33
of some chemical traits are known, the effects of other, particularly nonchemical,
plant traits on spectra are less well understood, in part due to overlapping spectral
features and challenges associated with accurately describing nonchemical traits
(Ustin and Jacquemoud, Chap. 14). Using the full spectral profile of plants in spectral diversity calculations provides a means to integrate chemical, structural, morphological, and anatomical variation and to acknowledge the many ways plants
differ from one another.
It is certainly more complicated to decipher the biological meaning of spectral
diversity calculated from spectral profiles than from measures of biodiversity that
are based on a specific set of plant traits or spectral bands or indices with known
links to specific traits. However, the variance that is explained by models based on
spectral profiles can be partitioned into known and unknown sources of variation.
This provides a means to assess the relative contribution of traits with known spectral characteristics and traits that are less well understood spectrally or that are of
yet-unrecognized importance. At the canopy level, when spectra are measured from
a distance, the question of what spectra and spectral diversity represent is further
complicated by the influences that plant architecture, soil, and other materials have
on the spectral characteristics of image pixels (Wang et al. 2018; Gholizadeh et al.
2018). Again, the degree to which these characteristics matter for a particular ecosystem needs to be evaluated in the particular context of the study. Some ecosystem
components such as shade, soil, rock, or debris, which influence remotely sensed
spectra, are biologically meaningful because they influence light availability and
microclimate and provide resources for other trophic levels.
The association between plant spectra and traits can be illustrated by plotting
spectral distances against functional distances or dissimilarity, as illustrated using
species from the Cedar Creek biodiversity experiment (Fig. 2.6d). Given that functional differences among species are expected to increase with evolutionary
divergence time (Fig. 2.1b), positive relationships are also expected among spectral
and phylogenetic distances. The observed associations among spectral, functional,
and phylogenetic dissimilarity (Fig. 2.6a, b) allow biodiversity metrics based on any
of these dimensions of biodiversity to explain a similar proportion of the total variability in aboveground productivity (Fig. 2.6c–e), which is known to increase with
the functional diversity of the plant community in this system (Cadotte et al. 2009).
The species in the biodiversity experiment at Cedar Creek are relatively functionally
dissimilar and distantly related, such that spectral, functional, and phylogenetic
diversity also predict species richness (not shown). One advantage of spectral diversity is that the metric can be calculated from remotely sensed image pixels without
depending on information about the distribution and abundance of species in an area,
their functional traits, or phylogenetic relationships (Schweiger et al. 2018). By
extracting a random number of high-resolution image pixels in each plant community, Schweiger et al. (2018) found that remotely sensed spectral diversity explained
the biodiversity effect on aboveground productivity about as well as spectral diversity calculated using leaf-level spectra (Fig. 2.6).
2 Applying Remote Sensing to Biodiversity Science
of some chemical traits are known, the effects of other, particularly nonchemical,
plant traits on spectra are less well understood, in part due to overlapping spectral
features and challenges associated with accurately describing nonchemical traits
(Ustin and Jacquemoud, Chap. 14). Using the full spectral profile of plants in spectral diversity calculations provides a means to integrate chemical, structural, morphological, and anatomical variation and to acknowledge the many ways plants
differ from one another.
It is certainly more complicated to decipher the biological meaning of spectral
diversity calculated from spectral profiles than from measures of biodiversity that
are based on a specific set of plant traits or spectral bands or indices with known
links to specific traits. However, the variance that is explained by models based on
spectral profiles can be partitioned into known and unknown sources of variation.
This provides a means to assess the relative contribution of traits with known spectral characteristics and traits that are less well understood spectrally or that are of
yet-unrecognized importance. At the canopy level, when spectra are measured from
a distance, the question of what spectra and spectral diversity represent is further
complicated by the influences that plant architecture, soil, and other materials have
on the spectral characteristics of image pixels (Wang et al. 2018; Gholizadeh et al.
2018). Again, the degree to which these characteristics matter for a particular ecosystem needs to be evaluated in the particular context of the study. Some ecosystem
components such as shade, soil, rock, or debris, which influence remotely sensed
spectra, are biologically meaningful because they influence light availability and
microclimate and provide resources for other trophic levels.
The association between plant spectra and traits can be illustrated by plotting
spectral distances against functional distances or dissimilarity, as illustrated using
species from the Cedar Creek biodiversity experiment (Fig. 2.6d). Given that functional differences among species are expected to increase with evolutionary
divergence time (Fig. 2.1b), positive relationships are also expected among spectral
and phylogenetic distances. The observed associations among spectral, functional,
and phylogenetic dissimilarity (Fig. 2.6a, b) allow biodiversity metrics based on any
of these dimensions of biodiversity to explain a similar proportion of the total variability in aboveground productivity (Fig. 2.6c–e), which is known to increase with
the functional diversity of the plant community in this system (Cadotte et al. 2009).
The species in the biodiversity experiment at Cedar Creek are relatively functionally
dissimilar and distantly related, such that spectral, functional, and phylogenetic
diversity also predict species richness (not shown). One advantage of spectral diversity is that the metric can be calculated from remotely sensed image pixels without
depending on information about the distribution and abundance of species in an area,
their functional traits, or phylogenetic relationships (Schweiger et al. 2018). By
extracting a random number of high-resolution image pixels in each plant community, Schweiger et al. (2018) found that remotely sensed spectral diversity explained
the biodiversity effect on aboveground productivity about as well as spectral diversity calculated using leaf-level spectra (Fig. 2.6).
2 Applying Remote Sensing to Biodiversity Science