Westward Expansion by Juniperus virginiana of the Eastern United States and Intersection with Western Juniperus Species in a Novel Assemblage

Abstract

Eastern redcedar (Juniperus virginiana L.) is increasing in density in the eastern United States and expanding in range to the west, while western Juniperus species also are increasing and expanding, creating the potential for a novel assemblage. I estimated range expansion and intersection by comparing recent USDA Forest Service Forest Inventory and Analysis surveys (mean year = 2009) to the oldest available surveys (mean year = 1981), with adjustments for sampling changes, and predicted climate envelopes during the following year ranges: 1500–1599, 1800–1849, 1850–1899, 1900–1949, and 1960–1989. During approximately 28 years, eastern redcedar range expanded by about 54 million ha (based on ≥0.5% of total stems ≥12.7 cm in diameter in ecological subsections). Combined range of western species of juniper did not expand. Range intersection of eastern redcedar and western Juniperus species totaled 200,000 km² and increased by 31,600 km² over time, representing a novel assemblage of eastern and western species. Predicted ranges during the other time intervals were 94% to 98% of predicted area during 1960–1989, suggesting major climate conditions have been suitable for centuries. The southern western Juniperus species and Rocky Mountain juniper (Juniperus scopulorum Sarg.) have the greatest potential for intersection with eastern redcedar, whereas eastern redcedar may have concluded westward expansion.

1. Introduction

During the past century, native tree and shrub species have changed in density within existing ranges and also changed in distribution within both forested ecosystems and non-forested ecosystems [1,2,3]. Spread of native woody vegetation is described as encroachment, densification, thicketization, and woodification, evoking alteration in the characteristics and condition of ecosystems that threaten native biodiversity [4]. Changes in density within existing or novel ranges occur because of changes in land use, natural disturbances, and climate change among other factors.

Eastern redcedar (Juniperus virginiana L.) is a native tree species of the eastern United States that is spreading, in part through deliberate introduction. Eastern redcedar has increased in density within eastern forests, doubling in percentage of all trees in the central eastern and northern southeastern regions [5]. More notably, eastern redcedar is expanding westward into the Great Plains grasslands of the central U.S. Eastern redcedar received an assist in migration into the central grasslands, as this species was, and continues to be, planted for windbreaks and fence rows, and additionally is dispersed widely by birds and small mammals [6,7].

Most area of the central grasslands is likely outside of the native distribution of eastern redcedar, but eastern redcedar has overcome historical boundaries to survive and reproduce in grasslands. Expansion of numerous tree species into the westernmost eastern grasslands was noted in historical accounts during the first half of the 1800s [8,9])and tree expansion into central grasslands was noted by 1860 [10,11]. However, eastern redcedar generally was delayed in expanding into central grasslands compared to other tree species or else largely remained unobserved, due to initial relative rarity at ecotones (e.g., almost no observations in historical tree surveys of Missouri and Illinois circa 1800 to 1860) [12,13]. Beilmann and Brenner [14] dated early eastern redcedar expansion by 1874 in Missouri, but Gleason [11] did not include eastern redcedar with numerous tree species that were expanding into grasslands. Bruner [15] mentioned eastern redcedar as being restricted to steep slopes in Oklahoma, but expansion occurred by 1950 [16].

The most common western Juniperus species are Utah (J. osteosperma (Torr.) Little), oneseed (J. monosperma (Engelm.) Sarg.), Ashe (J. ashei J. Buchholz), and Rocky Mountain (J. scopulorum Sarg.) junipers. Western Juniperus species can be associated with pinyon pines (e.g., P. edulis Engelm., P. monophyla Torr. and Frém.); however, this association appears to be enforced by historical disturbance regimes and new genus- and species-specific dynamics are emerging. Indeed, juniper species probably are benefiting from drought declines of pinyon pines [17,18,19]. Range expansion and increased density of pinyon and juniper have occurred across the Great Basin and Colorado Plateau during the past 200 years with a rapid increase in the late 1800s and peaking during the early 1900s [20]. Ashe juniper expansion was noted in Texas by 1917 [21]. Similarly to eastern redcedar, western juniper species have expanded from dry, rocky, and shallow soils, sometimes on steep slopes along stream channels, to productive sites with deep soils [22,23,24,25,26,27]. Eastward expansion specifically has not been evident for the western species; rather, expansion generally has been multidirectional, into riparian and aspen ecosystems [20], upslope in elevation into ponderosa pine forests (Pinus ponderosa Douglas ex P. Lawson and C. Lawson) [28], and downslope into shrub- or grass-dominated ecosystems [24]. Increases have been attributed to climate change, overgrazing due to livestock, and reduced frequency of surface fire in locations with enough continuous vegetation to support surface fires [20]. Regarding past climate change, woodlands typically increased when climate was wetter and declined when climate became warmer and drier [20].

Because eastern redcedar has been expanding westward while western Juniperus species also have been expanding, novel assemblages of eastern and western Juniperus species may be developing. Paleoecological evidence has demonstrated that novel tree species assemblages emerged during range shifts in response to climate change after the glacial maximum 20 thousand years ago, but recent novel tree assemblages of have not been documented in the United States, to my knowledge [29]. My objectives were to quantify (1) recent ranges of Juniperus by species, (2) expansion, or increases in range, of eastern redcedar and all combined western Juniperus species, (3) intersection of eastern redcedar and all combined western Juniperus species ranges in the central United States for documentation of a modern novel assemblage, and (4) potential range expansion for intersection. To establish potential ranges and range expansion, I predicted climate envelopes during the years 1500s (1500 to 1599), 1800–1849, 1850–1899, 1900–1949, and near present climate (1960–1989).

2. Materials and Methods

I used USDA Forest Service Forest Inventory and Analysis surveys (FIA) [30], which contain thousands of long-term plots located about every 2000–2500 ha across the country. From all states of the continental United States, I selected the most recently completed cycles in FIA surveys, ranging from years 2000 to 2013 (mean year = 2009) and the oldest FIA surveys, ranging from years 1966 to 2001 (mean year = 1981). Survey years uniquely vary by U.S. state, but surveys were initiated earlier in the eastern states. I grouped all plots within ecological subsections, which are the smallest ecological unit provided in FIA surveys (mean area = 642,000 ha, SD = 777,000) [31] and represent spatial area that share common ecological characteristics. I determined recent ranges where presence was ≥0.5 percent composition (i.e., percent of all stems) in ecological subsections and also quantified ranges at ≥10% composition and ≥30% composition.

Caveats include that a standardized national survey design was phased in beginning in 1999 to improve precision of inventory parameter estimates and variance; however, the national sampling design preserved consistency with regional surveys to allow continuity of trend estimation because the pre-standardized survey design provides valuable and robust information [32,33,34]. Surveys switched from periodic inventories in which all plots were visited during an undefined time interval to inventory cycles of typically 5 years (eastern states) or 10 years (western states) in which each annual survey is a representative sample and about half of plots were re-located, generally close to the original plot location [33]. To account for sampling changes over time, I used adjustment factors [35]; additionally, I summarized changes by ecological subsections [31], a larger scale than plots, and limited the study to established trees ≥12.7 cm in diameter at 1.37 m height that are surveyed completely in each plot.

To make comparisons among old and recent survey data, adjustments to the old survey data were required to account for differing sampling intensity and also populate non-sampled areas [35]. Old surveys generally sampled timberland only (i.e., a specified productivity level and available for timber harvesting) across all ownerships. To adjust the old survey data sets for non-sampled areas, new survey plot data were added to the old survey data sets for non-sampled areas; this produced a data set that sampled changes between the old and new surveys for only those lands sampled in both surveys, with no change in the non-sampled areas. A plot weighting factor of survey intensity accounted for differences caused by various sampling intensities between states and over time; the survey intensity was proportional to area from which the sample plots were randomly selected divided by the number of plots in the sample survey. Density by species was computed following procedures described in Bechtold and Patterson [32] that account for differences in plot design (variable compared to fixed area) and size used in each survey. The adjustments are meant to provide better comparisons for change detection; nonetheless, adjustments do alter values, resulting in divergence from recent inventories.

For live trees with diameters ≥12.7 cm at 1.37 m height, I determined change in range area at ≥0.5%, ≥10%, and ≥30% of all stems by ecological subsection and composition of Juniperus species and range intersection of eastern redcedar and western Juniperus species. I provided information about combined western Juniperus species to help reduce variation due to survey standardization. For subsections where either eastern redcedar or all combined western Juniperus species increased by ≥5% composition, I determined species that decreased by ≥5% composition. I retained species that that decreased in at least four subsections. I also excluded hickory and ash species, due to unidentified species of these genera in older surveys.

I extracted 11 climate variables [36,37] by ecological subsection to model potential presence ranges (≥0.5 percent composition in ecological subsections) during years 1960 to 1989, 1900–1949, 1850–1899, 1800–1849, and 1500–1599. Six temperature variables included growing degree days above 0 °C and 5 °C and mean minimum and maximum temperature annually and for the coldest and warmest months. Total annual precipitation, the coefficient of variation for precipitation, percent winter (December and January) precipitation, July vapor pressure, and the aridity index of total annual precipitation to potential evapotranspiration indicated changing dryness. In addition, I used soil moisture [38] and topographic roughness [39] variables, which represent only current conditions, but provide additional information.

For modeling presence ranges (≥0.5 percent composition in ecological subsections) under current climate, I applied random forests and extreme gradient boosting classifiers in the caret package [40,41] and trained the model with 10-fold cross-validation. Validation occurred on withheld testing data, 25% presence range samples, to determine how well the classifier assigned classes using explanatory variables. For this modeling, prevalence, or the number of present samples (≥0.5 percent composition in ecological subsections) to pseudoabsent samples (<0.5 percent composition in ecological subsections) was equal. Then, I predicted potential presence (≥0.5 percent composition) ranges for the different year intervals. I combined some species to increase sample size of ecological subsections with species presence to ≥74. I grouped the more southern species of oneseed, Ashe, alligator (Juniperus deppeana Steud.), Pinchot (Juniperus pinchotii Sudw.), and redberry (Juniperus coahuilensis (Martiñez) Gaussen ex R.P. Adams) junipers and the more western species of western juniper (Juniperus occidentalis Hook.) and the relatively widespread but uncommon California juniper (Juniperus californica Carrière). I also modeled two pinyon pine species (P. edulis, P. monophylla), which are somewhat associated with juniper.

3. Results

To summarize recent status in forest inventories, the most abundant western Juniperus species generally maintained spatially differentiated ranges, where western Juniperus species occurred at ≥10% of composition, or percent of all stems in ecological subsections (Figure 1). Utah juniper was the most common western Juniperus species according to FIA surveys, representing 42% of western Juniperus species percent composition and 2% of all combined trees, even though Rocky Mountain juniper covered a presence range (≥0.5% of all trees by ecological subsection) of 110 million ha compared to 95 million ha for Utah juniper (

Table 1. Summary statistics of Juniperus species for recent surveys, not limited by areas with older surveys. Percent values are percent of total stems.

Species	Scientific Name	% of Western Junipers	Range at ≥0.5% of All Trees (km2)	Range at ≥10% All Trees (km2)	Range at ≥30% All Trees (km2)
Utah juniper	J. osteosperma	42.2	946,571	764,094	499,995
Oneseed juniper	J. monosperma	17.5	757,882	472,246	194,949
Ashe juniper	J. ashei	11.8	293,600	192,553	81,157
Rocky Mountain juniper	J. scopulorum	10.9	1,103,090	399,009	131,736
Western juniper	J. occidentalis	6.1	289,313	136,213	97,493
Alligator juniper	J. deppeana	5.7	277,667	77,883	0
Pinchot juniper	J. pinchotii	2.9	385,019	95,775	37,071
Redberry juniper	J. coahuilensis	2.3	477,101	112,413	0
Eastern redcedar	J. virginiana	N/A	1,978,270	339,504	86,258

). Utah juniper was ≥10% of composition in about twice the area as Rocky Mountain juniper and Utah juniper was ≥30% of composition in almost four times the area as Rocky Mountain juniper. Eastern redcedar was the second most common Juniperus species, at 1% of all trees, and the most widespread Juniperus with a range of almost 200 million ha, but because most of eastern redcedar’s range was within forested ecosystems of the eastern United States, eastern redcedar did not occur at ≥10% of composition throughout about 85% of its range.

Comparing compositional change between oldest and most recent surveys, eastern redcedar doubled in number of trees. Survey adjustments equalized survey intensity and thereby improve trend detection but concurrently resulted in some divergence from recent inventory information. Utah juniper, the most abundant western juniper slightly decreased, from 1.8% to 1.7% of all recorded trees. Indeed, all western juniper species decreased, except for Rocky Mountain juniper, which increased slightly (from 0.61% to 0.66% of all recorded trees). For calculation of expansion, which also required survey adjustments, where eastern redcedar was present (≥0.5% of composition), eastern redcedar increased 125% in area, from 207.6 to 261.5 million ha. Where eastern redcedar occurred at 10% composition, eastern redcedar more than doubled in area (215%), from 48.5 million ha to 104.1 million ha. Where eastern redcedar occurred at 30% composition, eastern redcedar almost tripled in area (280%), from about 6.9 to 19.5 million ha. The areas where all combined western Juniperus species occurred at the three different percent composition remained stable (recent values were 0.98 to 1.05 of older areas). However, this represented only 19 years of change due to later surveys and areas without older surveys were filled in with current values. By species, results were similar. Where each of the western Juniperus species was present (≥0.5% of composition), maximum expansion was 103% of the older range; where each of the western Juniperus species occurred at 10% composition, the maximum expansion was 100%; where each of the western Juniperus species occurred at ≥30% composition, the maximum expansion was 101%, except for Rocky Mountain juniper that expanded 127%.

Intersection between eastern redcedar and all combined western Juniperus species (≥0.5% composition) increased in a net total of two ecological subsections between oldest and most recent surveys (Figure 2). Intersection increased in four ecological subsections covering 48,365 km² but decreased in two subsections covering 16,765 km² and remained stable in eleven subsections covering 176,845 km². In three subsections, only eastern redcedar was recorded previously; that is, Rocky Mountain juniper and Ashe juniper increased in the two or one subsections, respectively.

In areas where eastern redcedar increased over time, decreasing species primarily were pine and oak species (

Table 2. Species that decreased by ≥5% composition where Juniperus species have increased by ≥5% composition.

Species	Scientific Name	# Subsections	Change % Composition	Change % Juniperus Composition
Eastern redcedar
Shortleaf pine	Pinus echinata	13	−11.6	9.1
Boxelder	Acer negundo	12	−17.4	20.3
Eastern cottonwood	Populus deltoides	9	−24.0	24.2
Blackjack oak	Quercus marilandica	7	−6.9	10.4
Post oak	Quercus stellata	5	−7.8	10.3
Virginia pine	Pinus virginiana	5	−11.1	6.7
White oak	Quercus alba	5	−9.1	9.3
Ponderosa pine	Pinus ponderosa	4	−22.5	15.9
Sugar maple	Acer saccharum	4	−8.9	8.8
Western Juniperus
Ponderosa pine	Pinus ponderosa	12	−9.6	11.9
Quaking aspen	Populus tremuloides	9	−19.3	26.7
Two-needle pinyon	Pinus edulis	8	−9.0	9.50
Lodgepole pine	Pinus contorta	6	−23.3	25.4
Subalpine fir	Abies lasiocarpa	6	−15.9	27.4
Limber pine	Pinus flexilis	5	−13.1	15.6
Singleleaf pinyon	Pinus monophylla	4	−7.4	8.0

). In addition, decreasing species included boxelder (Acer negundo L.) and eastern cottonwood (Populus deltoides Bartram ex Marsh.) that are located along riparian channels. In areas where western Juniperus increased, primarily pine species decreased, particularly ponderosa pine and two-needle pinyon pine.

For potential ranges, modeled ranges were accurate, with a mean AUC (area under the curve; area under the receiver operating characteristic curve, which plots the true positive rate compared to false positive rate at all classification thresholds) value of 0.98, ranging from 0.96 to 1.0. The climate envelopes during 1500–1599 were relatively similar in area to near present climate envelopes, with a mean value of 94% of the near present area, ranging from 82% to 102%. Similarly, area during 1800–1849, 1850–1899, and 1900–1950 overall were slightly less than near present climate envelopes, with mean areas of 94% to 98% by time interval of near present climate envelopes, ranging from 87% to 107%. When compared to current realized ranges, the combined western juniper and California juniper had the greatest potential for expansion, with a 4.4-fold increase when comparing the near present climate envelope (years 1960–1989) to realized range (Figure 3A). The Rocky Mountain juniper climate envelope was 2.3-fold greater than the realized range (Figure 3B), whereas the climate envelopes of other species were almost double (1.8- to 1.9-fold) the realized ranges (Figure 3). However, potential for intersection was greater between eastern redcedar and the southern western junipers, albeit consisting of multiple species, at 656,845 km² than between eastern redcedar and Rocky Mountain juniper at 254,860 km² (Figure 4).

4. Discussion

Eastern redcedar has increased in relative density and expanded in range, one of the great range expansions in the United States during the past few decades, according to adjusted FIA surveys. Additionally, eastern redcedar has increased in relative density within its range, by two- to three-fold where it occurred at ≥10% composition (or percent of all stems) and ≥30% composition. Potential bias in comparing FIA surveys over time arises from survey standardization to a national standard, and while adjustments to equalize survey intensity may correct change detection, exact values are altered. Nonetheless, these results concur with studies reporting eastern redcedar expansion and in-filling that were documented as early as 1875 [14,42,43].

Although older surveys in the western United States are limited in time and space, recent recorded trends in western Juniperus species suggest stability, or even minor decreases, with Rocky Mountain juniper as the one species that slightly has increased and expanded overall. The FIA surveys offer a long-term record to detect species trends in composition and ranges for the entire United States, based on thousands of plots covering millions of ha, but some trends may vary in time and space or may be distorted by survey changes to a national standard. For western Juniperus, the time interval of measurement was only about 19 years, and coverage in older surveys was incomplete in some portions of the western United States, which resulted in conservative estimates. It may be that some western Juniperus species expanded but incomplete surveys, particularly in Washington, Oregon, southern California, and western Texas, prevented detection. Alternatively, it may be that reduced rates of expansion and spatiotemporal variability [20,44] resulted in the relative stability indicated by surveys. In addition, spatial scales of expansion may be small relative to expansion by more abundant species during the past few decades throughout the country. For example, western juniper (Juniperus occidentalis) has expanded by a few million ha over 100 years, which is important to invaded shrub steppe communities in Oregon and California [45]. Indeed, western juniper has the greatest potential to expand into appropriate climate conditions that match current conditions, as demonstrated by the area of the climate envelope that was greater by a factor of 4.4 than the area of the realized range.

In terms of intersection, eastern redcedar and western Juniperus species overlapped in ecological subsections representing about 200,000 km² during the older surveys and increased slightly by 31,600 km² over time. Despite greater expansion by eastern redcedar than western Juniperus species, it appears that western Juniperus species have expanded into the central United States in areas that intersect the range of eastern redcedar. Eastern redcedar expansion, particularly as planted windbreaks, may be greater in areas unoccupied by another juniper species than in occupied areas. Based on climate suitability, the southern western Juniperus species followed by Rocky Mountain juniper have the greatest potential for intersection with eastern redcedar.

Intersection of eastern and western junipers may represent the first new, viable assemblage during recent millennia in the United States. Although native tree species have changed in relative composition and become less spatially differentiated within regions during the past century or two due to land use change, including fire exclusion, native tree species shared at least limited spatial overlap within regions [46]. However, association of an eastern tree species with western species is novel, given that fires in the Great Plains grasslands historically prevented tree expansion and intermixing during the past 10,000 or more years [47,48]. Additionally, junipers are widespread and abundant, resulting in multiple subsections with overlap, and junipers have similar traits that may allow coexistence into an established long-term association, as indicated by increasing spatial intersection over time. In contrast, ponderosa pine, another western species, decreased where eastern juniper expanded (Table 2), and likely this trend indicates a non-viable, transient association during the replacement of ponderosa pine by eastern juniper. Novel assemblages documented by paleoecological evidence captured common associations that were stable during hundreds or thousands of years, rather than brief transitions [29]. Novel assemblages are expected to form under current climate change. Nonetheless, novel assemblages also can develop due to land use change rather than climate change.

If the climate envelopes are approximate depictions of climate limitations, eastern redcedar mostly has finalized westward expansion, with some scope for southwestern expansion across Texas (Figure 4). Without a fire regime, and particularly with aided migration by windbreak plantings and birds, climate may not limit western expansion of eastern redcedar. Potential eastern redcedar expansion throughout the eastern United States is possible, but competition from other expanding native tree species may prevent long-term establishment.

Junipers have become more abundant compared to other tree species, despite having traits of early-successional colonizers. Juniper densification and expansion reduced already fragmented or deteriorated ecosystems of open grassland and shrubland ecosystems and open pine and oak forests of savannas and woodlands that coexist with herbaceous vegetation, as demonstrated by reduction of fire-dependent pines and oaks (Table 1) [43,49,50]. Junipers replace herbaceous vegetation, which is important for a variety of animals associated with open ecosystems [23,25,45,51,52]. Species such as eastern cottonwood and boxelder that are present in riparian networks decreased where junipers increased. Broadleaf species are relatively important to wildlife in gymnosperm-dominated forests of the western United States [53].

Recent expansion of fire-sensitive species, such as eastern redcedar, has been attributed to multiple mechanisms. The very similar areas of climate envelope during the time intervals indicate that major climate indices, such as precipitation and aridity, may not have been limiting factors for expansion. Otherwise, eastern redcedar already would have expanded, rather than starting during the late 1800s [12]. In the eastern and central U.S., fire exclusion may be the main factor driving tree expansion and densification [2,6,7,42,43,46,48,49,54,55,56]. Fire exclusion, as opposed to fire suppression, incorporates a land use change component through reduction in herbaceous fine fuels by agriculture and intensive herbivory or high tree densities and reduction in fire spread by landscape fragmentation, particularly linear barriers of roads and trails. Similarly, fire reduction and intensive grazing have been implicated in Ashe juniper increases in Texas [23,25]. Moreover, species that decreased where eastern redcedar and western Juniperus increased generally were fire-dependent pines and oaks (e.g., shortleaf pine [Pinus echinata Mill.] and oaks in the eastern U.S., ponderosa pine in the western U.S.) or else post-fire-disturbance species, with traits such as prolific seeding or sprouting (e.g., quaking aspen, Populus tremuloides Michx.; lodgepole pine, Pinus contorta Douglas ex Loudon; Virginia pine, Pinus virginiana Mill.) that are competitive after overstory disturbance (Table 2).

Western Juniperus, particularly the southern species, may be expanding primarily into locations where fire exclusion has occurred, but fire exclusion with land use change may not be sufficient to explain expansion of western Juniperus into arid locations where surface fuels were not continuous enough to support frequent fire regimes [20]. Although tree expansion is a long process, weather oscillations may be influential, including severe droughts and pluvials [18,19,56]. However, the pluvial of the years 1905 to 1917 occurred after rapid juniper expansion was noted during the late 1800s and the pluvial was not unprecedented as an analogous pluvial occurred during the years 1825 to 1850 [57]. Major droughts also have occurred regularly, such as during years 1855 to 1865, but also with long gaps during which juniper expansion did not occur prior to the late 1800s, while more minor droughts occurred during 1870–1883 and 1897–1904 with juniper expansion [57]. Hydraulic characteristics of junipers may help confer greater drought tolerance relative to pinyon pines, which were decreasing species where western Juniperus increased. Junipers have been documented to continue leaf gas exchange and carbon assimilation with decreasing soil water potential during drought compared to other co-occurring tree species that experience declines in leaf gas exchange with declining soil moisture [18,19].

5. Conclusions

Trends revealed by large-scale, long-term FIA surveys, despite limitations, can provide the current status of tree species and inform future studies. Eastern redcedar had one of the greatest recent range expansions in the United States by increasing presence (≥0.5% composition) in about 54 million ha, along with increased density within the range. In contrast, western Juniperus species appeared relatively stable, which may be due to survey gaps or may represent stability during the past two decades. Range intersection between eastern redcedar and Rocky Mountain juniper or southern western Juniperus species totaled almost 200,000 km² and increased by 31,600 km² over time, representing one of the first new assemblages of native tree species for millennia in the United States. Based on climate suitability, the southern western Juniperus species and Rocky Mountain juniper have the greatest potentials for intersection with eastern redcedar; western juniper has the greatest potential for expansion, whereas eastern redcedar may have finalized westward expansion, excepting expansion across Texas. Predicted areas of climate envelopes during a range of time intervals were 94% to 98% of predicted area during the recent past, indicating that the major climate conditions have been suitable for juniper species and realized ranges may be due to other constraints. Juniper increases were at the expense of other ecosystems, such as shrublands, grasslands, riparian forests, and open pine and oak forests, with consequent impacts on plant and wildlife species that flourish in open or unique ecosystems.