Introduction
As a first approximation, ecologists often relate performance of plants or animals to environmental conditions in that time and place. Nonetheless, many ecological studies have highlighted the importance of past environmental conditions for current performance of individuals and populations. Such carry-over effects, which range from inter-life stage to inter-generational impacts, are well documented in life history literature (Pechenik 2006; Harrison et al. 2011; O’Connoret al. 2014; Liz & Ruiz-Herrera 2016). For example, the nutritional environment experienced in juvenile stages of development in animals can influence development times (Pechenik 2006) and aspects of adult fitness including body size (e.g. Pereboom et al. 2003) and fecundity (e.g. Lindström 1999). In such cases, changes in individual quality (vital rates) can have lasting demographic consequences. Carry-over effects can be an important component of long-term population cycles (Beckerman et al. 2002), as well as responses to environmental stochasticity (Sabo & Post 2008) and recovery from extreme resource pulses or dearth (Gratton & Denno 2003; Yang et al. 2008).
Effects of past conditions on future population dynamics have been particularly well studied in stage-structured populations (Ezardet al. 2010; Stott et al. 2011). In many populations, vital rates (such as individual growth, survival, and reproduction) depend on the size of individual plants or animals. If these rates are constant through time, the population reaches a stable size distribution, and grows (or declines) at a constant rate associated with that size distribution. In fluctuating environments, the population growth rate in a particular year depends approximately equally on the vital rates individuals experience in that year, and the size distribution of individuals at the beginning of that year (Ellis & Crone 2013; McDonald et al. 2016). In response to extreme changes in the environment, changes in size distribution can create persistent differences in abundance and transient multi-year changes in population growth rates, even after vital rates have returned to average levels (Gamelon et al. 2014). For example, a resource pulse might enable many individuals to grow faster during that pulse, leading to more large individuals and higher population growth rates for many years after the pulse is over. The increase in growth rate occurs simply because the population (after the pulse) started with more large individuals. Similarly, a resource dearth might reduce individual size, leading to a lack of large individuals and slower population growth rates after resources return. In structured population projection models, “transient dynamics” refer to the period between when a population’s size distribution is perturbed, and when it returns to the size distribution determined by the current environmental conditions. During the past two decades, a large body of literature has focused on how to quantify transient dynamics using simulation models (starting with Neubert & Caswell 1997; see reviews by (Ezard et al. 2010; Stottet al. 2011). However, few studies have tested these predictions with experiments (see Tenhumberg et al. 2009 for a notable exception).
We use bumble bees (Bombus vosnesnesnkii ) as a model system for understanding how current and historical environmental conditions affect population dynamics. Within-colony dynamics of bumble bees are useful for exploring population dynamics for several reasons. Bumble bees are annual eusocial insects with colonies that grow via the production of successive overlapping cohorts (generations) of worker bees. Colonies typically grow exponentially, often producing ~ 5-10 (or more) worker cohorts during its development. After this growth phase, the colony stops making new workers and switches to reproduction (i.e. production of males and new queens, (Duchateau & Velthuis 1988; Müller & Schmid‐Hempel 1992; Goulson 2010; Crone & Williams 2016). Colonies persist for several months; however, individual foragers live 1-4 weeks (Rodd et al. 1980; Goldblatt & Fell 2011; Malfi et al.2019) during which time they provide the resources to rear subsequent cohorts of workers (Alford 1975; Goulson 2010). Therefore, a single bumble bee colony can be considered as an exponentially growing population of workers, with the advantage that multi-cohort dynamics can be studied in nature in a single growing season.
Bumble bee colonies are also a tractable system for studying environmental drivers of population dynamics because it is straightforward to manipulate a known environmental driver of colony and population size: forage. Growth of bumble bee colonies is at least partly food-limited (Pelletier & McNeil 2003). Larvae require pollen and nectar to develop into adults, and adult size is affected by the amount of food they receive as larvae (Couvillon & Dornhaus 2009; Malfiet al. 2019). Adults require nectar to fuel various colony functions, including foraging itself. Colonies have a limited capacity to store food resources (Goulson 2010); as a result, the spatiotemporal availability of flowering plants strongly regulates colony growth and the size of the intra-season worker population (Westphal et al.2009; Rundlöf et al. 2014; Crone & Williams 2016; Kämperet al. 2016; Spiesman et al. 2017; Rundlöf & Lundin 2019). Therefore, it seems likely that short-term changes in forage resources would affect worker size distribution, leading to transient effects on population growth rate. Other carry-over effects (e.g., persistent effects on size-based survival or foraging ability of workers) are also possible (Malfi et al. 2019). In addition its feasibility as a model system, understanding the effects of forage resources on bumble bee colony growth and long-term population dynamics is of immediate practical importance. Bumble bees are important pollinators in natural and crop systems (Corbet et al. 1991; Thomson & Goodell 2001; Kremen et al. 2002; Kleijn et al.2015), multiple bumble bee species are in decline (Bommarco et al. 2012; Colla et al. 2012; Kerr et al. 2015; Woodet al. 2019), and millions of dollars are invested annually in planting forage resources to support bumble bees and other bee populations (USDA Farm Services Agency 2019). Revealing the mechanistic connections between bumble bee colony and population dynamics and their resource environment is important for conservation efforts (Roulston & Goodell 2011).
In this study, we experimentally varied the amount and timing of food available to free-foraging colonies of B. vosnesenskii and measured the effects on within-colony population dynamics. We supplemented colonies with abundant pollen and nectar resources for a limited duration (20 days), either earlier or later in the growth phase of colony development, simulating two resource environments that differed in their seasonal patterns of food availability. These supplementation treatments reflect the type of variation reported in real-world landscapes, with a resource “pulse” (e.g., natural resources or a mass-flowering crop) occurring either early or late in the season (Westphal et al. 2009; Williams et al. 2012; Rundlöf et al. 2014). Thus, we refer to our supplementation and non-supplementation periods as “pulse” and “off-pulse” periods, respectively. Using this experimental design, we evaluated the impacts of varying resources levels and their timing on colony growth and reproduction, specifically: (1) whether colony growth rates during the pulse differed depending on pulse timing relative to colony development (early vs. late); (2) whether the growth rate during the off-pulse period (i.e. ambient resources only) depended on whether this period came before or after the pulse; (3) whether early-pulse and late-pulse colonies achieved similar peak sizes by the end of the colony growth phase; (4) whether the timing of the pulse affected the switch point from growth to decline and reproduction; and (5) whether early-pulse and late-pulse colonies differed in reproductive success.