Edge Effects of Oil Pipeline Canopy Openings on Tree Community Structure and Dynamics in a Montane Atlantic Forest

The Atlantic forest has historically been severely deforested, and only fragments currently remain that are subject to a wide variety of anthropogenic impacts, including edge effects that can cause structural and functional degradation. The Tinguá Biological Reserve-RJ comprises approximately 26,000 hectares of well-preserved Atlantic Forest, but it is subject to impacts caused by two canopy openings along oil pipelines. Comparisons were made between pipeline edges and forest interiors to evaluate edge effects on the structure and dynamics of those tree communities. Tree densities were higher along forest edges, apparently increasing over time. Tree basal areas, on the other hand, have decreased along edges due to higher mortality rates. Linear canopy opening edges showed higher densities of small trees, while the interior had more very large trees, indicating changes in successional processes and community structural patterns due to edge effects.


Introduction
Fragmentation of tropical forests can promote edge effects that lead to accelerated increases in plant recruitment and mortality [1] [2].Small forest fragments showed changes in their structure, increased density, and reduced biomass [3] [4], resulting in both local and regional extinctions of tree species [5] [6] and impoverished edge communities [12].Some species, however, such as pioneer trees [7] and lianas [8], can be favored by the higher solar radiation levels found in disturbed sites, and show increases in recruitment and growth rates [9].The structural and functional changes caused by edge effects are currently viewed as responsible for many of the major alterations seen in tropical forests [10].
The Atlantic Forest is the second largest forest type in Brazil, originally covering an area of 1.1 million km 2 and extending from the states of Rio Grande do Norte to Rio Grande do Sul [11] [12] although it is now reduced to only 11% of its original size [13].This biome is counted among the 25 primary global "hotspots"-areas of high diversity and high concentrations of endemic species that are threatened by rapid habitat losses [14].The ecological relationships between species have been severely altered in many tropical forests and their biological communities are at risk of extinction [15] due to deforestation, selective cutting, highway construction, and urban expansion [16], all leading to habitat loss and fragmentation [14] [17] [18].
These anthropogenic processes create mosaics of small forest remnants within urban or agricultural matrices [19], and increases in forest edges are followed by structural and functional changes in communities that arise from alterations of abiotic and biotic conditions [20] [21].Increased light and wind flux near edges [22] can increase air temperatures and vapor pressure deficits, and decrease soil moisture levels [3], generating an edgeinterior microclimatic gradient.These changes invariably lead to negative biotic responses, including increased mortality, recruitment, damage, and tree falls [7] [8] [22].
While oil pipeline canopy openings represent a threat to tropical forests, they can also be viewed as potential large-scale field experiments [23], and we used such openings (created by two oil pipelines inside a protected area) to examine whether they create edge effects that impact the structures and dynamics of Brazilian montane Atlantic Forests.

Study Area
The Tinguá Biological Reserve is located at Rio de Janeiro State, Brazil, and covers 26,000 hectares of preserved Atlantic Forest fragments (22˚28' -39'S × 43˚13' -34'W).The landscape there is rugged, with mountains intersected by river valleys.The climate is classified as Cwb (Köeppen), with temperatures ranging from 13˚ to 23˚C, and average annual rainfall between 1500 -2600 mm.According to the classification proposed by [24], four vegetation types can be found in the reserve; submontane forests, montane forests, highland forests, and highland fields.
The local vegetation is relatively well-preserved, mainly because of the difficult access to the reserve, and its rivers supply water for part of Rio de Janeiro.Despite its importance, human impacts such as hunting and selective extraction are commonly observed (e.g., palm hearts-Euterpe edulis Mart), and two oil pipelines run through the area creating linear canopy openings [25], designated here as the Old Pipeline (OldP) and New Pipeline (NewP).The former was built following a very old (300 years) road, while the second was laid down approximately 40 years ago.

Sampling Design and Comparisons
To evaluate edge effects on the structures and dynamics of the tree communities, ten permanent plots (300 m 2 per plot) were installed in each of three forest situations (treatments); along the edges of the Old Pipeline (OldP) and the New Pipeline (NewP), and in the forest interior (more than 400 meters from any edge).All of the trees with diameters at breast height ≥ 5 cm (hereafter DBH) inside each plot were tagged and their diameters measured in 2003, 2004 and 2005.To verify whether the treatments differed in terms of their tree densities, basal areas, the numbers of dead trees, mortality, and recruitment, statistical comparisons were made between the treatment sites (OldP, NewP, and INT) using One-way ANOVA followed by the HSD Tukey test (p < 0.05).The mean rates of mortality and recruitment were calculated using the formula m = 1 − (N 1 /N 0 ) 1/t and r = 1 − (1 − nr/Nt) 1/t [42] [43].Diameter distributions (DBH) were compared using Hierarchical two-way ANOVA (Nested plot in treatment) followed by the Tukey test (p < 0.05) [45].Normality and homoscedasticity tests were performed and, when necessary, the data were log (x + 1) or sin −1 transformed (to percentage data) [44].In some cases, we analyzed small trees (5 ≤ DBH < 10 cm) and large trees (DBH ≥ 10 cm) separately to sharpen comparisons using ANOVA.

Results
The tree densities per hectare (n.ha −1 ) (2004/2005) were 1853/1857 in the forest interior (INT) and 2136/2173 and 2353/2380 along the edges of the Old Pipeline (OldP) and New Pipeline (NewP) respectively (Table 1).Thus, trees densities per plot were higher along the disturbance edges, with a significant difference mainly between the New Pipeline and the Interior in the years 2004 and 2005 (F = 4.25; p < 0.02 and F = 5.05; p < 0.01 respectively) (Figure 1).In 2005, the edges also showed more standing dead trees, with 5.5% (OldP) and 6.6%  (NewP) as opposed to 4% in the Interior (Figure 1).Surprisingly, no differences were found in terms of basal area between treatments during the study period.
Size structures considering all trees (DBH ≥ 5 cm) did not differ between the treatments (Figure 2), although considering small trees (5 ≤ DBH < 10 cm) and large trees (DBH ≥ 10 cm) separately showed distinct patterns.Small tree densities and basal areas per plot no showed differences between the treatments during the study years (p < 0.05).Large tree densities were higher along the New Pipeline as compared to the Interior in the years 2004 and 2005 (F = 5.23, p < 0.01 and F = 4.83, p < 0.02 respectively) (Table 2) 2).Tree densities along the New Pipeline were higher, however, mainly in terms of the size classes 10 -20, 20 -30, and 30 -40 cm, while the densities of very large trees (diameters > 40 cm) were always higher in the Interior (Figure 2).
Vegetation dynamics along the edges were more accelerated as compared to the interior, principally along the New Pipeline.Despite the general pattern of recruitment of new individuals between the years 2003-2005 (Old Pipeline: 1.5%; New Pipeline: 7.28%, and Interior: 3.23%), the edges showed reduced basal areas (Old Pipeline: −0.59%;New Pipeline: −3.97%) while Interior gained less basal area (3.12%).Furthermore, New Pipeline mortality rates were higher and differed significantly from the interior site during the period from 2004-2005 (Figure 3).
Edge effects can also be detected by observing the occurrence of individuals with multiple trunks, and higher frequencies of these individuals were seen along the edges (Figure 4).On the other hand, the palm tree Euterpe edulis exhibited strong reductions in numbers along forest edges.

Discussion
Edge effects due to disturbance regimes can cause profound changes in species interactions and resource Table 2. Comparisons between the communities of small trees (5 ≤ DBH < 10 cm) and large trees (DBH ≥ 10 cm) in the Forest Interior, and along the Old Pipeline and New Pipeline at the Tinguá Biological Reserve, Rio de Janeiro State, Brazil.The marked ( * ) F values indicate significant differences between treatments.Density means (±standard deviation) (n), basal areas (BA) and diameters (DBH) followed by the same letter (a or b, on the same lines) do not differ significantly (Tukeytype test, p < 0.05).Tree densities increased over the years, and were highest along the pipeline, due to greater light availability, which generally favors recruitment [28].Canopy openness near edges (with increased wind and light flux) can therefore favor recruitment, mortality [7] [8] [28] and the growth of small plants and/or [29].Their observed mortality and recruitment rates were similar, however, to other Tropical Forests [12] [30] [31].Some tropical forests lose from 15% to 35% of their biomass due to increased mortality in the first 5 -10 years after fragmentation [32].In the case of Tinguá Biological Reserve, the expected decrease in basal area was observed only along pipeline edges during the period from 2003-2005 (Old Pipeline: −0.59%,New Pipeline: −3.97%, Interior: +3.12%, Table 1).
Therefore, increases in the numbers of small pioneer trees [28] [33] [34] and changes in successional patterns due to edge effects [5] [35] resulted in higher densities of small trees along the forest edges.As small plants are generally more fragile and may exhibit higher mortality rates [36] [37] vegetation dynamics will often be accelerated along edges.Additionally, very large trees may be excluded depending on the disturbance scale and intensity [8] [15], which would explain why very large trees were only found in the Interior.
Edge effects may be attenuated over time [5], however, by the buffering effects of new vegetation originating at the edge [19], or by the establishment of certain vegetation communities such as vines and bamboos [38] along those disturbed areas [7] [12] [39] [40].The oldest edge (Old Pipeline) was found to be more similar to the interior.The maintenance of abrupt edges and the proximity of shrubby or herbaceous matrices, however, can favor continuing anthropogenic impacts [40] [41].Linear canopy openings may be less harmful, than fire-prone sites and agricultural fields, although they can facilitate the entry of hunters and collectors.The lower edge densities of the heart-palm Euterpe edulis (pers.observation) is related to illegal extraction and numerous cut trunks were observed during the course of the fieldwork.

Conclusion
Tree densities were higher along forest edges while tree basal areas decreased along edges due to higher mortality rates.On the other hand, edges showed higher densities of small trees, while the interior had more very large trees.Furthermore, pipelines can also favor the anthropogenic damage by providing access to hunters and collectors.Thus, these linear canopy openings promote edge effects, although they seem to be less detrimental than edges on fire-prone sites and agricultural fields.

Figure 1 .
Figure 1.Standing dead trees density per plot (A) and live trees density per plot (B) and on the treatments forest Interior, Old Pipeline and New Pipeline at Tinguá Biological Reserve.Means (±standard deviation) of density in the respective year followed by the same letter (a or b) do not differ significantly (p < 0.05).

.
Differences were observed between the Old Pipeline and the Interior in terms of the DBH values of small trees (5 ≤ DBH < 10 cm) in 2003 and 2004 (F = 3.52, p < 0.03 and F = 4.33, p < 0.01), and in 2005 both edges were different from the Interior (F = 4.93, p < 0.007), (Table 2) with the latter having higher values.The same pattern was observed for the size structures of the large trees, with the interior site having higher values than the New Pipeline in the years 2004 and 2005 (F = 3.96, p < 0.02 and F = 3.42, p < 0.03) (Table

Figure 2 .
Figure 2. Trees frequency diameter distributions (%) on the treatments forest Interior, Old Pipeline and New Pipeline in six DBH (cm) classes at Tinguá Biological Reserve in the years 2003, 2004 and 2005.

Figure 3 .
Figure 3. Recruitment (a) and mortality (b) annual rates on the treatments forest Interior, Old Pipeline and New Pipeline at Tinguá Biological Reserve in the periods 2003-2004 and 2004-2005.Treatments with means (±standard deviation) followed by the same letter (a or b) do not differ (p < 0.05).

Figure 4 .
Figure 4. Density of multiple trunks and Euterpe edulis palm per plot on the treatments forest Interior, Old Pipeline and New Pipeline at Tinguá Biological Reserve in the year 2005.Means (±standard deviation) of density followed by the same letter (a or b) do not differ significantly (p < 0.05).availability[4][5][7] [8] often leading to increases in mortality, recruitment, biotic damage, and tree falls[7] [8][22], and tropical forests subjected to edge effects often remain in early-successional stages[15] [26][27].Tree structures and their dynamics indicated that the forest communities in the Tinguá Biological Reserve were subject to edge effects, especially along the edge of the most recently installed pipeline.Tree densities increased over the years, and were highest along the pipeline, due to greater light availability, which generally favors recruitment[28].Canopy openness near edges (with increased wind and light flux) can therefore favor recruitment, mortality[7] [8][28] and the growth of small plants and/or[29].Their observed mortality and recruitment rates were similar, however, to other Tropical Forests[12] [30][31].Some tropical forests lose from 15% to 35% of their biomass due to increased mortality in the first 5 -10 years after fragmentation[32].In the case of Tinguá Biological Reserve, the expected decrease in basal area was observed only along pipeline edges during the period from 2003-2005 (Old Pipeline: −0.59%,New Pipeline: −3.97%, Interior: +3.12%, Table1).Therefore, increases in the numbers of small pioneer trees[28] [33][34] and changes in successional patterns due to edge effects[5] [35]  resulted in higher densities of small trees along the forest edges.As small plants are generally more fragile and may exhibit higher mortality rates[36] [37] vegetation dynamics will often be accelerated along edges.Additionally, very large trees may be excluded depending on the disturbance scale and intensity[8] [15], which would explain why very large trees were only found in the Interior.Edge effects may be attenuated over time[5], however, by the buffering effects of new vegetation originating at the edge[19], or by the establishment of certain vegetation communities such as vines and bamboos[38] along those disturbed areas[7] [12]  [39][40].The oldest edge (Old Pipeline) was found to be more similar to the interior.The maintenance of abrupt edges and the proximity of shrubby or herbaceous matrices, however, can favor continuing anthropogenic impacts[40] [41].Linear canopy openings may be less harmful, than fire-prone sites and agricultural fields, although they can facilitate the entry of hunters and collectors.The lower edge densities of the heart-palm Euterpe edulis (pers.observation) is related to illegal extraction and numerous cut trunks were observed during the course of the fieldwork.

Table 1 .
Densities and basal areas (BA) of trees (DBH > 5 cm) per hectare in the years 2003, 2004 and 2005 in the Forest Interior, and along the Old Pipeline and New Pipeline in the Tinguá Biological Reserve, Rio de Janeiro State, Brazil.