    Ediciones RIALP, S.A. Madrid. Pp. 567.

    Diaci, J. 2002. Regeneration dynamics in a Norway spruce plantation on a silver fir-

    beech forest site in the Slovenian Alps. Forest Ecology and Management. 161(1-3):

    27-38. doi:10.1016/S0378-1127(01)00492-3

    Drössler, L. and Lüpke, B. V. 2005. Canopy gaps in two virgin beech forest reserves

    in Slovakia. Journal of Forest Science. 51(10):446-457. doi:10.17221/4578-JFS

    Eastman, J.R. 2016. TerrSet. Geospatial Monitoring and Modeling System. Clark

    University. www.clarklabs.org.

    Page 29 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Esseen, P. A. 1994. Tree mortality patterns after experimental fragmentation of an old-

    growth conifer forest. Biological Conservation. 68(1):19–28. doi:10.1016/0006-

    3207(94)90542-8

    Fenn, M., de Bauer, M.L.I. and Hernández, T. (eds.). 2002. Urban Air Pollution and Forests:

    Resources at Risk in the Atmospheric Air Basin of Mexico. Springer-Verlag, New York.

    Ferrusquía-Villafranca, I. 1993. Geology of Mexico: a synopsis. In: Ramamoorthy,

    T.P., Bye, R., Lot, A. and Fa, J. (eds.). 1993. Biological Diversity of Mexico. Origins

    and Distribution. Oxford University Press, New York. Pp.3-107.

    Flores, M., Jiménez, J., Madrigal, S., Moncayo, R., and Takaki, T. 1971. Memoria del

    mapa de tipos de vegetación de la República Mexicana. Secretaría de Recursos

    Hidráulicos. México D.F. Pp. 59.

    Flores, P., López, M., Ángeles, G., de la Isla, M., Clava, G. 2011. Modelos para estimación

    y distribución de biomasa de Abies religiosa (Kunth) Schltdl. et Cham., en proceso de

    declinación. Revista Mexicana de Ciencias Forestales. 2(8): 9-20.

    Page 30 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Franklin, S. E. 2018. Pixel- and object-based multispectral classification of forest tree

    species from small unmanned aerial vehicles. Journal of Unmanned Vehicle Systems. 6(4):

    195-211. doi:10.1139/juvs-2017-0022

    Geburek, T., and Myking, T. 2018. Evolutionary consequences of historic anthropogenic

    impacts on forest trees in Europe. Forest Ecology and Management. 422: 23–32.

    doi:10.1016/j.foreco.2018.03.055

    Jensen, J. R. 1996. Introductory Digital Image Processing: A remote sensing perspective,

    2nd Edition. NJ: Prentice-Hall. Pp. 17-24.

    Johnston, C. A. 1998. Geographic Information Systems in Ecology. Methods in Ecology.

    First Edition. Blackwell Science Ltd. Malden, USA.

    Komonen, A., Penttilä, R., Lindgren, M., and Hanski, I. 2000. Forest fragmentation truncates

    a food chain based on an old-growth forest bracket fungus. Oikos. 90(1): 119–126.

    doi:10.1034/j.1600-0706.2000.900112.x

    Landis, J and Koch, G. G. 1977. The measurements of observer agreement for categorical

    data. Biometrics. 33(1): 159-174. doi:10.2307/2529310

    Page 31 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Lehvävirta, S., Vilisics, F., Hamberg, L., Malmivaara-Lämsä, M., and Kotze, D. J. 2014.

    Fragmentation and recreational use affect tree regeneration in urban forests. Urban Forestry

    and Urban Greening. 13(4): 869-877. doi:10.1016/j.ufug.2014.10.003

    Lertzman, K. P., Sutherland, G. D., Inselberg, A. and Saunders, S. C. 1996. Canopy Gaps

    and the Landscape Mosaic in a Coastal Temperate Rain Forest. Ecology. 77(4): 1254-1270.

    doi:10.2307/2265594

    Lobo, E. and Dalling, J. 2014. Spatial scale and sampling resolution affect measures of gap

    disturbance in a lowland tropical forest: implications for understanding forest regeneration

    and carbon storage. Proceedings of the Royal Society. Biological Sciences. 281(1778):

    20133218. doi:10.1098/rspb.2013.3218

    Manzanilla, H. 1974. Investigaciones epidométricas y silvícolas en bosques mexicanos de

    Abies religiosa, Dirección General de Información y Relaciones Públicas de la Secretaría

    de Agricultura y Ganadería, México.

    Marcantonio, M., Rocchini, D., Geri, F., Bacaro, G., and Amici, V. 2013. Biodiversity, roads,

    and landscape fragmentation: Two Mediterranean cases. Applied Geography. 42: 63-72.

    doi:10.1016/j.apgeog.2013.05.001

    Page 32 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Pawlik, Ł., Šamonil, P., Malik, I., Kroh, P., Ślęzak, A., and Danĕk, P. 2019. Geomorphic

    edge effects in response to abiotic and anthropogenic disturbances in forest ecosystems of

    the Gorce Mountains, Western Carpathians. Catena. 177: 134-148.

    doi:10.1016/j.catena.2019.02.013

    Pineda-Lopez, M. R., Sanchez-Velasquez, L. R., Vazquez-Dominguez, G., and Rojo-

    Alboreca, A. 2013. The effects of land use change on carbon content in the aerial biomass

    of an Abies religiosa (Kunth Schltdl. et Cham.) forest in central Veracruz, Mexico. Forest

    Systems. 22(1): 82-93. doi: 10.5424/fs/2013221-02756

    QGIS Development Team. 2018. QGIS Geographic Information System. Open Source

    Geospatial Foundation. URL http://qgis.osgeo.org

    Ruete, A., Snäll, T., and Jönsson, M. 2016. Dynamic anthropogenic edge effects on the

    distribution and diversity of fungi in fragmented old-growth forests. Ecological Applications.

    26(5): 1475-1485. doi: 10.1890/15-1271

    Rzedowski, J. and McVaugh, R. 1966. La vegetación de Nueva Galicia. University of

    Michigan Press, Ann Arbor.

    Rzedowski, J. 1998. Vegetación de México. Editorial Limusa, México. Pp 432.

    Page 33 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Sánchez-González. A. y López-Mata, L. 2003. Clasificación y ordenación de la vegetación

    del norte de la Sierra Nevada, a lo largo de un gradiente altitudinal. Anales del Instituto de

    Biología. (Serie Botánica). Universidad Nacional Autónoma de México. 74: 43-71.

    Senécal, J. F., Doyon, F. and Messier, C. 2018. Tree Death Not Resulting in Gap Creation:

    An Investigation of Canopy Dynamics of Northern Temperate Deciduous Forests. Remote

    Sensing. 10(1): 121. doi:10.3390/rs10010121

    Stehman, S. V. 1997. Selecting and interpreting measures of thematic classification

    accuracy. Remote Sensing, Environment. 62: 77-89. doi:10.1016/S0034-4257(97)00083-7

    Vanbianchi, C., Gaines, W.L., Murphy, M.A., Pither, J., and Hodges, K.E. 2017. Habitat

    selection by Canada lynx making do in heavily fragmented landscapes. Biodiversity and

    Conservation. 26(14): 3343-3361. doi:10.1007/s10531-017-1409-6

    Velázquez, A. 1994. Multivariate analysis of the vegetation of the volcanoes Tlaloc and

    Pelado, Mexico. Journal of Vegetation Science. 5: 263-270.

    Page 34 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Zhang, Y., and Hong, G. 2005. An IHS and wavelet integrated approach to improve pan-

    sharpening visual quality of natural colour IKONOS and QuickBird images. Information

    Fusion. 6(3): 225-234. doi:10.1016/j.inffus.2004.06.009

    Zhirin, V. M., Knyazeva, S. V. and Eidlina, S. P. 2019. Estimation of Linkages between

    Biometric Indexes of Forests and Pattern of Canopy Spaces on Super-High-Resolution

    Satellite Images. Contemporary Problems of Ecology. 11(7): 743-753.

    doi:10.1134/S1995425518070107

    Captions

    Figure 1. Polygon of the Abies religiosa forest in the Sierra Nevada, State of Mexico (INEGI,

    2017) and polygon modified by the authors from the image of supervised classification of a

    SPOT scene of 2018, done with SAGA software.

    Figure 2. General aspect of the areas used to generate the training sites. The location of the

    letters in the upper left image corresponds to the location of the photographs taken by the

    authors during field trips in the Abies religiosa forest of the Sierra Nevada, Mexico. The

    Page 35 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    upper left image has coordinates UTM 14N, Datum WGS84; it was generated by a

    composition of bands 2,3 and 4 SPOT images of the zone. Bands were initially

    pansharpened with SAGA software. A = forest openings formed by Abies religiosa cuttings;

    B = forest openings induced from the creation of a path; C = closed canopy in the Abies

    forest; D = forest openings corresponding to Pinus reforestation; E and F = area with Other

    Forest Cover, mainly Arbutus, Cupressus, Pinus and Quercus.

    Figure 3. Results of the supervised classification of SPOT images of the years 2005

    (December), 2010 (March), 2015 (February) and 2018 (March) in the Abies religiosa forest

    of the Sierra Nevada, Mexico. Image classifications were done with SAGA software.

    Figure 4. Patch size structure of different cover classes in the Abies religiosa forest of the

    Sierra Nevada, Mexico, during the period from December 2005 to March 2018. Green bars

    = Abies religiosa. Grey bars = Forest openings. Red bars = Other Forest Cover. Vertical

    lines below the bars indicate the occurrence of patches with the Ln(area) indicated. Plots

    were done with the ggplot2 library of R software.

    Figure 5. Patch frequency of different types of land coverage from 2005 to 2018 in the Abies

    religiosa forest of the Sierra Nevada, Mexico.

    Page 36 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Figure 6. Dynamics of land coverage changes from 2005 to 2018 in the Abies religiosa forest

    of the Sierra Nevada, Mexico. * Data in each cell indicate estimated surface in hectares.

    Data come from classified SPOT images reported in Figure 3. Calculations were done by

    means of the CROSSTAB command of Terrset version of IDRISI software.

    Figure 7. Vegetation coverage changes from 2005 to 2018 in the Abies religiosa forest of

    the Sierra Nevada, Mexico. Data come from classified SPOT images reported in Figure 3.

    Image processing was done by means of the CROSSTAB command of Terrset version of

    IDRISI software.

    Page 37 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Figure 1. Polygon of the Abies religiosa forest in the Sierra Nevada, State of Mexico (INEGI, 2017) and polygon modified by the authors from the image of supervised classification of a SPOT scene of 2018, done

    with SAGA software.

    Page 38 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Figure 2. General aspect of the areas used to generate the training sites. The location of the letters in the upper left image corresponds to the location of the photographs taken by the authors during field trips in the Abies religiosa forest of the Sierra Nevada, Mexico. The upper left image has coordinates UTM 14N, Datum

    WGS84; it was generated by a composition of bands 2,3 and 4 SPOT images of the zone. Bands were initially pansharpened with SAGA software. A = forest openings formed by Abies religiosa cuttings; B = forest openings induced from the creation of a path; C = closed canopy in the Abies forest; D = forest

    openings corresponding to Pinus reforestation; E and F = area with Other Forest Cover, mainly Arbutus, Cupressus, Pinus and Quercus.

    Page 39 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Figure 3. Results of the supervised classification of SPOT images of the years 2005 (December), 2010 (March), 2015 (February) and 2018 (March) in the Abies religiosa forest of the Sierra Nevada, Mexico.

    Image classifications were done with SAGA software.

    Page 40 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Figure 4. Patch size structure of different cover classes in the Abies religiosa forest of the Sierra Nevada, Mexico, during the period from December 2005 to March 2018. Green bars = Abies religiosa. Grey bars = Forest openings. Red bars = Other Forest Cover. Vertical lines below the bars indicate the occurrence of

    patches with the Ln(area) indicated. Plots were done with the ggplot2 library of R software.

    177x160mm (300 x 300 DPI)

    Page 41 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Figure 5. Patch frequency of different types of land coverage from 2005 to 2018 in the Abies religiosa forest of the Sierra Nevada, Mexico.

    Page 42 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Figure 6. Dynamics of land coverage changes from 2005 to 2018 in the Abies religiosa forest of the Sierra Nevada, Mexico. * Data in each cell indicate estimated surface in hectares. Data come from classified SPOT

    images reported in Figure 3. Calculations were done by means of the CROSSTAB command of Terrset version of IDRISI software.

    Page 43 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Figure 7. Vegetation coverage changes from 2005 to 2018 in the Abies religiosa forest of the Sierra Nevada, Mexico. Data come from classified SPOT images reported in Figure 3. Image processing was done by means

    of the CROSSTAB command of Terrset version of IDRISI software.

    Page 44 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research

  • Draft

    Table 1. Satellite images used for analyzing coverage dynamics in the Abies religiosa forest of the Sierra Nevada, State of Mexico, Mexico.Year Date Satellite Type* K/J2005 December 25th SPOT 5 P 589/3112005 December 25th SPOT 5 M 589/3112010 March 28th SPOT 5 P 589/3112010 March 28th SPOT 5 M 589/3112015 January 28th SPOT 6 P 589/3112015 January 28th SPOT 6 M 589/3112015 February 10th SPOT 7 P 589/3122015 February 10th SPOT 7 M 589/3122018 March 4th SPOT 7 P 589/3122018 March 4th SPOT 7 M 589/312*P= Panchromatic; M= Multiespectral

    Page 45 of 45

    https://mc06.manuscriptcentral.com/cjfr-pubs

    Canadian Journal of Forest Research