https://orcid.org/0000-0003-1827-8660
Abstract
In the southwestern Mediterranean Basin, cork oaks (Quercus suber L.) are periodically harvested for their cork. This natural product is valued by its homogeneity which heightens the importance of characterizing cork tissue discontinuities, or cork pores. Cork porosity profile in natural cork planks has been reported to be affected by forest management practices but, so far, has been scarcely addressed. We characterize the cork porosity profile in two contrasting cork oak woodland; at a mountain forest, in Western Algeria (absence of forest management) and at a peneplain “montado,” in southern Portugal (intensively managed toward the optimization of cork production). Image analysis techniques were applied on transverse sections of more than 40 cork samples from both woodland, and a stepwise discriminant analysis was used to discriminate between the cork pore features data-sets. Cork porosity profiles were similar between regions but; in the cork samples from Algeria, cork pores were having higher values for linear dimensions of pores (length and perimeter) and contrasting shape values (roundness) which depreciate cork quality, when compared to the cork samples from Portugal. However, improved woodland management strategies at Algeria should ensure adequate cork homogeneity and suitability for more valuable cork products.
Public Interest Statement
Cork is the outer bark of cork oak trees; one natural product generating US $2 billion annually, and, currently, the sixth most valuable global non-timber forest product. Cork oaks are strictly distributed in the western Mediterranean Basin, between the southwestern Europe and northern Africa. An increasingly important challenge for the later region is to apply adequate forest management practices toward the optimization of high-quality cork production. In their research, authors assessed cork quality of two contrasting cork-producing regions: Portugal, the world leader in cork production; and Algeria, one potential cork-producing region. Results showed strong similarities between cork quality profiles, but higher cork quality heterogeneity at Algeria. Clearly, at this region, adequate forest management practices such as thinning (for selecting the best cork-producing trees) or pruning (for optimizing trees’ cork-harvesting surface), and appropriate cork-harvesting cycles should improve the cork yield and quality, at medium term.
Competing Interests
The authors declare no competing interest.
1. Introduction
Cork oak (Quercus suber L.) is a strictly Mediterranean species distributed in the western Mediterranean Basin, between the southwestern Europe and northern Africa, a climatic, ecological, and socioeconomical sensitive region (Bugalho, Caldeira, Pereira, Aronson, & Pausas, Citation2011; Costa, Madeira, Santos, & Plieninger, Citation2014; Giorgi, Citation2006). Cork oak woodland are considered keystone ecosystems (Vicente & Alés, Citation2006), while enhance other important provisioning/regulating ecosystem services (Plieninger, van der Horst, Schleyer, & Bieling, Citation2014). Moreover, cork oak woodland are important biodiversity hotspots where cork oak is harvested throughout their lifetime for their bark (Oliveira & Costa, Citation2012; Ticktin, Citation2004), the cork, one valuable global non-timber forest product (FAO, Citation2013).
In Portugal and Algeria, the cork oak woodland ecosystems are under high human pressure; in Portugal, these woodland are iconic examples of domesticated nature featuring savannah-like ecosystems carefully managed for a sustainable cork yield (Costa & Oliveira, Citation2015; Costa et al., Citation2014). In Algeria, cork oak woodland are disturbed forest ecosystems with strong, political, economical, and technological underlying factors, leading to the depreciation of cork value in relation to other tree’s products such as acorn or (fire)wood (Messaoudène & Merouan, Citation2009). Despite the differences found between cork oak woodland conditions and management in Portugal and Algeria, a common increasingly important challenge for both regions is maintaining the trees producing high-quality cork. Cork quality depends on the homogeneity of the cork tissue, and on the presence of cork tissue discontinuities, pores or lenticular channels crossing radially the transverse sections of cork planks (Costa & Pereira, Citation2010).
Given the importance of the quality grading of cork planks in the industrial processing, namely for the production of natural disks and stoppers (Costa & Pereira, Citation2006; Pereira, Citation2007), the cork quality has long being determined in raw cork material by visual inspection (Costa & Pereira, Citation2010). In the surface of the transverse sections of cork planks, the high porosity variability in-between the transverse section of the cork samples have been reported in technical studies in industries, and a strong radial variability from the inner part (with lower porosity) to the outer part (with higher porosity) has been generally accepted. However, it is urgently needed a more comprehensive understanding on the extent of lenticular channels, i.e. on the cork porosity variability profile in the cork plank transverse sections, so far, scarcely addressed.
In this study, the goal is to reinforce the existent knowledge on the cork quality profile of cork planks by a large-scale assessment of cork yield in the Mediterranean area, including those of areas of northern Africa. In fact, given the ever-increasing demand for cork raw material, and with the Portuguese cork industry leading the world market with about a 65% share (APCOR, Citation2015), all the cork oak woodland areas should constitute potential cork-producing areas, contributing to the high cork-value chain.
We propose to use image vision inspection systems currently used in the industry for stoppers and disks quality classification, to identify and quantify the lenticular channels, in transverse sections of cork planks. The goal is to fill the existent knowledge gaps on the porosity profile of cork by generating comparable data on cork porosity in the cork plank’s transverse sections of the two Mediterranean Basin cork-producing regions, Portugal and Algeria. Similar porosity (i.e. cork tissue discontinuities attributed to lenticular channels) features will be addressed and we hypothesized that in both regions the cork porosity variability profile will be similar, with larger and more porosity in outer part of the transverse section of cork planks. Moreover, it is hypothesized that appropriate forest management practices should improve cork quality, given the presence of porosity features, at each study area.
2. Material and methods
2.1. Study areas
Cork sampling was made during the cork-harvesting season, in the summer of 2010, at two study areas in the western Mediterranena Basin: at the cork oak woodland of Hafir-Zarieffet (ZA) (34°50′ N, 1°23′ W), at Tlemcen Mountain, in northwestern Algeria, close to the Mediterranean Sea and; at the cork oak woodland of Benavente (CL) (38°04′ N, 8°40′ W), in the Tagus Basin peneplain, in southwestern Portugal, close to the Atlantic Ocean (Supplementary material, Figure S1).
The climate in both study areas is of Mediterranean type, smoothed by the influence of the Mediterranean Sea (at ZA) and the Atlantic Ocean (at CL). Mean annual temperature and annual rainfall are 13.5°C and 654 mm, and 15.3°C and 577 mm at ZA and CL, respectively (Figure ). The mean temperature of the coldest month (January) is 3.5 and 5.0°C and for the hottest month (July) is 30.7 and 28.7°C, for ZA and CL, respectively. The highest temperatures are in summer, when precipitation is the lowest, and a dry period (p < 2T) generally occurs extending from May to September (Figure ).
Figure 1. Climatic diagrams for the study areas, Hafir-Zarieffet (in black) and Benavente (in gray).
The study areas are within the potential vegetation area for cork oak, corresponding to the meso-Mediterranean (ZA) and thermo-Mediterranean (CL) thermotypes and to the sub-humid ombrotype (Capelo et al., Citation2007; Dehane, Citation2012).
At ZA, the elevation ranges from 1,000 to 1,220 m a.s.l. and the landscape is steeply undulated (with slopes ranging between 3 and 12%). The dominant Jurassic formations are related to the predominant shallow (less than 30 cm depth) carbonated soils, from sandy loam to sandy clay. At CL, Pliocene formations are dominant, occupying the flat or gently undulating (with slopes lower than 5%) areas adjacent to the Tagus river alluvial area. Mostly related to the nature of these geological formations, the predominant soils are deep sandy soils. In both study areas, ZA and CL, the soils are poor in nutrient and organic matter, and with low water storage capacity (Costa, Madeira, & Oliveira, Citation2008; Dehane, Citation2012).
At ZA, the cork oak woodland are mainly open woodland and scattered-tree woodland, with the occurrence of thorny trees due to fire recurrence (Letreuch-Belarouci, Letreuch-Belarouci, Benabdeli, & Medjahdi, Citation2009) and overgrazing. Exploited tree’s density is under 50 trees ha−1, and trees are old and poorly managed, within heterogeneous cork-production cycles, extending largely from typical Iberia’s 9 to 12-years. Cork oak management oriented to the cork production management is lacking.
At CL, the cork oak forest management practices, from planting or seeding to pruning, thinning, and cork harvesting, all are oriented toward optimizing cork productivity (Costa, Oliveira, Vidas, & Borges, Citation2010; Costa et al., Citation2008). The cork oak woodland are uneven-aged and resulted of assist natural regeneration complemented with local artificial seeding and plantation. Trees are pruned in juvenile ages for maximize harvesting surface, before the beginning of the 9-year cork production cycles, at the 20–25 years of age. Thinning occurs regularly, only to eliminate dead trees. The mean tree density was 86 trees ha−1 (with 62 trees ha−1 under exploitation) and an average annual cork production of 110 kg ha−1 year−1 was to be expected (Costa et al., Citation2010). The cork productivity at CL, Portugal, 110 kg ha−1 year−1, is higher than at ZA, Algeria with 33 kg ha−1 year−1 but mean annual cork growth is similar 2.9 and 3.1 mm year−1, respectively.
2.2. Image acquisition and cork quality assessment
A total of 41 cork samples were randomly selected in the two study areas. The cork samples, with dimensions of 10 × 10 cm2, were prepared for image acquisition. The preparation included boiling in water at 100°C for 1 h, and dried in open air until equilibrium (minimum one week), and an optical quality surface finishing: the cutting and the sanding of a plan surfaces in the transversal section in order that cork pores and cork annual rings were clearly visible (Costa, Nunes, Graça, & Spiecker, Citation2015). The image acquisition of the transversal sections was made through snapshot images scanned at a resolution of 300 dpi and stored in TIF graphic format (Supplementary material, Figure S2). Images were then analyzed using ImageProPlus® image-processing software.
Cork quality assessment was made at the transversal section within a defined rectangular area of interest (AOI). A set of 13 pore variables at cork sample level was selected for cork quality assessment (Table ). Pore data were filtered out by pore area, only pores with an area equal or superior to 0.5 mm2 were kept for analysis as small porosity is functionally and esthetically irrelevant and only brings higher variance and variability to the cork sample (Costa & Pereira, Citation2007).
Table 1. Pore variables measured in the transverse section of the cork samples at both regions
The exploratory analysis of cork porosity was made in the transverse section of the cork sample at an AOI-level; firstly, considering the AOI and as a whole and then considering the inner part of the AOI (one third of the AOI transect length, from cork belly to cork back); the middle AOI (second third); the external AOI (one third of the AOI length). At these subsections (inner, middle, and external AOI) of the cork sample transverse section, the found cork porosity will be therefore comparable and independent on the cork age, cork thickness, and cork production cycles.
2.3. Statistical analysis
Stepwise discriminant analysis (SDA) was used to investigate differences between the cork-producing regions, ZA and CL, the categorical dependent variable, and the cork porosity features, the independent pore-level metrics variables (Costa & Pereira, Citation2006; Gonzalez-Adrados, Lopes, & Pereira, Citation2000). The stepwise discriminant functions were evaluated using the Wilk’s λ value: the smaller the Wilk’s λ value, the more important is the independent variable or feature to the canonical discriminant function. The purpose of using SDA was to find the pore features that contribute the most to distinguish these regions (SPSS vs. 21.0). This statistical analysis creates a new (reduced) set of variables, called canonical discriminant functions, each one a linear function of the original independent (predictor) variables, in this case the cork porosity at pore level. These discriminant functions would account for most of the variance and would define the maximum possible difference between both regions. The relative importance of the predictor variables is evaluated through their correlation (Pearson coefficients) with each discriminant function.
3. Results
3.1. Cork porosity
The porosity profile of the transversal sections of the cork planks from Hafir-Zarieffet (ZA) and from Benavente (CL) showed similarities in the range of the values variation for all the selected features (Table ). Regarding the dimension features, the mean pore area (Mpa) ranged between 4.5 mm2 (at ZA) and 5.3 mm2 (at CL). The mean values for maximum pore area (Maxpa) ranged between 29.2 mm2 (at ZA) and 47.6 mm2 (at CL). At the later region, the Maxpa was almost the double, confirming the relatively higher Mpa found in CL.
Table 2. Cork porosity of the transverse section of cork samples
Lenticular channels are the dominant cork tissue discontinuities, mostly linear shape objects crossing radially the transverse section of the cork planks. Shape features such as aspect ratio, fractal dimension, or roundness showed compatible values in both regions, CL and ZA (Table ). The mean fractal dimension was 1.1 in average in both regions and the roundness and aspect ratio mean values at ZA, respectively, 8.9 and 6.2, were superior to those found at CL, respectively, 4.6 and 4.7. These differences are confirmed by the fact that at ZA, the lenticular channels are longer (6.2 mm mean length) and relatively thinner (1.1 mm mean width) than at CL (4.9 and 1.5 mm, respectively for length and width) (Table ).
In relation to the concentration features, ZA presents a porosity coefficient of 9.2% against 13.2% at CL. In average, the cork samples from CL showed higher porosity, at CL, the larger Mpa and higher Maxpa directly influences the porosity coefficient as CL presents smaller number of pores, 250 pores per 100 cm2 against 280 pores per 100 cm2 at ZA.
The distribution of cork porosity per classes of pore area showed similarities between the two regions of ZA and CL. However, at CL, a larger number of pores with area smaller than 2 mm2 (# 229 pores) is noticeable when compared to ZA (# 151) (Figure ) and the correspondent pore area (of pores with area <2 mm2) was higher (288 mm2) at CL when compared to the correspondent area of 186 mm2, at ZA. This difference could significantly increase the porosity coefficient (area fraction) of the cork samples at CL (see Table ). In addition, at CL, pores of extremely large dimension (>30 mm2) were noticed in the cork samples, increasing the range of pore area variation (Figure ).
Figure 2. Histogram for the pore area feature in all cork samples from Hafir-Zarieffet (ZA) and Benavente (CL).
3.2. Cork quality profile
The porosity from the cork samples of ZA and CL was represented in a pore centroid scatterplot (Figure ), showing the location of the total number of pores, 205 at ZA and 494 at CL, from cork back (minimum y-value) to cork belly (maximum y-value) and the correspondent area, represented by a proportional circle.
Figure 3. Representation of pore centroids in a XY-scatter plot for all cork samplers at Hafir-Zarieffet (ZA) and Benavente (CL).
The spatial distribution of pores in the transverse section of cork samples showed that the porosity at ZA ranged between y-values of 0 and 50 mm, while at CL ranged between y-values of 0 and 25 mm (Figure ), i.e. cork samples thickness at ZA are almost the double than the ones at CL. At CL, the pore area decreases from the cork back (outer cork) to cork belly (inner cork) (Table ), while this trend is not as consistent at ZA as there are relatively larger cork pores in the inner cork layers (Figure ).
The pore centroids of the larger pores at ZA and CL are located in the middle cork layers in result of the linear projection of the lenticular channels in a transverse section, but it is in the yearly cork layers (more than at the inner cork layers) that are located the smallest porosity in both regions (Table ). Pore area (pa), length (le) and fractal dimension (fr) of the lenticular channels at CL showed a effectively a consistent decreasing trend of from cork back (outer-cork layers) to cork belly (inner- or yearly cork layers). However, at ZA, none of these pore variables allowed such as consistent trend and it is noticeable in the spatial representation of the porosity (Figure ).
3.3. Cork quality regions
The SDA functions allowed distinguishing the porosity of the two regions (Wilks’ Lambda significant at p < 0.001). Two SDA functions (F1 and F2) explained about 98% of the variation in porosity of the transverse section of the cork planks: 77.5% variation was explained by F1 and 20.7% was explained by F2.
The strongest correlations between the first SDA function, F1, and the independent variables were found for pore length (equal to 0.698) and pore roundness (equal to 0.694). Between the second SDA function, F2, which explained less variability, and the independent variables the higher correlation was found with pore perimeter (equal to 0.727) (Table ).
Table 3. Eigenvalues, canonical correlation coefficients, and variance explained by the first two discriminant functions (F1 and F2)
The results (Table ) showed that, at the AOI-level, and considering the cork layers, the selected independent pore-level variables, with the highest correlation with the SDA functions were those quantifying pore linear dimension, as pore perimeter and pore length, and related to pore shape, as roundness.
The cork layer that was decisive to discriminate between the study regions, ZA and CL was the middle-cork layer. The lenticular channel-driven porosity be the most prominent feature of the transverse section of the cork planks. At ZA, the cork thickness is rather larger when compared to CL, and there are also some cork tissue discontinuities, such as cracks or heavy porosity, that were present in the transverse section, and this abnormal porosity lead to an increasing variability on porosity features, between cork layers when compared to the cork samples at CL (see Figure ).
4. Discussion
This study show the possibility of using image analysis techniques for quality evaluation of the cork planks, currently made based only on visual appreciation by an experienced operator. Despite the fact that the quality of the cork planks directly conditioned the quality yield profile of the cork products, such as cork stoppers or disks (Costa & Pereira, Citation2010), the image analysis in the raw material is not done only because as a natural raw material, cork is more heterogeneous and the image analysis is done in the end product.
The porosity values found in the transverse section of the cork samples were in agreement with the values found in cork planks by previous reports for both regions (Dehane, Bouhraoua, González-Adrados, & Belhoucine, Citation2011; Gonzalez-Adrados et al., Citation2000; Pereira, Lopes, & Graça, Citation1996), and porosity was in general mostly related to the radial development of the lenticular channels in the bark of the living tree (Graça & Pereira, Citation2004).
The intrinsic variability of porosity is higher in the ZA cork samples than in CL cork samples (Table ) and this could be in result of the difference in the area observed by image analysis, which is higher for the larger cork samples of ZA (Figure ). The importance of dimension of the observed area for porosity determinations by image analysis was previously addressed for cork planks (Pereira et al., Citation1996).
According to the results obtained, the high area fraction found at CL, when compared to ZA, was driven by small porosity (mostly inferior to 2 mm2). This so called small porosity should not be as detrimental for the cork planks quality as larger cork porosity. However, at CL there are also larger pores as in ZA and this is highly detrimental for cork quality. As previously reported (Costa & Pereira, Citation2010), small porosity could represent 75% of the total number of the pores and less than 20% of the total pore area.
At ZA, the longer and thinner lenticular channels are mainly related to the larger cork production cycles, larger cork thickness of the cork planks. These longer cork production cycles should be reconsidered in the cork oak woodland management as could also be detrimental to cork quality yield in the sense that the cork quality profile could not be as consistent as in CL. In Algeria regions, where the annual cork growth is relatively high, similar to the one found in the study area where cork samples were collected (annual cork growth rate between 3.1 and 3.6 mm year−1), forest management should include the shortening of the typical cork production cycles (to minimum of 9-year cork production cycle as in Benavente (Portugal). This way, the cork oak woodland would provide a more sustainable productive land use, maintaining their ecological resilience to cork harvest and a fine adaptation to the Mediterranean environments.
The porosity profile found in the transverse section of the cork samples is, however, consistent and emphasis the typical techniques when punching the natural cork stoppers, near the cork belly, in relation to the cork back, as in this region the lenticular channels, and porosity, are a more conspicuous feature. Thus, from the inner (recent) cork layers to the outer (older) cork layers, the porosity increases.
The results on the decision-making of the cork planks porosity discriminated by regions highlighted the importance of dimensional variables at pore level, length, and perimeter, in agreement with previous studies (Pereira et al., Citation1996). However, in contrast to these previous reports, the importance of the area of the pore was not decisive. This apparent discrepancy could be related to the fact that in this study the cork samples were more heterogeneous and large lenticular channels were found in the cork samples, which could be classified as cracks or heavy porosity, with a relatively poor match for a SDA analysis. Further studies should be made with more cork samples, in order to have a clear understanding on the cork quality profile for both regions, and mainly for Algeria, showing higher heterogeneity in the cork porosity features.
5. Conclusion
The porosity profile of cork samples of the two study areas, Benavente (CL) and Hafir-Zarieffet (ZA), were compared and results showed strong similarities related to lenticular channels presence and their correspondent cork tissue discontinuities. At both regions, CL and ZA, in the transverse section of the cork samples, larger and conspicuous porosity was found in outer part of the transverse section of cork planks, when compared to the one found in the inner part. However, when comparing cork quality of both regions, CL and ZA, clearly ZA had more heterogeneous cork quality, discriminated by porosity linear features, such as length and perimeter. Clearly, at Algeria (ZA), adequate thinning and pruning of cork oak woodland and appropriate cork production cycles should be applied toward the improvement of the cork quality production.
Funding
This research was partially supported by the Foundation for Science and Technology of the Portuguese Ministry of Education and Science (FCT-MEC) through the research projects [grant number UID/AMB/04085/2013], [grant number EXPL/AGR/FOR/1220/2012]. Augusta Costa’s contribution was funded by the FCT-MEC under [grant number SFRH/BPD/97166/2013]. Amina Ghalem’s internship was supported by University Abou Baker Belkaid of Tlemcen, as part of her PhD studies.
Supplementary_material.docx
Download MS Word (1.6 MB)Acknowledgments
Authors acknowledge the collaboration of Beni Mester Cork Industry (Hafir-Zarieffet, Algeria) and of Companhia das Lezírias, S.A. (Benavente, Portugal) for the implementation of the study areas. Authors acknowledge the comments of two anonymous reviewers.
Additional information
Notes on contributors
Amina Ghalem
The authors were motivated by the spatial variability of cork oak woodland across Mediterranean environments, the implications in the cork yield, and cork quality and, consequently, in the economic and ecological sustainability of these sensitive forest ecosystems. Authors used image analysis techniques to study cork porosity in cork planks, in a similar way that cork industry detects cork porosity in cork products such as natural cork stoppers and disks, the most valuable cork product in the cork industrial processing.
This study will be one first original report from a young research group, integrating young MSc and PhD students from the Center for Environmental and Sustainability Research (NOVA University of Lisbon, Portugal) and Tlemcen University (Algeria) (http://www.augustacosta.net/people.html) addressing a comprehensive understanding on cork planks quality and cork oak woodland management, working at INIAV, I.P.
References
- APCOR. (2015). Anuário Estatístico. http://www.apcor.pt/wp-content/uploads/2015/12/APCOR-Boletim-Estatistico.pdf
- Bugalho, M. N., Caldeira, M. C., Pereira, J. S., Aronson, J., & Pausas, J. G. (2011). Mediterranean cork oak savannas require human use to sustain biodiversity and ecosystem services. Frontiers in Ecology and the Environment, 9, 278–286.10.1890/100084
- Capelo, J., Mesquita, S., Costa, J., Ribeiro, S., Arsénio, P., Neto, C., ... Lousã, M. (2007). A methodological approach to potential vegetation modeling using GIS techniques and phytosociological expert-knowledge: Application to mainland Portugal. Phytocoenologia, 37, 399–415.10.1127/0340-269X/2007/0037-0399
- Costa, A., & Oliveira, G. (2015). Cork oak (Quercus suber L.): A case of sustainable bark harvesting in Southern Europe. In C. M. Shackleton, A. K. Pandey, & T. Ticktin (Eds.), Ecological sustainability for non-timber forest products dynamics and case studies of harvesting (pp. 179–198). New York, NY: Earthscan from Routledge.
- Costa, A., & Pereira, H. (2006). Decision rules for computer vision quality classification of wine natural cork stoppers. American Journal of Enology and Viticulture, 57, 210–219.
- Costa, A., & Pereira, H. (2007). Influence of vision systems, black and white, colored and visual digitalization, in natural cork stopper quality estimation. Journal of the Science of Food and Agriculture, 87, 2222–2228.10.1002/(ISSN)1097-0010
- Costa, A., & Pereira, H. (2010). Influence of cutting direction of cork planks on the quality and porosity characteristics of natural cork stoppers. Forest System, 19, 51–60.
- Costa, A., Madeira, M., & Oliveira, Â. C. (2008). The relationship between cork oak growth patterns and soil, slope and drainage in a cork oak woodland in Southern Portugal. Forest Ecology and Management, 255, 1525–1535.10.1016/j.foreco.2007.11.008
- Costa, A., Madeira, M., Santos, J. L., & Plieninger, T. (2014). Recent Dynamics of Mediterranean Evergreen Oak Wood Pastures in Southwestern Iberia. In T. Hartel, & T. Plieninger (Eds.), European wood-pastures in transition—A social-ecological approach (pp. 70–89). New York, NY: Earthscan from Routledge.
- Costa, A., Nunes, L. C., Graça, J., & Spiecker, H. (2015). Insights into the responsiveness of cork oak (Quercus suber L.) to bark harvesting. Economic Botany, 20(10), 1–14.
- Costa, A., Oliveira, A. C., Vidas, F., & Borges, J. G. (2010). An approach to cork oak forest management planning: A case study in southwestern Portugal. European Journal of Forest Research, 129, 233–241.10.1007/s10342-009-0326-y
- Dehane, B. (2012). Incidences de l’etat sanitaire des arbres du chêne-liége sur les accroissements annuels et la qualité du liège de deux suberaies oranaises: M’Sila (W Oran) et Zarieffet (W. Tlemcen) [Impact of health condition of cork oaks on cork annual increases and quality at cork oak forests in Oran: M’Sila (W Oran) and Zarieffet (W. Tlemcen)] (Dissertation). University of Tlemcen, Algeria.
- Dehane, B., Bouhraoua, R. T., González-Adrados, J. R., & Belhoucine, L. (2011). Caractérisation de la qualité du liège selon l’état sanitaire des arbres par la méthode d’analyse d’image Cas des forêts de M’Sila et de Zarieffet (Nord-Ouest Algérien) [Cork quality characterization assessed by image analysis according to the health status of the trees on the forests of M’Sila and Zarieffet (North-West of Algeria)]. Forêt méditerranéenne, 32, 39–50.
- FAO. (2013). State of Mediterranean forests. Rome. Retrieved from http://www.fao.org/docrep/017/i3226e/i3226e.pdf
- Giorgi, F. (2006). Climate change hot-spots. Geophysical Research Letter, 33, L08707.
- Gonzalez-Adrados, J. R., Lopes, F., & Pereira, H. (2000). Quality grading of cork planks with classification models based on defect characterisation. Holz als Roh- und Werkstoff, 58, 39–45.10.1007/s001070050383
- Graça, J., & Pereira, H. (2004). The periderm development in Quercus suber. IAWA Journal, 25, 325–335.10.1163/22941932-90000369
- Letreuch-Belarouci, A. M., Letreuch-Belarouci, N., Benabdeli, K., & Medjahdi, B. (2009). Impact des incendies sur la structure des peuplements de chêne-liège et sur le liège: le cas de la subéraie de Tlemcen (Algérie) [Impact of wildfires on the structure of cork oak stands and on cork yield: the case of the cork oak forests of Tlemcen (Algeria)]. Forêt méditerranéenne, 30, 231–238.
- Messaoudène, M., & Merouan, H. (2009). Site profile 1.1: Akfadou, Algeria. In J. Aronson, J. S. Pereira, & J. G. Pausas (Eds.), Cork oak woodlands on the edge ecology, adaptive management, and restoration (pp. 22–23). Washington DC: Island Press.
- Oliveira, G., & Costa, A. (2012). How resilient is Quercus suber L. to cork harvesting? A review and identification of knowledge gaps. Forest Ecology and Management, 270, 257–272.10.1016/j.foreco.2012.01.025
- Pereira, H. (2007). Cork: Biology, production and uses. Amsterdam: Elsevier.
- Pereira, H., Lopes, F., & Graça, J. (1996). The evaluation of the quality of cork planks by image analysis. Holzforschung, 50, 111–115.10.1515/hfsg.1996.50.2.111
- Plieninger, T., van der Horst, D., Schleyer, C., & Bieling, C. (2014). Sustaining ecosystem services in cultural landscapes. Ecology and Society, 19, 59. doi:10.5751/ES-06159-190259
- Ticktin, T. (2004). The ecological implications of harvesting non-timber forest products. Journal of Applied Ecology, 41, 11–21.10.1111/jpe.2004.41.issue-1
- Vicente, Á. M., & Alés, R. F. (2006). Long term persistence of dehesas. Evidences from history. Agroforestry Systems, 67, 19–28.10.1007/s10457-005-1110-8