The overarching message from this first remeasurement of the Pando aspen clone is one of limited stem recruitment resulting from persistent browsing (Figure 2). Mule deer browsing is the main factor affecting recruitment success, with domestic cattle also have significant impacts (Table 2). This conclusion is supported by significant ordinal relationships in axis 1 (Figure 2) between herbivore presence, browse level, surviving recruitment, and protection status (i.e., “fence”). A promising finding here is that many aspen stems within the 2013 fence have moved into the recruitment class, though recruitment is lagging in the 2014 Fence (Table 1, Figure 3d), likely related to the permeable condition of the 2014 enclosure prior to 2019 fortification. Overall, more than 80% of Pando’s 43 ha area has inadequate recruitment. A secondary finding here was that regeneration counts (ha-1) define a distinct gradient within the study area, explaining ~25% variance. The relationship between regeneration (axis 2) and recruitment (axis 1) is explored further below.
The present assessment at Pando indicates protection regimes (treatments) are driving this genetically uniform forest toward divergent ecological pathways based on an array of tree and herbivore measures (Table 3). This divergence was supported by previous work at Pando where dissimilar vegetation assemblages track the same protection regimes . Evidently past management—limiting or allowing herbivores differentially—is driving observed understorey and overstorey departures from a relatively consistent forest overstorey. Fencing to limit herbivory is a logical first-step after decades of failed recruitment, although barriers appear to be having unintended consequences.
This study suggests that regeneration and recruitment success are not equivalent, and that recruitment is a more meaningful indicator of long-term resilience [29,30]. The NMS ordination (Figure 2) indicates deviating gradients of recruitment, browse, and ungulate presence (axis 1) from regeneration (axis 2) where we would expect covariance in these factors. In self-replacing aspen, a predictable rate of attrition occurs in juvenile stems based on resource availability . However, aspen affected by chronic herbivory in the Rocky Mountains commonly have plentiful regeneration (< 2 m ht.) with few suckers attaining recruitment (>2 m ht.) size . This pattern seems to hold here.
In this study, direct and indirect human influences complicated initial result interpretations (Table 1). For instance, increased regeneration in unfenced areas of Pando defied expectations based on previous studies [13,14]. Further investigation found most of that unprotected regeneration was in the smallest height class (< 0.5 m), suggesting newly emerged unbrowsed suckers disproportionately influenced counts (S1). Successful regeneration-to-recruitment development has been absent outside fenced areas at Pando for decades . Thus, I was curious to investigate whether taller stems were surviving or whether this was simply a case of new regeneration that had gone unbrowsed in first-year emergence. After removing the smallest height class the No Fence and 2014 Fence treatments displayed significant differences in regeneration success and reduced browse (Figure 3b, 3e-f), contrasting with baseline results  and suggesting the 2019 fence reinforcement (2014 Fence) is showing positive effects (Figure 3b, 3f).
If regeneration success in protected areas progresses at Pando, we would expect signs of burgeoning recruitment. While long-term aspen recruitment may be episodic based on disturbance, climate, or other human factors [32,33], the decadal dearth of recruitment at Pando suggests recent management is driving successes and failures . A deeper examination was required. Viewing only recruitment, the older 1992 enclosure (Figure 1) seemed to have an undue influence on initial findings. This parcel, within the 2014 Fence, contains six plots where many recruitment stems have grown into the “mature tree” class, erroneously implying a loss of the recruitment here (Table 1). Though there were differences in recruitment overall, most of that consists of the large gap between fenced and unfenced groups (Figure 3c). Rerunning the analysis with the subgroup of six 1992 clearfell plots removed presents a different outcome; No Fence and 2014 Fence areas contain almost no recent recruitment (Figure 3d). Twenty-nine-year-old recruitment confounds our understanding of restoration success in the full Pando data set. When accounting for that in the context of regeneration shortfalls (Figure 3b), a stubborn trend of marginal stand replacement at Pando persists, particularly in the sizable unfenced zone. This conclusion leaves us with nagging questions regarding a path forward and whether recovery solely dependent on fencing is appropriate.
Aspen forests worldwide support outsized biodiversity [6,7,34], but at Pando overabundant browsers are usurping resilience with expected spiraling (or diverging) diversity outcomes. Results here bolster evidence of a faltering recovery when protection from herbivores is absent (Figure 3c-d). The dominant ecological gradient among 19 system indicators was an inverse relationship between recruitment success and browse pressure (Figure 2, Table 2). A secondary gradient of regeneration fecundity suggests the importance of this variable, though regeneration alone is a poor predictor of system resilience . Moreover, Pando is affectively “breaking up,” as evidence here shows distinct forests within the clone by protection status (Table 3) and previously by plant communities .
Excessive herbivory is present on a much larger landscape surrounding Pando and, in fact, is a major pressure on aspen communities regionally . Caretakers at Pando will need to address herbivory to ensure long-term system sustenance. Primarily mule deer, but also domestic cattle, are responsible for insufficient recruitment. This conclusion is reached by mule deer’s stronger linkages to browse (Figure 2). Additionally, there are unfenced areas of Pando off-limits to cattle during their brief annual occupancy (e.g., campground, near recreational cabins), though these areas still experience browse-induced aspen suppression. This should not suggest, however, that cattle have no impact on aspen recruitment (Figure 2); additionally, cattle are influencing broader floral diversity within Pando .
Pando is paradoxical: putatively earth’s largest organism, it is small as conservation challenges go. As an exemplar, however, it portends pathways for aspen diversity and resilience globally. In turn, circumboreal aspen forests contain ‘mega-conservation’ potential as these keystone forests support hundreds of dependent species . Given that ungulate populations, and to a degree movement, are controlled by land- and wildlife- managers, policy changes are needed to sustain Pando as well as aspen writ large. With high biodiversity value and visibility, wise choices backed by credible monitoring are required.
Current short-term strategies favor system control over more lasting process-based restoration [8,36]. Aesthetically, visitors may be disappointed to see an illustrious native forest fenced; perhaps symbolizing nature in captivity or managerial expediency over ecological integrity. Temporary secession of livestock grazing and iterative mule deer culling and/or dispersal will require multiagency coordination—no small request, but necessary for prioritizing long-term process-, rather than control-, based conservation. Current browsing pressure, alongside increasing human traffic, forecasts a bleak future for Pando. An adaptive monitoring approach  paired with greater stewardship agility offers a robust framework for a resilient Pando. Such recommended actions are not only needed for prolonging Pando’s status as the ‘World’s Largest Organism,’ but this prescription has implications for conservation writ large; lessons here may inform aspen resilience and biodiversity globally.