Breaking out of the comfort zone: El Niño-Southern Oscillation as a driver of trophic flows in a benthic consumer of the Humboldt Current ecosystem

Abstract

The trophic flow of a species is considered a characteristic trait reflecting its trophic position and function in the ecosystem and its interaction with the environment. However, climate patterns are changing and we ignore how patterns of trophic flow are being affected. In the Humboldt Current ecosystem, arguably one of the most productive marine systems, El Niño-Southern Oscillation is the main source of interannual and longer-term variability. To assess the effect of this variability on trophic flow we built a 16-year series of mass-specific somatic production rate (P/B) of the Peruvian scallop (Argopecten purpuratus), a species belonging to a former tropical fauna that thrived in this cold ecosystem. A strong increase of the P/B ratio of this species was observed during nutrient-poor, warmer water conditions typical of El Niño, owing to the massive recruitment of fast-growing juvenile scallops. Trophic ecology theory predicts that when primary production is nutrient limited, the trophic flow of organisms occupying low trophic levels should be constrained (bottom-up control). For former tropical fauna thriving in cold, productive upwelling coastal zones, a short time of low food conditions but warm waters during El Niño could be sufficient to waken their ancestral biological features and display massive proliferations.

1. Introduction

All living organisms require matter to build their bodies and energy to perform their activities. Fluxes of energy and matter are intimately bound and result from organismal, population, and community processes and their link with the abiotic environment. By quantifying the efficiency of energy transfer between trophic levels, Lindeman [1] provided the conceptual basis to study the energetics of food webs and catalysed our current understanding of the biological basis of productivity in terrestrial and aquatic ecosystems. However, an array of pervasive human influences is causing dramatic changes that greatly influence principal modes of climate variability (e.g. El Niño-Southern Oscillation, the North Atlantic Oscillation) and hence patterns of ecosystem productivity at a global scale [2]. Therefore, there is a need to understand the biological basis of changes in the trophic flow of energy and matter, now under a highly modified climate.

Quantifying and understanding changes in the flow of energy and matter through trophic compartments, which may be attributed to climate change, is an elusive task. Processes occurring at the population level are directly linked to changes at the organismal level. In turn, populations are structured, meaning that individuals cannot be treated as identical; they differ in size, age, sex, reproductive condition, etc., and all these differences ultimately determine birth and mortality rates, hence the flow of energy and matter. Therefore, long-term investment and time-consuming investigations are needed if we want to address such questions [3–5]. For heterotrophs, the quantity of matter of its energy budget potentially available as food (biomass) for the next trophic level is represented by somatic production. In turn, the annual production to biomass ratio (P/B) in a population indicates the turnover rate, and thus the speed of the biomass regeneration for a determined trophic compartment. Early researchers observed that P/B is empirically related to some life-history traits (notably mortality, lifespan, and body mass) and formulated models describing these relationships, to easily predict P/B (see Dolbeth et al. 2012 [6] for an overview). However, predictions based on empirical models are not accurate [7] and depend on assumptions that are rarely met in nature, particularly that of a ‘steady state population’, in which mean biomass and size structure remain constant year to year [6,8] 2012). When populations are unsteady and asynchronous, size class-based methods, specifically the Mass Specific Growth Rate Method [7] provides a more integrative and robust tool to estimate P/B. It provides a measure of population fitness, as it integrates population processes (body mass, reproduction, recruitment, growth rate, survivorship, lifespan, size structure), biotic interactions, and the environmental conditions in which populations evolve [6]. Therefore, assessing changes in P/B and determining the organismal and population processes explaining these is the most comprehensive way to measure population function in the ecosystem and assess responses to environmental stress [9].

El Niño-Southern Oscillation (ENSO cycle), with its warm El Niño phase (EN) and cold La Niña phase (LN) represent the strongest signal in the interannual variation of the global ocean-atmosphere system and, although uncertainties persist, climate models indicate that ENSO strength and frequency will be increased by future greenhouse warming [10–12]. The Peruvian-Chilean Humboldt Current Upwelling System (HCS) is probably the most heavily affected ecosystem during ENSO: the system is characterized by a predominant northward flow of surface subantarctic waters by the Humboldt Current and the strong, wind-driven upwelling of cool nutrient-rich subsurface waters of equatorial origin. As a result, the HCS is a rather cold system with a low temperature seasonality compared to other coastal areas at the same latitude range [13–16]. However, positive and negative thermal anomalies that may reach up to 10°C are brought about by the ENSO cycle [14,15]. But not only temperature is modified; species inhabiting the HCS are faced with a complex chain of concomitant oceanographic changes, including modified levels of dissolved oxygen, changes in the position of the thermocline and nutrient flux, and changes in the upwelling intensity and salinity, with dramatic and widespread biological responses such as bathymetric and latitudinal migrations, invasion by warm-water species, behavioural alterations, reproductive and recruitment failures, and mass mortalities or sudden increases in local populations (e.g. [14,16–18]).

Within the group of species displaying clear responses to the ENSO cycle in the HCS, the Peruvian scallop, Argopecten purpuratus (Lamarck 1819) is an emblematic one. For instance, annual landings of this species, one of the most important for Peruvian fisheries, increased from a few hundred tons to about 25 000 t after the warm EN 1982–1983. This promoted an ephemeral multimillion dollar business for 3 years, after which scallop divers had to readjust back to lower catches and income levels [19]. These ‘positive responses’ (from an anthropocentric perspective) are held by species from tropical origin that maintained warm-water characteristics during evolution in cold waters of the HCS [20–22]. Population outbreaks of A. purpuratus during exceptionally strong EN phases have been attributed to the enhanced reproduction process and subsequent recruitment: faster recovery between spawning events, increased larval production, increased settlement of larvae, and increased survival rate of juveniles [19,20,23,24]. However, we lack long-term and detailed data to fully understand responses to EN and LN phases of different magnitudes and how this variability influences both organismal (body condition, reproduction, individual growth) and population parameters (abundance, biomass, somatic production), which ultimately determine changes in the speed of biomass regeneration for this important trophic compartment.

Here, we present results from a 16-year study to monitor population variability of A. purpuratus in Bahía Independencia, Paracas National Reserve, Peru. The study aimed to evaluate the effects of climate variability related to the EN-LN cycle on the processes regulating the P/B ratio of the A. purpuratus population.

2. Material and methods

(a). Study area and sampling

The sampling was performed in La Pampa (figure 1), Bahía Independencia, central Peru, with monthly measurements from March 1992 to November 2007 (with a short gap between April and May 2001). La Pampa holds one of the most important populations of A. purpuratus in the country and the largest in the Bay. The population is distributed along a 3.65 km² area varying in depth (10–20 m) (figure 1). This population was chosen for its greater distance and depth compared to other populations, that were subjected to higher fishing pressure (MA Solis 1995, unpublished data). The sampling area was mapped and divided into 30 m² cells, resulting in a 4 055 cell-grid. A quadrat (1 m²) was used to take monthly samples by scuba diving in 10 randomly chosen cells within the grid. All the resulting animals in each quadrat were counted and stored in plastic bags for further laboratory processing. Samples in these quadrats were called quantitative samples and were used to assess changes in abundance, biomass, and length-frequency distributions. Additional samples were taken from the same 10 cells: 40–80 scallops in three size ranges, i.e. small, medium, and large, were picked by hand, to obtain equivalent numbers of animals through all body size ranges available per month. This was called the qualitative sample and was used to assess changes in the length–mass relationship.

Figure 1.

Location of the study area. Bahía Independencia, Peru. Dotted line shows the distribution of the scallop population in La Pampa.

(b). Environmental parameters

During El Niño-La Niña cycles, sea life in Peruvian coastal waters is affected by both local and remote factors, through teleconnections. Therefore, we included variables describing environmental variability at macro, regional, and local spatial scales. The multivariate ENSO index (MEI), which combines multiple meteorological and oceanographic factors in a single index was used as a descriptor of ocean-atmosphere weather variability in the tropical Pacific (data available from: http://www.esrl.noaa.gov/psd/enso/mei/table.html). El Niño-Coastal Index (ENCI), an index based on sea surface temperature anomalies in the El Niño 1 + 2 Region [25], was taken as a descriptor of interannual anomalies of sea surface temperature and upwelling strength, and thus nutrient availability, at a regional scale (i.e. Peruvian coastal waters). Moreover, daily measurements of sea surface temperature (SST) taken at a fixed station (13°42′27″ S; 76°13′14.88″ W) were provided by the Instituto del Mar del Perú; salinity (Sal) in surface waters for the Pisco Region was taken from Morón [26]. Finally, in situ measurements of sea surface temperature (isSST), sea bottom temperature (isSBT), and dissolved oxygen (isDO) were taken monthly at each sampling point using a SBE 19 CTD profiler.

(c). Annual length–mass relationship

Each animal in monthly qualitative samples was rinsed and cleaned. The anterior-posterior shell length (SL) was registered to the nearest 0.1 mm using a Vernier caliper (0.05 mm precision). Soft parts were removed and dried in an oven at 90°C to constant mass, and the shell-free dry mass was determined. Finally, ash-free dry mass (AFDM) was obtained by the ignition of dried mass at 600°C for 3 h in a furnace. These monthly data were then pooled for each year (January to December) and the length (SL)–mass (AFDM) relationship was modelled using:

2.1

were ‘a’ is the intercept and b is the curvature parameter of the power function.

(d). Reproductive cycle

Additionally, the length–mass relationship (equation (2.1)) was used to analyse monthly changes in body mass. Monthly changes in the length–mass relationship can be used to assess the reproductive cycle [16,27]: an increase in body mass through time may reflect a gonad production phase, while a sudden decrease in body mass between successive months may indicate a spawning event. However, these interpretations may be misleading as weight increases or decreases might also be caused by feeding conditions and thus be independent of the reproductive cycle. Therefore, changes in body mass are only a proxy of reproduction activity and can only be interpreted convincingly in combination with other population parameters. To easily describe the body mass cycle, the AFDM of a standard (hypothetic) individual of 65 mm was plotted against time.

(e). Individual growth

The TropFishR package [28] implemented in R [29] was used to estimate growth parameters for each year. For this, a modified version of the von Bertalanffy growth function VBGF, which accounts for oscillating growth, was used:

2.2

where L_t is the total length at age t; L_∞ the asymptotic total length, K the growth coefficient (y⁻¹); t₀ the total length for the hypothetical age t = 0; t_s = WP + 0.5, (WP being the time of year where the biggest delay in growth occurred) and C is the intensity of the growth oscillation. The ELEFAN_SA optimization procedure (electronic length-frequency analysis, simulated annealing [30]) was used to search over all parameters simultaneously. Instead of t₀, TropFishR return the parameter ‘t_anchor’, which describes the fraction of the year where yearly repeating growth curves cross length equal to zero. The ELEFAN_SA procedure requires confinement in the search space for growth parameter estimation. This was performed with the Powell–Wetherall method, also included in TropFishR, which allows a preliminary estimation of L_∞, thus restricting the search space range. Alternatively, the maximum length in the data or the maximum length class was used to restrict the search space range. The maximum time of the optimization procedure was set at 5 min. To assess growth variability among years, we used the Φ′ value [31], which represents an index for growth performance using K and L_∞from equation (2.2):

2.3

(f). Somatic production and P/B ratio

The mass-specific growth rate method [7] was used to calculate annual somatic production P (g AFDM m⁻² year⁻¹) for each year. This method uses data on mean abundance (ind m⁻²), length-frequency distributions from quantitative samples, the VBGF parameters, and the length–mass relationship for each year:

2.4

where N_i is the mean annual abundance in length class I; W_i the mean body mass in length class i, and G_i the mass-specific growth rate:

2.5

where b is the exponent of the annual length–mass relationship (equation (2.1)); K and L_∞ the parameters of the VBGF (equation (2.2)) and L_i the mean length in length class i. The mean annual biomass B (g AFDM m⁻²) of the population was estimated as:

2.6

Finally, the production-to-biomass ratio (P/B) was calculated as the ratio between P (equation (2.4)) and B (equation (2.6))

(g). Data analyses

Although the various indices and environmental parameters included in this study describe variability at different spatial scales (i.e. in situ, local, regional), it is expected that they concurrently affect the turnover rate (P/B) of A. purpuratus and could be highly correlated. This interdependence among environmental factors (i.e. multicollinearity), creates problems in the interpretation of a least-squares multiple regression analysis [32]. To address this problem, the primary data on environmental variability were first analysed by Principal Components Analysis (PCA). Data were normalized (subtracting the mean and dividing by the standard deviation for each variable) to account for the different scales among variables before calculating Euclidean distances. Moreover, a Draftsman plot was used to check that data are roughly symmetrically distributed across the range of each variable, thus achieving approximate (multivariate) normality underlying PCA analysis [33]. This procedure reduces the volume of data to a handful of orthogonal (i.e. mutually independent) principal components that account for a proportion of the variation in the environmental variables. Vector plots were superimposed on the bi-dimensional PCA plots and the length and direction of vectors were evaluated to assess the contribution of each environmental variable to each PC. Therefore, we assessed the relationship between P/B of A. purpuratus and the five most important principal components (PCA scores), using a forward stepwise multiple regression analysis. Compared to traditional approaches looking for correlations with multiple environmental variables but not considering collinearity, this approach that includes multicollinearity among independent variables is more robust [34].

Finally, differences in individual mass, monthly mean abundance, and mean shell length between neutral, EN, and LN periods were assessed using Kruskal–Wallis tests by ranks, because data did not meet the assumptions of the parametric test. When significant differences were found, we performed pairwise comparisons of mean ranks for all groups. A significance level of α = 0.05 was chosen for all the tests performed. Kruskal–Wallis tests and multiple regression analyses were performed using Statistica 10.0 for Windows (StatSoft, Inc 1984–2011, USA).

3. Results

(a). Changes in P/B

The trophic flow through the population of A. purpuratus, as represented by annual P/B ratio, varied nearly one order of magnitude (figure 2a), with the lowest ratio observed in 1996 (0.19) and the highest in 1997 (1.08). Generally, high ratios were observed during El Niño years and low ratios during La Niña years. The PCA used to evaluate the environmental factors driving the observed changes of the P/B of A. purpuratus showed that PC1 accounted for approximately 70% of the total variance in the dataset of environmental factors and PC2 and PC3 accounted for most of the remaining variability (table 1). The multiple regression analysis indicated that only PC1 was significant as a predictor of changes in P/B of A. purpuratus (table 2).

Figure 2.

Interannual changes in population parameters of Argopecten purpuratus in Bahía Independencia, Peru. (a) Production to Biomass (P/B) ratio; (b) annual changes in growth performance and somatic production. Years highlighted with a black star represent severe El Niño anomalies (Multivariate ENSO Index ≥ 2), whereas years highlighted with a white diamond denote years with strong La Niña anomalies (MEI ≤ −1).

Table 1.

Results of the Principal Components Analysis of environmental factors potentially affecting the P/B rate of Argopecten purpuratus. Coefficients in the linear combinations of variables making up the Principal Components (PC) are given for each variable.

PC	% variation	cumulative % variation	variable coefficient
			MEI	ENCI	SST	isSST	isSBT	isDO	Sal
PC1	67.7	67.7	−0.404	−0.43	−0.428	−0.423	−0.436	−0.195	−0.248
PC2	13.4	81.1	−0.249	−0.071	0.093	−0.092	0.078	0.899	−0.318
PC3	11.1	92.1	−0.045	−0.178	−0.018	−0.329	−0.068	0.261	0.886
PC4	4.2	96.3	−0.674	0.289	−0.421	0.276	0.419	−0.068	0.171
PC5	1.8	98.1	0.389	0.559	−0.63	−0.114	−0.276	0.225	−0.011

Table 2.

Multiple regression analysis to test the influence of environmental variability (reduced to five PCA scores) on the Production to Biomass ratio of Argopecten purpuratus. Significant effects are given in italic.

variables	R	R2	R change	F	significance
PC1	0.879	0.774	0.774	47.841	<0.001
PC1 + PC2	0.894	0.799	0.026	1.682	0.217
PC1 + PC2 + PC3	0.903	0.815	0.0156	1.018	0.332

The biplot of the two first PCs (figure 3a) showed strong interannual differences, mostly for the PC1 axis. Strongest differences in the PC1 axis occurred between warm El Niño years (1992, 1997, 1998) and cool La Niña years (1996, 1999, 2000, 2007), as depicted by the multivariate ENSO index. Vector plots indicated that these differences were best explained by changes in temperature-related factors and indices describing ENSO variability. In fact, the best fit of P/B ratio with a single variable was found with the ENCI (figure 3b). By contrast, smaller variations in the PC2 axis were mostly related to local changes in dissolved oxygen and salinity, not necessarily connected with the ENSO cycle.

Figure 3.

(a) Biplot of two principal components (PC1 and PC3) after the PCA analysis of environmental variability and the superimposed vectors (grey lines) of environmental factors (Sal: salinity in surface waters averaged for the Pisco region; isDO: in situ-dissolved oxygen; SST: sea surface temperature reported by IMARPE in a fixed station near the study area; MEI: multivariate ENSO index; isSBT: in situ-sea bottom temperature ENCI: El Niño-Coastal Index; isSST: in situ-sea surface temperature). Black stars denote years with severe El Niño anomalies (MEI ≥ 2); white diamonds denote years with strong La Niña anomalies (MEI ≤ −1), and grey squares represent neutral periods. (b) Correlation between the P/B ratio of Argopecten purpuratus and the El Niño-Coastal Index.

(b). Demographic processes underlying changes in P/B

Figure 2b shows variability in individual growth performance and concurrent changes in population somatic production. As expected, variability in these processes was generally coupled during the period 1992–1996. However, this coupling changed afterwards: during the 1997 El Niño, individual growth performance increased strongly, but population somatic production remained low. The opposite was observed during El Niño 1998; growth performance decreased, while population somatic production increased by more than one order of magnitude. Thereafter, growth performance fluctuated strongly and somatic production remained low, with only little changes.

Figure 4 shows monthly changes in individual mass of a standard (65 mm SL) scallop and the mean abundance in the population (figure 4a), and the mean shell size of the scallops in the population, with percentiles, maximum and minimum as descriptors of dispersion in body size (figure 4b). Firstly, while individual mass fluctuated strongly (from approximately 0.5 to 5 g AFDM), there was not a consistent seasonal pattern of increase and sudden decrease that would suggest regular gametogenesis-spawning seasons. Likewise, there were huge changes in mean abundance of the scallops (approx. 0–250 ind m²) and their mean shell length (25–85 mm), but regular seasonal changes that would reveal seasonal recruitment pulses after spawning events were not shown.

Figure 4.

Monthly changes in population parameters of Argopecten purpuratus in Bahía Independencia between March 1992 and November 2007; (a) mean abundance (black line) and individual mass (grey line) calculated for a standard individual (65 mm shell length) from the monthly length-mass relationship; (b) boxplot showing the distribution of body sizes (minimum, 25th percentile, mean, 75th percentile, and maximum shell length). Grey panels show periods of severe El Niño (EN) anomalies (Multivariate ENSO Index ≥ 2) and strong La Niña (LN) anomalies (Multivariate ENSO Index ≤ −1).

The variability depicted in figure 4 was mostly related to the El Niño-La Niña cycle. Individual mass was significantly lower during EN 1997–1998 than during LN or neutral periods (table 3, figure 5). Mean abundance also changed (dropped) significantly during EN in comparison with LN or neutral periods (table 3, figure 5). This result was counterintuitive, if we observe figure 4, but it can be explained by the fact that pulses in abundance occurred at the end of EN 1992 and EN 1997–1998, and remained for a few months, during neutral or La Niña conditions, thus suggesting lagged responses. The observed pulses in mean abundance coincided with an increased abundance of small animals in the population (figure 4b), suggesting strong spawning and recruitment pulses, mainly during El Niño. This was supported by the Kruskal–Wallis test showing a significant drop in mean shell length during EN in comparison with LN or neutral periods (table 3, figure 5).

Table 3.

Kruskal–Wallis ANOVA by Ranks testing differences in population parameters of Argopecten purpuratus between neutral, El Niño, and La Niña periods. Significant effects are given in italics.

parameter	N	Df	H	p
individual mass	225	2	15.168	<0.001
mean abundance	219	2	7.476	0.024
mean shell length	225	2	9.338	0.009

Figure 5.

Boxplots showing differences in abundance, individual mass, and shell length of Argopecten purpuratus between neutral, El Niño, and La Niña periods. Central markers represent the mean, box limits: ± s.e., whiskers: ± s.d. Different capital letters in boxplots indicate significant differences (p < 0.05) between periods after paired comparisons. (Online version in colour.)

4. Discussion

(a). Changes in P/B and potential environmental controls

The annual P/B ratio of a species has been assumed to be a distinctive, species-specific flow rate reflecting its trophic position and function within the ecosystem (e.g. [3,35]). This view fired efforts to describe spatial patterns of P/B for taxonomic groups or trophic guilds in diverse ecosystems [36]. Hence, P/B has been used as an indicator of ecological performance that is little or indirectly affected by changes in the environmental setting, unless they affect life-history traits like lifespan or voltinism [36–38]. Although it was early detected and recognized that the constancy of P/B was a convenient but inaccurate notion [39], only the availability of long-term data and the magnitude of climate and human-induced impacts is starting to show the bounciness of this crucial biological rate.

The interannual variability observed for A. purpuratus is large for a benthic bivalve inhabiting a rather stable environment. A compilation of available data regarding temporal changes in P/B ratios for molluscs in the literature (figure 6) shows that large intraspecific variability in P/B ratio becomes clear when highly contrasting climatic periods are compared or long-term data are available to assess how environmental, anthropogenic, or biotic sources of stress affect the speed of biomass regeneration for specific trophic elements. Indeed, responses to stressing factors can be site-specific: changes in annual P/B of the bivalve Cerastoderma edule (figure 6g,h) and the gastropod Peringia ulvae (figure 6i,j) were larger in a sandflat than those observed in a Zostera bed area.

Figure 6.

Compilation of studies reporting on interannual changes in P/B ratio of molluscs facing different environmental and mixed stressors. Black box represents the mean P/B ratio and error bars the range observed during the years of study (number of years in parentheses). ZB: Zostera bed; SA: sandflat area. References: a: this study; b: [40]; c: [41]; d: [42]; e: [29]; f: [43]; g–m: [5].

Ecological theory predicts that the transfer of matter and energy between trophic levels (i.e. ecosystem function) is regulated by an interplay of bottom-up and top-down processes [44]. Top-down control mechanisms, mainly predation, are more commonly observed at trophic levels near to top predators while nutrient supply to primary producers, i.e. bottom-up control, would be most apparent at low trophic levels. A decreased (increased) nutrient supply and thus primary production during EN (LN) episodes in the HCS is an established fact (e.g. [15]). Therefore, a decreased P/B ratio under EN and an increased P/B ratio under LN might be expected for this filter-feeder occupying a low trophic level which should be controlled by bottom-up mechanisms. Observed results show the opposite and can only be understood by considering the biogeography of the coastal biota in the HCS. In short, the establishment of the HCS during the mid-to-late Miocene involved both, the northward advance of a sub-Antarctic biota and the northward retraction of a former tropical/subtropical biota [45]. A. purpuratus represents one of the relicts of formerly warm water fauna that thrived into a cool, nutrient-rich system, but retained ancestral biological features that are favoured when ‘tropical’ EN conditions appear [20,46], with increased growth efficiency under warm waters as a key factor. This, and the modified conditions of competition and predation during EN (see below), apparently overshadow the decreased food availability in the water column. These findings challenge the intuitive view that trophic flow should be decreased in a highly modified habitat.

According to our analysis, long-term interannual changes in P/B ratio of A. purpuratus were mainly driven by ENSO-related environmental variability operating at regional spatial scales. This is an interesting finding adding to the discussion of the scale dependence of oceanographic processes and the analysis of ecological consequences. Coastal upwelling is regarded as the main mesoscale oceanographic process affecting the dynamics and the spatial structure of coastal ecosystems in the HCS [17]. Local upwelling-favourable winds and geomorphology determine flow structures (e.g. filaments, squirts, meanders, and eddies) and upwelling shadows that largely determine the dynamics of nutrient enrichment, primary production, larval retention and recruitment, and ultimately the structure and function of the coastal ecosystem [17,47,48]. Hence, our results suggest that i) large-scale oceanographic processes related to ENSO can overshadow the influence of prevailing mesoscale coastal features of the HCS on the structure of populations (hypothesized by Thiel et al. [17]) and ii) effects of ENSO on population dynamics can only be reliably assessed when long-term data series are available. It is clear, however, that different species will have different responses to ENSO forcing.

(b). Organismal versus population processes underlying changes in P/B

Observed changes in organismal and population parameters did not show a consistent seasonal pattern. It seems reasonable to suggest that this is related to the small seasonality in oceanographic parameters, particularly upwelling intensity, which characterize the HCS [15,17]. In contrast, significant changes were observed when EN and LN periods were compared to periods of neutral conditions.

(c). El Niño phase

When the size structure and abundance in a population remain relatively stable, a change in individual growth performance is the main organismal feature determining the variability in population somatic production. This seemed to be the case for A. purpuratus between 1992 and 1996. During the EN 1997–1998, the strongest warming event on records during the twentieth century [49], the mean abundance increased temporarily by two orders of magnitude and the P/B ratio was the highest in the data series. This episode seemingly shifted the coupling between growth performance and somatic production, suggesting that population processes exerted an influence on production and on the P/B ratio that lasted until the end of our study. Monthly changes in individual mass, abundance, and mean shell length depicted in figure 5 can be used to further analyse this shift. Early in 1997, the population was dominated by old animals, which seemingly spawned (drop in body mass) and died or were removed by the fishery, giving place to a population dominated by young scallops (approx. 35 mm) in June-July. Interestingly, this spawning was not followed by a recruitment pulse and young scallops grew to barely reach their size at first maturity (35 mm; [50]), spawned, and presumably died. This second spawning lead to a massive recruitment pulse that resulted in a population with a high abundance (250 ind m⁻²), heavily dominated by juveniles (approx. 27 mm) in March 1998. By the end of the EN episode in June 1998 the abundance was still high and juveniles had quickly grown to adult size. Obviously, a population dominated by juveniles and young adults growing faster than normal under anomalously high temperature led to an increase in population production and hence the P/B ratio.

These findings partially support the idea of increased gonad activity, settlement and survival rate of juveniles during strong EN periods [19,20,23,24]. There were two spawning events during EN 1997–1998 but settlement and recruitment only occurred after the second. This emphasizes the fact that physical-biological processes affecting larvae and settlers are complex and subject to a high degree of stochasticity. Modelling efforts to understand changes in trophic flow structure related to EN in Bahía Independencia suggest that the increased abundance of A. purpuratus under lowered primary production can be partially explained through the replacement of usually abundant polychaetes by scallops as dominant secondary consumers [51].

(d). La Niña phase

Efforts to understand effects of ENSO in coastal ecosystems have been focused on the warming EN phase, while little is known in relation to the cooling LN phase. Being a rather cool coastal system with little seasonality, thermal anomalies during LN are smaller than those observed during EN, which have led to the simplistic assumption that effects should be mild and opposite of those under EN [52]. The P/B ratio during LN 1996 was the lowest observed, reflecting the dominance of older scallops, a poor growth performance, and the lack of recruiting pulses after several spawning events. However, the P/B ratio was high during the stronger and longer LN 1999–2000 episode. While growth performance was lower, the dominance of medium size scallops and later of juveniles after a strong recruitment pulse during 1999 resulted in high P/B ratios. If increases in recruitment of A. purpuratus during EN are assumed to be caused by decreased competition with filter feeder polychaetes and predation by crabs, that are negatively affected by EN [50], A. purpuratus populations could still have recovered during LN 1999–2000, but there is no evidence to support this conjecture.

Finally, fisheries must have played a role in the observed patterns of abundance and temporal distribution of shell size, if we take into account the fact that the huge abundance of scallops during EN 1997–1998 attracted fishermen from northern areas of the country to Independencia Bay [46]. This may particularly explain the dominance of small size and juvenile scallops between 1997 and 1999.

Data accessibility

Supporting data are freely available at: http://dx.doi.org/10.5061/dryad.1436s [53].

Authors' contributions

M.A.S. coordinated the monitory programme. J.M.R., and M.A.S. created the databases and conceptualized the study; J.M.R., A.S.P., and M.A.S. analysed the data; J.M.R., M.A.S., and M.B. drafted the first version of the article; all authors revised and approved the final version.

Competing interests

All authors declare that they have no financial or non-financial competing interests

Funding

M.A.S. was supported by several funding schemes including the United States Agency for International Development (USAID), the Alfred Wegener Institute for Polar and Marine Research (AWI), Germany, and the National University of San Marcos (Peru).