# UAV Remote Sensing for Biodiversity Monitoring: Are Forest Canopy Gaps Good Covariates?

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

^{2}in size. Correlation and linear regression techniques were used to explore relationships between patch metrics and understory (density, development, and species diversity) or forest habitat biodiversity variables (density of micro-habitat bearing trees, vertical species profile, and tree species diversity). The results revealed that small openings in the canopy cover (75% smaller than 7 m

^{2}) can be faithfully extracted from UAV red, green, and blue bands (RGB) imagery, using the red band and contrast split segmentation. The strongest correlations were observed in the mixed forests (beech and turkey oak) followed by intermediate correlations in turkey oak forests, followed by the weakest correlations in beech forests. Moderate to strong linear relationships were found between gap metrics and understory variables in mixed forest types, with adjusted R

^{2}from linear regression ranging from 0.52 to 0.87. Equally strong correlations in the same forest types were observed for forest habitat biodiversity variables (with adjusted R

^{2}ranging from 0.52 to 0.79), with highest values found for density of trees with microhabitats and vertical species profile. In conclusion, this research highlights that UAV remote sensing can potentially provide covariate surfaces of variables of interest for forest biodiversity monitoring, conventionally collected in forest inventory plots. By integrating the two sources of data, these variables can be mapped over small forest areas with satisfactory levels of accuracy, at a much higher spatial resolution than would be possible by field-based forest inventory solely.

## 1. Introduction

^{2}, a minimum width of 2 m and a maximum vegetation height of 3 m’. Getzin et al [37], adopting a different definition, mapped canopy gaps of 1 m

^{2}from a true colors UAV product of 7 cm spatial resolution in beech-dominated deciduous and mixed deciduous-coniferous forests in Germany. Furthermore, Getzin et al. [8] extracted forest gaps and their spatial pattern from ten different plots of 1 ha. The authors recommend collecting aerial data in a cloudy condition in order to reduce the effect of the shadow that could lead to the misinterpretation of dark pixels as gaps.

## 2. Materials and Methods

#### 2.1. Study Site

^{2}), distributed according to a one-per-stratum systematic sampling design. This sampling scheme partitions the study area into square grids of 529 m

^{2}plots spatially aggregated into 50 equal size strata. One sample-site location is independently and randomly selected in each stratum, resulting in the spatial distribution displayed in Figure 1. There were 28, 13, and 9 plots in beech, turkey oak, and mixed forests, respectively.

#### 2.2. Field Data Processing

#### 2.3. UAV Data

^{2}at 2,000 m altitude, with a maximum flight time of 45 min.

#### 2.4. Methods

#### 2.4.1. Image Processing and Variable Selection

^{2}and greater than 2 m

^{2}. This leads to two times 95 variables for each plot. These two thresholds were set, because some ecological phenomena, such as understory structure dependencies, are only quantifiable if small gaps are taken into account, yet very small gaps may not affect at all the lower dependency phenomenon and therefore constitute noise [57].

#### 2.4.2. Statistical Analysis

^{2}> 0.25 as meaningful, since the predictor value leads to a great change if it explains over 25% of variance, we used a threshold of R

^{2}> 0.5. We then validated the regression by checking regression quality assumptions such as the normality of residuals, the homoscedasticity of residuals, and that the mean of the residuals is zero. For forest parameters that can be predicted with R

^{2}greater than 50%, we generalized the model to the whole forest type extent using the grid of 529 m

^{2}plots. We performed the validation with cross-validation (Leave-One-Out Cross-Validation), which produced root mean squared errors (RMSE).

^{2}. Bootstrapping is a method commonly used for that purpose [63,64,65], especially in the case of linear regression [66]. Bootstrap is a resampling method developed by Efron and Tibshirani [67]. We computed those statistics for variables with R

^{2}greater 0.50 using formulas given by Mura et al. [63]. Although it is suggested that the bootstrap sample size should be big enough (greater than 200), there is no consensus yet on what should be the actual size. Following Mura et al. [63], we set the bootstrap sample size to 500. The mapping of biodiversity attributes was achieved with ArcGIS 10.2. A graphical abstract of the methodology is presented in Figure 2.

## 3. Results

#### 3.1. Canopy Gaps Mapping

^{2}(1 pixel) to 122 m

^{2}. This range encompassed small openings or inter-crown cracks in the canopy cover and larger gaps generated by the fall of one or more canopy trees. The gap size distribution is roughly the same in the three forest types. When considering gaps greater than 1 m

^{2}, over 75% of gaps in the three forest types were less than 5 m

^{2}. When considering gaps greater than 2 m

^{2}, over 75% of gaps in the three forest types were less than 7 m

^{2}(Figure 3).

#### 3.2. Correlation between Gap Metrics and Understory Variables

^{2}= 0.87, p < 0.000) and the lowest was in Fagus forest (N_SPECIES, R

^{2}= 0.15, p < 0.05). The best results were found in Quercus and mixed forests, whereas the worst ones were in Fagus forest. I_SHANNON and MEAN_DBH in Quercus forest, and MEAN_HTOT and N_PLANTS in mixed forest were predicted with R

^{2}> 0.50. All patch metrics, except one (cv_Length in Fagus forest), correlated with the field parameters were related to gap shape.

^{2}was greater than 0.5, because the linear model slope was close to zero. The empty cells in the figure correspond to cells in which the prediction gives unreasonable values.

#### 3.3. Correlations between Gap Metrics and Living Tree Variables

^{2}= 0.79, p < 0.000) and the lowest was in Fagus forest (%HAB, R

^{2}= 0.11, p < 0.05). Best results were in Quercus forest (I_PRETZSCH, HAB, and %HAB) and mixed forest (HAB, %HAB, MEAN_DBH, and MEAN_HTOT). HAB, %HAB, MEAN_HTOT in mixed forest and HAB and I_PRETZSCH in Quercus forest had R

^{2}exceeding 0.50 (Table 5). Among the patch metrics correlated with living tree parameters, all but the length (cv_Length in mixed forest), which is gap size related patch metric, were gap shape-related patch metrics.

#### 3.4. Quality Assessment

^{2}was relatively close to the experimental R

^{2}(computed from the actual data) for both the understory and the living trees. The standard error and the bias are very low. The highest negative bias was related to I_SHANNON with a bias of −0.05. This suggests that the actual R

^{2}of I_SHANNON is lower than the one observed. Similarly, the highest positive bias was recorded with HAB in mixed forest (living trees with a value of 0.03)suggesting that the actual R

^{2}associated with habitat trees in mixed forest is higher than 0.79 (Table 6).

## 4. Discussion

#### 4.1. Mapping Forest Canopy Gaps

^{2}, corresponding either to intra-crown openings or to inter-crown cracks in the canopy cover. This fine-scale mapping of canopy openings is important, because some ecological phenomena, such as understory structure dependencies, are only quantifiable if small gaps are taken into account. Though it is known that different forests, depending on ecological and geographical contexts, exhibit different canopy structure, there is not yet a rule for a minimum gap size. As a consequence, different authors set, arguably, minimum gap size of 1 m

^{2}[8,37,57,68,69,70], 5 m

^{2}[71], 10 m

^{2}[33,72], and 50 m

^{2}[36]. To circumvent this problem, we did not set any gap size limit but let the ecological phenomenon under investigation dictate the appropriate gap size limit. Hobi et al. [73] used the same approach by not setting any gap size limit. The gap mapping used in this paper does not take into account the vegetation height. Therefore, our mapping, although consistent with the definition given by Brokaw [74], focuses only on shaded gaps or dark objects. Zielewska-Büttner et al. [33] reported as well that in their attempt to map forest canopy gaps, the shadow occurrence and forest height affected the mapping accuracy.

#### 4.2. Modeling Understory Variables through Canopy Gaps Covariates

^{2}), might have an influence on the growth and diversification of the understory in Quercus and mixed forest types. The same level of canopy openness is not sufficient to modify light availability under beech canopy cover. This is confirmed by the relatively low-density values of beech understory.

#### 4.3. Modeling Living Trees Biodiversity through Canopy Gaps Covariates

#### 4.4. Comparison with Other Studies and Implications

^{2}, when gaps smaller than 2 m

^{2}are not considered. Strong relationships between this micro-porosity of the canopy layer and ecological phenomena beneath and inside the forest canopy were found only in two out of the three examined forest types. This finding suggests that variability in canopy cover composition and the size of field inventory plots are to be taken into account when dealing with phenomena associated with spatial patterns of sunlight penetration into the canopy.

^{2}. In their study, Getzin et al. [37] found relevant relationships in forest areas with beech as canopy-dominant species using plots almost twenty times larger than the plot size adopted in this study.

## 5. Conclusions

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## Appendix A

Patch Metric | Formula | Values Range | Description |
---|---|---|---|

Border length (${b}_{v}$) | [0, ∞) | Is basically the perimeter of the gap | |

Length (${l}_{v}$) | $\sqrt{{p}_{v}\cdot {\gamma}_{v}}$ | [0, ∞) | ${p}_{v}$ is the total number of pixels contained in the patch v ${\gamma}_{v}$ is the length-width ratio of an image object v |

Length/Width (${\gamma}_{v}$) | - | [0, ∞) | The length-to-width ratio of an image object |

Width (${w}_{v}$) | $\frac{{p}_{v}}{{\gamma}_{v}}$ | [0, ∞) | The width of an image object is calculated using the length-to-width ratio |

Asymmetry | - | [0,1] | The Asymmetry feature describes the relative length of an image object, compared to a regular polygon. An ellipse is approximated around a given image object, which can be expressed by the ratio of the lengths of its minor and the major axes. The feature value increases with this asymmetry |

Border Index | $\frac{{b}_{v}}{2\cdot ({l}_{v}+{w}_{v})}$ | [1, ∞) 1 = ideal | The Border Index feature describes how jagged an image object is; the more jagged, the higher its border index. This feature is similar to the Shape Index feature, but the Border Index feature uses a rectangular approximation instead of a square. The smallest rectangle enclosing the image object is created, and the border index is calculated as the ratio between the border lengths of the image object and the smallest enclosing rectangle |

Compactness | - | [0, ∞) 1 = ideal | The Compactness feature describes how compact an image object is. It is similar to Border Index, but is based on area. However, the more compact an image object is, the smaller its border appears. The compactness of an image object is the product of the length and the width, divided by the number of pixels |

Density | - | [0, depending on shape of image object] | The Density feature describes the distribution in space of the pixels of an image object. In eCognition Developer 9.0, the most “dense” shape is a square; the more an object is shaped like a filament, the lower its density. The density is calculated by the number of pixels forming the image object divided by its approximated radius, based on the covariance matrix |

Elliptic Fit | - | [0,1]; 1 = complete fitting, 0 = <50% fit. | The Elliptic Fit feature describes how well an image object fits into an ellipse of similar size and proportions. While 0 indicates no fit, 1 indicates a perfect fit. The calculation is based on an ellipse with the same area as the selected image object. The proportions of the ellipse are equal to the length to the width of the image object. The area of the image object outside the ellipse is compared with the area inside the ellipse that is not filled by the image object |

Radius of Largest Enclosed Ellipse (${\epsilon}_{v}^{\mathrm{max}}$) | - | [0, ∞) | The Radius of Largest Enclosed Ellipse feature describes how similar an image object is to an ellipse. The calculation uses an ellipse with the same area as the object and is based on the covariance matrix. This ellipse is scaled down until it is totally enclosed by the image object. The ratio of the radius of this largest enclosed ellipse to the radius of the original ellipse is returned as a feature value |

Radius of Smallest Enclosing Ellipse (${\epsilon}_{v}^{\mathrm{min}}$) | - | [0, ∞) | The Radius of Smallest Enclosing Ellipse feature describes how much the shape of an image object is similar to an ellipse. The calculation is based on an ellipse with the same area as the image object and based on the covariance matrix. This ellipse is enlarged until it encloses the image object in total. The ratio of the radius of this smallest enclosing ellipse to the radius of the original ellipse is returned as a feature value |

Rectangular Fit | - | [0,1] ; where 1 is a perfect rectangle. | The Rectangular Fit feature describes how well an image object fits into a rectangle of similar size and proportions. While 0 indicates no fit, 1 indicates for a complete fitting image object. The calculation is based on a rectangle with the same area as the image object. The proportions of the rectangle are equal to the proportions of the length to width of the image object. The area of the image object outside the rectangle is compared with the area inside the rectangle |

Roundness | ${\epsilon}_{v}^{\mathrm{max}}-{\epsilon}_{v}^{\mathrm{min}}$ | [0, ∞); 0 = ideal | The Roundness feature describes how similar an image object is to an ellipse. It is calculated by the difference of the enclosing ellipse and the enclosed ellipse |

Shape Index | ${b}_{v}/4\sqrt{A}$ | [1,∞) ; 1 = ideal | The Shape index describes the smoothness of an image object border. The smoother the border of an image object is, the lower its shape index |

Gap shape complexity index (GSCI) | ${b}_{v}/\sqrt{4\pi A}$ | [1,∞) ; 1 = perfect circle | It is the ratio of a gap’s perimeter to the perimeter of a circular gap of the same area |

Patch fractal dimension (PFD) | $2\cdot ln({b}_{v})/ln(A)$ | - | - |

Fractal dimension (FD) | $2\cdot ln({b}_{v}/4)/ln(A)$ | - | - |

fractal dimension index (FDI) | $2\cdot ln({b}_{v}/\sqrt{4\pi})/ln(A)$ | - | - |

## Appendix B. Correlations with understory data

**Table B1.**Coefficient of correlations of Pearson and Spearman for some selected explicative and understory dependent variables in Quercus forest.

N_PLANTS | N_SPECIES | I_SHANNON | I_PIELOU | MEAN_DBH | MEAN_HTOT | V_TOT | G_TOT | |
---|---|---|---|---|---|---|---|---|

Mdn_GSCI | Sd_Rect.fit | Sd_Rect.fit | Sd_Density | Avg_Rect.fit | Mdn_B. Index | Mdn_Asy | Avg_Asy | |

Threshold | 1 m^{2} | 2 m^{2} | 2 m^{2} | 1 m^{2} | 1 m^{2} | 2 m^{2} | 1 m^{2} | 1 m^{2} |

Pearson | 0.73 | −0.66 | −0.75 | −0.64 | 0.79 | −0.58 | −0.64 | −0.57 |

Spearman | 0.70 | −0.73 | −0.88 | −0.68 | 0.75 | −0.67 | −0.57 | −0.55 |

**Table B2.**Coefficient of correlation of Pearson and Spearman for some selected explicative and dependent variables in mixed forest.

N_PLANTS | N_SPECIES | I_SHANNON | I_PIELOU | MEAN_DBH | MEAN_HTOT | G_TOT | V_TOT | |
---|---|---|---|---|---|---|---|---|

Avg_Round. | Mdn_RSE | Mdn_RSE | Avg_RLE | Sum_width | Avg_Asym. | Sd_Asym. | Avg_Comp. | |

Threshold | 2 m^{2} | 1 m^{2} | 1 m^{2} | 2 m^{2} | 2 m^{2} | 2 m^{2} | 2 m^{2} | 2 m^{2} |

Pearson | −0.81 | 0.73 | 0.69 | 0.71 | −0.78 | 0.94 | 0.72 | −0.67 |

Spearman | −0.83 | 0.70 | 0.70 | 0.75 | −0.70 | 0.92 | 0.72 | −0.77 |

**Table B3.**Coefficient of correlation of Pearson and Spearman for some selected explicative and dependent variables in Fagus forest.

N_SPECIES | I_SHANNON | I_PIELOU | MEAN_HTOT | |
---|---|---|---|---|

Cv_Round | Sd_Round | Mdn_PFD | Cv_Lenght | |

Threshold | 1 m^{2} | 1 m^{2} | 1 m^{2} | 2 m^{2} |

Pearson | −0.43 | 0.50 | 0.74 | 0.52 |

Spearman | −0.45 | 0.56 | 0.87 | 0.57 |

## Appendix C. Correlations with living trees data

**Table C1.**Coefficient of correlations of Pearson and Spearman for some selected explicative and living trees dependent variables in Quercus forest.

N_SPECIES | I_SHANNON | I_MARGALEF | I_PRETZSCH | MEAN_DBH | MEAN_HTOT | HAB | %HAB | |
---|---|---|---|---|---|---|---|---|

Sd_rect_fit | Cv_Density | Sd_rect_fit | Sd_Density | Sum_Rect.fit | Mdn_Rect.fit | Sum_Rect.fit | Sum_RLE | |

Threshold | 2 m^{2} | 2 m^{2} | 2 m^{2} | 2 m^{2} | 1 m^{2} | 1 m^{2} | 2 m^{2} | 2 m^{2} |

Pearson | −0.68 | −0.64 | −0.71 | −0.87 | 0.61 | 0.59 | −0.79 | −0.71 |

Spearman | −0.72 | −0.61 | −0.74 | −0.90 | 0.70 | 0.56 | −0.70 | −0.64 |

**Table C2.**Coefficient of correlations of Pearson and Spearman for some selected explicative and living trees dependent variables in mixed forest.

MEAN_DBH | MEAN_HTOT | HAB | %HAB | |
---|---|---|---|---|

Avg_Round. | Sd_Asym. | Cv_Length | Avg_RSE | |

Threshold | 2 m^{2} | 1 m^{2} | 2 m^{2} | 1 m^{2} |

Pearson | 0.74 | −0.84 | −0.90 | −0.83 |

Spearman | 0.82 | −0.95 | −0.92 | −0.92 |

**Table C3.**Coefficient of correlations of Pearson and Spearman for some selected explicative and living trees dependent variables in Fagus forest.

2m | HAB | %HAB |
---|---|---|

Cv_PFD | Avg_RSE | |

Pearson | 0.43 | −0.38 |

Spearman | 0.50 | −0.39 |

## References

- Muscolo, A.; Bagnato, S.; Sidari, M.; Mercurio, R. A review of the roles of forest canopy gaps. J. For. Res.
**2014**, 25, 725–736. [Google Scholar] [CrossRef] - Karsten, R.J.; Jovanovic, M.; Meilby, H.; Perales, E.; Reynel, C. Regeneration in canopy gaps of tierra-firme forest in the Peruvian Amazon: Comparing reduced impact logging and natural, unmanaged forests. For. Ecol. Manag.
**2013**, 310, 663–671. [Google Scholar] [CrossRef] - Stan, A.B.; Daniels, L.D. Growth releases across a natural canopy gap-forest gradient in old-growth forests. For. Ecol. Manag.
**2014**, 313, 98–103. [Google Scholar] [CrossRef] - Feldmann, E.; Drößler, L.; Hauck, M.; Kucbel, S.; Pichler, V.; Leuschner, C. Canopy gap dynamics and tree understory release in a virgin beech forest, Slovakian Carpathians. For. Ecol. Manag.
**2018**, 415–416, 38–46. [Google Scholar] [CrossRef] - Amir, A.A. Canopy gaps and the natural regeneration of Matang mangroves. For. Ecol. Manag.
**2012**, 269, 60–67. [Google Scholar] [CrossRef] - Muscolo, A.; Settineri, G.; Bagnato, S.; Mercurio, R.; Sidari, M. Use of canopy gap openings to restore coniferous stands in Mediterranean environment. iFor. Biogeosci. For.
**2017**, 10, 322–327. [Google Scholar] [CrossRef][Green Version] - Čater, M.; Diaci, J.; Roženbergar, D. Gap size and position influence variable response of Fagus sylvatica L. and Abies alba Mill. For. Ecol. Manag.
**2014**, 325, 128–135. [Google Scholar] [CrossRef] - Getzin, S.; Nuske, R.S.; Wiegand, K. Using unmanned aerial vehicles (UAV) to quantify spatial gap patterns in forests. Remote Sens.
**2014**, 6, 6988–7004. [Google Scholar] [CrossRef] - Nagendra, H. Using remote sensing to assess biodiversity. Int. J. Remote Sens.
**2001**, 22, 2377–2400. [Google Scholar] [CrossRef] - Tang, L.; Shao, G. Drone remote sensing for forestry research and practices. J. For. Res.
**2015**, 26, 791–797. [Google Scholar] [CrossRef] - Giannetti, F.; Chirici, G.; Gobakken, T.; Naesset, E.; Travaglini, D.; Puliti, S. A new set of DTM-independent metrics for forest growing stock prediction using UAV photogrammetric data. Remote Sens. Environ.
**2018**, 213, 195–205. [Google Scholar] [CrossRef] - Torresan, C.; Berton, A.; Carotenuto, F.; Di Gennaro, S.F.; Gioli, B.; Matese, A.; Miglietta, F.; Vagnoli, C.; Zaldei, A.; Wallace, L. Forestry applications of UAVs in Europe: a review. Int. J. Remote Sens.
**2017**, 38, 2427–2447. [Google Scholar] [CrossRef] - Puliti, S.; Olerka, H.; Gobakken, T.; Næsset, E. Inventory of Small Forest Areas Using an Unmanned Aerial System. Remote Sens.
**2015**, 7, 9632–9654. [Google Scholar] [CrossRef][Green Version] - Dandois, J.P.; Ellis, E.C. High spatial resolution three-dimensional mapping of vegetation spectral dynamics using computer vision. Remote Sens. Environ.
**2013**, 136, 259–276. [Google Scholar] [CrossRef] - Fardusi, M.J.; Chianucci, F.; Barbati, A. Concept to practices of geospatial information tools to assist forest management & planning under precision forestry framework: A review. Ann. Silvic. Res.
**2017**, 41, 3–14. [Google Scholar] [CrossRef] - Remondino, F.; Spera, M.G.; Nocerino, E.; Menna, F.; Nex, F. State of the art in high density image matching. Photogramm. Rec.
**2014**, 29, 144–166. [Google Scholar] [CrossRef] - Zahawi, R.A.; Dandois, J.P.; Holl, K.D.; Nadwodny, D.; Reid, J.L.; Ellis, E.C. Using lightweight unmanned aerial vehicles to monitor tropical forest recovery. Biol. Conserv.
**2015**, 186, 287–295. [Google Scholar] [CrossRef][Green Version] - Alonzo, M.; Andersen, H.E.; Morton, D.; Cook, B. Quantifying Boreal Forest Structure and Composition Using UAV Structure from Motion. Forests
**2018**, 9, 119. [Google Scholar] [CrossRef] - Messinger, M.; Asner, G.P.; Silman, M. Rapid assessments of amazon forest structure and biomass using small unmanned aerial systems. Remote Sens.
**2016**, 8, 1–15. [Google Scholar] [CrossRef] - Chianucci, F.; Disperati, L.; Guzzi, D.; Bianchini, D.; Nardino, V.; Lastri, C.; Rindinella, A.; Corona, P. Estimation of canopy attributes in beech forests using true colour digital images from a small fixed-wing UAV. Int. J. Appl. Earth Obs. Geoinf.
**2016**, 47, 60–68. [Google Scholar] [CrossRef][Green Version] - Lisein, J.; Michez, A.; Claessens, H.; Lejeune, P. Discrimination of deciduous tree species from time series of unmanned aerial system imagery. PLoS ONE
**2015**, 10. [Google Scholar] [CrossRef] [PubMed][Green Version] - Lehmann, J.R.K.; Nieberding, F.; Prinz, T.; Knoth, C. Analysis of unmanned aerial system-based CIR images in forestry-a new perspective to monitor pest infestation levels. Forests
**2015**, 6, 594–612. [Google Scholar] [CrossRef][Green Version] - Hall, R.J.; Castilla, G.; White, J.C.; Cooke, B.J.; Skakun, R.S. Remote sensing of forest pest damage: a review and lessons learned from a Canadian perspective. Can. Entomol.
**2016**, 148, S296–S356. [Google Scholar] [CrossRef][Green Version] - Michez, A.; Piégay, H.; Lisein, J.; Claessens, H.; Lejeune, P. Classification of riparian forest species and health condition using multi-temporal and hyperspatial imagery from unmanned aerial system. Environ. Monit. Assess.
**2016**, 188, 146. [Google Scholar] [CrossRef] [PubMed] - Myers, D.; Ross, C.M.; Liu, B. A Review of Unmanned Aircraft System (UAS) Applications for Agriculture. 2015 ASABE Int. Meet.
**2015**, 1. [Google Scholar] [CrossRef] - Lisein, J.; Linchant, J.; Lejeune, P.; Bouche, P.; Vermeulen, C. Aerial surveys using an Unmanned Aerial System (UAS): Comparison of different methods for estimating the surface area of sampling strips. Trop. Conserv. Sci.
**2013**, 6, 506–520. [Google Scholar] [CrossRef] - Colomina, I.; Molina, P. Unmanned aerial systems for photogrammetry and remote sensing: A review. J. Photogramm. Remote Sens.
**2014**, 7, 9632–9654. [Google Scholar] [CrossRef] - Sandbrook, C. The social implications of using drones for biodiversity conservation. Ambio
**2015**, 44, 636–647. [Google Scholar] [CrossRef] [PubMed][Green Version] - Zhang, J.; Hu, J.; Lian, J.; Fan, Z.; Ouyang, X.; Ye, W. Seeing the forest from drones: Testing the potential of lightweight drones as a tool for long-term forest monitoring. Biol. Conserv.
**2016**, 198, 60–69. [Google Scholar] [CrossRef] - Anderson, K.; Gaston, K.J. Lightweight unmanned aerial vehicles will revolutionize spatial ecology. Front. Ecol. Environ.
**2013**, 11, 138–146. [Google Scholar] [CrossRef][Green Version] - Paneque-Gálvez, J.; McCall, M.K.; Napoletano, B.M.; Wich, S.A.; Koh, L.P. Small drones for community-based forest monitoring: An assessment of their feasibility and potential in tropical areas. Forests
**2014**, 5, 1481–1507. [Google Scholar] [CrossRef] - Turner, D.; Lucieer, A.; Watson, C. An automated technique for generating georectified mosaics from ultra-high resolution Unmanned Aerial Vehicle (UAV) imagery, based on Structure from Motion (SFM) point clouds. Remote Sens.
**2012**, 4, 1392–1410. [Google Scholar] [CrossRef] - Zielewska-Büttner, K.; Adler, P.; Ehmann, M.; Braunisch, V. Automated Detection of Forest Gaps in Spruce Dominated Stands using Canopy Height Models Derived from Stereo Aerial Imagery. Remote Sens.
**2016**, 8, 1–21. [Google Scholar] [CrossRef] - Betts, H.D.; Brown, L.J.; Stewart, G.H. Forest canopy gap detection and characterisation by the use of high- resolution Digital Elevation Models. N. Z. J. Ecol.
**2005**, 29, 95–103. [Google Scholar] - Seidel, D.; Ammer, C.; Puettmann, K. Describing forest canopy gaps efficiently, accurately, and objectively: New prospects through the use of terrestrial laser scanning. Agric. For. Meteorol.
**2015**, 213, 23–32. [Google Scholar] [CrossRef] - Bonnet, S.; Gaulton, R.; Lehaire, F.; Lejeune, P. Canopy gap mapping from airborne laser scanning: An assessment of the positional and geometrical accuracy. Remote Sens.
**2015**, 7, 11267–11294. [Google Scholar] [CrossRef] - Getzin, S.; Wiegand, K.; Schöning, I. Assessing biodiversity in forests using very high-resolution images and unmanned aerial vehicles. Methods Ecol. Evol.
**2012**, 3, 397–404. [Google Scholar] [CrossRef] - Turner, D.; Lucieer, A.; Wallace, L. Direct georeferencing of ultrahigh-resolution UAV imagery. IEEE Trans. Geosci. Remote Sens.
**2014**, 52, 2738–2745. [Google Scholar] [CrossRef] - Czapski, P.; Kacprzak, M.; Kotlarz, J.; Mrowiec, K.; Kubiak, K.; Tkaczyk, M. Preliminary analysis of the forest health state based on multispectral images acquired by Unmanned Aerial Vehicle. Folia For. Pol.
**2015**, 57, 138–144. [Google Scholar] [CrossRef][Green Version] - Scoppola, A.; Caporali, C. Mesophilous woods with Fagus sylvatica L. of Northern Latium (Tyrrenian Central Italy): Synecology and syntaxonomy. Plant Biosyst.
**1998**, 132, 151–168. [Google Scholar] [CrossRef] - Winter, S.; Möller, G.C. Microhabitats in lowland beech forests as monitoring tool for nature conservation. For. Ecol. Manag.
**2008**, 255, 1251–1261. [Google Scholar] [CrossRef] - Assmann, E. The Principles of Forest Yield Study; Pergamon Press: Oxford, UK, 1970. [Google Scholar]
- Shannon, C. A mathematical theory of communication. Bell Syst. Tech. J.
**1948**, 27, 379–423. [Google Scholar] [CrossRef] - Pielou, E.C. Ecological Diversity; Wiley: New York, NY, USA, 1975. [Google Scholar]
- Clifford, H.T.; Stephenson, W. An Introduction to Numerical Classification; Academic Press: New York, NY, USA, 1975. [Google Scholar]
- Agisoft PhotoScan User Manual: Professional Edition, Version 1.3. 2017. Available online: http://www.agisoft.com/pdf/photoscan-pro_1_3_en.pdf (accessed on 11 April 2017).
- Kachamba, D.; Ørka, O.H.; Gobakken, T.; Eid, T.; Mwase, W. Biomass Estimation Using 3D Data from Unmanned Aerial Vehicle Imagery in a Tropical Woodland. Remote Sens.
**2016**, 8, 968. [Google Scholar] [CrossRef] - Lisein, J.; Pierrot-Deseilligny, M.; Bonnet, S.; Lejeune, P. A photogrammetric workflow for the creation of a forest canopy height model from small unmanned aerial system imagery. Forests
**2013**, 4, 922–944. [Google Scholar] [CrossRef] - Puliti, S.; Gobakken, T.; Ørka, H.O.; Næsset, E. Assessing 3D point clouds from aerial photographs for species-specific forest inventories. Scand. J. For. Res.
**2017**, 32, 68–79. [Google Scholar] [CrossRef] - Wallace, L.; Lucieer, A.; Malenovsky, Z.; Turner, D.; Vopenka, P. Assessment of forest structure using two UAV techniques: A comparison of airborne laser scanning and structure from motion (SfM) point clouds. Forests
**2016**, 7, 1–16. [Google Scholar] [CrossRef] - Dezso, B.; Fekete, I.; Gera, D.; Giachetta, R.; László, I.; Benczúr, A.; Dezs, B.; Fekete, I.; Gera, D.; Giachetta, R.; et al. Object-based image analysis in remote sensing applications using various segmentation techniques. Ann. Univ. Sci. Budapest. Sect. Comp.
**2012**, 37, 103–120. [Google Scholar] - Trimble Germany GmbH. Trimble Segmentation Algorithms: 9.01. Trimble Germany GmbH: Munich, Germany, 2014. Available online: https://www.scribd.com/document/332173213/Reference-Book (accessed on 31 August 2018).
- Koukoulas, S.; Blackburn, G.A. Quantifying the spatial properties of forest canopy gaps using LiDAR imagery and GIS. Int. J. Remote Sens.
**2004**, 25, 3049–3072. [Google Scholar] [CrossRef] - Moser, D.; Zechmeister, H.G.; Plutzar, C.; Sauberer, N.; Wrbka, T.; Grabherr, G. Landscape patch shape complexity as an effective measure for plant species richness in rural landscapes. Landsc. Ecol.
**2002**, 17, 657–669. [Google Scholar] [CrossRef] - Eysenrode, D.S.; Bogaert, J.; Van Hecke, P.; Impens, I. Influence of tree-fall orientation on canopy gap shape in an Ecuadorian rain forest. J. Trop. Ecol.
**1998**, 14, 865–869. [Google Scholar] [CrossRef] - Saura, S.; Carballal, P. Discrimination of native and exotic forest patterns through shape irregularity indices: An analysis in the landscapes of Galicia, Spain. Landsc. Ecol.
**2004**, 19, 647–662. [Google Scholar] [CrossRef] - Busing, R.T. Canopy cover and tree regeneration in old-growth cove forests of the Appalachian Mountains. Vegetatio
**1994**, 115, 19–27. [Google Scholar] - Main-Knorn, M.; Moisen, G.G.; Healey, S.P.; Keeton, W.S.; Freeman, E.A.; Hostert, P. Evaluating the remote sensing and inventory-based estimation of biomass in the western carpathians. Remote Sens.
**2011**, 3, 1427–1446. [Google Scholar] [CrossRef] - Legendre, P.; Legendre, L. Development in Environmental Modelling: Numerical Ecology; Elsevier: Amsterdam, The Netherlands, 1983. [Google Scholar]
- James, G.; Witen, D.; Hastie, T.; Tibshirani, R. An Introduction to Statistical Learning with Applications in R; Castilla, G., Fienberg, S., Olkin, I., Eds.; Springer: New York, NY, USA, 2013; ISBN 978-1-4614-7138-7. [Google Scholar]
- Hauke, J.; Kossowski, T. Comparison of values of Pearson’s and Spearman’s correlation coefficients on the same sets of data. Quaest. Geogr.
**2011**, 30, 87–93. [Google Scholar] [CrossRef] - Møller, A.P.; Jennions, M.D. How much variance can be explained by ecologists and evolutionary biologists? Oecologia
**2002**, 132, 492–500. [Google Scholar] [CrossRef] [PubMed][Green Version] - Mura, M.; McRoberts, R.E.; Chirici, G.; Marchetti, M. Statistical inference for forest structural diversity indices using airborne laser scanning data and the k-Nearest Neighbors technique. Remote Sens. Environ.
**2016**, 186, 678–686. [Google Scholar] [CrossRef] - Lyons, M.B.; Keith, D.A.; Phinn, S.R.; Mason, T.J.; Elith, J. A comparison of resampling methods for remote sensing classification and accuracy assessment. Remote Sens. Environ.
**2018**, 208, 145–153. [Google Scholar] [CrossRef] - McRoberts, R.E.; Magnussen, S.; Tomppo, E.O.; Chirici, G. Parametric, bootstrap, and jackknife variance estimators for the k-Nearest Neighbors technique with illustrations using forest inventory and satellite image data. Remote Sens. Environ.
**2011**, 115, 3165–3174. [Google Scholar] [CrossRef] - Rao, J.N.K. Jackknife and Bootstrap Methods for Small Area Estimation. 2007. Available online: https://ww2.amstat.org/sections/srms/Proceedings/y2007/Files/JSM2007-000358.pdf (accessed on 11 July 2018).
- Efron, B.; Tibshirani, R. Improvements on cross-validation: The 632 plus bootstrap method. J. Am. Stat. Assoc.
**1997**, 92, 548. [Google Scholar] [CrossRef] - Boyd, D.S.; Hill, R.A.; Hopkinson, C.; Baker, T.R. Landscape-scale forest disturbance regimes in southern Peruvian. Ecol. Appl.
**2013**, 23, 1588–1602. [Google Scholar] [CrossRef] [PubMed] - Senécal, J.F.; Doyon, F.; Messier, C. Tree death not resulting in gap creation: An investigation of canopy dynamics of northern temperate deciduous forests. Remote Sens.
**2018**, 10, 121. [Google Scholar] [CrossRef] - Senécal, J.F.; Doyon, F.; Messier, C. Management implications of varying gap detection height thresholds and other canopy dynamics processes in temperate deciduous forests. For. Ecol. Manag.
**2018**, 410, 84–94. [Google Scholar] [CrossRef] - Vepakomma, U.; St-Onge, B.; Kneeshaw, D. Spatially explicit characterization of boreal forest gap dynamics using multi-temporal lidar data. Remote Sens. Environ.
**2008**, 112, 2326–2340. [Google Scholar] [CrossRef] - Schliemann, S.A.; Bockheim, J.G. Methods for studying treefall gaps: A review. For. Ecol. Manag.
**2011**, 261, 1143–1151. [Google Scholar] [CrossRef] - Hobi, M.L.; Ginzler, C.; Commarmot, B.; Bugmann, H. Gap pattern of the largest primeval beech forest of Europe revealed by remote sensing. Ecosphere
**2015**, 6, 76. [Google Scholar] [CrossRef] - Brokaw, N.V.L. The Definition of Treefall Gap and Its Effect on Measures of Forest Dynamics. Biotropica
**1982**, 14, 158. [Google Scholar] [CrossRef] - White, J.C.; Tompalski, P.; Coops, N.C.; Wulder, M.A. Comparison of airborne laser scanning and digital stereo imagery for characterizing forest canopy gaps in coastal temperate rainforests. Remote Sens. Environ.
**2018**, 208, 1–14. [Google Scholar] [CrossRef] - Kane, V.R.; Gersonde, R.F.; Lutz, J.A.; McGaughey, R.J.; Bakker, J.D.; Franklin, J.F. Patch dynamics and the development of structural and spatial heterogeneity in Pacific Northwest forests. Can. J. For. Res.
**2011**, 41, 2276–2291. [Google Scholar] [CrossRef] - Asner, G.P.; Kellner, J.R.; Kennedy-Bowdoin, T.; Knapp, D.E.; Anderson, C.; Martin, R.E. Forest Canopy Gap Distributions in the Southern Peruvian Amazon. PLoS ONE
**2013**, 8. [Google Scholar] [CrossRef] [PubMed] - Zielewska-Büttner, K.; Adler, P.; Petersen, M.; Braunisch, V. Parameters Influencing Forest Gap Detection Using Canopy Height Models Derived From Stereo Aerial Imagery. Publ. DGPF
**2016**, 25, 405–416. [Google Scholar] - Perroy, R.L.; Sullivan, T.; Stephenson, N. Assessing the impacts of canopy openness and flight parameters on detecting a sub-canopy tropical invasive plant using a small unmanned aerial system. ISPRS J. Photogramm. Remote Sens.
**2017**, 125, 174–183. [Google Scholar] [CrossRef] - Lombard, L.; Ismail, R.; Poona, N. Modelling forest canopy gaps using LiDAR-derived variables. Geocarto Int.
**2017**, 6049, 1–15. [Google Scholar] [CrossRef] - Ma, L.; Zheng, G.; Wang, X.; Li, S.; Lin, Y.; Ju, W. Retrieving forest canopy clumping index using terrestrial laser scanning data. Remote Sens. Environ.
**2018**, 210, 452–472. [Google Scholar] [CrossRef] - Barbati, A.; Chirici, G.; Corona, P.; Montaghi, A.; Travaglini, D. Area-based assessment of forest standing volume by field measurements and airborne laser scanner data. Int. J. Remote Sens.
**2009**, 30, 5177–5194. [Google Scholar] [CrossRef][Green Version] - Popma, J.; Bongers, F. The effect of canopy gaps on growth and morphology of seedlings of rain forest species. Oecologia
**1988**, 75, 625–632. [Google Scholar] [CrossRef] [PubMed][Green Version] - Garbarino, M.; Borgogno Mondino, E.; Lingua, E.; Nagel, T.A.; Dukic, V.; Govedar, Z.; Motta, R. Gap disturbances and regeneration patterns in a Bosnian old-growth forest: a multispectral remote sensing and ground-based approach. Ann. For. Sci.
**2012**, 69, 617–625. [Google Scholar] [CrossRef][Green Version] - Runkle, J.R. Patterns of Disturbance in Some Old-Growth Mesic Forests of Eastern North America. Ecology
**1982**, 63, 1533–1546. [Google Scholar] [CrossRef] - Franklin, J.F.; Spies, T.A.; Pelt, R.V.; Carey, A.B.; Thornburgh, D.A.; Berg, D.R.; Lindenmayer, D.B.; Harmon, M.E.; Keeton, W.S.; Shaw, D.C.; Bible, K.; Chen, J. Disturbances and structural development of natural forest ecosystems with silvicultural implications, using Douglas-fir forests as an example. For. Ecol. Manag.
**2002**, 155, 399–423. [Google Scholar] [CrossRef] - Larrieu, L.; Paillet, Y.; Winter, S.; Bütler, R.; Kraus, D.; Krumm, F.; Lachat, T.; Michel, A.K.; Regnery, B.; Vandekerkhove, K. Tree related microhabitats in temperate and Mediterranean European forests: A hierarchical typology for inventory standardization. Ecol. Indic.
**2018**, 84, 194–207. [Google Scholar] [CrossRef]

**Figure 3.**Example of gap delineation using the Contrast Split algorithm (

**A**) and the size distribution of gaps limited to plots with minimum threshold of 1 m

^{2}(

**B**) and minimum threshold of 2 m

^{2}(

**C**).

**Figure 4.**Example of the correlation matrix of the most correlated and the least correlated predictor variables: (

**A**) Pearson’s Correlation of patch metrics variables associated with the sum and (

**B**) Pearson’s correlation from the patch metrics associated with coefficient of variation.

**Figure 5.**Spatial patterns of understory parameters with R

^{2}greater than 50% in mixed and Quercus forests. The predictor variables were (

**A**) standard deviation of rectangular fit, (

**B**) average of rectangular fit, (

**C**) average of asymmetry, and (

**D**) average of roundness.

**Figure 6.**Spatial representation of living tree parameters with R

^{2}greater than 50% in mixed and Quercus forests. The predictor variables were (

**A**) sum of rectangular fit, (

**B**) standard deviation of density, (

**C**) coefficient of variation of length, (

**D**) standard deviation of asymmetry, and (

**E**) average of radius of the smallest enclosed ellipse.

Indices | Formulae | Range of Variation | Description |
---|---|---|---|

Shannon index (${H}^{\prime}$) | $-{\displaystyle \sum}{p}_{i}\mathrm{ln}\left({p}_{i}\right)$ | [0,ln(S)] | The Shannon index expresses the frequency of the i-th species in a community; its values generally lie between 0 and 3.5; higher values correspond to higher species diversity. Its maximum value (MAX_SHANNON) is given by the natural logarithm of the number of species found in the test area and occurs when all species are equally present. |

Pielou index ($E$) | ${H}^{\prime}/\mathrm{ln}\left(S\right)$ | [0,1] | The Pielou index measures the relative abundance of species groups. The index can take values between 1 (all species are equally abundant) and 0 (there is only one species). |

Pretzsch index (${A}_{p}$) | $-{\displaystyle {\displaystyle \sum}_{i=1}^{S}}{\displaystyle {\displaystyle \sum}_{j=1}^{Z}}{p}_{ji}\mathrm{ln}\left({p}_{ij}\right)$ | [0,ln(SxZ)] | The Pretzsch index summarizes and quantifies species diversity and the vertical distribution of species in a forest. The index is lowest in one-story pure forests, whereas it rises for pure forests with two or more stories. Peak values are reached in mixed woodlands with heterogeneous structures. |

Margalef index ($D$) | $\left(S-1\right)\mathrm{ln}\left(N\right)$ | [0,∞) | It quantifies the presence of a number of species in a community. It also depends on the number of plants found in the sampling area. The index value increases with increasing species diversity. |

Normal Distributed Variables | ||

Variables | ANOVA | TUKEY |

Pielou index (I_PIELOU) | no significant difference | |

Mean total height (MEAN_HTOT) | no significant difference | |

Variables Non Normal | ||

Variables | KRUSKAL-WALLIS | MANN-WHITNEY |

Number of plants (N_PLANTS) | *** | (1 vs. 2) ***; (2 vs. 3) ** |

Number of species (N_SPECIES) | *** | (1 vs. 2) ***; (1 vs. 3) ** |

Shannon index (I_SHANNON) | *** | (1 vs. 2) ***; (1 vs. 3) ** |

Mean DBH (MEAN_DBH) | no significant difference | |

Total basal area (G_TOT) | *** | (1 vs. 2) *** |

Total volume (V_TOT) | ** | (1 vs. 2) ** |

**Table 3.**Results of linear regression with understory data (B = slope of linear regression; SE = standard error, N = sample size). Bold values refer to adjusted R

^{2}higher than 50%.

Quercus Forest | |||||||||||||||||||

N_PLANTS | N_SPECIES | I_SHANNON | I_PIELOU | ||||||||||||||||

N = 13 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||

Linear regr. | Linear regr. | Linear regr | Linear regr | ||||||||||||||||

Intercept | −28.85 | 16.8 | Intercept | 7.21 | 0.71 | 0.000 | Intercept | 1.70 | 0.13 | 0.000 | Intercept | 1.02 | 0.10 | 0.000 | |||||

Mdn_GSCI | 18.61 | 5.24 | 0.49 | 0.005 | Sd_rect.fit | −20.66 | 7.02 | 0.39 | 0.013 | Sd_rect.fit | −4.83 | 1.28 | 0.52 | 0.003 | Sd_Density | −1.10 | 0.41 | 0.34 | 0.021 |

MEAN_DBH | MEAN_HTOT | V_TOT | G_TOT | ||||||||||||||||

N = 13 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||

Linear regr. | Linear regr | Linear regr | Linear regr | ||||||||||||||||

Intercept | −23.27 | 7.38 | 0.009 | Intercept | 9.54 | 1.04 | Intercept | 5.34 | 1.48 | 0.004 | Intercept | 1.05 | 0.35 | 0.012 | |||||

Avg_rect.fit | 48.39 | 11.43 | 0.60 | 0.001 | Mdn_B.Index | −56.61 | 24.15 | 0.27 | 0.039 | Mdn_Asym. | −6.60 | 2.36 | 0.36 | 0.018 | Avg_Asym | −1.36 | 0.59 | 0.26 | 0.041 |

Mixed forest | |||||||||||||||||||

N_PLANTS | N_SPECIES | I_SHANNON | I_PIELOU | ||||||||||||||||

N = 9 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||

Linear regr. | Linear regr | Linear regr | Linear regr | ||||||||||||||||

Intercept | 47.02 | 9.62 | 0.002 | Intercept | 0.86 | 0.93 | Intercept | 0.11 | 0.35 | Intercept | 0.1 | 0.3 | |||||||

Avg_Round. | −23.79 | 6.40 | 0.62 | 0.007 | Mdn_RSE | 8.29 | 2.92 | 0.47 | 0.025 | Mdn_RSE | 2.73 | 1.08 | 0.40 | 0.040 | Avg_RLE | 0.42 | 0.17 | 0.43 | 0.045 |

MEAN_DBH | MEAN_HTOT | G_TOT | V_TOT | ||||||||||||||||

N = 9 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||

Linear regr. | Linear regr | Linear regr | Linear regr. | ||||||||||||||||

Intercept | 9.70 | 0.82 | 0.000 | Intercept | 0.90 | 0.82 | Intercept | −0.02 | 0.05 | Intercept | 3.48 | 1.23 | 0.026 | ||||||

Sum_width | −0.00 | 0.00 | 0.51 | 0.018 | avg_Asym | 10.39 | 1.40 | 0.87 | 0.000 | Sd_Asym. | 0.59 | 0.21 | 0.45 | 0.027 | Avg_Comp | −1.16 | 0.48 | 0.37 | 0.048 |

Fagus forest | |||||||||||||||||||

N_SPEIES | I_SHANNON | I_PIELOU | MEAN_HTOT | ||||||||||||||||

N = 28 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||

Linear regr. | Linear regr | Linear regr | Linear regr | ||||||||||||||||

Intercept | 3.13 | 0.77 | 0.000 | Intercept | 1.01 | 0.29 | 0.002 | Intercept | 0.26 | 0.18 | Intercept | 5.46 | 1.40 | 0.000 | |||||

Cv_Round. | −6.02 | 2.64 | 0.15 | 0.033 | Sd_Round. | −1.85 | 0.72 | 0.21 | 0.018 | Mdn_PFD. | 0.06 | 0.02 | 0.48 | 0.022 | Cv_Length | 11.14 | 4.20 | 0.23 | 0.016 |

Variables with Normal Distribution | ||

Variables | ANOVA | TUKEY |

Number of plants (N_PLANTS) | *** | (1 vs. 2) ***; (2 vs. 3) *** |

Pretzsch index (I_PRETZSCH) | *** | (1 vs. 2) ***; (1 vs. 3) *** |

Total basal area (G_TOT) | no significant difference | |

Total volume (V_TOT) | no significant difference | |

Number of habitat trees (HAB) | * | (1 vs. 2) ** |

Percentage of habitat trees (%_HAB) | ** | (1 vs. 2)* |

Non Normal Distributed Variables | ||

Variables | KRUSKAL-WALLIS | MANN-WHITNEY |

Number of species (N_SPECIES) | *** | (1 vs. 2) ***; (1 vs. 3) *** |

Margalef index (I_MARGALEF) | *** | (1 vs. 2) ***; (1 vs. 3) *** |

Shannon index (I_SHANNON) | *** | (1 vs. 2) ***; (1 vs. 3) *** |

Mean diameter at breast height (MEAN_DBH) | *** | (1 vs. 2) ***; (2 vs. 3) *** |

Mean total height (MEAN_HTOT) | ** | (1 vs. 2)** |

**Table 5.**Results of linear regression of living trees (B = coefficient of linear regression, SE = standard error, N = sample size). Bold values refer to adjusted R

^{2}higher than 50%.

Quercus forest | |||||||||||||||||||

N_SPECIES | I_SHANNON | I_MARGLEF | I _ PRETZSCH | ||||||||||||||||

N = 13 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||

Linear regr. | Linear regr | Linear regr | Linear regr | ||||||||||||||||

Intercept | 8.13 | 0.71 | 0.000 | Intercept | 1.81 | 0.14 | 0.000 | Intercept | 1.94 | 0.18 | 0.000 | Intercept | 2.32 | 0.1 | 0.000 | ||||

Sd_Rect.fit | −21.46 | 7.04 | 0.41 | 0.011 | Cv_Density | −2.61 | 0.95 | 0.35 | 0.019 | Sd_Rect.fit | −6.02 | 1.80 | 0.46 | 0.006 | Sd_Density | −2.68 | 0.45 | 0.74 | 0.000 |

MEAN_DBH | MEAN_HTOT | HAB | %HAB | ||||||||||||||||

N = 13 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||

Linear regr. | Linear regr | Linear regr | Linear regr | ||||||||||||||||

Intercept | 11.30 | 3.42 | 0.007 | Intercept | −15.73 | 11.99 | Intercept | 20.54 | 2.44 | 0.000 | Intercept | 41.87 | 5.45 | 0.000 | |||||

Sum_rect.fit | 1.29 | 0.50 | 0.32 | 0.02 | Mdn_rect.fit | 45.41 | 18.68 | 0.29 | 0.033 | Sum_rect.fit | −3.10 | 0.73 | 0.59 | 0.001 | Sum_RLE | −1.86 | 0.54 | 0.47 | 0.006 |

Mixed forest | |||||||||||||||||||

MEAN_DBH | MEAN_HTOT | HAB | %HAB | ||||||||||||||||

N = 9 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||

Linear regr. | Linear regr | Linear regr | Linear regr | ||||||||||||||||

Intercept | −10.86 | 13.32 | Intercept | 27.74 | 2.70 | 0.000 | Intercept | 14.34 | 1.20 | 0.000 | Intercept | 108.24 | 17.54 | 0.000 | |||||

Avg_Round | 25.92 | 8.86 | 0.49 | 0.022 | Sd_Asym. | −49.06 | 11.72 | 0.67 | 0.004 | Cv_Length | −14.01 | 2.51 | 0.79 | 0.000 | Avg_RSE | −208.99 | 53.19 | 0.64 | 0.005 |

Fagus forest | |||||||||||||||||||

HAB | %HAB | ||||||||||||||||||

N = 28 | B | SE (B) | R^{2} | p-value | B | SE (B) | R^{2} | p-value | |||||||||||

Linear regr. | Linear regr | ||||||||||||||||||

Intercept | 4.00 | 1.21 | 0.003 | Intercept | 61.35 | 7.88 | 0.000 | ||||||||||||

cv_PFD | 11.82 | 5.23 | 0.15 | 0.034 | Avg_RSE | −58.21 | 27.71 | 0.11 | 0.045 |

Forest Types | Parameters | Bootstrap R^{2} | Standard Error (95%) | R^{2} | Bias | |
---|---|---|---|---|---|---|

Understory | mixed | Mean total height | 0.877 | 0.006 | 0.872 | −0.005 |

Number of plants | 0.623 | 0.016 | 0.616 | −0.007 | ||

Mean DBH | 0.507 | 0.022 | 0.515 | 0.008 | ||

Quercus | Mean DBH | 0.577 | 0.017 | 0.600 | 0.024 | |

Shannon index | 0.576 | 0.010 | 0.523 | −0.053 | ||

Living trees | Quercus | Pretzsch index | 0.755 | 0.008 | 0.738 | −0.016 |

Number of habitat trees | 0.554 | 0.018 | 0.586 | 0.032 | ||

mixed | Mean total height | 0.682 | 0.012 | 0.674 | −0.008 | |

Number of habitat trees | 0.757 | 0.014 | 0.790 | 0.032 | ||

Percentage of habitat trees | 0.627 | 0.019 | 0.644 | 0.017 |

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Bagaram, M.B.; Giuliarelli, D.; Chirici, G.; Giannetti, F.; Barbati, A.
UAV Remote Sensing for Biodiversity Monitoring: Are Forest Canopy Gaps Good Covariates? *Remote Sens.* **2018**, *10*, 1397.
https://doi.org/10.3390/rs10091397

**AMA Style**

Bagaram MB, Giuliarelli D, Chirici G, Giannetti F, Barbati A.
UAV Remote Sensing for Biodiversity Monitoring: Are Forest Canopy Gaps Good Covariates? *Remote Sensing*. 2018; 10(9):1397.
https://doi.org/10.3390/rs10091397

**Chicago/Turabian Style**

Bagaram, Martin B., Diego Giuliarelli, Gherardo Chirici, Francesca Giannetti, and Anna Barbati.
2018. "UAV Remote Sensing for Biodiversity Monitoring: Are Forest Canopy Gaps Good Covariates?" *Remote Sensing* 10, no. 9: 1397.
https://doi.org/10.3390/rs10091397